NX Nastran Users Guid

NX Nastran
User’s Guide
www.cadfamily.com EMail:[email protected]
The document is for study only,if tort to your rights,please inform us,we will delete
Proprietary & Restricted Rights Notice
 2005 UGS Corp. All Rights Reserved. This software and related documentation are proprietary to UGS
Corp. LIMITATIONS TO U.S. GOVERNMENT RIGHTS. UNPUBLISHED - RIGHTS RESERVED UNDER
THE COPYRIGHT LAWS OF THE UNITED STATES. This computer software and related computer software
documentation have been developed exclusively at private expense and are provided subject to the following
rights: If this computer software and computer software documentation qualify as “commercial items” (as
that term is defined in FAR 2.101), their use, duplication, or disclosure by the U.S. Government is subject to
the protections and restrictions as set forth in the UGS Corp. commercial license for the software and/or
documentation as prescribed in FAR 12.212 and FAR 27.405(b)(2)(i) (for civilian agencies) and in DFARS
227.7202-1(a) and DFARS 227.7202-3(a) (for the Department of Defense).,or any successor or similar
regulation, as applicable or as amended from time to time. If this computer software and computer
documentation do not qualify as “commercial items,” then they are “restricted computer software,” and are
provided with “restrictive rights,” and their use, duplication or disclosure by the U.S. Government is subject
to the protections and restrictions as set forth in FAR 27.404(b) and FAR 52-227-14 (for civilian agencies), and
DFARS 227.7203-5(c) and DFARS 252.227-7014 (for the Department of Defense), or any successor or similar
regulation as applicable or as amended from time to time. UGS Corp. , Suite 600 - 5800 Granite Parkway,
Plano, Texas 75024.
NASTRAN is a registered trademark of the National Aeronautics and Space Administration. NX Nastran is
an enhanced proprietary version developed and maintained by UGS Corp.
MSC is a registered trademark of MSC.Software Corporation. MSC.Nastran and MSC.Patran are trademarks
of MSC.Software Corporation.
All other trademarks are the property of their respective owners.
C O N T E N T S
NX Nastran User’s Guide
NX Nastran Reference ManualNX

Nastran User’s Guide
1
Overview of NX ■ General Overview of NX Nastran, 2
Nastran ❑ General Capabilities, 2
❑ Modular Structure and DMAP, 2
❑ Pre-Defined Solution Sequences, 3
❑ Using DMAP for Programmability, 3
■ Overview of the NX Nastran Analysis Process, 3

❑ Creating the Input File, 3
❑ Submitting the Input File for Analysis, 4
❑ Reviewing the Results, 5
2
Structure and ■ Overview of Specifying Input Data, 8
Syntax for Input ■ General Input File Syntax Rules, 8
Data
❑ Understanding Types of Input Data, 8
❑ Using Consistent Unit Systems, 9
❑ Specifying Valid Input File Characters, 9
■ Understanding NX Nastran Input Records, 10

❑ Understanding Free Field Format, 11
❑ Understanding Small Field Format, 12
❑ Understanding Large Field Format, 13
■ Specifying Continuations for Bulk Data, 14

❑ Manually Specifying a Continuation with Identifiers, 14
❑ Automatically Specifying a Continuation, 15
■ Replicating Bulk Data Entries, 17

❑ Replicating Continuation Entries, 17
❑ Replication with MSGMESH, 17
■ Inserting External Files with INCLUDE, 20

❑ Using INCLUDE in Different Input File Sections, 20
❑ INCLUDE Statement Limitations, 20
■ Including Comments in the Input File, 21
3
Understanding the ■ Overview of the NX Nastran Input File Sections, 24
Input Data File ❑ Example: Using NX Nastran Input Files to Model a Truss, 24
■ The NASTRAN Statement (Optional), 27

■ File Management Section (Optional), 27
■ Executive Control Section (Required), 28
❑ Executive Control Format, 28
❑ Order Dependence for Certain DMAP-Related Statements, 29
❑ Commonly Used Executive Control Statements, 29
■ Case Control Section (Required), 32

❑ Case Control Format, 32
❑ Using the SET Command to Limit Output, 38
■ Bulk Data Section, 40

❑ Bulk Data Section Format Rules, 41
❑ Sorting Bulk Data, 41
❑ Storing Bulk Data in the Database for Restarts, 42
❑ Understanding the Relationship Between Bulk Data Entries and Case
Control Commands, 42
■ Using Parameters, 43
❑ Parameters in the Superelement Solution Sequences, 43
4
NX Nastran Output ■ Overview of the NX Nastran Output Files, 48
Files ■ Understanding the .f06 File, 49
❑ Controlling Printed Output, 49
❑ Understanding the General Analysis Output, 53
❑ Understanding Analysis Type Specific Output, 55
■ Understanding the .f04 File, 59

❑ General Information, 60
❑ Summary of Physical File Information, 60
❑ Memory Allocation Map, 61
❑ Day Log, 62
❑ Understanding the I/O MB Column, 63
❑ User Information Messages 4157 and 6439, 63
❑ Memory and Disk Usage Statistics, 65
❑ Database Usage Statistics, 65
❑ Summary of Physical File I/O Activity, 66
❑ Other Optional Diagnostic Information, 67
❑ Comparing Different F04 Files with F04REPRT, 68
■ Understanding the .log File, 68

■ Understanding the .plt and .pch Files, 69
■ Understanding the .DBALL and .MASTER Files, 69
❑ Specifying Whether to Save .DBALL and .MASTER Files, 69
❑ Manually Deleting .DBALL and .MASTER Files, 70
■ Understanding .xdb and .op2 Files, 70
5
Defining Grid ■ Overview of Grid Points and Scalar Points, 72
Points and Scalar ■ Defining Grid Points, 72
Points
❑ Format of the GRID Entry, 74
❑ Understanding the GRDSET Entry, 79
■ Specifying Grid Point Properties, 79

❑ Defining Mass, Damping, and Stiffness through Direct Input
Matrices, 79
❑ Defining Mass through the Concentrated Mass Element, 80
❑ Defining Temperatures at Grid Points, 80
❑ Assigning a Singularity Normal to a Degree-of-Freedom, 81
■ Understanding Scalar Points, 81

❑ Comparing Grid Points to Scalar Points, 81
❑ Defining Scalar Points, 81
❑ Using Scalar Points as Extra Points for Dynamics, 82
6
Understanding ■ Overview of Coordinate Systems in NX Nastran, 84
Coordinate ■ Understanding the Basic Coordinate System, 84
Systems
❑ ID of the Basic Coordinate System, 84
❑ Comparing the Basic and Global Coordinate Systems, 84
■ Defining a Local Coordinate System, 87

❑ Bulk Data Entries for Defining Coordinate Systems, 88
❑ Defining a Rectangular Coordinate System, 90
❑ Defining Cylindrical Coordinate Systems, 94
❑ Example: Using Cylindrical and Spherical Coordinate Systems, 100
■ Understanding the Element and Material Coordinate System, 101

❑ Special Considerations for Grid Points on the Polar Axis, 102
❑ Defining Element Offsets, 102
■ Other Special Coordinate System Entries, 102

❑ CORD3G, 102
❑ GMCORD, 103
7
Material Properties ■ Introduction, 106
❑ Elements and Material Input Data Types, 106
■Isotropic Materials, 107

❑ MAT1 Constituitive Relationships, 107
❑ Understanding the Isotropic Material Constants, 109
❑ MAT 1 Format, 110
❑ Example: Using MAT1 to Define a Structural Steel, 111
■ Orthotropic Materials, 112
❑ Understanding the MAT3 Entry, 112
❑ Understanding the MAT8 Entry, 114
■ Anisotropic Materials, 119

❑ Understanding MAT2, 120
■ Material Properties for Heat Transfer, 133

❑ Understanding MAT4 and MATT4, 134
❑ Understanding MAT5 and MATT5, 135
❑ Understanding RADM and RADMT, 135
■ Material Properties for Fluid Analysis, 136

❑ Understanding MFLUID, 136
■ Material Properties for Nonlinear Analysis, 136

❑ Understanding MATG, 137
❑ Understanding MATHP, 137
8
Applied Loads ■ Overview of Load Sets, 142
■ Defining and Combining Subcases for Loading, 142
❑ Example: Creating Two Load Subcases for the Truss, 142
❑ Using REPCASE, 145
❑ Using SUBCOM to Combine Subcases, 146
■ Defining Static Loads, 148

■ Point Loads, 149
❑ FORCE, FORCE1, FORCE2, 149
■ Distributed Loads, 156

❑ PLOAD, 156
❑ PLOAD1, 158
❑ PLOAD2, 167
❑ PLOAD4, 168
❑ PRESAX and PLOADX1, 172
■ Inertia Loads (Acceleration Loads), 172

❑ Using GRAV, 173
❑ Using RFORCE, 176
❑ Defining Mass in Your Model, 178
■ Thermal Loads, 182

■ Enforced Element Deformation Loads, 189
❑ DEFORM Example, 190
■ Loads Due to Enforced Motion, 194

❑ Using Residual Vectors (Modal Augmentation) in Modal Dynamic
Response Analysis, 196
■ Frequency-Dependent Loads, 197
■ Time-Dependent Loads, 198
■ Using Static Loading Data in Dynamic Analysis, 200
■ Defining Loads for Cyclic Symmetry Analysis, 201
■ Combining Loads, 202
❑ Using LOAD, 202
9
Constraints ■ Introduction to Constraints, 210
■ Single-Point Constraints, 210
❑ Defining Single Point Constraints in NX Nastran, 211
❑ Using GRID Entries, 212
❑ Using SPC and SPC1, 213
■ Automatically Applying Single-Point Constraints, 216

❑ Grid Point Singularity Processor, 216
❑ Understanding the Grid Point Singularity Table, 217
■ Enforced Displacements at Grid Points (SPCD, SPC), 220

■ Multipoint Constraints, 223
❑ Uses of Multi-point Constraints, 224
❑ Defining Multi-point Constraints in NX Nastran, 225
❑ Rules for Using Multi-point Constraints, 228
■ Rigid Body Supports, 229

■ Linear Contact Using Constraints, 232
❑ Introduction, 232
❑ Input, 233
❑ Output, 233
❑ Limitations, 235
❑ Examples, 235
10
Understanding ■ Overview of Sets and Matrix Operations, 244
Sets and Matrix ■ Understanding the Global Displacement Set and General Solution
Operations Process, 245
❑ Understanding Basic Partitioning Operations, 246
■ Static Condensation (Guyan Reduction), 248

❑ Using Static Condensation, 249
■ Sets for Dynamic Reduction, 257

■Sets for Aerodynamics, 259
■ Output Selection Via Set Specification, 259
11
Direct Matrix Input ■ Overview of Direct Matrix Input, 262
■ Using DMIG, 263
❑ DMIG Bulk Data User Interface, 264
❑ DMIG Case Control User Interface, 266
■ DMIG Examples, 266

❑ Stiffness Matrix Using DMIG, 267
❑ Mass Matrix Using DMIG, 273
❑ Load Matrix Using DMIG, 273
12
Element Data ■ Introduction to Element Data Recovery, 280
Recovery ■ Stress Recovery at Grid Points, 281
Resolved at
❑ Description of Method, 282
Grid Points
❑ Topological Method– Plate Elements and Solid Elements, 283
❑ Geometric Method – Surface Elements, 286
❑ Treatment of Exception Points – “BRANCHing”, 288
❑ User Information, 290
❑ Grid Point Stress Output Description, 291
❑ General Remarks, 293
■ Mesh Stress Discontinuities at Grid Points, 294
13
Solution ■ Understanding Solution Sequences, 302
Sequences ❑ Obtaining DMAP Listings for Solution Sequences, 305
❑ Using Alters to Modify Solution Sequences, 305
❑ Understanding Scheduled and Unscheduled Exits, 306
❑ Restart Procedures, 306
■ Data Processing and Matrix Operations by Functional Module, 308

❑ Geometry Processing in SubDMAP PHASEO, 308
❑ Matrix Assembly Operations in SubDMAP SEMG, 311
❑ Multipoint Constraint Operations in SubDMAPs SEKR and SEMR2 and
SEMRB, 313
❑ Single Point Constraint Operations in SubDMAPs SEKR, SEMR2 and
SEMRB, 316
❑ Constraint and Mechanism Problem Identification in
SubDMAP SEKR, 317
❑ Static Condensation in SubDMAPs SEKR and SEMR2, 321
❑ Static Solutions in SubDMAP SEKRRS, 324
❑ Static and Dynamic Load Generation, 328
❑ Dynamic Reduction and Component Mode Synthesis in SubDMAP
SEMR3, 331
❑ Formulation of Dynamic Equations in SubDMAP GMA, 334
❑ Data Recovery Operations in SubDMAP SEDISP, 338
❑ Data Recovery Operations in SubDMAP SEDRCVR, 344
■ Summary of Solution Sequence Operations, 345
❑ Static Solutions, 345
❑ Real Eigenvalue Analysis in SubDMAPs SEMR3 and MODERS, 353
❑ Complex Eigenvalue Analysis in SubDMAP CEIGRS, 367
❑ Frequency Response and Random Analysis in SubDMAP FREQRS, 368
❑ Transient Response in SubDMAPs SEDTRAN and SEMTRAN, 369
❑ Buckling Analysis in SubDMAP MODERS, 369
❑ Superelement Analysis, 371
❑ Nonlinear Heat Transfer Analysis, 387
14
Modeling ■ Introduction to Modeling Guidelines, 392
Guidelines ■ Choosing the Right Element, 392
❑ General Guidelines for Element Selection, 393
❑ Zero-Dimensional Elements, 393
❑ One-Dimensional Elements, 393
❑ Two-Dimensional Elements, 393
❑ Three-Dimensional Elements, 398
❑ R-Type Elements, 398
■ Determining Mesh Density, 399

❑ h-elements, 406
❑ p-elements, 406
■ Creating Mesh Transitions, 407

❑ Transition from a Coarse Mesh to a Fine Mesh, 407
❑ Mesh Transition Between Dissimilar Element Types, 417
❑ Shell-to-Solid Transition Element (RSSCON), 419
■ Reviewing Grid Point Stresses, 432

■ Defining Consistent Loading, 435
■ Symmetry, 437
❑ User Interface, 440
15
Model Verification ■ Introduction to Model Verification, 446
■Preprocessor Checks, 446
❑ Creating a Shrink Plot of the Elements, 446
❑ Hidden Line/Shaded Plots, 447
❑ Free Edge/Face, 448
❑ Zipper Effect, 450
❑ CBAR/CBEAM Orientation and Offset Check, 451
❑ Duplicate Grid Points/Elements, 451
❑ Properties/Material Plots, 451
❑ Using Consistent Normals, 451
■ Strain Energy Output, 454
■ Diagnostic Tools, 461
❑ Element Summary Output, 462
❑ Element Geometry Checks, 463
❑ Weight Center of Gravity and Moment of Inertia Check, 471
❑ Mechanisms and Singularities, 473
❑ Applied Loads Check, 476
❑ Verifying Reaction Loads, 478
❑ 1g Load, 480
❑ Output of Maximum, Minimum Grid Point-Based Responses, 480
❑ Performing an Unconstrained Equilibrium Check, 485
❑ Automatic Identification of Unintentional Grounding, 486
❑ Thermal Equilibrium Check, 488
❑ Grid Point Force Balance, 489
■ Stress Error Estimators, 494

❑ Grid Point Stress Discontinuities, 496
❑ Element Stress Discontinuities, 500
❑ Discussion of Error Measures, 503
■ Postprocessor Checks, 506

❑ Examining Displacements, 506
❑ Examining Stress Contours, 506
❑ Creating Grid Point Stress and Stress Discontinuity Plots, 510
❑ Viewing Element Strain Energy Plots, 511
16
Restarts ■ Introduction to Restarts, 514
■ Types of Restarts, 515
❑ Understanding Restart Terminology, 516
■ Read-Write Restarts, 516

❑ NASTRAN Statement Section, 516
❑ File Management Section (FMS), 517
❑ Executive Control Section, 518
❑ Case Control Section, 518
❑ Bulk Data Section, 518
■ Read-Only Restarts, 526

■ Mini-Database Restarts, 527
■ Determining the Version for a Restart, 527
17
Database ■ Introduction to Database Concepts, 532
Concepts
■ Understanding DBsets, 533
■ Delivery Database, 535
■ Deleting DBsets, 536
■ Database Autoassignment, 536
■ Database Project and Version, 537
■ Migrating Databases, 537
■ Database Archival, Compression, and Transfer, 538
❑ Database Archival and Retrieval, 538
❑ Database Compression, 539
❑ Transferring Databases Across Different Platforms, 540
■ Using the File Management Section with Databases, 541

❑ Using the INIT Statement, 541
❑ Using the ASSIGN Statement, 543
❑ Using the EXPAND Statement, 545
❑ Using the RESTART Statement, 546
❑ Using the DBCLEAN Statement, 547
❑ Using the DBDIR Statement, 548
❑ Using the INCLUDE Statement, 549
■ Guidelines for Large Problems, 550

❑ BUFFSIZE, 553
18
Inertia Relief in ■ Introduction to Inertia Relief, 558
Linear Static ■ Description of Inertia Relief Using PARAM,INREL,-1, 558
Analysis
■ Implementation of Inertia Relief Using PARAM,INREL,-1, 559
■ Automatic Inertia Relief, 571
❑ Limitations for the Automatic Inertia Relief Method, 573
19
Surface Contact ■ Introduction, 576
for SOL 101 ■ Understanding Contact Regions - BSURF, BCPROP, BSURFS, 577
■ Understanding Contact Region Parameters - BCRPARA, 577
■ Understanding Contact Pairs - BCTSET, 578
■ Understanding Source Regions and Target Regions, 579
■ Understanding the Contact Control Parameters Used in SOL 101 -
BCTPARM, 580
■ Tips for setting PENN and PENT, 582
■ Including CGAP elements in a contact solution - system cell OLDGAPS, 582
20
Plotting ■ Overview of Plotting in NX Nastran, 588
❑ Structural Plots, 588
❑ X-Y Plots, 588
❑ Flutter Analysis Plots, 589
■ Superelement Plotting, 589

■ Post Processors, 593
21
Linear Buckling ■ Introduction to Linear Buckling, 596
■ Finite Element Approach, 596
■ Eigenvalue Extraction Methods, 602
❑ Inverse Power Method (INV), 603
❑ Enhanced Inverse Power Method (SINV), 603
❑ Lanczos Method, 603
❑ Comparison of Methods, 603
■ Linear Buckling Assumptions and Limitations, 606

■ Buckling Examples, 607
❑ Example 1 – Classical Euler Beam Buckling, 607
❑ Example 2 – Lateral Buckling, 613
❑ Example 3 – Planar Frame Buckling, 618
❑ Example 4 – Buckling of a Stiffened Panel with Transverse Shear
Flexibility, 621
❑ Example 5 – Buckling of a Cylinder Under Uniform Axial Load, 625
❑ Theoretical Results for Example 5, 630
❑ Example 6 – Multiple Buckling Analyses in a Single Run, 631
22
Laminates ■ Overview of Laminated Composite Materials, 635
❑ Modeling Laminates in NX Nastran, 636
■ Understanding Classical Lamination Theory, 641

■ Understanding Laminate Failure Indices, 652
❑ Selecting a Failure Criterion, 652
■ Calculating Interlaminar Shear Stress and Strains, 656

■ Examples: Defining Laminate Material Properties, 657
❑ Example: Using PCOMP to Model a Honeycomb Plate, 657
❑ Example: Using PSHELL to Model a Honeycomb Section, 663
23
Coupled Fluid- ■ Overview of Coupled Fluid-Structure Interaction, 672
Structure ■ Performing a Coupled Fluid-Structural Analysis, 672
Interaction
❑ Defining the Fluid Model, 673
❑ Basic Fluid Equations, 677
■ Acoustic Cavity Modeling, 683

❑ Assumptions and Limitations, 685
24
Cyclic Symmetry ■ Understanding Cyclic Symmetry, 694
❑ Supported Solution Sequences, 694
❑ Comparing Cyclic Symmetry and Traditional Reflective Symmetry
Techniques, 695
■ Understanding Types of Symmetry, 696
■ Cyclic Symmetry Theory, 699
❑ Symmetrical Components, 699
❑ Understanding Rotational Symmetry, 701
❑ Dihedral Symmetry, 713
❑ Reflective Symmetry, 722
❑ Axisymmetry, 727
■ Using Cyclic Symmetry, 729

❑ Overview of Performing a Cyclic Symmetry Analysis, 729
❑ Bulk Data Entries Specific to Cyclic Symmetry, 732
❑ Case Control Commands Specific to Cyclic Symmetry, 740
■ Performing Axisymmetric Analysis, 745

❑ Solving Axisymmetric Problems, 745
❑ Axisymmetric Example, 747
25
p-Elements ■ Introduction to p-Elements, 754
❑ p-Elements in NX Nastran, 754
■ p-Version Capabilities, 756

❑ Solution Sequences Supported, 756
❑ Geometry, 756
❑ Elements, 757
❑ Materials, 757
■ p-Element Loads and Constraints, 757

❑ Defining p-Element Loads, 757
❑ Defining p-Element Constraints, 758
❑ p-Version Adaptivity, 760
❑ Data Recovery, 761
■ p-Element Example Problems, 763

❑ Square Plate with a Circular Hole, 764
❑ Spherical Structure Under Internal Pressure, 768
❑ Motor Blade – Normal Modes, 772
❑ Torsion of Elliptical Bar, 775
❑ Circular Solid Model, 786
❑ Clamped Rhombic Plate, 797
❑ Helix Modeled with CBEAM p-Elements, 799
❑ Plate with Stiffener, 801
■ Solid p-Elements in Linear Dynamic Solution Sequences, 803
❑ User Interface in NX Nastran, 804
❑ Limitations and Recommendations, 804
❑ Example, 804
■ p-Element Thermal Analysis, 806

❑ Introduction, 806
❑ Geometry, 806
❑ Elements, 806
❑ Materials, 806
❑ Loads and Boundary Conditions, 806
❑ p-Version Adaptivity, 807
❑ Data Recovery, 807
❑ Thermal Stress Analysis, 807
❑ Limitations and Recommendations, 807
NX Nastran Reference Manual
1
CHAPTER
Overview of NX Nastran
■ General Overview of NX Nastran
■ Overview of the NX Nastran Analysis Process
2 NX Nastran User’s Guide
1.1 General Overview of NX Nastran

NX Nastran is a general purpose finite element analysis computer program that lets
you solve a wide range of engineering problems. NX Nastran is written primarily in
FORTRAN and is optimized to run efficiently and provide identical results on a wide
variety of computers and operating systems.
General Capabilities
NX Nastran contains the following analysis capabilities:
• Linear statics (including inertia relief)

• Normal modes and buckling
• Heat transfer (steady-state and transient)
• Transient response
• Frequency response
• Response spectrum and random response
• Geometric and material nonlinear static and transient response
• Design optimization and sensitivity (including dynamic and shape
optimization)
• Composite materials
• Acoustic response
• Aeroelasticity
• Superelements
• Complex eigenanalysis
• Axisymmetric analysis
• Cyclic symmetry
• p-elements
Modular Structure and DMAP

The NX Nastran software is composed of a large number of building blocks called
modules. Each module is a collection of FORTRAN subroutines which is designed
to perform a specific task, such as processing model geometry, assembling matrices,
applying constraints, solving matrix problems, and calculating output quantities.
CHAPTER 1 3
Within NX Nastran, the modules are controlled by an internal language called the
Direct Matrix Abstraction Program (DMAP), which is a high-level programming
language with its own compiler and grammatical rules. A DMAP statement is like
a subroutine call statement within FORTRAN, with both input and output
information.
Pre-Defined Solution Sequences

Each type of analysis available in NX Nastran is called a “solution sequence.” Each
solution sequence is a pre-defined collection of hundreds or thousands of DMAP
commands. Once you select a solution sequence, its particular set of DMAP
commands sends instructions to the modules that are needed to perform the
requested solution.
NX Nastran also includes DMAP (Direct Matrix Abstraction Programming), a

high-level programming language that allows you to build custom solution
sequences or modify existing ones.
Using DMAP for Programmability

Although NX Nastran contains a number of pre-defined solutions, you can use
special DMAP “alters”to either modify exisitng solution sequences or create new
ones.
See also:
• NX Nastran DMAP Programmer’s Guide
1.2 Overview of the NX Nastran Analysis Process

To solve a finite element model in NX Nastran, you must:
• create the input file
• use the nastran command to submit the input file for batch processing
• examine your results
Creating the Input File

To perform an analysis with NX Nastran, you must generate an input file that
contains model data (information about the finite element model, including
geometry, elements, materials, and loads) as well as analysis data (information
about the type of analysis you want to perform, such as the analysis type and the
type of data you want output).
In NX Nastran, the input file is an ASCII text file which you can create:
• automatically using a preprocessor, such as I-DEAS MasterFEM

• manually using a text editor
Input File Sections

An NX Nastran input file contains five separate sections:
• the NASTRAN statement

• the File Management statement section (FMS)
• the Executive Control statement section
• the Case Control command section
• the Bulk Data entry section
Documentation for NX Nastran Input Options

Each section in the NX Nastran input file is described in more detail later in this
guide. Comprehensive documentation for each File Management and Executive
Control statement, Case Control command, and Bulk Data entry is located in the NX
Nastran Quick Reference Guide.
Input File Naming

In general, Nastran input files are given a .dat extension. However, NX Nastran will
try to process any file you submit, regardless of its file extensions. Other common NX
Nastran input file extensions include .bdf or .blk.
Submitting the Input File for Analysis

Once you’ve created the input file, you submit the file for execution as a batch process
in NX Nastran. Because NX Nastran isn’t an interactive program, once you submit
the input file, you won’t have any additional interaction with the software until the
job completes, unless you need to terminate the job.
To execute NX Nastran, you typically use the nastran system command (in some
cases, your system administrator may assign a different name to the command)
followed by the name of the input file. For example:
NASTRAN MODEL1A
The nastran command also allows you to specify certain keywords for controlling the
job execution. The format of the nastran command is:
nastran keyword1 = value1 keyword2 = value2 ...

See also:
• “Using the nastran Command” of the NX Nastran Installation and Operations

Guide
CHAPTER 1 5
• “nastran Command and NASTRAN Statement” on page 1 of the NX

Nastran Quick Reference Guide
Reviewing the Results

When NX Nastran completes the analysis, you can review your results. NX Nastran
automatically creates a number of different types of files, which are summarized in
the following table:
File Type Description

.DBALL Contains permanent data for database runs for restart
analyses.
.f04 Contains database file information and a module
execution summary, such as the start and stop time for
each module as well as the size of the database file (if
any).
.f06 Contains the results of your analysis, such as
displacements and stresses, as well as any diagnostic
messages.
.LOG Contains system information and any system level
error messages.
.MASTER Contains permanent data for database runs for restart
analyses.
In addition to these automatically generated files, you can manually request that NX
Nastran create the following files:
File Type Description

.pch Punch file.
.plt Plot file
.op2 Output2 file.
.xdb XDB file.
NX Nastran provides for direct support of interface to other products, such as NX

Nastran Access and I-DEAS Master FEM.
For MSC.Patran, the support is provided through the DBC module, which creates a
“graphics” database. The interface to the DBC module is described in the “POST”
on page 1471 of the NX Nastran Quick Reference Guide.
The graphics database created by the DBC module, can also be read by the NX
Nastran Access library of object routines. This object library can be linked with a
user-created program that extracts data from the database. This is an indirect
method for outside vendors to obtain NX Nastran model and results information.
See the NX Nastran Access Reference Manual for more information.
Communication with other software packages, such as I-DEAS Master FEM, is

through the NX Nastran OUTPUT2 module, which creates a FORTRAN readable file
that is converted by these outside programs to their own particular data formats. The
user interface and the supported data for these outside programs are described in
“POST” on page 1471 of the NX Nastran Quick Reference Guide(-1, -2, -4, and -5).
2
CHAPTER
Structure and Syntax for Input Data
■ Overview of Specifying Input Data
■ General Input File Syntax Rules
■ Understanding NX Nastran Input Records
■ Specifying Continuations for Bulk Data
■ Replicating Bulk Data Entries
■ Inserting External Files with INCLUDE
■ Including Comments in the Input File
2.1 Overview of Specifying Input Data

In NX Nastran, you submit all your data to the solver through the input file, which
is sometimes referred to an input “deck.” In most cases, you will automatically
generate your NX Nastran input file by exporting an existing finite element model
from a graphical pre-processing program. However, because you may need to
manually edit the contents of that input file, it’s important that you understand the
rules and conventions of entering data into an NX Nastran input file.
This chapter describes:
• the general syntax rules for input data in NX Nastran

• NX Nastran input records, including the different input data formats (free
field, small field, and large field)
• how to use the INCLUDE option to include external files in your NX
Nastran input file
• how to add comments to your NX Nastran input file
The NX Nastran Quick Reference Guide describes the input requirements and default
values for all NX Nastran data file input.
2.2 General Input File Syntax Rules

The following sections provide the general rules and conventions for inputting data
in NX Nastran.
Understanding Types of Input Data

In Nastran, you can input integer or real numbers, as well as character data.
• Integer numbers can’t contain a decimal point.
• Real numbers, including zero, must contain a decimal point.
You can enter real numbers in a variety of formats. For example, the
following are all acceptable versions of the real number seven:
7.0 .7E1 0.7+1

.70+1 7.E+0 70.-1
• Character data (sometimes called literal, or BCD—binary coded decimal

data) can be alphanumeric, but it must always begin with an alpha
character and contain 8 or fewer total characters.
Importantly, the different types of data aren’t interchangeable. If a particular field

requires an integer and you enter a real number instead, NX Nastran issues an error
message.
CHAPTER 2 9
Using Consistent Unit Systems

In NX Nastran, you must always use a consistent set of units when you develop
your finite element model in the input file. The actual unit system you use doesn’t
matter. However, the unit system must agree with itself. Table 2-1 shows examples
of consistent unit systems:
Table 2-1 Consistent Systems of Units
Quantity Input English SI

Input
Grid Point Geometry inch meter
2 2
Elastic Modulus lbf /inch Newton/meter
Applied Moment inch-lbf Newton-meter
Applied Force lbf Newton
2
Mass lbf-sec /inch Kilogram
Time second second
Output:
Displacements inch meter
2 2
Stresses lbf /inch Newton/meter
NX Nastran can’t detect inconsistent units within your input file. Consequently, if
your model contains inconsistent units, the software won’t generate any user
warning messages. It simply produces incorrect answers.
Specifying Valid Input File Characters

Table 2-2 lists the valid characters for NX Nastran.
Table 2-2 ASCII Subset for NX Nastran
Character ASCII (HEX) Character ASCII (HEX)
HT (Horizontal @ 40
Tabulation) 09 A 41
SP (Space B 42
“blank”) 20 C 43
! 21 D 44
" 22 E 45
# 23 F 46
Table 2-2 ASCII Subset for NX Nastran (continued)
Character ASCII (HEX) Character ASCII (HEX)

$ 24 G 47
% 25 H 48
& 26 I 49
‘ 27 J 4A
( 28 K 4B
) 29 L 4C
* 2A M 4D
+ 2B N 4E
’ 2C O 4F
- 2D P 50
. 2E Q 51
/ 2F R 52
0 30 S 53
1 31 T 54
2 32 U 55
3 33 V 56
4 34 W 57
5 35 X 58
6 36 Y 59
7 37 Z 5A
8 38 [ 5B
9 39 \ 5C
: 3A ] 5D
; 3B ^ 5E
< 3C - 5F
= 3D
> 3E
? 3F
2.3 Understanding NX Nastran Input Records

In NX Nastran, you input data records which are 80 characters (or columns) in length
on each line. Those records are then divided into fields. How those records are
divided into fields depends upon depending upon the data format type.
There are two basic categories of input data formats in NX Nastran:
CHAPTER 2 11
• “free” format data, in which the data fields are simply separated by
commas. This type of data is known as free field data.
• “fixed” format data, in which your data must be lined up in columns of
specific width. There are two subcategories of fixed format data that differ
based on the size of fixed column width:
• small field format, in which a single line of data is divided into ten
fields that can contain eight characters each
• large field format, in which a single line of input is expanded into
two lines The first and last fields on each line are eight columns
wide, while the intermediate fields are sixteen columns wide. The
large field format is useful when you need greater numerical
accuracy.
Which type of format data you use depends upon the section of the input file that
you’re working with.
• You must use the free field format for the NASTRAN statement and the
File Management, Executive Control, and Case Control sections.
• With the Bulk Data section you can use the free field, small field, or large
field formats.
The rules for entering free field, small field, and large field format data are described
in more detail below.
Understanding Free Field Format

In free field format, you separate data fields by commas.
Free Field Format Rules

• You must begin free field data entries in column 1.
• You can use commas to skip fields.
• To skip one field, use two commas in succession.
• To skip two fields, use three commas in succession (and so on).
• Integer or character data is limited to eight characters. If you include an
integer or character longer than 8 characters, NX Nastran issues a fatal
error.
• With free field format data, NX Nastran rounds off real numbers with
more than eight characters. Therefore, you may lose some precision. For
example, an entry of 1.2345678+2 becomes 123.4568. If you need more
significant digits, use the large field format.
• Free field data cannot contain embedded blanks. An example of a free field
embedded blank is shown below:
GRID,2,,1 0,-2.0,3.0,,136
Embedded blank
not allowed
• You use a dollar sign ($) to terminate the free field entry. You can follow the
dollar sign with comments.
Free Field Format Example

The following example shows the GRID Bulk Data entry in free field format:
GRID,2,,1.0,-2.0,3.0,,136
These two commas

indicate an empty field
Understanding Small Field Format

With the small field format, the software separates an 80 column line of Bulk Data
entry into ten equal fields of eight characters each:
1 2 3 4 5 6 7 8 9 10
<-8 Char-> <-8 Char-> <-8 Char-> <-8 Char-> <-8 Char-> <-8 Char-> <-8 Char-> <-8 Char-> <-8 Char-> <-8 Char->
Small Field Format Rules

• Fields 1 and 10 must be left justified.
• Fields 2 through 9 don’t need to be either right or left justified. However,
aligning the data fields is a good practice.
• You can’t include any embedded blanks in small field format input data.
An example of a small field embedded blank is shown below:
8 character field
7 . 6 5
Embedded blank
not allowed
CHAPTER 2 13
Small Field Format Example

The following example shows the GRID entry in small field format:
1 2 3 4 5 6 7 8 9 10
GRID 2 1.0 -2.0 3.0 136
Understanding Large Field Format

A high degree of numerical accuracy is required in some NX Nastran applications.
You can use large field format when small field format doesn’t provide enough
significant digits. For example, you can use large field format when you need to
input integers larger than eight characters, or you need to enter real numbers with
more significant digits than you can fit in an eight-character field.
Note: A minus sign, decimal point, and the “E” in scientific notation count as
characters.
Large Field Format Rules

Large field format requires (at least) two lines for each entry:
• the first and last field of each line contains eight columns
• the fields in between contain 16 columns
Large field entries are denoted by an asterisk (*). The asterisk must:
• immediately follow the character string in field 1A of the first line of the
entry
• immediately precede the character string in field 1B of the second line of
the entry
Large Field Format Example

The following is an example of the GRID Bulk Data entry in large field format:
First Line: (Left half of single field)
Field 1A 2 3 4 5 6
GRID* 2 1.0 -2.0 *GRID10
8 16 16 16 16 8
columns
Second Line: (Right half of single field)
Field 1B 6 7 8 9 10B
*GRID10 3.0 136
8 16 16 16 16 8
columns
2.4 Specifying Continuations for Bulk Data

Some Bulk Data entries require more than eight fields (72 columns) of data. In those
cases, you need to specify a continuation to continue the remainder of the data on
one or more subsequent lines. To do this, you follow the parent entry (the first line)
with one or more continuation entries on subsequent lines.
There are two ways to specify a continuation:
• manually
• automatically
Note: If you’re using the large field format, you must manually specify a
continuation. Automatic continuation generation is not available for large field
format data.
Manually Specifying a Continuation with Identifiers

You can manually specify a continuation by using a continuation identifier. A
continuation identifier is a special character (+ or *) that indicates that the data will
continue on another line. Once you’ve specified the continuation identifier, you then
follow the parent entry (the first line) with one or more continuation entries on
subsequent lines.
Rules for Manually Specifying a Continuation

• You must use a + or * in column 1, field 1 of a continuation entry (small field
only?).
• The alphanumeric character strings you use as the continuation identifiers
can’t contain the symbols *, =, or $.
Note: The + symbol used in column 1 of field 10 is for clarity.
• Field 10 contains the continuation identifier that must be unique with
respect to all other continuation identifiers.
• NX Nastran always ignores column 1 of field 10. It is not considered to be
part of the identifier.
CHAPTER 2 15
• The remaining contents in columns 2-8, field 1 of a continuation entry must

be identical to the entry in field 10 (columns 2-8) of either the parent entry
or the preceding continuation entry.
• You can use small and large field continuation entries together to define a
single data item entry.
Manual Continuation Example

For example, consider the following PBAR simple beam property entry:
Standard PBAR Entry:
1 2 3 4 5 6 7 8 9 10
PBAR PID MID A I1 I2 J NSM
C1 C2 D1 D2 El E2 F1 F2
K1 K2 I12
Continuation Example:
PBAR 39 6 2.9 1.86 2.92 .48 +PB1
+PB1 0. 0. 0. 1. 1. 1. 1. 0. +PB2
+PB2 .86 .86
In this example:
• +PB1 in field 10 of the parent entry is an arbitrary (and unique)

user-defined pointer to field 1 of the second line of the entry.
• +PB2 in field 1 of the second line points the third line.
Automatically Specifying a Continuation

You can also have NX Nastran automatically generate continuation fields if:
• your bulk data is in sorted order (i.e., the continuation lines must follow
the parent line)
• you are using the small field format; automatic continuation is not
available with large field format data.
In general, you should use the automatic continuation method when possible as it
allows you to avoid the inconvenience of using the continuation identifiers.
To have NX Nastran automatically generate continuations:
• Leave both field 10 of the parent entry and field 1 of the continuation line
blank.
• Place the continuation line directly following the parent entry.

The blank field indicates that it is a small-field continuation of the previous
line.
You can mix small-field and large-field format continuations; however, it is

generally not recommended because it becomes difficult to discern the locations of
the fields.
Rules for Automatically Generating Continuations

To have NX Nastran automatically generate a continuation:
• The continuation line(s) must immediately follow the parent Bulk Data
entry.
• You must leave fields 1 and 10 of the continuation line (or lines) blank.
• Your bulk data must be in sorted order.
Note: Setting SYSTEM CELL 357 to 1 tells NX Nastran to ignore the continuation
field. This eliminates the concern of duplicate continuation identifiers.
Automatic Continuation Example

This process is illustrated in the following example:
Input (.DAT) file:
CHEXA, 1, 10, 3, 5, 7, 1, 15, 17,

, 19, 13, 4, 6, 8, 2, 10, 11,
, 12, 9, 16, 18, 20, 14
Output (.f06) file:
GEAR TOOTH EXAMPLE JULY 10, 2003 NX NASTRAN 07/10/2003 PAGE 3
S O R T E D B U L K D A T A E C H O
ENTRY
COUNT . 1 .. 2 .. 3 .. 4 .. 5 .. 6 .. 7 .. 8 .. 9 .. 10 .
1- CHEXA 1 10 3 5 7 1 15 17 +000001
2- ++00000119 13 4 6 8 2 10 11 +000002
3- ++00000212 9 16 18 20 14 +000003
Note that column 1 of field 10 contains a + symbol. Since NX Nastran ignores this
column, the choice of the + is arbitrary. However, using the + symbol improves
readability and serves as a reminder for the format of the continuation line.
CHAPTER 2 17
2.5 Replicating Bulk Data Entries

To avoid having to input each bulk data entry individually, you can replicate the
repetitive fields from a single entry definition. You can use replication with bulk
data that is in small field, large field, and free field format.
• To replicate selected fields from the preceding entry, use the symbol =.
• To replicate all the remaining entries from the preceding entry, use the
symbol ==.
• To generate an incremented value from the previous entry, use *x or *(x),
where x is the value of the increment. “x” should be a real number for real
fields or an integer for integer fields.
• To repeatedly replicate a field, use =n or =(n), where n is the number of
images to be generated using the values of the increments on the preceding
entry.
Replicating Continuation Entries

If you’re manually creating continuation entries, you can replicate the continuation
fields (fields one and ten) using the following conventions:
• You can only use letters of the alphabet and integers. They are coded into
a base 36 number. That is, the sequence of numbers is 0,1, 2, ..., 8, 9, A, B, ...
• The software increments the continuation field by +1 regardless of the
value you specify.
• The software doesn’t increment the first character in field one or ten.
• The software won’t ever increase the number of characters in an
incremented field. For example, if the first field is “0", the thirty-seventh
field will also be “0", resulting in an illegal entry. A method to solve this
problem would be to start with a first field of “00". This will provide
thirty-six squared unique fields.
• At least one field in fields 2 through 8 of continuation entries must contain
data.
Replication with MSGMESH

Replication is a limited data generation capability. You can use replication with bulk
data that is either fixed (large or small field) or free field format.
• The MSGMESH capability includes the capabilities described here, plus

the following capabilities as long as NASTRAN MESH is specified in the
File Management Section.
• Continuation entry fields may be incremented or decremented.
• Repeated replication is indicated by coding =(n) in field 1, where n is

number of entry images to be generated using the values of increments
from the current or preceding replication entry.
Entered entries:
GRID,101 ,17,1 .0,10.5,,17,3456
=(4), *(1),=,*(0.2),==$
Generated entries:
GRID 101 17 1.0 10.5 17 3456
GRID 102 17 1.2 10.5 17 3456
GRID 103 17 1.4 10.5 17 3456
GRID 104 17 1.6 10.5 17 3456
GRID 105 17 1.8 10.5 17 3456
• A blank in field 1 indicates immediate continuation entry replication. The

default continuation entry increment is 1. Example:
BSET1,123,1,2,3,4,5,6,7
,,*7,*7,*7,*7,*7,*7,*7
=(3)
Generated entries:
BSET1 123 1 2 3 4 5 6 7 +00001
++00001 8 9 10 11 12 13 14 +00002
++00002 15 16 17 18 19 20 21 +00003
++00003 22 23 24 25 26 27 28 +00004
++00004 29 30 31 32 33 34 35 +00005
• A “=(D)” in field 1 indicates delayed continuation entry replication. A

maximum of 9 entries may be replicated as a group. The default
continuation entry increment is 10. Example:
Entered entries:
CTRIA3,10,1,1,10,11/+C1
=(D),*(1),=,=,*(1),*(1)/*(20)
+C1,,,2.0,1.0,1.0
=(2),==
Generated entries:
CTRIA3 10 1 1 10 11 +C1
+C1 2.0 1.0 1.0
CTRIA3 11 1 1 11 12 +C21
+C21 2.0 1.0 1.0
CHAPTER 2 19
CTRIA3 12 1 1 12 13 +C41
+C41 2.0 1.0 1.0
• Parentheses are optional on replication entries and an equal sign may

replace an asterisk.
The following is an example of the use of replication, automatic
continuation field generation, and the free field format:
GRID,101 ,17,1.0,10.5,,17,3456
=,*1,=,*0.2, *(0.1), == $ COMMENTS MAY APPEAR AFTER $
=3
EIGR,13,GIV,,30.
,MASS
CBAR,1 ,1 ,101 ,102,0.,0.,1.,,+0
=,*1,=,*1,*1====*1
+0,56
*1,=$
The above free-field entries will generate the following Bulk Data in
the 8-column format, as seen in the SORTED BULK DATA ECHO:
Note: A “,” should always be used after the “*1” for the
continuation increment even if fixed field format is being used.
CBAR 1 1 101 102 0. 0. 1. +0
+0 56
CBAR 2 1 102 103 0. 0. 1. +1
+1 56
EIGR 13 GIV 30. +0000

01
++000001 MASS
GRID 101 17 1.0 10.5 17 3456
GRID 102 17 1.2 10.6 17 3456
GRID 103 17 1.4 10.7 17 3456
GRID 104 17 1.6 10.8 17 3456
GRID 105 17 1.8 10.9 17 3456
The automatically generated continuation entries start with the number 1, are incremented
by 1, and are padded with zeros and plus signs as shown above. If this feature is used, it is
the user’s responsibility not to enter continuation entries that also use this convention. In
particular, data generated on another run and then written to the PUNCH file with the
ECHO=PUNCH, will cause problems when introduced into other data with blank
continuation fields.
2.6 Inserting External Files with INCLUDE

You can use the INCLUDE statement to insert external files into your NX Nastran
input file. You can specify INCLUDE in any part of the input file as long as the data
referenced by the INCLUDE statement is appropriate for that section of the input
file. This allows you to avoid repetitious input in the different sections of the input
file.
Using INCLUDE in Different Input File Sections

You can use INCLUDE in a variety of ways in the different sections of your input file.
• In the File Management section, you can use the INCLUDE statement to
assign multiple database files with one statement.
• In the Executive section, you can use the INCLUDE statement to include
alters to a solution sequence.
• In the Case Control section, the INCLUDE file can be useful if you want to
use the same subcase structure and/or the same output requests for
different models.
• In the Bulk Data section, the INCLUDE statement is very useful for
working with superelement data. You can use it to place each superelement
of a superelement model in a different physical file. You can then share the
files with other users and analyze them separately. Since superelement
models are usually very large and complex, using the INCLUDE statement
can greatly reduce the complexity of the input file.
INCLUDE Statement Limitations

• Although you can nest INCLUDE statements within the external file, the
nested depth level must not be greater than 10.
• The INCLUDE statement plus the filename must be less than 72 characters.
• You can’t use continuations with the INCLUDE statement.
Example: Using the INCLUDE Statement

Below is an example of the use of the INCLUDE statement. Note that the syntax for
the filenames is machine dependent. The format of lowercase filenames within
single quotes works on all computer operating systems that support NX Nastran.
INCLUDE ’nastran.dat’
INCLUDE ’database.dat’
SOL 101
INCLUDE ’sol101.alter’
CEND
CHAPTER 2 21
TITLE = Complete Model

SUBTITLE = Three load cases
INCLUDE ’output.dat’
SUBCASE 1
LOAD = 10
BEGIN BULK
INCLUDE ’se0.dat’
ENDDATA
Each of the INCLUDE files in the example contains different information. For this
example, “nastran.dat” contains NASTRAN system cells, “database.dat” contains
FMS statements, “sol101.alter” contains a DMAP Alter for SOL 101, “output.dat”
contains Case Control commands, and “se*.dat” contains modeling information for
superelements 0, 10, and 20.
See Also
• “INCLUDE” on page 95 of the NX Nastran Quick Reference Guide
2.7 Including Comments in the Input File

You can insert comments in any part input file to help organize your input data.
Comment entries begin with a dollar ($) sign in column one followed by any
characters out to column 80. Comments appear only in the unsorted echo of the
input file listing.
NX Nastran Reference Manual
3
CHAPTER
Understanding the Input Data File
■ Overview of the NX Nastran Input File Sections
■ The NASTRAN Statement (Optional)
■ File Management Section (Optional)
■ Executive Control Section (Required)
■ Case Control Section (Required)
■ Bulk Data Section
■ Using Parameters
3.1 Overview of the NX Nastran Input File Sections

The input file is comprised of five distinct sections and three required one-line
delimiter which must follow the sequence shown in Figure 3-1. These delimiters and
sections are described below.
Note: If any of the required delimiters are missing from your input file, NX Nastran
issues a fatal error message when you submit your job.
NASTRAN Optional
Statement
File Management
Optional
Statements
Executive Control
Statements Required Section
CEND Required Delimiter
Case Control
Commands Required Section
BEGIN BULK Required
Bulk Data
Entries Required Section
ENDDATA Required
Figure 3-1 NX Nastran Input File Structure
Example: Using NX Nastran Input Files to Model a Truss

Throughout this chapter, a simple truss structure, shown in Figure 3-2, is used to
illustrate different aspects of input files.
CHAPTER 3 25
2 Y
A = 4.0 in
6 2
E = 30.0 × 10 lb ⁄ in
1 3
4
J = 1.27 in 2
ν = 0.3 3
2
5 10 ft
Elements
1 X
Grid points
4 4
F =1000 lb
Pin joints
50 ft
Figure 3-2 Truss Example
The truss model consists of four grid points that represent the joints of the structure.
The locations of these grid points are given in the default coordinate system for NX
Nastran, which is known as the “basic coordinate system.” In this example, the
origin is located at grid point 1.
The truss structure consists of five members, each with a cross-sectional area (A) of
2 4
4 in and a polar moment of inertia (J) of 1.27 in . The material is steel with a
6
Young’s modulus E of 30 x 10 psi and a Poisson’s ratio of 0.3. All the connections
between the members are pin joints (i.e., they can transmit forces but not moments).
The goal of the analysis is to determine the displacement of the grid points and the
stresses and forces within the members.
Although the dimensions given in Figure 3-2 are in feet, the dimensions used for the
grid point locations in the input file (Listing 3-1 ) are in inches. For consistency, the
units for the cross-sectional area, torsional stiffness, and Young’s modulus in the
input file are also provided in inches.
Note: You must always ensure that you enter data in a consistent system of units.
Listing 3-1 shows the input file for the truss structure. Although this input file
represents a fairly simple structure, it is typical of all NX Nastran input files.
Listing 3-1 Truss Example Input File
$ FILENAME - TRUSS1.DAT
$ Executive
ID LINEAR,TRUSS1 Control
SOL 101
section
TIME 2
CEND
TITLE = LINEAR STATICS USER’S SAMPLE INPUT FILE
SUBTITLE = TRUSS STRUCTURE
LABEL = POINT LOAD AT GRID POINT 4 Case
LOAD = 10 Control
SPC = 11
DISPLACEMENT = ALL section
ELFORCE = ALL
ELSTRESS = ALL
BEGIN BULK
$
$ THE GRID POINTS LOCATIONS
$ DESCRIBE THE GEOMETRY
$
GRID 1 0. 0. 0. 3456
GRID 2 0. 120. 0. 3456
GRID 3 600. 120. 0. 3456
GRID 4 600. 0. 0. 3456
$
$ MEMBERS ARE MODELED USING
$ ROD ELEMENTS
$
CROD 1 21 2 3
CROD 2 21 2 4
CROD 3 21 1 3
CROD 4 21 1 4 Bulk Data
CROD 5 21 3 4
$ section
$ PROPERTIES OF ROD ELEMENTS
$
PROD 21 22 4. 1.27
$
$ MATERIAL PROPERTIES
$
MAT1 22 30.E6 .3
$
$ POINT LOAD
$
FORCE 10 4 1000. 0. -1. 0.
$
SPC1 11 123456 1 2
$
ENDDATA
CHAPTER 3 27
3.2 The NASTRAN Statement (Optional)

The NASTRAN statement lets you modify the default values for certain operational
parameters (also called system cells) at runtime to control internal solution
processing or provide specific diagnostics. For example, you can use the NASTRAN
statement to control aspects of working memory, datablock size, and datablock
parameters.
• You can include more than one NASTRAN statement in a single input file.
• You can combine the NASTRAN statement with the File Management
section.
• You can specify the NASTRAN statement in the Runtime Configuration
(RC) files at the system, user, and job level.
See also:
• “Using the NASTRAN Statement” of the NX Nastran Installation and

Operations Guide
• “nastran Command and NASTRAN Statement” on page 1 of the NX
Nastran Quick Reference Guide
• “BUFFSIZE” on page 553 of the NX Nastran User’s Guide
3.3 File Management Section (Optional)

The File Management section (FMS) of the input file is optional and typically
follows the NASTRAN statement(s). It ends with the specification of an Executive
Control statement. For example, the File Management section lets you:
• assign files
• attach or initialize the NX Nastran database sets (DBsets) and FORTRAN
files
• perform restart analyses
The initialization of a database includes specification of its maximum size, member
names, and physical filenames. The initialization of a FORTRAN file includes the
specification of its filename, FORTRAN unit numbers, and FORTRAN attributes.
The File Management section is commonly used in input files for larger analysis
problems. For many NX Nastran problems, no File Management statements are
required because a default File Management section is executed by the software at
the beginning of every run. For example, since the truss model shown in Listing 3-1
is small and the desired analysis is not a restart, a File Management section isn’t
included in the input file.
See Also:
• “File Management Statements” on page 35 of the NX Nastran Quick

Reference Guide
• “Database Concepts” on page 531 of the NX Nastran User’s Guide
• “Restarts” on page 513 of the NX Nastran User’s Guide
3.4 Executive Control Section (Required)

The Executive Control section is the first required section in the NX Nastran input file
and provides the overall control for the solution. For example, statements in the
Executive Control section let you:
• specify an optional ID statement to identify the job

• select the type of analysis (solution sequence) you want to perform
• specify an optional TIME statement to set limits on the maximum allowable
CPU time for the run
• request various diagnostics
• include complete DMAP sequences, if you want to use DMAP in the
analysis
For example, the Executive Control section for the truss example (Listing 3-1)
identifies the job, requests the static Structured Solution Sequence 101, and sets the
maximum time limit for the job to two minutes.
For the truss example, the Executive Control section specifies static solution SOL 101
and a TIME statement that specifies a maximum of 2 CPU minutes for the run. The
end of the Executive section is denoted by the CEND statement. The ID statement is
an optional statement to help document your input file. If used, it can be located
anywhere in the Executive Control section.
See Also
• “Executive Control Statements” on page 105 of the NX Nastran Quick

Reference Guide
Executive Control Format

The format of the Executive Control statements is free-field format between columns
1-72. The statements may begin in any column (except for comments). The
operands—that is, the input data following the name—are separated from the name
using a comma and/or one or more blanks. If more than one operand is needed, they
are also separated by a comma and/or one or more blanks.
You must use the CEND delimiter to mark the end of the Executive Control section .
CHAPTER 3 29
Order Dependence for Certain DMAP-Related Statements

Most statements within the Executive Control section are order independent.
However, the following statements, all related to the use of DMAP are order
dependent:
• COMPILE
• COMPILER
• ALTER
• ENDALTER
• LINK
Notably:
• The LINK statement must appear after all COMPILE statements.

• The COMPILER statement (or equivalent DIAGs) must appear before all
COMPILE statements. This statement sets the default print control for
subsequent COMPILE statements.
Commonly Used Executive Control Statements

Some commonly used Executive Control statements are:
• ID
• SOL
• ECHOON and ECHOOFF
• TIME
• DIAG
• CEND
The ID Statement (Optional)

You can use the ID statement to identify your job. Its format is:
ID i1, i2
where i1 and i2 are character strings. i1 may be 1 to 8 characters in length. i2 may

be of any length. The first character of each string must be alphabetic.
You can use only one ID statement, though you can specify it anywhere in the
Executive Control section The format of the ID statement is the keyword “ID”
followed by a delimiter and a comment. The comment may be any ASCII
characters. For example:
www.cadfamily.com EMail:[email protected] ID THIS IS RUN 2

See Also
• “ID” on page 139 of the NX Nastran Quick Reference Guide
The SOL Statement (Required)

The SOL statement specifies the type of analysis (solution sequence) you want to
perform. The format of the SOL statement is:
SOL n
where n is a positive integer identifying the solution type or the character name of
the solution procedure.
See Also
• “Understanding Solution Sequences” on page 302 of the NX Nastran

User’s Guide
• “SOL” on page 149 of the NX Nastran Quick Reference Guide
• “SOL 601,N or SOL 701” on page 154 of the NX Nastran Quick Reference
Guide
ECHO Statement (Optional)

The ECHO statement lets you control whether the software prints (“echoes”) the
Executive Control section to the output file. ECHOON is the default.
• Use ECHOON (default) to print the section to the output file.

• Use ECHOOFF to prevent the software from printing the section to the
output file.
In general, using ECHOOFF hinders the usability of your output file, especially if
you need to review the output file several months after your original analysis.
However, the ECHOOFF statement can be useful for security. For example, you can
use ECHOOFF, to prevent NX Nastran from printing proprietary portions of the
Executive Control section (such as user-written DMAP) to the output file.
You can use multiple ECHOON and ECHOOFF statements in an single input file.
See Also
• “ECHO” on page 132 of the NX Nastran Quick Reference Guide
Time Statement (Optional)

The TIME statement allows you to set the maximum CPU time and I/O time for the
job before NX Nastran terminates the run. Its format is
TIME t1, t2
CHAPTER 3 31
where:
t1 = maximum allowed execution time in CPU minutes (real or integer;

default=one minute, which is only adequate for very small jobs).
t2 = maximum allowable I/O time in seconds (default is infinity)
Importantly, the time limits you can set with the TIME statement are different from
the system time limit imposed by your system administrator. In general, you
should use the TIME statement to set the NX Nastran time limit to a value less than
the system time limit. If your job exceeds the system time limit during execution, the
system aborts the job, which may corrupt the database. Once the database is
corrupted, it is difficult, if not impossible, to restart the job.
However if the required execution time exceeds the time specified on the NX
Nastran TIME statement, NX Nastran cleanly terminates the job. This allows you to
restart the job later. Also, NX Nastran estimates the CPU time required to perform
certain operations during execution. If the time remaining isn’t sufficient to
complete a particular operation, the job terminates without wasting any additional
computer resources.
See Also
• “TIME” on page 158 of the NX Nastran Quick Reference Guide
DIAG Statement (Optional)

You can use the DIAG statement to request a variety of additional diagnostic
information or output and/or to modify the solution.
See Also
• “DIAG” on page 124 of the NX Nastran Quick Reference Guide
• “Other Optional Diagnostic Information” on page 67 of the NX Nastran
User’s Guide
The CEND Statement (Required)

The CEND statement is a required statement that designates the end of the
Executive Control section (and the beginning of the Case Control section). Its
format is simply:
CEND
For example:
ID SIMPLE,MODEL
SOL 101
TIME 5
www.cadfamily.com
CEND EMail:[email protected]
See Also
• “CEND” on page 116 of the NX Nastran Quick Reference Guide
3.5 Case Control Section (Required)

The Case Control section contains commands. It always follows the Executive
Control section and precedes the Bulk Data section. You use the Case Control section
to:
• define/select analysis subcases (multiple loadings in a single job execution)

• define SETS to specify and control the type of analysis output produced
(e.g., forces, stresses, and displacements)
• select output requests for printing, punching and plotting
• define titles, subtitles and labels for documenting the analysis
For the truss example, a title, subtitle, and label are defined. These labels are printed
on each page of the .f06 output file. The LOAD = 10 command instructs NX Nastran
to apply the loading defined by the FORCE entry with an ID of 10 in the Bulk Data
section. The SPC = 11 command instructs NX Nastran to apply the constraints
defined by the SPC1 entry with an ID of 11 in the Bulk Data section. Printed
displacements for all the grid points and the forces and stresses within each member
are requested with the DISPLACEMENT = ALL, FORCE = ALL, and STRESS = ALL
commands. By default, these output requests are printed to the .f06 output file.
See Also
• “Case Control Commands” on page 159 of the NX Nastran Quick Reference

Guide.
Case Control Format

The Case Control section must end with the BEGIN BULK delimiter.
Defining SUBCASES for Loading

The SUBCASE case control commands lets you define subcases within your analysis.
Subcases let you perform an analysis with different combinations of loads and
boundary conditions within a single run.
To denote a subcase, you use the SUBCASE command together with an ID operand
that identifies the subcase number, such as:
SUBCASE i
where i is an integer identifier for the subcase.
CHAPTER 3 33
Note: You can also use the MASTER Case Control command to define a master
subcase. A MASTER subcase applies to all of the subcases that follow until you
define a new MASTER subcase.
See Also
• “SUBCASE” on page 388 of the NX Nastran Quick Reference Guide
• “MASTER” on page 280 of the NX Nastran Quick Reference Guide
Defining Output
By default, NX Nastran doesn’t output any results. With NX Nastran, you must use
Case Control commands to explicitly request the results you want. NX Nastran
output is requested in the Case Control Section.
Echoing Bulk Data in the Output

An NX Nastran model resides in a .DAT text file. The .DAT file is submitted to NX
Nastran and an .f06 results file is produced.
The following Case Control command will produce sorted and unsorted model file
listings, called echoes, at the beginning of the .f06 file:
ECHO=BOTH
The unsorted input file is an exact copy of the Executive Control, Case Control, and
Bulk Data Sections of the input (.DAT) file, including comment ($) entries. The
sorted input file is a listing of the Bulk Data Section with entries rearranged in
alphabetical order and with comments removed. In addition, the sorted Bulk Data
is expanded to ten fields, each eight columns wide. Therefore, if the .DAT file is
entered in free field format, it will appear in small field format in the sorted Bulk
Data listing. This small field format listing is especially helpful when reviewing the
model’s Bulk Data Section.
Examples of unsorted and sorted input listings are shown in Listing 3-2 and
Listing 3-3.
Listing 3-2 Unsorted Listings
JULY 10, 2003 NX NASTRAN 7/10/2003 PAGE 1
N A S T R A N E X E C U T I V E C O N T R O L D E C K E C H O
ID MPM,EXAMPLE1
SOL 101
TIME 100
CEND
Listing 3-2 Unsorted Listings (continued)
SIMPLY SUPPORTED BEAM JULY 10, 2003 NX.NASTRAN 07/10/2003 PAGE 2

WITH CONCENTRATED FORCE
C A S E C O N T R O L D E C K E C H O
CARD
COUNT
1 ECHO=BOTH
2 DISP=ALL
3 STRESS=ALL
4 FORCE=ALL
5 SPCF=ALL
6 SPC=100
7 LOAD=10
8 TITLE=SIMPLY SUPPORTED BEAM
9 SUBTITLE=WITH CONCENTRATED FORCE
10 $
11 BEGIN BULK
SIMPLY SUPPORTED BEAM JULY 10, 2001 NX NASTRAN 7/10/2003 PAGE 3

I N P U T B U L K D A T A D E C K E C H O
.1 .. 2 .. 3 .. 4 .. 5 .. 6 .. 7 .. 8 .. 9 .. 10 .
$ DEFINE GRID POINTS
GRID,1,,0.,0.,0.
GRID,2,,10.,0.,0.
GRID,3,,20.,0.,0.
GRID,4,,30.,0.,0.
$
$ DEFINE CBAR ELEMENTS
CBAR,1,101,1,2,0.,1.,0.
CBAR,2,101,2,3,0.,1.,0.
CBAR,3,101,3,4,0.,1.,0.
$
$ DEFINE CBAR ELEMENT CROSS SECTIONAL PROPERTIES
PBAR,101,201,2.,.667,.1667,.458,,,+PB1
+PB1,1.,.5,-1.,.5
$
$ DEFINE MATERIAL PROPERTIES
MAT1,201,30.E6,,.3
$
$ DEFINE SPC CONSTRAINT SET
SPC1,100,12345,1
SPC1,100,12345,4
$
$ DEFINE CONCENTRATED FORCE
FORCE,10,3,,-100.,0.,1.,0.
$
ENDDATA
INPUT BULK DATA CARD COUNT = 26
CHAPTER 3 35
Listing 3-3 Sorted Listing

CARD
COUNT . 1 .. 2 .. 3 .. 4 .. 5 .. 6 .. 7 .. 8 .. 9 .. 10 .
1- CBAR 1 101 1 2 0. 1. 0.
2- CBAR 2 101 2 3 0. 1. 0.
3- CBAR 3 101 3 4 0. 1. 0.
4- FORCE 10 3 -100. 0. 1. 0.
5- GRID 1 0. 0. 0.
6- GRID 2 10. 0. 0.
7- GRID 3 20. 0. 0.
8- GRID 4 30. 0. 0.
9- MAT1 201 30.E6 .3
10- PBAR 101 201 2. .667 .1667 .458 +PB1
11- +PB1 1. .5 -1. .5
12- SPC1 100 12345 1
13- SPC1 100 12345 4
ENDDATA
TOTAL COUNT= 14
Other options for the ECHO command include:
ECHO = SORT Prints only sorted Bulk Data (this is the default)
ECHO = UNSORT Prints only unsorted Bulk Data
ECHO = NONE Turns off the Bulk Data listing
ECHO = PUNCH Prints a sorted echo of the Bulk Data to a separate file
While learning to use NX Nastran by using small models, you should set ECHO to
BOTH. The resulting listings in the .f06 file will not occupy very much space and
having the model information available can be useful. However, when the models
become very large it may be best to use ECHO = PUNCH or even ECHO = NONE.
Printing Bulk Data Using the ECHO Case Control Command

You can use the ECHO Case Control command to control how the Bulk Data section
is printed to the output file. For example, you can use ECHO to specify whether you
want NX Nastran to print the Bulk Data in sorted or unsorted format. For example,
to echo the Bulk Data in the unsorted form, you use the following ECHO command:
ECHO = UNSORT
See Also
• “ECHO” on page 230 of the NX Nastran Quick Reference Guide
The output requests, such as DISPLACEMENT, FORCE, STRESS, etc., are needed
whenever data recovery quantities are to be computed, even if they are not printed.
Such is the case when you use a postprocessor to view the results. Typical output
requests are as follows:
SET 1 = 5, 6, 7
SET 3 = 1, 5, 9
STRESS = ALL
DISP(PLOT) = 1 $ WHERE 1 IS THE ID OF A SET OF GRID POINTS
ELFORCE(PUNCH) = 3 $ WHERE 3 IS THE ID OF SET OF ELEMENTS
Types of Output
There are two main types of output that you can request in NX Nastran:
• grid point output

• element output
In a small or medium-sized model, it may be best to simply request output for all grid
points or elements. However, in a large model, you may want to be more selective.
Grid Point Output (DISP, SPCF, OLOAD, GPFORCE)
Requests for output quantities that occur at grid points include the following (n is a
SET ID, and ALL requests that all quantities be printed):
DISP Requests displacements for grid points. In general, you should

use DISP=ALL as displacement output takes up very little space.
SPCFORCE Requests forces of single point constraint (SPCFORCES). In

general, you should use SPCF = ALL so you can examine reaction
forces during the model validation process.
OLOAD Requests the set of applied loads in static analysis.
GPFORCE Generates a table containing a grid point force balance at the

selected grid points. This is useful for determining load paths,
contributions of applied loads to element response, and the effects
of initial thermal strain. Contributors to the grid point force
balance table include applied loads, SPC forces, and element
elastic forces.
www.cadfamily.comElement Output (STRESS, FORCE, STRAIN, ESE)

EMail:[email protected]
CHAPTER 3 37
Requests for element-based output quantities include the following:
STRESS Requests element stresses for a set of structural

elements.
FORCE Requests element forces to be calculated for a set

of structural elements.
STRAIN Requests the strains for a set of plate elements.
ESE Requests the strain energy for a set of elements.
CQUAD4 stresses, strains, and forces are available at corner grid points, with
output at the element center the default. CQUAD4 center and corner output is
obtained using the STRESS, STRAIN, and FORCE Case Control commands as
follows:
STRESS(CORNER) = {ALL or n}
STRAIN(CORNER) = {ALL or n}
FORCE(CORNER) = {ALL or n}
Where { } indicates that a choice of ALL or n is mandatory, but the braces are not
included.
Only one type of element output (center or corner) is supported per run..
Selecting the Destination File for the Output

For each type of results you request, you can specify where you want NX Nastran
to print the results.
• The PRINT option provides printed results in the .f06 output file. It is the
default option for most result types.
• The PUNCH option prints the results in the ASCII format .pch file using a
“punch” format (80 column width per line). Writing results to the .pch file
is useful if you want to export your results to other programs.
• The PLOT option instructs NX Nastran to calculate the requested results,
but not print them. In general, you use this option when you want to view
the results as plots or examine them with a post processing program. This
option is also useful when you’re working with large models where the
quantity of printed output would be excessive but you still need NX
Nastran to recover the specified data for post processing.
Formatting Printed Output

You can include up to three title lines per page of NX Nastran output with the TITLE,
SUBTITLE, and LABEL Case Control commands. You can use any character string.
Titles are optional but highly recommended as they can help you distinguish
between your jobs.
Example
These Case Control commands
TITLE=SIMPLY SUPPORTED BEAM

SUBTITLE=WITH CONCENTRATED FORCE
LABEL=NX NASTRAN USER’S GUIDE
produce the following headings on each page of output:

NX NASTRAN GETTING STARTED USER’S GUIDE
D I S P L A C E M E N T V E C T O R
POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 0.0 0.0 0.0 0.0 -2.221112E-04
2 G 0.0 -1.943473E-03 0.0 0.0 0.0 -1.388195E-04
3 G 0.0 -2.221112E-03 0.0 0.0 0.0 1.110556E-04
4 G 0.0 0.0 0.0 0.0 0.0 2.776390E-04
Using the SET Command to Limit Output

In many cases, particularly in larger models, you may want to limit the output to a
particular set of grid points or elements. In NX Nastran, you can use the SET Case
Control command to identify a group of grid points and/or elements as a specific set
ID.
For example, suppose you want to output all of the grid point displacements for
Subcase 1 and the grid point displacements for grid points 3 and 4 for Subcase 2. In
addition, you want the element force output for element 3 for Subcase 1 and elements
3 and 4 for Subcase 2. A Case Control section that meets these requirements is shown
in Listing 3-4.
CHAPTER 3 39
Listing 3-4 Controlling Output with Set Commands
$
$
ID LINEAR,TRUSS3
SOL 101
TIME 2
CEND
TITLE = SET EXAMPLE
LOAD = 10
SPC = 11
DISPLACEMENT = ALL
SET 1 = 3,4
SET 2 = 3
$
SUBCASE 1
LABEL = POINT LOAD AT GRID POINT 4 LOAD = 10
ELFORCE = 2
$
SUBCASE 2
LABEL = POINT LOAD AT GRID POINT 3
LOAD = 11
DISPLACEMENT =1
ELFORCE = 1
BEGIN BULK
Set Information from Getting Started!
A Case Control set is a collection of grid point IDs or element IDs for use in output
requests. Case Control sets are used to obtain output for a selected portion of the
model. Case Control sets are defined with the SET command according to the
following formats:
SET n = ALL
SET n = i 1 , i 2 , i 3 , ..., i 12 THRU i 28 , i 35 , ..
where n is the set identification number and i1 , i 2 , i3 ,...etc., are entity identification
numbers, e.g. grid point numbers.
Example
Consider the Case Control Section shown below:
CEND
TITLE=OUTPUT SELECTION EXAMPLE
SUBTITLE=ILLUSTRATES USE OF SETS
LOAD=15
SET 1=3,4,7,9,11
SET 5=2,9,15 THRU 21,23
DISP=1
FORCE=1
STRESS=5
BEGIN BULK
DISP is a grid point output quantity, so displacements for grid points 3, 4, 7, 9, and
11 will be printed. FORCE is an element output quantity, so forces for elements 3, 4,
7, 9, and 11 will be printed. Note that grid quantities and element quantities can
share the same set, since a set is simply a list of numbers. STRESS is an element
quantity, so the stresses in elements 2, 9, 15 through 21 (inclusive), and 23 will be
calculated and printed.
3.6 Bulk Data Section

The Bulk Data section always follows the Case Control section and begins with the
BEGIN BULK delimiter. In NX Nastran, Bulk Data entries contain all the information
necessary to define the finite element model, such as:
• geometry
• coordinate systems
• finite elements
• element properties
• loads
• boundary conditions
• material properties
The Bulk Data section generally constitutes the majority of the content of the input
file. The order of Bulk Data entries is not critical (continuations without continuation
identifiers are an exception to this rule and are discussed later).
Field 10 of the Bulk Data entry is used for two purposes. If the Bulk Data entry does
not have a continuation line, Field 10 may be used as an optional comment field. If
the Bulk Data entry has a continuation line, field 10 is used for the continuation
identifier. The continuation identifier must be unique with respect to all the other
identifiers in your Bulk Data Section.
If you leave a data field blank, the software uses the appropriate default. Any
applicable defaults for a particular field are listed in the NX Nastran Quick Reference
Guide. There are certain situations that allow either an integer or a real number in a
data field. However, in these situations, whether you specify an integer or a real
number in a data field can affect the problem’s solution.
Consider the definition for grid point 2 of the truss model shown in Figure 3-3. The
name of the entry is GRID, which begins in column 1 of field 1. The grid point ID (2
in this case) must be an integer (no decimal point) greater than 0. Fields 3 and 7
represent coordinate system IDs and also must be integers. Since these fields are
CHAPTER 3 41
blank, the default of 0 is used. Fields 4, 5, and 6 represent the physical location of
the grid point, and they must be entered as real numbers. Optional fields are field
8, which is used to define permanently constrained degrees of freedom, and field 9,
the superelement ID field. If they are used, only integers are acceptable. Since the
GRID entry does not have a continuation line, field 10 may be used as a comment,
if desired.
1 2 3 4 5 6 7 8 9 10
GRID ID CP X1 X2 X3 CD PS SEID
GRID 2 0.0 120. 0. 2345
Figure 3-3 A Bulk Data Example: the GRID Entry
See Also:
• “Bulk Data Entries” on page 755 of the NX Nastran Quick Reference Guide
Bulk Data Section Format Rules

You must begin the Bulk Data section with the BEGIN BULK delimiter. The last
entry must be the ENDDATA delimiter. The BEGIN BULK entry is in free field
format. The ENDDATA delimiter must begin in Column 1 or Column 2. The
ENDDATA delimiter also signifies the end of the NX Nastran input file.
Note: You must include both BEGIN BULK and ENDDATA in your input file, even
if you’re not introducing any new Bulk Data is being introduced into the problem
or if all of the Bulk Data is coming from an alternate source, such as user-generated
input.
NX Nastran Bulk Data contains ten fields per input data entry:
• The first field contains the character name of the Bulk Data item, such as
GRID, CBAR, MAT1.
• The second through ninth fields contain data input information for the
Bulk Data entry.
• The tenth field never contains data. It is reserved for entry continuation
information, if applicable.
Sorting Bulk Data

You can enter Bulk Data entries in any order since NX Nastran automatically sorts
them (using the XSORT module) prior to the execution of the Input File Processor
(IFP) module.
However, for very large problems, this sorting process can be CPU intensive. If you
anticipate submitting the same input file multiple times, you should use the
ECHO=PUNCH Case Control command to generate a sorted Bulk Data file in the
first analysis. Then, you can use the sorted Bulk Data input for subsequent analyses.
Note: Alternatively, the sorted Bulk Data is stored on the database and is available
using the Restart feature.
See Also
• “XSORT” on page 1511 of the NX Nastran DMAP Programmer’s Guide

• “IFP” on page 1077 of the NX Nastran DMAP Programmer’s Guide
Storing Bulk Data in the Database for Restarts

To minimize the handling of a large number of entries, NX Nastran solution
sequences let you store the Bulk Data in the database. That way, you can easily
modify it on subsequent runs.
For any initial analysis (also referred to as a “cold start”), you must submit the entire
Bulk Data section of the input file. For subsequent analyses, if you save the database
from any previous run, the Bulk Data exists in the database in sorted form where you
can modify it and reuse it in a restart analysis.
When you perform a restart, the software adds the Bulk Data entries contained in the
input file to the Bulk Data stored in the database. You can use a ”/” entry to remove
entries are from the database.
See Also
Understanding the Relationship Between Bulk Data

Entries and Case Control Commands
Most of the entries in the Bulk Data section do not need to be called out by a Case
Control command to be included in the model. However, for loads and constraints,
the Bulk Data and Case Control sections of the input file are carefully interrelated.
• In the Bulk Data section, you specify actual load and constraint entries.
• In the Case Control section, you specify which of the load and constraint
entries you defined in the Bulk Data section to use in a given analysis.
In other words, NX Nastran only includes the load and constraint entries you define
in the Bulk Data section if they’re specified in the Case Control section. Therefore,
you can have load and constraint entries in the Bulk Data section that aren’t specified
in the Case Control. However, those entries won’t be used during the analysis.
CHAPTER 3 43
3.7 Using Parameters

Each NX Nastran solution sequence is designed with a specific set of default values
and solution paths. In the majority of the cases, the default values are the ones best
suited for your model. There are occasions, however, when you may want to
deviate from the default values. Several methods are available in NX Nastran to
perform this task: DMAP (Direct Matrix Abstraction Program) modules,
NASTRAN statements, and parameters.
Using parameters is the most straightforward method to change defaults in NX

Nastran. Depending on the parameter, you can specify it either as a PARAM Bulk
Data entry and/or a PARAM Case Control command. The format of each is as
follows:
PARAM,xname,xvalue
where xname and xvalue are the name and value of the PARAMeter, respectively.
PARAM,WTMASS,0.00259
For example, with PARAM,WTMASS, the software multiplies the terms of the
structural mass matrix are multiplied by the value of this parameter.
In NX Nastran, parameters are used extensively in the solution sequences for input
of scalar values and for requesting special features. You can specify values for
parameters using either the PARAM Bulk Data entry or the PARAM Case Control
command.
See Also
• “PARAM” on page 326 of the NX Nastran Quick Reference Guide (for the
PARAM Case Control command)
• “PARAM” on page 1503 of the NX Nastran Quick Reference Guide (for the
PARAM bulk data entry)
• “Parameters” on page 563 of the NX Nastran Quick Reference Guide (for a
complete description of all parameters)
• NX Nastran DMAP Programmer’s Guide (for a description of the manner
in which NX Nastran processes parameters)
Parameters in the Superelement Solution Sequences

The specification of PARAMeters differs between the rigid formats and the
superelement solutions sequences. These differences require different user action in
placing the PARAM entries in the NX Nastran section, and in changing PARAMeter
values on restarts.
The program module reads the Case Control section for PARAM entries. If PARAM
entries exist, the VPS table is updated with the values found in the Case Control
section. The PARAM values are updated at the beginning of the generation,
assembly, reduction, and the data recovery loops of the superelement solution
sequences. This allows the values of some user-supplied parameters to change for
each superelement during the course of one run.
In the nonlinear solution sequences the PARAM values are updated at the beginning
of the iteration loop. This allows the values of some user-set parameters to vary
between subcases. These parameters may appear in either the Case Control and/or
Bulk Data section. Other PARAMeters must appear in the Bulk Data section. If they
appear in the Case Control section, they will not cause a preface error, but will be
ignored. Other parameters may be reset to default values during the course of the
run. This type should be set only in the subcase of the Case Control section.
See Also
• Table 5-1 and Table 5-2 in the NX Nastran Quick Reference Guide (for
parameter applicability)
The PARAM Case Control command is treated in the same manner as constraint and
output commands. Parameters set above the subcase level are effective in every
subcase. If there is also a parameter value set in the subcase, it takes precedence. Only
one value of a parameter may be current at any time during the execution of the run.
These rules imply an order of precedence if several forms are used in the same run.
The rules are illustrated in the following example.
A partial subcase structure is listed on the following page. The subcases are listed in
ascending superelement number order. PARAM,GRDPNT is varied between them.
CHAPTER 3 45
Subcase Structure, Sorted by SElD

SUBCASE 1
SUPER = 1
...
SUBCASE 2
SUPER = 2
PARAM,GRDPNT,100
...
SUBCASE 3
SUPER = 3
PARAM,GRDPNT,200
...
SUBCASE 4
SUPER = 4
...
SUBCASE 100 $ RESIDUAL STRUCTURE
PARAM,GRDPNT,-1
...
BEGIN BULK
PARAM,GRDPNT,0
Processing Order Value of GRDPNT

SEID
1 0
2 100
3 200
4 0
0 –1
The Bulk Data entries are processed at the beginning of the run. Any PARAM
entries in the Bulk Data section set the value of that parameter until it is changed by
a Case Control command.
• Superelement 1 is processed first. Since it has no PARAM,GRDPNT entry
the value of 0, as set by the Bulk Data entry PARAM,GRDPNT,0 is used.
• The value for superelement 2 is set at 100 by a Case Control command.
• Superelement 4 is processed next. Its subcase contains no PARAM entry,
so that the value of the parameter that remains is set by the bulk data.
• The values for superelement 3 and the residual structure are set by Case
Control commands.
An alternate method of parameter specification is to set the value most used above
the subcase level, and exceptional values within the subcase. For example, if
PARAM,GRDPNT,0 were placed above the subcase level and the values of 100, 200,
and -1 placed in subcases 2, 3 and 100, respectively, the value used in each
superelement will be the same as described above, regardless of processing order.
For the unstructured solution sequences, NX Nastran doesn’t store the VPS table in
the database. Therefore, it is not recovered on restarts. Therefore, parameters set by
the user must be in the NX Nastran section if the default values are not wanted.
4
CHAPTER
NX Nastran Output Files
■ Overview of the NX Nastran Output Files
■ Understanding the .f06 File
■ Understanding the .f04 File
■ Understanding the .log File
■ Understanding the .plt and .pch Files
■ Understanding the .DBALL and .MASTER Files
■ Understanding .xdb and .op2 Files
4.1 Overview of the NX Nastran Output Files

When you submit an input file to NX Nastran for execution, the software generates
a number of different output files. Some files are generated automatically, while
other files are generated only by specific requests you make in the input file or on the
command line when you submit the job.The following table shows the different
types of NX Nastran output files.
File File
How Created Brief Description
Extension Format
.f06 ASCII automatically Main output file with printed output,
such as displacements and stresses.
.f04 ASCII automatically Provides history of assigned files, disk
space usage, and modules used during
the analysis.
.log ASCII automatically Gives a summary of the command line
options used and the execution links
.DBALL binary automatically A database containing the input file,
assembled matrices, and solutions.
Used also for restarting the run for
additional analysis or output.
.MASTER binary automatically Contains the master directory of the
files used by the run and the physical
location of those files on your system.
Needed for a restart.
.plt binary by request Contains the plot information requested
by the plotter commands specified in
input file.
.pch ASCII by request Contains the punch output as requested
in the input file.
.op2 binary by request Graphics database file used for
postprocessing.
.xdb binary by request Graphics database file used for
postprocessing.
Table 4-1 Types of Files Output by NX Nastran
In addition to the files listed above, the software also generates several temporary
(scratch) files during the analysis which are automatically deleted when the solve
www.cadfamily.comcompletes.
CHAPTER 4 49
See Also
• “Automatic Deletion of Scratch Data Blocks” on page 40 of the NX

Nastran DMAP Programmer’s Guide
Note: By default, the software writes the .f06, .f04, and .log files separately.
However, you can use the “append” keyword (append=yes) on the command line
when you run your job to have the software combine them into a single file with a
.out extension.
See Also
• “append” of the NX Nastran Installation and Operations Guide

The following sections describe the different output file types in more detail.
4.2 Understanding the .f06 File

The output file that you will use most frequently is the .f06 file. The software writes
this file to FORTRAN unit 6. The .f06 file contains all the requested results from your
analysis, such as the displacements, stresses, and element forces, as well as any
diagnostic messages. The information in the .f06 file is critical for model checkout
and debugging.
• The output from data recovery and plot modules is all optional and its
selection is controlled by commands in the Case Control section.
Although most of the solution sequence output is optional and requested in the
Case Control Section, some of the printer output is automatic or under control of
DIAG statements in the Executive Control Section.
See Also
• “Flushing .f04 and .f06 Output to Disk (UNIX)” in the NX Nastran
Installation and Operations Guide
Controlling Printed Output

In NX Nastran, there are two different formats in which the software prints results:
SORT1 and SORT2.
Understanding the SORT1 Format

With the SORT1 format, the software presents the analysis outputs as a tabular
listing of the selected output for each grid point in each individual subcase. With
SORT1, the software prints the output for each loading condition on any page. This
is the default for static analysis, frequency response, steady state heat transfer, real
and complex eigenvalue analysis, flutter analysis, and buckling analysis.
If you select SORT1 in a transient solution for one or more of the commands listed
below, then the software outputs all remaining commands in the input file in SORT1
format:
ACCE, DISP, ENTH, FORC, HDOT, MPCF, OLOA, SPCF, STRA, STRE,
VELO
Understanding the SORT2 Format

With the SORT2 format, the software presents the analysis output as a tabular listing
of the subcases for each selected output option. With SORT2, the software prints the
output for each grid point or element on a new page. Consequently, the SORT2
option can produce a large number of output pages. Therefore, SORT2 should
generally be used only when you’re performing dynamic solutions.
The SORT2 format is the default output format in transient response analysis
(structural and heat transfer). The SORT2 option isn’t available when you’re
performing real eigenvalue (including buckling) analyses, complex eigenvalue
analyses, or flutter analyses.
If you select SORT2 in a frequency response solution for one or more of commands
listed below, then the software outputs all remaining commands in the input file in
SORT2 format:
ACCE, DISP, FORC, MPCF, OLOA, SPCF, STRA, STRE, VELO
Controlling General Output Options

The printer output is designed for 132 characters per line, with the lines per page
controlled by the LINE command in the Case Control Section. The default is LINE =
50 for 11-inch paper.
Optional titles are printed at the top of each page from information in the Case
Control Section. You can define these titles at the subcase level. All pages of the
output are automatically dated and numbered.
A few printer output items are under the control of PARAM Bulk Data entries.
See Also
• “Parameters” on page 563 of the NX Nastran Quick Reference Guide

The following output is either automatically or optionally provided during
execution:
• NX Nastran title page – automatic

• File Management Section echo – automatic if statements are present
• Executive Control Section echo – automatic
CHAPTER 4 51
• Case Control Section echo – automatic

• Unsorted Bulk Data Section echo – optional, selected in the Case Control
Section
• Sorted Bulk Data Section echo – automatic, unless suppressed in the Case
Control Section
By default, the output file contains an echo (listing) of the input file with the Bulk
Data sorted alphabetically as shown in Figure 4-1.
ID LINEAR,TRUSS1
SOL 101
TIME 2
CEND
LINEAR STATICS USER’S GUIDE JUNE 18, 2004 NX NASTRAN 6/13/04 PAGE 3
TRUSS STRUCTURE
POINT LOAD AT GRID POINT 4
C A S E C O N T R O L D E C K E C H O
CARD
COUNT
1 TITLE = LINEAR STATICS USER’S GUIDE
2 SUBTITLE = TRUSS STRUCTURE
3 LABEL = POINT LOAD AT GRID POINT 4
4 LOAD = 10
5 SPC = 11
6 DISPLACEMENT = ALL
7 ELFORCE = ALL
8 ELSTRESS = ALL
9 BEGIN BULK
INPUT BULK DATA CARD COUNT = 33
NX NASTRAN USER’S GUIDE JUNE 18, 2004 NX NASTRAN 6/13/04 PAGE 4

TRUSS STRUCTURE
CARD
COUNT . 1 .. 2 .. 3 .. 4 .. 5 .. 6 .. 7 .. 8 .. 9 .. 10 .
1- CROD 1 21 2 3
2- CROD 2 21 2 4
3- CROD 3 21 1 3
4- CROD 4 21 1 4
5- CROD 5 21 3 4
6- FORCE 10 4 1000. 0. -1. 0.
7- GRID 1 0. 0. 0. 3456
8- GRID 2 0. 120. 0. 3456
9- GRID 3 600. 120. 0. 3456
10- GRID 4 600. 0. 0. 3456
11- MAT1 22 30.E6 .3
12- PROD 21 22 4. 1.27
13- SPC1 11 123456 1 2
ENDDATA
TOTAL COUNT= 14
TRUSS STRUCTURE
USER INFORMATION MESSAGE
ORIGIN OF ASSEMBLY BASIC COORDINATE SYSTEM WILL BE USED AS REFERENCE LOCATION.
RESULTANTS ABOUT ORIGIN OF ASSEMBLY BASIC COORDINATE SYSTEM IN ASSEMBLY BASIC SYSTEM COORDINATES.
OLOAD RESULTANT
T1 T2 T3 R1 R2 R3
1 0.0000000E+00 -1.0000000E+03 0.0000000E+00 0.0000000E+00 0.0000000E+00 -6.0000000E+05
TRUSS STRUCTURE
*** USER INFORMATION MESSAGE 5293 FOR DATA BLOCK KLL
LOAD SEQ. NO. EPSILON EXTERNAL WORK EPSILONS LARGER THAN .001 ARE FLAGGED WITH ASTERISKS
1 -5.2155330E-15 6.4518387E+01
Figure 4-1 truss1.f06 Output File
CHAPTER 4 53
TRUSS STRUCTURE


1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 .0 .0 .0 .0
3 G 1.247578E-02 -1.285377E-01 .0 .0 .0 .0
4 G -1.252422E-02 -1.290368E-01 .0 .0 .0 .0

TRUSS STRUCTURE

F O R C E S I N R O D E L E M E N T S ( C R O D )
ELEMENT AXIAL ELEMENT AXIAL

ID. FORCE TORQUE ID. FORCE TORQUE
1 2.495156E+03 .0 2 2.554449E+03 .0
3 -2.544570E+03 .0 4 -2.504844E+03 .0
5 4.990313E+02 .0

TRUSS STRUCTURE
S T R E S S E S I N R O D E L E M E N T S ( C R O D )
ELEMENT AXIAL SAFETY TORSIONAL SAFETY ELEMENT AXIAL SAFETY TORSIONAL SAFETY
ID. STRESS MARGIN STRESS MARGIN ID. STRESS MARGIN STRESS MARGIN
1 6.237891E+02 .0 2 6.386123E+02 .0
3 -6.361426E+02 .0 4 -6.262109E+02 .0
5 1.247578E+02 .0
Figure 4-2 Truss1.f06 Output File (Continued)
Understanding the General Analysis Output

OLOAD Resultant
After the input file echo, the software prints the “OLOAD RESULTANT.” This
shows the resultant of the loads about the origin of the basic coordinate system in
this case.
User Information Message 5293

Next is the USER INFORMATION MESSAGE (UIM) 5293 which shows the epsilon
and external work for the run. The epsilon is a measure of the numeric conditioning
of the model, while the external work is the work due to the applied loads. Both of
these values are useful for model checkout.
See Also
• “Model Verification” on page 445 of the NX Nastran User’s Guide
In the truss example, output is requested for displacements, element forces, and
element stresses. The displacement request is the only grid point output. Element
forces and stresses are element output.
The last summary is the maximum of the applied loads.
After the summaries, the displacement for all of the grid points, the element forces,
and the element stresses is printed as requested in the Case Control Section.
For static analysis of the truss model, just one applied load is used. You can,
however, specify multiple loading conditions to be analyzed in a single input file if
desired. If you do specify multiple loading conditions, the OLOAD RESULTANT
and EXTERNAL WORK are shown for each selected loading conditions. The
external work is calculated as 1 ⁄ 2 { P l } T [ U l ] . In the case of enforced displacement, the
equivalent load is calculated as the constraint force required to impose the enforced
displacement. The external work printout in this case is meaningless.
Accuracy Diagnostic Information for Global Iterative Solver

When you’re using NX Nastran’s global iterative solver, the .f06 file also contains
important accuracy diagnostic information.
See Also
• “Accuracy Diagnostics” on page 104 of the NX Nastran Numerical Methods

User’s Guide
Grid Point Singularity Table

If singularities remain in the stiffness matrix at the grid point level, the software
automatically outputs a Grid Point Singularity Table (GPST) after the Grid Point
Singularity Processor (GPSP) executes. The table lists singular degrees-of-freedom in
the global coordinate system. You can have the software automatically constrain
singular degrees-of-freedom using the PARAM,AUTOSPC,YES option.
See Also
• “Constraint and Mechanism Problem Identification in

SubDMAP SEKR” on page 317 of the NX Nastran User’s Guide.
• “Understanding the Grid Point Singularity Table” on page 217 of the NX
• “AUTOSPC” on page 570 of the NX Nastran Quick Reference Guide
Note: You can print any of the matrices or tables that are prepared by the functional
modules using selected utility modules described in the NX Nastran DMAP
Programmer’s Guide. These utility modules can be scheduled at any point in a solution
sequence by using the alter feature. In general, they should be scheduled
CHAPTER 4 55
immediately after the functional module that generates the table or matrix to be
printed. However, you should check the calling sequence for the utility module, in
order to be certain that all required inputs have been generated prior to this point.
Understanding Analysis Type Specific Output

The following descriptions relate to automatic output that is associated with
particular types of solution sequences.
Detailed information on the force and stress output available for each element type
is given in the NX Nastran Element Library.
Real Eigenvalue Analysis

Each eigenvalue is identified with a mode number determined by sorting the
eigenvalues by their magnitude. The following summary is also automatically
printed for all methods:
• Mode number
• Extraction order (this is irrelevant under the Lanczos method)
• Eigenvalue
• Frequency (radians)
• Frequency (Hz)
• Generalized mass
• Generalized stiffness
Note: For all methods except the Inverse Power Method, if the orthogonality
criterion fails, then the software issues User Warning Message 3034.
Lanczos Method Additional Summary Content

• Block size used
• Number of decompositions
• Number of roots found
• Number of (block) solves required
• One of the following termination messages:
Required number of eigenvalues found
Not all the roots found within the range
All eigenvalues found within the range
Insufficient time to find more eigenvalues
Inverse Power Method Additional Summary Content

• Number of eigenvalues extracted
• Number of starting points used
• Number of starting point moves
• Number of triangular decompositions
• Number of vector iterations
• The reason for termination is indicated by one of the following numbers:
Reason
Description
Number
1 Two consecutive singularities encountered while
performing triangular decomposition
2 Four shift points while tracking a single root
3 All eigenvalues found in the frequency range specified
4 Three times the number of roots estimated in the
frequency range have been extracted
5 All eigenvalues that exist in the problem have been found
6 The number of roots desired have been found
7 One or more eigenvalues have been found outside the
frequency range specified
8 Insufficient time to find another root
9 Unable to converge
• Largest off-diagonal modal mass term identified by mode pair

• Number of off-diagonal mass terms failing orthogonality criterion
Complex Eigenvalue Analysis (SOL 110?)

Each complex eigenvalue is identified with a root number determined by sorting the
complex eigenvalues by magnitude. The following summary of the complex
eigenvalues extracted is automatically printed for all methods:
• Root number
• Extraction order (this is irrelevant under the Lanczos method)
• Real and imaginary parts of the eigenvalue
• The coefficients f (frequency) and g (damping coefficient) in the following
representation of the eigenvalue ( p = α + iω ) :
CHAPTER 4 57
p = 2πf ( i – 1 ⁄ 2 g ) Eq. 4-1
Therefore,
f = ω Eq. 4-2
------
2π
– 2α C
g = ---------- = 2 -------- Eq. 4-3
ω C cr
Complex Lanczos Method

If you use the complex Lanczos method, the software also issues User Information
Messages 5444, 5445, and 5453.
Determinant Method
If you use the Determinant method, the software also prints:
• number of eigenvalues extracted

• number of passes through starting points.
• number of criteria changes
• number of starting point moves
• number of triangular decompositions
• number of failures to iterate to a root
• number of predictions outside region
• the reason for termination is indicated by one of the following integers:
Reason
Description
Number
2 All predictions for eigenvalues are outside the regions
specified
4 Matrix is singular at the first three starting points
• swept determinant functions for each starting point
Inverse Power Method

The following summary is automatically printed for each region specified when the
Inverse Power method is used:
• Number of eigenvalues extracted

• Number of starting points used
• Number of starting point or shift point moves
• Number of triangular decompositions
• Number of vector iterations
• The reason for termination is indicated by one of the following integers:
Reason
Description
Number
1 Two consecutive singularities encountered while performing
triangular decomposition
2 Four starting point moves while tracking a single root
3 All eigenvalues found in the region specified
4 Three times the number of roots estimated in the region have
been extracted
5 All eigenvalues that exist in the problem have been found
7 One or more eigenvalues have been found outside the region
specified
9 Unable to converge
Superelement Analysis
• Superelement mapping information that lists the membership of grid
points and elements in the superelements, the order in which the
superelements will be processed, and timing and space estimates for each
superelement
• A page header that identifies the superelement for which the printed
information on the page pertains
• Matrices wherein mechanisms are identified
• Resultants of external loads
• Maximum applied loads
• Maximum displacements
• Maximum SPC forces
CHAPTER 4 59
For SOL 101, if PARAM,INREL,-1 or PARAM,INREL,-2, the following data is

printed automatically:
INTERMEDIATE MATRIX...QRR
This is the rigid body mass matrix of the total structure, including superelements. It
is presented as a 6 × 6 matrix, measured at the reference grid point, in its global
coordinates. Masses on scalar points are ignored.
INTERMEDIATE MATRIX...QRL
This is the resultant of the applied loads, measured at the reference point. There is
one column for each loading condition.
INTERMEDIATE MATRIX...URACCEL
This is the user-supplied rigid body acceleration, input on the DMIG,UACCEL
entry. This output does not appear if the entry is not present.
INTERMEDIATE MATRIX...URA
This is the rigid body acceleration matrix, computed from the applied loads.
4.3 Understanding the .f04 File

The .f04 file contains information that helps you monitor and tune the performance
of your analysis job. For example, the .f04 file contains:
• a summary of physical file information that describes the files used for
DBsets
• a map showing memory allocation
• a “day log” that provides a DMAP execution summary
• a listing of memory and disk usage statistics
• a listing of database usage statistics that summarizes the I/O activity for
the DBsets
• a summary of physical file I/O activity for each database file
• module execution and statistical information
• other optional diagnostic information as requested by the DIAG executive
control command in the input file
If you’re performing an analysis using NX Nastran’s global iterative solver, the .f04
file also contains important performance diagnostic information about the solver.
The software writes the .f04 file to FORTRAN unit 4.
See Also
• “Flushing .f04 and .f06 Output to Disk (UNIX)” in the NX Nastran

Installation and Operations Guide
• “Performance Diagnostics” on page 106 of the NX Nastran Numerical
Methods User’s Guide
General Information
The .f04 file begins with a section that lists general information about the run.
• DAY TIME indicates the time of day as the run progresses.

• ELAPSED indicates the number of elapsed “wall clock” minutes:seconds at
the beginning on the job. This number may vary greatly depending on the
system load in a time-sharing environment.
• I/O SEC is an internally generated measure of I/O. This measure is a count
of the number of BUFFSIZE blocks transferred, divided by the value of
SYSTEM(84) (the estimated rate of number of blocks moved per second).
This number doesn’t include I/O for loading the executable file or I/O for
FORTRAN files.
• DEL I/O is the I/O seconds spent in the previous DMAP statement.
• CPU SEC shows the number of CPU seconds that had elapsed when this
line was printed.
In general, you should examine the above entries and look for large jumps
in the CPU SEC number. For example, the cost for “formatting” the output
(modules SDR2 and OFP) may be greater than the cost of generating the
numbers. In such situations, simple techniques such as reducing output
requests may save more time than other more well-known techniques such
as sequencing, OMITs, better matrix methods, etc. In short, you should be
aware of the expensive operations on each large run.
• DEL CPU is the CPU time spent in the previous DMAP statement.
Summary of Physical File Information

This summary table describes the physical files used for the DBSets. A sample of this
table, located near the top of the .f04 file, is shown below.
S U M M A R Y O F P H Y S I C A L F I L E I N F O R M A T I O N
ASSIGNED PHYSICAL FILE NAME RECL (BYTES) MODE FLAGS

------------------------------------------------------------ ------------ ---- -----
/tmp/65872_57.SCRATCH 8192 R/W L
/tmp/65872_57.OBJSCR 8192 R/W L
/tmp/65872_57.MASTER 8192 R/W L
/tmp/65872_57.DBALL 8192 R/W L
/tmp/65872_57.DBALL2 8192 R/W L
/tmp/65872_57.SCR300 8192 R/W L
www.cadfamily.com/nxnr/aix/SSS.MASTERA
EMail:[email protected] 8192 R/O L
CHAPTER 4 61
/nxnr/aix/SSS.MSCOBJ 8192 R/O L
FLAG VALUES ARE --

F FFIO INTERFACE USED TO PROCESS FILE
H HPIO INTERFACE USED TO PROCESS FILE
L FILE HAS BEEN LOCKED
M FILE MAPPING USED TO PROCESS FILE
R FILE BEING ACCESSED IN ’RAW’ MODE
** PHYSICAL FILES LARGER THAN 2GB FILES ARE NOT SUPPORTED ON THIS PLATFORM
In this summary:
• ASSIGNED PHYSICAL FILENAME is the physical FILENAME with any

symbols translated.
• RECL is the record length in bytes.
• MODE is the file access mode.
• R/W is read-write mode.
• R/O is read-only mode.
• The FLAGS column contains various letters depending on the capabilities
of the platform and user requests. The text below the table indicates
possible flag values for your specific platform.
• BUFFSIZE lists the buffer size of each DBset. This buffer size may be
defined by the INIT command for each DBset (File Management
Statement).
• CLUSTER SIZE lists the unit of allocation for GINO and executive blocks.
This unit may be defined on the INIT command.
• The TIME STAMP column lists the time the file was created. The format
is YYMMDDHHMMSS.
In this example, an INIT statement was used to create the DBALL DBSet with two
files using the logical names DBALL and DBALL2.
Below the summary is a message that indicates whether large files are available on
the platform on which you’re running.
Memory Allocation Map

The next section of the file is a map of memory allocation.
** MASTER DIRECTORIES ARE LOADED IN MEMORY.
USER OPENCORE (HICORE) = 3804612 WORDS
EXECUTIVE SYSTEM WORK AREA = 78605 WORDS
MASTER(RAM) = 30000 WORDS
SCRATCH(MEM) AREA = 204900 WORDS ( 100 BUFFERS)
BUFFER POOL AREA (GINO/EXEC) = 76183 WORDS ( 37 BUFFERS)
TOTAL NX NASTRAN MEMORY LIMIT = 4194300 WORDS

In this memory map:
• USER OPENCORE is the amount of memory available to the module for

computation purposes.
• EXECUTIVE SYSTEM WORK AREA is the space reserved for the executive
system.
• MASTER(RAM) is the space reserved to cache datablocks from the
MASTER DBSet.
• SCRATCH(MEM) AREA is the space reserved to cache datablocks from the
SCRATCH and SCR300 DBSets.
• BUFFER POOL AREA is the space reserved for the buffer pool.
• TOTAL NX Nastran MEMORY LIMIT is the total space allocated to NX
Nastran’s open core using the “memory” keyword.
See Also
• “Managing Memory” in the NX Nastran Installation and Operations Guide
Day Log
The Day Log portion of the .f04 is a DMAP execution summary. This log, in table
format, contains the vast majority of the information in the .f04. The beginning an
example Day Log is shown below:
DAY TIME ELAPSED I/O MB DEL_MB CPU SEC DEL_CPU SUB_DMAP/DMAP_MODULE MESSAGES
10:32:16 0:16 13.6 .3 .8 .0 SESTATIC 20 IFPL BEGN

10:32:16 0:16 13.7 .1 .8 .0 IFPL 29 IFP1 BEGN
10:32:16 0:16 13.7 .0 .8 .0 IFPL 39 XSORT BEGN
In the Day Log:
• DAY TIME is the time of day of the entry.

• ELAPSED is the elapsed time since the start of the job.
• I/O MB is the megabytes of I/O to the databases since the start of the job
• DEL_MB is the delta I/O since the previous entry.
• CPU SEC is the total CPU seconds since the start of the job.
• DEL_CPU is the delta CPU since the previous entry.
• SUB_DMAP/DMAP_MODULE” indicates the DMAP statement being
executed.
• MESSAGES are any messages issued by the module.
• BEGN and END indicate the start and end of the module or subDMAP.
CHAPTER 4 63
The number that precedes the module name on lines that terminate with
BEGN or END is the DMAP statement number. This number corresponds
to the statement number in the DMAP listing for the solution sequence
being used. The DMAP sequence listing will be printed as a part of the
output if you include either DIAG 14 or COMPILER LIST in the Executive
Control section of your input file.
Understanding the I/O MB Column

• The software computes the “I/O MB” value by multiplying SYSTEM(85),
which is incremented by one for each GINO I/O, by BUFFSIZE. This value
loses accuracy if the DBSets don’t have the same BUFFSIZE.
• If SYSTEM(84) is set to 0, the “I/O MB” column will be the number of
GINO I/Os.
• The software scales the “I/O MB” column by gigabytes. It appends a “G”
after each number if the value is greater than or equal to 100 000.
User Information Messages 4157 and 6439

The UIM 4157 text provides decomposition estimates when the preface of the
decomposition module completes. The format of this message varies slightly
depending on whether you selected either sparse or parallel methods.
This counterpart to UIM 4157 is UIM 6439, which provides actual information from
the completed decomposition process. These two messages are interspersed within
the Day Log at each decomposition.
The following is an example from a sparse decomposition:

*** USER INFORMATION MESSAGE 4157 (DFMSA) ---PARAMETERS FOR SPARSE DECOMPOSITION OF DATA
BLOCK KLL ( TYPE=RDP ) FOLLOW
MATRIX SIZE = 64 ROWS NUMBER OF NONZEROES = 260 TERMS
NUMBER OF ZERO COLUMNS = 0 NUMBER OF ZERO DIAGONAL TERMS = 0
CPU TIME ESTIMATE = 0 SEC I/O TIME ESTIMATE = 0 SEC
MINIMUM MEMORY REQUIREMENT = 20 K WORDS MEMORY AVAILABLE = 3804 K
WORDS
MEMORY REQR’D TO AVOID SPILL = 30 K WORDS
EST. INTEGER WORDS IN FACTOR = 1 K WORDS EST. NONZERO TERMS = 1 K
TERMS
ESTIMATED MAXIMUM FRONT SIZE = 11 TERMS RANK OF UPDATE = 16
*** USER INFORMATION MESSAGE 6439 (DFMSA) ---ACTUAL MEMORY AND DISK SPACE REQUIREMENTS FOR
SPARSE SYM. DECOMPOSITION
SPARSE DECOMP MEMORY USED = 30 K WORDS MAXIMUM FRONT SIZE = 11
TERMS
INTEGER WORDS IN FACTOR = 1 K WORDS NONZERO TERMS IN FACTOR = 1 K
TERMS
SPARSE DECOMP SUGGESTED MEMORY = 30 K WORDS
In UIM 4157:
• MINIMUM MEMORY REQUIREMENT is an estimate of the open core

memory that will allow the decomposition to run, but with heavy spilling
to disk.
• MEMORY REQR’D TO AVOID SPILL is the estimate of memory necessary

for the decomposition to run in “in core”, without spilling to disk.
Note: These two values represent memory requirement extremes. The
amount of memory necessary for optimal CPU performance is somewhere
between the two.
• ESTIMATED MAXIMUM FRONT SIZE is a function of the model and
affects the memory estimates described above.
• MINIMUM MEMORY REQUIREMENT is a function of the front
size.
• MEMORY REQR’D TO AVOID SPILL is a function of the square of
the front size.
• NUMBER OF NONZEROES is the size of the input matrix, multiply this
value by 8 to estimate the size of the input file in bytes.
• The sum of EST. INTEGER WORDS IN FACTOR and EST. NONZERO
TERMS is the size of the output matrix.
Note: Multiply the integer value by 4 and the nonzero value by 8 to estimate
the size of the output file in bytes.
• RANK OF UPDATE is the number of rows that will be simultaneously
updated during the decomposition. You can set by either the “rank”
keyword or SYSTEM(205).
Note: Setting SYSTEM(69)=64 causes NX Nastran to terminate after printing
UIM 4157. This can be useful for determining a job’s memory and disk
space requirements.
In UIM 6439:
• SPARSE DECOMP MEMORY USED states the actual memory used in the
decomposition process.
• SPARSE DECOMP SUGGESTED MEMORY is based on the execution of
the module. Using this value will result in optimal throughput
performance.
CHAPTER 4 65
Memory and Disk Usage Statistics

NX Nastran writes the memory and disk usage statistic tables after the job
completes. These tables indicate the maximum memory used by any sparse
numerical module and the maximum disk used by any module during the job. For
example:
+---------- SPARSE SOLUTION MODULES -----------+ +------------- MAXIMUM DISK USAGE -------------+
HIWATER SUB_DMAP DMAP HIWATER SUB_DMAP DMAP
(WORDS) DAY_TIME NAME MODULE (MB) DAY_TIME NAME MODULE
517786 04:35:44 SEKRRS 18 DCMP 15.625 04:35:48 SESTATIC 186 EXIT
In the sparse solution modules table:
• HIWATER WORDS is the maximum amount of open core used by certain

sparse numerical modules.
• DAY_TIME is the time of day the module ran.
• SUB_DMAP NAME is the name of the SUBDmap.
• DMAP MODULE indicates the line number and module name that made
the maximum request.
In the maximum disk usage table:
• HIWATER (MB) is the maximum amount of disk space used by any

module.
• DAY_TIME is the time of day the module ran.
• SUB_DMAP NAME is the name of the SUBDmap.
• DMAP MODULE indicates the line number and module name that made
the maximum request.
Database Usage Statistics

These statistics, provided in table format, summarize the I/O activity for the
DBSets.
*** DATABASE USAGE STATISTICS ***
+------------------ LOGICAL DBSETS ------------------+ +------------------------- DBSET FILES -------------------------+

DBSET ALLOCATED BLOCKSIZE USED USED FILE ALLOCATED HIWATER HIWATER I/O TRANSFERRED
(BLOCKS) (WORDS) (BLOCKS) % (BLOCKS) (BLOCKS) (MB) (GB)
MASTER 5000 2048 143 2.86 MASTER 5000 143 1.117 .010
DBALL 250000 2048 9 .00 DBALL 250000 9 .070 .000
DBALL2 300 1 .008 .000
OBJSCR 5000 2048 121 2.42 OBJSCR 5000 121 .945 .003
SCRATCH 500100 2048 19 .00 (MEMFILE 100 81 .633 .000)
SCRATCH 250000 1 .008 .000
SCR300 250000 1 .008 .000
============
TOTAL: .013
This statistical table contains two parallel tables:
• The LOGICAL DBSETS table lists each DBset

• The DBSET FILES tables lists the component files of the DBSet.
In these tables:
• DBSET is the name of the DBSet.

• ALLOCATED is the NX Nastran DBSet size limit in blocks.
• BLOCKSIZE is BUFFSIZE of the DBSet minus one.
• USED (BLOCKS) and USED % are the number of GINO and executive
blocks used at the end of the job as well as the percent of the DBSet actually
used.
• FILE is the file’s logical name associated with the DBSet to the left.
• ALLOCATED is the number of blocks allocated by NX Nastran to the file.
• HIWATER (BLOCKS) and HIWATER (MB) are the maximum number of
GINO and executive blocks and actually used during the execution.
• I/O TRANSFERRED is the amount of I/O to the file. The last line of this
column lists the total I/O transferred.
In this example, the MASTER and OBJSCR DBSets are each composed of one file. The
DBALL DBSet is composed of two files, DBALL and DBALL2; and the SCRATCH
DBSet has three components, MEMFILE, SCRATCH, and SCR300.
You can use this database statistics table to determine whether the DBSets and files
are appropriately sized and to assess the amount of I/O activity associated with each
file. You can obtain the best elapsed time performance can be obtained if the files
with the greatest activity are on different physical devices (and better yet, separate
I/O controllers or busses).
Summary of Physical File I/O Activity

This section of the .f04 file describes the physical file I/O for each database file.
*** SUMMARY OF PHYSICAL FILE I/O ACTIVITY ***
ASSIGNED PHYSICAL FILENAME RECL (BYTES) READ/WRITE COUNT MAP WSIZE (NUM) MAP COUNT
------------------------------------------------------------ ------------ ----------- --------------- -----------
/tmp/65872_57.SCRATCH 8192 1 128KB ( 4) 1
/tmp/65872_57.OBJSCR 8192 378 128KB ( 4) 24
/tmp/65872_57.MASTER 8192 1247 128KB ( 4) 11
/tmp/65872_57.DBALL 8192 26 128KB ( 4) 1
/tmp/65872_57.SCR300 8192 1 128KB ( 4) 1
/nxnr/aix/SSS.MASTERA 8192 162 N/A N/A
/nxnr/aix/SSS.MSCOBJ 8192 202 N/A N/A
In this summary:
• ASSIGNED PHYSICAL FILENAME, RECL, and MAP WSIZE and NUM

are repeated from the “Summary of Physical FIle Information” table.
• READ/WRITE COUNT is the number of GINO reads and writes that were
performed on the file.
CHAPTER 4 67
• MAP COUNT is the number of times the map window had to be

remapped (these columns are only present on systems supporting mapped
I/O).
You can use this summary to tune I/O performance. For mapped I/O systems, if the
map count approaches the number of reads and writes, you should increase the map
size and/or the number of maps.
• Increasing the number of maps is suggested if a module simultaneously

accesses more data blocks or matrices in a file than there are windows.
• Increasing the size of the windows is suggested if a file contains very large
data blocks or matrices.
You can obtain the best elapsed time performance, with or without mapping, if the
files with the greatest activity are on different physical devices, and better yet, on
separate I/O controllers or busses.
Other Optional Diagnostic Information

The DIAG Executive Control command has a number of different options that you
can use to have NX Nastran print additional, diagnostic information to the .f04 file.
Several of these options are described in detail below.
See Also
• “DIAG” on page 124 of the NX Nastran Quick Reference Guide
DIAG 8
If you include DIAG 8 in the Executive Control section of your input file, the
software prints matrix trailers as the matrices are generated. Terms in the DIAG 8
output include:
• the number of GINO blocks in the matrix (BLOCKS);
• the average number of terms per string (STRL)
• the number of strings in a matrix data block (NBRSTR)
• the number of bank terms in a matrix data block on the average (BNDL)
• the number of banks in a matrix data block (NBRBND)
• the average first nonzero row position (ROWL)
• the average of full bank width (BNDAVG)
• the maximum full bank width of all columns of data blocks (BNDMAX)
• the number of null columns in a data block (NULCOL)
DIAG 13
If you include DIAG 13 in your input file, the software includes the message REAL
CORE IS xxxx throughout the day log. It shows the value of the NASTRAN keyword
REAL. The message only appears after module executions which use the value. The
open core (i.e., working space) for other modules is set via HICORE
DIAG 19
If you include DIAG 19 in your input file, the software prints data for MPYAD and
FBS method selection in the Execution Summary table at the top of the .f04 file.
DIAG 49
If you include DIAG 49, the software includes a detailed analysis of timing data
(statistical information on module expected/actual CPU item) at the end of the .f04
file in three separate sections.
• The first section contains information about each module and submodule.
• The second section reports statistical information about the submodules
(FBS, MPYAD, and DCMP) grouped by specific methods. Instances where
the CPU estimate doesn’t pass the following criteria are output:
Estimated CPU – Actual CPU
------------------------------------------------------------------------------------ < 0.05
Actual CPU
or
Estimated CPU – Actual CPU < 5% of the total CPU of the job
• The third section contains information about the modules listed in
alphabetical order.
Comparing Different F04 Files with F04REPRT

The NX Nastran F04RERPT utility program contains a Perl script that allows you to
summarize and/or compare the contents of different .f04 files. This utility can be
helpful if you want to compare the relative performance of one or more jobs under
various configurations.
See Also
• “F04REPRT” in the NX Nastran Installation and Operations Guide
4.4 Understanding the .log File
The .log file contains system information, such as the name of the computer on which
the analysis was run. It also contains any system errors encountered during the
analysis.
CHAPTER 4 69
4.5 Understanding the .plt and .pch Files

The .plt file is a binary file that contains the plotting information generated by the
NX Nastran plotter. If the NX Nastran plotter is not used, run1.plt is deleted
following the completion of the run.
The .pch file (“punch file”) is a binary file that contains “punched” output. It is
written to FORTRAN unit 7. In the Case Control section, you can specify whether
you want the software to output specified results to the punch file.
4.6 Understanding the .DBALL and .MASTER Files

The .DBALL and .MASTER files are database files (DBsets) that NX Nastran
generates automatically during a solve. The files are generated in the scratch
directory. These files are necessary if you want to later perform a restart analysis.
• The .MASTER file is a directory of the database. It contains the names of

all DBsets, DBset members and their physical file names, and a directory
of projects, versions, data blocks, parameters, DMAP source and object
files. It also contains the Nastran Data Definition Language (NDDL)
scheme used to describe the database. The default maximum size for
.MASTER is 5,000 blocks
• The .DBALL file contains all the DMAP data blocks you can save for reuse
in a subsequent run. In the solution sequences that don’t allow restarts,
such as SOL 101, this file is empty. The default size is for .DBALL is 250,000
blocks
Specifying Whether to Save .DBALL and .MASTER Files

The “scratch” option on the nastran command line allows you to specify whether
you want the software to save the files .DBALL and .MASTER files. By default, NX
Nastran deletes the files at the end of the run.
• If you specify “scratch=yes,” the software deletes the files.

• If you specify “scratch=no,” the software keeps the files.
• If you specify “scratch=mini,” the software retains a reduced size database
that you can use to perform data recovery restarts.
See Also
• “scratch” on page 8 of the NX Nastran Quick Reference Guide

• “Mini-Database Restarts” on page 527 of the NX Nastran User’s Guide
• “Understanding DBsets” on page 533 of the NX Nastran User’s Guide
Manually Deleting .DBALL and .MASTER Files

In some cases, you may need to manually delete the .DBALL and .MASTER files. For
example:
• If your solve doesn’t complete successfully, NX Nastran may not

automatically delete the .DBALL and .MASTER files, even though you
specified “scratch=yes.” In those cases, you must delete the files manually
before resubmitting your job.
• If you specified “scratch=no,” but you decide that you want to submit a .dat
file a second time as a cold start (you don’t use any restart commands), you
must first delete the corresponding .DBALL and .MASTER files before you
resubmit the job.
4.7 Understanding .xdb and .op2 Files

The .xdb and .op2 files are binary files that contain NX Nastran results for use with
certain types of postprocessors.
• The .xdb file is known as the “results database” file. It contains results that
can be postprocessed by certain postprocessors.
• The .op2 file is known as an “OUTPUT2” file. It is written to FORTRAN unit
12. Also, if an .op2 file already exists for the run, it too should be deleted
prior to resubmitting the run. It contains results that can be postprocessed
by certain postprocessors, such as NX MasterFEM (formerly known as I-
DEAS MasterFEM), NX Scenario, and MSC.Patran.
The setting for the PARAM,POST option in the Bulk Data section of the input file
controls whether NX Nastran creates an .xdb or op2 file. For example, to have the
software create an .xdb file for use with NX MasterFEM, you would specify
PARAM,POST,-2. Additionally, the options you specify for PARAM,POST control
whether the file contains geometry.
See Also
• “POST” on page 627 of the NX Nastran Quick Reference Guide

Note: Before you perform a restart analysis, you should delete the .xdb and .op2 files
(if any) from your original run.
5
CHAPTER
Defining Grid Points and Scalar Points
■ Overview of Grid Points and Scalar Points
■ Defining Grid Points
■ Specifying Grid Point Properties
■ Understanding Scalar Points
5.1 Overview of Grid Points and Scalar Points

Whenever you analyze a finite element model using NX Nastran, you must generate
a set of grid points (nodes). Those grid points, together with the elements, define the
size and shape of your model. The physical behavior of your finite element model is
represented by the elements connecting the grid points as well as the associated
constraints and their coordinate system.
When you create a finite element model of a structure, you’re actually constructing a
mathematical model that represents your structure in matrix form. The unknowns
in the matrix equation are the displacements in the model. These displacements
consist of:
• the six components for each of the grid points

• one component for each of the scalar points
In general, the displacements at the grid points and the scalar points are referred to
as the model’s degrees of freedom.
This chapter describes grid points, grid point properties, and scalar points. Detailed
information about coordinate systems is provided in the next chapter.
For many practical modeling situations, you may have several different output
coordinate systems in a model.
See Also
• “Understanding Coordinate Systems” on page 83 of the NX Nastran

User’s Guide
5.2 Defining Grid Points

Each grid point has six degrees of freedom (DOF): three translations and three
rotations. The six degrees of freedom are identified as 1, 2, 3, 4, 5, and 6, as shown
below.
CHAPTER 5 73
1 4
3
6
Figure 5-1 DOF Nomenclature
Other commonly used terms for the components of displacement at a grid point are
DOF 1 = T1 = U = translation in direction 1

1

2

3
DOF 4 = R1 = θ = rotation about direction 1

1

2

3
When you define a location of the grid point in space using these components, you
are inherently defining the point relative to a coordinate system.
In NX Nastran, you use the GRID bulk data entry to define a grid point. GRID
allows you to:
• assign a unique ID to the grid point
• specify the location of the grid point in space with respect to a reference
coordinate system
• assign permanent constraints to the grid point
• define the directions of motions at the grid point
Format of the GRID Entry

The format of the grid point entry is as follows:
1 2 3 4 5 6 7 8 9 10
Field Contents
lD Grid point identification number.
CP Identification number of coordinate system in which the
location of the grid point is defined.
X1, X2, X3 Location of the grid point in coordinate system CP.
CD Identification number of coordinate system in which the
displacements, degrees of freedom, constraints, and
solution vectors are defined at the grid point.
PS Permanent single-point constraints associated with the
grid point.
SEID Superelement identification number.
See Also
• “GRID” on page 1335 of the NX Nastran Quick Reference Guide
Specifying the Grid Point ID (Field 1)

Field 1 lets you assign an ID to the grid point. That ID must be unique with respect
to all other grid points and scalar points in your model. In NX Nastran, grid point
IDs must begin at 1, however, they don’t have to be contiguous.
You will often find it helpful to assign groups of IDs to the different sections of your
model. For example, if you’re working on an automotive FE model, you may assign
IDs in the range from 1 to 1000 for the grid points for the driver’s door and assign
IDs in the range from 5001 to 6000 for the grid points for the hood. A clear and logical
scheme for numbering grid points can be very helpful on complex models.
Defining Grid Point Coordinate Systems (Fields 3 and 5)

In NX Nastran, each grid point refers to two coordinate systems, CP and CD:
CHAPTER 5 75
• The CP system that you specify in field 3 is used to locate the grid point in
space. This is known as the “position” coordinate system. It defines the
grid point’s location in space.
• The CD system that you specify in field 7 is used to establish the grid
point’s displacement (output) coordinate system. NX Nastran uses the CD
system to measure to defines the direction of displacements, constraints,
and other grid point related quantities such as reaction forces. It is also the
system in which the software outputs the results.
NX Nastran has two main categories of coordinate system:
• the basic coordinate system (the default for the CP and CD fields)
• local coordinate systems, which you define yourself. These can be
rectangular, cylindrical, or spherical
With the CP and CD fields, you can use either the basic coordinate system or a local
coordinate system.
Specifying the Location Coordinate System in the CP Field

To specify the coordinate system for the location of the grid point, enter the ID of
that coordinate system in the CP field (Field 3). If you leave the CP field blank, or
enter a 0 in the CP field, the software defines the grid point in the basic coordinate
system.
Specifying the Output Coordinate System in the CD Field

To specify the coordinate system in which you want NX Nastran to generate and
output the results at grid points (displacements, grid point forces, etc.), enter the ID
of the coordinate system in the CD field (Field 7). For many practical modeling
situations, you may have several different output coordinate systems in a single
model.
Although the CD field is referred to as the displacement coordinate system, it is

much more than just the coordinate system used to output the results. In fact, it has
a much more fundamental role in NX Nastran. The collection of all the output
coordinate systems--that is, all the coordinate systems defined on the CD field of all
the grid points-constitutes the global coordinate system. It is important to note that
the global coordinate system is not necessarily a single coordinate system but rather
the collection of all of your CD coordinate systems. NX Nastran assembles the
global stiffness matrix in the global coordinate system.
See Also
• “Comparing the Basic and Global Coordinate Systems” on page 84 of the

Understanding How the CP and CD Fields Work Together

Although you can have the CP and CD fields reference the same coordinate system,
you can also have them reference different coordinate systems. For example, you
may want to specify a local coordinate system to define the location of the grid point
in the CP field, and then leave the CD field blank or enter a 0 to output your results
in the basic coordinate system. On the other hand, if you need to output the
displacement information in particular coordinate system, you enter the ID of that
local coordinate system in the CD field.
Defining the Directions of Displacement (Fields 4-6)

X1, X2, and X3 in fields 4-6 let you define the location of the grid point in the CP
coordinate system you specified in field 3.
In NX Nastran, each grid point has six degrees of freedom or “displacement

components”: three translational and three rotational. The degrees of freedom are
denoted as u1, u2, u3, θ1, θ2, and θ3, or as T1, T2, T3, R1, R2, and R3.
• T1, T2, and T3 are the three orthogonal components of translation parallel
to the 1, 2, and 3 directions of the grid point’s displacement coordinate
system, respectively.
• R1, R2, and R3 are the components of rotation in the same directions.
X1, X2, and X3 have the following meanings for different types of coordinate
systems:
Type X1 X2 X3
Rectangular X Y Z
Cylindrical R θ (degrees) Z
Spherical R θ (degrees) φ (degrees)
The coordinate system you specify in the CD field determines the orientation of these
displacement components:
• If the system you specify in the CD field is rectangular, the displacement

components are parallel to the local system and are independent of the grid
point location as indicated in (a) below.
CHAPTER 5 77
• If the system you specify in the CD field is cylindrical, the displacement

components are in the radial, tangential, and axial directions as indicated
in (b) below.
• If the system you specify in the CD field is spherical, the displacement
components are in the radial, meridional, and azimuthal directions as
indicated in (c) below.
z u3
(a)Rectangular
u2
Grid Point
u1
Local System y
z
(b)Cylindrical u3 – z direction
r u2 – θ direction
u1 – r direction
Local System Grid Point
u1 – r direction
(c)Spherical
u3 – φ direction
θ
u2 – θ direction
Local System r Grid Point
Figure 5-2 Displacement Components
Specifying Permanent Single Point Constraints (Field 8)

You use field 8 (PS) of the GRID entry to specify permanent single point constraints
to constrain any or all the degrees of freedom associated with the grid point. In this
field, you enter the integers (1-6) that correspond to the degrees of freedom you want
to constrain.
The constraints that you define in field 8 are considered permanent because you can’t
change them during the run. Any degrees of freedom you specify in the PS field are
constrained for all subcases. Constraints specified on the GRID entry are usually
restricted to those degrees-of-freedom that will not be elastically constrained and
hence must be removed from the model in order to avoid singularities in the stiffness
matrix.
In many situations, you want to analyze your structure using more than one set of
constraints. For these situations, don’t specify the constraints on the GRID entry.
Instead, specify the constraints using an SPC bulk data entry (Single-Point
Constraint). The only difference from using the GRID entry is that constraints
applied using an SPC entry must be selected through the Case Control section.
Therefore, you can apply different SPC entries for different subcases.
NX Nastran applies the constraints you define in field 8 in the output coordinate
system (CD), which you defined in field 7. This gives you greater flexibility for
modeling a variety of joints and boundary conditions.
Note: NX Nastran applies all constraints, whether you define them with an SPC bulk
data entry or using the PS field of the GRID entry, in the output coordinate system.
See Also
• “SPC” on page 1749 of the NX Nastran Quick Reference Guide

• “Single-Point Constraints” on page 210 of the NX Nastran User’s Guide
CHAPTER 5 79
Defining the Superelement ID (Field 9)

The SEID option in field 9 lets you define superelements within the bulk data section
of your input file.
See Also
• “Superelement Definition” on page 50 of the NX Nastran Superelement

Users’s Guide
Understanding the GRDSET Entry

If many or all of the GRID entries in your model have the same values in the location
coordinate system (CP), displacement coordinate system (CD), single point
constraint (PS) fields, or superelement ID (SEID), you can use the GRDSET bulk data
entry to avoid having to repeatedly specify the same information. If you specify a
value for any of these fields on the GRDSET entry, the software then uses those
values to replace corresponding blank fields on any GRID entry in your model.
GRDSET is useful in the case of such problems as space trusses where you want to
remove all the rotational degrees-of-freedom. It is also helpful with plane structures
where you want to remove all out-of-plane or all in-plane motions.
Note: You can only include a single GRDSET entry in the bulk data section of your
input file.
See Also
• “GRDSET” on page 1333 of the NX Nastran Quick Reference Guide
5.3 Specifying Grid Point Properties
In NX Nastran, you specify some of the characteristics of the model as properties of
grid points, rather than as properties of elements.
Defining Mass, Damping, and Stiffness through Direct

Input Matrices
You use direct input matrices to define mass, damping, and stiffness properties
either in part or entirely, as properties of grid points.
• You can use the DMIG bulk data entry define direct input matrices. You
can associate these matrices with components of geometric grid points,
scalar points or extra points introduced for dynamic analysis.
• You can use the TF bulk data entry define transfer functions that NX
Nastran automatically converts to direct matrix input.
• The DMIAX entry is an alternate form of direct matrix input that is used for
hydroelastic problems (see ).
See Also
• “DMIG” on page 1128 of the NX Nastran Quick Reference Guide

• “TF” on page 1831 of the NX Nastran Quick Reference Guide
• “DMIAX” on page 1125 of the NX Nastran Quick Reference Guide
• “Overview of Coupled Fluid-Structure Interaction” on page 672 of the NX
Defining Mass through the Concentrated Mass Element

You can also use the concentrated mass element to input mass properties as
properties of grid points.
• You can use the CONM1 entry to define a 6 × 6 matrix of mass coefficients
at a geometric grid point in any selected coordinate system.
• You can use the CONM2 entry to define a concentrated mass at a geometric
grid point in terms of its mass, the three coordinates of its center of gravity,
the three moments of inertia about its center of gravity, and its three
products of inertia, referred to any selected coordinate system.
See Also
• “CONM1” on page 980 of the NX Nastran Quick Reference Guide

• “CONM2” on page 982 of the NX Nastran Quick Reference Guide
Defining Temperatures at Grid Points

You can define thermal fields by specifying the temperatures at grid points.
• You use TEMP bulk data entry to specify the temperature at grid points for
use in connection with thermal loading and temperature-dependent
material properties.
• You use the TEMPD bulk data entry to specify a default temperature to
avoid a large number of duplicate entries on a TEMP entry when the
temperature is uniform over a large portion of the structure.
• You use the TEMPAX bulk data entry for conical shell problems.
See Also
• “TEMP” on page 1815 of the NX Nastran Quick Reference Guide

CHAPTER 5 81
• “TEMPD” on page 1820 of the NX Nastran Quick Reference Guide

• “TEMPAX” on page 1817 of the NX Nastran Quick Reference Guide
Assigning a Singularity Normal to a Degree-of-Freedom

The SNORM Bulk Data entry and/or the PARAM,SNORM assigns the singularity
normal of a shell element to a degree-of-freedom. NX Nastran then uses the
AUTOSPC capability to remove this singularity.
See Also
• “SNORM” on page 1747 of the NX Nastran Quick Reference Guide

• “Automatically Applying Single-Point Constraints” on page 216 of the
5.4 Understanding Scalar Points

In NX Nastran, a scalar point has only one degree of freedom. A scalar point is
similar to a grid point in the way that NX Nastran assembles it into the matrix
equation. However, a scalar point has no spatial orientation, so it does not require
a coordinate system. It is simply an additional degree of freedom that you can define
in your model. Scalar points are useful for modeling with scalar elements,
multipoint constraints (MPCs), computing relative and average motions, and
defining the warping coefficients of a thin-walled beam.
Comparing Grid Points to Scalar Points

While you can use a grid point as a scalar point, a grid point requires more input
and requires you to constrain five of the six degrees of freedom. Consequently,
using scalar points instead of a grid point is more efficient, both computationally
and in terms of the actual input required.
Defining Scalar Points

In NX Nastran, you define use the SPOINT bulk data entry to define a scalar point.
You can specify up to eight scalar points on a single SPOINT entry. You can also
implicitly define scalar points on a connection entry for a scalar element SPOINT
entries are used primarily to define scalar points appearing in constraint equations
but to which no structural elements are connected.
See Also
• “SPOINT” on page 1775 of the NX Nastran Quick Reference Guide
Using Scalar Points as Extra Points for Dynamics

You can also introduce special scalar points, called “extra points,” for dynamic
analyses. Extra points are used in connection with transfer functions and other forms
of direct matrix input used in dynamic analyses and are defined on EPOINT entries.
See Also
• “DMIGs, Extra Points, and Transfer Functions” on page 22 of the NX

Nastran Advanced Dynamic Analysis User’s Guide
6
CHAPTER
Understanding Coordinate Systems
■ Overview of Coordinate Systems in NX Nastran
■ Understanding the Basic Coordinate System
■ Defining a Local Coordinate System
■ Understanding the Element and Material Coordinate System
6.1 Overview of Coordinate Systems in NX Nastran

In NX Nastran, there are two general types of coordinate systems:
• the “basic” coordinate system, which is an implicitly defined reference

coordinate system
• “local” coordinate systems, which you define explicitly
Both of these different types of coordinate system is described in more detail below.
6.2 Understanding the Basic Coordinate System

NX Nastran has a built-in Cartesian system called the basic coordinate system. It is
also called the default coordinate system. The basic system is a rectangular system,
except when you’re using axisymmetric elements.
Figure 6-1 The Basic Coordinate System
ID of the Basic Coordinate System

In NX Nastran, all coordinate systems have a coordinate system identification
number (CID). The basic coordinate system’s identification number is zero (0) or
blank.
Comparing the Basic and Global Coordinate Systems

Many finite element textbooks refer to a “global coordinate system.” However, this
term means something different in NX Nastran. It is important to understand the NX
Nastran definition of the global coordinate system as it differs from the classical
textbook approach. These differences in terminology can often be confusing to new
NX Nastran users.
In the classical approach, individual element stiffness matrices are generated for all
the elements. Those individual elemental matrices are then transformed into a single
stiffness matrix commonly referred to as the global stiffness matrix. Each row and
column of this global stiffness matrix represents a degree of freedom of the structural
model. The coordinate system associated with each one of these degrees of freedom
is the same and is known as the global coordinate system.
CHAPTER 6 85
In NX Nastran, the software generates the element stiffness matrix similarly to the
classical approach. NX Nastran assembles all the individual matrices into the global
stiffness matrix. However, instead of transforming the element stiffness matrices
into a unique coordinate system, NX Nastran transforms these matrices into the
coordinate system that you specified in the CD field of the GRID entry for the grid
points to which each element is attached. NX Nastran then assembles the individual
elemental stiffness matrices to form the global stiffness matrix.
Therefore, in NX Nastran, the “global” coordinate system isn’t a unique coordinate

system. It is simply the collection of all the CD coordinate systems specified on all
of the GRID entries. This approach facilitates a variety of elegant modeling
techniques. If you prefer to have the global coordinate system be the same as the
basic coordinate system, which is a unique system, you can leave the CD field blank
for all the grid points in your model.
Figure 6-2 shows the coordinate system definition where:
G2, (B1, B2,
G3, (C1, C2,

G13
G1, (A1, A2, j
G2
i G1
G3
x
zb
yb
Figure 6-2 Coordinate System Definition
local (user defined) coordinate system (x, y, z) or (r, θ , z) or

(i, j, k) =
(r, θ , φ )
( x b , y b , z b ) = basic coordinate system
k = ( G2 – G1 ) normalized
G13 = ( G3 – G1 )
j = ( k × G13 ) normalized
i = ( j × k)
CHAPTER 6 87
The GRID bulk data entry lets you define six rectangular displacement components
(three translations and three rotations) are defined at each grid point. You also
specify both the location coordinate system and the displacement coordinate system
are specified on the GRID entry for each geometric grid point.
See Also
• “Defining Grid Points” on page 72 of the NX Nastran User’s Guide

6.3 Defining a Local Coordinate System
Because you may need to create your model, or portions of your model, in a
coordinate system other than the basic coordinate system, NX Nastran also lets you
construct alternate coordinate systems. In NX Nastran, such coordinate systems are
known as “local” coordinate systems. Local coordinate systems may be:
• rectangular
• cylindrical
• spherical
You can relate local coordinate systems either directly or indirectly to the basic
coordinate system. Local coordinate systems play an important role in generating
your models, solving for the solution, and displaying the results.
For example, consider the farm silo in Figure 6-3. It is cylindrical with a
hemispherical dome. It is offset from the origin of the basic coordinate system.
Creating local cylindrical ( r, θ, z ) and spherical ( r, θ, φ ) systems makes it easier to
define model geometry or review displacement results.
φ
Local θ
Spherical
System
r
Z
Basic
Coordinate Y Local
System Cylindrical
X
System
r
Figure 6-3 Example of Local Coordinate Systems to Aid in Establishing Model
Geometry
Note: In any coordinate system, all angular input is in degrees, but output (such as
rotational displacement) is in radians.
Bulk Data Entries for Defining Coordinate Systems

In NX Nastran, there are six different bulk data entries that you can use to define local
coordinate systems:
CORD1R
Rectangular
CORD2R
CORD1C
Cylindrical
CORD2C
CORD1S
Spherical
CORD2S
With these different bulk data entries, you can define coordinate systems by
referencing:
• three existing grid points
• three existing grid points in an existing reference coordinate system
CHAPTER 6 89
Defining a Coordinate System by Referencing Existing Grid Points

The CORD1R, CORD1C, and CORD1S entries let you define a local coordinate
system by referencing three defined grid points. This is also known as the “Method
1” approach of coordinate system definition in NX Nastran.
The method for defining a coordinate system with CORD1R, CORD1C, or CORD1S
is very similar. With each of these bulk data entries, you must:
• Specify a unique coordinate system ID (CID) in field 2 of the entry.

• Specify the IDs for each of the grid points that you want to reference. The
three points must be noncolinear and can’t be coincident. NX Nastran’s
geometry processor checks the points to ensure that they’re not colinear.
Additionally, with the CORD1R and CORD1C entries, you can define up to two
coordinate systems on a single entry.
Understanding the Importance of Using Fully Constrained, Separate Grid

Points
Importantly, with Method 1s, if your model is later modified and any of the
reference grid point locations change, your local coordinate system orientation will
also change. To avoid this type of problem when using Method 1, you should
always define three fully constrained grid points that are not part of the structure.
Then, if your model is modified, your local coordinate system remains unchanged.
This example illustrates the problems you can encounter if you use Method 1 to
define a coordinate system with the CORD1R entry using three existing grid points
instead of defining three new ones.
Consider the truss structure shown in Figure 6-4. Suppose you want the positive T1
component of the displacements for all the grid points to be in the negative
Xb-direction as shown in (a).
In this case, we want to define CORD1R using existing grid points 2, 4, and 1 and
assigning it an ID of 11. You enter coordinate system 11 in the CD field of the GRID
entry for all the grid points. The model solves without problems and all results are
in the correct direction.
Problems will occur if you need to subsequently modify the model. For example,
suppose you need to move the location of grid point 4 as shown in (b). However, in
the process of moving grid point 4, you also inadvertently rotate the local coordinate
system 11.
While the deformed shape may look accurate if you post process the model, if you
examine the results in the output file, you will notice that they are not in the correct
direction.
Y Y
3 4 3 4
Z11 Z11
X11 2 X11
1 X 1 2 X
(a) (b)
Figure 6-4 Model Change
Defining a Coordinate System By Referencing an Existing

Coordinate System
The CORD2R, CORD2C, and CORD2S entries let you define a local coordinate
system by specifying the location of three points in a previously defined reference
coordinate system. This is also known as the “Method 2” approach of coordinate
system definition in NX Nastran. This reference coordinate system can either be the
basic coordinate system or another local coordinate system. Importantly, if this
reference coordinate system is modified, the associated local coordinate system will
also change.
The method for defining a coordinate system with CORD2R, CORD2C, or CORD2S
is very similar. With each of these bulk data entries, you must:
• Specify a unique coordinate system ID (CID) in field 2 of the entry.

• Specify the coordinate system ID (RID) in field 3 of the reference coordinate
system you’re using. If you enter a 0 or a blank in this field, NX Nastran uses
the basic coordinate system for the reference coordinate system.
• Enter the coordinates of three grid points in the reference coordinate system
you defined in field 3.
Defining a Rectangular Coordinate System

You use the CORD1R and CORD2R Bulk Data entries to define a rectangular
coordinate system.
Understanding CORD1R
When you define a rectangular coordinate with CORD1R, you must specify three
reference grid points: G1, G2, and G3.
• G1 defines the origin of the coordinate system.
• A vector taken from G1 to G2 defines the Z-axis
CHAPTER 6 91
• G3, together with the Z-axis, define the XZ plane.

• The X-axis is defined to be in this XZ plane and is perpendicular to the
Z-axis.
• The Y-axis is generated from the X- and Z-axes using the right-hand rule.
CORD1R Format
1 2 3 4 5 6 7 8 9 10
CORD1R CIDA G1A G2A G3A CIDB G1B G2B G3B
Field Contents
CIDA, CIDB Coordinate system identification number.
GiA, GiB Grid point identification numbers.
See Also
• “CORD1R” on page 993 of the NX Nastran Quick Reference Guide
Example: Adding Three Additional Grid Points to Define CORD1R

To better understand the use of a CORD1R entry, consider the truss structure
introduced earlier in this book. The input file has been modified so that the location
of grid points 1 through 4 are defined in a rectangular coordinate system with an ID
of 11 as shown in Figure 6-5.
2 Y
A = 4.0 in
6 2
E = 30.0 × 10 lb ⁄ in
1 3
4
J = 1.27 in 2
3
ν = 0.3 2 11
13 X 11 12 5 10 ft
Elements Y1
1 X
Grid points
4 4
F =1000 lb
Pin joints
50 ft
Reference Grid Points
Figure 6-5 Grid Point Locations Using a Local Coordinate System
The new rectangular coordinate system with an ID of 11 has an origin located at the
center of the truss and is rotated 180 degrees about the basic Z-axis.
The corresponding input file for the structure is given in Listing 6-1. Note that it is
necessary to add three additional grid points to the input file. Importantly, these
additional grid points don’t affect the overall stiffness of the structure.
• Grid point 11 is located at the center of the truss structure, but it isn’t
attached to the structure. It is fully constrained.
• Grid point 12 is located 10 inches above grid point 11 in the basic
Z-direction. The Z-axis for the new coordinate system is therefore parallel
to the basic Z-axis. The choice of 10 inches is arbitrary since any positive
value produces the same results.
• Grid point 13 defines the local XZ plane from which the X- and Y-axes of
coordinate system 11 are determined. The choice of 0 inches for the X
location of grid point 13 is also somewhat arbitrary since it is used only to
define the XZ plane. Any real number less than 300 inches produces the
same results. If the X-component of grid point 13 is greater than 300 inches,
then the X-axis will be in the basic X-direction instead of the direction
shown.
Listing 6-1 Truss Using CORD1R
$
$
BEGIN BULK
$
$ LOCAL COORDINATE SYSTEMS ARE DEFINED IN TERMS
$ GRID POINTS FOR METHOD 1
$
CORD1R 5 11 12 13
$
$ DESCRIBE THE GEOMETRY IN CP FIELD COORDINATE SYSTEMS
$
GRID 1 5 -300. -60. 0. 3456
GRID 2 5 -300. 60. 0. 3456
GRID 3 5 300. 60. 0. 3456
GRID 4 5 300. -60. 0. 3456
$
$ THE FOLLOWING GRID POINTS ARE ADDED ONLY FOR DEFINING
$ LOCAL COORDINATE SYSTEM 5 AND ARE FULLY CONSTRAINED
$
GRID 11 300. 60. 0. 123456
GRID 12 300. 60. 10. 123456
GRID 13 0. 60. 5. 123456
$
.
.
.
ENDDATA
CHAPTER 6 93
Understanding CORD2R
With CORD2R, you define the locations of points A, B, and C in the same manner
as with the CORD1R entry. You define these points in space using coordinates
relative to a reference coordinate system that you specify in Field 3. You must
define the reference coordinate system independently.
CORD2R Format
1 2 3 4 5 6 7 8 9 10
CORD2R CID RID A1 A2 A3 B1 B2 B3
C1 C2 C3
Field Contents
CID Coordinate system identification number.
RID Identification number of a coordinate system that is
defined independently from the CID coordinate
system.
Ai, Bi, Ci Coordinates of three points in coordinate system
defined in field 3.
z
uz
G2 or B
P
G3 or C uy
G1 or A ux Z
y
X
Y
x
Figure 6-6 CORDIR and CORD2R Definitions
See Also
• “CORD2R” on page 1000 of the NX Nastran Quick Reference Guide
Defining Cylindrical Coordinate Systems

The CORD1C and CORD2C bulk data entries allow you to define cylindrical
coordinate systems.
Understanding CORD1C
The CORD1C entry is used to define a cylindrical coordinate system using three grid
points to define its location and orientation in space as shown in Figure 6-7.
• Grid point G1 defines the origin of the cylindrical system.

• A vector from grid point G1 to grid point G2 defines the Z-axis.
• Grid point G3 together with the Z-axis defines the θ = 0 plane.
z
uz
G2
uθ
G3
G1 Z
ur
θ
x R y
Figure 6-7 CORD1C Definition
CORD1C Format
1 2 3 4 5 6 7 8 9 10
CORD1C CIDA G1A G2A G3A CIDB G1B G2B G3B
Field Contents
CIDA, CIDB Coordinate system identification number.
See Also
• “CORD1C” on page 991 of the NX Nastran Quick Reference Guide
CHAPTER 6 95
Understanding CORD2C
The CORD2C entry uses the location of three points A, B, and C in the same manner
as the three grid points are used for the CORD1C entry. These points are defined in
space using coordinates relative to the reference coordinate system that you specify
in Field 3.
• Point A defines the origin.

• Point B defines the direction of the z-axis.
• Point C lies in the plane of the azimuthal origin.
You must independently define the reference coordinate system (RID).
The location of a grid point (P in Figure 6-8) in this coordinate system is given by (r,
θ , z) where θ is measured in degrees. If this coordinate system is used to specify the
displacement directions they are given by u r , uθ and u z (the displacements are u r ,
u θ and u z).
CORD2C Format
1 2 3 4 5 6 7 8 9 10
CORD2C CID RID A1 A2 A3 B1 B2 B3
C1 C2 C3
Field Contents
CID Coordinate system identification number. (Integer > 0)
RID Identification number of a coordinate system that is defined
independently from this coordinate system. ( Integer ≥ 0 ;
Default (zero) is the basic coordinate system)
Ai, Bi, Ci Coordinates of three points in coordinate system defined in field
3. (Real)
uz
RID B
coordinate uθ
system
C P ( r ,θ ,z )
A
z
ur
x r
y
Figure 6-8 CORD2C Definition
See Also
• “CORD2C” on page 997 of the NX Nastran Quick Reference Guide
Example: Using the CORD2C Entry

The following example illustrates the use of the CORD2C entry. An arch with a
semicircular top is shown in Figure 6-9. To facilitate grid point input, we want to
establish a local cylindrical coordinate system for grid points 3 through 7.
CHAPTER 6 97
5
4 6
3 7
r = 15 in
2
8
50 in
Y
origin of the
basic system
1 9
X
Basic
Coordinate
System 30 in
Figure 6-9 Arch Structure
A cylindrical coordinate system with an ID of 100 is defined as shown in

Figure 6-10.
Basic
Coordinate Z z
System
B
Y
X
A y
C r
θ
x
Figure 6-10 Definition of Local Cylindrical Coordinate System
The local cylindrical coordinate system is defined by referencing the basic

coordinate system. Three points are required to define the local system:
• Point A is at the origin

• Point B lies on the z-axis of the new system.
• Point C defines the reference axis at which θ = 0° .
This coordinate system has an identification number in field CID which will be
referenced by other entities, such as the CP and CD fields of a GRID entry. In this
example, the coordinates of points A, B, and C are in the basic coordinate system.
1 2 3 4 5 6 7 8 9 10
CORD2C CID RID A1 A2 A3 B1 B2 B3
C1 C2 C3
CORD2C 100 15. 50. 0. 15. 50. 1.
16. 50. 0.
The associated GRID entries are defined as follows:
1 2 3 4 5 6 7 8 9 10
GRID 1 0. 0. 0.
GRID 2 0. 50. 0.
GRID 3 100 15. 150. 0.
GRID 4 100 15. 120. 0.
GRID 5 100 15. 90. 0.
GRID 6 100 15. 60. 0.
GRID 7 100 15. 30. 0.
GRID 8 30. 50. 0.
GRID 9 30. 0. 0.
Grid points 3 through 7 use (r, θ , z) coordinates with r = 15.0 inches and θ varying
from 30 degrees (GRID 7) to 150 degrees (GRID 3). The output for all grid points is
in the basic rectangular coordinate system since field 7 is left blank.
Understanding CORD1S
The CORD1S entry is similar to the CORD1C entry since three grid points (G1, G2,
and G3) are used to define the new coordinate system. As shown in Figure 6-11:
• Grid point G1 defines the origin of the spherical system.

• The vector starting at grid point G1 and passing through grid point G2 is the
θ = 0 axis.
• This axis, together with grid point G3, determines the φ = 0 plane.
• The φ = 0 axis is in this plane and is oriented using the right-hand rule.
CHAPTER 6 99
CORD1S Format
1 2 3 4 5 6 7 8 9 10
CORD1S CIDA G1A G2A G3A CIDB G1B G2B G3B
Field Contents
CIDA, CIDB Coordinate system identification numbers.
See Also
• “CORD1S” on page 995 of the NX Nastran Quick Reference Guide
Understanding CORD2S
The CORD2S entry uses the location of three points A, B, and C in the same manner
as the three grid points are used for the CORD1S entry. These points are defined in
space using coordinates relative to the reference coordinate system (Field 3). The
reference coordinate system must be independently defined.
CORD2S Format
1 2 3 4 5 6 7 8 9 10
CORD2S CID RID A1 A2 A3 B1 B2 B3
C1 C2 C3
Field Contents
RID Identification number of a coordinate system that is defined
independently from this coordinate system.
Ai, Bi, Ci Coordinates of three points in coordinate system defined in field 3.
G2,B
uφ
θ
ur
G3,C P
G1,A R
φ uθ
x
y
Figure 6-11 CORD1S and CORD2S Definitions
Note: θ for the spherical coordinate system is different than the θ in the
cylindrical coordinate system.
Example: Using Cylindrical and Spherical Coordinate

Systems
To see the use of cylindrical and spherical coordinate systems, consider the grain silo
shown below.
• The cylindrical portion of the silo is modeled using a cylindrical coordinate

system.
• The roof of the silo is modeled using a spherical coordinate system.
25
B2 C2
Spherical Cap
θ
2 φ
A2
Basic Coordinate
System Cylindrical Body
Y B1 C1 50
Z
1 R
X
A1
100
Figure 6-12 Silo Model
The cylindrical coordinate system has an ID of 1 and is defined with the CORD2C
entry. It is defined with respect to the basic coordinate system.
CHAPTER 6 101
• The origin is given by point A1, and located by the vector (100,0,0) as
shown in Figure 6-12.
• The Z-axis is defined by the line extending from point A1 to point B1, which
is located at (100,1,0).
• The Z-axis along with point C1, which is located at (101,1,0), defines the θ
= 0 plane.
• The R-axis is defined in this plane and is perpendicular to the Z-axis.
• The direction of the θ-axis is determined using the right-hand rule.
The spherical coordinate system has an ID of 2 is defined with the CORD2S entry.
It is defined with respect to the reference coordinate system 1.
• The location of origin of the spherical coordinate system 2 is given by point
A2 at (0,0,50). This location is with respect to coordinate system 1.
• The θ = 0 axis is the line extending from A2 to B2, which is located at
(0,0,51).
• The φ = 0 plane is located by the φ = 0 axis and point C2, which is located
at (1,0,51).
Since coordinate system 1 is defined in terms of the basic coordinate system,

coordinate system 2 is also indirectly defined in terms of the basic coordinate
system. You could also directly define both coordinate systems in terms of the basic
coordinate system. However, since coordinate system 2 is linked to coordinate
system 1, adjusting the position of the silo (for example, due to a change in the
module’s geometry) is only a matter of changing the locations of the grid points on
the CORD2C entry.
Listing 6-2 Grain Silo Coordinate Systems
BEGIN BULK
$
$ THE CYLINDRICAL SYSTEM USED FOR THE SILO BODY
$
CORD2C 1 0 100. 0. 0. 100. 1. 0. +COR1
+COR1 101. 1. 0.
$
$ THE SPHERICAL SYSTEM USED FOR THE SILO CAP
$
CORD2S 2 1 0. 0. 50. 0. 0. 51. +COR2
+COR2 1. 0. 51.
.
.
.
ENDDATA
6.4 Understanding the Element and Material

Coordinate System
Every element in the NX Nastran library has its own unique element coordinate
system that is used to output the element forces, moments, and stresses as well as to
orient section properties. For some elements, such as the CROD element, the element
coordinate systems are implicitly defined based on the element connectivity. For
other elements, you must explicitly define the orientation of the element coordinate
system. For further discussion of element coordinate systems, refer to the NX
Nastran Element Library.
In addition to the element coordinate system, an element may also have an optional
material coordinate system that can be used to define the orientation of orthotropic
or anisotropic material.
Special Considerations for Grid Points on the Polar Axis

If the displacements of a grid point are defined to be in a spherical or cylindrical
coordinate system and the grid point is located on, or relatively close to, the polar
axis, then it will be replaced by the “defining” rectangular coordinate system. The
z-axis of the “defining” rectangular coordinate system is coincident with the polar
axis and the x-axis is coincident with θ = 0 in cylindrical system, or θ = 0 and φ = 0
in the spherical system. In this case, you should explicitly specify a rectangular
coordinate system.
Defining Element Offsets

Element offsets are defined by a Cartesian coordinate system, which is also parallel
to the displacement coordinate system at the connecting grid point. However, the
components of the offsets are always defined in units of translation, even if the
displacement coordinate system is cylindrical or spherical.
6.5 Other Special Coordinate System Entries

In addition to the main coordinate system bulk data entries discussed above, there
are some additional entries that you can use when defining coordinate systems in NX
Nastran.
CORD3G
CORD3G lets you define a general coordinate system using three rotational angles as
functions of coordinate values in the reference coordinate system. You use the
CORD3G entry with the MAT9 entry to orient the material principal axes for 3D
composite analyses.
See Also
• “CORD3G” on page 1005 of the NX Nastran Quick Reference Guide

CHAPTER 6 103
GMCORD
GMCORD lets you define a convective/follower coordinate system on a FEEDGE,
GMCURV, FEFACE, or GMSURF bulk data entry for use in adaptive (p-element)
analysis.
See Also
• “GMCORD” on page 1303 of the NX Nastran Quick Reference Guide
7
CHAPTER
Material Properties
■ Introduction
■ Isotropic Materials
■ Orthotropic Materials
■ Anisotropic Materials
■ Material Properties for Heat Transfer
■ Material Properties for Fluid Analysis
■ Material Properties for Nonlinear Analysis
7.1 Introduction
Material definitions describe the stiffness or constitutive relationships, strength,
density, and thermal expansion characteristics of a material. NX Nastran supports
isotropic, orthotropic and anisotropic materials.
This chapter describes materials for general structural analysis as well as thermal and
fluid material properties.
Elements and Material Input Data Types

Table 7-1 summarizes the relationships between elements and material input data
types.
Table 7-1 Correspondence Between Element and Material Type
Hyper- Stress Temperature

Element Material Type Composite
elastic Dependent Dependent
Type MAT1 MAT2 MAT3 MAT8 MAT9 MATHP MATS1 MATTi
CBAR X X
CBEAM X X X X
CBEND X X
CCONEAX X X X
CRAC2D X X
CRAC3D X X
CHEXA X X X X X
CPENTA X X X X X
CQUAD X
CQUAD4 X X X X X X X
CQUAD8 X X X X X X
CQUADR X X X X
CQUADX X
CROD X X X
CSHEAR X X
CTETRA X X X X X
CTRIA3 X X X X X X X
CTRIA6 X X X X X X
CTRIAR X X X X
CTRIAX X
CHAPTER 7 107
Material Properties
Table 7-1 Correspondence Between Element and Material Type (continued)
Hyper- Stress Temperature

Element Material Type Composite
elastic Dependent Dependent
Type MAT1 MAT2 MAT3 MAT8 MAT9 MATHP MATS1 MATTi
CTRIAX6 X X
CTUBE X X X
7.2 Isotropic Materials

In NX Nastran, you model isotropic materials with the MAT1 bulk data entry. An
isotropic material is defined as a material that has the same material properties in
any direction. In NX Nastran, you use the MAT1 entry to describe the isotropic
material properties for general, plain stress, plain strain, and axisymmetric isotropic
material behavior.
This section describes the MAT1 Bulk Data entries used to define isotropic materials
and its corresponding constitutive relations.
MAT1 Constituitive Relationships

The general three-dimensional isotropic material stress-strain constitutive
relationship is:
E(1 – ν) νE νE
-------------------------------------- -------------------------------------- -------------------------------------- 0 0 0
( 1 + ν ) ( 1 – 2ν ) ( 1 + ν ) ( 1 – 2ν ) ( 1 + ν ) ( 1 – 2ν )
 σx  E(1 – ν) νE
-------------------------------------- -------------------------------------- 0 0 0
σ  ( 1 + ν ) ( 1 – 2ν ) ( 1 + ν ) ( 1 – 2ν )
 y
  E(1 – ν)
 σz  -------------------------------------- 0 0 0
= ( 1 + ν ) ( 1 – 2ν ) Eq. 7-1
 
 τ xy  E
-------------------- 0 0
  2(1 + ν)
 τ yz 
  symmetric E
0
 τ zx  --------------------
2(1 + ν)
E
--------------------
2( 1 + ν )
  εx  A 
 ε  A 
 y   
   
  εz  A 
   – ( T – T ref )  
  γ xy  0 
   
  γ yz  0 
   
  γ zx  0 
where:
{σ} stress
{ε} strain
{A} thermal expansion coefficients
E Young’s modulus
ν Poisson’s ratio
τ shear stress
γ shear strain
( T – T ref ) change in element temperature used
to calculate initial element thermal
expansion
Constitutive Relationship for Plane Stress

Isotropic material constitutive relationship for plane stress behavior is:
E νE
---------------- ---------------- 0
2 2
 σx  1–ν 1–ν   εx  A 
  νE νE    
 σy  = ---------------- ---------------- 0   ε y  – ( T – T ref )  A   Eq. 7-2
  2 2    
1–ν 1–ν
 τ xy    γ xy   0 
E
0 0 ----------------------
2(1 + ν)
Constitutive Relationship for Plane Strain

Isotropic material constitutive relationship for plane strain behavior is:
E(1 – ν) νE
-------------------------------------- -------------------------------------- 0
σ
 x ( 1 + ν ) ( 1 – 2ν ) ( 1 + ν ) ( 1 – 2ν )   εx  A 
  νE E(1 – ν)    
 σ y  = -------------------------------------- -------------------------------------- 0   ε y  – ( T – T ref ) ( 1 + ν )  A   Eq. 7-3
  ( 1 + ν ) ( 1 – 2ν ) ( 1 + ν ) ( 1 – 2ν )    
 τ xy  E   γ xy   0 
0 0 --------------------
2(1 + ν)
Constitutive Relationship for Transverse Shear

When MAT1 is used for transverse shear:
CHAPTER 7 109
Material Properties
 σ zx γ
G 0  zx
  =   Eq. 7-4
 σ zy 0 G  γ zy
Understanding the Isotropic Material Constants

An isotropic material is fully described by only two material constants. These two
constants may be any combination of E, G, and ν . When you enter only two
constants, the software computes the third constant using the following equation:
E
G = ---------------------- Eq. 7-5
2(1 + ν)
where:
E Young’s modulus
G shear modulus
Note:
• For line elements, E is the modulus of elasticity, and G is the shear
modulus for torsion and transverse shear if it is present in the element.
• For plate and solid elements, E, G, and ν are used to develop a material
matrix for the element.
If you enter all three constants and they don’t satisfy this relationship, the software
prints a warning message indicating that the isotropic relationship has been
violated. Because of this, we recommend that you input only two of the three
constants on the MAT1 bulk data entry.
Importantly, NX Nastran doesn’t apply some of the isotropic material constants in

the stiffness formulation of certain elements. Table 7-2 shows which elastic
constants are used for the different element types.
Table 7-2 MAT1 – Material Constant Usage for the Different Element Types
Element Entry E NU G
CROD
Extension and Torsion
CBEAM Not Used
Bending Transverse Shear
CBAR
CQUADi
CTRIAi Membrane and Bending Transverse Shear
CCONEAX
Table 7-2 MAT1 – Material Constant Usage for the Different Element Types
Element Entry E NU G
CSHEAR Not Used Shear
CRAC2D All Terms Not Used
CHEXA
CPENTA
All Terms Not Used
CTETRA
CRAC3D
Radial, Axial,
CTRIAX6 All Coupled Ratios Shear
Circumferential
MAT 1 Format
The MAT1 entry has the following format:
1 2 3 4 5 6 7 8 9 10
MAT1 MID E G NU RHO A TREF GE
ST SC SS MCSID
Field Contents
MID Material identification number. (Integer > 0)
E Young’s modulus. ( Real ≥ 0.0 or blank)
G Shear modulus. ( Real ≥ 0.0 or blank)
NU Poisson’s ratio. (-1.0 < Real ≤ 0.5 or blank)
RHO Mass density. (Real)
A Thermal expansion coefficient. (Real)
TREF Reference temperature for the calculation of thermal loads, or a
temperature-dependent thermal expansion coefficient. (Real;
Default = 0.0 if A is specified)
GE Structural element damping coefficient. (Real)
ST, SC, SS Stress limits for tension, compression, and shear used only to
compute margins of safety in certain elements; they have no
effect on the computational procedures. (Real)
MCSID Material coordinate system identification number. Used only
for PARAM,CURV processing. See the “Parameters” on
page 563 of the NX Nastran Quick Reference Guide. ( Integer ≥ 0
www.cadfamily.com or blank)
CHAPTER 7 111
Material Properties
• The material identification number (MID) connects the MAT1 entry to an

element property entry (e.g., PBAR, PSHELL, or PSOLID).
• You only have to enter values for two of the three properties E, G, and NU.
NX Nastran automatically calculates the remaining value, if it’s required,
as described above. Note that you can’t leave both the E and G fields blank.
• Mass density RHO is used for GRAV loads as well as problems in dynamic
analysis.
• The thermal expansion coefficient A and reference temperature TREF are
used only in thermal analysis problems.
• The structural damping coefficient (GE) is only used in dynamic analysis.
• The ST, SC, and SS entries let you perform optional margin of safety
calculations by supplying yield values for the materials. These are extra
arithmetic calculations and don’t affect your basic NX Nastran results.
• MCSID lets you specify a material coordinate system ID number. Material
coordinate systems are used with two- and three-dimensional elements to
orient orthotropic and anisotropic material axes with respect to the
element coordinate system.
Defining Temperature- and Stress-Dependent Properties for MAT1

With the MAT1 entry, you can use related bulk data entries to define temperature-
or stress-dependent properties.
• You can use the MATT1 entry to designate certain material properties on
the MAT1 entry as temperature-dependent by referencing TABLEMi
entries.
• In SOLs 106 and 129, you can use the MATS1 entry to designate certain
material properties on the MAT1 entry as stress-dependent.
See Also
• “MAT1” on page 1360 of the NX Nastran Quick Reference Guide

• “MATT1” on page 1400 of the NX Nastran Quick Reference Guide
• “MATS1” on page 1394 of the NX Nastran Quick Reference Guide
Example: Using MAT1 to Define a Structural Steel

For example, suppose you needed to create a MAT1 entry for a linear static analysis
problem using a mild structural steel. The material properties you need to define
are:
6 2
• E = 30 × 10 lbf ⁄ in
• ν = 0.3
–4 2 4
• mass density ρ m = 7.0 × 10 lb f –sec /in
Given these parameters, you would create a MAT1 entry (small field format) as
follows:
1 2 3 4 5 6 7 8 9 10
MAT1 5 30.E6 0.3 7.0E-4
The same MAT1 entry in free field format would be:
MAT1,5,30.E6,,0.3,7.0E-4
7.3 Orthotropic Materials

In NX Nastran, you use the MAT3 and MAT8 bulk data entries to define the
orthotropic material properties for two-dimensional and three-dimensional element
orthotropic material behavior.
• MAT3 lets you for define orthotropic material properties for three-
dimensional elements.
• MAT8 lets you define orthotropic material properties for two-dimensional
elements.
This section describes the MAT3 and MAT8 Bulk Data entries and their
corresponding constitutive relations. See Table 7-1 for a summary of elements that
use MAT3 and MAT8 entries.
Understanding the MAT3 Entry

MAT3 lets you define an orthotropic three-dimensional relationship in a
cross-sectional coordinate system (x, θ , z) for an axisymmetric CTRIAX6 element.
The stress-strain relationship as given by:
1 – ν θx – ν zx
------ ------------ ----------- 0
Ex Eθ Ez
 εx  – ν xθ – ν zθ  σx   Ax 
  1    
------------ ------ ----------- 0
 εθ  Ex Eθ Ez  σθ   Aθ 
  =   + ( T – T ref ) ⋅   Eq. 7-6
 εz  – ν xz – ν θz 1  σz   Az 
  ----------- ----------- ----- 0    
 γ zx  Ex Eθ Ez  τ zx   0 
1
0 0 0 ---------
G zx
CHAPTER 7 113
Material Properties
where:
{σ} stress
{ε} strain
E Young’s modulus
G shear modulus
τ shear stress
γ shear strain
expansion
To preserve symmetry, the following relationships must hold:
ν xθ ν θx ν xz ν zx ν θz ν zθ
-------- = -------- ; -------- = -------- ; -------- = -------- Eq. 7-7
Ex Eθ Ex Ez Eθ Ez
Only the νθx , νxz , and νzθ terms are used in the input for the MAT3 bulk data entry.
MAT3 Format
The format of the MAT3 bulk data entry is as follows:
1 2 3 4 5 6 7 8 9 10
MAT3 MID EX ETH EZ NUXTH NUTHZ NUZX RHO
GZX AX ATH AZ TREF GE
Field Contents
MID Material identification number.
EX, ETH, EZ Young’s moduli in the x, θ , and z directions,
respectively.
xx Poisson’s ratios.
RHO Mass density.
GZX Shear modulus.
AX, ATH, AZ Thermal expansion coefficients.

TREF Reference temperature.
See Also
Understanding the MAT8 Entry

You use the MAT8 entry to define orthotropic material matrix for plate and shell
elements. In NX Nastran, you can use MAT8 in conjunction with the PCOMP
property bulk data entry to define lamina properties for thin shell elements.
See Also
• “PCOMP” on page 1549 of the NX Nastran Quick Reference Guide
The stress-strain relationship as given by:
1 – ν yx
------ ----------- 0 0 0
Ex Ey
ε
 x  – ν xy 1  σx   Ax 
ε  ----------- ------ 0 0 0    
 y  Ex Ey  σy   Ay 
  1    
 γ xy  = 0 0 --------- 0 0  τ xy  + ( T – T ref )  0  Eq. 7-8
  G xy    
 γ xz   τ xz  0 
  1    
 γ yz  0 0 0 --------- 0  τyz  0 
G xz
1
0 0 0 0 ---------
G yz
where:
{σ} stress
{ε} strain
E Young’s modulus
τ shear stress
γ shear strain
CHAPTER 7 115
Material Properties
G shear modulus
expansion
Where ν yx ⁄ E y = ν xy ⁄ E x must be satisfied to preserve symmetry. Only the ν yx term is

used in the input.
When either G xz or G yz is input as zero, it implies that the corresponding shear

modulus is infinite. When this is the case, the transverse shear flexibility effects are
neglected in the shell formulation.
The in-plane stress-strain relationship is defined by:
1 – ν 12
------ ----------- 0
  E1 E1  
 ε1   σ1   A1 
  – ν 12 1    
 ε2  = ----------- ------ 0  σ 2  + ( T – T R EF )  A 2  Eq. 7-9
  E1 E2    
 γ 12   τ 12   0 
  1  
0 0 ---------
G 12
The transverse shear stress-transverse shear strain relationship is defined by:
   
 τ 1z  G 1z 0  γ 1z 
  =   Eq. 7-10
 τ 2z  0 G 2z  γ 2z 
   
MAT8 Format
The format of the MAT8 Bulk Data entry is as follows:
1 2 3 4 5 6 7 8 9 10
MAT8 MID E1 E2 NU12 G12 G1Z G2Z RHO
A1 A2 TREF Xt Xc Yt Yc S
GE F12 STRN
Field Contents
E1 Modulus of elasticity in the longitudinal direction.
E2 Modulus of elasticity in the lateral direction.

NU12 Poisson’s ratio ( ( ε 2 ⁄ ε 1 ) for uniaxial loading in 1-direction).
G12 In-plane shear modulus.

G1Z Transverse shear modulus for shear in the 1-Z plane.
G2Z Transverse shear modulus for shear in the 2-Z plane.
RHO Mass density.
Ai Thermal expansion coefficient in the i-direction.
Xt, Xc Allowable stresses or strains in tension and compression,
respectively, in the longitudinal direction.
Yt, Yc Allowable stresses or strains in tension and compression,
respectively, in the lateral direction.
S Allowable stress or strain for in-plane shear.
GE Structural damping coefficient.
F12 Interaction term used in the Tsai-Wu theory.
STRN Request for maximum strain theory.
Note: With the MAT8 entry, you can’t define any temperature-dependent material
properties.
See Also
MAT8 Example
The composite tube structure shown in Figure 7-1 illustrates the modeling of an
orthotropic material. The longitudinal direction is the fiber direction and is oriented
at an angle of 45 degrees relative to the tube axes as shown. The tube is held fixed at
one end, and a uniform axial load is applied at the other end.
CHAPTER 7 117
Material Properties
θ = 90°
R, θ = 0
G1 G4
600 N
Fixed End
Z
G2 G3
R, θ = 0
Principal fiber directions
Xmat are laid at 45 degrees to
Ymat
the tube axis.
G1 G4
Tube Dimensions:
Mean Diameter = 30 mm
T = 3 mm
Length = 90 mm
Orthotropic Material Properties:

G2 G3
Typical Element Ey
ν yx = υ xy ------
Ex
θ mat = 135 degrees
Figure 7-1 Composite Tube with Orthotropic Properties
A sample of the tube material has been tested, and the material properties are
determined in the longitudinal (Xmat) and transverse (Ymat) directions. The material
properties are shown in Table 7-3.
Table 7-3 Composite Tube Material Properties
Longitudinal Transverse
Shear
Direction Direction
4 N 4 N
E 25 ⋅ 10 ------------ 10 ⋅ 10 ------------ –
2 2
mm mm
ν 0.4 0.16 –
N
G12 – – 80,000 ------------
2
mm
Listing 7-1 shows the input file. The geometry has been defined with a local
cylindrical coordinate system for convenience. This doesn’t affect the orientation of
the material coordinate system.
Listing 7-1 Orthotropic Tube Input File
$
$ GEOMETRY DEFINED AND OUTPUT IN THE CYL COORD SYS 1
$
CORD2C 1 0 0.0 0.0 0.0 0.0 0.0 1
1. 0.0 0.0
$
GRID 110 1 15. 150. 30. 1
GRID 111 1 15. 150. 60. 1
GRID 114 1 15. -150. 30. 1
GRID 115 1 15. -150. 60. 1
$
CQUAD4 8 1 110 114 115 111 135.
$
PSHELL 1 1 3. 1
$
Note: Only element 8 and its connected grid points are shown.
The finite element model for the tube is shown in Figure 7-2. Element 8 is highlighted
to illustrate how the material coordinate system is used. The material coordinate
system is oriented at an angle of 135 degrees (field 8 of the CQUAD4 entry). The 135
degrees is with respect to a line drawn from G1 (which has a grid ID of 110) to G2
(which has a grid ID of 114). The resulting material coordinate system is therefore the
same as that shown in Figure 7-1. If you look at the deformed model, it is interesting
to note that the tube rotates about the z-axis even though a uniform axial load is
applied. This rotation is due to the orthotropic material properties and the fact that
the fiber direction oriented in a different direction than the applied load.
CHAPTER 7 119
Material Properties
110 111
114 115
X Z
1 Undeformed
110
111
114
115
Y
X Z
Deformed
2
Figure 7-2 Finite Element Model of the Orthotropic Tube
7.4 Anisotropic Materials

This section describes the MAT2 and MAT9 entries that define anisotropic materials
and their corresponding constitutive relations.
• MAT2 lets you define anisotropic material properties for two-dimensional
elements.
• MAT9 lets you for define anisotropic material properties for three-
dimensional elements.
See Table 7-1 for a summary of elements that use MAT2 and MAT9 entries.
Understanding MAT2
The MAT2 entry is used to define two-dimensional anisotropic constitutive stress-
strain relationship using the anisotropic material matrix, [G]. You can only use the
MAT2 entry with plate and shell elements. With MAT2, you define in-plane material
properties with respect to an element material coordinate system. You can also
include transverse shear material properties.
The stress-strain relationship is given by:
 σ1  G 11 G 12 G 13   ε 1   A1  
     
 σ2  = G 12 G 22 G 23   ε 2  – ( T – T ref )  A 2   Eq. 7-11
     
 τ 12  G 13 G 23 G 33   γ 12   A 12  
where:
{σ} stress
{ε} strain
E Young’s modulus
G shear modulus
expansion
The MAT2 entry is also used to define two-dimensional orthotropic constitutive

stress-strain relation using the anisotropic material matrix, [ G ] . The stress-strain
relationship is given by:
 σ1  G 11 G 12 0   ε1   A1  
     
 σ2  = G 12 G 22 0   ε 2  – ( T – T ref )  A 2   Eq. 7-12
     
 τ 12  0 0 G 33   γ 12   0 
where the elastic coefficients are given by engineering constants used to describe
orthotropic material properties as:
E1
G 11 = --------------------------------- Eq. 7-13
( 1 – ν 12 ν 21 )
CHAPTER 7 121
Material Properties
E2
G 22 = --------------------------------- Eq. 7-14
( 1 – ν 12 ν 21 )
ν 12 E ν 21 E 1
2
G 12 = --------------------------------- = --------------------------------- Eq. 7-15
( 1 – ν 12 ν 21 ) ( 1 – ν 12 ν 21 )
When MAT2 is used for transverse shear, the following equation defines the
transverse shear stress-strain relationship:
 σ zx G 11 G 12  γ zx
  =   Eq. 7-16
 σ zy G 12 G 22  γ zy
G 33 must be set to 0.
MAT2 Format
1 2 3 4 5 6 7 8 9 10
MAT2 MID G11 G12 G13 G22 G23 G33 RHO
A1 A2 A3 TREF GE ST SC SS
MCSID
Field Contents
Gij The material property matrix.
RHO Mass density.
Ai Thermal expansion coefficient vector.
ST, SC, SS Stress limits for tension, compression, and shear.
MCSID Material coordinate system identification
number.
When you perform composite analysis using the PCOMP entries, NX Nastran
automatically generates the MAT2 entries.

• You can use the MATT2 entry to designate certain material properties on the
MAT2 entry as temperature-dependent by referencing TABLEMi entries.
See Also

Understanding MAT9
The MAT9 entry is used to define general three-dimensional anisotropic constitutive
stress-strain relationship using the anisotropic material matrix, [ G ] . You can use
MAT9 to define an anisotropic material property for the CHEXA, CPENTA, and
CTETRA solid elements. You can also use MAT9 to define a three-dimensional
orthotropic material.
The stress-strain constitutive relationship is:
σ    A 
 x G 11 G 12 G 13 G 14 G 15 G 16   ε x   1 
σ   ε  A 
 y G 22 G 23 G 24 G 25 G 26   y   2 
     
 σz  G 33 G 34 G 35 G 36   ε z   A3  
  =   – ( T – T ref )    Eq. 7-17
 τ xy  G 44 G 45 G 46   γ xy   A4  
     
 τ yz  symmetric G 55 G 56   γ yz   A5  
     
 τ zx  G 66   γ zx   A6  
     
where:
{σ} stress
{ε} strain
E Young’s modulus
CHAPTER 7 123
Material Properties
G shear modulus
expansion
The MAT9 entry is also used to define three-dimensional orthotropic constitutive

stress-strain relation using the anisotropic material matrix [ G ] . The stress-strain
relationship is given by:
σ   ε  
 x G 11 G 12 G 13  x  0 0 0
 A1  
σ     
 y G 22 G 23 0 0 0   εy   A2  
     
 σz  G 33 0 0 0   εz  A 
  =   – ( T – T ref )  3   Eq. 7-18
τ
 xy  G 44 0 0   γ xy   0 
     
 τ yz  symmetric G 55 0   γ yz   0 
     
 τ zx  G 66   γ zx   0 
    
If you want to use the MAT9 entry to define an orthotropic material, the terms of the
material matrix can be generated by using Eq. 7-19:
1 – υ yz υ zy υ yx + υ zx υ yz υ zx + υ yx υ zy
G 11 = --------------------------- G 12 = --------------------------------- G 13 = ---------------------------------
Ey Ez ∆ Ey Ez ∆ Ey Ez ∆
1 – υ xz υ zx υ zy + υ xy υ zx
G 22 = --------------------------- G 23 = --------------------------------- Eq. 7-19
Ex Ez ∆ Ex Ez ∆
1 – υ xy υ yx
G 33 = ----------------------------
Ex Ey ∆
where:
υ ij = Poisson’s ratios
E x, E y, E z = Young’s modulus in the x-, y- and z-directions
G xy, G yz, G zx = shear moduli
1 – υ xy υ yx – υ yz υ zy – υ xz υ zx – 2υ yx υ zy υ xz
∆ = ------------------------------------------------------------------------------------------------------------------
-
Ex Ey Ez
also
G 44 = G xy
G 55 = G yz
G 66 = G zx
and
G 14 = G 15 = G 16 = 0.0
G 24 = G 25 = G 26 = 0.0
G 34 = G 35 = G 36 = 0.0
G 45 = G 46 = G 56 = 0.0
MAT9 Format
The format of the MAT9 Bulk Data entry is as follows:
1 2 3 4 5 6 7 8 9 10
MAT9 MID G11 G12 G13 G14 G15 G16 G22
G23 G24 G25 G26 G33 G34 G35 G36
G44 G45 G46 G55 G56 G66 RHO A1
A2 A3 A4 A5 A6 TREF GE
Field Contents
Gij Elements of the 6 ⋅ 6 symmetric material property
matrix in the material coordinate system.
RHO Mass density.
Ai Thermal expansion coefficient.
When you use the MAT9 entry, you should define a material coordinate system on
the PSOLID entry (using field 4). For solid elements, NX Nastran outputs stresses in
the material coordinate system, which by default, is the basic coordinate system. In
general, for solid elements, it isn’t easy to determine the orientation of the element
coordinate system.
CHAPTER 7 125
Material Properties

• You can use the MATT9 entry to designate certain material properties on
the MAT9 entry as temperature-dependent by referencing TABLEMi
entries.
See Also

• “PSOLID” on page 1615 of the NX Nastran Quick Reference Guide
MAT9 Example
The following example illustrates the use of the anisotropic material property using
the MAT9 material entry. This example consists of three parts as follows:
1. Develop a material property matrix using a pseudo-anisotropic model

consisting of a single CHEXA element with some bars along the edges and
across the diagonals.
2. Verify that the material matrix developed in part 1 is correct.
3. Use the material matrix in a distorted element.
This three-part approach is analogous to constructing an anisotropic material test

coupon and then testing that coupon in the test lab to obtain a strain matrix. From
the strain matrix, you develop the material matrix. Then, you verify the material
matrix by modeling the test coupon. Finally, you use the newly developed material
matrix in the actual structure.
Part 1 – Develop the Material Matrix

To simulate a three-dimensional anisotropic material, this example uses a single
CHEXA element with CBAR elements of various cross-sectional areas and bending
stiffnesses along the edges and across the diagonals as shown in Figure 7-3(a). The
cross sections of the bars aren’t important, but if you’re interested in the actual
dimensions, refer to the input file “anis1.dat” in the Test Problem Library located on
the delivery media. The dimensions of the cube are 1 ⋅ 1 ⋅ 1 inches. To constrain
the cube, one of the corner grid points is fixed in all six component directions.
Six self-equilibrating load cases are used to represent each of the six stress
components. Each direction was applied as a separate subcase. A self-equilibrating
normal load consists of applying equal and opposite forces to opposite faces of the
cube (i.e., forces F1, F2, and F3).
For the shear load, a self-equilibrating load consists of applying four forces around
the cube in order to place the cube in a state of pure shear. Note that for all
self-equilibrating load cases, the net resultant should be zero; hence, the SPC force
should also be zero. Because of the dimensions chosen for CHEXA, each load case
represents a unit stress, and the resulting strains are the strains due to unit stress.
F2 F2
F4 F4
F5 F5
F4
F4
F5 F1 F5
F6 F1
F6
F3 F4 F6
F3
Y Y
Z Z
X X
(a) (b)
Figure 7-3 Pseudo-Anisotropic Model
The strains at the center of the CHEXA were extracted from the output file and
summarized in Table 7-4.
Table 7-4 Strains Due to Unit Loads on Pseudo Anisotropic Model
Applied Unit Load
FX FY FZ FXY FYZ FZX
εX 0.155 -0.034 -0.056 -0.016 0.002 -0.011
εY -0.034 0.072 -0.018 -0.005 0.001 -0.003
Resulting εZ -0.056 -0.018 0.107 -0.008 0.003 -0.005

Strains
–6)
(10 γ XY -0.016 -0.005 -0.008 0.197 -0.011 0.009
γ YZ 0.002 0.001 0.003 -0.011 0.219 -0.021
γ ZX -0.011 -0.003 -0.005 0.009 -0.021 0.180

CHAPTER 7 127
Material Properties
Table 7-4 represents a strain matrix that, when inverted, is the material matrix [ G ]
introduced in Eq. 7-19. Note that since the strain matrix is symmetric, the material
matrix [ G ] is also symmetric.
The result is the material matrix [ G ] for our anisotropic material as shown in
Eq. 7-20.
10.67 6.91 6.89 1.25 – 0.09 0.86

6.91 19.03 6.90 1.27 – 0.12 0.83
[G] = 6.89 6.90 14.28 1.23 – 0.15 0.82 ⋅ 10 6 Eq. 7-20
1.25 1.27 1.23 5.27 0.21 – 0.11
– 0.09 – 0.12 – 0.15 0.21 4.63 0.52
0.86 0.83 0.82 – 0.11 0.52 5.72
Part 2 – Verify the Material Matrix

Now that the material matrix has been generated, the next step is to verify that the
matrix is entered on the MAT9 entry correctly. To do this, we return to the 1 ⋅ 1 ⋅ 1
inch cube. However, this time the analysis will not include the CBAR entries as
shown in Figure 7-3(b). The same six load cases are applied as in Part 1.
If the material matrix is correct, the strains produced should be numerically the
same as those shown in Table 7-4.
The input file for the verification model is shown in Listing 7-2.
Listing 7-2 Anisotropic Verification Input File
$
$ FILENAME - ANIS2.DAT
$
ID LINEAR,ANIS2
SOL 101
TIME 5
CEND
TITLE = ANISOTOPIC TEST ELEMENT
SPC = 1
SET 1 =13
STRAIN = 1
SPCFORCE=ALL
SUBCASE 1
LOAD = 1
SUBCASE 2
LOAD = 2
SUBCASE 3
LOAD = 3
SUBCASE 4
LOAD = 4
SUBCASE 5
LOAD = 5
SUBCASE 6
LOAD = 6
EGIN BULK$
GRID 1 0.0 0.0 0.0
GRID 2 1. 0.0 0.0
GRID 3 0.0 1. 0.0
GRID 4 1. 1. 0.0
GRID 6 0.0 0.0 1.
GRID 8 1. 0.0 1.
GRID 10 0.0 1. 1.
GRID 12 1. 1. 1.
CHAPTER 7 129
Material Properties
Listing 7-2 Anisotropic Verification Input File (continued)
CHEXA 13 1 1 6 8 2 3 10 +
+ 12 4
$
PLOAD4 1 13 1. 2 12 +
+ 1.
PLOAD4 1 13 1. 10 1 +
+ -1.
PLOAD4 2 13 1. 10 4 +
+ 0.0 1.
PLOAD4 2 13 1. 6 2 +
+ 0.0 -1.
PLOAD4 3 13 1. 6 12 +
+ 0.0 0.0 1.
PLOAD4 3 13 1. 1 4 +
+ 0.0 0.0 -1.
PLOAD4 4 13 1. 10 4 +
+ 1. 0.0
PLOAD4 4 13 1. 6 2 +
+ -1. 0.0
PLOAD4 4 13 1. 12 2 +
+ 1.
PLOAD4 4 13 1. 1 10 +
+ -1.
PLOAD4 5 13 1. 10 4 +
+ 1.
PLOAD4 5 13 1. 6 2 +
+ -1.
PLOAD4 5 13 1. 6 12 +
+ 1.
PLOAD4 5 13 1. 1 4 +
+ -1.
PLOAD4 6 13 1. 6 12 +
+ 1.
PLOAD4 6 13 1. 1 4 +
+ -1.
PLOAD4 6 13 1. 2 12 +
+ 1.
PLOAD4 6 13 1. 1 10 +
+ -1.
$
PSOLID 1 2 0
$
MAT9 2 10.67+6 6.91+6 6.89+6 1.25+6 -0.09+6 0.86+6 19.03+6
6.90+6 1.27+6 -0.12+6 0.83+6 14.28+6 1.23+6 -0.15+6 0.82+6
5.27+6 0.21+6 -0.11+6 4.63+6 0.52+6 5.72+6
SPC1 1 123 6
SPC1 1 12 1
SPC1 1 2 2
ENDDATA
You should notice a few items in the verification model input file. The six loads are
applied separately, each with their own subcase. The constraints are applied to
three separate grid points and are non-redundant. The non-redundant constraint
set allows the element to expand in all directions without imposing any constraint
forces. However, rigid body modes are constrained, which is a requirement for static
analysis. The last item to note is the use of the basic coordinate system for the material
definition.
The resulting center strains for the verification model are shown in Figure 7-4. As can
be seen, there is good agreement compared to the strain matrix shown in Table 7-4.
S T R A I N S I N H E X A H E D R O N S O L I D E L E M E N T S ( H E X A )
CORNER ------CENTER AND CORNER POINT STRAINS--------- DIR. COSINES MEAN
ELEMENT-ID GRID-ID NORMAL SHEAR PRINCIPAL -A- -B- -C- PRESSURE VON MISES
13
SUBCASE 1
CENTER X 1.549806E-07 XY -1.567790E-08 A 1.554435E-07 LX 1.00 0.02-0.04 -2.147629E-08 1.345509E-07
Y -3.437740E-08 YZ 2.221636E-09 B -5.634964E-08 LY-0.04-0.04-1.00
Z -5.617430E-08 ZX -1.076345E-08 C -3.466499E-08 LZ-0.03 1.00-0.04
SUBCASE 2
CENTER X -3.437740E-08 XY -5.186951E-09 A 7.192971E-08 LX-0.02 1.00 0.09 -6.659116E-09 6.601998E-08
Y 7.186195E-08 YZ 1.194498E-09 B -3.456629E-08 LY 1.00 0.02 0.01
Z -1.750721E-08 ZX -2.957457E-09 C -1.738608E-08 LZ 0.01 0.09-1.00
SUBCASE 3
CENTER X -5.617430E-08 XY -7.530764E-09 A 1.065883E-07 LX-0.01 1.00 0.10 -1.095211E-08 9.830458E-08
Y -1.750721E-08 YZ 2.773456E-09 B -5.656718E-08 LY 0.01 0.10-1.00
Z 1.065378E-07 ZX -4.683692E-09 C -1.716476E-08 LZ 1.00 0.01 0.01
SUBCASE 4
CENTER X -1.567790E-08 XY 1.970898E-07 A 8.825833E-08 LX 0.69 0.72-0.05 9.465206E-09 1.142492E-07
Y -5.186951E-09 YZ -1.062715E-08 B -1.095836E-07 LY 0.73-0.69 0.04
Z -7.530764E-09 ZX 8.945698E-09 C -7.070323E-09 LZ-0.01-0.07-1.00
SUBCASE 5
CENTER X 2.221636E-09 XY -1.062715E-08 A 1.126186E-07 LX-0.10-0.03-0.99 -2.063197E-09 1.271659E-07
Y 1.194498E-09 YZ 2.189892E-07 B -1.076343E-07 LY 0.70 0.71-0.09
Z 2.773456E-09 ZX -2.101742E-08 C 1.205294E-09 LZ 0.71-0.71-0.05
SUBCASE 6
CENTER X -1.076345E-08 XY 8.945698E-09 A 8.237462E-08 LX 0.69 0.71-0.12 6.134867E-09 1.046499E-07
Y -2.957457E-09 YZ -2.101742E-08 B -9.874503E-08 LY-0.05-0.11-0.99
Z -4.683692E-09 ZX 1.796267E-07 C -2.034187E-09 LZ 0.72-0.69 0.04
Note: Only the center strains are shown here to reduce space. The complete output
also includes the corner strains.
Figure 7-4 Verification Model Strain Output
Part 3 - Use the Anisotropic Material in the Actual Structure

The last step in the example is to use the material properties to model an actual
structure as shown in Figure 7-5. The structure is very coarse since it is used for
demonstration purposes only. The model consists of four elements with their
material coordinate system oriented 45 degrees with respect to the basic coordinate
system.
CHAPTER 7 131
Material Properties
Note: Ymat is along Yb and 70

Xmat and Zmat are F
rotated 45 degrees with F
68 Zmat
respect to Xb and Zb 69
F Ymat
F
67
Fiber
Zb Direction Xmat
Yb F = 100 lb
All elements are
Xb 5X5X5 inches.
Figure 7-5 Anisotropic Solids Model
The input file is shown in Listing 7-3. Note the local coordinate system used to
define the material orientation. The displacement results are shown in Figure 7-6.
Listing 7-3 Structure with Anisotropic Material Input File
$ FILENAME - ANIS3.DAT
$
ID LINEAR, ANIS3
SOL 101
TIME 5
CEND
TITLE = USING THE MATERIAL MATRIX
LOAD = 1
SPC = 1
DISP = ALL
BEGIN BULK
PARAM POST 0
PARAM AUTOSPC YES
$ MATERIAL COORDINATE SYSTEM
$
CORD2R 1 0 0.0 0.0 0.0 .70710680.0 .7071068+
.70710680.0 -.707107
GRID 51 0.0 5. 0.0
GRID 52 5. 5. 0.0
GRID 53 0.0 5. 5.
GRID 54 5. 5. 5.
GRID 55 0.0 0.0 0.0
GRID 56 5. 0.0 0.0
GRID 57 0.0 0.0 5.
GRID 58 5. 0.0 5.
GRID 59 0.0 5. 10.
GRID 60 5. 5. 10.
GRID 61 0.0 0.0 10.
GRID 62 5. 0.0 10.
GRID 63 0.0 5. 15.
GRID 64 5. 5. 15.
GRID 65 0.0 0.0 15.
GRID 66 5. 0.0 15.
GRID 67 10. 5. 10.
GRID 68 10. 5. 15.
GRID 69 10. 0.0 10.
GRID 70 10. 0.0 15.
$
CHEXA 1 1 51 55 56 52 53 57 +
+ 58 54
CHEXA 2 1 53 57 58 54 59 61 +
+ 62 60
CHEXA 3 1 59 61 62 60 63 65 +
+ 66 64
CHEXA 4 1 60 62 69 67 64 66 +
+ 70 68
$
$
FORCE 1 69 0 100. 1. 0.0 0.0
FORCE 1 67 0 100. 1. 0.0 0.0
FORCE 1 68 0 100. 1. 0.0 0.0
FORCE 1 70 0 100. 1. 0.0 0.0
$
PSOLID 1 2 1 GRID SMECH
$
MAT9 2 10.67+6 6.91+6 6.89+6 1.25+6 -0.09+6 0.86+6 19.03+6
6.90+6 1.27+6 -0.12+6 0.83+6 14.28+6 1.23+6 -0.15+6 0.82+6
5.27+6 0.21+6 -0.11+6 4.63+6 0.52+6 5.72+6
$
SPC 1 51 123456 0.0
SPC 1 52 123456 0.0
SPC 1 56 123456 0.0
SPC 1 55 123456 0.0
$
ENDDATA
CHAPTER 7 133
Material Properties

51 G 0.0 0.0 0.0 0.0 0.0 0.0
52 G 0.0 0.0 0.0 0.0 0.0 0.0
53 G 1.232738E-04 3.251065E-07 8.268995E-05 0.0 0.0 0.0
54 G 1.204173E-04 1.783952E-06 -8.797066E-05 0.0 0.0 0.0
55 G 0.0 0.0 0.0 0.0 0.0 0.0
56 G 0.0 0.0 0.0 0.0 0.0 0.0
57 G 1.002035E-04 2.286116E-05 7.448530E-05 0.0 0.0 0.0
58 G 9.797833E-05 -2.485534E-05 -7.772400E-05 0.0 0.0 0.0
59 G 3.605538E-04 1.530884E-05 1.199663E-04 0.0 0.0 0.0
60 G 3.610890E-04 -1.436989E-05 -1.294236E-04 0.0 0.0 0.0
61 G 3.284150E-04 1.772471E-05 1.189002E-04 0.0 0.0 0.0
62 G 3.284143E-04 -1.659684E-05 -1.265683E-04 0.0 0.0 0.0
63 G 6.271876E-04 1.717129E-05 1.256454E-04 0.0 0.0 0.0
64 G 6.300415E-04 -1.526032E-05 -1.351714E-04 0.0 0.0 0.0
65 G 5.944516E-04 1.910164E-05 1.262980E-04 0.0 0.0 0.0
66 G 5.975869E-04 -1.427919E-05 -1.336644E-04 0.0 0.0 0.0
67 G 3.671078E-04 -5.154453E-05 -4.031157E-04 0.0 0.0 0.0
68 G 6.358424E-04 -4.903843E-05 -4.012304E-04 0.0 0.0 0.0
69 G 3.339844E-04 -4.717636E-05 -4.023441E-04 0.0 0.0 0.0
70 G 6.039479E-04 -4.626662E-05 -3.997554E-04 0.0 0.0 0.0
Figure 7-6 Output for Structure with Anisotropic Material
7.5 Material Properties for Heat Transfer

You define material properties for heat transfer analysis are supplied on MAT4,
MAT5, MATT4, MATT5, RADM and RADMT entries.
• The MAT4, MAT5, and RADM entries provide temperature independent

(constant valued) material properties.
• The MATT4, MATT5, and RADMT entries supply information about
temperature-dependent properties through reference to material tables
(TABLEMi (i = 1, 2, 3, 4)).
With heat transfer material properties:
• Although the surface characteristics for free convection (heat transfer

coefficient) and radiation (emissivity and absorptivity) are not normally
considered to be material properties, they are included in the bulk data
entries described here.
• Temperature dependence of the quantity of interest is ultimately defined
on a TABLEMi entry. For example, the table is connected to the MAT4
through the MATT4 entry. Most material properties are directed toward a
structural element and are referenced by the property entry for the element
of interest.
Understanding MAT4 and MATT4

The MAT4 bulk data entry lets you define thermal properties for temperature-
independent isotropic materials.
The format of the MAT4 Bulk Data entry is:
1 2 3 4 5 6 7 8 9 10
MAT4 MID K CP ρ H µ HGEN REFENTH
TCH TDELTA QLAT
The following table describes the fields on the MAT4 Bulk Data entry.
Symbol Property Consistent Units

K(T) Thermal Conductivity* W/m °C
CP(T) Specific Heat* J/kg °C
Density* 3
ρ kg/m
H(T) Free Convection Heat Transfer Coefficient 2
W/m °C
Dynamic Viscosity* 2
µ (T) N sec/m
HGEN(T) Volumetric Internal Heat Generation 3
W/m
REFENTH Reference Enthalpy J/kg
TCH Lower Temperature Limit for Phase Change °C
TDELTA Temperature Range for Phase Change °C
QLAT Latent Heat J/kg
* Thermal conductivity, specific heat, density, and dynamic viscosity cannot be
time dependent. Thermal heat flux loads can reference explicit functions of time.
Thermal boundary conditions for convection and radiation utilize control node
techniques for specifying time dependent behavior.
The MATT4 entry lets you define temperature-dependent material properties in

conjunction with MAT4. NX Nastran always multiplies the basic quantities you
specify on the MAT4 entry by the corresponding tabular function (TABLEMi bulk
data entries) that you reference with the MATT4 entry.
See Also
• “Convection and Radiation Boundary Conditions” in the NX Nastran
Thermal Analysis User’s Guide.
CHAPTER 7 135
Material Properties

Understanding MAT5 and MATT5

The MAT5 bulk data entry lets you define thermal properties for temperature-
independent anisotropic materials. The MATT5 entry lets you define temperature
temperature-dependent material properties in conjunction with MAT5. NX Nastran
always multiplies the basic quantities you specify on the MAT5 entry by the
corresponding tabular function (TABLEMi bulk data entries) that you reference
with the MATT5 entry.
See Also
Understanding RADM and RADMT

The RADM bulk data entry lets you define the radiation properties of a boundary element
for heat transfer analysis. The format of the RADM Bulk Data entry is as follows:
1 2 3 4 5 6 7 8 9 10
RADM RADMID ABSORP EMIS1 EMIS2 EMIS3 EMIS4 EMIS5 EMIS6
EMIS7 -etc.-
The following table describes the fields for the RADM Bulk Data entry.
Field Description Unit

ABSORP(T) Absorptivity for Directional Heat Flux and Nondimensional
Radiation Boundary Condition (QVECT and
RADBC)
EMIS(λ,T) Emissivity for Radiation Boundary Condition Nondimensional
and Enclosure Radiation (RADBC and
RADSET)
The RADMT entry lets you specify table references for properties in a RADM entry
that you want to be temperature-dependent.
See Also
• “RADM” on page 1654 of the NX Nastran Quick Reference Guide
• “RADMT” on page 1656 of the NX Nastran Quick Reference Guide
7.6 Material Properties for Fluid Analysis

In NX Nastran, the MAT10 and MFLUID bulk data entries allow you to define
material properties for different types of fluid analyses.
Understanding MAT10
The MAT10 entry lets you define material properties for fluid elements for use in a
coupled fluid-structural interaction analysis.
See Also
• “Coupled Acoustic Analysis” on page 136 of the NX Nastran Advanced
Dynamic Analysis User’s Guide
• “Overview of Coupled Fluid-Structure Interaction” on page 672 of the NX
Understanding MFLUID
The MFLUID entry lets you define properties of an incompressible fluid volume so
that you can generate a virtual mass matrix.
See Also
• “MFLUID” on page 1413 of the NX Nastran Quick Reference Guide

• “Virtual Fluid Mass” on page 128 of the NX Nastran Advanced Dynamic
Analysis User’s Guide
7.7 Material Properties for Nonlinear Analysis

In NX Nastran, there are several different types of nonlinear analysis:
• With a nonlinear elastic analysis, you can use the MAT1 bulk data entry in
conjunction with the MATS1 and TABLES1 entries to define the stress-
dependent material properties.
• With an elastic-plastic analysis, you can use the MAT1 bulk data entry in
conjunction with the MATS1 and TABLES1 entries to define the stress-
dependent material properties.
• With a viscoelastic analysis, you can define elastoplastic properties to
consider plastic deformation using the MATS1 bulk data entry in
conjunction with MAT1.
CHAPTER 7 137
Material Properties
• With a temperature dependent analysis, you can specify temperature-

dependent properties by using the MATTi, TABLEST, and TABLES1 bulk
data entries.
• With a SOL 601 analysis, you can use the MATG entry (described below)
to define gasket materials.
• With a hyperelastic material analysis, you can use the MATHP entry
(described below) to define the material properties.
With a geometric nonlinear analysis, you can use either MAT2 or MAT9 to define
material properties.
See Also
• “Material and Geometric Nonlinearity” on page 61 of the NX Nastran

Basic Nonlinear Analysis User’s Guide
Understanding MATG
The MATG entry lets you define material properties for gasket type materials for
use in a SOL 601 analysis.
See Also
• “MATG” on page 1383 of the NX Nastran Quick Reference Guide
Understanding MATHP
In NX Nastran, you can use the MATHP bulk data entry to specify material
properties for use in a fully nonlinear hyperelastic analysis.
The generalized Mooney-Rivlin strain energy function may be expressed as follows:
NA ND
i j 2i
U ( J, I 1 , I 2 ) = ∑ Aij ( I 1 – 3 ) ( I 2 – 3 ) + ∑ Di ( J – 1 – AV ( T – T 0 ) )
i+j = 1 i = 1
A00 = 0
and 2D1 = K and 2 ( A10 + A01 ) = G at small strains.
where:
I1 first distortional strain invariant
I2 second distortional strain invariant
Aij material constants related to shear deformation (distortion)

Di material constants related to volumetric deformation
T current temperature
T0 initial temperature
K bulk modulus
G shear modulus
The model reduces to a Mooney-Rivlin material NA = 1 and to a Neo-Hookean

material if NA = 1 and A01 = 0.0 .
Using MATHP to Define Material Constants

The MATHP entry consists of five continuation entries that let you define the Aijand
Di material constants of the strain energy function. You obtain the values to enter in
the Aij and Di fields from least squares fitting of experimental data.
If you don’t know the material constants, enter one of the following strings in the sixth
continuation entry of MATHP to invoke NX Nastran’s curve-fitting algorithm:
TAB1, TAB2, TAB3, TAB4, and TABD
These TABi fields reference experimental data that is contained in TABLES1 bulk data
entries.
• You can use one or more of four experiments (TAB1 to TAB4 options) to
obtain Aij.
• You can use pure volumetric compression data (TABD option) to obtain Di.
If ND = 10 (order of the volumetric strain energy polynomial function) and you
either enter a nonzero value for D1 or a nonzero value for D1 is calculated from
experimental data in TABD, then the parameter estimation of the material constants
Aij takes compressibility into account in the cases of simple tension/compression,
equibiaxial tension, and general biaxial deformation. Otherwise, the software
assumes full incompressibility in estimating the material constants.
See Also
• “Defining Hyperelastic Materials in NX Nastran” on page 65 of the NX
Nastran Basic Nonlinear Analysis User’s Guide
Using MATHP to Model Mooney-Rivlin or Neo-Hookean Materials

If you’re modeling a Mooney-Rivlin or Neo-Hookean material, you don’t need to
specify a continuation entry for MATHP.
CHAPTER 7 139
Material Properties
Conventional Mooney-Rivlin and Neo-Hookean materials are fully incompressible.

With the MATHP entry, you can’t define full incompressibility. However, you can
simulate full incompressibility by specifying a large enough value for the D1 entry
(field 5). However, entering a value of D1 greater than 10 3 ⋅ ( A10 + A01 ) is not
recommended.
See Also
• “MATHP” on page 1390 of the NX Nastran Quick Reference Guide
8
CHAPTER
Applied Loads
■ Overview of Load Sets
■ Defining and Combining Subcases for Loading
■ Defining Static Loads
■ Point Loads
■ Distributed Loads
■ Inertia Loads (Acceleration Loads)
■ Thermal Loads
■ Loads Due to Enforced Motion
■ Frequency-Dependent Loads
■ Time-Dependent Loads
■ Using Static Loading Data in Dynamic Analysis
■ Defining Loads for Cyclic Symmetry Analysis
■ Combining Loads
8.1 Overview of Load Sets

Load sets in NX Nastran are collections, or lists, of loads. You define the loads using
bulk data entries. You then use case control commands to specify which of the
loading entries you want to use in a given analysis.
8.2 Defining and Combining Subcases for Loading

NX Nastran allows you to efficiently analyze multiple load cases in a single analysis
using the SUBCASE Case Control command. Each subcase defines a unique loading
condition. When you use create subcases:
• Each SUBCASE represents a separate, static loading condition, including

different boundary conditions.
• You can select different constraints, loads, and output in each SUBCASE.
• You must list SUBCASE ids within the Case Control section in ascending
order. However, the ids don’t need to be sequential. For example, you could
have SUBCASE 1, 14, 31, and 50.
Within the Case Control section, any title information, output requests, or Bulk Data
selections that you make prior to defining the first SUBCASE command are the
defaults for all subcases. These defaults apply to all of the subcases unless you
specifically override them with a different request within a subcase.
If you don’t include a SUBCASE command in the Case Control section of your input
file, the software assumes that the model contains a single subcase with an ID of 1.
See Also
• “Defining SUBCASES for Loading” on page 32 of the NX Nastran Reference

Manual
Example: Creating Two Load Subcases for the Truss

For example, suppose we add a second load case to the truss structure as shown in
Figure 8-1 so that we can perform two separate analyses during a single run.
• Subcase 1 consists of the 1000-pound force acting at grid point 4 in the

negative Y-direction.
CHAPTER 8 143
Applied Loads
• Subcase 2 consists of the 2000-pound force acting at grid point 3 in the

X-direction.
2 Y
A = 4.0 in
6 2
E = 30.0 × 10 lb ⁄ in
1 3
4 Subcase 2
J = 1.27 in 2 F =2000 lb
ν = 0.3 3
2
5 10 ft
Elements
1 X
Grid points
4 4
Subcase 1
Pin joints F =1000 lb
50 ft
Figure 8-1 Truss Example with Two Loading Conditions.
In the input file for this analysis (shown in Listing 8-1). In this input file, we placed
a displacement request (DISPLACEMENT - ALL) above the first subcase but not
within either of the two subcases. Therefore, NX Nastran outputs displacements for
all the grid points for both subcases. Since we included an element force request
only in Subcase 2, NX Nastran doesn’t generate any element force output for
Subcase 1. It only outputs the element forces for the grid points for Subcase 2.
Listing 8-1 Truss Input File with Two Loadings
$ Filename - TRUSS2.DAT
ID LINEAR,TRUSS2
SOL 101
TIME 2
CEND
TITLE = LINEAR STATICS USER’S GUIDE - TWO SUBCASES
SPC = 11
DISPLACEMENT = ALL
$
SUBCASE 1
LOAD = 10
$
SUBCASE 2
LOAD = 11
ELFORCE = ALL
$
BEGIN BULK
$
$ DESCRIBE THE GEOMETRY
$
GRID 1 0. 0. 0. 3456
GRID 2 0. 120. 0. 3456
GRID 3 600. 120. 0. 3456
GRID 4 600. 0. 0. 3456
$
$ MEMBERS ARE MODELED USING
$ ROD ELEMENTS
$
CROD 1 21 2 3
CROD 2 21 2 4
CROD 3 21 1 3
CROD 4 21 1 4
CROD 5 21 3 4
$
$ PROPERTIES OF ROD ELEMENTS
$
PROD 21 22 4. 1.27
$
$ MATERIAL PROPERTIES
$
MAT1 22 30.E6 .3
$
SPC1 11 123456 1 2
$
$ POINT LOAD SUBCASE 1
FORCE 10 4 1000. 0. -1. 0.
$
$ POINT LOAD SUBCASE 2
FORCE 11 3 2000. 1. 0. 0.
ENDDATA
CHAPTER 8 145
Applied Loads
Using REPCASE
For large models, you may need to further partition the results in the output files in
a certain manner or repeat the same output in different subcases. You can do this
withe the REPCASE Case Control command. The REPCASE command lets you
organize the output based on the element and/or grid point ID numbers.
Listing 8-2 shows how you can use REPCASE separate the results for the truss
example into three different sections in the .f06 output file. In this example, the
REPCASE 2 command provides for additional output from the solution generated
in SUBCASE 1 without performing any additional analysis.
For the Case Control shown in Listing 8-2, SUBCASE 1 output consists of the
displacement for all of the grid points and the element forces for elements 1 and 2.
The output for REPCASE 2 consists of the element force for elements 3 through 5.
Note that the default output request for the grid point displacements is overridden
by the DISPLACEMENT = NONE command. SUBCASE 3 is a new loading
condition with its own output requests. A second static solution is performed for
this subcase. Also note that only two static solutions are performed for this input
file.
Listing 8-2 Truss Input File with the REPCASE Command
$ FILENAME TRUSS4.DAT
$
ID LINEAR,TRUSS4
SOL 101
TIME 2
CEND
TITLE = LINEAR STATICS USER’S GUIDE - REPCASE EXAMPLE
LOAD = 10
SPC = 11
DISPLACEMENT = ALL
SET 1 = 3,4
SET 2 = 1,2
SET 3 = 3 THRU 5
$
SUBCASE 1
LOAD = 10
ELFORCE = 2
$
REPCASE 2
ELFORCE = 3
DISPLACEMENT = NONE
$
SUBCASE 3
LOAD = 11
DISPLACEMENT =1
ELFORCE = 1
BEGIN BULK
See Also
• “REPCASE” on page 331 of the NX Nastran Quick Reference Guide
Using SUBCOM to Combine Subcases

When you’re performing a linear statics analysis, you can use the SUBCOM
command to create new subcases by combining other subcases. When you use
SUBCOM to combine subcases, you must use the SUBSEQ command to specify the
coefficients for combining the previous subcases. A SUBSEQ command is required
for each SUBCOM subcase.
The use of the SUBCASE and SUBSEQ commands is best illustrated with an example.
Returning to the truss structure, the two load cases were previously treated as
separate loads, each with their own subcase. Now, suppose a third load case is
desired that is a linear combination of the first two subcases. The input file shown in
Listing 8-3 illustrates this.
Listing 8-3 SUBCOM Input File
$
ID LINEAR,TRUSS5
SOL 101
TIME 2
CEND
TITLE = LINEAR STATICS USER’S GUIDE SUBCOM EXAMPLE
LOAD = 10
SPC = 11
DISPLACEMENT = ALL
$
SUBCASE 1
LOAD = 10
$
SUBCASE 2
LOAD = 11
ELFORCE = ALL
$
SUBCOM 3
SUBSEQ 1.0,0.5
ELFORCE = ALL
$
SUBCOM 4
SUBSEQ 1.0,-0.5
ELFORCE =ALL
$
BEGIN BULK
For this Case Control section, the SUBCOM 3 output is as follows:

CHAPTER 8 147
Applied Loads
 SUBCOM 3   SUBCASE 1   SUBCASE 2 

  = 1.0   + 0.5  
 OUTPUT   OUTPUT   OUTPUT 
The output for SUBCOM 4 is as follows:
 SUBCOM 4   SUBCASE 1   SUBCASE 2 

  = 1.0   – 0.5  
 OUTPUT   OUTPUT   OUTPUT 
The coefficients 1.0, and 0.5 for SUBCOM 3 are input on a SUBSEQ command. These
coefficients used in the SUBSEQ command refer to the previous Subcases 1 and 2.
The 1.0 value in SUBCOM 3 is the coefficient corresponding to SUBCASE 1 and the
0.5 is the coefficient associated with SUBCOM 2. SUBCOM 3 is not considered a
subcase. Likewise, the coefficients 1.0 and –0.5 for SUBCOM 4 refer to SUBCASES
1 and 2. If you specify a third SUBSEQ coefficient for either SUBCOM, the job will
fail.
You can use SUBCOM commands to combine any number of previous subcases.
However, you must define a coefficient on the SUBSEQ command for each subcase.
If you want to exclude a subcase from the summation, set the coefficient to 0.0 for
that particular subcase.
Note: You can define the combined loads in their own subcases and not use
SUBCOMs. If you only want to generate the output of the combined loads (so you
only need two subcases), this is the preferred way. However, using SUBCOMs is
helpful if you want to review the structural response to the individual loads in
addition to the combined load. The SUBCOM command sums up the results of the
previous subcases multiplied by the appropriate scale factors. One of the best
applications of the SUBCOM command is combining the results generated with
different constraints.
See Also
• “SUBCOM” on page 389 of the NX Nastran Quick Reference Guide
Using SYMCOM to Create a Symmetry Combination Subcase

In a static analysis, you can use the SYMCOM case control command to create a
symmetry combination subcase. When you use the SYMCOM command to combine
selected SUBCASES, you must also use a SYMSEQ command to specify the
coefficients for combining the symmetry subcases.
See Also
• “SYMCOM” on page 404 of the NX Nastran Quick Reference Guide

• “SYMSEQ” on page 406 of the NX Nastran Quick Reference Guide
Temperature Loads and Enforced Deformations in Subcase

Combinations (SUBCOMS)
You can use the SUBCOM and SYMCOM case control commands to combine the
results of two or more immediately preceding subcases or SYM cases respectively. In
NX Nastran, the results for a combined loading condition are efficiently obtained by
forming the linear combination of the displacement vectors obtained in the previous
subcases. NX Nastran calculates the element forces and stresses in the usual manner
from the displacement vector for the combined loading condition. This technique is
quite straightforward when the several loading conditions are composed of forces
and moments.
However, if any of the subcases that you’re combining contain thermal loads and/or
enforced element deformations, you must exercise some care to obtain correct
element stresses and forces for the combined loading condition.
See Also
• “Using Temperature Loads with Subcases” on page 184 of the NX Nastran
User’s Guide
8.3 Defining Static Loads

In NX Nastran, you can apply static loads to geometric and scalar points in a variety
of ways, including
• Loads applied directly to grid points

• Pressure loads on surfaces
• Distributed and concentrated loads on CBAR, CBEAM, and CBEND
elements
• Gravity loads
• Centrifugal loads due to steady rotation
• Tangential loads due to angular acceleration
• Loads resulting from thermal expansion
• Loads resulting from enforced deformations of a structural element
• Loads resulting from enforced displacements at a grid point
While you can define multiple load sets in the bulk data section of your input file, the
software only uses the sets that you explicitly select in the case control section to solve
the model.
See Also
• “Defining SUBCASES for Loading” on page 32 of the NX Nastran Reference

Manual
CHAPTER 8 149
Applied Loads
• Section 3.6 of The NASTRAN Theoretical Manual

8.4 Point Loads
This section shows how to apply concentrated forces and concentrated moments at
grid points.
• You can use the FORCE, FORCE1, and FORCE2 entries to apply
concentrated forces directly to grid points.
• You can use the MOMENT, MOMENT1, and MOMENT2 entries to apply
a apply concentrated force in the translational degrees of freedom (T1, T2,
and T3) at a grid point.
• Use FORCEAX and MOMAX to define forces or moments on conical shell
elements.
• Use SLOAD to define a concentrated static load at a scalar point.
FORCE, FORCE1, FORCE2

You can apply concentrated forces directly to grid points with the FORCE, FORCE1,
and FORCE2 entries.
• Use a FORCE entry if you want to define a static, concentrated force at a

grid point by specifying a vector.
• Use a FORCE1 entry if the direction is determined by a vector connecting
two grid points.
• Use a FORCE2 entry if the direction is specified by the cross product of two
such vectors.
Using FORCE
The FORCE entry lets you specify the magnitude and direction of a force vector in
any coordinate system as shown below. You can use the FORCE entry to define a
static load applied to a geometric grid point in terms of components defined by a
local coordinate system. The orientation of the load components depends on the
type of local coordinate system you used define the load. The directions of the load
components are the same as those indicated on Figure 5-2 of “Defining Grid Points
and Scalar Points” on page 71 for displacement components.
Consider a force F acting on a cantilever beam:
F = 10 lbf Y
X
Z
1 2
A concentrated force F is applied to a CBAR element connecting grid points 1 and 2

using the FORCE entry. Its format is as follows:
1 2 3 4 5 6 7 8 9 10
FORCE SID G CID F N1 N2 N3
Field Contents
SID Load set identification number. (Integer > 0)
G Grid point identification number. (Integer > 0)
CID Coordinate system identification number. ( Integer ≥ 0 ;
Default = 0)
F Scale factor. (Real)
Ni Components of a vector measured in coordinate system
defined by CID. (Real; at least one Ni ≠ 0.0 )
In our example, the FORCE entry may be written as:
FORCE 2 2 10. 0. -1. 0.
or, in free field format,
FORCE,2,2,,10.,0.,-1.,0.
The load set identification number (SID in column 2) refers to a command defined in
the Case Control Section (LOAD=2 in this example; the integer value 2 is arbitrarily
chosen). Leaving column 4 blank means that the basic (default) coordinate system is
used to specify the orientation of the load. The (0., -1., 0.) entries in columns 6, 7, and
8 refer to a vector in the -Y direction, defining the direction of application of the load.
The force applied to the grid point is f , given by
f = FN Eq. 8-1
where N = (N1,N2,N3)
Thus, the value of F in column 5 is the full value of the applied load of 10 lbf because
vector N (in this example) is of unit length.
CHAPTER 8 151
Applied Loads
The magnitude of the applied force is the scale factor (field 5) times the magnitude
of the vector defined in fields 6 through 8. For example, the force applied with the
following two FORCE entries is the same.
1 2 3 4 5 6 7 8 9 10
FORCE 1 3 0 -100. 1. 0. 2.
1 2 3 4 5 6 7 8 9 10
FORCE 1 3 0 2. -50. 0. -100.
See Also
• “FORCE” on page 1262 of the NX Nastran Quick Reference Guide
Using FORCE1 and FORCE2

• The FORCE1 entry lets you define a force by specifying a magnitude and
two grid points (not necessarily the loaded grid point) to determine the
direction.
• The FORCE2 entry lets you define a force by specifying a magnitude with
the direction defined by the vector product of two other vectors.
The format of the FORCE1 and FORCE2 entries is as follows:
1 2 3 4 5 6 7 8 9 10
FORCE1 SID G F G1 G2
1 2 3 4 5 6 7 8 9 10
FORCE2 SID G F G1 G2 G3 G4
Field Contents
SID Load set identification number.
G Grid point identification number.
F Magnitude of the force.
Gi Grid point identification numbers.
FORCE1 and FORCE2 Example

While the FORCE1 and FORCE2 forms are not used extensively, they are useful for
some modeling situations as shown in the following example.
Suppose you wish to model a curved structure with edge loads as shown in
Figure 8-2. The edge loads consist of 10 N per element (acting normal to the
elements) and 8 N per element along the edge of each element (acting as a shear load).
The FORCE1 and FORCE2 entries are the ideal choice to apply these loads. You
could use the FORCE entry, but it would be more difficult because you would need
to compute the normals for each of the loaded CQUAD4 elements. Using FORCE1
and FORCE2 entries for the shear and normal loads, respectively, makes the task
much easier.
The 10 N normal force for each element is divided equally between the two grid
points along the edge. The directions of the resulting 5 N forces are applied using the
FORCE2 entry by taking the cross product of the vectors defined by the edges to
which the grid points are attached. The result is a 5 N load acting normal to the plate.
The 8 N shear force for each element is also divided up equally between the two grid
points along the edge. The resulting 4 N loads are applied with the FORCE1 entry.
The directions of these loads are determined by the line connecting the edge grid
points.
The input file is shown in Listing 8-4. Note that the normal forces and the shear
forces are applied as one load condition. If you want the forces to act as separate
loadings, you can use different IDs for the FORCE1 and FORCE2 entries and call
them out using two separate subcases in your input file.
F1
10 11 12
9
7 8 F2
F1
4 5 6 F2
F1 F2
1 2 3
F 1 = 10 N
F2 = 8 N
Figure 8-2 Edge Forces on a Plate Element
CHAPTER 8 153
Applied Loads
Listing 8-4 Edge Loads
$ FILENAME - EDGE.DAT
$
OLOAD=ALL
LOAD=3
.
.
.
BEGIN BULK
$
$ UNITS ARE CM AND N
$
FORCE1 3 3 4. 3 6
FORCE1 3 6 4. 3 6
FORCE1 3 6 4. 6 9
FORCE1 3 9 4. 6 9
FORCE1 3 9 4. 9 12
FORCE1 3 12 4. 9 12
FORCE2 3 3 5. 3 6 3 2
FORCE2 3 6 5. 3 6 3 2
FORCE2 3 6 5. 6 9 6 5
FORCE2 3 9 5. 6 9 6 5
FORCE2 3 9 5. 9 12 9 8
FORCE2 3 12 5. 9 12 9 8
$
GRID 1 0.0 -60. -2.
GRID 2 60. -60. -2.
GRID 3 120. -60. -2.
GRID 4 0.0 -30. 0.0
GRID 5 60. -30. 0.0
GRID 6 120. -30. 0.0
GRID 7 0.0 30. 0.0
GRID 8 60. 30. 0.0
GRID 9 120. 30. 0.0
GRID 10 0.0 60. -2.
GRID 11 60. 60. -2.
GRID 12 120. 60. -2.
$
PSHELL 1 4 .5 4
MAT1 4 7.1E6 .3
$
CQUAD4 1 1 1 2 5 4
CQUAD4 2 1 2 5 6 3
CQUAD4 3 1 4 5 8 7
CQUAD4 4 1 5 6 9 8
CQUAD4 5 1 7 8 11 10
CQUAD4 6 1 8 9 12 11
$
SPC1 1 123456 1 4 7 10
$
ENDDATA
Note that the two loads are applied at the same grid points. This is perfectly
acceptable. The total force acting at the grid points is the sum of all the applied
loads. The OLOAD (applied load output request) output, which you request
through the Case Control section, shows the applied force resultant acting at a set of
user specified grid points. The OLOAD output is useful in verifying that the loads
are applied correctly. A partial listing of the output file showing the OLOAD output
is shown in Figure 8-3.
L O A D V E C T O R

3 G 0.0 3.658545E+00 5.255002E+00 0.0 0.0 0.0
6 G 0.0 7.658545E+00 1.025500E+01 0.0 0.0 0.0
9 G 0.0 8.323735E+00 9.722850E+00 0.0 0.0 0.0
12 G 0.0 4.323736E+00 4.722850E+00 0.0 0.0 0.0
Figure 8-3 OLOAD Output
See Also
• “FORCE1” on page 1264 of the NX Nastran Quick Reference Guide

• “FORCE2” on page 1266 of the NX Nastran Quick Reference Guide
Using MOMENT, MOMENT1, MOMENT2

MOMENT, MOMENT1, and MOMENT2 are similar to the three FORCE entries. The
only difference is that you use force entries to apply concentrated force in the
translational degrees of freedom (T1, T2, and T3), while you use moment entries
apply concentrated moments in the rotational degrees of freedom (R1, R2, and R3).
• With MOMENT, you define a static, concentrated moment at a grid point

by specifying a scale factor and a vector that determines the direction
• With MOMENT1, the software uses two grid points to determine direction
( m = Mn , where n is a unit vector parallel to the vector from grid 1 to grid 2).
• With MOMENT2, the software uses four grid points to determine direction
( m = Mn , where n is the unit vector parallel to the cross product of the
vectors from G1 to G2, and G3 to G4).
The application of concentrated moments is analogous to forces. MOMENT has the

following format:
1 2 3 4 5 6 7 8 9 10
MOMENT SID G CID M N1 N2 N3
Field Contents
G Grid point identification number at which the moment is
applied. (Integer > 0)
CHAPTER 8 155
Applied Loads
Field Contents
CID Coordinate system identification number. ( Integer ≥ 0 or
blank)
M Scale factor. (Real)
Ni Components of the vector measured in the coordinate system
defined by CID. (Real; at least one Ni ≠ 0.0 )
Consider moment M acting about the basic Z axis of the simply supported beam
shown below:
M = 18.6 in-lbf
Y
1 3 X
2
Z
In this case, the MOMENT entry could be written as
1 2 3 4 5 6 7 8 9 10
MOMENT 6 2 -18.6 0.0 0.0 1.
MOMENT,6,2,,-18.6,0.,0.,1.
The applied moment m is given by
m = MN Eq. 8-2
where N is the vector (N1, N2, N3). As was the case with FORCE entry, M is the full
magnitude of the moment since N is a vector of unit length. The direction of the
applied moment is given by the sign of M according to the right-hand rule because
N = (0., 0., 1.) is a vector in the direction of the positive Z axis direction. Note that
specifying M = 18.6 and N = ( 0., 0., -1. ) would produce an equivalent result.
See Also
• “MOMENT” on page 1423 of the NX Nastran Quick Reference Guide
• “MOMENT1” on page 1425 of the NX Nastran Quick Reference Guide
• “MOMENT2” on page 1427 of the NX Nastran Quick Reference Guide
Using FORCEAX, MOMAX

You can use the FORCEAX and MOMAX entries to define the loading of specified
harmonics on rings of conical shell elements. You can use FORCE and MOMENT
entries to apply concentrated loads or moments to conical shell elements if you
defined such points with a POINTAX entry.
See Also
• “FORCEAX” on page 1268 of the NX Nastran Quick Reference Guide

• “MOMAX” on page 1421 of the NX Nastran Quick Reference Guide
Using SLOAD
You can use the SLOAD entry to define a concentrated static load at a scalar point or
a grid point. With SLOAD, you can only specify the magnitude since only one
component of motion exists at a scalar point.
See Also
• “SLOAD” on page 1746 of the NX Nastran Quick Reference Guide
8.5 Distributed Loads
There are three Bulk Data entries available to apply distributed loads to element
surfaces.
• Use PLOAD define a uniform static pressure load on a triangular or

quadrilateral surface.
• Use PLOAD1 and PLOAD2 to apply a distributed load to the CQUAD4,
CTRIA3, and CSHEAR elements only.
• Use PLOAD4 to apply distributed loads on any of the two-dimensional
elements and on the surfaces of three-dimensional elements.
PLOAD
You can use the PLOAD entry to define a uniform normal static pressure load on
triangular or quadrilateral surfaces by using grid points. You can apply PLOAD to
2-D (surface) or 3-D (solid) elements.
The PLOAD entry is different from the other PLOADi entries because it references
three or four grid points rather than the element itself. When you use PLOAD to load
a CTRIA3 element, the total force acting on the element is divided evenly among the
three corner points. When you use PLOAD to load a CQUAD4 element, the total
force acting on the element is distributed to corner grid points using the geometric
CHAPTER 8 157
Applied Loads
shape of the element. The resultant of the applied corner loads acts through the
centroid of the element. The direction of the pressure load is determined using the
right-hand rule by the numbering sequence of the grid points on the PLOAD entry.
See Also
• “PLOAD” on page 1576 of the NX Nastran Quick Reference Guide

The PLOAD entry has the following form:
1 2 3 4 5 6 7 8 9 10
PLOAD SID P G1 G2 G3 G4
Field Contents
P Pressure. (Real)
Gi Grid point identification numbers. (Integer > 0; G4 may be zero
or blank)
Grid Points G1, G2, G3, and G4 define either a triangular or quadrilateral surface; if
G4 is zero or blank, the software assumes that the surface is triangular. The direction
of the pressure load is determined by applying the right-hand rule to the grid point
ordering sequence of the surface.
P P
G3 G4
G8 G3
G1
G1
G2
G2
G5 G7
G6
Pressure is applied in the opposite direction by making the value of P negative.
PLOAD1
You can use the PLOAD1 entry to apply a distributed load to a CBAR, CBEAM, or
CBEND element. You can use the PLOAD1 entry for both concentrated and linearly
distributed forces.
• For the CBAR and CBEAM elements, you can apply the linearly distributed
force between any two locations on the element (or off the element if you
wish.)
• For the CBEND element, you can only apply the linearly distributed loads
to linearly varying forces and moments between the end points.
You can use PLOAD1 entry to define either distributed or concentrated loads at
intermediate points on CBAR and CBEAM elements. You define applied loads at the
end points of CBAR and CBEAM elements with either FORCE or MOMENT entries.
You can also use PLOAD1 to define only distributed loads for the CBEND element.
For the CBEND element, you can only use PLOAD1 to define
You can use the CBARAO entry is used to define intermediate points on CBAR
elements where you want to obtain stress and/or force output. You can also use the
PLOAD1 to define only distributed loads for the CBEND element. The distributed
load is always applied along the entire length of the element and does not define
loads at intermediate points on the CBEND. The PLOAD1 entry cannot be used to
define concentrated loads at intermediate stations on the CBEND element.
You can apply the load can be applied along the entire element length, a segment of
the length, or at a point along the length. The form of the PLOAD1 entry is shown
below. The meanings of X1, X2, P1, and P2 are shown in Figure 8-4.
Format:
1 2 3 4 5 6 7 8 9 10
PLOAD SID EID TYPE SCALE X1 P1 X2 P2
Field Contents
EID CBAR, CBEAM, or CBEND element identification number.
(Integer > 0)
TYPE Load type. (Character: “FX”, “FY”, “FZ”, “FXE”, “FYE”,
“FZE”, “MX”, “MY”, “MZ”, “MXE”, “MYE”, “MZE”)
SCALE Determines scale factor for X1, X2. (Character: “LE”, “FR”,
“LEPR”, “FRPR”)
CHAPTER 8 159
Applied Loads
Field Contents
X1, X2 Distances along the CBAR, CBEAM, or CBEND element axis
from end A. (Real; X2 may be blank, 0 ≤ X1 ≤ X2 )
P1, P2 Load factors at positions X1, X2. (Real or blank)
P2
P1
GA GB
X1
X2
l
Figure 8-4 PLOAD1 Convention in CBAR or CBEAM Elements
The TYPE option in field 4 lets you specify the type of the load . The load can be a
concentrated force or moment in the basic or element coordinate system. If the
applied load is to be a concentrated load, leave fields 8 and 9 blank and the
concentrated load will be applied at the X1 location. If both X1 and X2 are input, the
load will be taken as a linearly varying load between X1 and X2.
The SCALE option in field 5 lets you specify the location of the load on the CBAR
and CBEAM elements using field 5 of the PLOAD1 entry.
• With the “LE” or “LEPR” methods, you specify the actual start and end
positions of the load as measured from end A of the element. When using
this method, the distances X1 and X2, as shown in Figure 8-5, are in the
same units as the dimensions used for the model.
• With the “FR” or “FRPR” methods, you specify the percent (using
“X/XB”) along the element where the load starts and ends. A value of 0.0
defines end A, while a value of 1.0 defines end B. Again, the start of the
load is measured from end A.
For both methods of describing the location of the loads, you can also specify
whether the applied load is to be a direct load (scale “LE” and “FR”) or a projected
load (scale “LEPR” or “FRPR”). For the projected loads, distributed loads are
entered in terms of the projected length of the element as shown in Figure 8-5.
Remember that if you input the loads in terms of the basic coordinate system (“FX”,
“FY”, “FZ”, “MX”, “MY”, or “MZ”), then the projected angle a is with respect to the
basic coordinate system, not the element coordinate system.
PLOAD1Examples
This section contains a series of examples that illustrate different uses of the PLOAD1
entry.
PLOAD1 Example 1
In this example, we used PLOAD1 to apply a uniformly distributed load over the full
length of a CBAR element using fractional (normalized) scaling.
Y
Basic System
12.6 lbf./inch
X
Z
1 2 CBAR Element ID = 52
2.0 inches
Note that P1 = P2 = 12.6 lbf/inch

X1 = 0.0
fractional scaling
X2 = 0.0
1 2 3 4 5 6 7 8 9 10
PLOAD1 SID EID TYPE SCALE X1 P1 X2 P2
PLOAD1 36 52 FY FR 0.0 -12.6 1.0 -12.6
Note: You can also use PLOAD1 to apply a uniformly distributed load over the full
length of a CBAR element using length scaling.
1 2 3 4 5 6 7 8 9 10
PLOAD1 36 52 FY LE 0.0 -12.6 2.0 -12.6
PLOAD1 Example 2
In this example, we used PLOAD1 to apply a linearly varying distributed load to the
interior of a CBAR element using length scaling:
CHAPTER 8 161
Applied Loads
8.4 lbs./inch Y
Basic System
3.1 lbs./inch X
Z
1 2 CBAR Element
ID = 52
1.0 in.
2.0 in.
3.0 in.
1 2 3 4 5 6 7 8 9 10
PLOAD1 36 52 FY LE 1.0 -3.1 2.0 -8.4
PLOAD Example 3
The example demonstrating the use of the PLOAD1 entry consists of applying a
direct linearly varying load to the three-bar structure shown in Figure 8-5.
N
45000 ----
m
N
23000 ----
m
x
1 10 2 11 3 12 4
0.5 m
0.3 m
1m 1m 1m
Figure 8-5 Distributed Load on CBAR Elements
The input file for this example is shown in Listing 8-5.
Listing 8-5 PLOAD1 Loading
$
$ FILENAME - DISTRIB.DAT
$
ID LINEAR,DISTRIB
SOL 101
TIME 2
CEND
LOAD = 1
FORCE = ALL
BEGIN BULK
$
$ UNITS ARE MM AND N
$
GRID 1 0. 0.0 0.0 123456
GRID 2 1000. 0.0 0.0 345
GRID 3 2000. 0.0 0.0 345
GRID 4 3000. 0.0 0.0 123456
$
CBAR 10 1 1 2 1.0 0.0 1.0
CBAR 11 1 2 3 1.0 0.0 1.0
CBAR 12 1 3 4 1.0 0.0 1.0
CBARAO 11 FR 19 .05 .05
$
PBAR 1 1 100.0 100.0 100.0 100.0
$
MAT1 1 4.E4 .3
$
PLOAD1 1 11 FY LE 300. 23. 800. 45.
$
ENDDATA
Note the that the starting location for the linearly varying load starts and ends at 0.3
m (300 mm) and 0.8 m (800 mm), respectively. This is the distance measured from
end A. One of the most common mistakes is to make these distances 1.3 m and 1.8
m, respectively, while thinking that it is the actual distance measured in the basic
coordinate system. It is not; it is the distance measured from end A of the loaded
element-in this case, grid point 2.
When you use the PLOAD1 entry to define a distributed load, the CBARAO (CBAR
Additional Output) entry is helpful. You use the CBARAO entry to request stress
and force output at intermediate locations along the CBAR element. The CBARAO
output for element 2 is shown in Figure 8-6.
CHAPTER 8 163
Applied Loads
F O R C E D I S T R I B U T I O N I N B A R E L E M E N T S ( C B A R )
ELEMENT STATION BEND-MOMENT SHEAR FORCE AXIAL
ID. (PCT) PLANE 1 PLANE 2 PLANE 1 PLANE 2 FORCE TORQUE
10 .000 .0 -5.982597E+06 .0 -7.851454E+03 .0 .0
10 1.000 .0 1.868857E+06 .0 -7.851454E+03 .0 .0
11 .000 .0 1.868857E+06 .0 -7.851454E+03 .0 .0
11 .050 .0 2.261429E+06 .0 -7.851454E+03 .0 .0
11 .100 .0 2.654002E+06 .0 -7.851454E+03 .0 .0
11 .150 .0 3.046575E+06 .0 -7.851454E+03 .0 .0
11 .200 .0 3.439148E+06 .0 -7.851454E+03 .0 .0
11 .250 .0 3.831720E+06 .0 -7.851454E+03 .0 .0
11 .300 .0 4.224293E+06 .0 -7.851454E+03 .0 .0
11 .350 .0 4.587198E+06 .0 -6.646454E+03 .0 .0
11 .400 .0 4.887105E+06 .0 -5.331454E+03 .0 .0
11 .450 .0 5.118511E+06 .0 -3.906454E+03 .0 .0
11 .500 .0 5.275917E+06 .0 -2.371454E+03 .0 .0
11 .550 .0 5.353823E+06 .0 -7.264532E+02 .0 .0
11 .600 .0 5.346729E+06 .0 1.028546E+03 .0 .0
11 .650 .0 5.249135E+06 .0 2.893546E+03 .0 .0
11 .700 .0 5.055540E+06 .0 4.868546E+03 .0 .0
11 .750 .0 4.760446E+06 .0 6.953546E+03 .0 .0
11 .800 .0 4.358353E+06 .0 9.148546E+03 .0 .0
11 .850 .0 3.900926E+06 .0 9.148546E+03 .0 .0
11 .900 .0 3.443498E+06 .0 9.148546E+03 .0 .0
11 .950 .0 2.986070E+06 .0 9.148546E+03 .0 .0
11 1.000 .0 2.528644E+06 .0 9.148546E+03 .0 .0
12 .000 .0 2.528644E+06 .0 9.148546E+03 .0 .0
12 1.000 .0 -6.619903E+06 .0 9.148546E+03 .0 .0
Figure 8-6 CBARAO Output for the Center Element
The CBARAO entry is one of the best tools available for model checkout when
you’re using linearly varying loads on the CBAR elements. The CBARO output lets
you easily generate a shear and moment diagram as shown in Figure 8-7 below.
10
Shear Force (N)

5
3
10
–5
–10
–1
0 1 2 3 4 5 6 7 8 9 10 • 10
Distance from End A of Element 11 (m)
6
Bending Moment (N-mm)
5
4
3
6
10
2
1 –1
0 1 2 3 4 5 6 7 8 9 10 • 10
Distance from End A of Element 11 (m)
Figure 8-7 Shear and Moment Diagram Generated from CBARAO Output
For CBEAM elements, you can use the PBEAM entry to directly request intermediate
output along the element.
PLOAD1 Example 4
You can also use PLOAD1 to apply a concentrated load at an interior point of a CBAR
element using fractional scaling.
Y
1000 lbs. Basic System
X
Z
1 2
1.0 in. 1.0 in. CBAR Element ID = 52
CHAPTER 8 165
Applied Loads
1 2 3 4 5 6 7 8 9 10
PLOAD1 36 52 FY FR 0.5 -1000.
PLOAD1 Example 5
Another feature of the PLOAD1 entry is the ability to apply projected loads. To
apply a load as a projected load, you choose a scale (field 5) of “LEPR” for the actual
length from end A or “FRPR” for the fractional distance from end A. This example
shows illustrates how you can use PLOAD1 to define a projected load using a snow
load on a truss structure as shown in Figure 8-8.
N
N 4 Snow Load = 500 ----
E = 4 ⋅ 10 -------------- m
2
mm
9
7 2.5 m
6
8 11
1 5 8
6
6.5 m
13 m
Y el 2
58 mm A = 1856 mm
Y
5 4
Z el I 1 = 5.2 × 10 mm
X 32 mm = grids
= cbar
Typical Cross Section
Figure 8-8 Planar Truss Structure with a Snow Loading
The snow load on the truss is 500 N/m acting in the basic Y-direction. The input file
containing the grid points and PLOAD1 entries is shown in Listing 8-6.
Listing 8-6 Truss with Snow Loading
$
$ FILENAME - PLOAD1.DAT
$
$ UNITS ARE MM AND N
$
GRID 1 0.0 0.0 0.0 345
GRID 6 13000. 0.0 0.0 345

GRID 8 3250. 1250. 0.0 345
GRID 9 6500. 2500. 0.0 345
GRID 11 9750. 1250. 0.0 345
$
CBAR 5 1 1 8 1.0 1. 0.0
CBAR 6 1 8 9 1.0 1. 0.0
CBAR 7 1 9 11 1.0 1. 0.0
CBAR 8 1 11 6 1.0 1. 0.0
$
PLOAD1 1 5 FY FRPR 0.0 -.5 1. -.5
PLOAD1 1 6 FY FRPR 0.0 -.5 1. -.5
PLOAD1 1 7 FY FRPR 0.0 -.5 1. -.5
PLOAD1 1 8 FY FRPR 0.0 -.5 1. -.5
To see the effect of using projected forces, the resulting SPC forces are shown in
Figure 8-9. The total reaction of the loads obtained by adding the SPC forces is 6500
N in the Y-direction. The total reaction load is equal to the projected length of 13
meters times the distributed load of 500 N/m. If the projected option had not been
used, the applied load would have been 6964.2 N, which is the total length of the top
members times the distributed load.
F O R C E S O F S I N G L E - P O I N T C O N S T R A I N T

1 G -2.511155E-05 3.250000E+03 .0 .0 .0 .0
6 G .0 3.250000E+03 .0 .0 .0 .0
Figure 8-9 SPC Forces Due to the Snow Loading
Note: With PLOAD1, if you’re using a concentrated load and don’t need the
additional features that the CBEAM element offers, you should use the CBAR
element because of the way the equivalent end loads are generated.
• For the CBAR element, the equivalent end loads are generated explicitly.
• For the CBEAM element, the end loads are generated by numerical
integration along the length of the element, which may not be as accurate as
the method used for the CBAR element.
A linearly varying load applied with a PLOAD1 entry is accurate when used with
either element.
CHAPTER 8 167
Applied Loads
PLOAD2
You can use the PLOAD2 entry to apply a normal uniform pressure load to
CQUAD4, CSHEAR, or CTRIA3 2-D (surface) elements using element IDs. The
positive direction of the loading is determined by the order of the grid points on the
element connection entry, using the right hand rule. The software computes the
magnitude and direction of the load from the value of the pressure and the
coordinates of the connected grid points. NX Nastran applies the load to the
connected grid points.
The PLOAD2 entry has two forms:
1 2 3 4 5 6 7 8 9 10
PLOAD2 SID P EID1 EID2 EID3 EID4 EID5 EID6
Alternate form:
PLOAD2 SID P EID1 “THRU” EID2
Field Contents
P Pressure value. (Real)
EIDi Element identification number. (Integer > 0 or blank; for the
“THRU” option, EID1 < EID2)
The direction of the pressure is determined using the connected GRID points in the
same right-hand rule sense as the PLOAD entry (i.e., with respect to the positive
element z axis). In addition, you use the “THRU” option, all referenced elements
must actually exist.
The PLOAD2 entry is similar to the PLOAD entry except that the PLOAD2
references the element ID instead of the grid points where the element is attached.
The PLOAD2 is usually preferred over the PLOAD entry because it is easier to use.
The PLOAD2 entry, similar to the PLOAD entry, is limited to pressure acting
normal to the element surface. The direction of the pressure load is determined by
the numbering sequence of the grid points on the connectivity entry (CQUAD4,
CTRIA3, etc.).
See Also
• “PLOAD2” on page 1584 of the NX Nastran Quick Reference Guide
PLOAD4
The PLOAD4 entry lets you create the most general pressure definition. You can use
PLOAD4 to apply pressures and/or tractions to any of the two-dimensional elements
and the surfaces of the three-dimensional elements. With PLOAD4, you can create a
pressure load that’s either normal to the surface or that contains a traction (not
normal to the surface) component. In addition, you can define a different value of
pressure at each corner.
You can use PLOAD4 with CQUAD4, CQUAD8, CQUADR, CTRIA3, CTRIA6, and
CTRIAR, CHEXA, CPENTA, and CTETRA elements. Since the surface hyperelastic
elements CQUAD4, CQUAD8, CQUAD, CTRIA3, and CTRIA6 are plane strain
elements, you can’t apply pressure loads on them.
The software automatically computes the magnitude and direction of the equivalent
grid point forces using the isoparametric shape functions of the element to which the
load has been applied.
Load intensities P1, P2, P3, (and P4) act at corner points G1, G2, G3, (and G4) for
triangular (and quadrilateral) elements. The default direction of positive pressure for
faces of solid elements is inward.
You can set the direction of the pressure can be set by one of two methods.
• By default, the software uses the element normal to determine the direction
of the pressure. For two-dimensional elements, the direction of positive
pressure is in the direction of the outward normal as is determined by
applying the right-hand rule to the ordering sequence of the grid points on
the connectivity entry. For surfaces of solid elements, the direction of
positive pressure is inward toward the center of the element. The face of the
solid to which you are applying the pressure is determined by specifying
the appropriate corner grid points.
• Alternatively, you can input the direction of the pressure defined by the
PLOAD4 entry using an optional coordinate system and a vector entered on
the continuation line. Using a local coordinate system, you can define a
pressure acting at any angle to the surface.You can also apply loads acting
parallel to the surface (tractions).
Another feature unique to the PLOAD4 entry is the ability to apply a nonuniform
pressure. You can enter the pressure at each of the corner grid points to create a
linearly varying pressure load. If you enter the pressure for the first grid point only,
the software assumes that the pressure is constant over the element.
CHAPTER 8 169
Applied Loads
The format of the PLOAD4 entry is:
1 2 3 4 5 6 7 8 9 10
PLOAD4 SID EID P1 P2 P3 P4 G1 G3 or
G4
CID N1 N2 N3
The alternate format is
1 2 3 4 5 6 7 8 9 10
PLOAD4 SID EID1 P1 P2 P3 P4 “THRU” EID2
CID N1 N2 N3
Field Contents
EID
EID1 Element identification number.
EID2
P1, P2, P3, P4 Load per unit surface area (pressure) at the corners of the face
of the element.
G1 Identification number of a grid point connected to a corner of
the face. Required data for solid elements only.
G3 Identification number of a grid point connected to a corner
diagonally opposite to G1 on the same face of a CHEXA or
CPENTA element. Required data for the quadrilateral faces
of CHEXA and CPENTA elements only. G3 must be omitted
for a triangular surface on a CPENTA element.
G4 Identification number of the CTETRA grid point located at the
corner; this grid point may not reside on the face being
loaded. This is required data and is used for CTETRA
elements only.
N1, N2, N3 Components of the vector measured in coordinate system
defined by CID. Used to define the direction (but not the
magnitude) of the load intensity.
If P2, P3, and P4 are blank fields, the load intensity is uniform and equal to P1; P4 is
left blank for a triangular face. In addition, for pressure that acts normal to the face,
the continuation entry is not used.
See Also
• “PLOAD4” on page 1586 of the NX Nastran Quick Reference Guide
PLOAD4 Example 1
Consider the curved plate shown in Figure 8-10. PLOAD4 entries are to be used to
apply a normal pressure to each of the six CQUAD4 elements.
N
P = .5 ----------
2
cm
10 11 12
5 6
7 9
3 8 4
4 5
6
1 2
1 2 3
Figure 8-10 Pressure Loads on the CQUAD4s
This input file is given in Listing 8-7. When using the PLOAD4 entry, you may
specify a PLOAD4 entry for every element or use the alternate method of specifying
several elements that have the same pressure. The alternate method is used for this
example.
CHAPTER 8 171
Applied Loads
Listing 8-7 CQUAD4 Elements with Pressure Loads
$ FILENAME - PRESS.DAT
$ UNITS ARE CM, N
PLOAD4 3 1 -.5 THRU 6
$
GRID 1 0.0 -60. -2.
GRID 2 60. -60. -2.
GRID 3 120. -60. -2.
GRID 4 0.0 -30. 0.0
GRID 5 60. -30. 0.0
GRID 6 120. -30. 0.0
GRID 7 0.0 30. 0.0
GRID 8 60. 30. 0.0
GRID 9 120. 30. 0.0
GRID 10 0.0 60. -2.
GRID 11 60. 60. -2.
GRID 12 120. 60. -2.
$
CQUAD4 1 1 1 2 5 4
CQUAD4 2 1 2 5 6 3
CQUAD4 3 1 4 5 8 7
CQUAD4 4 1 5 6 9 8
CQUAD4 5 1 7 8 11 10
CQUAD4 6 1 8 9 12 11
$
SPC1 1 123456 1 4 7 10
$
PSHELL 1 4 .5 4
MAT1 4 7.1E6 .3
The resulting deflection is shown in Figure 8-11. The deflected shape is not what
you would expect. Element 2 appears to bend up. The problem occurs because the
normal for element 2 is reversed (see the ordering of the grid point IDs on the
CQUAD4s). The pressure on this element is acting upward. This type of mistake is
quite common.
Figure 8-11 Displacements Due to Pressure Load
PLOAD4 Example 2
Specify the PLOAD4 entry for a uniform normal pressure load applied to the
CHEXA solid element shown in the following figure:
2
P = 26.4 ( lb f ⁄ in )
5 8
6 7
Element
100 1 4
2 3
1 2 3 4 5 6 7 8 9 10
PLOAD4 SID EID P1 P2 P3 P4 G1 G3 or G4
CID N1 N2 N3
PLOAD4 12 100 26.4 5 7
This load is selected in the Case Control Section with the command LOAD = 12.
Leaving P2, P3, and P4 blank assigns a uniform pressure value of 26.4 lb/in2 .
PRESAX and PLOADX1

• You can use the PRESAX entry to define pressure loading on the conical
shell element.
• You can use the PLOADX1 entry to define pressure loading on the
CTRIAX6 element and on the hyperelastic CTRIAX and CQUADX
elements.
See Also
• “PRESAX” on page 1603 of the NX Nastran Quick Reference Guide

• “PLOADX1” on page 1590 of the NX Nastran Quick Reference Guide
8.6 Inertia Loads (Acceleration Loads)
This section describes how to apply “static” acceleration loads to your model, such
as gravity (the response of a structure to its own weight) and vehicle maneuver loads.
The acceleration causes a static load and not a dynamic one - the structural response
is steady state (static) with the transient part being zero. With an acceleration load,
you must specify both the magnitude of the acceleration and the direction in which
it acts.
CHAPTER 8 173
Applied Loads
Using GRAV
You can use the GRAV entry to define the direction and magnitude of a gravity
vector in any user-defined coordinate system. The software multiplies components
of the gravity vector by the mass matrix to obtain the components of the gravity
force at each grid point. Since the mass matrix is used to compute the forces, you
must have mass in your model. Typically, you would define that mass by density
you specify on a material bulk data entry. The software doesn’t calculate the
gravitational acceleration =at scalar points. You must introduce gravity loads at
scalar points directly.
Gravity loading is one the best loadings to use when you’re performing model
checkout because the software applies the force at every point in the model that has
mass. If you have modeling errors where elements or grids points aren’t attached to
the structure, their displacement is unbounded, and the job fails. If this occurs, the
software prints a diagnostic message indicating the grid point and component,
where the singularity is detected.
The format of the GRAV Bulk Data entry is:
1 2 3 4 5 6 7 8 9 10
GRAV SID CID A N1 N2 N3 MB
Field Contents
SID Set identification number. (Integer > 0)
CID Coordinate system identification number. ( Integer ≥ 0 ;
Default = 0)
A Acceleration vector scale factor. (Real)
Ni Acceleration vector components measured in coordinate
system CID. (Real; at least one Ni ≠ 0.0 )
MB Used only in superelement analysis.
The direction and magnitude of acceleration are given by a = A N where the vector
N = (N1, N2, N3) gives the direction. The magnitude of a is equal to A times the
magnitude of N . Note, for example, that entering the value of A in in/sec 2 indicates
that other specifications of the model that involve length units (element length,
moments of inertia, modulus of elasticity, etc.) must also be in inches to preserve
unit consistency.
For models that have only GRAV loads, the load set ID (SID) is selected in the Case
Control Section with a LOAD = SID Case Control command. For models that also
contain other types of loads, such as FORCE or PLOAD entries, you must use the
LOAD entry to combine them with the GRAV entry.
See Also
• “Combining Loads” on page 202 of the NX Nastran User’s Guide

Finally, the model must include mass density information to use acceleration loads.
You enter mass density on a material property bulk data entry. For example, for a
model of typical structural steel using English units, the mass density might be
–4 2 4
ρ m = 7.0 × 10 lb f . sec /in
Thus, the MAT1 Bulk Data entry might look like the following:
1 2 3 4 5 6 7 8 9 10
MAT1 MID E G NU RHO A TREF GE
MAT1 12 30.E6 0.3 7.0E-4
GRAV Example 1
This example shows how to calculate the tip deflection of the cantilever beam due to
its own weight.
X
Z
uy = ?
First, notice that the force of gravity (g) acts in the -Y direction (the direction of the
center of the earth). Thus, the vector N can be written as:
N = ( 0., –1. , 0. ) Eq. 8-3
The acceleration due to gravity on the Earth’s surface is approximately 32.2 ft/sec 2
( 386.4 in/sec2 ) or 9.8 m/sec 2 . Units must be consistent throughout the model (this
example uses inches).
CHAPTER 8 175
Applied Loads
Thus, the GRAV entry is as follows:
1 2 3 4 5 6 7 8 9 10
GRAV 15 386.4 0. –1. 0.
GRAV,15,,386.4,0.,-1.,0.
The Case Control command required to apply this load is LOAD = 15.
GRAV Example 2
Next, we want to calculate the tip deflection of the cantilever beam due to its own
weight and a concentrated force. Assume that a concentrated force is added to the
beam as shown:
F
Y
X
g
Z
uy = ?
To combine gravity loading with the concentrated force, the following approach
must be used:
In Case Control: LOAD = 15

In Bulk Data: LOAD 15 1.0 1.0 12 1.0 13
and
GRAV 12
FORCE 13
Importantly, the set ID on a GRAV entry can’t be the same as the Set ID on any other
load entry.
GRAV Example 3
As a final example, suppose you are analyzing an instrument package subjected to
inertial loads in specific directions, say:
2 2
1.3 g in the x-direction ( 1.3g = 1.3 ⋅ 9.8 m/sec = 12.7 m/sec ).
2 2
2.8 g in the y-direction ( 2.8g = 2.8 ⋅ 9.8 m/sec = 27.4 m/sec ).
2 2
0.3 g in the z-direction ( 0.3g = 0.3 ⋅ 9.8 m/sec = 2.9 m/sec ).
Three separate GRAV entries can be written as:
1 2 3 4 5 6 7 8 9 10
GRAV SID CID A N1 N2 N3 MB
GRAV 15 12.7 1. 0. 0.
GRAV 15 27.4 0. 1. 0.
GRAV 15 2.9 0. 0. 1.
Using RFORCE
You can use the RFORCE entry to define a static loading condition due to angular
velocity and/or acceleration. These loads are specified by the designation of a grid
point that lies on the axis of rotation and by the components of rotational velocity or
angular acceleration in any defined coordinate system.
There are two methods to compute these loads.
• In the first method, the mass matrix is regarded as pertaining to a set of

distinct rigid bodies connected to grid points. Deviations from this
viewpoint, such as the use of scalar points or the use of mass coupling
between grid points, can result in errors.
• The second method allows mass coupling, but does not allow offsets. The
second method is the only method used for angular acceleration.
On the RFORCE entry, you input the components of a spin vector that are used
internally to compute centrifugal forces. Each component of the spin vector is
multiplied by the same scale factor. The format of the RFORCE entry is as follows:
1 2 3 4 5 6 7 8 9 10
RFORCE SID G CID A R1 R2 R3 METHOD
RACC
Field Contents
CHAPTER 8 177
Applied Loads
G Grid point identification number through which the rotation

vector acts.
CID Cartesian coordinate system defining the components of the
rotation vector.
A Scale factor of the angular velocity in revolutions per unit time.
R1, R2, R3 Rectangular components of rotation vector R . The vector

defined passes through point G.
METHOD Method used to compute centrifugal forces due to angular
velocity.
RACC Scale factor of the angular acceleration in revolutions per unit
time squared.
zCID
zbasic
R
yCID
G
R1 R3
ra R2
ri Gi F
xCID
ybasic
xbasic
Figure 8-12 RFORCE Vector at Grid Point Gi
{ F } i = [ m ]i [ ω X ( ω X ( ri – ra ) ) + α X ( ri – ra ) ]
where:
angular velocity = ω = 2πA ⋅ R (radians/unit time)
angular acceleration = a = 2πRACC ⋅ R (radians/unit time squared)
[ m ] i = 3 × 3 translational mass matrix at grid point Gi
Note: The equation for F i has additional terms if the mass is offset and METHOD
= 1 is selected.)
You must select one of two methods for the internal calculation of the loading vector
(field 9).
• Method=1 yields correct results only when there is no coupling in the mass
matrix. This occurs when the lumped mass option is used with or without
the ZOFFS option (see the CQUAD4 entry for a description of ZOFFS).
• Method=2 yields correct results for lumped or consistent mass matrix only
if the ZOFFS option is not used. The acceleration terms due to the mass
offset (X1, X2, X3) on the CONM2 entry are not computed with method=2.
All possible combinations of mass matrices and offset and the correct method to be
used are shown in Table 8-1.
Table 8-1 Restrictions When Using the RFORCE Entry
No Offset Offset
Lumped Method=1 or Method=1
Method=2
Coupled Method=2 Neither
In addition, for problems with elements that have edge grid points (CQUAD8,
CTRIA6, CTRlAX6, CHEXA, CPENTA, and CTETRA), the software only produces
correct centrifugal loads if you include the parameter PARAM,COUPMASS,x (where
x is greater than 1) in the input file and you select Method 2.
See Also
• “RFORCE” on page 1673 of the NX Nastran Quick Reference Guide
• “COUPMASS” on page 578 of the NX Nastran Quick Reference Guide
Defining Mass in Your Model

There are several different ways that you can define mass in your model in NX
Nastran:
• the concentrated mass elements and scalar points (CMASSi and CONMi)
• the mass density option on the material entries (MAT1, MAT2, etc.)
• nonstructural mass defined on the property entries
You define mass density on the material entries in terms of mass/unit volume. The
mass unit must be consistent with the other units in the model. For example, in the
English system (in, lb, sec), the mass density of steel is approximately
–4 2 4
7.32 ⋅ 10 lb-sec ⁄ in .
CHAPTER 8 179
Applied Loads
The software adds the nonstructural mass you define on the property entries to the
structure in addition to the structural mass from the elements.
• For one-dimensional elements, the units are mass/unit length.
• For two-dimensional and three-dimensional elements, the units are
mass/unit area and mass/unit volume, respectively.
Using the WTMASS Parameter to Express Mass in Weight Units

In some cases, you may want to use the WTMASS parameter to express the mass in
terms of weight units instead of mass units. The WTMASS parameter multiplies the
assembled mass matrix by the scale factor you specify with the WTMASS
parameter.
For the steel example, you can enter the mass density a weight density of 0.283 lb ⁄ in 3
with a WTMASS parameter of 0.00259 (which is 1/386.4). However, if you enter
any of the mass in terms of weight, you must enter all the mass in terms of weight.
The WTMASS multiplies all the mass in the model by the same scale factor, with the
only exception being the mass you enter in Direct Matrix Input.
See Also
• “WTMASS” on page 661 of the NX Nastran Quick Reference Guide
WTMASS Example
For example, consider the two rotating masses shown in Figure 8-13. The masses
2
are accelerated at constant angular acceleration of 20 rev/sec from 0 to 200
rev/min. The goal is to determine the axial force in the CBAR as a function of
angular velocity.
z
Constant Acceleration
from 0 to 200 rev/min
MASS = 0.1 MASS = 0.1

x
3 2
10 in 10 in
Figure 8-13 Rotating Concentrated Masses
The input file is shown in Listing 8-8. Multiple load cases are used to determine
axial forces at six rotational speeds. Using multiple load cases for this problem is
much more efficient than submitting separate runs because NX Nastran doesn’t
have to regenerate the stiffness and mass matrices. Another interesting note is that
the model doesn’t actually rotate; it is fixed at the center. If the constraints permitted
the model to rotate, the analysis would fail.
Listing 8-8 Axial Force in the Rotating Arms
$
$ FILENAME ROTATE.DAT
$
ID LINEAR,ROTATE
SOL 101
TIME 2
CEND
$
FORCE = ALL
SUBCASE 1
LABEL = 0 REV/min
LOAD = 1
SUBCASE 2
LABEL = 40 REV/min
LOAD = 2
SUBCASE 3
LABEL = 80 REV/min
LOAD = 3
SUBCASE 4
LABEL = 120 REV/MIN
LOAD = 4
SUBCASE 5
LABEL = 160 REV/MIN
LOAD = 5
SUBCASE 6
LABEL = 200 REV/MIN
LOAD = 6
BEGIN BULK
PARAM GRDPNT 0
CHAPTER 8 181
Applied Loads
Listing 8-8 Axial Force in the Rotating Arms (continued)
$
GRID 1 0.0 0.0 0.0 123456
GRID 2 10. 0.0 0.0
GRID 3 -10. 0.0 0.0
$
CBAR 10 1 1 2 1.0 0.0 1.0
CBAR 11 1 1 3 1.0 0.0 1.0
$
PBAR 1 1 1.0 1.0 1.0 1.0
$
MAT1 1 30.E6 .3
$
RFORCE 1 1 0. 0.0 0.0 1.0 1
20.
RFORCE 2 1 .666667 0.0 0.0 1.0 1
20.
RFORCE 3 1 1.33333 0.0 0.0 1.0 1
20.
RFORCE 4 1 2.0 0.0 0.0 1.0 1
20.
RFORCE 5 1 2.66667 0.0 0.0 1.0 1
20.
RFORCE 6 1 3.33333 0.0 0.0 1.0 1
20.
$
CONM2 101 2 .1
CONM2 102 3 .1
$
ENDDATA
Figure 8-14 shows a plot summarizing the axial force. The plot isn’t linear because
centrifugal forces aren’t linear with angular velocity. Consequently, using a
SUBCOM command to combine the subcases with RFORCE entries can lead to
misleading results. In other words, the element axial forces for Subcase 3 can’t be
obtained by merely doubling the element axial forces of Subcase 2 by using a
SUBCOM/ SUBSEQ combination. This occurs because the axial force is a function
2
of ω , and not 2ω.
500
400
Axial Force (lb)

300
200
100
0
0 40 80 120 160 200
Rotational Speed (rad/min)
Figure 8-14 Axial Force in the Bar as a Function of Angular Velocity
8.7 Thermal Loads

You can define thermal loads on a structure to perform stress analysis or to determine
thermal expansion. In NX Nastran, you must define a temperature distribution via
TEMPij Bulk Data entries and thermal expansion coefficients.
• Thermal expansion coefficients are specified on the material Bulk Data

entries.
• Temperatures can be specified at grid points (TEMP and TEMPD entries)
and interpolated to grid points within elements.
You can also specify temperature data on an element-by-element basis as shown in

Table 8-2.
Table 8-2 Bulk Data Entries Used for Temperature Definition on Elements
Elements Temperature Data Bulk Data

Entry
CROD, CONROD, Average temperature at ends A and B. TEMPRB
CTUBE
CBAR, CBEAM, Average temperature and cross-sectional TEMPRB
CBEND temperature gradients at ends A and B.
CQUAD4, CTRIA3 Average temperature and gradient in the TEMPP1
CQUAD8, CTRIA6 thickness direction.
Any average temperatures you specify directly for an element take precedence over
the temperatures that the software interpolates from the element’s connected grid
points. Solid elements obtain their temperatures only by interpolation from
CHAPTER 8 183
Applied Loads
connected grid points. Note that interpolated grid point temperatures provide
temperature gradients over the neutral surface of shell elements, whereas the
TEMPPi entries do not.
NX Nastran uses the temperature data and thermal expansion coefficients to

calculate equivalent forces and moments acting at the grid points.
The TEMP(INIT) and TEMP(LOAD) Case Control commands specify the initial
temperature and applied temperature, respectively. The TEMP(INIT) command
must appear either above the first subcase or inside the first subcase. As with other
types of loads, NX Nastran only applies a thermal load if you specifically request it
in the case control section.
Defining Temperatures for Grid Points

You can use the TEMP and TEMPD bulk data entries to define grid point
temperatures.
• TEMP lets you define a scalar value of temperature at selected grid points.
• TEMPD lets you define a default temperature for all grid points whose
temperatures aren’t defined by a TEMP entry.
See Also
• “TEMP” on page 1815 of the NX Nastran Quick Reference Guide

• “TEMPD” on page 1820 of the NX Nastran Quick Reference Guide
Defining Temperatures for Elements

You can define temperatures for selected elements.
• You can specify temperatures for CROD, CBAR, CBEAM, CBEND,

CONROD, or TUBE elements on the TEMPRB bulk data entry. TEMPRB
specifies the average temperature on both ends. With CBAR, CBEAM and
CBEND elements, you can use TEMPRB to define temperature gradients
over the cross section.
• You can specify temperatures for two-dimensional plate and membrane
elements on a TEMPP1, TEMPP2, or TEMPP3 bulk data entry. The user-
defined average temperature over the volume is used to produce in-plane
loads and stresses. You can use thermal gradients over the depth of the
bending elements, or the resulting moments to produce bending loads and
stresses.
If you don’t define any thermal element data for an element, the software averages
temperatures of the connected grid points from the TEMP, TEMPD, or TEMPAX
entries to calculate an average temperature for the element. The thermal expansion
coefficients are defined on the material definition entries. Regardless of the type of
thermal data, if the material coefficients for an element are temperature dependent
by use of the MATTi entry, the software always calculates them from the “average”
temperature of the element.
Using Temperature Loads with Subcases

If temperature loads are present in one or more of the subcases in your input file, the
software must take thermal strain into account in the calculation of element stresses
and forces. That is, a body subjected to a temperature field will deform in a stress free
manner in the absence of constraints. However, if the deformation is resisted through
boundary conditions and/or the presence of applied loads, the total actual strain, at
a point must be reduced by the thermal strain, α ( T – To ) , to obtain the stress at the
point. For example, the stress-strain relation for a rod that undergoes thermal
expansion and also sustains an axial force is written in the form
σ
ε – α ( T – T o ) = --- Eq. 8-4
E
NX Nastran calculates element stresses and forces on an element-by-element basis

from the displacement vector and the temperature field. Therefore, if you’re using
SUBCOM or SYMCOM to combine selected subcases, you must supply a definition
of the temperature field whenever element stresses and forces are required.
Consider the following example where a temperature load is specified in SUBCASE

2.
.
.
.
SUBCASE 1
LOAD=2
SUBCASE 2
TEMP(LOAD)=4
SUBCASE 10
LOAD=10
SUBCOM 20
SUBSEQ=.5,1.,1.6
TEMP(LOAD)=4
.
.
.
CHAPTER 8 185
Applied Loads
Here, the multiplier for SUBCASE 2 is unity so that the temperature distribution
required to calculate element forces and stresses for the combined loading case is
precisely TEMP(LOAD)=4. If, on the other hand, the SUBSEQ entry is modified as
indicated below, an additional TEMP(LOAD) must be defined.
.
.
.
SUBCASE 1
LOAD=1
SUBCASE 2
TEMP(LOAD)=4
SUBCASE 10
LOAD=10
SUBCOM 20
SUBSEQ=.5,.75,1.6
TEMP(LOAD)=40
Here, TEMP(LOAD)=40 must reflect a temperature distribution of the form
( T 40 – T o ) = .75 ( T 4 – T o ) Eq. 8-5
T 40 = T o + .75 ( T 4 – T o ) Eq. 8-6
You must supply this new TEMP(LOAD) through standard Bulk Data entries such
as TEMPD, TEMP, TEMPP1, and TEMPP3 and them call them out in the Case
Control section with the SUBCOM command.
In general, with n temperature loading conditions in the subcases immediately

preceding a SUBCOM, the temperature distribution that must be defined under the
SUBCOM if you want element force or stress output is of the form
n
T SUBCOM = T o + ∑ ai ( Ti – To ) Eq. 8-7
i = 1
where the coefficients a i are defined on the SUBSEQ entry under the SUBCOM in
question. The reference temperature T o , defined on MATi Bulk Data entries, can
conveniently be defined as zero in order to simplify the calculation of the
temperature distribution for the combined loading condition of the SUBCOM.
The same approach is used with combined loading condition defined through a
SYMCOM Case Control entry. However, the use of SYM subcases and SYMCOM
entries implies the use of the concepts of reflective symmetry, in which case you
should use the DIH option of cyclic symmetry. NX Nastran internally generates the
appropriate boundary conditions, and the you simply define the actual temperature
distribution on each portion of the structure without regard to decomposing the

temperatures and/or mechanical loads into symmetric and antisymmetric
components.
See Also
• “Understanding Cyclic Symmetry” on page 694 of the NX Nastran User’s

Guide
This is best illustrated with the example problem shown in Listing 8-9.
Listing 8-9 Using SUBCOMs with Temperature Loads
.
.
.
CEND
$
TEMP(INIT) = 1
$
SUBCASE 1
TEMP(LOAD) = 10
$
SUBCASE 2
TEMP(LOAD) = 15
$ THE FOLLOWING SUBCOM WILL COMBINE
$ 100% OF THE RESULTS FROM SUBCASE 1
$ WITH 50% OF THE RESULTS FROM SUBCASE 2
SUBCOM 3
SUBSEQ = 1.0,0.5
TEMP(LOAD) = 20
$
BEGIN BULK
$
TEMPD,1, 100.
TEMPD,10,150.
TEMPD,15,250.
$ "TEMPD,20" IS CALCULATED AS FOLLOW:
$ 100 + 1.0*(150-100) + 0.5*( 250 - 100 ) = 225.
$
TEMPD,20,225.
$ REST OF THE BULK DATA ENTRIES
ENDDATA
As an example of thermal loads, consider the umbrella tent shown in Figure 8-15.
The tent is subjected to thermal radiation from partial exposure to the sun. The
ground is held at 40 degrees Fahrenheit and the air temperature is held at 50 degrees
Fahrenheit. Can the tent withstand the thermal stresses imposed by solar radiation?
This problem involves conduction, convection, and radiation, which is beyond the
scope of this user’s guide, but the stress analysis of the resulting temperature profile
is not. For further discussion on the subject of heat transfer, refer to the NX Nastran
Thermal Analysis User’s Guide.
CHAPTER 8 187
Applied Loads
10 11 12
7 8 9
4 5
6
Shadow
z
1 3
2
y
Figure 8-15 Umbrella Tent With Thermal Loads
To analyze this type of problem, you need to run the heat transfer Solution 153 to
produce a temperature profile for all of the grid points in the model. A Solution 153
input file is generated for this model and is included in the Test Problem Library
with the name ‘heat.dat’.
In Solution 153, you have the option of creating a temperature file that can be
included directly into the Solution 101 input file. The Solution 101 input file used
for the stress analysis is shown in Listing 8-10. The temperature file from the
Solution 153 run is shown in Listing 8-11. Note the use of the INCLUDE statement,
which inserts the temperature input file at that location when you submit the job.
Listing 8-10 Umbrella Tent Thermal Analysis
$ FILENAME - THERMAL.DAT
$
ID LINEAR,THERMAL
SOL 101
TIME 10
CEND
TITLE = UMBRELLA TENT WITH THERMAL LOADING
$ DEFINE INITIAL TEMPERATURE
TEMP(INIT) = 2
$ DEFINE THERMAL LOADING
TEMP = 1
STRESS = ALL
SPC = 15
BEGIN BULK
$
INCLUDE ’heat.pch’
$
TEMPD 2 40.0
$
MAT1 1 10.E6 .3 2.59E-3 21.E-6
25000. 25000. 15000.
MAT1 4 2.E6 .3 2.59E-5 21.E-6
25000. 25000. 15000.
$
$ GRID, CBAR, CQUAD4, PBAR, PSHELL, AND SPC ENTRIES
$ REMOVED TO SAVE SPACE
ENDDATA
Listing 8-11 Temperature Distribution Generated with Solution 153
$
$FILENAME - HEAT.PCH
$
$TITLE = TENT THERMAL ANALYSIS SUBJECTED TO RADIATION 1
$SUBTITLE=THE BASE TEMPERATURE AT 40 DEGREE F 2
$LABEL = 3
$DISPLACEMENTS 4
$REAL OUTPUT 5
$SUBCASE ID = 1 6
TEMP* 1 1 4.011359E+01 7
TEMP* 1 2 5.999791E+01 8
TEMP* 1 3 1.194475E+02 9
TEMP* 1 4 4.001704E+01 10
TEMP* 1 5 4.705626E+01 11
TEMP* 1 6 6.816640E+01 12
TEMP* 1 7 4.000030E+01 13
TEMP* 1 8 4.004408E+01 14
TEMP* 1 9 6.814966E+01 15
TEMP* 1 10 4.005281E+01 16
TEMP* 1 11 4.012182E+01 17
TEMP* 1 12 1.193867E+02 18
TEMP* 1 99 4.000000E+01 19
TEMP* 1 101 7.500000E+01 20
TEMP* 1 103 7.500000E+01 21
TEMP* 1 110 7.500000E+01 22
TEMP* 1 112 7.500000E+01 23
TEMP* 1 999 0.000000E+00 24
CHAPTER 8 189
Applied Loads
The OLOAD output shows the applied loads generated by the imposed
temperature. The thermal stresses for this example are quite small, so the structure
should have no difficulty surviving the thermal loading.
8.8 Enforced Element Deformation Loads

NX Nastran includes a limited capability to define enforced axial deformations. You
can use the DEFORM Bulk Data entry to apply an enforced extension (units of
length) to CBAR, CBEAM, CONROD, CROD, and CTUBE elements.
Enforced deformations, for example, are useful in the simulation of misfit or

misalignment in engineering structures. As in the case of thermal expansion, the
software calculates the equivalent loads by separate subroutines for each type of
structural element and are applied to the connected grid points. The magnitude of
the axial deformation is specified on a DEFORM entry.
You can only use DEFORM in SOLs 101, 105, 114, and 200. To apply enforced
displacements in other solution sequences, use SPCD.
See Also
• “DEFORM” on page 1103 of the NX Nastran Quick Reference Guide
• “SPCD” on page 1755 of the NX Nastran Quick Reference Guide
With DEFORM, the enforced deformation is analogous to a turnbuckle that can be
adjusted to provide a desired axial deformation in an element. The appropriate
stress-strain relation for a rod element of length that undergoes an enforced axial
deformation, δ is given by the following equation:
δ σ
ε – -- = --- Eq. 8-8
l E
Consider this example:.
.
.
.
SUBCASE 1
LOAD=1
SUBCASE 2
DEFORM=5
SUBCASE 10
LOAD=10
SUBCOM 20
SUBSEQ=.5,.75,1.6
DEFORM=50
.
.
.
Here DEFORM = 50 reflects a set of enforced deformations of the form
δ 50 = .75δ 5 Eq. 8-9
For n DEFORM loading conditions in the subcases that immediately precede a

SUBCOM, the DEFORM loading condition that must be defined under the SUBCOM,
if element force and stress output is desired, is of the form
n
δ SUBCOM = ∑ ai δi Eq. 8-10
i = 1
The format of the DEFORM entry is shown below:
1 2 3 4 5 6 7 8 9 10
DEFORM SID EID1 D1 EID2 D2 EID3 D3
Field Contents
SID Deformation set identification number. (Integer > 0)
EIDi Element number. (Integer > 0)
Di Deformation. (Real; positive value represents elongation;
negative value represents contraction)
You must use the DEFORM=SID to select the DEFORM entry in the Case Control
section.
When you use DEFORM, you must remember that you’re not enforcing a strain or an
actual extensional length to the element. Rather, you’re applying a force to the
element that produces the specified extension if the element is free to expand without
internal forces being generated. The software adds this computed force to the other
forces in the model. Since most elements in your model aren’t free to expand, the
extension value you specify may not be achieved. This feature is best illustrated with
an example.
DEFORM Example
Suppose you want to use the DEFORM entry to enforce a strain on the three-member
structure shown in Figure 8-16. The goal is to impose an initial strain equal to 100
µcm/cm in the center member.
CHAPTER 8 191
Applied Loads
6 N
E = 7.1 × 10 ----------
2
cm
1 2 3 4
10 11 12
10 cm 10 cm 10 cm
Figure 8-16 Three-Bar Structure with Initial Strain
Consider: (1) the DEFORM entry can only impose a force, (2) from static
equilibrium, if no forces are imposed on grid points 2 and 3, the force in all three
members must be the same. With this in mind, the following procedure is used.
1. Determine the force required in member 11 that produces the 100 mcm/cm
strain.
σ 11 = Eε
6 –6 N
= ( 7.1 × 10 ) ( 100 × 10 ) = 710 ----------
2
cm
P = Aσ 11
= ( 1.5 ) ( 710 ) = 1065 N
2. Since there are no external forces acting on the elements, the force of 1065
N is same in all the members. Using the 1065 N force, the stresses
computed in the other members.
P 1065 N
σ 10 = --------- = ------------ = 1065 ----------
A 10 1 cm
2
P 1065 N
σ 12 = --------- = ------------ = 532.5 ----------
A 12 2 cm
2
3. From the stresses found in part 2, the element strains are computed for
elements 10 and 12. Multiplying the strain within each element by its
length yields the extensional change of the element. These extensional
changes are the deform values required to produce a 100 µcm/cm strain in
element 11.
σ 10 –6
ε 10 = -------- = 150 ⋅ 10
E
σ 12 –6
ε 12 = -------- = 75.0 ⋅ 10
E
Therefore,
–6 –3
∆L 10 = L 10 ε 10 = ( 10 ) ( 150 ⋅ 10 ) = 1.5 ⋅ 10 cm
–6 –3
∆L 11 = L 11 ε 11 = ( 10 ) ( 150 ⋅ 10 ) = 1.0 ⋅ 10 cm
–6 –3
∆L 12 = L 12 ε 12 = ( 10 ) ( 75 ⋅ 10 ) = 0.75 ⋅ 10 cm
These values are entered on the DEFORM Bulk Data entry as shown in Listing 8-12.
To help understand the implementation of the DEFORM entry, two configurations
are being analyzed. The first configuration (Subcase 1) specifies the boundary
conditions as shown in Figure 8-17(a). For this case, the structure is free to expand
without inducing any element forces. The results of this subcase should be the
extension of the members as entered on the DEFORM entry. As mentioned
previously, a force is applied to the element that produces the specified extension if
the element is free to expand.
The second subcase, shown in Figure 8-17(b), is the constrained configuration that is
of interest. If the calculations are correct, the strain in the center element should be
the required 100 µcm/cm.
2 2 2
A = 1 cm A = 1.5 cm A = 2 cm
(a)
10 11 12
10 cm 10 cm 10 cm
(b)
Figure 8-17 Three-Bar Structures with Initial Strain
CHAPTER 8 193
Applied Loads
Listing 8-12 Initial Strain Input File
$ FILENAME - STRAIN.DAT
ID LINEAR,STRAIN
SOL 101
TIME 2
CEND
$
DISP = ALL
STRESS = ALL
DEFORM = 1
SUBCASE 1
SUBCASE 2
SPC = 1
BEGIN BULK
$
SPC 1 4 1
$
GRID 1 0.0 0.0 0.0 123456
GRID 2 10. 0.0 0.0 23456
GRID 3 20. 0.0 0.0 23456
GRID 4 30. 0.0 0.0 23456
$
CBAR 10 1 1 2 1.0 0.0 1.0
CBAR 11 2 2 3 1.0 0.0 1.0
CBAR 12 3 3 4 1.0 0.0 1.0
$
PBAR 1 1 1.0
PBAR 2 1 1.5
PBAR 3 1 2.0
$
MAT1 1 7.1E6 .3
$
DEFORM 1 10 1.5E-3 11 1.E-3 12 .75E-3
ENDDATA
The stresses and displacements for the two subcases are shown in Figure 8-18. As
can be seen for the free end configuration, the extension of all of the members is
consistent with the DEFORM entry.
Note that the extensions for the elements aren’t the same as the values specified on
2 6
the DEFORM entry. The stress in member 11 is 710 N/cm . Since E = 7.1 x 10 , the
strain in element 11 is equal to ε = E/σ = 100 µcm/cm strain. The stress output is
requested for this example because strain output is not available for
one-dimensional elements.
1 G .0 .0 .0 .0 .0 .0
2 G 1.500000E-03 .0 .0 .0 .0 .0
3 G 2.500000E-03 .0 .0 .0 .0 .0
4 G 3.250000E-03 .0 .0 .0 .0 .0
SUBCASE 2
1 G .0 .0 .0 .0 .0 .0
2 G -1.791005E-11 .0 .0 .0 .0 .0
3 G 8.955025E-12 .0 .0 .0 .0 .0
4 G .0 .0 .0 .0 .0 .0
SUBCASE 1
S T R E S S E S I N B A R E L E M E N T S ( C B A R )
ELEMENT SA1 SA2 SA3 SA4 AXIAL SA-MAX SA-MIN M.S.-T
ID. SB1 SB2 SB3 SB4 STRESS SB-MAX SB-MIN M.S.-C
10 .0 .0 .0 .0 .0 .0 .0
.0 .0 .0 .0 .0 .0
11 .0 .0 .0 .0 .0 .0 .0
.0 .0 .0 .0 .0 .0
12 .0 .0 .0 .0 .0 .0 .0
.0 .0 .0 .0 .0 .0
SUBCASE 2
S T R E S S E S I N B A R E L E M E N T S ( C B A R )
ELEMENT SA1 SA2 SA3 SA4 AXIAL SA-MAX SA-MIN M.S.-T
ID. SB1 SB2 SB3 SB4 STRESS SB-MAX SB-MIN M.S.-C
10 .0 .0 .0 .0 -1.065000E+03 -1.065000E+03 -1.065000E+03
.0 .0 .0 .0 -1.065000E+03 -1.065000E+03
11 .0 .0 .0 .0 -7.100000E+02 -7.100000E+02 -7.100000E+02
.0 .0 .0 .0 -7.100000E+02 -7.100000E+02
12 .0 .0 .0 .0 -5.325000E+02 -5.325000E+02 -5.325000E+02
.0 .0 .0 .0 -5.325000E+02 -5.325000E+02
Figure 8-18 Stress and Displacement Output of the Three Bar Structure
8.9 Loads Due to Enforced Motion

Statics
You can specify enforced zero displacements on GRID, SPC, or SPC1 entries.
• If you use GRID entries, the constraints become part of the structural model
and modifications cannot be made at the subcase level.
• SPC or SPC1 entries are usually used for defining zero displacements,
which result in nonzero forces of constraint.
You can specify nonzero enforced displacements on SPC or SPCD entries.
• The SPC entry lets you specify both the component to be constrained and
the magnitude of the enforced displacement.
CHAPTER 8 195
Applied Loads
• The SPCD entry lets you specify only the magnitude of the enforced grid
point displacement. When you use an SPCD entry, you must specify the
component to be constrained on either an SPC or SPC1 entry. Using SPCD
avoids the decomposition of the stiffness matrix when changes are only
made in the magnitudes of the enforced displacements.
NX Nastran calculates the equivalent loads resulting from enforced displacements

of grid points and adds them to the other applied loads. You specify the magnitudes
of the enforced displacements on SPC entries (SPCAX in the case of conical shell
problems) in the global coordinate system. NX Nastran automatically applies the
load when you select the associated SPC set in the Case Control section.
See Also
• “SPCAX” on page 1754 of the NX Nastran Quick Reference Guide
Dynamics
In NX Nastran, you can use the:
• TLOAD1 and TLOAD2 entries to define a time-dependent dynamic load

or enforced motion for use in a transient response analysis
• RLOAD1 and RLOAD2 entries to define a frequency-dependent dynamic
load for use in a frequency response analysis
You can use the TYPE field (field5 on TLOAD1/TLOAD2 and field 8 on RLOAD1
and RLOAD2) to specify the type of dynamic excitation you’re creating.
• If the TYPE field on the TLOAD1/2 entries indicates an enforced motion,
the software first assumes that the EXCITEID field points to SPC-type data.
If not present, NX Nastran then assumes the excitation is enforced motion
using large masses and will then look for DAREA and various static
loading data, just as in the case of applied loads.
• With RLOAD1 and RLOAD2 entries, the software only look for SPC-type
data in the case of enforced motion.
For all entries (TLOAD1/2, RLOAD1/2), if the TYPE field indicates an applied
load, the program will search only for static loading data.
See Also
• “TLOAD1” on page 1835 of the NX Nastran Quick Reference Guide

• “TLOAD2” on page 1839 of the NX Nastran Quick Reference Guide
• “RLOAD1” on page 1681 of the NX Nastran Quick Reference Guide
• “RLOAD2” on page 1684 of the NX Nastran Quick Reference Guide
Specifying Enforced Motion Data

In NX Nastran, you specify enforced motion data on the SPC/SPC1/SPCD Bulk Data
entries. (Note that components specified in SPCD data must be also be referenced on
SPC or SPC1 entries.)
You can define enforced motion with SPC data (without the use of SPCDs) alone.
However, such usage is discouraged in favor of SPC/SPC1 data in connection with
SPCDs. In this way, you select SPCD entries via the EXCITEID field in the dynamic
load data, while the software uses the SPC/SPC1 sets to identify the constrained
displacement set in the Case Control section.
With SPC data alone, enforced motion is applied solely via the Case Control section.
The entries referenced by the EXCITEID field on the TLOAD1/2 or RLOAD1/2
entries do not have to exist. This method of specifying enforced motion is less elegant
and somewhat more difficult to interpret.
See Also
• “SPC1” on page 1751 of the NX Nastran Quick Reference Guide
Using Residual Vectors (Modal Augmentation) in Modal

Dynamic Response Analysis
When enforced motion is employed in modal dynamic response analysis, using
modal augmentation vectors is highly recommended in order to improve solution
accuracy. Without these vectors, your obtained results may be unacceptable.
Qualitatively, the enforced motion shapes, or “modes,” may not necessarily be

represented by the component modes of the structure, leading to a truncation of the
enforced motion shapes during projection to the modal space. To obtain modal
residual vectors from applied loads using modal augmentation vectors, specify
PARAM,RESVEC,YES in the Bulk Data section. NX Nastran automatically derives
the necessary static loads from the enforced motion data; you don’t need to specify
any additional loads.
See Also
• “RESVEC” on page 638 of the NX Nastran Quick Reference Guide
In summary, the procedure for applying enforced motion in a dynamic analysis:
CHAPTER 8 197
Applied Loads
1. Specify the appropriate type of enforced motion to be applied via the TYPE
field in RLOAD1/RLOAD2 (field 8) or TLOAD1/TLOAD2 (field 5) Bulk
Data entry, as appropriate.
2. Define the desired enforced motion using the SPCD Bulk Data entry. The
set IDs of these SPCD data must match the IDs appearing in the EXCITEID
fields of the corresponding dynamic load data in step 1.
3. Ensure that the components referenced in the SPCD Bulk Data entry above
are also specified in the SPC1 Bulk Data entry and this SPC1 is also selected
in the Case Control Section.
4. Ensure PARAM,RESVEC is set to YES in the Bulk Data Section if a modal
dynamic solution approach is used.
8.10 Frequency-Dependent Loads

You use the DLOAD entry to define linear combinations of frequency dependent
loads that are defined on RLOAD1 or RLOAD2 entries. The RLOAD1 entry defines
a frequency dependent load of the form
i ( θ – 2πfτ )
{ P ( f ) } = A [ C ( f ) + iD ( f ) ]e Eq. 8-11
where A is defined on a DAREA entry, C ( f ) and D ( f ) are defined on TABLEDi

entries, θ is defined on a DPHASE entry, τ is defined on a DELAY entry, and f is
defined on a FREQ, FREQ1 or FREQ2.entry. The RLOAD2 entry defines a
frequency-dependent load of the form
i { φ ( f ) + θ – 2πfτ }
{ P ( f ) } = AB ( f )e Eq. 8-12
where A is defined on a DAREA entry, B ( f ) and φ ( f ) are defined on TABLEDi

entries, θ is defined on a DPHASE entry, and τ is defined on a DELAY entry. The
coefficients on the DAREA, DELAY, and DPHASE entries may be different for each
loaded degree-of-freedom. The loads are applied to the specified components in the
global coordinate system.
Static loading data may be used to supply DAREA values. Using a LOADSET Case
Control request with LSEQ Bulk Data entries, you can build DAREA load vectors
from any valid static load set. This option is available in the frequency response
solutions and in superelement analysis.
The RANDPS entry defines load set power spectral density factors for use in
random analysis of the form
S jk ( f ) = ( X + iY )G ( f ) Eq. 8-13
where G ( f ) is defined on a TABRNDi entry. The subscripts j and k define the subcase
numbers of the load definitions. If the applied loads are independent, only the
diagonal terms ( j = k ) need be defined. You use the RANDT1 entry to specify the
time lag constants for use in the computation of the auto correlation functions.
See Also
• “Frequency Response Analysis” in the NX Nastran Basic Dynamic Analysis

User’s Guide (for a discussion of frequency response calculations)
• Section 12.2 of The NASTRAN Theoretical Manual (for a discussion of random
response calculations)
8.11 Time-Dependent Loads

You can use the DLOAD entry to define linear combinations of time-dependent loads
that are defined on TLOAD1 and TLOAD2 entries. The TLOAD1 entry defines a
time-dependent load of the form
{ P ( t ) } = { AF ( t – τ ) } Eq. 8-14
where A is defined on a DAREA entry, τ is defined on a DELAY entry, and F ( t – τ ) is

defined on a TABLEDi entry. The TLOAD2 entry defines a time-dependent load of
the form

 0 , t < ( T 1 + τ ) or t > ( T 2 + τ )
{P(t)} =  Eq. 8-15
 A t̃ B e C t̃ cos ( 2πF t̃ + P ) , ( T1 + τ ) ≤ t ≤ ( T2 + τ )

where T1 and T2 are time constants, t̃ = t – T1 – τ , A and τ are the same as on the
TLOAD1 entry, B is the growth coefficient, C is the exponential coefficient, F is the
frequency, and P is the phase angle.
The coefficients on the DAREA and DELAY entries may be different for each loaded
degree-of-freedom. The loads are applied to the specified components in the global
coordinate system.
Two methods are available to automatically create DAREA values from static load
data.
• Simple static loads that supply load values directly to grid points (such as
FORCE, MOMENT, SLOAD, and PLOAD Bulk Data entries) will add to the
DAREA factors when the load identification matches the requested set.
CHAPTER 8 199
Applied Loads
• More complicated loads may be generated using the LOADSET Case

Control command. The requested LSEQ Bulk Data entry matches DAREA
sets to static loading sets. Any type of static load (e.g., LOAD, GRAV,
RFORCE, and SPCD Bulk Data entries) may be generated for either
superelement or residual only (non-superelement) models.
Nonlinear effects are treated as an additional applied load vector, for which the
components are functions of the degrees-of-freedom. This additional load vector is
added to the right side of the equations of motion and treated along with the applied
load vector during numerical integration. It is required that the points to which the
nonlinear loads are applied and the degrees-of-freedom on which they depend be
members of the solution set; i.e., that they cannot be degrees-of-freedom eliminated
by constraints. It is further required that, if a modal formulation is used, the points
referenced by the nonlinear loads be members of the set of extra scalar points
introduced for dynamic analysis.
• “Transient Response Analysis” in the NX Nastran Basic Dynamic Analysis
User’s Guide (for a discussion of transient response calculations)
• Section 11.2 of The NASTRAN Theoretical Manual
There are four different types of nonlinear transient forcing functions:
• arbitrary function generators

• multiplies
• positive power functions
• negative power functions
For a discussion of nonlinear forcing functions, see Section 11.2 of The NASTRAN
Theoretical Manual.
The NOLIN1 entry defines a nonlinear load of the form
Pi ( t ) = Si T ( uj ) Eq. 8-16
where P i is the load applied to u i , S i is a scale factor, T ( u j ) is a tabulated function

defined with a TABLEDi entry, and u j is any permissible displacement component
or velocity component.
Pi ( t ) = Si uj uk Eq. 8-17
where u j and u k are any permissible pair of displacement components and/or

velocity components. They may be the same.

 –S ( u )A , u > 0
 i j j
Pi ( t ) =  Eq. 8-18
 0 , uj ≤ 0

where A is an exponent and uj is any permissible displacement component or

velocity component.
 –S ( u )A , u < 0
 i j j
Pi ( t ) =  Eq. 8-19
 0 , uj ≥ 0

where A is an exponent and uj is any permissible displacement component or

velocity component.
8.12 Using Static Loading Data in Dynamic Analysis

To include static loading data in a dynamic analysis, all static loads and thermal
loads that have the same ID as the DAREA ID on the dynamic load entry are
automatically selected, as shown in the example below.
.
CEND
.
$ THE FOLLOWING REQUEST SELECTS RLOAD1 ENTRY 150
DLOAD = 150
.
BEGIN BULK
$ THE FOLLOWING PLOAD4 IS AUTOMATICALLY SELECTED
$ BECAUSE ITS ID OF 100 MATCHES THE EXCITEID ID OF THE
$ SELECTED RLOAD1 ENTRY
PLOAD4,100,…
RLOAD1,150,100,…
$ THE ABOVE USAGE IS EQUIVALENT TO THE USER
$ SELECTING AN LSEQ BULK DATA OF THE FORM:-
$ LSEQ,SET_ID,100,100,100
.
The DAREA Bulk Data entry is used to specify point loads in dynamic analysis. In
the case of grid points, these loads are implicitly assumed to be in the displacement
(or local) coordinate systems of those points. The software automatically converts all
DAREA Bulk Data entries for grid and scalar points into the appropriate equivalent
FORCE/MOMENT/SLOAD Bulk Data entries.
Two additional, important advantages of the above enhancement follow.
CHAPTER 8 201
Applied Loads
• When you perform dynamic analysis using the modal approach, you may
want to employ residual vectors to improve the quality of the solution. In
addition to specifying PARAM,RESVEC,YES in the Bulk Data Section, you
must also specify static loads at those points that are dynamically excited.
However, with the above automatic conversion feature, it isn’t necessary
to explicitly specify static loads for the purpose of residual vector
calculations. Such loads are automatically generated by NX Nastran.
• With the automatic conversion feature, you can use the DAREA entry to be
used in static analysis as well. This is particularly advantageous when you
want to apply loads at grid points in the displacement (or local) coordinate
systems of those grid points.
When the software performs the automatic conversion, it issues you a message at
the end of the Preface module that contains an image of each converted DAREA
Bulk Data entry as well as the corresponding FORCE/MOMENT/SLOAD Bulk
Data entry into which it has been converted.
See Also
• “DAREA” on page 1096 of the NX Nastran Quick Reference Guide

8.13 Defining Loads for Cyclic Symmetry Analysis
In NX Nastran, there are several different bulk data entries that you can use to
define loading for cyclic symmetry analyses.
• You can use the LOADCYH entry to define the harmonic coefficients of a
load in cyclic symmetry analysis.
• You can use the LOADCYN entry to define a physical load in cyclic
symmetry analysis.
• You can use the LOADCYT entry to specify a tabular load input for
axisymmetric cyclic symmetry problems.
See Also
• “LOADCYH” on page 1352 of the NX Nastran Quick Reference Guide

• “LOADCYN” on page 1354 of the NX Nastran Quick Reference Guide
• “LOADCYT” on page 1356 of the NX Nastran Quick Reference Guide
• “Understanding Cyclic Symmetry” on page 694 of the NX Nastran User’s
Guide
8.14 Combining Loads

You can use the LOAD case control command to combine the loads you specified on
other entries to form new loading conditions. You can use LOAD to define a static
loading condition that’s a linear combination (superposition) of load sets consisting
of loads applied directly to grid points, pressure loads, gravity loads, and centrifugal
forces.
Once you’ve used the loading bulk data entries to define the load, you can use case
control commands to specify which of the loading entries you want to use in a given
analysis. The case control command you use depends on the type of load or loads
you’re applying:
LOAD=n
DEFORM=n
TEMP(LOAD)=n
SPC=n
where n is a unique, user-defined set identification number (SID) in field 2 of the load
Bulk Data entry.
For a given analysis, the total load the software applies is the sum of external
(LOAD), thermal (TEMP(LOAD)), element deformation (DEFORM), and
displacement (SPC) loads.
See Also
• “LOAD” on page 274 of the NX Nastran Quick Reference Guide
• “TEMPERATURE” on page 407 of the NX Nastran Quick Reference Guide
• “DEFORM” on page 217 of the NX Nastran Quick Reference Guide
Using LOAD
You must use LOAD if you want to combine gravity loads with loads applied directly
to grid points, pressure loads, or centrifugal forces, even if there is only one loading
condition.
You can define up to 300 (Si, Li) pairs with a single LOAD entry.
For any particular solution, the total static load will be the sum of the applied loads
(grid point loading, pressure loading, gravity loading, and centrifugal forces) and the
equivalent loads.
The resulting combined load is determined by:
CHAPTER 8 203
Applied Loads
LOAD = S ∑ Si ⋅ { Li } Eq. 8-20

i
Where:
• (S) is the overall scale factor you define on the LOAD entry
• (Si) is the scale factor you define on the LOAD entry for each individual
load set
• {Li} is the applied load vector corresponding to load set ID Li
You request subcases, which may or may not include different loading conditions,
in the Case Control Section. Three Case Control commands (LOAD, TEMP(LOAD),
and DEFORM) are used in selecting loading conditions for subcases. Each of these
commands identifies a unique set whose SID is the same as the SID of one or more
Bulk Data entries. Figure 8-19 shows the Bulk Data entries you can select with each
of the three Case Control commands.
Case Control LOAD = SID1 TEMP(LOAD) = SID2 DEFORM = SID3
Bulk
LOAD Data
Entries
SID = SID1 SID = SID2 SID = SID3
FORCE TEMP DEFORM

FORCE1 TEMPD
FORCE2 TEMPP1
MOMENT TEMPRB
MOMENT1
Bulk MOMENT2
Data PLOAD
Entries PLOAD1 Notes: 1. A LOAD Bulk Data entry combines
PLOAD2 FORCE, MOMENT, etc., entries with
PLOAD4 different SIDs.
PLOADX 2. SID1, SID2, and SID3 must be different
RFORCE numbers.
SLOAD
The GRAV entry cannot have the same SID as other
GRAV loads. Thus, a LOAD Bulk Data entry is required to
combine the gravity force with other loads.
SPCD
If the SPCD entries are the only static loads selected, User Warning
Message 3204 will be printed. The message may be safely ignored.
See Chapter 7 for further discussion of the use of the SPCD entry.
Figure 8-19 Load Combinations
CHAPTER 8 205
Applied Loads
LOAD Example 1
As an example of the LOAD entry combining two loads, consider the umbrella tent
Wind load 0.02 psi

acting at an angle of 5°
1 g gravity load below the horizon
(the X Y plane)
10 11 12
Framing
1.25 in tube Legs
7 8 9
T = 0.125 in 1.5 in tube
6 4 5
E = 10 ⋅ 10 psi 6 T = 0.25 in
Plastic sheet E = 10 ⋅ 10 6 psi
T = 0.05 in 1 2 3
6
E = 2. ⋅ 10 psi
Y
X
Figure 8-20 Umbrella Tent Subjected to Gravity and Wind Loads
The input file containing the combined loading is shown in Listing 8-13. To apply
the wind pressure in the correct direction, a local coordinate system is defined so
that the local Y axis makes an angle of 5 degrees with the ground. This coordinate
system is referenced on the PLOAD4 continuation entry. Since all of the elements
in the model are exposed to the wind load, the simplest method is to use the thru
option on the PLOAD4 entry (fields 8 and 9).
The gravity load of 386.4 entered on the GRAV entry corresponds to a 1 G load in
the English system. Whenever you are working with an acceleration vector in terms
of Gs, make sure that you convert the load such that its units are consistent with
your structure.
Listing 8-13 Umbrella Tent with Wind and Gravity Loads
$
$ FILENAME - TENT.DAT
$
:
LOAD=5
:
BEGIN BULK
$
CORD2R 1 0 0. 0. 0. 0. .08716 1.
1. .08716 1.
$
GRAV 1 0 386.4 0.0 0.0 -1.0
PLOAD4 2 1 .02 THRU 6
1 0. 1. 0.
LOAD 5 1. 1. 1 1. 2
$
PARAM POST 0
PARAM AUTOSPC YES
PARAM GRDPNT 0
$
PSHELL 1 4 .05
$
MAT1 1 10.E6 .3 2.59E-3
25000. 25000. 15000.
$
MAT1 4 2.E6 .3 2.59E-5
25000. 25000. 15000.
$
$ GRID, ELEMENT, AND CONSTRAINTS ARE NOT SHOWN
$
ENDDATA
The gravity load is defined by the GRAV entry with a loading set ID of 1. The wind
loading is defined using the PLOAD4 entry with a loading set ID of 2. The LOAD
entry (ID of 5) is used to combine the gravity and wind loadings. The LOAD entry is
selected by the LOAD = 5 command in the Case Control Section. If the GRAV and
the PLOAD4 entries were assigned the same ID, a fatal error would have resulted.
LOAD Example 2
Assume that your model has one concentrated force of 15.2 lbs in the y direction
applied to grid point 12, and one concentrated moment of 6.4 inch-lbs about the
x-axis applied to grid point 127. It is required to double the value of force and triple
the value of moment for your next analysis.
The LOAD Bulk Data entry may be written with an overall scale factor (S) of 1.0 and
loadset scale factors (Si) of 2.0 for force and 3.0 for moment. Thus,
CHAPTER 8 207
Applied Loads
LOAD = S ∑ Si ⋅ { Li }
i
= 1.0 [ 2.0 { L1 } + 3.0 { L2 } ]
In Case Control:
LOAD = 22
In Bulk Data:
1 2 3 4 5 6 7 8 9 10
LOAD 22 1.0 2.0 30 3.0 40
FORCE 30 12 15.2 0. 1. 0.
MOMENT 40 127 6.4 1. 0. 0.
9
CHAPTER
Constraints
■ Introduction to Constraints
■ Single-Point Constraints
■ Automatically Applying Single-Point Constraints
■ Enforced Displacements at Grid Points (SPCD, SPC)
■ Multipoint Constraints
■ Rigid Body Supports
■ Linear Contact Using Constraints
9.1 Introduction to Constraints

A constraint is the enforcement of a prescribed displacement (i.e., component of translation
or rotation) on a grid point or points. There are two basic types of constraints in NX
Nastran: single point constraints (SPCs) and multipoint constraints (MPCs).
• A single point constraint is a constraint applied to an individual grid point. Single

point constraints can enforce either zero displacement or nonzero displacement.
• A multipoint constraint is a mathematical constraint relationship between one
grid point and another grid point (or set of grid points).
The boundary conditions of a static structure (fixed, hinged, roller support, etc.) typically
require that various degrees of freedom be constrained to zero displacement. For example,
consider a grid point fixed in a rigid wall. All six displacement degrees of freedom—three
translational directions and three rotational directions—must be constrained to zero to
mathematically describe the fixed boundary condition.
Real world structures often don’t have simple or ideal boundary conditions. Because a
model’s constraints greatly influences its response to loading, you must try to constrain
your model as accurately as possible.
This chapter describes how you apply constraints. To understand how constraints are
processed, you need to be familiar with the NX Nastran set notation and matrix operations.
9.2 Single-Point Constraints

A single point constraint (SPC) applies a fixed value to a translational or rotational
component at an individual grid point or to a scalar point. Common uses of single point
constraints are to specify the boundary conditions of a structural model by fixing the
appropriate degrees-of-freedom and to eliminate unwanted degrees-of-freedom with zero
stiffness. Multiple sets of single point constraints can be provided in the Bulk Data Section,
with selections made at execution time by using the subcase structure in the Case Control
Section. This procedure is particularly useful in the solution of problems having one or
more planes of symmetry.
You can use single-point constraints to:
• To constrain a structure to ground.

• To apply symmetric or antisymmetric boundary conditions by restraining the
degrees of freedom that must have a zero value to satisfy symmetry or
antisymmetry.
• To remove degrees of freedom that are not used in the structural analysis (i.e., are
not connected to any structural elements or otherwise joined to the structure).
Constraints 211
• To remove degrees of freedom that are very weakly coupled to the structure.
This condition can occur, for example, to the rotations about the normal of a
slightly curved shell. In this case, a judgment must be made whether to remove
the degree of freedom using an SPC (in which case the structure may be
over-constrained), or to leave it in the problem (in which case the stiffness
matrix is nearly singular). A reasonable rule is to constrain the degree of
freedom if its stiffness is less than 10-8 as large as the stiffness in another
direction at the same grid point.
The elements connected to a grid point may not provide resistance to motion in certain
directions, causing the stiffness matrix to be singular. Single point constraints are used to
remove these degrees-of-freedom from the stiffness matrix.
For example, consider a planar structure composed of membrane and extensional

elements. The translations normal to the plane and all three rotational degrees-of-
freedom must be constrained since the corresponding stiffness matrix terms are all zero.
If a grid point has a direction of zero stiffness, the single point constraint doesn’t need to
be exactly in that direction. It only needs to have a component in that direction. This
allows the use of single point constraints for the removal of such singularities regardless
of the orientation of the global coordinate system. Although the displacements will
depend on the direction of the constraint, the internal forces will be unaffected.
Defining Single Point Constraints in NX Nastran

There are several different ways that you can define single point constraints:
• You can define single point constraints on SPC, SPC1, SPCADD and SPCAX
bulk data entries.
• You can define single point constraints on the GRID and GRDSET entries.
• You can use the PARAM,AUTOSPC,YES option to have the software
automatically create SPCs to remove degrees of freedom that are either
unconnected or very weakly coupled.
You can obtain structural reaction forces called single point constraint forces (SPCFs) at
grid points constrained by SPCs. You can request printed output for single point forces
in the Case Control section. The software prints all nonzero forces, whether they originate
from SPCi entries, the PS field on GRID entries, or PARAM,AUTOSPC,YES.
When you apply a single-point constraint to remove a singularity, the restrained

component of motion to doesn’t need to be aligned exactly with the singular direction of
motion (however, it is highly recommended). Consider the pair of co-linear
pin-connected rods, shown in Figure 9-1, that permit unrestrained motion at point G in
any direction perpendicular to the axis of the rods.
y
n
CROD 1
x
G
CROD 2
Figure 9-1 Pin-Connected Rods
After SPCs are applied to the translation in the z-direction and to all three components of
rotation, unrestrained motion is still possible in the n-direction. An SPC in either the x- or
the y-direction removes the remaining singularity without disturbing the axial forces in the
rods (in this special case). Note, however, that the motion at point G has a zero component
in the restrained direction, which may produce incorrect results for the displacements. For
this reason, you should select a local coordinate system whose coordinate axes align with
the singular components of motion.
Using GRID Entries

You can also define single point constraints using the PS option (field 8) on the GRID and
GRDSET bulk data entries. When you define constraints this way, the constraints are
permanent and are constrained for all subcases; you can’t select them in the Case Control
section. Their primary use, in addition to applying permanent supports, is to remove
unconnected or very weakly connected degrees of freedom.
The default value for enforced displacement on points that you constrain on GRID entries
is zero. You can override the default value at the subcase level with SPC entries.
See Also
• “Specifying Permanent Single Point Constraints (Field 8)” on page 78 of the NX

• “GRID” on page 1335 of the NX Nastran Quick Reference Guide
• “GRDSET” on page 1333 of the NX Nastran Quick Reference Guide
Constraints 213
Grid Example 1: Cantilever Beam

No translations
or rotations
1
CBAR element
Z X
Free—no
2 constraints
Y
required
In this example, we constrained Grid 1 in all six DOFs using field 8 (permanent single
point constraint) of the GRID Bulk Data entry:
1 2 3 4 5 6 7 8 9 10
GRID 1 0. 0. 0. 123456
Or, in free field format,
GRID,1,,0.,0.,0.,,123456
Note that with this option, you don’t need to specify a constraint set in the Case Control
section.
Using SPC and SPC1

You can define single point constraints on SPC, SPC1, SPCADD and SPCAX bulk data
entries.
• The SPC entry is the most general way of specifying single point constraints.
• The SPC1 entry is a less general entry that is more convenient when several grid
points have the same components constrained to a zero displacement.
• The SPCADD entry defines a union of single point constraint sets specified with
SPC or SPC1 entries.
• The SPCAX entry is used only for specifying single point constraints in
problems using conical shell elements.
You should use the SPC and SPC1 Bulk Data entries if you need to apply different sets of
boundary conditions in different subcases. The constraints specified on the SPC and
SPC1 entries belong to sets identified by set identification numbers (SIDs) that must be
selected in the Case Control Section to be used. You can use the SPCADD entry in the
Case Control
www.cadfamily.com section to merge individual SPC and SPC1 sets into combined sets.
The format of the SPC entry is as follows:
1 2 3 4 5 6 7 8 9 10
SPC SID G1 C1 D1 G2 C2 D2
Field Contents
SID Identification number of the single-point constraint set.
Gi Grid or scalar point identification number.
Ci Component number.
Di Value of enforced displacement for all degrees of freedom designated
by Gi and Ci.
The SPC entry can also be used to specify a nonzero displacement constraint. The values
specified in fields 5 and 8 are enforced displacements for the components specified in fields
4 and 7, respectively. When using an enforced constraint within multiple subcases, it is
more efficient to use the SPCD entry.
See Also
• “SPC1” on page 1751 of the NX Nastran Quick Reference Guide
• “SPCADD” on page 1753 of the NX Nastran Quick Reference Guide
SPC Example 1
fixed edge
3 6 9 Z
Y
fixed 5 fixed X
2 8
edge edge
CQUAD4
1 4 7 elements
fixed edge
Constraints 215
Grids 1, 2, 3, 4, 6, 7, 8, and 9 must be constrained in all six DOFs. Assume that a single
point constraint selection is defined in the Case Control Section (SPC = 42). Applying the
same constraint on numerous individual GRID Bulk Data entries is not particularly
convenient-an SPC1 entry is a better choice:
1 2 3 4 5 6 7 8 9 10
SPC1 SID C G1 G2 G3 G4 G5 G6
G7 G8 G9 -etc.-
SPC1 42 123456 1 2 3 4 6 7
8 9
SPC1,42,123456,1,2,3,4,6,7
,8,9
Note that if the “THRU” (alternate format) option were used (1 THRU 9), grid point 5
would have been incorrectly fixed, and no displacement would occur anywhere in the
model.
SCP Example 2: Fixed Hinged Beam

Z
No constraints on
these grid points
X
1 2 3 4 5
Y
No translations No translations
or rotations No rotations about x or z axes
Permit rotation about y axis
Constrain DOFs Constrain DOFs 1, 2, 3 for translation

1 through 6 and 4, 6 for rotation
Here, an SPC entry is used to constrain both ends of the beam. Assume that a single point
constraint set selection is defined in the Case Control Section (SPC = 42). The values of
enforced displacement (Di) are left blank since zero displacement is the default.
1 2 3 4 5 6 7 8 9 10
SPC SID G1 C1 D1 G2 C2 D2
SPC 42 1 123456 5 12346
SPC,42,1,123456,,5,12346
9.3 Automatically Applying Single-Point Constraints

NX Nastran automatically identifies and constrains singularities in the stiffness matrix
using a parameter called AUTOSPC. When you add PARAM,AUTOSPC,YES to either the
bulk data or case control sections of your input file, NX Nastran automatically removes
degrees of freedom that are either unconnected or very weakly coupled to the finite
element model.
With AUTOSPC, you use the PARAM,EPZERO option to specify the minimum stiffness
ratio the software should use to determine a singularity. The stiffness ratio is the ratio of
the stiffness in the weakest direction to the stiffness in the strongest direction at the grid
point (considering all possible directions rather than just the coordinate directions). The
software considers translational and rotational stiffness separately.
See Also
• “AUTOSPC” on page 570 of the NX Nastran Quick Reference Guide

• “EPZERO” on page 590 of the NX Nastran Quick Reference Guide
Grid Point Singularity Processor

When you use the AUTOSPC parameter, the software uses the Grid Point Singularity
Processor (GPSP) module to identify potential singularities. When you set the AUTOSPC
parameter to YES, the GPSP constrains the degrees of freedom associated with the obvious
singularities. The GPSP identifies local singularities in the stiffness matrix after the
software processes all the SPC and MPC constraints in your model.
The software begins identifying local grid point singularities by examining the 3 ⋅ 3
matrices that represent the three translations and three rotations of each grid point. For
each 3 ⋅ 3 matrix, the software performs an eigensolution to obtain the principal values
(eigenvalues) and the principal directions (eigenvectors). The principal values are
normalized to obtain a measure R i as shown in Eq. 9-1:
Ki
R i = ------------------- if K MAX > 0
K MAX
i = 1, 2, 3
R i = 0.0 if K MAX = 0
Eq. 9-1
Constraints 217
where:
K i = the diagonal stiffness term corresponding to the i-th degree of freedom
K MAX = the maximum principal stiffness
Any principal value for which R i < EPZERO represents a singularity. The software
examines the principal direction corresponding to the singularity and identifies the
component with the largest ratio as the failed direction for the grid point. The software
doesn’t constrain any degrees of freedom that you specify as belonging to the a-set.
See Also
• “Understanding Sets and Matrix Operations” on page 243 of the NX Nastran

User’s Guide (for information on a-sets)
• “Constraint and Mechanism Problem Identification in SubDMAP SEKR” on
page 317
Understanding the Grid Point Singularity Table

When the GPSP examines the stiffness matrix for singularities at the grid point level, it
lists any singularities in the Grid Point Singularity Table (GPST). The software prints the
GPST in the .f06 file. The GPST lists all singular degrees-of-freedom, in the global
coordinate system, and the ratio of stiffness between the softest and stiffest degree-of-
freedom for the grid point. It also lists any degrees-of-freedom that the software
automatically constrained.
You should always review the Grid Point Singularity Table carefully to understand
which DOFs were singular and why. Automatic constraint generation can mask actual
modeling errors, such as your forgetting to specify bending stiffness (via MID2 on the
PSHELL entry).
The output includes a grid point singularity table with the following format:
• The TYPE specifies whether the failed degree of freedom is a component of a
grid point (G) of a scalar point (S).
• The component of motion that failed. Constraints are applied in the global
coordinate direction closest to the weakest direction as defined above.
• If possible, the failed degrees of freedom are placed in the sb-set if AUTOSPC is
set to YES (the default for Solution 101). If AUTOSPC is set to NO, the
singularity is identified but not constrained and is not placed in the sb-set.
• An asterisk (*) is printed at the end of the line and indicates that action was taken
by NX Nastran.
To illustrate how failing to inspect the GPSP output can lead to a potential problem,
consider theEMail:[email protected]
www.cadfamily.com single CBAR element example shown in Figure 9-2.
Note: Local coordinate system 1

is rotated –45ο about the X b axis.
F = 100 N
Y1
Z1
End B
X1
End A
Yb
I 1 = 0.667
Xb
I 2 = 0.167
Zb
Figure 9-2 Single Element with Vertical Load
The CBAR element is fully constrained at end A and is loaded with a 100 N force in the
Yb-direction at end B. The input file for this model is shown in Listing 9-1. Both grid points
are defined in the basic coordinate system. The displacement coordinate system for grid
point 1 is the basic system, but the displacement coordinate system for grid point 2 is in the
local coordinate system 1. It is typical to define a local coordinate system when you need a
displacement component in a specific direction.
A sample Grid Point Singularity Table is shown in Figure 9-3. .
JULY 10, 2003 NX NASTRAN 7/10/2003 PAGE 7
G R I D P O I N T S I N G U L A R I T Y T A B L E
POINT TYPE FAILED STIFFNESS OLD USET NEW USET

ID DIRECTION RATIO EXCLUSIVE UNION EXCLUSIVE UNION
2 G 1 0.00E+00 B F SB SB *
2 G 3 0.00E+00 B F SB SB *
2 G 4 0.00E+00 B F SB SB *
2 G 5 0.00E+00 B F SB SB *
2 G 6 0.00E+00 B F SB SB *
Figure 9-3 Grid Point Singularity Table
In this table, grid point 2, the free end of the rod, failed (i.e. had no stiffness connection) in
five of its six possible DOFs. DOF 2 (translation in the Y direction) of grid point 2 wasn’t
singular because it is attached to the extensional stiffness of the rod. CROD elements
support extension and torsion, but in this example no value of J, the torsional constant, was
provided in field 5 of the PROD element property entry. As a result, the CROD element
had no torsional stiffness. The resulting unconnected DOF was detected and constrained
by AUTOSPC.
Constraints 219
Listing 9-1 Single Element with a Vertical Load
$ FILENAME - BAR3A.DAT
ID LINEAR,BAR3A
SOL 101
TIME 5
CEND
TITLE = BAR WITH I1 AND I2
SPC = 1
LOAD = 1
FORCE = ALL
DISP = ALL
BEGIN BULK
PARAM POST 0
PARAM AUTOSPC YES
CORD2R 1 0 0.0 0.0 0.0 0.0 .7071068.7071068+
1. 0.0 0.0
$
GRID 1 0.0 0.0 0.0
GRID 2 10. 0.0 0.0 1
$
CBAR 1 1 1 2 1. 1. 0.0
$
SPC 1 1 123456 0.0
$
PBAR 1 1 1. .667 .167
FORCE 1 2 100. 0. 1. 0.
MAT1 1 1.+7 .3
ENDDATA
For this example, the displacement should be exclusively in the Yb-direction. Since the
displacement is computed and output in the CD coordinate system, you expect that the
displacement components T2 (the Y1-direction) and T3 (the Z1-direction) for grid point 2
are the same for this model. The displacement and force results shown in Figure 9-4
confirm this assumption.
2 G 4 0.00E+00 B F SB S *

1 G .0 .0 .0 .0 .0 .0
2 G .0 3.533767E-03 3.533767E-03 .0 -5.300650E-04 5.300650E-04
F O R C E S I N B A R E L E M E N T S ( C B A R )
ELEMENT BEND-MOMENT END-A BEND-MOMENT END-B - SHEAR - AXIAL
ID. PLANE 1 PLANE 2 PLANE 1 PLANE 2 PLANE 1 PLANE 2 FORCE TORQUE
1 1.000000E+03 1.761739E-14 4.263256E-13 -4.944563E-14 1.000000E+02 6.706302E-15 .0 .0
Figure 9-4 Force and Displacement at Grid Point 2 with a Nonzero I2
However, suppose that when you construct the model, you know that the deflection is in
plane 1 of the element only, so you leave the I2 field blank. The resulting grid point
singularity table, element forces, and displacements are shown in Figure 9-5
2 G 2 0.00E+00 B F SB S *
2 G 4 0.00E+00 B F SB S *
2 G 5 0.00E+00 B F SB S *
1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 7.067534E-03 .0 .0 1.060130E-03
F O R C E S I N B A R E L E M E N T S ( C B A R )
ELEMENT BEND-MOMENT END-A BEND-MOMENT END-B - SHEAR - AXIAL
ID. PLANE 1 PLANE 2 PLANE 1 PLANE 2 PLANE 1 PLANE 2 FORCE TORQUE
1 1.000000E+03 .0 4.263256E-13 .0 1.000000E+02 .0 .0 .0
Figure 9-5 Force and Displacement at Grid Point 2 with a Zero I2
As can be seen, the GPSP module identified a singularity that is caused by the missing
bending stiffness of the element. However, the constraint is placed on the Y1 and θy1 the
degrees of freedom, not the Y1 and θyb degrees of freedom. The force in the CBAR element
is correct, but the displacements are wrong. You made a common assumption that for a
symmetric structure with in-plane loads, the out-of-plane properties are not needed.
However, the choice of the output coordinate system and parameter AUTOSPC resulted in
a modeling error. If you chose the output coordinate system for grid point 2 as the basic
system, the displacements would be correct. Also, for this model, if parameter AUTOSPC
is set to NO, the job will fail.
If this had been a production size model, the error would not have been so obvious and
may well have gone undetected. Always look at the Grid Point Singularity Table results to
make sure that the constrained degrees of freedom are acceptable.
9.4 Enforced Displacements at Grid Points (SPCD, SPC)

There are two methods available to you for specifying an enforced displacement at a
component. The first method is to enter the value of the enforced displacement directly on
an SPC entry.
The alternate method to enforce a displacement at a component is to use the SPCD Bulk
Data entry. The SPCD entry is actually a force, not a constraint, but it is used in conjunction
with an SPC entry to enforce the displacement. When you use an SPCD entry, internal
forces are computed that are applied to the structure to produce the desired enforced
displacements.
The format of the SPCD entry is as follows:
1 2 3 4 5 6 7 8 9 10
SPCD SID G1 C1 D1 G2 C2 D2
Constraints 221
Field Contents
SID Identification number of a static load set.
Gi Grid or scalar point identification number.
Ci Component numbers.
Di Value of enforced displacement for at Gi and Ci.
Grid points with an enforced displacement using the SPCD entry must also appear on an
SPC or an SPC1 Bulk Data entry.
The SPCD method of enforcing a nonzero constraint is more efficient than using an SPC
entry alone when you’re using multiple subcases that specify different constraint
conditions. Note also that when you use an SPCD entry, the displacement values entered
on the SPC entry are ignored. The software only uses the SPCD values.
As an example of enforced displacement, consider the flat plate shown in Figure 9-6. The
plate is modeled with 2500 CQUAD4 elements.
Figure 9-6 Flat Plate with Various Corner Constraints
The purpose of this example is to compare the efficiency of using the SPC/SPCD
combination to that of using the SPC entry only. For the plate model, four runs are made,
each having two subcases with different constraint conditions. For the input files
“const1.dat” and “const2.dat”, the locations of the constraints are the same for both
subcases, but the magnitudes of the constraints change. The file “const1.dat” uses a
different SPC for both subcases. The file “const2.dat” specifies the same SPC for both
subcases, but
www.cadfamily.com it has different SPCD entries.
The input files “const3.dat” and “const4.dat” are similar to the first two; however, the
locations of the constraints in the two subcases are different. Table 9-1 summarizes the
constraints and CPU time for all four analyses.
Table 9-1 Comparison of the SPC Versus the SPC/SPCD Combination
Location of Enforced Displacement
const1.dat const2.dat const3.dat const4.dat
SPC Only SPC/SPCD SPC Only SPC/SPCD
Corner Subcase Subcase Subcase Subcase Subcase Subcase Subcase Subcase

Disp (in) 1 2 1 2 1 2 1 2
Grid 1 0 0 0 0 - - - -
Grid 51 0 5 0 5 - 0 - 0
Grid 2551 5 5 5 5 0 5 0 5
Grid 2601 5 0 5 0 5 - 5 -
CPU 7.3 (sec) 5.1 (sec) 7.3 (sec) 7.3 (sec)

Time*
– free to move
* CPU time on Windows 2000
The SPC/SPCD combination is more efficient for both modeling situations. It is most
advantageous to use this combination when the magnitudes but not the locations of the
constraints change. The larger improvement seen by the magnitude only change is
reasonable because only one SPC needs to be processed; the multiple boundary conditions
are represented as multiple loads. In general, whenever you have multiple subcases with
different constraint conditions, the SPC/SPCD combination is the preferred method for
efficiency.
As mentioned previously, the SPCD is a force entry, not a constraint. Therefore, the SPCD
is called out in the Case Control Section with the LOAD command. The input file for
“const2.dat” is shown in Listing 9-2.
Constraints 223
Listing 9-2 The SPC/SPCD Combination
$ FILENAME - CONST2.DAT
ID LINEAR,CONST2
SOL 101
TIME 200
CEND
TITLE = ENFORCED DISPLACEMENTS USING SPC1 ONLY
DISP(PLOT) = ALL
SPC = 1
SUBCASE 1
LOAD = 11
SUBCASE 2
LOAD = 12
BEGIN BULK
PARAM POST 0
$
SPC1 1 123456 1 51 2551 2601
$
SPCD 11 1 3 0.0
SPCD 11 51 3 0.0
SPCD 11 2551 3 5.0
SPCD 11 2601 3 5.0
$
SPCD 12 1 3 0.0
SPCD 12 51 3 5.0
SPCD 12 2551 3 5.0
SPCD 12 2601 3 0.0
$
INCLUDE ’const.bulk’
$
ENDDATA
As a final comment, if the SPCD entry is the only loading in the model, the software issues
a warning message stating that no loading is specified for your model. If you intend for
the SPCD entry to be the only loading condition, you can ignore the message.
9.5 Multipoint Constraints

In NX Nastran, you can create linear constraint relationships between several degrees of
freedom known as multiple-point constraints or MPCs. Each multipoint constraint is
described by a single equation that specifies a linear relationship for two or more degrees-
of-freedom.
See Also
• Section 5.4 of The NASTRAN Theoretical Manual.

In NX Nastran, the term multiple-point constraint (MPC) describes an imposed linear
relationship between two or more degrees of freedom that are expressed in the form
∑ Rj uj = 0 Eq. 9-2
j
where:
u j = any degree of freedom defined by a grid point or an S point
R j = user-defined scale factor
Uses of Multi-point Constraints

Multi-point constraints have many important practical applications and can be used to:
• Define the relative motion between two grid points as a degree of freedom.
• Define the average of several motions as a degree of freedom.
• Supply hinges or sliding joints between parts of a structure.
• Join dissimilar elements, for example, to join elements with rotational degrees of
freedom to elements that have only translational degrees of freedom (e.g., to join
shell elements to solid elements).
• Obtain the resultant of forces on a structure or on part of a structure.
• Distribute a force to several points in a structure. This is particularly useful if the
force is an unknown force, for example, the force required to compress a fluid.
• Join elements with noncoincident grid points, for example, to change mesh size
within a structure.
• Replace extremely stiff structural members with rigid connections. This modeling
technique should only be used when necessary to improve the numerical
conditioning of the stiffness matrix. MPCs can be used for this application, but
they are not recommended. The preferred method is to use an R-type element
because this element is simpler, and therefore less prone to error.
• Define a component of motion at a grid point that isn’t aligned with the axes of
the global coordinate system, for example, to let you supply a constraint in such
a direction with an additional SPC entry.
• Describe rigid elements and mechanisms such as levers, pulleys, and gear trains.
In this application, the degrees-of-freedom associated with the rigid element that
are in excess of those needed to describe rigid body motion are eliminated with
multipoint constraint equations. Treating very stiff members as rigid elements
eliminates the ill-conditioning associated with their treatment as ordinary elastic
elements.
• Enforce zero motion in directions other than those corresponding with
components of the global coordinate system. In this case, the multipoint
constraint will involve only the degrees-of-freedom at a single grid point. The
constraint equation relates the displacement in the direction of zero motion to the
www.cadfamily.com displacement components in the global system at the grid point.
Constraints 225
• Generate nonstandard structural elements and other special effects with scalar
elements.
• Describe parts of a structure by local vibration modes (see Section 14.1 of The
NASTRAN Theoretical Manual) where the matrix of local eigenvectors represents
a set of constraints relating physical coordinates to modal coordinates.
Defining Multi-point Constraints in NX Nastran

In NX Nastran, you can define multi-point constraints on MPC, MPCADD, and MPCAX
bulk data entries.
• MPC is the basic entry for defining multipoint constraints. The first component
specified on the entry is the dependent degree-of-freedom, i.e., that degree-of-
freedom that is removed from the equations of motion. Dependent degrees-of-
freedom may appear as independent terms in other equations of the set;
however, they may appear as dependent terms in only a single equation.
• The MPCADD entry defines a union of multipoint constraints.
• The MPCAX defines multipoint constraints in conical shell problems.
You can output multipoint forces of constraint with the MPCFORCE Case Control
command in Solution Sequences 101 through 200.
See Also
• “MPC” on page 1431 of the NX Nastran Quick Reference Guide

• “MPCADD” on page 1433 of the NX Nastran Quick Reference Guide
• “MPCAX” on page 1434 of the NX Nastran Quick Reference Guide
•
In general, with these bulk data entries, you must provide the coefficients in the multi-
point constraint equations. However, you can also use several rigid elements to generate
the MPC equations automatically for some applications:
RROD A pin-ended rod which is rigid in extension.

RBAR A rigid bending element with six degrees-of-freedom at
each end.
RTRPLT A rigid triangular plate with six degrees-of-freedom at
each vertex.
RBE1 A rigid body connected to an arbitrary number of grid
points. The independent and dependent degrees-of-
freedom can be arbitrarily selected by the user.
RBE2 A rigid body connected to an arbitrary number of grid

points. The independent degrees-of-freedom are the six
components of motion at a single grid point. The
dependent degrees-of-freedom are specified at an
arbitrary number of grid points.
RBE3 Defines the motion at a reference grid point as the
weighted average of the motions at a set of other grid
points.
RSPLlNE Defines multipoint constraints for the interpolation of
displacements at grid points.
RSSCON Defines multipoint constraints to model clamped
connections of shell-to-solid elements.
The rigid elements will always meet equilibrium and continuity requirements, whereas
this your responsibility if you use the bulk data entries to define the MPC equations.
Using MPC
The format of the MPC entry is as follows:
1 2 3 4 5 6 7 8 9 10
MPC SID G1 C1 A1 G2 C2 A2
G3 C3 A3 -etc.-
Field Contents
SID Set identification number.
Gj Identification number of grid or scalar point.
Cj Component number.
Aj Coefficient.
As an illustration of the use of the MPC entry, consider the following example.
Consider the model shown in Figure 9-7. The goal is to determine the relative motion
between component 1 of grid point 1 and component 1 of grid point 2 using an MPC entry
and a scalar point.
Constraints 227
1 g gravity load
1 2
x1 x2 Y
4 5 6 X
7
Figure 9-7 Using an MPC to Compute Relative Displacement
Let the components of motion at the two grid points whose difference is desired be
designated as and as given by Eq. 9-3. The difference is represented by scalar point 3
defined on an SPOINT entry.
x3 = x2 – x1 Eq. 9-3
The equation of constraint can be in any of the forms shown in Eq. 9-4.
x3 – x2 + x1 = 0 (a)
x2 – x1 – x3 = 0 (b) Eq. 9-4
x1 – x2 + x3 = 0 (c)
Listing 9-3 shows the implementation of Eq. 9-4(a) to obtain the relative displacement.
The first degree of freedom on the MPC entry is the dependent degree of freedom-which
in this case is scalar point 3. It can just as well be made a component of motion at a grid
point defined on a GRID entry. Grid points 1 and 2 are the independent degrees of
freedom. Note that the MPC set ID (SID) is selected in the Case Control Section.
The first listed degree of freedom on the MPC entry (i.e., x 3 in this example) is designated
as dependent and is placed in the m-set. The software eliminates the degrees of freedom
in the m-set early in the solution process. If you want the relative motion to appear as an
independent degree of freedom, express the multipoint constraint in either the second or
third forms of Eq. 9-4.
Listing 9-3 Relative Displacement Using an MPC Entry
$ FILENAME - RELDIFF.DAT
ID LINEAR,RELDIFF
SOL 101
TIME 5
CEND
TITLE = RELATIVE DISPLACEMENT
SET 1 = 1,2,3
DISP = 1
MPC = 1
LOAD = 1
SPC = 1
BEGIN BULK
$
$ ONLY THE GRIDS, SPOINT AND MPC ENTRIES ARE SHOWN
$
SPOINT 3
MPC 1 3 0 1. 2 1 -1.
1 1 1.
$
GRID 1 -10. 20. 0.0
GRID 2 10. 20. 0.0
GRID 4 -10. 10. 0.0
GRID 5 0.0 10. 0.0
GRID 6 10. 10. 0.0
GRID 7 0.0 0.0 0.0
ENDDATA
Rules for Using Multi-point Constraints

There are three important rules for using MPCs (or R-type elements):
1. A member of the m-set can’t be a member of any exclusive subset of the g-set, for
example, it cannot be a degree of freedom that is removed by single-point
constraints.
See Also
• “Understanding Sets and Matrix Operations” on page 243 of the NX
2. A given degree of freedom cannot be designated as a member of the m-set
(dependent) more than once. Thus, it cannot appear as the first listed component
on more than one MPC entry.
3. You must avoid redundant constraints. This redundancy occurs, for example, if
two of the forms in Eq. 9-4 are supplied simultaneously on MPC entries. The
technical difficulty that occurs is that the matrix of constraint coefficients for the
m-set is singular so that the dependent degrees of freedom cannot be evaluated in
terms of the independent ones. This problem also occurs when R-type elements
are interconnected in a closed loop.
Constraints 229
Multipoint constraints simultaneously supply relationships between components of

motion and relationships between the forces of constraint. This can be demonstrated by
using the principle that the forces acting on a rigid body can do no work. Consider, for
example, the equation of constraint among three degrees of freedom shown in Eq. 9-5.
R1 u1 + R2 u2 + R3 u3 = 0 Eq. 9-5
The work done by the corresponding forces of constraint is given by Eq. 9-6.
W = 1 ⁄ 2 ( F1 u1 + F1 u2 + F3 u3 ) = 0 Eq. 9-6
In order for both Eq. 9-5 and Eq. 9-6 to be valid, it is necessary that the following equation
also be valid.
F1 F2 F3
------ = ------ = ------ Eq. 9-7
R1 R2 R3
Thus, the forces of constraint are proportional to the coefficients in the equations of
constraint. As an example, consider the use of a constraint to evaluate the average of four
motions, i.e.,
u5 = 1 ⁄ 4 ( u1 + u2 + u3 + u4 ) Eq. 9-8
From Eq. 9-8, the forces of constraint are given by:
F1 = F2 = F3 = F4 = 1 ⁄ 4 F5 Eq. 9-9
It is clear, from Eq. 9-9, that a force applied to u 5 is distributed equally to u 1 , u 2 , u 3 , and
u4 .
In general, any constraint can be viewed either as a means to constrain a component of

displacement or as a means to distribute the net force acting on the constrained
component of displacement (i.e., the resultant of applied force and internal forces).
9.6 Rigid Body Supports

In the following discussion, a free body is defined as a structure that is capable of motion
without internal stress; i.e., it has one or more rigid body degrees-of-freedom. The
stiffness matrix for a free body is singular with the defect equal to the number of stress-
free, or rigid body modes. A solid three-dimensional body has up to six rigid body modes.
Linkages and mechanisms can have a greater number. In order to permit the analysis of
mechanisms, no restriction is placed in the program on the number of stress-free modes.
You can define free-body supports with a SUPORT or SUPORT1 entry. With conical shell
elements, you use the SUPAX entry. You must define free-body supports in the global
coordinate system. You must select the SUPORT1 bulk entry with the SUPORT1 Case
Control command
www.cadfamily.com in your input file.
In static analysis by the displacement method, the rigid body modes must be restrained in
order to remove the singularity of the stiffness matrix. The required constraints may be
supplied with single point constraints, multipoint constraints, or free body supports.
If you use free body supports, NX Nastran calculates the rigid body characteristics and
checks the sufficiency of the supports. The software checks the supports by calculating the
rigid body error ratio and the strain energy as defined in the Rigid Body Matrix Generator
operation.
See Also
• “Static Solutions in SubDMAP SEKRRS” on page 324 of the NX Nastran User’s

Guide.
This error ratio and the strain energy are automatically printed following the execution of
the Rigid Body Matrix Generator. The error ratio and the strain energy should be zero, but
may be nonzero for any of the following reasons:
• Round-off error accumulation.

• Insufficient free body supports have been provided.
• Redundant free body supports have been provided.
The redundancy of the supports may be caused by improper use of the free body supports
themselves or by the presence of single point or multipoint constraints that constrain the
rigid body motions.
Static analysis with inertia relief is necessarily performed on a model that has at least one
rigid body motion. Such rigid body motion must be constrained by the use of free body
supports. These supported degrees-of-freedom define a reference system, and the elastic
displacements are calculated relative to the motion of the support points. The element
stresses and forces will be independent of any valid set of supports.
Rigid body vibration modes are calculated by a separate procedure provided that you
supply a set of free body supports. This is done to improve efficiency and, in some cases,
reliability. The determinant method, for example, has difficulty extracting zero frequency
roots of high multiplicity, whereas the alternate procedure of extracting rigid body modes
is both efficient and reliable. If you don’t specify free body supports (or you specify an
insufficient number of them), the (remaining) rigid body modes are calculated by the
method selected for the finite frequency modes, provided zero frequency is included in the
range of interest. If you don’t provide free body supports, and if zero frequency isn’t
included in the range of interest, the rigid body modes won’t be calculated.
Constraints 231
You must specify free body supports if the mode acceleration method of solution
improvement is used for dynamic problems having rigid body degrees-of-freedom. This
solution improvement technique involves a static solution, and although the dynamic
solution can be made on a free body, the static solution cannot be performed without
removing the singularities in the stiffness matrix associated with the rigid body motions.
See Also
• “Rigid-body Modes” in the NX Nastran Basic Dynamic Analysis User’s Guide

• “SUPORT Entry” in ge 102 of the NX Nastran Basic Dynamic Analysis User’s
Guide
9.7 Linear Contact Using Constraints

Please note: This section describes an older approach to defining linear contact conditions
with the use of constraints. When this procedure was created, the options for linear contact
in SOL 101 were limited and required some creative modeling techniques. Since there are
now better procedures to define linear contact conditions in SOL 101, the use of this
procedure is no longer recommended. The documentation for this procedure is included
here for those interested in the technique, and in order to support legacy input files with
this type of definition. For a full discussion on defining the recommended linear contact
conditions for SOL 101, see the chapter “Surface Contact for SOL 101” in this book.
Introduction
Traditionally, GAP elements were used to model contact-type problems and had required
the use of nonlinear solution sequences—specifically, SOL 106 or 129. In contrast to the gap
element available for nonlinear analysis (SOL 106 and SOL 129), the “linear” gap
capabilities in SOL 101 did not require estimation of gap stiffness. It used an iterative
constraint type approach. The constraints were applied to grid points or SPOINTs. The
constraints ensured that:
1. The displacement (UR) cannot be negative. This is to ensure that there is no

penetration. Therefore, the chosen degree-of-freedom must be perpendicular to
the contact surface and positive in the opening direction.
2. The force of constraint (QR) cannot be negative. This is to ensure that there is no
tension.
The constraints are satisfied by an iterative technique that is built into SOL 101. The
iterative process starts with a random vector. This random vector assumes certain grids to
be in contact and other grids to be in an open state. A solution is obtained when all the gap
constraints are satisfied, i.e., there's no penetration and no tension forces. If a limit cycle
(return to a previous state) is detected during the iterations, a new random start vector is
tried.
This approach provides an alternate method to the use of GAP elements in SOL 106. Some
experiments have shown that the cost of analysis will be about the same. The advantage is
that you do not have to learn how to calculate the GAP stiffness nor how to control SOL
106. Multiple load conditions are allowed, and each will be solved separately.
If the constraint is between a finite element model and a fixed boundary, then arrange to
have one of the degrees-of-freedom at the boundary grid points represent motion
perpendicular to the boundary. A positive displacement represents motion away from the
boundary. If, on the other hand, the constraint represents relative motion between two
bodies, MPC equations are needed to define a relative motion degree-of-freedom, which is
then constrained to have a non-negative displacement.
Constraints 233
Input
A SUPORT entry along with PARAM,CDITER are required. These two items plus some
optional parameters are described below.
SUPORT entry Selects constrained degrees-of-freedom. These points must be in the

a-set of the residual structure. This means they must not be
dependent in an MPC equation, constrained by SPC, partitioned by
OMIT, or in an upstream superelement.
PARAM,CDITER,x CDITER Constraints will be applied if CDITER is greater than zero.
The value x is the maximum number of iterations allowed. (Default
= 0).
PARAM,CDPRT Controls the printing of constraint violations during the iterations.
The sparse matrix printer prints UR (negative displacements) and
QR (negative forces of constraint) for constrained degrees-of-
freedom. (Default = `YES’)
DMIG,CDSHUT Optional DMIG input of the vector defining the state of the
constrained degrees-of-freedom. A one indicates closed and a zero
indicates open. If DMIG,CDSHUT is not provided, all SUPORT
degrees of freedom will assumed to be closed initially.
PARAM,CDPCH Controls the PUNCH output of DMIG,CDSHUT records for the
final state of the constrained degrees-of-freedom. (Default = `NO’).
This output will be created in a separate .pch file.
Output
The output is similar to standard SOL 101 output. Forces for closed degrees-of-freedom
will appear in the SPCFORCE output. In addition, optional diagnostic information for
the iterations can be requested (PARAM,CDPRT). A final state vector may also be sent to
a .pch file (PARAM,CDPCH).
The .f06 file should be examined to ensure that the iterations have converged, since the
results of the last iteration will be output. The last iteration should have zero changes. A
typical converged output is shown below. More detailed explanation are included in the
actual examples.
Constraints 235
Limitations
The following is a list of limitations for the linear gap elements.
• The only nonlinearity allowed is the constrained displacements.

• There is no sliding friction.
• Free bodies—using the SUPORT entry to define rigid body modes— cannot be
used in the same model with linear gap elements. The parameters INREL and
CDITER are mutually exclusive. A fatal message is issued if both parameters are
present.
• No constraint changes are allowed between subcases.
• There is no guarantee that the solution will converge or that all systems will
follow the same path.
Examples
Two examples are shown below to illustrate the use of the linear GAP element.
EXAMPLE 1(cd_0.dat)
This model consists of a cantilever beam with a stopper and a vertical load of 30 lbs
applied at the free-end. There is a clearance of 0.05 inch between the free-end and the
stopper. The purpose of this problem is to find out if 30 lbs is sufficient to cause the free-
end to hit the stopper.
y P = 30.
12 U 12
x G i = .05 U 13
13
L = 60.
The first step in solving this problem is to define a variable ( S 51 ) to monitor the opening
between the free-end (grid point 12) and the stopper (grid point 13). The following
equation
S 51 = U 12 + G i – U 13 Eq. 9-10
where:
U 12 = the displacement at grid point 12

U 13 = the displacement at grid point 13
Gi = the initial gap opening
S 51 = the relative distance between grids points 12 and 13 including the
effect of the initial opening
accomplishes this goal. Grid point 13 can be removed from the equation since the stopper
does not move. Eq. 9-10 can be simplified and recast as
– U 12 + S 51 – G 101 = 0 Eq. 9-11
so that an MPC equation can be written. Two scalar points are introduced in this case:
Scalar points 101 for G 101 ( G i ) and 51 for S51 . Note that neither S 51 nor G 101 can be defined
as dependent since they already belong to the r-set and s-set, respectively. A copy of the
input file is shown in Listing 9-4.
Listing 9-4 Input file for Cantilever Beam with Stopper.

$
$ cd_0.dat
$
sol 101
cend
Title = Gap elements, Cantilever beam
mpc = 77
spc = 88
load = 300
disp = all
spcf = all
mpcf = all
$
begin bulk
$
grid 11 0. 0. 0.
grid 12 60. 0. 0.
cbar 15 1 11 12 0. 1. 0.
pbar 1 2 1. 1. 1. 1.
mat1 2 3.e7 0.3
force 300 12 -30. 0. 1. 0.
spoint 51 101
suport 51 0
spc 88 11 123456
spc 88 101 0 .05
mpc 77 12 2 -1.0 51 0 1.
Eq. 9-11 is represented by MPC,77. Scalar point 51 must also be placed on the SUPORT
entry. In addition, an enforced displacement of 0.05 must also be imposed on scalar point
101. PARAM,CDITER,10 requests that a maximum of 10 iterations be performed for this
Constraints 237
analysis. If more than 10 iterations are required for this problem, the results at the end of
the 10th iteration will be printed. PARAM,CDPRT,YES is optional in this case since it is
the default.
An abridged output is shown in Eq. 9-7. The large non-zero values for epsilon and strain
energy for the SUPORT degree of freedom can be safely ignored as we are not performing
inertia relief analysis. The solution converges in 1 iteration as indicated by the following
output.
^^^ END ITERATION 1, CHANGES 0
The gap is in contact as indicated by the following scalar point 51 output.
0
0 SHUT
POINT VALUE POINT VALUE POINT VALUE POINT VALUE POINT VALUE
COLUMN 1
51 S 1.00000E+00
The displacement T2 at grid point 12 is –0.05, which implies that the cantilever beam hits
the stopper. The load required to close the gap is:
3 ⋅ EI ⋅ U 12 3 ( 3E7 ) ( 1 ) ( 0.05 )
P g = ------------------------------ = -------------------------------------------- = 20.833 Eq. 9-12
L
3 60 ⋅ 60 ⋅ 60
which is what the spcforce output at grid point 11 shows. The rest of the applied load 9.167
(30-20.833) is reacted by the stopper as indicated by spcforce at grid point 51 and, similarly,
the mpcforce at grid point 12.
0
*** SYSTEM INFORMATION MESSAGE 6916 (DFMSYN)
DECOMP ORDERING METHOD CHOSEN: BEND, ORDERING METHOD USED: BEND
*** USER INFORMATION MESSAGE 3035 (SOLVER)
FOR DATA BLOCK KLR
SUPPORT PT.NO. EPSILON STRAIN ENERGY EPSILONS LARGER THAN 0.001 ARE FLAGGED WITH
ASTERISKS
1 2.5000000E-01 2.0833333E+02 ****
*** USER INFORMATION MESSAGE 5293 (SSG3A)
FOR DATA BLOCK KLL
LOAD SEQ. NO. EPSILON EXTERNAL WORK EPSILONS LARGER THAN 0.001 ARE FLAGGED WITH
ASTERISKS
1 0.0000000E+00 1.5625000E+00
^^^ USER INFORMATION MESSAGE 9077 (CSTRDISP)
^^^ CONSTRAINED DISPLACEMENT ITERATIONS
^^^ CONSTRAINED DISPLACEMENT DIAGNOSTICS, UR= NEGATIVE DISPLACEMENT, QR= NEGATIVE FORCE
^^^ END ITERATION 1, CHANGES 0
^^^ USER INFORMATION MESSAGE 9097 (CSTRDISP)
^^^ CONSTRAINED DISPLACEMENT ITERATIONS CONVERGED LOAD CASE = 1
1 GAP ELEMENTS, CANTILEVER BEAM JULY 31, 2001 MSC.NASTRAN 4/ 9/01 PAGE
7
0
0 SHUT
POINT VALUE POINT VALUE POINT VALUE POINT VALUE POINT
VALUE
COLUMN 1
51 S 1.00000E+00
8
.
0 .

11 G 0.0 0.0 0.0 0.0 0.0 0.0
12 G 0.0 -5.000000E-02 0.0 0.0 0.0 -1.250000E-03
51 S 0.0
101 S 5.000000E-02
11
0

11 G 0.0 2.083333E+01 0.0 0.0 0.0 1.250000E+03
51 S 9.166666E+00
101 S -9.166666E+00
13
F O R C E S O F M U L T I P O I N T C O N S T R A I N T

12 G 0.0 9.166666E+00 0.0 0.0 0.0 0.0
51 S -9.166666E+00
101 S 9.166666E+00
Figure 9-8 Abridged Output of Cantilever Beam with Stopper.

Constraints 239
EXAMPLE 2(cd_1.dat)
The second example is a thick pad supported on a rigid base with a vertical point load
applied at the center. Due to symmetry, a half model is used by applying the appropriate
boundary condition at the center. The problem is to determine where the lift off occurs
along the base (grid points 1 through 7.) An abridged input file for this example is shown
in Listing 9-5. A planar problem (Figure 9-9a) is considered in this case by constraining
3456 on the GRDSET entry. SPC1,200 applies the symmetric boundary condition at the
center line of the structure. The potential contact points are specified on the seven
SUPORT entries—one for each grid point at the bottom. In this case, since the contact is
between a structure and a rigid surface, and there is no initial gap, the use of the MPC is
not required. A maximum of up to 20 iterations (PARAM,CDITER,20) are allowed for
this case. Furthermore, an initial guess of where the contact may occur is provided
(DMIG,SHUT.) In this case, grid points 1 and 7 are assumed to be shut and grid points 2
through 6 are assumed to be opened initially. PARAM,CDPCH,YES requests that the
final open/shut configuration be created in a separate punch file. PARAM,CDPRT,YES
requests that the printing of the constraint violations during the iterations, which is the
default.
Listing 9-5 Input File for Pad on Rigid Foundation.
$ cd_1.dat
$
SOL 101
CEND
TITLE = DEMONSTRATE CONSTRAINED DISPLACEMENT IN SOL 101, #1 CD_1
SPC = 200
LOAD = 300
DISPL = ALL
OLOAD = ALL
SPCFO = ALL
BEGIN BULK
$
$ SYMMETRIC BOUNDARY AT X=0, DOWNWARD LOAD AT TOP CENTER
$
SPC1 200 1 1 11 21 31 41
GRDSET 3456
FORCE 300 41 1. -1.
$
$ CONSTRAIN DISPLACEMENTS ON BOTTOM EDGE, OPTIONAL CDSHUT INPUT
$
$2345678 2345678
SUPORT 1 2
SUPORT 2 2
SUPORT 3 2
SUPORT 4 2
SUPORT 5 2
SUPORT 6 2
SUPORT 7 2
$
PARAM CDITER 20
PARAM CDPRT YES
PARAM CDPCH YES
$
DMIG CDSHUT 0 9 1 0 1
Constraints 241
Figure 9-9 Undeformed Plot of the Thick Pad.
Figure 7-8(cont.) Deformed Plot of the Thick Pad.
From Figure 9-9b, it is obvious that grid points 6, 7, and possibly 5 have lifted off the
foundation due to the downward vertical load. Looking at the output (Figure 9-10), one
can see that grid points 1, 2, 3, and 4 are in contact as indicated under the “SHUT” output.
Furthermore, the displacement output illustrates that grid points 1, 2, 3, and 4 have zero
displacement in the T2 direction—indicating that these four points are in contact. The T2
displacements for grid points 5, 6, and 7 are positive—indicating that these three points
have lifted off the foundation. The applied vertical load is reacted at the four grid points
(1-4) as illustrated by the spcforce output.
0
0 SHUT
POINT VALUE POINT VALUE POINT VALUE POINT VALUE POINT
VALUE
COLUMN 1
1 T2 1.00000E+00 2 T2 1.00000E+00 3 T2 1.00000E+00 4 T2 1.00000E+00
1 DEMONSTRATE CONSTRAINED DISPLACEMENT IN SOL 101, #1 CD_1 SEPTEMBER 10, 2001 MSC.NASTRAN 4/9/01
PAGE 12

1 G 0.0 0.0 0.0 0.0 0.0 0.0
2 G 2.815052E-05 0.0 0.0 0.0 0.0 0.0
3 G 4.985067E-05 0.0 0.0 0.0 0.0 0.0
4 G 6.308859E-05 0.0 0.0 0.0 0.0 0.0
5 G 7.122347E-05 8.332880E-06 0.0 0.0 0.0 0.0
6 G 7.456562E-05 3.158361E-05 0.0 0.0 0.0 0.0
7 G 7.511267E-05 5.419435E-05 0.0 0.0 0.0 0.0
.
.
L O A D V E C T O R

41 G 0.0 -1.000000E+00 0.0 0.0 0.0 0.0
1 DEMONSTRATE CONSTRAINED DISPLACEMENT IN SOL 101, #1 CD_1 SEPTEMBER 10, 2001 MSC.NASTRAN 4/9/01 PAGE
15

1 G -8.072071E-02 2.248780E-01 0.0 0.0 0.0 0.0
2 G 0.0 3.992608E-01 0.0 0.0 0.0 0.0
3 G 0.0 2.689573E-01 0.0 0.0 0.0 0.0
4 G 0.0 1.069040E-01 0.0 0.0 0.0 0.0
11 G -1.541045E-01 0.0 0.0 0.0 0.0 0.0
21 G -1.733866E-01 0.0 0.0 0.0 0.0 0.0
31 G -1.259178E-01 0.0 0.0 0.0 0.0 0.0
41 G 5.341296E-01 0.0 0.0 0.0 0.0 0.0
Figure 9-10 Abridged Output of Thick Pad.
10
CHAPTER
Understanding Sets and Matrix
Operations
■ Overview of Sets and Matrix Operations
■ Understanding the Global Displacement Set and General Solution

Process
■ Static Condensation (Guyan Reduction)
■ Sets for Dynamic Reduction
■ Sets for Aerodynamics
■ Output Selection Via Set Specification
10.1 Overview of Sets and Matrix Operations

When you define a finite element model in NX Nastran, the software writes
equations to simulate the necessary structural properties. These equations are rows
and columns in the matrix equations that describe your model’s structural behavior.
Certain data entries (i.e., SPCs, MPCs, ASETs, etc.) cause matrix operations to be
performed in the various stages of the solution process.
To organize the matrix operations, NX Nastran assigns each degree of freedom

(DOF) in your model to a displacement set. NX Nastran uses the sets to successively
eliminate variables during the solution process.
Most matrix operations used in a structural analysis involve partitioning, merging,

and/or transforming matrix arrays from one subset to another. For example, all
components of motion of a given type form a vector set that is distinguished from
other vector sets. A given component of motion can belong to many combined sets
but can belong to only one exclusive set.
See Also
“Data Processing and Matrix Operations by Functional Module” on page 308 of

the NX Nastran User’s Guide.
In NX Nastran, there are two basic types of degree-of-freedom sets:
• mutually exclusive sets

In NX Nastran, mutually exclusive sets are the data sets that are removed
by a constraint or reduction operation during the analysis. If a DOF is a
member of a mutually exclusive set, it can’t be a member of any other
mutually exclusive set. For example, a degree-of-freedom may not be in the
m-set on two RBAR entries, or on a selected SPC entry and a SUPORT
entry.
• combined sets
Combined sets are formed through the union (combination) of two or more
sets. Each degree-of-freedom is also a member of one or more combined
sets called “supersets.” Supersets are formed when the software combines
the mutually exclusive sets.
The contents of both of these types of sets is listed in the NX Nastran Quick Reference
Guide.
See Also
• “Degree-of-Freedom Sets” on page 747 of the NX Nastran Quick Reference

Guide
CHAPTER 10 245
Understanding Sets and Matrix Operations
10.2 Understanding the Global Displacement Set and

General Solution Process
A finite element model’s degrees of freedom include all the components of
displacement at the grid points and scalar points. Each grid point has six degrees of
freedom, while each scalar point has one. The union of all the local displacement
components at all the grid points and scalar points is called the global displacement
set or g-set. The g-set is the top level set in NX Nastran and represents an
unconstrained set of structural equations.
The g-set displacement is represented as either u g or { u g } . Its significance is that the

mathematical expression for the static equilibrium of the finite element model can
be expressed in matrix form as:
[ K gg ] { u g } = { P g } Eq. 10-1
where [ K gg ] is the global stiffness matrix and { P g } is the vector of explicit loads (e.g.,
loads generated from FORCE entries) or implicit loads (e.g., loads generated from
GRAV entries) applied to all of the grid points and scalar points.
The stiffness matrix [ K gg ] is formed by generating and assembling the stiffness

matrices for all of the elements. The load vector is formed by assembling all of the
implicitly and explicitly defined load components within a given loading condition.
Loads defined within elements are automatically transferred to the appropriate grid
points in a consistent manner.
The number of terms in the { u g } and { P g } vectors is equal to six times the number
of grid points plus the number of scalar points. [ K gg ] is a real and symmetric matrix.
is usually singular; therefore, the equations of equilibrium cannot be solved

[ K gg ]
in the form shown by Eq. 10-1.
At the next step in the solution process, NX Nastran partitions the g-set into two
subsets: the m-set and n-set.
The dependent DOFs of all multipoint constraint relations (MPCs, RBEs, etc.) define
the m-set. During the solution process, the contributions of these m-set DOFs are
condensed into the set of independent DOFs, the n-set. The n-set represents all the
independent DOFs that remain after the dependent DOFs are removed from the
active set of equations. The m-set represents the DOFs that are linearly dependent
upon the n-set DOFs.
Using the n-set, SPCs are then applied to the independent equations to further
partition the equations.
• Degrees of freedom identified by PARAM,AUTOSPC and/or SPCi entries

are assigned to the sb-set.
• Permanent constraints, those DOFs specified in field 8 of the GRID entry,
are assigned to the sg-set.
• The union of the sb- and sg-sets compose the s-set.
After reducing the n-set by applying the s-set constraints, the f-set remains. The f-set
represents the “free” DOFs of the structure. If applied constraints are properly
specified, then the f-set equations represent a statically stable solution (i.e, a
nonsingular stiffness matrix).
If you want the software to perform static condensation (Guyan reduction), the f-set
is further reduced by partitioning it into the o- and a-set. The o-set DOFs are those
that are to be eliminated from the active solution by static condensation. The a-set is
the “analysis “ set and is often the partition at which the solution is performed. By
default, all DOFs in the f-set are assigned to the a-set. In other words, the o-set is null,
by default.
If you use the SUPORT entry, the software places the degrees-of-freedom defined on
the SUPORT entry are placed in the r-set or “reference” set. This set aids in the
calculation of inertia forces, balances the applied loads, and helps determine rigid
body modes. When the r-set is partitioned from the t-set, the l-set remains. This final
set is termed the "leftover" set and is the lowest level of partitioning performed in
NX Nastran static analysis. The l-set partition is the matrix on which the final
solution is performed.
See Also
• “Static Condensation (Guyan Reduction)” on page 248 of the NX Nastran

User’s Guide
• “Inertia Relief in Linear Static Analysis” on page 557 of the NX Nastran
User’s Guide
Understanding Basic Partitioning Operations

NX Nastran partitions the combined sets and removes one set from the analysis.
Table 10-1, for example, lists the basic set partitioning operations. The the m-, s-, o-,
and r-sets are the mutually exclusive sets.
Table 10-1 Basic Set Partitioning Operations
g-set - m-set = n-set

n-set - s-set = f-set
CHAPTER 10 247
Table 10-1 Basic Set Partitioning Operations
f-set - o-set= a-set

a-set - r-set= l-set
Physically, the mutually exclusive set partitioning ensures that operations can’t be
performed on a DOF that’s no longer active. For example, if you apply an SPC to a
DOF that is a dependent degree of freedom on an RBAR, NX Nastran issues User
Fatal Message (UFM) 2101A. Using an SPC entry moves a DOF to the sb-set, but this
cannot occur if the same DOF is already a member of the m-set. Both the m-set and
sb-set are mutually exclusive.
When a particular set has no DOFs associated with it, it is a null set. If a null set
exists, NX Nastran applies a partition and moves the DOFs to the subsequent
partition.
The final set of degrees of freedom remaining after the exclusive sets are eliminated
is called the l-set (degrees of freedom left over). The static solution is performed on
the u l set. The reduced equations of equilibrium are expressed in the matrix form
[ K ll ] { u l } = { P l } Eq. 10-2
The reduced stiffness matrix [ K ll ] is symmetric and nonsingular if all mechanisms

and rigid body motion are removed by the partitioning of the m-set, s-set, and r-set.
Eq. 10-2 is solved for { u l } by Gaussian elimination (decomposition and
forward-backward substitution). All of the other subsets of vector—namely u r , u o ,
u s , and u m —can be obtained subsequently by successive substitution into their
defining equations. By default, all f-set degrees of freedom are assigned to the a-set.
If you don’t have a rigid body support (r-set) in your model, then the a-set is
identical to your l-set.
Determining Set Membership

You can output the set membership of all of the degrees of freedom with the
PARAM,USETPRT and PARAM,USETSEL Bulk Data entries. Use this parameter
with caution as it can produce a substantial amount of output for a large model.
See Also
• “USETPRT” on page 658 of the NX Nastran Quick Reference Guide

• “USETSEL” on page 658 of the NX Nastran Quick Reference Guide
10.3 Static Condensation (Guyan Reduction)

The terms static condensation and Guyan reduction represent the same matrix
operations and are used interchangeably. Guyan Reduction involves a
reformulation of the stiffness and mass matrices based on partitions of the stiffness
matrix. If you start with the stiffness equation in terms of the set of the
unconstrained (free) structural coordinates, you have
[ K ff ] { u f } = { P f } Eq. 10-3
Partitioning the statically independent free degrees of freedom into two subsets of
the f-set, you obtain
 ua 
{ u f } =  -----  Eq. 10-4
 uo 
where:
u a = the a-set variables (the “analysis” set)
u o = the o-set variables, which are removed by static condensation (the “omitted”
set)
Rewriting the static equation for u f in terms of o-set and a-set partitions, you obtain
K aa K ao  u a   Pa 
[ K ff ] [ u f ] =  -----  =  ------  Eq. 10-5
K oa K oo  u o   P0 
Expanding the bottom equation, you arrive at
{ u o } = [ G oa ] { u a } + { u oo } Eq. 10-6
where:
–1
[ G oa ] = – [ K oo ] [ K oa ]
{ u oo } = [ K oo ] – 1 [ P o ]
Expanding the top part of Eq. 10-5 and substituting u o with Eq. 10-6, you arrive at
the equation
–1
{ u a } = [ K aa ] { Pa } Eq. 10-7
CHAPTER 10 249
where:
[ K aa ] = [ K aa + K ao G oa ]
T
{ P a } = [ P a + G oa P o ]
The solution to the f-set degrees of freedom is obtained in three phases. The first
phase is to reduce the f-set to the a-set using Guyan reduction. The second phase is
to solve for { u a } in Eq. 10-7, which is the analysis set. Once this is done, { uo } can
then be solved by Eq. 10-6, which is the omitted set. The conventional method
involves solving Eq. 10-3 in a single step.
In static analysis, the results using static condensation are numerically exact. The
partitioned solution merely changes the order of the operations of the unpartitioned
solution. However, Guyan reduction isn’t generally recommended for static
analysis. This recommendation stems from the cost of the Guyan reduction process
since it creates reduced matrices [ K aa ] and [ M aa ] , which are typically smaller but
denser (i.e., matrix bandwidth is destroyed). Additional decomposition of the
matrix [ K oo ] is also required. In other words, even though the solution size is
smaller, the solution time may take longer and the disk space requirement may be
larger due to a denser matrix. Furthermore, it requires additional interaction on
your part. However, some understanding of the concept of Guyan reduction in
static analysis is quite helpful if you ever decide to apply this feature in dynamic,
cyclic symmetry, or superelement analyses.
Guyan reduction has special applications in dynamic analysis. In dynamic analysis,

the reduction is approximate; the term { u oo } is ignored. The reduction is based
solely on static transformation and is exact provided that no loads are applied to the
o-set degrees of freedom.
Using Static Condensation

In NX Nastran, the ASETi and OMITi bulk data entries define the degrees-of-
freedom in the analysis and omitted sets, respectively. You can use either the
ASET/ASET1 or the OMIT/OMIT1 entries to specify only the a-set (with
ASET/ASET1 entries) or o-set (with OMIT/OMIT1 entries) degrees of freedom.
The software automatically places the remaining f-set DOFs in the complementary
set, as described below. However, if you specify both the a-set and o-set DOFs, then
the unspecified remaining f-set DOFs are automatically placed in the o-set. If you
specify the same DOF on both the a-set and o-set, the job fails with UFM 2101A.
The a-set and the o-set are created in one of the following ways:
1. If OMITi entries are present, the o-set consists of degrees-of-freedom listed

explicitly on OMITi entries. The remaining f-set degrees-of-freedom are
placed in the b-set. They are then also members of the a-set.
2. If ASETi or QSETi entries are present, the a-set consists of all degrees-of-
freedom listed on ASETi entries and any entries listing its subsets, such as
QSETi, SUPORTi, CSETi, and BSETi entries. The remaining f-set degrees-
of-freedom are placed in the o-set. Note, however, there must be at least
one explicit ASETi or QSETi entry type present for the o-set to exist, even
if it is a redundant specification.
3. If there are no ASETi or QSETi entries , but there are SUPORT, BSETi, or
CSETi entries, the entire f-set is placed in the a-set, and the o-set isn’t
created.
4. There must be at least one explicitly defined ASETi, QSETi, or OMITi entry
for the o-set to exist, even if the ASETi, QSETi, or OMITi entry is
redundant.
Notes:
• If you’re performing an analysis using conical shell elements, you use the
OMITAX entry instead of the OMITi entry.
• In superelement analysis, exterior (“boundary”) degrees-of-freedom are
automatically determined and placed in the a-set of the superelement.
Exterior degrees-of-freedom for superelements are therefore equivalent to
degrees-of-freedom listed on ASETi entries.
The a-set defines a combined set. Degrees-of-freedom listed on Bulk Data entries
that define its subsets, the q-, r-, c-, and b-set, may also be listed on ASETi Bulk Data
entries, although this is a redundant specification and isn’t necessary. However,
automatic restarts are more reliable if ASETi entries are present for all degrees-of-
freedom listed on SUPORT entries.
Since the reduction process is performed on an individual degree of freedom, it is

possible to have some of the degrees of freedom at a grid point in the a-set and other
degrees of freedom at a grid point in one of the other mutually exclusive sets. No
additional user input is required.
Format for ASET

You can use ASET in conjunction with ASET1. The formats for the ASET/ASET1 are
as follows:
1 2 3 4 5 6 7 8 9 10
ASET G1 C1 G2 C2 G3 C3 G4 C4
CHAPTER 10 251
Field Contents
Gi The i-th grid or scalar point identification number
Ci Component number(s) for the i-th grid or scalar point
Example:
ASET 5 3 30 345 8 12345
In this example, component 3 of grid point 5, components 3, 4, and 5 of grid point

30, and components 1,2,3,4, and 5 of grid point 8 are assigned to the a-set. All other
degrees of freedom that aren’t otherwise specified (e.g., on SPCi or MPC entries) are
placed in the omitted set (o-set).
Format for ASET1
1 2 3 4 5 6 7 8 9 10
ASET1 C G1 G2 G3 G4 G5 G6 G7
G8 G9 G10 G11 -etc.-
Field Contents
C Component number(s).
Gi The i-th grid or scalar point identification number.
Example:
ASET1 345 3 4 1 16 8 9 5
13 20 25
In this example, components 3,4, and 5 of grid points 1, 3, 4, 5, 8, 9, 13, 16, 20, and 25
are assigned to the a-set. All other degrees of freedom that are not otherwise
specified (e.g., on SPCi or MPC entries) are placed in the omitted set (o-set).
Alternate Format for ASET1
1 2 3 4 5 6 7 8 9 10
ASET1 C ID1 “THRU” ID2
Field Contents
ID1 The first grid or scalar point identification number in the
range.
ID2 The last grid or scalar point identification number in the
range.
Example:
ASET1 123456 10 THRU 909
In this example, components 1,2,3,4,5, and 6 of grid point 10 through 909 are
assigned to the a-set. As long as at least one grid point between 10 and 909 exists, all
the grid points between grid points 10 and 909 don’t need to exist . NX Nastran
places all other degrees of freedom that are outside of grid points 10 through 909 and
not otherwise specified (e.g., on SPCi or MPC entries) in the omitted set (o-set).
Format for OMIT

OMIT can be used in conjunction with OMIT1. The formats for the OMIT/OMIT1
are as follows:
1 2 3 4 5 6 7 8 9 10
OMIT G1 C1 G2 C2 G3 C3 G4 C4
Field Contents
Ci Component number(s) for the i-th grid or scalar point.
Example:
OMIT 5 3 30 345 8 123456
For the above example, component 3 of grid point 5, components 3, 4, and 5 of grid
point 30, and components 1,2,3,4,5, and 6 of grid point 8 are assigned to the o-set. All
other degrees of freedom that are not otherwise specified (e.g., on SPCi or MPC
entries) are placed in the analysis set (a-set).
CHAPTER 10 253
Format for OMIT1
1 2 3 4 5 6 7 8 9 10
OMIT1 C G1 G2 G3 G4 G5 G6 G7
G8 G9 G10 G11 -etc.-
Field Contents
Example:
OMIT1 345 3 4 1 16 8 9 5
13 20 25
For the above example, components 3, 4, and 5 of grid points 1, 3, 4, 5, 8, 9, 13, 16,
20, and 25 are assigned to the o-set. All other degrees of freedom that aren’t
otherwise specified (e.g., on SPCi or MPC entries) are placed in the analysis set
(a-set).
Alternate Format for OMIT1
1 2 3 4 5 6 7 8 9 10
OMIT1 C ID1 “THRU” ID2
Field Contents
ID1 The first grid or scalar point identification number in the range.
ID2 The last grid or scalar point identification number in the range.
Example:
OMIT1 123456 10 THRU 909
For this example, components 1,2,3,4,5, and 6 of grid points 10 through 909 are
assigned to the o-set. All grid points between grid points 10 and 909 need not exist
as long as at least one of them exists. All other degrees of freedom that are outside
of grid points 10 through 909 and not otherwise specified (e.g., on SPCi or MPC
entries) are placed in the analysis set (a-set).
The choice of whether to use the ASET/ASET1 or OMIT/OMIT1 is really a matter

of convenience.
Consider a simple example using Guyan reduction. The cantilever beam model in
Figure 10-1 is used for this purpose. The common Bulk Data entries for this model
are contained in Listing 10-1.
x
1 1 2 2 3 3 4 4 5 5 6
Figure 10-1 Cantilever Beam Using Guyan Reduction
Listing 10-1 Input File for Common Bulk Data Entries
$ filename - bulkm.dat
$
CBAR 1 1 1 2 10
CBAR 2 1 2 3 10
CBAR 3 1 3 4 10
CBAR 4 1 4 5 10
CBAR 5 1 5 6 10
FORCE 10 6 100. 0. 0. 1.
$
GRID 1 0. 0. 0.
GRID 2 1. 0. 0.
GRID 3 2. 0. 0.
GRID 4 3. 0. 0.
GRID 5 4. 0. 0.
GRID 6 5. 0. 0.
GRID 10 0. 0. 10.
MAT1 1 7.1+10 .33 2700.
PBAR 1 1 2.654-3 5.869-7 5.869-7 9.9-7
SPC1 10 123456 1
$
CHAPTER 10 255
The first run is a conventional run without any reduction (Listing 10-2). Since
Guyan reduction isn’t used, then, by default, all degrees of freedom that aren’t
otherwise constrained are assigned to the a-set. In other words, degrees of freedom
1 through 6 for grid points 2 through 6 are assigned to the a-set. Components 1
through 6 of grid point 1 are assigned to the s-set due to the SPC Case Control
command and SPC1 Bulk Data entry.
Listing 10-2 Input File for Using Conventional Analysis
$ filename - full.dat
id bar aset
sol 101
time 10
cend
$
spc = 10
disp = all
$
subcase 1
load = 10
$
begin bulk
$
include ’bulkm.dat’
enddata
The second run uses Guyan reduction by specifying degrees of freedom 1 through
6 at grid points 2, 4, and 6 to the a-set by using the ASET1 entries (Listing 10-3). Due
to this action, the remaining unconstrained degrees of freedom (DOF 1 through 6
for grid points 3 and 5) are assigned to the o-set. Components 1 through 6 of grid
point 1 are assigned to the s-set due to the SPC Case Control command and SPC1
Bulk Data entry.
Listing 10-3 Input File Specifying A-Set DOFs
$ filename - aset.dat
id bar aset
sol 101
time 10
cend
$
title =
subtitle =
spc = 10
disp = all
$
subcase 1
load = 10
$
begin bulk
$
aset1,123456,2,4,6
$
enddata
The third run is similar to the second run except the omitted degrees of freedom are
specified rather than the analysis degrees of freedom (Listing 10-4). Degrees of
freedom 1 through 6 for grid points 3 and 5 are assigned to the o-set by using the
OMIT1 entries. Due to this action, degrees of freedom 1 through 6 for grid points 2,
4, and 6 are automatically assigned to the a-set. Components 1 through 6 of grid
point 1 are assigned to the s-set due to the SPC Case Control command and SPC1
Bulk Data entry.
Listing 10-4 Input File Specifying O-Set DOFs
$ filename - omit.dat
id bar omit
sol 101
time 10
cend
$
title =
subtitle =
spc = 10
disp = all
$
subcase 1
load = 10
$
begin bulk
$
omit1,123456,3,5
$
enddata
CHAPTER 10 257
Since Guyan reduction is numerically exact in static analysis, all three runs should
yield the same results. Figure 10-2 summarizes the results from the three runs-they
are identical, as expected.
***** PARTIAL OUTPUT FROM MODEL WITHOUT USING GUYAN REDUCTION *****

1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 5.599566E-03 .0 -1.079916E-02 .0
3 G .0 .0 2.079839E-02 .0 -1.919851E-02 .0
4 G .0 .0 4.319665E-02 .0 -2.519804E-02 .0
5 G .0 .0 7.039454E-02 .0 -2.879777E-02 .0
6 G .0 .0 9.999225E-02 .0 -2.999767E-02 .0
10 G .0 .0 .0 .0 .0 .0
-----------------------------------------------------------------------------------------------------------------
***** PARTIAL OUTPUT FROM MODEL WITH GUYAN REDUCTION USING ASET ENTRIES *****

1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 5.599566E-03 .0 -1.079916E-02 .0
3 G .0 .0 2.079839E-02 .0 -1.919851E-02 .0
4 G .0 .0 4.319665E-02 .0 -2.519804E-02 .0
5 G .0 .0 7.039454E-02 .0 -2.879777E-02 .0
6 G .0 .0 9.999225E-02 .0 -2.999767E-02 .0
10 G .0 .0 .0 .0 .0 .0
-----------------------------------------------------------------------------------------------------------------
***** PARTIAL OUTPUT FROM MODEL WITH GUYAN REDUCTION USING OMIT ENTRIES *****

1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 5.599566E-03 .0 -1.079916E-02 .0
3 G .0 .0 2.079839E-02 .0 -1.919851E-02 .0
4 G .0 .0 4.319665E-02 .0 -2.519804E-02 .0
5 G .0 .0 7.039454E-02 .0 -2.879777E-02 .0
6 G .0 .0 9.999225E-02 .0 -2.999767E-02 .0
10 G .0 .0 .0 .0 .0 .0
Figure 10-2 Results Comparison With and Without Guyan Reduction
Note that the only differences in the three input files are the ASET1/OMIT1entries.
The common Bulk Data entries are inserted in the run stream by using the
INCLUDE statement.
10.4 Sets for Dynamic Reduction

One of the more important applications of partitioning is Guyan reduction. This
technique is a means for reducing the number of degrees-of-freedom used in
dynamic analysis with minimum loss of accuracy. Its basis is that many fewer grid
points are needed to describe the inertia of a structure than are needed to describe
its elasticity with comparable accuracy. The error in the approximation is small
provided that the set of displacements used for dynamic analysis is judiciously
chosen. Its members should be uniformly dispersed throughout the structure, and
all large mass items should be connected to grid points that are members of the
analysis set.
Importantly, the matrix operations associated with this partitioning procedure tend
to both create nonzero terms and fill what were previously very sparse matrices. The
partitioning option is most effectively used if the members of the omitted set are
either a very large fraction or a very small fraction of the total set. In most of the
applications, the omitted set is a large fraction of the total and the matrices used for
analysis, while small, are usually full. If the analysis set isn’t a small fraction of the
total, a solution using the large, but sparser, matrices may be more efficient.
In dynamic analysis, additional vector sets are obtained by a modal transformation

derived from real eigenvalue analysis of the a-set. These sets are as follows:
ξ o = rigid body (zero frequency) modal degrees-of-freedom
ξ f = finite frequency modal degrees-of-freedom
ξ i = ξ o + ξ f , the set of all modal degrees-of-freedom
One vector set is defined that combines physical and modal degrees-of-freedom:
u h = ξ i + u e , the set of all physical and modal degrees-of-freedom
In dynamic analysis, the a-set is further partitioned into the q-, r-, c-, and b-subsets.
• The q-set stores the coefficients for the generalized coordinates determined
by dynamic reduction and/or component mode calculations.
• The r-set is again used to determine rigid body modes.
• The c-set contains coordinates considered free to move while computing
the generalized coordinate functions.
• The b-set contains coordinates fixed during this process.
The d-set (“dynamic”) for direct formulations is formed from the union of the
structural degrees-of-freedom in the a-set and the extra points of the e-set. The d-set
can be used to input unsymmetrical terms into the equations of motion using DMIG
and TF Bulk Data entries. In the modal formulations, the h-set is composed of the
union of the ci (modal) set and the e-set. Again, you can use the DMIG and TF Bulk
Data entries to introduce unsymmetrical matrix coefficients.
See Also
• “Dynamic Reduction” in the NX Nastran Basic Dynamic Analysis User’s
Guide
CHAPTER 10 259
• “The Set Notation System Used in Dynamic Analysis” in the NX Nastran

Basic Dynamic Analysis User’s Guide
• “Direct Matrix Input” on page 261 of the NX Nastran User’s Guide
10.5 Sets for Aerodynamics
Aerodynamic calculations are made in a Cartesian aerodynamic coordinate system.
By convention, the flow is in the positive x direction. The basic coordinate system
was not chosen, since it would place an undesirable restriction upon the description
of the structural model. Any NX Nastran Cartesian system may be specified, and
flow will be defined in the direction of its positive x-axis. All element and
aerodynamic grid point data, computed initially in the basic coordinate system, will
be converted to the aerodynamic coordinate system. The global (displacement)
coordinate system of the aerodynamic grid points will have its T1 direction in the
flow direction. T3 is normal to the element for boxes, and parallel to the
aerodynamic T3 in the case of bodies. Coordinate system data are generated for the
aerodynamic grid points.
The grid points are physically located at the centers of the boxes and body elements.
Permanent constraints are generated for the unused degrees-of-freedom. A second
set of grid points, used only for undeformed plotting, is located at the element
corners. All six degrees-of-freedom associated with each grid point in this second
set are permanently constrained. Grid point numbers are generated based upon the
element identification number. For any panel, the external grid point numbers for
the boxes start with the panel identification number and increase consecutively.
Aerodynamic degrees-of-freedom, along with the extra points, are added after the
structural matrices and modes have been determined.
See Also
• “Aerodynamic Data Input and Generation” on page 18 of the NX Nastran

Aeroelastic Analysis User’s Guide
10.6 Output Selection Via Set Specification

The membership of grid points and their components and scalar points in the
various sets is stored in the USET table. In NX Nastran, sets are defined by their
name. For example, if you want the TABPRT module to print the M, R, N, SG, K,
and SA sets, then specify:
PARAM,USETSTR1, M:R:N:SG
PARAM,USETSTR2, K:SA.
The data block naming convention in the solution sequences for matrices related to
the sets is based on the set names. For example, the system stiffness matrix [ Kgg ] has
the name KGG. Some sets are used that do not appear in the USET table. The h-set,
for example, is the set used for modal formulations. Its size is not known until after
the USET table is formed, so it does not appear in the table. The j-set, a set
synonymous with the g-set, is used in superelement analysis. There are several other
sets used that also do not appear in the USET table.
Some names may be used for other purposes. You can input the name H, for
example, to the MATGPR module when you want to label rows or columns of
matrices sequentially, rather than by their external sequence numbers. This feature
may be used to number vectors in load matrices or in other instances not related to
modal analysis, even though “H” is defined for modal analysis.
11
CHAPTER
Direct Matrix Input
■ Overview of Direct Matrix Input
■ Using DMIG
■ DMIG Examples
11.1 Overview of Direct Matrix Input

You can use the Bulk Data entry DMIG to input a stiffness (or mass) matrix which
connects specified degrees-of-freedom. The matrix so defined will be added to the
stiffness (or mass) matrix computed from finite element properties.
The DMIG entry includes provisions for unsymmetrical terms and complex values,
both of which are useful in dynamic analysis. These provisions should not be used in
static or normal modes. Note that an entry in the Case Control Section is required
(K2GG = NAME for a stiffness matrix or M2GG= NAME for a mass matrix).
The primary application of the DMIG Bulk Data entry is to enter stiffness and mass
data for parts of the structure which are obtained from another computer run. The
format is cumbersome (two matrix terms per continuation entry) and the matrix
should be input to high precision. For stiffness matrices only, the GENEL Bulk Data
entry is an alternative for manually inputting data.
The finite element approach simulates the structural properties with mathematical
equations written in matrix format. Once you provide the grid point locations,
element connectivities, cross-sectional properties, material properties, applied loads,
and boundary conditions, NX Nastran then automatically generates the appropriate
structural matrices. The structural behavior is then obtained by solving these
equations.
If these structural matrices are available externally, you can input these matrices
directly without providing all the modeling information. Normally this is not a
recommended procedure since it requires additional effort. However, there are
occasions where the availability of this feature is very useful and in some cases is
extremely crucial. Some possible applications are listed below:
• Suppose you are a subcontractor to a classified project. The substructure
that you are analyzing is attached to the main structure built by the primary
contractor. The flexibility of this main structure is crucial to the response of
your component, but the geometry of the main structure is classified. The
main contractor, however, can provide you with the stiffness matrix of the
classified structure. By reading in this stiffness matrix and adding it to your
NX Nastran model, you can account for the flexibility of the attached
structure without compromising the security. The stiffness matrix is the
inverse of the flexibility matrix.
• Suppose you are investigating a series of design options on a component
attached to an aircraft bulkhead. Your component consists of 500 DOFs and
the aircraft model consists of 100,000 DOFs. The flexibility of the backup
structure is somewhat important. You can certainly analyze your
component by including the full aircraft model (100,500 DOFs). On the
other hand, if the flexibility at the attachment points on the aircraft can be
CHAPTER 11 263
Direct Matrix Input
measured experimentally, then you can add the experimental backup

structure stiffness to your component without including the whole aircraft
model. The experimental backup structure stiffness matrix is the inverse
of the measured flexibility matrix. This way your model size remains at
500 DOFs, and you still have a good approximation of the backup structure
stiffness.
• The same concept can be applied to a component attached to a test fixture.
The stiffness of the fixture at the attachment locations can be read in as a
stiffness matrix. Once again, the experimental test fixture stiffness matrix
at the attachment points is the inverse of the measured flexibility at these
points.
There are several ways that these matrices can be read in, such as DMIG, GENEL,
and INPUTT4. Only DMIG is covered here. The DMIG and the INPUTT4 options
offer alternate methods for inputting large matrices. Note that INPUTT4 provides
more precision than the DMIG input; the DMIG yields more precision than the
GENEL on a short word machine.
See Also
• “General Element Capability (GENEL)” on page 148 of the NX Nastran

Element Library
• “INPUTT4” on page 1098 of the NX Nastran DMAP Programmer’s Guide
11.2 Using DMIG
You can use the DMIG feature to directly input a stiffness, mass, or loading matrix
to the grid and/or scalar points in static analysis. These matrices are referenced in
terms of their external grid point IDs and DOFs. The symbols for these g-type
matrices in standard mathematical format are
2 2 2
[ K gg ], [ M gg ], and [ P g ]
All three matrices must be real. Furthermore,
2 2
[ K gg ] and [ M gg ]
must also be symmetric. These matrices are implemented on the g-set level. In other
words, these terms are added to the finite element model at the specified DOFs prior
to the application of constraints. This section focuses on the DMIG features that are
used in static analysis. The DMIG has additional features that can be used in
dynamic analysis.
See Also
• “DMIGs, Extra Points, and Transfer Functions” on page 22 of the NX

Nastran Advanced Dynamic Analysis User’s Guide
DMIG Bulk Data User Interface

In the Bulk Data section, the DMIG matrix is defined by a single DMIG header entry
followed by a series of DMIG data entries. Each of these DMIG data entries contains
a column containing the nonzero elements for the particular matrix that you are
entering.
Header Entry Format
1 2 3 4 5 6 7 8 9 10
DMIG NAME “0" IFO TIN TOUT NCOL
Field Contents
NAME Name of the matrix. It consists of one to eight alphanumeric
characters, the first of which must be alphabetic. (Required)
IFO Form of matrix input: Integer, required
1 = square
9 or 2 = rectangle
6 = symmetric
TIN Type of matrix that is input: Integer, required
1 = real, single precision
2 = real, double precision
TOUT Type of matrix that is created: Integer
0 = set by your machine precision (default)
1 = real, single precision
2 = real, double precision
NCOL Number of columns in a rectangular matrix. Used only for IFO = 9.
(Integer > 0).
Remarks:
1. One header entry is required for each input matrix.
2. The matrix name used in field 2 must be unique.
3. Field 3 of the header entry must contain an integer 0.
4. The default value of 0 for TOUT is recommended.
CHAPTER 11 265
Direct Matrix Input
5. If the input matrix is rectangular, using IFO = 9 is recommended instead of

IFO = 2.
Column Data Entry Format
1 2 3 4 5 6 7 8 9 10
DMIG NAME GJ CJ G1 C1 A1
G2 C2 A2 -etc.-
Field Contents
NAME This is the same name that is used with its
corresponding header entry as described above.
GJ The external grid or scalar point identification number
for the column. (Integer > 0).
CJ The component number for the GJ grid point. (0 <
Integer < 6 if GJ is a grid point; blank or zero if GJ is a
scalar point).
Gi The external grid or scalar point identification number
for the row. (Integer > 0).
Ci The component number for the Gi grid point. (0 <
Integer < 6 if Gi is a grid point; blank or zero if Gi is a
scalar point).
Ai The value of the matrix term.
Remarks:
1. Each non-null column (a column with at least one nonzero term) starts
with a GJ, CJ pair. The entries (Gi, Ci pairs) for each row of that column
follow. The number of Gi, Ci pairs equals the number of nonzero terms in
that column. The terms may be input in arbitrary order. High precision is
required when using DMIG; therefore, in general, the use of large field
input is recommended. Each non-null column of your matrix makes up a
DMIG column data entry. Therefore, the total number of DMIG entries is
equal to the number of non-null columns plus one. The extra entry is the
DMIG header entry.
2. For symmetric matrices (IFO = 6), you can input a given off-diagonal term
either below or above the diagonal. While the upper and lower triangle
terms may be mixed, a fatal error results if the corresponding symmetric
element is input both below and above the diagonal. However, to keep it
simple, for a given matrix, you should decide to input the terms either
completely below the diagonal or above the diagonal, but not both. The
symmetric option is generally used for the mass and stiffness matrices.
3. Rectangular matrices can be input using the option IFO = 9. This option is
generally used to input load matrices.
See Also
• “DMIG” on page 1128 of the NX Nastran Quick Reference Guide
DMIG Case Control User Interface

In order to include the direct input matrices you define with DMIG, the Case Control
Section must contain the appropriate K2GG, M2GG, or P2G command.
See Also
• “K2GG” on page 270 of the NX Nastran Quick Reference Guide

• “M2GG” on page 278 of the NX Nastran Quick Reference Guide
• “P2G” on page 324 of the NX Nastran Quick Reference Guide
Examples
K2GG = mystif
The above Case Control command adds the terms of the matrix, which are defined
by the DMIG entries with a name of “mystif”, to the g-set stiffness matrix.
M2GG = yourmass
by the DMIG entries with a name of “yourmass”, to the g-set mass matrix.
P2G = hisload
by the DMIG entries with a name of “hisload”, to the g-set load matrix.
11.3 DMIG Examples

The following examples of using DMIG features use the cantilever beam model
CHAPTER 11 267
Direct Matrix Input
P1
x
1 2 3 4 5 6
1 2 3 4 5
l1 l2 l3 l4 l5
Figure 11-1 Planer Cantilever Beam
Stiffness Matrix Using DMIG

First, we must analyze this problem using the conventional method with two
loading conditions—a vertical tip load and a gravity load. Listing 11-1 contains the
input file for this run. Note that for illustrative purposes, a planar problem (xz
plane) is shown. However, this same concept is applicable to a general
three-dimensional structure.
Listing 11-1 Input File for Conventional Analysis
$ FILENAME - DMIGFULL.DAT
ID BAR DMIG
SOL 101
TIME 10
DIAG 8
CEND
$
TITLE = PLANAR CANTILEVER BEAM
SPC = 10
DISP = ALL
$
SUBCASE 1
LABEL = TIP LOAD AT END
LOAD = 10
$
SUBCASE 2
LABEL = GRAVITY LOAD
LOAD = 20
$
BEGIN BULK
$
CBAR 1 1 1 2 10
CBAR 2 1 2 3 10
CBAR 3 1 3 4 10
CBAR 4 1 4 5 10
CBAR 5 1 5 6 10
FORCE 10 6 100. 0. 0. 1.
GRAV 20 9.8 0. 0. 1.
GRID 1 0. 0. 0. 246
GRID 2 1. 0. 0. 246
GRID 3 2. 0. 0. 246
GRID 4 3. 0. 0. 246
GRID 5 4. 0. 0. 246
GRID 6 5. 0. 0. 246
GRID 10 0. 0. 10. 123456
MAT1 1 7.1+10 .33 2700.
PBAR 1 1 2.654-3 5.869-7
SPC1 10 123456 1
$
ENDDATA
The next run removes CBAR element number 5 from the input file and uses the
DMIG feature to directly read in the bar element stiffness matrix. The bar element
stiffness matrix for a planar element and ignoring the axial stiffness is as follows:
6 – 3l – 6 – 3l
2 2
2EI
[ K ] e = --------- – 3l 2l 3l l
3
l – 6 3l 6 3l
2 2
– 3l l 3l 2l
CHAPTER 11 269
Direct Matrix Input
Since the matrix is symmetric, either the lower or upper triangular portion of the
matrix must therefore be provided. For E = 7.1 ⋅ 1010 N ⁄ m 2 , I = 5.869 ⋅ 10– 7 m 4 , and
l 5 = 1m , the stiffness matrix for CBAR element number 5 is as follows:
5, 3 5, 5 6, 3 6, 5
5, 3 500038.3
5, 5 – 250019.4 166679.6 SYM
[ K ]e = Eq. 11-1
6, 3 – 500038.8 250019.4 500038.8
6, 5 – 250019.4 83339.8 250019.4 166679.6
In this case, you chose UGSTIF as the name of the input stiffness for bar element
number 5. Therefore, in order to bring in this stiffness matrix and add it to the
global stiffness matrix, the Case Control command K2GG= UGSTIF is required.
Note these stiffness terms are additions to the global stiffness matrix at that location
and are not a replacement of the stiffness terms at that location. In the Bulk Data
Section, five DMIG entries are required-one for the header entry and four for the
data column entries—since there are four non-null columns in the above matrix.
For the header entry, the same name UGSTIF must be used to match the name called
out in the Case Control Section. The third field is 0, which must be the value used
for the header entry. The fourth field (IFO) is set to 6 to denote a symmetric matrix
input. The fifth field (TIN) is set to 1 to denote that the matrix is provided as real,
single precision.
Once again, the terms in the matrix are referenced in terms of their external grid IDs
when using the DMIG feature. Physically, each term in a particular column of the
stiffness matrix ( k ij ) represents the induced reactive load in the i-th degree of
freedom due to a unit displacement in the j-th direction with all other displacement
degrees of freedom held to zero. Since the matrix is symmetric, only the lower
triangular portion of the matrix is read.
The first DMIG data column entry reads the first column of the above matrix. Field
2 of this DMIG entry must have the same name UGSTIF as referenced by the Case
Control K2GG = UGSTIF command. Fields 3 and 4 of this entry identify this column
in terms of its external grid ID and corresponding degree of freedom, respectively.
In this case, it is grid point number 5, degree of freedom 3 (z-translation at grid
number 5).
Once this column is defined and by following the format description as described in
the section for column data entry format, you can then input the four terms in this
column row by row. These four terms are defined by sets of three fields. They are
the external grid point ID number, corresponding degree of freedom, and the actual
matrix term, respectively. The first row of column one is defined by external grid
point ID 5, degree of freedom 3 (z-translation at grid point number 5) with a stiffness
value EMail:[email protected]
www.cadfamily.com of 500039. The second row of column one is defined by external grid point ID
5, degree of freedom 5 (y-rotation at grid point number 5) with a stiffness value of

-250019. The third row of column one is defined by external grid point ID 6, degree
of freedom 3 (z-translation at grid point number 6) with a stiffness value of -500039.
The fourth row of column one is defined by external grid point ID 6, degree of
freedom 5 (y-rotation at grid point number 6) with a stiffness value of -250019.
The next DMIG entry reads the second column of the above matrix starting with the
diagonal term. Fields 3 and 4 of this entry identify this column in terms of its external
grid point ID and corresponding degree of freedom, respectively. In this case, it is
grid number 5, degree of freedom 5 (y-rotation at grid point number 5). The rest of
the procedure is similar to that of column one with the exception that only three rows
need to be read due to symmetry (rows two through four).
The next two DMIG entries read columns three and four of the stiffness matrix,
respectively. Note that due to symmetry, one less row needs to be read for each
additional column.
The input file using DMIG is contained in Listing 11-2. The results as compared with
the conventional analysis are tabulated in the first two sections of Figure 11-2. The
comparison is performed for Subcase 1 with a vertical load applied to the tip of the
cantilever beam. In this case, the results are accurate to within three to four digits.
As mentioned previously, a high degree of precision should be retained when

inputting a matrix with the DMIG method. In this case, since the small-field input
format is used, the maximum number of characters that can be used are eight
(including sign and decimal place). The same job was then rerun for the DMIG input
with additional significant digits added using double-precision and large-field
format (Listing 11-3). In this case, the results are identical by comparing the results
of the first and last sections of Figure 11-2. Note that even the double-precision input
on short-word machines does not contain all of the significant digits of the machine.
CHAPTER 11 271
Direct Matrix Input
Listing 11-2 Input File for Reading a Stiffness Matrix

Using DMIG with Small-Field Format
$ FILENAME - DMIGSTFS.DAT
ID BAR DMIG
SOL 101
TIME 10
DIAG 8
CEND
$
TITLE = PLANER CANTILEVER BEAM
SUBTITLE = USE DMIG TO BRING IN STIFFNESS MATRIX OF ELEMENT # 5 USING SMALL FIELD
K2GG = UGSTIF
SPC = 10
DISP = ALL
$
SUBCASE 1
LOAD = 10
$
BEGIN BULK
$
CBAR 1 1 1 2 10
CBAR 2 1 2 3 10
CBAR 3 1 3 4 10
CBAR 4 1 4 5 10
$ DMIG HEADER ENTRY
$
DMIG UGSTIF 0 6 1
$
$ DMIG DATA COLUMN ENTRIES
$
DMIG UGSTIF 5 3 5 3 500039. +000001
++0000015 5 -250019. 6 3 -500039. +000002
++0000026 5 -250019. +000003
$
DMIG UGSTIF 5 5 5 5 166680. +000004
++0000046 3 250019. 6 5 83340. +000005
$
DMIG UGSTIF 6 3 6 3 500039. +000006
++0000066 5 250019. +000007
$
DMIG UGSTIF 6 5 6 5 166680.
$
FORCE 10 6 100. 0. 0. 1.
GRAV 20 9.8 0. 0. 1.
GRID 1 0. 0. 0. 246
GRID 2 1. 0. 0. 246
GRID 3 2. 0. 0. 246
GRID 4 3. 0. 0. 246
GRID 5 4. 0. 0. 246
GRID 6 5. 0. 0. 246
GRID 10 0. 0. 10. 123456
MAT1 1 7.1+10 .33 2700.
PBAR 1 1 2.654-3 5.869-7
SPC1 10 123456 1
$
ENDDATA
Listing 11-3 Input File for Reading a Stiffness Matrix with Large-Field
Format
$ FILENAME - DMIGSTFL.DAT
ID BAR DMIG
SOL 101
TIME 10
DIAG 8
CEND
TITLE = PLANER CANTILEVER BEAM
SUBTITLE = DMIG TO BRING IN STIFFNESS MATRIX FOR ELEMENT # 5 USING LARGE FIELD
K2GG = UGSTIF
SPC = 10
DISP = ALL
$
SUBCASE 1
LOAD = 10
$
BEGIN BULK
$
CBAR 1 1 1 2 10
CBAR 2 1 2 3 10
CBAR 3 1 3 4 10
CBAR 4 1 4 5 10
$
$ DMIG HEADER ENTRY
$
DMIG UGSTIF 0 6 2
$
$ DMIG DATA COLUMN ENTRIES
$
DMIG* UGSTIF 5 3 *A
*A 5 3 5.000388 D+5 *B
*B 5 5 -2.500194 D+5 *C
*C 6 3 -5.000388 D+5 *D
*D 6 5 -2.500194 D+5
$
DMIG* UGSTIF 5 5 *A2
*A2 5 5 1.666796 D+5 *B2
*B2 6 3 2.500194 D+5 *C2
*C2 6 5 8.33398 D+4
$
*A3 6 3 5.000388 D+5 *B3
*B3 6 5 2.500194 D+5
$
*A4 6 5 1.666796 D+5
$
FORCE 10 6 100. 0. 0. 1.
GRAV 20 9.8 0. 0. 1.
GRID 1 0. 0. 0. 246
GRID 2 1. 0. 0. 246
GRID 3 2. 0. 0. 246
GRID 4 3. 0. 0. 246
GRID 5 4. 0. 0. 246
GRID 6 5. 0. 0. 246
GRID 10 0. 0. 10. 123456
MAT1 1 7.1+10 .33 2700.
PBAR 1 1 2.654-3 5.869-7
SPC1 10 123456 1
ENDDATA
CHAPTER 11 273
Direct Matrix Input
***** PARTIAL OUTPUT FROM DMIGFULL.F06 - FULL MODEL *****


1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 5.599566E-03 .0 -1.079916E-02 .0
3 G .0 .0 2.079839E-02 .0 -1.919851E-02 .0
4 G .0 .0 4.319665E-02 .0 -2.519804E-02 .0
5 G .0 .0 7.039454E-02 .0 -2.879777E-02 .0
6 G .0 .0 9.999225E-02 .0 -2.999767E-02 .0
10 G .0 .0 .0 .0 .0 .0
***** PARTIAL OUTPUT FROM DMIGSTIFS.F06 - DMIG TO INPUT STIFFNESS FOR ELEM # 5 USING SMALL FIELD

1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 5.598505E-03 .0 -1.079704E-02 .0
3 G .0 .0 2.079415E-02 .0 -1.919427E-02 .0
4 G .0 .0 4.318711E-02 .0 -2.519168E-02 .0
5 G .0 .0 7.037757E-02 .0 -2.878928E-02 .0
6 G .0 .0 9.996621E-02 .0 -2.998813E-02 .0
10 G .0 .0 .0 .0 .0 .0
***** PARTIAL OUTPUT FROM DMIGSTIFL.F06 - DMIG TO INPUT STIFFNESS FOR ELEM # 5 USING LARGE FIELD

1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 5.599566E-03 .0 -1.079916E-02 .0
3 G .0 .0 2.079839E-02 .0 -1.919851E-02 .0
4 G .0 .0 4.319665E-02 .0 -2.519804E-02 .0
5 G .0 .0 7.039454E-02 .0 -2.879777E-02 .0
6 G .0 .0 9.999225E-02 .0 -2.999767E-02 .0
10 G .0 .0 .0 .0 .0 .0
Figure 11-2 Results Comparison Between the Conventional Method Versus

DMIG
Mass Matrix Using DMIG

You can also include a mass matrix with DMIG in a similar manner as the stiffness
matrix with the following exceptions:
• You use the Case Control command M2GG = “name”, instead of K2GG=
“name,” where “name” is the name of the mass matrix referenced in field
2 of the DMIG Bulk Data entries.
• The matrix defined in the DMIG entries represents the mass matrix terms
instead of the stiffness matrix terms.
Load Matrix Using DMIG

The load vectors can also be read using the DMIG feature if desired, although this
is not a commonly used feature. It is similar to the format for reading in the stiffness
or mass matrix, with the following exceptions:
• You use the Case Control command P2G = “name”, instead of K2GG =
“name” or M2GG = “name.”
• The matrix defined in the DMIG entries are columns of the load vectors
instead of the stiffness or mass matrix terms.
• The load matrix, in general, is rectangular and unsymmetrical (i.e., IFO = 9
on the header entry).
• The column number (GJ) you specify on the DMIG input must match the
load sequence number. The GJ field is the load sequence number, and the
CJ field isn’t used at all for a rectangular matrix. For example, if there is a
total of ten load conditions in your model and you want to add load vectors
to the third and seventh subcases using DMIG, then the NCOL field should
be assigned as 10 on the header entry. Two additional DMIG Data entries,
representing the two load vectors added to the third and seventh subcases,
are needed. The GJ field for the first of these two data entries must have a
value of 3 to denote the third load sequence. The remaining data for this
entry are the nonzero rows of this load vector. Their input format is
identical to that of the stiffness or mass matrix. The GJ field for the second
of these two data entries must have a value of 7 to denote the seventh load
sequence number. Again, the remaining data for this entry are the nonzero
rows of this load vector.
Again the problem in Figure 11-1 is used to illustrate this feature. The input file is
shown in Listing 11-4. In this example, the DMIG feature is used to read in the load
for the vertical tip load in Subcase 10. In the Case Control Section, the P2G =
UGLOAD command is used to activate the DMIG entries with the name of
UGLOAD. In the Bulk Data Section, the DMIG load header entry is similar to that of
the stiffness and mass header entry with the exception of the IFO and NCOL fields.
In general, the matrix is rectangular; therefore, IFO is defined as 9 instead of 6. NCOL
is the total number of load cases for the problem.
Since there are two load cases, NCOL must be defined as 2. The number of DMIG
load data entries is equal to the number of external load columns added. In this case,
you only need one DMIG load data entry since only one load condition is added to
the first subcase. The GJ field of 1 denotes that you are adding the load to the first
subcase (first load sequence). Note that GJ references the load sequence (1) rather
than the SUBCASE ID (10). The CJ field should be blank when using the DMIG load
feature. The rest of the fields are read in the same manner as the stiffness and mass
matrices—only nonzero entries need to be defined. Note that if you have reversed
your subcase setup—that is, if you apply your gravity load as the first subcase and
bring in the tip load as the second subcase—then field three of the second DMIG
entry must then be modified to 2 instead of 1.
CHAPTER 11 275
Direct Matrix Input
Listing 11-4 Input File for Reading in a Load Vector Using DMIG
$ FILENAME - dmigload.dat
ID BAR DMIG
SOL 101
TIME 10
DIAG 8
CEND
$
TITLE = USE DMIG TO BRING IN LOAD VECTOR FOR SUBCASE 1
P2G = UGLOAD
SPC = 10
DISP = ALL
$
SUBCASE 10
$
SUBCASE 20
LABEL = GRAVITY LOAD
LOAD = 20
$
BEGIN BULK
$
CBAR 1 1 1 2 10
CBAR 2 1 2 3 10
CBAR 3 1 3 4 10
CBAR 4 1 4 5 10
CBAR 5 1 5 6 10
$
DMIG UGLOAD 0 9 1 2
DMIG UGLOAD 1 6 3 100.
$
GRAV 20 9.81 0. 0. 1.
GRID 1 0. 0. 0. 246
GRID 2 1. 0. 0. 246
GRID 3 2. 0. 0. 246
GRID 4 3. 0. 0. 246
GRID 5 4. 0. 0. 246
GRID 6 5. 0. 0. 246
GRID 10 0. 0. 10. 123456
MAT1 1 7.1+10 .33 2700.
PBAR 1 1 2.654-3 5.869-7
SPC1 10 123456 1
$
ENDDATA
***** PARTIAL OUTPUT FROM DMIGFULL.F06 - FULL MODEL *****
TIP LOAD AT END SUBCASE 1


1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 5.599566E-03 .0 -1.079916E-02 .0
3 G .0 .0 2.079839E-02 .0 -1.919851E-02 .0
4 G .0 .0 4.319665E-02 .0 -2.519804E-02 .0
5 G .0 .0 7.039454E-02 .0 -2.879777E-02 .0
6 G .0 .0 9.999225E-02 .0 -2.999767E-02 .0
10 G .0 .0 .0 .0 .0 .0
GRAVITY LOAD SUBCASE 2


1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 9.278418E-03 .0 -1.729160E-02 .0
3 G .0 .0 3.233388E-02 .0 -2.783526E-02 .0
4 G .0 .0 6.326194E-02 .0 -3.331796E-02 .0
5 G .0 .0 9.784514E-02 .0 -3.542669E-02 .0
6 G .0 .0 1.335530E-01 .0 -3.584844E-02 .0
10 G .0 .0 .0 .0 .0 .0
***** PARTIAL OUTPUT FROM DMIGLOAD.F06 - DMIG TO INPUT LOAD CONDITION # 1 *****
LOAD CONDITION # 2 IS APPLIED BY THE CONVENTIONAL MANNER
TIP LOAD AT END SUBCASE 10


1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 5.599566E-03 .0 -1.079916E-02 .0
3 G .0 .0 2.079839E-02 .0 -1.919851E-02 .0
4 G .0 .0 4.319665E-02 .0 -2.519804E-02 .0
5 G .0 .0 7.039454E-02 .0 -2.879777E-02 .0
6 G .0 .0 9.999225E-02 .0 -2.999767E-02 .0
10 G .0 .0 .0 .0 .0 .0
GRAVITY LOAD SUBCASE 20


1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 9.278418E-03 .0 -1.729160E-02 .0
3 G .0 .0 3.233388E-02 .0 -2.783526E-02 .0
4 G .0 .0 6.326194E-02 .0 -3.331796E-02 .0
5 G .0 .0 9.784514E-02 .0 -3.542669E-02 .0
6 G .0 .0 1.335530E-01 .0 -3.584844E-02 .0
10 G .0 .0 .0 .0 .0 .0
Figure 11-3 Results Comparison Between Conventional Load Input Versus

DMIG
CHAPTER 11 277
Direct Matrix Input
Inputting large matrices by hand using the DMIG method can be time consuming,
error prone, and quite tedious. However, in most instances, the DMIG input may
be automatically generated by a DMAP module when using NX Nastran or other
external programs. You can also use PARAM,EXTOUT,DMIGPCH to generate
these DMIG entries.
See Also
• “EXTOUT” on page 591 of the NX Nastran Quick Reference Guide
12
CHAPTER
Element Data Recovery Resolved at
Grid Points
■ Introduction to Element Data Recovery
■ Stress Recovery at Grid Points
■ Mesh Stress Discontinuities at Grid Points
12.1 Introduction to Element Data Recovery

Element stresses may be averaged at grid points and then used to calculate estimates
of stress discontinuities in the model mesh. Resolving element data at the grid points
is supported only for selected linear elastic element types (see Table 12-1 for
supported element types). The calculation of average stresses at grid points and the
calculation of estimates of stress discontinuities use a common mesh definition
method.
You define the mesh in the OUTPUT(POST) section of the Case Control section using
one of two methods:
• SURFACE input for 2-D shell element meshes

• VOLUME input for 3-D solid element meshes
The averaging of grid point stresses (and strains) and stress discontinuities are
calculated independently for meshes composed of only 2-D shell elements and those
composed of only 3-D solid elements. You use the STRESS Case Control command
to request the element stress and strain data used in the calculation of grid point
stresses and stress discontinuities. To substitute the calculation of strains for stresses,
use the STRAIN Case Control command in place of the STRESS command in the Case
Control Section.
You request grid point stresses for reports with the GPSTRESS Case Control
command. The calculation of stress discontinuities is supported for both elements
and grid points associated with an element mesh by the ELSDCON and GPSDCON
commands, respectively.
See Also
• “STRESS” on page 383 of the NX Nastran Quick Reference Guide
• “STRAIN” on page 380 of the NX Nastran Quick Reference Guide
• “ELSDCON” on page 236 of the NX Nastran Quick Reference Guide
• “GPSDCON” on page 256 of the NX Nastran Quick Reference Guide
CHAPTER 12 281
Element Data Recovery Resolved at Grid Points
The following sections explain in more detail the requesting and calculation of grid
point stresses and stress discontinuities, respectively.
Table 12-1 Elements Types Supported in Element Data Resolved at Grid Point
Calculations (Hyperelastic Elements Not Included)
Element 2-D Shell 3-D Solid Stress

Type Stress Stress Discontinuity
CQUAD4 x x
CQUADR x x
CQUAD8 x x
CTRIA3 x x
CTRIAR x x
CTRIA6 x x
CHEXA x x
CPENTA x x
CTETRA x x
12.2 Stress Recovery at Grid Points

You can have NX Nastran compute stresses at grid points:
• on surfaces of two-dimensional plate elements, namely, the CQUAD4,

CQUADR, CQUAD4, CTRIA3, CTRIAR, and CTRIA6
• in volumes containing CHEXA, CPENTA, and CTETRA solid elements
By default, NX Nastran outputs stresses at the center of each of these elements and
also at the vertices of CQUADR, CQUAD8, CTRIAR, and CTRIA6 shell elements
and the CHEXA, CPENTA, and CTETRA solid elements. However, the center
output isn’t always adequate. For example, when you’re working with the surfaces
of CQUAD4 and/or CTRIA3 elements, you may also need realistic stress values at
the grid points. Also, in the case of surfaces of CQUADR, CQUAD8, CTRIAR, and
CTRIA6 elements or of volumes containing CHEXA, CPENTA, and CTETRA
elements, the vertex stresses output for the elements that connect to a common grid
point are often not identical.
This section describes a rational method that yields accurate and unique stresses at
each vertex grid point in user-defined surfaces or volumes. Grid point stress
recovery is only available in linear static analysis, real eigenvalue analysis, and
transient analysis. Grid point stress recovery isn’t available for lamina stresses.
You can also use NX Nastran to estimate the probable error in these grid point
stresses (see “Mesh Stress Discontinuities at Grid Points” on page 294). Estimates
are generated for each stress component. The root mean square error of these
estimates is then computed to provide a single measure of the error in the
computation of the stresses at a grid point. This data provides a gross indication of
the adequacy of a finite element model to represent the physical phenomena under
investigation. The error estimator is available in SOLution Sequence 101.
Description of Method
A scheme to interpolate and/or extrapolate over a surface from a known set of
stresses (the element stresses) requires that these known stress components be
transformed into a consistent coordinate system prior to interpolation and
extrapolation. This coordinate system is defined by the analyst and it should be
generally compatible with the surface of interest in order to obtain accurate grid
point stresses. Generally, such a coordinate system will be the natural coordinate
system that the analyst used in the generation of the model of the surface or volume
in question. Calculated grid point stress components are output in this coordinate
system, called the output coordinate system of the surface or volume.
Each stress component is treated independently of other stress components in the

calculation of average grid point stress components. The stress invariants at grid
points (i.e., principal stresses, the vonMises stress, and the mean pressure) are
evaluated from the average values of the stress components at grid points. The grid
point stress components are output in the output coordinate system of the surface or
volume and the stress invariants are oriented relative to this output coordinate
system.
Two methods are used to calculate grid point stress components for plate and shell
elements: topological and geometric. A single, simplified, topological method is
used to calculate grid point stress components in volumes containing solid elements.
However, prior to a discussion of these methods, it is necessary to define the four
categories into which grid points are divided.
1. Interior Grid Points

Grid points that are connected only to interior line segments; i.e., line
segments that are coincident with the edges of two or more elements.
2. Corner Grid Points
Grid points that are only connected to a single element.
3. Edge Grid Points
Grid points that are connected to both interior and exterior line segments.
www.cadfamily.com An exterior line segment is coincident with the edge of only one element.
CHAPTER 12 283
4. Exception Grid Points

Grid points at which discontinuities in stress can occur. Exception points
will be described more completely under “Treatment of Exception Points
– “BRANCHing”” on page 288.
Stresses at interior, corner, or edge grid points are considered to be continuous

between directly connected elements. Stresses at exception points may be
discontinuous between connected elements and, as will be seen, a different grid
point stress may be output for each of the connected elements. You must exercise
good engineering judgment when you assess the validity of grid point stresses at
exception points.
Topological Method– Plate Elements and Solid Elements

The topological method for the evaluation of grid point stresses only recognizes the
existence of interior, edge, and corner points; i.e., exception points are not
considered as a distinct category of grid point (exception points arise as a result of
geometrical rather than topological characteristics). The relations used to compute
average grid point stresses in surfaces of plate elements by the topological method
are described in Table 12-2. The relations used to compute the average grid point
stresses in volumes containing solid elements are described in Table 12-4. Only the
topological method is available for solid elements.
Table 12-2 Evaluation of Average Grid Point Stresses for Plate and Shell
Elements by the Topological Method
Type of
Average Grid Point Stress Component Remarks
Grid Point
Interior N
e If CQUAD4 and/or
1 CTRIA3 elements
σ I = ------
Ne ∑ σ ei connect to the same
i=1 grid point as CQUAD8
where: and/or CTRIA6
σ I = average stress at interior grid point elements, then only the
CQUAD8 and or
N e = number of directly connected elements CTRIA6 elements are
used in the calculation
 element center stresses if only CQUAD4
 and/or CTRIA3 elements are connected of average grid point
 stresses.
σ ei = 
 element vertex stresses if CQUAD8 and/or
 CTRIA6 are connected

Edge N
e
N
I
If an edge point is not
2 1 connected to an interior
σ E = ------
Ne ∑ σei – N-----I ∑ σIi point by line segments,
i=1 i=1 then N I = 0 , the
where: second term in the
relation σ E is set to
σ E = average stress at edge grid point
zero, and the factor 2 in
N I = number of interior points connected by line the first term is
segments replaced by 1.
Same as remark for
σ Ii = average stress at ineterior grid points
interior grid points.
CHAPTER 12 285
Elements by the Topological Method (continued)
Type of
Average Grid Point Stress Component Remarks
Grid Point
Corner Corner points connected to CQUAD4 element with Same as remark for
the grid point diagonally opposite the corner point, an interior points
interior point, or an edge point.
σ c = 2σ 1 – σ D
where:
σ c = average stress at corner grid point
σ i = CQUAD4 center stress
σ D = average stress at grid point diagonally

opposite corner point
(If point diagonally opposite corner point is not an
interior or edge point, then σ c = σ 1 .
Corner Corner points connected to CTRIA3 elements and the Same as remark for
other two grid points, namely, E1 and E2, of the interior points
connected CTRIA3 are edge points:
σ c = 3σ 1 – ( σ E1 + σ E2 )
where:
σ c = average stress at corner grid point
σ 1 = CTRIA3 center stress
σ E1 , σ E2 = stresses at edge points E1 and E2
(if either E1 or E2 is not an edge point, then σ c = σ 1 )

Corner points connected to one CQUAD4 or one
CTRIA6 element:
σc = σ r
where:
σ r = stress at element vertex connected to corner
point
Corner points connected to two or more CTRIA6
elements:
N
c
1
σ c = ------
Nc ∑ σ ri
i=1
where:
N c = number of CTRIA6 elements connected to
corner point.
Table 12-3 Evaluation of Average Grid Point Stress for Solid Elements by the
Topological Method
Type of
Grid Average Grid Point Stress Component Remarks
Point
Interior and Ne Mixtures of CHEXA,
Edge 1 CPENTA, and CTETRA
σ I = ------
Ne ∑ σei elements may connect to an
i=1 interior or edge grid point.
where:
σ I = average stress at interior or edge grid points
N e = number of directly connected elements
σ ei = element vertex stresses
Corner σ c = σr The same relation is used

whether a CHEXA,
where: CPENTA, or CTETRA
σ r = stress at element vertex connected to corner element is connected to the
grid point.
point
Geometric Method – Surface Elements

In the geometric method, grid point stresses for points which connect to CQUAD8 or
CTRIA6 elements use the average of the element corner stresses. If any CQUAD4 or
CTRIA3 elements connect to these points, their influence is not considered.
The geometric method for the determination of average grid point stresses which
connect only to CQUAD4 or CTRIA3 elements utilizes the numerical technique of
least squares. The “best fit” value of the stress Σ at a point ( x , y ) is functionally
denoted as:
∑ ( x,y ) = A0 + A1 x + A2 y Eq. 12-1
where the coefficients A 0 , A 1 x , and A 2 must be determined. The coordinates x and y

are the locations of points at which known stresses exist. In accordance with the least
squares technique, it is required to minimize the function:
T
1   
f = ---  σ –
2 Σ  σ –
 
Σ  Eq. 12-2
with respect to { A } . Here σ represents the known element stresses.
To perform the minimization, Eq. 12-1 is written in matrix form as:
CHAPTER 12 287
 


Σ  = [T]{A} Eq. 12-3
where:
1 X1 X2
[T] = .. .. ..
. . .
i Xi Xi
and Eq. 12-2 is differentiated with respect to the coefficients A i .
∂f T T
-------- = [ T ] [ T ] { A } – [ T ] { σ } = 0 Eq. 12-4
∂A i
The coefficients { A } are then determined from Eq. 12-4 to be:
T –1 T
{A} = ([T] [T]) [T] {σ} Eq. 12-5
The average grid point stress at a given grid point can then be calculated by
substituting Eq. 12-5 into the matrix representation of Eq. 12-1 as follows:
–1
Σ T
= [ 1 Xo Yo ] ( [ T ] [ T ] )
T
[T] {σ} = [Q]{σ} Eq. 12-6
where:
Q = matrix of interpolation factors

X o Y o = locations of grid points at which stress is required
–1 1
Note that if ( [ T ]T [ T ] ) is singular, the matrix of interpolation factors is taken to be -----
NI
so that Eq. 12-6 becomes:
NI
Σ 1
= ------
NI ∑ σ ei Eq. 12-7
i = 1
where N I = number of known stress points associated with the grid point. In most
cases, Eq. 12-7 is equivalent to the topological interpolation method.
The strategy used in the evaluation of Eq. 12-6 for interior, edge, and corner grid
points is presented in Table 12-4.
As previously mentioned, the coordinates x and y of Eq. 12-1 are the locations of
points at which known stresses exist. These locations are actually taken as the
projected coordinates of the known stress points on the average surface evaluated at
the grid point at which the stress is to be computed. This average surface is
determined to be the surface whose normal is the average of the unit normals to the
elements connected to the grid point of interest. The local x-axis used in the
evaluation of the geometric coefficients of the matrix [ T ] is taken as the projection of
the x-axis of the output coordinate system on the average surface evaluated at the
grid point of interest.
Elements by the Geometric Method
Type of
Grid Number of Elements Involved in Stress Calculation
Point
Interior All directly connected elements
Edge Two directly connected elements:
Use the two known element stresses plus the stresses at all interior
points connected by line segments of the directly connected
elements.
Three or more directly connected elements:
Use the known element stresses of the directly connected elements.
Corner Same as topological method (see Table 12-1).
Treatment of Exception Points – “BRANCHing”

Exception points have been defined as grid points at which stress discontinuities
may occur. Such discontinuities may occur because a substantial difference in slope
exists between adjacent elements directly connected to the same line segment or if
more than two elements are directly connected to the same line segment. Examples
of these two cases are illustrated below.
CHAPTER 12 289
Figure 12-1 Three or More Elements Connected to the Same Line Segment
Figure 12-2 Slope Difference between Adjacent Elements

Connected to the Same Line
Note: That these two situations are geometric rather than topological in
nature and, therefore, are only considered under the geometric
interpolation option.
Since exception points are defined as grid points at which stress discontinuities may
exist, it is conceivable that a computed average grid point stress at an exception
point may be different for each of the directly connected elements. However, it is
more common to find that a subset of the elements that are directly connected to an
exception point forms a subsurface on which stresses may be considered to be
continuous. A unique average grid point stress may then be computed for the
elements
www.cadfamily.com of the subsurface.
The calculation of average grid point stresses for a subsurface that contains two or
more elements is performed with the geometric interpolation method used for edge
points. If a subsurface consists of a single element, the calculation of an average grid
point stress is performed in the same manner as utilized in the calculation of average
grid point stresses at corner points. It must be noted, however, that σD , σ E1 , and σ E2
may be interior, exterior, or corner points (see Table 12-4).
User Information
The use of the grid point stress recovery option is demonstrated in this section
through the consideration of several example problems. These problems are
discussed separately after the presentation of the requirements that must be met in
the Executive Control Section, the Case Control Section, and the Bulk Data Section.
Executive Control Section

A typical Executive Control Section would be of the following form:
SOL 101
CEND
Case Control Section

The option to obtain grid point stress output is exercised through commands in the
Case Control Section. The general request for the recovery of grid point stresses is
accomplished using commands with the form:
GPSTRESS = ALL
or
SET j = k, l, m, . . .
GPSTRESS = j
The k, l, m, . . . reference the identification numbers of the surfaces or volumes of

elements that are defined in the OUTPUT(POST) section of the Case Control Section.
Grid point stress data will be output only for those surfaces or volumes referenced
through the GPSTRESS Case Control command.
Note: If grid point stress data is desired, an ELSTRESS (or STRESS) request
for all elements in the surfaces or volumes of interest must be
included in the Case Control Section. Such requests will cause the
output of element stress data as well as grid point stress data. If the
output of element stress data is not desired, the requests
ELSTRESS(PLOT) = or STRESS(PLOT) = may be used.
CHAPTER 12 291
In static analysis, one may place the GPSTRESS command above the subcase level
or in individual subcases. If the GPSTRESS command is above the subcase level, it
will be utilized by all subcases and subcoms unless overridden by a GPSTRESS
command in a SUBCASE or SUBCOM.
The OUTPUT(POST) section of the Case Control Section contains specific requests
for grid point stress data for the surfaces or volumes of elements. Any number of
element surfaces or volumes may be defined, but only those surfaces or volumes
that are referenced through the GPSTRESS command will have grid point stress
data generated and output.
Element surfaces and volumes are defined through a SURFACE and VOLUME
commands in the OUTPUT(POST) section of the Case Control Section.
Grid Point Stress Output Description

Listing 12-1 through Listing 12-3 are examples of grid point stress output produced
for two-dimensional plate elements requested on the SURFACE Case Control
command. Listing 12-1 is the most common form of output and is always the result
when the TOPOLOGICAL method is used, and when the GEOMETRIC method is
used and no surface “BRANCHing” is detected. Note that when the “ELEMENT
ID” value is zero, it always indicates that all elements in the SURFACE connected to
the grid point contribute to the average grid point stress and lie in the “best”
subsurface. Listing 12-2 and Listing 12-3 show examples of the output when the
GEOMETRIC method is used and when “BRANCHing” occurs. The negative value
for the “ELEMENT ID” indicates the “best” average grid point stress and the
element identification number of an element associated with “best” subsurface.
Similarly a positive element identification numbers indicate another average grid
point stress for a “less best” grid point stress and the element identification number
associated with it’s subsurface. When TOLERANCE is set positive as in Listing 12-2
only the “best” grid point stress is calculated when “BRANCHing” is detected and
when TOLERANCE is set negative all subsurface average grid point stress are
calculated. See “SURFACE” on page 396 of the NX Nastran Quick Reference Guide.
Listing 12-1 Grid Point Stress Output – Typical
SQUARE PLATE WITH CIRCULAR HOLE JULY 23, 2003 NX NASTRAN 7/23/03 PAGE 27
GRID POINT STRESS TEST CASE
UNIFORM LOAD ALONG X=5.0. SUBCASE 1
SUBCASE = 1
S T R E S S E S A T G R I D P O I N T S - - S U R F A C E 4
SURFACE X-AXIS X NORMAL(Z-AXIS) Z REFERENCE COORDINATE SYSTEM FOR SURFACE DEFINITION CID 1
GRID ELEMENT STRESSES IN SURFACE SYSTEM PRINCIPAL STRESSES MAX
ID ID FIBER NORMAL-X NORMAL-Y SHEAR-XY ANGLE MAJOR MINOR SHEAR VON MISES
10304 0 MID 4.965E+03 4.041E+03 -6.431E+03 -42.9450 1.095E+04 -1.944E+03 6.447E+03 1.204E+04
10305 0 MID 6.132E+03 3.839E+03 -5.518E+03 -39.1310 1.062E+04 -6.503E+02 5.636E+03 1.096E+04
10306 0 MID 7.033E+03 3.408E+03 -4.246E+03 -33.4429 9.837E+03 6.037E+02 4.617E+03 9.549E+03
10400 0 MID 6.101E+02 1.631E+04 -1.038E+03 -86.2353 1.638E+04 5.418E+02 7.919E+03 1.612E+04
10401 0 MID 1.463E+03 1.417E+04 -3.402E+03 -75.9185 1.503E+04 6.098E+02 7.209E+03 1.473E+04
10402 0 MID 2.228E+03 1.211E+04 -5.354E+03 -66.3558 1.446E+04 -1.156E+02 7.286E+03 1.452E+04
10403 0 MID 2.809E+03 1.019E+04 -6.381E+03 -60.0211 1.387E+04 -8.721E+02 7.371E+03 1.433E+04
10404 0 MID 3.159E+03 8.514E+03 -6.308E+03 -56.5007 1.269E+04 -1.016E+03 6.852E+03 1.323E+04
10405 0 MID 3.307E+03 7.318E+03 -5.482E+03 -55.0445 1.115E+04 -5.249E+02 5.837E+03 1.142E+04
Listing 12-2 Grid Point Stress Output with “BRANCHing” and TOLERANCE > 0
INTERSECTING PLATES JULY 23, 2003 NX NASTRAN 7/23/2003 PAGE 34

SUBCASE = 1
10304 0 Z1 9.107E+02 1.209E+03 2.799E+03 46.5224 3.863E+03 -1.743E+03 2.803E+03 4.969E+03
Z2 1.052E+03 1.571E+03 2.913E+03 47.5464 4.236E+03 -1.613E+03 2.925E+03 5.233E+03
MID 9.813E+02 1.390E+03 2.856E+03 47.0453 4.049E+03 -1.678E+03 2.863E+03 5.099E+03
10305 0 Z1 1.906E+03 5.699E+02 1.789E+03 34.7594 3.147E+03 -6.714E+02 1.909E+03 3.531E+03
Z2 2.032E+03 5.648E+02 1.934E+03 34.6114 3.367E+03 -7.698E+02 2.068E+03 3.811E+03
MID 1.969E+03 5.674E+02 1.861E+03 34.6824 3.257E+03 -7.206E+02 1.989E+03 3.671E+03
10306 0 Z1 2.771E+03 -1.983E+02 2.678E+02 5.1132 2.795E+03 -2.223E+02 1.508E+03 2.912E+03
Z2 3.024E+03 -1.118E+02 4.889E+02 8.6589 3.098E+03 -1.862E+02 1.642E+03 3.196E+03
MID 2.897E+03 -1.550E+02 3.783E+02 6.9615 2.944E+03 -2.012E+02 1.572E+03 3.049E+03
10400 -119 Z1 -5.775E+04 1.181E+03 -2.060E+04 -72.5215 7.667E+03 -6.423E+04 3.595E+04 6.839E+04
Z2 -5.821E+04 1.207E+04 -1.647E+04 -77.4454 1.573E+04 -6.188E+04 3.880E+04 7.106E+04
MID -5.798E+04 6.623E+03 -1.853E+04 -75.0777 1.156E+04 -6.292E+04 3.724E+04 6.942E+04
10401 -119 Z1 -1.856E+04 7.060E+03 -4.939E+03 -79.4558 7.979E+03 -1.948E+04 1.373E+04 2.446E+04
Z2 -1.714E+04 1.656E+04 -3.881E+03 -83.5131 1.700E+04 -1.758E+04 1.729E+04 2.995E+04
MID -1.785E+04 1.181E+04 -4.410E+03 -81.7177 1.245E+04 -1.849E+04 1.547E+04 2.696E+04
CHAPTER 12 293
Listing 12-3 Grid Point Stress Output with “BRANCHing” and TOLERANCE < 0
INTERSECTING PLATES JULY 23, 2003 NX NASTRAN 7/23/2003 PAGE 40

SUBCASE = 1
10304 0 Z1 9.107E+02 1.209E+03 2.799E+03 46.5224 3.863E+03 -1.743E+03 2.803E+03 4.969E+03
Z2 1.052E+03 1.571E+03 2.913E+03 47.5464 4.236E+03 -1.613E+03 2.925E+03 5.233E+03
MID 9.813E+02 1.390E+03 2.856E+03 47.0453 4.049E+03 -1.678E+03 2.863E+03 5.099E+03
10305 0 Z1 1.906E+03 5.699E+02 1.789E+03 34.7594 3.147E+03 -6.714E+02 1.909E+03 3.531E+03
Z2 2.032E+03 5.648E+02 1.934E+03 34.6114 3.367E+03 -7.698E+02 2.068E+03 3.811E+03
MID 1.969E+03 5.674E+02 1.861E+03 34.6824 3.257E+03 -7.206E+02 1.989E+03 3.671E+03
10306 0 Z1 2.771E+03 -1.983E+02 2.678E+02 5.1132 2.795E+03 -2.223E+02 1.508E+03 2.912E+03
Z2 3.024E+03 -1.118E+02 4.889E+02 8.6589 3.098E+03 -1.862E+02 1.642E+03 3.196E+03
MID 2.897E+03 -1.550E+02 3.783E+02 6.9615 2.944E+03 -2.012E+02 1.572E+03 3.049E+03
10400 -119 Z1 -5.775E+04 1.181E+03 -2.060E+04 -72.5215 7.667E+03 -6.423E+04 3.595E+04 6.839E+04
Z2 -5.821E+04 1.207E+04 -1.647E+04 -77.4454 1.573E+04 -6.188E+04 3.880E+04 7.106E+04
MID -5.798E+04 6.623E+03 -1.853E+04 -75.0777 1.156E+04 -6.292E+04 3.724E+04 6.942E+04
219 Z1 -5.772E+04 1.152E+03 2.064E+04 72.4819 7.667E+03 -6.423E+04 3.595E+04 6.839E+04
Z2 -5.819E+04 1.204E+04 1.652E+04 77.4058 1.573E+04 -6.188E+04 3.880E+04 7.106E+04
MID -5.795E+04 6.597E+03 1.858E+04 75.0381 1.156E+04 -6.292E+04 3.724E+04 6.942E+04
10401 -119 Z1 -1.856E+04 7.060E+03 -4.939E+03 -79.4558 7.979E+03 -1.948E+04 1.373E+04 2.446E+04
Z2 -1.714E+04 1.656E+04 -3.881E+03 -83.5131 1.700E+04 -1.758E+04 1.729E+04 2.995E+04
MID -1.785E+04 1.181E+04 -4.410E+03 -81.7177 1.245E+04 -1.849E+04 1.547E+04 2.696E+04
120 Z1 -1.856E+04 7.060E+03 -4.939E+03 -79.4558 7.979E+03 -1.948E+04 1.373E+04 2.446E+04
Z2 -1.714E+04 1.656E+04 -3.881E+03 -83.5131 1.700E+04 -1.758E+04 1.729E+04 2.995E+04
MID -1.785E+04 1.181E+04 -4.410E+03 -81.7177 1.245E+04 -1.849E+04 1.547E+04 2.696E+04
219 Z1 -1.855E+04 7.053E+03 4.957E+03 79.4163 7.979E+03 -1.948E+04 1.373E+04 2.446E+04
Z2 -1.713E+04 1.655E+04 3.905E+03 83.4735 1.700E+04 -1.758E+04 1.729E+04 2.995E+04
MID -1.784E+04 1.180E+04 4.431E+03 81.6781 1.245E+04 -1.849E+04 1.547E+04 2.696E+04
220 Z1 -1.855E+04 7.053E+03 4.957E+03 79.4163 7.979E+03 -1.948E+04 1.373E+04 2.446E+04
Z2 -1.713E+04 1.655E+04 3.905E+03 83.4735 1.700E+04 -1.758E+04 1.729E+04 2.995E+04
MID -1.784E+04 1.180E+04 4.431E+03 81.6781 1.245E+04 -1.849E+04 1.547E+04 2.696E+04
General Remarks
Each element stress component (independently of all other stress components) is
transformed into the output coordinate system before interpolation. It is necessary
that this output coordinate system vary slowly over the surface for the interpolated
stresses to be accurate. This approach is certainly valid for stress interpolation over
elements that are in a common plane and it should be adequate for engineering
analysis when the difference in slope between adjacent elements is small. However,
when the elements do vary substantially from a smooth surface, the analyst must
utilize engineering judgment relative to the accuracy of the resultant stress data.
Average grid point stresses are only computed at element vertices. For elements
with defined midpoints, you can estimate the stress at these midpoints to be the
average of the computed stresses at the grid point at the ends of the edge in
question.
12.3 Mesh Stress Discontinuities at Grid Points

Error Estimates for Grid Point Stress Data
This section describes the error estimation tools that can help you identify areas of
your model that may require refinement.
The essence of the previous averaging procedures used to obtain grid point stress
data is to:
1. Convert these local element stress components into a common user-defined

coordinate system.
2. Average the several values of each stress component to obtain a unique
value of the stress component that is to be associated with the grid point in
question.
3. Compute the stress invariants at the grid points from the stress components
at the grid points.
In the general case, the stress components are σ x , σ y , σ z , τ xy , τ xz , and τ yz .
For discussion purposes, the averaging process used to compute the stress
components at the grid points can be represented in the form of Eq. 12-8.
N
e
σg = ∑ ( W i σ ei ) Eq. 12-8
i = 1
where:
σ g = weighted mean value of the stress component computed at the grid

point.
σ ei = value of the stress component in the ith element (i = 1,2,..., N e ) in the
neighborhood of the grid point. δgi is in the same coordinate system as
δg .
W i = weighting factor assigned to the ith element. The sum of the Ne values
of Wi must equal 1. (This requirement assures that all computed
statistics will be unbiased.) The attribute of being unbiased implies
that the variance is equal to the mean square error. Equal weighting
(i.e., W i = 1 ⁄ Ne ) is assumed.
CHAPTER 12 295
An estimate of the error in a particular component of stress at a grid point can then
be computed by assuming that the values of the corresponding stress components
computed for the elements in the neighborhood of the grid point are data points
with uncorrelated random errors. It then follows that an estimate of the probable
error in the stress component δg at the grid point is:
N
e
2
δg = ∑ ( W i δ ei )
i = 1
Ne
2
∑ ( δ ei )
1 i = 1
= ----------- ---------------------------
-
Ne Ne
Eq. 12-9
where δei = σei – σg . Thus, the probable error δg is the root mean square error in δ ei
divided by N e .
It should be noted that the standard deviation or root mean square error is a
reasonable measure of precision in many practical cases, but it is easy to provide
examples in which the standard deviation is a poor measure of the concentration of
the distribution about the mean. Eq. 12-9 is assumed to provide an approximate
error estimator for the grid point stress data.
Element and Grid Point Stress Discontinuity Output Requests

It should be clear that element and grid point stress discontinuity output can only
be obtained if the analyst has requested grid point stress output via the GPSTRESS
command in the Case Control Section of the NX Nastran input file. The GPSTRESS
command, of course, requires that the analyst define all required SURFACEs and
VOLUMEs in the OUTPUT(POST) portion of the Case Control Section.
To output Grid Point Stress DisCONtinuities:
GPSDCON = ALL
or
SET j = k, l, m, ....
GPSDCON = j
To output ELement Stress DisCONtinuities:
ELSDCON = ALL
or
SET j = k, l, m, ....
ELSDCON = j
The following remarks should be noted relative to the use of the GPSDCON and/or
ELSDCON commands.
1. The GPSDCON and ELSDCON commands are honored only in SOLution

Sequences 101, 114, 144, and 200.
2. The GPSDCON and ELSDCON commands may be placed above the
SUBCASE level or in individual SUBCASEs and/or SUBCOMs. The use of
these commands above the SUBCASE level causes stress discontinuity data
to be output for ALL SUBCASEs in the NX Nastran input file. If the analyst
wishes to restrict stress discontinuity output to specific SUBCASEs,
GPSDCON and/or ELSDCON commands should appear only under these
specific SUBCASEs.
3. The analyst is cautioned that these commands produce a substantial
amount of output.
4. The GPSDCON and/or ELSDCON commands may be effectively used in
restarts. For example, one could include requests for GPSTRESS data
during the initial run of a project and then obtain data on stress
discontinuities on a subsequent restart by including GPSDCON and/or
ELSDCON commands in the Case Control Section.
5. Stress discontinuity output will only be provided when grid point stresses
have been previously computed by the methods described in this article.
Thus, stress continuity will not be computed when, for example, both plate
and solid elements are connected to a grid point that is involved in stress
discontinuity calculations.
Discussion of Error Measures

Large values of error estimators for CQUAD4 and CTRIA3 elements aren’t
uncommon occurrences in statistical error measures. For example, the mean and
variance have direct analogs in engineering mechanics. The mean is analogous to the
centroid of a body and is thus a measure of where the mass is centered. The variance
is the second moment about the mean, and it tends to be small if the majority of the
mass is concentrated about the centroid. As the mass is dispersed further from the
centroid, the moment of inertia tends to increase. If the mass is concentrated at the
centroid (as it might be in some idealized lumped mass models), the moment of
inertia becomes zero.
CHAPTER 12 297
Thus, the error estimates for CQUAD4 and CTRIA3 elements will tend toward
larger values because the data used to compute the estimators are dispersed
relatively far from the mean. For those elements that provide stress data at vertices,
the estimators will tend toward smaller values because the stress data tend to be
concentrated about the mean in well-designed finite element models.
Generally, the goal is to design a finite element mesh for static analysis so that all
important stress gradients will be adequately represented. If the mesh is not
sufficiently detailed, the stress data at element vertices or element centroids will
result in the relatively large values of the error estimators.
Very inaccurate values of these error estimators may be obtained at the edges of
defined SURFACEs and on the faces of defined VOLUMEs. Such edges and
surfaces will often correspond with the boundaries of superelements.
In summary, error estimators can, in some cases, be highly inaccurate.

Nevertheless, error estimate data is quite useful when properly interpreted by the
analyst.
Grid Point Stresses and Mesh Stress Discontinuities

Grid point stresses may be requested for grid points connected by CTRIA3, CTRIA6,
CTRIAR, CQUAD4, CQUAD8, CQUADR, CHEXA, CPENTA, and CTETRA
elements (with linear material properties only). In linear static analysis only, a mesh
stress discontinuity error analysis may then be performed based on the computed
element and grid point stresses.
Grid point stresses are not computed for laminar stresses.
The grid point stresses are calculated in the user-specified coordinate system. The
following real membrane stresses are output on request:
• Normal stresses in the x and y directions.
• Shear stresses on the x face in the y direction.
• Angle between the x axis and the major principal axis.
• Major and minor principal stresses.
• von Mises equivalent or maximum shear stress (see STRESS Case Control
command).
The following real stresses are output on request for grid points connected to solid
elements:
• The three normal stresses.

• The three shear stresses.
• Magnitude and direction of the three principal stresses.

• The mean pressure.
• The von Mises equivalent or octahedral stress (see STRESS Case Control
command).
Only real stresses are available at the grid points. The user has the following options:
• The capability to vary output requests by SUBCASE.
• Choice of elements in stress field.
• Choice of topological or geometric interpolation (surface fields only).
• Choice of output coordinate system.
• Choice of output stress fiber locations (surface fields only).
The grid point stresses are requested in the Case Control Section for printing and
postprocessing using the Case Control commands GPSTRESS (Grid Point Stress) and
STRFIELD (Stress Field). The user is required to request element stress output
(STRESS = N) for all elements referenced by selected SURFACE and VOLUME
commands. The grid point stress requests a set of surface or volume fields which are
defined in the OUTPUT(POST) part of the Case Control Section on the SURFACE or
VOLUME commands, respectively. The postprocessing output requests are
separated from case control and plot requests by the OUTPUT(POST) command.
The general rules for Case Control syntax are the same as for plotting. The grid point
stress output is defined on the SURFACE and VOLUME commands. The SURFACE
command defines the following: output coordinate system, list of elements in the
stress surface field, output stress fiber location for stress evaluation and the choice of
topological or geometric interpolation method used to determine the grid point
stresses from the elements. The VOLUME command defines the following: output
coordinate system, list of elements in the stress volume, and the choice of DIRECT
and/or PRINCIPAL stresses.
In the linear static structured solution sequences only (SOLutions 101, 114, 144, and
200) mesh stress discontinuities may be requested to estimate errors due to the
coarseness of the finite element mesh. The estimation of the error is computed for the
element and grid point stresses and requested by the ELSDCON and GPSDCON
Case Control commands, respectively. These commands select the desired surfaces
and volumes which are defined in the OUTPUT(POST) section of the case control.
This output option also requires a corresponding GPSTRESS or STRFIELD request
for the desired surfaces and volumes and a STRESS case control request for the
elements in those surfaces and volumes.
The following is computed and output for stress discontinuities on shell elements:
CHAPTER 12 299
• Normal stress discontinuities in the x and y directions.

• Shear stress discontinuities on the xy direction.
• Major and minor principal stress discontinuities.
• von Mises and maximum shear stress discontinuities.
• Error estimate.
The following is computed and output for stress discontinuities on solid elements:
• Normal stress discontinuities in the x, y, and z directions.

• Shear stress discontinuities on the xy, yz, and xz directions.
• Major and minor principal stress discontinuities.
• von Mises stress and mean pressure discontinuities.
• Error estimate.
13
CHAPTER
Solution Sequences
■ Understanding Solution Sequences
■ Data Processing and Matrix Operations by Functional Module
■ Summary of Solution Sequence Operations
13.1 Understanding Solution Sequences

NX Nastran is comprised of a large number of building blocks called modules. A
module is a collection of FORTRAN subroutines designed to perform a specific task,
such as processing model geometry, assembling matrices, applying constraints,
solving matrices, calculating output quantities, conversing with the database,
printing the solution, and so on. The modules are controlled by an internal language
called the Direct Matrix Abstraction Program (DMAP).
Each type of analysis available in NX Nastran is called a solution sequence. Each

solution sequence is a prepackaged collection of hundreds or thousands of DMAP
commands that you select with a SOL Executive Control statement. Once you select
a solution sequence, the software automatically sends that sequence’s particular set
of DMAP commands to the modules that are needed to perform the requested
solution.
The major type is called a Solution Sequence. They are called with numbers ranging
from 100 to 200. The solution sequences are listed in Table 13-1.
Note: You can also select a solution sequence by using the SOL keyword followed by
the name of the solution sequence. For example, SOL SESTATIC is equivalent to
SOL 101. Using the solution numbers is generally recommended rather than their
names because the solution numbers are more easily recognized.
CHAPTER 13 303
Solution Sequences
Solution Sequence
SESTATIC
DMAP Introduction SubDMAP PHASE1DR Matrix Operation Functional Module

DMAP (Accessible)
SubDMAP PHASE1A S S
Solution Sequence Operations

SubDMAP . . . SubDMAP SEMG
General Operations
Functional Module
EMG
(Not Accessible)
Source
S S S
Figure 13-1 Hierarchy in the Structural Solution Sequences
Table 13-1 Structured Solution Sequences
SOL
SOL
Numb Description References
Name
er
101 SESTATIC Statics with options: page 324
Linear Heat Transfer page 369
Alternate Reduction
Inertia Relief
Design Sensitivity – Statics
103 SEMODES Normal Modes with option: page 353
Design Sensitivity – Modes
105 SEBUCKL Buckling with options: page 369
Static Analysis
Design Sensitivity –
Buckling
106 NLSTATIC Nonlinear Statics See the NX Nastran Basic
Nonlinear Analysis User’s
Guide
107 SEDCEIG Direct Complex Eigenvalues page 367
108 SEDFREQ Direct Frequency Response page 368
109 SEDTRAN Direct Transient Response page 369
110 SEMCEIG Modal Complex Eigenvalues page 367
111 SEMFREQ Modal Frequency Response page 368
112 SEMTRAN Modal Transient Response page 369
114 CYCSTATX Cyclic Statics with option: page 693
Alternate Reduction
115 CYCMODE Cyclic Normal Modes page 693
118 CYCFREQ Cyclic Direct Frequency page 693
Response
129 NLTRAN Nonlinear Transient Response See the NX Nastran Basic
Nonlinear Analysis User’s
Guide
144 AESTAT Static Aeroelastic Response See the NX Nastran
Aeroelastic Analysis
User’s Guide
145 SEFLUTTR Aerodynamic Flutter See the NX Nastran
User’s Guide
146 SEAERO Aeroelastic Response See the NX Nastran
www.cadfamily.com EMail:[email protected] User’s Guide
CHAPTER 13 305
Solution Sequences
Table 13-1 Structured Solution Sequences (continued)
SOL
SOL
Numb Description References
Name
er
153 NLSCSH Static Structural and/or page 387
Steady State Heat Transfer
Analysis
159 NLTCSH Transient Structural and/or page 387
Transient Heat Transfer
Analysis
190 DBTRANS Database Transfer page 531
200 DESOPT Design Optimization See the NX Nastran
Design Sensitivity and
Optimization User’s
Guide
Obtaining DMAP Listings for Solution Sequences

You can obtain the DMAP listings for any of the solutions sequences by including a
COMPILER LIST statement in the Executive Control Section.
• A description of the data processing and matrix operations for the solution
sequences is given in “Summary of Solution Sequence Operations” on
page 345.
• The complete details for each of the functional modules are given in
Chapter 4 of the NX Nastran Programmer’s Manual.
• Those DMAP modules which are of general use (matrix operations and
utility modules) are described along with examples of their use in “Data
Processing and Matrix Operations by Functional Module” on page 308.
Using Alters to Modify Solution Sequences

If you want to modify a DMAP sequence, you can use the alter feature described in
NX Nastran DMAP Programmer’s Guide. For example, you can use an alter to:
• schedule an EXIT prior to completion in order to check intermediate

output
• schedule the printing of a table or matrix for diagnostic purposes
• delete or add a functional module to the DMAP sequence
To make alterations to a solution sequence, you should be familiar with the rules for
DMAP programming
A library of DMAP alters is delivered with NX Nastran. You can use these alters to
request options not available in the solution sequences. You can also use them to
perform exceptional operations that do not fit in with the normal flow of the solution
sequences.
Understanding Scheduled and Unscheduled Exits

Scheduled exits can be requested at any point in a structured solution sequence
format by means of the SEMG, SEKR, SELG, SELR, SEMR, and SEALL Case Control
commands. If none of these commands are specified, then SEALL = ALL is the
default.
Unscheduled exits are usually caused by errors on input entries or errors in the
structural model resulting from missing or inconsistent input data. When such
errors are detected, an unscheduled exit is performed accompanied with the output
of the applicable user error messages. Following the correction of the input data
errors, a modified restart can be performed.
Unscheduled exits may also occur because of machine failure or insufficient time
allowance. In these cases, an unmodified restart is usually made. In some cases,
where a portion of the problem has been completed, including the output for the
completed portion, a modified restart must be made following an unscheduled exit
due to insufficient time allowance. The situations are discussed under case control
requirements in the sections dealing with the individual solution sequences.
The initial execution of any problem must be made with complete NX Nastran input
data, including all of the bulk data. However, all or part of the bulk data may be
assembled from alternate input sources.
Restart Procedures
NX Nastran also has an automatic restart capability. The automatic restart is not
only driven by modifications to the Case Control and Bulk Data Sections but also by
modifications in upstream superelements.
See Also

In NX Nastran, restarts are accomplished with the NASTRAN Data Definition
Language feature, a file that lists the names and attributes of data types saved for
restart. The evidence that a Solution Sequence has restart capability can be found by
the existence of TYPE statements in its DMAP compile listings. It uses the SubDMAP
feature, which allow breaking a large collection of DMAP statements into more
manageable data structures similar to subroutines in other programming languages.
CHAPTER 13 307
Solution Sequences
For a restart analysis, the Bulk Data Section consists only of delete “/” entries and
any new entries you want to add. The previous Bulk Data is read from the database.
All other parts of the NX Nastran Input Data (including the File Management
statements, the Executive Control statements, the Case Control commands, the
BEGIN BULK command and the ENDDATA entry) must be resubmitted even
though no changes are made in the case control and no new bulk data is added.
When changing solution sequences, the solution number (SOL) must be changed to
the number or name of the new solution sequence.
Any changes in the Case Control Section associated with Bulk Data entry selection
or subcase definition, or changes in the Bulk Data Section.
Restarts always start at the beginning of a specified DMAP or SOL sequence. In a

restart analysis, NX Nastran queries the database in two phases.
Phase 1: The software marks all appropriate existing database data blocks
and parameters as existing for the current restart run.
Phase 2: The software checks the current input against the version from
which the restart initiates and deletes any data blocks with
modified input from the Phase 1 determined database data blocks
and parameters. This phase is performed by the RESTART
module(s) contained within the solution’s DMAP.
Restart only executes DMAP Modules for which some or all of the output does not
exist on the current version database. The S-type parameter is considered as an
output data block for restart purposes, hence if it is not listed in the NDDL, then the
module that contains it will be re-executed. All DMAP modules are executed until
the first RESTART module after which output checking is performed. Forced
execution of modules after this point may be manually controlled by SYSTEM (109)
flag.
If no such changes are made, the executive system performs an unmodified restart.
If changes have been made only in the output requests, the restart is considered
unmodified. However, some modules may have to be re-executed in order to
prepare the output. Output requests that were satisfied, plots made, etc., should be
removed if they are not to be output again.
For modified restarts, a number of previously executed DMAP instructions may

have to be re-executed, depending on the nature of the modifications made by the
user. The DMAP instructions that need to be executed in a modified restart are
automatically determined within the program by comparing all changes made in
Case Control commands and Bulk Data entries.
13.2 Data Processing and Matrix Operations by

Functional Module
The following sections will refer specifically to the structural analysis matrices such
as stiffness, damping, and mass. However, this information also applies to heat
transfer analysis for conductance and capacitance. There is no analogy for mass in
heat transfer analysis.
See Also
• NX Nastran DMAP Programmer’s Guide (for additional information

regarding the use of DMAP modules for matrix operations)
• Chapter 4 or the NX Nastran Programmers Manual contains detailed
descriptions of the individual functional modules.
Geometry Processing in SubDMAP PHASEO

NX Nastran begins the initial processing of your total model in PHASEO. In
PHASEO, the software takes data from the input files and stores it in data structures
that feed more expensive processes in later SubDMAPS. Flags are set to control
branching for special features. All superelement inputs are processed. Tables are
indexed for faster processing in post-processors. Parameters are read from Case
Control and Bulk Data files by module PVT. The shell element normals at each grid
point are computed in module TASNP2, and output in table GPSNT.
Tables for internal program use are generated by the geometry processors from the
grid point entries, coordinate system definition entries, sequence entries, connection
entries, static load entries, temperature definition entries, and constraint entries. The
computer time for all geometry processing is usually only a small part of the time
required for problem solution.
The basic geometry processing is performed by GP1 (Geometry Processor – Phase 1).
The software assembles a list of all grid points and scalar points in internal order for
use in relating internal and external identification numbers. Matrices for
transformation from basic to all defined global coordinate systems are computed,
and all grid point locations are transformed to the basic coordinate system.
The element connection tables (ECT) are generated by GP2 (Geometry Processor –
Phase 2) from the connection entries. The external grid point numbers are converted
to the internal indices defined in GP1.
CHAPTER 13 309
Solution Sequences
The temperature data and static loads are processed by GP3 (Geometry Processor –
Phase 3). The temperature data is collected by sets and the external grid point
numbers. Element numbers are converted to the internal indices defined in GP1 and
GP2. In the case of static analysis, the static loads are also collected by sets and the
external identification numbers are converted to internal indices.
The constraint data is processed by GP4 (Geometry Processor – Phase 4). The
displacement set definition table (USET) is generated from the constraint definition
entries. USET contains one coded word for each degree-of-freedom in the model to
identify the set or sets of coordinates to which the degree-of-freedom belongs (see
“Understanding Sets and Matrix Operations” on page 243 for the definition of
degree-of-freedom sets). The multipoint constraint matrix [ R mg ] is formed from the
MPC and rigid element Bulk Data entries. In static analysis, the enforced
displacement matrix [ Y s ] is formed for use in the generation of equivalent static
loads.
The BNDSPC module moves SPCs on boundary grid points to downstream

superelements.
Model Checkout
A checkout procedure is provided for the development of new models. It includes
the bookkeeping and control blocks, and part of the generation block of Figure 13-3.
It is requested by the PARAM,CHECKOUT,YES option.
See Also
• “Superelement Analysis” on page 371 of the NX Nastran User’s Guide.
SEP1 Module
Generates the SEMAP table, which lists superelement membership of grid points
and elements. Provides estimates for computation costs and storage requirements.
This module will identify any structural partitioning errors.
There are two paths through the superelement capability. If the bulk data entries
have a BEGIN SUPER and/or BEGIN SUPER = [SEID] entries the data flow goes
through the partitioned superelement capability modules, SEP1X and SEP2X. If
only a BEGIN BULK entry is present the grid point list capability modules SEP1 and
SEP2 are used. The SEP1X path also allows grid point list superelements, with the
exception of fluid structure analysis, SOL 200 (Design Optimization), and the
RELEASE entry. Models using these features must use the grid point list path.
See Also
• “Superelement Analysis” on page 371 of the NX Nastran User’s Guide.
SEP2, SEP2CT Modules

Partitions out the data for any superelement selected by SE-type Case Control
commands. These modules identify load and rigid element partitioning errors, and
errors in the Case Control Section.
Constraint Modules
The GP1 and GP4 modules process all selected superelements. They check for the
presence of requested coordinate systems; for illegal duplicate specifications of
mutually exclusive sets, such as SPCi and ASETi entries referencing the same point;
and for correct specifications on SPCi, xSETi, and MPCi entries, and rigid elements.
You may print the user set table with PARAM,USETPRT, when
PARAM,CHECKOUT,YES is present. If rigid elements or selected MPC entries exist,
the constraint matrix [ R mg ] (see “Multipoint Constraint Operations in SubDMAPs
SEKR and SEMR2 and SEMRB” on page 313) is given three tests:
1. A matrix of rigid body vectors [ u ogh ] is calculated by the VECPLOT module.

This matrix includes zeroes for the rows corresponding to scalar points.
The product
o
[ E mh ] ← [ R mg ] [ u gh ]
is calculated. Terms of [ E mh ] larger than PARAM,TINY are printed. If the

constraint matrix contains internal constraints, nonzero terms will exist in
[ E mh ] . If an MPC equation includes scalar points, nonzero terms may exist
in the corresponding row of [ E mh ] although the equation may be in
equilibrium.
2. The product
g T
[ R mm ] ← [ R mg ] [ R mg ]
is calculated. The product matrix is decomposed by the DCMP module.

Diagonal terms of the factor matrix that are larger than
PARAM,MAXRATIO are printed. These terms indicate that the printed
row is linearly dependent on prior rows. Unless the constraint matrix is
changed, it is likely that the model will encounter fatal errors due to
singularity in the MCE1 module, or will have a poorly-conditioned
constraint elimination process.
3. The product
m T
[ R mm ] ← [ R mm ] [ R mm ]
CHAPTER 13 311
Solution Sequences
is calculated, where [ R mm ] is the partition of [ R mg ] described in Section .

This product is also processed by the DCMP module. If a row appears
here that also appears in the messages from the prior step, it confirms that
the constraint matrix is inherently ill-conditioned. If a row appears here
that did not appear in the prior step, it indicates that a change in the m-set
exists that can also make the [ R mm ] matrix well conditioned.
The DBC module stores data for use in postprocessors. It runs at many points in the
Solution Sequences. In this SubDMAP it stores basic structural geometric data. The
AEROE SubDMAP generates the geometry of the aerodynamic model. The
Automated Component Mode Synthesis (ACMS) capability controls branches at
many places in the Solution Sequences. The RESTART module controls the modules
that will be run on restart, based on changes in the input files. The run exits after
these checks are performed.
Matrix Assembly Operations in SubDMAP SEMG

The stiffness, mass, damping, differential stiffness, and heat conduction matrices
are formed during the matrix assembly operations in subDMAP SEMG.
Stiffness is generated from the following sources:
• Structural elements defined on connection entries (e.g., CBAR and

CROD).
• General elements defined on GENEL entries.
• Scalar springs defined on CELASi entries.
Mass is generated from the following sources:
• A 6 × 6 matrix of mass coefficients at a grid point defined on a CONM1

entry.
• A concentrated mass element defined on a CONM2 entry in terms of its
mass and moments of inertia about its center of gravity.
• Structural mass for all elements, except plate elements without membrane
stiffness, using the mass density on the material definition entry.
• Nonstructural mass for all elements specifying a value on the property
entry.
• Scalar masses defined on CMASSi entries.
Damping is generated from the following sources:
• Viscous rod elements defined on CVISC entries.

• Scalar viscous dampers defined on CDAMPi entries.
• Element structural damping by multiplying the stiffness matrix of an

individual structural element by a damping factor obtained from the
material properties (MATi) entry for the element.
These matrices are formed from the connection and geometric tables prepared by the
geometry processors and the material and element property tables prepared in the
Phase 0. The matrix assembly operations will usually represent a significant portion
of the total solution time. The computer time for matrix assembly operations is linear
with the number and type of elements. Unit times for stiffness matrix assembly are
given in the NX Nastran Installation and Operations Guide. Assembly times for
coupled mass matrices, differential stiffness matrices, and heat conduction matrices
are about the same as those for stiffness matrices. Assembly times for lumped mass
matrices and damping matrices are usually relatively small.
The matrix assembly process begins with the generation of various tables in TA1
(Table Assembler) from the geometric data, connection data and property data.
These tables are used in data recovery operations as well as in the matrix assembly
operations.
The matrix assembly operations are performed by EMG (Element Matrix Generator)
and EMA (Element Matrix Assembler) modules. All of the element matrices (KELM,
KDICT, MELM, MDICT, BELM, and BDICT) are generated by EMG. The stiffness
matrix exclusive of general elements, [ Kxgg ] , as well as the structural damping matrix,
4
[ K gg ] , the mass matrix, [ M gg ] , and the viscous damping matrix, [ B gg ] , are assembled
in separate executions of EMA.
The parameter COUPMASS is used to select either coupled mass or lumped mass
generation in EMG. If the parameter WTMASS is present, the mass matrices
generated in EMG are multiplied term by term by the value in field 3 of the PARAM
entry. This operation allows you to specify mass data in units of weight.
The structural damping matrix [ K4gg ] is assembled from elements with structural
damping. Structural damping is specified in the GE field of the MATi entry.
NE
4
[ K gg ] = ∑ GE e [ K e ] Eq. 13-1
e = 1
where:
NE = number of elements
GE e = element structural damping coefficient from MATi entries
[ K e ] = element stiffness matrix
CHAPTER 13 313
Solution Sequences
The direct matrix inputs requested by the K2GG, M2GG, and B2GG Case Control
commands defined the stiffness matrix [ K2gg ] , mass matrix [ M 2gg ] , and damping
matrix [ B 2gg ] , and are assembled by the MTRXIN module. The outputs from EMA
and MTRXIN are combined by the ADD module.
The stiffness contributions from the general elements (GENEL Bulk Data entry) are
assembled in SMA3 (Structural Matrix Assembler – Phase 3). The final stiffness
matrix [ K gg ] is formed by combining [ K xgg ] prepared by the EMA module with the
contributions from general elements generated by SMA3.
Weight and balance information is calculated from the mass matrix [ Mgg ] by GPWG
(Grid Point Weight Generator). The execution of GPWG is controlled by the
parameter GRDPNT. The details of the output are described in “GRDPNT” on
page 603 of the NX Nastran Quick Reference Guide.
Multipoint Constraint Operations in SubDMAPs SEKR

and SEMR2 and SEMRB
The multipoint constraint operations apply constraints to the stiffness matrices
using the linear relationships among the displacements you specify on MPC,
MPCAX, and rigid element Bulk Data entries.
The multipoint constraint equations are initially expressed in the form,
[ Rg ] { ug } = 0 , Eq. 13-2
where you supply the coefficients. You also specify the degree-of-freedom that is
made dependent by each equation of constraint, so that the { u g } matrix may
immediately be partitioned into two subsets,
 un 
{ ug } =   , Eq. 13-3
 um 
where the set, u m , is the set of dependent degrees-of-freedom. The matrix of

constraint coefficients is similarly partitioned
[ Rg ] = [ Rn Rm ] , Eq. 13-4
so that Eq. 13-2 becomes
[ Rn ] { un } + [ Rm ] { um } = 0 . Eq. 13-5
is a nonsingular matrix. We can, therefore, form the multipoint constraint

[ Rm ]
matrix,
–1
[ Gm ] = –[ Rm ] [ Rn ] , Eq. 13-6
so that Eq. 13-4 may be stated as
{ um } = [ Gm ] { un } . Eq. 13-7
Prior to the imposition of constraints, the structural problem may be written as
[ K gg ] { u g } = { P g } , Eq. 13-8
or, partitioning in terms of the coordinate sets, un and u m
K nn K nm  u n   Pn 
  =   Eq. 13-9
K nm K mm  u m 
T  Pm 
Bars over symbols are used to designate arrays that are replaced in the reduction
process.
The addition of constraints to the structure requires that the forces of constraint be
added to the equilibrium equations. Thus, writing the equilibrium and constraint
equations together in partitioned form,
T    
K nn K nm G m  u n   Pn 
   
=  P , Eq. 13-10
K nm K mm – I  m 
 u
T
 m 
Gm –I 0  q m   0 
 
where { q m } is the vector of constraint forces on { u m } . Straightforward elimination of

u m and q m gives
T T T T
[ K nn + K nm G m + G m K mm + G m K mm G m ] { u n } = { P n } + [ G m ] { P m } , Eq. 13-11
or
[ K nn ] { u n } = { P n } , Eq. 13-12
where
T T T
K nn = K nn + K nm G m + G m K nm + G m K mm G m , Eq. 13-13
and
T
Pn = Pn + Gm Pm . Eq. 13-14
CHAPTER 13 315
Solution Sequences
The initial partition of K gg and the operations indicated by Eq. 13-6, Eq. 13-13 and
Eq. 13-14 are performed by appropriate modules of the program. The multipoint
constraint matrix, G m , is used in structural matrix reduction (Eq. 13-13), load vector
reduction, (Eq. 13-14) and data recovery (Eq. 13-7).
The multipoint constraint equations, R mg , formed in the GP4 module, are

partitioned by the MCE1 module (Multipoint Constraint Eliminator – Phase 1) as
follows:
[ R mg ] = R mm R mn Eq. 13-15
MCE1 also solves the equation
[ R mm ] [ G mn ] = – [ R mn ] Eq. 13-16
for the transformation matrix [ G mn ] . An alternate path is available for module

MCE1 in SubDMAP XMCE1, requested by PARAM,MCE1,NEW. It contains a
1
partitioned solution for R –mm ⋅ R mn , where the part of R mm that is diagonal is
partitioned from the coupled part. It was developed at a time when the partitioned
solution was faster than the single-pass solution used in MCE1. The single pass
solution was improved so that it is now faster than the partitioned solution. The
partitioned solution is maintained to provide an alternate solution when the single-
pass solution is implausible, or does not complete in a normal manner.
The MCE2 module (Multipoint Constraint Eliminator – Phase 2) partitions the

stiffness matrix
K nn K nm
[ K gg ] = Eq. 13-17
K mn K mm
and performs matrix reduction
T T
[ K nn ] = [ G mn ] [ K mm G mn + K mn ] + [ K mn G mn + K nn ] Eq. 13-18
The other structural matrices, [ K4gg ] , [ B gg ] and [ Mgg ] , are transformed by formulas
that are identical in form to Eq. 13-13, in subDMAPs SEMR2 and SEMRB.
The MCE2 module is also used in subDMAP SEMR2 to obtain [ M nn ] .
The UPARTN module partitions [ Kgg ] to [ Kmg ] form which is equivalent to
[ K mg ] = [ K mn K mm ] Eq. 13-19
and is used to obtain multipoint forces of constraint. See “Data Recovery

Operations in SubDMAP SEDISP” on page 338.
Single Point Constraint Operations in SubDMAPs SEKR,

SEMR2 and SEMRB
The single point constraint operations apply constraints to the structural matrices
using the single point constraints you specify on SPCi Bulk Data entries, the PS field
on GRID entries, or constrained by the AUTOSPC feature described in “Constraint
and Mechanism Problem Identification in SubDMAP SEKR” on page 317.
Single point constraints are applied to the set, us , in the form
{ us } = { Ys } , Eq. 13-20
where { Y s } is a vector of enforced deformations, any or all of whose elements may

be zero. The set, u n , is partitioned into us and uf (the free or unconstrained set)
 uf 
{ un } =   Eq. 13-21
 us 
The stiffness matrix, Knn , is similarly partitioned
K ff K fs
[ K nn ] = Eq. 13-22
T
K fs K ss
The complete structural equations including the single point forces of constraint, q s ,
may be written in partitioned matrix form as
   
K ff K fs 0  u f   Pf 
   
K fs K ss – I  u s  =  P s 
T Eq. 13-23
   
0 I 0  q s   Y s 
   
Straightforward elimination gives
[ K ff ] { u f } = { P f } – [ K fs ] { Y s } = { P f } Eq. 13-24
The forces of constraint are recovered by means of the middle row of Eq. 13-23, i.e.,
T
{ q s } = – { P s } + [ K fs ] { u f } + [ K ss ] { u s } Eq. 13-25
The distinct partitions Kfs and K ss , are needed in subsequent calculations of SPC
forces in static analysis. For the other structural matrices ( K 4nn , B nn , and M nn ) the (sf)
and (ss) partitions are saved, for dynamic SPC forces calculation.
CHAPTER 13 317
Solution Sequences
The partially constrained stiffness matrix [ K nn ] is partitioned by UPARTN module

as follows:
K ff K fs
[ K nn ] = Eq. 13-26
K sf K ss
Similar partitioning operations are performed on [ Mnn ] in subDMAP SEMR2 to

obtain [ M ff ] , [ Msf ] and Mss . In subDMAP SEMRB, the structural and viscous
damping matrices [ K 4sf ] , [ K 4ss ] , [ B sf ] and [ B ss ] are obtained from the MATREDU
module. All (sf) and (ss) matrices are required for the recovery of single point
constraint reaction forces described in “Data Recovery Operations in SubDMAP
SEDISP” on page 338.
Constraint and Mechanism Problem Identification in

SubDMAP SEKR
A singularity in a linear equation solution leads to a condition where a unique
solution is not possible. For example, if [ K ] { u } = { F } , and [ K ] is zero, then if
{ F } is not zero, no value of { u } will satisfy the equation. However, if { F } is zero,
any value of { u } will satisfy the equation. The equation can always be satisfied by
prescribing { u } , rather than { F } . This is equivalent to constraining u with a
single-point constraint.
There are two types of singularity:
1. The singularity can be identified by considering the stiffness terms of only

one grid point. This is called a grid point singularity.
2. The singularity requires consideration of stiffness terms of more than one
grid point. This is called a mechanism type of singularity.
Grid Point Singularity

Grid point singularities are automatically identified by the GPSP module. They
may be automatically constrained, under user control. First, multipoint constraints
are eliminated, leading to the matrix [ K nn ] (see “Multipoint Constraint Operations
in SubDMAPs SEKR and SEMR2 and SEMRB” on page 313). Singularities are
identified by grid point and component number of the failed direction. The ratio of
stiffness at the degree-of-freedom to the largest stiffness at that grid point is listed.
If automatic constraint is selected (see “AUTOSPC” on page 570 of the NX Nastran
Quick Reference Guide), the set membership of the singular degrees-of-freedom is
listed before the automatic constraint and after the automatic constraint. Both the
mutually exclusive set and the highest combined set identical to the mutually
exclusive set are listed for each identified singularity.
The output
www.cadfamily.com will include a Grid Point Singularity Table with the following format:
POINT TYPE FAILED STIFFNESS OLD NEW

ID DIRECTION RATIO USET USET
3801 G 1 0.00E+00 L S
*
3801 G 2 0.00E+00 L S
*
3801 G 3 0.00E+00 L S
*
3801 G 4 0.00E+00 L S
*
3801 G 5 0.00E+00 L S
*
3801 G 6 0.00E+00 L S
*
In this table,
1. G (under the heading TYPE) stands for grid point (S for scalar point is also
output).
2. FAILED DIRECTION is the degree-of-freedom component in the
displacement coordinate system and closest to the weakest direction.
3. The STIFFNESS RATIO is the ratio of the stiffness in the weakest direction
to the stiffness in the strongest direction at the grid point (considering all
possible directions rather than just the coordinate directions).
Translational and rotational stiffnesses are considered separately. The
default value for the failure criterion is 10– 8 , which may be changed by the
PARAM,EPZERO entry.
4. Note that, for the example shown, the failed degrees-of-freedom were
originally in the ul set and were placed in the u s set. This may not always
be the case, for reasons noted earlier, or because you didn’t request the
AUTOSPC feature.
5. You should carefully inspect the Grid Point Singularity Table because a
singularity is very often the result of a modeling error, which should be
corrected.
CHAPTER 13 319
Solution Sequences
User Options
The identification of singularities is an automatic feature; however, you may
exercise the following options through PARAM Bulk Data entries described in
“Parameters” on page 563 of the NX Nastran Quick Reference Guide.
Parameter
Description
Name
AUTOSPC Controls automatic constraint of a singularity.
EPPRT Specifies the maximum value of a singularity to print.
EPZERO Specifies the maximum value of a singularity.
PRGPST Controls printout of singularity table.
SPCGEN Writes SPC entries to the PUNCH file based on the singularity table.
The singular degrees-of-freedom are shifted to other sets by the following logic if
PARAM,AUTOSPC,YES is specified:
o-set Set Before Set After

Condition Constraint Constraint
If o-set exists o s
c b
If no o-set c s
b s
Other mutually exclusive sets not listed above will not be constrained. In addition,
singular degrees-of-freedom on CYAX, CYJOIN, and SPCOFFi entries are
identified but not constrained. The most usual cause of grid point singularity is
degrees-of-freedom not constrained by elements because they are not of interest to
the modeler. An example is out-of-plane rotations on a component made of plates
lying in a plane. The automatic constraint provides the correct action for this type
of singularity. Another cause is elements left out by oversight. Here, one typically
observes that many or all degrees-of-freedom at a grid point are singular. The
automatic constraint may mask a modeling error in this instance. For this reason,
all singularity messages should be carefully inspected during the modeling
checkout phase to ensure that modeling errors are not being masked.
When using the K2PP, M2PP, B2PP, or TFL Case Control command,
PARAM,AUTOSPC,NO is recommended.
Mechanism-Type Singularities
A more complicated type of instability arises when two or more grid points
participate in a singularity in a linear equation solution. This is known as a
mechanism type of instability. Again, for the simple example shown in Figure 13-2,
the equations of equilibrium can not be satisfied at all, or may have nonunique
solutions for special loading conditions. Note, however, that the structure of this
example will pass the grid point singularity checks.
u1 u2
Figure 13-2 A Mechanism Type of Instability
The static equilibrium equation for the example in Figure 13-2 is
u P
k –k  1  =  1  Eq. 13-27
   
–k k  u2   P2 
If P 1 = – P 2 , many solutions are possible, where ( u 1 – u2 ) = P 1 ⁄ k , but u 1 may have any

value. If P 1 and P 2 are arbitrary, the equation cannot be satisfied.
This condition is detected during the matrix decomposition phase of the linear
equation solution process by the DCMP module. The stiffness matrix is decomposed
into a lower triangular matrix, [ L ] , and a diagonal matrix, [ D ] . As discussed in
“User Interface” on page 76 of the NX Nastran Numerical Methods User’s Guide a
diagonal term of [ D ] (“ d i ”) will approach zero for every mode of instability of the
structure. (In this example there is one instability.) These instabilities are identified
by dividing all di ’s into their corresponding stiffness term, then listing all ratios
larger than a large number. See “MAXRATIO” on page 611 of the NX Nastran Quick
Reference Guide for a description of several other parameters used to control this
operation.
T
Let [ K ] = [ L ] [ D ] [ L ]
Then [ D ] = k 0 , [L] = 1 0
0 (k – k) –1 1
CHAPTER 13 321
Solution Sequences
– 10
If ( k – k ) = 0.0 , replace d i with 10
Ratio i = k ii ⁄ d i
Identify Grid Component for All Ratios > PARAM,MAXRATIO.
A corrective action for this example would be to constrain one of the grid points or
to add springs connecting this spring to other parts of the structure (not shown).
Since, in general, there may be many grid points active in a singularity, there is little
reason for picking one over another for connection. The usual cause of this type of
instability is elements left out by oversight. While this may be your intention, it
cannot be detected by the program. For these reasons, there is no automatic method
to constrain mechanisms. The discussion of “MAXRATIO” on page 611 of the NX
Nastran Quick Reference Guide describes the various actions and user options that are
available in the several contexts where singularities may be detected. Diagnosis
techniques and corrective actions are described in the NX Nastran Numerical
Methods User’s Guide.
Static Condensation in SubDMAPs SEKR and SEMR2

Structural partitioning is used to refer structural matrices to the boundary points in
substructuring operations and to perform a static condensation in dynamic
analyses.
At user option the set of free coordinates, u f , may be partitioned into two sets, u o
and u a , such that the u o set is eliminated first. Thus
 ua 
{ uf } =   Eq. 13-28
 ou 
The equilibrium equations after the elimination of constraints (Eq. 13-24) may be
written in partitioned form as
K aa K ao  u a   Pa 
  =   Eq. 13-29
T
K ao K oo  u o   Po 
Rearrange the bottom half of Eq. 13-29:
T
[ K oo ] { u o } = { P o } – [ K ao ] { u a } , Eq. 13-30
and solve for { u o } :
–1 –1 T
{ u o } = [ K oo ] { P o } – [ K oo ] [ K ao ] { u a } Eq. 13-31
Substitute for u o into the top half of Eq. 13-29:
–1 T –1
[ K aa – K ao K oo K ao ] { u a } = { P a } – [ K ao ] [ K oo ] { Po } Eq. 13-32
It is convenient to define the matrix
–1 T
[ G o ] = – [ K oo ] [ K ao ] , Eq. 13-33
so the Eq. 13-32 becomes
T
[ K aa + K ao G o ] { u a } = { P a } + [ G o ] { P o } , Eq. 13-34
where advantage is taken of the symmetry of [ K oo ] .

Following the practice of condensation established in preceding subsections,
[ K aa ] { u a } = { P a } , Eq. 13-35
where
[ K aa ] = [ K aa ] + [ K ao ] [ G o ] , Eq. 13-36
T
{ Pa } = { Pa } + [ Go ] { Po } Eq. 13-37
The [ G o ] matrix defined in Eq. 13-33 is obtained practically from the solution of
T
[ K oo ] [ G o ] = – [ K ao ] , Eq. 13-38
where [ K ao ]T is treated as a set of load vectors. Each such vector produces a column
of [ G o ] .
Once { u a } is obtained the set of omitted coordinates, { u o } , is obtained as follows.

 
Define the set  u oo  as the solution of
 
 o
[ K oo ]  u o  = { P o } Eq. 13-39
 
Then, using Eq. 13-33 and Eq. 13-39 in Eq. 13-30,
 o
{ uo } =  uo  + [ Go ] { ua } Eq. 13-40
 
In subDMAP SEKR, the constrained stiffness matrix, [ K ff ] , is partitioned by

UPARTN as follows:
CHAPTER 13 323
Solution Sequences
K aa K ao
[ K ff ] = Eq. 13-41
K oa K oo
DCMP and FBS are used to solve the equation
t
[ K oo ] [ G oa ] = – [ K oa ] Eq. 13-42
for the static transformation matrix, [ G toa ] , and MPYAD performs the matrix
reduction
t
[ K aa ] = [ K aa ] + [ K ao ] [ G o a ] Eq. 13-43
If PARAM,ALTRED,YES is specified then the [ K ff ] matrix is partitioned and

reassembled to form [ Kxff ] which has the a-set degrees-of-freedom sequenced last.
x
[ K ff ] is partially decomposed
x
[ K ff ] → [ L fo ], [ K aa ] Eq. 13-44
where [ L fo ] is a trapezoidal matrix and [ K aa ] is the boundary stiffness matrix.
The [ Lfo ] matrix is partitioned
L oo
[ L fo ] = Eq. 13-45
L ao
While [ L ao ] T is of the same dimension as the [ G toa ] matrix, it is on the average half
as dense, thereby reducing the storage requirements.
In subDMAP SEMR2, UPARTN partitions the constrained mass matrix [ M ff ] as

follows:
M aa M ao
[ M ff ] = Eq. 13-46
M oa M oo
and successive executions of MPYAD and SMPYAD modules perform the matrix
reduction
t T t T t
[ M aa ] = [ G oa ] [ M oo G oa + M oa ] + [ M oa G oa + M aa ] Eq. 13-47
In subDMAP SEMRB, the structural and viscous damping matrices [ B ff ] and [ K4ff ]
are condensed by the MATREDU module.
The partitioned solution for statics is an exact solution. It merely changes the order
of operations of the unpartitioned solution. The static reduction for dynamics is
exact for o-set degrees-of-freedom that are massless, but represents an approximate
relumping of o-set masses to the a-set. You must select enough a-set degrees-of-
freedom to adequately define the eigenvectors of the system. The dynamic
reduction technique discussed in “Dynamic Reduction and Component Mode
Synthesis in SubDMAP SEMR3” on page 331 provides a way of improving the
approximation by computing the normal modes for the o-set degrees-of-freedom.
Static Solutions in SubDMAP SEKRRS

There are two major branches through this subDMAP, standard static analysis and
static analysis with inertia relief effects.
Inertia Relief Not Selected

When inertia relief is not selected and there are no SUPORTi entries are present the
a-set equations are solved directly
K aa ⋅ u a = P a Eq. 13-48
is decomposed into its factor matrices L aa and D a . These factors are passed on to
K aa
SubDMAP STATRS for solution of u a .
When inertia relief is not selected but there are SUPORTi entries are present the a-set
equations are partitioned into the r-set, defined on the SUPORTi entries, and the l-
set, what is “left over” after the r-set is removed.
K ll K lr ul P1
≥ Eq. 13-49
K rl K rr ur Pr
In this context, the r-set is treated as an alternate form of single-point constraints. The
partitions of K involving r are discarded. u r is set to zero. u l is found from the first
row of this equation. Loads in P r go directly to ground, and are output with the SPC
forces, along with the product –K rl ⋅ u l .
Inertia Relief Selected

The basic assumption of inertia relief is that the structure is in a state of steady
acceleration, and that all transients due to dynamic effects are negligible. The
solution is obtained in a moving reference frame from which all displacements are
measured. For inertia relief analysis inertial loads are computed automatically that
exactly balance the applied loads. A support system is applied to make the system
numerically stable. The inertial loads assume rigid body accelerations when
computing the loads.
CHAPTER 13 325
Solution Sequences
The derivations given here assume that the system mass matrix M gg has been
reduced to Maa and is available, as described elsewhere. In the actual DMAP,
equivalent methods that never generate M aa are used instead, in the interests of
efficiency. Maa can be a larger size data block, and costly to compute. It can be
shown that the results provided by the two derivations are identical. For example,
the rigid body mass matrix Mrr , defined below as
M rr = D ar ⋅ M aa ⋅ D ar Eq. 13-50
can just as easily be formed from the g-set equations
M rr = D gr ⋅ M gg ⋅ D gr Eq. 13-51
Similar comments can be made about computing the inertial loads that balance the
applied loads. The derivatives are presented here in the form of a-set equations that
are more readily understood, if less efficient in practice, than the g-set equations.
You can find equations for the more efficient method TANs 4002 and 4854 or in
DMAP compilations. Current research indicates that this capability may shift to a-
set operations in future versions to obtain more automation.
There are at present two forms of inertia relief: automatic selection of a support
system (“auto-support”); and manual selection of the support system, using
SUPORTi entries. Both react to PARAM,GRDPNT, the parameter used to locate a
reference point for grid point weight summaries. The more automatic method is
described first.
Auto-Support (PARAM,INREL,-2)
Vectors of rigid body motion D gr for six unit motions of a reference point are
computed from grid point geometry in module VECPLOT. The D ar component is
partitioned from this matrix. D ar is used to compute the rigid body mass M rr about
the reference point.
MD ar = M aa ⋅ D ar Eq. 13-52
M rr = D ar ⋅ MD ar Eq. 13-53
M rr is printed out with the matrix name of QRR.
The acceleration of the reference point ar due to the applied loads is computed,
M rr ⋅ ar ≥ D ar ⋅ P a Eq. 13-54
The accelerations ar are printed with the matrix name of URA.
The resulting inertial loads P ai and total loads P a are

P ai = M aa ⋅ ar Eq. 13-55
P a1 = P a – P ai Eq. 13-56
A constraint equation is written that states that the average weighted motion of the
system is zero,
MD ar ⋅ u ar = 0 Eq. 13-57
This equation is appended to the a-set to define the 1-set equations in their
augmented form,
K ll ⋅ u l = P l Eq. 13-58
K aa MD ar ua P a1
= Eq. 13-59
MD ar 0 rr –qr 0r
qr are the unknown forces required on the system to produce equilibrium. They
should be computational zeros, and are output with the SPC forces. The equation
involving Kll is solved directly, with u a and q r partitioned from u l .
Manual Support (PARAM, INREL, -1 and SUPORTi)

You prescribe a set of DOFs that constrain the structure in a statically-determinate
manner on SUPORTi entries. These entries define the r-set. There must be six or less
DOFs listed, and they must constrain all rigid body modes that are not constrained
by SPCs or other modeling devices. The rigid body modes are computed from the
statics equation, where there are no loads applied, and u rr is the identity matrix.
Given
K ll K lr D lr 0 lr
= Eq. 13-60
K le K rr I rr 0 rr
D lr is found by solving the first row of this equation, then merged with I rr to form
D ar ,
K ll ⋅ D lr ≥ – K lr ⋅ I rr Eq. 13-61
The rigid body mass, reference point accelerations, and inertial loads are formed
with the same equations used for auto-support. The loads are partitioned into the l-
and r-sets, and the partitioned form of K ll is used for solution for displacements,
K ll ⋅ u l = P ll Eq. 13-62
CHAPTER 13 327
Solution Sequences
ua is merged from u l and a null partitioned for u r .
A Comparison of Auto- and Manual-Support Solution

If one analyzes the same model with both methods, you should find that stresses,
constraint forces, and other internal quantities are identical except for truncation
errors, but the displacements may differ appreciably. This is because they use a
different reference frame. For the many supports, the reference frame is set up so
that the motion at the r-set DOFs are identically zero. One can think of a reference
frame drawn through these points, and all displacements measured relative to it.
This reference frame is moving through space at a constant acceleration rate, so the
displacements are relative, rather than inertial displacements.
For the auto-support option, one can visualize the reference frame as a set of axes
connected to every point with mass with a very complicated set of levers that allows
the structure to “breathe”, as loads are applied to it, but keep the weighted average
motion to a zero value. If the model happens to have a grid point at the reference
point, you will observe very small motion there, computational zeros but not the
binary zeros of the manual support method. This is because this motion is
computed for the auto-support method, but assumed to be zero for the manual
method.
The choice of an r-set as defined by SUPORTi DOFs requires meeting both a

geometric requirement and a stiffness requirement. The geometric requirement is
that the set should form a statically determinate tie to ground. This can be difficult
to do with complicated geometries. The stiffness requirement is that the DOFs
should be stiff enough to provide a stable solution. A method that has been used by
some modelers is to pick points stiff enough that they could be used to lift the
structure without damaging it. Sometimes such points do not exist. The difficulties
in picking a stable, reliable support system was one of the main motivations for
development of the more modern auto-support systems.
PARAM, GRDPNT, GID is optional for both methods when superelements are not
present, but required by the manual method when superelements are present. The
auto-method does not support superelement analysis. The default for GRDPNT is
GID = -1, which is a flag to use the basic origin as the reference point when
computing rigid body modes. Use of a GID near the c.g. of the structure produces
more esthetically pleasing results. The rigid body accelerations produced by the
solution are then measured near the center of the structure, rather than at a point
that may be outside of the structure.
Static and Dynamic Load Generation
Applied Static Loads in SubDMAPs SELG, SELR, and SELRRS

The static load vectors { P g } are generated by SSG1 (Static Solution Generator –
Phase 1) in subDMAP SELG from your input of applied loads, temperature fields,
and enforced deformations.
The direct matrix input requested by the P2G Case Control command define the load
matrix [ P g2 ] which is assembled by the MTRXIN module and is added to the output
from SSG1.
If PARAM,INREL,-1 is specified, then

1. The rigid body transformation matrix [ D gr ] is generated with the
VECPLOT module. Its columns contain the motion of all degrees-of-
freedom due to unit motion of the corresponding reference point degree-of-
freedom.
2. Loads due to unit rigid body accelerations are computed,
{ P ji } = – [ M jj ] [ D gr ]
3. The loads above are appended to the applied loads { P j } ,
{ Pj } ← { Pj P ji }
The SELA module assembles { P j } into { P g } .
The constraints are applied to the static load vectors by SSG2 (Static Solution
Generator – Phase 2) in subDMAP SELR as follows:
 Pn 
{ Pg } =   Eq. 13-63
 Pm 
T
{ P n } = { P n } + [ G mn ] { P m } Eq. 13-64
 Pf 
{ Pn } =   Eq. 13-65
 Ps 
{ P f } = { P f } – [ K fs ] { Y s } Eq. 13-66
– 2α  P a C
- = 2 --------
g {=P ---------
f }ω =  C Eq. 13-67
 P o  cr
CHAPTER 13 329
Solution Sequences
T
{ P a } = { P a } + [ G oa ] { P o } Eq. 13-68
If PARAM,ALTRED,YES is specified, the diagonal extracted from [ Loo ] → [ D oo ]
and the reduced load matrix is computed from
{ P a } = { P a } – [ L ao ] [ D oo ] [ u ox ] Eq. 13-69
where { u ox } is obtained on a forward pass from the equation
[ L oo D oo ] { u ox } = { P o } Eq. 13-70
for PARAM,ALTRED,NO, the SSG3 module (Static Solution Generation – Phase 3)

solves the equation
[ K oo ] { u oo } = { P o } Eq. 13-71
for { u oo } , the displacements of the omitted coordinates.
SSG3 also calculates the residual vector, δP o , and the residual vector error ratio, ε o ,
for the omitted coordinates
{ δP o } = [ K oo ] { u oo } – { P o } Eq. 13-72
T
{ u oo } { δP o }
ε o = ---------------------------------- Eq. 13-73
T
{ P o } { u oo }
Except for round-off error, the error ratio εo should be zero. Large values of these
error ratios usually indicate singularities in the stiffness matrix. The residual load
vector, RUOV, may be output by use of PARAM,IRES,1.
The quantity
T
1 ⁄ 2 { Po } { uo } Eq. 13-74
is calculated by the SSG3 module and printed under the heading “External Work.”
This component of strain energy includes effects of thermal loads, element
deformations, and enforced displacements that may be subtracted later in the
solution process. For example, an enforced displacement that causes zero strain
will result in external work.
If a r-set is present, subDMAP SELRRS calculates determinate forces of reaction
T
[ q r ] = – [ D ar ] { P a } Eq. 13-75
where:
D er
[ D ar ] =
I rr
Applied Dynamic Loads

Dynamic loads are produced in many subDMAPs. They are generally of the form:
P(x) = ∑ PO i ⋅ F i ( x ) Eq. 13-76

i
The PO i vectors are spatial functions that describe what DOFs are loaded and their
scale factors. The F i functions are temporal functions that define how the loads
change with time in transient analysis x = t , or with excitation frequency in
frequency response analysis x = f . All spatial loads except those on extra points are
calculated in the static analysis load generator module, SSG1, to obtain the P g vector
described in the prior paragraphs.
Enforced motions are produced similarly.
Y(x) = ∑ YO i ⋅ F i ( x ) Eq. 13-77

i
The YO i vectors are again the spatial distribution, and the F i ( x ) temporal functions.
The YO i vectors are produced by the GP4 module to produce the Y g vector. It is the
same vector described for static analysis in Eq. 13-66, except that it is merged to g-set
size with null terms from Ys.
The static load and enforced motion vectors are converted into dynamic loading
tables in DPD module, then passed on to the dynamic load generators, FRLG and
TRLG. These modules produce the applied loads and enforced motions at the p-set
level, Pp ( x ) and Yp ( x ) , as a function of excitation frequency or time. For transient
analysis velocity and acceleration vectors are also provided, Yp· ( x ) and Yp·· ( x ) .
A reduction matrix Rpx is made in the FDRMGEN module, where the subscript “x”
implies the d-set, for direct analysis, and h, for modal analysis. It combines the
operations shown in Eq. 13-63 through Eq. 13-68 previously. In subDMAP
CFORCE2 Eq. 13-66 is modified to add velocity and acceleration effects,
P f = P – K fs ⋅ Y s – B fs ⋅ Y· s – M fs ⋅ Y·· s Eq. 13-78

f
In frequency response the velocity and acceleration effects can be computed from Ys
by multiplying it by i ⋅ ω once or twice, where ω is the excitation frequency.
CHAPTER 13 331
Solution Sequences
Pf is reduced to P d or P h by the same techniques used in static analysis.
For modal analysis with enforced motion it is essential that PARAM,RESVEC,YES

be used. It causes static solutions to be formed from K fs ⋅ Y s -type terms. When the
structure is tied to ground in a statically-determinate manner, these solutions are
linear combinations of the rigid body modes that would exist if the ties to ground
were removed. If the ties to ground are redundant, a condition named multi-base
input, the resulting shapes will have significant stresses associated with them. They
represent the effect of low-frequency excitation on the structure when
PARAM,RESVEC,YES is not present, for the statically determinate case the element
stresses will be correct but the displacement-type data will be missing the effect of
the static shapes. For the multi-base input case, both stresses and displacement-type
output will be incorrect.
Dynamic Reduction and Component Mode Synthesis in

SubDMAP SEMR3
These capabilities use several subsets of the a-set not used by other solution
techniques. The v-set matrix is formed in subDMAP SEKR
K oo K or K oc
K vv = K ro K rr K rc Eq. 13-79
K co K cr K cc
The v-set mass matrix is formed similarly in subDMAP SEMR3. The v-set degrees-
of-freedom are free to vibrate during dynamic reduction and component mode
computation. They consist of the omitted terms (o-set), the free boundary points
used for rigid-body mode designation (r-set), and the other free boundary points
(c-set). The [ Mv ] matrix is formed in a similar manner.
If the METHOD command is specified for superelements, then the component

modes are computed. The auto-omit operations described in “Real Eigenvalue
Analysis in SubDMAPs SEMR3 and MODERS” on page 353 are carried out
resulting in the reduced eigensolution equation
[ K yy – λM yy ] [ φ yz ] = 0 Eq. 13-80
where N z is the number of eigenvectors requested by the EIGR or EIGRL Bulk Data
entries. The back-transformations for the auto-omit-spc are performed, resulting in
the expanded eigenvector matrix [ φ yz ] . This matrix is used to purify the
approximate eigenvectors found in dynamic reduction,
[ φ vz ] = [ φ vy ] [ φ yz ] Eq. 13-81
The uncoupled eigensolution components are assembled from [ φ vz ]
 φ oz 
 
[ φ vz ] =  φ rz  Eq. 13-82
 
 φ cz 
 φ rz 
 
[ φ az ] =  φ cz  Eq. 13-83
 
 φ bz 
where [ φ bz ] = [ 0 ]
and [ φ az ] are stored for use with the uncoupled solution obtained using
[ φ oz ]
PARAM,FIXEDB,-1 as described in “Data Recovery Operations in SubDMAP
SEDISP” on page 338.
The residual vector capability is requested with PARAM,RESVEC,YES. Details on

this capability are given in “Matrix Assembly Operations in SubDMAP SEMG” on
page 311, where real eigensolutions are discussed. The RESVEC SubDMAP returns
auxiliary functions u o2 . They are appended to the eigenvectors, and are given all of
the operations given to eigenvectors,
φ oz u o2
H vz = φ rz 0 r2 Eq. 13-84
φ cz 0 c2
The inertia relief mode shapes are calculated by statically applying the rigid-body
inertial loads to the structure.
i
K oo K oa Ho M oo M oa Vo
= Eq. 13-85
K ao K aa i
Ha M ao M aa Va
Setting [ H ia ] to zero and solving for [ H io ] results in,
i –1
[ H o ] = [ K oo ] [ [ M oo ] [ V o ] + [ M oa ] [ V a ] ] Eq. 13-86
The [ G oq ] matrix is formed,
1
[ G oq ] ← [ G oz H oi 0] Eq. 13-87
CHAPTER 13 333
Solution Sequences
The [ G oq ] matrix is truncated or padded with null columns to provide Nq columns,

where Nq is the q-set size. [ G oq ] is expanded to a-size,
[ G oq q ]
0 ot ] → [ G oa Eq. 13-88
The total transformation matrix is formed by adding dynamic and static

components and is reassembled whenever needed,
t ] + [Gq ]
[ G oa ] = [ G oa Eq. 13-89
oa
The generalized stiffness coefficients are formed from the dynamic transformation,
T
[ K qq ] = [ G oq ] [ K oo ] [ G oq ] Eq. 13-90
The physical and generalized degrees-of-freedom are not stiffness coupled. [ Kqq ] is
merged to N a size,
l ] =
K qq 0
[ K aa Eq. 13-91
0 0
where superscript l refers to matrices formed from dynamic reduction. [ Kqq ] is

stored in the database. When the total stiffness boundary matrix is needed, it is
formed from its partitions,
m ] =
K qq 0
[ K aa Eq. 13-92
0 K tt
or
m l ]
[ K aa ] = [ K aa ] + [ K aa Eq. 13-93
The physical and generalized degrees-of-freedom are mass-coupled. The terms are
formed from the equations,
T
[ M qq ] = [ G oq ] [ M oo ] [ G oq ]
T t ]
[ M qt ] = [ G oq ] [ M ot + M oo G oa
Eq. 13-94
l
M qq M qt
[ M aa1 ] =
M tq 0
Virtual mass effects, if any, are added to [ M aa1

l ] to form [ M l ] .
aa
The total
www.cadfamily.com mass matrix is formed by:
m ] = [M ] + [Ml ]
[ M aa Eq. 13-95
aa aa
For the rigid formats [ Maa ] is formed directly from its partitions,
M qq M qt
[ M aa ] = Eq. 13-96
M tq M tt
The a-set matrices do not contain static effects in the t-set components, and dynamic
effects in the q-set components. Virtual mass effects, if any, are added to [ M aa ] to
form [ M aa
m ].
The damping matrices [ B gg ] and [ K gg

4 ] are reduced to the a-set by the MATREDU
module which performs the following operations:
1. Eliminate multipoint constraints
B nn B nm
[ B gg ] ⇒
B mn B mm
T T
[ B nn ] = [ G mn ] [ B mn G mn + B mn ] + [ B mn ] [ G mn ] + [ B nn ]
2. Eliminate single point constraints
B ff B fs
[ B nn ] ⇒
B sf B ss
3. Partition omitted degrees-of-freedom
B aa B ao
[ B ff ] ⇒
B oa B oo
4. Perform static condensation
T T
[ B aa ] = [ G oa ] [ B oo G oa + B oa ] + [ B oa ] [ G oa ] + [ B aa ]
Formulation of Dynamic Equations in SubDMAP GMA

The tables prepared by the geometry processors are modified by DPD (Dynamic
Pool Distributor) to include the extra points (e-points) introduced for dynamic
analysis.
CHAPTER 13 335
Solution Sequences
m ] , [M m ] , [B ] ,
[ K aa and [ K aa
4 ] are expanded to the d-set to form [ K 1 ] , [ M 1 ] , [ B 1 ] ,
aa aa dd dd dd
and [ K dd4 ] . The direct input matrices [ K 2 ] , [ M 2 ] , and [ B 2 ] requested by the K2PP,
pp pp pp
M2PP, and B2PP Case Control commands are reduced to the d-set to form [ K dd 2 ],
2 ] , and [ B 2 ] , respectively.
[ M dd dd
The general dynamic equation used in the direct methods is
2
[ [ M dd ]p + [ B dd ]p + [ K dd ] ] { U d } = { P d } Eq. 13-97
p is the derivative operator
 Ua 
{ Ud } =   Eq. 13-98
 Ue 
where:
{ U d } = the extra points
For direct frequency response and complex eigenvalue analysis the stiffness,
damping, and mass matrices are generated as follows:
1 ] + [ K 2 ] + i[ K 4 ]
[ K dd ] = ( 1 + ig ) [ K dd Eq. 13-99
dd dd
1 ] + [B2 ]
[ B dd ] = [ B dd Eq. 13-100
dd
and
1 ] + [M2 ]
[ M dd ] = [ M dd Eq. 13-101
dd
For direct transient response the stiffness, damping and mass matrices are
generated as follows:
1 ] + [K2 ]
[ K dd ] = [ K dd Eq. 13-102
dd
g
1 ] + [ B 2x ] + ------ 1
1 ] + ------
[ B dd ] = [ B dd dd [ K dd [K4 ] Eq. 13-103
ω3 ω 4 dd
and
1 ] + [M2 ]
[ M dd ] = [ M dd Eq. 13-104
dd
where:
1
• is the reduced structural stiffness matrix plus the reduced direct input
[ K dd ]
K2GG (symmetric).
2
• is the reduced direct input matrix K2PP plus the reduced transfer
[ K dd ]
function input (symmetric oar unsymmetrical).
4
• is the reduced structural damping matrix obtained by multiplying the
[ K dd ]
stiffness matrix [ K e ] of an individual structural element by an element
damping factor g e and combining the results for all structural elements
(symmetric).
1
• is the reduced viscous damping matrix plus the reduced direct input
[ B dd ]
B2GG (symmetric).
2
• is the reduced direct input matrix B2PP plus the reduced transfer
[ B dd ]
function input (symmetric or unsymmetrical).
1
• is the reduced mass matrix plus the reduced direct input M2GG
[ M dd ]
(symmetric).
2
• is the reduced direct input matrix M2PP plus the reduced transfer
[ M dd ]
function input (symmetric or unsymmetrical).
• g is the structural damping coefficient on the PARAM Bulk Data entry G.
The frequencies ω 3 and ω 4 are specified on the PARAM Bulk Data entries
as W3 and W4, respectively.
For direct solutions [ K dd ] , [ Mdd ] , and [ B dd ] are examined to identify rows and
columns which are null in all three matrices. If the parameters ASING = 0 (the
default) the singularities caused by the null rows and columns are removed as
follows:
1. For direct frequency and direct transient response, the stiffness matrix
x ] is formed from [ K ] by placing unity on the diagonal for each null
[ K dd dd
row and column.
2. For direct complex eigenvalue analysis null rows and columns are
discarded from [ Kdd ] , [ M dd ] , and [ B dd ] to form [ K xx
d ] , [ M d ] , and [ B d ] . This
xx xx
is performed in subDMAP DCEIGRS. See “Complex Eigenvalue Analysis
in SubDMAP CEIGRS” on page 367.
If parameter ASING = -1 and null rows and/or columns exist, a fatal error will result.
In modal frequency response, and complex eigenvalue analysis, structural damping

and viscous damping are included in the dynamic matrices as follows:
2 ] = [ K 2x ] + ig [ K 1 ] + i [ K 4 ]
[ K dd Eq. 13-105
dd dd dd
and
CHAPTER 13 337
Solution Sequences
2 ] = [ B 1 ] + [ B 2x ]
[ B dd Eq. 13-106
dd dd
In modal transient response, the structural damping and viscous damping are
included as follows:
2 ] = [ K 2x ]
[ K dd Eq. 13-107
dd
and
g
2 ] = [ B 1 ] + [ B 2x ] + ------ 1.0 4
[ B dd dd dd [ K 1 ] + ------- [ K dd ] Eq. 13-108
ω 3 dd ω4
The general dynamic equation used in the modal method is:
2
[ [ M hh ]p + [ B hh ]p + [ K hh ] ] { U h } = { P h } Eq. 13-109
where p is the derivative operator
 ζi 
{ Uh } =   Eq. 13-110
 Ue 
ζi are the modal coordinates.
The transformation from { U h } to { U d } is
{ U d } = [ φ dh ] { U h } Eq. 13-111
where:
φ ai 0
[ φ dh ] =
0 I ee
The dynamic matrices expressed in modal coordinates are assembled in GKAM

(General K Assembler-Modal) as follows:
T 2 ][φ ]
[ K hh ] = [ k ] + [ φ dh ] [ K dd dh Eq. 13-112
T2 ][ φ ]
[ M hh ] = [ m ] + [ φ dh ] [ M dd dh Eq. 13-113
and
T 2 ][ φ ]
[ B hh ] = [ b ] + [ φ dh ] [ B dd dh Eq. 13-114
where, if PARAM,KDAMP = 1 (default)

m i = modal mass
b i = 2πf i g ( f i )m i
k i = 4π 2 f i2 m i
If PARAM,KDAMP = –1 in complex eigenvalue analysis, modal frequency response,

aerodynamic flutter, and aeroelastic response the viscous modal damping will be
incorporated into the complex stiffness matrix as follows:
m i = modal mass
bi = 0 Eq. 13-115
k i = [ 1 + ig ( f i ) ]4π 2 f i2 m i
Values of g ( fi ) are specified on the TABDMP1 Bulk Data entry, selected by the
SDAMPING Case Control command. If KDAMP = 1 (default), the matrices [ b ] and
[ k ] are formulated as in modal transient response.
In SubDMAP CMPMODE damping of CMS (o-set) modes is calculated by similar

techniques. This q-set damping is expanded to g-set size, and placed in the B aa
matrix for summing into downtown superelements.
Data Recovery Operations in SubDMAP SEDISP

The dependent displacements and the single-point forces of constraint are calculated
in SDR1 (Stress Data Recovery – Phase 1). The inputs to SDR1 are columns of the
solution vectors corresponding to each eigenvalue, loading condition, or output time
step. In the case of transient analysis, there are three columns, corresponding to
displacement, velocity and acceleration, for each output time step.
Load Modification
If PARAM,INREL,-1 or PARAM,INREL,-2 are specified, then any component of the
load vector is first multiplied by [ C rx ] to produce a set of loads that are a linear
combination of applied and inertial loads called { P li }
{ P li } ← [ P l ] [ C rx ] Eq. 13-116
Static Analysis
In the case of static analysis and PARAM,ALTRED,NO, SDR1 recovers dependent
displacements
 ul 
  ⇒ { ua } Eq. 13-117
 ur 
CHAPTER 13 339
Solution Sequences
 ua 
  ⇒ { uf } Eq. 13-118
 uo 
where
t ]{ u } + { u o }
{ u o } = [ G oa Eq. 13-119
a o
If PARAM,ALTRED,YES is specified, then the displacements { u o } are obtained

from a backward pass from the equation
T
[ L oo ] { u o } = { u oy } Eq. 13-120
where { u oy } = { uox } – [ Lao ] T { ua }

Enforced displacements are combined with u f
 uf 
  ⇒ { un } Eq. 13-121
 Ys 
 un 
  ⇒ { ug } Eq. 13-122
 um 
where
{ u m } = [ G mn ] { u n } Eq. 13-123
and recovers single point forces of constraint
T
{ q s } = [ K fs ] [ u f ] + [ K ss ] { Y s } – { P s } Eq. 13-124
and multipoint forces of constraint
I mm
q mg = [ [ K mg ] { u g } – { P m } ] Eq. 13-125
t
– G mn
In superelement static analysis, if PARAM,FIXEDB,-1 is present, then u a is set to

zero, and data recovery is performed using the modified equation
{ u o } = { u oo } Eq. 13-126
This is equivalent to obtaining a solution with all exterior (“boundary”) degrees-of-

freedom fixed. Displacements are due only to interior displacements due to interior
loads, that is, the { u oo } component.
Eigenvalue Analysis
In the case of eigenvalue analysis SDR1 recovers dependent components of the
eigenvectors
{ φ o } = [ G oa ] { φ a } Eq. 13-127
 φa 
 -----  = { φf } Eq. 13-128
 φo 
 φf 
-  = { φn }
 ---- Eq. 13-129
 φs 
{ φ m } = [ G mn ] { φ n } Eq. 13-130
 φn 
-  = { φg }
 ------ Eq. 13-131
 φm 
and single-point forces of constraint
T
{ q s } = [ K fs ] { φ f } Eq. 13-132
If PARAM,DYNSPCF,NEW (default) then the single-point forces are computed as

follows:
{ q s } = [ [ K sf ] – ω 2 [ M sf ] ] { φ f } Eq. 13-133
In superelement analysis, if PARAM,FIXEDB,–1 is present, the uncoupled

eigenvectors are recovered from the modified equations
{ φ a } = { φ az } Eq. 13-134
{ φ o } = { φ oz } Eq. 13-135
with { φ az } and { φ oz } from “Superelement Analysis” on page 371.
The multipoint forces of constraint qm q are recovered by
I mm 2
qm q = ( – ω M mg + K mg ) Eq. 13-136
T
– G mn
CHAPTER 13 341
Solution Sequences
Dynamic Analysis
In the case of dynamic analysis with the direct formulation the extra points are
partitioned out of { u d } .
 ua 
 -----  = { ud } Eq. 13-137
 ue 
SDR1 recovers the dependent components
{ u o } = [ G oa ] { u a } Eq. 13-138
 ua 
 -----  = { uf } Eq. 13-139
 uo 
and recovers single point forces of constraint
T
{ q s } = [ K fs ] { u f } – { P s } Eq. 13-140
If PARAM,DYNSPCF,NEW (default) then the forces are computed as follows.
In transient response analysis, the formulation is
x u·· + B x u· + K x u – P
q s = M sf Eq. 13-141
f sf f sf f s
where:
x = M + M 2pp
M sf sf sf
g
x = B x + B 2pp + ------ 1
B sf sf sf K + ------ K 4gg
ω 3 sf ω 4 sf
x = K + K 2pp
K sf sf sf
In frequency response and complex eigenvalue analysis, the formulation is
q s = ( – ω 2 M sf
x + iωB x + K x )u – P
sf sf f s Eq. 13-142
where:
x = M + M 2pp
M sf sf sf
x = B x + B 2pp
B sf sf sf
x = ( 1 + ig )K + K 2pp + iK 4gg
K sf sf sf sf
The multipoint forces of constraint qm g are recovered
I mm
qm g = ( Xm – Pm ) Eq. 13-143
T
– G mn
where qm g is the reactions expanded to the g-set and X m and P m are the internal and
external loads on the m-set degrees-of-freedom, respectively. (See also Eq. 9 in
Section 3.5.1 in The NASTRAN Theoretical Manual).
In transient response analysis,
x u·· + B x u· + K x u
X m = M mg Eq. 13-144
g mg g mg g
where:
x
M mg = M mg + M mg
2pp
x g
2pp + ----- 1
B mg = B mg + B mg -K + ------K 4gg
ω mg ω mg
3 4
x
K mg = K mg + K mg
2pp
In frequency response and complex eigenvalue analysis,
2
x + iωB x + K x )u
X m = ( – ω M mg mg mg g Eq. 13-145
where:
x
M mg 2pp
= M mg + M mg
x
B mg 2pp
= B mg + B mg
x
K mg 2pp + iK 4gg
= ( 1 + ig )K mg + K mg mg
For frequency response there is one complex vector for each loading condition. For
complex eigenvalue analysis there is one complex eigenvector for each complex root
and there is no loading term { P s } associated with the determination of the single
point constraints. For transient analysis there is a displacement, velocity, and
acceleration associated with each output time step.
In the case of dynamic analysis with the modal formulation, you may obtain
improved element stresses by requesting the mode acceleration data recovery
technique.
The use of the mode acceleration technique requires the decomposition of the l -set
stiffness matrix [ K ll ] . This decomposition is not otherwise performed in dynamic
www.cadfamily.comanalysis unless the model has rigid body modes. Also, the technique requires a
CHAPTER 13 343
Solution Sequences
forward/backward substitution operation with a number of righthand sides equal

to the number of solution vectors. A subset of the solution vectors can be obtained
can be obtained by using the OTIME or OFREQUENCY entries in the Case Control
Section. The subset of solution vectors is prepared by the MODACC functional
module.
In the case of dynamic analysis with the modal formulation, two methods of data
recovery are available. The default procedure is usually much more efficient for
data recovery operations when the modal formulation is used for dynamic analysis.
In this procedure, the dependent components of the eigenvectors used in the modal
formulation are first recovered by SDR1 using the same sequence of operations
described for problems in real eigenvalue analysis (after eliminating unwanted
modes). The complete eigenvectors are used as the input to SDR2 in order to
determine the requested forces and stresses in terms of modal coordinates.
In dynamic analysis, you may restrict the output request to the response of selected
points in the solution set. In this case, the time-consuming operations in the
recovery of the dependent degrees-of-freedom can be avoided. This solution set
processing is performed by VDR (Vector Data Recovery).
The modal data generated by SDR1 and SDR2 are formed into a data matrix by
DDRMM (Dynamic Data Recovery Matrix Method). The output quantities are
determined by multiplying the data matrix by the modal solution vectors. The
computer time for this matrix multiplication is proportional to the number of
output times, the number of modes used in the modal formulation, and the number
of output frequencies or output time steps. If both the number of output items and
the number of modes are small, the computer time required will not be significant
when compared to the total problem solution time. Although the DDRMM method
of data recovery is more efficient, a complete { u g } is not formed. For this reason the
method does not permit the creation of deformed structure plots, grid point force
balance or the use of the mode acceleration technique.
The alternate procedure may be invoked using PARAM,DDRMM,-1. With this
method the solution vectors are first transformed from modal to physical
coordinates
{ u d } = [ φ dh ] { u h } Eq. 13-146
This operation is performed by MPYAD in subDMAP MODACC. Having

determined the vectors { ud } in physical coordinates, the data recovery operations
can be completed using the same sequence of operations described for the direct
formulation of dynamics problems.
Most of the operations in the recovery of the dependent components do not require
significant amounts of computer time. However, in the case of the multiply-add
operation associated with the recovery of the omitted coordinates
u a }, [ G oa ] { φ a } ) a significant amount of computer time will be required if there
( [ G oa ] {EMail:[email protected]
www.cadfamily.com
is a large number of vectors and a large number of components in the a-set. This
operation is likely to be especially costly for transient problems because there is
usually a large number of output time steps and there are three vectors for each time
step.
Data Recovery Operations in SubDMAP SEDRCVR

NX Nastran calculates the element forces and stresses in SDR2 (Stress Data Recovery
– Phase 2). The matrix operations in this phase of data recovery are similar to those
performed in EMG for stiffness matrix generation. The computer time will be
proportional to the number of elements for which output is requested.
If static loads are applied to intermediate points on the CBAR element, you can
request the output of stresses and forces at a selected set of intermediate points. The
additional calculations required for output at intermediate points on CBAR elements
is performed in functional module SDRX.
The solutions and data recovery operations for transient analysis are initially sorted
by the time step (SORT1). A transpose operation is performed by SDR3 (Stress Data
Recovery – Phase 3) in order to prepare the requested output sorted by external
point identification number or external element identification number (SORT2). In
frequency response problems, you may request either SORT1 or SORT2 output. The
SORT2 output will usually be desirable if the solution is made for a large number of
frequencies. In any event, the SORT2 output must be prepared if XY-plots are
requested.
Computer time for the preparation of SORT2 output will be directly related to the
number of items requested for output. If a large number of items are requested for
output, the transpose operation will require a significant amount of computer time.
You may request the output of a grid point force balance table for a selected set of
points. This table lists the forces acting at each selected point due to element forces,
single point forces of constraints and applied loads. The grid point force balance
table is prepared by GPFDR (Grid Point Force Data Recovery).
You may also request the output of an element strain energy table and/or an element
kinetic energy table for a selected set of elements. The tables include the energy for
each element, the percentage of the total energy for each element, and the energy
density for each element. The element energy is computed from the equations
1 T
V e = --- { u e } [ K ee ] { u e } Eq. 13-147
2
1
V e = --- { u e } [ M ee ] { u e } Eq. 13-148
2
CHAPTER 13 345
Solution Sequences
where the subscript e implies the data for one element. The energy density is
computed by dividing the element energy by the element volume. Some elements
do not have enough data input to allow calculation of element volume, such as
CELASi elements. The density output for these elements is set to zero. Simplified
equations for other element volumes may be used. For example, offsets are ignored
for bar-like elements, and midside node locations are ignored for plate and solid
elements. Element volumes are calculated by the ELPRT module, and can be output
by use of PARAM,EST,1. The strain energy table is prepared by functional module
GPFDR.
13.3 Summary of Solution Sequence Operations

This section describes the data processing and algorithms that are performed by
special purpose functional modules, sequences of functional modules (subDMAPs)
and complete solution sequences.
See Also
• NX Nastran DMAP Programmer’s Guide (for detailed descriptions of
functional modules)
• Chapter 4 of the NX Nastran Programmer’s Manual
Static Solutions
There are several major paths for static solutions and some minor ones. For
standard static analysis the stiffness matrix is decomposed in SubDMAP SEKRRS,
in preparation for a linear solution in SubDMAP STATRS. There is also a form of
static analysis called inertia relief analysis.
There are now several paths through inertia relief analysis, reflecting that it is still
in a state of development, with more changes likely in future systems. This revision
of this manual will discuss inertia relief in SOL 101 only. Each path has advantages
and limitations. There is also a similar but different method used in SOL 1, using
obsolete modules and with all of the limitations of SOL 1, most notably lack of
superelement capability. That method has not changed in many years, and is
adequately described in the first edition of this manual. It will not be documented
here. There is still another method in SOL 144, Static Aeroelasticity. It is
documented elsewhere.
All methods of inertia relief analysis take the standard static analysis equation
K aa ⋅ u a = P a Eq. 13-149
where P a are the applied loads, and add a set of constraint equations R ra and
automatically-computed inertial loads P i . The constraints are required because K aa
is for a free structure and cannot be solved reliably without some form of constraint.
K aa ⋅ u a = P a – P i = P ai Eq. 13-150
R ra ⋅ u a = 0 Eq. 13-151
In theory, if P ai is a set of balanced loads (no resultant) the solution of u a is possible

without the need for constraint equations. In practice, P ai may have some unbalance
due to numerical truncation effects. Even a small resultant load will provide a
displacement solution at infinity in theory, or with very large numbers in practice,
usually. The same truncation effects that unbalance the loads may provide enough
“ground springs” in the stiffness matrix to allow a solution, but it will be a poor
quality solution. Everything will first move 1.E5 units in the same direction. Stresses
will be determined by subtracting numbers of this size to find element displacement
perhaps on the order of 1.E-3. Ironically, if all goes well the forces in the support
system are miniscule, but it must be present to control stability when truncation
errors are present.
Types of Inertia Relief

Inertia Relief analysis is a quasi-dynamic analysis. It is assumed that the structure is
not tied to ground and is in a state of steady acceleration caused by the applied loads.
The displacements measured relative to ground are constantly increasing and
cannot be determined. Displacements are therefore determined with respect to an
accelerating reference frame that is attached to the structure. The applied loads are
balanced by inertial loads that are calculated automatically from the steady
acceleration acting on the mass properties of the model. Inertia relief analysis is
requested with PARAM,INREL,[x], where [x] may have the values -1 or -2.
The reference frame may be obtained by two different methods.

• For the manual support method (PARAM,INREL,-1) you select 6 or fewer
DOFs using SUPORTi entries. The reference frame is defined from these
DOFs.
• For the auto support method (PARAM,INREL,-2) 6 constraint equations
are written that constrain the structure to zero motion, on average, while
leaving it free to “breath” and deflect due to the applied loads and inertial
loads.
Both methods find it convenient to have “rigid body modes shapes” that represent a
set of vectors that represent all possible strain-free motion of the structure when not
loaded. The technique used to obtain rigid body mode shapes differs between the
two methods.
CHAPTER 13 347
Solution Sequences
Rigid Body Modes Using Manual Support (PARAM,INREL,-1 and

SUPORTi)
You prescribe a set of DOFs that constrain the structure in a statically-determinate
manner on SUPORTi entries. These entries define the r-set. There must be six or less
DOFs listed and they must constrain all rigid body modes that are not constrained
by SPCs or other modeling devices. The rigid body modes are computed from the
static analysis equation. There are no loads applied and I rr is the identity matrix.
Given
K ll K lr D lr O lr
= Eq. 13-152
T I rr O rr
K lr K rr
D lr is found by solving the first row of this equation, then merged with I rr to form
D ar ,
K ll ⋅ D lr ≤ – K lr ⋅ I rr Eq. 13-153
D lr
[ D ar ] = Eq. 13-154
I rr
Rigid Body Modes Using Auto-Support (PARAM,INREL,-2)

Vectors of rigid body motion D gr for six unit motions of a reference point are
computed from grid point geometry in module VECPLOT. The D ar component is
partitioned from this matrix.
The Basic Equation of Motion for Inertia Relief

Inertia relief analysis starts with the a-set equation
K aa ⋅ u a ≤ P ai Eq. 13-155
and adds a constraint equation
T
R ar ⋅ u a = 0 Eq. 13-156
The last equation implies a similar equation on a generalized constraint force

variable vector, q r
R ar ⋅ q r = P ai – K aa ⋅ u a Eq. 13-157
If P ai represents balanced loads, q r is zero. If there is a small unbalance due to

numerical truncation effects, qr is then small, but finite. There are two methods for
defining R ar , as discussed below. Both have the property of defining some part of u a
to have zero motion, so that the other parts of ua are in effect measured relative to
the constrained DOFs.
On of the solution methods partitions Kaa into its r-set and l-set components. The
first equation is partitioned, and the a-set equations and the constraint equation may
combine into one equation,
T
K ll K lr R rl ul P ail
T T ⋅ ur = P air Eq. 13-158
K lr K rr R rr
R rl R rr O rr –qr Yr
The methods of computing P i are described below. Yr , the generalized displacement

input, is set to zero.
Constraint Equations Using Manual Supports

In the constraint equation R rr = I rr , the identity matrix. R rl is null. The last row can
then be solved for u r = Yr = 0 . The second equation can then be used to eliminate q r ,
T
K lr ⋅ u l – I rr ⋅ q r = P air Eq. 13-159
T
q r = K lr ⋅ u l – P air Eq. 13-160
Because R rl is null, the first equation then becomes
K ll ⋅ u l = P ail Eq. 13-161
This equation can be solved for u l reduced from u a directly. It will be shown later
that P r terms are transferred to P l terms in the inertial loads. u a is formed from a
merge of u l and null terms for u r .
Constraint Equations Using Automatic Supports

A constraint equation is written that states that the average weighted motion of the
system is zero,
T
R ra = D ar ⋅ M aa Eq. 13-162
R ra ⋅ u ar = 0 Eq. 13-163
CHAPTER 13 349
Solution Sequences
is generally a dense matrix, with non-zero terms at most DOFs with mass. It
R ra
could be used to reduce the K aa matrix with equations similar to those used for
multi-point constraints, but that would produce a dense K ll matrix, much more
expensive to decompose and solve than the Kll used in the manual support method.
A solution method is used which increases solution costs only slightly.
The r-set l-set partitioning is not needed. The constraint equation is appended to the
a-set equations to define the l-set equations in their augmented form,
K ll ⋅ u l = P l Eq. 13-164
T ua P al
K aa R ra
= Eq. 13-165
R ra 0 rr –qr 0r
qr are the unknown forces required on the system to produce equilibrium. They
should be computational zeros, and are output with the SPC forces. The equation
involving the augmented Kll is solved directly for u l . u a and q r are then partitioned
from u l . In superelement analysis the constraints are applied in the residual
structure only.
Matrix Decomposition in SubDMAP SEKRRS

There are two major branches through this SubDMAP, standard static analysis and
static analysis with inertia relief effects.
Inertia Relief Not Selected
When inertia relief is not selected and there are no SUPORTi entries present the a-
set is decomposed directly in the DCMP module,
T
K aa = L aa ⋅ D aa ⋅ L aa Eq. 13-166
L aais lower triangular, and D aa is diagonal. Both are stored in an unconventional

(“packed”) format, and output in a single data block named LLL. This name is used
because the l-set and the a-set are synonymous when no SUPORTi entries exist. The
factor matrices Laa and D aa are passed on to SubDMAP STATRS for solution of ua .
When inertia relief is not selected but there are SUPORTi entries present, the a-set
equations are partitioned into the r-set, defined on the SUPORTi entries, and the l-
set, what is “left over” after the r-set is removed.
K ll K lr ul Pl
≥ Eq. 13-167
K rl K rr ur Pr
In this context, the r-set is treated as an alternate form of single-point constraints. The
partitions of K involving r are discarded. u r is set to zero. u l is found from the first
row of this equation. Loads in P r go directly to ground, and are output with the SPC
forces, along with the product – Krl ⋅ u l from the second equation. K ll is decomposed
with the same equations used for conventional static analysis.
Decomposition of K ll
For the manual support method K ll is defined above. Its factors, used to compute the
rigid body modes are also used later to compute the static solution. For the auto
support method the factors of the expanded form of Kll are passed on.
Solution in SubDMAP STATRS

Computation of Inertial Loads
The basic assumption of inertia relief is that the structure is in a state of steady
acceleration and that all transients due to dynamic effects are negligible. The solution
is obtained in an accelerating reference frame from which all displacements are
measured. Inertial loads are automatically computed that exactly balance the
applied loads. A support system is applied to make the system numerically stable.
The inertial loads assume rigid body accelerations when computing the loads.
The derivations given here assume that the system mass matrix Mgg has been
reduced to M aa and is available, as described elsewhere. In the actual DMAP,
equivalent methods that never generate Maa are used instead, in the interests of
efficiency. Unit inertial loads are computed from Mgg , then reduced by load
reduction techniques, avoiding the need for Maa . M aa can be a large size data block,
and costly to compute. It can be shown that the results provided by the two
derivations are identical. For example, the rigid body mass matrix Mrr , defined
below as
T
M rr = D ar ⋅ M aa ⋅ D ar Eq. 13-168
can just as easily by formed from the g-set equations
T
M rr = D gr ⋅ M gg ⋅ D gr Eq. 13-169
D ar is used to compute the rigid body mass M rr about the reference point.
T
M rr = D ar ⋅ M aa ⋅ D ar Eq. 13-170
M rr is printed out with the matrix name of QRR.
The acceleration of the reference point a r due to the applied loads is computed,
CHAPTER 13 351
Solution Sequences
T
M rr ⋅ a r = D ar ⋅ P a Eq. 13-171
The acceleration a r are printed with the matrix name of URA.
The resulting inertial loads P ai and total loads P al are
P i = M aa ⋅ a r Eq. 13-172
P ai = P a – P i Eq. 13-173
For both methods, the mass matrix is reduced from M gg size to M rr size with load
vector operations that can be shown to be identical to the M aa operations described
here, except for the order of operations. The load and mass transformations are
based on rigid body modes produced from geometry.
A Comparison of Auto- and Manual-Support Solutions

When the same model is analyzed with both methods you should find that stresses,
constraint forces, and other internal quantities are identical except for truncation
errors, but the displacements may differ appreciably. This is because they use a
different reference frame. For the manual supports, the reference frame is set up so
that the motion at the r-set DOFs are identically zero. Think of a reference frame
drawn through these points, and all displacements measured relative to it. This
reference frame is moving through space at a constant acceleration rate, so that
displacements are relative rather than inertial displacements. There are many sets
of SUPORT i entries valid for one model. If you change SUPORTi DOFs you may
also expect to see changes in displacements, but not in internal loads or element
stresses.
For the auto-support option, one can visualize the reference frame as a set of axes
connected to every point with mass with a very complicated set of levers that allow
the structure to “breathe” as loads are applied to it, but keep the weighted average
motion to a zero value. If the model happens to have a grid point at the reference
point, you will observe very small motion there, computational zeros but not the
binary zeros of the manual support method. This is because this motion is
computed for the auto-support method, but assumed to be zero for the manual
method.
The choice of an r-set as defined by SUPORTi DOFs requires meeting both a

geometric requirement and a stiffness requirement. This geometric requirement is
that the set should form a statically determinate tie to ground. This can be difficult
to do with complicated geometries. The stiffness requirement is that the DOFs
should be stiff enough to provide a stable solution. A method that has been used by
some modelers is to pick points stiff points do not exist. The difficulties in picking
a stable, reliable support system was one of the main motivations for development
of the more modern auto-support system. SUPORTi entries are not allowed for the
auto support method, and will cause a fatal error exit when present.
PARAM,GRDPNT,GID is optional for both methods when superelements are not

present. It is required when superelements are present, and must be on a boundary
point attached to all superelements. This same point should be used for the SUPORTi
DOFs for the manual support method. It defines the reference point used for
generating the geometry-based rigid body vectors. The default for GRDPNT is
GID=-1, which is a flag to use the basic origin as the reference point when computing
rigid body modes. Use of a GID near the c.g. of the structure produces more
esthetically pleasing results. The rigid body accelerations produced by the solution
are then measured near the center of the structure, rather than at a point that may be
outside the structure.
Applications of Inertia Relief

Although this capability could be used in an environment where the reference frame
is accelerating steadily, such as a free-falling elevator, the more common usage is
when the modeler attempts to apply loads in equilibrium. He wishes the net
acceleration of the structure to be zero. A ship floating in the sea, with pressure loads
balancing gravity loads, is an example. Developing balanced load sets can be tedious
on large models with complex geometry. The accelerations printed out in the URA
matrix can be used as a measure of load quality. If the accelerations are negligibly
small, the load balance is adequate. If the accelerations are large, some adjustment of
the loads sets may be necessary. For good inertia relief analysis a good mass
distribution is needed. As the structural weight is usually only a small proportion of
the total weight, this means that a realistic payload mass distribution must be
modeled also.
Another use of inertia relief analysis becoming more prevalent with the advent of
auto support is in checkout of a new model. Although six and only six rigid body
modes are provided, they are provided in the form of a “flying waterbed”. To take
an extreme example, suppose that a structure is made up of two components bolted
together, but the bolts were not modeled, inadvertently. This model has twelve rigid
body modes before the constraints are applied. Both components sit on the same
waterbed. This means that all singularities will be constrained, and the model will
run to completion on the first pass. If loads are applied to one component only, the
constraints will cause the other component to move too, in a strain-free manner. A
plot of the motion of the model should reveal that there is no displacement
continuity between the two components. This should make the cause of the
unexpected behavior apparent. As the goal of this analysis is model checkout only
the structure mass distribution is usually adequate for stable solution. The payload
masses may be added later.
CHAPTER 13 353
Solution Sequences
Real Eigenvalue Analysis in SubDMAPs SEMR3 and

MODERS
The basic equation solved in these SubDMAPs is
[K – λ ⋅ M] ⋅ φ = 0 Eq. 13-174
where K and M are real and symmetric, and M is positive semi-definite. This
equation is defined for the a-set variables when called from MODERS for overall
structural modes of vibration, and for o-set variables when called from SEMR3
during CMS (Component Mode Synthesis) analysis. Many steps are taken in
DMAP to better automate the removal of DOFs that are ill-defined, and to augment
the mode shapes with residual flexibility effects. The SubDMAPs that perform
these operations are described below.
Although a knowledge of DMAP is not necessary for dynamic analysis, such a

study can increase the understanding of the steps that are performed.
Massless Mechanism Identification in SubDMAP MMFIND

SubDMAP MMFIND identifies potential instabilities in the matrices used for real
eigensolutions. The K and M matrices presented to the READ module for real
eigensolutions may contain massless mechanisms (MMs). A massless mechanism
has a constraint shape X such that
T
X ⋅ K ⋅ X = εk Eq. 13-175
T
X ⋅ M ⋅ X = εm Eq. 13-176
T
X ⋅ X > 0.0 Eq. 13-177
If φ contains the eigenvectors of the system, then
T
X ⋅φ = 0 Eq. 13-178
T
X ⋅M⋅φ = 0 Eq. 13-179
The constraint shapes have non-zero terms only where the eigenvectors have zero
terms. The massless mechanism shapes can be thought of as "deficient"
eigenvectors describing parts of the system that are unconnected (or very weakly
connected) to other parts of the structure that have both stiffness and mass that are
active in the well-conditioned eigenvectors. ε k and ε m may be binary zeroes, or
computational zeroes due to truncation effects. An "eigenvalue" λ associated with
massless mechanism shapes is
λ = εk ⁄ εm Eq. 13-180
λ is indeterminate in the limit as the ε i approach zero. It can be a very large number,
a very small number, and either negative or positive. It is likely to change a large
amount when small changes are made to the model. This is the definition of an
unstable eigensolution. The assumption taken here is that the cause of this condition
is either stiffness and/or mass effects inadvertently left out of this part of the model,
with the result that the load paths are ill-defined. This load path is constrained, and
the software issues a warning message. That is, this mode of motion is later removed
from the K and M matrices prior to eigensolution by steps equivalent to single
point constraint elimination.
The massless mechanisms are defined from the shifted stiffness matrix
A = K + s ⋅ M , where s is a shift that you may change from its default value with
the parameter PARAM, SHIFT1, (default=1.234 Hz.). The default is chosen to make
detection of rigid body modes as MMs unlikely. A is decomposed into its factors,
A = L⋅D⋅L Eq. 13-181
The ratios of the diagonal terms of A over D are stored in a vector Ry. Terms of D
approach a value of zero at the ends of massless mechanisms. Terms of Ry larger in
magnitude than user parameter MAXRATIO are reset to unity and are kept in vector
R1. It can be shown that each of these terms is at the end of a massless mechanism
constraint shape. R1 is expanded into a set of loading vectors P, with one unit term
per column. P is used along with the factors to find the constraint shapes,
A⋅u ≤P Eq. 13-182
The constraint shapes are normalized, and small numbers are discarded. Each
column of u is then a potential constraint mode. It could also be a rigid body mode.
The rigid body modes are detected by computing the diagonal terms of the
generalized mass matrix for these shapes,
Mg = u ⋅ M ⋅ u Eq. 13-183
Terms of Mg ( i, i ) greater than PARAM, MECHFIL, default value = 1.e-6 are an

indication that the corresponding column is a rigid body mode. Columns of this
type in u are discarded to produce u1, the matrix of constraint mode shapes. The
location of the end of the constraint shapes is passed back in R. The constraint mode
shapes themselves are passed back in MECHU. R is used later to remove rows and
columns in K and M to constrain the massless mechanisms. The nonzero terms in
MECHU are used to indicate which DOFs will discard loads applied to them,
because they will have zero coefficients in the true eigenvectors of the system,
leading to a null contribution to generalized force for all modes. MECHLD contains
P, and is intended for a feature not yet implemented.
CHAPTER 13 355
Solution Sequences
Massless Mechanism Control in SubDMAP XREAD

At present, the auto-omit DMAP is duplicated at several places, followed by a call
to XREAD. It is therefore an implicit part of the XREAD operations now, and may
be moved to XREAD in a later version to reduce the duplication of code.
SubDMAP XREAD modifies the input matrices used for eigensolution, when
needed, then restores the modified solutions back to the original size of the
matrices. Solutions for both structural vibration and buckling analyses are
provided. This discussion describes vibration analysis first, with a concluding
section on changes made for a buckling solution.
Auto-Omit Operation
All the eigensolution methods require that each column of [ K ] plus [ M ] have some
nonnull terms. Any degree-of-freedom that has only null terms in both matrices is
regarded as undefined and causes a fatal error. The GIV and HOU methods of
eigenvalue extraction also require that the mass matrix be positive-definite. The
MGIV, MHOU, and the Lanczos methods allow null columns in the mass matrix but
are generally faster if the massless degrees-of-freedom are removed by the static
reduction method described in “Static Solutions in SubDMAP SEKRRS” on
page 324. However, the static reduction tends to increase the number of active
columns in the a-set matrices. The INV, SINV, and Lanczos methods, by contrast,
are generally slower if the static reduction is performed on massless degrees-of-
freedom, because these methods take advantage of sparse matrices, whereas the
other methods do not.
Three options are provided to deal with the various null column possibilities. If
PARAM,ASING,-1 is used, any degrees-of-freedom with null columns in both the
stiffness and mass matrices (“undefined degrees-of-freedom”) are identified and a
fatal error exit is taken. This is an option to use to identify modeling errors before
performing an expensive eigenvalue solution.
If ASING = 0 (the default) the method used depends on the eigenvalue extraction
method. In the tridiagonal methods of eigenvalue extraction, degrees-of-freedom
with null columns in the mass matrix are identified and placed in the w-set.
M xx M xw
[M] = Eq. 13-184
M wx M ww
where [ M ww ] and [ M wx ] are null by definition. Degrees-of-freedom in the r-set

(SUPORT Bulk Data entry) must have mass, or a fatal error will result. The stiffness
matrix is similarly partitioned.
K xx K xw
[K] = Eq. 13-185
K wx K ww
with until diagonal terms placed on null columns of [ K ww ] . The static condensation
matrices are formed
[ K ww ] [ G wx ] = – [ K wx ] Eq. 13-186
and
T
[ K xx ] = [ K xx ] + [ K wx ] [ G wx ] Eq. 13-187
This method will prevent errors due to mass matrix singularly in the tridiagonal
methods if the singularity is caused by null columns in [ M ] . It will not detect
singularities due to mechanisms (caused, for example, by a point mass offset from a
grid point). This type of singularity does not necessarily prevent obtaining
eigensolutions, but will cause poor numerical stability. The MGIV or MHOU
methods do not suffer from numerical instability due to singular or nearly singular
mas matrices. They are therefore more reliable than the GIV HOU methods. In well
conditioned problems, however, it is unnecessary to ? automatically select the
appropriate method. The AHOU method is the most modern, reliable method.
The static reduction will reduce the cost of solution and minimize the number of
uninteresting high roots found at the value of computed infinity due to the
singularity of the mass matrix. Note that the static reduction used here does not
introduce any approximation, as the “omitted” that is w-set terms have null mass
values. The introduction of unit diagonal terms on null columns of [ K ww ] does not
introduce approximations because the terms that multiply these diagonal terms (the
corresponding rows of [ G ?? ] are null. The same results can be achieved at higher cost
by first eliminating null rows and columns of [ Kww ] by the single point constraint
operations of “Single Point Constraint Operations in SubDMAPs SEKR, SEMR2
and SEMRB” on page 316 followed by static reduction.
The solution for the system [ K xx – λ ⋅ Mxx ] ⋅ φx = 0 is provided. Rigid body shapes D xr
are provided when SUPORTi entries exist, causing an r-set. The factor of K xx , named
LLL, is used for computing approximate solutions for buckling analysis, to better set
the range of eigenvalues to be searched. For the tridiagonal methods, K and M have
DOFs with zero mass removed by an auto-omit method in a prior subDMAP. For
other methods of analysis, K aa and Maa are used directly for residual structure
modes, and K vv and Mvv for CMS (Component Mode Synthesis) modes.
The modes of the system are computed in the READ eigensolver module. If the
solution appears to be of good quality the subDMAP returns with no other
operations performed. If the modes can not be found, or appear to be of low quality,
CHAPTER 13 357
Solution Sequences
the module sets a flag (Neigv parameter). This flag causes the MMFIND subDMAP
to be called. Given K and M, it finds a high-ratio vector R and a matrix of constraint
mode shapes MMECHA. These terms are defined in the MMFIND section above.
There are non-zero terms in R at the end of massless mechanisms. These

mechanisms can be removed with SPC-like operations using R as a partitioning
vector,
K xx = K22 K21 Eq. 13-188

K12 K11
is partitioned similarly, to obtain M22. Partitions involving the "1" subscript are
M xx
discarded. The modified eigenproblem solved is then [ K22 – λ ⋅ M22 ] ⋅ φ 2 = 0 . φ2 is
merged to the size of the original K matrix by inserting null rows for the "1-set"
DOFs,
φ2
φx = Eq. 13-189
01
The auto-omit reduction is backed out by the operations
φ w = G wx ⋅ φ x Eq. 13-190
φx
φ = Eq. 13-191
φw
Real Eigenvalue Analysis in the READ Module

Real eigenvalues and their associated eigenvectors are determined in READ (Real
Eigenvalue Analysis – Displacement). In the case of normal mode analysis, READ
extracts real eigenvalues from the equation
[ K – λM ] { u } = 0 Eq. 13-192
There are seven basic eigensolution methods. The Inverse (INV and SINV) Method
obtains eigensolutions by iterations based on the equation
[ K – λi M ] { φi + 1 } = [ M ] [ φi ] Eq. 13-193
where λ i is an estimate of the eigenvalue. This method is best suited to large

problems with sparse matrices, where only a few eigenvectors are desired.
The tridiagonal methods, Givens (GIV and MGIV) and Householder (HOU and
MHOU), first mass-scale the equation
T
[M] = [L ][L] Eq. 13-194
where [ L ] is a lower-triangular matrix. The [ K ] and [ M ] matrices are pre- and

T
post-multiplied by [ L ] – 1 and [ L– 1 ] , respectively, to obtain the standard form
[ J – λI ] { φ } = 0 Eq. 13-195
T
where [ J ] = [ L – 1 ] [ K ] [ L– 1 ] and [ I ] is the identity matrix. Note that the mass matrix is
in effect inverted, so that it must be nonsingular. The [ J ] matrix is transformed to
tridiagonal form for economical computation of eigenvalues and eigenvectors. The
eigenvectors in the tridiagonal basis are then back-transformed to the initial basis.
The Modified Givens (MGIV) and Modified Householder (MHOU) Methods use the
shifted matrix
[ K + λs M ] Eq. 13-196
in place of [ M ] in the discussion above, thereby allowing the mass matrix to be

singular if there are compensating terms in the stiffness matrix. The GIV, MGIV, and
MHOU methods are best suited to small problems, or for large problems where
many eigenvectors are required, after extensive static reduction or dynamic
reduction is performed.
The Lanczos method uses a block shifted algorithm. Sets of vectors obtained by a
recursive form are used to reduce the problem to a reduced block tridiagonal form.
The eigensolutions are computed in the reduced basis by a QL algorithm, then back-
transformed to the original basis. This is currently the most modern method, and
should be considered for all large-size problems. Since the method takes advantage
of sparsity in the input matrices, it is most economical when used without static or
dynamic reduction. It is requested with either the EIGRL or EIGR Bulk Data entries.
The AHOU method combines the best features of all of the tridiagonal methods, and
is the recommended method at present. It first decomposes M. If the factors are
stable, the HOU method is used. If the factors may be unstable, the MHOU method
is used instead.
Modal Solution Preparation in the MODERS subDMAP

MODERS is called when solving for the eigensolution of the residual structure, with
a-set operations. It is used for both structural mode and buckling analysis.
In buckling, the r-set caused by SUPORTi entries are treated as equivalent SPCs. The
rows and columns associated with r-set DOFs are discarded before eigensolution.
For structural analysis, the auto-omit operations take place for the tridiagonal
methods. See the XREAD Section for details.
CHAPTER 13 359
Solution Sequences
There are also operations for ACMS and indexing of output tables present in
MODERS. They can also be found in many other SubDMAPs. Briefly, ACMS
(Automated Component Mode Synthesis) is based on breaking a large model into
superelements automatically, solving it in pieces, then re-assembling the pieces so
that the output appears to be from a conventional model. Some of the partitioning
burden that was placed on you in manual superelement analysis proved difficult to
automate, so more automatic alternate methods have been developed for ACMS
that require less user input. The automatic support system described in the
RESVEC SubDMAP section is a case in point. Most of the ACMS innovations are
moved to conventional analysis, over time, after they have been tested in field
usage.
Table indexing is a method to allow direct access to data in large sequential files, to
speed up interactive analysis. This work is being done as part of the API project.
Residual Flexibility in the SubDMAP RESVEC

Residual flexibility augments the mode shapes with shape functions found from
static analysis with various types of loading functions. These auxiliary
displacement shapes reduce errors due to modal truncation, particularly near the
area of load application. These shapes are modified or discarded to make them
linearly independent of the mode shapes. They are then made orthogonal to the
mode shapes with a second eigensolution in a reduced basis. This augmented
system provides increased accuracy for modal dynamic analysis.
The RESVEC subDMAP is called in several contexts. When used in the residual
structure, it operates on a-set matrices such as K aa . When used in CMS, it operates
on o-set matrices such as Koo . The naming convention used in the subDMAP is to
use the generic set name x. For example, K xx implies K aa when called from the
residual structure operations, and Koo during CMS. In this section we will use the
a-set nomenclature, to make the discussion less abstract. The notation "In CMS"
below implies that we are then talking about operations unique to CMS-only in that
paragraph, that is, o-set operations. Other operations for CMS operations are
similar to those for the a-set.
Residual flexibility has proven to be of great value in the Automated Component

Mode Synthesis capability ("ACMS") because the structure is partitioned
automatically based on economic criteria, rather than engineering criteria.
Economic criteria sometimes pick boundaries that are non-optimal for accuracy.
The residual flexibility operations recover the lost accuracy. This ability has driven
development in this area. The new capabilities generated for ACMS also carry
forward to more conventional analysis techniques. The DMAP implementation
has several branches unique to ACMS, and a reset of default values. The main
change is that RESVEC is automatically requested when ACMS is requested, but
must be requested explicitly for conventional analysis. The method to request this
capability is by adding PARAM, RESVEC, YES. The default values for parameters
described below is for the conventional (non-ACMS) option. Most options are
automatically turned on when ACMS is requested.
As in other subDMAPs, fluid-structure analysis of modal data is broken into two

parts, a part for structural modes only, and a part for fluid modes only. These effects
are coupled after RESVEC. A review of the subDMAP compilation shows the
complication resulting from this partitioning and the looping through several
operations, once for fluid and once for structure. The complexity is ignored in this
section by largely ignoring fluid analysis. The equations describe the structural path.
The fluid path can be inferred to be similar.
Finally, some of the steps described in this section are actually performed in other
SubDMAPs. They are described here for clarity, with a notation of the SubDMAP
where the actual DMAP statements may be found.
Parameters Unique To Residual Flexibility Operations

The comment section of the DMAP source code is reproduced here, with extensions
such as the default value of NDDL parameters. The recommended parameters to use
are param, resvec, yes, and param, resviner, yes. The first is required to select any
type of residual vector calculation. The second adds load cases for loads due to
inertia effects from rigid body acceleration to the applied load cases. It has been
found to improve results for a small incremental cost.
$ resvec default='no'
$ The main user parameter. Selects all resvec operations, as qualified
$ by the other resv* parameters. Set to 'yes' for any resvec
$ operations.
Tuning parameters. Used only for unusual models. The defaults were set after
running many test problems
$ resvrat default = 1.e8
$ filter value for discarding trial vectors with little
$ independent content. Make it smaller to discard more vectors
$ resvpgf default = 1.e-6

$ filter on partitioning vector. Make it larger to dicard more vectors
$ resvofxs default = .001

$ [in the DMAP, but no longer in use]
Intermediate Output Control

$ resvse default='no'
$ The default 'NO' will skip the printing of the strain
$ energy of the static shapes (labeled EXTERNAL WORK).
$ Use RESVSE='YES' to print the strain energy.
CHAPTER 13 361
Solution Sequences
Miscellaneous
$ resvsli default = 'yes'
$ The default 'YES' will test for linearly
$ independent shapes. Use RESVSLI='NO' to skip the tests.
$ resviner default = 'no'

$ The default (NO) is to NOT use inertial loads to obtain trial vectors.
$ Use 'yes' to use inertial loads as a source for
$ trial vectors
$ resvso default = 'yes'

$ The default 'YES' will clean up the static vectors to
$ insure orthogonality to modes. If the stiffness
$ matrix is well conditioned this is unnecessary, use
$ RESVSO='NO' to skip the reorthogonalization.
$ resmeth default = 'ahou'
$ method used for eigensolution in reduced basis.
$ 'lan' is an alternate method, somewhat more robust but slower for this
$ class of reduced basis problem.
The research legacy of this capability is evident in the many user parameters for
specific functions, some of which appear only in TYPE statements. Most are used
only when you suspect problems in your analysis. Their defaults, and even names,
are likely to change from version to version. The res*-type parameters current with
the version you are using can be determined by compiling the RESVEC SubDMAP,
and searching for, y, parameter names in TYPE statements. The default values of
NDDL parameters can be determined compiling the NDDL statements with an
input file like this:
$ file nddl*.dat
compile nddl=nddl nolist noref deck $
cend
begin bulk
enddata
The .pch file output format, provided when "deck" appears on the compile
statement, is more readable with an editor than the .f06 output file. An example of
some lines of this output is:
PARAM RESVEC ='NO' TYPE=CHAR8 PATH=SEID LOCATION=MASTER
PARAM RESVSE ='NO' TYPE=CHAR8 PATH=SEID LOCATION=MASTER
PARAM RESVSLI ='YES' TYPE=CHAR8 PATH=SEID LOCATION=MASTER
PARAM RESVINER ='NO' TYPE=CHAR8 PATH=SEID LOCATION=MASTER
PARAM RESVSO ='YES' TYPE=CHAR8 PATH=SEID LOCATION=MASTER
PARAM RESVRAT =1.E+8 TYPE=RS PATH=SEID LOCATION=MASTER
PARAM RESVPGF =1.E-6 TYPE=RS PATH=SEID LOCATION=MASTER
Load Generation
Three types of loads are used to generate auxiliary shapes. Their selection is
controlled by user parameters and other bulk data input. The parameter name and
its default are listed by each load type. Adding new load columns has little
incremental cost, so provisions are made for many types of loading conditions, to
avoid the need for re-calculation of residual flexibility effects on restarts. The types
of loads chosen are those most likely to contain effects that may be truncated by
modal analysis.
• Point Loads (USET,U6 Bulk Data entry; PARAM, RESVEC,YES)

This is a provision for cases where the DOFs to be loaded in later runs are
known a priori, but the magnitudes of the loads are not known. This type
of loading is described not by conventional load bulk data entries, but by
an entry used to define a special set, the u6 set. This option is best suited to
loads on a few DOFs. This set is not used for other purposes in the solution
sequences, so it is used in this context to define places where unit point
loads will be applied later. Each u6 DOF causes one additional loading
condition consisting of a point load at DOF u6. These loads are labeled P g1
in the subDMAP.
• Applied Loads (Conventional dynamic loading entries; PARAM,
RESVEC,YES)
All dynamic loads defined on DAREA Bulk Data entries are converted to
equivalent FORCE entries at the start of the program. All loads selected
from a DLOAD Case Control command that apply to grid and scalar points
are processed for static loads. These load entries are processed by the static
load generator (SSG1 module) to produce the P g load vectors. These are the
spatial loads that will be multiplied later by time-varying functions in
transient analysis, or frequency-varying functions in frequency response
analysis.
Appended, Reduced Loads

Loads from the prior two possible sources are appended into one load matrix,
P g2 = P g1 P g Eq. 13-197
The loads are partitioned and reduced,
Po
P g2 = Eq. 13-198
Pax
T
P a = R ga ⋅ P g2 Eq. 13-199
where R ga symbolizes the several-step load reduction described for static loads
earlier. P a is carried forward when processing residual structure modes. P o is
carried forward when processing CMS modes. They will be given the generic name
P a in the remaining discussion.
CHAPTER 13 363
Solution Sequences
Inertia loads due to unit rigid body acceleration. (param,RESVINER, YES; PARAM,
RESVEC,YES)
Rigid body displacements D gr for 6 unit motions of a reference point are computed
from geometry. The a-set DOFs are partitioned out to form D ar . The inertia loads
due to unit accelerations of these shapes are computed,
P aa = M aa ⋅ D ar Eq. 13-200
These loads are appended to all prior selected loads to make the final load matrix
P a1 , consisting of u6-set effects, applied load effects, and rigid body inertia load
effects.
Free Boundary Effects (CMS only)

When free boundary points are defined by CSETi entries, the mode shapes are
computed for the v-set, which includes the o-set and the c-set. It is necessary to
reduce the v-set effects to o-set effects. The v-set may include rigid body modes that
the c-set constrains. The derivation and motivation for these steps is described
elsewhere. The resulting steps implemented in the subDMAP are:
• Discard any rigid body modes by counting the number of r-set points.
• Reduce the mode shapes from v-set to o-set size,
P o2 = P o1 – G ox ⋅ φ x Eq. 13-201
where x symbolizes the union of the c- and r-sets, locally.

P o2 replaces P a1 in the ensuing discussion.
Load Sweeping
Later steps will discard shape functions that are not linearly independent of the
mode shapes and prior shape functions. Some of these shapes that would be
discarded can be identified by inspecting the loads that produce them. These loads
can then be discarded before their residual vectors are calculated, saving the
expense of calculating shapes that will eventually be discarded.
Loads in P a1 are normalized to produce P an . Some portion of the loads in P an can

be expressed as modal loads, and therefore do not need to be carried further. The
derivation of the sweeping equation used to eliminate these loads starts by
expressing the loads as a linear combination of the inertial modal loads
P = M aa ⋅ φ a ⋅ y Eq. 13-202
where y is unknown. Multiplying both sides by φTa , and solving for y ,
–1 T
y = Mi ⋅ φa ⋅ P Eq. 13-203
–1
where Mi – 1 = [ φTa ⋅ Maa ⋅ φa ] , the inverse of the modal mass.
These equations are used to solve for the swept loads P a2 , based on the normalized
loads P an ,
–1
P a2 = P a n – M aa ⋅ φ a2 ⋅ M i ⋅ φ a2 ⋅ P an Eq. 13-204
If the modes adequately express the generalized load caused by a column of P an , the
corresponding column of P a2 will be numerical noise, consisting of numbers near
computational zero. Your input used to set this filter is PARAM,RESVPGF, with a
default value of 1.E-6. Columns with little content are discarded from P a2 , leading
to the swept load vectors P af .
Static Solution for Shape Functions

If the structure contains rigid body modes it is not stable for static analysis. This
problem is solved by one of two methods:
Manual SUPORT (SubDMAP SEKRRS)

If you supply SUPORTi entries, the r-set DOFs it produces are used as "temporary"
SPCs, applied to allow solving for the static shapes, but removed before computing
modes. Partitions involving "r" are discarded.
K ll K lr
K aa ≥ Eq. 13-205
K rl K rr
P l2
P a2 ≥ Eq. 13-206
P r2
K ll ⋅ ul ≤ P l2 Eq. 13-207
u l2
u a2 = Eq. 13-208
0 r2
Auto SUPORT (SubDMAPs MODEFSRS, SEMR3) ACMS Only

If no SUPORTi entries or an inadequate set are supplied for constraining all rigid
body modes, and rigid body modes are found when solving [ K aa – λ ⋅ M aa ] ⋅ φ a = 0 , the
rigid body modes φr are separated from the flexible modes based on their natural
frequency being smaller than a user parameter, PARAM, FZERO, (default = 1.E-4
Hz). φ r is used to make constraint equations MP ar ,
CHAPTER 13 365
Solution Sequences
MP ar = M aa ⋅ φ r Eq. 13-209
These equations are appended to Kaa using Lagrange Multiplier Techniques, and
solved directly,
K aa MP ar u a2 P a2
= Eq. 13-210
T – q r2 O r2
MP ar O rr
The matrix on the left is named KLL in the subDMAP. It is similar to the KLL
produced above by partitioning the larger K aa matrix in that it is stable for static
solution, but differs in that it is larger than K aa by the addition of the constraint
equations. These equations enforce the condition expressed by the bottom row,
T
MP ar ⋅ u a2 = 0 Eq. 13-211
This is a statement that the c.g. of the structure is at rest on average, but the structure
is free to deflect about it. This approach will produce different u a2 vectors than
those produced by the partitioned solution. In the partitioned solution one set of
points is grounded, and all loads are transmitted through the structure to them. A
load on an explicit r-set point, for example, will go directly to ground, and will
cause zero deflection. This implies that some skill is required in selecting DOFs for
SUPORTi entries. With the LMT approach the constraints to ground are
distributed, so that loads applied anywhere will tend to load the entire structure.
Solution for Trial Vectors by Static Analysis

Inertia relief analysis is NOT performed, despite the presence of SUPORTi entries,
usually used in the context of inertia relief. In this context, SUPORTi data or its
equivalent for the auto-support feature is used for SPC-type operations. In CMS at
present, Koo may not contain rigid body modes. There is no provision to constrain
them with SUPORT-type inputs. DOFs may be removed from the o-set to make it
stable. For example, in superelement analysis, an interior point active in the rigid
body mode, which by default is in the o-set, can be made an exterior point. This
changes it to the a-set, which is held fixed when computing CMS modes. If the rigid
body modes still exist downstream, they can be constrained in residual structure
calculations.
The u a2 trial vectors, or "raw static shapes", as they are called in the SubDMAP
comments, may not be linearly independent of the mode shapes, and are certainly
not orthogonal to the mode shapes or each other. An orthogonalization process is
now used to insure that the resulting shapes are orthogonal to the mode shapes and
each other. Shapes that have little information after this process are discarded.
Orthogonalization
The trial vectors u a2 may not be linearly independent of each other or the mode
shapes. They are processed to find a set of combined mode shapes and linear
combinations of the trial shapes that are all orthogonal to each other with respect to
the mass matrix. This reduces the risk of ill-conditioning when using these matrices
for solution later.
The trial vectors are appended to φa to form the combined set φua1
φ ua1 = φ a u a1 Eq. 13-212
An optional pre-sweeping of trial vectors is done when requested by PARAM,

RESVSO, YES. The derivation of the sweeping equation is similar to that for
sweeping loads above, with the equation for x , the vector of linear combination
coefficients, being
x = M i – 1 ⋅ φ a ⋅ M aa ⋅ u a2s Eq. 13-213
The swept displacement shape vectors φ ua1 are then determined by eliminating x ,
–1
φ ua1 = u a2s – φ a2 ⋅ M i ⋅ φ a2 ⋅ M aa ⋅ u a2s Eq. 13-214
If the optional sweeping is not selected φ a1 is merely equal to φ a2 .
The u a1 vectors are mass-normalized into φ ua such that the diagonal terms of the
product on the left side are unity
φ ua ⋅ M aa ⋅ φ ua = M zz Eq. 13-215
diag ( M zz ) = I zz Eq. 13-216
M zzis generally fully coupled at this point. It is decomposed with an LTL

factorization,
M zz = L ⋅ D ⋅ L , where L is lower triangular and D is diagonal.
The ratio of D over the diagonal terms of Mzz is calculated in the vector R ,
 diag ( M zz ) 
R =  ---------------------------  Eq. 13-217
 diag ( D ) 
Rows of R whose value exceeds PARAM, RESVRAT (default=1.e8) indicate columns

of φ ua that have little new content, and are discarded to produce U a3 and M zz3 . Mzz3
is diagonal in the region of the eigenvectors, but coupled in the region of the
auxiliary shapes. It is then re-named as Mzz .
CHAPTER 13 367
Solution Sequences
The generalized stiffness corresponding to M zz is calculated,
K zz = u a3 ⋅ K aa ⋅ u a3 Eq. 13-218
The eigensolution of the z-set is obtained,
[ K zz – λ z ⋅ M zz ] ⋅ φ z = 0 Eq. 13-219
The eigenvectors φ z of this reduced basis are used as a transformation to

orthogonalize U a3 ,
φa = φu a ⋅ φz Eq. 13-220
The eigenvalue table of the combined system, λ z , replace the prior eigenvalue table
that contained modes only. λ z includes both structural modes and the effects of
orthogonalized residual flexibility vectors.
Model Partitioning in the MODEFSRS subDMAP

MODEFSRS is used in the residual structure to separate the fluid and structural
components of the a-set. The modes of these components are computed
independent of one another to find φ sa and φ fa , the structural and fluid modes,
respectively. Once computed, residual flexibility effects are added with calls to the
RESVEC SubDMAP.
The eigenvectors with residual flexibility effects appended to them φ a are then
merged together to produce one eigenvector matrix,
φ sa 0
φa = Eq. 13-221
0 φ fa
For the cases where only fluid or only structural modes are requested, the identity
matrix is substituted for the eigenvector matrix that is not calculated.
Complex Eigenvalue Analysis in SubDMAP CEIGRS

Complex eigenvalues and the associated eigenvectors are calculated in CEAD
(Complex Eigenvalue Analysis – Displacement) using the inverse power method,
the determinant method, the Hessenberg method, or the Lanczos method, as you
request on the EIGC Bulk Data entry. For direct complex eigenvalue analysis and
if PARAM,ASING = 0 (default) then null rows and columns are discarded from
[ K dd ] , [ M dd ] , and [ B dd ] to form [ K xx
d ] , [ M d ] , and [ B d ] .
xx xx
In the case of a direct formulation CEAD extracts the eigenvalues from the
following equation.
d p2 + B d p + K d ] { u d } = 0
[ M xx Eq. 13-222
xx xx x
{ u xd } is then augmented with null rows to form { u d } .
In the case of a modal formulation the following equation is used:
[ M hh p 2 + B hh p + K hh ] { u h } = 0 Eq. 13-223
CEAD also normalizes the eigenvectors according to one of the following user
requests:
• Unit magnitude of a selected coordinate (“POINT”).
• Unit magnitude of the largest component (“MAX”).
as specified on the EIGC Bulk Data entry.
The complex Lanczos method is recommended for large problems with sparse
matrices. The HESS method is best suited to small dense matrices, typical of small
models or those typical in a modal formulation. The INV method is obsolete, and is
retained as an alternate method when all else fails.
Frequency Response and Random Analysis in SubDMAP

FREQRS
The frequency-dependent loads are generated in FRLG (Frequency Response Load
Generation) from the RLOAD1 and RLOAD2 Bulk Data entries. Constraints are
applied to reduce the loads to the d-set, and in the case of a modal formulation, the
loads are transformed to the h-set.
The solution for the response is performed in FRRD1 or FRRD2 (Frequency Response
Solution). In the case of a direct formulation the following equation is solved.
[ – M dd ω 2 + iB dd ω + K dd ] { u d } = { P d } Eq. 13-224
and in the case of a modal formulation, the following equation is used in:
[ – M hh ω 2 + iB hh ω + K hh ] { u h } = { P h } Eq. 13-225
The arithmetic used in the solution may be real or complex and the solution
procedure may be symmetric or unsymmetrical. The choice of arithmetic and
solution procedure is made by the program depending on the form of the dynamic
matrices.
Power spectral density functions and autocorrelation functions are computed in
RANDOM (Random Analysis) according to the information on RANDPS and
RANDT1 Bulk Data entries. Random analysis calculations are made for selected
CHAPTER 13 369
Solution Sequences
displacements, loads, single point constraint forces, element stresses and element
forces. The selection of items for random analysis is made in the XY-output request
part of the Case Control Section.
Transient Response in SubDMAPs SEDTRAN and

SEMTRAN
The linear time-dependent loads are generated in TRLG (Transient Response Load
Generation) from the TLOAD1 and TLOAD2 Bulk Data entries. The nonlinear
loads are generated from the NOLINi Bulk Data entries. Constraints are applied to
reduce the loads to the d-set, and in the case of a modal formulation, the loads are
transformed to the h-set.
The solution for the response is performed in TRD1 (Transient Response Solution).
In the case of a direct formulation the following equation is integrated over the time
periods specified via the TSTEP Bulk Data entry in subDMAP DTRANRS:
[ M dd p 2 + B dd p + K dx d ] { u d } = { P d } + { P dnl } Eq. 13-226
in order to determine the displacement velocity and acceleration response.
In the case of a modal formulation, the following equation is used in subDMAP

MTRANRS:
[ M hh p 2 + B hh p + K hh ] { u h } = { P h } + { P hnl } Eq. 13-227
If there are no direct input matrices and no nonlinear loads, an analytical

integration procedure is used rather than numerical integration. The numerical
integration procedure may use either symmetric or unsymmetrical routines. The
choice is made by the program depending on the form of the dynamic matrices.
Buckling Analysis in SubDMAP MODERS

In linear buckling, a static load is applied to the structure and the following
eigenvalue problem is solved subsequently,
d ]{φ } = 0
[ K aa + λ i K aa Eq. 13-228
i
where [ K aa ] is the linear stiffness, [ K aa

d ] is the differential stiffness, λ are the
i
eigenvalues and { φ i } are the eigenvectors (buckling modes). The critical buckling
loads are
{ P icrit } = λ i { P a } Eq. 13-229
where { P a } is the vector of applied static loads including reaction forces.

The basic assumption in linear buckling is that the differential stiffness [ Kaa
d ] is
proportional to the applied load { P a } . The assumption implies a linear material law
and small deformations. For buckling in nonlinear problems see the NX Nastran
Basic Nonlinear Analysis User’s Guide.
The linear stiffness [ Kaa ] is derived from the variation of the stresses, the differential
stiffness [ K aa
d ] is derived from the variation of the strains keeping the stresses
constant.
SOL 105 is the recommended solution sequence for linear buckling in static analysis.
The Case Control Section must contain at least two subcases. A static loading
condition is applied in the first subcase. The load is defined with a LOAD,
TEMP(LOAD), or DEFORM Case Control command. The load may also be defined
with enforced deformations on the SPC Bulk Data entries. In the second subcase, a
METHOD command must appear to solve the eigenvalue problem. The Lanczos
method (EIGR Bulk Data entry) is the recommended eigenvalue solver for linear
buckling.
Linear buckling is available in the following solution sequences.
SOL 105 Linear buckling in statics with superelements,

including original design sensitivity (DSA).
SOL 5 Linear buckling in statics.
The linear buckling capability is available for the following elements: CONROD,
CROD, CTUBE, CBAR, CBEAM, CBEND, CQUAD4, CTRIA3, CTRIA6, CSHEAR,
CHEXA, CPENTA, AND CTETRA.
The differential stiffness formulation for the CQUAD4 and CTRIA3 elements was
changed in Version 68. Numerous spurious modes appeared in linear buckling of
thin shell structures using versions prior to Version 68. The improved results of
Version 68 can be produced with a version earlier than Version 68 by overlapping
two CQUAD4 or CTRIA3 elements, one with bending stiffness only and one with
membrane stiffness only. The pre-Version 68 method of calculating the differential
stiffness may be activated in Version 68 by setting the system cell NASTRAN
SYSTEM(170)=1 in the NASTRAN statement. The Version 68 method is the default,
NASTRAN SYSTEM(170)=0. The pre-Version 68 method remains in the code to
recover old results.
The following recommendations apply to linear buckling analysis.

• Use at least 5 elements per half sine wave to get reasonable results in the
buckling load.
• For buckling of 3-D shell structures, use PARAM,K6ROT,100. to assign a
stiffness to the sixth degree-of-freedom. The default is PARAM,K6ROT,0.
www.cadfamily.comTheEMail:[email protected]
following restrictions apply to linear buckling analysis.
CHAPTER 13 371
Solution Sequences
• Offsets should not be used in beam, plate or shell elements. The buckling
loads for structures with offsets are incorrect.
• Follower force effects are not included in the differential stiffness. The
calculated buckling loads are incorrect. The Bulk Data entries FORCE1,
FORCE2, MOMENT1, MOMENT2, PLOAD, PLOAD2, PLOAD4,
RFORCE, TEMP, TEMPD, TEMPP1, TEMPP3, TEMPRB describe follower
forces.
The theoretical aspects of superelement analysis are discussed in the NX Nastran
Superelement User’s Guide. Superelements are mathematically equivalent to
substructures. The major difference is in a more convenient user interface and in
the ability to process a much broader class of problems than is practical for most
other substructuring capabilities.
There are now two distinct paths through superelements in the solution sequences.
There are also two major types of superelements.
• When a BEGIN SUPER entry is present the program uses the more
modern SEP1X module to make the SEMAP table used to control partition
of superelements.
• When there is a BEGIN BULK entry but no BEGIN SUPER entries a
parallel path using the older SEP1 module is used instead.
While the newer module supports most features of both part and grid list
superelements, it does not have some of the advanced features supported in the
older module. When starting a new project you should decide at modeling initiation
which path is most suitable because transitioning from one to another can be
painful, particularly when the model is in development. If one starts on the old
path, for example, and then adds a new feature that causes the program to
automatically switch to the new path, some of your modeling input may be
invalidated. This may cause considerable, confusing, and unnecessary work to
transition to the new path.
The short answer is that the new path should be the best path for most projects,
based on the appeal of its advanced features and the likelihood that future
improvements in NX Nastran may require use of the new path. The new path is
forced by replacing the BEGIN BULK entry with a BEGIN SUPER entry in your
Main Bulk Data file. It should by the first entry in your bulk data. It allows grid list
superelements, but they need not be present.
The exception for the path chosen is when your project requires the use of features
available in the old path but not yet enabled in the new path. Some of these features
may be used in the new paths by special avoidance techniques, but these techniques
can be cumbersome. It may be more convenient to stay with the older methods until
the newer methods catch up.
Terminology
• Newest Option: The partitioned bulk data file superelements (parts) are
the form introduced in Version 69. They are sometimes called
substructures. A major feature of parts is ID independence. That is, element
IDs and grid points IDs may be duplicated across different parts, although
they must be unique within a part.
• Intermediate: The Main Bulk Data Superelements (superelements)
introduced in Version 47 are sometimes called grid list superelements (list
s.e.s). When used in the new path, they have more restrictive rules than in
the old path.
• Oldest Path: List s.e.s only are used in the old path. There are no formal
names at present to differentiate the intermediate and old flavors. Some
advanced features are available only in this path, at present.
A subtle difference between the two paths is that the new path always uses a single
level configuration for list s.e.s if there is no SETREE-type entry. It does this by
defining new boundary points in the residual structure when some are needed but
not supplied by you. The old path automatically converts to a multilevel
configuration when required, regardless of the presence of DTI, SETREE entries. It
has no ability to generate new boundary points. This type of change when switching
from the old path to the new, on purpose or by accident can be confusing.
Criteria for Selecting Part Versus List Superelements

Each of the two major forms has overwhelming advantages in certain contexts. Some
of these advantages are of a temporary nature, as the rest of the technology in
supporting FE tools catch up to the newest methods. However, other relative
advantages are due to the inherent difference in concept between the two major
branches that will be relevant into the foreseeable future. It is likely that both
branches will be maintained and extended because of their unique capabilities. The
following criteria should be considered when making a choice:
• Converting an existing non-superelement model to a part configuration is

a labor-intensive process at present. No migration aides are provided.
Converting it to a list superelement requires addition of only SESET entries,
a trivial task by comparison.
CHAPTER 13 373
Solution Sequences
• When debugging a new model it is much easier to comprehend and more

efficient when it is in a superelement configuration. Errors in one area will
not obscure errors in other areas, and the model solved is smaller, leading
to faster turnaround. It may be more economical to run it in a non-
superelement configuration for production runs after the model is stable
and requires no more changes. If it is a list superelement this means
merely removing the SESET entries, after which it is automatically an
exactly equivalent non-superelement model if advanced features such as
those provided by the RELEASE and CSUPEXT entries are avoided. If it is
a part superelement there can be a large labor cost for conversation, as all
duplicated IDs will need to be changed, duplicate points at boundaries
will need to be resolved, etc. A list s.e. model, to make virtue of necessity,
enforces rules such as unique IDs when updating the model that make re-
conversion to a non-s.e. configuration an almost automatic process.
• When starting a new modeling project where more than one person
provides input, and most particularly when more than one organization
contributes to the model, the higher degree of data control for part
superelements is attractive. There is no need to legislate acceptable ID
ranges for each part. It is possible to lock the description of a part into its
boundary matrices, in a data base marked with a time stamp, so that there
is little questions about who is responsible for the behavior of each part.
• Many GUIs can be used to modify list superelements because the rules for
ID independence are the same for GUIs as they are for list s.e.s Part
superelements can be modeled in GUIs only one at a time. This makes
boundary matching, in particular, cumbersome.
Features Not Supported in the New Path

• Fluid Structure Analysis: This capability is supported in the new path with
the “old” path rules:
a. All fluid elements must be in a special list superelement, and/or in the
residual structure
b. This special superelement may have fluid elements only; structural
elements or grid points are not allowed
c. One or several fluid cavities may be in the fluid superelement
d. Fluid elements may be connected to structural elements in list
superelements, or in the residual structure.
e. Elements in part superelements may not be connected to fluid
elements.
• p-Elements: Not supported at all. It may be possible to segregate the

portions of the model containing p-elements into external superelements,
but this will be a labor-intensive process. No documentation is available on
this topic at present.
• RELEASE Entries: Allowed only in the old path. There are semi-heroic
modeling techniques that may be used to simulate this feature in the new
path, but they are not documented.
• SExSET Entries: Most set entries for non-superelement models such as
ASET have synonyms for use in list s.e.s such as SEASET. The synonym is
not accepted in PARTs, but the simpler format of ASET, for example, is
accepted.
Features Not Supported in the Old Path

In general, features associated with automatic boundary control are not supported.
The QRG is not explicit on this issue. A list of entries not supported in the old path is:
SETREE, SENQSET, SEBNDRY, SEBULK, SECONCT, SEEXCLD, SELABLE, SELOC,

SEMPLN
The DTI, SETREE entry is used in place of SETREE. The action of the SEELT entry is
somewhat different. For the new path, the SEMAP will be changed to account for the
action of SEELT entries. For the old path, it is assumed that the element being moved
is one that can be moved without changing the SEMAP, such as an element attached
only to boundary points. If the SEELT requires change of the SEMAP the program
processes conventionally until it encounters the s.e. with the “illegal” SEELT entry,
then gives a fatal error exit stating that the connectivity of the superelement is
incorrect.
CHAPTER 13 375
Solution Sequences
Check List for Transitioning an Old Model to the New Path

• Replace the BEGIN BULK entry with a BEGIN SUPER entry.
• Check the SEMAP table from an old run to determine if it is in a multilevel
state. When it is, add SETREE entries that define the tree in the old run, or
convert it to a single level model if multilevel is not essential. If you have
many multilevel models to convert, there is a parameter that will request
that SETREE entries be made from the SEMAP, and placed in the .pch
(punch) file.
• Discard all RELEASE and CSUPEXT entries and p-elements. If you cannot
find reasonable replacements for them you may be forced to stay in the old
path.
Decide if this is the right time to convert to a PART model, or whether the list s.e.
capability in the new path is more advantageous, based on the discussion above.
The Case Control commands for superelement processing selection all begin with
the letters SE (SEMG, SELG, SEKR, SEMR, SELR, SEALL, and SEEX). The SEFINAL
command is used to control the order of processing of the superelements. A
subcase structure is required that defines constraints, loads, and data recovery
options for each superelement. Each subcase includes the Case Control command
SUPER in order to relate the subcase to a particular superelement and load
condition. Examples of Case Control (including plot requests) are given in the NX
Nastran Superelement User’s Guide. A set of plotting commands is required for
each superelement to be plotted, separated by the plotter commands SEPLOT and
SEUPPLOT.
Each superelement is given an identification number (SEID). You assign each grid
point to a superelement on a GRID or SESET Bulk Data entry. Identical image,
mirror image, and external superelements are defined by the CSUPER and optional
SEQSEP Bulk Data entries. The SEELT Bulk Data entry may be used to change the
superelement membership of boundary elements. The RELEASE Bulk Data entry
may be used to disconnect selected degrees-of-freedom from exterior grid points of
a superelement.
The mesh generator option, called by MESH on the NASTRAN statement, supports
the superelement capability fully, in that all GRID entries generated for a field can
be given the same SEID, and the SESET Bulk Data entry can change these SEIDs at
boundaries. The automatic grid point resequencing features are fully implemented
for superelement models.
The superelement partitioning module passes on resequencing data from SEQGP

Bulk Data entries only for grid points of a superelement which are not exterior
points for any superelement. This class of grid points can be determined from the
SEMAP table printed by the program, but may change if the order of processing is
changed. The automatic resequencing option processes each superelement
separately.
Component mode synthesis uses SEBSETi, SECSETi, SESUP and SEQSETi entries to
define the boundary condition when computing component modes. Uncoupled
solutions for statics and modes can be obtained by use of PARAM,FIXEDB,-1.
Superelement Bulk Data entries may be specified in nonsuperelement solution

sequences, but those entries pertaining to the superelement capability (e.g., the
CSUPER entry) are ignored, as well as some fields on other entries (e.g., Field 9 on
the GRID entry). However, better automatic resequencing may be possible if the
SEID data is removed.
Superelement Processing in the Structured Solution Sequences

An overview of the superelement processing in SOLutions 101-200 is shown in
Figure 13-3. Phase 0 performs the preparation and bookkeeping function. The input
data is stored on the database and scanned to prepare the SEMAP table. This table
lists the superelement to which each Bulk Data entry is to be applied. Phase 0 also
partitions out the data that defines each superelement and performs automatic
restart processing. In Phase I this data is used to generate the boundary matrices to
be assembled when combining the superelements into the total structure. Phase II
performs calculations done only on the residual structure (the superelement
processed last), such as transient response calculations or eigensolutions. Phase III
expands boundary solutions to the interior grid points of the superelement,
performs the data recovery operations for internal forces, stresses, and strain energy,
and outputs these quantities. Phase IV combines plot vectors from upstream
superelements to provide deformed structural plots for all or part of the structure.
CHAPTER 13 377
Solution Sequences
SOLs 101 through 200

Bulk Data Entries, Case Control Commands
Bookkeeping and Control (Phase 0) SubDMAP PHASE0

Generate IFP data blocks, resequence, and store.
Generate SEMAP SLIST (SEP1, SEP3)
Index to next superelement and repeat. (SEP2DR)
Partition out data for one superelement. (SEP2 and SEP2CT)
Undeformed plots.
Automatic restart processing.
Superelement model checkout.
Generation, Assembly, and Reduction (Phase I) SubDMAP PHASE1DR

Generate system structural and load matrices in subDMAP PHASE1A.
Assemble upstream stiffness, loads (statics), mass and damping (dynamics) in SubDMAP
PHASE1A.
Reduce to exterior points, dynamic reduction in subDMAPs SEKR, SEMRM, and SELR.
Eliminate ur set in subDMAPs SELRRS and SEMRRS. (residual structure only)
Residual Structure-Only Operations (Phase II) Various SubDMAPs

Formulate dynamic equations.
Eigensolutions, transient, frequency response calculations.
Data Recovery (Phase III) SubDMAP SUPER3

Generate DRLIST (SEP4)
Index to next superelement and repeat. (SEDRDR)
Partition out boundary solution. (SEDR)
Expand boundary displacements to interior points in subDMAP SEDISP.
Compute and add in fixed boundary displacements (statics only) in subDMAP SEDISP.
Compute and output stresses, strain energy, etc. in subDMAP SEDRCVR.
Upstream Plots (Phase IV) SubDMAP SUPER3

Index to next superelement and repeat. (PLVCDR)
Assemble upstream plot vectors (PLTVEC)
Plot deformed structure plots.
(Superelement module names are in parentheses)
Figure 13-3 Structured Solution Sequence Flow Chart
The input file processor operations are carried out in a conventional manner. The
grid point resequencer module resequences each superelement individually.
Some of the geometry processor operations of “Geometry Processing in SubDMAP

PHASEO” on page 308 are carried out to prepare tables of all grid points in the entire
model, useful for generation of the SEMAP table. After the SEMAP is available, the
SLIST table is assembled. This list controls the operations to be performed in that
particular run for generating boundary matrices, as selected with Case Control
commands. The GP4 module generates constraint data in the form of the USET table,
as described in “Geometry Processing in SubDMAP PHASEO” on page 308. A
separate USET table is generated for each superelement. It is similar to the USET
table of the older rigid formats except that the exterior points are automatically
placed in the a-set.
Assembly and Reduction (Phase I) in SubDMAP Phase1DR

The following operations may be carried out, as you select with superelement-type
Case Control commands. By default, all superelements and the residual structure are
given these operations serially. If requests for the SEMA (or SEMR) operation are
made on an upstream superelement and the residual structure, all superelements in
the tree between the selected superelement and the residual structure will also be
given the SEMA operation.
Superelement Matrix Generation (SEMG) in SubDMAP PhaselA

The element structural matrices (stiffness, mass and damping) are generated and
stored. The direct matrix input option (DMIG Bulk Data entry) can be used to add
stiffness, mass or damping terms to the system matrices in this operation. These
terms are selected by the Case Control commands K2GG, M2GG, and B2GG. The
parameters CB1, CB2, CK1, CK2, CM1 and CM2 can be used to scale the structural
matrices.
The SEMA module adds in stiffness terms from any upstream superelements. The
SEEX Case Control command may be used to eliminate stiffness contributions from
selected upstream superelements. The superelements that are upstream are defined
by the current SEMAP. The number of rows and columns of the upstream
superelements must be consistent with the number of exterior grid points, as defined
by the SEMAP In order to complete execution of the SEMA module, the boundary
matrices of all upstream superelements must be present in the database, and
consistent with the current SEMAP.
The SELA module adds loads from upstream superelements to the load vector for
the current superelement. The SEEX Case Control command may be used to
eliminate load contributions from upstream superelements. The requirements for
conformity of load data in the database with the current state of the SEMAP are
similar to those for stiffness, except that only rows, not columns of the load vectors
are mapped using data in the SEMAP table. The requirement on columns of all load
matrices is that they be equal to the number of columns (number of load condition)
in the residual structure. If the number of columns in all superelements and the
CHAPTER 13 379
Solution Sequences
residual structure is not the same, a user fatal error results. Static loads may be
applied to any grid point, whether exterior or interior to a superelement. Dynamic
loads may be applied directly to residual structure points. Dynamic load entries
may also reference static load sets applied to the superelements via the LOADSET
Case Control command and LSEO Bulk Data entries.
Superelement Load Generation (SELG) in SubDMAP PHASE1A

The load generation module SSG1 is executed in the static and dynamic response
solutions. This operation is not executed in the normal modes and complex
eigenvalue solution sequences. As loads may depend on structural matrices (i.e.,
thermal or inertial loads), an SELG operation must be preceded by an SEMG
operation. The resultants of these loads about the grid point listed on the
PARAM,GRDPNT Bulk Data entry are output.
Superelement Stiffness Reduction (SEKR) in SubDMAPs SEKR and SEKRRS

You may change SPCs and/or MPCs between subcases (“boundary condition
changes”). The GPSP module identities grid point singularities, and, optionally,
provides SPCs for singular degrees-of-freedom. This is followed by the constraint
eliminations described in “Multipoint Constraint Operations in SubDMAPs
SEKR and SEMR2 and SEMRB” on page 313, “Single Point Constraint
Operations in SubDMAPs SEKR, SEMR2 and SEMRB” on page 316, and
“Constraint and Mechanism Problem Identification in SubDMAP SEKR” on
page 317. If the omitted degrees-of-freedom contain mechanism-type singularities,
the grid and component number of all downstream ends of the mechanism are
output in the matrix named MECHS, and a user fatal error stops execution. The
approximate fixed-boundary natural frequencies of the superelement are printed in
the FAPPROX matrix. The boundary stiffness matrix [ K aa ] is stored in the database.
Mass and damping matrices are not processed in this phase.
SUPORTi Bulk Data entries, which define the r-set, may reference only grid points
in the residual structure. The rigid body operations for the stiffness are performed
as described in “Static Solutions in SubDMAP SEKRRS” on page 324.
Superelement Load Reduction (SELR) in SubDMAP SELR

This operation processes loads in the static and dynamic response solution
sequences. This operation is not executed in the normal modes and complex
eigenvalue solution sequences. The static load reduction of “Static Condensation
in SubDMAPs SEKR and SEMR2” on page 321 is applied, stopping after
generation of the [ P a ] matrix for superelements and [ P l ] for the residual structure.
Superelement Mass Assembly and Reduction (SEMR) in SubDMAP SEMRM

In the dynamics solution sequences, the constraint operations of “Multipoint
Constraint Operations in SubDMAPs SEKR and SEMR2 and SEMRB” on
page 313, “Single Point Constraint Operations in SubDMAPs SEKR, SEMR2 and
SEMRB” on page 316, and “Constraint and Mechanism Problem Identification in
SubDMAP SEKR” on page 317 are applied to the mass and damping matrices. The
rigid body operations of “Static Solutions in SubDMAP SEKRRS” on page 324 are
applied to the mass of the residual structure only. Dynamic reduction and
component mode calculations are performed. This allows a mass-change-only type
of restart by use of the SEMG and SEMR options and avoids the costly reduction of
the stiffness matrix in the SEKR operation if no stiffness changes are made.
The optional Bulk Data entry PARAM,ERROR,0 affects operations in Phase I. If it is

not used, the default action is to terminate the run after the first user fatal error, as in
the older rigid formats. If the parameter is used, a user fatal error causes a branch to
the end of Phase I for the superelement, and continuation to the generation,
assembly, and reduction operations for any remaining superelements. The five SE-
type commands may be executed on one run, or may be done separately. The
following sequence must be observed if these commands are executed separately:
SEKR after SEMG

SELG after SEMG
SELR after SEKR, SELG, SEMR
SEMR after SEKR
Residual Structure-Only (Phase II)

The operations on the superelements described above are also performed on the
residual structure. This is followed by operations unique to the residual structure.
Eigensolutions and direct and modal forced response calculations take place at this
point. Direct input matrices, when selected by the Case Control commands M2PP,
K2PP, and B2PP are constrained and added as in the “Formulation of Dynamic
Equations in SubDMAP GMA” on page 334.
Bulk Data entries used for static loads (i.e., FORCE, GRAV, LOAD, etc.) may be used
in direct and modal dynamic analysis. They are interchangeable with DAREA Bulk
Data entries and may be referenced by LSEQ Bulk Data entries.
Data Recovery Operations (Phase III) in SubDMAP SUPER3

The total solution vectors of the downstream superelement are partitioned to
generate the boundary solution vectors for the superelement being processed. The
boundary solution is expanded to all interior points of the superelement. The fixed
boundary component is computed and added.
CHAPTER 13 381
Solution Sequences
If SPC force output is requested, the resultants of the SPC forces about
PARAM,GRDPNT or the basic coordinate system origin are also output. The data
recovery operations of “Static Solutions in SubDMAP SEKRRS” on page 324 are
carried out. (Note that solution sequences involving complex eigenvalue or
frequency response analysis may produce complex displacement vectors. GPFDR
module output (grid point force balance, element strain energy) may be requested
for complex vectors, but only the real component of the vector is processed.)
Deformed structure plots of only elements in the superelement are generated here.
Upstream Plots (Phase IV) in SubDMAP SUPER3

The displacement vectors of a superelement and all of its upstream members are
combined and plotted at your request.
Superelement Processing
An overview of the superelement processing is shown in Figure 13-4. Phase 0
performs the preparation and bookkeeping function. The input data is stored on the
database and scanned to prepare the SEMAP table. This table lists the superelement
to which each Bulk Data entry is to be applied. Phase I partitions out the data that
defines one superelement and processes this data to generate the boundary
matrices to be assembled when combining the superelements into the total
structure. Phase II performs calculations done only on the residual structure (the
superelement processed last), such as transient response calculations or
eigensolutions. Phase III expands boundary solutions to the interior grid points of
the superelement, performs the data recovery operations for internal forces,
stresses, and strain energy, and outputs these quantities. Phase IV combines plot
vectors from upstream superelements to provide deformed structural plots for all
or part of the structure.
The input file processor operations are carried out in a conventional manner. The
grid point resequencer module resequences each superelement individually.
Some of the geometry processor operations of “Geometry Processing in

SubDMAP PHASEO” on page 308 are carried out to prepare tables of all grid
points in the entire model, useful for generation of the SEMAP table. After the
SEMAP is available, the SLIST table is assembled.
This list controls the operations to be performed in that particular run for
generating boundary matrices as selected with Case Control options.
Assembly and Reduction (Phase I)

The following operations may be carried out as you select on superelement-type
Case Control commands. By default, all superelements and the residual structure
are given these operations serially. If requests for the SEMA (or SEMR) operation
are made on an upstream superelement and the residual structure, all superelements
in the tree between the selected superelement and the residual structure will also be
given the SEMA operation.
Superelement Matrix Generation (SEMG)

The element structural matrices (stiffness, mass and damping) are generated and
stored. The direct matrix input option (DMIG Bulk Data entry) can be used to add
stiffness, mass or damping terms to the system matrices in this operation. These
terms are selected by the Case Control commands K2GG, M2GG, and B2GG. The
parameters CB1, CB2, CK1, CK2, CM1, and CM2 can be used to scale the structural
matrices.
Superelement Load Generation (SELG)

The load generation module SSGI is executed in the statics and dynamic response
solutions. This operation is not executed in the normal modes and complex
eigenvalue solution sequences. As loads may depend on structural matrices (i.e.,
thermal or inertial loads), an SELG operation must be preceded by an SEMG
operation. The resultants of these loads about the grid point listed on the
PARAM,GRDPNT Bulk Data entry are output.
CHAPTER 13 383
Solution Sequences
Bulk Data Entries, Case Control Commands
Bookkeeping and Control (Phase 0)

Generate IFP data blocks, resequence, and store.
Generate SEMAP, SLIST (SEP1, SEP3)
Generation, Assembly, and Reduction (Phase I)

Partition out data for one superelement. (SEP2 and SEP2CT)
Undeformed plots.
Generate system structural and load matrices.
Assemble upstream stiffness, loads (statics), mass and damping (dynamics).
Reduce to exterior points, dynamic reduction.
Eliminate Uf set. (residual structure only)
Residual Structure-Only Operations (Phase II)

Forrnulate dynamic equations.
Eigensolutions, transient, frequency response calculations.
Data Recovery (Phase III)

Generate DRLIST (SEP4)
Index to next superelement and repeat. (SEDRDR)
Partition out boundary solution. (SEDR)
Expand boundary displacements to interior points.
Compute and add in fixed boundary displacements (statics only).
Compute and output stresses, strain energy, etc.
Upstream Plots (Phase IV)

Index to next superelement and repeat. (PLVCDR)
Assemble upstream plot vectors (PLTVEC)
Plot deformed structure plots.
(Superelement module names are in parentheses)
Figure 13-4 Solution Sequence Flow Chart
Superelement Stiffness Matrix Assembly and Reduction (SEKR)

The SEMA module adds in stiffness terms from any upstream superelements. The
SEEX Case Control command may be used to eliminate stiffness contributions from
selected upstream superelements. The superelements that are upstream are
defined by the current SEMAP. The number of rows and columns of the upstream
superelements must be consistent with the number of exterior grid points, as
defined by the SEMAP. In order to complete execution of the SEMA module, the
boundary matrices of all upstream superelements must be present in the database,
and consistent with the current SEMAP. The GP4 module generates constraint data
in the form of the USET table, as described in “Geometry Processing in SubDMAP
PHASEO” on page 308. A separate USET table is generated for each superelement.
It is similar to the USET table of the older rigid formats except that the exterior points
are automatically placed in the a-set.
You may change SPCs and/or MPCs between subcases (“boundary condition
changes”) in SOLs 101 and 103 only. The GPSP module identifies grid point
singularities, and at user’s option provides SPCs for singular degrees-of-freedom.
This is followed by the constraint eliminations described in “Multipoint Constraint
Operations in SubDMAPs SEKR and SEMR2 and SEMRB” on page 313, “Single
Point Constraint Operations in SubDMAPs SEKR, SEMR2 and SEMRB” on
page 316, and “Constraint and Mechanism Problem Identification in
SubDMAP SEKR” on page 317. If the omitted degrees-of-freedom contain
mechanism-type singularities, the grid and component number of all downstream
ends of the mechanism are output in the matrix named MECHS, and a user fatal
error stops execution. The approximate fixed-boundary natural frequencies of the
superelement are printed in the FAPPROX matrix. The boundary stiffness matrix
[ KAA ] is stored in the database. Mass and damping matrices are not processed in
this phase.
SUPORT Bulk Data entries, which define the r-set, may reference only grid points in
the residual structure. The rigid body operations for the stiffness are performed as
described in “Static Solutions in SubDMAP SEKRRS” on page 324.
Superelement Load Assembly and Reduction (SELR)

This operation processes loads in the static and dynamic response solution
sequences. This operation is not executed in the normal modes and complex
eigenvalue solution sequences. The SELA module adds loads from upstream
superelements to the load vector for the current superelement. The SEEX Case
Control command may be used to eliminate load contributions from upstream
superelements. The requirements for conformity of load data in the database with
the current state of the SEMAP are similar to those of the SEKR operation above,
except that only rows, not columns of the load vectors are mapped using data in the
SEMAP table. The requirement on columns of all load matrices is that they be equal
to the number of columns (number of load condition) in the residual structure. If the
number of columns in all superelements and the residual structure is not the same,
a user fatal error results. Static loads may be applied to any grid point, whether
exterior or interior to a superelement. Dynamic loads may be applied directly to
residual structure points. Dynamic load entries may also reference static load sets
applied to the superelements via the LOADSET Case Control command and LSEQ
CHAPTER 13 385
Solution Sequences
Bulk Data entries. The static load reduction of “Static Condensation in

SubDMAPs SEKR and SEMR2” on page 321 is applied, stopping after generation
of the [ P a ] matrix for superelements and [ P l ] for the residual structure.
Superelement Mass Assembly and Reduction (SEMR)

In the dynamics solution sequences, the constraint operations of “Multipoint
Constraint Operations in SubDMAPs SEKR and SEMR2 and SEMRB” on
page 313, “Single Point Constraint Operations in SubDMAPs SEKR, SEMR2 and
SEMRB” on page 316 and “Constraint and Mechanism Problem Identification in
SubDMAP SEKR” on page 317 are applied to the mass and damping matrices. The
rigid body operations of “Static Solutions in SubDMAP SEKRRS” on page 324 are
applied to the mass of the residual structure only. Dynamic reduction and
component mode calculations of “Static and Dynamic Load Generation” on
page 328 are performed. This allows a mass-change-only type of restart by use of
the SEMG and SEMR options and avoids the costly reduction of the stiffness matrix
in the SEKR operation if no stiffness changes are made.
The optional Bulk Data entry PARAM,ERROR,0 affects operations in Phase I. If it

is not used, the default action is to terminate the run after the first user fatal error,
as in the older rigid formats. If the parameter is used, a user fatal error causes a
branch to the end of Phase I for the superelement, and continuation to the
generation, assembly, and reduction operations for any remaining superelements.
The five SE-type commands may be executed in one run, or may be done separately.
The following sequence must be observed if these commands are executed
separately:
SEKR after SEMG

SELG after SEMG
SELR after SEKR, SELG, SEMR
SEMR after SFKR
Residual Structure-Only (Phase II)

The operations on the superelements described above are also performed on the
residual structure. This is followed by operations unique to the residual structure.
Eigensolutions and direct and modal forced response calculations take place at this
point. Direct input matrices, when selected by the Case Control commands M2PP,
K2PP, and B2PP are constrained and added as in “Formulation of Dynamic
Equations in SubDMAP GMA” on page 334.
Bulk Data entries used for static loads (i.e., FORCE, GRAV, LOAD, etc.) may be
used in direct and modal dynamic analysis. They are interchangeable with DAREA
Bulk Data entries and may be referenced by LSEQ Bulk Data entries.
Data Recovery Operations (Phase Ill)

The total solution vectors of the downstream superelement are partitioned to
generate the boundary solution vectors for the superelement being processed. The
boundary solution is expanded to all interior points of the superelement. The fixed
boundary component is computed and added.
If SPC force output is requested, the resultants of the SPC forces about
PARAM,GRDPNT or the basic coordinate system origin are also output. The data
recovery operations of “Data Recovery Operations in SubDMAP SEDISP” on
page 338 are carried out. (Note that solution sequences involving complex
eigenvalue or frequency response analysis may produce complex displacement
vectors. GPFDR module output (grid point force balance, element strain energy)
may be requested for complex vectors, but only the real component of the vector is
processed.) Deformed structure plots of only elements in the superelement are
generated here.
Upstream Plots (Phase IV)

The displacement vectors of a superelement and all of its upstream members are
combined and plotted at user's request.
Superelement Alternate Statics Solution

SOL 101 provides an alternate method of performing the superelement static
condensation. PARAM,ALTRED,YES must be entered in the Bulk Data Section. In
some cases, the alternate solution can result in significant reduction in computer cost
for static analysis.
The stiffness and load reduction and displacement recovery are described in:
• “Static and Dynamic Load Generation” on page 328

• “Static Condensation in SubDMAPs SEKR and SEMR2” on page 321
• “Data Recovery Operations in SubDMAP SEDISP” on page 338
Static Superelement Analysis With lnertia Relief

Inertia relief analysis is available in SOL 101. PARAM,INREL,-1 or PARAM,INREL,
-2 must be present for either of these options. The following method transforms
static loads based on model geometry rather than the stiffness matrix as described in
“Static Solutions in SubDMAP SEKRRS” on page 324.
The additional operations required to perform inertia relief are described in “Static
and Dynamic Load Generation” on page 328 and “Static Solutions in SubDMAP
SEKRRS” on page 324.
CHAPTER 13 387
Solution Sequences
Assumptions and Limitations

1. The model must have six and only six rigid body modes for a continuous,
three-dimensional structure.
2. Masses on scalar points are ignored in calculating inertia loads.
3. The structural model, when given enforced displacements at the reference
point, is assumed to produce the same motion as a rigid body model. This
will be true for any structural or rigid element. It is possible, if
unconventional, to use CELASi or general elements, MPC equations, or
DMIG coefficients that will not meet this assumption.
4. Identical and mirror image superelements are not processed correctly.
Nonlinear Heat Transfer Analysis

In NX Nastran, you can perform:
• steady state heat transfer analysis

• transient heat transfer analysis
Steady State Heat Transfer Analysis

The steady state heat balance equation is given by
4
[ K ] { u } + [ ℜ ] { u + T abs } = {P} + {N} Eq. 13-230
where:
[ K ] = a heat conduction matrix

[ ℜ ] = a radiation exchange matrix
{ P } = a vector of applied heat flows that are constant
{ N } = a vector of nonlinear heat flows that depend on temperature
{ u } = a vector of grid point temperature
T abs = the absolution temperature
The components of the applied heat flow vector, { P } , are associated either with
surface heat transfer or with heat generated inside the volume heat conduction
elements. The vector of nonlinear heat flows, { N } , results from boundary
radiation, surface convection, and temperature dependent thermal loads.
The equilibrium equation is solved by a Newton iteration scheme, where the

tangential stiffness matrix is approximated by
i
i i i 3  ∂N 
[ KT ] ≈ [ K ] + 4 [ ℜ ] { ui + T abs } –  ------  Eq. 13-231
 ∂u 
and the residual vector is
i i i i i 4
{R} = { P } + { N } – [ K ] { u } – [ ℜ ] { u i + T abs } Eq. 13-232
Transient Heat Transfer Analysis

The general equation solved in transient analysis has the form
4
[ B ] { u· } + [ K ] { u } + [ ℜ ] { u + T abs } = { P } + { N } Eq. 13-233
To take phase change into consideration, the heat diffusion equation is converted
into
{ H· } + [ K ] { u } + [ ℜ ] { u + T abs } = { P } + { N }
4
Eq. 13-234
where:
[ B ] = heat capacity matrix

[ K ] = heat conduction matrix
[ ℜ ] = radiation matrix
{ P } = vector of applied heat flows that are constant or functions of
time
{ N } = vector of nonlinear heat flows that depend on temperature
{ H } = enthalpy vector
{ H· } = { dH ⁄ dt }
{ u } = grid point temperatures
{ u· } = { du ⁄ dt }
T abs = absolute temperature scale factor
The equilibrium equation is solved by Newmark’s method with adaptive time

stepping. Base on this one-step integration scheme, the time derivative of the nodal
temperatures at the ( i + 1 ) th iteration of the time step ( n + 1 ) is expressed as
i+1 1 1
{ u· n + 1 } = --------- { u ni ++ 11 – u n } +  1 – --- { u· n } Eq. 13-235
θ∆t θ
where
CHAPTER 13 389
Solution Sequences
i+1 i i
{ un + 1 } = { u n + 1 } + { ∆u n + 1 } Eq. 13-236
and
1
--- = 2 – 2η Eq. 13-237
θ
The parameter η is specified on the PARAM,NDAMP Bulk Data entry. When

η = 0 ( θ = 0.5 ) , no numerical damping is requested. In this case, Newmark’s
method is equivalent to the Crank-Nicolson method.
14
CHAPTER
Modeling Guidelines
■ Introduction to Modeling Guidelines
■ Choosing the Right Element
■ Determining Mesh Density
■ Creating Mesh Transitions
■ Reviewing Grid Point Stresses
■ Defining Consistent Loading
■ Symmetry
14.1 Introduction to Modeling Guidelines

With finite element modeling, the quality of your results depends directly upon the
quality of your model. One of the more common errors that a beginning finite element
analyst makes in modeling is to simply simulate the geometry rather than to simulate both
the geometry and the physical behavior of the real structure. This section covers the
following finite element modeling topics and guidelines:
• Choosing the right element

• Mesh density
• Mesh transitions
• Grid point stresses
• Consistent loading
• Symmetry
It is also good modeling practice to simulate and validate a new capability or a feature that
you haven’t used before with a small prototype model before applying this feature to your
production model. Model verification techniques are covered in “Model Verification” on
page 445.
14.2 Choosing the Right Element

NX Nastran contains a large library of structural elements. Often, you could use several
elements to model the same structural effects. The criteria for the selecting an element
may include its capabilities (for example, whether it supports anisotropic material
properties), its cost (in general, the more DOF an element has, the more expensive it is),
and/or its accuracy.
In many cases, the choice of the best element for a particular application may not be
obvious. For example, in the model of a space frame, you may choose to use CROD
elements if end moments are unimportant or to use CBAR elements if end moments are
important. You may choose to use CBEAM elements with warping if the members have
open cross sections and torsional stresses are estimated to be significant. You may even
choose to represent the members with assemblies of plate or solid elements. The choice of
which type and number of elements to use depends primarily on your assessment of the
effects that are important to represent in your model and on the cost and accuracy you are
willing to accept.
Given this, you should have a fairly good idea of how the structure will behave prior to
generating your finite element model. In other words, understanding the load path is
crucial in the selection of the appropriate element. Additionally, a few hand calculations
can usually provide a rough estimate of stress intensities. If you don’t have a fairly good
idea of how the structure will behave, you may be misled by incorrect results due to errors
or incorrect assumptions in your input data preparation.
CHAPTER 14 393
Modeling Guidelines
General Guidelines for Element Selection

Always experiment with a small test model when using elements that you aren’t familiar
with. This practice is cheaper than experimenting with a large production model, and it
gives you a better understanding of an element’s capabilities and limitations prior to
applying it to a large production model.
Avoid using obsolete elements. Elements are classified as obsolete only if their utility is
completely superseded by newer elements and after the newer elements are in service
long enough to be reliable. In most cases, they are retained in NX Nastran only to satisfy
the requirements for upward compatibility. Elements that are considered obsolete are no
longer included in any updated documentation.
Zero-Dimensional Elements
When you use CELASi elements to represent concentrated springs between two
components of translation, the directions of the two components must be coaxial. Even
small deviations in direction can induce a significant moment to your model that does not
exist in your physical structure. When you use a CELASi element, the locations of the two
end points should be coincident to avoid this type of problem. If the two end points aren’t
coincident, consider using a CROD or CBUSH element instead.
One-Dimensional Elements
If only an axial and/or torsion load is to be transmitted in an element, then the CROD is
the easiest element to use.
A CBAR is easier to use than a CBEAM element. The I1 and/or I2 values can be set to
zero. Use the CBEAM element instead of the CBAR element if any of the following
features is important:
• The cross-sectional properties are tapered.
• The neutral axis and shear center do not coincide.
• The effect of cross-sectional warping on torsional stiffness is critical.
• The difference in the mass center of gravity and the shear center is significant.
The formulation for the CBEAM element is based on a flexibility approach; the element
stiffness matrix is generated by inverting the flexibility matrix. For this reason, I1 and I2
must not be zero for the CBEAM element.
Two-Dimensional Elements
In general, quadrilateral elements (CQUAD4 and CQUAD8) are preferred over the
triangular elements (CTRIA3 and CTRIA6). The CTRIA3 element is a constant strain
element. It is excessively stiff, and when used alone, it is generally less accurate than the
CQUAD4 element, particularly for membrane strain. Whenever feasible, you should use
the CQUAD4 element instead of the CTRIA3 element. CTRIA3 should only be used when
necessary for geometric or topological reasons, for example, mesh transition between
regions of quadrilateral elements with different meshes or near the polar axis of a
spherical shell.
Additoinally, you should avoid using CTRIA3s in locations where the membrane stresses
are changing rapidly, for example, in the web of an I-beam. Since CTRIA3 has constant
membrane stresses, a large number of them may be needed to obtain acceptable accuracy.
It is better to use quadrilateral elements or CTRIA6 elements, if possible.
Don’t use plate or shell elements (CQUADi, CTRIAi) in stiffened shell structures with very
thin panels that can buckle. You should use shear panels (CSHEAR) in this case or in any
situation where direct stresses cannot be supported, such as in a very thin curved panel.
Avoid highly skewed elements (see Figure 14-1). The angle α should be as close to
90 degrees as possible.
Figure 14-1 Highly Skewed Element
For the CTRIA3 element, the skew test is based upon the three vertex angles.
Aspect ratio is defined as l ⁄ ω (length/width). Very high aspect ratio (see Figure 14-2)
should also be avoided, although it is no longer true that accuracy degrades rapidly with
aspect ratios as it once did with some of the obsolete elements.
Figure 14-2 Element with High Aspect Ratio
Warping is a measure of the amount the element deviates from being planar (see
Figure 14-3). Element warping should be minimized.
CHAPTER 14 395
Modeling Guidelines
Element Mid-Plane
Figure 14-3 Highly Warped Element
For the CQUAD8 elements, if midside nodes are present, they should be located within
the middle third of the edge. If a midside node is located at one-fourth the distance of the
edge as measured from either corner node on the edge, the internal strain field becomes
singular at the corners of the element. For best results, the midside node should be
located as close to the center of the edge as possible. If you want to use midside nodes,
you should include all of them. A CQUAD8 element with midside nodes deleted is
excessively stiff and inferior to a CQUAD4 element.
Figure 14-4 CQUAD8 with Missing Midside Node
For single curved structures (e.g., a cylinder), the CQUAD8, in general, yields better
results than the CQUAD4 element. For doubly curved structures (e.g., a spherical dome),
the CQUAD4 element, in general, performs better than the CQUAD8.
You should turn on the shell normal (PARAM,SNORM,X) option when using the
CQUAD4, CTRIA3, CQUADR, or CTRIAR element. See the next section on shell normals
for further details.
The membrane properties for the CQUADR and CTRIAR elements are less sensitive to
the element shape than the CQUAD4 and CTRIA3. The same thickness should be used
for all four corners (T1 = T2 = T3 = T4, or use the T Field on the PSHELL entry). The
CQUADR and CTRIAR elements can be used in conjunction with each other, but not with
other elements.
Shell Normals
By default, the direction of the normal rotation vector for flat plate elements is assumed to
be perpendicular to the plane of each element. If the model is curved, the shell bending
and twist moments must change direction at the element intersection. If transverse shear
flexibility is present, the deformations may be too large. (Because elements using
low-order formulations ignore the edge effect, this rarely causes any problems-a large
value of the parameter K6ROT partially cures the problem.) With the unique normal
(SNORM) option, the rotational degrees of freedom at each corner of an element are
measured relative to the specified normal vector direction. Thus, all elements connected
to a grid point will use a consistent direction for defining shell bending and twisting
moments.
In CQUAD4 and CTRIA3 elements, the stiffness matrices of the elements are modified to
eliminate the undesirable small stiffness in the rotational motions about the shell normal
vector. In effect, the transformation replaces the normal moments with in-plane forces. No
changes were made to the basic element stiffness matrix, and therefore flat plate models
will not be affected. The objective of the new transformation was to remove a potential
weakness in curved shell models and allow the automatic constraint process to remove the
true singularity in the assembled stiffness matrix.
The CQUADR and CTRIAR elements are also improved for some types of shell problems.
Because of their extra degrees of freedom, these elements are more sensitive to the
coupling between in-plane and out-of-plane motion due to curvature. The new
formulation provides more consistency between adjacent elements in a curved shell.
Shell normals are available for CQUAD4, CQUADR, CTRIA3, and CTRIAR elements.
Normals are activated if the actual angle between the local element normal and the unique
grid point normal is less than 20° , the default value for β (see Figure 14-5.) The default
for β can be changed by setting PARAM,SNORM, β to the desired real value up to 89
degrees. The unique grid point normal is the average of all local shell element normals at
a specific grid point. Generated grid point normals may be overwritten by user-defined
normals.
CHAPTER 14 397
Modeling Guidelines
Grid Point
Normal
Shell 1 Normal Shell 2 Normal
β β
Shell 1 Shell 2
Figure 14-5 Unique Grid Point Normal
A shell normal defines a unique direction for the rotational degrees of freedom of all
adjacent elements. A shell normal vector is created by averaging the normal vectors of
the attached elements. If the actual angle is greater than the value defined on
PARAM,SNORM, (default = 20.) the edge is assumed to be a corner, and the old method
is used. Shell normals improve the accuracy of the results in curved shells where in-plane
shear and twisting moments act together.
If the actual shell normal vector is known, such as in a cylinder or sphere, the
automatically calculated values may be superseded by the actual vectors using the Bulk
Data entry
SNORM,GID,CID,V1,V2,V3
where:
GID is a unique grid point,
CID is the coordinate system for defining the shell normal vector,
V1, V2, and V3 are unscaled components of the vector in the coordinate system.
A second parameter, SNORMPRT, controls the print or punch of values of the internal
shell normals. The output format is the same as the SNORM Bulk Data so that the
individual values may be used and modified on a subsequent job.
Remarks and Recommendations for Shell Normals

1. The type of structure that exhibits the most change in results is a thick curved
shell with large in-plane shear forces and twisting moments.
2. Most other problems, such as flat plates and curved shells with pressure loads,
show changes in results of less than 1%. More degrees of freedom may be
constrained using this improved formulation. This formulation results in zero
in-plane rotational stiffness values.
3. It has been observed from testing that the automatic normal vector calculations
produced answers nearly equal to and as accurate as those using the explicit
SNORM vector input. In other words, the results were insensitive to small
differences in the direction of the vectors. The important fact is that the shell
normal vector requires the connected elements to use a consistent normal
direction.
4. The CQUAD8 and CTRIA6 elements are not included in the shell normal
processing. If they are modeled correctly, they do not require shell normal
processing. Connecting these elements to the lower-order flat elements is not
recommended.
5. Curved shell elements have no stiffness in the rotational degree of freedom when
the new formulation is used-consequently, mechanisms may be introduced.
Mechanisms occur when elements, RBEs, or MPCs are connected to the
out-of-plane rotation of the shell grid points. Note that PARAM,AUTOSPC,YES
does not constrain these mechanisms.
6. In linear solution sequences, the values of “param,k6rot,0.” and
“param,snorm,20.” are the default.
7. Transverse shear flexibility (MID3 on the PSHELL property entry) should be left
on when normals are used.
Three-Dimensional Elements
Although the CHEXA and CPENTA elements are designed to behave reasonably well as
thin shell elements, in general, you shouldn’t use them in this capacity. The high ratio of
extensional stiffness in the direction normal to the effective transverse shear stiffness can
produce significant round-off errors.
As in the case of plate elements, if you include midside nodes for the solid elements, you
should locate them as close to the center of the edge as possible. Additionally, if midside
nodes are desired, you should include all of them.
R-Type Elements
A high degree of precision must be maintained when specifying coefficients for MPCs to
avoid introduction of unintentional constraints to rigid body motions. You should use
rigid elements (e.g., the RBE2, RBAR, etc.) whenever possible because their constraint
coefficients are internally calculated to a high precision. Furthermore, these R-type
elements require much less user interaction.
CHAPTER 14 399
Modeling Guidelines
14.3 Determining Mesh Density

The mesh density in a finite element model has important implications for both accuracy
and cost. Frequently, the minimum number of elements is set by topological
considerations, for example, one element per member in a space frame or one element per
panel in a stiffened shell structure. Historically, when problem size was more severely
limited, it was not uncommon to lump two or more frames or other similar elements in
order to reduce the size of the model. With computers becoming faster and cheaper, the
current trend is to represent all major components individually in the finite element
model.
If the minimum topological requirements are easily satisfied, the question remains as to
how fine to subdivide the major components. The question is particularly relevant for
elastic continua, such as slabs and unreinforced shells. In general, the accuracy of your
results increases with the mesh density. The mesh density required can be a function of
many factors. Among them are the stress gradients, the type of loadings, the boundary
conditions, the element types used, the element shapes, and the degree of accuracy
desired.
The grid point spacing should typically be the smallest in regions where you expect stress
gradients to be the steepest. Figure 14-6 shows a typical example of a stress concentration
near a circular hole. The model is a circular disk with an inner radius = a and an outer
radius = b. A pressure load p i is applied to the inner surface. Due to symmetry, only half
of the disk is modeled. In the example, both the radial stress and the circumferential stress
decrease as a function of l ⁄ r 2 from the center of the hole. The error in the finite element
analysis arises from differences between the real stress distribution and the stress
distribution within the finite elements.
Three different mesh densities are used in this example as shown in Figure 14-6.
• The first one is a coarse mesh model with the elements evenly distributed.
• The second model consists of the same number of elements; however, the mesh
is biased toward the center of the hole.
• The third model consists of a denser mesh with the elements evenly distributed.
These three models are then analyzed with three different element types-CQUAD4,
CQUAD8, and CQUAD4 with the corner stress option. The circumferential stress at the
inner radius is always greater than pi , which is the applied pressure load at the inner
radius, and approaches this value as the outer radius becomes larger. The theoretical
circumferential stress σ θ (see Timoshenko and Goodier, Theory of Elasticity, McGraw-Hill
Book Company, Third Ed., 1970) is given by the following equation:
2
2 b 
p i a  1 + -----
 2
r 
σ θ = ----------------------------------
2 2
(b – a )
where:
a = inner radius
b = outer radius
r = radial distance as measured from the center of the disk
p i = pressure applied at the inner radius
The stresses are then plotted as a function of the radius in a nondimensional fashion-stress
p i / versus r / a . The results are summarized in Table 14-1 and Figure 14-7.
Coarse Even Mesh
pi Coarse Biased Mesh
z x
pi
Fine Even Mesh 2a
2b
pi
Figure 14-6 Circular Disk with Different Meshes
CHAPTER 14 401
Modeling Guidelines
Legend
(Stress/ p )
i
r/a
(a)
Figure 14-7 Stresses with Different Elements and Meshes
(Stress/ p )
i
r/a
(b)
Figure 9-7 Stresses with Different Elements and Meshes (continued)
CHAPTER 14 403
Modeling Guidelines
(Stress/ p )
i
r/a
(c)
Figure 9-7 Stresses with Different Elements and Meshes (continued)
Table 14-1 Stresses Close to r = a for a Circular Disk
Case Element Type DOF Description Stress

No. (L-Set) Theoretical
1 CQUAD4 194 Coarse, Even Mesh 0.594
2 CQUAD4 194 Coarse, Biased Mesh 0.881
3 CQUAD4 868 Fine, Even Mesh 0.801
4 CQUAD8 550 Coarse, Even Mesh 0.929
5 CQUAD8 550 Coarse, Biased Mesh 1.041
6 CQUAD8 2538 Fine, Even Mesh 1.015
7 CQUAD4 with 194 Coarse, Even Mesh 1.142
Corner Option
8 CQUAD4 with 194 Coarse, Biased Mesh 1.047
Corner Option
9 CQUAD4 with 868 Fine, Even Mesh 1.127
Corner Option
10 THEORETICAL – – 1.000
For this particular case, since the stresses are proportional to 1/r2, you expect the highest
stress to occur at the inner radius. In order to take advantage of this piece of information,
the obvious thing to do is to create a finer mesh around the inner radius. Looking at the
results for the first two cases in Table 14-1, it is quite obvious that just by biasing the mesh,
the results are 30% closer to the theoretical solution with the same number of degrees of
freedom.
A third case is analyzed with a finer but unbiased mesh. Note that for case number 3, even
though it has more degrees of freedom, the result is still not as good as that of case number
2. This poor result is due to the fact that for the CQUAD4 element, the stresses, by default,
are calculated at the center of the element and are assumed to be constant throughout the
element. This assumption is best illustrated by the results shown in Figure 14-7(a).
Looking at Figure 14-6, it is also obvious that the centers of the inner row of elements are
actually further away from the center of the circle for case number 3 as compared to case
number 2. The results for case number 3 can, of course, be improved drastically by biasing
the mesh.
You can request corner outputs (stress, strain, and force) for CQUAD4 in addition to the
center values. Corner results are extrapolated from the corner displacements and
rotations by using a strain rosette analogy with a cubic correction for bending. The same
CHAPTER 14 405
Modeling Guidelines
three models are then rerun with this corner option—their results are summarized in
cases 7 through 9. Note that the results can improve substantially for the same number
degrees of freedom.
See Also
• “CQUAD4 and CTRIA3 Elements” on page 101 of the NX Nastran Element
Library
You select corner output by using a corner output option with the STRESS, STRAIN, and
FORCE Case Control commands. When you select one of these options, output is
computed at the center and four corners for each CQUAD4 element, in a format similar
to that of CQUAD8 and CQUADR elements.
There are four corner output options available:
• CORNER
• CUBIC
• SGAGE
• BILIN
These different options provide for different approaches to the stress calculations. The
default option is CORNER, which is equivalent to BILIN. BILIN has been shown to
produce better results for a wider range of problems.
To carry it a step further, the same three models are then rerun with CQUAD8 (cases 4
through 6). In this case, the results using CQUAD8 are better than those using the
CQUAD4. This result is expected since CQUAD8 contains more DOFs per element than
CQUAD4. Looking at column three of Table 14-1, you can see that due to the existence
of midside nodes, the models using CQUAD8 contain several times the number of DOFs
as compared to CQUAD4 for the same number of elements. The results using CQUAD4
can, of course, be improved by increasing the mesh density to approach that of the
CQUAD8 in terms of number of DOFs.
It is important to realize that the stresses are compared at different locations for Cases 1
through 3 versus Cases 4 through 9. This difference occurs because the stresses are
available only at the element centers for Cases 1 through 3, but the stresses are available
at the corners as well as the element centers for Cases 4 through 9. When looking at your
results using a stress contour plot, you should be aware of where the stresses are being
evaluated.
How fine a mesh you want depends on many factors. Among them is the cost you are
willing to pay versus the accuracy you are receiving. The cost increases with the number
of DOFs. The definition of cost has changed with time. In the past, cost is generally
associated with computer time. With both hardware and software becoming faster each
day, cost is probably associated more with the time required for you to debug and
interpret your results. In general, the larger the model is, the more time it takes you to
debug and interpret your results. As for acceptable accuracy, proceeding from case 8 to
case 6, the error is reduced from 4.7% to 1.5%; however, the size of the problem is also
increased from 194 to 2538 DOFs. In some cases, a 4.7% error may be acceptable. For
example, in cases in which you are certain of the loads to within only a 10% accuracy, a
4.7% error may be acceptable. In other cases, a 1.5% error may not be acceptable.
In general, if you can visualize the form of the solution beforehand, you can then bias the
grid point distribution. However, this type of information is not necessarily available in
all cases. If a better assessment of accuracy is required and resources are available (time
and money), you can always establish error bounds for a particular problem by
constructing and analyzing multiple mesh spacings of the same model and observe the
convergences. This approach, however, may not be realistic due to the time constraint.
You can use the stress discontinuity feature, as described in “Model Verification” on
page 445 of the NX Nastran User’s Guide, to assess the quality of the mesh density for the
conventional h-version elements.
With the software’s p-version element and p-version adaptivity capabilities, you can
perform the mesh refinement in NX Nastran.
h-elements
In traditional finite element analysis, as the number of elements increases, the accuracy of
the solution improves. The accuracy of the problem can be measured quantitatively with
various entities, such as strain energies, displacements, and stresses, as well as in various
error estimation methods, such as simple mathematical norm or root-mean-square
methods. The goal is to perform an accurate prediction on the behavior of your actual
model by using these error analysis methods. You can modify a series of finite element
analyses either manually or automatically by reducing the size and increasing the number
of elements, which is the usual h-adaptivity method. Each element is formulated
mathematically with a certain predetermined order of shape functions. This polynomial
order does not change in the h-adaptivity method. The elements associated with this type
of capability are called the h-elements.
p-elements
A different method used to modify the subsequent finite element analyses on the same
problem is to increase the polynomial order in each element while maintaining the
original finite element size and mesh. The increase of the interpolation order is internal,
and the solution stops automatically once a specified error tolerance is satisfied. This
method is known as the p-adaptivity method. The elements associated with this
capability are called the p-elements.
See Also
www.cadfamily.com •EMail:[email protected]
“p-Elements” on page 753 of the NX Nastran User’s Guide
CHAPTER 14 407
Modeling Guidelines
14.4 Creating Mesh Transitions

You can use mesh transitions, for example, to refine the mesh in a particular area, connect
different element types (for example, a CBAR element to a solid element), or provide
transitions required to model the geometry of the structure.
• Never place a mesh transition in an area of interest or in an area where there is
a large variation in stress.
• Mesh transitions should be located away from the areas of interest in a region.
Due to incompatibilities between finite element types, any transition between different
element types (even a transition from CQUAD4 to CTRIA3 elements) can result in local
stress anomalies. Normally, these stress anomalies are localized and dissipate quickly as
you move away from the transition. However, a problem arises when the transition
occurs in an area of interest. In this case, the local stress rises (or decreases) due to the
effect of the transition; in other words, the results may be conservative (or
unconservative) in an area near a transition. However, if this localized stress variation
occurs away from areas of interest, the increase (or decrease) in stress caused by the
transition should cause no concern.
Transition from a Coarse Mesh to a Fine Mesh

Creating a transitions between corase and fine meshes can be challenging. One common
method of performing a transition is to use an intermediate belt of triangular elements as
shown in Figure 14-8(a). It is also very tempting to selectively delete midside nodes of
higher-order elements (e.g., CQUAD8) and use them as transition elements as shown in
Figure 14-8(b). However, this method isn’t recommended because it severely distorts the
stress distribution in the elements adjacent to the change in mesh size.
Q4 Q4 T3 Q4 Q4
T3 Q4 Q8 Q4
Q4 Q4 T3 Q4 Q4
Q4 Q4 T3 Q4 Q4
T3 Q4 Q8 Q4
Q4 Q4 T3 Q4 Q4
(a) (b)
Figure 14-8 Transition Methods
When using a higher-order element, such as a CQUAD8 or a CHEXA (nine to 20 nodes)

for any purpose, you should include all midside nodes. A CQUAD8 element with all
midside nodes deleted is excessively stiff and therefore is inferior to a CQUAD4. For the
CHEXA element, you should use it with eight nodes (no midside nodes), or use it with
20 nodes (all midside nodes). The midside nodes, if used, should be located as close to the
center of the edge as possible.
As an alternative mesh transition method, a spline element can be used. In NX Nastran,

the RSPLINE element is an elastic interpolation element that connects N end points.
Displacements for intermediate grid points are interpolated from the equations of an
elastic beam passing through these grid points. The spline element assumes a linear
interpolation for displacement and torsion along the axis of the spline, a quadratic
interpolation for rotations normal to the axis of the spline, and a cubic interpolation for
displacements normal to the axis of the spline.
Consider an example problem as shown in Figure 14-9. This problem involves a plate
fixed at the left end with a dimension of 14 in by 9 in, and modeled with 1 in by 1 in
CQUAD4 elements. For clarity, only grid points of interest are labeled. Five different
loadings-as shown in Table 14-2 are applied to this structure along the free edge (right
end). For convenience, a RBE3 element is created along the free edge by defining a
dependent point (grid point 500) at the center. This grid point is connected to all the edge
grid points. The load can then be applied to grid point 500 and distributed to the rest of
the edge grid points.
See Also
• “The RBE3 Element” on page 211 of the NX Nastran Element Library

Hypothetically, suppose that you’re interested in the stresses along the line one inch from
the fixed edge. Since you’re not interested in the stresses outside of this region, you can
represent the rest of the region with a coarser mesh. Two different transition methods will
be used for this purpose-one with CTRIA3s (Figure 14-10) and one with RSPLINEs
(Figure 14-11). The transitions are performed along the line five inches away from the
fixed edge. An additional transition is performed at x = 11 inches in order to maintain the
same grid point density on the right edge for all three models. This condition assures
identical load applications for all three models. Normally, this additional transition is
seldom done in a real model.
CHAPTER 14 409
Modeling Guidelines
Table 14-2 Loadings Used for the Mesh Transition Test
Loading Description
1. Membrane Loads Z
500 Fx
Y
RBE3 Element
X Along Edge
2. In-Plane Shear Loads Fy

(parabolic distribution) Z
500
Y
RBE3 Element
X Along Edge
3. In-Plane Bending Loads Mz

Z
Y 500
RBE3 Element
X Along Edge
4. Vertical Loads
Fz
Z
500
Y
RBE3 Element
X Along Edge
5. Twisting Loads
Z
500 Mx
Y
RBE3 Element
X Along Edge
Interested in Stresses Along This Line, x = 1.0 in

20 60
19 59
18 58
17 57
16 56
15 55 9 in
14 54
13 53
12 52
11 51
Y
1 in 4 in 9 in
X
Figure 14-9 Baseline Model
20 60
19
18
17 57
16
15
14 54
13
12
11 51
Figure 14-10 Transition with CTRIA3
CHAPTER 14 411
Modeling Guidelines
Transition Transition
Region Region
20 60
19 59
18
58
17 57
16
56
15 55
14 54
13
53
12
52
11 51
Figure 14-11 Transition with RSPLINE
The model with the mesh transition using CTRIA3s, as shown in Figure 14-10, is quite
straightforward and needs no further explanation. However, further discussion is
warranted in the case of mesh transition using the RSPLINE element.
The input format for the RSPLINE element is as follows:
1 2 3 4 5 6 7 8 9 10
RSPLINE EID D/L G1 G2 C2 G3 C3 G4
C4 G5 C5 G6 -etc.-
Field Contents
EID Element identification number.
Gi Grid point identification number.
Ci Components to be constrained.
Remarks:
1. The first and last grid points must be independent.
2. Displacements are interpolated from the equations of an elastic beam passing
through the grid points.
A single RSPLINE is used to connect grid points 51 through 60 as shown in Figure 14-11.
Grid points 51, 54, 57, and 60 are considered to be independent, while the other six points
are dependent (grid points 52, 53, 55, 56, 58, and 59). This RSPLINE interpolates the
motions at the intermediate points, indicated by circles in Figure 14-11 (grid points 52, 53,
55, 56, 58, and 59), from the cubic equations of a beam spline passing through the points
indicated by squares (grid points 51, 54, 57, and 60). The input for the RSPLINE element
connecting grid points 51 through 60 is as follows:
1 2 3 4 5 6 7 8 9 10
RSPLINE 1000 51 52 12345 53 12345 54
55 12345 56 12345 57 58
12345 59 12345 60
The results using these two transition methods are compared with the baseline model with
no transition. These results are evaluated at two locations: at x = 1.0 in (location of
interest) and at x = 5.0 in (transition region). The five different loading conditions
mentioned previously were used. The Hencky-von Mises stresses using the grid point
stress feature (“Reviewing Grid Point Stresses” on page 432) and the CQUAD4 corner
stress option are used for comparison. The results are shown in Figure 14-12 through
Figure 14-16. As you can see, the results compare well, using either method of transition,
with the baseline model at the area of interest (away from the transition region) for all five
loading conditions.
At the transition region, depending on the loading and location, the transition using
CTRIA3 elements compares better with the baseline model in some cases, whereas the
transition using RSPLINE elements compares better with the baseline model in other
cases. Note that there are only four stress values available in the case of using CTRIA3s
since there are only four grid points along the line joining grid points 51 through 60 in
Figure 14-10.
The CTRIA3 elements used in this model are all regularly shaped elements; in most real
life problems, it is very likely that the transition areas may consist of CTRIA3 elements that
are more distorted. In this case, the transition using the distorted CTRIA3 elements does
not perform as well as the model in Figure 14-9 consisting of regularly shaped CTRIA3
elements.
In either case, there are noticeable differences in the stresses at the transition region, as
expected. However, at the area of interest and away from the transition region, the
stresses are virtually the same for both cases as compared to the baseline model. Once
again, mesh transitions should always be modeled away from the area of interest.
CHAPTER 14 413
Modeling Guidelines
Figure 14-12 Resultant Stresses for Membrane Loads
Figure 14-13 Resultant Stresses for In-Plane Shear Loads
CHAPTER 14 415
Modeling Guidelines
Figure 14-14 Resultant Stresses for In-Plane Bending Loads
Figure 14-15 Resultant Stresses for Vertical Loads
CHAPTER 14 417
Modeling Guidelines
Figure 14-16 Resultant Stresses for Twisting Loads
Mesh Transition Between Dissimilar Element Types

Attaching a plate or bar element to a solid element is a case of transition between
dissimilar element types. This process is more involved than it appears at first glance.
Solid elements have stiffness only in the translational DOFs at the attachment grid points;
they have no stiffness for rotational DOFs. A simple visualization is to think of the
attachment of a solid element to a grid point as a “ball-and-socket” joint, that is,

translational forces may be transmitted, but no moment may be transmitted through the
connection.
This incompatibility of the element stiffness matrices represents a modeling problem

whenever plate or bar elements are attached to solid elements. Both plate and bar
elements have stiffness for rotational DOFs (although the plate element may not have a
stiffness for the normal rotation). Therefore, special modeling must be performed
whenever a plate or bar is connected to a solid element. Otherwise, the connection
becomes a hinge (for plate elements) or a pinned connection (for bar elements).
Several methods are available to handle the transition between these elements. These
methods range from adding extra elements (for example, adding an additional plate or bar
that continues into the solid element) to using special (R-type) elements for the transition.
One method of handling this transition is to use RBE3 elements. The RBE3 is an
interpolation element, which is ideally suited for this application. By using RBE3s, the
rotations of the attached grid points is simply slaved to the translations of the adjacent grid
points.
See Also
• “The RBE3 Element” on page 211 of the NX Nastran Element Library

Examples of using RBE3 elements to connect a solid element to a plate element and bar
element are shown in Figure 14-17. The RBE3 elements attach the rotational DOFs to the
translational DOFs on the solid element.
3 2 10
4 1 9
13 12 15
Plate to Solid Connection 14 11
Y 22
X Bar to Solid Connection

Z
Figure 14-17 Typical Transition Between Dissimilar Elements
CHAPTER 14 419
Modeling Guidelines
For the plate to solid connection, two RBE3 elements suffice:
1 2 3 4 5 6 7 8 9 10
$RBE3 EID REFGRID REFC WT1 C1 G1,1 G1,2
RBE3 901 1 456 1.0 123 2 3
RBE3 902 2 456 1.0 123 1 3
For the bar to solid connection, one RBE3 element can make the connection:
RBE3 903 12 456 1.0 123 11 13
22
These RBE3 elements transmit the loads to the independent DOFs. If RBE2 elements are
used, then the connection is “rigid.”
Remember, when handling these connections, the solid elements have no stiffness for
rotational DOFs, whereas the real structure does. This means that a special modeling
effort is needed when any element with bending stiffness is connected to a solid element.
When using the RBE3 element, care must be taken to ensure that the independent DOFs
are sufficient to transfer any applied loads. For the bar-to-solid connection in
Figure 14-16, if only two independent grid points are used, the element is “unstable,” that
is, since only the translational DOFs are used as independent DOFs in the sample, the
element is unstable for rotation about the axis connecting the two points. Therefore, three
non-colinear grid points are used. A simple way to remember this is to ask, “If I constrain
the DOFs that I list as independent on the RBE3, can I prevent any possible rigid body
motion?” If the answer to this question is “yes,” then the RBE3 element is capable of
transferring any applied loads. In this way, you can avoid possible problems in
processing the RBE3 elements.
Shell-to-Solid Transition Element (RSSCON)

The RSSCON method is another way to model shell-to-solid transitions (see
Figure 14-18). This capability conveniently eliminates the need to define RBE3s or MPCs
(multipoint constraints) to constrain a shell element’s translational and rotational degrees
of freedom to a solid element’s translational degrees of freedom. When using the
RSSCON capability, the shell element mesh must line up with the solid element mesh so
that there is an exact element-to-element correspondence.
Upper Edge
Solid Lower Edge
Shell
Figure 14-18 Clamped Connection of a Shell to a Solid
The RSSCON Bulk Data entry supports both p-adaptive elements and h-elements. The
p-adaptive elements are outside the scope of this user’s guide. Through the RSSCON Bulk
Data entry, the shell elements CQUAD4, CQUAD8, CTRIA3, CTRIA6, CQUADR, and
CTRIAR can be connected to the solid elements CHEXA, CPENTA, and CTETRA.
Elements with midside nodes are also supported.
RSSCON generates a multipoint constraint, which puts the shell degrees of freedom in the
dependent set (m-set). The three translational degrees of freedom and the two rotational
degrees of freedom of the shell edge are connected to the three translational degrees of
freedom of the upper and lower solid edge. Poisson’s ratio effects and temperature loads
are modeled correctly. The generated multipoint constraints produce six zero-energy
modes for rigid-body motion.
The RSSCON Bulk Data entry defines the connection of a shell element to a solid element.
Within NX Nastran, however, there are two options for making this connection using the
RSSCON Bulk Data entry, as shown in Figure 14-19:
1. The ELEM option, which allows you to specify the element ID of the shell and
the element ID of the solid to which the shell is to be connected.
2. The GRID option, which allows you to specify the grid point ID of the shell and
the upper and lower grid point IDs of the solid. Two triplets of grid point IDs
may be specified on one RSSCON Bulk Data entry.
CHAPTER 14 421
Modeling Guidelines
31 Solid Element 302
301 102
101
202
201 Shell Element
21
RSSCON, 100, ELEM, 21, 31

or
RSSCON, 200, GRID, 101, 201, 301, 102, 202, 302
Figure 14-19 ELEM and GRID Option on the RSSCON Bulk Data Entry
The best modeling practice is illustrated in Figure 14-20. The height of the connected
solid element should be chosen equal to the thickness of the shell. If the height of the
connected solid element is much larger than the thickness of the shell element, then the
connection modeled with RSSCON will be stiffer than the continuum model. For
example, in a mesh where shell grid points are identical or coincide with solid grid points,
the RSSCON Bulk Data entry may model a connection that is too stiff (see
Figure 14-21[a]). Note also that in the model of Figure 14-21[b], the GRID option must be
used to connect a shell element to more than one solid element.
The RSSCON connector element can accommodate a variety of mesh topologies, as

31 Solid
21 Shell
RSSCON, 100, ELEM, 21, 31
Figure 14-20 Best Modeling Practice for RSSCON
102
100
101 200 201
202 [b]
201 101
[a]
RSSCON, 200, GRID, 101, 201, 101, 102, 202, 102

301
or
RSSCON, 200, ELEM 200, 100 RSSCON, 100, GRID, 101, 201, 301
Figure 14-21 RSSCON in Meshes Where Solid and

Shell Elements Share Grid Points
CHAPTER 14 423
Modeling Guidelines
23 Slanted
202
22
12 23
HEXA 208
22
11
24 12
102
11 TRIA 103
21
104 24
RSSCON, 100, ELEM, 102, 202
or 21
RSSCON, 100, GRID, 11, 21, 22, 12, 24, 23
RSSCON, 200, ELEM, 103, 208
RSSCON, 201, ELEM, 104, 208 (not necessary)
or
RSSCON, 200, GRID, 11, 21, 22, 12, 24, 23
220 23
HEXA20
26 12 PENTA 206 23
22
13 22
11 25 24 12
108
QUAD 11 QUAD
21 101
RSSCON, 200, ELEM, 108, 220

or 21
RSSCON, 200, GRID, 11, 21, 22, 12, 24, 23 RSSCON, 200, ELEM, 101, 206
RSSCON, 201, GRID, 13, 25, 26 or
RSSCON, 200, GRID, 11, 21, 22, 12, 21, 23
Figure 14-22 Modeling Options with RSSCON
The only modeling requirement is that the shell grid point must lie on the line connecting
the lower and upper solid grid points. By default, the RSSCON allows the shell grid point
to have an offset from this line of up to 5% of the distance between the two solid grid
points. If a shell grid point does not lie on the line joining the two solid grid points, but
within this 5% tolerance, the shell grid point will be moved to lie on the line. On the other
hand, if a shell grid point lies outside this 5% tolerance, the run terminates with a fatal
message. You can modify the default 5% tolerance by adding “PARAM,TOLRSC, ε ” to
the Bulk Data Section, where ε is the desired tolerance percentage. The default value for
ε is 0.05. You can print the old and new location of the moved shell grid points with the
parameter “PARAM,SEP1XOVR,16".
When you use the GRID option, NX Nastran does not verify that the grid points are valid
shell or solid grid points. When the ELEM option is used, if a solid-shell connection has
horizontal edges that are curved in the shell plane, NX Nastran assumes that the geometry
of the shell element is compatible with that of the solid element. In the case of vertical
edges, the shell element grid point must lie on the line between the upper and lower solid
element grid point. A default offset tolerance of 5% (user modifiable via "param,tolrsc, ε ")
of the distance between the solid grid points is allowed. For limitations of the RSSCON
related to the p-elements, see:
• “Introduction to R-Type Elements” on page 194 of the NX Nastran Element

Library
By default, the moved shell grid points aren’t printed.
δ≥ε⋅h
h
δ
Shell 21
Solid 31
RSSCON, 100, ELEM, 21, 311
Figure 14-23 RSSCON Modeling Practice Parameters
RSSCON Modeling Recommendations

The RSSCON connector produces excellent results if you follow good modeling practices.
The geometry of the RSSCON connector is checked, and fatal messages are issued for
invalid connections. Additional recommendations are given below.
1. Midside-noded elements should only be connected to other midside-noded

elements. For example, a CQUAD8 element connected to a CHEXA20 element is
acceptable; however, a CQUAD4 element connected to a PENTA15 element is
not allowed. Elements with missing midside grid points are not allowed.
2. For midside-noded elements, avoid using the RSSCON where shells connect to
triangular solid faces.
CHAPTER 14 425
Modeling Guidelines
3. Do not mix coordinate systems when using midside nodes. In other words,
when using midside nodes, do not define a local coordinate system containing
all the shell grid points and the basic coordinate system to define the solid grid
points. Otherwise, incorrect answers will be generated.
4. Only one edge of a single shell element may connect to a solid surface.
5. Do not connect more than one shell element to one solid element when using the
ELEM option.
6. Do not attempt solid-shell connections to severely warped (greater than 20%)
midside-noded solid elements.
7. The plane of the shell element should not be parallel to the plane of the
connecting face of the solid element.
8. The connecting edge of the shell element must lie in the face of the solid element.
In addition, the vertex grid points of the shell edge connecting to a quadrilateral
solid face must line up with its opposite edges; for triangular solid faces, the
connections may be made between any of the edges. The tolerance in the
geometry may be modified; see PARAM,TOLRSC, ε described above.
9. When using the ELEM option in the case of partitioned superelement Bulk Data,
you must ensure that the solid element, RSSCON connector, and shell element
which define a shell-solid connection are all contained within the same
superelement.
Example - Cantilever Clamped to a Wall

A cantilever (4.1 in long, 0.1 in thick, 0.5 in wide) is clamped to a wall and loaded at the
tip. This example demonstrates how accurate the local stress distribution is when a
RSSCON entry is used to model the clamped condition. Four different finite element
models are used, and their results are compared with classical beam theory. The
deflection and bending stress from beam theory are 0.1103 in and 9,840 psi, respectively,
at the clamped end of the cantilever.
1. The first model is a plate model consisting of all CQUAD4s. The stress contours
of this model are shown in .
2. The second model is a plate model consisting of all CHEXAs. The stress
contours of this model are shown in Figure 14-26.
3. The third model is a plate-solid model using MPCs to transition between the
plate and solid model. The input file and stress contours of this model are
shown in Listing 14-1 and Figure 14-27, respectively.
4. The fourth model is a plate-solid model using RSSCONs to transition between
the plate and solid model. The input file and stress contours of this model are
shown in Listing 14-2 and Figure 14-29, respectively.
An exploded view of the transition used in the third and fourth models are shown in
Figure 14-24. Both the deflections and stress contours of the four models are in good
agreement with the theoretical results, as shown in Table 14-3.
7
200
6
10
9 3
300 12
2
11
Figure 14-24 Stress Distribution for CQUAD4 Model
Table 14-3 Tip Displacements and Maximum Stresses at the Clamped End
Cantilever Clamped to Wall Displacement at Tip Stress at Clamped End (psi)

Beam Theory (Rigid Wall) 0.1103 9840.
All CQUAD4s 0.1103 9720.
All CHEXAs 0.1087 9720.
Plate-to-Solid Transition Using 0.1103 9720.
MPCs
Plate-to-Solid Transition Using 0.1103 9720.
RSSCONs
CHAPTER 14 427
Modeling Guidelines
Listing 14-1 Input File for Solid-Plate Model Using MPCs for Transition
$
$ mpc1.dat
$
TIME 100
SOL 101
CEND
TITLE= cant cantilever clamped with HEXA
$
ECHO=BOTH
$SEALL=ALL
DISPLACEMENT =ALL
stress(corner) = all
OLOAD =ALL
spc= 100
mpc= 200
$
SUBCASE 10
LOAD = 100
BEGIN BULK
$
param,post,-1
$
grid,1,, 0., 0., 0.
grid,2,,0.1, 0.,0.
grid,3,,0.1,0.5,0.
grid,4,,0.,0.5,0.
grid,5,,0.,0.,0.1
grid,6,,0.1,0.,0.1
grid,7,,0.1,0.5,0.1
grid,8,,0.,0.5,0.1
grid,9,,0.1,0.,0.05
grid,10,,0.1,0.5,0.05
Listing 14-1 Input File for Solid-Plate Model Using MPCs for Transition
grid,11,,1.1,0.,0.05
grid,12,,1.1,0.5,0.05
grid,13,,2.1,0.,0.05
grid,14,,2.1,0.5,0.05
grid,15,,3.1,0.,0.05
grid,16,,3.1,0.5,0.05
grid,17,,4.1,0.,0.05
grid,18,,4.1,0.5,0.05
$
spc1, 100, 123, 1, 4, 5, 8
$
MPC, 200, 9,5, 1., 6, 1, -10.,,+MP01
+MP01, , 2,1, 10.
MPC, 200,10,5, 1., 7, 1, -10.,,+MP02
+MP02, , 3,1, 10.
MPC, 200, 9,4, 1., 2, 2, -10.,,+MP03
+MP03, , 6,2, 10.
MPC, 200,10,4, 1., 3, 2, -10.,,+MP04
+MP04, , 7,2, 10.
MPC, 200, 9,1, 1., 6, 1, -0.5,,+MP05
+MP05, , 2,1, -0.5
MPC, 200, 9,2, 1., 6, 2, -0.5,,+MP06
+MP06, , 2,2, -0.5
MPC, 200, 9,3, 1., 6, 3, -0.5,,+MP07
+MP07, , 2,3, -0.5
MPC, 200,10,1, 1., 7, 1, -0.5,,+MP08
+MP08, , 3,1, -0.5
MPC, 200,10,2, 1., 7, 2, -0.5,,+MP09
+MP09, , 3,2, -0.5
MPC, 200,10,3, 1., 7, 3, -0.5,,+MP10
+MP10, , 3,3, -0.5
$
chexa, 200, 210, 1, 2, 3, 4, 5, 6, +HX1
+HX1, 7, 8
CQUAD4, 300, 310, 9, 11, 12, 10
CQUAD4, 301, 310,11, 13, 14, 12
CQUAD4, 302, 310,13, 15, 16, 14
CQUAD4, 303, 310,15, 17, 18, 16
$
PSOLID 210100
PSHELL 310 100 .1 100 1.100
MAT1 100 1.+7 0.
$
FORCE 100 17 1. 0. 0.0 -1.
FORCE 100 18 1. 0. 0.0 -1.
$
ENDDATA
CHAPTER 14 429
Modeling Guidelines
Listing 14-2 Input File for Solid-Plate Model Using

RSSCONS for Transition
$ rsscon1.dat
$
TIME 100
SOL 101
CEND
TITLE = CANTILEVER CLAMPED WITH HEXA
$
ECHO=BOTH
DISPLACEMENT = ALL
stress(corner) = all
OLOAD = ALL
SPC = 100
$
SUBCASE 10
LOAD = 100
$
BEGIN BULK
$
param,post,-1
$
GRID,1,,0.0,0.0,0.0
GRID,2,,0.1,0.0,0.0
GRID,3,,0.1,0.5,0.0
GRID,4,,0.0,0.5,0.0
GRID,5,,0.0,0.0,0.1
GRID,6,,0.1,0.0,0.1
GRID,7,,0.1,0.5,0.1
GRID,8,,0.0,0.5,0.1
GRID,9,,0.1,0.0,0.05
GRID,10,,0.1,0.5,0.05
GRID,11,,1.1,0.0,0.05
GRID,12,,1.1,0.5,0.05
GRID,13,,2.1,0.0,0.05
GRID,14,,2.1,0.5,0.05
GRID,15,,3.1,0.0,0.05
GRID,16,,3.1,0.5,0.05
GRID,17,,4.1,0.0,0.05
GRID,18,,4.1,0.5,0.05
$
SPC1, 100, 123, 1, 4, 5, 8
$
RSSCON, 200, elem, 300, 200
$
CHEXA, 200, 210, 2, 3, 7, 6, 1, 4, +HX1
+HX1, 8, 5
CQUAD4, 300, 310, 9, 11, 12, 10
CQUAD4, 301, 310,11, 13, 14, 12
CQUAD4, 302, 310,13, 15, 16, 14
CQUAD4, 303, 310,15, 17, 18, 16
$
PSOLID 210 100
$
PSHELL 310 100 .1 100 1. 100
MAT1 100 1.+7 0.
$
FORCE 100 17 1. 0. 0.0 -1.
FORCE 100 18 1. 0. 0.0 -1.
$
ENDDATA
Figure 14-25 Stress Distribution for CQUAD4 Model
CHAPTER 14 431
Modeling Guidelines
Figure 14-26 Stress Distribution for CHEXA Model
Figure 14-27 Stress Distribution for Solid-Plate Model

Using MPCs for Transition
Figure 14-28 Stress Distribution for Solid-Plate Model

Using RSSCONs for Transition
14.5 Reviewing Grid Point Stresses

When you’re working with a structure that’s modeled with either plate or solid elements,
you will generally want to review the element component stresses. If this is the case, extra
care must be taken on your part to ensure that the component stresses, which you are
evaluating, are output in some consistent manner.
Consider a simple problem as shown in Figure 14-29. This plate, which is 0.1 inch thick,
is subjected to a uniaxial tensile load of 1 lb/in, and the boundary condition is as shown
in Figure 14-29. This model is represented with a 3 x 3 mesh of CQUAD4 elements. If the
elements are connected in a manner as shown in Table 14-4, then all the element σ x values
are in the direction of the applied loads. In this case, σx = 10 psi for all nine elements. On
the other hand, if you were to change, for instance, the connectivity order for element 5
from grid points 6-7-11-10 to 7-11-10-6, the component stresses and would be swapped.
Note that this is the only difference betweenTable 14-4 and Table 14-5. However, if you
are not aware of it and you are making a contour plot of σx , the results will be quite
different for the two cases.
CHAPTER 14 433
Modeling Guidelines
5 lb
13 14 15 16
l = 10 in 7 8 9
10 lb
9 10 11 12
l = 10 in 4 5 6
10 lb
5 6 7 8
l = 10 in 2 3
1
5 lb
1 2 3 4
y t = 0.10 in
Figure 14-29 Uniaxial Load for a 3 x 3 Mesh
Table 14-4 Element Component Stresses

CQUAD4,7,1,9,10,14,13 σ x = 10 σ x = 10 σ x = 10
CQUAD4,8,1,10,11,15,14
σy = 0 σy = 0 σy = 0
CQUAD4,9,1,11,12,16,15
$ σ v = 10 σ v = 10 σ v = 10
CQUAD4,4,1,5,6,10,9 σ x = 10 σ x = 10 σ x = 10
CQUAD4,5,1,6,7,11,10
σy = 0 σy = 0 σy = 0
CQUAD4,6,1,7,8,12,11
$ σ v = 10 σ y = 10 σ v = 10
CQUAD4,1,1,1,2,6,5 σ x = 10 σ x = 10 σ x = 10
CQUAD4,2,1,2,3,7,6
σy = 0 σy = 0 σy = 0
CQUAD4,3,1,3,4,8,7
$ σ v = 10 σ v = 10 σ v = 10
Table 14-5 Element Component Stresses

CQUAD4,7,1,9,10,14,13 σ x = 10 σ x = 10 σ x = 10
CQUAD4,8,1,10,11,15,14
σy = 0 σy = 0 σy = 0
CQUAD4,9,1,11,12,16,15
$ σ v = 10 σ v = 10 σ v = 10
CQUAD4,4,1,5,6,10,9 σ x = 10 σ x = 10 σ x = 10
CQUAD4,5,1,7,11,10,6
σy = 0 σy = 0 σy = 0
CQUAD4,6,1,7,8,12,11
$ σ v = 10 σ y = 10 σ v = 10
CQUAD4,1,1,1,2,6,5 σ x = 10 σ x = 10 σ x = 10
CQUAD4,2,1,2,3,7,6
σy = 0 σy = 0 σy = 0
CQUAD4,3,1,3,4,8,7
$ σ v = 10 σ v = 10 σ v = 10
To avoid this type of surprise, it may be easier and more meaningful to look at the
invariant stress quantities, such as Hencky-von Mises stresses, rather than the component
stresses. The grid point stress (GPSTRESS) option in NX Nastran offers you another
alternative. The grid point stress option calculates the stresses at the grid points from the
adjoining plate and solid elements in a coordinate system defined by you.
This option enables you to request the output of stresses at grid points in surfaces of
two-dimensional plate elements—namely, the CQUAD4, CQUADR, CQUAD8, CTRIA3,
CTRIAR, and CTRIA6 elements—and in volumes containing CHEXA, CPENTA, and
CTETRA solid elements.
See Also
• “Element Data Recovery Resolved at Grid Points” on page 279 of the NX
CHAPTER 14 435
Modeling Guidelines
14.6 Defining Consistent Loading

In finite element analysis, you often apply loads to the elements rather than to the grid
points. Examples of such loadings include the edge load on a CQUAD4 element or a
pressure load on one of the faces of a solid element.
When you convert these element loads to grid point loads, a common tendency is to
spread them equally at the connected grid points. Let us look at an example of a typical
plate element with uniform in-plane loading at one of the edges as shown in Figure 14-30.
The total load (P) is to be uniformly distributed along the edge of the element with an
intensity of P/l, where l is the length of the edge. This same load is to be applied to six
different elements-CQUAD4, CTRIA3, CQUAD8, CTRIA6, CQUADR, and CTRIAR. If
you lump these loads equally to the grid points, the load at each node is P/2 for the
CQUAD4, CTRIA3, CQUADR and CTRIAR, and P/3 for the CQUAD8 and CTRIA6.
Since the geometry and boundary condition are identical for the six models, and the loads
are “seemingly equivalent,” you will expect the results (displacements, stresses, etc.) to be
identical. However, in general, this is not the case. The results for the cases using
CQUAD8, CTRIA6, CQUADR, and CTRIAR are different from the results using
CQUAD4 and CTRIA3. The reason for this is because these “seemingly equivalent” loads
are actually not quite equivalent. They are considered as lumped loads. In order to obtain
the correct load distribution, these loads need to be converted to consistent loads.
The equivalent loads at the grid points computed from the element loads are known as
consistent loads, and they are calculated by applying the principle of virtual work. The
same shape function that is used in deriving the element stiffness is used for arriving at
this load-and hence the word consistent load. They are a function of the element types
and the applied loads.
Depending on the element, these lumped loads in general are not equal to the consistent
loads. The consistent loads for each of these elements for a uniform in-plane edge load of
P/l are shown next to each element. In this case, the lumped loads and consistent loads
are the same for the CQUAD4 and CTRIA3. They are not the same for the CQUAD8,
CTRIA6, CQUADR and CTRIAR. As you can see, they are substantially different from
the lumped loads that were discussed in the previous paragraph. It is interesting to note
that the loads are distributed as 1/6, 4/6, and 1/6 along the edge for the CQUAD8 and
CTRIA6, which is quite different from the lumped load approach. For the CQUADR and
CTRIAR, an additional moment Pl/12 is needed to arrive at the consistent loads. When
the corresponding consistent loads shown in Figure 14-30 are applied to each of the
respective elements, the force distributions are then equivalent for all the elements.
P/2 P/2
CTRIA3
CQUAD4
CTRIA3
P/2 P/2
(a) (b)
P/6 P/6
CTRIA6
CQUAD8 4P/6 4P/6
CTRIA6
P/6 P/6
(c) (d)
Pl/12 Pl/12
P/2 P/2
CTRIAR
l CQUADR
CTRIAR
P/2 P/2
(e) (f)
Pl/12 Pl/12
Figure 14-30 Consistent Loads Due to Uniform In-Plane Element Edge Load
Consider another example with solid elements. A load P is to be applied evenly as an

outward pressure load (P/A) to a surface of each of the four solid elements shown in
Figure 14-31. Again, if you lump these loads equally to the grid points, the lumped load
at each node is P/4 for the CHEXA (with eight nodes) and CPENTA (with six nodes), and
P/8 for the CHEXA (with 20 nodes) and CPENTA (with 15 nodes). The consistent loads
are shown in Figure 14-31. The consistent and lumped loads are the same for the CHEXA
(with eight nodes) and the CPENTA (with six nodes) for this case. The consistent loads
for the CHEXA (with 20 nodes) and CPENTA (with 15 nodes) are shown in
Figure 14-31(c) and Figure 14-31(d), respectively. The load distribution along each edge
is –.0833P, .3333P and –.08333P, or a –1/4/–1 ratio, which is quite counter intuitive,
especially for the sign change. Fortunately, you do not have to calculate this consistent
load for the pressure load. This consistent load is generated automatically inside NX
Nastran by using the PLOAD4 entry.
CHAPTER 14 437
Modeling Guidelines
P/4 P/4
CPENTA
P/4 P/4
P/4 (6 Nodes) P/4
CHEXA
(8 Nodes)
CPENTA
P/4 (6 Nodes) P/4
(a) (b)
.08333P .08333P
.3333P .3333P
.3333P .3333P
.08333P CPENTA .08333P

(15 Nodes) .08333P
.08333P
CHEXA .3333P .3333P
(20 Nodes) CPENTA

.3333P .3333P
.08333P (15 Nodes) .08333P
(c) (d)
Figure 14-31 Consistent Loads Due to Uniform Loads on a Solid Element Face
As you can see, consistent loads are functions of the element types and applied loads.
14.7 Symmetry
When you think of a symmetrical structure, you most likely think of a structure that has
one or more planes of reflective symmetry. Although there are other kinds of symmetry
available in NX Nastran, reflective symmetry is the only type that is discussed in this
section. NX Nastran also provides a series of special solution sequences that automates
some of these other types of analyses. They are known as the cyclic symmetry features.
If a structure is symmetric, then the size of your finite element model can be reduced,
which, in turn, reduces the time and cost of your analysis. For each plane of symmetry
that you have in your model, the model size can be reduced by a factor of approximately
two. Figure 14-32(a) and Figure 14-32(b) illustrate structures that contain one and two
planes of symmetry, respectively. In the first case, only half the model needs to be
represented. In the second case, only a quarter of the model needs to be represented.
Plane of Plane of
Symmetry Symmetry
Plane of
Symmetry
(a) (b)
Figure 14-32 Structures with Planes of Symmetry
If the loads applied to the structure are symmetric relative to the plane of symmetry, then
the full model can be replaced with half the model by applying a symmetric boundary
condition. On the other hand, if the loads are antisymmetric, the same simplification can
be achieved by applying the antisymmetric boundary condition.
A symmetric boundary condition implies that the displacements normal to the plane of
symmetry and rotations about the axes in the plane of symmetry are zero at the plane of
symmetry. An antisymmetric boundary condition implies that the displacements in the
plane of symmetry and rotations normal to the plane of symmetry are zero at the plane of
symmetry.
CHAPTER 14 439
Modeling Guidelines
Any general loading condition can be broken into a combination of symmetric and
antisymmetric loads relative to the plane of symmetry. You can work through an
example of applying symmetry using the model in Figure 14-33(a). This load can be
broken into two separate loads as shown in Figure 14-33(b) and Figure 14-33(c).
5000 lb 2500 lb 2500 lb 2500 lb 2500 lb
4
2 3
1 5
CL CL CL
y
(a) (b) (c)
Figure 14-33 Decomposing a General Loading into Symmetric

and Antisymmetric Loadings
Note that Figure 14-33(a) is a case of a general loading applied to a structure with one
plane of symmetry. Figure 14-33(b) and Figure 14-33(c) represent a symmetric loading
and an antisymmetric loading, respectively, applied to the same structure. By using
symmetry, only one-half of the structure, as shown in Figure 14-33(b), needs to be
modeled by applying the symmetric boundary condition at the plane of symmetry (see
Figure 14-34).
2500 lb 2500 lb 2500 lb
2 3 4 2 3 SPC DOFs 1,5,6

at Grid Point 3
for Symmetric
Boundary
1 5 1 Y
CL CL X
(a) (b)
Figure 14-34 Symmetric Model
Similarly, only one-half of the structure shown in Figure 14-33(c) needs to be modeled by
applying the antisymmetric boundary condition as shown in Figure 14-35.
2500 lb 2500 lb 2500 lb
2 3 4 2 3 SPC DOFs 2,3,4

at Grid Point 3
for Antisymmetric Boundary
5 Y
1 1
CL CL X
(a) (b)
Figure 14-35 Antisymmetric Model
User Interface
No special user interaction is required in the Executive Control Section or Bulk Data
Section. Depending on the number of planes of symmetry, approximately one-half or less
of the structure needs to be modeled. For the above frame model, only one-half of the
model needs to be analyzed.
This feature is basically activated by Case Control commands. The above frame example
is used to illustrate the required Case Control commands. Listing 14-3 contains the input
file for this problem.
CHAPTER 14 441
Modeling Guidelines
Listing 14-3 Input File for Using Symmetric and Antisymmetric

Boundary Conditions
$
$ FILENAME - symbar.dat
$
ID SYM ANTI
TIME 5
SOL 101
CEND
$
TITLE = SYMMETRIC AND ANTISYMMETRIC
SUBCASE 1
LABEL = SYMMETRIC CONSTRAINTS - Y LOAD
SPC = 1
LOAD = 2
$
SUBCASE 2
LABEL = ANTISYMMETRIC CONSTRAINTS - Y LOAD
SPC = 2
LOAD = 2
$
SUBCOM 3
LABEL = LEFT SIDE OF MODEL - Y LOAD
SUBSEQ 1.0, 1.0
DISP=ALL
$
SUBCOM 4
LABEL = RIGHT SIDE OF MODEL - Y LOAD
SUBSEQ 1.0, -1.0
DISP=ALL
$
BEGIN BULK
CBAR 1 100 1 2 -1.0 0.0 0.0
CBAR 2 100 2 3 0.0 1.0 0.0
FORCE 2 2 2500. 0.0 -1. 0.0
GRID 1 0.0 0.0 0.0 123456
GRID 2 0.0 10.0 0.0
GRID 3 5.0 10.0 0.0
MAT1 1 3.+7 0.3
PBAR 100 1 5.0 5.0 5.0 10.
SPC1 1 156 3
SPC1 2 234 3
ENDDATA
The first subcase calls out the symmetric boundary condition and load as shown in
Figure 14-34(b). The second subcase calls out the antisymmetric boundary condition and
load as shown in Figure 14-35(b).
The third subcase is a “results combination” subcase. No boundary or load conditions

need to be applied for this loading condition (see “Modeling Guidelines” on page 391 ).
This subcase produces results for the portion of the structure that you have modeled (the
left-hand side of the frame in this case) by performing a linear combination of the first two
subcases. This combination is achieved by adding 100% of the results for the first subcase
to 100% of the second subcase using the SUBSEQ command. Note that in this case of
results combination, the SUBCOM instead of the SUBCASE command is used.
The fourth subcase is also a “results combination” subcase. It produces results for the
portion of the structure that you did not model (the right-hand side of the frame in this
case). These results are achieved by subtracting 100% of the results of the second subcase
from 100% of the results from the first subcase. As in Subcase 3, the SUBCOM command
instead of the SUBCASE command is used in this case. Once again, no load or boundary
condition is needed for this subcase; this is an optional subcase. It is not required if you
do not want to obtain results for the other half of the structure that you did not model.
The displacements for SUBCOM 3 and SUBCOM 4 are shown in Figure 14-36 along with
the results obtained from a full model run. The results for SUBCOM 3 (left-hand side of
the frame) correlate with the full model results. The results for SUBCOM 4 (right-hand
side of the frame) correlate with the full model results except for the sign changes for
components x, θy, and θz. The sign changes occur because the results of the right half
(reflected half) are produced in terms of its left half using the left-hand coordinate system.
Note that these are the same degrees of freedom that are constrained for the symmetric
boundary condition.
LEFT SIDE OF MODEL - Y LOAD SUBCOM 3

1 G .0 .0 .0 .0 .0 .0
2 G -1.381215E-04 -3.278085E-04 .0 .0 .0 2.762431E-05
3 G -1.381215E-04 -1.666667E-04 .0 .0 .0 3.453039E-05
RIGHT SIDE OF MODEL - Y LOAD SUBCOM 4
1 G .0 .0 .0 .0 .0 .0
2 G 1.381215E-04 -5.524853E-06 .0 .0 .0 -2.762431E-05
3 G 1.381215E-04 -1.666667E-04 .0 .0 .0 -3.453039E-05
FULL MODEL - NEGATIVE Y LOAD SUBCASE 1

1 G .0 .0 .0 .0 .0 .0
2 G -1.381215E-04 -3.278085E-04 .0 .0 .0 2.762431E-05
3 G -1.381215E-04 -1.666667E-04 .0 .0 .0 3.453039E-05
4 G -1.381215E-04 -5.524862E-06 .0 .0 .0 2.762431E-05
5 G .0 .0 .0 .0 .0 .0
Figure 14-36 Results Comparison Using Symmetry Versus a Full Model
In the past, when computers were not as fast as they are now, the use of symmetry to
reduce the finite element model size was more popular than it is today. The advantage of
using symmetry is obviously the reduction in model size. The disadvantage is that it
requires more effort on your part and it is more prone to errors since you have to provide
CHAPTER 14 443
Modeling Guidelines
equivalent symmetric/antisymmetric boundaries and loading conditions to your model.

In NX Nastran, the use of symmetry is built into some of the solution sequences-they are
known as the cyclic symmetry solution sequences. Similar approaches can be used with
superelement analysis.
The SYM/SYMCOM Case Control command combination can be used instead of the
SUBCASE/SUBCOM combination. If you use this combination, the only changes you
have to make is to replace the SUBCASE and SUBCOM Case Control commands with the
SYM and SYMCOM Case Control commands, respectively. When the
SUBCASE/SUBCOM combination is used, output is available in both the SUBCASE and
SUBCOM subcases. On the other hand, when the SYM/SYMCOM combination is used,
output requests are only available in the SYMCOM subcases. There is no real advantage
to using the SYM/SYMCOM command as compared to the SUBCASE/SUBCOM
command.
15
CHAPTER
Model Verification
■ Introduction to Model Verification
■ Preprocessor Checks
■ Strain Energy Output
■ Diagnostic Tools
■ Stress Error Estimators
■ Postprocessor Checks
15.1 Introduction to Model Verification

This chapter is dedicated to various tools that are available for verifying and improving
the quality of your models. They are broken into five different categories:
• Preprocessor checks.
• Strain energy output.
• Diagnostic tools.
• Stress error estimators.
• Postprocessor checks.
Some of these categories overlap each other. Most of these checks provide you with a
tremendous amount of information, but yet their usages are quite straightforward and
require very little effort on your part.
15.2 Preprocessor Checks

Preprocessors are used extensively for the generation of finite element models. They also
contain features that can help you to improve the quality of your model and detect errors
in your model prior to analysis. These features can result in substantial savings in terms
of time and money. Depending on the graphics package that you use, some of these
features may or may not be available to you.
Creating a Shrink Plot of the Elements

The shrink option allows you to shrink your elements by a specified percentage. This
feature is an excellent tool for identifying missing elements. As an example, let us look at
a plate model with 12 CQUAD4 elements surrounded by 31 CBAR elements as shown in
Figure 15-1. (A) shows the regular plot, while (B) shows the plot with the shrink option
turned on. The CBAR elements aren’t visible since they lie on top of the edges of the
CQUAD4 elements.
If you turn on the shrink option, as shown in Figure 15-1(b), you can easily see that you’re
missing a CBAR element at the top and a CQUAD4 element close to the center. You may
argue that had the element labels been turned on, the missing elements would probably
have been noticed. This is certainly a valid argument in this case since the model is a
simple one. However, for complex models, the picture would have become too crowded
to be useful had all the labels been turned on. This same shrink feature can also be applied
to solid elements in a similar manner.
CHAPTER 15 447
Model Verification
(a) (b)
Figure 15-1 Shrink Plot
Hidden Line/Shaded Plots

For a complicated model, looking at a wireframe model by itself can be quite confusing
since you lose the depth of field. However, hidden line and shaded plots can serve as the
perfect complement to the wireframe plots. The box model in Figure 15-2 is a perfect
example. This model is made of CHEXA elements and has a hole in the front. By looking
at the wireframe plot on the top left-hand corner of Figure 15-2, you may not notice that
there is a hole in the front. However, the hidden line (top right), shaded (bottom left), and
shaded with shrink option (bottom right) plots all reveal the hole in front of the box.
Furthermore, when using these features, you can see the structure being plotted from
back to front, which gives you a better perspective of what the structure looks like. As the
structure becomes more complicated, you will appreciate these features even more.
Figure 15-2 Wireframe, Hiddenline, and Shaded Plots
Free Edge/Face
A free edge is an edge that is connected by one single two-dimensional element (e.g.,
CQUAD4). The existence of free edges does not necessarily indicate a modeling problem.
Figure 15-3 represents a crude wing model. The bold lines around the outside are
examples of legitimate free edges. On the other hand, whether the internal vertical bold
line is a legitimate free edge or not depends on your design intent. This free edge indicates
that elements 12 and 14, 11, and 13 are not connected to each other. If this is not your
intention, then this free edge indicates a potential modeling error.
This can happen, for example, if you create this wing with four separate surfaces as shown
in Figure 15-4. Most preprocessors typically create geometric surfaces and then generate
the finite element mesh for each of these surfaces. Since elements 12 and 14 are created
from two separate geometric surfaces, their connecting grid points have distinct IDs. After
these elements are generated, if you want them to share the same edge, then you must
perform some sort of equivalencing operation. The exact operation depends on the
graphics package itself. This free edge indicates that you have either forgotten to perform
this equivalencing operation or the equivalencing tolerance level is not tight enough for
the program to perform this task.
CHAPTER 15 449
Model Verification
16 14 12 10 8 6 4 2
15 13 11 9 7 5 3 1
Figure 15-3 Wing Free Edge Check
Surface Surface Surface Surface

1 2 3 4
Figure 15-4 Wing Geometric Surfaces
The free face concept is similar to the free edge concept except that it applies to
three-dimensional elements instead of two-dimensional elements. A free face is a face
that is occupied by one single three-dimensional element (e.g., CHEXA). The example in
Figure 15-5 contains two volumes. After these two volumes are meshed with solid
elements (e.g., CHEXAs), a free face check is performed. The outside surfaces of these two
volumes indicate that they are free faces that are legitimate. In addition, it indicates that
the adjoining surface (hashed area) is also a free surface. This indicates that the two
volumes are not connected, which may or may not be your intention. If it is not your
intention, there is probably a modeling error.
Figure 15-5 Possible Modeling Error as Indicated by Free Face
Zipper Effect
When adjoining surfaces are connected at very few points, the interface has a tendency to
open up when it is loaded. This looks similar to a zipper and hence is classified as the
“zipper effect.” Use the surfaces in Figure 15-4 and create new meshes. Surfaces 1, 2, 3,
and 4 contain meshes of 4 X 2, 4 X 2, 3 X 2, and 3 X 2 , respectively. As you can see in
Figure 15-6 the interface between surfaces 2 and 3 is only connected at two grid points—
grid points A and B. The loads can only be transferred between these two portions of the
structure through these two locations. In other words, the load path may be quite different
from what the actual structure does.
Figure 15-6 Model with Zipper Effect
The “zipper effect” can occur in both plate and solid element models. Therefore, when
generating meshes for a structure with multiple surfaces and/or volumes, you should
keep this in mind so that the interfaces are connected properly. You should always
perform a quick check by zooming in at all the interface locations.
CHAPTER 15 451
Model Verification
CBAR/CBEAM Orientation and Offset Check

One common error that many users make is the incorrect orientation of the
CBAR/CBEAM inertia properties. If the graphics package that you are using plots these
orientation vectors, you should take advantage of this feature to ensure that you have
oriented these elements properly. You should also check the CBAR/CBEAM element
offset plots, if they are available.
See Also
• “CBAR Element” on page 22 of the NX Nastran Element Library

• “CBEAM Element” on page 45 of the NX Nastran Element Library
Duplicate Grid Points/Elements

The “duplicate grid point check” feature highlights grid points that occupy the same
location. Having grid points occupy the same space may or may not be your intention.
An example of unintentional grid points occupying the same location can be a result of
not performing an equivalencing operation after meshing more than one curve, surface,
or volume. An application of intentional duplicate grid points can be a model used to
simulate a bolted joint. This effect can be achieved by connecting a stiff spring between
these two points that occupy the same location but are connected to different portions of
the structure.
The “duplicate element check” feature is similar to the “duplicate grid check” except the
check is done on the elements rather than the grid points. These errors are often due to
unintentionally meshing the same line, surface, or volume more than once.
Properties/Material Plots
You can also assign different colors to elements based on their property IDs (e.g., PSHELL
IDs). If you have assigned an incorrect property, graphically this becomes quite obvious.
Differentiating by color can also be used to highlight material properties (e.g., MATi IDs).
Using Consistent Normals

When you create a model, you should always generate the elements in a consistent
manner. In some cases, if you don’t generate your model in a consistent manner, you
might inadvertently apply the loads in the wrong directions.
An example of such a case is when the applied load is a pressure load The plate model,
consisting of four CQUAD4s as shown in Figure 15-7, is used to illustrate this point. As
mentioned in Chapter 4, each CQUAD4 has a positive normal direction associated with it.
This direction is defined by the way you connect the element using the right-hand rule.
The symbols used to denote the directions of the normals are shown in Figure 15-8.
7 8 9
3 4
4 5 6
1 2
1 2 3
Figure 15-7 CQUAD4 Model
Now apply a 100 psi pressure load to these four elements. The direction of the pressure
load is pointing into the paper. Listing 15-1 contains the relevant partial input file for this
job. In this case, the grid points for each element are defined in a consistent manner. They
are connected in a clockwise direction. Therefore, the positive normals are all pointing
into the paper as indicated by Figure 15-8(a).
Listing 15-1 Consistent CQUAD4 Connectivity
ID RUN1 PLOAD4
SOL 101
CEND
TITLE = CONSISTENT ELEMENT NORMAL DIRECTIONS
LOAD = 100
.
.
BEGIN BULK
$
CQUAD4,1,10,1,4,5,2
CQUAD4,2,10,2,5,6,3
CQUAD4,3,10,4,7,8,5
CQUAD4,4,10,5,8,9,6
$
PLOAD4,100,1,100.0,,,,THRU,2
PLOAD4,100,3,100.0
PLOAD4,100,4,100.0
.
.
ENDDATA
CHAPTER 15 453
Model Verification
7 8 9 7 8 9
4 5 6 4 5 6
1 2 3 1 2 3
(a) (b)
– Positive Normal Pointing – Positive Normal Pointing
into the Paper out of the Paper
Figure 15-8 CQUAD4s with Pressure Loads
On the other hand, perhaps for some reason, you connect element 3 in a counterclockwise
direction (Figure 15-8(b)) instead of a clockwise direction like the other three elements. If
this is the case, you then need to change the sign of the pressure load for element number 3
in order to have identical loads for both cases (see Listing 15-2). Determining the positive
direction of the pressure load for the CQUAD4 element can be achieved by checking the
connectivity order using the right-hand rule. If the sign for the pressure load applied to
element number 3 is not switched, then you have three elements with the pressure load
pointing into the paper and the fourth element with the pressure load pointing out of the
paper.
Listing 15-2 Inconsistent CQUAD4 Connectivity
ID RUN1 PLOAD4
SOL 101
CEND
TITLE = INCONSISTENT ELEMENT NORMAL DIRECTIONS
LOAD = 100
.
.
BEGIN BULK
$
CQUAD4,1,10,1,4,5,2
CQUAD4,2,10,2,5,6,3
CQUAD4,3,10,4,5,8,7
CQUAD4,4,10,5,8,9,6
$
PLOAD4,100,1,100.0,,,,THRU,2
PLOAD4,100,3,-100.0
PLOAD4,100,4,100.0
.
.
ENDDATA
This type of inconsistency can occur if the whole model is generated with a preprocessor
using multiple surfaces and the consistency of the connectivity direction is not retained.
This problem is also likely to occur if you edit some of the elements manually and forget
to account for the sign change in the pressure load. Many graphics packages offer this
element normal check feature. This type of check requires very little effort, and in many
cases the graphics processor may also allow you to reverse the direction of the normals.
The CQUAD4 element outputs are in the element coordinate system. Different
connectivity orders affect the way the results are printed. If you are not aware of this fact,
you may interpret the results improperly (see the example in “Stress Error Estimators” on
page 494).
15.3 Strain Energy Output

Frequently you are faced with the decision of modifying your structure in order to meet
stringent design criteria. In most instances, your objective is to maximize the benefits with
minimal changes. The element strain energy output is an excellent tool for identifying
areas of modification that will reap the most benefits for design changes.
The element strain energy is basically the elastic energy stored in the structural element.
As an example, if you hold onto one end of the spring and push slowly on the other end
starting from rest, the load deflection curve looks something like Figure 15-9 for small
deflections.
CHAPTER 15 455
Model Verification
Area underneath the

curve = strain energy
F
δ
Figure 15-9 Load Deflection Curve
This energy is defined as
U = 0.5 ⋅ F ⋅ δ
Since F = k < d,
2
U = 0.5 ⋅ kδ Eq. 15-1
By working through a couple of examples, you can see how the element strain energy
output is used to identify areas to be modified in order to reduce deflections.
The problem of interest is shown in Figure 15-10. It consists of two springs in a series with
a tip load applied at grid point 2. The stiffnesses of the springs and the load applied are
as shown in Figure 15-10. The goal is to reduce the tip deflection at grid point 2. Without
performing any calculation, it is quite obvious that stiffening K 2 is more efficient than
stiffening K1 . The next step is to calculate the strain energy and see if it also guides you
in the same direction.
K 1 = 10 K2 = 1
P = 1
1 2
Figure 15-10 Spring Sample Problem
The deflections at grid points 1 and 2 can be calculated as follows:
P P P
δ 1 = ------ ; δ 2 = ------ + ------ Eq. 15-2
K1 K1 K2
From Eq. 15-1 and Eq. 15-2,
2 1 2
U 1 = 0.5 ( K 1 ) ( δ 1 ) = 0.5 ( 10 )  ------ = 0.05
10
2 1 1 1 2
U 2 = 0.5 ( K 2 ) ( δ 2 – δ 1 ) = 0.5 ( 1 )  ------ +  --- –  ------ = 0.5
 10  1  10
As you can see, U 2 is an order of magnitude larger than U 1 . Therefore, it concurs with our
intuition that stiffening K2 is more effective than stiffening K 1 for reducing the deflection
at the tip.
The second example is a classic cantilever beam with a vertical tip load applied at the end
as shown in Figure 15-11. This finite element model is made up of five bar elements of
equal length with square cross-sectional properties (0.05m x 0.05m) as shown in
Table 15-1.
Table 15-1 Cross-Sectional Properties of a Cantilever Beam
Elem. L W D A I1 I2 J
No. (m) (m) (m) (m2) (m4) (m4) (m4)
1 1.0 0.05 0.05 2.5E-3 5.208E-7 5.208E-7 8.789E-7

2 1.0 0.05 0.05 2.5E-3 5.208E-7 5.208E-7 8.789E-7
3 1.0 0.05 0.05 2.5E-3 5.208E-7 5.208E-7 8.789E-7
4 1.0 0.05 0.05 2.5E-3 5.208E-7 5.208E-7 8.789E-7
5 1.0 0.05 0.05 2.5E-3 5.208E-7 5.208E-7 8.789E-7
z
N
E = 7.1 E10 -------
2
m 1N
x
1 2 3 4 5 6
1 2 3 4 5
5m
Figure 15-11 Cantilever Beam
As is commonly known, the largest deflection occurs at the tip of the cantilever beam. The
vertical deflection at grid point 6 is equal to 1.126833 ⋅ 10 – 3 meters. The goal is to minimize
this deflection at grid point 6. To make the problem more interesting, hypothetically
impose the following constraint such that all that is available is one structural member that
CHAPTER 15 457
Model Verification
is 0.06m ⋅ 0.06m and 1m long. In other words, you can only replace one of the existing
five CBAR elements with this new CBAR element. Which one should you replace to
minimize the deflection at grid point 6? Since grid point 6 has the largest deflection,
combined with our experience with the spring problem, you may be tempted to replace
CBAR element number 5 with this new element. If you do this, the vertical displacement
at grid point 6 is then reduced from 1.126833 ⋅ 10 – 3 meters down to 1.122165 ⋅ 10 – 3 meters, a
reduction of only 0.414%. Now run this problem using NX Nastran and see what the
element strain energy output suggests. Listing 15-3 contains a listing of the NX Nastran
input file used for this purpose. Note that the only additional request required is the
following command in the Case Control Section:
ESE = ALL
or
SET a = x,y,......
ESE = a
or
ESE(PLOT) = ALL
For large models, the element strain energy request can potentially generate a large
amount of printout. If this is not desired, you can use the “plot” option shown above to
create a postprocessing file containing element strain energy data without generating
printed output (see Section ). The rest of the input file is standard NX Nastran input.
Listing 15-3 Input File for the Strain Energy of a Cantilever Beam
$
$ FILENAME - secant.dat
$
ID CANT ESE
SOL 101
TIME 10
CEND
$
TITLE = BASELINE MODEL
SUBTITLE =
$
SPC = 10
LOAD = 10
$
SUBCASE 1
SET 1 = 6
DISP = 1
ESE = ALL
$
BEGIN BULK
$
CBAR 1 10 1 2 10
CBAR 2 100 2 3 10
CBAR 3 100 3 4 10
CBAR 4 100 4 5 10
CBAR 5 20 5 6 10
FORCE 10 6 1. 0. 0. 1.
GRID 1 0. 0. 0.
GRID 2 1. 0. 0.
GRID 3 2. 0. 0.
GRID 4 3. 0. 0.
GRID 5 4. 0. 0.
GRID 6 5. 0. 0.
GRID 10 0. 0. 10.
MAT1 1 7.1+10 .33 2700.
PBAR 10 1 2.5-3 5.208-7 5.208-7 8.789-7
PBAR 20 1 2.5-3 5.208-7 5.208-7 8.789-7
$
$ REPLACE PBAR,10,...
$ WITH THE FOLLOWING FOR .06 X .06 CROSS SECTION
$
$PBAR,10,1,3.6-3,1.08-6,1.08-6,1.823-6
$
PBAR 100 1 2.5-3 5.208-7 5.208-7 8.789-7
SPC1 10 123456 1
$
ENDDATA
The NX Nastran element strain energy output for this run is summarized in Figure 15-12.
As it turns out, element number 1 has the highest element strain energy among the five
elements.EMail:[email protected]
www.cadfamily.com In fact, it contains 48.8% of the total strain energy of the whole structure. On the
CHAPTER 15 459
Model Verification
other hand, element number 5 contains the lowest element strain energy among the five
elements—only 0.8%. In other words, element number 5 is the least effective element to
modify if you wish to increase the stiffness of the structure for the same amount of weight
increase.
This job is now rerun by replacing element number 1 with the new 0.06 × 06 element.
The deflection at grid point 6 is now reduced from 1.126833 × 10 – 3 meters down to
–4
8.421097 × 10 meters, or a reduction of 25.27% as compared to a mere 0.414% reduction if
you had replaced element number 5 instead.
E L E M E N T S T R A I N E N E R G I E S
ELEMENT-TYPE = BAR * TOTAL ENERGY OF ALL ELEMENTS IN PROBLEM = 5.634164E-04

SUBCASE 1 TOTAL ENERGY OF ALL ELEMENTS IN SET -1 = 5.634164E-04
*
ELEMENT-ID STRAIN-ENERGY PERCENT OF TOTAL STRAIN-ENERGY-DENSITY
1 2.749472E-04 48.8000 1.099789E-01
2 1.667712E-04 29.6000 6.670850E-02
3 8.563929E-05 15.2000 3.425572E-02
4 3.155132E-05 5.6000 1.262053E-02
5 4.507331E-06 .8000 1.802932E-03
TYPE = BAR SUBTOTAL 5.634164E-04 100.0000
Figure 15-12 Element Strain Energy Output for a Cantilever Beam Model
As you can see, the element strain energy output is an extremely useful tool in helping you
to identify the most efficient locations for modification. The same concept used for the
previous two simple examples can be applied to complex models in the same manner.
Again, the only additional input needed is the “ESE” request in the Case Control Section.
The output format is identical to that of Table 15-2. Furthermore, most modern plot
packages also support element strain energy output. For plate and solid elements, these
strain energy plots look similar to stress contour plots.
A car model (courtesy of Lapcad Engineering), containing CQUAD4 and CTRIA3

elements, is used to illustrate the element strain energy contour plot. The Executive
Control and Case Control Sections are shown in Listing 15-4. As you can see, the amount
of user input required is quite small. The Bulk Data for this model is inserted into the
input file using the INCLUDE statement. A 1g loading is applied to each of the three
orthogonal axes. The element strain energy contour plot for a 1g loading in the z-direction
is shown on a deformed plot in Figure 15-13.
Listing 15-4 Input File for the Strain Energy of a Car Model
$ FILENAME - secar.dat
$
$ MODEL COURTESTY OF LAPCAD ENGINEERING
$
ID LSUG, CAR
SOL 101
TIME 60
CEND
TITLE = CAR MODEL WITHOUT SUSPENSION SPRINGS
ECHO = UNSORT
SPC = 10
DISP(PLOT) = all
ESE(PLOT) = all
$
SUBCASE 10
LABEL = 1G LOAD IN X-DIRECTION
LOAD = 1
$
SUBCASE 20
LABEL = 1G LOAD IN Y-DIRECTION
LOAD = 2
$
SUBCASE 30
LABEL = 1G LOAD IN Z-DIRECTION
LOAD = 3
$
BEGIN BULK
$
INCLUDE ’carbulk.dat’
$
ENDDATA
Figure 15-13 Element Strain Energy Contour Plot of an Automobile Model

CHAPTER 15 461
Model Verification
(Geometry model courtesy of LAP CAD Engineering)
15.4 Diagnostic Tools

There are numerous diagnostic tools available in NX Nastran for debugging and
understanding the solution process. Some of them are standard output; others are
available with requests in the Executive Control Section, Case Control Section, and/or
Bulk Data Section.
This section covers many of these tools. Whenever possible, the problem in Listing 15-5
is used to demonstrate these features. The problem in Listing 15-5 is similar to the
cantilever beam shown in Figure 15-5 (Listing 15-3) with the exception of an additional
load case (gravity load) and output request. The modifications and additions are
indicated by the shaded regions.
Listing 15-5 Input File for Illustrating Diagnostic Checks
$ FILENAME - diag.dat
SOL 101
TIME 10
CEND
$
TITLE = DIAGNOSTIC TOOLS
$
SPC = 10
DISP = ALL
SPCF = ALL
OLOAD = ALL
$
SUBCASE 1
LABEL = VERTICAL TIP LOAD
LOAD = 10
$
SUBCASE 2
LABEL = GRAVITY LOADS
LOAD = 20
$
BEGIN BULK
$
PARAM,GRDPNT,0
$
CBAR 1 10 1 2 10
CBAR 2 10 2 3 10
CBAR 3 10 3 4 10
CBAR 4 10 4 5 10
CBAR 5 10 5 6 10
FORCE 10 6 100. 0. 0. 1.
GRAV 20 9.8 0. 0. 1.
GRID 1 0. 0. 0.
GRID 2 1. 0. 0.
GRID 3 2. 0. 0.
GRID 4 3. 0. 0.
GRID 5 4. 0. 0.
GRID 6 5. 0. 0.
GRID 10 0. 0. 10.
$ 2 3 4 5 6
MAT1 1 7.1+10 .33 2700.
PBAR 10 1 2.5-3 5.208-7 5.208-7 8.789-7
SPC1 10 123456 1
$
ENDDATA
Element Summary Output

The ELSUM Case Control command provides a quick summary table of properties for the
elements in the model. These properties include element-id, material-id, length (or
thickness), area, volume, structural mass, non-structural mass, total mass, and weight
(weight mass * total mass). ELSUM provides an extension of the element summaries
previously obtained with PARAM,EST,1.
CHAPTER 15 463
Model Verification
The following examples highlight some of the new features available via the ELSUM
command. Note that the ELSUM command on line 11 in subcase 100 requests summary
information for those elements defined in SET 1. The corresponding rod element output
follows.
0
0 CASE CONTROL ECHO
COMMAND
COUNT
1 $
2 SET 1 = 401 THRU 403
3 DISP(PRINT,PUNCH)=ALL
4 ELST(PRINT,PUNCH)=ALL
5 ELFO(PRINT,PUNCH)=ALL
6 SPCF(PRINT,PUNCH)=ALL
7 ESE(PRINT,PUNCH)=ALL
8 SPC=50
9 $
10 SUBCASE 100
11 ELSUM=1
12 LOAD=100
13 ESE(PRINT)=ALL
14 $
15 BEGIN BULK
0
E L E M E N T P R O P E R T Y S U M M A R Y
0ELEMENT TYPE = ROD
ID MID LENGTH AREA VOLUME SM NSM TM WEIGHT
401 202 1.000000E+01 6.000000E+00 6.000000E+01 4.800000E+02 5.000000E+00 4.850000E+02 4.850000E+02
402 201 1.000000E+01 4.000000E+00 4.000000E+01 1.600000E+02 8.000000E+00 1.680000E+02 1.680000E+02
403 202 1.000000E+01 6.000000E+00 6.000000E+01 4.800000E+02 5.000000E+00 4.850000E+02 4.850000E+02
SUBTOTAL MASS = 1.120000E+03 1.800000E+01 1.138000E+03 1.138000E+03
TOTAL MASS = 1.120000E+03 1.800000E+01 1.138000E+03 1.138000E+03
See Also
• “ELSUM” on page 237 of the NX Nastran Quick Reference Guide
Element Geometry Checks

Mathematical models are usually composed of several different types of finite elements
depending upon the simulation being performed and the type of structure being
investigated. For each of these finite element formulations, thorough evaluations of the
geometry presented are performed prior to and during element matrix generation
operations.
These geometry tests are typically used to determine whether (1) the element geometry is
adequate for finite element matrix generation at all, and if so, (2) how close the geometry
could be to producing poor formulations. Taking a beam element for example, a fatal
geometry condition would be locations of the two end points that result in an element
with zero length. An example of the second type of geometry check for a beam would be
an excessive ratio of the length of the beam with offset vectors to the length of the beam
without them. In the first case, the geometry check is fatal because the finite element
cannot be formulated from the given geometry and the simulation cannot be done. In the
second case, the geometry check is informational and the simulation analysis can proceed.
It is your responsibility to inspect these informational messages and determine if the
results are acceptable.
Several different basic geometry tests are performed; for example, duplicate grid point
location tests, mid-side node location tests, warped element tests, and aspect ratio tests.
One or more of these types of tests is performed using the geometry supplied for each
finite element. Obviously, not all tests apply to every element type. Each one of the
geometry tests is accompanied by a message that can generate up to four lines of
informational output for an element if the test tolerance is violated. For models that
employ a small number of elements, this amount of output is not overwhelming.
However, the size and complexity of mathematical simulation models have continued to
grow over time.
NX Nastran reports the results for all tests for a single element are reported on a single line
together with a visual indication of which tests have exceeded their tolerances and the
severity of the failure. Default values for all test tolerances are defined such that geometry
tests result in the same type of pass/fail decision as previous releases. In a few cases,
additional tests are performed. You can also use GEOMCHECK executive statement to
have more control over the test tolerances, their severity levels, and the number of
messages output.
Using the GEOMCHECK Statement

The GEOMCHECK statement lets you override the default values supplied for all tests.
Except for a very few isolated instances, bad geometry that doesn’t allow the formulation
of finite element matrix data is always fatal and isn’t under user control. In these cases,
every test that fails produces a message for every element.
For the cases where geometry doesn’t prevent finite element matrix generation, NX
Nastran performs additional tests that evaluate various geometry parameters and reports
the results. For these tests, you can control both the test tolerance and the severity level of
test failure.
Using the BAR element as an example, there is a test performed that evaluates the ratio of
the BAR’s length with offset effects to the length without offset effects. The tolerance for
this test is 15%, meaning that if the length with offset is different from the length without
the offset by more than fifteen percent, an informational message is issued and element
processing continues. The user can change the tolerance if desired. The severity of the test
failure can also be changed from informational to fatal if desired. In that case, any BAR
element that exceeds the offset ratio test tolerance will cause the job to stop after element
matrix generation has been attempted for all elements.
CHAPTER 15 465
Model Verification
You can use the GEOMCHECK statement to modify geometry test activities for the
following types of elements:
• QUAD4 and QUADR

• TRIA3 and TRIAR
• BAR
• BEAM
• HEXA
• PENTA
• TETRA
As stated previously, only informational tests can be affected using the GEOMCHECK
statement. Results of geometry generated by NX Nastran that prevent the generation of a
finite element matrix cannot be modified using the GEOMCHECK statement.
There are several keywords for GEOMCHECK that let you control the number of
messages to be generated (MSGLIMIT) as well as the severity of the message (MSGTYPE)
and its associated effect on the job. Only a severity of FATAL causes the job to abort after
the element matrix generation module has executed. In this case, the identifier associated
with failed tests is FAIL and the run will also terminate.
See Also
• “GEOMCHECK” on page 134 of the NX Nastran Quick Reference Guide
Geometry Test Descriptions

NX Nastran performs several different types of geometry tests. For each finite element
type, the geometry must be tested to ensure that it is adequate to allow generation of
element matrices. Those tests are not described here. This section only describes optional
tests that certain characteristics of the geometry. These tests fall into several categories
depending upon the element type. Thus, not all tests are applicable to every element type.
Optional Tests for BAR and BEAM Elements

The only optional test available for these element types is the offset test. The length of the
element offset is compared to the original length of the element. If the ratio of these
lengths is greater than the tolerance, an informational message is issued identifying the
element and its length with and without the effects of offset.
Optional Tests for CQUAD4 and CQUADR Elements

Four optional tests are performed on the quadrilateral shaped elements. These tests are:
1. Interior Angle test: This test evaluates the interior angles measured at each of the
four corner grid points. If any one of the four angles exceeds minimum or
maximum tolerance levels, an informational message is produced.
2. Taper test: The QUAD4 taper ratio test compares the area of the triangles formed
at the four corner nodes (the corner node and the two nodes attached to it via the
two element edges) to the total area of the QUAD4. The minimum of the four
values is taken as the taper ratio. If the ratio exceeds the tolerance, an
informational message is produced.
4 3 4 3
A4 A3
A2 A1
1 2 1 2
J i = 0.5 A i J a = 0.25 ( J 1 + J 2 + J 3 + J 4 )
Ji – Ja
Taper message is issued if ----------------- > 0.5
Ja
Figure 15-14 CQUAD4 Taper
3. Skew test: This test evaluates the distortion or “parallelogram shape effect” by
measuring the angle between lines that join the midpoints of opposite sides of the
element. If the angle exceeds the tolerance, an informational message is
produced.
4 3
α Skew message is issued if α < 30°
1 2
Figure 15-15 CQUAD4 Skew
CHAPTER 15 467
Model Verification
4. Warp test: This test evaluates how far out of plane the four corner grid points are
by measuring the distance of each point from a “mean” plane passing through
the locations of the four points. The corner points are alternately H units above
and H units below this mean plane. If the lengths of the diagonals of the element
are denoted by D1 and D2, the warping coefficient is obtained from the equation
WC = H / 2(D1+D2). If this value exceeds the tolerance, an informational
message is produced.
Optional Tests for CTRIA3 and CTRIAR Elements

Only one optional test is performed on the triangular shaped planar elements. It is the
same as the interior angle test discussed previously under the CQUAD4 and CQUADR
test description.
Optional Tests for CHEXA, CPENTA and CTETRA Elements

There are several different optional tests that you can perform on solid elements. Not all
tests are applicable to every element type. These tests are:
• Aspect Ratio test: This test evaluates the ratio of the longest length (edge or
height) to the shortest length (edge or height) encountered in the element. If the
ratio exceeds the tolerance, an informational message is produced.
• Edge Point Length Ratio test: This test evaluates the location of the “mid-side”
nodes if any are present. The node should be located on the line connecting the
two adjacent corner grid points at approximately the mid-way point. This
determination is made in two ways. First, the distance of the node from the two
corner nodes is measured and if the ratio of the distance from one node to the
other exceeds the tolerance, an informational message is issued. Next, the angles
between the lines connecting the mid-side node to its two adjacent corner nodes,
and the line connecting the corner nodes themselves, is evaluated. If either of the
angles exceeds 30 degrees, an informational message is issued.
• Integration Point Jacobian Determinant test: This test evaluates the determinant
of the Jacobian at each integration point. If it is zero, or changes sign from
integration point to integration point, an informational message is issued. Note
that detection of a zero value should always be fatal because element matrices
calculated from such geometry rarely produce satisfactory analysis results.
• Warped Face test: This test is applicable to CHEXA and CPENTA elements. It
evaluates the warping coefficient of the quadrilateral faces of such elements in a
manner similar to that discussed previously under the CQUAD4 element
warping test description. If the warping coefficient exceeds the tolerance, an
informational message is issued.
d
c
a
Figure 15-16 CHEXA - Warping
• Grid Point Jacobian Determinant test: This test is applicable to CTETRA elements
only. The test is the same as that done for the integration point test discussed
previously except it uses the location of the corner grid points to perform the test.
If any determinant is zero or changes sign, an informational message is
produced.
User Interface:
The geometry check is requested by the GEOMCHECK Executive Control Statement and
the format is as follows:
FATAL
GEOMCHECK test_keyword [= tol_value], [MSGLIMIT=n], MSGTYPE = INFORM , [ SUMMARY ], [ NONE ]
WARN
The following table summarizes the acceptable specifications for the test_keyword:
Value
Name Default Comment
Type
Q4_SKEW Real > 0.0 30.0 Skew angle in degrees
Q4_TAPER Real > 0.0 0.50 Taper ratio
Q4_WARP Real > 0.0 0.05 Surface warping factor
Q4_IAMIN Real > 0.0 30.0 Minimum Interior Angle in degrees
Q4_IAMAX Real > 0.0 150.0 Maximum Interior Angle in degrees
T3_SKEW Real > 0.0 10.0 Skew angle in degrees
T3_IAMAX Real > 0.0 160.0 Maximum Interior Angle in degrees
CHAPTER 15 469
Model Verification
Value
Name Default Comment
Type
TET_AR Real > 0.0 100.0 Longest edge to shortest edge aspect ratio
TET_EPLR Real > 0.0 0.50 Edge point length ratio
TET_DETJ Real 0.0 | J | minimum value
TET_DETG Real 0.0 | J | minimum value at vertex point
HEX_AR Real > 0.0 100.0 Longest edge to shortest edge aspect ratio
HEX_EPLR Real > 0.0 0.50 Edge point length ratio
HEX_DETJ Real 0.0 | J | minimum value
HEX_WARP Real > 0.0 0.707 Face warp coefficient
PEN_AR Real > 0.0 100.0 Longest edge to shortest edge aspect ratio
PEN_EPLR Real > 0.0 0.50 Edge point length ratio
PEN_DETJ Real 0.0 | J | minimum value
PEN_WARP Real > 0.0 0.707 Quadrilateral face warp coefficient
BEAM_OFF Real > 0.0 0.15 CBEAM element offset length ratio
BAR_OFF Real > 0.0 0.15 CBAR element offset length ratio
Examples:
1. GEOMCHECK Q4_SKEW=15.0,MSGLIMIT=50
Set the tolerance for the CQUAD4 element skew angle test to 15.0 degrees and
limit the number of messages to 50.
2. GEOMCHECK SUMMARY
Requests summary table output only using the default tolerance values.
Sample Output:
The following output from a small example provides a sample of the message formats that
are produced for geometry tests that can be controlled using the GEOMCHECK
statement. For this output, most of the default test tolerances were modified so that the
effect could be observed in the tolerance information lines of the messages that were
produced.
Note that a small table is generated at the end of all messages that summarizes the number
of tests that actually exceeded the tolerance value for each element type as well as a list of
the elements that produced the worst violations. The summary table itself can be
produced by using the keyword SUMMARY on the GEOMCHECK statement.
*** USER INFORMATION MESSAGE 7555 (GMTSTD)

FINITE ELEMENT GEOMETRY CHECK RESULTS EXCEED TOLERANCE LEVELS FOR THE FOLLOWING ELEMENTS.
User Action: Use the GEOMCHECK (Executive Control Statement) keyword=value to change tolerance values if desired.
A MINIMUM OF 100 OFFSET LENGTH RATIO TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY xxxx.
TOLERANCE LIMIT IS: BAR_/BEAM_OFF(SET) = .15 (xxxx = LIMIT VIOLATED)

ELEMENT TYPE ID LENGTH W/O OFFSET LENGTH W/ OFFSET OFFSET LENGTH RATIO
BAR 7101 1.00000E+00 3.66367E+00 2.66 xxxx

A MINIMUM OF 4 OFFSET LENGTH RATIO TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY WARN.
TOLERANCE LIMIT IS: BAR_/BEAM_OFF(SET) = -.02 (WARN = LIMIT VIOLATED)

ELEMENT TYPE ID LENGTH W/O OFFSET LENGTH W/ OFFSET OFFSET LENGTH RATIO
BEAM 8101 1.00000E+00 1.10372E+01 10.04 WARN
BEAM 8102 1.00000E+00 1.00000E+00 .00 WARN
*** USER INFORMATION MESSAGE 7555 (EHEXGD)

A MINIMUM OF 100 EDGE LENGTH ASPECT RATIO (AR) TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY xxxx.
A MINIMUM OF 100 EDGE NODE POINT LENGTH RATIO (ER)TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY xxxx.
EDGE ANGLE VALUES ARE INDICATED BY AN * AFTER THE VALUE. THE TOLERANCE LIMIT IS 30 DEGREES.
ALL NEG/ZERO DET(JACOBIAN) VALUE (DJ)TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY FAIL.
A MINIMUM OF 100 WARPED FACE COEFFICIENT TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY xxxx.
TOLERANCE LIMITS ARE: HEX_AR = 300.00, HEX_EPLR = .40, HEX_DETJ = .01, HEX_WARP = .65 (xxxx = LIMIT VIOLATED)
LONGEST SHORTEST ASPECT EDGE POINT MIN. JACOBIAN FACE WARP
ELEMENT TYPE ID EDGE EDGE RATIO LENGTH RATIO DETERMINANT COEFFICIENT
HEXA 8602 4.00 1.00 4.00 N/A- ONLY 8 NODE .92 .52 xxxx
HEXA 8620 4.00 1.00 4.00 .60xxxx .50 1.00

A MINIMUM OF 100 SKEW ANGLE (SA) TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY xxxx.
A MINIMUM OF 100 MIN INT. ANGLE (IA) TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY xxxx.
A MINIMUM OF 100 MAX INT. ANGLE (IA) TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY xxxx.
A MINIMUM OF 100 WARPING FACTOR (WF) TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY xxxx.
A MINIMUM OF 100 TAPER RATIO (TR) TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY xxxx.
TOLERANCE LIMITS ARE: SA = 30.00, IA(MIN) = 30.00, IA(MAX) = 150.00, WF = .05, TR = .50 (xxxx = LIMIT VIOLATED)
ELEMENT TYPE ID SKEW ANGLE MIN INT. ANGLE MAX INT. ANGLE WARPING FACTOR TAPER RATIO
QUAD4 105 29.74 xxxx 29.05 xxxx 150.95 xxxx .00 .05

A MINIMUM OF 100 SKEW ANGLE (SA) TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY xxxx.
A MINIMUM OF 100 MIN INT. ANGLE (IA) TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY xxxx.
A MINIMUM OF 100 MAX INT. ANGLE (IA) TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY xxxx.
A MINIMUM OF 100 WARPING FACTOR (WF) TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY xxxx.
A MINIMUM OF 100 TAPER RATIO (TR) TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY xxxx.
CHAPTER 15 471
Model Verification
TOLERANCE LIMITS ARE: SA = 30.00, IA(MIN) = 30.00, IA(MAX) = 150.00, WF = .05, TR = .50 (xxxx = LIMIT VIOLATED)
ELEMENT TYPE ID SKEW ANGLE MIN INT. ANGLE MAX INT. ANGLE WARPING FACTOR TAPER RATIO
QUADR 1105 29.74 xxxx 29.05 xxxx 150.95 xxxx .00 .05
*** USER INFORMATION MESSAGE 7555 (ETETGD)

A MINIMUM OF 4 EDGE LENGTH TO HEIGHT RATIO (AR)TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY WARN.
A MINIMUM OF 4 EDGE NODE POINT LENGTH RATIO (ER)TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY WARN.
EDGE ANGLE VALUES ARE INDICATED BY AN * AFTER THE VALUE. THE TOLERANCE LIMIT IS 30 DEGREES.
A MINIMUM OF 4 NEG/ZERO DET(JACOBIAN) VALUE (DJ)TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY WARN.
A MINIMUM OF 4 NEG DET(JAC) AT VERTEX NODE (DG)TOLERANCE LIMIT VIOLATIONS WILL BE IDENTIFIED AND INDICATED BY WARN.
TOLERANCE LIMITS ARE: TET_AR = 351.00, TET_EPLR = .45, TET_DETJ = -.01, TET_DETG = -.02 (xxxx = LIMIT VIOLATED)
LONGEST SHORTEST ASPECT EDGE POINT MIN. JACOBIAN MIN. DET(JAC)
ELEMENT TYPE ID EDGE HEIGHT RATIO LENGTH RATIO DETERMINANT AT VERTEX
TETRA 6601 1.41 .58 2.45 .25 WARN .14 -.20 WARN
E L E M E N T G E O M E T R Y T E S T R E S U L T S S U M M A R Y
TOTAL NUMBER OF TIMES TOLERANCES WERE EXCEEDED
ASPECT/ MINIMUM MAXIMUM SURFACE/FACE EDGE POINT JACOBIAN
ELEMENT TYPE SKEW ANGLE TAPER RATIO INTER. ANGLE INTER. ANGLE WARP FACTOR OFFSET RATIO LENGTH RATIO DETERMINANT
BAR N/A N/A N/A N/A N/A 0 N/A N/A
BEAM N/A N/A N/A N/A N/A 0 N/A N/A
HEXA N/A 0 N/A N/A 1 N/A 1 0
PENTA N/A 0 N/A N/A 0 N/A 1 0
QUAD4 10 0 10 10 0 N/A N/A N/A
QUADR 5 0 5 5 0 N/A N/A N/A
TETRA N/A 0 N/A N/A N/A N/A 1 1
TRIA3 0 N/A N/A 0 N/A N/A N/A N/A
TRIAR 0 N/A N/A 0 N/A N/A N/A N/A
N/A IN THE ABOVE TABLE INDICATES TESTS THAT ARE NOT APPLICABLE TO THE ELEMENT TYPE AND WERE NOT PERFORMED.
FOR ALL ELEMENTS WHERE GEOMETRY TEST RESULTS HAVE EXCEEDED TOLERANCES,
HEXA ELEMENT ID 8620 PRODUCED SMALLEST EDGE COS(ANGLE) OF .60 (TOLERANCE = .87).
HEXA ELEMENT ID 8602 PRODUCED SMALLEST FACE WARP FACTOR OF .52 (TOLERANCE = .65).
PENTA ELEMENT ID 3915 PRODUCED SMALLEST EDGE COS(ANGLE) OF .78 (TOLERANCE = .87).
QUAD4 ELEMENT ID 109 PRODUCED SMALLEST SKEW ANGLE OF 24.94 (TOLERANCE = 30.00).
QUAD4 ELEMENT ID 109 PRODUCED SMALLEST INTERIOR ANGLE OF 24.44 (TOLERANCE = 30.00).
QUAD4 ELEMENT ID 109 PRODUCED LARGEST INTERIOR ANGLE OF 155.56 (TOLERANCE = 150.00).
QUADR ELEMENT ID 1109 PRODUCED SMALLEST SKEW ANGLE OF 24.94 (TOLERANCE = 30.00).
QUADR ELEMENT ID 1109 PRODUCED SMALLEST INTERIOR ANGLE OF 24.44 (TOLERANCE = 30.00).
QUADR ELEMENT ID 1109 PRODUCED LARGEST INTERIOR ANGLE OF 155.56 (TOLERANCE = 150.00).
TETRA ELEMENT ID 6601 PRODUCED LARGEST EDGE POINT LR OF .25 (TOLERANCE = .45).
TETRA ELEMENT ID 6601 PRODUCED SMALLEST VERTEX DET(JAC) -UWM 6828- OF -.20 (TOLERANCE = -.02).
Weight Center of Gravity and Moment of Inertia Check

You can use the Grid Point Weight Generator (GPWG) to check whether the weight of
your model is correct. To turn on the Grid Point Weight Generator, include the parameter
GRDPNT in the Bulk Data or Case Control Section of your input file.
The data entry format is
PARAM,GRDPNT,i
where “i” is an integer value defining a reference point. The value of “i” can be any grid
point in the model, or if it is set to zero, the reference point is the origin of the basic
coordinate system.
The output from the GPWG includes a rigid body mass matrix, various coordinate
transformations, and the location of the center of mass. The output from the GPWG is
generally more than what you need. The mass and center of gravity (CG) location is
typically all that is used. A partial output for the problem in Listing 15-5 is shown in
Figure 15-17.
DIRECTION
MASS AXIS SYSTEM (S) MASS X-C.G. Y-C.G. Z-C.G.

X 3.375000E+01 0.000000E+00 0.000000E+00 0.000000E+00
Y 3.375000E+01 2.500000E+00 0.000000E+00 0.000000E+00
Z 3.375000E+01 2.500000E+00 0.000000E+00 0.000000E+00
Figure 15-17 Partial Output from the Grid Point Weight Generator
NX Nastran does not keep track of your units. Therefore, you must input all of your
properties using a consistent set of units. For example, if you are using meters (m) for
defining locations on your grid entries, then your properties, such as areas (A), should be
in terms of m2. NX Nastran expects your mass input (MATi, CONMi, etc.) to be in terms
of mass units. However, if you would rather input your mass in terms of weight units,
then you must add the following entry to your Bulk Data Section:
PARAM,WTMASS,x
In the parameter statement above, x is the appropriate conversion factor based on the units
that you are using with a default value of 1.0. In other words, the weight density is related
to the mass density by the relationship
1
ρ m =  --- ⋅ ρ w = WTMASS ⋅ ρ w
 g
where:
ρ w = weight density
ρ m = mass density
g = gravitational acceleration constant
In the example below, the mass came from the density entered on the MAT1 entry
associated with the CBAR elements. In this case, the density is in terms of mass units.
Since the grid point locations are defined in terms of meters, the density should then be in
units of kg/m3. Using a consistent set of units, the Young’s modulus (E) and density ( ρ )
are 7.1E10 N/m2 and 2700 kg/m3 as indicated by Fields 3 and 6, respectively, on the
following MAT1 entry.
$ MASS DENSITY (kg/m3)

$
MAT1 1 7.1+10 .33 2700.
On the other hand, if you want to input your density in weight units, then you should
replace the above MAT1 entry with the following two entries.
CHAPTER 15 473
Model Verification
$ WEIGHT DENSITY (N/m3)

$ ACCELERATION (m/sec2)
$
MAT1 1 7.1+10 .33 2.65+4
PARAM WTMASS 0.1019
This density and all the other masses (excluding M2PP or M2GG input) are multiplied by
this scale factor. In this case, ( 0.1019 ⋅ 2.65E4 ) = 2700 . The results (e.g., displacements,
stresses, etc.) are the same using either of the above mass or weight units.
The only difference is in the grid point weight generator output. The GPWG output is in
the same units of your density or point masses. In other words, if your input is in mass
units, then the GPWG output is also in mass units. If your input is in weight units, then
the GPWG output is also in weight units.
If you are using English units, then the following two sets of entries yield the same results.
Of course, other units in your model must also be consistent, e.g., the grid point locations
and cross-sectional properties.
$ MASS DENSITY (lb-sec2/in4)

$
MAT1 1 1.0 +7 .33 2.51-4
$ WEIGHT DENSITY (lb/in3)

$ ACCELERATION (in/sec2)
$
MAT1 1 1.0 +7 .33 0.097
PARAM WTMASS 0.002588
The output from GPWG is for informational purposes only and is not used in any
subsequent steps in the solution process.
See Also
• “Grid Point Weight Generator” in the NX Nastran Basic Dynamic Analysis User’s
Guide
Mechanisms and Singularities

When performing the solution to a system of linear equations, singularities lead to
conditions in which a unique solution is not possible. NX Nastran considers two types of
singularities:
• Grid point singularity that is identified by considering the stiffness terms of only
one grid point.
• A mechanism type of singularity that requires the consideration of the stiffness

terms of more than one grid point.
Singularities cause ill-conditioned matrices that can be detected in several phases of the
NX Nastran execution.
After matrix assembly, the grid point singularities are detected. At each grid point, a 3 X 3
partition of the stiffness matrix for each of the three translational and three rotational
DOFs is solved as an eigenvalue problem to determine the principal stiffnesses. Each
stiffness term is compared to the principal stiffness using the formula
K ii
ε = -------------
K max
where Kii is the term in the i-th row and i-th column of the matrix and Kmax is the principal
stiffness. If ε is less than the value of PARAM,EPZERO, the global direction nearest i is
considered singular. The default value for EPZERO is 10–8. A list of potential singularities
is printed in the grid point singularity table (see “Strain Energy Output” on page 454 for
further details).
PARAM,AUTOSPC. If PARAM,AUTOSPC,YES is specified (this is the default in the

Structured Solution Sequences, except SOLs 106 and 129), the potential singularities are
automatically constrained if possible.
During decomposition, mechanisms can be detected based on the maximum ratio of the
matrix diagonal to the factor diagonal
K ii
MAXRATIO = -------
D ii
where K ii is the i-th diagonal term of the original stiffness matrix and D ii is the i-th
diagonal term of the factor diagonal matrix. For a symmetric matrix K it can be
represented as
T
[K] = [L][D][L ]
where:
[ L ] = lower triangular factor

[ D ] = factor diagonal matrix
PARAM,MAXRATIO. Refer to the NX Nastran Numerical Methods User’s Guide for further
details on the subject of system of linear equations. All terms whose ratio exceed the value
of PARAM,MAXRATIO are printed. The default for MAXRATIO is 107. UIM 4158 prints
the statistics for the decomposition that include the number of negative terms on the factor
diagonal and the maximum ratio of matrix diagonal to factor diagonal at a specified row
CHAPTER 15 475
Model Verification
number and its corresponding grid point. User Warning Message (UWM) 4698 prints the
degrees of freedom that have a factor diagonal ratio greater than the MAXRATIO value
or have negative terms on the factor diagonal. Both of these messages are issued by the
DECOMP module and are shown below:
*** USER INFORMATION MESSAGE 4158---STATISTICS FOR SYMMETRIC DECOMPOSITION OF DATA BLOCK KLL FOLLOW
NUMBER OF NEGATIVE TERMS ON FACTOR DIAGONAL = 1
MAXIMUM RATIO OF MATRIX DIAGONAL TO FACTOR DIAGONAL = 7.2E+15 AT ROW NUMBER 16
*** USER WARNING MESSAGE 4698. STATISTICS FOR DECOMPOSITION OF MATRIX KLL
.
THE FOLLOWING DEGREES OF FREEDOM HAVE FACTOR DIAGONAL RATIOS GREATER THAN 1.00000E+05 OR HAVE
NEGATIVE TERMS ON THE FACTOR DIAGONAL.
GRID POINT ID DEGREE OF FREEDOM MATRIX/FACTOR DIAGONAL RATIO MATRIX DIAGONAL
6714 T1 -7.19297E+15 6.02908E+07
^^^ DMAP FATAL MESSAGE 9050 (SEKRRS) - RUN TERMINATED DUE TO EXCESSIVE PIVOT RATIOS
IN MATRIX KLL. USER PARAMETER BAILOUT MAY BE USED TO CONTINUE THE RUN.
If the MAXRATIO value is exceeded, your job terminates with DMAP Fatal Message 9050.
You can override this fatal message by inserting “PARAM,BAILOUT,–1" in your input
file. You should, however, be aware that a large value of the MATRIX/FACTOR
DIAGONAL may be an indication of a potential modeling problem. Taking the log10 of
MAXRATIO indicates how many significant digits may have been lost during the
decomposition. The MAXRATIO may change slightly if a different sequencer is used.
This change is due to the fact that the order of operation may change with a different
sequencer that yields a different numerical roundoff.
After decomposition, a singularity may lead to an incorrect solution. In static analysis, NX

Nastran solves
Ku = P
to obtain u (displacements). Using these displacements, NX Nastran then calculates a

“residual” loading vector as follows:
Ku – P = δP
This residual vector should theoretically be null but may not be null due to numeric
roundoff. To obtain a normalized value of the residual loading, an error measure ε is
calculated by
T
u ⋅ δP
ε = -------------------
T
u ⋅P
δP can be printed by including PARAM,IRES,1 in your Bulk Data Section. The value ε
(epsilon) is printed in User Information Message (UIM) 5293 as shown below. Epsilon
values that are greater than 0.001 are flagged for a possible loss of accuracy due to numeric
conditioning as shown below:
*** USER INFORMATION MESSAGE 5293 FOR DATA BLOCK KLL
1 -4.5961966E-14 5.6341639E+00
2 -9.2748066E-15 9.4523830E+00
One ε is generated for each loading condition. An acceptable value of ε depends on the
model complexity and the machine that it runs on. An epsilon value of |ε| in the
neighborhood of less than 10-9 is generally considered acceptable.
Some general causes for singularity can include:
• Degrees of freedom without stiffness because of missing elements.

• A 2-D plate problem with the normal rotation unconstrained.
• A solid model with rotational DOFs at the corners unconstrained.
• Incorrect modeling of offset beams.
• Incorrect multipoint constraints.
• Mechanisms and free bodies, such as sloped plates, beam to plate connections,
beam to solid connections, and plate-to-solid connections.
• Low stiffness in rotation.
• A stiff element adjacent to a very flexible element.
Applied Loads Check

There are two checks in NX Nastran to let you verify that loads are being applied correctly
to the structure. The first one is the summation of the total loads applied to the structure
about the reference point x as specified on the PARAM,GRDPNT,x entry. If
PARAM,GRDPNT,x is not specified, this summation is performed about the origin of the
basic coordinate system. This quantity is known as the OLOAD RESULTANT output and
is part of the standard NX Nastran output. It consists of seven lines of output per load
case. The OLOAD RESULTANT output for the problem in Listing 15-5 is shown in
Figure 15-17. Note that loads from scalar points are not included in the OLOAD
RESULTANT output.
CHAPTER 15 477
Model Verification
0 OLOAD RESULTANT
SUBCASE/ LOAD
DAREA ID TYPE T1 T2 T3 R1 R2 R3
0 1 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 0.000000E+00 ---- 0.000000E+00 ---- 0.000000E+00
FZ ---- ---- 1.000000E+02 0.000000E+00 -5.000000E+02 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
TOTALS 0.000000E+00 0.000000E+00 1.000000E+02 0.000000E+00 -5.000000E+02 0.000000E+00
0 2 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 0.000000E+00 ---- 0.000000E+00 ---- 0.000000E+00
FZ ---- ---- 3.307500E+02 0.000000E+00 -8.268750E+02 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
TOTALS 0.000000E+00 0.000000E+00 3.307500E+02 0.000000E+00 -8.268750E+02 0.000000E+00
Figure 15-18 OLOAD Resultant
Look at Subcase 1 with a vertical tip load applied to grid point 6 in the +z direction. Since
there are no loads applied to the structure in the x- or y-direction, the loads in those two
directions are also equal to zero. The total z-load is equal to 100 Newtons contributed
solely by the applied load at grid point 6, which is the only applied load for the whole
structure. The total moment about the origin of the basic coordinate system is
– 5 ⋅ ( 100 N ) = – 500 Nm about the y-axis. This R2 moment is contributed by the Z load
as indicated by the -500 in row FZ, column R2. There is no moment about the other two
axes. The next seven lines in the OLOAD RESULTANT output is due to the gravity load,
the seconding loading condition.
So far, only the resultant applied load was discussed. You can also request more detailed
information regarding the exact location on the structure where loads are applied. This
can be done with the OLOAD request in the Case Control Section. The format is as
follows:
OLOAD = All
or
SET 10 = x,y,z
OLOAD = 10
In this particular case, a printout of all the applied loads is requested. As shown in
Figure 15-19, there is one load applied to grid point 6 in the +z direction for Subcase 1.
However, for the gravity load in Subcase 2, each grid point that has mass associated with
it has a 1g load applied to it. Each CBAR element contributes {(2700)(1.0)(2.5E-3)(9.8)/2}
= 33.075 N to each end of its connecting grid points. Therefore, a 33.075 N load is applied
to the end points and a 66.15 N load is applied to each of the intermediate points since
there are two CBAR elements connected to each of these intermediate grid points. As you
can see, for a large model, OLOAD output for gravity loads can potentially generate many
lines of output.
VERTICAL TIP LOAD SUBCASE 1
L O A D V E C T O R

6 G .0 .0 1.000000E+02 .0 .0 .0
GRAVITY LOADS SUBCASE 2
L O A D V E C T O R

1 G .0 .0 3.307500E+01 .0 .0 .0
2 G .0 .0 6.615000E+01 .0 .0 .0
3 G .0 .0 6.615000E+01 .0 .0 .0
4 G .0 .0 6.615000E+01 .0 .0 .0
5 G .0 .0 6.615000E+01 .0 .0 .0
6 G .0 .0 3.307500E+01 .0 .0 .0
Figure 15-19 OLOAD Output
Verifying Reaction Loads

There are two checks available in NX Nastran to let you check whether the reaction loads
are correct. The first one is the summation of the total reaction loads about the reference
point specified on the PARAM,GRDPNTx entry. If PARAM,GRDPNT,x is not specified,
this summation is performed about the origin of the basic coordinate system. This
operation is known as the SPCFORCE RESULTANT output and this output is obtained if
the SPCFORCE output is requested. It consists of seven lines of output per load case. The
SPCFORCE RESULTANT output for the above problem (Listing 15-5) is shown below:
=0 SPCFORCE RESULTANT
SUBCASE/ LOAD
0 1 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 0.000000E+00 ---- 0.000000E+00 ---- 0.000000E+00
FZ ---- ---- -1.000000E+02 0.000000E+00 0.000000E+00 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 5.000000E+02 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
TOTALS 0.000000E+00 0.000000E+00 -1.000000E+02 0.000000E+00 5.000000E+02 0.000000E+00
0 2 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 0.000000E+00 ---- 0.000000E+00 ---- 0.000000E+00
FZ ---- ---- -3.307500E+02 0.000000E+00 0.000000E+00 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 8.268750E+02 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
TOTALS 0.000000E+00 0.000000E+00 -3.307500E+02 0.000000E+00 8.268750E+02 0.000000E+00
Figure 15-20 SPCFORCE RESULTANT
CHAPTER 15 479
Model Verification
Since the SPCFORCE RESULTANT and OLOAD RESULTANT are summed about the
same reference point, you would expect the total applied loads and total reaction loads to
be equal and opposite of each other. As indicated by Figure 15-18 and Figure 15-20, this
is indeed the case.
Similar to the applied loads, you can also request more detailed information regarding
where exactly on the structure is reacting to these applied loads (grid point locations and
directions). These are also known as boundary loads. This operation can be done with
the SPCFORCE request in the Case Control Section. The format is as follows:
SPCFORCE = All
or
SET 10 = 15, 16, 18, 25

SPCFORCE = 10
VERTICAL TIP LOAD SUBCASE 1

1 G .0 .0 -1.000000E+02 .0 5.000000E+02 .0
GRAVITY LOADS SUBCASE 2

1 G .0 .0 -3.307500E+02 .0 8.268750E+02 .0
Figure 15-21 Single-Point Constraint Forces
In this case, since there is only one reaction point (fixed at grid point 1) in the model, the
SPCFORCE output only contains one line of output per subcase (see Figure 15-21). Some
guidelines to consider when examining the reaction loads include:
1. Are they equal to the applied loads? Is the structure in equilibrium?

2. Are the boundary loads coming out in the directions and magnitudes that you
are expecting? As an example, if one of your reaction loads is to simulate a shear
bolt and you are getting a tension reaction load at that location, the chances are
that you have modeled the boundary condition incorrectly.
3. Is the load distribution over the structure what you expect? If not, you may want
to go back and check your modeling of your reaction points to make sure that
they are what you intended.
4. Grid points with reaction force equal to 0.0 on all six components are suppressed
from the printed output.
1g Load
A 1g gravitation load applied separately in each of the three orthogonal directions is
suggested for model checkout. This check is accomplished with three separate subcases.
This is a basic check to determine grid points on the structure that are either loosely
connected or have very small stiffnesses. A displacement output indicates that these
points are moving excessively as compared to other parts of the model, especially if you
are using a postprocessor. To use this feature, you must ensure that your structure
contains masses-either by direct masses (e.g., CONM2) or indirect masses (e.g., from the
density on the MATi entries), or nonstructural mass (e.g., a nonzero value on the NSM
field on the PBAR entry).
Output of Maximum, Minimum Grid Point-Based Responses

Documenting the validation of a mathematical simulation model can be a very tedious and
time consuming process. A portion of this process usually requires the generation of
reports to completely describe the results of the performed analysis. Although a
substantial amount of the reporting requirements can be satisfied using graphical results
plots, tabular data that summarizes the information depicted in the graphics are often
required. Also, during the validation process itself, many tools are employed that assist
users in evaluating the results produced by the simulation. Maximum/minimum surveys
of output quantities are one such tool provided by the MAXMIN Case Control command.
See Also
• “MAXMIN” on page 284 of the NX Nastran Quick Reference Guide

Historically, NX Nastran users have been able to obtain simple surveys of grid point
related results output such as displacements, applied loads, and forces of single point
constraint. The output consists of a summary that identifies the grid point at which the
maximum and minimum values occurred in each of the six output directions. This
information is provided for each subcase.
The max/min information appears as typical grid point related output. You can request a
number of the highest and lowest values for a particular component. All components of
the grid related quantities are output together with the one component that is being
surveyed. For example, if 10 maximum and 10 minimum values of T3 displacements are
requested, the output will also contain the T1, T2, R1, R2, and R3 displacement values
associated with the grid points identified in the T3 component survey. The output
heading clearly identifies the component being surveyed. Output can be surveyed for
SUBCASEs and SUBCOMs. You can also specify sets of points to be surveyed and
coordinate system to be used for the survey output. Only SORT1 output is available.
If output coordinate systems are specified for any of the grid points being surveyed, the
user has the option of specifying one of several coordinate system views of the max/min
output. The data can be displayed in the basic system, a local system, or even the global
system. EMail:[email protected]
www.cadfamily.com Any grid point that is output in the max/min survey will have its output
CHAPTER 15 481
Model Verification
presented in both the survey coordinate frame of reference and its original frame of
reference if they are different. In this way, the user can see both the relative position of
the grid point component in the max/min output as well as the original output quantities
without having to research other portions of the displacement output produced by the
analysis.
Example: Using the MAXMIN Command

The following sample output demonstrates some features available with the MAXMIN
Case Control command.
Inspection of the Case Control Echo reveals that the MAXMIN command at line 7 requests
print and punch of max/min survey results for all available output quantities generated
(via keywords PRINT, PUNCH and ALL). It also specifies that survey results be limited
to component directions T1 and T2 for grid points defined in case control set 100. Because
the command has been placed above the subcase level, it applies to all subcases and
subcoms.
The MAXMIN command at line 11 requests a survey of the forces of single point
constraint output (keyword SPCF) for component direction T1 only (keyword T1) in the
basic coordinate system frame of reference (keyword CID selects BASIC). It also requests
that the highest and lowest 15 values be output considering ALL grid points. Because it
has been placed within subcase 1000, it only pertains to that subcase.
The MAXMIN command at line 14 requests print and punch of max/min survey results
for single point forces of constraint. A vector magnitude survey is requested (keyword
VMAG) for the four highest and lowest values. The results for these types of surveys
appear in the columns of output normally containing component directions T1 and R1.
Note that the specification of component directions to be surveyed (keyword COMP) has
no effect in this case. All grid points will be included in the survey.
0 C A S E C O N T R O L E C H O
COMMAND
COUNT
1 TITLE = EXAMPLE OUTPUT FROM THE NEW MAXMIN CASE CONTROL COMMAND
2 SUBTITLE = ONLY GRID POINT RELATED OUTPUT QUANTITIES CAN BE PROCESSED
3 LABEL = COMPONENT DOF SPECIFICATIONS ARE POSSIBLE
4 SPC=100
5 SET 100 = 2,51,59
6 DISP=100
7 MAXMIN (PUNCH,PRINT,ALL ,COMP=T1/T2) = 100 $
8 SPCF=ALL
9 OLOAD=ALL
10 SUBCASE 1000
11 MAXMIN (SPCF,CID=BASIC,COMP=T1,BOTH=15) = ALL
12 LOAD=1000
13 SUBCASE 2000
14 MAXMIN (PUNCH,PRINT,SPCF,VMAG=4,COMP=T1/T3) = ALL
15 LOAD=1000
16 SUBCOM 3000
17 SUBSEQ = 0.5,0.5
18 BEGIN BULK
Examination of the following actual MAXMIN survey output results reveals that the three
heading lines contain important information that describe the options used to produce the
results. The first line identifies the component direction that was searched, the type of
output being searched, and the SUBCASE/SUBCOM identification number. The second
line summarizes the options specified on the MAXMIN Case Control command used to
perform the search. The third line contains column headings for the data values that
follow in the output listing. Note that the field, TYPE, identifies whether it is a grid point
(G) or scalar point (S) and the field, POINT ID, identifies the grid point or scalar ID
number. In addition, the coordinate system reference frame for the output values is
identified. If data values for a point were transformed to a reference frame different from
the one originally defined in the output, then the data values for both frames of reference
are output. This is shown in the output for the first grid point (ID = 2) in the sample
output. Here it is seen that values for the forces of single point constraint are provided for
the point in the basic and CID 100 systems. This occurs because the output for the point
has been specified in CID=100, but the max/min survey was requested in the basic
coordinate system. See the MAXMIM Case Control command for further details.
1 EXAMPLE OUTPUT FROM THE NEW MAXMIN CASE CONTROL COMMAND AUGUST 22, 2004 NX NASTRAN 8/21/04 PAGE 19
ONLY GRID POINT RELATED OUTPUT QUANTITIES CAN BE PROCESSED
0 COMPONENT DOF SPECIFICATIONS ARE POSSIBLE
0 *** T1 *** A P P L I E D L O A D M A X / M I N V A L U E S U M M A R Y RESULTS FOR SUBCASE 1000

MAXMIN OPTIONS: SET=100, CID=BASIC, MAX=1, MIN=2, COMP=T1
POINT ID. TYPE CID *** T1 *** T2 T3 R1 R2 R3
2 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
2 G 100 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
51 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
59 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00

POINT ID. TYPE CID T1 *** T2 *** T3 R1 R2 R3
2 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
2 G 100 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
51 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
59 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00

2 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
2 G 100 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
51 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
59 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
CHAPTER 15 483
Model Verification

2 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
2 G 100 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
51 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
59 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
0 *** T1 *** A P P L I E D L O A D M A X / M I N V A L U E S U M M A R Y RESULTS FOR SUBCOM 3000

2 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
2 G 100 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
51 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
59 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
0 *** T2 *** A P P L I E D L O A D M A X / M I N V A L U E S U M M A R Y RESULTS FOR SUBCOM 3000

2 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
2 G 100 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
51 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
59 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
0 *** T1 *** S P C F O R C E M A X / M I N V A L U E S U M M A R Y RESULTS FOR SUBCASE 1000

MAXMIN OPTIONS: SET=ALL, CID=BASIC, MAX=15, MIN=15, COMP=T1
3 G BASIC -1.118073E+03 0.000000E+00 0.000000E+00 5.236134E+01 2.360384E-16 1.280828E+02
3 G 100 0.000000E+00 0.000000E+00 -1.118073E+03 1.280828E+02 -2.360384E-16 5.236134E+01
101 G BASIC -1.000002E+03 -1.733791E-11 -1.208690E-09 0.000000E+00 0.000000E+00 8.325034E-02
251 G BASIC -1.000000E+03 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
201 G BASIC -1.000000E+03 -1.731096E-11 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
151 G BASIC -9.999980E+02 0.000000E+00 -6.541988E-14 0.000000E+00 0.000000E+00 -2.454053E+02
1 G BASIC -4.012542E+02 1.588410E-11 1.125718E-15 -5.499307E+00 -2.439500E-01 7.725024E+01
51 G BASIC -3.474817E+02 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 -4.013108E+01
2 G BASIC -1.331908E+02 0.000000E+00 0.000000E+00 -4.686203E+01 2.439500E-01 2.724495E+02
2 G 100 0.000000E+00 0.000000E+00 -1.331908E+02 2.724495E+02 -2.439500E-01 -4.686203E+01
209 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
208 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
207 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
206 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
205 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
204 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
203 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
55 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 -2.368870E-02
56 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 -7.088747E-04
57 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 -4.525901E-04
58 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 -1.690151E-04
59 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 1.056287E-04
60 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 8.333347E+01
6 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 1.338836E-03
102 G BASIC 0.000000E+00 0.000000E+00 1.209136E-09 0.000000E+00 0.000000E+00 2.414764E+02
103 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 -3.667573E+00
104 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 -1.456001E+00
105 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 -8.364819E+01
5 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 1.390992E-01
152 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 6.736528E+00
153 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 1.944593E+00
700 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
.
. .
78 S 0.000000E+00
88 S 0.000000E+00
0 *** T1 *** S P C F O R C E M A X / M I N V A L U E S U M M A R Y RESULTS FOR SUBCASE 2000

MAXMIN OPTIONS: SET=ALL, CID=BASIC, VMAG=4, VMAG=4, COMP=T1
POINT ID. TYPE CID ***TMAG*** T2 T3 R1 R2 R3
700 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
257 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
202 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
203 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
251 G BASIC 1.000000E+03 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
201 G BASIC 1.000000E+03 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
101 G BASIC 1.000002E+03 0.000000E+00 0.000000E+00 8.325034E-02 0.000000E+00 0.000000E+00
3 G BASIC 1.118073E+03 0.000000E+00 0.000000E+00 1.383724E+02 0.000000E+00 0.000000E+00
3 G 100 0.000000E+00 0.000000E+00 -1.118073E+03 1.280828E+02 -2.360384E-16 5.236134E+01
78 S 0.000000E+00
88 S 0.000000E+00
0 *** R1 *** S P C F O R C E M A X / M I N V A L U E S U M M A R Y RESULTS FOR SUBCASE 2000

MAXMIN OPTIONS: SET=ALL, CID=BASIC, VMAG=4, VMAG=4, COMP=R1
POINT ID. TYPE CID T1 T2 T3 ***RMAG*** R2 R3
700 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
256 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
210 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
209 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
52 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 2.110957E+02 0.000000E+00 0.000000E+00
102 G BASIC 1.209136E-09 0.000000E+00 0.000000E+00 2.414764E+02 0.000000E+00 0.000000E+00
151 G BASIC 9.999980E+02 0.000000E+00 0.000000E+00 2.454053E+02 0.000000E+00 0.000000E+00
2 G BASIC 1.331908E+02 0.000000E+00 0.000000E+00 2.764504E+02 0.000000E+00 0.000000E+00
2 G 100 0.000000E+00 0.000000E+00 -1.331908E+02 2.724495E+02 -2.439500E-01 -4.686203E+01
0 *** T1 *** S P C F O R C E M A X / M I N V A L U E S U M M A R Y RESULTS FOR SUBCOM 3000

51 G BASIC -3.474817E+02 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 -4.013108E+01
2 G BASIC -1.331908E+02 0.000000E+00 0.000000E+00 -4.686203E+01 2.439500E-01 2.724495E+02
2 G 100 0.000000E+00 0.000000E+00 -1.331908E+02 2.724495E+02 -2.439500E-01 -4.686203E+01
59 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 1.056287E-04
0 *** T2 *** S P C F O R C E M A X / M I N V A L U E S U M M A R Y RESULTS FOR SUBCOM 3000

51 G BASIC -3.474817E+02 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 -4.013108E+01
2 G BASIC -1.331908E+02 0.000000E+00 0.000000E+00 -4.686203E+01 2.439500E-01 2.724495E+02
2 G 100 0.000000E+00 0.000000E+00 -1.331908E+02 2.724495E+02 -2.439500E-01 -4.686203E+01
59 G BASIC 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 1.056287E-04
0 *** T1 *** D I S P L A C E M E N T M A X / M I N V A L U E S U M M A R Y RESULTS FOR SUBCASE 1000

2 G BASIC 0.000000E+00 -3.568517E-05 -2.158988E-07 0.000000E+00 0.000000E+00 0.000000E+00
2 G 100 -2.158988E-07 3.568517E-05 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
51 G BASIC 0.000000E+00 1.852034E-04 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
59 G BASIC 1.939060E-02 -2.595003E-02 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
CHAPTER 15 485
Model Verification

59 G BASIC 1.939060E-02 -2.595003E-02 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
2 G BASIC 0.000000E+00 -3.568517E-05 -2.158988E-07 0.000000E+00 0.000000E+00 0.000000E+00
2 G 100 -2.158988E-07 3.568517E-05 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
51 G BASIC 0.000000E+00 1.852034E-04 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00

2 G BASIC 0.000000E+00 -3.568517E-05 -2.158988E-07 0.000000E+00 0.000000E+00 0.000000E+00
2 G 100 -2.158988E-07 3.568517E-05 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
51 G BASIC 0.000000E+00 1.852034E-04 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
59 G BASIC 1.939060E-02 -2.595003E-02 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00

59 G BASIC 1.939060E-02 -2.595003E-02 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
2 G BASIC 0.000000E+00 -3.568517E-05 -2.158988E-07 0.000000E+00 0.000000E+00 0.000000E+00
2 G 100 -2.158988E-07 3.568517E-05 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
51 G BASIC 0.000000E+00 1.852034E-04 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
0 *** T1 *** D I S P L A C E M E N T M A X / M I N V A L U E S U M M A R Y RESULTS FOR SUBCOM 3000

2 G BASIC 0.000000E+00 -3.568517E-05 -2.158988E-07 0.000000E+00 0.000000E+00 0.000000E+00
2 G 100 -2.158988E-07 3.568517E-05 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
51 G BASIC 0.000000E+00 1.852034E-04 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
59 G BASIC 1.939060E-02 -2.595003E-02 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
0 *** T2 *** D I S P L A C E M E N T M A X / M I N V A L U E S U M M A R Y RESULTS FOR SUBCOM 3000

59 G BASIC 1.939060E-02 -2.595003E-02 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
2 G BASIC 0.000000E+00 -3.568517E-05 -2.158988E-07 0.000000E+00 0.000000E+00 0.000000E+00
2 G 100 -2.158988E-07 3.568517E-05 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
51 G BASIC 0.000000E+00 1.852034E-04 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
Performing an Unconstrained Equilibrium Check

If a structure is truly unconstrained, then movement at one point of this structure should
cause the whole structure to move as a rigid body. The following procedure can be used
to perform this check.
1. Remove all the constraints.
2. Apply a unit enforced displacement in the x-direction at one selected grid point
while constraining the other five components at this selected grid point to zero.
This grid point should be close to the center of gravity (C.G.) of the structure,
although this is not a requirement. The x-displacements at all the other grid
points should also be equal to unity if the structure is truly unconstrained. If the
displacement at a certain grid point is not equal to unity in the x-direction, it is
very likely that it overstrained. The likely causes for this can include incorrect
modeling of rigid elements, offset beams, non-colinear CELASi, etc. See
“Constraints” on page 209 for details regarding enforced displacement.
3. Repeat the same procedure for the other two orthogonal directions. You can
perform similar checks for rotations; however, it is more difficult to interpret the
results.
At the conclusion of this check, you should remember to put your original constraints back
into your model.
Automatic Identification of Unintentional Grounding

In addition to the unconstrained equilibrium check mentioned above, you may also
request a more robust and more automatic grounding check of the stiffness matrix. This
grounding check is also referred to as rigid body check because the stiffness matrix is
multiplied by the rigid body transformation matrix. The grounding check identifies
unintentional constraints and ill-conditioning in the stiffness matrix.
The grounding check is requested with the GROUNDCHECK Case Control command.
 PRINT  
 , PUNCH, SET =   G, N, N + AUTOSPC, F, A 
 NOPRINT  ALL   
GROUNDCHECK =  YES 
  NO 
GRID = gid, THRESH = e,DATAREC = YES , ( RTHRESH = r )
NO 
For example, the following command
GROUNDCHECK=YES
will request a grounding check of the g-set. The following command
GROUNDCHECK(SET=ALL)=YES
will request a grounding check of all DOF sets.
CHAPTER 15 487
Model Verification
The check may be performed at any or all stages of the stiffness reduction based on the
specification of the SET keyword
SET
DOF Set Description
Keyword
G g-set before single point, multipoint constraints, and rigid
elements are applied
N n-set after multipoint constraints and rigid elements are
applied
N+AUTO n-set with same as the n-set with the rows/columns in the
AUTOSPC stiffness matrix corresponding to degrees-of-freedom
constrained by the PARAM,AUTOSPC operation
zeroed out
F f-set after single point, multipoint constraints, and rigid
elements are applied
A a-set after static condensation
The THRESH keyword specifies the maximum strain energy that passes the grounding
check. DATAREC=YES requests data recovery of grounding forces. The RTHRESH=r
keyword prints the grounding forces larger than r percent of the largest grounding force
if DATAREC is set to YES.
User Information Message 7570 is issued by GROUNDCHECK for each DOF set
requested. The strain energy is computed in each direction of the rigid body motion and
by default, if the strain energy exceeds the tolerance, then “FAIL” is printed out for that
directory. The tolerance is set by dividing the largest stiffness term by 1.E10. The THRESH
keyword can be used to further control the amount of output (see Appendix A). Possible
reasons for failure are also printed after the strain energies. Here is an example for the g-
set that indicates the stiffness matrix passes the grounding check in all six rigid body
directions:
RESULTS OF RIGID BODY CHECKS OF MATRIX KGG (G-SET) FOLLOW:

PRINT RESULTS IN ALL SIX DIRECTIONS AGAINST THE LIMIT OF 2.725275E-04
DIRECTION STRAIN ENERGY PASS/FAIL
--------- ------------- ---------
1 1.484295E-09 PASS
2 2.182787E-10 PASS
3 1.637090E-11 PASS
4 1.619810E-10 PASS
5 2.727802E-10 PASS
6 1.054841E-07 PASS
SOME POSSIBLE REASONS MAY LEAD TO THE FAILURE:
1. CELASI ELEMENTS CONNECTING TO ONLY ONE GRID POINT;
2. CELASI ELEMENTS CONNECTING TO NON-COINCIDENT POINTS;
3. CELASI ELEMENTS CONNECTING TO NON-COLINEAR DOF;
4. IMPROPERLY DEFINED DMIG MATRICES;
Here is an example for the f-set that indicates the stiffness matrix fails the grounding check
in all six rigid body directions:
*** USER INFORMATION MESSAGE 7570 (GPWG1D)

RESULTS OF RIGID BODY CHECKS OF MATRIX KFF (F-SET) FOLLOW:
PRINT RESULTS IN ALL SIX DIRECTIONS AGAINST THE LIMIT OF 1.464858E-04
DIRECTION STRAIN ENERGY PASS/FAIL
--------- ------------- ---------
1 2.564102E+05 FAIL
2 7.326008E+05 FAIL
3 2.237437E+03 FAIL
4 6.057062E+02 FAIL
5 4.015165E+03 FAIL
6 8.747863E+04 FAIL
SOME POSSIBLE REASONS MAY LEAD TO THE FAILURE:
1. CONSTRAINTS WHICH PREVENT RIGID-BODY MOTION.
If the DATAREC keyword is specified and the rigid body check fails then data recovery
will be performed to compute and print the grounding forces to make it easy to locate the
source of the failure. By default, only those grounding forces larger than 10 percent of the
largest grounding force will be printed (see RTHRESH keyword). Here is an example of
grounding forces created by moving the model in rigid body direction 1 associated with
the f-set failure shown above:
DIRECTION 1
G R O U N D C H E C K F O R C E S ( F - S E T )

3 G 2.564102E+05 -2.884615E+05 .0 .0 .0 .0
4 G 2.564102E+05 2.884615E+05 .0 .0 .0 .0
Thermal Equilibrium Check

The following procedure can also be used to check for unconstrained thermal expansion if
the structure is going to be subjected to thermal loads.
1. Remove the actual boundary conditions and apply a set of statically determinate
constraints. Typically this procedure is done by constraining all six DOFs at a
single grid point. If a single grid point is used, make sure this single grid point
contains six degrees of freedom. A single grid point of a model consisting of all
solid elements, for example, cannot satisfy this requirement since each grid point
of a solid element contains only three degrees of freedom.
2. Change all the thermal coefficients of expansion to a single value.
3. Apply a uniform ∆T to the structure.
If the model is “clean,” then the structure should be strain free; in other words, there
should be no reaction loads, element forces, or stresses. If this is not the case, then you may
want to investigate around the vicinity where the element forces or stresses are nonzero.
Incorrect modeling of rigid elements or offsets is a common cause of these types of errors.
Once you are satisfied with your model, remember to change the boundary condition,
thermal coefficients of expansion, and ∆T back to their original values.
CHAPTER 15 489
Model Verification
Grid Point Force Balance

Frequently, there are several elements connected to a single grid point. When this
happens, you may want to know how the loads are distributed among these elements at
the common grid point. In other words, you want to understand the load path. Is the load
path correct? These questions can be answered by using the GPFORCE (grid point force)
option in NX Nastran.
Grid point force balance is computed only from linear stiffness elements, the sum of
applied loads and thermal loads, and SPC forces. Effects not accounted for include those
from mass elements in dynamic analysis (inertia loads), rigid elements and MPCs, general
elements, DMIG entries, and boundary loads from upstream superelements. These
effects may lead to an apparent lack of equilibrium at the grid point level. Table 15-2
summarizes those effects that are considered and those effects that are ignored in the
calculation of grid point forces in the global coordinate system.
Table 15-2 Contributions Used for the Grid Point Force Balance
Contributions Included Contributions Ignored

Applied Loads Boundary Loads from Upstream Superelements
SPC Forces GENEL Forces
Element Elastic Forces DMIG and DMI Forces
Thermal Loads
The model in Figure 15-22 illustrates this feature. This model consists of nine CQUAD4
and six CBAR elements and is subjected to vertical loads applied at the free end. The
input file is shown in Listing 15-6.
EXAMPLE
50 lb
13 14 15 16
10 in 7 8 9
9 10 11 12
100 lb VIEW A-A
104 104 105
10 in 4 5 6
5 6 7 8
100 lb
101 102 103
10 in 1 2 3
1 2 3 4
50 lb
y
10 in 10 in 10 in
x
2
0.5
A = 0.50 in2
I1 = 0.0417 in4
1 1 1.0
I2 = 0.01042 in4
J = 0.0286 in4
2 t = 0.10 in
Section A-A
Figure 15-22 Clamped Plate with Stiffeners
CHAPTER 15 491
Model Verification
Listing 15-6 Input File for the Demonstration of

Grid Point Force Balance
$ FILENAME - q4bargpf.dat
$
SOL 101
TIME 5
CEND
TITLE = ILLUSTRATE USE OF GPFORCE
SUBTITLE = INCORRECT PROPERTIES
DISP = ALL
GPFORCE = ALL
STRESS(CORNER) = ALL
SPC = 1
LOAD = 1
PARAM AUTOSPC YES
PARAM POST 0
$
$ THIS SECTION CONTAINS BULK DATA
$
GRID 1 0.0 0.0 0.0
GRID 2 10. 0.0 0.0
GRID 3 20. 0.0 0.0
GRID 4 30. 0.0 0.0
GRID 5 0.0 10. 0.0
GRID 6 10. 10. 0.0
GRID 7 20. 10. 0.0
GRID 8 30. 10. 0.0
GRID 9 0.0 20. 0.0
GRID 10 10. 20. 0.0
GRID 11 20. 20. 0.0
GRID 12 30. 20. 0.0
GRID 13 0.0 30. 0.0
GRID 14 10. 30. 0.0
GRID 15 20. 30. 0.0
GRID 16 30. 30. 0.0
$
CBAR 101 100 5 6 0.0 0.0 1.
+C101 0. 0. .55 0. 0. .55
CBAR 102 100 6 7 0.0 0.0 1.
+C102 0. 0. .55 0. 0. .55
CBAR 103 100 7 8 0.0 0.0
+C103 0. 0. .55 0. 0. .55
CBAR 104 200 9 10 0.0 0.0
+C104 0. 0. .55 0. 0. .55
CBAR 105 200 10 11 0.0 0.0
+C105 0. 0. .55 0. 0. .55
CBAR 106 200 11 12 0.0 0.0
+C106 0. 0. .55 0. 0. .55
$
CQUAD4 1 1 1 2 6 5
CQUAD4 2 1 2 3 7 6
CQUAD4 3 1 3 4 8 7
CQUAD4 4 1 5 6 10 9
CQUAD4 5 1 6 7 11 10
CQUAD4 6 1 7 8 12 11
CQUAD4 7 1 9 10 14 13
CQUAD4 8 1 10 11 15 14
CQUAD4 9 1 11 12 16 15
$
Listing 15-6 Input File for the Demonstration of

Grid Point Force Balance (continued)
$ THIS SECTION CONTAINS THE LOADS and CONSTRAINTS

$
$
FORCE 1 4 0 50. 0. 0. -1.
FORCE 1 8 0 100. 0. 0. -1.
FORCE 1 12 0 100. 0. 0. -1.
FORCE 1 16 0 50. 0. 0. -1.
$
SPC1 1 123456 1 5 9 13
$
$ THIS SECTION CONTAINS THE PROPERTY AND MATERIAL BULK DATA ENTRIES
$
PSHELL 1 1 .1 1
$
PBAR 100 1 .50 .0417 .01042 .0286
+P100 .5 .25 -.5 .25 -.5 -.25 .5 -.25
PBAR 200 1 .01042 .0286
+P200 .5 .25 -.5 .25 -.5 -.25 .5 -.25
$
$ CORRECT PBAR ENTRY WITH A AND I1
$
$PBAR 200 1 .50 .0417 .01042 .0286
$
$
MAT1 1 1.+7 .3
ENDDATA
If everything is done correctly, expect the stress contour plot to look something like
Figure 15-23 (a). However, say that you made a modeling mistake somewhere, and the
stress contour plot is coming out as shown in Figure 15-23(b). You know that something
is wrong because the stress contour plot is not symmetrical. Since the structure, boundary
conditions, and applied loads are all symmetrical, you also expect the stress contour plot
to be symmetrical. By inspecting the grid point force balance output (Figure 15-24), you
notice that the stiffeners (CBAR 104, 105, and 106) are not picking up any axial, vertical, or
out-of-plane bending loads at grid points 9, 10, 11, and 12.
A review of the input file explains the reason for this result. The area (A) and the
out-of-plane bending moment of inertia (I1) were "accidentally" left out for CBAR
elements 104, 105, and 106 (PBAR 200). Hence, all the axial and out-of-plane bending loads
at grid points 9, 10, 11, and 12 were picked up completely by the surrounding CQUAD4
elements. By correcting the PBAR element properties, note that CBAR elements 104, 105,
and 106 now pick up the correct loads as shown in Figure 15-25.
CHAPTER 15 493
Model Verification
Figure 15-23 Stress Contour Plots of a Cantilever Plate
G R I D P O I N T F O R C E B A L A N C E
POINT-ID ELEMENT-ID SOURCE T1 T2 T3 R1 R2 R3
1 F-OF-SPC 2.974426E+03 9.828222E+02 1.589665E+01 1.144652E+02 1.100169E+01 .0
1 1 QUAD4 -2.974426E+03 -9.828222E+02 -1.589665E+01 -1.144652E+02 -1.100169E+01 .0
1 *TOTALS* 4.547474E-13 .0 -8.881784E-15 -4.263256E-14 -2.486900E-14 .0
9 F-OF-SPC 2.149695E+03 -7.652444E+02 4.273880E+01 5.070187E+02 -2.392951E+02 -1.157442E+02
9 104 BAR .0 1.266284E+01 .0 -7.117958E+02 .0 1.157442E+02
9 4 QUAD4 -1.788234E+03 6.868671E+02 -5.524031E+01 1.670983E+02 1.526720E+02 .0
9 7 QUAD4 -3.614603E+02 6.571449E+01 1.250151E+01 3.767887E+01 8.662309E+01 .0
9 *TOTALS* 1.136868E-13 1.705303E-13 -1.136868E-13 -1.350031E-13 4.263256E-13 .0
10 104 BAR .0 -1.266284E+01 .0 7.117958E+02 .0 1.088424E+01
10 105 BAR .0 2.289362E-01 .0 -9.452994E+02 .0 -1.088424E+01
10 4 QUAD4 1.194813E+03 -1.018952E+02 6.047494E+01 -1.471362E+02 4.924568E+01 .0
10 5 QUAD4 -1.501499E+03 3.827450E+02 -6.910983E+01 3.978395E+02 -4.706585E+01 .0
10 7 QUAD4 7.029081E+02 -2.641614E+02 -1.464156E+01 -7.215542E+01 -5.501759E+01 .0
10 8 QUAD4 -3.962218E+02 -4.254534E+00 2.327645E+01 5.495581E+01 5.283775E+01 .0
10 *TOTALS* 8.526513E-13 -8.970602E-14 4.476419E-13 -1.627143E-12 -1.946887E-12 -5.329071E-15
11 105 BAR .0 -2.289362E-01 .0 9.452994E+02 .0 1.317360E+01
11 106 BAR .0 -1.317360E+00 .0 -7.749208E+02 .0 -1.317360E+01
11 5 QUAD4 5.103351E+02 -5.226174E+02 9.304677E+01 -4.448763E+02 -8.715293E+01 .0
11 6 QUAD4 -8.479946E+02 6.063411E+02 -8.722735E+01 4.217841E+02 3.954052E+01 .0
11 8 QUAD4 4.692538E+02 -8.115437E+01 -3.112581E+01 -1.103274E+02 2.823395E-01 .0
11 9 QUAD4 -1.315942E+02 -1.023029E+00 2.530639E+01 -3.695915E+01 4.733007E+01 .0
11 *TOTALS* -3.439027E-12 2.285061E-12 -2.501110E-12 5.400125E-12 -7.815970E-13 6.572520E-14
12 APP-LOAD .0 .0 -1.000000E+02 .0 .0 .0
12 106 BAR .0 1.317360E+00 .0 7.749208E+02 .0 -4.178994E-14
12 6 QUAD4 -1.224487E+02 -1.046283E+01 1.404051E+02 -5.812056E+02 3.421728E+00 .0
12 9 QUAD4 1.224487E+02 9.145473E+00 -4.040514E+01 -1.937151E+02 -3.421728E+00 .0
12 *TOTALS* 7.389644E-13 -6.270540E-13 -1.371347E-12 1.023182E-12 -4.396483E-12 -4.178994E-14
16 APP-LOAD .0 .0 -5.000000E+01 .0 .0 .0
16 9 QUAD4 -4.145872E-14 -8.016339E-13 5.000000E+01 -3.638509E-12 2.948251E-12 .0
16 *TOTALS* -4.145872E-14 -8.016339E-13 1.691092E-12 -3.638509E-12 2.948251E-12 .0
Figure 15-24 Grid Point Force Balance Output for a Model with
Incorrect CBAR Properties
G R I D P O I N T F O R C E B A L A N C E
POINT-ID ELEMENT-ID SOURCE T1 T2 T3 R1 R2 R3

1 F-OF-SPC 1.695155E+03 6.360602E+02 1.951741E+01 8.563783E+01 -2.406043E+01 .0
1 1 QUAD4 -1.695155E+03 -6.360602E+02 -1.951741E+01 -8.563783E+01 2.406043E+01 .0
1 *TOTALS* .0 .0 3.552714E-15 1.421085E-14 -3.552714E-15 .0
9 F-OF-SPC -1.695155E+03 -4.214514E+02 1.304826E+02 1.775948E+02 -4.475939E+03 -4.778485E+01

9 104 BAR 4.887901E+03 5.637726E+00 -1.451854E+02 -2.707557E+02 4.427706E+03 4.778485E+01
9 4 QUAD4 -1.885757E+03 3.980623E+02 -1.594168E+00 1.076575E+01 2.244092E+01 .0
9 7 QUAD4 -1.306990E+03 1.775138E+01 1.629696E+01 8.239518E+01 2.579244E+01 .0
9 *TOTALS* -6.821210E-13 7.815970E-14 -1.101341E-13 5.684342E-14 -2.984279E-13 .0
10 104 BAR -4.887901E+03 -5.637726E+00 1.451854E+02 2.707557E+02 -2.975852E+03 8.592416E+00
10 105 BAR 2.834930E+03 -2.358537E-02 -1.454878E+02 -2.992984E+02 2.890022E+03 -8.592416E+00
10 4 QUAD4 1.885757E+03 2.042302E+02 1.594168E+00 1.819319E+01 -6.499233E+00 .0
10 5 QUAD4 -1.123152E+03 -1.147349E+02 -1.597489E+00 3.257575E+01 2.503494E+01 .0
10 7 QUAD4 1.960801E+03 -4.728610E+02 -1.916611E+01 -9.600525E+01 -2.932959E+00 .0
10 8 QUAD4 -6.704347E+02 3.890270E+02 1.947184E+01 7.377895E+01 7.022765E+01 .0
10 *TOTALS* -1.023182E-12 3.410605E-13 -3.623768E-13 -2.842171E-14 -4.263256E-13 1.598721E-14
11 105 BAR -2.834930E+03 2.358537E-02 1.454878E+02 2.992984E+02 -1.435144E+03 8.356563E+00
11 106 BAR 8.836555E+02 -8.356562E-01 -1.431583E+02 -1.784329E+02 1.377845E+03 -8.356563E+00
11 5 QUAD4 1.123152E+03 -2.516147E+01 1.597489E+00 4.088165E+01 -9.060055E+00 .0
11 6 QUAD4 -3.665653E+02 1.245855E+02 -1.571911E+00 5.274417E+01 2.618646E+01 .0
11 8 QUAD4 1.240412E+03 -2.789030E+02 -2.937232E+01 -1.661472E+02 1.650724E+01 .0
11 9 QUAD4 -4.572457E+01 1.802911E+02 2.701724E+01 -4.834404E+01 2.366565E+01 .0
11 *TOTALS* -6.849632E-12 -3.126388E-13 -1.712408E-12 2.771117E-13 3.197442E-12 5.151435E-14
12 APP-LOAD .0 .0 -1.000000E+02 .0 .0 .0
12 106 BAR -8.836555E+02 8.356562E-01 1.431583E+02 1.784329E+02 5.373812E+01 -3.213094E-14
12 6 QUAD4 3.665653E+02 4.705300E+02 1.571911E+00 5.770844E+01 -1.046736E+01 .0
12 9 QUAD4 5.170903E+02 -4.713657E+02 -4.473020E+01 -2.361413E+02 -4.327076E+01 .0
12 *TOTALS* 3.069545E-12 -8.526513E-13 -4.902745E-13 -8.810730E-13 1.975309E-12 -3.213094E-14
16 APP-LOAD .0 .0 -5.000000E+01 .0 .0 .0
16 9 QUAD4 2.966625E-13 3.832166E-14 5.000000E+01 -1.397169E-12 1.610404E-12 .0
16 *TOTALS* 2.966625E-13 3.832166E-14 8.171241E-13 -1.397169E-12 1.610404E-12 .0
Figure 15-25 Grid Point Force Balance Output for a Model with
Correct CBAR Properties
Both Figure 15-24 and Figure 15-25 are reduced versions of the full output. Only the grid
point force balance output as related to grid points 1, 9, 10, 11, 12, and 16 are shown in
these two figures for clarity. Notice that for grid points 1 and 9, in addition to the elastic
element forces, the SPC forces are also printed. Similarly, for grid points 12 and 16, the
applied loads are also printed. As you can see, the grid point force balance feature can
provide you with a tremendous insight regarding the load path of your structure.
15.5 Stress Error Estimators

“Reviewing Grid Point Stresses” on page 432 is devoted to a discussion of the averaging
procedures utilized to provide meaningful stresses at the grid points of finite element
models in NX Nastran. However, when you look at a stress distribution, frequently you
may not be sure of whether further model refinement is necessary. This section focuses
upon the description of error estimators that help you identify regions of your model that
may require refinement.
The essence of the averaging procedures used to obtain grid point stress data are to:
CHAPTER 15 495
Model Verification
1. Convert these local element stress components into a common coordinate

system defined by you.
2. Average several values of each stress component to obtain a unique value of the
stress component that is to be associated with the grid point in question.
3. Compute the stress invariants at the grid points from the stress components at
the grid points.
In the general case, the stress components are σx , σy , σz , τ xy , τxz , and τ yz .
For discussion purposes, the averaging process used to compute the stress components at
the grid points can be represented in the form
N
e
σg = ∑ ( W i σ ei ) Eq. 15-3
i = 1
where:
σ g = the weighted mean value of the stress component computed at the grid point.
σ ei = the value of the stress component in the i-th element i = 1, 2, …, N e

connected to the grid point. σ ei is in the same coordinate system as σ g .
W i = the weighting factor assigned to the i-th element. The sum of the values of
must equal 1. This requirement assures that all computed statistics are
unbiased. The attribute of being unbiased implies that the variance is equal to
the mean square error. Equal weighting, i.e., W i = 1 ⁄ N e , is assumed in NX
Nastran.
An estimate of the error in a particular component of stress at a grid point can then be
computed by assuming that the values of the corresponding stress components computed
by NX Nastran for the elements in the neighborhood of the grid point are data points with
uncorrelated random errors. It then follows that an estimate of the probable error in the
stress component δg at the grid point is
N
e
2
N
e ∑ ( δ ei )
2 1 i = 1
δg = ∑ ( W i δ ei ) = ----------- ---------------------------
Ne Ne
- Eq. 15-4
i = 1
where δei = σei – σ g . Thus, the probable error δ g is the root mean square error in δei
divided by N e .
It should be noted that the root mean square error is a reasonable measure of precision in
many practical cases, but it is easy to provide examples in which this value is a poor
measure of the concentration of the distribution about the mean. This subject is discussed
further in the section "Discussion of Error Measures." Eq. 15-4 is assumed to provide an
approximate error estimator for the grid point stress data output by NX Nastran.
This stress error assessment is often referred to as stress discontinuity, and it is evaluated
in two different ways in NX Nastran: as grid point and element stress discontinuities.
Grid Point Stress Discontinuities

To illustrate the computations described in the previous section, consider the following
portion of a finite element model consisting of CQUAD4 elements. For clarity, only a
portion of the model is plotted. These elements lie on a surface (defined as surface 91 in
this case).
102 103
Y = 3.0
59 60 61
Y = 1.5 61 62
19 20 21
Y = 0.0
X = 27 X = 28.5 X = 30 X = 31.5
Figure 15-26 Partial Model for the Stress Discontinuity Calculations
The following data is extracted from the NX Nastran output. Table 15-3 and Table 15-4
contain partial listings of the element stresses and grid point stresses for the above model.
The grid point stresses are obtained following the procedures described in “Reviewing
Grid Point Stresses” on page 432.
CHAPTER 15 497
Model Verification
Table 15-3 Element Stresses for CQUAD4
Stresses in the Element Coordinate System

Element ID NORMAL-X NORMAL-Y SHEAR-XY
19 –4.006485E+03 –8.949428E-09 –1.045506E+02
20 –3.820136E+03 2.255547E-09 –1.045506E+02
.
.
.
59 –1.335495E+03 8.658390E-09 –2.287827E+02
60 –1.273379E+03 –1.818989E-09 –2.287827E+02
Table 15-4 Grid Point Stresses for Surface 91
Stresses in the Element System

Grid ID NORMAL-X NORMAL-Y SHEAR-XY
20 –5.218E+03 –6.730E–09 –4.243E+01
21 –4.969E+03 1.692E–09 –4.243E+01
.
.
.
60 –2.733E+03 –3.274E–10 –1.667E+02
61 –2.609E+03 –3.638E–11 –1.667E+02
62 –2.485E+03 1.273E–10 –1.667E+02
63 –2.360E+03 –5.457E–11 –1.667E+02
.
.
.
102 –6.104E–05 6.803E-09 –2.909E+02
103 –9.155E–05 –1.437E–09 –2.909E+02
The probable error for stresses at both grid points 20 and 61 can be computed using
Eq. 15-4. The probable error δg is expected to provide some measure of the error in the
magnitude of the grid point stress values. You must first compute the values δei = σei – σ g
for each stress component and then compute the probable error, δg . This calculation is
done for both grid points 20 and 61. The symbols “GID” and “EID”, which represent grid
point ID and element ID, respectively, are used extensively throughout this section.
Consider an example calculation using the NORMAL-X stress component.
NORMAL-X Stress
GID = 20
EID σ ei σg δ ei
19 –4.006485E+03 –5.218E+03 1.212E+03

20 –3.820136E+03 –5.218E+03 1.398E+03
Using Eq. 15-4,
2 2
1 ( 1212 ) + ( 1398 )
δ g = ------- ------------------------------------------------ = 924.9
2 2
GID = 61
EID σ ei σg δ ei
19 –4.006485E+03 –2.609E+03 -1.397E+03

20 –3.820136E+03 –2.609E+03 -1.211E+03
59 –1.335495E+03 –2.609E+03 1.274E+03
60 –1.273379E+03 –2.609E+03 1.336E+03
Again, using Eq. 15-4,
2 2 2 2
1 ( – 1397 ) + ( – 1211 ) + ( 1274 ) + ( 1336 )
δ g = ------- -------------------------------------------------------------------------------------------------------------- = 653.1
4 4
The other stress components ( σy and τxy ) can then be obtained in a similar manner.
This same finite element model is then remodeled using different element types—
CQUAD8, CHEXA(8) as well as CQUAD4 elements with the corner output option. The
values of δg for each of the three stress components are then calculated at two locations:
x = 28.5 and y = 0.0, as well as x = 28.5 and y = 1.5. Note that these two locations
correspond to GID 20 and GID 61, respectively, for the above CQUAD4 model. This
process is then repeated for each of the above element types. The results ( δ g and ERROR
CHAPTER 15 499
Model Verification
ESTIMATE) from NX Nastran are summarized in Table 15-5. Note that these values may
be slightly different than the hand-calculated values since the computer provides more
precision than the hand calculation.
The probable error δg provides an estimated error for each of the three stress components
at each of the two grid point locations considered. It is probably more useful and more
desirable to combine these three estimated errors into a single representative error
measure at each grid point. An approximate root mean square value of the three
estimated errors for each of the three stress components offers such a representative error
measure. That is,
N
c
2
∑ ( δ gi )
i = 1
ERROR ESTIMATE = ----------------------------
Nc
Eq. 15-5
where N c , which is the number of stress components, is equal to three for plate elements
and six for solid elements.
Table 15-5 The Values of δ g and ERROR ESTIMATE for Different Element Types
(X = 28.5, Y = 0.0)
X = 28.5 Y = 1.5
NORMAL-X NORMAL-Y SHEAR-XY ERROR
( δg ) ( δg ) ( δg ) ESTIMATE
Element Type
CQUAD4 9.247E+02 4.628E–09 4.392E+01 5.345E+02
CQUAD8 0.000E+00 1.204E–09 0.000E+00 9.840E–10
CHEXA (8) 8.839E+01 2.841E+01 0.000E+00 5.729E+01
CQUAD4 (CORNER) 9.653E+01 2.896E+01 3.548E+02 2.129E+02
(X = 28.5, Y = 1.5)
X = 28.5 Y = 1.5
( δg ) ( δg ) ( δg ) ESTIMATE
Element Type
CQUAD4 6.531E+02 3.196E-09 3.106E+01 3.775E+02
CQUAD8 0.000E+00 3.634E-09 0.000E+00 2.098E-09
(X = 28.5, Y = 1.5)
CHEXA (8) 3.125E+01 1.497E+01 3.125E+01 2.148E+01

The following remarks should be noted in regard to Table 15-5 and Table 15-6.
1. By default, the stress components for the CQUAD4 elements are only output at
the element centroids, while the stress components for the CQUAD8, and
CHEXA(8) elements are output at the element centroids as well as the element
vertices.
2. If vertex stresses are also desired for the CQUAD4 elements, then use the
STRESS(CORNER) = x option in the Case Control Section. This output
corresponds to the last rows of Table 15-5 and Table 15-6.
3. The calculation of δg for the CQUAD8 and CHEXA(8) elements involves grid
point and element vertex stresses that are defined at the same points.
4. The calculation of δg for the CQUAD4 elements involve grid point stresses and
element stresses provided at a different geometric location if the default center
option is requested. They are evaluated at the same location when using the
corner option.
5. The ERROR ESTIMATE for the CHEXA(8) also includes the contribution of σ z ,
σ xz , and σ yz . These three values are not listed in Table 15-5 and Table 15-6. The
procedure is identical to that of the plate element with the exception that it
includes six instead of three stress components.
Table 15-5 and Table 15-6 indicate that the smaller values of δg and the ERROR
ESTIMATE are obtained from the models composed of elements for which vertex stresses
are computed by NX Nastran. A detailed discussion of the error estimator used by NX
Nastran is presented later in the section “Discussion of Error Measures.”
In addition to values of δg for each stress component and the ERROR ESTIMATE of
Eq. 15-5 NX Nastran output also contains values of δ g for each of the stress invariants.
These quantities are output for each grid point in the defined SURFACE (plate elements)
or VOLUME (solid elements).
Element Stress Discontinuities

Other statistical error estimates may be generated by associating the estimates with the
elements rather than with the grid points. For example, root mean square errors for each
stress component of an element may be computed from the values of δei that are computed
for each of the Ng connected vertex grid points where δei = ( σ e – σgi ) . This computation is
done for each stress component for all of the elements of interest.
CHAPTER 15 501
Model Verification
Next is an example calculating the NORMAL-X stress components for elements 20 and 60
in Figure 15-26.
NORMAL-X Stress
EID=20
GID σe σ gi δ ei
20 –3.820136E+03 –5.218E+03 1.398E+03

21 –3.820136E+03 –4.969E+03 1.149E+03
62 –3.820136E+03 –2.485E+03 –1.335E+03
61 –3.820136E+03 –2.609E+03 –1.211E+03
Ng
2
∑ ( δ ei )
i = 1
δe = ---------------------------
-
Ng
2 2 2 2
( 1398 ) + ( 1149 ) + ( – 1335 ) + ( – 1211 )
δg = -------------------------------------------------------------------------------------------------------------- = 1277
4 Eq. 15-6
EID = 60
GID σe σ gi δ ei
61 -1.273379E+03 -2.609E+03 1.336E+03

62 -1.273379E+03 -2.485E+03 1.212E+03
103 -1.273379E+03 -9.155E-05 -1.273E+03
102 -1.273379E+03 -6.104E-05 -1.273E+03
2 2 2 2
( 1336 ) + ( 1212 ) + ( – 1273 ) + ( – 1273 )
δe = -------------------------------------------------------------------------------------------------------------- = 1274
4
The other two components ( σ y and τ xy ) can be calculated in a similar manner.
The above data provides an estimated error for each of the three stress components for
each of the two elements considered. As discussed in the previous section, it is desirable
to combine these three estimated errors into a single representative error estimate for each
element. The root mean square value of the three error estimates for each of the three
stress components offers such a representative error measure. In other words,
N
c
2
∑ ( δ ei )
i = 1
ERROR ESTIMATE = ---------------------------
- Eq. 15-7
Nc
(The results ( δ e and ERROR ESTIMATE) from NX Nastran for this model, using CQUAD4,
CQUAD8, and CQUAD4 with the corner option, are summarized in Table 15-7 and
Table 15-8.
Table 15-7 The Values of δ e and ERROR ESTIMATE for Different Plate Element Types
Location at EID = 20
( δe ) ( δe ) ( δe ) ESTIMATE
Element Type
CQUAD4 1.277E+03 4.757E–09 6.212E+01 7.382E+02
CQUAD8 0.000E+00 4.014E–09 0.000E+00 2.318E-09
Table 15-8 The Values of δ e and ERROR ESTIMATE for Different Plate Element
Types
Location at EID = 60
( δe ) ( δe ) ( δe ) ESTIMATE
Element Type
CQUAD4 1.274E+03 4.520E–09 6.212E+01 7.365E+02
CQUAD8 5.038E-10 3.362E–09 0.000E+00 2.020E-09
The CHEXA(8) element is not included in the above tables; however, the calculations are
similar with the exception that six rather than three components of stress need to be
considered. In addition to values of δe for each stress component and the ERROR
CHAPTER 15 503
Model Verification
ESTIMATE of Eq. 15-7, NX Nastran output also contains values of for each of the stress
invariants. For plate elements, all of these quantities are output at the neutral plane and
at fiber distances Z1 and Z2 for each element in the defined SURFACE.
Discussion of Error Measures

An examination of Table 15-5 through Table 15-8 clearly shows that the error estimates
for CQUAD4 elements without vertex stress data are substantially larger than the
corresponding error estimates for those elements that provide vertex stress data as a
default. In view of the fact that the finite model under discussion was constructed with a
rather high mesh density to produce results that agreed very closely with theory, you may
perhaps be surprised by the large values of the error estimates for those elements that only
provide stress data at the element centroid.
These seemingly large values of error estimators for CQUAD4 elements are not
uncommon occurrences in statistical error measures. For example, the mean and variance
have direct analogies in engineering mechanics. The mean is analogous to the centroid of
a body and is thus a measure of where the mass is centered. The variance is the second
moment about the mean, and it tends to be small if the majority of the mass is
concentrated about the centroid. As the mass is dispersed further from the centroid, the
moment of inertia tends to increase. If the mass is concentrated at the centroid (as it might
be in some idealized lumped mass models), the moment of inertia becomes zero.
Thus, the error estimates for CQUAD4 elements tend toward larger values because the
data used to compute the estimators is dispersed relatively far from the mean. For those
elements that provide stress data at vertices, the estimators tend toward smaller values
because the stress data tends to be concentrated about the mean in well-designed finite
element models.
Generally, you should design a finite element mesh for static analysis so that all important
stress gradients are adequately represented. If the mesh is not of sufficient detail, the
stress data whether available at element vertices or at element centroids results in the
relatively large values of the error estimators discussed here.
You are cautioned that very inaccurate values of these error estimators may occur at the
edges of defined SURFACEs and on the faces of defined VOLUMEs. In summary, the
error estimators under discussion can, in some cases, be highly inaccurate. Nevertheless,
these data are quite useful when interpreted properly.
User Interface
The element and grid point stress discontinuity output can only be obtained if the grid
point stress output is requested via the STRFIELD Case Control Command. The
STRFIELD Case Control command is also used for graphical postprocessing of grid point
stress, element stress discontinuity, and grid point stress discontinuity. The STRFIELD
command, however, does not provide printed output for the grid point stress. The
GPSTRESS Case Control command is needed if printed output for the grid point stress is
desired. Similar to the GPSTRESS command, the STRFIELD command also requires that
you define all applicable SURFACEs and VOLUMEs in the OUTPUT(POST) portion of the
Case Control Section. The element stresses must also be requested for those elements that
lie on the SURFACEs and/or VOLUMEs. For a more detailed description of GPSTRESS,
refer to “Reviewing Grid Point Stresses” on page 432.
Two other Case Control commands are needed–GPSDCON for grid point stress
discontinuity and ELSDCON for element stress discontinuity. Note that you can also
request grid point stress discontinuity without requesting element stress discontinuity,
and vice versa, by omitting the appropriate Case Control command (ELSDON or
GPSDCON).
The following remarks should be noted relative to the use of the GPSDCON and/or
ELSDCON commands.
1. The GPSDCON and ELSDCON commands are supported only in Solution
Sequence101.
2. The GPSDCON and ELSDCON commands may be placed above the SUBCASE
level or in individual SUBCASEs and/or SUBCOMs. The use of these commands
above the SUBCASE level causes the stress discontinuity data to be output for all
SUBCASEs in the NX Nastran input file. If you wish to restrict stress
discontinuity output to specific SUBCASEs, GPSDCON and/or ELSDCON
commands should appear only under these specific SUBCASEs.
3. You are cautioned that these commands can produce a substantial amount of
data. Therefore, you should be judicious in their use.
4. Stress continuity data is not provided when both plate and solid elements are
connected to a grid point that is involved in stress discontinuity calculations.
The following is a listing of the input file used to generate the results for the CQUAD4
element model used in Table 15-5 through Table 15-8. The example problem in
Listing 15-7 contains the full list of Case Control commands required to obtain Grid
Point/Element Stress Discontinuity output.
CHAPTER 15 505
Model Verification
Listing 15-7 Input File for the Error Estimators for CQUAD4
$
$ FILENAME - q4sdcon.dat
$
ID STRESS DISCON
TIME 25
SOL 101
CEND
TITLE = CANTILEVER BEAM WITH PLATES
$
SPC=1
LOAD=1000
$
$
$ ELEMENT OUTPUT REQUIRED
$
SET 22 = 19,20,21,59,60,61
$
ELSTRESS = 22
$
$ FOLLOWING REFERENCES SURFACE 91
$
SET 21 = 91
GPSTRESS=21
STRFIELD=21
$
$ FOLLOWING ARE FOR STRESS DISCONTINUITIES
$
GPSDCON=21 $ FOR GRID POINT DISCONTINUITIES
ELSDCON=21 $ FOR ELEMENT DISCONTINUITIES
$
$ FOLLOWING REQUIRED FOR GRID POINT STRESS
$
OUTPUT(POST)
SET 95 = 19,20,21,59,60,61
$
SURFACE 91 SET 95 NORMAL Z SYSTEM ELEMENT
$
BEGIN BULK
$
$ BRING IN THE REST OF THE BULK DATA FILE
$
INCLUDE 'q4sdcon.blk'
$
$
ENDDATA
15.6 Postprocessor Checks

Once you have completed your NX Nastran job, you must interpret the results. For a large
model, it is impractical to go through every page of your output to look at the results.
Graphical postprocessors can help you more easily examine and interpret large amounts
of output data. This section describes some of the basic output quantities that you may
want to consider when using a postprocessor.
Examining Displacements
When you post process your model, you should create a displacement plot for each of your
loading conditions. If there are abrupt changes in the displacements at a certain region,
you may want to zoom in to inspect that local region and see if they are justifiable. One
possible cause can be improper modeling at that location or certain elements not being
connected.
Examining Stress Contours

Stress contour plots are calculated using a stress averaging technique of neighboring
elements. Once these average stresses are calculated, the postprocessor assigns different
colors to these stresses based on the stress ranges. These stresses can then be plotted as
stress contour plots with various colors designating different stress ranges. This feature is
probably the most frequently used postprocessing option. When used properly, it can
help you quickly identify high stress regions and provide you with insight regarding your
model.
When you create stress contour plots of component stresses, you should ensure that the
component that you are plotting is in a consistent direction for all the elements that are
included in your plot. The default component direction may be a function of element type.
Some postprocessing packages may also transform them into a common coordinate
system. The following example problem in Figure 15-27 illustrates the importance of
consistent direction. The stresses in a CQUAD4 element are in terms of the element
coordinate system.
See Also
• “CQUAD4 and CTRIA3 Elements” on page 101 of the NX Nastran Element

Library
CHAPTER 15 507
Model Verification
4 5 6
100 lbs
1 2
y 100 lbs
1 2 3
Figure 15-27 CQUAD4 Model
The complete input file used for this problem is sown in Listing 15-8.
Listing 15-8 Input File for CQUAD4 Stress Contour Plots

with Consistent Connectivity
$
$ FILENAME - (q4consc.dat)
$
SOL 101
TIME 5
CEND
TITLE = CONSISTENT CQUAD4 CONNECTIVITIES
SPC =1
LOAD =1
STRESS = ALL
BEGIN BULK
$
$
PARAM,POST,0
$
$ THIS SECTION CONTAINS GRID POINT LOCATIONS
$
GRID 1 0.0 0.0 0.0
GRID 2 10. 0.0 0.0
GRID 3 20. 0.0 0.0
GRID 4 0.0 10. 0.0
GRID 5 10. 10. 0.0
GRID 6 20. 10. 0.0
$
Listing 15-8 Input File for CQUAD4 Stress Contour Plots

with Consistent Connectivity (continued)
$ THIS SECTION CONTAINS ELEMENT CONNECTIVITIES

$
CQUAD4 1 10 1 2 5 4
CQUAD4 2 10 2 3 6 5
$
$CQUAD4 2 10 3 6 5 2
$
$ LOADS
$
FORCE 1 3 100. 1.
FORCE 1 6 100. 1.
$
$ CONSTRAINTS
$
SPC 1 1 12345 0.0
SPC 1 4 1345 0.0
$
$ SECTION AND MATERIAL PROPERTIES
$
PSHELL 10 10 .1 10
MAT1 10 1.+7 .32
$
ENDDATA
In this example, the connectivities for both CQUAD4 elements are described in a
consistent manner as shown in Figure 15-28. Due to this consistency, the corresponding
component stresses (e.g., ) point in the same direction for both elements. Note that for this
model is equal to zero for both elements.
σy = 0
σy = 0
4 5 5 6
σ x = 200 σ x = 200 σ y = 200 σ x = 200
1 2
1 2 2 3
σy = 0 σy = 0
y σ von = 200 σ von = 200
1 2
Figure 15-28 Consistent Connectivity
If you perform a stress contour on this model using σx , the contour plot for σx across the
two elements
www.cadfamily.com can be calculated as follows:
CHAPTER 15 509
Model Verification
( σ x1 + σ x2 ) ( 200 + 200 )
σ x = ------------------------------- = ------------------------------- = 200.
2 2
The contour plot for σvon across the two elements can be calculated as follows:
( σ von1 + σ von 2 ) ( 200 + 200 )

σ von = ------------------------------------------- = ------------------------------- = 200.
2 2
On the other hand, if the two elements are not connected in a consistent matter as shown
in Figure 15-29, then the corresponding component stresses (e.g. σx , ) do not necessarily
point in the same directions for both elements. As shown in Figure 15-29, the σ x
component stresses point in the directions of the x- and y-axis for elements 1 and 2,
respectively.
σx = 0
σy = 0
4 5 5 6
σ x = 200 σ x = 200 σ y = 200 σ y = 200
1 2
1 2 2 3
σy = 0 σx = 0
y σ von = 200 σ von = 200
1 2
Figure 15-29 Inconsistent Connectivity
Once again, if you perform a stress contour on this model using , the value used in the
contour plot for across the two elements in this case is as follows:
( σ x1 + σ x2 ) ( 200 + 0 )
σ x = ------------------------------- = ------------------------- = 100.
2 2
As you can see, the σx value used in the contour plot for this case is basically meaningless
and could lead you to an erroneous conclusion.
The von Mises stresses used for the contour plot can be calculated as follows:
( σ von1 + σ von 2 ) ( 200 + 200 )

σ von = ------------------------------------------- = ------------------------------- = 200.
2 2
As you can see, σ von remains the same since it is an invariant quantity and is independent
of the order of connectivity. Note that the input file used for Figure 15-29 is identical to
Listing 15-8 with the exception of the shaded CQUAD4 entry. In other words, replace the
following CQUAD4 entry:
1 2 3 4 5 6 7 8 9 10
CQUAD4 2 10 2 3 6 5
with the following:
1 2 3 4 5 6 7 8 9 10
CQUAD4 2 10 3 6 5 2
and the two input files are then identical.
As you have probably observed by now, using contour plots of component stresses can
lead to misleading results unless they are transformed into a common coordinate system.
Therefore, in most instances, it is generally easier and more meaningful to look at the
invariant stress quantities, such as von Mises stresses, instead of the component stresses.
However, there are situations where you may want to look at component stresses. An
example is a cylindrical model whereby you are interested in looking at the hoop stress.
In this case, you should be aware of the consequences if your model is not created in a
consistent manner. Alternate invariant stress quantities are the principal stresses.
Creating Grid Point Stress and Stress Discontinuity Plots

The grid point stress feature enables you to obtain stress output at the grid points instead
of the elements. If the contour plot using grid point stress is substantially different from
the contour plot using element stress, it is probably an indication that the model needs to
be refined or that you may have inconsistent element orientations as described in the
previous section.
See Also
• “Reviewing Grid Point Stresses” on page 432
Stress discontinuity is a measure of the probable error of the stresses across the grid points
or elements. This type of plot assists you in the identification of regions of your model that
may require refinement. This feature is discussed in detail in “Stress Error Estimators” on
page 494.
CHAPTER 15 511
Model Verification
Viewing Element Strain Energy Plots

The element strain energy plot is an excellent tool for identifying areas that are most
effective in influencing your design. For a large model, this output request can generate
a large amount of printed output. The plot option can be used to generate plots without
generating printed output. This feature is discussed in detail in “Strain Energy Output”
on page 454.
16
CHAPTER
Restarts
■ Introduction to Restarts
■ Types of Restarts
■ Read-Write Restarts
■ Read-Only Restarts
■ Mini-Database Restarts
■ Determining the Version for a Restart
16.1 Introduction to Restarts

A restart is a logical way of continuing from a previous run without having to start
from the beginning. The savings can be in terms of both time and money. In the case
of a static analysis, the most expensive and time consuming part of the run is the
decomposition of the stiffness matrix, especially with large models. Now suppose
after completing the original run, you want to obtain additional output (e.g., stresses,
displacements, etc.) or add more load cases. You can always start from the beginning
and redo the whole problem, or you can perform a restart at a fraction of the time and
cost. In the case of additional output requests and additional load conditions, the
decomposition of the stiffness matrix, which was performed in the previous run, is
not redone if you use the restart feature.
However, a restart may not always be the best option compared to rerunning the job.
Determining whether to save the database for restarting or rerun the analysis is
based on:
• the amount of available disk space

• computer speed
• the types of restarts
• the size of your model
For conventional static analysis runs (i.e, those not involving superelements), if a
restart involves model changes, the savings is probably minimal. However, in the
case of additional output requests or additional load cases, the savings can be
substantial as indicated by Table 16-1.
In addition to needing to rerun a job, you may also want to use NX Nastran’s restart
capabilities if you need to access data from an old file and the input file is no longer
available. If you still have the database archive for that solve, the following sample
run ( Listing 16-1) shows how you can use restart to either print or punch a copy of
the Bulk Data from the database that you can subsequently use to create a new input
file.
CHAPTER 16 515
Restarts
Listing 16-1 Input File for Printing/Punching a Listing of

Bulk Data File from the Database
$ FILE - bulklist.dat
$
restart
assign master=’archive1.MASTER’
sol 101
time 10
cend
title = run to get listing of bulk data
echo = sort
$
$ replace echo=sort with echo=punch if a
$ punched output is desired instead of a listing
$
param,serst,semi
begin bulk
$
enddata
See Also
• “Restart Procedures” on page 306 of the NX Nastran User’s Guide

16.2 Types of Restarts
There are two types of restarts available in NX Nastran:
• automatic
• manual
With automatic restarts, NX Nastran automatically determines which operations
need to be performed during the current run based on the changes that are made in
your input file. Automatic restarts are easy to use, and they can potentially save you
a substantial amount of time—both CPU and elapsed time—by not having to
resolve a problem from the beginning.
With manual restarts, in addition to providing the necessary input file changes, you
are also responsible for telling NX Nastran which major operations need to be
performed. Manual restarts are more difficult to use and are more prone to user
error.
Restarts can use the same solution sequence or a different solution sequence.
Restarting from a statics run and requesting stress output is an example of a restart
using the same solution sequence. Restarting from a statics run into a normal modes
run is an example of a restart across solution sequences.
Within automatic restart, there are two different options:

• a read-write restart
• a read-only restart
Each of these is described below.
Understanding Restart Terminology

When you’re performing restart analyses in NX Nastran, your analyses are
categorized into cold start and restart runs.
Cold Start Runs

Your initial run is known as a “cold start” run and is identical to a regular NX
Nastran job, except that you must save your database. In addition to your normal
output files (e.g., the .F06 file), NX Nastran creates four database files as a result of
this run. The naming convention for the filenames is machine dependent-you should
refer to NX Nastran Installation and Operations Guide for the exact syntax for your
machine. For a typical UNIX machine with an NX Nastran input file called
“stat1.dat”, the following sample submittal command can be used:
nastran stat1 scr=no
In this case, nastran is the name of the shell script for executing NX Nastran. The
“scr=no” option in the above submittal command is not required if you have not
modified the default value of the submittal command as provided on the delivery
media. By default, the following two database files are created as a result of the
above command:
stat1.DBALL
stat1.MASTER
Restart Runs
Once you’ve performed the cold start run and saved the database, subsequent runs
are referred to as “restart” runs.
16.3 Read-Write Restarts

The following sections describe the changes you’ll need to make to your input file to
restart the analysis for a read-write restart.
NASTRAN Statement Section

This section is normally the same as your cold start run. You shouldn’t change the
BUFFSIZE in a restart run.
CHAPTER 16 517
Restarts
File Management Section (FMS)

This section tells NX Nastran that you are performing a restart run. In the FMS
Section, only two statements are needed: the “restart” statement to tell NX Nastran
that this current run is a restart run and the “assign” statement that tells NX Nastran
the name of the database you want to attach.
The RESTART statement is required in any restart run. The general format for the
RESTART statement is:
RESTART VERSION=a,b
where, “a” is the version from which you restart (the default value for “a” is LAST),
and “b” indicates whether version “a” will be kept (KEEP) or deleted (NOKEEP) at
the end of the run. The default value for “b” is NOKEEP. Due to the default values,
the following two restart statements are identical:
RESTART VERSION=LAST,NOKEEP
RESTART
Whenever you perform, the software automatically creates a new version number
in the database. For each restart, the current version number is incremented by one,
regardless of whether the job ran successfully or not. However, there are two
exceptions to this rule (which are discussed later on in the chapter).
Restart–Method 1
You also need to tell NX Nastran the database that you want to attach to your
current run. There are two ways of accomplishing this goal. One way is to use the
ASSIGN statement. For example, if you are restarting from the database created by
stat1.dat, the following FMS statements can be used in your current run
RESTART
assign master=’stat1.MASTER’
$
$ the following statement is optional
$
assign dball=’stat1.DBALL’
For UNIX machines, the filenames are case sensitive. They should be entered
exactly as they were created in the cold start run and enclosed with a single quotes
as shown above. Note that the use of the second ASSIGN statement is optional since
the “master” DBset can find the associated DBALL database.
Restart–Method 2
An alternate way to attach a restart database is to use the DBS keyword on the
submittal command instead of the ASSIGN statement. Assuming the current run is
called stat2.dat, then the equivalent submittal command is as follows:
nastran stat2 dbs=stat1
The ASSIGN statement is not needed in this case; however, the RESTART statement
is needed using either option. While the first method requires more input than the
second method, it also provides you with information about the database used for
the restart run although this information is also provided in the Execution Summary
Table (.F04 for most machines).
Executive Control Section

In a restart analysis, the executive control section is identical to one in a cold start
analysis.
Case Control Section

When you’re performing a restart analysis, the software’s automatic restart logic
compares the modifications you made to the Case Control and Bulk Data Sections to
determine which operations need to be processed upon restart. Therefore, you must
be very careful in the changes that you make in your restart run. Adhering to the
following rules will avoid unnecessary reprocessing of previously completed
operations:
1. You must include in your restart run all solution-type related Case Control
commands that are unchanged as compared to the cold start run. In other
words, do not make unnecessary load, SPC, or MPC command changes or
remove them from the Case Control Section unless these are actual changes.
This process is clarified later on with the example problems.
2. Output requests can be modified. A typical example can be a request of the
element stress output that was not requested in the cold start run.
Bulk Data Section

The software stores a copy of the Bulk Data for each version. Your restart run must
not contain any Bulk Data entries that were included with the previous version that
you are restarting from since they are already saved in the database for that
particular version. The Bulk Data section in the current restart run should only
contain the addition of new entries, the modification of old entries, or the deletion of
CHAPTER 16 519
Restarts
old entries from the database. The following is a series of three runs illustrating an
application of a restart. The first run (Listing 16-2) is a sample statics run with one
loading condition, and it requests displacement output.
Listing 16-2 Input File for a Cold Start Run
$
$ filename - stat1.dat
$
ID ROD,TEST
SOL 101
TIME 20
DIAG 8
CEND
TITLE= COLD START RUN
$
SPC = 1
$
SUBCASE 1
LOAD = 1
DISP = ALL
$
BEGIN BULK
$
GRID,1,,0.,0.,0.
GRID,2,,5.,0.,0.
GRID,3,,10.,0.,0.
CROD,1,1,1,2
CROD,2,1,2,3
PROD,1,1,.2
MAT1,1,1.+7,,.32
FORCE,1,2,,1000.,1.,0.,0.
SPC1,1,123456,1
PARAM,AUTOSPC,YES
$
ENDDATA
The second run (Listing 16-3) is a restart run with an additional load condition.
Displacement and stress outputs are desired for this new loading condition.
Furthermore, stress output for all the elements is desired for Subcase 1. However,
since you already printed the displacement for Subcase 1 in your original run, you
may elect not to reprint this information by removing the displacement request
from Subcase 1. There are no model changes as compared to the original run. Note
that the Bulk Data entries that were present in the original run must not be included
in the current run. The only entry in the Bulk Data Section is the new FORCE entry
for Subcase 2, which was not included in the original run. Since this load is a new
loading condition, the FORCE entry must have an unique ID (2 in this case) as
compared to the existing FORCE entry ID (1 in this case) that is stored in the
database.
This type of restart run constitutes an efficient restart since the re-decomposition of
the stiffness matrix is not performed; for a large problem, this can result in
substantial CPU savings.
Listing 16-3 Input File for a Restart Run

with an Additional Load Case
$
restart
ID ROD,TEST
SOL 101
TIME 20
DIAG 8
CEND
TITLE= RESTART RUN
$
SPC = 1
$
SUBCASE 1
LABEL = ADDITIONAL STRESS OUTPUT REQUEST
LOAD = 1
STRESS = ALL
$
SUBCASE 2
LABEL = ADDITIONAL LOAD CASE
LOAD = 2
DISP = ALL
STRESS = ALL
$
BEGIN BULK
$
$
$ NOTE THAT THE FORCE ID IS UNIQUE
$ WITH RESPECT TO EXISTING LOAD ID IN
$ THE DATABASE.
$
FORCE,2,2,,-500.,1.,0.,0.
$
ENDDATA
CHAPTER 16 521
Restarts
The third run (Listing 16-4) is a restart run with a modification in the material
property. The same two loading conditions are used as run number two. A new
material entry is added in this run and the old material entry (the eighth sorted Bulk
Data from the stat2.dat run) is deleted. The general format for deleting a range of
entries from the database is
/,x,y
where x and y are the range of the first and last sorted Bulk Data count to be deleted
from the database, respectively. You can have as many of these entries as you need.
If you are only deleting one entry (y=x), then either one of the following two formats
can be used:
/,x,x
/,x
The sorted Bulk Data is output by default. If the ECHO Case Control command is
specified in the cold start run, it should be set to output the sorted Bulk Data (e.g.,
ECHO=SORT).
There may be times when you may be making substantial changes to the model. In
this case, it may be easier to delete the whole Bulk Data listing from the database and
insert a complete new Bulk Data file as shown below:
/,1,z
$ include a complete new bulk data listing
$
In this case, z stands for an integer larger than the total number of Bulk Data entries
(e.g., 999999). Alternatively, you could delete the database files and run the
problem as a cold start run.
Listing 16-4 Input File for a Restart Run

with Modified Material Property Entry
restart
ID ROD,TEST
SOL 101
TIME 20
DIAG 8
CEND
TITLE= RESTART RUN
$
SPC = 1
$
SUBCASE 1
LABEL = DISPLACEMENT AND STRESS OUTPUT REQUEST
LOAD = 1
DISP = ALL
STRESS = ALL
$
SUBCASE 2
LABEL = ADDITIONAL LOAD CASE
LOAD = 2
DISP = ALL
STRESS = ALL
$
BEGIN BULK
$
/,8
mat1,1,3.+7,,.32
$
ENDDATA
In this example, the old MAT1 entry is the eighth sorted Bulk Data (from run
stat2.dat) as shown by the following partial output:
CHAPTER 16 523
Restarts
Listing 16-5 Sorted Bulk Data Listing from a Cold Start Run
CARD
COUNT . 1 .. 2 .. 3 .. 4 .. 5 .. 6 .. 7
1- CROD 1 1 1 2
2- CROD 2 1 2 3
3- FORCE 1 2 1000. 1. 0.
4- FORCE 2 2 -500. 1. 0.
5- GRID 1 0. 0. 0.
6- GRID 2 5. 0. 0.
7- GRID 3 10. 0. 0.
8- MAT1 1 1.+7 .32
9- PARAM AUTOSPC YES
10- PROD 1 1 .2
11- SPC1 1 123456 1
ENDDATA
TOTAL COUNT= 12
This type of restart (run number 3) requires the reformulation of the stiffness matrix
since the material was changed. For a nonsuperelement run, this type of restart does
not save you much in terms of runtime, if any. This run (run number 3) is included
merely to show you how to delete information from the database. However, for
superelement type runs, this type of restart can still save you time since you may
only be resolving a portion of the structure.
The model used in the above series of runs is quite simple. In most cases, such
simple models do not warrant the use of a restart since the savings are insignificant
for such a small model. This simple example is chosen to illustrate the mechanics of
how simple the restart feature is without being distracted by the details of a complex
model.
The same restart procedure can now be applied to a larger model. The model used
is “BCLL16". “BCLL16" is one of the standard test problems that is run for each new
version of NX Nastran. It is basically a 16 × 16 × 16 cube of CHEXA elements
overlaid with a CQUAD4 elements on the face of each CHEXA element to simulate
a relatively dense stiffness matrix. The model consists of 29,400 degrees of freedom.
The following table illustrates the benefit of using restarts for large models.
Table 16-1 BCLL16 Performance Summary
Run Description CPU Time

Number (Sec)
1 Static analysis with one load case and the following 372.1
output request:
SET n = a,b,c,d
DISP = n
STRESS (PLOT) = All
2 After the completion of run number 1, you now 364.5
decide that you want additional printed stress
output for a few selective elements. One option is
to rerun the same job from the beginning with the
following modified output request in the Case
Control Section:
SET m = x,y,z
STRESS = m
3 The second option is to restart from run number 1 5.0
with the following modified output request in the
Case Control Section:
SET m = x,y,z
STRESS = m
Note the tremendous savings in the CPU time (364.5 versus 5 seconds) between the
run numbers 2 and 3 with both runs accomplishing the same goal.
CHAPTER 16 525
Restarts
The cold start and restart input files used for run numbers 1 and 3 are shown in
Listing 16-6 and Listing 16-7.
See “Mini-Database Restarts” on page 527 for efficient method of restarts when
only data recovery is desired.
Listing 16-6 Input File for a BCELL16 Cold Start Run
$ FILENAME - BCLL16A.DAT
INIT DBALL,LOGICAL=(DBALL(200000))
INIT SCRATCH,LOGICAL=(SCRATCH(200000))
ID NXN, BCLL16A
TIME 300
SOL 101
CEND
TITLE = STATIC ANALYSIS OF A CELLULAR CUBE BCLL16A
SUBTITLE = 16 X 16 X 16 CUBE
ECHO=NONE
STRESS(PLOT)=ALL
SET 100 = 30080,40088,70880,80888
DISP = 100
LOAD=100
SPC=10
BEGIN BULK
$ BRING IN THE REST OF THE BULK DATA FILE
$
include ’bcllblk.dat’
ENDDATA
Listing 16-7 Input File for a BCELL16 Restart Run
$ FILENAME - RBCLL16A.DAT
$
RESTART
ASSIGN MASTER=’bcll16a.MASTER’
$
$ THE ABOVE STATEMENT ASSUMES THAT YOUR
$ COLDSTART INPUT FILENAME AS
$
$ bcll16a.dat
$
ID NXN, BCLL16A
TIME 100
SOL 101
CEND
TITLE = STATIC ALALYSIS OF A CELLULAR CUBE BCLL16A
SUBTITLE = 16 X 16 X 16 CUBE
ECHO=NOSORT
SET 100 = 30080,40088,70880,80888
set 200 = 1001,2001,3001,4001
$DISP = 100
stress = 200
LOAD=100
SPC=10
BEGIN BULK
$
$ SINCE THERE IS NO MODEL CHANGES, NO BULK
$ DATA ENTRIES ARE NEEDED
$
ENDDATA
Note the user interface is quite straightforward considering the tremendous benefits
in terms of CPU reduction.
16.4 Read-Only Restarts

Read-only restart is very similar to read-write restart, except that original database is
accessed in a read-only mode. No new data is added to the existing database. Instead,
all new data is written to the new database, leaving the original database intact. You
can delete the new database or save it for another restart. The new database created
during the restart run retains an internal link to the original database.
To use read-only restart for the same example problem as in Listing 16-3, all you
need to do is modify the FMS statements from:
Restart
Assign master=’stat1.MASTER’
to:
Restart logical=run2
CHAPTER 16 527
Restarts
Assign run2=’stat1.MASTER’
In this case, run2 is an arbitrary logical name.
16.5 Mini-Database Restarts

You can save on the database only those data blocks that are required for data
recovery. The size of the database, called a mini-database, is on average about 25%
the size of a full database. For pure data recovery run (no SPC or loading changes),
this type of restart is very efficient.
To use this feature, use the following submittal command in the cold start run:
nastran jobname scr=mini
In the restart run, no special input is required because NX Nastran automatically

detects a mini-database and will process any output request from the table below.
The mini-database restart processes the following Case Control commands and user
parameters in SOL 101:
DISPLACEMENT STRAIN
ELSDCON STRFIELD
ESE STRESS
FORCE SUBCOM
GPFORCE SUBSEQ
GPSDCON SURFACE
GPSTRESS SYMCOM
MPCFORCES SYMSEQ
PARAM,POST THERMAL
REPCASE VOLUME
SPCFORCES
For cyclic symmetry analysis, you can’t change the HOUTPUT and NOUTPUT in
the restart run.
16.6 Determining the Version for a Restart

Not all versions in the database are restartable. For each run, NX Nastran prints a
message near the top of the .F04 file indicating the version number of the current job.
In general, if a job completes without any fatal messages, and the database is saved,
then that particular version is restartable. It is a good idea to keep a brief log of all
the restartable versions since this is probably the most reliable source of information
regarding whether a version is restartable. If a restart job fails, e.g., due to Bulk Data
error, then this newly created version is not restartable, and it is indicated as such by
the following error message at the bottom of the .F06 file.
*** USER WARNING MESSAGE 784 (XCLNUP)

VERSION = yyy PROJECT = “zzz” OF THIS DATA BASE IS NOT VALID FOR RESTART
PURPOSES.
USER ACTION:
SUBSEQUENT RESTARTS SHOULD REFERENCE VERSION = xxx OR A PRIOR VALID VERSION
In this case, you must restart from a previous valid version. The “xxx” and “yyy”
above denote version numbers. The “zzz” denotes a project description provided by
you. This project description is alphanumeric and can contain up to 40 characters
with the default being blank. The project description is rarely used and is an optional
statement.
If for some reason the records for the old runs are no longer available, then the
DBDIR FMS statement can be used to query the database contents to find out which
versions are stored in the database. The following simple setup is all that is required
for this purpose:
ASSIGN MASTER=’ddddd.MASTER’
DBDIR VERSION=*,PROJECT=*
ENDJOB
Neither the Executive, Case Control, nor Bulk Data Section is required in this case.
Furthermore, a new version is not created because the ENDJOB statement is present.
Near the top of the .F06 output, a PROJECT VERSION DIRECTORY TABLE is
printed listing all the versions in the database. A “**” next to a version number
indicates that this particular version was deleted from the database. This deletion
may be due to the “NOKEEP” option or the use of the DBCLEAN statement when
performing a restart run. A version number with a “**” next to it is not restartable.
A version number without a “**” next to it is restartable if the run that created the
version did not terminate with UFM 784. This DBDIR FMS statement can also be
used to check the database directory for the existence of data blocks (e.g., UG). If the
job fails very early in the run (e.g., error in the FMS section), then a new version may
not be created.
Always back up your database on a regular basis. If the system aborts the run, then
your database may get corrupted. Another good practice, which ensures that only
good models are retained in the database, is to perform the following:
1. Use RESTART VERSION=a,KEEP

where “a” is a version number.
CHAPTER 16 529
Restarts
2. If a version contains errors, or is no longer of interest, then you can use the
FMS statement DBCLEAN to remove obsolete or incorrect versions of the
model from the database. Using DBCLEAN allows the Executive System
to reuse some of this space for new versions.
17
CHAPTER
Database Concepts
■ Introduction to Database Concepts
■ Delivery Database
■ Deleting DBsets
■ Database Autoassignment
■ Database Project and Version
■ Migrating Databases
■ Database Archival, Compression, and Transfer
■ Transferring Databases Across Different Platforms
■ Guidelines for Large Problems
17.1 Introduction to Database Concepts

All NX Nastran runs require a database. A database allows for the storage and
retrieval of data: data blocks, parameters, scratch files, and DMAP object and source
files.
During a run, data may be written onto only one database. This is called the primary
database which is attached automatically or by the INIT MASTER statement.
Additional databases may be attached for “read-only” purposes. This includes a
“delivery” database and one or more “located” databases.
• The delivery database contains solution sequences and is automatically

attached by the SOL Executive statement or the ACQUIRE FMS statement.
• “Located” databases are attached by the DBLOCATE statement.
Before presenting the details of the database structure, the definitions for some of the
basic terms that are used throughout this chapter need to be presented.
DBset Database set. DBset consists of an NX Nastran logical name, such

as MASTER, DBALL, etc., which refers to an entity within the NX
Nastran database. It consists of a collection of physical files.
Database The collection of all DBsets assigned to a run.
Data block A matrix or table (e.g., KAA, LAMA) that is stored in the
database.
Logical name The local internal name (log-name) used by NX Nastran for a file
or DBset.
Word For 32- and 64-bit machines, each word is equivalent to four and
eight bytes, respectively.
BUFFSIZE The length of an NX Nastran buffer in terms of words (32- or
64-bit words). It controls the physical record size for data
storage/transfer that is contained in many NX Nastran logical
units. The default and maximum allowable BUFFSIZE is
machine dependent. The default value is recommended except
for large problems. It can be modified by using the following
NASTRAN statement:
NASTRAN BUFFSIZE = xxxxx
Block A block is often referred to as an NX Nastran GINO block. Each
GINO block contains one NX Nastran buffer. The size of each
block is equal to BUFFSIZE-1 words. (GINO is an acronym for
general input/output.)
{} A brace indicates that the quantity within this bracket is
mandatory. The underlined item within { } is the default value.
CHAPTER 17 533
Database Concepts
[] A square bracket indicates that the quantity within this bracket is

optional.
_____ The underlined value is the default value for a particular option.
17.2 Understanding DBsets

A database is divided into several DBsets which are initialized by the INIT file
management statement. There are two types of DBsets:
• permanent DBsets, which you can save at the end of the run and reuse in
a restart run
• scratch DBsets, which are automatically deleted at the end of the run
A DBset may be composed of up to twenty concatenated physical files or DBset
members. Each member is assigned to a physical file by the ASSIGN statement.
For a typical UNIX-based workstation with an NX Nastran input file called

“stat1.dat”, the following sample submittal command can be used:
nastran stat1 scr=no
In the command above, nastran is the name of the shell script used to execute NX
Nastran. The following two permanent physical database files are created as a
result of the above command.
stat1.MASTER
stat1.DBALL
Note that the temporary scratch files are allocated and used by NX Nastran during
the solve and are deleted at the end. Unless otherwise stated, the input filename is
assumed to be “stat1.dat” in this chapter.
By default, there are four DBsets—two permanent and two scratch—which are
predefined and automatically assigned by the program. The DBset names of the
predefined permanent DBsets are: MASTER and DBALL. The DBset-names of the
predefined scratch DBsets are: SCRATCH and OBJSCR. The DBsets are briefly
described below:
• The MASTER DBset is the directory of the database. It contains the names
of all DBsets, DBset members and their physical file names, and a
directory of projects, versions, data blocks, parameters, DMAP source and
object files. It also contains the NDDL scheme used to describe the
database. The default maximum size for MASTER is 5,000 blocks.
• The DBALL DBset contains all the DMAP data blocks which may be saved
permanently for reuse in a subsequent run. In the solution sequences in
which restarts are not possible, (e.g., SOL 1) this DBset is empty.
• The SCRATCH DBset is a temporary DBset for all scratch data blocks and
files. It has two partitions: one for DMAP data blocks and one for DMAP
module internal scratch files.
• The SCR300 DBset is used as the temporary workspace for the modules.
This space is released at the end of the module execution to be reused by
subsequent modules. The default size is 250,000 blocks
• The OBJSCR DBset is a temporary DBset for DMAP compilation.
• If you want to create your own DMAP source and object files, you need to
allocate the USROBJ and USRSOU DBsets. They contain DMAP source and
object files that may be saved permanently for execution in a subsequent
run. See the COMPILE Executive Control statement.
If an INIT and ASSIGN statement aren’t specified for one or more of the predefined
DBsets, then the corresponding statement(s) are provided by default.
ASSIGN MASTER=’dbs-name.MASTER’
INIT MASTER(RAM) LOGICAL=(MASTER(5000))
ASSIGN DBALL=’dbs-name.DBALL’
INIT DBALL LOGICAL=(DBALL(25000))
ASSIGN USROBJ=’dbs-name.USROBJ’
INIT USROBJ LOGICAL=(USROBJ(5000))
ASSIGN USRSOU=’dbs-name.USRSOU’
INIT USRSOU LOGICAL=(USRSOU(5000))
ASSIGN SCRATCH=’temp-name.SCRATCH’
ASSIGN SCR300=’temp-name.SCR300’
INIT SCRATCH(MEM) LOGICAL=(SCRATCH(175000)),
SCR300=(SCR300(175000))
ASSIGN OBJSCR=’temp-name.OBJSCR’
INIT OBJSCR LOGICAL=(OBJSCR(5000))
The numbers inside the parentheses are the maximum sizes of the DBsets in GINO
blocks. By default, one member with the same log-name as the DBset-name is
assigned to each DBset. (For a description of a GINO block, DBset-name and
log-name, see the INIT statement). For example, the DBALL DBset has one member
also called DBALL. The dbs-name is determined from the dbs keyword on the
nastran command and temp-name is generated by the command procedure. The
scratch DBsets are assigned special names on a scratch disk. (For a description of the
dbs-name and temp-name, see the NX Nastran Installation and Operations Guide).
The filenames above are those that would be generated on UNIX-type computers.
For example, if the name of the input file is called MYJOB.DAT and the dbs keyword
isn’t specified, then the default filenames for the permanent DBsets would be:
MYJOB.MASTER
MYJOB.DBALL
Also, they would have the same directory or path as the input file.
CHAPTER 17 535
Database Concepts
A physical filename is automatically created for any new DBsets specified by an

INIT statement. For example, to create a new DBset called DBUP for split database
operations, specify:
INIT DBUP
The physical filename will then be: MYJOB.DBUP. For another example, to create
two members for DBALL, specify
INIT DBALL LOGI=(DB1 ,DB2)

Their physical filenames will be MYJOB.DB1 and MYJOB.DB2.
You can use the ASSIGN statement o override these default names. For example, to
rename MYJOB.DBUP to FENDER.DBUP, specify
ASSIGN DBUP=’FENDER.DBUP’
INIT DBUP
To rename MYJOB.DB1 and MYJOB.DB2, specify
ASSIGN DB1 =’FENDER.DBALL1’

ASSIGN DB2=’FENDER.DBALL2’
INIT DBALL LOGI=(DB1,DB2)
The INIT statement may be used to change the maximum size of the predefined
DBsets. (The size may be specified in blocks, kilowords, megawords, kilobytes, or
megabytes.) For example, to specify a size of 50,000 blocks for DBALL,

or 50,000 bytes,
INIT DBALL LOGICAL=(DBALL(50KB))

The INIT and ASSIGN statement may be used to specify more than one member for
a DBset which may exist on different physical devices.
ASSIGN DB1 =’physical file name of DB1 on disk 1’

ASSIGN DB2=’physical file name of DB2 on disk 2’
INIT DBALL LOGICAL=(DB1(5000),DB2(5000))
17.3 Delivery Database

The Delivery Database contains the solution sequences. This database, in addition
to the primary database, is automatically assigned in order to execute an NX
Nastran solution sequence. For example, on UNIX-type computers, the filenames
of the database are
SSS.MASTERA
SSS.MSCOBJ
SSS.MSCSOU
If the solution sequence is not an NX Nastran solution sequence, as in the case of a

user’s solution sequence; then the ACQUIRE FMS statement may be used to select
the appropriate database. ACQUIRE NDDL selects the SSS database. The ACQUIRE
statement is required when the DBLOAD FMS statement is used with the ENDJOB
FMS statement.
17.4 Deleting DBsets

There are several different ways that you can control what happens to permanent
DBSets in the primary database at the end of a solve.
• If scr=yes is specified on the nastran command or the INIT MASTER(S)

statement appears in the FMS Section, then NX Nastran deletes all DBsets
in the primary database.
• An individual DBset member is automatically deleted at the end of the run
by specifying the TEMP keyword on the ASSIGN statement. For example,
to delete the USROBJ and USRSOU DBsets at the end of the run, specify:
ASSIGN USROBJ=’A’ TEMP
ASSIGN USROBJ=’B’ TEMP
Note: If you use this method to delete a DBset, you can’t recreate it for a
future run.
• The DBSETDEL FMS statement provides a better method than the TEMP
keyword on the ASSIGN statement except that the DBsets are deleted at the
beginning of the run. The deleted DBsets may be recreated in a future run.
For example:
DBSETDEL USROBJ,USRSOU
17.5 Database Autoassignment
When a database is created, the physical filenames of all DBset members are loaded
into the MASTER DBset. Therefore, when using DBLOCATE or RESTART, it is
sufficient to assign the MASTER DBset member. All other DBset members previously
initialized, i.e., DBALL, DBUP, etc., are automatically assigned by the program. For
example, on restart, the following input will automatically assign the DBALL,
USROBJ, and USRSOU DBsets:
ASSIGN MASTER=’filename of the MASTER DBset member’

RESTART
If DBLOCATE is used, then:
ASSIGN MSTR1 =’filename of the MASTER DBset member’

DBLOCATE LOGI=MSTR1
CHAPTER 17 537
Database Concepts
The autoassignment feature cannot assign the physical file of a DBset member that
has been renamed. This is because the MASTER DBset member contains the old
physical filename. Therefore, an additional ASSIGN statement is required to specify
the new filename of the DBset member. If in the previous restart example DBALL
is renamed, then the following input is required:
ASSIGN MASTER=’filename of the MASTER DBset member’

ASSIGN DBALL=’new filename of the DBALL DBset member’
RESTART
In split database operations, it is often necessary to assign only some of the DBsets
because the others are “offline.” The NASTRAN AUTOASGN statement is used to
specify the types of databases that will be autoassigned:
1. AUTOASGN = 0: No databases will be autoassigned. This includes the

primary database, Delivery Database, and any located databases.
2. AUTOASGN = 1: All databases will be autoassigned. This is the default.
3. AUTOASGN = 2: Only the Delivery Database will be autoassigned.
4. AUTOASGN = 4: Only located databases will be autoassigned.
If some combination of the options above is desired, then sum their values. For
example, if the delivery and located databases are to be autoassigned, then specify
NASTRAN AUTOASGN = 6.
17.6 Database Project and Version

All data stored on the primary database is assigned to a particular project
identification string (ID) and version identification number (ID). The current project
ID is specified on the PROJECT FMS statement. The current version ID is an integer
value assigned by the program each time the database is attached as the primary
database. When a primary database is initialized or a new project is specified, then
all items stored on the database during the run are stored under a version ID of
integer 1. Then, in subsequent runs a new version is created with an ID equal to the
last version ID incremented by one. If the RESTART FMS statement is specified,
then a new version of data is created which is equivalent to the last version. In other
words, all items in the old version are made available for reuse in the new version.
17.7 Migrating Databases

Databases between versions in general aren’t compatible and must be converted, or
migrated, before you can use them in a higher version of the software. This is
required primarily because NDDL sequences change. For example, some qualifiers
may have been added to the paths of some data blocks in order to support new
capabilities in the solution sequences. Also, NDDL description of some data blocks
change.
Importantly, a database created in nonlinear analysis, heat transfer analysis, or

optimization analysis cannot be fully migrated. This is because of changes in Bulk
Data entry formats and/or data blocks. Also, it would be inefficient to recreate
because it is created by modules which perform the bulk of a nonlinear analysis.
You can migrate a database in a single run which consists of two phases.
• In the first phase, the DBLOCATE FMS statement with the CONVERT and
COPY keywords is used to copy most of the data blocks and all of the
parameters from the earlier version database to the current version primary
database.
• In the second phase, the remaining data blocks which cannot simply be
copied into the current version have to be regenerated. This involves the
execution of a short solution sequence that regenerates those data blocks:
specify SOL DBTRANS (or SOL 190).
For example, the input file necessary to migrate a database created by one of the
structured solution sequences (SOLs 100 through 200) is:
ASSIGN MIG=’physical filename of the MASTER DBset’

RFINCL DBTRANS
SOL DBTRANS
CEND
With the DBTRANS solution sequence, it is assumed that the early version database
contains all of the data from a completed and successful analysis. If not, then the
solution sequences may fail and the migration may not be successful. In this case, it
is possible to alter DBTRANS with the ALTER statement to obtain the appropriate
data.
If the earlier version database wasn’t created by an NX Nastran solution sequence,

then it may be necessary to develop new DBLOCATE statements and a new
migration solution sequence.
17.8 Database Archival, Compression, and Transfer

Database archival, compression and transfers are performed using the DBUNLOAD
and DBLOAD FMS statements. You can select all items on the database or selected
items according to the WHERE clause.
Database Archival and Retrieval

The DBUNLOAD FMS statement is used to archive the primary database. It writes
out the data blocks and parameters stored in the database to a sequential file in a
binary format which is similar to that of the OUTPUT2 module (see NX Nastran
DMAP Programmer’s Guide). A database can only be unloaded in a run separate from
the run in which it is created or modified. The default action is to archive all data
CHAPTER 17 539
Database Concepts
blocks and parameters under all projects and versions in the database. The ASSIGN
FMS statement is also required to assign a FORTRAN unit on which to unload the
database. The default unit number is 50. This unit number and other default
attributes are predefined under the logical keyword DBUNLOAD. For example, a
simple request to unload or archive a database is as follows:
ASSIGN DBUNLOAD=’physical filename of archive file’

DBUNLOAD
ENDJOB
The DBLOAD FMS statement is used to read an archived database and copy it to the
primary database. The ACQUIRE statement is required to select the appropriate
NDDL sequence. The ASSIGN FMS statement is also required to assign a
FORTRAN unit to the unloaded database. The default unit number is 51. This unit
number and other default attributes are predefined under the logical keyword
DBLOAD. For example, a simple request to reload the archived database above is
as follows:
ACQUIRE NDDL
ASSIGN DBLOAD=’physical filename of archive file’
DBLOAD
ENDJOB
The example above makes use of the ENDJOB statement, which means that only the
FMS statements will be processed in the run. A new BUFFSIZE may be assigned to
the retrieved database with the NASTRAN statement. SOL DBTRANS or
DBTRANU must also be executed:
NASTRAN BUFFSIZE=xxxx
ASSIGN DBLOAD=’physical filename of archive file’
DBLOAD
SOL DBTRANS
CEND
By default, the DBLOAD and DBUNLOAD statements will load and unload all the
data under all projects and versions found in the database. The WHERE clause may
be used to be more selective according to the project ID, version ID, item name,
qualifier values, and/or DBset. Also, the CONVERT clause may be used to modify
the project ID, version ID, item name, qualifier values, and/or DBset.
Database Compression
When data is deleted from the database, its space is released for the storage of new
data. This old space will be reused before any new space is used. However, even if
the space is not reused, the size of the database is not reduced. After several restarts
or the execution of the DBCLEAN FMS statement, the database may contain a
significant amount of released or “dead” space. There are two methods of removing
dead space or “compressing” the database. The first method involves the
DBUNLOAD and DBLOAD FMS statements, and the examples shown above will
also perform a database compression. The second and simplest method involves the
DBLOCATE statement specified with the COPY keyword. For example, if the
database was created with SOLs 100 through 200, then the input file is:
ACQUIRE NDDL
ASSIGN MYDB=’physical file name of the MASTER DBset’
DBLOCATE LOGI=MYDB COPY
ENDJOB
It is also possible to compress an individual DBset and change its maximum size. For
example, the input file is:
ACQUIRE NDDL
ASSIGN MYDB=’physical file name of the MASTER DBset’
DBLOCATE LOGI=MYDB COPY WHERE(DBSET=’DBALL’)
ENDJOB
Transferring Databases Across Different Platforms

You can also use the DBUNLOAD and DBLOAD statements to copy a database from
one computer platform to another.
• On the first computer, the DBUNLOAD statement converts the database to

a “neutral” file on the first computer. This neutral file is now suitable for a
simple copy to the second computer or may be accessed through a network
if the two computers are so connected.
• On the second computer, the DBLOAD statement converts the neutral file
back to a database. It is also necessary to execute the solution sequence
DBTRANS (or 190) as described under Database Migration, because some
data blocks cannot be converted to a neutral format. (In SOL DBTRANS;
data blocks EST, EMAP, ETT, PTELEM, SLT, KDICT, KELM, MDICT,
MELM, BDICT, BELM, and ACPT are regenerated.)
The ASSIGN statement is also required to assign the physical filename of the neutral
file and specify that the file is formatted since the default format (UNFORMATTED)
is not transferable across dissimilar computers.
The following example unloads a database created by one of the unstructured

solution sequences on the first computer to a neutral file:
ASSIGN DBUNLOAD=’physical filename of neutral file’ FORMATTED

DBUNLOAD FORMAT=NEUTRAL
ENDJOB
Then the neutral file is copied to, or accessed through a network by, the second
computer:
CHAPTER 17 541
Database Concepts
ASSIGN DBLOAD=’physical filename of neutral file’ FORMATTED

DBLOAD FORMAT=NEUTRAL
SOL DBTRANU
CEND
With the DBTRANS solution sequence it is assumed that the neutral file contains all
of the data from a completed and successful analysis. If not, then the solution
sequences may fail and the transfer may not be successful. In this case, it is possible
to alter DBTRANS with the ALTER statement to obtain the appropriate data.
If the database was created in a heat transfer analysis then NASTRAN HEAT = 1
must be specified.
If the database on the first computer was not created by a NX Nastran solution
sequence, then it may be necessary to develop a new transfer solution sequence.
17.9 Using the File Management Section with Databases

The File Management Section (FMS) is intended primarily for the attachment and
initialization of NX Nastran database sets (DBsets) and FORTRAN files. For many
problems due to the default values, the FMS is handled automatically and is not
required to be specified by you.
You can allocate the database file size in any of the following units:
Bytes
Words
Megabytes (1,000,000 bytes)
Megawords (1,000,000 words)
Kilobytes (1,000 bytes)
Kilowords (1,000 words)
NX Nastran GINO blocks (default)
Mixture of the above units
Internally, NX Nastran converts these units to GINO blocks. If an FMS statement is

longer than 72 characters, it must be continued on the next line. You can have a
maximum of 200 files (FORTRAN and DBsets) assigned (online) at a time. This limit
may be less for some machines due to their operating system limit.
Using the INIT Statement

INIT creates/initializes permanent and/or temporary DBsets. The INIT statement
has two basic formats: one for all the DBsets and one specifically for the SCRATCH
DBsets.
Format (Simplified) for All DBsets Except SCRATCH DBsets

INIT DBset-name [LOGICAL=(log-name1(max-size1) ,
log-name2(max-size2),...log-namei(max-sizei) ,
...log-namen(max-sizen))]
DBset-name The logical name of the DBset being used (e.g., DBALL).
log-namei The i-th logical name for the DBset-name referenced in this
INIT statement. You can have up to 20 logical names for
each DBset (1 < n < 20 ). An ASSIGN statement may be used
to attach a physical file to the i-th logical file.
max-sizei The maximum allowable size that may be written to the i-th
file.
Example A
The following statement creates the DBALL DBset with a logical name of DBALL and
the maximum size of 50,000 NX Nastran blocks instead of 250,000 blocks, which is
the default value for DBALL:
Assuming that your input file is called “stat1.dat”, the physical file has the name
stat1.DBALL unless an ASSIGN statement is also used.
The following statement creates the DBALL DBset with logical names of DB1 and
DBTWO:
INIT DBALL LOGICAL=(DB1(35000),DBTWO(60000))
The two physical files stat1.DB1 and stat1.DBTWO are created with a maximum of
35,000 and 60,000 NX Nastran blocks, respectively.
Format (Simplified) for the SCRATCH DBset

INIT SCRATCH [LOGICAL=(log-name1(max-size1) ,
log-name2(max-size2),...log-namei(max-sizei)) ,
SCR300=(log-namei+1(max-sizei+1),...log-namen(max-sizen))]
Log-name1 through log-namei are allocated for regular scratch files as temporary
workspace. This space is not released until the end of the job. SCR300 is a special
keyword that indicates the log-names are members reserved for DMAP module
internal scratch files. The space occupied by these files is for the duration of the
execution of the module. This space is released at the end of the module execution.
You can have up to a combined total of 20 logical names for the SCRATCH DBset (1
< n < 20).
CHAPTER 17 543
Database Concepts
Example B
The following statement creates the SCRATCH DBset with logical names of SCR1,
SCR2, SCRA, and SCRB:
INIT SCRATCH LOGICAL=(SCR1(150MW),SCR2(100MW)) ,

SCR300=(SCRA(250MW),SCRB(300MW))
The two physical files stat1.SCR1 and stat1.SCR2 are created with maximum sizes
of 150 and 100 megawords, respectively. These two files are regular scratch files.
Two additional physical files stat1.SCRA and stat1.SCRB are created with
maximum sizes of 250 and 300 megawords, respectively. These last two files are
SCR300-type files.
Using the ASSIGN Statement

Use ASSIGN to assign a physical filenames to logical filenames or special
FORTRAN files that are used by other FMS statements or DMAP modules.
Format (Simplified) to Assign Logical Files

ASSIGN log-namei=’filenamei’ [TEMP DELETE]
log-namei The i-th logical name for the DBset created by the INIT
statement.
TEMP Requests that filenamei be deleted at the end of the job.
DELETE Requests that filenamei be deleted if it exists before the start
of the run. This is optional; however, if this option is not used
and the file exists prior to the current run, then the job may
fail with the following messages:
*** USER FATAL MESSAGE 773 (DBDEF)

THE FOLLOWING PHYSICAL FILE ALREADY EXISTS.
LOGICAL NAME = xxxx
PHYSICAL FILE = yyyy
USER INFORMATION: NO ASSOCIATED DEFAULT FILES OR ASSIGNED DBSETS CAN
EXIST PRIOR TO THE DATA BASE INITIALIZATION RUN.
USER ACTION: DELETE THIS FILE AND RESUBMIT THE JOB.
DELETE is not a suggested option if you are using RESTART since you can delete
your database inadvertently. Manual deletion of unwanted databases is a safer
approach.
Example C
The following statements create a logical name DB1 for the DBset DBALL in the
current directory:
ASSIGN DB1=’sample.DB1’
INIT DBALL LOGICAL=(DB1(50000))
The physical file sample.DB1 is created in this case. Without the ASSIGN statement,
the physical filename created is called stat1.DB1, assuming once again that your
input file is called stat1.dat.
The following statements create the two logical names DB1 and DB2 for the DBset
DBALL:
ASSIGN DB1=’/mydisk1/se/sample.DB1’
ASSIGN DB2=’/mydisk2/sample.DB2’
INIT DBALL LOGICAL=(DB1(50000),DB2(40000))
DB1 points to a physical file called sample.DB1 that resides in the file system
(directory) /mydisk1/se. DB2 points to a physical file called sample.DB2 that resides
in the file system (directory) /mydisk2.
Format (Simplified) to Assign FORTRAN Files

ASSIGN log-key=’filenamef’ [STATUS={new,old},UNIT=u ,
FORM={FORMATTED,UNFORMATTED},DELETE]
log-key This is the logical keyword for the FORTRAN file being assigned.
This file may already exist. The default value depends on the
keyword. Acceptable keywords are
DBC,DBMIG,INPUTT2,INPUTT4,OUTPUT2,OUTPUT4,
DBUNLOAD,DBLOAD, and USERFILE.
See the NX Nastran Quick Reference Guide for detailed descriptions of these keywords
and their default values.
filenamef The physical name of the FORTRAN file.

STATUS Specifies whether the FORTRAN file is created (STATUS = new)
or is an existing file (STATUS = old).
UNIT Specifies the FORTRAN unit (e.g., UNIT = 12).
CHAPTER 17 545
Database Concepts
FORM Specifies whether the file written is in ASCII (FORM =

FORMATTED) or binary (FORM = UNFORMATTED) format.
Note that if the log-key is OUTPUT2, the FORMATTED output is
in compressed ASCII format.
DELETE Requests that filenamef be deleted, if it exists before the start of
the run.
Example D
The following example creates a new FORTRAN file to be used for OUTPUT2
operations:
ASSIGN OUTPUT2=’sample.out’,STATUS=NEW,UNIT=11,FORM=FORMATTED
This FORTRAN file is in compressed ASCII format with a filename of sample.out

and is assigned to unit 11. Note that a compressed ASCII file can be transferred
directly across machines, but it should not be edited.
Using the EXPAND Statement

Expand concatenates files into an existing DBset in order to increase the allowable
disk space available for NX Nastran. The EXPAND statement is normally used in a
restart run when you exceed the disk space allocation in your previous run.
Format
EXPAND DBset-name LOGICAL=(log-namei(max-sizei),... )
DBset-name The logical name of the DBset to be expanded by the addition of

new members to an existing DBset previously defined with an
INIT statement.
log-namei The logical name of the i-th member of the DBset. An ASSIGN
statement should be used to point this logical name to a physical
file.
max-sizei The maximum size of the i-th member.
Only one EXPAND statement is allowed per run.
Example E
The original run creates a database with the name stat1.DBALL. However, this
database was filled and the job failed with the following error messages in the .F06
file:
*** USER FATAL MESSAGE 1012 (GALLOC)

DBSET DBALL IS FULL AND NEEDS TO BE EXPANDED.
For small to medium problems, it is best to rerun the job from the beginning with a
larger file allocation. For large problems, if rerunning the job is not practical, then the
database can be expanded with the following FMS statements:
RESTART
ASSIGN MASTER=’stat1.MASTER’
ASSIGN DBADD=’morespace.DB’
EXPAND DBALL LOGICAL=(DBADD(50000))
These statements assign an additional member, with a logical name of DBADD, to

the existing DBset DBALL. This member points to a new physical file called
morespace.DB, which may contain up to a maximum of 50,000 NX Nastran blocks.
You are restarting from “stat1.MASTER” in this case.
The EXPAND statement cannot be used for the scratch files since they are deleted at
the end of each job.
Using the RESTART Statement

RESTART allows you to continue from the end of the previous run without solving
the problem from the beginning.
Format
RESTART [PROJECT=’proj-ID’, VERSION={version-ID,LAST},
{KEEP,NOKEEP}]
proj-ID Project identifier used in the original run. It can have up to 40

characters. This option is optional and is normally not used.
The default proj-ID is blank.
version-ID The version number that you are restarting from.
CHAPTER 17 547
Database Concepts
KEEP If this option is used, then the version that you are restarting
from is also saved at the end of the current run.
NOKEEP If this option is used, then the version that you are restarting
from is deleted at the end of the current run.
Example F
The following statement causes the current run to use the last version in the
database for the restart:
RESTART
At the end of the run, this last version that you are restarting from is deleted from
the database. This is probably the most commonly used form for RESTART.
The following statement instructs the current run (version 6 or higher) to use
version 5 in the database for restart:
RESTART VERSION=5,KEEP
At the end of the run, version 5 is also retained in the database. This format is used
most often when you want to ensure that a specific version is saved in the database
(e.g., a large run from which you may want to request additional data recovery in
the future).
The following statement specifies that the current run (version 4 or higher with a
proj-ID of xyz) uses version 3 with a proj-ID of xyz in the database for restart:
RESTART PROJ=’xyz’ VERSION=3
At the end of the run, version 3 with a proj-ID of xyz is deleted from the database.
Using the DBCLEAN Statement

DBCLEAN deletes unwanted versions from the database. The DBCLEAN
statement does not reduce the size of the database. This statement deletes the data
blocks for the specified versions to allow other data blocks to reuse this space for
subsequent restarts without potentially increasing the database size. Up to ten
DBCLEAN statements can be used in the FMS for each run.
Format
DBCLEAN VERSION={version-ID,*} [PROJECT={’project-ID’,*}]
version-ID The version number that you want to remove from the database.
project-ID The identifier of the project to be deleted. The default is blank,
which is the most commonly used form.
* This is a wildcard command. It removes all versions or projects
from the database.
Example G
The following statements delete versions 3 and 6 with a blank proj-ID from the
database in the current run:
DBCLEAN VERSION=3
DBCLEAN VERSION=6
Using the DBDIR Statement

The DBDIR statement prints the database directory.
Format
DBDIR [VERSION={version-ID,*}] [PROJECT={proj-ID,*}]
[FORMAT={format-no}]
version-ID Version number.

proj-ID Project number.
* This is a wildcard command. It refers to all versions or projects.
format-no This controls the type of directory information printed. The
value is the sum of the desired types listed below. The default
value is 63 (sum of first six values), which means the first six
types will be printed.
1 = Project version table

2 = NDDL data blocks
4 = NDDL parameters
8 = NDDL empty data blocks
16 = NDDL data block trailers
32 = NDDL path value table for NDDL entries
64 = Scratch data blocks
CHAPTER 17 549
Database Concepts
Example H
The following statement causes the printing of the database directory for all
versions of the current project-ID in the database: DBDIR
Using the INCLUDE Statement

The INCLUDE statement inserts an external file at the location where the INCLUDE
statement is used.
Inserts an external file at the location where this INCLUDE statement is used. It is
not a pure FMS statement because it can be used anywhere in the input file, not just
in the FMS section.
Format
Include ’filename’
filename Physical filename of external file to be inserted at this location.

The following run reads a file called sub1.dat with all the Case Control commands
contained in it:
Sol 101
cend
include ’sub1.dat’
begin bulk
$
include ’bulk1.dat’
include ’bulk2.dat’
$
$ rest of bulk data file
$
.
.
enddata
This run also brings two additional files (bulk1.dat and bulk2.dat) into the Bulk Data
Section. You may, for example, want to include all your grid entries in file bulk1.dat
and all your element connectivities in bulk2.dat. As you can see, the INCLUDE
statement can be a handy tool. For parametric studies, you can potentially save a
tremendous amount of disk space by using the INCLUDE statement instead of
having multiple files with duplicate input data.
17.10 Guidelines for Large Problems

In general, you can solve small to medium sized problems in NX Nastran with very
little knowledge of FMS statements and database structure. However, when you’re
solving large problems, knowledge of FMS statements and database structure can
help you to allocate and optimize your computer resources.
When running large problems, you should have some idea of the resource
requirements—both disk space and computer time—prior to engaging in the actual
analysis. A utility program called ESTIMATE provides a set of recommended
resource values for running a particular job.
See Also
• “ESTIMATE” of the NX Nastran Installation and Operations Guide
For most small to medium problems, the default values for all the file assignments
should be more than adequate. However, for a large problem, larger disk space
allocation may be necessary. Several examples are shown in this section that
illustrate various options and consequences with each option. The submittal
command for each example is shown followed by a partial listing of the
corresponding input file. In each case, it is assumed that you are running on a UNIX
machine with “nastran” as the name of the shell script executing NX Nastran.
Example I
The following submittal command and FMS statements allocate 300,000 and 5,000
NX Nastran GINO blocks for DBALL and MASTER, respectively:
nastran runa scr=no
$
$ filename - runa.dat
$
INIT SCRATCH,LOGICAL=(SCRATCH(200000)) ,
SCR300=(SCR300(200000))
$
$ THE REST OF YOUR INPUT FILE
$
Furthermore, the following physical files are created and saved in your current
directory:
runa.DBALL
runa.MASTER
CHAPTER 17 551
Database Concepts
These statements also allocate 200,000 and 200,000 NX Nastran GINO blocks for
SCRATCH and SCR300, respectively. These two files are allocated to the default
scratch directory during the course of the run and are deleted at the end of the run.
Since the database is saved, this job is restartable. Other than the allocation size of
the databases, this is identical to the default setup.
When using the “scr = yes” option on the submittal command, the permanent data
blocks that are normally written to the DBALL DBset are now written to the
SCRATCH DBset.
Example J
The submittal command and FMS statements below allocate 50, 500, and 800
megabytes for DBALL, SCRATCH, and SCR300, respectively. Furthermore, by
default, 5,000 NX Nastran GINO blocks are allocated each for the MASTER dbset.
During the course of the run, all of these files are allocated to your default scratch
directory. Because you are using the “scr=yes” option on the submittal command,
the permanent data blocks that are normally written to the DBALL DBset are now
written to the SCRATCH DBset. In other words, you must allocate the disk space to
the SCRATCH DBset, which you would normally allocate to the DBALL DBset,
when you use the “scr=yes” option on the submittal command. At the end of the
run, all of these files are deleted automatically; this action is different from the
previous versions of NX Nastran. Since the database is not saved, this job is not
restartable.
nastran runb scr=yes
$
$ filename - runb.dat
$
INIT DBALL,LOGICAL=(DBALL(50MB))
INIT SCRATCH,LOGICAL=(SCRATCH(500MB)) ,
SCR300=(SCR300(800MB))
$
$
Example K
If you have sufficient disk space in your current and default scratch file systems
(disk packs), the method shown in “Example A” on page 542 or “Example B” on
page 543 is the preferred procedure since it is the simplest method. However, if you
do not have sufficient disk space in these two file systems, but have other file
systems (e.g., /disk2, /disk3, and /disk4) mounted to your system, then the
procedure listed below can be used. You must have read and write privileges in
order to use these file systems. For this example, the three new file systems are
/disk2, /disk3, and /disk4.
nastran runc scr=no
$
$ filename - runc.dat
$
ASSIGN DBALL1=’runc.db1’
ASSIGN DBALL2=’/disk2/user_guide/statics/runc.db2’
ASSIGN SCRATCH2=’/disk3/user_guide/statics/runc.scrtch2’
ASSIGN SC3B=’/disk4/user_guide/statics/runc.sc3b’
$
INIT DBALL,LOGICAL=(DBALL1(200000KB),DBALL2(100000KB))
INIT SCRATCH,LOGICAL=(SCRATCH1(100000),SCRATCH2(100000)) ,
SCR300=(SC3A(150000),SC3B(50000))
$
$
The permanent database is broken up into two separate logical files (DBALL1 and
DBALL2). The above run allocates 200,000 and 100,000 kilobytes for DBALL1 and
DBALL2, respectively. The run also allocates 5,000 GINO blocks for the MASTER
dbset. DBALL1 is allocated to the current directory that you are running on with the
physical filename of “runc.db1". DBALL2 is allocated to the
“/disk2/user_guide/statics” directory with a physical filename of “runc.db2". The
following additional physical file is also created in your current directory:
runc.MASTER
These three files are saved at the end of the run and the job is restartable.
The regular SCRATCH and SCR300 files are each divided into two separate files.
Two logical files (SCRATCH1 and SCRATCH2) are assigned to the regular
SCRATCH file. SCRATCH1 is assigned to the default scratch directory requesting
100,000 GINO blocks. SCRATCH2 is assigned to the “/disk3/user_guide/statics”
directory with a filename of “runc.scrtch2" and requesting 100,000 GINO blocks. The
SCR300 file is also divided into two files (SC3A and SC3B). SC3A requests 150,000
blocks and assigns this disk space to the default scratch directory. SC3B requests
50,000 blocks and assigns this disk space to the “/disk4/user_guide/statics”
directory with a filename of “runc.sc3b”. All four of these scratch files are deleted at
the end of the run.
nastran rund scr=yes
CHAPTER 17 553
Database Concepts
Example L
Even though the setup in “Example D” on page 545 looks virtually the same as that
in “Example C” on page 544, the “scr=yes” keyword causes the permanent data
blocks-that are normally written to the DBALL DBset-to be written to the SCRATCH
DBset. In other words, except for the amount used for overhead, the space that you
have allocated to DBALL1 and DBALL2 is not used at all. You must adjust the disk
space allocation to the SCRATCH DBset that you normally would allocate to the
DBALL DBset when you use the “scr=yes” option on the submittal command. The
correct action in this case is to increase the disk space allocation for SCRATCH1 and
SCRATCH2, and decrease the disk space allocation for DBALL1 and DBALL2.
Once again, since the database is not saved, this job is not restartable.
$
$ filename - rund.dat
$
ASSIGN DBALL1=’rund.db1’
ASSIGN DBALL2=’/disk2/user_guide/statics/runc.db2’
ASSIGN SCRATCH2=’/disk3/user_guide/statics/runc.scrtch2’
ASSIGN SC3B=’/disk4/user_guide/statics/runc.sc3b’
$
INIT DBALL,LOGICAL=(DBALL1(200000KB),DBALL2(100000KB))
INIT SCRATCH,LOGICAL=(SCRATCH1(100000),SCRATCH2(100000)) ,
SCR300=(SC3A(150000),SC3B(50000))
$
$
BUFFSIZE
BUFFSIZE is the length of an NX Nastran buffer in terms of words (32- or 64-bit
words). It controls the physical record size for data storage/transfer that is
contained in many NX Nastran logical units. The default and maximum allowable
BUFFSIZE is machine dependent. The default value is recommended except for
large problems. Each 32-, and 64-bit word contains 4 and 8 bytes, respectively. This
feature can be invoked by including the following NASTRAN statement in your
input file.
NASTRAN BUFFSIZE=xxxxx
where xxxxx = ( n ⋅ disk-block-size ) + 1 < limit and n is a positive integer.

The acceptable values of disk-block-size and limit are machine dependent and are
defined in Table 17-1 for most machines.
A larger BUFFSIZE takes fewer requests of the computer operating system to

transfer the same amount of data, with each transfer involving a larger physical
record size. This can reduce the I/O time which, in turn, reduces the elapsed time.
In general, the effect on CPU time is insignificant. A larger BUFFSIZE decreases the
number of I/O operations but may increase the database size. Since each data block
uses at least one buffer and the read/write operation requires a minimum of one
buffer, the default BUFFSIZE is recommended except for large problems. See NX
Nastran Installation and Operations Guide for the default BUFFSIZE for various
machines.
Disk space allocation is affected by the BUFFSIZE if the disk space units are in terms
of GINO blocks. GINO blocks are related to words by the following equation:
Number of words = Number of GINO blocks ⋅ ( BUFFSIZE – 1 ) Eq. 17-1
“Example E” on page 546 illustrates the use of a non-default BUFFSIZE.
Example M
Assume the machine that you are running on has a default BUFFSIZE of 2049. If you
increase the BUFFSIZE to 17291, using Eq. 17-1, this example allocates the same
maximum amount of physical disk space as “Example A” on page 542, which uses
the default BUFFSIZE.
$
$ filename - rune.dat
$
NASTRAN BUFFSIZE=17291
INIT SCRATCH,LOGICAL=(SCRATCH(23691)),SCR300=(SCR300(23691))
$
$
Assuming you have disk space, is there any other reason why you would not want
to use a larger BUFFSIZE for a large problem? The answer is “no” for most machines,
provided you have sufficient memory available on your machine. For some
machines, optimum values are preselected, and you should use these default values.
See Reference 16. for further details regarding your particular machine. Total
memory allocation can be controlled by the “mem” keyword on the submittal line.
The following example allocates 100 mb of total memory for the run.
nastran stat1 mem=100mb

A portion of this total memory is allocated to the Executive System. The rest of the
memory is then available to the functional modules in NX Nastran. Assuming that
you are using the default SCRATCH(MEM) and BUFFERPOOL, the portion of the
memory that is available for your problem is denoted as “User OPENCORE
(HICORE)”, and can be estimated by Eq. 17-2.
User OPENCORE ≅ mem – ( BUFFSIZE ⋅ 148 ) – 89,315 Eq. 17-2

CHAPTER 17 555
Database Concepts
As you can see from Eq. 17-2, the amount of memory allocated to the Executive
System is a function of the BUFFSIZE. As the BUFFSIZE increases, a larger portion
of the total memory is allocated to the Executive System. A summary of the NX
Nastran memory utilization is printed near the beginning of the .F04 file. Some
typical examples of the memory summary tables are shown in Table 17-1,
Table 17-2, and Table 17-3. You can also let NX Nastran estimate the amount of
required memory by using the following submittal command:
nastran stat1 mem=estimate
Table 17-1 Memory Summary Table

(mem = 2MW, buffsize = 2049)

BUFFER POOL AREA (EXEC) = 76183 WORDS ( 37 BUFFERS)




18
CHAPTER
Inertia Relief in Linear Static Analysis
■ Introduction to Inertia Relief
■ Description of Inertia Relief Using PARAM,INREL,-1
■ Implementation of Inertia Relief Using PARAM,INREL,-1
■ Automatic Inertia Relief
18.1 Introduction to Inertia Relief

Inertia relief is an advanced option that allows you to simulate unconstrained
structures in a static analysis. Typical applications of inertia relief include modeling
an aircraft in flight, an automobile on a test track, or a satellite in space. The inertia
relief method is a tool that you may not be familiar with. To help understand the
method, the first section provides a brief description of inertia relief. The last section
shows its implementation within NX Nastran.
Although the internal processing is somewhat complex, the interface to the inertia
relief option is quite straightforward. There are two ways to invoke the inertia relief.
• The first method is to specify the SUPORT entry explicitly by including

“PARAM,INREL,-1” in the Bulk Data Section.
• The second method (recommended) is to let NX Nastran select the SUPORT
degrees of freedom automatically by including “PARAM,INREL,-2” in the
Bulk Data Section.
An optional “PARAM,GRDPNT,x”— where x is a grid point ID— can be specified in

the model. If “PARAM,GRDPNT,x” is used, the loads and accelerations will be
summed about this point. If “PARAM,GRDPNT,x” is not specified, then the loads
and accelerations will be summed about the origin of the basic coordinate system.
18.2 Description of Inertia Relief Using PARAM,INREL,-1

Static analysis by the finite element method assumes that the model contains no
mechanisms and may not move as a rigid body (strain free). If either of these
conditions exists in a conventional finite element analysis, the stiffness matrix for the
model becomes singular. When NX Nastran attempts to decompose a singular
matrix, a fatal message or unreasonable answers result.
Consequently, you can’t perform conventional finite element static analysis on

unconstrained structures. However, a method called inertia relief is provided in NX
Nastran for analyzing these conditions. A simple description of inertia relief is that
the inertia (mass) of the structure is used to resist the applied loadings, that is, an
assumption is made that the structure is in a state of static equilibrium even though
it is not constrained. Two examples are a spacecraft in orbit or an aircraft in flight.
In these cases, the structure is in state of static equilibrium, although it is capable of
unconstrained motion.
To invoke inertia relief, you must provide a SUPORT Bulk Data entry with a list of
up to six non-redundant degrees of freedom that describe the possible unconstrained
motion. The easiest way to describe how to use the SUPORT entry in static analysis
is if you hold the SUPORT degrees of freedom constrained, there is no possible rigid
CHAPTER 18 559
body motion. If all possible rigid body motion is not described on the SUPORT
entry, then the stiffness matrix is singular, and the problem either fails in
decomposition or gives unreasonable answers.
When you specify inertia relief, NX Nastran calculates the forces that result from a
rigid body acceleration about the point specified on the “PARAM,GRDPNT,x” in
the specified directions. NX Nastran then calculates the summation of all applied
loadings in the same directions. Accelerations are applied to the structure in the
appropriate directions to “balance” the applied loadings. The structure is now in a
state of static equilibrium, i.e., the summation of all applied loads is 0.0. Since the
problem is not constrained, rigid body displacement is still possible.
NX Nastran next constrains the SUPORT degrees of freedom to a displacement of

0.0 and provide the relative motion of all other grid points with respect to that
reference point. Hence, the term “reference” degree of freedom is used to describe
the SUPORT degrees of freedom in NX Nastran. The set of degrees of freedom
described on the SUPORT entry belong to the r-set or reference set for the solution.
The computed solution is the correct one, and it is relative to any rigid body motion
that is occurring. A simple way to think of this is that the solution coming from NX
Nastran represents the deformation of the structure you would see if you were
standing at the SUPORT degrees of freedom.
18.3 Implementation of Inertia Relief Using

PARAM,INREL,-1
Solution 101 requires that the model does not contain mechanisms. You specify the
reference degrees of freedom using the SUPORT entry as follows:
1 2 3 4 5 6 7 8 9 10
SUPORT ID1 C1 ID2 C2 ID3 C3 ID4 C4
Field Contents
IDi Grid or scalar point identification number.
Ci Component numbers.
In addition to the PARAM,INREL,-1 entry, a SUPORT entry is also needed. If
PARAM,GRDPNT,x is specified, the rigid body mass matrix is calculated about grid
point x; otherwise, the rigid body mass is calculated about the origin of the basic
coordinate system. The inertia relief method requires that a realistic mass
distribution exists, and the degrees of freedom listed on the SUPORT entry must be
connected elastically to the model in all specified degrees of freedom. (For example,
degrees of freedom 4, 5, and 6 on a grid point with only solid elements attached
cannot be used since solid elements have no stiffness in these degrees of freedom.)
Whenever a SUPORT entry is used in static analysis, the epsilon and strain energy
printed in the table from UIM 3035 should be checked. The values printed for epsilon
and strain energy in the UIM 3035 table should all be numeric zero. The strain energy
printed in this table for matrix KLR represents the strain energy in the model when
the one SUPORT degree of freedom is moved 1.0 unit, while all other SUPORT
degrees of freedom are constrained. If the SUPORT degrees of freedom are properly
specified, the model should be capable of rigid body motion (strain-free) with no
strain energy.
The values printed for the strain energy indicate the ability of the model to move as
a rigid body. These values should always be checked. If the structure is not
constrained, the values should be numeric zero, but roundoff almost always results
in a small nonzero value. Acceptable values are a function of the units, size of the
structure, and precision of the hardware; therefore, a recommended value is not
provided in this user’s guide.
An additional feature allowed in Solution 101 is the solution of a problem under

uniform acceleration. This problem is posed using the DMIG,UACCEL entry in
addition to the previously mentioned requirements. In this case, uniform
accelerations are applied to the model, and the solution is found. Uniform
accelerations are useful for situations, such as spacecraft liftoff and landing loadings,
which are often specified as static accelerations.
As an example of inertia relief, consider the three CBEAM model shown in

Figure 18-1.
1 2 3 4
10 in 10 in 10 in
Y
X
Figure 18-1 Inertia Relief Analysis of a CBEAM Model
The three CBEAM structure is to be analyzed as a free-free structure with a line load
acting in the Y-direction as shown and using inertia relief. To show the effect of the
SUPORT point, two runs are made, each with a different SUPORT grid point. The
www.cadfamily.cominput file is given in Listing 18-1. Note that the SUPORT information is shown for
CHAPTER 18 561
both runs; however, one set is commented out. For the first run, “inertia1.dat”, the
SUPORT point is located at grid point 1. All six degrees of freedoms for the
SUPORT point are placed on one grid point as required for Solution 101. The
parameter INREL is set to –1, and the parameter GRDPNT is set to the SUPORT
point, which is grid point 1 in this case.
Listing 18-1 Three CBEAM Model with Inertia Relief
$ FILENAME - INERTIA1.DAT
ID LINEAR,INERTIA
SOL 101
TIME 5
CEND
TITLE = BAR WITH SUPORT ENTRY
LOAD = 1
DISP = ALL
SPCF = ALL
STRESS = ALL
BEGIN BULK
PARAM POST 0
$
$
$ SUPORT ENTRY FOR INERTIA1.DAT
$
SUPORT 1 123456
PARAM GRDPNT 1
PARAM INREL -1
$
$ SUPORT ENTRY FOR INERTIA2.DAT
$
$SUPORT 3 123456
$PARAM GRDPNT 3
$PARAM INREL -1
$
GRID 1 0.0 0.0 0.0
GRID 2 10. 0.0 0.0
GRID 3 20. 0.0 0.0
GRID 4 30. 0.0 0.0
$
CBEAM 1 1 1 2 1. 1. 0.0
CBEAM 2 1 2 3 1. 1. 0.0
CBEAM 3 1 3 4 1. 1. 0.0
$ .5 .5 .5 .5
PBEAM 1 1 1. .667 .167 .1
.5 .1 1.5 -.1 -.5 .1 -.5 -.1
PLOAD1 1 1 FY FR 0. 1000. 1. 1000.
PLOAD1 1 2 FY FR 0. 1000. 1. 1000.
PLOAD1 1 3 FY FR 0. 1000. 1. 1000.
$
MAT1 1 1.+7 .3 7.43E-3
ENDDATA
A partial listing of the output is shown in Figure 18-2.
1 BAR WITH SUPORT ENTRY OCTOBER 16, 2004 NX NASTRAN 4/ 9/04 PAGE 7
O U T P U T F R O M G R I D P O I N T W E I G H T G E N E R A T O R
0 REFERENCE POINT = 1
M O
* 2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 *
* 0.000000E+00 2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00 3.343500E+00 *
* 0.000000E+00 0.000000E+00 2.229000E-01 0.000000E+00 -3.343500E+00 0.000000E+00 *
* 0.000000E+00 0.000000E+00 0.000000E+00 1.858986E-01 0.000000E+00 0.000000E+00 *
* 0.000000E+00 0.000000E+00 -3.343500E+00 0.000000E+00 7.058500E+01 0.000000E+00 *
* 0.000000E+00 3.343500E+00 0.000000E+00 0.000000E+00 0.000000E+00 7.058500E+01 *
S
* 1.000000E+00 0.000000E+00 0.000000E+00 *
* 0.000000E+00 1.000000E+00 0.000000E+00 *
* 0.000000E+00 0.000000E+00 1.000000E+00 *
DIRECTION
X 2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00
Y 2.229000E-01 1.500000E+01 0.000000E+00 0.000000E+00
Z 2.229000E-01 1.500000E+01 0.000000E+00 0.000000E+00
I(S)
* 1.858986E-01 0.000000E+00 0.000000E+00 *
* 0.000000E+00 2.043250E+01 0.000000E+00 *
* 0.000000E+00 0.000000E+00 2.043250E+01 *
I(Q)
* 1.858986E-01 *
* 2.043250E+01 *
* 2.043250E+01 *
Q
* 1.000000E+00 0.000000E+00 0.000000E+00 *
* 0.000000E+00 1.000000E+00 0.000000E+00 *
* 0.000000E+00 0.000000E+00 1.000000E+00 *
0 RESULTANTS ABOUT 1 IN SUPERELEMENT BASIC SYSTEM COORDINATES.

0 OLOAD RESULTANT
SUBCASE/ LOAD
0 1 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 3.000000E+04 ---- 0.000000E+00 ---- 4.500000E+05
FZ ---- ---- 0.000000E+00 0.000000E+00 0.000000E+00 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 5.456968E-11
TOTALS 0.000000E+00 3.000000E+04 0.000000E+00 0.000000E+00 0.000000E+00 4.500000E+05
0 2 FX -2.229000E-01 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 0.000000E+00 ---- 0.000000E+00 ---- 0.000000E+00
FZ ---- ---- 0.000000E+00 0.000000E+00 0.000000E+00 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
TOTALS -2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
0 3 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- -2.229000E-01 ---- 0.000000E+00 ---- -3.343500E+00
FZ ---- ---- 0.000000E+00 0.000000E+00 0.000000E+00 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
TOTALS 0.000000E+00 -2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00 -3.343500E+00
0 4 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 0.000000E+00 ---- 0.000000E+00 ---- 0.000000E+00
FZ ---- ---- -2.229000E-01 0.000000E+00 3.343500E+00 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
Figure 18-2 Inertia Relief Output When Grid Point 1 Is the SUPORT Point
CHAPTER 18 563
TOTALS 0.000000E+00 0.000000E+00 -2.229000E-01 0.000000E+00 3.343500E+00 0.000000E+00

0 5 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 0.000000E+00 ---- 0.000000E+00 ---- 0.000000E+00
FZ ---- ---- 0.000000E+00 0.000000E+00 0.000000E+00 ----
MX ---- ---- ---- -1.858986E-01 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
TOTALS 0.000000E+00 0.000000E+00 0.000000E+00 -1.858986E-01 0.000000E+00 0.000000E+00
0 6 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 0.000000E+00 ---- 0.000000E+00 ---- 0.000000E+00
FZ ---- ---- 3.343500E+00 0.000000E+00 -7.058500E+01 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
TOTALS 0.000000E+00 0.000000E+00 3.343500E+00 0.000000E+00 -7.058500E+01 0.000000E+00
0 7 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- -3.343500E+00 ---- 0.000000E+00 ---- -7.058500E+01
FZ ---- ---- 0.000000E+00 0.000000E+00 0.000000E+00 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
TOTALS 0.000000E+00 -3.343500E+00 0.000000E+00 0.000000E+00 0.000000E+00 -7.058500E+01
0
FOR DATA BLOCK KLR
SUPPORT PT.NO. EPSILON STRAIN ENERGY EPSILONS LARGER THAN 0.001 ARE FLAGGED WITH ASTERISKS
1 2.3344220E-15 0.0000000E+00
2 2.3344220E-15 7.2759576E-12
3 2.3344220E-15 0.0000000E+00
4 2.3344220E-15 0.0000000E+00
5 2.3344220E-15 0.0000000E+00
6 2.3344220E-15 5.3551048E-09
0
INTERMEDIATE MATRIX ... QRR
COLUMN 1
1 2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 6
COLUMN 2
1 0.000000E+00 2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00 3.343500E+00 6
COLUMN 3
1 0.000000E+00 0.000000E+00 2.229000E-01 0.000000E+00 -3.343500E+00 0.000000E+00 6
COLUMN 4
1 0.000000E+00 0.000000E+00 0.000000E+00 1.858986E-01 0.000000E+00 0.000000E+00 6
COLUMN 5
1 0.000000E+00 0.000000E+00 -3.343500E+00 0.000000E+00 7.058500E+01 0.000000E+00 6
COLUMN 6
1 0.000000E+00 3.343500E+00 0.000000E+00 0.000000E+00 0.000000E+00 7.058500E+01 6
0
INTERMEDIATE MATRIX ... QRL
COLUMN 1
1 0.000000E+00 -3.000000E+04 0.000000E+00 0.000000E+00 0.000000E+00 -4.500000E+05 6
0
INTERMEDIATE MATRIX ... URA
COLUMN 1
1 0.000000E+00 1.345895E+05 0.000000E+00 0.000000E+00 0.000000E+00 8.246463E-
13 6

FOR DATA BLOCK KLL
LOAD SEQ. NO. EPSILON EXTERNAL WORK EPSILONS LARGER THAN 0.001 ARE FLAGGED WITH ASTERISKS
1 -1.6152625E-14 1.5617192E+02

1 G 0.0 0.0 0.0 0.0 0.0 0.0
2 G 0.0 -6.246877E-02 0.0 0.0 0.0 -1.249375E-02
3 G 0.0 -2.498751E-01 0.0 0.0 0.0 -2.498751E-02
4 G 0.0 -5.622189E-01 0.0 0.0 0.0 -3.748126E-02

1 G 0.0 -8.807738E-13 0.0 0.0 0.0 -6.462349E-27
0
S T R E S S E S I N B E A M E L E M E N T S ( C B E A M )
STAT DIST/
ELEMENT-ID GRID LENGTH SXC SXD SXE SXF S-MAX S-MIN M.S.-
T M.S.-C
0 1
1 0.000 0.0 0.0 0.0 0.0 0.0 0.0
2 1.000 0.0 0.0 0.0 0.0 0.0 0.0
0 2
2 0.000 0.0 0.0 0.0 0.0 0.0 0.0
3 1.000 0.0 0.0 0.0 0.0 0.0 0.0
0 3
3 0.000 0.0 0.0 0.0 0.0 0.0 0.0
4 1.000 0.0 0.0 0.0 0.0 0.0 0.0
(continued)
The GPWG (Grid Point Weight Generator) output is calculated with respect to the
grid point specified on the parameter GRDPNT. The total mass of the structure in
this model is 0.229, which is used by NX Nastran to develop the inertia loads.
Inspection of the GPWG output should be part of your routine model checkout.
CHAPTER 18 565
The OLOAD RESULTANT contains seven sections—consisting of seven lines per

section. The first section is the resultant of the applied loads about the GRDPNT,
which is the basic coordinate system in this case. The remaining six sections are the
loads necessary to impose unit acceleration about the SUPORT point. Subcase 2
corresponds to the X-direction; Subcase 3 corresponds to Y-direction, etc. The first
six lines of each section represent the detailed contributions, whereas the seventh
line represents the total.
Following the OLOAD output is User Information Message 3035 showing the strain
energy and epsilon due to the imposed unit accelerations about the SUPORT point.
As discussed earlier, the epsilons and strain energies should be small, which is the
case for this example. Small epsilons and strain energy tell you that you do not have
any unwanted constraints or poorly defined MPCs causing a constraint in your
model. Always inspect User Information Message 3035 to make sure that the
epsilons and strain energies are small.
The intermediate matrix QRR is printed following the UIM 3035. The QRR matrix
is the total rigid body mass of the total structure. It is a 6 x 6 matrix measured about
the PARAM,GRDPNT point in the global coordinate system. Masses on the scalar
points are not included. Following the QRR matrix is the QRL matrix. The QRL
matrix is the resultant of the “apparent reaction loads,” measured at the SUPORT
point. This resultant is equal and opposite to the OLOAD resultant shown earlier
in the output, if the SUPORT point is the same as the PARAM,GRDPNT point.
There is one column for each loading condition. The last matrix output is the URA
matrix, which is the rigid body acceleration matrix that is computed from the
applied loads.
The displacement and stress output shown is the standard output as requested
through the Case Control Section. Note that the displacement at the SUPORT point
is exactly 0.0. This SUPORT point should be 0.0 since this point is constrained for
the solution, and the forces of constraint should be numeric zero. The SPC forces
are shown in the output to confirm that the loads are balanced at the SUPORT point.
The displacement of all of the other points in the model are relative to the SUPORT
point.
An interesting effect occurs when the SUPORT point is changed in the previous
example. Suppose you change the location of the SUPORT point from 1 to 3 (see the
commented SUPORT entries in Listing 18-1).
The resulting displacement and stress output is shown in Figure 18-3.
O U T P U T F R O M G R I D P O I N T W E I G H T G E N E R A T O R
0 REFERENCE POINT = 3
M O
* 2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 *
* 0.000000E+00 2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00 -1.114500E+00 *
* 0.000000E+00 0.000000E+00 2.229000E-01 0.000000E+00 1.114500E+00 0.000000E+00 *
* 0.000000E+00 0.000000E+00 0.000000E+00 1.858986E-01 0.000000E+00 0.000000E+00 *
* 0.000000E+00 0.000000E+00 1.114500E+00 0.000000E+00 2.600500E+01 0.000000E+00 *
* 0.000000E+00 -1.114500E+00 0.000000E+00 0.000000E+00 0.000000E+00 2.600500E+01 *
S
* 1.000000E+00 0.000000E+00 0.000000E+00 *
* 0.000000E+00 1.000000E+00 0.000000E+00 *
* 0.000000E+00 0.000000E+00 1.000000E+00 *
DIRECTION
X 2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00
Y 2.229000E-01 -5.000000E+00 0.000000E+00 0.000000E+00
Z 2.229000E-01 -5.000000E+00 0.000000E+00 0.000000E+00
I(S)
* 1.858986E-01 0.000000E+00 0.000000E+00 *
* 0.000000E+00 2.043250E+01 0.000000E+00 *
* 0.000000E+00 0.000000E+00 2.043250E+01 *
I(Q)
* 1.858986E-01 *
* 2.043250E+01 *
* 2.043250E+01 *
Q
* 1.000000E+00 0.000000E+00 0.000000E+00 *
* 0.000000E+00 1.000000E+00 0.000000E+00 *
* 0.000000E+00 0.000000E+00 1.000000E+00 *
0 OLOAD RESULTANT
SUBCASE/ LOAD
0 1 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 3.000000E+04 ---- 0.000000E+00 ---- -1.500000E+05
FZ ---- ---- 0.000000E+00 0.000000E+00 0.000000E+00 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 5.456968E-11
TOTALS 0.000000E+00 3.000000E+04 0.000000E+00 0.000000E+00 0.000000E+00 -1.500000E+05
0 2 FX -2.229000E-01 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 0.000000E+00 ---- 0.000000E+00 ---- 0.000000E+00
FZ ---- ---- 0.000000E+00 0.000000E+00 0.000000E+00 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
TOTALS -2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00
0 3 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- -2.229000E-01 ---- 0.000000E+00 ---- 1.114500E+00
FZ ---- ---- 0.000000E+00 0.000000E+00 0.000000E+00 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
TOTALS 0.000000E+00 -2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00 1.114500E+00
0 4 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 0.000000E+00 ---- 0.000000E+00 ---- 0.000000E+00
FZ ---- ---- -2.229000E-01 0.000000E+00 -1.114500E+00 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
TOTALS 0.000000E+00 0.000000E+00 -2.229000E-01 0.000000E+00 -1.114500E+00 0.000000E+00
0 5 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 0.000000E+00 ---- 0.000000E+00 ---- 0.000000E+00
FZ ---- ---- 0.000000E+00 0.000000E+00 0.000000E+00 ----
MX ---- ---- ---- -1.858986E-01 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
CHAPTER 18 567
TOTALS 0.000000E+00 0.000000E+00 0.000000E+00 -1.858986E-01 0.000000E+00 0.000000E+00

0 6 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 0.000000E+00 ---- 0.000000E+00 ---- 0.000000E+00
FZ ---- ---- -1.114500E+00 0.000000E+00 -2.600500E+01 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
TOTALS 0.000000E+00 0.000000E+00 -1.114500E+00 0.000000E+00 -2.600500E+01 0.000000E+00
0 7 FX 0.000000E+00 ---- ---- ---- 0.000000E+00 0.000000E+00
FY ---- 1.114500E+00 ---- 0.000000E+00 ---- -2.600500E+01
FZ ---- ---- 0.000000E+00 0.000000E+00 0.000000E+00 ----
MX ---- ---- ---- 0.000000E+00 ---- ----
MY ---- ---- ---- ---- 0.000000E+00 ----
MZ ---- ---- ---- ---- ---- 0.000000E+00
TOTALS 0.000000E+00 1.114500E+00 0.000000E+00 0.000000E+00 0.000000E+00 -2.600500E+01
0
FOR DATA BLOCK KLR
SUPPORT PT.NO. EPSILON STRAIN ENERGY EPSILONS LARGER THAN 0.001 ARE FLAGGED WITH ASTERISKS
1 8.8574680E-16 0.0000000E+00
2 8.8574680E-16 1.4551915E-11
3 8.8574680E-16 0.0000000E+00
4 8.8574680E-16 0.0000000E+00
5 8.8574680E-16 1.1641532E-10
6 8.8574680E-16 3.2596290E-09
0
INTERMEDIATE MATRIX ... QRR
COLUMN 1
1 2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 6
COLUMN 2
1 0.000000E+00 2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00 -1.114500E+00 6
COLUMN 3
1 0.000000E+00 0.000000E+00 2.229000E-01 0.000000E+00 1.114500E+00 0.000000E+00 6
COLUMN 4
1 0.000000E+00 0.000000E+00 0.000000E+00 1.858986E-01 0.000000E+00 0.000000E+00 6
COLUMN 5
1 0.000000E+00 0.000000E+00 1.114500E+00 0.000000E+00 2.600500E+01 0.000000E+00 6
COLUMN 6
1 0.000000E+00 -1.114500E+00 0.000000E+00 0.000000E+00 0.000000E+00 2.600500E+01 6
0
INTERMEDIATE MATRIX ... QRL
COLUMN 1
1 0.000000E+00 -3.000000E+04 0.000000E+00 0.000000E+00 0.000000E+00 1.500000E+05 6
0
COLUMN 1
1 0.000000E+00 1.345895E+05 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 6

FOR DATA BLOCK KLL
LOAD SEQ. NO. EPSILON EXTERNAL WORK EPSILONS LARGER THAN 0.001 ARE FLAGGED WITH ASTERISKS
1 -2.6191993E-15 1.5617192E+02
(continued)

1 G 0.0 -2.498751E-01 0.0 0.0 0.0 2.498751E-02
2 G 0.0 -6.246877E-02 0.0 0.0 0.0 1.249375E-02
3 G 0.0 0.0 0.0 0.0 0.0 0.0
4 G 0.0 -6.246877E-02 0.0 0.0 0.0 -1.249375E-02
0
S T R E S S E S I N B E A M E L E M E N T S ( C B E A M )
STAT DIST/
ELEMENT-ID GRID LENGTH SXC SXD SXE SXF S-MAX S-MIN M.S.-T M.S.-C
0 1
1 0.000 0.0 0.0 0.0 0.0 0.0 0.0
2 1.000 0.0 0.0 0.0 0.0 0.0 0.0
0 2
2 0.000 0.0 0.0 0.0 0.0 0.0 0.0
3 1.000 0.0 0.0 0.0 0.0 0.0 0.0
0 3
3 0.000 0.0 0.0 0.0 0.0 0.0 0.0
4 1.000 0.0 0.0 0.0 0.0 0.0 0.0
(continued)
Note that the displacement vector changes as the displacements are now relative to
grid point 3, which is the new SUPORT point. However, the stresses in the elements
are the same because the stresses are based on the relative displacement between the
grid points, which is independent of the grid point used for the SUPORT point.
A final comment on this simple beam model: suppose you used CBAR elements
instead of CBEAM elements for this model. The run would fail because the CBAR
element does not have any torsional inertia mass, whereas the CBEAM element does.
This model is basically a one-dimensional structure so that the rotational inertia
about the X-axis is the mass inertia of the elements. When the inertia relief method
is used, the structure must have mass in all six directions. In general, however, there
are no restrictions on using a CBAR element with inertia relief. Most structures
include CBARs that are not co-linear. The example above was introduced merely to
emphasize the need to have mass in all six degrees of freedom.
Suppose you wish to impose a 10 g acceleration at grid point 2 of the beam structure
shown in Figure 18-1. This acceleration can be applied using the DMIG,UACCEL
Bulk Data entry as shown below.
1 2 3 4 5 6 7 8 9 10
DMIG UACCEL “0" “9" TIN
DMIG UACCEL L G1 C1 X1
G2 C2 X2 G3 C3 X3
CHAPTER 18 569
Field Contents
TIN Type of matrix being input.
L Load sequence number.
Gi Grid point identification number of a single reference point.
Ci Component number for Gi in the basic coordinate system.
Xi Value of enforced acceleration term in the basic coordinate
system.
The input file for this example is shown in Listing 18-2.
Listing 18-2 Imposing a 10 g Acceleration Using the PARAM,UACCEL
$ FILENAME - UACCEL
ID LINEAR,UACCEL
SOL 101
TIME 5
CEND
TITLE = BAR WITH UACCEL INPUT
OLOAD = ALL
DISP = ALL
BEGIN BULK
PARAM POST 0
$
$
$ SUPORT INFORMATION
$
SUPORT 2 123456
PARAM GRDPNT 2
PARAM INREL -1
DMIG UACCEL 0 9 1
DMIG UACCEL 1 2 2 3864.0
$
GRID 1 0.0 0.0 0.0
GRID 2 10. 0.0 0.0
GRID 3 20. 0.0 0.0
GRID 4 30. 0.0 0.0
$
CBEAM 1 1 1 2 1. 1. 0.0
CBEAM 2 1 2 3 1. 1. 0.0
CBEAM 3 1 3 4 1. 1. 0.0
$
PBEAM 1 1 1. .667 .167 .1
.1 .1 .1 -.1 -.1 .1 -.1 -.1
$
MAT1 1 1.+7 .3 7.43E-3
ENDDATA
The acceleration of 10 g is entered on the DMIG,UACCEL entry. When a DMIG

(Direct Matrix Input) entry is used, UACCEL, the first entry is the header entry,
which is denoted by the 0 in field 3. Field 4 must be a 9, and field 5 is either a 1 for
single precision input (the most common) or a 2 for double precision input. For this
example, the input is in single precision.
The second entry is where the applied acceleration is entered. Field 3 must be a 1,
indicating that this is the first load case (and the only load case in this model). The
grid point component and value of the enforced acceleration are entered in fields 6,
7, and 8. If acceleration is be to enforced in more than one component, then the
continuation entries are used. For this example, the acceleration of 3864. in/sec2 is
entered in field 8. The 3864.0 is 10 g in the English system-you must ensure that the
units are consistent.
When using the DMIG,UACCEL, the intermediate matrix URACCEL is output as

shown in Figure 18-4. This matrix is the rigid body acceleration that you input using
the DMIG,UACCEL entry. Also shown is the OLOAD output and the displacement
vector. As can be seen, the SUPORT point is constrained, which is what you would
expect, and the total load applied to the structure corresponds to the mass of the
structure (see the GPWG output in Figure 18-4) times the enforced acceleration.
Although not commonly done, you can perform multiple inertia relief analyses in a
single run. This feature can be activated with the SUPORT1 entries instead of the
SUPORT entry. Unlike the SUPORT entry, which is automatically activated, the
SUPORT1 entry must be called out by the SUPORT1 Case Control command in order
for it to be applied.
CHAPTER 18 571
O U T P U T F R O M G R I D P O I N T
W E I G H T G E N E R A T O R
REFERENCE POINT = 2
M O
* 2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 *
* 0.000000E+00 2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00 1.114500E+00 *
* 0.000000E+00 0.000000E+00 2.229000E-01 0.000000E+00 -1.114500E+00 0.000000E+00 *
* 0.000000E+00 0.000000E+00 0.000000E+00 1.858986E-01 0.000000E+00 0.000000E+00 *
* 0.000000E+00 0.000000E+00 -1.114500E+00 0.000000E+00 2.600500E+01 0.000000E+00 *
* 0.000000E+00 1.114500E+00 0.000000E+00 0.000000E+00 0.000000E+00 2.600500E+01 *
S
* 1.000000E+00 0.000000E+00 0.000000E+00 *
* 0.000000E+00 1.000000E+00 0.000000E+00 *
* 0.000000E+00 0.000000E+00 1.000000E+00 *
DIRECTION
X 2.229000E-01 0.000000E+00 0.000000E+00 0.000000E+00
Y 2.229000E-01 5.000000E+00 0.000000E+00 0.000000E+00
Z 2.229000E-01 5.000000E+00 0.000000E+00 0.000000E+00
I(S)
* 1.858986E-01 0.000000E+00 0.000000E+00 *
* 0.000000E+00 2.043250E+01 0.000000E+00 *
* 0.000000E+00 0.000000E+00 2.043250E+01 *
I(Q)
* 1.858986E-01 *
* 2.043250E+01 *
* 2.043250E+01 *
Q
* 1.000000E+00 0.000000E+00 0.000000E+00 *
* 0.000000E+00 1.000000E+00 0.000000E+00 *
* 0.000000E+00 0.000000E+00 1.000000E+00 *

COLUMN 1
1 0.000000E+00 3.864000E+03 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 6
1 G .0 -7.547017E-03 .0 .0 .0 1.076069E-03
2 G .0 .0 .0 .0 .0 .0
3 G .0 -3.340174E-02 .0 .0 .0 -5.380345E-03
4 G .0 -9.475221E-02 .0 .0 .0 -6.456414E-03
L O A D V E C T O R
1 G .0 -1.435476E+02 .0 .0 .0 .0
2 G .0 -2.870952E+02 .0 .0 .0 .0
3 G .0 -2.870952E+02 .0 .0 .0 .0
4 G .0 -1.435476E+02 .0 .0 .0 .0
Figure 18-4 Partial Output of the UACCEL Example
18.4 Automatic Inertia Relief

In the previous sections, manual inertia relief is discussed where the SUPORT entry
must be specified explicitly along with PARAM,INREL,-1. The SUPORT entry lists
the DOFs in the r-set, the reference set. Some skill is required to use this entry
correctly. When the r-set is chosen such that structure is not constrained in a
statically determinate manner, or if the stiffness attached to these points is
inadequate, the rigid body mode shapes may be poor, leading to low accuracy in
inertial load calculations. A poor static analysis solution may also result from the
poor constraints, independent of the loading problems. Both effects may result in
poor overall accuracy. Poor results can occur with little warning when the SUPORT
entry is misused.
The automatic inertia relief method is the recommended inertia relief method. With
the automatic method, the specification of the SUPORT entry is no longer needed. To
turn it on, simply add PARAM,INREL,-2 to the input file. The reference frame is
selected automatically, in a manner that poor solutions are unlikely because of the
choice of reference frame variables.
The constraints associated with the reference frame are distributed to all points with
mass. This means that structures with modeling errors, such as a region of elements
left out through oversight, will still give reasonable results that can aid in diagnosing
the modeling errors. A model in development may contain many disjoint parts,
inadvertently. Any part with three non co-linear points with mass will be adequately
constrained for solution.
In the manual support option, the free stiffness matrix Kaa is constrained by
removing the r-set DOFs from the a-set and imposing zero motion on them,
[ K aa ] ⋅ [ u a ] = { P a } Eq. 18-1
[ ur ] = { 0 } Eq. 18-2
With the automatic inertia relief, the u r variables are no longer a subset of the u a
variables. The constraint equation (Eq. 18-2) is replaced by a more general constraint
equation,
T
[ M aa ⋅ D a6 ] ⋅ { u a } = { u r } = { 0 } Eq. 18-3
The variables u r are a set of generalized coordinates that represent the average
motion of the u a variables, as weighted by the mass matrix. The zero value for this
equation implies that the average motion of all a-set points is zero, although the
displacement variables are free to move relative to this least-squares fit reference
frame.
The manual style of inertia relief allows use of PARAM,GRDPNT to define the origin
used in determining the shape functions D a6 . The presence of this parameter causes
the output of the Grid Point Weight Generator (GPWG) table, which lists the c.g.
location of the model, its moments of inertia, and other related data. This point in
space is used as the reference point when computing rigid body mass and moments
of inertia, and rigid body accelerations. If the parameter is not present, the basic
CHAPTER 18 573
origin is used, in the basic coordinate system. It is good practice (but not required
by NX Nastran) to use the same grid point on PARAM,GRDPNT and on the
SUPORT entry.
Unlike the manual option, PARAM,GRDPNT is not used as the inertia relief
reference point for the automatic support option. The basic origin is used. If the
PARAM,GRDPNT is present, it causes the generation of the GPWG table, but does
not influence the inertia relief calculations in any way.
The constraint forces caused by the distributed constraints are printed with the
SPCFORCE case control command. They are computational zeros for a well-
defined model. PARAM,TINY (default value is 1.E-3) is traditionally used to
discard element strain energies with values less than “tiny”. The same parameter
requests that small SPC forces of the reference frame constraint type be replaced
with binary zeros, for the automatic support option only. This filtering is not used
on conventional SPC forces from selected SPC entries or AUTOSPC, and manual
support DOFs.
Limitations for the Automatic Inertia Relief Method

The following features aren’t supported for the Automatic Inertia Relief Method:
• The DMIG,UACCEL option

• The Iterative Solver
EXAMPLE
Let’s revisit the CBEAM model in Figure 18-1 using the automatic inertia relief
method. This is accomplished by removing the SUPORT entry and replacing
PARAM,INREL,–1 with PARAM,INREL,–2. Figure 18-5 summarizes the
displacements, spcforces, and element stresses output.
The spcforces and stresses are both zeros, similar to the other two runs. As for the
displacements, the mass-weighted average motion of all a-set points must be zero.
Since the mass is evenly distributed, u r can be calculated for the T2 components as
follows:
U r = { 0.5 ⋅ ( – 8.329169E-2 ) + 1.0 ⋅ ( 4.164584E-2 ) + 1.0 ⋅ ( 4.164584E-2 ) + 0.5 ⋅ ( – 8.329169E-2 ) } = 0
Eq. 18-4
See TAN 4002 for further details regarding the Automatic Inertia Relief Method.
F
i
g
u
r
e
1
1
-
5
.
A
b
r
i
e
d
Figure 18-5 Abridged Output Using the Automatic Inertia Relief Method
19
CHAPTER
Surface Contact for SOL 101
■ Introduction
■ Understanding Contact Regions - BSURF, BCPROP, BSURFS
■ Understanding Contact Region Parameters - BCRPARA
■ Understanding Contact Pairs - BCTSET
■ Understanding Source Regions and Target Regions
■ Understanding the Contact Control Parameters Used in SOL 101 -

BCTPARM
■ Tips for setting PENN and PENT
■ Including CGAP elements in a contact solution - system cell

OLDGAPS
19.1 Introduction
NX Nastran provides a surface to surface contact capability that you can use with the
SOL 101 linear statics solution. Including contact conditions in your bulk data deck
allows the SOL 101 solution to search and detect when element faces come into
contact. The software then creates contact elements, thus preventing the faces from
penetrating and allowing finite sliding with optional friction effects.
The solver uses pre-defined regions of element free faces to detect contact conditions
in the model. From each element free face, it projects a normal, then checks to see if
any of the normals intersect with other element free faces. A contact element is
created during the solution if:
• NX Nastran finds an intersection between element faces, and
• the distance between the two faces is equal to or less than a distance that
you specify.
Note: The term contact element is used to describe a transient element created by the
solver to detect and analyze contact. You cannot manually create an NX Nastran
element called a contact element, and it is not documented in the NX Nastran Element
Library.
CHAPTER 19 577
19.2 Understanding Contact Regions - BSURF,

BCPROP, BSURFS
A contact region is a collection of element free faces in a section of the model where
you expect contact to occur. These regions can be created using shell elements
(BSURF and BCPROP) and using solid element free faces (BSURFS).
• The BSURF entry is defined by its own unique ID and is a list or range of
shell element IDs to include in the region.
• The BCPROP entry is defined by its own unique ID and is a list of shell
element property IDs. Shell elements which use any of these listed
property IDs will be included in the region.
• The BSURFS entry is defined by its own unique ID and is a list of solid
element IDs each followed by 3 grid points defining which face of the 3-D
element to include in the contact region.
The IDs used in any of the region definitions above must be unique to all other
BSURF, BCPROP and BSURFS entries.
A contact surface can be defined as any type of shell or any face of a solid element.
Although parabolic faces with omitted midside nodes are permitted, their use could
affect accuracy.
19.3 Understanding Contact Region Parameters -

BCRPARA
Contact region parameters are defined using the BCRPARA bulk entry, which is
required when surface contact is defined with SOL 101. The CRID field on
BCRPARA must match the ID used on one of the BSURF, BCPROP and BSURFS
bulk entries in order to be considered by the solution. There are two contact region
parameters which can be used with SOL 101, the SURF and OFFSET fields.
• The SURF field is important if a region you’ve created is made of shell

elements. Shell elements have a top and bottom side associated with them.
The top side is determined by the order and direction of grid points on the
element’s definition and the right-hand rule. (See chapter 15 of the NX
Nastran User’s Guide for information on using consistent normals) When
shell elements are used as the source in a contact pair, by default, the solve
will use the top side to project the element normal while searching for the
target elements. In the case where the target is on the bottom side, you can
set the SURF option to “BOT”, and the normal will be projected from the
bottom side. Understanding shell element normals, and making sure they
are consistent in the source regions is very important to ensuring that
contact elements will be created as expected.
The figure below shows examples of how the SURF option is used to
modify the contact search projection direction when the source element
normals vary. The arrow on each face shows the element’s top side. Source
regions are shaded while target regions are white.
Note that the target region normals in these examples are valid regardless
of their top or bottom status, since a normal is not projected from the target
side of a pair.
• Use OFFSET field to account for a rigid layer which might occur between
two faces coming into contact. For example, suppose you have a model
which has two metal surfaces coming into contact, and one of these has a
ceramic coating. If the ceramic material stiffness is not significant enough to
be included in the analysis, it may not have been specifically modeled, but
the thickness it adds to the face of the metal may be important when
considering the contact problem. Add the thickness of the ceramic to the
OFFSET field, and it will be included by the contact portion of the solve.
You can also use the OFFSET field to analyze an interference fit problem if
unconnected elements are modeled coincident. The offset value in this
example can represent the theoretical interference of these faces.
19.4 Understanding Contact Pairs - BCTSET

A contact pair is a way to combine two contact regions, source and target, in which
contact will be analyzed during the solution. Each contact pair can have its own
unique friction value (if desired) and search distance.
CHAPTER 19 579
The BCTSET bulk entry is used to define each contact pair. Its CID field will need to
match the value of ‘n’ on the BCSET case control entry for the solution to recognize
this contact definition. The SIDi and TIDi fields refer to regions created by the
BSURF, BCPROP and BSURFS entries, and are used to define source and target
regions respectively for a pair. As many pairs as desired can be included on a single
BCTSET entry, and each pair can have a unique friction value (optional), a
minimum search distance, and a maximum search distance.
Enter the optional Coefficient of Friction field (FRICi) for each defined contact pair
if you expect finite sliding to occur in this region of the model. When contact is
detected, the solver uses this value to calculate any tangential contact forces by
multiplying the normal contact force by FRICi.
The minimum and maximum search distance fields (MINDi and MAXDi) define a
range in which the solver can initially determine if the distance between element
faces in a particular pair are within the threshold for creating contact elements.
These values are only used once, at the beginning of SOL 101, to determine where
contact elements need to be initially created. Recall that NX Nastran projects
normals from element faces and then checks to see if any of these normals intersect
with another element free face. If the projected normal intersects an element face,
and the distance between the two element faces is within the range defined in the
MINDi and MAXDi fields, a contact element is created. Since SOL 101 is used for
linear problems with small deflection, there are no geometry nonlinear updates
which occur. This includes updating the contact conditions, therefore the number of
contact elements created in this initial step will remain the same for the rest of the
solution. The minimum distance can be negative if there is an interference fit
condition modeled as overlapping surfaces.
19.5 Understanding Source Regions and Target Regions

It’s important to understand how contact elements are created when selecting
which region will be the source and which the target, since the two can be
interchangeable. The solver projects vector normals from the element integration
points from the source region to the target region. It then creates contact elements
when these normals intersect elements in the target region and are within the search
distance criteria for the contact pair. This means that when the two regions of a pair
do not have corresponding one-to-one elements, the number of contact elements
that the solver creates can change depending on which region it projects the
elements from and which region it projects them to.
In general, of the two contact regions you use for the pair, choose the one with the
finer mesh for the source region. When the source and target regions have different
mesh densities, more elements on the source region will mean that more contact
elements are created, which will produce a more accurate solution.
For example, the two regions below are both composed of linear shells. The source
region (A) has one element and the target region (B) has four elements.
When creating the contact elements between these regions, the software projects
contact elements from the single element on the source region to the four elements in
the target region. This results in the creation of a single contact element.
However, if you were to use the region with four elements as the source (C, below)
and the region with one element as the target (D), the solution will create 4 contact
elements.
19.6 Understanding the Contact Control Parameters

Used in SOL 101 - BCTPARM
The contact control parameters on the BCTPARM bulk entry (optional) can help you
adjust the contact algorithm when you are having problems getting a solution to
converge and complete, or when the contact results are not as expected (see “Tips for
setting PENN and PENT” in section 19.7). To understand these parameters and how
the solution uses them, it is helpful to understand the contact algorithm. For most
solutions, the default settings are appropriate, thus the BCTPARM entry is not
required.
CHAPTER 19 581
• MAXF: The contact algorithm iterates in an inner and outer loop. This
parameter specifies the maximum number of iterations of the inner or
force loop. The inner loop forces a condition of zero penetration between
the contacting bodies. (default = 10)
• MAXS: Specifies the maximum number of iterations for the outer or status
loop. This loop determines which contact locations (or contact elements)
are active. (default = 20)
• PENN: Controls the normal penetration stiffness (also called penalty
factors) of the faces coming into contact. Generally, a larger value reduces
penetration and speeds convergence. However, convergence may not
occur if this value is too large. (default = 10.0) (See “Tips” below)
• PENT: Controls the convergence of friction forces when the friction is non-
zero. Generally, this value should be 10 to 100 times smaller than the
normal penalty factor (PENN). (default = 1.0) (See “Tips” below)
• CTOL: This is a contact convergence tolerance (Euclidean norm) found by
comparing changes in traction between the two contacting bodies. When
the convergence tolerance is less than the CTOL factor that is specified
here, the solve considers the contact algorithm to be converged. (default =
0.01)
• NCHG: Specifies a number of contact elements which can be active from
one iteration to the next, yet the software will consider the solution
converged. (default = 0)
• MPER: Sets the minimum percentage of contact elements considered
active. If you set this value to 0, only those contact elements initially in
contact are considered active. If you set this value to 100, all elements are
considered active, even if they’re initially open. (default = 100)
• SHLTHK: This is the shell thickness offset flag. If you set this value to 0, the
contact surface is assumed to be offset t/2 from the nodes defining a shell
element. If the FE mesh represents the outer surface, set this value to 1 so
the thickness offset will be ignored. (default = 0)
• RESET: This is a flag to indicate if the contact status for a specific subcase
is to start from the final status of the previous subcase. If you set this value
to 0, the contact status will start from a previous subcase. If you set it to 1,
the contact status will start from an initial state. (default = 0)
• AVGSTS: Determines the averaging method for contact pressure/traction

results. If you set this value to 0, the averaging of pressure/traction values
for a contact grid will include the results from ALL contact elements
attached to the grid regardless of whether they are active or inactive in the
contact problem. If you set this value to 1, the averaging of the
pressure/traction values for a contact grid will exclude those contact
elements which are not active in the contact solution. (default = 0)
• INIPENE: Controls definition of initial gap or penetration of the generated
contact elements. If you set this value to 0 or 1, the contact solution will use
the value calculated from the grid coordinates. If you set the value to 2, it
will initially work the same as above, but if penetration is detected, the
gap/penetration will be set to zero. If you set the value to 3, the
gap/penetration will be zero for all contact elements. Option 3 is very
useful when the contact surfaces are intended to be coincident but due to
irregularity in the meshes, some grids are not exactly on the surface. This
can lead to contact pressures that have a few hot spots rather than a smooth
distribution. This option will ensure that a smooth distribution is obtained
and in many cases the number of iterations will be reduced along with run
time. (default = 0)
19.7 Tips for setting PENN and PENT

• Penalty factors in the range of 1000 to 10,000 work well for most problems.
• Penalty factors in the range of 10 to 100 work best for coincident surfaces
with unmatched meshes. Lower penalty factors tend to smooth out
irregularities caused by the displacement incompatibilities of dissimilar
meshes. If contact stresses are important, you’ll obtain the best results if the
meshes on the source and target regions match.
• Very high penalty factors (in excess of 1E9) aren’t recommended and will
cause numerical problems, even for flat surfaces with very regular meshes.
• Very low penalty factors (less than 1) produce extremely slow convergence
rates and aren’t recommended.
19.8 Including CGAP elements in a contact solution -

system cell OLDGAPS
CGAP elements can be included in a model where contact is defined, and the system
cell (412) OLDGAPS can be used to control how CGAPs are treated in SOL 101.
CHAPTER 19 583
• If a CGAP is included in a model where contact is defined, SOL 101 will

treat it as a contact element, and thus it will be included in the contact
solution. If you don’t want the CGAP to be treated as a contact element and
want it handled as in previous releases as described in the NX Nastran
Element Library, set the system cell OLDGAPS = 1.
• If a CGAP is included in a model where contact is not defined (no BCSET
card), it will be treated as described in the NX Nastran Element Library,
regardless of the value of the system cell OLDGAPS. In other words, the
system cell OLDGAPS is only considered when contact is defined for SOL
101.
T
h
e
v
a
r
i
a
b
l
e
s
a
r
e
CHAPTER 19 585
20
CHAPTER
Plotting
■ Overview of Plotting in NX Nastran
■ Superelement Plotting
■ Post Processors
20.1 Overview of Plotting in NX Nastran

NX Nastran can generate the following types of plots:
• structural plots
• X-Y graphs of transient response or frequency response.
• V-F and V-G graphs for flutter analysis.
• Contour plots of displacements, temperature and stress on the structure.
Structural Plots
NX Nastran can generate the following types of structural plots:
• Undeformed geometric projections of the structural model.

• Static deformations of the structural model by either displaying the
deformed shape (alone or superimposed on the undeformed shape), or
displaying the displacement vectors at the grid points (superimposed on
either the deformed or undeformed shape).
• Modal deformations resulting from real or complex eigenvalue analysis by
the same options stated in 2 above. Complex modes for flutter analysis may
be plotted for any user-chosen phase lag.
• Deformations of the structural model for transient response or frequency
response by displaying either vectors or the deformed shape for specified
times or frequencies.
See Also
• “OUTPUT(PLOT) Commands” of the NX Nastran Quick Reference Guide
X-Y Plots
Requests for structure plots or X-Y plots are made in the Case Control Section by
submitting a structure plot request or an X-Y output request.
See Also
• “X-Y PLOT Commands” of the NX Nastran Quick Reference Guide

The optional PLOTID command is considered to be part of the plot request although
it must precede any OUTPUT(PLOT), OUTPUT(XYOUT) or OUTPUT(XYPLOT)
commands.
See Also
• “PLOTID” of the NX Nastran Quick Reference Guide
CHAPTER 20 589
Plotting
Plot requests are separated from Case Control by the OUTPUT(PLOT),

OUTPUT(XYPLOT) or OUTPUT(XYOUT) commands. Data above this command
(except PLOTID) will not be recognized by the plotter, even though it may have
the same name (for example, the SET command).
Flutter Analysis Plots

In NX Nastran, you can use the aeroelasticity capabilities to perform a flutter
analysis to determine the stability of an aeroelastic system. You can then create
graphs of your results.
See Also
• “Flutter Analysis” of the NX Nastran Aeroelastic Analysis User’s Guide

20.2 Superelement Plotting
Plotting can occur at four different places in the superelement solution sequences.
There are two plot commands, SEPLOT and SEUPPLOT, used with other Case
Control and PARAM commands to control the type of plot to be prepared. A flow
chart of the solution process is given in Figure 20-1, Figure 20-2, and Figure 20-3.
Undeformed structure plots are made early during Phase 0 in the structured
solution sequences and during Phase l in the unstructured solution sequences.
They can be made for either one superelement only, or for a superelement and all
of its upstream members, as controlled by PARAM,PLOTSUP (see Figure 20-1 or
Figure 20-2). Plots are made for superelements selected by the SEMG command
(see Figure 20-2, block I), and listed in a plot request headed by SEPLOT SElD,
where SEID is the superelement identification number. In the structured solution
sequences, the SEMG command is not required and all superelements which
appear on SEPLOT commands will be plotted (see Figure 20-1, block 0).
Undeformed plots can be used to check geometry and connectivity, and do not
require the presence of property or material commands. In the unstructured
solution sequences, a branch to the end of the loop immediately after the plot
module can be made by using the PARAM,PLOT,-1 command.
In dynamic analysis, solution set XY-plots are requested by use of SEPLOT 0 (see
Figure 20-3, block II). In the data recovery phase, XY-plots and deformed structure
plots for elements in one superelement only are requested by the SEPLOT
command (see Figure 20-3, block III). Deformed structure plots for a superelement
and all its upstream plots are requested with the SEUPPLOT command (see
Figure 20-3, block IV).
XY-plots and deformed structure plots are regarded as output requests and will
result in automatic execution of the data recovery loop as is required to produce
the plots requested. For example, the command SEUPPLOT 0 will result in data
recovery being performed on the entire model, even in the absence of any other
output requests. For SEUPPLOT requests, if SUBCOM or SYMCOM subcases are
used, each superelement must have identical SUBCOM and SYMCOM structure.
START
0. Bookkeeping and Control Case Control

Commands
PARAM,PLOTSUP,0 PARAM,PLOTSUP,-1
Prepare plotting data Prepare plotting SEPLOT SEID

for superelement and data for one
all upstream members superelement only
(default)
1. Undeformed Plots
I. Generation, Assembly,
Reduction Loop
Generate, assemble,
and reduce one
superelement
(Continued on Figure 20-2)
Figure 20-1 Phase 0 and I Superelement Plot Control In the

Structured Solution Sequences
CHAPTER 20 591
Plotting
I. Generation, Assembly, Case Control

Reduction Loop Commands
PARAM,PLOTSUP,0 PARAM,PLOTSUP,-1
Prepare plotting data Prepare plotting SET n=SEID(s)

for superelement and data for one SEMG=n
all upstream members superelement only SEPLOTSEID(s)
(default)
1. Undeformed Plots
PARAM,PLOT,-1
PARAM,PLOT,0
(Default)
Generate, assemble,
and reduce one
superelement
(Continued on Figure 20-3)
Figure 20-2 Phase l Superelement Plot Control In the

Unstructured Solution Sequences
II. Residual Structure Only Phase
Eigensolutions,
Forced Response
SEPLOT 0
2. Solution Set XY Plots
III. Data Recovery Loop
Recover interior
displacements
and stresses
SEPLOT SEID
3. XYPLOTS
Deformed structure
plots (local)
IV. Upstream Plot Loop
Assemble upstream
plot vectors
SEUPPLOT
SEID
4. Deformed structure
plots (upstream)
END
Figure 20-3 Phase II, Ill, and IV Superelement Plot Control in All Solution
Sequences
CHAPTER 20 593
Plotting
20.3 Post Processors

The plotps post processor reads plotting commands from a single NX Nastran
binary or neutral format plot file and produces a file which can be printed or
viewed on a PostScript device. Each image will be oriented to best correlate the
plot’s page size with the printable page size (assumed to be 7.5 inches wide and
10.0 inches tall). Command line arguments can be specified in any order,
command keywords can be abbreviated by truncation.
See Also
• “PLOTPS” of the NX Nastran Installation and Operations Guide
On UNIX systems, you can omit the plot file type(‘.plt’ for binary files and ‘.neu’
for neutral files).
Examples
• Translate a binary format plot file into PostScript:
plotps example.plt
• Translate a neutral format plot file into PostScript:
plotps example.neu
21
CHAPTER
Linear Buckling
■ Introduction to Linear Buckling
■ Finite Element Approach
■ Eigenvalue Extraction Methods
■ Linear Buckling Assumptions and Limitations
■ Buckling Examples
21.1 Introduction to Linear Buckling

In linear static analysis, a structure is normally considered to be in a state of stable
equilibrium. As the applied load is removed, the structure is assumed to return to its
original position. However, under certain combinations of loadings, the structure
may become unstable. When this loading is reached, the structure continues to
deflect without an increase in the magnitude of the loading. In this case, the structure
has actually buckled or has become unstable; hence, the term “instability” is often
used interchangeably with the term “buckling.”
Only linear buckling or elastic stability is considered in this user’s guide; in other
words, assume there is no yielding of the structure and the direction of the forces do
not change (i.e., follower force effects are ignored). Other assumptions of elastic
stability are discussed in “Linear Buckling Assumptions and Limitations” on
page 606.
This chapter is organized into the following sections:
• Finite element approach

• Eigenvalue extraction method
• Assumptions and limitations of linear buckling analysis
• Examples
21.2 Finite Element Approach
In finite element analysis, the problem of linear buckling is addressed by including
the effect of the differential stiffness to the linear stiffness matrix. The differential
stiffness results from including the higher-order terms of the strain-displacement
relationships. From a physical standpoint, the differential stiffness matrix represents
the linear approximation of softening (reducing) the linear stiffness matrix in the case
of a compressive axial load, and stiffening (increasing) the linear stiffness matrix in
the case of a tensile axial load. Since the differential stiffness either adds or subtracts
from the linear stiffness, often it is also called the incremental stiffness. The element
linear and differential stiffness matrices are denoted as [ k a ] i and [ k d ] i , respectively.
The differential stiffness matrix is a function of the geometry, element type, and
applied loads. A look at the differential stiffness matrix of a single planar bar element
as shown in Figure 21-1 shows how a linear buckling analysis is handled using a
finite element approach. For clarity, only the y and θ z degrees of freedom at each end
are retained for this example.
CHAPTER 21 597
Linear Buckling
x Pa
Figure 21-1 Single Planar Bar Element
From The Nastran Theoretical Manual, the differential stiffness for this planar bar
element can be represented as
6F x –Fx – 6F x –Fx
i i i i
----------- ----------- -------------- -----------
5l i 10 5l i 10
– F x 2l i F x Fx –li Fx
i i i i
----------- --------------- ------- ---------------
10 15 10 30
[ kd ] = Eq. 21-1
i
– 6F x Fx 6F x Fx
i i i i
-------------- ------- ----------- -------
5l i 10 5l i 10
–Fx – lF x Fx 2l i F x
i i i i
----------- ------------- ------- ---------------
10 30 10 15
where F x is the axial force in the CBAR element. In this case, P a (applied load) =
i
F x since there is only one element in the model and the applied load is in line with
i
the element axis. In the general case, F x is proportional to P a as long as the structure
i
remains linear; in other words, if P a is increased by scale factor a i , then F x also
i
increases by the same scale factor. The value “i” stands for the i-th element. Note
also that the differential stiffness matrix is dependent only on the element type, the
applied forces, and the geometry of the structure. This is the reason why the
differential stiffness is also often called the geometric stiffness matrix. Eq. 21-1 can,
therefore, be rewritten as follows:
6α i –αi – 6α i –αi
--------- -------- ------------ --------
5l i 10 5l i 10
– α i 2l i α i αi –li αi
-------- ------------ ------ ------------
10 15 10 30
[ kd ] = Pa = Pa [ kd ]i Eq. 21-2
i
– 6α i αi 6α i αi
------------ ------ --------- ------
5l i 10 5l i 10
–αi –li αi αi 2l i α i
-------- ------------ ------ ------------
10 30 10 15
One can view α i as the distribution factor of the applied load to the i-th element.
Each element in the structure that supports differential stiffness (see “Linear
Buckling Assumptions and Limitations” on page 606) has an element differential
stiffness matrix associated with it. Each of these differential stiffness matrices has a
scale factor similar to Eq. 21-2. In general, each αi is different for each element in the
structure. The value of each αi depends on the element type, the orientation of the
element relative to the overall structure, and the applied load.
The system linear stiffness matrix can then be represented as

n
[ Ka ] = ∑ kai
i
Similarly, the system differential stiffness matrix can be represented as

n
[ Kd ] = ∑ kdi
i
In general, the individual [ k d ] i is more complicated than Eq. 21-1; however, the
concept is the same. The overall system stiffness matrix is represented by Eq. 21-3.
[ K ] = [ Ka ] + [ Kd ] Eq. 21-3
The total potential energy is equal to
T T
[ U ] = 0.5 { u } [ K a ] { u } + 0.5 { u } [ K d ] { u } Eq. 21-4
In order for the system to achieve static equilibrium, the total potential must have a
stationary value; in other words, the relationship in Eq. 21-5 must be satisfied
CHAPTER 21 599
Linear Buckling
∂[U]
------------- = [ K a ] { u } + [ K d ] { u } = { 0 } Eq. 21-5
∂u i
where u i is the displacement of the i-th degree of freedom.
It is convenient to rewrite Eq. 21-5 as
[ [ Ka ] + P [ Kd ] ] { u } = { 0 } Eq. 21-6
a
where [ K d ] = P a [ Kd ] and P a is the applied load. In order for Eq. 21-6 to have a
non-trivial solution, the following relationship must be true:
[ Ka ] + Pa [ Kd ] = {0} Eq. 21-7
where || stands for the determinant of the matrix. Eq. 21-7 is only satisfied for
certain values of P a . These values of are the critical buckling loads.
A real structure has an infinite number of degrees of freedom. The finite element
model approximates the behavior of the structure with a finite number of degrees
of freedom. The number of buckling loads obtainable for your finite element model
is equal to the number of degrees of freedom of your model. In other words,
P cr = λi ⋅ Pa Eq. 21-8
i
Eq. 21-7 can, therefore, be rewritten as
[ Ka ] + λi [ K d ] = [0] Eq. 21-9
Eq. 21-9 is in the form of an eigenvalue problem. Once you obtain the eigenvalues
λ i , the buckling loads can then be obtained using Eq. 21-8. The values λ i are the
scale factors by which the applied load P a is multiplied to produce the critical
buckling loads P cr . As you can see from Eq. 21-8, the magnitude of the applied load
i
P a is arbitrary for arriving at the correct P cr . As an example, if P a is increased by a
i
factor of 10, then the calculated λ i values in Eq. 21-9 are reduced by a factor of 10; in
other words, their resulting products P cr remain the same.
i
In general, only the lowest buckling load is of any practical interest. The structure
will fail prior to reaching any of the higher buckling loads.
Solution 105. In NX Nastran you can solve a linear buckling problem by using
Solution 105 and following the procedure listed below.
1. Apply the static loads to the first n subcases (n is usually equal to one) and
treat them as static analysis. The distribution of element forces due to these
applied loads is generated internally. The actual magnitude of these
applied loads is not critical.
2. You can perform buckling analysis on any or all of the loading conditions
used in Step 1. One additional subcase is needed for each buckling analysis.
3. The n+1 to the n+m subcases must each request an eigenvalue method from
the Bulk Data Section to solve the eigenvalue problem shown in Eq. 21-9. In
this case, m is equal to the number of buckling analyses that you want to
perform. Each buckling subcase may call out a unique eigenvalue solution.
4. The differential stiffness matrix is automatically generated for each element
that supports differential stiffness. See “Linear Buckling Assumptions
and Limitations” on page 606 for a list of elements that support differential
stiffness.
5. You must then multiply the eigenvalues obtained in Step 3 by the
appropriate applied loads to obtain the buckling loads (Eq. 21-8) for each
buckling analysis.
6. Each subcase may have a different boundary condition.
A typical input file used to calculate the buckling loads is shown in Listing 21-1. In
most applications, only one static and one buckling analysis is performed per run.
Example 6 in “Buckling Examples” on page 607 contains an application of multiple
static and buckling analyses.
CHAPTER 21 601
Linear Buckling
Listing 21-1 Input File for a Typical Buckling Analysis
$
SOL 105
TIME 10
CEND
TITLE = SAMPLE INPUT FOR BUCKLING ANALYSIS
SPC = 10
DISP = ALL
$
$ STATIC SUBCASE TO GENERATE INTERNAL ELEMENT FORCES
$
SUBCASE 1
LOAD = 10
$
$ EIGENVALUE CALCULATION TO OBTAIN SCALE FACTORS BY WHICH
$ THE APPLIED LOAD(S) IS MULTIPLIED BY
$
SUBCASE 2
METHOD = 20
$
BEGIN BULK
$
$ APPLIED LOAD IN THE STATIC SUBCASE - SUBCASE ONE
$
FORCE,10,100,,-100.,1.0,0.,0.
$
$ LANCZOS EIGENVALUE METHOD REQUESTING THE LOWEST EIGENVALUE
$
EIGRL,20,,,1
$
$ BRING IN THE REST OF THE BULK DATA ENTRIES
$
include ’bulk.dat’
$
ENDDATA
You should also adhere to the following guidelines when solving a linear buckling
problem in NX Nastran.
• The Case Control Section must contain at least two subcases.

• A METHOD command must appear in each buckling subcase to select the
appropriate eigenvalue extraction method (EIGRL or EIGB entry) from the
Bulk Data Section. If there are two or more buckling subcases, then a
STATSUB = x command must be placed in each buckling subcase to select
the appropriate static subcase. If there is only one buckling subcase, then
the STATSUB = x command is optional if x references the first static subcase
ID.
• A static loading condition must be defined with a LOAD, TEMP(LOAD), or
DEFORM selection unless all loading is specified by grid point
displacements on SPC entries. All static subcases must be placed before the
first buckling subcase.
• Correct SPC sets must be selected for all subcases.
• Output requests that apply to only a particular subcase must be placed
inside that subcase.
• Output requests that apply to all subcases may be placed above the subcase
level.
• Some type of output request (e.g., disp(plot) = all) must be requested for at
least one subcase.
“Buckling Examples” on page 607 contains six example problems that illustrate
some of these guidelines.
21.3 Eigenvalue Extraction Methods

Seven methods of real eigenvalue extraction are available in NX Nastran:
• Givens
• Modified Givens
• Householder
• Modified Householder
• Inverse power
• Enhanced inverse power
• Lanczos
Three out of these seven methods can be used for linear buckling analysis:
CHAPTER 21 603
Linear Buckling
• Inverse power
• Enhanced inverse power
• Lanczos
A brief description of each eigenvalue method is presented in the following sections.
For further details, see the NX Nastran Numerical Methods User’s Guide.
Inverse Power Method (INV)

The inverse power method is a tracking method since it attempts to extract the
lowest eigenvalue and eigenvector in the desired range first. Their effects are then
“swept” out. The next higher mode is then found, and its effects are “swept” out,
and so on—hence, the term “tracking,” which means that one root at a time is found.
Furthermore, each root is found via an iterative process. (The classical literature
often refers to this method as the “inverse iteration method with sweeping.”)
Unfortunately, depending on your selection of the eigenvalue range, the inverse

power method can miss modes, which makes it unreliable. For this reason, the
inverse method is not recommended. The enhanced inverse power method is, in
general, a more reliable tracking method.
Enhanced Inverse Power Method (SINV)

This method is similar to the inverse power method except that it uses Sturm
sequence logic to ensure that all modes are found within the specified eigenvalue
range. The Sturm sequence informs you of the number of modes below each trial
eigenvalue. See the NX Nastran Numerical Methods User’s Guide and the NX Nastran
Basic Dynamic Analysis User’s Guide for further details regarding the Sturm sequence
check.
Lanczos Method
The Lanczos method overcomes the limitations and combines the best features of
the other methods. It is efficient, and if an eigenvalue cannot be extracted within the
range that you specify, a diagnostic message is issued. This method computes
accurate eigenvalues and eigenvectors. Furthermore, it prints meaningful user
diagnostics and supports parallel processing computers.
Comparison of Methods
The best method for a particular model depends on four factors:
The size of the model (the total number of degrees of freedom)

The number of eigenvalues desired
The available real memory on your computer

How well you can estimate the range of your eigenvalues
The enhanced inverse power method (SINV) can be a good choice if the model is too
large to fit into memory, only a few modes are needed, and you have a reasonable
idea of your eigenvalue range of interest. It is useful for models in which only the
lowest few modes are desired. This method is also useful as a backup method to
verify the accuracy of other methods.
For medium to large models, the Lanczos method is the recommended method.
Furthermore, the Lanczos method takes full advantage of sparse matrix methods that
can substantially increase computational speed and reduce disk space usage. For
overall robustness, the Lanczos method is the recommended method.
User Interface
A METHOD command is required in a subcase of the Case Control Section to select
the appropriate eigenvalue extraction method in the Bulk Data Section. The Bulk
Data entry is different depending on whether you are using the inverse power (INV),
enhanced inverse power (SINV), or Lanczos method. The EIGRL entry is used for
the Lanczos method, and the EIGB entry is used for the INV and SINV methods.
EIGRL Format
The EIGRL entry has the following format:
1 2 3 4 5 6 7 8 9 10
EIGRL SID V1 V2 ND
The SID field is the set identification number, which is referenced by the METHOD
command in the Case Control Section. The V1 field defines the lower eigenvalue
bound λ l , and the V2 field defines the upper eigenvalue bound λ u . The ND field
specifies the number of roots desired. Table 21-1 summarizes the action NX Nastran
takes depending on the values specified for V1, V2, and ND.
Table 21-1 Number and Type of Roots Found with EIGRL Entry
Case V1 V2 ND Number and Type of Roots Found

1 x x x Lowest ND or all in range, whichever is
smaller
2 x x All in range
3 x x Lowest ND in range [V1, + ∞ ]
CHAPTER 21 605
Linear Buckling
4 x Lowest root in range [V1, + ∞ ]

5 x Lowest ND roots in [- ∞ , + ∞ ]
6 Lowest root
7 x x Lowest ND roots below V2
8 x All below V2
EIGRL Example
1 2 3 4 5 6 7 8 9 10
EIGRL 10 1
The above example selects the Lanczos method requesting the lowest root for your
model.
In order for this entry to be used, the METHOD = 10 command must be specified in
the Case Control Section.
EIGB Format
The EIGB entry has the following format:
1 2 3 4 5 6 7 8 9 10
EIGB SID METHOD L1 L2 NEP
The SID field is the set identification number, which is referenced by the METHOD
command in the Case Control Section. The METHOD field selects the desired
eigenvalue extraction method (SINV or INV). The L1 field defines the lower
eigenvalue bound λ l , and the L2 field defines the upper eigenvalue bound λ µ . The
NEP field estimates the number of positive roots in the range and is used only for
the INV method.
EIGB Example
1 2 3 4 5 6 7 8 9 10
EIGB 10 SINV 0.5 1.2
The above example selects the enhanced inverse power method. All eigenvalues
between 0.5 and 1.2 are desired. In order for this entry be used, the METHOD = 10
command must be specified in the Case Control Section.
EIGB Example
EIGB 20 INV 0.5 1.2 2
The above example selects the inverse power method. There are two estimated roots
in this eigenvalue range. In order for this entry be used, the METHOD = 20
command must be specified in the Case Control Section.
21.4 Linear Buckling Assumptions and Limitations

The following assumptions and limitations apply to linear buckling analysis:
1. The deflections must be small.

2. The element stresses must be elastic.
3. The differential stiffness is supported for the following elements:
CONROD, CROD, CTUBE, CBAR, CBEAM, CBEND, CQUAD4, CQUAD8,
CTRIA3, CTRIA6, CSHEAR, CHEXA, CPENTA, and CTETRA.
4. A minimum of five grid points per half sine wave (buckled shape) is
recommended.
5. The distribution of the internal element forces due to the applied loads
remains constant.
6. Offsets should not be used in beam, plate, or shell elements for buckling
analysis.
7. For 3-D buckling problems, the use of PARAM,K6ROT is recommended for
CQUAD4 and CTRIA3 elements. A value of 100 is recommended.
8. For structures that exhibit nonlinear material or large deflection
deformations, the linear buckling load obtained from Solution 105 may be
different than the actual buckling load. For structures with significant
nonlinearities, it is recommended that you perform a nonlinear buckling
analysis using Solution 106.
See Also
• “Performing Nonlinear Buckling Analysis” on page 127 of the NX Nastran

Basic Nonlinear Analysis User’s Guide
CHAPTER 21 607
Linear Buckling
21.5 Buckling Examples

This section contains six example problems pertaining to buckling. Each problem is
chosen to illustrate either a particular type of failure mode, the effect of an element
type, the effect of a structural configuration, or the method of performing multiple
buckling analyses in a single run. The examples are as follows:
1. Classical Euler beam buckling.

2. Lateral buckling.
3. Planar frame buckling.
4. Buckling of a stiffened panel with transverse shear flexibility.
5. Buckling of a cylinder under uniform axial load.
6. Multiple buckling analyses in a single run.
In each example, the results produced by NX Nastran are compared with known
theoretical solutions.
Example 1 – Classical Euler Beam Buckling

Consider a classical Euler buckling problem. This problem is shown in Figure 21-2.
A
P
x
A
1
l = 10 m
2
E = 7.1 E10 N ⁄ m 0.10 m 2 2
2
G = 2.67 E10 N ⁄ m
2
A = 0.002 m
1
4
I2 = 1.667 E-6 m
4 0.02 m
I1 = 6.667 E-8 m
Section A-A
4
J = 2.332 E-7 m
Figure 21-2 Euler Beam
The initial buckling occurs in the xz plane since I1 is smaller than I2.
NX Nastran Results for Example 1

The input file used for this problem is shown in Listing 21-2. In this case, a 1000 N
compressive load is applied at the roller end to generate internal loads for the
structure. The Lanczos eigenvalue extraction method is used to request the first five
modes. Either the CBAR or CBEAM elements can be chosen for this problem; the
CBAR elements are used in this case.
CHAPTER 21 609
Linear Buckling
Listing 21-2 Input File for Euler Beam
$ FILENAME - bukeuler.dat
ID EULER BEAM
SOL 105
TIME 10
CEND
$
TITLE = EULER BEAM
SUBTITLE = METRIC UNITS
SPC = 10
DISP = ALL
$
SUBCASE 1
LOAD = 10
$
SUBCASE 2
METHOD = 10
$
BEGIN BULK
$
PARAM POST 0
EIGRL 10 5
FORCE 10 11 -1000. 1. 0. 0.
$
CBAR 1 1 1 2 100
CBAR 2 1 2 3 100
CBAR 3 1 3 4 100
CBAR 4 1 4 5 100
CBAR 5 1 5 6 100
CBAR 6 1 6 7 100
CBAR 7 1 7 8 100
CBAR 8 1 8 9 100
CBAR 9 1 9 10 100
CBAR 10 1 10 11 100
$
GRID 1 0. 0. 0.
GRID 2 1. 0. 0.
GRID 3 2. 0. 0.
GRID 4 3. 0. 0.
GRID 5 4. 0. 0.
GRID 6 5. 0. 0.
GRID 7 6. 0. 0.
GRID 8 7. 0. 0.
GRID 9 8. 0. 0.
GRID 10 9. 0. 0.
GRID 11 10. 0. 0.
GRID 100 0. 0. 10.
$
MAT1 1 7.1+10 .33 2700.
$
$ RECTANGULAR SECTION OF DIMENSION .1m x .02m
$
PBAR 1 1 .002 6.667-8 1.667-6 2.332-7
$
SPC1 10 123 1
SPC1 10 23 11
SPC 10 1 4
ENDDATA
As indicated on the EIGRL entry, the first five modes are requested for this problem,
and the eigenvalue table is shown in Figure 21-3. The first eigenvalue λ 1 in this case
is equal to 0.4672, while the applied load in SUBCASE 1 is equal to –1000 N.
Therefore, the lowest buckling load is equal to
P 1 = ( λ 1 ) ( P a ) = ( .46719 ) ( – 1000 ) = – 467.2 N
R E A L E I G E N V A L U E S
MODE EXTRACTION EIGENVALUE RADIANS CYCLES GENERALIZED GENERALIZED
NO. ORDER MASS STIFFNESS
1 1 4.671909E-01 6.835136E-01 1.087846E-01 4.934669E+02 2.305432E+02
2 2 1.869135E+00 1.367163E+00 2.175907E-01 2.181382E+03 4.077298E+03
3 3 4.209068E+00 2.051601E+00 3.265225E-01 4.431894E+03 1.865414E+04
4 4 7.498919E+00 2.738415E+00 4.358323E-01 4.971707E+03 3.728243E+04
5 5 1.168153E+01 3.417825E+00 5.439637E-01 4.934669E+02 5.764446E+03
Figure 21-3 Eigenvalue Table for a Euler Beam
A partial output that contains the first, second, and fifth eigenvectors is shown in
Figure 21-4. The mode shapes for the first five modes are plotted in Figure 21-5.
CHAPTER 21 611
Linear Buckling
EIGENVALUE = 4.671909E-01
R E A L E I G E N V E C T O R N O . 1
1 G .0 .0 .0 .0 -3.141592E-01 -8.436650E-13
2 G 2.344155E-14 -3.535500E-13 3.090170E-01 3.617327E-13 -2.987832E-01 5.605364E-13
3 G 2.295983E-14 1.997131E-13 5.877852E-01 -5.413820E-14 -2.541602E-01 -5.748283E-13
4 G 3.762660E-14 -8.750774E-15 8.090170E-01 -8.143186E-13 -1.846582E-01 -5.513294E-13
5 G 5.222389E-14 1.995187E-14 9.510565E-01 2.312171E-13 -9.708054E-02 1.283237E-14
6 G 6.208775E-14 -6.846741E-14 1.000000E+00 -2.349688E-13 6.485303E-13 -2.091279E-13
7 G 9.229606E-14 1.712600E-13 9.510565E-01 -1.026823E-12 9.708054E-02 1.438084E-12
8 G 8.977146E-14 -7.130678E-14 8.090170E-01 7.968065E-14 1.846582E-01 -5.453334E-13
9 G 6.058189E-14 1.523950E-14 5.877852E-01 1.248126E-13 2.541602E-01 1.957519E-12
10 G 8.648088E-14 -1.169763E-13 3.090170E-01 5.284149E-13 2.987832E-01 -1.513853E-12
11 G 9.207012E-14 .0 .0 -1.129664E-12 3.141592E-01 2.010604E-12
100 G .0 .0 .0 .0 .0 .0
EIGENVALUE = 1.869135E+00
1 G .0 .0 .0 .0 -6.606488E-01 1.696698E-11
2 G -5.494490E-13 8.408953E-12 6.180340E-01 -8.098147E-12 -5.344761E-01 -1.038079E-11
3 G -5.359018E-13 -4.648322E-12 1.000000E+00 1.386206E-12 -2.041517E-01 1.248417E-11
4 G -8.844888E-13 -1.029370E-13 1.000000E+00 1.719914E-11 2.041517E-01 1.241602E-11
5 G -1.226278E-12 -3.770107E-13 6.180340E-01 -5.748548E-12 5.344761E-01 -2.901899E-12
6 G -1.454765E-12 1.844125E-12 -1.042989E-12 3.660258E-12 6.606488E-01 5.452428E-12
7 G -2.164086E-12 -4.046730E-12 -6.180340E-01 2.005433E-11 5.344761E-01 -3.277508E-11
8 G -2.103291E-12 1.503905E-12 -1.000000E+00 -2.061866E-12 2.041517E-01 1.278820E-11
9 G -1.422088E-12 -3.211627E-13 -1.000000E+00 -3.318442E-12 -2.041517E-01 -4.639094E-11
10 G -2.025011E-12 2.810148E-12 -6.180340E-01 -1.061518E-11 -5.344761E-01 3.454609E-11
11 G -2.158100E-12 .0 .0 2.222183E-11 -6.606488E-01 -4.772200E-11
100 G .0 .0 .0 .0 .0 .0
1 G .0 .0 .0 .0 2.954037E-12 3.141592E-01
2 G -1.530020E-14 3.090170E-01 -4.216155E-12 -8.584071E-11 6.817444E-12 2.987832E-01
3 G -3.407930E-14 5.877852E-01 -1.068567E-11 8.455609E-11 4.070235E-12 2.541602E-01
4 G -4.763809E-14 8.090170E-01 -7.933438E-12 -6.013640E-11 -8.832388E-12 1.846582E-01
5 G -5.124878E-14 9.510565E-01 3.385932E-12 -8.368475E-11 -1.058233E-11 9.708054E-02
6 G -2.379340E-14 1.000000E+00 1.107325E-11 8.325122E-11 -5.506725E-12 2.671122E-13
7 G -4.143882E-15 9.510565E-01 1.763771E-11 -4.475530E-11 -8.492263E-12 -9.708054E-02
8 G -2.010117E-15 8.090170E-01 2.572503E-11 -1.136649E-11 -4.849835E-12 -1.846582E-01
9 G 3.640430E-15 5.877852E-01 2.352012E-11 -2.112232E-10 9.559306E-12 -2.541602E-01
10 G -1.566881E-14 3.090170E-01 1.092055E-11 5.283983E-11 1.326611E-11 -2.987832E-01
11 G -2.601413E-14 .0 .0 1.037907E-10 9.724306E-12 -3.141592E-01
100 G .0 .0 .0 .0 .0 .0
Figure 21-4 Eigenvectors for an Euler Beam
Figure 21-5 Mode Shapes for an Euler Beam
Note that the first four system buckling modes are in the xz plane and the first
buckling mode in the xy plane is the fifth overall system mode. The lowest buckling
load in the xy plane is, therefore, equal to
P 5 = ( λ 5 ) ( P a ) = ( 11.6815 ) ( – 1000 ) = – 11681.5 N
Theoretical Results for Example 1

This Euler buckling load is calculated as follows (S.P. Timoshenko and J.M. Gere,
Theory of Elastic Stability, Engineering Societies Monograph Series, Second Ed., 1961):
2
π EI 2 2 10 –8
P 1 = --------------- = π ( 7.1 ⋅ 10 ) ( 6.667 ⋅ 10 ) ⁄ 100 = 467.2 N
2
l
For this particular problem, the buckling load in the xy plane is equal to
CHAPTER 21 613
Linear Buckling
2
π EI 2 2 10 –6
P 5 = --------------- = π ( 7.1 ⋅ 10 ) ( 1.667 ⋅ 10 ) ⁄ 100 = 11681.4 N
2
l
Table 21-2 compares the NX Nastran results with theoretical results.
Table 21-2 Euler Beam Results Comparison
Mode Number NX Nastran (N) Theoretical (N) % Difference

1 (XZ Plane) 467.2 467.2 0
5 (XY Plane) 11681.5 11681.4 0.001
As you approach the higher modes, the mode shapes become less smooth. This
effect occurs because for the same mesh density, the higher modes have fewer grid
points per buckled sine wave than the lower modes. If you are interested in these
mode shapes, you should increase your model mesh density to follow the general
guideline of maintaining a minimum of five grid points per half sine wave.
However, these higher failure modes are seldom of interest in real world
applications.
Example 2 – Lateral Buckling

The problem shown in Figure 21-6 illustrates an interesting case of lateral buckling.
The model consists of a cantilever beam with a vertical tip load applied at the free
end. The dimensions of the beam are identical to Example 1. In this example, the
structure buckles in the lateral direction (xz plane) perpendicular to the applied load
rather than in the plane of the applied load (xy plane).
P 1
A
X
0.10 m 2 2
A
l = 10 m
1
0.02 m
Section A-A
Figure 21-6 Lateral Buckling

Listing 21-3 contains the corresponding NX Nastran model for this lateral buckling
problem. The CBEAM element is used in this case, although the CBAR element can
also be used since this beam consists of a constant cross section. A vertical 1000 N
load is applied in the negative y-direction to generate the element internal loads in
Subcase 1. Once again, the Lanczos eigenvalue extraction method is employed
requesting two modes. Figure 21-8 contains the eigenvalue table and corresponding
eigenvectors. The static deflection and buckling mode shapes are shown in
Figure 21-10.
It is interesting to note that the static deflection is in the plane of the applied load (xy
plane), while the buckling mode shape is in the plane perpendicular to the applied
load (xz plane). Two modes were requested on the EIGRL entry to illustrate another
subtle point for this problem. The two eigenvalues are identical in magnitude but
different in sign. What is the significance of a negative eigenvalue in a buckling
analysis? In buckling analysis, a negative eigenvalue implies that the load that
causes the structure to buckle is opposite to the direction of the applied load.
Since both the structure and loading are symmetric, it does not matter whether the
applied load is in the +y or –y direction. This fact brings up another subtle point. If
you specify V2 only on the EIGRL entry, then NX Nastran attempts to extract all the
roots below V2. In this case, these roots include all the negative eigenvalues. For a
large model, calculating a large number of modes can be expensive especially when
you are only interested in the lowest modes. Note that these negative roots are
legitimate both from a physical and mathematical standpoint. In general, this
situation only occurs if the applied buckling load is in the opposite direction of the
buckling load. For this particular problem, there is no right or wrong direction.
Therefore, in order to avoid this type of surprise, it is best to request the lowest
number of desired roots (ND) on the EIGRL entry when you are not sure of the
buckling load direction.
CHAPTER 21 615
Linear Buckling
Listing 21-3 Input File for Lateral Buckling
$ FILENAME - bucklat.dat
SOL 105
TIME 10
CEND
$
TITLE = LATERAL BUCKLING OF CANTILEVER BEAM
SPC = 10
DISP = ALL
$
SUBCASE 1
LOAD = 10
$
SUBCASE 2
METHOD = 10
$
BEGIN BULK
$
EIGRL 10 2
$
FORCE 10 11 -1000. 0. 1. 0.
$
CBEAM 1 1 1 2 100
CBEAM 2 1 2 3 100
CBEAM 3 1 3 4 100
CBEAM 4 1 4 5 100
CBEAM 5 1 5 6 100
CBEAM 6 1 6 7 100
CBEAM 7 1 7 8 100
CBEAM 8 1 8 9 100
CBEAM 9 1 9 10 100
CBEAM 10 1 10 11 100
$
GRID 1 0. 0. 0.
GRID 2 1. 0. 0.
GRID 3 2. 0. 0.
GRID 4 3. 0. 0.
GRID 5 4. 0. 0.
GRID 6 5. 0. 0.
GRID 7 6. 0. 0.
GRID 8 7. 0. 0.
GRID 9 8. 0. 0.
GRID 10 9. 0. 0.
GRID 11 10. 0. 0.
GRID 100 0. 0. 10.
$
MAT1 1 7.1+10 .33 2700.
$
PBEAM 1 1 .002 6.667-81.667-6 2.332-7
$
SPC1 10 123456 1
ENDDATA
1 1 2.183456E-01 4.672747E-01 7.436907E-02 1.749877E+02 3.820780E+01
2 2 -2.183456E-01 4.672747E-01 7.436907E-02 -1.749877E+02 3.820780E+01
EIGENVALUE = 2.183456E-01
1 G .0 .0 .0 .0 .0 .0
2 G 2.537741E-11 -3.892257E-09 2.499774E-03 3.424457E-02 -7.371472E-03 -4.395732E-09
3 G -1.277730E-11 -1.073776E-08 1.867549E-02 6.424069E-02 -2.656322E-02 -1.484095E-08
4 G 1.189500E-10 -2.919947E-08 5.795542E-02 8.779415E-02 -5.277112E-02 -1.207688E-08
5 G 8.478357E-11 -2.620355E-08 1.250618E-01 1.045299E-01 -8.153865E-02 1.648216E-08
6 G 1.273718E-10 -2.596014E-09 2.207153E-01 1.152518E-01 -1.093766E-01 2.570351E-08
7 G 9.853920E-11 2.034603E-08 3.427072E-01 1.213348E-01 -1.339076E-01 2.049627E-08
8 G 8.265020E-11 4.163538E-08 4.869550E-01 1.242675E-01 -1.537219E-01 2.487351E-08
9 G 1.443917E-10 6.310452E-08 6.483413E-01 1.253649E-01 -1.681112E-01 8.187754E-09
10 G 1.136984E-10 4.827729E-08 8.212749E-01 1.256189E-01 -1.767931E-01 -3.784262E-08
11 G 1.540406E-10 -4.905004E-09 1.000000E+00 1.256360E-01 -1.796903E-01 -6.074886E-08
100 G .0 .0 .0 .0 .0 .0
EIGENVALUE = -2.183456E-01
1 G .0 .0 .0 .0 .0 .0
2 G 7.576565E-11 2.472828E-09 2.499775E-03 -3.424457E-02 -7.371472E-03 -3.871140E-10
3 G -4.478584E-12 -3.649991E-09 1.867549E-02 -6.424069E-02 -2.656322E-02 -9.834048E-09
4 G 1.762648E-11 -1.338523E-08 5.795542E-02 -8.779415E-02 -5.277112E-02 -7.743097E-09
5 G -7.876090E-11 -1.461366E-08 1.250618E-01 -1.045299E-01 -8.153865E-02 8.549041E-09
6 G -1.393435E-10 8.698958E-10 2.207153E-01 -1.152518E-01 -1.093766E-01 1.680920E-08
7 G -9.326392E-11 6.915506E-09 3.427072E-01 -1.213349E-01 -1.339076E-01 -1.117986E-08
8 G -1.053673E-10 -1.124932E-08 4.869550E-01 -1.242675E-01 -1.537219E-01 -7.835656E-09
9 G -8.121134E-11 -4.722858E-09 6.483413E-01 -1.253649E-01 -1.681112E-01 7.580319E-09
10 G -1.725437E-10 -1.078508E-08 8.212749E-01 -1.256188E-01 -1.767931E-01 -2.140314E-08
11 G -1.482953E-10 -4.360446E-08 1.000000E+00 -1.256359E-01 -1.796903E-01 -3.865582E-08
100 G .0 .0 .0 .0 .0 .0
Figure 21-7 Eigenvalue Table and Eigenvectors for Lateral Beam Buckling
CHAPTER 21 617
Linear Buckling
Figure 21-8 Static and Buckling Shapes
The lowest buckling load calculated by NX Nastran is
P cr = ± 218 ( 1000 ) = ± 218N

1

The theoretical buckling load can be calculated as follows (S.P. Timoshenko and J.M.
Gere, Theory of Elastic Stability, Engineering Societies Monograph Series, Second Ed.,
1961):
4.013
P cr = ------------- EIGJ
2
l
4.013 10 –8 10 –7
= ------------- 7.1 ⋅ 10 ( 6.667 ⋅ 10 ) ( 2.67 ⋅ 10 ) ( 2.332 ⋅ 10 )
100
= 218 N
Table 21-3 compares the NX Nastran results to the theoretical results.

Table 21-3 Lateral Buckling Results Comparison
NX Nastran (N) Theoretical (N) % Difference

218 218 0
Example 3 – Planar Frame Buckling

This example is a planar frame subjected to a compressive load at the top as shown
in Figure 21-9. At first glance, since both the structure and loading are symmetrical,
you may be tempted to conclude that the structure will buckle in a symmetric mode.
As illustrated in Figure 21-10(a) (from NX Nastran) and S. P. Timoshenko and J. M.
Gere, Theory of Elastic Stability, 1961, this isn’t the case. The lowest buckling mode is
actually antisymmetric. The second buckling mode is symmetric (Figure 21-10(b)).
P E 1, I 1 P
4
I 1 = I 2 = 1.0 in
7 2
E 1 = E 2 = 1.0 ⋅ 10 lb/in
E 2, I 2 l
Y b = l = 100.0 in
P = 5000 lb
X
b
Figure 21-9 Planar Frame Buckling
CHAPTER 21 619
Linear Buckling
Figure 21-10 Frame Buckling

Listing 21-4 shows a portion of the input file for this problem.
Listing 21-4 Input File for Frame Buckling
$
$ INPUT FILE - bukframe.dat
$
SOL 105 $
TIME 5
CEND
TITLE = FRAME BUCKLING PROBLEM
SUBTITLE = PLANAR MODEL
DISP = ALL
SPC = 1
$
SUBCASE 1
LABEL = STATIC LOAD CASE
LOAD = 1
$
SUBCASE 2
LABEL = BUCKLING CASE
METHOD = 10
$
BEGIN BULK
$
$ BRING IN REST OF BULK DATA FILE
$
INCLUDE ’bucklat.dat’
$
$ LOADS BULK DATA ENTRIES
$
FORCE 1 11 0 5000. -1.
FORCE 1 22 0 5000. -1.
$
$ CONSTRAINTS BULK DATA ENTRIES
$
SPC1 1 123456 1 33
$
$ PROPERTY AND MATERIAL BULK DATA ENTRIES
$
PBAR 1 1 3.46 1. 1.
MAT1 1 1.+7 .3
$
$ FOR CALCULATING BUCKLING MODES
$
EIGRL 10 3
ENDDATA
P cr = ( λ 1 ) ( P a ) = 1.4756 ( – 5000 ) = – 7378 lb

1

From Theory of Elastic Stability, the buckling load P cr can be calculated as
CHAPTER 21 621
Linear Buckling
2
P cr = k E 2 I 2 Eq. 21-10
where the value of k can be obtained by solving the following transcendental

equation
kl –6 l I1
-------------- = ----------------- Eq. 21-11
tan kl b I2
Solving Eq. 21-11, the lowest value of k is equal to
k 1 = 0.02716 Eq. 21-12
Note that all the roots from the above transcendental equation only yield the
antisymmetrical modes. In this case, the lowest buckling load is
2 7
P cr = ( .02716 ) ( 1 ⋅ 10 ) ( 1.0 ) = – 7377 lb
1
Table 21-4 compares the theoretical results to the NX Nastran results for the
antisymmetric frame buckling.
Table 21-4 Frame Buckling Results Comparison
NX Nastran (lb) Theoretical (lb) % Difference

7378 7377 0.01
Example 4 – Buckling of a Stiffened Panel with

Transverse Shear Flexibility
This problem is selected to illustrate the effect of the transverse shear flexibility in
the buckling failure of a stiffened panel. Figure 21-11 consists of a panel reinforced
with stiffeners at both the top and bottom. The panel is subjected to a distributed
compressive load at the right edge. The failure mode of interest in this case is the
in-plane (xy plane) buckling of the stiffened panel.
Stiffeners A 1
h 40 X 10 Mesh of CQUADs
l x
Stiffeners A 2 RBE3
2
A 1 = A 2 = 0.6 in ; h = 2.0 in
2 2
A t = 1.2 + 0.4 = 1.6 in ; A w = t ( h ) = 0.2 ( 2.0 ) = 0.4 in
7 4
E = 1.0 ⋅ 10 psi ; I = 1.333 in
l = 20 in 6
; G = 3.79 ⋅ 10 psi
t = 0.2 in ; n = At ⁄ Aw = 4
Figure 21-11 Buckling of a Stiffened Panel

The corresponding NX Nastran input file for this problem is shown in Listing 21-5.
The panel is modeled with CQUAD4s. Since only the in-plane buckling failure mode
is of interest, only the membrane stiffness is requested for the CQUAD4s (MID1
only). The stiffeners are modeled with CRODs. Since this is a planar model, the
out-of-plane motion is constrained (3, 4, 5, and 6 DOFs).
The panel is subjected to a distributed compressive load at the right end. As an

alternative modeling technique, an RBE3 element is connected to the grid points at
the right edge so that the load can be applied to a single grid point with the RBE3
spreading the loads to the other grid points.
The SINV method of eigenvalue extraction method is used for this problem.
Therefore, the EIGB entry instead of the EIGRL entry is used. Field 3 of the EIGB
entry designates the selected method. Fields 4 and 5 provide the range of the
eigenvalue of interest. A reduced output showing the eigenvalue table and lowest
www.cadfamily.combuckling mode shape is included in Figure 21-13 and Figure 21-14, respectively.
CHAPTER 21 623
Linear Buckling
Listing 21-5 Input File for the Buckling of Stiffened Panels
$
$ FILENAME - stffqud4.dat
$
SOL 105
TIME 10
CEND
TITLE = PANEL WITH STIFFENERS
DISP = ALL
STRESS = ALL
SPC = 1
$
SUBCASE 1
LABEL = COMPRESSIVE LOAD
LOAD = 1
$
SUBCASE 2
LABEL = BUCKLING SUBCASE
METHOD = 10
$
BEGIN BULK
$
$EIGRL10 1
EIGB 10 SINV .7 .9
PARAM POST 0
$
RBE3 1000 1000 123456 2. 12 82 123 +
+ 164 205 287 328 369 410 2. 12345 +
+ 246 1.0 12 41 451
$
FORCE 1 1000 -1.0 100000.
$
$ THIS SECTION CONTAINS THE LOADS, CONSTRAINTS, AND CONTROL BULK DATA ENTRIES
$
PSHELL 1 1 .2
PROD 2 1 .6
$
MAT1 1 1.+7 .32
$
$ BRING IN THE REST OF THE MODEL
$
include ’stffqud4.blk’
$
ENDDATA
1 1 7.811782E-01 8.838429E-01 1.406680E-01 6.161843E+03 4.813498E+03
2 3 4.946603E+00 2.224096E+00 3.539759E-01 1.096963E+07 5.426240E+07
3 2 5.002625E+00 2.236655E+00 3.559746E-01 1.383715E+04 6.922209E+04
Figure 21-12 Eigenvalues for a Stiffened Panel
Figure 21-13 Stiffened Panel Buckling Mode
P cr = ( λ 1 ) ( P a ) = 0.78118 ( – ( 100,000 ) ) = – 78,118 lb

1

If this problem is treated as an Euler beam, the critical buckling load P e is equal to
CHAPTER 21 625
Linear Buckling
2
π EI
P e = ------------ = 82,245 lb Eq. 21-13
2
4 l
If the effect of the transverse shear flexibility is included, then it can be calculated as
follows (Timoshenko and Gere, Theory of Elastic Stability, 1961):
4nPe
1 + ------------- – 1
At G
P cr = ---------------------------------------- = 78,210 lb Eq. 21-14
2n
----------
At G
If the transverse shear flexibility is ignored, then the buckling load deviates by 5.2%.
Table 21-5 contains a comparison between the theoretical results versus the NX
Nastran results for the buckling of the stiffened panel. Note that the transverse
shear flexibility of this structure is automatically included when you are performing
a buckling analysis in NX Nastran. There is no additional input required on your
part.
Table 21-5 Stiffened Panel Buckling Results Comparison
NX Nastran (lb) Theoretical (lb) % Difference

78,118 78,210 0.1
A review of the stresses also indicates that the structure will yield prior to reaching
the linear buckling load level. In other words, the critical failure mode may be due
to yielding rather than to the linear buckling of the structure. The knowledge of this
linear buckling load level can still be of design significance. If you are interested in
detailed stresses for this problem, then a nonlinear analysis using Solution 106 may
be more appropriate in this case.
Example 5 – Buckling of a Cylinder Under Uniform Axial

Load
The next example is the buckling of a cylinder subjected to a distributed
compressive load at one end and simply supported at the other end. The cylinder
has a diameter of 20 inches and a length of 20 inches with a wall thickness of 0.03
inches as shown in Figure 21-14. This problem illustrates the phenomenon of the
buckling of a thin curved shell structure.
Figure 21-14 Buckling of a Cylinder Under a Compressive Load

The CQUAD4 elements are used for the NX Nastran model. A mesh of 20 elements
along the length and 72 elements around the circumference is used for this problem.
The corresponding input file containing the pertinent entries is shown in
Listing 21-6.
CHAPTER 21 627
Linear Buckling
Listing 21-6 Input File for a Cylindrical Buckling Problem
$ FILENAME - buckcy20r.dat
$
ID CYLIN BUCKLING
SOL 105
TIME 200
CEND
TITLE = BUCKLING OF CYLINDER - SIMPLY SUPPORTED
SUBTITLE = 20" x 20" - t=.03" - CP IN CYL COORD SYSTEM - K6ROT = 100.
DISP = ALL
SPC = 1
$
SUBCASE 1
LABEL = STATIC LOAD
LOAD = 1
SPCF = ALL
STRESS = ALL
$
SUBCASE 2
LABEL = EIGENVALUE CALCULATION
METHOD = 1
$
BEGIN BULK
$
PARAM K6ROT 100.
PARAM POST -1
EIGRL 1 2
$
PSHELL 1 1 .03 1
$
MAT1 1 1.+7 .3
$
include ’full_cp1.blk’
$
$ THIS SECTION CONTAINS ALL DEFINED COORDINATE SYSTEMS
$
CORD2C 1 0 0.0 0.0 0.0 0.0 0.0 1. +
1. 0.0 0.0
$
RBE3 5000 5000 123456 1. 123 381 382 +
+ 383 384 385 386 387 388 389 390 +
+ 391 392 393 394 395 396 397 398 +
+ 399 781 782 783 784 785 786 787 +
+ 788 789 790 791 792 793 794 795 +
+ 796 797 798 1180 1181 1182 1183 1184 +
+ 1185 1186 1187 1188 1189 1190 1191 1192 +
+ 1193 1194 1195 1196 1197 1579 1580 1581 +
+ 1582 1583 1584 1585 1586 1587 1588 1589 +
+ 1590 1591 1592 1593 1594 1595
$
FORCE 1 5000 0 100000. -1.
$
ENDDATA
Once again, an RBE3 element is used to distribute the load from a single grid
point to the circumference of the cylinder.
A static load of –100,000 lb is applied in Subcase 1 to generate the internal

forces for the structure. This load is applied to the RBE3, which, in turn,
distributes the loads around the circumference of the cylinder. The static
deflection is shown in Figure 21-15.
The Lanczos eigenvalue extraction method is used in this case to request the
lowest mode. The lowest calculated eigenvalue is equal to 0.35986. The
buckling load is, therefore, equal to
P cr = ( .34262 ) ( – 100,000 ) = – 34,262 lb
The corresponding buckling mode shape is shown in Figure 21-16. As you

can see from the plots, there are two grid points per half sine wave (four
grid points per sine wave), which is below the recommended value of a
minimum of five grid points per half sine wave.
This problem is then remeshed with 40 elements along the z-direction

keeping the same mesh density in the circumferential direction. The new
eigenvalue calculated in this case is equal to 0.34914. The revised buckling
load is, therefore, equal to
P cr = ( .34234 ) ( – 100,000 ) = – 34,234 lb
The corresponding mode shape is shown in Figure 21-17. In this case, there
are three grid points per half sine wave (six grid points per sine wave),
which is still below the minimum requirement. To obtain better accuracy,
you can certainly further refine this model until you meet the minimum
number of grid points requirement. However, this mesh is sufficient for
demonstrating the basic linear buckling features of NX Nastran.
CHAPTER 21 629
Linear Buckling
Figure 21-15 Static Deflection of a Cylinder Due to a Compressive Load
Figure 21-16 Buckling Shapes of a Cylinder with 20 Elements Along the

z-Direction
Figure 21-17 Buckling Shape of a Cylinder with 40 Elements Along the

z-Direction

The first buckling load (Timoshenko and Gere, Theory of Elastic Stability, 1961) can be
calculated as follows:
2 7 2
E(h) 1x 10 ( .03 )
N x = -------------------------------- = ------------------------------------- = 544.7 lb/in
2 2
r 3(1 – ν ) 10 3 ( 1 – .3 )
P cr = N x ( 2 ) ( π )r = 544.7 ( 2 ) ( π ) ( 10 ) = 34,225 lb
Table 21-6 contains a comparison between the results obtained with NX Nastran
versus the theoretical results.
Table 21-6 Thin Cylinder Buckling Results Comparison
Mesh Density NX Nastran (N) Theoretical (N) % Difference

72 x 20 34,262 34,225 .11
72 x 40 34,234 34,225 .03
CHAPTER 21 631
Linear Buckling
Example 6 – Multiple Buckling Analyses in a Single Run

So far, a single static analysis was considered followed by a single buckling analysis.
Example 6 shows you how to run multiple static and buckling analyses in a single
run. In fact, what is done in this example is to combine the Euler beam buckling
(Example 1) and lateral buckling (Example 2) problem into a single run. The
corresponding input file is shown in Listing 21-7.
The model used is the same geometric beam model used in Examples 1 and 2. The
first subcase (Subcase 2) is a static subcase consisting of a cantilever beam with a
vertical tip load applied to the free end (see Figure 21-6). The second subcase
(Subcase 5) is a static subcase consisting of a simply supported beam with an axial
load applied to the roller end (see Figure 21-2). The third subcase (Subcase 11) is for
a lateral buckling analysis. The Case Control command (STATSUB = 2 in Subcase
11) tells NX Nastran that you want to generate the differential stiffness matrix from
the first static subcase (Subcase 2). The fourth subcase (Subcase 21) is for a Euler
beam buckling analysis. The Case Control command (STATSUB = 5 in Subcase 21)
tells NX Nastran that you want to generate the differential stiffness matrix from the
second static subcase (Subcase 5). The Bulk Data entries are similar to Examples 1
and 2. Note that different boundary conditions are allowed for different subcases.
Listing 21-7 Input File for Multiple Buckling Analyses
$ FILENAME - MULTBUCK.DAT
$
SOL 105
TIME 10
CEND
$
TITLE = LATERAL BUCKLING OF CANTILEVER BEAM
DISP = ALL
$
SUBCASE 2
LABEL = CANTILEVER BEAM
LOAD = 10
SPC = 10
$
SUBCASE 5
LABEL = SIMPLY SUPPORTED BEAM
LOAD = 20
SPC = 20
$
SUBCASE 11
LABEL = LATERAL BUCKLING OF CANTILEVER BEAM
METHOD = 10
SPC = 10
STATSUB = 2
$
SUBCASE 21
LABEL = EULER BUCKLING OF SIMPLY SUPPORTED BEAM
METHOD = 10
SPC = 20
STATSUB = 5
$
BEGIN BULK
$
EIGRL 10 2
$
FORCE 10 11 -1000. 0. 1. 0.
FORCE 20 11 -1000. 1. 0. 0.
$
CHAPTER 21 633
Linear Buckling
Listing 21-7 Input File for Multiple Buckling Analyses (continued)
CBEAM 1 1 1 2 100
CBEAM 2 1 2 3 100
CBEAM 3 1 3 4 100
CBEAM 4 1 4 5 100
CBEAM 5 1 5 6 100
CBEAM 6 1 6 7 100
CBEAM 7 1 7 8 100
CBEAM 8 1 8 9 100
CBEAM 9 1 9 10 100
CBEAM 10 1 10 11 100
$
GRID 1 0. 0. 0.
GRID 2 1. 0. 0.
GRID 3 2. 0. 0.
GRID 4 3. 0. 0.
GRID 5 4. 0. 0.
GRID 6 5. 0. 0.
GRID 7 6. 0. 0.
GRID 8 7. 0. 0.
GRID 9 8. 0. 0.
GRID 10 9. 0. 0.
GRID 11 10. 0. 0.
GRID 100 0. 0. 10.
$
MAT1 1 7.1+10 .33 2700.
$
$
PBEAM 1 1 .002 6.667-81.667-6 2.332-7
$
SPC1 10 123456 1
$
SPC1 20 123 1
SPC1 20 23 11
SPC 20 1 4
$
ENDDATA
22
CHAPTER
Laminates
■ Overview of Laminated Composite Materials
■ Understanding Classical Lamination Theory
■ Understanding Laminate Failure Indices
■ Calculating Interlaminar Shear Stress and Strains
■ Examples: Defining Laminate Material Properties
CHAPTER 22 635
Laminates
22.1 Overview of Laminated Composite Materials

The term composite refers to an engineering material that is made up of more than
one material. Classical lamination theory takes the term composite another step to
mean a material that is composed of stacks of ply layers or lamina, with each lamina
having its own orthotropic properties. A laminate is a stack of these individual
lamina arranged with the individual lamina having different orientations of the
principal material directions. The laminae are bonded together with a thin layer of
bonding material that is considered to be of zero thickness.
(1)
Y mat (2)
Y mat
(1) (2)
X mat X mat
(3)
Y mat
(3)
X mat
Figure 22-1 Laminae Arranged to Form a Laminate
Each ply or lamina may be considered as a group of unidirectional fibers. The ability
to orient the fibers in particular directions lets you tailor the mechanical properties
of a composite material to match the loading environment. The principal material
axes for the lamina are parallel and perpendicular to the fiber directions. The
principal directions are referred to as:
• “longitudinal” or the 1-direction of the fiber

• “transverse” or the 2-direction for the perpendicular direction (matrix
direction)
Figure 22-2 shows an exploded view of three cross-ply laminated plate. The n-
laminae (n = 1,2,3,4) of each of the three configurations are normal to the z-axis of the
indicated coordinate system and the 1- and 2-axes appended to the individual lamina
denote principal material axis directions. The directions of the principal material
axes of each lamina alternate as implied by the use of the word "cross-ply" to describe
the configuration. The xy-plane of the coordinate axes is defined in the geometric
middle plane of the laminae.
x
2 1
x
1 2
1 2
y 2 1
1
y 2
Two-Ply
Three-Ply
2
1 x
1
2
2
y 1
1
2
Four-Ply
Figure 22-2 Exploded View of Three Cross-Ply Laminated Plates
Plies are typically made of fibers glued in place by a matrix. If the ply is a tape, all
the fibers are oriented in the same direction. Cloth plies have fibers woven in two
directions. Many different materials can be used as fibers or matrices. Examples of
fibers include graphite, glass, boron, silicon carbide, and tungsten. Examples of
matrices include epoxy and aluminum.
Modeling Laminates in NX Nastran

In NX Nastran, you can use the CQUAD4, CQUAD8, CTRIA3, and CTRIA6 elements
to model composite and nonuniform materials. This lets you model plates with
coupled membrane and bending elastic behavior. NX Nastran uses the assumptions
of classical lamination theory in formulating shell behavior for nonuniform and
composite laminate element properties.
CHAPTER 22 637
Laminates
You can model an entire stack of laminae with a single plate or shell element
because the material properties of the stack are completely reflected in the matrices
of elastic moduli for the element. The software automatically calculates these
matrices from user-supplied definitions of the thickness, the material properties,
and the relative orientation of these properties for the individual lamina. Once the
software calculates the matrices of elastic moduli, the analysis proceeds.
Because the material properties of the laminated composite plate are completely
contained in the matrices of elastic moduli, you can use standard data recovery
methods to have the software calculate stresses in individual laminae and the forces
sustained by the laminate.You can have NX Nastran:
• evaluate stresses or strains and appropriate failure indices in individual
laminae
• calculate interlaminar shear stresses and strains and a bonding failure
index
If the loading on a structure is sufficient to exceed the linear elastic limit of the
material, then nonlinear methods are required to predict the nature of the plastically
(permanently) deformed state. Nonlinear methods are also required if the material
is nonlinear in its elastic range.
For a two-dimensional analysis involving composite material properties, there are

two different methods you can use to define a laminate.
• You can use the PSHELL bulk data entry to directly input membrane,
bending, membrane-bending coupling and transverse shear constitutive
relationships.
• You can use the PCOMP entry to define the composite laminate on a ply-
by-ply basis. When you use PCOMP, NX Nastran computes the equivalent
PSHELL and MAT2 entries for you.
Both the PSHELL and PCOMP methods are described in the following sections.
Note: For three-dimensional composite analyses, you must develop the anisotropic
material matrix yourself. Typically, you would use the MAT9 bulk data entry to
define this material matrix.
See Also

• “Understanding MAT9” on page 122 of the NX Nastran User’s Guide
Using the PSHELL Method

With the PSHELL method, you use the PSHELL entry with the appropriate material
definitions and geometric parameters to directly define a laminate. For example, you
can use PSHELL to define properties such as membrane thickness, material
properties, and bending and transverse shear parameters.
On the PSHELL entry, there are different four material ID (MID) fields as shown in
Table 22-1.
Table 22-1 Material Properties on the PSHELL Entry
ID PSHELL Field Purpose

MID1 3 Membrane material property.
MID2 5 Bending material property.
MID3 7 Transverse shear stiffness material property.
MID4 4 (cont.) Material property used to define coupling
between the membrane and the bending
deformation.
You can specify MAT1, MAT2, or MAT8 as the material type for any of the material
(MIDi) fields. For isotropic materials, use the same MAT1 identification for MID1
and MID2, and leave the MID3 and MID4 fields blank.
• Use MID1 if you want to include membranes only.

• Use MID2 (and optionally, MID3) if you want to include bending only.
• Use MID3 to have the software use thick plate theory, which includes the
transverse shear flexibility, to develop the element stiffness matrix. In
general, for thin or curved surfaces, you shouldn’t use MID3.
• Use MID4 to input a coupling relationship between the in-plane forces and
bending moments. This coupling only occurs in plates that are
nonsymmetric about the neutral plane or where the neutral plane is offset
from the grid points. Typical applications include reinforced skins and
aluminum bonded to fiberglass. You should leave the MID4 field blank if
the element cross section is symmetric.
While the PSHELL method is sufficient for defining simple composites, determining
the appropriate material definitions can be difficult for more complex composites.
Additionally, when you use the PSHELL method, you can only recover smeared
element data (running loads and moments, and equivalent strains and curvatures).
See Also
CHAPTER 22 639
Laminates
• “PSHELL” on page 1611 of the NX Nastran Quick Reference Guide
Using the PCOMP Method

The PCOMP entry provides a convenient way of entering appropriate material
properties for a plate consisting of layers of unidirectional fibers. The input on the
PCOMP entry consists of the layer-by-layer material definition. You can use
PCOMP to define the thickness, orientation, and the material identification number
of each of the individual lamina.
When you solve your model, NX Nastran calculates the membrane, bending,
transverse shear, and coupled membrane-bending material properties of the
laminate as a whole from the data you specified on the PCOMP entry. The software
outputs these calculated properties in the form of one equivalent PSHELL and four
MAT2 entries, as shown in Figure 22-3. If you specify ECHO = PUNCH in the Case
Control section, the software prints the generated PSHELL and MAT2 entries in the
PUNCH file after the sorted bulk data echo
Since the MIDi fields use more than eight digits, the software uses the large field
format for the generated PSHELL and MAT2 entries. You can then modify your
input file with these new entries and proceed with your analysis.
CQUAD4
PCOMP
MAT1 MAT2 MAT8
EQUIV PSHELL
Output if
ECHO = SORT
MID1 MID2 MID3 MID4
(the default)
MAT2 MAT2 MAT2 MAT2
Figure 22-3 Equivalent PSHELL and MAT2 Entries Are Generated
The ID numbers of the generated MAT2 entries are important.
• The MID1 material has an ID in the range of 100000000 to 199999999.

• The MID2 material has an ID in the range of 200000000 to 299999999, etc.
These MIDi ranges alert NX Nastran that the material is part of a composite analysis.
Therefore, if you’re using the equivalent properties in a future analysis instead of
using the PCOMP entry, and you’re entering a thermal coefficient of expansion, don’t
change the ID numbers.
Selecting a Failure Theory

When you use the PCOMP method, you can use the FT field (field 6) to specify a
failure theory for the laminate.
See Also
• “Understanding Laminate Failure Indices” on page 652 of the NX Nastran

User’s Guide
Selecting Output Options

The output you may request for a composite analysis includes:
• Stresses and strains for the equivalent plate

• Force resultants
• Stresses and/or strains in the individual laminate including approximate
interlaminar shear stresses in the bonding material output
• A failure index table
Note that if you want NX Nastran to calculate stress and/or the failure indices for the
composite elements, you must use the ELSTRESS Case Control command to request
stress output for the appropriate composite elements.
One benefit of the PCOMP entry is that it gives you more output options than the
equivalent PSHELL and MAT2 entries. For example, PCOMP lets you obtain:
• individual layer stresses and strains, interlaminar stress, and failure indices
in Solution Sequences 101, 103, 105, 106, 114, 144, and 200
• element force and equivalent strain output in the other solution sequences
If you use the equivalent PSHELL and MAT2 entries, you won’t be able to obtain the
laminae stress or the failure index table.
You can use PARAM,NOCOMPS to control the computation and printout of

composite element ply stresses and failure indices.
See Also
• “PCOMP” on page 1549 of the NX Nastran Quick Reference Guide

• “STRESS” on page 383 of the NX Nastran Quick Reference Guide
• “NOCOMPS” on page 616 of the NX Nastran Quick Reference Guide
CHAPTER 22 641
Laminates
22.2 Understanding Classical Lamination Theory

Classical lamination theory makes the following assumptions regarding the
behavior of the laminae:
• The laminate consists of perfectly bonded laminae.

• The bonds are infinitesimally thin and nonshear-deformable; i.e.,
displacements are continuous across laminae boundaries so that no lamina
can slip relative to another.
• There is linear variation of strain through the laminate thickness.
Deformation in the X-Y plan of the plate at any point C at a distance z in the normal
direction to plate middle surface is
U = U 0 + zθ y Eq. 22-1
V = V 0 + zθ x Eq. 22-2
where U , V , and W are the displacements along the X, Y, and Z directions in the
element coordinate system, and θx , θy are the rotations.
The strain-displacement-middle surface strain and curvatures relationship is given

by:
∂U 0 ∂θ y
 ----------   -------- 
 ∂ x   ∂x   ε0 
 εx       x   χx 
   ∂V 0   ∂θ x   0  
 εy  =  --------- + z  – --------  =  ε y  – z  χ y  Eq. 22-3
   ∂y   ∂y     
 γ xy   ∂U V 0   ∂θ   0   χ xy 
 ---------0- + ∂---------  -------y- – ∂-------
θx 
-  γ xy 
 ∂y ∂x   ∂y ∂x
where the ε 0 ’s and χ ’s are the middle surface strains and curvatures, respectively.
The stress resultants for an N-layer laminate are obtained by integration of the
stresses in each lamina through the laminate thickness as:
 Nx  t
---
 σx  N
z  σx 
  2   k  
 Ny  =
 
∫- --t-  σ y dz = ∑ ∫z k – 1  σ y  dz Eq. 22-4
 N xy  2 τ  k = 1  τ xy k
xy
 Mx  t  σx  N  σx 
--
- zk
  2    
 M y  = – ∫ t  σ y z dz = – ∑ ∫  σ y  z dz Eq. 22-5
  - ---   z
2 τ  k – 1 
 M xy  xy
k = 1  τ xy k
The stress resultant to strain relationship is:
  ε0  χx  
 Nx  N  z  x  z  
   k  0 k   
 Ny  = ∑ [ G ]k  ∫  ε y  dz – ∫  χ y  z dz  Eq. 22-6
   zk – 1   zk – 1   
N
 xy  k = 1   0   χ xy k 
  γ xy k 
M   ε0  χx  
 x  N  z  x  zk
 
   k  0   2 
 My  = ∑ [ G ]k  – ∫  ε y  z dz + ∫  χ y  z dz  Eq. 22-7
   z z
k – 1  k – 1  
 M xy  k = 1   0   χ xy k 
  γ xy k 
where [ G ] k is the material matrix transformed from the laminate coordinate system
into the lamina coordinate system.
These relations can be written in the following form used to describe composite
elements:
F  0
A B ε  Eq. 22-8
  =  
M  B D χ
where:
N
[A] =
∑ [ G ]k ( zk – z k – 1 )
k=1
N
[ B ] = 1--- 2 2
2 ∑
[ G ]k ( z k – zk – 1 )
k=1
N
[ D ] = 1--- 3 3
3 ∑
[ G ]k ( zk – zk – 1 )
k=1
are named in composite element literature as the membrane, membrane-coupling,

and bending matrices, respectively.
In the shell element formulation in NX Nastran (CQUAD4, CQUAD8, CTRIA3, and

CTRIA6), these relationships take the following form:
CHAPTER 22 643
Laminates
2
TG 1 T G 4 0
N   ε0 
  3  
M  = 2 T χ Eq. 22-9
  T G 4 ------ G 2 0  
12
Q  γ 
0 0 Ts G3
where:
[ A ] = TG1 (3x3 matrix)
[ B ] = – T2 G 4 (3x3 matrix)
3
T
[ D ] = -----
- G 2 (2x2 matrix)
12
 Qx 
{Q} =   = transverse shear resultants
 Qy 
 γx 
{ γ } =   = transverse shear strains
 γy 
T = nominal plate thickness

T s = effective transverse shear material thickness
G 3 = effective transverse shear material matrix
If you use the PSHELL entry, you can directly input G 1 , G 2 , G 4 , T , G 3 , and Ts
(G1=MID1, G2=MID2, G3=MID3, G4=MID4 on the PSHELL entry) If you use the
PCOMP entry, you can have NX Nastran calculate the composite equivalent
material matrices from the data you supply.
Equivalent thermal properties are determined as follows:
 αε   Tα 1  –1  Gα ε 
 0   3  A B 0  0 
  T   
 α χ  =  -----
- α2  = B D 0  Gα χ  Eq. 22-10
   12   
α
 χε   2  0 0 Ts G3  Gα χε 
 0  T α3   0
The following coefficients are used to determine equivalent thermal properties:
N
 
 Gα ε  =
 0 ∑ [ G ]k { α }k ( zk – zk – 1 )
k = 1
N
1 3 3
{ Gα χ } = ---
3 ∑ [ G ]k { α }k ( zk – zk – 1 )
k = 1
N
  1 2 2
 Gα χε  = --2-
 0 ∑ [ G ]k { α }k ( zk – zk – 1 )
k = 1
These coefficients are used to calculate the equivalent thermal properties as follows:
  –1  
 α ε  = [ A ]  Gα ε 
 0  0
and
–1
{ αx } = [ D ] { Gα x }
where { αξ } and { α x } are the membrane and bending equivalent thermal properties.
0
Note that { α ε } is not directly calculated, but is determined from { Gα ε } when the
0 0
PCOMP input is used when the MID4 field on the PSHELL is > 400,000,000. Note that
{ G xξ } cannot be input directly using PSHELL and { α ε } can be input only if [ B ] is
0 0
invertible (which is generally not true).
The thermal expansion relationships in the shell element formulation take the
following form:
αε = Tα 1
0
3
T
α χ = ------ α 2
12
2
α χε = T α3
0
where α 1 , α 2 , and α 3 are the thermal expansion inputs on the materials referenced
by the MID1, MID2, and MID4 fields on the PSHELL entry. If you use PCOMP, NX
Nastran automatically calculates these relationships.
The terms G 1 , G 2 , and G 4 are defined by the following integrals:
CHAPTER 22 645
Laminates
1
G 1 = --- ∫ [ G e ] dz
T
1 2
G 2 = --- ∫ z [ G e ] dz Eq. 22-11
I
1
G 4 = ------ ∫ ( – z ) [ G e ] dz
2
T
The limits on the integration are from the bottom surface to the top surface of the
laminated composite. The matrix of material moduli, [ G e ] , has the following form
for isotropic materials:
E νE
---------------- ---------------- 0
2 2
1–ν 1–ν
νE E
[ Ge ] = ---------------- ---------------- 0 Eq. 22-12
l 2 2
1–ν 1–ν
E
0 0 ----------------------
2( 1 + ν)
For orthotropic materials, the matrix, [ G e ] , is written as follows:
E1 ν1 E2
---------------------- ---------------------- 0
1 – ν1 ν2 1 – ν1 ν2
ν2 E1 E2
[ Ge ] = ---------------------- ---------------------- 0 Eq. 22-13
0 1 – ν1 ν2 1 – ν1 ν2
0 0 G 12
Here, ν 1 E 2 = ν2 E 1 to satisfy the requirement that the matrix of elastic moduli be

symmetric. In general, the analyst may supply element properties with respect to a
particular orientation which does not necessarily correspond to the principal
material axes. In this case, the analyst must also supply the value of the angle, θ or
material coordinate system that orients the element material axis relative to the side
G1-G2 of the element. The material elastic modulus matrix is then transformed by
the program into the element modulus matrix through the relation
t
[ Ge ] = [ U ] [ Gm ] [ U ] Eq. 22-14
where:
2 2
cos θ sin θ cos θ sin θ
[U] = sin θ
2
cos θ
2
– cos θ sin θ Eq. 22-15
2 2
– 2 cos θ sin θ 2 cos θ sin θ cos θ – sin θ
The finite element model for a structure composed of composite materials requires
the evaluation of the matrix of elastic moduli for each plate element of the model. The
characteristics of the composite media are totally contained in these matrices.
To illustrate evaluation of these matrices, consider the cross-ply laminates shown in

Figure 22-2. Here, all three configurations are represented by a single quadrilateral
plate element. The coordinate axes are coincident with the element coordinate axes.
Then, if we assume that each lamina of the n-ply laminates is of thickness T ⁄ n ,
where T is the total thickness of each of three configurations, the matrices of elastic
moduli may be evaluated from the following relations:
 – T--- + T--- T 2T
– --- + ------
T
--- 
 2 n 2 n 2 
1 2 
[ G 1 ] = ---  ∫ [ G e ] 1 z dz + ∫ [ G e ] 2 dz + … + ∫ [ G e ] n dz  Eq. 22-16
T 
 – T--- T T T ( n – 1 )T 
– --- + --- – --- + --------------------
 2 2 n 2 n 
 – T--- + T--- T 2T
– --- + ------
T
--- 
 2 n 2 n 2 
1 2 2 
[ G 2 ] = ---  ∫ [ G e ] 1 dz + ∫ [ G e ]2 z dz + … + ∫ [ G e ] n z dz  Eq. 22-17
T 
 – T--- T T T ( n – 1 )T 
– --- + --- – --- + --------------------
 2 2 n 2 n 
– T T
--- + ---
T 2T
– --- + ------
 2 n 2 n
1
[ G 4 ] = ------ 
T 
2 ∫ [ G e ] ( – z ) dz +
1 ∫ [ G e ] ( – z ) dz
2
Eq. 22-18
 T
– ---
T T
– --- + ---
 2 2 n
T
--- 
2 

+…+ ∫ [ G e ] ( – z ) dz 
n

T ( n – 1 )T 
--- + ----------------------
2 n 
These relations reflect the assumption that the xy-plane of the element coordinate
system is coincident with the geometric middle plane of the laminate. The xy-plane
of the element coordinate system is defined in the mean plane of the element so that
any offset between the mean plane of the connected grid points and the geometric
middle plane of the laminate would be reflected in the integration limits of the
www.cadfamily.compreceding relations.
CHAPTER 22 647
Laminates
Note that I = T3 ⁄ 12 in the evaluation of [ G 2 ] , i.e., the value ( 12I ) ⁄ Γ2 , will be assigned
the default value of 1.0 on the plate element property entry.
The matrix of elastic moduli for transverse shear, [ G 3 ] m is defined as a two-by-two

matrix of the form
G 11 G 12
[ G3 ] = Eq. 22-19
m
G 21 G 22
and the corresponding matrix transformed into an element coordinate system is

given by
T
[ G3 ] = [ W ] [ G3 ] [ W ] Eq. 22-20
e m
where [ W ] = cos θ sin θ

– sin θ cos θ
The mean value of the transverse shear modulus G for the laminated composite is
defined in terms of the transverse shear strain energy, U , through the depth
2 2
1V 1 (τ(z))
U = --- -------- = --- ∫ ------------------ dz Eq. 22-21
2 GT 2 G(z)
A unique mean value of the transverse shear strain is assumed to exist for both the
x and y components of the element coordinate system, but for ease of discussion,
only the evaluation of an uncoupled x component of the shear moduli will be
illustrated here. From Eq. 22-21 the mean value of transverse shear modulus may
be written in the following form
N zi
2
1 T ( τ zx ( z ) )
------ = --------- ∑ ∫ ----------------------- dz Eq. 22-22
Gx Vx
2 ( Gx )
i
i = 1z
i–1
where G is an “average” transverse shear coefficient used by the element code and
( G x )i is the local shear coefficient for layer i . To evaluate Eq. 22-13, it is necessary to
obtain an expression for ( τ zx ( z ) ) . This can be accomplished by assuming that the x-
and y-components of stress are decoupled from one another. This assumption
allows the desired equation to be deduced through an examination of a beam unit
cross-sectional width.
z
σx σ x + dσ x
Vx τ xz
M
τ zx
τ xz
The equilibrium conditions in the horizontal direction and for total moment are
∂τ xz ∂σ x
---------- + --------- = 0 Eq. 22-23
∂z ∂x
∂M x
V x + ---------- = 0 Eq. 22-24
∂x
Now, if the location of the neutral surface is denoted by zx and ρ is the radius of
curvature of the beam, the axial stress E x may be expressed in the form
Ex ( zx – z )
σ x + -------------------------- = 0 Eq. 22-25
( EI ) x
Relation 20 may be differentiated with respect to x combined with Eq. 22-23 and
Eq. 22-24. In a region of constant E x the result may be integrated to yield the
following expression
Vx 2
z
τ xz = C i + --------------  z x z – ----- E xi zi – 1 < z < zi Eq. 22-26
( EI ) x 2
Eq. 22-30 is particularly convenient to use in the analysis of n-ply laminates because
sufficient conditions exist to determine the constants Ci (i = 1,2,...,n) and the
“directional bending center” z x . For example, consider the following laminated
configuration
CHAPTER 22 649
Laminates
i=3
i=2 x
i=1
At the bottom surface ( i = 1 , z = z0 , and τxz = 0 )
2
–Vx  z0 
C 1 = --------------  z x z 0 – -------- E x1 Eq. 22-27
( EI ) x  2
and for the first ply at the interface between plies i = 1 and i = 2 ( z = z1 )
Vx 1 2 2
( τ xz ) = + -------------- z x ( z 1 – z 0 ) --- [ z 1 – z 0 ] E x1 Eq. 22-28
1 ( EI ) x 2
At this interface between plies i = 1 and i = 2 ,
2
Vx  z1 
( τ xz ) = C 2 + --------------  z x z 1 – -------- E x2 Eq. 22-29
2 ( EI ) x  2 
and as ( τ xz ) 2 = ( τ xz ) 1 at z = z1 ,
V x E x2 1 2
C 2 = ( τ xz ) – --------------- z x z 1 – --- z 1 Eq. 22-30
1 ( EI ) x 2
Then, in the ply z 1 < z < z 2 the shear is
V x E x2 1 2 2
τ xz ( z ) = ( τ xz ) --------------- z x ( z – z 1 ) – --- ( z – z 1 ) Eq. 22-31
1 ( EI )
x
2
In general, for any ply, Z i – 1 < z < z i , the shear is
V x E xi 1 2 2
τ xz ( z ) = ( τ xz ) -------------- z x ( z – z i – 1 ) – --- ( z – ( z i – 1 ) ) Eq. 22-32
i – 1 ( EI )
x
2
At any ply interface, z i , the shear is therefore
i
Vx 1
( τ xz ) = --------------
i ( EI ) x
∑ E xj T j z – --- ( z j + z j – 1 )
2
Eq. 22-33
j = 1
where T j = zj – ( zj – 1 ) .
Note that the shear at the top face, ( τxz ) n , is zero and therefore
n n
Vx
( τ xz ) = ----------- z x
n ( EI )
∑ E xj T j – ∑ E xj T j ( z j + z j – 1 ) ⁄ 2 = 0 Eq. 22-34
j = 1 j = 1
Eq. 22-34 proves that if z x is the bending center, the shear at the top surface must be
zero.
Eq. 22-32 could be substituted into Eq. 22-23 and integrated. A better form of
Eq. 22-34, for this purpose is
V x E xi 1 2 2
( τ xz ( z ) ) = -------------- f xi + z x ( z – z i – 1 ) --- ( z – ( z i – 1 ) ) Eq. 22-35
i ( EI ) x 2
where
i–1
1 1
f xi = --------
E xi ∑ E xj T j z x – --- ( z j + z j – 1 )
2
Eq. 22-36
j = 1
Substituting Eq. 22-35 into Eq. 22-26 and integrating the results, we obtain
N
1 T 1
------ = -------------- ∑ -------- R xi Eq. 22-37
Gx 2 G
( EI ) x i = 1 xi
where
2  1 2
R xi = ( E xi ) T i  f xi + ( z x – z i – 1 )T i – --- T i f xi Eq. 22-38
 3 
1 1  2 1 2 1 1 2 2
+  --- ( z x – 2 z i – 1 ) – --- T i z x T i +  --- z i – 1 + --- z i – 1 T i + ------ T i T i
3 4  3 4 20 
This expression for the inverse shear modulus for the x-direction may be generalized
to provide for the calculation of each term in the two-by-two matrix of shear moduli.
CHAPTER 22 651
Laminates
n –1
T i –1
[ G kl ] = ------------------- ∑ [ G kl ] R ki Eq. 22-39
2
( EI ) kk i = 1
where:
k = 1,2
l = 1,2
–1
Note that if no shear is given, [ G i ] = 0, and also that in Eq. 22-32
where [ G 2 ] ∗ is calculated in the same manner as [ G 2 ] except that Poisson’s Ratio is

set to zero.
The moduli for individual plies are provided through user input because, in
general, G 12 ≠ G 21 will be used for the coupling terms. Finally,
G 11 ( G 12 ) avg
[ G3 ] = Eq. 22-40
( G 12 ) avg G 22
As an example, consider a single layer element. For this case let zi – 1 = – T ⁄ 2 , z = 0 ,

3
f o = 0 , and EI = ET ⁄ 12 . Evaluating Eq. 22-39 we obtain
2 5
2 5 1 1 1 E T
Ri = E T ------ – --- + ------ = -------------
12 8 20 120
and
1  ( 12 ) 2 T  E 2 T 5  6
---- =  ------------------ ⋅  ------------------ = ----------
G 2
 E T   6 120 G 1 5G 1
which provides the correct factor for a nonuniform shear distribution in a plate.
The coefficients of thermal expansion derived for membrane-bending coupling,

which appear in the A1, A2, and A12 fields of MAT2 entry and correspond to the
MID4 Field on the PSHELL, require special interpretation. They are given by:
{ α MAT2 } = [ G ij ] { α ACTUA L } Eq. 22-41

MAT2
To obtain the actual values of A1, A2, and A12 { αMAT2 } , you must solve Eq. 22-41.
22.3 Understanding Laminate Failure Indices

When you use the PCOMP entry, you can optionally output a failure index for
individual laminae. Failure indices assume a value of one on the periphery of a
failure surface in stress space.
• If the failure index is < 1, the lamina stress is interior to the periphery of the
failure surface and the lamina is assumed to be safe.
• If the failure index is > 1, the lamina stress is exterior to the periphery of the
failure surface and the lamina is assumed to have failed.
These failure indices represent a phenomenological failure criterion in that only an

occurrence of a failure is indicated and not the mode of failure. In the present context,
concern is with the analytical definition of a failure surface in stress or strain space
for use with laminae under biaxial loading. NX Nastran provides four commonly
used definitions of the failure surface: Hill's Theory, Hoffman's Theory, Tsai-Wu
Theory, and maximum strain theory.
Selecting a Failure Criterion

NX Nastran produces a very large amount of output if you request the stresses for all
elements for all laminae. To help you to review this information, you can have the
software calculate a failure index for each ply. This failure index is obtained by
considering the failure criteria for unidirectional fiber composites as in the
commonly used failure theories. You can select one of the following failure criteria
for composites:
• Hill's Theory
• Hoffman's Theory
• Tsai-Wu Theory
• Maximum Strain
Note: If you specify a failure theory (using the FT option in field 6 of the PCOMP
entry), you must also specify the allowable shear stress of the bonding material
(using the SB option in field 5).
In the analysis of isotropic materials, strength is independent of the orientation of the

body under load and one may compare the largest computed principal stress with an
allowable stress to establish the integrity of the structure. Laminated composites, on
the other hand, are orthotropic materials and may exhibit unequal properties in
tension and compression. Thus, the strength of these orthotropic laminae is a
function of body orientation relative to the imposed stresses.
CHAPTER 22 653
Laminates
As the evaluation of the matrices of material moduli for laminated composites

provides sufficient information to determine the actual stress field sustained by the
material, the determination of structural integrity will depend on the definition of
an allowable stress field. The basic ingredient of this definition is the establishment
of a set of allowable stresses or strengths in the principal material directions.
Allowable tensile stress in the principal x (or 1) direction of the
Xt =
material.
Allowable compressive stress in the principal x (or 1) direction of
Xc =
the material.
Allowable tensile stress in the principal y (or 2) direction of the
Yt =
material.
Allowable compressive stress in the principal y (or 2) direction of
Yc =
the material.
S = Allowable shear stress in the principal material system.
Hill’s Theory (HILL)

Hill’s failure theory for orthotropic materials that have the same strength in tension
and compression, i.e., x t = x c and y t = yc can be expressed as:
σ 12 σ 1 σ 2 σ 22 τ 12 2
Failure Index = ------ – ------------- + ------ + -------- Eq. 22-42
X2 X2 Y2 S2
where:
X is allowable stress in 1-direction
Y is allowable stress in 2-direction
S is allowable stress in shear
and X = X t if σ1 is positive or X = X c if σ 1 is negative and similarly for Y and σ 2 .

For the interaction term σ 1 σ2 ⁄ X 2 , X = X t if σ 1 σ2 is positive or X = X c if σ1 σ 2 is
negative.
A plot of the above equation obtained by setting the failure index to 1 on the σ1 - σ2
plane yields an ellipse and is the anisotropic yield criterion of Hill (modified later
by Tsai, and hence also sometimes known as the Tsai-Hill theory). Therefore, if the
failure index so calculated is less than 1, the ply stresses are inside the yield ellipse
and the ply is said to be “safe”; conversely, if the failure index is greater than 1, the
ply stresses are outside the yield ellipse and the ply has failed.
Hoffman’s Failure Theory (HOFF)

Hoffman’s theory for an orthotropic lamina in a general state of plane stress with
unequal
www.cadfamily.com tensile and compressive strengths is given by
1 1 1 1 σ 12 σ 22 σ 122 σ1 σ2
 ----
- – ------ σ 1 +  ----- – ----- σ 2 + ------------ + ----------- + --------- – ------------- = 1 Eq. 22-43
X X c Y Y c Xt Xc Yt Yc S2 Xt Xc
t t
The failure index is obtained by evaluating the left-hand side of the above equation.
Note that this theory takes into account the difference in tensile and compressive
allowable stresses by using linear terms in the equation.
Tensor Polynomial Theory of Tsai-Wu (TSAI)

The theory of strength for anisotropic materials proposed by Tsai and Wu specialized
to the case of an orthotropic lamina in a general state of plane stress with unequal
tensile and compressive strengths is
F 1 σ 1 + F 2 σ 2 + F 11 σ 12 + F 22 σ 22 + 2F 12 σ 1 σ 2 + F 66 σ 12
2 = 1 Eq. 22-44
where:
1 1
F1 = X
----- – ------
t Xc
1 1
F 2 = Y---- – ----
-
t Yc
1
F 11 = + -----------
Xt Xc
1
F 22 = + ----------
Y t Yc
1
F 66 = + ----
-
2 S
and F 12 is to be evaluated experimentally.

The magnitude of F 12 is, however, constrained by the following inequality called a
“stability criterion”:
2 > 0
F 11 F 22 – F 12 Eq. 22-45
The necessity of satisfying the stability criterion, together with the requirement that
F 12 be determined experimentally from a combined stress state, poses difficulties in
1
the use of this theory. Narayanaswami and Adelman have suggested that F 12 be set
to zero and that use of Hoffman's theory or Tensor Polynomial theory with F 12 = 0 is
a preferred alternative to the experimental determination of F 12 . If you have a value
1
Narayanaswami, R., and H. M. Adelman, “Evaluation of the Tensor Polynomial
and Hoffman Strength Theories for Composite Materials,” Journal of Composite
www.cadfamily.com Materials, Vol. II, 1977, p. 366.
CHAPTER 22 655
Laminates
for use with F 12 in the theory, you can input that value in the MAT8 Bulk Data
entry; otherwise, with this failure theory, NX Nastran sets F 12 to 0.0. The left-hand
side of the above equation will be evaluated as the failure index by this theory.
NX Nastran calculates the failure index of bonding material as the maximum

interlaminar shear stress divided by the allowable bonding stress. NX Nastran
writes the failure indices for all the plies into the OEFIT (Output Element Failure
Index Table) output if stresses are requested. The failure index for the element is the
largest value of the failure indices for all plies of the element.
Maximum Strain Theory (STRN)

The midplane strains and curvatures are available in the element coordinate system.
From these, the stresses and strains in each individual lamina along the fiber
direction and transverse direction can be easily calculated. You can use the STRAIN
case control command to request the output of lamina strains.
The maximum strain criteria has no strain interaction terms. The strain allowables
specified on the MAT8 entry for each lamina include
Xt , Xc Allowable strains in tension and compression,

respectively, in the longitudinal direction.
Yt , Yc Allowable strains in tension and compression,

respectively, in the transverse direction.
S Allowable strain for inplane shear.
The Failure Index is calculated using
ε1 ε1
----- (or ------ if ε 1 compressive)
Xt Xc
ε2 ε2
----- (or ----- if ε 2 compressive)
Xt Yc
and
γ 12
-------
S
ε ε γ 12
i.e., the Failure Index = MAX  ----1- ,  ----2- ,  ---------
- .
X Y S 
Because you must understand which mode of Failure Index is critical; i.e.,
longitudinal (1), transverse (2), or shear (12), NX Nastran prints the mnemonic 1, 2
or 12 alongside the FP value to indicate the critical direction.
Note: There is no change in the way the Failure Index is calculated for interlaminar
shear stresses.
When you use the maximum strain theory, you may want to specify lamina stress
allowables instead of strain allowables on the MAT8. To do this, leave the STRN field
on the MAT8 entry blank.
For this case, NX Nastran calculates the failure indices using
ε 1 ⋅ E 11 ε 1 ⋅ E 11
-------------------- (or -------------------- if ε 1 compressive)
Xt Xc
ε 2 ⋅ E 22 ε 2 ⋅ E 22
-------------------- (or -------------------- if ε 2 compressive)
Yc Yc
and
γ 12 ⋅ G 12
---------------------------
S
ε 1 ⋅ E 11 ε 2 ⋅ E 22 γ 12 ⋅ G 12
i.e., the Failure Index = MAX  -----------------
- ,  ------------------ ,  ------------------------
X Y S
22.4 Calculating Interlaminar Shear Stress and Strains

The two-dimensional plate theory used in the CQUAD4, CQUAD8, CTRIA3 or
CTRIA6 elements doesn’t allow for the exact calculation of interlaminar stresses.
High values of these interlaminar stresses can lead to failures that are unique to
composite materials.
NX Nastran uses an approximate technique to determine the interlaminar shear

stresses. The basic assumption in this approximate technique is that the x- and y-
components of stress are decoupled from one another. The approximation consists of
neglecting the effects of twisting moments in the computation of interlaminar shear
stresses. For the majority of composite material structural analyses, this
approximation generally yields satisfactory results for all elements in the model
except those at the edges of the structure.
Eq. 22-33 is used to calculate ( τ xz ) i for the i-th lamina and z x is calculated from
Eq. 22-34. Similar expressions are used to calculate ( τ yz ) and zy .
The interlaminar shear strains are calculated by
τ 1z
γ 1z = ---------
G 1z
and
CHAPTER 22 657
Laminates
τ 2z
γ 2z = ---------
G 2z
22.5 Examples: Defining Laminate Material Properties

The following examples demonstrate how to use PSHELL and PCOMP to model a
composite honeycomb section.
Example: Using PCOMP to Model a Honeycomb Plate

As an example of the PCOMP entry, consider the cantilevered honeycomb plate
shown in Figure 22-4. Although the honeycomb structure is not considered a
composite layup, it can be analyzed effectively using the PCOMP entry.
Fixed Edge
t = 0.42 in
6
E = 30 x 10 psi 1000 lb
2000 lb
Z
Y 1000 lb 10 in
30 in
Top Aluminum Face Sheet - T = 0.02 in

Honeycomb Core - T = 0.35 in
Lower Aluminum Face Sheet - T = 0.05 in
Figure 22-4 Honeycomb Cantilever Plate
The material properties of the honeycomb section are listed in Table 22-2. Note that
the face sheets are not of the same thickness. The result is a nonsymmetric plate
(Figure 22-5), which can be noted by the MAT2 entry for the MID4 field of the
equivalent PSHELL.
Table 22-2 Honeycomb Material Properties
Modulus of Tensile Shear

Compression
Material Elasticity Limit 3 Limit
6 3 Limit (10 psi) 3
(10 psi) (10 psi) (10 psi)
Aluminum 10.0 35 35 23
Face Sheets
Core 0.0001 0.05 0.3 0.2
Bonding – – – 0.1
Material
Part of the input file for the honeycomb plate is shown in Listing 22-1. The load on
the plate is a 4000-pound uniform load acting in the plane of the plate. The output
requests include the element force, laminae stress, laminae strain, and failure index
table.
A selected portion of the output is shown in Figure 22-5.
Listing 22-1 Honeycomb Plate
$
$ FILENAME - PCOMP1.DAT
$
CQUAD4 1 100 1 2 6 5
CQUAD4 2 100 2 3 7 6
CQUAD4 3 100 3 4 8 7
CQUAD4 4 100 5 6 10 9
CQUAD4 5 100 6 7 11 10
CQUAD4 6 100 7 8 12 11
$
FORCE 1 12 0 1. 1000.0 0.0 0.0
FORCE 1 4 0 1. 1000.0 0.0 0.0
FORCE 1 8 0 1. 2000.0 0.0 0.0
$
SPC1 1 123456 1 5 9
$
PCOMP 100 100. STRN
120 0.05 0. YES 130 0.35 0.0 YES
120 0.02 0. YES
$
MAT1 120 10.+6 0.3
35.E3 35.E3 23.E3
MAT1 130 100. 0.3
50. 300. 200.
CHAPTER 22 659
Laminates
*** USER INFORMATION MESSAGE 4379, THE USER SUPPLIED PCOMP BULK DATA CARDS ARE REPLACED BY THE FOLLOWING PSHELL AND MAT2 CARDS.
PSHELL 100 100000100 4.2000E-01 200000100 1.0000E+00 300000100 1.0000E+00 0.0000E+00
-2.1000E-01 2.1000E-01 400000100
MAT2 100000100 1.8316E+06 5.4948E+05 0.0000E+00 1.8316E+06 0.0000E+00 6.4106E+05 0.0000E+00
0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00
0
MAT2 200000100 4.4895E+06 1.3469E+06 0.0000E+00 4.4895E+06 0.0000E+00 1.5713E+06 0.0000E+00
0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00
0
MAT2 300000100 3.9189E+01 0.0000E+00 0.0000E+00 3.9189E+01 0.0000E+00 0.0000E+00 0.0000E+00
0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00
0
MAT2 400000100 3.2705E+05 9.8115E+04 0.0000E+00 3.2705E+05 0.0000E+00 1.1447E+05 0.0000E+00
0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00
0

8 G 1.004209E-02 2.155610E-17 -3.155122E-01 -3.609077E-17 2.089842E-02 .0
HONEYCOMB COMPOSITE CANTILEVER PLATE JULY 1, 2004 NX NASTRAN 6/28/04 PAGE 9
F O R C E S I N Q U A D R I L A T E R A L E L E M E N T S ( Q U A D 4 )
ELEMENT - MEMBRANE FORCES - - BENDING MOMENTS - - TRANSVERSE SHEAR FORCES -

ID FX FY FXY MX MY MXY QX QY
4 2.000000E+02 2.784165E+01 -9.330303E+00 -4.593548E-15 -3.536165E-03 4.504795E-03 -7.892992E-17 -9.747001E-04
S T R E S S E S I N L A Y E R E D C O M P O S I T E E L E M E N T S ( Q U A D 4 )
ELEMENT PLY STRESSES IN FIBRE AND MATRIX DIRECTIONS INTER-LAMINAR STRESSES PRINCIPAL STRESSES (ZERO SHEAR) MAX
ID ID NORMAL-1 NORMAL-2 SHEAR-12 SHEAR XZ-MAT SHEAR YZ-MAT ANGLE MAJOR MINOR SHEAR
4 1 2.08187E+03 2.89631E+02 -9.68898E+01 -2.03948E-16 -2.51853E-03 -3.09 2.08710E+03 2.84409E+02 9.01344E+02
4 2 3.49114E-02 4.86145E-03 -1.63057E-03 -2.03936E-16 -2.51839E-03 -3.10 3.49996E-02 4.77323E-03 1.51132E-02
4 3 4.79470E+03 6.67919E+02 -2.24262E+02 -2.82614E-24 -3.48998E-11 -3.10 4.80685E+03 6.55768E+02 2.07554E+03
F A I L U R E I N D I C E S F O R L A Y E R E D C O M P O S I T E E L E M E N T S ( Q U A D 4 )
ELEMENT FAILURE PLY FP=FAILURE INDEX FOR PLY FB=FAILURE INDEX FOR BONDING FAILURE INDEX FOR ELEMENT FLAG
ID THEORY ID (DIRECT STRESSES/STRAINS) (INTER-LAMINAR STRESSES) MAX OF FP,FB FOR ALL PLIES
4 STRAIN 1 .0570 -1
.0000
2 .0007 -1
.0000
3 .1313 -1
.1313
S T R A I N S I N L A Y E R E D C O M P O S I T E E L E M E N T S ( Q U A D 4 )
ELEMENT PLY STRAINS IN FIBRE AND MATRIX DIRECTIONS INTER-LAMINAR STRAINS PRINCIPAL STRAINS (ZERO SHEAR) MAX
ID ID NORMAL-1 NORMAL-2 SHEAR-12 SHEAR XZ-MAT SHEAR YZ-MAT ANGLE MAJOR MINOR SHEAR
4 1 1.99499E-04 -3.34931E-05 -2.51914E-05 -5.30264E-23 -6.54819E-10 -3.09 2.00177E-04 -3.41721E-05 2.34350E-04
4 2 3.34529E-04 -5.61196E-05 -4.23949E-05 -5.30233E-18 -6.54781E-05 -3.10 3.35676E-04 -5.72665E-05 3.92943E-04
4 3 4.59433E-04 -7.70492E-05 -5.83081E-05 -7.34797E-31 -9.07396E-18 -3.10 4.61012E-04 -7.86288E-05 5.39641E-04
Figure 22-5 Composite Output of the Honeycomb Plate
The plate results indicate that the plates are bending downward due to the axial load,
which is expected from a nonsymmetric cross section. The laminae stresses show
that the face sheets are highly stressed while the core has virtually no stress. The
failure index table shows the failure for each laminae and the bonding material to be
less than 1, indicating that the structure will not fail. Note, however, that since each
lamina is assumed to be in a state of plane stress, the failure index doesn’t take into
account the crushing of the core.
As a final example showing the use of the PCOMP entry, consider the model shown
in Figure 22-6.
Figure 22-6 shows an example of the data entries required to define a five layer
composite material. A typical element (with grid points 101, 102, 103, and 104) in the
model is magnified to illustrate the individual layers.
CHAPTER 22 661
Laminates
Y5
Element
X5
Ply Angle Directions
Z5
103 –45 90
Projected onto the 0 45
element surface.
zm = ze
5 102
ym 4
3
2
90 1 Ply Number
xm .007
–45 .005
104
0 .005
45 .005
0 .003
Ply Angle 3
Ply Thickness
1
1 101
1
2
Ply Material
CQUAD4, 101, 1, 101, 102, 103, 104, 5
PCOMP, 1, , , 10000.0, STRN,
+, 2, .003, 0.0, YES, 1, .005, 45.0, YES,
+, 1, .005, 0.0, YES, 1, .005, -45.0, YES,
+, 3, .007, 90.0, YES
MAT8, 1, 2.0E+7, 2.0E+6, 0.35, 1.0E+6, 1.0E+6, 1.0E+6, 0.0
+, 0.0, 0.0, 0.0, .007, .006, .007, .006, 15000.,
+, , ,1.0
MAT8, 2, 1.0E+7, 1.0E+7, 0.05, 1.0E+6, 1.0E+6, 1.5E+6, 0.0
+, 0.0, 0.0, 0.0, .007, .006, .007, .006, 15000.,
+, , ,1.0
MAT8, 3, 0.8E+7, 0.8E+7, 0.05, 0.7E+6, 0.7E+6, 1.3E+6, 0.0
+, 0.0, 0.0, 0.0, .006, .005, .006, .005, 12000.,
+, , ,1.0
CORD2R, 5, , 0.,0.,0., 0.,-1.,0.,
+, 0.,0.,1.
Figure 22-6 Typical Composite Layup

In the example above, the direction of the x-axis of the material coordinate system
(xm) is defined by the projection of the x-axis of coordinate system 5 onto the element.
Using this technique, the direction of material coordinate system is independent of
the shape of the element.
The material direction of all the elements in the model can be defined by referencing
a single coordinate system. The MCID field on the CQUAD4 Bulk Data entry defines
the ID of this coordinate system. The coordinate system is shown in the figure as the
x5-, y5-, z5-axis and is defined on the CORD2R Bulk Data entry.
• The x-axis of the material coordinate system can also be defined with an
angle. If the THETA or MCID field on the CQUAD4 Bulk Data entry is an
integer, then the value is taken as a coordinate system. If it is a real number
, then the value is taken as an angle.
• The z-axis (zm) of the material coordinate system is in the direction of the
z-axis of the element coordinate system (ze).
• The direction of the y-axis (ym) follows the right-hand rule.
The CQUAD4 entry references the PCOMP entry, which defines the following layup:
Ply Material Thickness Angle

1 2 0.003 0.0
2 1 0.005 45.0
3 1 0.005 0.0
4 1 0.005 -45.0
5 3 0.007 90.0
The composite material is made of 5 plies. Ply number 1 is at the most negative zm.
Note that since zm is in the direction of ze, the element grid point connectivity order
determines which is the top and bottom surface of the composite and thus, the
stacking direction. Ply angle 0 is in the direction of xm. The positive sense of the ply
angle is in the direction from xm to ym. The PCOMP entry references the MAT8 data
entries by the MID field.
The example uses three different materials. Note that strain allowables are specified
on the MAT8 entry. To let NX Nastran know that these values are strain allowable,
the STRN field must be set equal to 1.0.
• Material 1 is typical of a graphite epoxy tape. A tape has continuous fibers

that all point in the same direction. Note that E1 is ten times the value of E2.
• Material 2 is also a tape.
CHAPTER 22 663
Laminates
• Material 3 is a cloth since E1 equals E2. A cloth made of woven fibers has
approximately the same number of fibers going in the 1 direction as in the
2 direction.
Directions 1 and 2 refer to the ply coordinate system. The ply material
properties on the MAT8 entry are in the ply coordinate system.
• The 1-axis of the ply coordinate system is in the direction of the ply
angle.
• The 3-axis is in the direction of zm (or ze).
• The 2-axis follows the right-hand rule.
Ply results are output in the ply coordinate system.
Example: Using PSHELL to Model a Honeycomb Section

If you’re only interested in the overall behavior of the honeycomb composite
section, then you can use an equivalent PSHELL/MAT2 or PSHELL/MAT1
combination to model the PCOMP/MAT8 effect. This representation yields the
correct equivalent stiffness and, consequently, the correct displacements. However,
the stresses in general are meaningless since the information regarding the
individual layers are not available to NX Nastran with the PSHELL/MATi
combination. If you want to obtain the stresses, you should use the PCOMP/MAT8
combination.
This example shows how to model the honeycomb section shown in Figure 22-7
using PSHELL/MAT1. The assumptions are that the membrane and bending loads
are carried by the facesheets and the transverse shear loads are carried by the
honeycomb core. The cross section consists of two 0.01 in thick aluminum
facesheets sandwiching a 0.75 in thick honeycomb core. The properties of the
section are shown in Figure 22-7.
Aluminum Facesheets Honeycomb Core

E1=E2=1.E7 E1=E2=1000.
G12=G13=G23=3.85E6 G12=100., G13=1.E5,
G23=5.E4
Aluminum Facesheet — T=0.01"
Aluminum Honeycomb Core — T=.75"
Aluminum Facesheet — T=0.01"
www.cadfamily.comFigure 22-7 Honeycomb Cantilever Plate Represented by PSHELL/MAT1

The calculation of the equivalent properties is shown below:
2 2 –3
I = 2y A = 2 ( .38 ) ( .01 ) = 2.888 ⋅ 10
T = 0.02
–3
12I 12 ( 2.88 ⋅ 10 )
-------- = --------------------------------------- = 4332
3 3
T ( 0.02 )
T s ≈ .75 + .02 = .77
Ts .77
----- = ------- = 38.5
T .02
5 4
1 ⋅ 10 + 5 ⋅ 10 4
G core ≈ ------------------------------------------- = 7.5 ⋅ 10
2
The transverse shear moduli (G13 and G23) are different in the two transverse
directions, which is typical of honeycomb core. Since the MAT1 entry only allows a
single shear modulus, an average Gcore is used.
A more accurate representation of G13 and G23 is to use the MAT2 entry instead of
the MAT1 entry. Listing 22-2 and Listing 22-3 show the input files using the
PCOMP/MAT8 and PSHELL/MAT1 entries, respectively. As shown in the
corresponding abridged output files in Figure 22-8 and Figure 22-9, the
displacements calculated by using the PSHELL/MAT1 combination are comparable
with the displacements obtained using the PCOMP/MAT8 combination.
CHAPTER 22 665
Laminates
Listing 22-2 Honeycomb Sandwich Using PCOMP
$ filename - pcomp2.dat
ID LINEAR,PCOMP2
SOL 101
TIME 5
CEND
TITLE = HONEYCOMB COMPOSITE CANTLEVER PLATE
SUBTITLE = SYMMETRIC - .01/.75/.01
$
SPC = 1
DISP = ALL
$
SUBCASE 1
LABEL = AXIAL LOAD
LOAD = 1
$
SUBCASE 2
LABEL = VERTICAL SHEAR LOAD
LOAD = 2
$
SUBCASE 3
LABEL = VERTICAL TWIST LOAD
LOAD = 3
$
BEGIN BULK
$
PARAM POST -1
$
GRDSET 6
GRID 1 0.0 -10. 0.0
GRID 2 10. -10. 0.0
GRID 3 20. -10. 0.0
GRID 4 30. -10. 0.0
GRID 5 0.0 0.0 0.0
GRID 6 10. 0.0 0.0
GRID 7 20. 0.0 0.0
GRID 8 30. 0.0 0.0
GRID 9 0.0 10. 0.0
GRID 10 10. 10. 0.0
GRID 11 20. 10. 0.0
GRID 12 30. 10. 0.0
$
CQUAD4 1 100 1 2 6 5
CQUAD4 2 100 2 3 7 6
CQUAD4 3 100 3 4 8 7
CQUAD4 4 100 5 6 10 9
CQUAD4 5 100 6 7 11 10
CQUAD4 6 100 7 8 12 11
$
FORCE 1 12 0 1. 1000.0 0.0 0.0
FORCE 1 4 0 1. 1000.0 0.0 0.0
FORCE 1 8 0 1. 2000.0 0.0 0.0
$
FORCE 2 12 0 1. 0. 0. 1.
FORCE 2 4 0 1. 0. 0. 1.
FORCE 2 8 0 1. 0. 0. 2.
$
Listing 22-2 Honeycomb Sandwich Using PCOMP (continued)
FORCE 3 12 0 1. 0. 0. 1.
FORCE 3 4 0 1. 0. 0. -1.
$
SPC1 1 123456 1 5 9
$
PCOMP 100
120 0.01 0. YES 130 0.75 0.0 YES
120 0.01 0. YES
$
MAT8 120 10.+6 10.+6 0.3 3.85+6
MAT8 130 1000. 1000. 100. 1.+5 0.5+5
ENDDATA
Listing 22-3 Honeycomb Sandwich Using PSHELL/MAT1
$ pshell2.dat
ID LINEAR,PSHELL2
SOL 101
TIME 5
CEND
TITLE = EQUIVALENT HONEYCOMB COMPOSITE CANTLEVER PLATE
SUBTITLE = SYMMETRIC - .01/.75/.01
$
SPC = 1
DISP = ALL
$
SUBCASE 1
LABEL = AXIAL LOAD
LOAD = 1
$
SUBCASE 2
LABEL = VERTICAL SHEAR LOAD
LOAD = 2
$
SUBCASE 3
LABEL = TWIST LOAD
LOAD = 3
$
BEGIN BULK
PARAM POST -1
$
GRDSET 6
GRID 1 0.0 -10. 0.0
GRID 2 10. -10. 0.0
GRID 3 20. -10. 0.0
GRID 4 30. -10. 0.0
GRID 5 0.0 0.0 0.0
GRID 6 10. 0.0 0.0
GRID 7 20. 0.0 0.0
GRID 8 30. 0.0 0.0
CHAPTER 22 667
Laminates
Listing 22-3 Honeycomb Sandwich Using PSHELL/MAT1 (continued)
GRID 9 0.0 10. 0.0

GRID 10 10. 10. 0.0
GRID 11 20. 10. 0.0
GRID 12 30. 10. 0.0
$
CQUAD4 1 100 1 2 6 5
CQUAD4 2 100 2 3 7 6
CQUAD4 3 100 3 4 8 7
CQUAD4 4 100 5 6 10 9
CQUAD4 5 100 6 7 11 10
CQUAD4 6 100 7 8 12 11
$
FORCE 1 12 0 1. 1000.0 0.0 0.0
FORCE 1 4 0 1. 1000.0 0.0 0.0
FORCE 1 8 0 1. 2000.0 0.0 0.0
FORCE 2 12 0 1. 0. 0. 1.
FORCE 2 4 0 1. 0. 0. 1.
FORCE 2 8 0 1. 0. 0. 2.
FORCE 3 12 0 1. 0. 0. 1.
FORCE 3 4 0 1. 0. 0. -1.
$
SPC1 1 123456 1 5 9
PSHELL 100 100 .02 100 4332. 200 38.5 +PSLL
+PSLL -.385 .385
MAT1 100 10.+6 0.3
MAT1 200 1000. .75+5
$
ENDDATA
1 HONEYCOMB COMPOSITE CANTLEVER PLATE JANUARY 13, 2004 NX NASTRAN 12/10/04 PAGE 10
SYMMETRIC - .01/.75/.01
0 AXIAL LOAD SUBCASE 1

1 G .0 .0 .0 .0 .0 .0
2 G 9.938049E-03 3.192859E-03 .0 .0 .0 .0
3 G 1.951274E-02 3.057678E-03 .0 .0 .0 .0
4 G 2.948488E-02 2.936741E-03 .0 .0 .0 .0
5 G .0 .0 .0 .0 .0 .0
6 G 9.154656E-03 -1.381969E-17 .0 .0 .0 .0
7 G 1.958685E-02 -4.734998E-17 .0 .0 .0 .0
8 G 2.954503E-02 -8.476228E-17 .0 .0 .0 .0
9 G .0 .0 .0 .0 .0 .0
10 G 9.938049E-03 -3.192859E-03 .0 .0 .0 .0
11 G 1.951274E-02 -3.057678E-03 .0 .0 .0 .0
12 G 2.948488E-02 -2.936741E-03 .0 .0 .0 .0
0 VERTICAL SHEAR LOAD SUBCASE 2

1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 8.081975E-03 2.360743E-04 -1.524400E-03 .0
3 G .0 .0 2.971653E-02 2.389541E-04 -2.685955E-03 .0
4 G .0 .0 5.863158E-02 1.268257E-04 -2.985823E-03 .0
5 G .0 .0 .0 .0 .0 .0
6 G .0 .0 9.109288E-03 -7.489637E-18 -1.693645E-03 .0
7 G .0 .0 3.090988E-02 -1.128154E-17 -2.562822E-03 .0
8 G .0 .0 5.927901E-02 -1.922195E-17 -3.002027E-03 .0
9 G .0 .0 .0 .0 .0 .0
10 G .0 .0 8.081975E-03 -2.360743E-04 -1.524400E-03 .0
11 G .0 .0 2.971653E-02 -2.389541E-04 -2.685955E-03 .0
12 G .0 .0 5.863158E-02 -1.268257E-04 -2.985823E-03 .0
0 VERTICAL TWIST LOAD SUBCASE 3

1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 -1.255258E-03 1.145127E-04 2.226868E-04 .0
3 G .0 .0 -3.594521E-03 3.597052E-04 2.259500E-04 .0
4 G .0 .0 -5.950768E-03 5.769896E-04 2.259347E-04 .0
5 G .0 .0 .0 .0 .0 .0
6 G .0 .0 -1.634409E-17 1.417446E-04 2.934695E-18 .0
7 G .0 .0 -5.219971E-17 3.523428E-04 4.029600E-18 .0
8 G .0 .0 -9.308500E-17 5.806674E-04 4.117684E-18 .0
9 G .0 .0 .0 .0 .0 .0
10 G .0 .0 1.255258E-03 1.145127E-04 -2.226868E-04 .0
11 G .0 .0 3.594521E-03 3.597052E-04 -2.259500E-04 .0
12 G .0 .0 5.950768E-03 5.769896E-04 -2.259347E-04 .0
Figure 22-8 Displacement Outputs for Honeycomb Sandwich Using PCOMP
CHAPTER 22 669
Laminates
1 EQUIVALENT HONEYCOMB COMPOSITE CANTLEVER PLATE JANUARY 13, 2004 NX NASTRAN 12/10/04 PAGE 9
SYMMETRIC - .01/.75/.01
0 AXIAL LOAD SUBCASE 1

1 G .0 .0 .0 .0 .0 .0
2 G 9.977354E-03 3.215835E-03 .0 .0 .0 .0
3 G 1.958953E-02 3.081032E-03 .0 .0 .0 .0
4 G 2.960244E-02 2.958689E-03 .0 .0 .0 .0
5 G .0 .0 .0 .0 .0 .0
6 G 9.187398E-03 1.784324E-18 .0 .0 .0 .0
7 G 1.966428E-02 9.149946E-18 .0 .0 .0 .0
8 G 2.966328E-02 2.280977E-17 .0 .0 .0 .0
9 G .0 .0 .0 .0 .0 .0
10 G 9.977354E-03 -3.215835E-03 .0 .0 .0 .0
11 G 1.958953E-02 -3.081032E-03 .0 .0 .0 .0
12 G 2.960244E-02 -2.958689E-03 .0 .0 .0 .0
0 VERTICAL SHEAR LOAD SUBCASE 2

1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 8.084896E-03 2.362314E-04 -1.527135E-03 .0
3 G .0 .0 2.977032E-02 2.404278E-04 -2.690023E-03 .0
4 G .0 .0 5.873456E-02 1.270834E-04 -2.989493E-03 .0
5 G .0 .0 .0 .0 .0 .0
6 G .0 .0 9.143887E-03 -1.200619E-18 -1.694704E-03 .0
7 G .0 .0 3.096434E-02 -3.004362E-18 -2.565396E-03 .0
8 G .0 .0 5.937611E-02 -4.946220E-18 -3.006373E-03 .0
9 G .0 .0 .0 .0 .0 .0
10 G .0 .0 8.084896E-03 -2.362314E-04 -1.527135E-03 .0
11 G .0 .0 2.977032E-02 -2.404278E-04 -2.690023E-03 .0
12 G .0 .0 5.873456E-02 -1.270834E-04 -2.989493E-03 .0
0 TWIST LOAD SUBCASE 3

1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 -1.277012E-03 1.157813E-04 2.219504E-04 .0
3 G .0 .0 -3.622530E-03 3.624341E-04 2.244598E-04 .0
4 G .0 .0 -5.978552E-03 5.825486E-04 2.239124E-04 .0
5 G .0 .0 .0 .0 .0 .0
6 G .0 .0 -3.011317E-18 1.428981E-04 5.786049E-19 .0
7 G .0 .0 -1.073759E-17 3.549739E-04 9.055212E-19 .0
8 G .0 .0 -2.047949E-17 5.861556E-04 1.015648E-18 .0
9 G .0 .0 .0 .0 .0 .0
10 G .0 .0 1.277012E-03 1.157813E-04 -2.219504E-04 .0
11 G .0 .0 3.622530E-03 3.624341E-04 -2.244598E-04 .0
12 G .0 .0 5.978552E-03 5.825486E-04 -2.239124E-04 .0
Figure 22-9 Displacement Outputs for Honeycomb Sandwich Using

PSHELL/MAT1
23
CHAPTER
Coupled Fluid-Structure Interaction
■ Overview of Coupled Fluid-Structure Interaction
■ Performing a Coupled Fluid-Structural Analysis
■ Acoustic Cavity Modeling
23.1 Overview of Coupled Fluid-Structure Interaction

NX Nastran provides you with several different methods to analyze the effects of
fluids upon a structure.
• NX Nastran contains an axisymmetric hydroelastic modeling capability

that lets you solve a variety of fluid problems with small motion,
compressibility, and gravity effects.
• NX Nastran allows you to compute virtual fluid mass.
• NX Nastran lets you perform a fully coupled fluid-structural analysis.
• NX Nastran lets you perform acoustic cavity modeling.
Hydroelastic modeling and analyzing virtual fluid mass are described in detail in the
NX Nastran Advanced Dynamic Analysis User’s Guide. Coupled-fluid structural
analysis and acoustic cavity modeling are described in the following sections.
See Also
• “Axisymmetric Fluids in Tanks” on page 114 of the NX Nastran Advanced

Dynamic Analysis User’s Guide
• “Virtual Fluid Mass” on page 128 of the NX Nastran Advanced Dynamic
Analysis User’s Guide
23.2 Performing a Coupled Fluid-Structural Analysis

NX Nastran has lets you perform a fully coupled fluid-structure analysis. Its
principal application is in the area of acoustic and noise control analysis; for example,
in the passenger compartments of automobiles and aircraft.
The approach used in this analysis is called the Pressure Method, which is analogous
to the Displacement Method in structural analysis except that pressures, instead of
displacements, are computed at the fluid points. The velocities and accelerations of
the fluid points are analogous to forces in structural analysis.
You can model the fluid with existing three-dimensional elements: CHEXA,
CPENTA, and CTETRA. These elements can assume the properties of irrotational
and compressible fluids suitable for acoustic analysis or other types of analyses
governed by the three-dimensional wave equation. In addition, two new elements
are available for acoustic analysis, the barrier and absorber. The absorber and barrier
elements may be used to analyze acoustic noise control devices.
Optionally, you can model the interface between the fluid and the structure so that
the grid points of the fluid are coincident with those of the structure. This is called a
matching mesh. If you don’t create coincident grid points at the interface between
the fluid and the structure, then it is called a nonmatching mesh. In either case and
CHAPTER 23 673
by default, NX Nastran automatically computes coupling for the stiffness and mass.
If you don’t want the software to compute coupling, include PARAM,ASCOUP,NO
in your input file.
Coupled fluid-structural analysis is available in the dynamic solution sequences

using the Direct Method (SOLs 107 through 109 and 200 with ANALYSIS =
DFREQ), and the Modal Method (SOLs 110 through 112 and 200 with ANALYSIS =
MFREQ and ANALYSIS = MTRAN). It should be noted that in SOLutions 110
through 112 and 200, the normal modes are computed separately for the fluid and
structural parts of the model; in other words, the uncoupled modes of the fluid and
structure are used in the modal formulation of the stiffness, mass, and damping. In
SOLs 110 through 112, the SDAMPING Case Control command and the parameters
G and W3 are applied only to the structural portion of the model. Design
Sensitivities may be computed in SOL 200.
In SOLs 103 and 106 (PARAM,NMLOOP), normal modes are computed separately
for the fluid and structure portions of the model and then put together in the
eigenvalue table and displacement printouts in the .f06 file output. Please note that
while the fluid and structure eigenvalues and eigenvectors are printed together,
they are really independent of each other and not true combined eigenvalues. The
combined normal modes computed in SOL 103 may not be used in a restart into
SOLs 110, 111, 112, or 200.
Defining the Fluid Model

You define fluid grid points on the GRID Bulk Data entry by specifying a value of -
1 for CD in field 7. You can define fluid elements on the CHEXA, CPENTA, and
CTETRA Bulk Data entries. Also, on the referenced PSOLID entry, you must specify
the character value PFLUID for FCTN in field 8 and MID, field 3, must reference a
MAT10 material entry. The MAT10 entry defines the bulk modulus and the mass
density properties of the fluid. If a PSOLlD entry defines fluid elements (FCTN =
PFLUID), then lSOP, field 7, defaults to 1 (FULL), resulting in a full integration
scheme.
Using the PANEL Bulk Data Entry

It is sometimes useful to find the contribution of a set of grid points to the noise level
in an acoustic cavity. For example, in an automobile the entire roof assembly
contributes significantly to the noise level inside the passenger cabin; as such it is
necessary to compute and minimize the contribution of this “panel.” A “panel” may
be defined with the PANEL entry, which references a set of structural grid points
listed on a SET1 Bulk Data entry.
Acoustic Absorber
The acoustic absorber element is simulated by an assembly consisting of a mass
attached to a spring and damper which are in parallel. The element and its properties
are defined by the CHACAB and PACABS entries, respectively. The element
consists of 8 to 16 structural grid points defining two surfaces.
The PACABS entry references three TABLEDi entries which define the resistance, Y1;
reactance, Y2; and the confidence level, Y3 of the material. These properties are a
function of frequency and are measured in the Standing Wave Tube Test (ASTM
Standard). This data is used to calculate the value of the mass, spring, and damper
used in the structural model that will emulate the behavior of the absorber material.
The last field on the PACABS entry defines the cutoff frequency which determines
the extent of the data used in the least square fit operations. This is needed because
finite element analysis is not applicable to acoustic analysis in the high frequency
range. In fact, it is recommended that this methodology be restricted to frequencies
with wavelengths much larger than the absorber thickness. Nevertheless, the user
may override this cutoff frequency so that the experimental results of short
wavelengths are used in the calculations.
Frequency Dependent Acoustic Absorber

The acoustic absorber element CACABSF is an element connecting 1 to 4 fluid grids
at the fluid-structure interface. Its frequency dependent impedance is specified on
the PAABSF entry.
If one grid point is given, then the impedance Z ( f ) = ZR + i ZI is the total impedance
at the point. If two grids are specified, then the impedance is the impedance per unit
length. If three or four points are specified, then the impedance is the impedance per
unit area. Z R ( f ) = TZREID ( f ) + B and ZI ( f ) = TZRIMD ( f ) – K ⁄ ( 2πf ) .
The resistance represents a damper quantity B. The reactance represents a quantity

of the type ( ωM – K ⁄ ω ) . The impedance is defined as Z = p ⁄ u· where p is the pressure
and u· is the velocity. The scale factor S is used in computing element stiffness and
damping terms as:
A 2πf Z I ( f )
k = --- ⋅ ------------------------ ∫ (of shape functions)
S Z R2 + Z I2
A ZR( f )
b = --- ⋅ --------------------- ∫ (of shape functions)
S Z R2 + Z I2
The absorption coefficient is calculated as:
CHAPTER 23 675
4 ( Z R ⁄ ρc )
α = -----------------------------------------------------------------------
2 2
( Z R ⁄ ρc + 1 ) + ( Z I ⁄ ρc )
Acoustic Barrier
The acoustic barrier element is synthesized with two masses sandwiching a spring.
The element and its properties are defined by the CHACBR and PACBAR entries,
respectively. The element consists of 8 to 16 structural grid points defining two
surfaces.
The PACBAR entry defines the mass-per-unit area of the backing and the septum,
and the resonant frequency of the barrier. The resonant frequency of the sandwich
is defined as the frequency with the lowest value of transmission loss (ASTM test).
This frequency will be used to calculate the value of the spring used in the structural
model that will emulate the behavior of the barrier material.
Loading
Loads on fluid elements are, in most cases, analogous to enforced displacements on
structural elements. The types of loading available for fluid elements are:
1. Constant, frequency-dependent, or time-dependent enforced pressure at

the grid points.
2. An acoustic source characterized by a volumetric flow rate and
corresponding to a power spectral density function.
Enforced Pressures
The Large Mass approach that is available for structural analysis is also used to
define an enforced pressure. This is not recommended for enforcing motion on
more than one point.
Acoustic Source
An acoustic source is assumed to be a pulsating sphere in infinite space and is
defined on the ACSRCE Bulk Data entry. The ACSRCE entry is selected by the
DLOAD Case Control command and contains the material properties of the source
and references a DAREA and TABLEDi entry. The TABLEDi entry defines the
power-versus-frequency curve characterizing the acoustic source. The ACSRCE
entry may also define a delay time and phase angle which is useful whenever
multiple sources are present.
Single-Point Constraints
Single-point constraints of the fluid (P = 0.0) may be enforced using the SPC entry or
the PS field on the GRID entry. However, only one degree-of-freedom per grid point
may be constrained. This type of boundary condition occurs at free surfaces.
Fluid-Structure Interface
The fluid-structure interface is determined automatically by the program.
A matching interface is defined by entering IDENT in the INTER field of the

ACMODL Bulk Data entry. This also requires that coincident grid points be defined
for the structure and fluid elements at the interface. The wetted structural elements
are determined by comparing grid point locations corresponding to structural
elements that are within a tolerance specified by FSTOL on the ACMODL entry. If
no structural elements are found, a warning message will be issued.
A nonmatching interface is defined by entering DIFF in the INTER field on the

ACMODL Bulk Data entry. You shouldn’t use an ACMODL Bulk Data entry if you
want a nonmatching interface. If ACMODL is present then 1. ≤ FSTOL ≤ 10.
Normal Modes Analysis

Since the fluid portion of the model may resonate within a different frequency range
than the structural portion, it is possible to specify a different eigenvalue extraction
method for each in the residual structure only. In modal analysis (SOLs 110 through
112 and 200), this is achieved by the specification of a keyword on the METHOD Case
Control command. The METHOD(FLUID) command references the ElGR or ElGRL
entry to be used for the fluid portion; METHOD(STRUCTURE) references the same
entries for the structural portion. Either one or both may be specified in the Case
Control Section. The default is STRUCTURE.
Fluid-structure models may be defined using superelements with the following
restrictions:
1. A superelement may contain either fluid or structural points, but not both.
(The residual structure may contain both.)
2. The grid points at the fluid-structure interface may be assigned to the
residual structure only.
This requires the specification of q-set points using the SEQSETi and ElGR or ElGRL
Bulk Data entries and the METHOD Case Control command.
CHAPTER 23 677
Output and Plotting

You can use the DISPLACEMENT or PRESSURE Case Control command to request:
• Pressure and peak sound pressure levels at the fluid points.

• RMS sound pressure level is requested by the PARAM,ACOUT,RMS Bulk
Data entry. This output is available in Modal and Direct Frequency
Response analysis only (SOLs 111, 108, and 200).
• Sound pressure level in dB and dBA—a peak reference pressure must be
specified on the PARAM, PREFDB Bulk Data entry. The dB level is
defined as dB = 20 log(P/PREFDB) (for air PREFDB = 20 × 10 – 6 Pascal).
In SOLutions 110, 111, 112, and 200, the SVECTOR, SDISPLACEMENT,

SVELOCITY, SACCELERATION Case Control commands result in separate output
for the fluid and structural portion of the model. The first output is for the structural
grid points and the second for the fluid grid points.
You can use the FORCE Case Control command to request particle velocity.
Basic Fluid Equations

In NX Nastran, the acoustic equations are based on small motion theory with
negligible convective momentum terms and locally linear pressure-density
relationship. Thus Euler’s equation is
·· 1
u f = – ----- ∇p Eq. 23-1
ρf
and the Continuity equation is
p = – β∇ ⋅ u f Eq. 23-2
with the compressibility β defined as
2
β = c ρf Eq. 23-3
Take the second time derivative of continuity to get
·· 1
∇ ⋅ u f = – --- p·· Eq. 23-4
β
and operate on Euler’s equation with the del operator to get
·· 1
∇ ⋅ u f = – ----- ∇ ⋅ ∇p
ρf
or
1 1
--- p·· – ----- ∇ ⋅ ∇p = 0 Eq. 23-5
β ρf
integrate this equation over the fluid volume and multiply by a virtual pressure δp
to get the integral
1 1
∫ --- p·· – ----- ∇ ⋅ ∇p dVδp = 0 Eq. 23-6
β ρf
V
This represents the virtual work expression for acoustic pressure. Integrating the
second term by parts using Green’s theorem and the identities:
∇ ⋅ ( αυ ) = ∇α ⋅ υ + α∇ ⋅ υ
α = δp
υ = ∇p
∇δp = δ∇p
δ  ∇p ⋅ ∇ p = 2δ∇p ⋅ ∇p
we get
1 1 1
∫ --β- p·· δp dV + δ ∫ -------
2ρ f
- ∇p ⋅ ∇pdV – ∫ ----- ∇p ⋅ dSδp
ρf
= 0 Eq. 23-7
V V S
The trick is to factor out the δ operator from each integral. The first term in the above
equation, however, contains p·· which prevents a simple factoring out of δp . Thus no
simple variational statement is directly possible. Because of this, it is standard
practice at this point to apply the Galerkin method. This is equivalent to assuming
that p but not p·· is subject to variation. This allows us to simply factor out the δp
from the integral containing p·· . Doing this and also defining u·· fn as the acceleration of
o
the fluid in the direction of the outward fluid normal, we obtain for the integral:
1 1 f
Π = ∫ --- pp·· + -------- ∇p ⋅ ∇p dV + ∫ u·· n p dS Eq. 23-8
β 2ρ f o
V S
and the variational statement becomes
CHAPTER 23 679
δΠ = 0 Eq. 23-9
Let the fluid pressure p at any point be defined as
p = Nf { p }
and
p·· = N f { p·· }
where N f is the row matrix of pressure shape functions and { p } is a vector of

nodal pressures. The expression for Π can be expressed as:
1
Π = p ∫ --β- { N f } N f dV { p·· }
V
1
+ p ∫ -------
2ρ f
- { ∇N f } ∇N f dV { p } Eq. 23-10
f
+ p ∫ { N f }u·· no dS
S
If n 0 is the fluid outward normal at the fluid-structure and n is the structure

boundary normal pointing into the fluid, then n0 = – n and the surface integral can
be written as
f s
p ∫ { N f }u·· n0 dS = – p ∫ { N f }u·· n dS
S S
where u·· sn is the normal acceleration of the structure.
If [ C ] is the direction cosine matrix for n and [ N s ] is the matrix of shape functions
for the structure, we can define
s
u·· n = C [ N s ] { u·· s }
Eq. 23-11
= N s { u·· s }
where { u·· s } is the matrix of nodal accelerations at the boundary and Ns is the row
matrix of shape functions for the structural normal.
s
– p ∫ { N f }u·· n dS = – p ∫ { Nf } N s dS { u·· s } Eq. 23-12
S S
We may then compute δΠ and define the following integrals:
1
[ Mf ] = ∫ --β- { N f } N f dV Eq. 23-13
V
1
[ Kf ] = ∫ ----
ρf
- { ∇N f } ∇ N f dV Eq. 23-14
V
T
[A ] = ∫ { Nf } N s dS Eq. 23-15
S
and arrive at the relationship
T
[ M f ] { p·· } + [ K f ] { p } – [ A ] { u·· s } = 0 Eq. 23-16
NX Nastran allows for the effect of acoustic source density, which represents a term
Q on the right-hand side of the continuity equation. When included into the virtual
work expression, this will yield an acoustic load { P f } on the right-hand side of
Eq. 23-16. Also, NX Nastran allows for frequency dependent wall impedance
z = p ⁄ u· f at the structure-fluid [ B f ] { p· } . Thus, Eq. 23-16 becomes:
T
[ M f ] { p·· } + [ B f ] { p· } + [ K f ] { p } – [ A ] { u·· s } = { P f } Eq. 23-17
The fluid pressure on the structure boundary causes surface tractions on the
structure represented by the standard relationship
T
{ Fs } = ∫ [ Ns ] { Φ } dS Eq. 23-18
S
But, the traction Φ is related to the fluid pressure by the relation
{ Φ } = –p { C } Eq. 23-19
or
T
{ F s } = – ∫ [ N s ] { C }p dS
S
Eq. 23-20
= – ∫ { N s } N f dS { p }
S
= –[ A ] { p }
The structure equations can now be written as:
[ M s ] { u·· s } + [ B s ] { u· s } + [ K s ] { u s } = { P s } – [ A ] { p } Eq. 23-21
CHAPTER 23 681
The fluid and structure equations can now be combined as
Ms 0  u··  B s 0  u· s  Ks A  us   Ps 
s
 +  +   =   Eq. 23-22
– A M f  p··  0 B f  p· 
T 0 Kf  p   Pf 
This is the fluid structure equation solved in NX Nastran. Note that this equation is
nonsymmetric. By default, NX Nastran solves a symmetric version of this equation
developed by Everstine as follows:
Let the velocity potential q be defined as
p = q· Eq. 23-23
and note the following:
du· s
u·· s = --------
dt
dp·
p·· = ------
dt
dp
p· = ------
dt
dq
p = ------
dt
and define G as
t
G = – ∫ P f ( τ ) dτ Eq. 23-24
0
Substitute q· for p in the structure equation and for the fluid equation replace the
vector terms with their derivative equivalents. Then multiplying the fluid equation
by -1 and integrating with respect to time and recombing, we get the symmetric
equation
Ms 0  u·· s  Bs A  u· s  Ks 0 us   Ps 
 +  +   =   Eq. 23-25
0 – M f  q··  A – B f  q· 
T 0 –Kf  q  G
The output from the solution is, however, expressed in terms of u s and p .
In the frequency domain, we assume a harmonic function of the form q = Q 0 e iωt as

a solution. Then the load integral becomes ∫ G 0 e iωt dτ = G 0 ⁄ ( iω )e iωt . The pressure is
recovered from the relationship p = iωq .
Participation Factor Equations

Following Sung and Nefske, the participation factors are obtained from Eq. 23-22 as
follows: Define the transformations
{ us } = [ Φs ] { ξ s } Eq. 23-26
{ p } = [ Φf ] { ξf } Eq. 23-27
where [ Φs ] are the uncoupled, undamped structural modes and [ Φ f ] are the
uncoupled, undamped, rigid-wall acoustic modes. The vectors { ξ s } and { ξ f } are the
modal amplitudes. Substituting these relations into Eq. 23-22 and pre-multiplying by
the modal matrices, we get the equation
T ·· T · T T  T 
Φs Ms Φs 0  ξs  Φ s Bs Φs 0  ξ s  Φ s K s Φ s Φ s AΦ f  ξs   Φs Ps 
 ··  + ·  +   =  T  Eq. 23-28
–Φf A Φs Φ Mf Φf  ξf 
T T T
0
T
Φ Bf Φf  ξf  0
T
Φf K f Φf  ξf  Φ P 
 f f
or
ms 0  ξ·· s  ·
bs 0  ξs  ks a  ξs   Qs 
 ··  +  +   =   Eq. 23-29
T
–a mf  ξf  0 b f  ξ· f  0 kf  ξf   Qf 
Define the following harmonic solution forcing frequency ω .
 ξs  iωt
{ξ} =   = e Eq. 23-30
 ξf 
· iωt
{ ξ } = iωe Eq. 23-31
·· 2 iωt
{ ξ } = –ω e Eq. 23-32
For the bottom equation of Eq. 23-29, we get:
2 T 2
ω [ a ] { ξ s } + [ – ω [ m f ] + iω [ b f ] + [ k f ] ] { ξ f } = { Q f } Eq. 23-33
Define [ Z2 ] as
2 –1
[ Z 2 ] = [ – ω [ m f ] + iω [ b f ] + [ k f ] ] Eq. 23-34
then
CHAPTER 23 683
2 T
{ ξf } = – ω [ Z2 ] [ α ] { ξs } + [ Z2 ] { Qf } Eq. 23-35
The fluid mode participation is defined as
[ Pf ] = [ Φf ] [ { ξf } ] Eq. 23-36
where [ { ξ f } ] is the diagonalized vector of fluid modal amplitudes per excitation

frequency.
The structure mode participation is defined as
2 T
[ Ps ] = –ω [ Φf ] [ Z2 ] [ a ] [ { ξs } ] Eq. 23-37
where [ { ξ s } ] is the diagonalized vector of structural modal amplitudes per

excitation frequency.
The fluid load participation is defined as
{ Pl } = [ Φf ] [ Z2 ] { Qf } Eq. 23-38
The fluid-structure panel participation is defined as
2 T T
[ P p ] = – ω [ Φ f ] [ Z 2 ] [ Φ f ] [ A ] panel [ Φ s ] [ { ξ s } ] Eq. 23-39
The fluid-structure panel-grid participation is defined as
2 T  T
[ Pg ] = –ω [ Φf ] [ Z2 ] [ Φf ]  Ab  [ Φs ] [ ξs ] Eq. 23-40
i
 
 
where  A Tb  are the columns extracted from the b-th boundary panel for panel grid
i and Φ s are the rows of the structural modal matrix corresponding to panel grid i.
23.3 Acoustic Cavity Modeling

With acoustic cavity modeling, you define a mesh over the longitudinal cross
section of an acoustic cavity. You connect finite elements are connected between
these points to define the enclosed volume.
You define the points with GRlDF data entries for the axisymmetric central fluid
cavity and by GRIDS for the radial slots. The GRlDF points are interconnected by
finite elements via the CAXlF2, CAXlF3, and CAXlF4 data entries to define a cross-
sectional area of the body of rotation. The CAXIF2 element data entry defines the
area of the cross section between the axis and two points off the axis (the GRlDF
points may not have a zero radius). The CAXIF3 and CAXlF4 data entries define
triangular or quadrilateral cross sections and connect three or four GRIDF points,
respectively. The density and/or bulk modulus at each location of the enclosed fluid
may also be defined on these entries.
The GRIDS points in the slot region are interconnected by finite elements via the
CSLOT3 and CSLOT4 data entries. These define finite elements with triangular and
quadrilateral cross-sectional shapes, respectively. The width of the slot and the
number of slots may be defined by default values on the AXSLOT data entry. If the
width of the slots is a variable, the value is specified on the GRIDS entries at each
point. The number of slots, the density, and/or the bulk modulus of the fluid may
also be defined individually for each element on the CSLOT3 and CSLOT4 entries.
The AXSLOT Bulk Data entry is used to define the overall parameters for the system.
Some of these parameters are called the “default” values and may be selectively
changed at particular cross sections of the structure. The values specified on the
AXSLOT entry will be used if the corresponding fields on the GRIDS, CAXlFi, or
CSLOTi entries are left blank. The fields RHOD and RHO (density) and BD and B
(bulk modulus) are properties of the fluid. If the value given for bulk modulus is
zero, the fluid is considered incompressible by the program. The fields MD and M
(number of slots) and WD (slot width) are properties of the geometry. The fields MD
and M define the number of equally spaced slots around the circumference, with the
first slot located at φ = 0° . The field N (harmonic number) is selected by the user
to analyze a particular set of acoustic modes. The pressure is assumed to have the
following distribution:
p ( r,z,φ ) = ( r,z ) cos Nφ
If N = 0 , the breathing and longitudinal modes will result. If N = 1 , the pressure at

φ = 180° will be the negative of the pressure at φ = 0° . If N = 2 , the pressures at
φ = 90° and φ = 270° will be the negative of that at φ = 0° . Values of N larger than
M/2 have no significance.
The interface between the central cavity and the slots is defined with the SLBDY Bulk
Data entries. The data for each entry consist of the density of the fluid at the interface,
the number of radial slots around the circumference, and a list of GRIDS points that
are listed in the sequence in which they occur as the boundary is traversed. In order
to ensure continuity between GRlDF and GRIDS points at the interface, the GRIDF
points on the boundary between the cylindrical cavity and the slots are identified on
the corresponding GRIDS data entries rather than on GRlDF entries. Thus, the
locations of the GRlDF points will be exactly the same as the locations of the
corresponding GRIDS points.
Various standard Bulk Data entries may be used for special purposes in acoustic
analysis. You can use the SPCi Bulk Data entry to constrain the pressures to zero at
specified points, such as at a free boundary. Dynamic load entries, direct input
www.cadfamily.commatrices, and scalar elements may be introduced to account for special effects.
CHAPTER 23 685
Assumptions and Limitations

The accuracy of the acoustic model depends upon the selection of the mesh of finite
elements. The assumption for each element is that the pressure field has a linear
variation over the cross section and a sinusoidal variation around the axis in the
circumferential direction. In areas where the pressure gradient changes are large,
such as near a sharp corner, the points in the mesh should be placed closer together
so that large changes in flow may be defined accurately by the finite elements.
The shape of the finite elements play an important part in the accuracy of the results.
It has been observed that long, narrow elements produce disproportionate errors.
Cutting a large square into two rectangles will not improve the results, whereas
dividing the square into four smaller squares may decrease the local error by as
much as a factor of ten.
The slot portion of the cavity is limited to certain shapes because of basic
assumptions in the algorithms. The cross section of the cavity normal to the axis
must have a shape that is reasonably well defined by a central circular cavity having
equally spaced, narrow slots. Various shapes are shown in Figure 23-1 in the order
of increasing expected error.
It is recommended that shapes such as the cloverleaf and square cross section be
analyzed with a full three-dimensional technique. The assumption of negligible
pressure gradient in the circumferential direction within a slot is not valid in these
cases.
The harmonic orders of the solutions are also limited by the width of the slots. The
harmonic number, N, should be no greater than the number of slots divided by two.
The response of the higher harmonics is approximated by the slot width correction
terms discussed in Section 17.1 of The NASTRAN Theoretical Manual.
SMALL
ERRORS
LARGE
ERRORS
Figure 23-1 Modeling Errors for Various Shapes
The output data for the acoustic analysis consist of the values of pressure in the
displacement vector selected via the Case Control command PRESSURE = n. The
velocity vector components corresponding to each mode may be optionally
requested by the Case Control command STRESS = n where n is the set number
indicating the element numbers to be used for output, or by the words STRESS =
ALL.
You can request plots of the finite element model and/or of the pressure field with
the NX Nastran plot commands. The central cavity cross section will be positioned
in the XY plane of the basic coordinate system. The slot elements are offset from the
XY plane by the width of the slot in the +Z direction. The radial direction
corresponds to X and the axial direction corresponds to the Y direction. Pressures
will be plotted in the Z direction for both the slot points and the central cavity points.
The PLOTEL elements are used for plotting the acoustic cavity shape. The plot
request entry SET n INCLUDE PLOTEL must be used where n is a set number.
CHAPTER 23 687
Acoustic Cavity Example Problem

Table 23-1 contains a listing of the data entries used as a simple example of acoustic
cavity analysis. The problem to be solved is illustrated in Figure 23-2. The model
was subdivided into only ten finite elements in order to limit the number of data
entries. For reasonable engineering accuracy, this model should be subdivided into
at least four times that number of elements.
1” 8” 4”
Rigid
Boundaries
2” 4” Slot
Nozzle
Central Cavity
4”
Symmetry P=0
Z
A
Symmetry 14” 2”
Section A-A
Parameters: Density: RHOD = 1.2 × 10 – 7 lb-sec 2 /in 4
Bulk Modulus: BD = ra 2 = gRT = 21.0 lb/in 2
Harmonic: N = 1
Number of Slots: MD = 4
Finite Element Model:

r
GRIDS points 6 9
3
GRIDF points
1 2
1 Element IDs 5 8 3
2 12
1 11
4 7 7
8
4 5 6 13
10 9
z
Figure 23-2 Description of Acoustic CavIty Example Problem
Each data entry in Table 23-1 is given a number on the left side. The format for each
type of Bulk Data entry is given in parentheses above the group of that type. The
following
www.cadfamily.com is a brief description of each part of the input data:
Lines
1-4 Each statement in the Executive Control Section has the format of a
request word and a selection separated by blanks or a comma. The CEND
statement is last, but the statements above may appear in any order.
Solution Sequence 103, normal modes analysis, was chosen. A limit of 2
minutes CPU time was set (TIME2).
5-7 The TITLE, SUBTITLE, and LABEL commands may contain any list of
letters and numbers following the (=) sign. This list will appear on the
first three lines of each output page.
8 The method of eigenvalue extraction is selected with the METHOD
command. The number 11 refers to the identification number of an
EIGRL Bulk Data entry which appears below as line 32.
9-11 A simple output request is illustrated with these commands. PRES = ALL
will result in the printout of all pressures at the GRIDF and GRIDS points.
STRESS = ALL will result in the printout of all velocities in the elements.
This printout will occur for all extracted eigenvectors. Selected points or
elements can be printed via the SET command.
12 The BEGIN BULK command denotes the beginning of the Bulk Data
Section. The Bulk Data entries may occur in any order. Putting these
entries in alphabetic sort will save sorting time in large problems,
however.
13 In this problem, all the parameters except slot width WD are constant
throughout the volume. The values on AXSLOT will be used whenever a
corresponding field in the following entries is blank.
14-20 The location of points on the slot are defined with these entries. Entries
14, 16, 18, and 20 serve a dual purpose by defining a GRIDS point
identification number in field 2 and a GRlDF point identification number
in field 6. The two types of points thereby are forced to have the same
locations at the interface.
21-22 The location of points within the axisymmetric fluid cavity are described
by the GRIDF entry. No points are allowed to have a zero or negative
radius.
23-31 These entries describe the elements shown in Figure 23-2. Each element is
given a unique identification number and a list of the connected GRIDS or
GRlDF points. Since the parameters p and B are constants, these fields are
left blank, so the values on the AXSLOT entry will be used.
CHAPTER 23 689
32 The ElGRL entry is used to define parameters for eigenvalue extraction

(resonant frequencies). More than one of these entries may appear. The
method to be used is selected with the METHOD command (command 8).
With this particular command, the Lanczos method was selected with
three (ND = 3) output mode shapes specified.
33 The SLBDY entry defines the boundary between the slot and the central
cavity. Both the density (RHO and the number of radial slots (M) are
-7
blank, so the AXSLOT defaults are used; i.e., RHOD = 1.2 x 10 and M = 4.
Only four GRIDS points are on the boundary, so a continuation entry is
not necessary. Field 8 is left blank to signify the last entry.
34 The ENDDATA entry is required to denote the end of the Bulk Data
Section. Any following entries will be ignored.
Table 23-1 Acoustic Cavity Example Problem Data
Line
No.
1 $DEMO PROBLEM D03D92 WITH SOL 103 AND
2 LANCZOS
3 SOL 103 Executive Control
4 TIME 2 statements
CEND
5 TITLE = ACOUSTIC CAVITY EXAMPLE PROBLEM

6 SUBTITLE = FIRST HARMONIC
7 LABEL = NORMAL MODES (LANCZOS)
8 METHOD = 11
9 OUTPUT Case Control commands
10 PRES = ALL
11 STRESS = ALL
12 BEGIN BULK
CHAPTER 23 691
Table 23-1 Acoustic Cavity Example Problem Data
Line
No.
(AXSLOT, (RHOD, BD, N, WD, MD)
13 AXSLOT,1.2-7,21.0,1,,4
(GRIDS,ID,R,Z,W,ID
14 GRIDS,2,4.0,0.0,0.23 01,1
15 GRIDS,3,8.0,.0,1.0
16 GRIDS,5,4.0,4.0,2.0,4
17 GRIDS,6,8.0,4.0,1.0
18 GRIDS,8,4.0,8.0,2.0,7
19 GRIDS,9,8.0,8.0,1.0
20 GRIDS,12,4.0,1.2+1,2.0,11
(GRIDF,ID,R,Z )
21 GRIDF,10,2.0,12.0
22 GRIDF,13,2.0,1.4E1
(CSLOT4,1D,P1,P2,P3,P4,RHO,,M )
23 CSLOT4,1 ,2,3,6,5
24 CSLOT4,2,5,6,9,8
(CSLOT3,1D,P1,P2,P3,,RHO,, )
Bulk Data entries
25 CSLOT3,3,8,9,12
(CAXIF2,1D,P1,P2,,,RHO )
26 CAXIF2,4,1 ,4
27 CAXIF2,5,4,7
28 CAXIF2,6,7,10
29 CAXIF2,9,10,13
(CAXIF3,ID,P1,P2,P3,,RHO,B )
30 CAXIF3,7,7,10,11
31 CAXIF3,8,10,11,13
(EIGRL,SID,V1,V2,ND )
32 EIGRL,11,,,3
(SLBDY,RHO,M,ID1,ID2,ID4,etc. )
33 SLBDY,,,12,8,5,2
34 ENDDATA
24
CHAPTER
Cyclic Symmetry
■ Understanding Cyclic Symmetry
■ Understanding Types of Symmetry
■ Cyclic Symmetry Theory
■ Using Cyclic Symmetry
■ Performing Axisymmetric Analysis
24.1 Understanding Cyclic Symmetry

Many structures, including pressure vessels, rotating machines and antennas for
space communications, are made up of virtually identical segments that are
symmetrically arranged with respect to an axis.
When you’re performing analyses on symmetrical structures, you can exploit that
symmetry to reduce greatly reduce the time needed to create and solve an FE model.
One advantage is that you will be required to provide an analytical model for only
the “fundamental regional” of a structure, and this fundamental region is generally
substantially smaller than the analytical model required by more traditional analysis
techniques. Analyzing only the fundamental region lets you:
• reduce the input data describing the finite element model to just that of the
fundamental region
• restrict the computation of g-size stiffness and mass matrices to just the
fundamental region
• exploit the orthogonality of symmetrical component sets, such as the
symmetrical and antisymmetrical sets in reflective symmetry, to reduce the
cost of matrix operations
For example, if a body consists of N identical segments in the sense that the entire
body can be generated through reflections and/or rotations of one of these N
segments (the fundamental region) , you only have to define a finite element model
for the fundamental region and provide a list of the boundary grid points for this
fundamental region. With NX Nastran’s cyclic symmetry option (the CYSYM bulk
data entry), the software performs all these tasks automatically for each of the three
supported subclasses of cyclic symmetry: rotational symmetry (ROT), dihedral
symmetry (DIH), and axisymmetry (AXI).
Supported Solution Sequences

In NX Nastran, you can use cyclic symmetry techniques in the following solution
sequences:
SOLution 114 Superelement Static Analysis

SOLution 115 Superelement Normal Modes Analysis
SOLution 116 Cyclic Buckling
SOLution 118 Superelement Direct Frequency Response
CHAPTER 24 695
Cyclic Symmetry
Comparing Cyclic Symmetry and Traditional Reflective

Symmetry Techniques
The following table compares the major steps in the analysis procedures required
with traditional reflective techniques and with cyclic symmetry techniques and
highlights some of the many advantages of cyclic symmetry.
Reflective Symmetry Cyclic Symmetry

Analytical Reduces the size of the analytical Reduces the size of the model to a
Model model by one-half for each plane “fundamental region,” which is generally
of reflective symmetry. substantially smaller than the entire model
Boundary In the general case of p planes of You supply a list of the grid points on the
Conditions reflective symmetry, you must boundary of the fundamental region.
explicitly define 2 p sets of Boundary conditions are automatically
boundary conditions with SPCi imposed by the program.
Bulk Data entries.
Applied You must decompose the various Applied loads are defined to act at arbitrary
Loads loading conditions into symmetric locations throughout the entire structure
and antisymmetric components. respective of whether any or all of these
locations are in the explicitly defined
fundamental region.
Computer For each set of defined boundary A static condensation is performed once to
Solution conditions, a lengthy matrix reduce the size of the analysis set. In static
decomposition and equation analysis, all but the boundary degrees-of-
solution are required. Static freedom should be omitted. In real eigenvalue
condensations aren’t generally analysis, buckling analysis, and dynamic
economically feasible in static analysis, the boundary degrees-of-freedom
analysis. and selected internal points are kept in the
analysis set. The static condensation involves
a lengthy decomposition which should be less
lengthy than any one of the decompositions
required by traditional techniques because it
only involves the fundamental region. Several
additional decompositions of relatively trivial
magnitude are required after the static
condensation.
Data You must manually combine the NX Nastran automatically provides the
Recovery results for each boundary response for the entire structure without any
condition with SUBCOM or special user action. SUBCOM and SYMCOM
SYMCOM commands to obtain the entries aren’t necessary and shouldn’t be used.
complete response of the structure.
24.2 Understanding Types of Symmetry

There are a number of different types of geometric symmetry, as shown in
Figure 24-1.
Cyclic symmetry is a subclass of the general class of geometric symmetry. Cyclic

symmetry is defined by the rotation of a fundamental region about an axis. Within
the category of cyclic symmetry, there are a number of different subtypes. You can
use NX Nastran to perform cyclic symmetry analyses on the following types of
geometric symmetry:
• rotational symmetry
• dihedral symmetry (a combination of reflective plus rotational symmetry)
• axisymmetry (with and without skew)
(a) Reflective Symmetry
Fundamental
Region
One Plane, Two Planes

or Three Planes
(b) Rotational Symmetry
Two Segments
Three Segments
(c) Dihedral Symmetry
2R 1L
n=2 n=1 Three Segments
Six Half-Segments
2L 1R
3R 3L
n=3
CHAPTER 24 697
Cyclic Symmetry
Figure 24-1 Types of Geometrical Symmetry
(d) Translational Symmetry
One-D
Two-D
Three-D
(e) Axisymmetry Fundamental

Region
Without Skew With Skew
Figure 16-1 Types of Geometrical Symmetry (continued)
Most of the types of symmetry shown in Figure 24-1 are subclasses of cyclic
symmetry. There are inter-relationships between these different subclasses. For
example, translational symmetry in one direction, see Figure 24-1(d), can be
considered to be a limiting case of rotational symmetry in which the axis of
symmetry is infinitely far away. Likewise, axisymmetry without skew is a limiting
case of dihedral symmetry as the number of segments is increased without limit,
and axisymmetry with skew is a limiting case of rotational symmetry.
The relationship of the different classes of symmetry to the general class called cyclic
symmetry is summarized in Table 24-1. With few exceptions, all important types
of symmetry are subclasses of cyclic symmetry, and they can, therefore, be analyzed
with the cyclic symmetry capability in NX Nastran.
Table 24-1 Types of Symmetry
Subclass of
Type Cyclic
Symmetry
Reflective Symmetry
One Plane Yes
Two Planes Yes
Three Planes No
Rotational Symmetry (Repetition about an axis) Yes
Dihedral Symmetry (Reflective symmetry plus rotational Yes
symmetry)
Translational Symmetry
One-D Yes
(special case)
Two-D No
Three-D No
Axisymmetry
With Skew Yes
Without Skew Yes
Note that in more general cases, a model may be susceptible to both dihedral and
rotational symmetry. For example, consider the three regular N-gons: the equilateral
triangle, the square, and the pentagon.
D
D c C d c
C
E C
0
c b d b
0 0 e b
A a B A a B A a B
N=3 N=4 N=5
CHAPTER 24 699
Cyclic Symmetry
In these figures, the internal solid lines in conjunction with the edges define
fundamental regions, e.g., A0B, that may be permuted through rotational symmetry
operations. The axis of rotation is normal to the plane of the page and passes
through the point 0.
In addition, the dotted internal lines such as a0 may be used to, for example, reflect
A0a into a0B. Also, a0B may be reflected through B0 to obtain B0b, and so on around
each N-gon until the identity of A0a is reached. Clearly, the fundamental region
A0a for the dihedral symmetry operations is smaller than the fundamental region
A0B for rotational symmetry operations. Importantly, both of these fundamental
regions are substantially smaller than the fundamental region required by
traditional reflective symmetry techniques.
24.3 Cyclic Symmetry Theory

NX Nastran’s cyclic symmetry capabilities rely on the development of
transformation relations that let you manipulate the mathematical representation of
the fundamental region in accordance with the appropriate symmetry operations.
The following operations are performed:
1. You create an NX Nastran input file for the fundamental region.

2. You define the mathematical representation of the model in terms of
physical quantities, such as [ K ] { u } = { P }
3. The software transforms the mathematical representation of the model to
cyclic quantities, such as [ K ] { u } = { P }
4. The software solves the problem in terms of cyclic quantities.
5. The results are transformed from the cyclic quantities to physical
quantities.
Steps 3 and 5 both depend on the existence of the necessary transformations. Step
4 is also of interest in that it must be demonstrated that problem solution procedures
in terms of cyclic quantities are valid and require less computer time than more
traditional solution procedures. These matters are addressed in the following
sections.
Symmetrical Components
Consider an arrangement of N identical objects symmetrically disposed about a
normal. The angle between these objects is
2π
a = ------ Eq. 24-1
N
and, in addition, let U 1 ( r,t ) be any variable displacement, force, stress, temperature,
etc. in object 1, the fundamental region. The variables r and t denote distance and
time respectively. The corresponding quantity in the n-th object is U n ( r,t ) and as a
single valued solution is required
Un = Un + N n = 1,2,...,N Eq. 24-2
The various U n 's are, in general, unsymmetrical quantities that are not related in a
simple manner.
Fortescue’s symmetrical component theory states that a general vector field on N-

vectors with arbitrary phase relationships may be resolved into N-sets of N-
symmetrically phased vectors of equal magnitude. For N = 3, the vectors may be
illustrated by the following diagram.
x 10
U2 x 20
x 30
U1
x 31 x 11
U3
x 21
x 32
x 12
x 22
Now, in accordance with the theory of symmetrical components, the following vector
sums may be stated.
U 1 = x 10 + x 11 + x 12
U 2 = x 20 + x 21 + x 22 Eq. 24-3
U 3 = x 30 + x 31 + x 32
CHAPTER 24 701
Cyclic Symmetry
Here, the U n 's represent physical quantities and the x nk terms represent symmetrical
components. The superscripts k denote the order of the symmetric components.
The subscripts n are taken to be the number of a particular vector in the vector field.
In the general context of cyclic symmetry, the subscripts n denote the number
applied to each of the N identical objects under consideration. The object n = 1 is
arbitrarily taken as the fundamental region.
Understanding Rotational Symmetry

In rotational symmetry, Figure 24-1(b), the structure is comprised of identical
segments that are symmetrically arranged about an axis. Common examples of
structures with rotational symmetry are skew slabs and many types of rotating
machinery including gears, electric motors, turbines, and helicopter rotors.
With rotational symmetry, if the structure has only two segments, the second
segment is created by rotating the first segment through 180 about an axis A. In
general, the n-th segment is created by rotating the fundamental region (i.e., the first
segment) through 360(n-1)/N degrees about axis A where N is the total number of
segments.
Mathematically, a structure has rotational symmetry if each of its relevant physical

properties Q repeats at even angular intervals around an axis A. Thus, referring to
the structure shown in Figure 24-2,
Q ( θ + θn ) = Q ( θ ) Eq. 24-4
where:
2π ( n – 1 )
θ n = -------------------------- and n = 1, 2, 3, ..., N.
N
Segment 2
e2
Segment 1
Sid
e1
(Fundamental
Sid
2
Region)
e
Sid
Side 1
T2
T1 Segment N
Conformable Interface
Figure 24-2 Conforming Boundaries for the ROT Option
You can apply rotational symmetry by using the following procedure:
1. Model one segment. Segment boundaries may be curved surfaces.

2. Ensure each segment has its own coordinate system that rotates with the
segment. The local displacement coordinate system must conform at the
joining points.
3. Specify a paired list of points on side 1 and side 2 that are to be joined.
Transforming Physical Components to Cyclic Components (ROT)

These transformations for the cyclic symmetric components in terms of the general
field vectors may be written in summation form for the general case of N vectors.
N
1
x 1k = ---- ∑ U n e i ( n – 1 )ka Eq. 24-5
N
n = 1
where a is given by Eq. 24-1.
CHAPTER 24 703
Cyclic Symmetry
Transforming Cyclic Components Back to Physical Components

(ROT)
After the solution phase of the problem, it will normally be necessary to transform
the results from cyclic components back into physical components.
Eq. 24-3 may be written in the following form.
N–1
Un = ∑ x 1k e – 1 ( n – 1 )ka
k = 0
= x 10 + [ x 11 e – 1 ( n – 1 ) a + x 1( N – 1 ) e i ( n – 1 )a ]
+ [ x 12 e – 1 ( n – 1 )a + x 1( N – 2 ) e i ( n – 1 )2a ] Eq. 24-6
+ x 13 e – 1 ( n – 1 )a + x 1( N – 3 ) e i ( n – 1 )3a
+ ,...,
(n – 1)
+ x 1N ⁄ 2 ( – 1 )
Here,
e – 1 ( n – 1 ) ( N – k )a = e – 1 ( n – 1 )ka e – 1 ( n – 1 )Na = e – 1 ( n – 1 )ka
and
n–1
e –1 ( n – 1 )( N ⁄ 2 ) a = ( –1 )
If the exponentials in the bracketed terms in the above summation are replaced by
the relations e ±ix = cos x ± sin x , the typical bracketed term takes the following form:
[ ( x 1k + x 1( N – k ) ) cos ( n – 1 )ka – i ( x ik – x 1( N – k ) ) sin ( n – 1 )ka ]
The purpose for the somewhat arbitrary introduction of Eq. 24-6 is now recognized,
and the typical bracketed term assumes the following form:
[ U kc cos ( n – 1 )ka + U ks sin ( n – 1 )ka ]
The final form of the transformation from cyclic components to physical

components is therefore,
K
L
n–1
Un = x 10 + ∑ [ U kc cos ( n – 1 )ka + U ks sin ( n – 1 )ka ] + ( – 1 ) X 1N ⁄ 2
Eq. 24-7
k = 1
n = 1, 2, 3, …, N
In Eq. 24-7, U n represents physical quantities such as displacement force, stress,

temperature in the n-th segment in terms of the cyclic components in the
fundamental region. The following points should be noted:
1. K L = (-----------------
N – 1)
and x 1N ⁄ 2 is nonexistent when N is odd.
2
2. K L = (-----------------
N – 2)
when N is even.
2
3. The notation for cyclic components of order k = 0 and k = N ⁄ 2 is

distinct from the notation for cyclic components of order 1 ≤ k ≤ k L because
of the somewhat peculiar nature of these two components.
Boundary Conditions and Problem Solution (ROT)

The transformation equations of “Cyclic Symmetry Theory” on page 699 are
assumed to allow one to transform a problem that is mathematically defined in terms
of physical components into a set of mathematical statements written in terms of
cyclic components. Specifically, the mathematical description of the n replications of
the fundamental region may be written in the form
{ Fn } = [ Yn ] { Un } Eq. 24-8
This expression represents the mathematical description of the system in terms of

physical coordinates.
The application of previously defined transform equations may now be introduced

to obtain the desired mathematical description of the system in terms of cyclic
components. First, however, consider the matrix representation of the transform
Eq. 24-5 and Eq. 24-7, respectively. Eq. 24-5 assumes the form
CHAPTER 24 705
Cyclic Symmetry
1 1 1 1
---- ---- ---- … ----
N N N N
0  U1 
 x1  2 2 2 2  
  ---- ---- cos a ---- cos 2a … ---- cos ( N – 1 )a  U2 
N N N N
 U1c   
  2 2 2 U 
 U 1s  0 ---- sin a ---- sin 2a … ---- sin ( N – 1 )a  3
  N N N  .. 
 U2c  = 2 2 2 2  . 
 ..  ---- ---- cos 2a ---- cos 4a … ---- cos ( N – 1 )2a  . 
 .  N N N N  .. 
 K s …  
U L   ... 
 N ⁄ 2 2 2 2
 x1  0 ---- sin K L a ---- sin 2K L a … ---- sin ( N – 1 )K L a  
N N N UN 
 
1 1 1 1 N–1
---- – ---- ---- … ---- ( – 1 )
N N N N
{ U k } = [ T xU ] { U n }
Eq. 24-9
Eq. 24-7 assumes the form
 U1  1 1 0 1 … 0 1  x 01 
   
 U2  1 cos a sin a cos 2a … sin K L a –1  U 1c 
   
 U3  1 cos 2a sin 2a cos 4a … sin 2K L a 1  
   1s 
  =  U 
   
   
   K S
   U L 
U  (N – 1)  xN ⁄ 2 
1 cos ( N – 1 )a sin ( N – 1 )a cos ( N – 1 )2a … sin ( N – 1 )K L a ( – 1 )
 N  1 
{ U n } = [ T Ux ] { U k }
Eq. 24-10
It is easily demonstrated that
[ T xU ] [ T Ux ] = [ I ]
Return now to Eq. 24-8, the mathematical representation of the system in terms of
physical coordinates, and represent the energy of the system in the form
U = [ Un ] T { Fn } Eq. 24-11
From Eq. 24-8, Eq. 24-10, and Eq. 24-11
 { U }T { Y } { U }
 n n n
U =  Eq. 24-12
 { U k }T [ T ]T [ Y ] [ T ] { U k }
 Ux n Ux
The system energy in terms of cyclic components may be written as
T
U = { U k } [ Yk ] { U k } Eq. 24-13
Since the energy must be invariant to the transformations, Eq. 24-13 and Eq. 24-14
may be equated
T
[ Y k ] = [ T Ux ] [ Y n ] [ T Ux ] Eq. 24-14
That is, the equations of the system in terms of physical components are transformed
into a set of equations in terms of cyclic components. As there is no coupling between
the N identical objects of the system, the matrix Yn may be written so that it consists
of N identical blocks along the diagonal. These blocks represent, in general, the
equations of motion of each of the N identical objects of the system.
As an example, assume N = 4 ( K L = 1 ) , and expand Eq. 24-14. Note that
Y1 = Y2 = Y3 = Y4 .
1 1 1 1 1 1 0 1
[ Y k ] = Y 1 1 cos a cos 2a cos 3a I
1 cos a sin a –1
0 sin a sin 2a sin 3a 1 cos 2a sin 2a 1
1 –1 1 –1 1 cos 3a sin 3a –1
4 0 0 0
= Y1 0 2 0 0
0 0 2 0
0 0 0 4
In general, for N objects the above equation may be written
Y k = Y 1 N for k = 0, N ⁄ 2
N
Y k = ---- Y 1 for 1 ≤ k ≤ K L
2
These equations imply that the equations of motion in terms of cyclic components are
either N or N ⁄ 2 times the corresponding equations written in terms of physical
components. This argument provides an explanation for the presence of the
coefficients that appear in the terms of Eq. 24-9.
To this point, it has been found that except for boundary conditions, the equations for
motion in terms of cyclic components ( k = 0, 1, ..., N ⁄ 2 ) may be developed from
the equations of motion of the fundamental region ( n = 1 ). In essence, an
arbitrarily large physical system may be investigated in terms of a fundamental
CHAPTER 24 707
Cyclic Symmetry
region which has been transformed into a number of models defined in terms of
cyclic components. The boundary conditions between cyclic models must now be
defined.
The only connection between the n identical objects in question exists along the
boundaries between adjacent segments. Specifically, consider three of the N objects
as indicated in the following figure.
(j + 1)
u 1,j + 1 Side 1
u 2,j Side 2
u 1,j
(j)
u 2,j – 1 Side 1
Side 2
(j – 1)
These objects are displaced from one another to more clearly define the boundary
conditions between the objects. Note the designation of the two sides of each of the
elements as side 1 and side 2 as one proceeds in the counterclockwise direction.
From the requirement of displacement compatibility, it is necessary that the typical
boundary degrees-of-freedom, u , satisfy the equations
u 2,j – 1 = u 1,j
Eq. 24-15
u 2,j = u 1,j + 1
These physical degrees-of-freedom may be transformed into cyclic components of

order K through Eq. 24-5.
1
( x k ) side 1 = ---- u 1,j e i ( j – 1 ) ka
N
Eq. 24-16
1
( x k ) side 2 = ---- u 1,j + 1 e ijka
N
A comparison of Eq. 24-16 provides the result
( x k ) side 2 = ( x k ) side 1 e ika Eq. 24-17

for 1 ≤ k ≤ k L . For k = 0 ,
( x k ) side 2 = ( x k ) side 1 Eq. 24-18
and for k = N
---- ,
2
( x k ) side 2 = – ( x k ) side 1 Eq. 24-19
The cyclic component of Eq. 24-19 may be written in terms of the variables U kc and
U ks in accordance with the definitions.
( U kc ) side 1 = [ ( x k ) side 1 + ( x N – k ) side 1 ]

( U ks ) side 1 = – i [ ( x k ) side 1 – ( x N – k ) side 1 ]
( U kc ) side 2 = [ ( x k ) side 2 + ( x N – k ) side 2 ]
( U ks ) side 2 = – i [ ( x k ) side 2 – ( x N – k ) side 2 ]
Eq. 24-17 may now be introduced to yield the equations
( U kc ) side 2 = [ ( x k ) side 1 e ika + ( x N – k ) side 1 e – ika ]

( U kc ) side 2 = – i [ ( x k ) side 1 e ika – ( x N – k ) side 1 e – ika ]
Upon the replacement of the exponentials with Euler's equation, the appropriate
boundary conditions may be represented as follow:
kc kc
 ( U ) side 2  cos ka – sin ka  ( U ) side 1 
 ks  =   Eq. 24-20
 ( U ) side 2  sin ka cos ka  ( U ks ) side 1 
These boundary conditions are of the form of constraint equations in which the
variables on side 2 are dependent and the variables on side 1 are independent.
Indeed, these equations may be viewed as MPC equations that couple kc and ks
terms of side 1 to the corresponding terms of side 2.
The equations of motion in terms of cyclic components may be written in the

following form.
CHAPTER 24 709
Cyclic Symmetry
N  x0   F0 
----    
2  U 1c   F 1c 
N    
 U 1s   F 1s 
N    
N  U 2c   F 2c 
   
[ Y1 ] .  .  =  .  Eq. 24-21
   
.  .   . 
.  .   . 
   
N  UK Ls   FKLs 
   
N  N⁄2  ⁄ 
---- x N 2
2   F 
From the previous discussion on boundary conditions, the above equation may be
rewritten in terms of the following disjoint problems.
k = 0 [ Y0 ] { X 0 } = F 0
Y K 0  U kc   F kc 
1 ≤ k ≤ KL  ks  =  ks  Eq. 24-22
0 Yk  U  F 
N [ YN ⁄ 2 ] { x N ⁄ 2 } = { F N ⁄ 2 }
k = ----
2
The boundary Eq. 24-18, Eq. 24-19, and Eq. 24-20 must be introduced to remove the
dependent variables that are defined to be on side 2 of the region. If you also use
the OMIT feature, the only independent degrees-of-freedom in the above disjoint
problems are the degrees-of-freedom on side 1 that have not previously been
declared dependent through MPC, SPCi, and/or OMITi Bulk Data entries. Thus,
the matrices in the above disjoint problems, although dense, are of small size
relative to, say, the matrices required to perform the same analysis by more
traditional methods.
Understanding the Harmonic Index

The harmonic index k plays an important role in the analysis of structures with
cyclic symmetry. The harmonic index k has the values k = 1, 2, ..., N/2. If the number
of segments is large, the higher harmonics may contribute very little to the total
solution. In NX Nastran you can select the harmonics (values of k) to use in the
analysis.
If the only nonzero symmetrical load components are for a particular value of k, then
the only nonzero symmetrical response components are for the same value of k,
provided only that the structure is linear and satisfies Eq. 24-4. This fact leads
directly to an efficient method of analysis in which each symmetric load component

is computed from Eq. 24-24 and applied to a finite element model that is very much
smaller than the complete structure.
The symmetrical components are evaluated from the physical components by means
of the following summations that replace the Fourier integrals used in axisymmetric
analysis. The argument (p) is dropped for convenience.
N
0 1 n
u = ----
N ∑ u
n = 1
N
ks 2 n
u = ----
N ∑ u sin ( kθ n ) k = 1, 2, … , k L
n = 1
Eq. 24-23
N
kc 2 n
u = ----
N ∑ u cos ( kθ n ) k = 1, 2, …, k L
n = 1
N
N⁄2 1 n–1 n
u = ----
N ∑ ( –1 ) u
n = 1
Note that stiffness damping, and mass matrices become uncoupled for u k .
Eq. 24-23 also applies to the expansion of applied loads into symmetrical component
sets. Thus
N
0 1 n
P = ----
N ∑ P
n = 1
N
ks 2 n
P = ----
N ∑ P sin ( kθ n ) k = 1, 2, …, k L
n = 1
Eq. 24-24
N
kc 2 n
P = ----
N ∑ P cos ( kθ n ) k = 1, 2, …, k L
n = 1
N
N⁄2 1 n–1 n
P = ----
N ∑ ( –1 ) P
n = 1
CHAPTER 24 711
Cyclic Symmetry
The implied coordinate systems in the physical segments remain fixed in the
segments and are rotated with them. To satisfy continuity between segments, the
grid point on side 2 of segment 1 must coincide with the grid point on side 1 of
segment 2 and their coordinate systems must be aligned (see Figure 24-2). As a
result, you must align the grid points on sides 1 and 2 of the fundamental region so
that their positions and displacement coordinate systems differ only by a rotation
through the angle a = 2π ⁄ N about the axis of symmetry. This is most easily
accomplished by using a cylindrical coordinate system for the grid points on sides
1 and 2. If this is done, the boundary conditions are satisfied by imposing the
following constraints on the symmetrical components of displacement at
corresponding points on sides 1 and 2 (indicated by subscripts).
0 0
u1 = u2
kc ks kc
u 1 cos ( ka ) + u 1 sin ( ka ) = u 2
kc ks kc
– u 1 sin ( ka ) + u 1 cos ( ka ) = u 2 Eq. 24-25
k = 1, 2, …, k L
N⁄2 N⁄2
–u1 = u2
Rotational Symmetry Example

As an example of the application, consider the simple problem illustrated in
Figure 24-3. The skew slab is divided into two identical segments. The
displacement coordinate system for the second segment is rotated 180° from the
first. A cylindrical coordinate system is used (at least along sides 1 and 2) to ensure
the compatibility of displacement components at the points in the boundaries
between segments.
1
P
θ R
Side 2
Side 1
n = 2 n = 1
Side 1
Side 2
R θ
2
P
0 1 1 2
P = --- ( P + P )
2
k = 0 Model
Side 2
Boundary Condition:
k = 0
Side 1
0 0
u1 = u2
1 1 1 2
P = --- ( P – P )
2
k = 1 Model
Side 2
Boundary Condition:
k = 1
1 1
Side 1
u1 = –u2
Combination of Solutions:
1 0 1 2 0 1
u = u +u u = u –u
Figure 24-3 Example of Rotational Symmetry
Since N = 2, there are two symmetrical component sets u 0 and u N ⁄ 2 = u 1 . The

symmetrical load components, as computed from Eq. 24-24, are
CHAPTER 24 713
Cyclic Symmetry
0 1 1 2
P = --- ( P + P )
2
Eq. 24-26
1 1 1 2
P = --- ( P – P )
2
0 1
P and P are applied to separate copies of the finite element model for the
fundamental region labeled the K = 0 model and the K = 1 model. The boundary
condition linking all six degrees of freedom at corresponding pairs of grid points on
sides 1 and 2, as computed from Eq. 24-25, are
0 0
u1 = u2
Eq. 24-27
1 1
u1 = –u2
These boundary conditions can be supplied in Solution 101 with multipoint

constraints; however, they are provided automatically in NX Nastran’s cyclic
symmetry capability.
The K = 0 and K = 1 models are solved in the usual way to obtain the symmetrical
response components u 0 and u 1 . As a final step, the physical responses are
evaluated as follows from Eq. 24-23
1 0 1
u = u +u
Eq. 24-28
2 0 1
u = u –u
Note: You could use the static condensation procedure in this problem. With that
procedure, all the interior degrees of freedom are placed in the u o set and
eliminated, resulting in a greatly reduced stiffness matrix for the remaining degrees
of freedom on the boundary (sides 1 and 2). Since this only needs to be done once,
there is a net saving of computer time.
See Also
• “Static Condensation (Guyan Reduction)” on page 248 of the NX Nastran

User’s Guide
Dihedral Symmetry
Dihedral symmetry, Figure 24-1(c), is a special case of rotational symmetry that
combines rotational and reflective symmetry, since each segment has its own plane
of reflective symmetry. Thus, in Figure 24-1(c), half-segment 1L is the mirror image
of half-segment 1R (the fundamental region), and these two halves of segment 1 are
rotated together through 120° to form segment 2 and through 240° to form segment
3. Water towers and reinforced domed roofs are familiar examples of structures with
dihedral symmetry.
Dihedral symmetry (DIH) can be useful because:
• DIH takes advantage of reflective symmetry to reduce the fundamental

region by one-half relative to the size of the fundamental region required in
the ROT option.
• You can use the DIH option for analyses that would normally be handled
with traditional reflective symmetry techniques.
Figure 24-4 and Figure 24-5 show two different examples of dihedral symmetry and
the notation system used for the DIH option in NX Nastran.
1L
e2
1R
Sid
2R
Side 1
3L
2L
3R
3 Segments
6 Half-Segments
Figure 24-4 Dihedral Symmetry
CHAPTER 24 715
Cyclic Symmetry
Segment 2L
Segment 2R
Segment 1L
Side 1
Side 2
Side 2 Segment 1R
Side 1 (Fundamental Region)
Plane of Overall
T2 T2 Reflective Symmetry
T1
Segment NL
Figure 24-5 Dihedral Symmetry
The fundamental region, for which you create a finite element model, is the right
half of the first segment (1R). With dihedral symmetry, all segment boundaries
must be planar.
• The plane that forms the boundary between the right and left halves of
segment 1 is labeled side 2.
• The other boundary of the fundamental region, which must also be planar,
is labeled side 1.
Note that:
• Local displacement systems axes, associated with the inter-segment
boundaries, must be in the plane or normal to the plane.
• In the specific case of a structure with a single overall plane of reflective
symmetry (i.e., the complete structure is represented by segments 1R and
1L) the reflective plane may be defined as side 1 or side 2.
• The two halves of a segment that join at a reflective plane are denoted as
the right, “R”, and left, “L”, halves. The L half segments are mirror images
of the R half segments.
• R halves use a right-handed coordinate system and L halves use a left-

handed coordinate system. Therefore, at each boundary a degree-of-
freedom is associated with its mirror image on the adjacent segment so that
compatibility always involves a left- and a right-hand coordinate system.
The degrees-of-freedom on the boundaries are further categorized into odd
and even components in accordance with the following table.
Component Degrees-of-Freedom
Even Displacement vectors parallel to the boundary and rotation vectors
normal to the boundary.
Odd Displacement vectors normal to the boundary and rotation vectors
parallel to the boundary.
This has important implications for how you input of loads data and
interpret the output. The coordinate systems for other half-segments in
Figure 24-4 are either rotated, or rotated and reflected as indicated in
Table 24-2.
Table 24-2 Coordinate Systems for Half Segments
Type of Rotation About the

Half-Segment Coordinate System Axis of Symmetry
1R Right-handed 0°
1L Left-handed 0°
2R Right-handed 120°
2L Left-handed 120°
3R Right-handed 240°
3L Left-handed 240°
The typical procedure that you follow when using dihedral symmetry is:
1. You model one half-segment (1R).

2. At a boundary between segments or half-segments, the displacement
components must be parallel and perpendicular to the boundary. The
boundaries are planes passing through the axis of symmetry.
3. You provide separate lists of the points on sides 1 and 2 which must be
joined.
CHAPTER 24 717
Cyclic Symmetry
As in rotational symmetry, the response quantities at corresponding points in the

physical segments can be expressed as a truncated Fourier series with the polar
angle of the n-th segment θ n = 2π ( n – 1 ) ⁄ N as the argument where N is the number
of segments. The series for response quantities in the right half-segments is
k
L
nR kc kc* ks ks*
u = ∑ [(u +u ) cos ( kθ n ) + ( u +u ) sin ( kθ n ) ] Eq. 24-29
k = 0
where k L is the largest integer less than or equal to N/2 (depending on whether n is
odd or even) and u kc , u kc* , u ks , and u ks* are symmetrical component sets. The series
for response quantities in the left half-segments is
k
L
nL kc kc* ks ks*
u = ∑ [(u –u ) cos ( k ( θ n + a ) ) - ( u –u ) sin ( k ( θ n + a ) ) ] Eq. 24-30
k = 0
where, in addition to previously defined terms, a = 2π ⁄ N .
The symmetrical component sets are evaluated from the physical components as
follows:
  N
 u kc  δ nR nL
 - ∑ [ u cos ( kθ n ) ± u cos ( k ( θ n + a ) ) ]
 = --- Eq. 24-31
 u kc*  N
  n = 1
  N
 u ks  δ nR nL
 - ∑ [ u sin ( kθ n ) −
 = --- + u sin ( k ( θ n + a ) ) ] Eq. 24-32
 u ks*  N
  n = 1
where, in addition to previously defined terms, δ = 1 ⁄ 2 for k = 0 and N ⁄ 2 ,

and δ = 1 for all other values of k.
The number of symmetrical components sets is equal to the number of

half-segments in the structure (neglecting the identically null sets u ks and uks* for
k = 0, N ⁄ 2 ). As in the case of rotational symmetry, the symmetrical load components
for a particular value of k produce only the symmetrical response components for
the same value of k, or, to put it another way, the analyses for the different values of
k are uncoupled.
Furthermore, the analyses for the starred (*) and unstarred symmetrical
components are uncoupled. For K = 0 and k = N ⁄ 2 , the complete finite
element model for the unstarred terms consists of a single copy of the fundamental
region (1 R). For 0 < k < N/2, the complete finite element model consists of two copies
of the fundamental region with coupling between their boundaries. The boundary
conditions are different for each value of k. The finite element models for the starred
(*) terms are identical to those for the unstarred terms for k > 0. Only the symmetrical
load components, which are computed by Eq. 24-31 and Eq. 24-32, are different.
As an example, consider the case of a structure with two planes of reflective

symmetry as shown in Figure 24-6. Applying dihedral symmetry to this case, you
can see that
• The number of segments N = 2.

• The segment angles a = n.
• θ 1 = 0.
• θ 2 = π.
• The harmonic indices k = 0, 1.
Substitution of these values into Eq. 24-31 and Eq. 24-32 gives
0c 1 1R 1L 2R 2L
u = --- ( u + u + u + u )
4
1c 1 1R 1L 2R 2L
u = --- ( u – u – u + u )
4
0c* 1 1R 1L 2R 2L Eq. 24-33
u = --- ( u – u + u – u )
4
1c* 1 1R 1L 2R 2L
u = --- ( u + u – u – u )
4
0s 0s* 1s 1s*
u = u = u = u = 0
Since all of the signs are positive in the equation for u0c , it follows that u 0c is
symmetrical with respect to both side 1 and side 2 in Figure 24-6. The other nonzero
symmetrical components are antisymmetrical with respect to one or both of the sides
as shown in Table 24-3.
Table 24-3 Symmetrical Components for Reflective Symmetry Example
Symmetry
Symmetrical
Side 1 Side 2
Component
0c
u S S
1c
u S A
CHAPTER 24 719
Cyclic Symmetry
Table 24-3 Symmetrical Components for Reflective Symmetry Example
Symmetry
Symmetrical
Side 1 Side 2
Component
0c*
u A A
1c*
u A S
Regardless of the number of segments, the starred terms are always

antisymmetrical, and the unstarred terms are always symmetrical with respect to
side 1 of the fundamental region. You can take advantage of this to reduce the
amount of calculation when it can be determined beforehand that the applied
loading has a plane of symmetry or antisymmetry that coincides with side 1 of the
fundamental region, or, in other words, if the problem has an overall plane of
symmetry. The amount of computation can be reduced still further if the problem
has a second overall plane of symmetry perpendicular to the first. The second
overall plane of symmetry can exist only if there is an even number of segments.
Summary of Transformations, Boundary Conditions, and Problem

Solution (DIH)
The form of the transformations between cyclic and physical components and the
boundary conditions between the segments of the object in question depend
somewhat on the notation system.
When you select the DIH option of cyclic symmetry, you can take advantage of
overall planes of symmetry when the applied loads are symmetrically and/or
antisymmetrically disposed with respect to these planes. In these cases, you only
need to specify loads for segments in the first half plane when there is one plane of
overall symmetry and in the first quadrant when there are two planes of overall
symmetry.
You specify symmetrical or antisymmetrical boundary conditions on the DSYM

Case Control command. You can only include one DSYM command per subcase. A
single plane of reflective symmetry must be with respect to Side 1, and if a second
plane of reflective symmetry exists, it must be perpendicular to Side 1. In the case
of two planes of symmetry, the model must contain an even number of whole
segments.
See Also
• “DSYM” on page 229 of the NX Nastran Quick Reference Guide
The following sections describe several of the equations of interest for the DIH
option, along with the equation number for the corresponding expression under the
ROT option. In the following equations, the starred (*) terms donate anti-symmetric
motions and unstarred terms denote symmetric motions. The subscripts R and L
refer to the right and left halves respectively.
Transforming Physical Components to Cyclic Components (DIH)

(See Eq. 24-7 for the corresponding equations for the ROT option.)
U kc  δ
 = ---- ∑ [ U n,R cos ( n – 1 )ka ± U n,L cos ( N – n )ka ]
U kc *  N
n
Eq. 24-34
U ks  δ
- ∑ [ U n,R sin ( n – 1 )ka ± U n,L sin ( N – n ) ka ]
 = ---
ks
U * N
n
Here,
 1 ⁄ 2 for k = 0, N ⁄ 2
δ =  Eq. 24-35
 1 for 1 ≤ k ≤ K L
and the positive sign is associated with the unstarred terms.
Transforming Cyclic Components to Back to Physical Components

(DIH)
(See Eq. 24-20 for the corresponding relation for the ROT option.)
U n,R = ∑ ( U kc + U kc * ) cos ( n – 1 )ka + ( U ks + U ks * ) sin ( n – 1 )ka

k
U n,L = ∑ ( U kc + U kc * ) cos ( N – n )ka + ( U ks + U ks * ) sin ( N – n )ka Eq. 24-36

k
N
0 ≤ k ≤ ----
2
Boundary Conditions and Problem Solution (DIH)

(See Eq. 24-7 for the corresponding boundary conditions for the ROT option.)
Side 1, EVEN
U kc * = 0
Eq. 24-37
U ks = 0
www.cadfamily.comSideEMail:[email protected]
1, ODD
CHAPTER 24 721
Cyclic Symmetry
U kc = 0
Eq. 24-38
U ks * = 0
Side 2, EVEN
ka ka
U kc sin ------ + U ks cos ------ = 0
2 2
Eq. 24-39
ka ka
U kc * cos ------ – U ks * sin ------ = 0
2 2
Side 2, ODD
ka ka
U kc cos ------ – U ks sin ------ = 0
2 2
Eq. 24-40
ka ka
U kc * sin ------ + cos ------ = 0
2 2
In static analysis with cyclic symmetry, the equation
K K K
[ K aa ] { u a } = { Pa }
is solved in SSG3 (Static Solution Generator – Part 3) for { u a } K
K T K G ] + [GT K G ]
[ K aa ] = [ G ck aa ck sk aa sk
and
K T ]{P} KC T ]{ P } ks
{ Pa } = [ G ck + [ G sk
are formed in CYCLIC3 (Cyclic Symmetry Transformation – Module 3).

kc ks x
The symmetrical components of the loads { P } and { P } are subsets of { P } which
is obtained in CYCLIC3 by multiplying the load vector formed in SSG1 by the
GFORE matrix formed in CYCLIC1 module
x
{P} = { P l } { GFORE }
where { P l } is formed in SSG2 (Static Solution Generator – Part 2) and the

transformation matrix [GFORE] is formed in CYCLIC1 module.
The cyclic component constraint data (CYCD) used in CYCLIC3 for the formulation
of the transformation matrices [ G ck ] and [ G sk ] are generated in CYCLIC2 (Cyclic
Symmetry – Module 2).
The solution { ua } K is transformed to symmetric components in CYCLIC4 as follows:

kc K
{ ux } = [ G ck ] { u a }
kc K
{ ux } = [ G sk ] { u a }
The symmetric components for each harmonic index are appended to form { u } x in
the CYCLIC4 module. The physical segment displacements are obtained as follows:
x
{ u a } = { u } [ GBACK ]
where the transformation matrix is formed in the CYCLIC1 module.
In the case of vibration analysis with cyclic symmetry the following equation is used
in READ to determine the eigenvalues and eigenvectors,
K
[ K aa – λM aa ] { u } = 0
K
where [ M aa ] is formed in a manner similar to [ K aa ] in CYCLIC3. The symmetrical
components of the eigenvectors are recovered in CYCLIC4 and the physical segment
data is recovered using the GBACK1 matrix created in CYCLIC4 module.
Since all the transformation matrices are extremely sparse, none of the matrix
multiplications indicated above will require large amounts of computer time. The
most significant operation is the triple product associated with the determination of
K
[ K aa ] . In the usual application this matrix is dense and approximately a-size. The
computer time for the triple product can be estimated as ten matrix packing
operations on a full matrix of a-size.
Reflective Symmetry
Reflective symmetry with one or two planes of symmetry is a special case of dihedral
symmetry ( Figure 24-7).
• Reflective symmetry with one plane of symmetry is identical to dihedral
symmetry with one segment.
• Reflective symmetry with two planes of symmetry is identical to dihedral
symmetry with two segments.
CHAPTER 24 723
Cyclic Symmetry
Fundamental
Region
1 Segment
1L 1R
2 Half-Segments
1L 1R
2 Segments
4 Half-Segments
2R 2L
Figure 24-6 Reflective Symmetry with One or Two Planes of Symmetry –

A Special Case of Dihedral Symmetry
Mathematically, a structure has reflective symmetry with respect to a yz plane if

each of its relevant physical properties Q (such as mass density and elastic modulus)
satisfies the equation
Q ( x, y, z ) = Q ( – x, y, z ) Eq. 24-41
or in other words, if the physical properties are identical at the image points on the
two sides of the yz plane. The plane of symmetry divides the structure into two
halves that are arbitrarily called the right half and the left half. The right half is also
called the fundamental region because, in analysis, it is the half for which a finite
element model is prepared, whereas the finite element model for the left half is only
implied.
The physical properties included in Q do not include the loads applied to the
structure or its responses (displacements, internal forces, or stresses). Let the
symbols u r , and u l represent the values of some particular response u at some point
in the right half and at its image point in the left half of the structure. u r and u l don’t
necessarily satisfy the reflective symmetry property Eq. 24-41, but their average u s
clearly does. Thus,
1
u s = --- ( u r + u l ) Eq. 24-42
2
so thatEMail:[email protected]
www.cadfamily.com
1
u s ( x, y, z ) = --- ( u ( x, y, z ) + u ( – x, y, z ) ) Eq. 24-43
2
and
1
u s ( – x, y, z ) = --- ( u ( – x, y, z ) + u ( x, y, z ) ) Eq. 24-44
2
Since the right-hand sides of Eq. 24-43 and Eq. 24-44 are equal, it follows that
u s ( x, y, z ) = u s ( – x, y, z ) Eq. 24-45
The quantity u s is called the symmetrical part of the response. The antisymmetrical
part of the response u a is defined as one-half of the difference of u r , and ul . Thus,
1
u a = --- ( u r – u l ) Eq. 24-46
2
The antisymmetrical part of the response has the property that
u a ( x, y, z ) = – u a ( – x, y, z ) Eq. 24-47
Eq. 24-42 and Eq. 24-46 show how to obtain the symmetrical and antisymmetrical
parts of the response from the values of the response for the left and right halves of
the structure. They may be written more compactly in matrix form
   
 us  1 1 1 ur 
  = --2-   Eq. 24-48
 ua  1 
–1  u l 
  
The vector
 
 us 
 
 ua 
 
is called the vector of symmetrical components, and the vector
 
 ur 
 
 ul 
 
is called the vector of physical components. The physical components may be found
from the symmetrical components by solving for them from Eq. 24-48.
CHAPTER 24 725
Cyclic Symmetry
   
 ur  1 1 1  us 
  = --
-   Eq. 24-49
 ul  2
  1 – 1  a 
 u
Eq. 24-48 and Eq. 24-49 apply to any response quantity, and they also apply to the
applied loads. Thus, any particular applied load P has symmetrical components P s
and P a evaluated as follows:
   
 Ps  1 1 1  Pr 
  = --2-   Eq. 24-50
 Pa  1 
–1  P l 
  
In order to include vectors as well as scalars in u and P, it is necessary to select the

coordinate systems in the right and left halves to be mirror images of each other as
shown in Figure 24-7. The coordinate system in the right half is right-handed, to
conform with standard NX Nastran practice. Note, however, that the coordinate
system in the implied left half is left-handed. This means, for example, that loads
applied indirectly to the left half, by listing the images of their points of application
in the right half, must be expressed in the particular left-hand coordinate system
Left Half Right Half
z, θ z, w
y, θ y, v
z, θ z , w
y, θ y, v
x , θ x, u
x , θ x, u
Left-Hand Right-Hand
Coordinate Coordinate
System System
Left-Hand Rule Right-Hand Rule
Symmetric Boundary Conditions: u = 0; θ y = 0; θ z = 0
Antisymmetric Boundary Conditions: v = 0; w = 0; θ x = 0
Figure 24-7 Reflective Symmetry
If the loads applied to a symmetrical structure are symmetrical, the response is

symmetrical. The responses also include the displacements in the plane of
symmetry, i.e., u, v, w, θ x , θ y , and θz at x = 0. Since the positive directions of u, θy ,
and θz are opposite in the two halves, continuity requires that they be zero at x = 0 in
a symmetrical solution. Thus, u = 0, θy = 0, and θ z = 0 are boundary conditions that
can be applied with SPCs if only the right half is represented by a finite element
model. The boundary conditions for v, w, and θx are natural, i.e., their reaction forces
in the plane of symmetry are zero by symmetry, and they may be left unconstrained.
In like manner, if the loads applied to the structure are antisymmetrical, the response
is antisymmetrical. By similar reasoning it can be shown that the antisymmetrical
boundary conditions for a finite element model of the right half are v = w = θx = 0.
CHAPTER 24 727
Cyclic Symmetry
If the applied loads are neither symmetrical nor antisymmetrical, the response may
be obtained by the following procedure (which is only valid if the response is a
linear function of the applied loads):
1. Compute the symmetrical and antisymmetrical load sets P s and P a by

means of Eq. 24-50.
2. Apply P s to a finite element model of the right-hand side with symmetric
boundary conditions. The response u s is obtained.
3. Apply P a to a finite element model of the right-hand side with
antisymmetric boundary conditions. The response u a is obtained.
4. Find the physical responses u r and u l by means of Eq. 24-49.
Everything that has been said about one plane of symmetry can be extended to two
or three orthogonal planes of symmetry. The steps that were outlined are
performed automatically in NX Nastran’s cyclic symmetry capability, using the
identity between reflective symmetry with one or two planes of symmetry and
dihedral symmetry with one or two segments, respectively. The steps may also be
performed using basic NX Nastran procedures but with considerable
nonautomated work.
See Also
• “Symmetry” on page 437 of the NX Nastran User’s Guide
Axisymmetry
In axisymmetry, Figure 24-1(e), the object is generated by rotating the fundamental
region continuously about an axis, thereby forming a solid of revolution. There are
two subclasses of axisymmetry:
• In the first subclass, the fundamental region is a plane surface. This
subclass is known as axisymmetry without skew.
• In the second subclass, the fundamental region is a skewed surface. This
subclass is known as axisymmetry with skew.
These two subclasses are distinguishable only if the mechanical properties of the
fundamental region are anisotropic. Examples of structures with skewed
axisymmetry may be found in nature (e.g., the trunks of pine trees) and in
spiral-wound reinforced shells.
Mathematically, a structure is axisymmetric (axisymmetric without skew) if each of

its relevant physical properties Q (such as mass density and elastic modulus)
satisfies the equation
Q ( r, θ, z ) = Q ( r, 0, z ) Eq. 24-51
where r, θ , and z are the components of position in a cylindrical coordinate system.

The cross section for θ = 0 is the fundamental region whose properties are mapped
into the entire structure by Eq. 24-51.
The response of the structure and its applied loads do not necessarily satisfy
Eq. 24-51. Let u ( r, θ, z ) be some particular response quantity (displacement,
internal force, or load). If u is a component of a vector u , let the coordinate system
that expresses be the cylindrical coordinate system used in connection with
Eq. 24-51. Then the Fourier series representation of u ( r, θ, z ) may be written as
∞
0 ks kc
u ( r, θ, z ) = u ( r, z ) + ∑ ( u ( r, z ) sin ( kθ ) + u ( r, z ) cos ( kθ ) ) Eq. 24-52
k = 1
or, dropping the explicit dependence on r and z,
∞
0 ks kc
u(θ) = u + ∑ (u sin ( kθ ) + u cos ( kθ ) ) Eq. 24-53
k = 1
The harmonic response coefficients u 0 , u ks , u kc and may be regarded as symmetrical

component sets that replace u ( θ ) in structural analysis. They are evaluated from the
physical component u ( θ ) by the following Fourier integrals:
2π
0 1
u = ------
2π ∫ u ( θ ) dθ
0
2π
ks 1
u = ---
π ∫ u ( θ ) sin ( kθ ) dθ Eq. 24-54
0
2π
kc 1
u = ---
π ∫ u ( θ ) cos ( kθ ) dθ
0
Applied loads P ( θ ) can also be expanded in a Fourier series. The symmetrical

component load sets are, by analogy with Eq. 24-54,
CHAPTER 24 729
Cyclic Symmetry
2π
0 1
P = ------
2π ∫ P ( θ ) dθ
0
2π
ks 1
P = ---
π ∫ P ( θ ) sin ( kθ ) dθ Eq. 24-55
0
2π
kc 1
P = ---
π ∫ P ( θ ) cos ( kθ ) dθ
0
Fourier series are often used in the analysis of axisymmetric structures. The first
step is to find the harmonic load coefficients from Eq. 24-55. The next step is to find
the harmonic response coefficients as functions of the harmonic load coefficients. It
can be shown from a consideration of the properties of trigonometric functions that,
0
in linear analysis, u 0 depends only on P that the harmonic response coefficients for
ks
a particular harmonic u ks and ukc depend only on the harmonic load coefficients P
kc
and P for the same harmonic index k. Computational effort can be further reduced
by noting that higher harmonics where is problem dependent do not contribute
significantly to the solution and need not be computed. Once the harmonic
response coefficients are found, the physical components can be computed by
means of Eq. 24-52.
24.4 Using Cyclic Symmetry

Overview of Performing a Cyclic Symmetry Analysis
In a cyclic symmetry analysis, your total model consists of N identical segments
which are numbered consecutively from 1 to N. In your input file, you use the
CYSYM bulk data entry to specify the type of cyclic symmetry (rotational, dihedral,
or axisymmetric) and the number of segments in the model.
You then generate the model for one segment, using regular elements and standard
modeling techniques. You can use any element in the NX Nastran element library,
except axisymmetric elements.
NX Nastran automatically rotates all other segments and their coordinate systems
to equally spaced positions about the polar axis. The segment boundaries must be
conformable; i.e., when the segments are put together, the grid points and the
displacement coordinate systems of adjacent segments must coincide. This is
easiest to ensure if you use a cylindrical or spherical coordinate system.
The two boundaries are called Sides 1 and 2.

• In the case of rotational symmetry, Side 2 of segment n is connected to Side
1 of segment n+1, as shown in Figure 24-2.
• In the case of dihedral symmetry, Side 1 is on the boundary of the plane of

overall reflective symmetry, and Side 2 is on the midplane or reflective
plane of the segment, as shown in Figure 24-5.
When you create your input file, you specify the grid point numbers on Sides 1 and
2 on the CYJOIN Bulk Data entry. You also list grid points that lie on the axis of
symmetry on the CYAX Bulk Data entry.
Specifying Loading and Constraints

Complete generality in the definition of loading conditions is achieved by allowing
you to independently specify loads, enforced motions, and/or temperatures on each
of the N segments, even though the fundamental region is the only segment explicitly
represented in the analysis. Output requests are handled with the same degree of
generality in that solutions are provided for each of the N segments.
If your loading has the same symmetry as the structure, only a single analysis is
required for a symmetrical subregion of the structure with symmetrical boundary
conditions. If your loading is unsymmetrical, the following additional tasks must be
performed:
1. Separate applied loads into symmetry sets (e.g., symmetrical and

antisymmetrical sets for a structure with one plane of symmetry).
2. Apply separate constraint sets to a subregion of the structure in order to
model the boundary conditions corresponding to each symmetry condition.
3. Compute separate solutions for each symmetry condition.
4. Combine the results for each symmetry condition to obtain the complete
solution for the complete structure.
NX Nastran also includes a way of handling the special case of unsymmetrical loads
on a structure with one or two planes of symmetry. With this method:
• Tasks 1 and 2 (listed above) you perform by hand by hand.

• Task 3 is accomplished by creating subcases for each symmetry condition,
with different SPC sets and loading conditions for each subcase.
• Task 4 is accomplished with SYMCOM and SYMSEQ Case Control
commands.
In a cyclic symmetry analysis, you apply loads to the model with the following bulk
data entries:
• LOADCYH (harmonic load)

• LOADCYN (physical segment load)
• LOADCYT (AXI option) Bulk Data entry
CHAPTER 24 731
Cyclic Symmetry
You only need to define a single subcase is needed for each loading condition. That
subcase must contain a LOAD, TEMP(LOAD), or DEFORM command that selects a
LOADCYH, LOADCYN, or LOADCYT Bulk Data entry (see Figure 24-8). The
LOADCYH and LOADCYN entries in turn reference FORCE, MOMENT, PLOAD,
SLOAD, SPCD, DEFORM, and TEMP entries. The LOADCYH entry may also
reference an RFORCE or GRAV entry in order to specify centrifugal or gravity
loading, respectively. The LOADCYT entry is used only with the AXI option to
specify loads as a function of azimuth angle.
Importantly, you can’t use the SUBCOM/SUBSEQ commands to combine loads in

a cyclic symmetry analysis. To combine loads, use the LOAD Bulk Data entry.
Case
Control LOAD = l
Section
Bulk Segment 1 Segment 2

Data ETC.
Section
LOADCYN LOADCYN
LOAD
FORCE FORCE
MOMENT SPCD
SPCD ETC.
ETC.
ETC.
HARMONIC (0) HARMONIC (1C)
LOADCYH LOADCYH LOADCYH
GRAV LOAD
RFORCE
FORCE
MOMENT
SPCD
ETC.
Figure 24-8 Cyclic Symmetry Flowchart
There are no restrictions on using single-point constraints, multipoint constraints, or

the OMIT feature. The software automatically applies constraints between segments
to the degrees-of-freedom at grid points specified on CYJOIN Bulk Data entries
which aren’t otherwise constrained. You can also use the SPCD Bulk Data entry to
vary the magnitude of enforced displacements for each of the segments.
In static analyses, you can use the OMIT feature to remove all degrees-of-freedom at
internal grid points without losing accuracy. Since this reduction is applied to a
single segment prior to the symmetry transformations, it can greatly reduce the
amount of subsequent calculation. In the case of vibration analysis, you can use the
OMIT feature to reduce the size of the analysis set with the typical approximations.
You can’t use the SUPORTi entries for the free bodies in a cyclic symmetry analysis.
Specifying Harmonics
You specify the harmonics to use in the cyclic symmetry analysis on the
HARMONICS Case Control command. The HARMONICS command must be
present in Case Control and it must appear above the subcase level. In buckling
analysis, the program limits the static preload to the zero harmonic and the
information on the HARMONICS command is used only for the eigenvalue analysis.
Bulk Data Entries Specific to Cyclic Symmetry

Table 24-4. lists bulk data entries that are unique to cyclic symmetry analyses
Table 24-4 Bulk Data Entries Unique to Cyclic Symmetry
Bulk Data
Function
Entry
CYJOIN Defines the grid points on sides 1 and 2 of the fundamental
region.
CYSYM Specifies the number of segments in the entire model and the
type of symmetry. Allowable symmetry types are ROTational,
DIHedral, and AXIsymmetric.
CYAX List of grid points that lie on the axis of symmetry.
CYSUP List of supported components at a single grid point that are
“constrained” in order to prevent rigid body motion for free
bodies.
LOADCYN Defines applied loads in terms of physical components.
LOADCYH Defines applied loads in terms of their harmonic components.
LOADCYT Specifies loads as a function of azimuth angle for the AXI
symmetry type.
CHAPTER 24 733
Cyclic Symmetry
These entries are described below.
Using CYSYM
The CYSYM entry lets you define important parameters for cyclic symmetry
analysis. For example, you use it to select the type (STYPE) of cyclic symmetry to
perform in the analysis, such as dihedral (DIH). The format of the CYSYM entry is:
1 2 3 4 5 6 7 8 9 10
CYSYM NSEG STYPE
Field Contents
NSEG Number of segments that comprise the total structure.
STYPE Symmetry type (“ROT”, “AXI”, “DIH”).
• For the ROT and AXI options, the polar angle θ subtended by the element
model for the fundamental region, must be such that θ = 360°/NSEG
exactly. For the AXI option, the NSEG value should be large enough
(typically 360 segments) to approximate true axisymmetry.
• For the DIH option, θ must equal 180°/NSEG exactly if there are any
points on side 1. In this case, the polar angle θ is smaller because the
fundamental region is the right half of segment 1. If there are no points on
side 1, θ must be less than or equal to 180°/NSEG in the DIH option
because otherwise the segments would overlap.
• To properly apply loads and interpret output data, you must be aware of
the orientation of the coordinate system associated with the individual
segments. For the ROT and AXI options, the coordinate system used to
define the model of the fundamental region is rotated through the
symmetry operation that “generates” each of the other identical segments.
For the DlH option, the coordinate system used to define the model of the
fundamental region (segment 1 R) is rotated through the necessary
symmetry operations that generate the R-half of each of the identical
segments. The coordinate system for the L-half of a segment is the mirror
image of the coordinate system associated with the R-half of the segment
in question; thus, the L-half of any segment has a left-handed coordinate
system.
See Also
• “CYSYM” on page 1095 of the NX Nastran Quick Reference Guide
Using CYJOIN
The boundary points of a segment in cyclic symmetry problems using the CYJOIN
entry are as follows:
1 2 3 4 5 6 7 8 9 10
CYJOIN SIDE C G1 G2 G3 G4 G5 G6
G7 G8 G9 -etc.-
Field Contents
SIDE Side identification.
C Type of coordinate system used on boundaries of dihedral
or axisymmetry problems.
Gi Grid or scalar point identification numbers.
One CYJOIN entry plus all necessary continuation entries are used to specify the
GRID and/or scalar points (SPOINT), on side 1 of the fundamental region. A second
CYJOIN entry plus all necessary continuation entries are used to specify the GRID
and/or scalar points (SPOINT) on side 2 of the fundamental region. As will be
shown, points that lie on the axis of symmetry appear on a CYAX Bulk Data entry but
not on a CYJOIN entry.
• For the ROT option, the two CYJOlN lists must be of the same length and,
since the boundary conditions for the fundamental region involve these two
lists, it is required that the two lists be ordered. That is, the n-th point on
side 1 is related through boundary conditions to the n-th point of side 2.
Side 2 is related to side 1 as indicated in Figure 24-2.
Note that displacement compatibility must be maintained at the
boundaries. This requirement necessitates that the displacement
coordinate systems for the boundary points be either cylindrical or
spherical.
• The AXI option is a subset of the ROT option in which all the grid points in
the fundamental region lie either on side 1 or on side 2. The rules for the
CYJOIN entry are identical for both the ROT and AXI options of cyclic
symmetry.
• For the DlH option, side 1 denotes the boundary between segments, and
side 2 denotes the side that is contained in the plane of reflective symmetry
of the segment. Since the boundary conditions do not relate sides 1 and 2, it
is not necessary that the two lists be of equal length. It is, however,
necessary under the DlH option to provide an entry in the third field for the
www.cadfamily.com parameter C. The appropriate entries are listed in Table 24-5.
CHAPTER 24 735
Cyclic Symmetry
Unlike the ROT option (in which case, one and only one logical CYJOIN entry must
exist for each of side 1 and side 2), any number of CYJOIN entries may be used to
define the points on a side of the fundamental region in the DIH option. Thus, each
point along a boundary may appear on a separate CYJOIN entry if, for example,
each point has a different coordinate system. For the DIH option, the displacement
coordinate systems associated with boundary points must be oriented so that the
components of motion are normal and parallel to the boundary planes.
Table 24-5 Parameter C for the DIH Option
Type of Point C Description

Grid T1 T1 component of translation motion normal to
boundary.
T2, R, or C T2 component of translation motion normal to
boundary.
T3 or S T3 component of translation motion normal to
boundary.
SCALAR T1 If the user-defined translation is normal to the
boundary or if the user-defined rotation vector is
in the plane containing the boundary.
Blank, T2, If the user-defined translation is parallel to the
or T3 boundary or if the user-defined rotation vector is
normal to the boundary.
See Also
• “CYJOIN” on page 1090 of the NX Nastran Quick Reference Guide
Using CYAX
The CYAX Bulk Data entry lets you list points on the axis of symmetry. The format
of the CYAX entry is:
1 2 3 4 5 6 7 8 9 10
CYAX G1 G2 G3 G4 G5 G6 G7 G8
G9 G10 -etc.-
Field Contents
Gi A list of grid points on the axis of symmetry.
• The coordinate system that locates points on the axis of symmetry and the
displacement coordinate system for these points must be rectangular with
the z-component aligned with the axis of symmetry. In addition, if the DlH
option is selected, the y-axis must be perpendicular to side 1 of the
fundamental region.
• Grid points on the axis of symmetry may not appear on a CYJOlN Bulk Data
entry.
• Grid points on the axis of symmetry may not be referenced as dependent
points by MPCs and/or rigid elements.
• Grid points on the axis of symmetry may have SPCs applied in accordance
with the following rule that is necessary to satisfy symmetry. If NSEG > 3,
SPCs must be applied to both components 1 and 2 or to neither. The same
rule applies to components 4 and 5.
• Grid points listed on CYAX Bulk Data entries must be in the a-set.
See Also
• “CYAX” on page 1088 of the NX Nastran Quick Reference Guide
Using CYSUP
The CYSUP bulk data entry lets you define rigid body supports for cyclic symmetry
analysis. All supported degrees-of-freedom for a rigid body must be specified at a
single grid point. The format of the CYSUP entry is:
1 2 3 4 5 6 7 8 9 10
CYSUP GID C
Field Contents
GID Grid point identification number.
C Component numbers.
The total structure can have, at most, six rigid body degrees of freedom. Supports
that remove only the rigid body motions are applied to the k = 0 and k = 1 harmonics
of the components of motion listed on this entry. These components may not be
constrained by SPCs, may not be referenced as dependent degrees of freedom by
MPCs and/or rigid elements, and may not be referenced on OMIT or OMIT1 Bulk
Data entries.
Rigid body motion is possible only for harmonics k = 0 and k = 1. Translation along
and rotation about the axis of symmetry are the only possible rigid body motions for
k = 0, while translations normal to the axis of symmetry and rotations about these
CHAPTER 24 737
Cyclic Symmetry
translational directions are the only possible rigid body motions for k = 1. The NX
Nastran Quick Reference Guide lists special rules relative to the proper support of
rigid body motion for CYSUP.
See Also
• “CYSUP” on page 1092 of the NX Nastran Quick Reference Guide
Using LOADCYN
The LOADCYN entry lets you identify the segments on which referenced loads,
defined in terms of physical components, are to be applied. The format of the
LOADCYN entry is:
1 2 3 4 5 6 7 8 9 10
LOADCYN SID S SEGID SEGTYPE S1 L1 S2 L2
Field Contents
S Scale factor.
SEGID Segment identification number.
SEGTYPE Segment type.
Si Scale factors.
Li Load set ID numbers.
The loads are referenced by listing the identification numbers of one or two LOAD
Bulk Data entries. Scale factors on the LOADCYN Bulk Data entry are used to form
a linear combination of the load sets referenced on the LOAD Bulk Data entries.
Alternatively, the LOADCYN Bulk Data entries may directly reference load sets
comprised of the types of static load definition entries shown in Table 24-6.
Table 24-6 Static Load Definitions for Cyclic Symmetry
Point Distributed Temperature Enforced

Loads Loads Loads Displacement
FORCE PLOAD1 TEMP SPCD
FORCE1 PLOAD2 TEMPP1
FORCE2 PLOAD4 TEMPRB
MOMENT TEMD
MOMENT1
Table 24-6 Static Load Definitions for Cyclic Symmetry
Point Distributed Temperature Enforced

Loads Loads Loads Displacement
MOMENT2
SLOAD
The loads listed on these entries are applied at the images in the segment referenced
in field 4 of the LOADCYN entry of the grid points and elements listed on these
entries. The coordinate systems for image points are rotated and/or reflected from
the coordinate system in the fundamental region.
Note that GRAV and RFORCE Bulk Data entries are not included in the above list.
These static load entries are handled automatically in a very straightforward fashion
through the LOADCYH Bulk Data entry.
See Also
• “LOADCYN” on page 1354 of the NX Nastran Quick Reference Guide
Using LOADCYH
The LOADCYH entry provides for the identification of referenced loads that are
defined in terms of harmonic components. The format of the LOADCYH entry is as
follows:
1 2 3 4 5 6 7 8 9 10
LOADCYH SID S HID HTYPE S1 L1 S2 L2
Field Contents
S Scale factor.
HID Harmonic component.
HTYPE Harmonic type.
Si Scale factor on Li.
Li Load set identification number.
This option is particularly useful for the definition of gravity (GRAV) and centrifugal
loads (RFORCE) because you only need to specify GRAV or RFORCE under the
HTYPE field on the entry and the correct harmonic components of the specified load
are automatically generated. The component of gravity parallel to the axis of
CHAPTER 24 739
Cyclic Symmetry
symmetry produces harmonic loads of the order k = 0, and components of gravity

perpendicular to the axis of symmetry produce loads of the order k = 1.
Components of centrifugal force produce harmonic loads of the order k = 0 if the
spin axis coincides with the axis of symmetry, and of the order k = 0, k = 1, and k = 2
if it does not.
See Also
• “LOADCYH” on page 1352 of the NX Nastran Quick Reference Guide
Using LOADCYT
You use the LOADCYT Bulk Data entry with the AXI option to specify loads in
terms of physical components as a tabular function of the azimuth angle (measured
in degrees). The format of the LOADCYT entry is:
1 2 3 4 5 6 7 8 9 10
LOADCYT SID TABLEID1 LOADSET1 METHOD1 TABLEID2 LOADSET2 METHOD2
Field Contents
TABLEIDi Table ID for table load input for load set Li.
LOADSETi Load set Li.
METHODi Method of interpolation.
0 interpolate the load with the Fourier coefficients
specified in the table up to the specified number of
harmonics.
1 interpolate the magnitude of the load at
corresponding grid points in all segments.
This capability is implemented by referencing a set of load definition entries under
the heading LOADSET on the LOADCYT entry.
These entries define a magnitude for the load. The azimuthal variation of the
applied load is specified on the TABLEDi entry referenced under the heading
TABLEID. The net applied load is the product of the magnitude given on the load
definition entry times the appropriate value on the TABLEDi entry. The heading
METHODi on the LOADCYT entry is utilized to specify the type of interpolation to
be used in assigning loads to degrees of freedom.
See Also
• “LOADCYT” on page 1356 of the NX Nastran Quick Reference Guide
Case Control Commands Specific to Cyclic Symmetry

Table 24-7 lists Case Control commands that are unique to cyclic symmetry analyses.
Table 24-7 Case Control Commands Unique to Cyclic Symmetry
Command Function
HARMONICS Specifies harmonics for which the computation is desired.
DSYM Provides for either one or two planes of overall symmetry
under the DIH option.
NOUTPUT Controls requests for the output of physical components.
HOUTPUT Controls requests for the output of harmonic components.
With cyclic symmetry, note that:
• References to SPC and/or MPC sets in the Case Control Section must
appear above the subcase level, i.e., before the first subcase.
• You must have one subcase per loading condition. Loads, enforced
deformations, and enforced displacements are requested in the standard
manner, i.e., through LOAD, TEMP(LOAD), DEFORM and SPC set
selections. As described above, loads can be input in terms of physical
components and/or harmonic components. With the AXI option, you can
also specify loads in terms of physical components that are tabulated as a
function of the azimuth angle (in degrees).
Using HARMONICS Case Control Command

A HARMONICS command must appear above the subcase level (i.e., before the
commands that describe the first subcase) to specify the harmonics for which
solutions are to be computed. The HARMONIC command cannot reference a set
definition that specifies “ALL”. If you need to use all the harmonics, specify:
HARMONICS = ALL
See Also
• “HARMONICS” on page 264 of the NX Nastran Quick Reference Guide
Requesting Output
The NOUTPUT and HOUTPUT Case Control commands let you request output from
a cyclic symmetry analysis.
CHAPTER 24 741
Cyclic Symmetry
• You can use the NOUTPUT command to request physical output within a
subcase for all output quantities, such as forces and displacements,
requested in the subcase.
• You can use the HOUTPUT command to request harmonic output within
a subcase for all output quantities requested in the subcase.
These commands let you specify the segments for which you want output and, with
HOUTPUT, the cyclic components for which output is desired.
See Also
• “HOUTPUT” on page 267 of the NX Nastran Quick Reference Guide
• “NOUTPUT” on page 311 of the NX Nastran Quick Reference Guide
In the case of thermal or deformation loading, you can only request element force
and stress output that’s consistent with the form of the loading. In other words, you
can only request physical output if you defined the loads with a LOADCYN Bulk
Data entry and harmonic output if you defined the loads with a LOADCYH Bulk
Data entry.
Using a single subcase for each loading condition introduces the need to refer to
“coded” subcases for SORT2 output and plotting purposes. Since a single subcase
now refers to NSEG segments (in the case of ROT and AXI options) and two times
NSEG segments (including Right and Left halves, in the case of DIH option), you
need a unique coded subcase ID to refer to a particular segment of the structure for
plotting purposes in PLOT commands wherever a subcase ID is to be supplied. This
coded ID is obtained as follows:
Coded Subcase ID for a Particular Segment

= SUBCASE ID ⋅ 10000 + SEGMENT ID ⋅ 10 + zero (for ROT or AXI option)
 1 ( if R segment ) 
= SUBCASE ID ⋅ (10000) + SEGMENT ID ⋅ 10 + 
 2 ( if L segment ) 
Coded Subcase ID for a Particular Harmonic
1 ( if component = C ) 
 
2 ( if component = S ) 
= SUBCASE ID ⋅ 10000 + HARMONIC ID ⋅ 10 +  
3 ( if component = C* ) 
4 ( if component = S* ) 

Note:
• If a subcase contains TEMP(LOAD) and/or DEFORM commands,

irrespective of the presence of a LOAD command, element force and stress
output are only correct in terms of physical components if the
TEMP(LOAD) and the DEFORM loads are defined in terms of physical
components. Similarly, if the TEMP(LOAD) and DEFORM data are
provided in terms of harmonic components, element force and stress output
are only correct in terms of harmonic components.
• If, under the DIH option of cyclic symmetry, a DSYM command is explicitly
present under a subcase, the output is, by default, limited to the segments
in the first half plane if there is one plane of overall symmetry, and to the
segments in the first quadrant if there are two planes of overall symmetry.
This default applies irrespective of the segments listed on NOUTPUT
commands.
You can request undeformed and deformed plots of the fundamental region.
Typical Case Control Section Examples

To illustrate some of the rules described above, typical Case Control sections for the
cylinder (Figure 24-9) and square plate (Figure 24-10) problems are shown in
Listing 24-2 and Listing 24-3, respectively.
CHAPTER 24 743
Cyclic Symmetry
c
z
b
Finite
Element
Model
d
10.0 θ
5° 5°
40 a
Plane of
Symmetry
20 – CQUAD4 Elements
Boundary Conditions:
Edge ad
uz = 0
θr = θθ = 0
Edge bc
Free
Figure 24-9 Cylinder Example
x
a
b
x'
d l
y --- l = 10.0
2
l T = 0.20
y' --- c
2
Figure 24-10 Square Plate Example

Listing 24-1 ROT or AXI Symmetry Type
TITLE=CYCLIC SYMMETRY-STATICS
SUBTITLE=FIVE DEGREE SEGMENT OF CYLINDER
LABEL=AXISYMMETRIC LOADING
SPC=1
SET 1=0
SET 2=1
HARMONICS=1
NOUTPUT=2
DISP=ALL
STRESS=ALL
FORCE=ALL
SUBCASE 1
LABEL=BAND LOADING
LOAD=1
SUBCASE 2
LABEL=CENTRIFUGAL LOADING
LOAD=2
BEGIN BULK
Listing 24-2 DIH Symmetry Type
TITLE=CYCLIC SYMMETRY-STATICS
SUBTITLE=SIMPLY-SUPPORTED SQUARE PLATE
SPC=1
HARMONICS=ALL
DISPLACEMENT=ALL
NOUTPUT=ALL
DSYM=SS
SUBCASE 1
LABEL=LOAD AT CENTER OF PLATE
LOAD=1
SUBCASE 2
LABEL=UNIFORM PRESSURE LOAD
LOAD=2
SUBCASE 3
LABEL=TEMPERATURE LOAD
TEMP(LOAD)=3
SUBCASE 4
LABEL=INPLANE GRAVITY LOAD
LOAD=4
DSYM=SA
BEGIN BULK
The following comments apply to the Case Control sections above:
• In Listing 24-1, the HARMONICS command specifies that solutions are to

be computed only for the zero harmonic. If the applied loads are
axisymmetric, no higher harmonics need to be considered. For Listing 24-2,
HARMONICS = ALL specifies that solutions are to be computed for all
meaningful harmonics.
CHAPTER 24 745
Cyclic Symmetry
• A DSYM command appears in all four subcases of Listing 24-2. The first
three subcases specify that the boundary conditions for both side 1 and
side 2 are to be symmetric. This specification has two effects: 1) loads need
only to be specified for the first quadrant of the structure (the fundamental
region in this case), and 2) output is only processed for the segments in the
first quadrant of the model irrespective of the analyst’s requests for output
in other segments. The particular lateral gravity condition considered in
the fourth subcase is symmetric with respect to side 1 of the fundamental
region and antisymmetric with respect to a plane perpendicular to side 1
of the fundamental region. It must be emphasized that the appropriate
boundary conditions for each of the symmetry conditions involved in this
example are internally applied by the program.
• Note that the specifications of loading conditions in the Case Control
Section conform to the procedural rules for static analysis in NX Nastran.
• The NOUTPUT Case Control command controls output requests for
selected segments. For Listing 24-1, output requests are limited to the first
segment, i.e., the fundamental region. For Listing 24-2, output requests
for all four subcases are limited to segments in the first quadrant because
of the presence of the DSYM commands, even though output for ALL
right-handed segments is requested. The specification of requests for
specific output quantities in the Case Control Section conform with the
procedural rules for static analysis in NX Nastran.
24.5 Performing Axisymmetric Analysis

You can use NX Nastran’s cyclic symmetry capabilities to solve axisymmetric
problems without defining any special axisymmetric elements.
Solving Axisymmetric Problems

Use the following procedure to solve static load problems:
1. Select the AXI symmetry option on the CYSYM Bulk Data entry.
2. Select a large number of segments (for example, 180) on the CYSYM Bulk
Data entry.
3. Construct a model with two rows of grid points and a single row of
elements in the meridional direction.
4. Define a set of harmonics on a SET Case Control command. The set of
harmonics should be reasonably small for efficiency, but sufficient in size
to yield reasonably accurate results.
5. Select the defined set of harmonics on the HARMONICS Case Control

command. Convergence of the harmonic output towards zero for the
higher harmonics will verify that the number of harmonics specified on the
HARMONICS Case Control command and used in the analysis is sufficient.
6. Select a static loading condition with the LOAD, TEMP(LOAD), and
DEFORM Case Control commands which reference LOADCYN,
LOADCYH, or LOADCYT Bulk Data entries.
7. Place static loads only on Side 1 or on elements. The boundary conditions
will take care of loading on Side2.
8. Prepare the referenced TABLED1 entries if LOADCYT entries are used to
specify loads as a function of azimuth angle. The number of x-y pairs in a
given table should be sufficient to produce reasonably accurate results
when linearly interpolated.
9. Select the physical output for a reasonably small number of segments with
the NOUTPUT Case Control command or ALL of the harmonic output with
the HOUTPUT Case Control command.
Use the following procedure to obtain the vibration modes of an axisymmetric

structure:
1. Select the AXI symmetry option on the CYSYM Bulk Data entry.
2. Select a large number of segments (for example, 90) on the CYSYM Bulk
Data entry.
3. Construct a model with two rows of grid points and a single row of
elements in the meridional direction.
4. Define a set of harmonics on a SET Case Control command where the
number of harmonics is reasonably small for efficiency, but sufficient to
cover the frequency range of interest.
5. Select the defined set of harmonics on the HARMONICS Case Control
command.
6. Select all the harmonic output with the HOUTPUT Case Control command.
This will give the eigenvectors, in a concise form, which may be interpreted
by the formula
U ( φ ) = U kc cos kφ + U ks sin kφ
where φ is the azimuth angle measured from Side 1 of the fundamental
region and K is the harmonic index.
CHAPTER 24 747
Cyclic Symmetry
Axisymmetric Example
The following cyclic symmetry example problem uses the same cylinder model as
“Example 5 – Buckling of a Cylinder Under Uniform Axial Load” on page 625.
A five degrees strip, as shown in Figure 24-11, is modeled using the axisymmetric
features.
Figure 24-11 A Five Degrees Strip of a Cylinder Using Axisymmetry
A copy of the input file is shown in Figure 24-7. Since only a five degrees strip is
modeled, seventy-two segments are required to generate the full cylinder model
internally. The CYSYM entry accomplishes this purpose. Furthermore, the
boundary points for segment one are defined on the CYJOIN entries—one each for
side one and side two. Note that with axisymmetry, no interior points are allowed.
The applied load is a uniform compressive load applied at the top of the cylinder,
using the LOADCYT/TABLED1/FORCE combinations. The TABLED1 defines the
spatial distribution around the circumference of the cylinder-which in this case is

evenly distributed. Since the structure is axisymmetric, and the load is constant
around the circumference, only the zeroth harmonic is required for the analysis. The
displacements at the top of the cylinder and stresses at the top and bottom of the
cylinders are shown in Figure 24-12. The same output using the full model is shown
in Figure 24-13 for comparison.
Listing 24-3 Input File for Five Degrees Strip of Cylinder
$ filename - cyclic1.dat
$
SOL 114
TIME 600
CEND
SUBCASE 1
TITLE= A 5 DEG STRIP
SET 100 = 381,382
SET 200 = 1,343
SET 1000 = 1
SET 1 = 0
HARMONICS = 1
NOUTPUT = 1000
LOAD = 1
SPC = 1
DISPLACEMENT= 100
SPCFORCES=ALL
STRESS= 200
BEGIN BULK
PARAM POST -1
PARAM,SNORM,20.
$ Elements and Element Properties for region : pshell.1
CYSYM,72,AXI
CYJOIN,1,T2,1,20,39,58,77,96
,115,134,153,172,191,210,229,248,
,267,286,305,324,343,362,381
$
CYJOIN,2,T2,2,21,40,59,78,97
,116,135,154,173,192,211,230,249,
,268,287,306,325,344,363,382
$
SPC1,1,123,1,2
SPC1,1,12,381,382
$
LOADCYT,1,10,11,1
TABLED1,10
,0.,1.,360.,1.,ENDT
FORCE,11,381,1,1388.889,,,-1.
$
CHAPTER 24 749
Cyclic Symmetry
PSHELL 1 1 .03 1 1
CQUAD4 1 1 1 2 21 20 0. 0.
CQUAD4 19 1 20 21 40 39 0. 0.
CQUAD4 37 1 39 40 59 58 0. 0.
CQUAD4 55 1 58 59 78 77 0. 0.
CQUAD4 73 1 77 78 97 96 0. 0.
CQUAD4 91 1 96 97 116 115 0. 0.
CQUAD4 109 1 115 116 135 134 0. 0.
CQUAD4 127 1 134 135 154 153 0. 0.
CQUAD4 145 1 153 154 173 172 0. 0.
CQUAD4 163 1 172 173 192 191 0. 0.
CQUAD4 181 1 191 192 211 210 0. 0.
CQUAD4 199 1 210 211 230 229 0. 0.
CQUAD4 217 1 229 230 249 248 0. 0.
CQUAD4 235 1 248 249 268 267 0. 0.
CQUAD4 253 1 267 268 287 286 0. 0.
CQUAD4 271 1 286 287 306 305 0. 0.
CQUAD4 289 1 305 306 325 324 0. 0.
CQUAD4 307 1 324 325 344 343 0. 0.
CQUAD4 325 1 343 344 363 362 0. 0.
CQUAD4 343 1 362 363 382 381 0. 0.
$
MAT1 1 1.+7 3.84+6 .3 0. 0. 0. 0.
$ Nodes of Group : group5
GRID 1 1 10. 0. 0.1
GRID 2 1 10. 5. 0.1
GRID 20 1 10. 0. 1.1
GRID 21 1 10. 5. 1.1
GRID 39 1 10. 0. 2.1
GRID 40 1 10. 5. 2.1
GRID 58 1 10. 0. 3.1
GRID 59 1 10. 5. 3.1
GRID 77 1 10. 0. 4.1
GRID 78 1 10. 5. 4.1
GRID 96 1 10. 0. 5.1
GRID 97 1 10. 5. 5.1
GRID 115 1 10. 0. 6.1
GRID 116 1 10. 5. 6.1
GRID 134 1 10. 0. 7.1
GRID 135 1 10. 5. 7.1
GRID 153 1 10. 0. 8.1
GRID 154 1 10. 5. 8.1
GRID 172 1 10. 0. 9.1
GRID 173 1 10. 5. 9.1
GRID 191 1 10. 0. 10.1
GRID 192 1 10. 5. 10.1
GRID 210 1 10. 0. 11.1
GRID 211 1 10. 5. 11.1
GRID 229 1 10. 0. 12. 1

GRID 230 1 10. 5. 12. 1
GRID 248 1 10. 0. 13. 1
GRID 249 1 10. 5. 13. 1
GRID 267 1 10. 0. 14. 1
GRID 268 1 10. 5. 14. 1
GRID 286 1 10. 0. 15. 1
GRID 287 1 10. 5. 15. 1
GRID 305 1 10. 0. 16. 1
GRID 306 1 10. 5. 16. 1
GRID 324 1 10. 0. 17. 1
GRID 325 1 10. 5. 17. 1
GRID 343 1 10. 0. 18. 1
GRID 344 1 10. 5. 18. 1
GRID 362 1 10. 0. 19. 1
GRID 363 1 10. 5. 19. 1
GRID 381 1 10. 0. 20. 1
GRID 382 1 10. 5. 20. 1
$
CORD2C 1 0. 0. 0. 0. 0. 1.
+ A
+ A 1. 0. 0.
ENDDATA
0 SUBCASE = 1 SEGMENT = 1

381 G .0 .0 -1.058190E-01 2.021944E-16 -2.435790E-02 5.526086E-19
382 G .0 .0 -1.058190E-01 1.009687E-16 -2.435790E-02 1.583684E-17
0 HARMO
1 A 5 DEG STRIP FEBRUARY 4, 2004 NX NASTRAN 1/23/04 PAGE 49
0 SUBCASE = 1 SEGMENT = 1
S T R E S S E S I N Q U A D R I L A T E R A L E L E M E N T S ( Q U A D 4 )
ELEMENT FIBRE STRESSES IN ELEMENT COORD SYSTEM PRINCIPAL STRESSES (ZERO SHEAR)
ID. DISTANCE NORMAL-X NORMAL-Y SHEAR-XY ANGLE MAJOR MINOR VON MISES
0 1 -1.500000E-02 -7.481326E+03 -5.592213E+04 2.894851E-10 .0000 -7.481326E+03 -5.592213E+04
5.258216E+04
1.500000E-02 -5.769141E+03 -5.021485E+04 2.903178E-10 .0000 -5.769141E+03 -5.021485E+04
4.759325E+04
0 343 -1.500000E-02 -7.481326E+03 -5.592213E+04 -5.481948E-10 .0000 -7.481326E+03 -5.592213E+04
5.258216E+04
1.500000E-02 -5.769141E+03 -5.021485E+04 -5.431988E-10 .0000 -5.769141E+03 -5.021485E+04
4.759325E+04
Figure 24-12 Abridged Output Using Axisymmetry
CHAPTER 24 751
Cyclic Symmetry

381 G .0 .0 -1.058189E-01 .0 -2.433120E-02 3.729655E-16
382 G .0 .0 -1.058189E-01 .0 -2.442414E-02 -5.255163E-16
S T R E S S E S I N Q U A D R I L A T E R A L E L E M E N T S ( Q U A D 4 )
ELEMENT FIBRE STRESSES IN ELEMENT COORD SYSTEM PRINCIPAL STRESSES (ZERO SHEAR)
ID. DISTANCE NORMAL-X NORMAL-Y SHEAR-XY ANGLE MAJOR MINOR VON
MISES
0 1 -1.500000E-02 -7.482788E+03 -5.592399E+04 -1.636876E-10 .0000 -7.482788E+03 -5.592399E+04
5.258343E+04
1.500000E-02 -5.769486E+03 -5.021298E+04 -1.697938E-10 .0000 -5.769486E+03 -5.021298E+04
4.759125E+04
0 343 -1.500000E-02 -7.482788E+03 -5.592399E+04 -4.949596E-10 .0000 -7.482788E+03 -5.592399E+04
5.258343E+04
1.500000E-02 -5.769486E+03 -5.021298E+04 -2.932691E-10 .0000 -5.769486E+03 -5.021298E+04
4.759125E+04
Figure 24-13 Abridged OUtput Using Full Model
25
CHAPTER
p-Elements
■ Introduction to p-Elements
■ p-Version Capabilities
■ p-Element Loads and Constraints
■ p-Element Example Problems
■ Solid p-Elements in Linear Dynamic Solution Sequences
■ p-Element Thermal Analysis
NX Nastran User’s Guide Introduction to p-Elements

25.1 Introduction to p-Elements
One of the main applications of p-version elements is detailed stress analysis. The p-
elements have higher-order polynomials, which provide better representation of
complex stress fields. For these complex stress fields, the geometry, loads, and
boundary conditions must be represented accurately. This includes modeling fillets
instead of sharp corners, distributed loads and constraints instead of point loads and
constraints, etc. A more detailed model leads to more detailed results.
p-Elements have several modeling advantages. With the higher-order polynomials,

the geometry and loads can be represented more accurately. Generally, fewer p-
version elements with curvature than h-version elements with straight edges are
required.
The accuracy of the analysis is controlled primarily by the polynomial level, not by
the element size. Generally the user needs to only use the minimum number of
elements necessary to adequately model the geometry, independent of expected
result characteristics. Polynomials levels can then be assigned based on areas of the
model in which the user is interested, and areas of the model in which the user is not.
(Of course the same polynomial level can be assigned everywhere for a minimum of
user involvement, but this is not as efficient.) If a more accurate answer is necessary,
the polynomial levels can be increased, either manually by the user or automatically
by the program, until the answers reach the specified accuracy, without changing the
mesh. Adding degrees-of-freedom until the error decreases to a specified level is
known as adaptivity. By adding higher-order polynomials instead of refining the
mesh, a faster rate of convergence, and therefore fewer iterations, can be achieved.
p-Elements in NX Nastran
In NX Nastran, p-elements can be used in linear static (SOL 101) and normal modes
(SOL 103) analysis only. They allow different polynomial orders in the three
coordinate directions, and use bubble functions at p = 1 and optimum integration
algorithms at p ≥ 2 to achieve better accuracy. The elements will also pass the patch
tests and rigid body motion at any allowable p-level. These additional features
improve the accuracy over conventional p-elements.
Compatibility with elements of different p-orders, and with the existing solid
elements, is automatically enforced. The user does not have to write constraint
equations or use special elements to provide displacement continuity. Loads and
boundary conditions can be applied to the elements themselves, or to the geometry
on which they are defined.
The definition of geometry has been added to NX Nastran. This geometry can be
defined with midside points, as in the existing elements; with rational parametric
cubics or equations; or even with any external geometry. A facility has been set up
CHAPTER 25 755
p-Elements
to invoke remote procedure calls to any independent program and database that
provides a toolkit for geometry queries, which would allow proprietary geometries
of other pre-processors or CAD packages to be used. Associativity has been added
so that the elements are defined on the geometry, and the element geometry
calculated accordingly.
The error estimator requires only a single analysis, not the difference between two
analyses. For linear elements, it is based on the grid-point stress discontinuity; for
higher polynomial orders, it considers the contribution of the additional terms for
the individual edge, face, and body functions. Therefore it provides efficient
information because of the directional sensitivity within the elements.
The adaptivity method is very flexible, working on an element-by-element basis.

Using the error estimator, the polynomial level within each element can be varied
independently in the three directions. At the most basic level, the edges are varied,
since they determine the face and body functions. Different error criteria may be
used in different regions, allowing areas of primary interest to have lower error
tolerances. Multiple subcases may be used.
Combining p- and h-Elements

The main advantage of implementing p-elements in NX Nastran is the ability to
combine them with the existing h-elements for global/local analysis. The mesh
could consist mostly of h-version elements, with p-version elements in the areas of
interest. This method of global/local interface is the most accurate, because it
provides both stiffness and loads information, rather than transferring just
boundary displacements or tractions into the local model.
h-Version elements still have several advantages. They are better for global
behavior, such as loads analysis, where the exact geometry is not used. They are
better for strongly singular problems, such as nonlinear problems.
Approximations, such as sharp corners, point loads, point constraints, MPCs, etc.
may be used. For a given iteration, they tend to have shorter solution times and use
less disk space. Finally, h-elements are a very mature technology.
Putting both approaches in the same program combines the advantages for general
problems. The p-elements may be connected directly to the existing h-elements,
and continuity is automatically enforced. This provides global/local analysis in a
single run.
Most general problems have both some global and some local aspects; i.e., regions
where h-elements and p-elements, respectively, are appropriate. Therefore the best
approach is combined h- and p-adaptivity, with different criteria in different parts
of the model.
25.2 p-Version Capabilities

Following is a limited description of the Bulk Data entries and Case Control
commands for the p-version elements. Table 25-1 contains a comparison with the
conventional h-version elements. A full description of the entries may be found in
the NX Nastran Quick Reference Guide.
Solution Sequences Supported

Solution Sequences 101, 103, 107, 108, 109, 110, 111, 112, and 101 (Heat Transfer).
Geometry
With the h-version elements, geometry is determined by the addition of midside grid
points to the element definition. With the p-version elements, you can define the
edges and faces of elements to lie on curves and surfaces, which can span an arbitrary
number of elements. You don’t have to define midside grid points.
The Bulk Data entries to define the associativity between the finite element model
and the geometrical model are the FEEDGE and FEFACE entries for the edges and
faces, respectively. These entries reference the grids that make up the edges and
faces, not the elements themselves, and therefore do not have to be repeated for each
element that shares the edge or face. The old method of midside grid points is still
available for the FEEDGE entry, which would reference one or two GRIDs or
POINTs. The POINT entry is very similar to the GRID entry, except that it defines a
location in space with no associated degrees-of-freedom.
The more useful way of defining the FEEDGEs and FEFACEs is to have them
reference geometrical curves and surfaces, defined by the GMCURV and GMSURF
entries. GMCURVs and GMSURFs are functions of one or two parametric variables,
respectively, and can be defined as rational parametric cubics or general equations.
For instance, a hole could be defined with the parametric equations for a cylinder,
which all the element faces on the hole would then reference. The GMCURVs and
GMSURFs could also reference external geometry from a CAD or other database,
using remote procedure calls with an external programming interface. When
geometry has been specified, it is approximated as cubics for the element-level
computations. Using cubics ensures that the elements pass the rigid-body and patch
tests.
The two different methods of defining the geometry are shown in Figure 25-1.
Convective coordinate systems may be defined on the FEEDGEs, FEFACEs,

GMCURVs, and GMSURFs. The directions of the systems are normal and tangent to
the entities at every point. These coordinate systems can then be used for material
properties, loads, boundary conditions, etc.
CHAPTER 25 757
p-Elements
POINT option GMCURV option

Figure 25-1 Methods of Defining Element Geometry
Elements
Only the CHEXA, CPENTA, CTETRA, CQUAD, CTRIA, and CBEAM elements
may be defined as p-elements. Midside grid points may not be specified on p-
version elements; curvature is defined by the FEEDGE or FEFACE entries, as
previously described.
The standard Bulk Data entries, PSOLID, PSHELL, and PBEAM are used to defined
the element properties. The p-version properties are defined in the PVAL and
ADAPT entries, which are described later.
The p-version elements may be joined to existing h-version elements with a few
considerations. FEEDGEs and FEFACEs may not be defined on the interface, since
the h-elements can not accept them. Midside grid points may not be defined on the
interface, since the p-elements can not accept them. The polynomial levels of the p-
elements will be lowered automatically so that displacement continuity is enforced.
Materials
The standard isotropic, orthotropic, and anisotropic materials are available for p-
elements. Orthotropic and anisotropic material properties may be defined in a
curvilinear material system.
25.3 p-Element Loads and Constraints

Defining p-Element Loads
For p-version elements in detailed stress analysis, it is important to use distributed
loads. Concentrated forces cause singularities in the stress field, and therefore
should be avoided.
Line and Pressure Loads in p-Element Analysis

The GMLOAD Bulk Data entry has been defined for the p-version elements. You can
use the GMLOAD entry to specify line and pressure loads by:
• Applying directly to the finite elements entities, FEEDGEs and FEFACEs.

• Applying to the geometry entities GMCURV and GMSURF.
The load can be distributed as constant, linear, quadratic, cubic, equation, or table,
and can be given in any coordinate system.
Thermal Loads
For thermal loads, the TEMPF entry allows specification of an equation or table to
describe the variation of the temperature field over the elements. The TEMP and
TEMPD entries may also be used, but will result in linear interpolation of the
temperature from the grids. The TEMPP1 entry defines a constant temperature and
temperature gradient distribution in a shell element. The TEMPRB entry defines a
linear temperature and temperature gradient distribution in a beam element.
Gravity Loads
You can use the GRAV entry to define gravity loads for p-elements.
Pressure Loads
You can use the PLOAD4 entry to define a pressure load on a p-element.
Point Loads
You can use the FORCE entry to define point loads applied to grid points. No point
loads can be applied to finite element edges or faces. You should only use FORCE
where the adjacent elements have a fixed p-level and the results aren’t of interest.
Defining p-Element Constraints

As with the loads, it is important to use distributed boundary conditions for p-
version elements in detailed stress analysis. Concentrated boundary conditions also
cause singularities in the stress field and should be avoided.
The GMSPC and GMBC Bulk Data entries have been defined for the p-version
elements. These can apply distributed boundary conditions to GRIDs, FEEDGEs,
FEFACEs, GMCURVs, or GMSURFs. Note that they should be applied to GRIDs
only to prevent rigid-body motion. The imposed boundary conditions can be
distributed as constant, linear, quadratic, cubic, equation, or table, and can be given
CHAPTER 25 759
p-Elements
in any coordinate system. The current SPC and SPC1 entries can be used, but should
only be used to prevent rigid-body motion or used where the adjacent elements
have a fixed p-level and the results are not of interest.
Boundary Conditions
The SPC and SPCD entries are used for the point constraints which are allowed only
on corner GRID points.
The GMSPC entry is used to define zero constraints for FEEDGEs, FEFACEs,
GMCURVs, and GMSURFs.
For non-zero constraints, the GMBC entry is used.
Resolution of Conflicting Zero Constraints

In general, the hierarchy set to resolve multiple input data (constraints and enforced
displacements) for GRID, FEEDGE, FEFACE, GMCURV, and GMSURF entities is:
1. GRIDs
2. FEEDGEs
3. GMCURVs
4. FEFACEs
5. GMSURFs
The rules for nonzero boundary constraints are:
• Displacement values specified for each component of a given GMSURF

entry are applied to the same component of all GRID, FEEDGE and
FEFACE degrees-of-freedom that lie within the GMSURF.
• Displacement values specified for each component of different GMSURF
entries are averaged and applied to the same component of all GRID,
FEEDGE degrees-of-freedom which are shared by (common to) the
multiple GMSURFs.
• Displacement values specified for a given FEFACE entry are applied to all
GRID, FEEDGE and FEFACE degrees-of-freedom that lie within the
FEFACE. This data overrides the data that is specified for all the
components of the given GRID, FEEDGE and FEFACE degrees-of-
freedom that lie within the FEFACE by using GMSURF entries.
• Displacement values specified for each component of different FEFACE
entries are averaged and applied to the same component of all GRID,
FEEDGE degrees-of-freedom which are shared by (common to) the
multiple FEFACEs. This data overrides the data that is specified for all the
components of the given FEEDGE and GRIDs by using GMSURF entries.
• Displacement values specified for each component of a given GMCURV

entry are applied to the same component of all GRID, FEEDGE, degrees-of-
freedom that lie within the GMCURV. This data overrides the data for all
components that is specified for the given FEEDGE and GRIDs by using
GMSURF and FEFACE entries.
• Displacement values specified for each component of different CURV
entries are averaged and applied to the same component of all GRID
degrees-of-freedom which are shared by (common to) the multiple
GMCURVs. This data overrides the data for all the components that is
specified for the given FEEDGE and GRIDS by using GMSURF and
FEFACE entries.
• Displacement values specified for each component of a given FEEDGE
entry are applied to the same component of all GRID, FEEDGE, degrees-of-
freedom that lie within the FEEDGE. This data overrides the data for all
components that is specified for the given FEEDGE by using GMCURV or
FEFACE or GMSURF entries.
• Displacement values specified for each component of different FEEDGE
entries are averaged and applied to the same component of all GRID
degrees-of-freedom which are shared by (common to) the multiple
FEEDGEs. This data overrides the data for all the components that is
specified for the given FEEDGE by using GMCURV or FEFACE or
GMSURF entries.
• Grids have the highest priority, i.e., any value/property specified using a
GRID entry overrides all other information associated with that GRID. If
multiple entries are used for a given GRID, e.g., multiple SPCs, then the
existing rules of NX Nastran govern (SPCs are combined, FORCE fare
added, SPCDs for the same component are not allowed).
• It is important to recall that these displacements are assumed to be in the
global coordinate system and that the interconsistency of the output
coordinate systems of the various GRIDs, FEEDGEs, FEFACEs is not
checked.
p-Version Adaptivity
In order for an element to be recognized as a p-version element, it must be assigned
the polynomial values in the PVAL Bulk Data entry, which is referenced in the
ADAPT Bulk Data entry. In order for the p-element to be adaptive, the adaptivity
parameters must be assigned in the ADAPT entry.
The PVAL Bulk Data entry assigns the polynomial levels for the three directions to a
single element or a set of elements, and is referenced in the ADAPT Bulk Data entry.
By default, the three directions are along the element edges. If a coordinate system
CHAPTER 25 761
p-Elements
is specified, the polynomial level of each edge is set to a weighted combination of

the coordinate directions. There may be multiple PVALs with the same ID; all of the
elements do not have to use the same entry. Each p-version element must have a
starting, minimum, and maximum PVAL ID; however, these do not need to be
unique. Since the polynomial levels can not decrease, the starting and minimum
PVAL IDs can be the same. If the starting and maximum PVAL IDs are the same,
the element will have a fixed p-level.
After the p-value distribution has been defined, all of the values will be resolved to
ensure displacement continuity. This includes selecting the highest p-level
specified for elements with common edges, decreasing the p-level for p-elements
adjacent to h-elements, and increasing the p-level on curved edges to adequately
map them.
The ADAPT Bulk Data entry controls the p-adaptivity. The ADAPT Bulk Data entry
is referenced with the ADAPT Case Control command, and there may only be one
unique ADAPT command in the Bulk Data file. The ADAPT entry contains the IDs
for the starting, minimum, and maximum p-value distributions and the maximum
number of iterations. It also contains the adaptivity parameters, which may be
assigned differently to different sets of elements. These sets of elements do not have
to correspond with any sets used in the PVAL entries. The adaptivity type may be
element-by-element p-adaptive, uniform p-adaptive, no change, or list of p-
distributions. For each set in the ADAPT entry, the error estimator method, error
tolerance, and stress and strain tolerances are specified. If a p-version element is not
included in one of the sets, it is not adaptive.
For a non-adaptive p-version solution, the ADAPT Bulk Data entry may specify a
maximum of one iteration, or may specify an adaptivity type of no change for all the
elements. A third way is to use the PSET Bulk Data entries, which have the same
format as the PVAL entry, and reference them with the ADAPT Case Control
command.
Data Recovery
Since the p-version analysis may have multiple iterations, the output control must
have the capability of differing among the iterations. For this reason, the OUTPUT
Bulk Data entry was added. For a given set of elements, the conventional
displacements, stresses, and strains, as well as the element errors and polynomial
values, may be printed, plotted, or punched for the given iteration. The OUTPUT
entry is referenced by the DATAREC Case Control command. The conventional
Case Control commands still apply to all iterations, but the OUTPUT entry has
precedence.
The p-version elements also tend to be larger than h-version elements, and have
higher-order distributions of displacements, stresses, and strains. Therefore results
only atEMail:[email protected]
www.cadfamily.com the grid points are not adequate, especially for most plotting packages,
which use linear interpolation. For data recovery, the p-version elements are divided
into view-grids and view-elements in order to better visualize the results. The
OUTRCV Bulk Data entry defines the number of view-elements in each element
direction and the output coordinate system for the view-grids and view-elements.
The OUTRCV entry is referenced by the OUTRCV Case Control command. The
coordinates of the view-grids may be output using the VUGRID Case Control
command.
Table 25-1 Bulk Data Entries for h-Elements and p-Elements
h-Version Elements p-Version Elements

Geometry GMCURV
GMSURF
Geometrical Associativity FEEDGE
FEFACE
Coordinate Systems CORDij CORDij
GMCORD
Grids GRID GRID
Points POINT
Elements CHEXA CHEXA
CPENTA CPENTA
CTETRA CTETRA
CQUAD CQUAD
CTRIA CTRIA
CBEAM CBEAM
Element Properties PSOLID PSOLID
PSHELL PSHELL
PBEAM PBEAM
Material Properties MATi MATi
Loads GMLOAD
PLOAD4 PLOAD4
FORCE FORCE*
Thermal Loads TEMPF
TEMP, TEMPD TEMP, TEMPD
TEMPP1 TEMPP1
TEMPRB TEMPRB
Gravity Loads GRAV GRAV
CHAPTER 25 763
p-Elements
Table 25-1 Bulk Data Entries for h-Elements and p-Elements (continued)
h-Version Elements p-Version Elements

Constraints GMSPC
SPC, SPC1 SPC, SPC1*
Boundary Conditions GMBC
SPCD SPCD*
Polynomial Values PVAL
PSET
Adaptivity Control ADAPT
Data Recovery OUTPUT
OUTRCV
Miscellaneous DEQATN
TABLE3D
Sets Definition SETS DEFINITION
* Should only be used under specific conditions.
25.4 p-Element Example Problems

This section demonstrates the p-element and p-adaptivity capabilities. Several
examples are included in this section to show how p-elements are used. Most of the
input files of these examples have been annotated and are included in the Test
Problem Library (TLP) directory. The detailed descriptions of the input commands
and the interpretation of the results can be found in the NX Nastran Quick Reference
Guide.
Examples
Each example contains the following parts:
Objectives Outlines the problem objectives.

Keywords Serves as index and show what p-element related Bulk
Data entries being used.
Problem Description Describes the geometry and data of the problem.
Theoretical Results Shows the closed form solution, if available.
Setup Describes what and how to use the Bulk Data entries.
Results Compares the calculated results with theory.
References
Square Plate with a Circular Hole

Objectives
This example computes the stress concentration factor of a square plate with a
circular hole under axial loads. The objective of this problem is to validate the stress
concentration factor for a square plate of 2.73 around the hole.
Problem Description
Plate size = 10 in x 10 in
Thickness = 0.1 in
Hole radius = 2.0 in
Modulus of Elasticity = 1.0E7 psi
Poisson's ratio = 0.3
Figure 25-2 shows the quarter model that requires only two CHEXA solid
p-elements.
Figure 25-2 Geometry of a Square Plate with a Circular Hole

CHAPTER 25 765
p-Elements
A portion of the file which has been annotated is shown in Listing 25-1. The
complete input file, PEV0101.dat, can be found in the TPL directory.
Theoretical Results
Stress concentration factor of 2.73 around the hole as described in Reference [1].
Setup
The Bulk Data entry FEEDGE is used to describe the circular arc of the hole. The
quadratic order is used for demonstration purposes; a cubic order curve can be
easily accommodated with an additional point for each curve. The load and
boundary conditions are also specified using FEFACE.
Since the file has been heavily annotated, the description of the Bulk Data entries is
not repeated here. Please refer either to the following partial list or the complete
input file in the TPL directory for detailed descriptions.
Listing 25-1 Partial Bulk Data Entries
TITLE = PLATE WITH HOLE -- QUARTER MODEL (p-element model; Point option)
SUBTITLE = p-adaptivity; maximum allowable p-value 5X5X3
$
$ p-elements Case Control commands
$
$ Select Bulk Data ADAPT ID for adaptivity control
$
ADAPT = 150
$
$ Request output destination(s) of locations (i.e. coordinates)
$ for p-element data recovery points
$
VUGRID(PLOT,PUNCH)=all
$
$ Request output (STRESS, STRAIN, DISP, PVAL) for p-elements only
$ DATAREC = n, where n refers to a Bulk Data OUTPUT ID
$
DATAREC = 301
$
$ Select output options for p-elements
$ OUTRCV = n, where n defines an OUTRCV ID in Bulk Data
$
OUTRCV = 401
$
$ Select set(s) of elements for adaptivity and stress output
$ SETS DEFINITION must be defined at the end of the Case Control
$ SET 501 defines a set of all elements
$ referenced by ELSETID in ADAPT entry for ERRTOL of 10%
$
SETS DEFINITION
set 501 = all
$
BEGIN BULK
$
$ p-elements Bulk Data

$
$ Adaptivity control
$ PSTRTID specifies the PVAL ID 120 for the starting pvalues p1, p2
and p3
$ for the element SET 501 (all elements)
$ PMINID specifies the PVAL ID 120 for the minimum pvalues p1, p2 and
p3
$ PMAXID specifies the PVAL ID 150 for the maximum pvalues p1, p2 and
p3
$
$ADAPT SID PSTRTID PMINID PMAXID
$ PART ELSETID
$
ADAPT 150 120 120 150
PART=ALL,ELSET=501
$
$ Specify p-levels for element SET 501 (all elements)
$ POLY1, POLY2 and POLY3 correspond to p1, p2 and p3 respectively and
$ can have different p-values
$ p-values specified by PMINID should be = PSTRTID < PMAXID
$ starting p-level p1=p2=p3=2
$ minimum p-level p1=p2=p3=2
$ maximum p-level p1=p2=5 p3=3
$
$PVAL ID POLY1 POLY2 POLY3 SETTYP ID
PVAL 120 2 2 2 SET 501
PVAL 150 5 5 3 SET 501
$
$ Select various output options;
$ Displacements, Stresses, Strains and p-values
$
$ SID
$ ELSET=n, DISP=(PRINT,PLOT,PUNCH),
$ STRESS=(PRINT,PLOT,PUNCH), STRAIN=(PRINT,PLOT,PUNCH),
$ PVAL=(PRINT,PLOT,PUNCH), BY=n
$
OUTPUT 301
ELSET=501, DISP=PRINT, STRESS=PRINT
STRAIN=PLOT, PVAL=PRINT, BY=1
$
$ Define various output options;
$ for Displacements, Stresses and Strains
$
$ SID SETID
$ CID= 0 or n (default 0; basic coordinate system)
$ VIEW=i*j*k (default 3*3*3)
$ For less output, reduce i,j and k to 2 or 1
$
OUTRCV 401 501
VIEW=2*2*2
$
CHEXA 1 1 1 25 28 4 101 125
+ 128 104
CHEXA 2 1 32 4 28 56 132 104
+ 128 156
$
CHAPTER 25 767
p-Elements
$ The following geometry specifications demonstrate the use of

$ different options on specifying actual geometry
$ to achieve more accurate results
$
$ Define FEEDGEs and FEFACEs to specify
$ geometry, loading and constraints
$
$ Use POINT, FEEDGE to describe curves 104, 108, 204 and 208
$
POINT 3 1.7305341.0026220.0
POINT 30 1.0026221.7305340.0
POINT 103 1.7305341.0026221.0
POINT 130 1.0026221.7305341.0
$
$ EDGEID G1 G2 GEOMIN ID1
FEEDGE 104 4 1 POINT 3
FEEDGE 204 4 32 POINT 30
FEEDGE 108 104 101 POINT 103
FEEDGE 208 104 132 POINT 130
$
$ Specify FEFACE for BC and loads
$
$ FACEID G1 G2 G3 G4
FEFACE 1001 1 25 125 101
FEFACE 1002 25 28 128 125
FEFACE 2002 56 156 132 32
$
$ Use GMSPC,GMLOAD and FEFACE to specify BC and loads
$ on geometry instead of element data
$
$ Boundary Conditions
$
$ SID C ENTITY ID
GMSPC 999 2 FEFACE 1001
GMSPC 999 1 FEFACE 2002
$
$ LOADS
$
$ LID N1 N2 N3 ENTITY ID METHOD
GMLOAD 998 1.0 0.0 0.0 FEFACE 1002 CONSTANT
90.
$
ENDDATA
Results
Table 25-2 compares NX Nastran results with the theory, and they agree very well
with only two p-elements. Closer results could have been obtained with more
elements or higher p-values.
Table 25-2 Comparison of NX Nastran and Theoretical Results
Stress Concentration Factor Percent

Theory NX Nastran Difference
2.73 2.77 1.5%
References
[1] Roark, R. J. and Young, W. C. Formulas for Stress and Strain. 5th edition. McGraw-
Hill, Inc., 1975, p. 594.
Spherical Structure Under Internal Pressure

Objectives
This example computes the effects of internal pressure on a spherical structure. The
deflections at the inner and the outer surfaces are compared with the theoretical
results.
Problem Description
Internal pressure (P) = 3000.0 psi
Inner radius (b) = 5.0 in
Outer radius (a) = 10.0 in
Modulus of Elasticity (E) = 1.0E7 psi
Poisson ratio ( ν ) = 0.3
Theoretical Results
The theoretical results on radial deflections can be found in reference 1 and are listed
as follows:
Pb ( 1 – ν ) ( a 3 + 2b 3 )
Inner radial deflection = ------- ------------------------------------------------ + ν
E 2( a3 – b3 )
Pa 3 ( 1 – ν )b 3
Outer radial deflection = ------- ----------------------------
www.cadfamily.com EMail:[email protected] E 2( a3 – b3 )
CHAPTER 25 769
p-Elements
Setup
This example demonstrates how the higher cubic order of the geometry for a
spherical structure can be specified using the EQUATION option of the GMCURV
and GMSURF Bulk Data entries. An octant model is analyzed with four CPENTA
elements. The boundaries of the CPENTA elements are specified along the great
circles. Figure 25-3 shows the geometry and the element boundaries.
Partial Bulk Data entries are given in Listing 25-2. Equations of the inner circle
along the sphere are described using Bulk Data entries GMCURV and DEQATN.
The equations (IDs 1, 2, and 3) are the spatial locations of the curve along the great
circle (circumference of a sphere) expressed in terms of the radius and an angle (U).
The first and second derivatives of the equations are described by DEQATN entries
4 to 9. The first and second derivatives are optional inputs. If they are missing, they
will be calculated by the program.
The GMCURV entry specifies the great circle geometry on the x-y plane with
equations 1 through 9 for the curve ID 1, and ranges from 0 to 90 degrees (0 to 1.57
radians). Similarly, the geometry on the y-z and x-z planes, curves 2 and 3, can be
specified using the same equations with different orders. Note that the labels and
arguments of the equations may be chosen for user convenience.
Figure 25-3 Spherical Structure
Listing 25-2 Partial Bulk Data Entries
$
$ ************** CURVE AND SURFACE GEOMETRY *************
$
$ Inner circle equations

$
$
DEQATN 1 X(U) = 5.0*COS(U)
DEQATN 2 Y(U) = 5.0*SIN(U)
DEQATN 3 Z(U) = 0.0
DEQATN 4 XP(U) = -5.0*SIN(U)
DEQATN 5 YP(U) = 5.0*COS(U)
DEQATN 6 ZP(U) = 0.0
DEQATN 7 X2P(U) = -5.0*COS(U)
DEQATN 8 Y2P(U) = -5.0*SIN(U)
DEQATN 9 Z2P(U) = 0.0
$
$ Inner circle on x-y plane
$
GMCURV 1 MSCGRP1 0 0 +
+ EQUATION 0.0 1.570796327 +
+ 1 2 3 4 5 6 7 8 9
$
$ Inner circle on y-z plane
$
+ EQUATION 0.0 1.570796327 +
+ 3 2 1 6 5 4 9 8 7
$
$ Inner circle on x-z plane
$
+ EQUATION 0.0 1.570796327 +
+ 2 3 1 5 6 4 8 9 7
$
Geometry of the inner surface is described using GMSURF and DEQATN entries.
The equations (IDs 41, 42, and 43) are the spatial locations of the surface using a
spherical coordinate system expressed in terms of radius and two angles (U and V).
The first and second derivatives of the equations are described by DEQATNs 44 to
59. Again, the first and second derivatives are optional inputs.
The GMSURF entry specifies the surface geometry with equations 41 through 59 for
the surface ID 1, and ranges from 0 to 90 degrees in two directions. Besides
describing geometry, the specification of this surface facilitates the application of
internal pressure with the GMLOAD command. The GMLOAD specifies a constant
pressure load of 3000 psi on the surface GMSURF 1. The commands are shown
below.
$
$ Inner sphere
$
DEQATN 41 SX(U,V) = 5.0*SIN(U)*COS(V)
DEQATN 42 SY(U,V) = 5.0*SIN(U)*SIN(V)
DEQATN 43 SZ(U,V) = 5.0*COS(U)
DEQATN 44 SXU(U,V) = 5.0*COS(U)*COS(V)
DEQATN 45 SYU(U,V) = 5.0*COS(U)*SIN(V)
DEQATN 46 SZU(U,V) = -5.0*SIN(U)
DEQATN 47 SXV(U,V) = -5.0*SIN(U)*SIN(V)
CHAPTER 25 771
p-Elements
DEQATN 48 SYV(U,V) = 5.0*SIN(U)*COS(V)

DEQATN 49 SZV(U,V) = 0.
DEQATN 51 SXUU(U,V) = -5.0*SIN(U)*COS(V)
DEQATN 52 SYUU(U,V) = -5.0*SIN(U)*SIN(V)
DEQATN 53 SZUU(U,V) = -5.0*COS(U)
DEQATN 54 SXVV(U,V) = -5.0*SIN(U)*COS(V)
DEQATN 55 SYVV(U,V) = -5.0*SIN(U)*SIN(V)
DEQATN 56 SZVV(U,V) = 0.
DEQATN 57 SXUV(U,V) = -5.0*COS(U)*SIN(V)
DEQATN 58 SYUV(U,V) = 5.0*COS(U)*COS(V)
DEQATN 59 SZUV(U,V) = 0.
$
GMSURF 1 MSCGRP1 0 0 +
+ EQUATION 0.0 1.570796327 0.0 1.570796327 +
+ 41 42 43 44 45 46 47 48 49 +
+ 51 52 53 54 55 56 57 58 59
$
$ THIS SECTION CONTAINS LOADS
$
GMLOAD 1 1 1. 0. 0. GMSURF 1 CONST +
+ 3000.
$
Since only an octant model is analyzed, the symmetric boundary conditions are
required. The following partial commands demonstrate how GMSPC, GMSURF,
and DEQATN work together. The GMSURF 201 is expressed in terms of radius (R)
and an angle (U), ranging from 5 to 10 inches and 0 to 1.57 radians. GMSPC specifies
the constraint in the basic z-direction.
$
$ Surface on x-y plane
$
DEQATN 61 SX(R,U) = R*COS(U)
DEQATN 62 SY(R,U) = R*SIN(U)
DEQATN 63 SZ(R,U) = 0.
DEQATN 64 SXU(R,U) = COS(U)
DEQATN 65 SYU(R,U) = SIN(U)
DEQATN 66 SZU(R,U) = 0.
DEQATN 67 SXV(R,U) = -R*SIN(U)
DEQATN 68 SYV(R,U) = R*COS(U)
DEQATN 69 SZV(R,U) = 0.
DEQATN 71 SXUU(R,U) = 0.
DEQATN 72 SYUU(R,U) = 0.
DEQATN 73 SZUU(R,U) = 0.
DEQATN 74 SXVV(R,U) = -R*COS(U)
DEQATN 75 SYVV(R,U) = -R*SIN(U)
DEQATN 76 SZVV(R,U) = 0.
DEQATN 77 SXUV(R,U) = - SIN(U)
DEQATN 78 SYUV(R,U) = COS(U)
DEQATN 79 SZUV(R,U) = 0.
$
GMSURF 201 MSCGRP1 0 0 +
+ EQUATION 5.0 10.0 0.0 1.570796327 +
+ 61 62 63 64 65 66 67 68 69 +
+ 71 72 73 74 75 76 77 78 79
$
$ x-y plane in xyz basic

$
GMSPC 1 3 GMSURF 201
$
Other inputs, such as FEEDGE and FEFACE, are used to specify the geometry, the
element boundaries, loads, and boundary conditions. For example, the following
FEEDGE command is used to specify that the element edge between grids 3 and 4
will follow the geometry of GMCURV ID 1. The complete input file has been
annotated and stored in the TPL directory for review.
$
$ Inner radius x-y plane
$
FEEDGE 1 3 4 0 GMCURV 1
$
Results
The p-element model requires two adaptivity cycles to converge to the desired error
tolerance of 10% at p-value of4. Table 25-3 tabulates the results with the theory.
Table 25-3 Comparison of Displacements
Results Percent
Deflections Theory NX Nastran Difference
Inner 1.20E-03 1.19E-03 1.0%

Outer 4.50E-04 4.45E-04 1.1%
References
[1] Roark, R. J. and Young, W. C. Formulas for Stress and Strain. 5th edition. McGraw-
Hill,Inc., 1975, p. 506.
Motor Blade – Normal Modes

Objectives
A normal mode analysis of a simplified motor blade is analyzed and compared using
the h- and p-elements. The natural frequencies between 0 and 2000 Hz are computed.
CHAPTER 25 773
p-Elements
Problem Description
Figure 25-4 shows the geometry of the motor blade consisting of a base and a blade.
The whole assembly part is made of aluminum.
Modulus of Elasticity = 1.0E7 psi

Poisson's ratio = 0.3
3
Mass density = 2.59E-4 lb/in
Figure 25-4 Motor Blade – p-Element Model
Figure 25-5 Motor Blade – Fine h-Model
Theoretical Results
There is no theoretical result. Results of three different meshes of h-elements—
coarse, medium, and fine—are compared with a p-element model.
Setup
Figure 25-4 and Figure 25-5 show the p-model and the h-model with fine mesh
respectively. Since only the natural frequencies are requested, the geometry of both
the h- and p-meshes are simplified and the boundaries are modeled with straight
edges. If a detailed stress analysis is required, the geometry should be specified more
precisely with higher-order geometry.
Results
All models produce six rigid body modes, so they are not discussed here. Table 25-4
tabulates the computed elastic natural frequencies between 0.0 to 2000.0 Hz, the
number of elements, and the number of degrees-of-freedom with respect to different
models. The p-element results agree very well with the fine meshed h-model
showing the power and usefulness of the p-elements.
CHAPTER 25 775
p-Elements
Table 25-4 Comparison of Natural Frequencies of a Motor Blade
Descriptions No. of No. of Elastic Modes

Mesh Elements DOFs 1st 2nd 3rd 4th
h-model Coarse 4 90 622.5 651.2 1204.9 –
Medium 80 441 604.5 954.7 1289.6 1956.2
Fine 1250 4878 659.7 1051.5 1404.8 1744.7
p-adaptive cycle=1 10 276 737.7 1064.8 1526.9 –
model cycle=2 1161 675.8 1070.8 1426.9 1796.6
cycle=3 2112 674.6 1071.2 1427.8 1765.2
Torsion of Elliptical Bar

Objectives and Problem Description
This example illustrates torsion of an elliptical bar.
2 in.
4 in.
10 in.
y
Material:
aluminum
z E = 1.0E7 lb/in 2
x ν = 0.3
Figure 25-6 Octant of the Elliptical Bar Model
Setup
The model is constructed from four CPENTA elements. There are two coordinate
systems used: the cylindrical coordinate system (CORD2C entry) to define grids of
the model, and the convective coordinate system (GMCORD entry) to define loads
applied to the bar.
The elliptical cross section of the model is described using GMCURV and GMSURF
entries, both with the equation method. For the GMCURV entry:
The general ellipse equation:
x2 y2
----- + ----- = 1
a2 b2
as shown in Figure 25-7.
Define x and y in polar coordinates:
a
x = r cos θ x = ------------------------------------------------ cos θ
y = r sin θ a2
cos 2 θ + ----- sin 2 θ
ab b2
r = ------------------------------------------------------
a
a 2 sin 2 θ + b 2 cos 2 θ y = ------------------------------------------------ sin θ
a2
a
= ------------------------------------------------ cos 2 θ + ----- sin 2 θ
b2
a2
cos 2 θ + ----- sin 2 θ
b2
x2 y2
----- + ----- = 1
a2 b2
R(θ)
b
θ
1
a
Figure 25-7
For the GMSURF entry, three different types of parameterization were used. In the
first method,
v π
x s = ------------------------------------------------ cos u 0.0 ≤ u ≤ ---
a2 2
cos 2 u + ----- sin 2 u
b2
v
y s = ------------------------------------------------ sin u 0.0 ≤ v ≤ a
a2
cos 2 u + ----- sin 2 u
b2
as shown in Figure 25-8.
In the second method,
CHAPTER 25 777
p-Elements
π
x s = av cos u 0.0 ≤ u ≤ ---
2
y s = bv sin u 0.0 ≤ v ≤ 1.0
b
v ---
a
x2 y2
----- + ----- = 1
a2 b2
R ( u,v )
v
Figure 25-8
The third method, using hyperbolic functions,
x + iy = e ( cosh r cos t + i sinh r sin t )

= e cosh ( r + it )
x = e cosh r cos t 0.0 < r < R

π
y = e sinh r sin t 0.0 < t < ---
2
b
e = a2 – b2 R = tanh – 1  ---
 a
as shown in Figure 25-9, with contours of t displayed.
π
t = ---
2
r = R
t = 0.0
e
a
Figure 25-9
For the GMSURF entry, the order of the input IDs of DEQATN entries for the surface
is important. First, we input the IDs of DEQATN entries providing equations for the
x, y, and z coordinates of the surface. Then, the IDs of DEQATN entries of the first
derivatives of x, y, and z functions with respect to the surface parameter u . Then, the
IDs of DEQATN entries of the first derivatives of x, y, and z functions with respect to
the surface parameter v . Then, the IDs of DEQATN entries of the second derivatives
of x, y, and z functions with respect to the surface parameter u . Then, the IDs of
DEQATN entries of the second derivatives of x, y, and z functions with respect to the
surface parameter v . Finally, the IDs of DEQATN of the mixed second derivatives of
the x, y, and z functions with respect to both surface parameters u and v .
The torsion in this example is defined by applying an edge load (GMLOAD entry)
using the convective coordinate system (GMCORD entry). The load is applied
tangent to the GMCURV 3. This is illustrated in Figure 25-10. Note that an edge load
on solid elements can cause singularities, but it is used here for demonstration
purposes.
z
y
x
Figure 25-10
The input file of this model is listed in Listing 25-3.
Listing 25-3 Input File for the Model
ID TORSION, STATIC
TIME 500
SOL 101 $ STATIC ANALYSIS
CEND
TITLE = TORSION, ELLIPSE CSA
SUBTITLE = METHOD 2
$
$ DEFINE ANALYSIS ENTRY
$
ADAPT = 100
$
LOAD = 1
SPC = 1
$
$ DEFINE OUTPUT REQUEST OPTION
$
DATAREC = 100
OUTRCV = 200
VUGRID (PUNCH) = ALL
www.cadfamily.com$ EMail:[email protected]
CHAPTER 25 779
p-Elements
$ DEFINE THE SETS OF ELEMENTS REQUIRED FOR THE ADAPTIVITY ANALYSIS

$ THIS IS REQUIRED WHEN SETTYP=SET IN PVAL ENTRY
$ (REFER TO PVAL ENTRY ALSO)
$
SETS DEFINITION
SET 120 = ALL
$
BEGIN BULK
PARAM VUPENTA CPENTA
$
$ BOTH THE ADAPT AND PVAL ENTRY ARE USED TO
$ DEFINE ADAPTIVITY OPTION ENTRY FOR ALL ELEMENTS
$
$ FOR THE ADAPT ENTRY:
$ ID = ADAPT ENTRY ID SELECTED IN CASE CONTROL
$ ADGEN = ID OF THE FIRST PVAL ENTRY GENERATED IN THE ADAPTIVE PROCESS;
$ MUST BE LARGER THAN PSTRID,PMINID, AND PMAXID
$ MAXITER = NUMBER OF ITERATIONS REQUESTED BEFORE THE ADAPTIVE PROCESS
STOPPED
$ PSTRTID = ID OF THE STARTING PVAL ENTRY
$ PMINID = ID OF THE MINIMUM PVAL ENTRY
$ PMAXID = ID OF THE MAXIMUM PVAL ENTRY
$ PART = ANY NAME CHOSEN FOR THE ELEMENTS;
$ USED FOR ERROR ESTIMATOR REFERENCE
$ ELSETID = ID OF SETS OF ELEMENTS REQUESTED FOR ADAPTIVE PROCESS
$ TYPE = EBEP : THE P-ORDER IS INCREASED ELEMENTS-BY-ELEMENT
$ ERREST = ERROR ESTIMATOR METHOD USED FOR ADAPTIVE PROCESS
$ ERRTOL = ERROR TOLERANCE (IN FRACTION), REQUIRED IF MAXITER IS NOT
DEFINED
$ SIGTOL = STRESS TOLERANCE
$
$ SID ADGEN MAXITER PSTRID PMINID PMAXID
ADAPT 100 1001 10 10 20 30
+000001
++000001PART=ALL,ELSET=120,TYPE=EBEP,ERREST=1,ERRTOL=0.02,SIGTOL=5.0
+000002
$
$ DEFINE THE POLYNOMIAL ORDER DISTRIBUTION:
$ PVAL ID = ID OF PVAL DEFINED IN THE ADAPT ENTRY ABOVE
$ PX = THE POLYNOMIAL ORDER DISTRIBUTION IN X-DIRECTION
$ PY = THE POLYNOMIAL ORDER DISTRIBUTION IN Y-DIRECTION
$ PZ = THE POLYNOMIAL ORDER DISTRIBUTION IN Z-DIRECTION
$ SETTYP = TYPE OF SET PROVIDED. IT CAN BE EITHER SET OR ELID
$ ID = SET ID OR ELEMENT ID, WHICHEVER IS APPROPRIATE
$
$ ID Px Py Pz SETTYP ID
PVAL 10 2 2 2 SET 120
PVAL 20 2 2 2 SET 120
PVAL 30 4 4 4 SET 120
$
$ DEFINE THE OUTPUT REQUEST OPTION
$ REQUEST DISPLACEMENT AND STRESS OUTPUT FOR SET 120 FOR THE LAST
$ ADAPTIVITY ANALYSIS
$
OUTPUT 100
+000003
++000003ELSET=120,DISP=PRINT,STRESS=PRINT,STRAIN=NONE
+000004
++000004FIRST=NO,LAST=YES,BY=2
+000005
$
$ THE OUTPUT COORDINATE SYSTEM IS THE BASIC COORDINATE SYSTEM (CID=0)
$
OUTRCV 200 120 +000003
++000003CID=0,VIEW=2*2*2
+000004
$
$ DEFINE THE CYLINDRICAL COORDINATE SYSTEM FOR
$ GRID POINTS
$
CORD2C 1 0 0.0 0.0 0.0 0.0 0.0 1.0
+000005
++0000051.0 0.0 0.0
+000006
$
$ DEFINE THE CONVECTIVE COORDINATE SYSTEM FOR THE APPLIED LOAD
$ IF GMCURV IS USED AS THE ENTITY TO DEFINE THE COORDINATE SYSTEM,
$ TWO ID'S ARE REQUIRED. THE FIRST ID IS THE GMCURV ID AND THE SECOND
$ ONE IS THE CORRESPONDING GMSURF ID
$
$ CID ENTITY ID 1 ID 2
GMCORD 10 GMCURV 3 2
$
$ DEFINE PENTA ELEMENTS
$
CPENTA 1 1 12 1 2 11 4 5
CPENTA 2 1 12 2 3 11 5 6
CPENTA 3 1 11 4 5 10 7 8
CPENTA 4 1 11 5 6 10 8 9
$
$ DEFINE GRID POINTS
$
GRID 1 1 4. 0.0 0.0
GRID 2 1 2.52982245. 0.
GRID 3 1 2.0 90.0 0.0
GRID 4 1 4.0 0.0 5.0
GRID 5 1 2.52982245.0 5.0
GRID 6 1 2.0 90.0 5.0
GRID 7 1 4.0 0.0 10.0
GRID 8 1 2.52982245.0 10.0
GRID 9 1 2.0 90.0 10.0
GRID 10 0.0 0.0 10.0
GRID 11 0.0 0.0 5.0
GRID 12 0.0 0.0 0.0
$
$ DEFINE FEEDGES
$ FEEDGES BY DEFAULT ARE STRAIGHT LINES
$ GMCURVS ARE REQUIRED FOR FEEDGES THAT DEFINE CURVED EDGES
$
$ FEEDGE 18 AND FEEDGE 19 MAKE GMCURV 1
$
CHAPTER 25 781
p-Elements
$ ID GRID 1 GRID 2 CID GEOMIN ID

FEEDGE 1 12 1 0
FEEDGE 2 12 2 0
FEEDGE 3 12 3 0
FEEDGE 4 11 4 0
FEEDGE 5 11 5 0
FEEDGE 6 11 6 0
FEEDGE 7 10 7 0
FEEDGE 8 10 8 0
FEEDGE 9 10 9 0
FEEDGE 10 12 11 0
FEEDGE 11 1 4 0
FEEDGE 12 2 5 0
FEEDGE 13 3 6 0
FEEDGE 14 11 10 0
FEEDGE 15 4 7 0
FEEDGE 16 5 8 0
FEEDGE 17 6 9 0
$
$ DEFINE CURVE GEOMETRY USING EQUATION METHOD
$
DEQATN 1 X(U) = 4.0*COS(U)/(SQRT(1.0+3.0*SIN(U)*SIN(U)))
DEQATN 2 Y(U) = 4.0*SIN(U)/(SQRT(1.0+3.0*SIN(U)*SIN(U)))
DEQATN 3 Z(U) = 0.0
DEQATN 4 Z1(U) = 5.0
DEQATN 5 Z2(U) = 10.0
DEQATN 6 XP(U) = -16.0*SIN(U)/((1.0+3.0*SIN(U)*SIN(U))**1.5)
DEQATN 7 YP(U) = 4.0*COS(U)/((1.0+3.0*SIN(U)*SIN(U))**1.5)
DEQATN 8 ZP(U) = 0.0
DEQATN 9 XPP(U) =(-16.0*COS(U)+96.0*SIN(U)*SIN(U)*COS(U))/
+000009
++000009 (1.0+3.0*SIN(U)*SIN(U))**2.5
+000010
DEQATN 10 YPP(U) =(-4.0*SIN(U)*(10.0-6.0*SIN(U)*SIN(U)))/
+000011
++000011 ((1.0+3.0*SIN(U)*SIN(U))**2.5)
+000012
DEQATN 11 ZPP(U) =0.0
$
$
GMCURV 1 MSCGRP1 0 0
+000013
++000013EQUATION,0.0,1.5708,1,2,3,6,7,
+000014
++0000148 9 10 11
+000015
+000016
++000016EQUATION,0.0,1.5708,1,2,4,6,7,
+000017
++0000178 9 10 11 +000018
+000019
++000019EQUATION,0.0,1.5708,1,2,5,6,7,
+000020
++0000208 9 10 11 +000021
$
$ DEFINE FEFACES
$
$ ID GRID 1 GRID 2 GRID 3 GRID 4 CID SURFID
FEFACE 1 12 1 2 0 1
FEFACE 2 12 2 3 0 1
FEFACE 3 11 4 5 0
FEFACE 4 11 5 6 0
FEFACE 5 10 7 8 0 2
FEFACE 6 10 8 9 0 2
FEFACE 7 12 1 4 11 0
FEFACE 8 12 3 6 11 0
FEFACE 9 11 4 7 10 0
FEFACE 10 11 6 9 10 0
$
$ DEFINE SURFACE GEOMETRY USING EQUATION METHOD
$
DEQATN 13 X(U,V) = 4.0*V*COS(U)
DEQATN 14 Y(U,V) = 2.0*V*SIN(U)
DEQATN 15 Z1(U,V) = 0.0
DEQATN 16 Z2(U,V) = 10.0
DEQATN 17 XU(U,V) = -4.0*V*SIN(U)
DEQATN 18 YU(U,V) = 2.0*V*COS(U)
DEQATN 19 ZU(U,V) = 0.0
DEQATN 20 XV(U,V) = 4.0*COS(U)
DEQATN 21 YV(U,V) = 2.0*SIN(U)
DEQATN 22 ZV(U,V) = 0.0
DEQATN 23 XUU(U,V)= -4.0*V*COS(U)
DEQATN 24 YUU(U,V)= -2.0*V*SIN(U)
DEQATN 25 ZUU(U,V)= 0.0
DEQATN 26 XVV(U,V)= 0.0
DEQATN 27 YVV(U,V)= 0.0
DEQATN 28 ZVV(U,V)= 0.0
DEQATN 29 XUV(U,V)= -4.0*SIN(U)
DEQATN 30 YUV(U,V)= 2.0*COS(U)
DEQATN 31 ZUV(U,V)= 0.0
$
$
GMSURF 1 MSCGRP1 0 0
+000022
++000022EQUATION,0.0,1.5708,0.0,1.0,13,14,15,
+000023
++00002317,18,19,20,21,22,23,24,
+000024
++00002425,26,27,28,29,30,31
+000025
GMSURF 2 MSCGRP1 0 0
+000026
++000026EQUATION,0.0,1.5708,0.0,1.0,13,14,16,
+000027
CHAPTER 25 783
p-Elements
++00002717,18,19,20,21,22,23,24,
+000028
++00002825,26,27,28,29,30,31
+000029
$ DEFINE CONSTRAINT IN THE X-DIRECTION
$ SID C ENTITY ID
GMSPC 1 1 FEFACE 1
GMSPC 1 1 FEFACE 2
GMSPC 1 1 FEFACE 7
GMSPC 1 1 FEFACE 9
$
$ DEFINE CONSTRAINT IN THE Y-DIRECTION
$
$ SID C ENTITY ID
GMSPC 1 2 FEFACE 1
GMSPC 1 2 FEFACE 2
GMSPC 1 2 FEFACE 8
GMSPC 1 2 FEFACE 10
$
$ DEFINE CONSTRAINT IN THE Z-DIRECTION
$
$ SID C ENTITY ID
GMSPC 1 3 FEFACE 7
GMSPC 1 3 FEFACE 8
GMSPC 1 3 FEFACE 9
GMSPC 1 3 FEFACE 10
$
$ DEFINE THE MATERIAL PROPERTY (ALUMINUM)
$
MAT1 1 1.+7 0.3
PSOLID 1 1 GRID SMECH
$
$ CONSTANT LOAD OF 100 LB/INCH IS APPLIED TANGENTIALLY TO
$ GMCURV 3
$
$ LID = LOAD IDENTIFICATION NUMBER DEFINED IN CASE CONTROL SECTION
$ CID = COORDINATE SYSTEM USED TO DEFINE THE APPLIED LOAD
$ N1,N2,N3 = DIRECTION OF THE LOAD VECTOR
$ ENTITY = ENTITY AT WHICH LOAD IS APPLIED TO
$ ID = ID OF THE ENTITY
$ METHOD = METHOD USED TO DEFINE THE LOAD
$
$ LID CID N1 N2 N3 ENTITY ID METHOD
GMLOAD 1 10 1.0 0.0 0.0 GMCURV 3 CONSTANT
+000007
$
$ VALUE OF LOAD DENSITY
++000007100.0
+000008
$
ENDDATA
Results
The OUTPUT entry requests the displacements and stress output solely for the last
adaptivity iteration. An abridged output for this model is shown in Figure 25-11.
1 TORSION, ELLIPSE CSA JULY 23, 2003 NX NASTRAN 07/23/03 PAGE 73

METHOD 2
0*** USER INFORMATION MESSAGE 5293 FOR DATA BLOCK KLL
1 5.2309875E-15 1.6239751E-02
*** USER INFORMATION MESSAGE 6502 (ADAPT)
THE CURRENT SUPERELEMENT REQUIRES NO FURTHUR P-ADAPTIVE ANALYSIS BECAUSE THERE ARE NO CHANGES IN P-VALUES.
0*** USER WARNING MESSAGE 4594. THE POSTCDB CONTAINS NO LEGITIMATE SURFACE OR VOLUME DEFINITIONS.
METHOD 2 ADAPTIVITY INDEX= 3
0

1 G .0 .0 .0 .0 .0 .0
2 G .0 .0 -8.692895E-07 .0 .0 .0
3 G .0 .0 .0 .0 .0 .0
4 G .0 1.357096E-05 .0 .0 .0 .0
5 G -3.062882E-06 -2.173771E-06 -1.839771E-06 .0 .0 .0
6 G 1.593498E-05 .0 .0 .0 .0 .0
7 G .0 1.513064E-04 .0 .0 .0 .0
8 G 1.887783E-06 -4.328616E-05 1.990558E-05 .0 .0 .0
9 G 1.121671E-04 .0 .0 .0 .0 .0
10 G .0 .0 .0 .0 .0 .0
11 G .0 .0 .0 .0 .0 .0
12 G .0 .0 .0 .0 .0 .0
201001001 G .0 .0 -6.008400E-23 .0 .0 .0
201001002 G .0 .0 -3.919389E-23 .0 .0 .0
201001003 G .0 .0 -1.764301E-22 .0 .0 .0
201001004 G .0 .0 -4.865105E-07 .0 .0 .0
201001005 G .0 .0 -8.088551E-07 .0 .0 .0
201001006 G .0 .0 -8.692895E-07 .0 .0 .0
201001007 G -3.961429E-23 1.151926E-22 6.248783E-24 .0 .0 .0
201001008 G -1.317319E-22 2.320618E-06 -1.101103E-22 .0 .0 .0
201001009 G -1.197587E-22 3.720583E-06 -2.934319E-24 .0 .0 .0
201001010 G -7.882488E-07 1.256443E-06 -2.743236E-07 .0 .0 .0
201001011 G -1.074418E-06 2.716110E-06 -1.095060E-06 .0 .0 .0
201001012 G -1.967381E-06 1.468333E-06 -7.467912E-07 .0 .0 .0
201001013 G -9.475001E-23 4.980112E-22 -3.150599E-23 .0 .0 .0
201001014 G -3.235471E-22 3.547284E-06 -1.786648E-22 .0 .0 .0
201001015 G -4.842875E-22 1.357096E-05 6.322074E-23 .0 .0 .0
201001016 G -8.565316E-07 4.229675E-06 -7.164187E-07 .0 .0 .0
201001017 G -3.000400E-06 4.619390E-06 -5.842737E-07 .0 .0 .0
201001018 G -3.062882E-06 -2.173771E-06 -1.839771E-06 .0 .0 .0
201002001 G .0 .0 3.004200E-23 .0 .0 .0
201002002 G .0 .0 -4.865105E-07 .0 .0 .0
201002003 G .0 .0 -8.692895E-07 .0 .0 .0
201002004 G .0 .0 2.921973E-23 .0 .0 .0
201002005 G .0 .0 5.445964E-07 .0 .0 .0
201002006 G .0 .0 4.219457E-22 .0 .0 .0
201002007 G -1.446019E-22 -5.887475E-23 -3.124391E-24 .0 .0 .0
201002008 G -7.882488E-07 1.256443E-06 -2.743236E-07 .0 .0 .0
201002009 G -1.967381E-06 1.468333E-06 -7.467912E-07 .0 .0 .0
201002010 G -8.927043E-07 3.961755E-23 -6.052314E-24 .0 .0 .0
201002011 G -2.660616E-06 7.911209E-07 5.446515E-08 .0 .0 .0
201002012 G -7.240385E-08 1.330381E-22 -1.194728E-22 .0 .0 .0
201002013 G -8.723581E-22 -3.184499E-22 1.575299E-23 .0 .0 .0
201002014 G -8.565316E-07 4.229675E-06 -7.164187E-07 .0 .0 .0
201002015 G -3.062882E-06 -2.173771E-06 -1.839771E-06 .0 .0 .0
201002016 G -1.335821E-06 1.364713E-22 2.932544E-23 .0 .0 .0
201002017 G -7.250181E-06 1.251031E-06 1.185184E-06 .0 .0 .0
201002018 G 1.593498E-05 2.834062E-22 4.533070E-22 .0 .0 .0
201003001 G -9.475001E-23 4.980112E-22 -3.150599E-23 .0 .0 .0
201003002 G -3.235471E-22 3.547284E-06 -1.786648E-22 .0 .0 .0
Figure 25-11 Abridged Output for the Model

CHAPTER 25 785
p-Elements

0 PVAL ID= 1002 (P1,P2,P3= 3 2 3), SUBCASE 1
S T R E S S E S I N P - V E R S I O N P E N T A H E D R O N S O L I D E L E M E N T S ( P E N T A )
VIEW-ELEMENT ID=100001001, P-ELEMENT ID = 1, OUTPUT COORD. ID= 0, P OF EDGES = 3 3 4 4 3 4 4 4 4
VUGRID SIGMA-X/1 SIGMA-Y/2 SIGMA-Z/3 TAU-XY TAU-YZ TAU-XZ MEAN-PRESSURE VON-MISES
201001001 2.507561E-15 2.507561E-15 5.850975E-15 -2.450845E-31 -3.490772E+00 3.045360E-16
3.490772E+00 -3.490772E+00 2.881587E-15 -4.069153E-15 6.046195E+00
201001002 -9.512720E-17 -6.704689E-16 3.533783E-16 7.719806E-32 7.080933E-01 -7.649862E-16
7.080933E-01 -7.080933E-01 -2.876032E-16 2.068900E-16 1.226454E+00
201001004 5.034403E+00 5.034403E+00 1.174694E+01 4.981801E-15 -1.109684E+01 -4.317494E+00
2.076182E+01 -3.980472E+00 5.034403E+00 -7.271916E+00 2.168871E+01
201001007 1.006225E+01 2.347857E+01 1.006225E+01 -2.168649E+00 3.630430E-01 -1.325106E-16
2.382975E+01 9.711069E+00 1.006225E+01 -1.453436E+01 1.394641E+01
201001008 -3.267787E+00 -7.624836E+00 -3.267787E+00 -1.073696E+00 -6.464741E-01 -1.551473E-15
-2.933002E+00 -7.959620E+00 -3.267787E+00 4.720137E+00 4.867867E+00
201001010 -1.527811E+00 -2.566048E+00 -9.243130E-01 -5.258331E-02 -3.754325E-01 -2.206367E+00
1.018136E+00 -3.538538E+00 -2.497769E+00 1.672724E+00 4.135702E+00
201001002 -9.512720E-17 -6.704689E-16 3.533783E-16 7.719806E-32 7.080933E-01 -7.649862E-16
7.080933E-01 -7.080933E-01 -2.876032E-16 2.068900E-16 1.226454E+00
201001003 2.474018E-14 5.246847E-14 2.999879E-14 -1.088513E-15 2.914146E+00 -2.396744E-15
2.914146E+00 -2.914146E+00 2.316292E-14 -3.525450E-14 5.047449E+00
201001005 -7.025239E-01 -7.025239E-01 -1.639222E+00 3.874331E-15 1.205412E+01 -6.993091E+00
1.277274E+01 -1.511449E+01 -7.025239E-01 1.014757E+00 2.415559E+01
201001008 -3.267787E+00 -7.624836E+00 -3.267787E+00 -1.073696E+00 -6.464741E-01 -1.551473E-15
-2.933002E+00 -7.959620E+00 -3.267787E+00 4.720137E+00 4.867867E+00
201001009 6.025518E+00 1.405954E+01 6.025518E+00 7.571502E-01 6.719486E+00 7.331966E-16
1.790772E+01 2.177344E+00 6.025518E+00 -8.703526E+00 1.420281E+01
201001011 2.021004E+00 -5.901449E+00 -1.849734E+00 2.802238E+00 2.882026E+00 -1.129690E+00
2.922884E+00 -8.277588E+00 -3.754748E-01 1.910060E+00 9.969281E+00
0 PVAL ID= 1002 (P1,P2,P3= 3 2 3), SUBCASE 1
201001005 -7.025239E-01 -7.025239E-01 -1.639222E+00 3.874331E-15 1.205412E+01 -6.993091E+00
1.277274E+01 -1.511449E+01 -7.025239E-01 1.014757E+00 2.415559E+01
201001004 5.034403E+00 5.034403E+00 1.174694E+01 4.981801E-15 -1.109684E+01 -4.317494E+00
2.076182E+01 -3.980472E+00 5.034403E+00 -7.271916E+00 2.168871E+01
201001002 -9.512720E-17 -6.704689E-16 3.533783E-16 7.719806E-32 7.080933E-01 -7.649862E-16
7.080933E-01 -7.080933E-01 -2.876032E-16 2.068900E-16 1.226454E+00
201001011 2.021004E+00 -5.901449E+00 -1.849734E+00 2.802238E+00 2.882026E+00 -1.129690E+00
2.922884E+00 -8.277588E+00 -3.754748E-01 1.910060E+00 9.969281E+00
201001010 -1.527811E+00 -2.566048E+00 -9.243130E-01 -5.258331E-02 -3.754325E-01 -2.206367E+00
1.018136E+00 -3.538538E+00 -2.497769E+00 1.672724E+00 4.135702E+00
201001008 -3.267787E+00 -7.624836E+00 -3.267787E+00 -1.073696E+00 -6.464741E-01 -1.551473E-15
-2.933002E+00 -7.959620E+00 -3.267787E+00 4.720137E+00 4.867867E+00
201001004 5.034403E+00 5.034403E+00 1.174694E+01 4.981801E-15 -1.109684E+01 -4.317494E+00
2.076182E+01 -3.980472E+00 5.034403E+00 -7.271916E+00 2.168871E+01
201001005 -7.025239E-01 -7.025239E-01 -1.639222E+00 3.874331E-15 1.205412E+01 -6.993091E+00
1.277274E+01 -1.511449E+01 -7.025239E-01 1.014757E+00 2.415559E+01
201001006 -3.200166E-02 -3.200166E-02 -7.467054E-02 -8.733983E-14 1.493832E+01 -4.544772E+00
1.556104E+01 -1.566771E+01 -3.200166E-02 4.622462E-02 2.704490E+01
201001010 -1.527811E+00 -2.566048E+00 -9.243130E-01 -5.258331E-02 -3.754325E-01 -2.206367E+00
1.018136E+00 -3.538538E+00 -2.497769E+00 1.672724E+00 4.135702E+00
201001011 2.021004E+00 -5.901449E+00 -1.849734E+00 2.802238E+00 2.882026E+00 -1.129690E+00
2.922884E+00 -8.277588E+00 -3.754748E-01 1.910060E+00 9.969281E+00
201001012 2.319424E+00 -3.111391E+00 -1.438776E+00 -7.330344E+00 8.622318E+00 -9.914967E+00
1.681246E+01 -1.124202E+01 -7.801181E+00 7.435812E-01 2.650212E+01
Figure 25-11 Abridged Output for the Model (continued)
Circular Solid Model

Objectives and Problem Description
This example is a circular solid with a simply supported edge. It is made of
aluminum and has the following properties:
E = 1.0E7 lb/in 2
ρ w = 0.098 lbf/in 3
ν = 0.3
It is illustrated in Figure 25-12.
0.5 in.
z
y 5 in.
11 10
3 2
12 9
4 1
Top View
5 8
y 13 16
6 7
x 14 15
z
element 1 thru 8: CPENTA elements

element 9 thru 16: CHEXA elements (boundary elements)
Figure 25-12 The Circular Solid Model
Theoretical Results
The behavior of this circular solid model can be compared to the theoretical circular
plate model described in Reference [1]. From Reference [1],
CHAPTER 25 787
p-Elements
K n Dg
f = ------ ---------
2π wr 4
where:
Et 3
D = -------------------------
2
-
12 ( 1 – ν )
w = uniform load per unit area including own weight

( ρ ) ( Volume )
= ------------------------------------
w
Area
( ρ ) ( Area ) ( t )
= -----------------------------------
w
-
Area
= ( ρm ) ( g ) ( t )
Kn Et 2
f = ------ ---------------------------------------------------- cycles/sec
2π 12 ( 1 – ν 2 ) ( ρ ) ( r 4 )
m
where:
E = modulus of elasticity
K n = constant, n refers to the mode of vibration
t = thickness of the plate

r = radius of the plate
ν = Poisson's ratio
For aluminum,
E = 1.0E7 lb/in2
ν = 0.3
ρ w = 0.098 lbf/in3
ρ m = 2.538E-4 lbm/in3
For the circular plate, radius ( r ) = 5 inches and thickness ( t ) = 0.5 inches. The
theoretical natural frequencies for all modes:
Kn ( 10 7 ) ( 0.5 2 )
f = ------ ----------------------------------------------------------------------------------- = 191.2 K n cycles/sec
2π 12 ( 1 – 0.3 2 ) ( 2.538 × 10 – 4 ) ( 5 4 )
The theoretical natural frequencies for the first several modes are calculated and
tabulated in Table 25-5.
Table 25-5 Theoretical Natural Frequency Values for Circular Plate
Natural Frequency
Mode Kn
(cycles/sec)
Fundamental 4.99 954.1
One Nodal Diameter 13.9 2657.7
Two Nodal Diameters 25.7 4913.9
One Nodal Circle 29.8 5697.8
Setup
The model is constructed from 8 CPENTA elements and 8 CHEXA elements. The
CHEXA elements are the boundary elements. The circular edge of the model is
described using the circle equation (GMCURV entry, method=equation). Since the
GMCURV can not be a closed curve, the full circle must be described with multiple
GMCURV entries. In this model, one full circle is defined by two GMCURV entries.
The first GMCURV entry has an input range from 0.0 to 3.1416 radians, and the
second GMCURV entry has an input range from 3.1415 to 6.2832 radians. Note that
there is some overlap, so that both segments include π . In addition, since the circle
equation is a straightforward equation, the first and the second derivatives of the
equation are not necessarily required. In this case, the derivatives are then computed
numerically.
For the boundary conditions, GMSPC entries are used. Due to the hierarchy order
which places grids as the highest priority, GMSPC entries that are imposed on grid 4
and 8 in the x-direction will override GMSPC entries on the adjacent FEFACES in the
z-direction. Thus, it is necessary to add two more GMSPC entries for grids 4 and 8 in
the z-direction to restore the constraints in the z-direction.
For the material property, the MAT1 entry is used. In the MAT1 entry, the weight
density term ρ w must be converted to the mass density ρ m for consistency of units.
Therefore,
ρ m = ( 0.098 lbf/in 3 ) ( 1 ⁄ 32.174 ft/s 2 ) ( 1 ⁄ 12 in/ft )
= 2.538E-4 lbm/in 3
The input file for this model is shown in Listing 25-4.
CHAPTER 25 789
p-Elements
Listing 25-4 Input File for the Circular Solid Model
ID NORMAL, MODES
TIME 200
SOL 103 $ NORMAL MODE ANALYSIS
CEND
TITLE = CIRCULAR PLATE
SUBTITLE = EDGE SIMPLY SUPPORTED
METHOD = 100
$
$ DEFINE ADAPTIVITY ANALYSIS ENTRY
$
ADAPT = 1
$
$ OUTPUT REQUEST
$
DISPLACEMENT = ALL
ECHO = BOTH
$
SPC =1
$
$ REQUEST FOR VIEW COORDINATES OF THE P-ELEMENTS TO BE
$ PRINTED IN THE PUNCH FILE. THE VIEW COORDINATES
$ ARE THE POINTS AT WHICH THE DISPLACEMENTS ARE
$ CALCULATED.
$
VUGRID(PUNCH) = ALL
$
$ DEFINE THE SETS OF ELEMENTS REQUIRED FOR THE ADAPTIVITY
$ ANALYSIS. THIS IS REQUIRED WHEN SETTYP=SET IN PVAL ENTRY
$ ( REFER TO PVAL ENTRY ALSO )
$
SETS DEFINITION
SET 10 = ALL
$
BEGIN BULK
PARAM POST 0
$
PARAM VUHEXA CHEXA
PARAM VUPENTA CPENTA
$
$ BOTH THE ADAPT AND PVAL ENTRY ARE USED TO
$ DEFINE ADAPTIVITY OPTION ENTRY FOR ALL ELEMENTS,
$
$ FOR THE ADAPT ENTRY:
$ ID = ADAPT ENTRY ID SELECTED IN CASE CONTROL
$ ADGEN = ID OF THE FIRST PVAL ENTRY GENERATED IN THE ADAPTIVE PROCESS;
$ MUST BE LARGER THAN PSTRID, PMINID, AND PMAXID
$ MAXITER = NUMBER OF ITERATIONS REQUESTED BEFORE THE ADAPTIVE PROCESS STOPPED
$ PSTRTID = ID OF THE STARTING PVAL ENTRY
$ PMINID = ID OF THE MINIMUM PVAL ENTRY
$ PMAXID = ID OF THE MAXIMUM PVAL ENTRY
$ PART = ANY NAME CHOSEN FOR THE ELEMENTS;
$ USED FOR ERROR ESTIMATOR REFERENCE
$ ELSETID = ID OF SETS OF ELEMENTS REQUESTED FOR ADAPTIVE PROCESS
$ TYPE = EBEP : THE P-ORDER WILL INCREASE ONLY IN THE ELEMENTS WHOSE ERROR
$ IS GREATER THAN ERROR TOLERANCE.
$ ERREST = ERROR ESTIMATOR METHOD USED FOR ADAPTIVE PROCESS
$ ERRTOL = ERROR TOLERANCE (IN FRACTION), REQUIRED IF MAXITER IS NOT DEFINED
$ SIGTOL = STRESS TOLERANCE
$
$ ID ADGEN MAXITER PSTRTID PMINID PMAXID
ADAPT 1 1001 5 10 20 30 +000001

$
++000001PART=ALL,ELSET=10,TYPE=EBEP,ERREST=1,ERRTOL=0.02,SIGTOL=0.0 +000002
$
$ DEFINE THE POLYNOMIAL ORDER DISTRIBUTION:
$ PVAL ID = ID OF PVAL DEFINED IN THE ADAPT ENTRY ABOVE
$ Px = THE POLYNOMIAL ORDER DISTRIBUTION IN X-DIRECTION
$ Py = THE POLYNOMIAL ORDER DISTRIBUTION IN Y-DIRECTION
$ Pz = THE POLYNOMIAL ORDER DISTRIBUTION IN Z-DIRECTION
$ SETTYP = TYPE OF SET PROVIDED. IT CAN BE EITHER SET or ELID
$ ID = SET ID or ELEMENT ID WHICHEVER IS APPROPIATE
$
$ ID Px Py Pz SETTYP ID
PVAL 10 2 2 2 SET 10
PVAL 20 4 4 2 SET 10
PVAL 30 6 6 2 SET 10
$
$ DEFINE A CYLINDRICAL COORDINATE SYSTEM
$ USED IN DEFINING THE GRIDS
$
CORD2C 1 0 0.0 0.0 0.0 0.0 0.0 1.0 +000003
++0000031.0 0.0 0.0
$
$ DEFINE SOLID ELEMENTS
$ HEXA ELEMENTS ARE THE BOUNDARY ELEMENTS
$
CPENTA 1 1 1 21 22 10 29 30
CPENTA 2 1 1 22 23 10 30 31
CPENTA 3 1 1 23 24 10 31 32
CPENTA 4 1 1 24 25 10 32 33
CPENTA 5 1 1 25 26 10 33 34
CPENTA 6 1 1 26 27 10 34 35
CPENTA 7 1 1 27 28 10 35 36
CPENTA 8 1 1 28 21 10 36 29
CHEXA 9 1 2 3 22 21 11 12 +000005
++00000530 29 +000006
CHEXA 10 1 3 4 23 22 12 13 +000007
++00000731 30 +000008
CHEXA 11 1 4 5 24 23 13 14 +000009
++00000932 31 +000010
CHEXA 12 1 5 6 25 24 14 15 +000011
++00001133 32 +000012
CHEXA 13 1 6 7 26 25 15 16 +000013
++00001334 33 +000014
CHEXA 14 1 7 8 27 26 16 17 +000015
++00001535 34 +000016
CHEXA 15 1 8 9 28 27 17 18 +000017
++00001736 35 +000018
CHEXA 16 1 9 2 21 28 18 11 +000019
++00001929 36
$
$DEFINE EIGRL CARD (LANCZOS METHOD)
$
EIGRL 100 10
$
$DEFINE THE CURVE GEOMETRY, METHOD = EQUATION, U IS THE CURVE PARAMETER
$
$DEFINE THE OUTER BOTTOM CIRCLE OF RADIUS=5.0 AT Z=0.0 USING DEQATN 1,2,AND 3.
$THE OUTER BOTTOM CIRCLE IS DEFINED BY TWO GMCURV ENTRIES:
$GMCURV 1 WITH THE RANGE OF CURVE PARAMETER,U = 0.0 TO 3.1416 RADIANS
$GMCURV 2 WITH THE RANGE OF CURVE PARAMETER,U = 3.1415 TO 6.2832 RADIANS
$
www.cadfamily.comDEQATN
1 X(U) = 5.0*COS(U)
CHAPTER 25 791
p-Elements

DEQATN 3 Z(U) = 0.0
$
GMCURV 1 MSCGRP1 0 0 +000021
++000021EQUATION,0.0,3.1416,1,2,3 +000022
++000025EQUATION,3.1415,6.2832,1,2,3 +000026
$
$ DEFINE THE TOP OUTER CIRCLE USING THE SAME EQUATION, DEQATN 1 AND DEQATN 2,
$ BUT WITH Z=0.5 (DEQATN 4). THIS TOP OUTER CIRCLE IS DEFINED BY TWO GMCURV
$ ENTRIES:
$ GMCURV 3 WITH THE RANGE OF CURVE PARAMETER,U = 0.0 TO 3.1416 RADIANS
$
$
++000023EQUATION,0.0,3.1416,1,2,4 +000024
++000027EQUATION,3.1415,6.2832,1,2,4 +000028
$
$ DEFINE THE BOTTOM INNER CIRCLE OF RADIUS = 4.0 AT Z=0.0 USING DEQATN 5,6,
$ AND 7. THE BOTTOM INNER CIRCLE IS DEFINED BY TWO GMCURV ENTRIES:
$ GMCURV 6 WITH THE RANGE OF CURVE PARAMETER U = 3.1415 TO 6.2832 RADIANS
$
DEQATN 5 X(U) = 4.0*COS(U)
DEQATN 7 Z(U) = 0.0
$
++000029EQUATION,0.0,3.1416,5,6,7 +000030
++000033EQUATION,3.1415,6.2832,5,6,7 +000034
$
$ DEFINE THE TOP INNER CIRCLE USING THE SAME EQUATION, DEQATN 5 AND DEQATN 6,
$ BUT WITH Z=0.5 (DEQATN 8). THIS TOP INNER CIRCLE IS DEFINED BY TWO GMCURV
$ ENTRIES:
$ GMCURV 8 WITH THE RANGE OF CURVE PARAMETER U = 3.1415 TO 6.2832 RADIANS
$
$
++000031EQUATION,0.0,3.1416,5,6,8 +000032
++000035EQUATION,3.1415,6.2832,5,6,8 +000036
$
$ DEFINE FEEDGES
$ FEEDGE BY DEFAULT IS A STRAIGHT LINE.
$ GMCURVS ARE REQUIRED FOR FEEDGES THAT DEFINE CURVED EDGES
$
$ FEEDGES 1, 2, 3 AND 4 MAKE GMCURV 1
$
$ ID GRID 1 GRID 2 CID GEOMIN ID
www.cadfamily.com
FEEDGEEMail:[email protected]
1 2 3 0 GMCURV 1
$
$ DEFINE FEFACES FOR BOUNDARY CONDITION
$
$ ID GRID 1 GRID 2 GRID 3 GRID 4 CID
FEFACE 1 2 3 12 11 0
FEFACE .2 3 4 13 12 0
FEFACE 3 4 5 14 13 0
FEFACE 4 5 6 15 14 0
FEFACE 5 6 7 16 15 0
FEFACE 6 7 8 17 16 0
FEFACE 7 8 9 18 17 0
FEFACE 8 9 2 11 18 0
$
$ DEFINE CONSTRAINTS IN THE Z-DIRECTION
$
$ SID = SPC ID THAT IS DEFINED IN CASE CONTROL
$ C = CONSTRAINT COMPONENT NUMBER IN GLOBAL COORDINATE SYSTEM.
$ IT HAS TO BE A SINGLE INTEGER (1 OR 2 OR 3).
$ ENTITY = ENTITY AT WHICH THE CONSTRAINT IS APPLIED
$ ID = ID OF THE ENTITY SELECTED ABOVE
$
$ SID C ENTITY ID
GMSPC 1 3 FEFACE 1
GMSPC 1 3 FEFACE 2
GMSPC 1 3 FEFACE 3
GMSPC 1 3 FEFACE 4
GMSPC 1 3 FEFACE 5
GMSPC 1 3 FEFACE 6
GMSPC 1 3 FEFACE 7
GMSPC 1 3 FEFACE 8
$
www.cadfamily.com$ DEFINE
CONSTRAINTS IN THE Y-DIRECTION
CHAPTER 25 793
p-Elements
$
GMSPC 1 2 GRID 1
$
$ DEFINE CONSTRAINTS IN THE X-DIRECTION AND Z-DIRECTION
$ FOR BOTH GRID 4 AND GRID 8. MULTIPLE GMSPC ENTRIES ARE
$ REQUIRED FOR CONSTRAINING MULTIPLE COMPONENTS.
$
GMSPC 1 1 GRID 4
GMSPC 1 3 GRID 4
GMSPC 1 1 GRID 8
GMSPC 1 3 GRID 8
$
$ DEFINE CONSTRAINTS FOR ROTATION FOR ALL GRIDS
$
GRDSET 0 0 456
$
$ DEFINE GRIDS
$
GRID 1 0.0 0.0 0.0
GRID 2 1 5.0 0.0 0.0
GRID 3 1 5.0 45.0 0.0
GRID 4 1 5.0 90.0 0.0
GRID 5 1 5.0 135.0 0.0
GRID 6 1 5.0 180.0 0.0
GRID 7 1 5.0 225.0 0.0
GRID 8 1 5.0 270.0 0.0
GRID 9 1 5.0 315.0 0.0
GRID 10 0.0 0.0 0.5
GRID 11 1 5.0 0.0 0.5
GRID 12 1 5.0 45.0 0.5
GRID 13 1 5.0 90.0 0.5
GRID 14 1 5.0 135.0 0.5
GRID 15 1 5.0 180.0 0.5
GRID 16 1 5.0 225.0 0.5
GRID 17 1 5.0 270.0 0.5
GRID 18 1 5.0 315.0 0.5
GRID 21 1 4.0 0.0 0.0
GRID 22 1 4.0 45.0 0.0
GRID 23 1 4.0 90.0 0.0
GRID 24 1 4.0 135.0 0.0
GRID 25 1 4.0 180.0 0.0
GRID 26 1 4.0 225.0 0.0
GRID 27 1 4.0 270.0 0.0
GRID 28 1 4.0 315.0 0.0
GRID 29 1 4.0 0.0 0.5
GRID 30 1 4.0 45.0 0.5
GRID 31 1 4.0 90.0 0.5
GRID 32 1 4.0 135.0 0.5
GRID 33 1 4.0 180.0 0.5
GRID 34 1 4.0 225.0 0.5
GRID 35 1 4.0 270.0 0.5
GRID 36 1 4.0 315.0 0.5
$
$ DEFINE THE MATERIAL PROPERTIES (ALUMINUM)
$
MAT1 1 1.+7 0.3 2.538-4
PSOLID 1 1 GRID SMECH
$
ENDDATA
This example illustrates a normal modes analysis, SOL 103, using the Lanczos
method (EIGRL entry). The output requests are eigenvectors for all modes
(DISPLACEMENT = ALL). Adaptivity analysis is required for all the elements. For
this purpose, both the ADAPT and PVAL entries are used in conjunction with the
ADAPT command in Case Control Section.
Results
1 CIRCULAR PLATE JULY 23, 2003 NX NASTRAN 07/23/03 PAGE 929
EDGE SIMPLY SUPPORTED
0
0
E I G E N V A L U E A N A L Y S I S S U M M A R Y (LANCZOS ITERATION)
BLOCK SIZE USED ...................... 7
NUMBER OF DECOMPOSITIONS ............. 2
NUMBER OF ROOTS FOUND ................ 10
NUMBER OF SOLVES REQUIRED ............ 10
TERMINATION MESSAGE : REQUIRED NUMBER OF EIGENVALUES FOUND.
1 CIRCULAR PLATE DECEMBER 13, 1993 NX NASTRAN 07/23/03 PAGE 930
0
1 1 3.465334E+07 5.886709E+03 9.368988E+02 1.000000E+00 3.465334E+07
2 2 1.182291E+08 1.087332E+04 1.730543E+03 1.000000E+00 1.182291E+08
3 3 1.483814E+08 1.218119E+04 1.938696E+03 1.000000E+00 1.483814E+08
4 4 2.502755E+08 1.582010E+04 2.517847E+03 1.000000E+00 2.502755E+08
5 5 2.654028E+08 1.629119E+04 2.592823E+03 1.000000E+00 2.654028E+08
6 6 3.027782E+08 1.740052E+04 2.769379E+03 1.000000E+00 3.027782E+08
7 7 8.642510E+08 2.939815E+04 4.678860E+03 1.000000E+00 8.642510E+08
8 8 8.645740E+08 2.940364E+04 4.679734E+03 1.000000E+00 8.645740E+08
9 9 1.166431E+09 3.415306E+04 5.435628E+03 1.000000E+00 1.166431E+09
10 10 2.009605E+09 4.482862E+04 7.134696E+03 1.000000E+00 2.009605E+09

0
*** USER INFORMATION MESSAGE 6502 (ADAPT)
THE CURRENT SUPERELEMENT REQUIRES NO FURTHUR P-ADAPTIVE ANALYSIS BECAUSE ALL P-ELEMENTS REACHED MAXIMUM P-VALUES.
0*** USER WARNING MESSAGE 4594. THE POSTCDB CONTAINS NO LEGITIMATE SURFACE OR VOLUME DEFINITIONS.

EDGE SIMPLY SUPPORTED ADAPTIVITY INDEX= 5
0
CYCLES = 9.368987E+02 R E A L E I G E N V E C T O R N O . 1

1 G 6.661354E-06 .0 -1.868621E+01 .0 .0 .0
2 G 1.333745E+00 -2.678128E-05 .0 .0 .0 .0
3 G 9.431428E-01 9.430584E-01 .0 .0 .0 .0
4 G .0 1.333743E+00 .0 .0 .0 .0
5 G -9.430543E-01 9.431386E-01 .0 .0 .0 .0
6 G -1.333741E+00 2.663938E-05 .0 .0 .0 .0
7 G -9.431665E-01 -9.430221E-01 .0 .0 .0 .0
8 G .0 -1.333743E+00 .0 .0 .0 .0
9 G 9.430305E-01 -9.431752E-01 .0 .0 .0 .0
10 G 6.681371E-06 -6.129073E-08 -1.868621E+01 .0 .0 .0
11 G -1.333732E+00 -3.440074E-04 .0 .0 .0 .0
12 G -9.428723E-01 -9.433200E-01 .0 .0 .0 .0
13 G 3.471338E-04 -1.333743E+00 .0 .0 .0 .0
14 G 9.433249E-01 -9.428771E-01 .0 .0 .0 .0
15 G 1.333754E+00 3.440111E-04 .0 .0 .0 .0
16 G 9.429217E-01 9.432840E-01 .0 .0 .0 .0
17 G -3.258019E-04 1.333743E+00 .0 .0 .0 .0
18 G -9.432753E-01 9.429132E-01 .0 .0 .0 .0
21 G 1.343029E+00 -8.329611E-05 -5.443803E+00 .0 .0 .0
22 G 9.497321E-01 9.495923E-01 -5.443810E+00 .0 .0 .0
23 G 8.442988E-05 1.343021E+00 -5.443810E+00 .0 .0 .0
24 G -9.495862E-01 9.497259E-01 -5.443810E+00 .0 .0 .0
25 G -1.343013E+00 8.315749E-05 -5.443817E+00 .0 .0 .0
26 G -9.497337E-01 -9.495776E-01 -5.443811E+00 .0 .0 .0
27 G -7.570248E-05 -1.343021E+00 -5.443810E+00 .0 .0 .0
28 G 9.495844E-01 -9.497408E-01 -5.443809E+00 .0 .0 .0
29 G -1.343016E+00 -2.135180E-04 -5.443803E+00 .0 .0 .0
30 G -9.495108E-01 -9.498014E-01 -5.443810E+00 .0 .0 .0
31 G 2.196022E-04 -1.343021E+00 -5.443810E+00 .0 .0 .0
32 G 9.498075E-01 -9.495170E-01 -5.443810E+00 .0 .0 .0
33 G 1.343026E+00 2.135190E-04 -5.443817E+00 .0 .0 .0
34 G 9.495386E-01 9.497868E-01 -5.443811E+00 .0 .0 .0
35 G -2.027499E-04 1.343021E+00 -5.443810E+00 .0 .0 .0
36 G -9.497795E-01 9.495314E-01 -5.443809E+00 .0 .0 .0
Figure 25-13 Abridged Output from the Circular Solid Model

CHAPTER 25 795
p-Elements
201001001 G 1.333745E+00 -2.678128E-05 .0 .0 .0 .0

201001002 G 1.288204E+00 3.459260E-01 .0 .0 .0 .0
201001003 G 1.155554E+00 6.657130E-01 .0 .0 .0 .0
201001004 G 9.431428E-01 9.430584E-01 .0 .0 .0 .0
201001005 G 1.354012E+00 -1.118739E-07 -1.801449E+00 .0 .0 .0
201001006 G 1.307760E+00 3.511871E-01 -1.801487E+00 .0 .0 .0
201001007 G 1.173140E+00 6.758680E-01 -1.801314E+00 .0 .0 .0
201001008 G 9.574580E-01 9.574031E-01 -1.801455E+00 .0 .0 .0
201001009 G 1.357469E+00 -5.671428E-05 -3.626504E+00 .0 .0 .0
201001010 G 1.311071E+00 3.521154E-01 -3.626570E+00 .0 .0 .0
201001011 G 1.176151E+00 6.776471E-01 -3.626329E+00 .0 .0 .0
201001012 G 9.599283E-01 9.598196E-01 -3.626513E+00 .0 .0 .0
201001013 G 1.343029E+00 -8.329611E-05 -5.443803E+00 .0 .0 .0
201001014 G 1.297096E+00 3.483891E-01 -5.443859E+00 .0 .0 .0
201001015 G 1.163644E+00 6.704805E-01 -5.443617E+00 .0 .0 .0
201001016 G 9.497321E-01 9.495923E-01 -5.443810E+00 .0 .0 .0
Figure 25-13 Abridged Output from the Circular Solid Model (continued)
The mode shapes for modes 1, 5, 8, and 9 are shown in Figure 25-14. Mode 1 is the
fundamental mode, mode 5 is the one nodal diameter mode, mode 8 is the two
nodal diameters mode, and mode 9 is one nodal circle mode.
Mode 1: Fundamental Mode
Mode 5: One Nodal Diameter Mode
Figure 25-14 The Mode Shapes for Modes 1, 5, 8, and 9

Mode 8: Two Nodal Diameter Mode
Mode 9: One Nodal Circle Mode
Figure 25-14 The Mode Shapes for Modes 1, 5, 8, and 9 (continued)

Comparison of the natural frequencies obtained from the adaptivity analysis for the
solid circular model to the theoretical natural frequencies for the circular plate is
shown in Table 25-6. The discrepancies between the two values decrease as the p-
order increases.
Table 25-6 Comparison of the Natural Frequencies
Theoretica
Numerical
Mode l
(Hertz)
(Hertz)
Fundamental 954.1 936.9
One Nodal Diameter 2657.7 2592.8
CHAPTER 25 797
p-Elements
Table 25-6 Comparison of the Natural Frequencies
Theoretica
Numerical
Mode l
(Hertz)
(Hertz)
Two Nodal Diameters 4913.9 4679.7
One Nodal Circle 5697.8 5435.6
Further studies were performed in which the p-order was kept constant and the
thickness of the solid model was varied. Results showed that as the thickness of the
circular solid model decreases, the solid model behaves like a plate model, which is
to be expected.
Reference:
[1] Roark, R. J. and Young, W. C. Formulas for Stress and Strain. 5th edition.
McGraw-Hill, Inc., 1975, p. 578.
Clamped Rhombic Plate

The first eigenfrequency of a rhombic plate is calculated using SOL 103.
Figure 25-15 shows a plate clamped at the left end and the first eigenmode.
Compare the convergence behavior of three different models:
• One CHEXA p-element with p-orders varying from p = 1 to p = 8
• One CQUAD p-element with p-orders varying from p = 1 to p = 8
• Multiple CQUAD h-elements with varying meshes ( 1 × 1 , 2 × 2 , 4 × 4 ,
8 × 8 , 16 × 16 , 32 × 32 )
The convergence behavior of the three models is shown in Figure 25-16. The first
eigenfrequency is normalized with 2.932 rad/sec, which is the value from a series
solution using Kirchhoff plate theory. All three models exhibit excellent
convergence behavior.
The solid p-element and the shell p-element models converge when the p-level is
increased. The CQUAD4 h-element model converges when the mesh is refined. If
the error rate in the first eigenfrequency must be under 0.2%, then either: (a) a single
shell or solid p-element with p = 8, or (b) 16 × 16 shell h-elements should be used.
The shell p-element needs the smallest number of degrees-of-freedom to achieve the
required accuracy. For a comparison of the degrees-of-freedom, see Table 25-7.
10” t = 0.5″
E = 10.6 × 10 6 psi
ν = 0.3
10”
30°
First
Eigenmode
Figure 25-15 Rhombic Plate
1.15
1.10
Normalized 1. Eigenfrequency
CHEXA-p
1.05
1.00
CQUAD4-p
0.95
CQUAD4
0.90
0.85
p=1 2 3 4 5 6 7 8 p-Order
1×1 2×2 4×4 8 × 8 16 × 1 6 32 × 32 Mesh Size
Figure 25-16 Normalized First Eigenfrequency for Different p-Levels and Mesh
Sizes
CHAPTER 25 799
p-Elements
Table 25-7 Comparison of the Number of Degrees-of-Freedom for 0.2%

Accuracy in the First Eigenfrequency
Normalized
Mesh Size Degrees of
Element Eigenfrequenc
and p-Level Freedom
y
CQUAD4 16 × 16 0.9977 1445
CQUAD4-p p=8 1.0012 264

CHEXA-p p=8 1.0017 510
Helix Modeled with CBEAM p-Elements

A 360° helix is modeled with nonadaptive beam h-elements and with adaptive
beam
p-elements. The helix is clamped at the bottom and free at the top, with a single
force applied at the top. The undeformed and deformed helix is shown in
Figure 25-17. The normalized deflection at the tip is shown in Figure 25-18 for
different mesh sizes.
Both beam h- and p-elements converge well. If the error in the deflection must be
under 0.2%, then 8 beam p-elements with p = 3 ( 45° angle for each element) or 64
beam h-elements should be used. A comparison of the degrees-of-freedom (see
Table 25-8) shows that the beam p-elements require a smaller number of degrees-
of-freedom than the number required by the beam h-elements in order to achieve
the required accuracy.
θ = 360° CHAN1 cross-section

H = 9. height E = 1.5e+7
R = 5. radius ν = 0.33
Figure 25-17 Helix
CHAPTER 25 801
p-Elements
1.000
0.990
CBEAM p = 3
Normalized Tip Displacement

0.980
0.970
0.960 CBEAMh
0.950
0.940
0.930
0.920
0.910
0.900
90° 45° 22.5° 11.25° 5.625° 2.812°
Angle θ
Tip displacement normalized with u z = 5.930e-3
Figure 25-18 Normalized Tip Displacement for Different Mesh Sizes
Table 25-8 Comparison of the Number of degrees-of-freedom for 0.2%

Accuracy in the Tip Displacement
Normalized Tip Degrees of

Element and Mesh
Displacement Freedom
CBEAM h, θ = 5.6° ; 64 Elements 0.9970 390
CBEAM p, θ = 45° ; p = 3,8 Elements 0.9978 150
Plate with Stiffener

In stiffened structures, an efficient practice is to model the stiffeners using beam
elements with offsets. Typically, nonadaptive h-elements are used. The example
below demonstrates the benefits of p-elements for this class of problems.
A plate with stiffener is loaded in the middle and simply supported at the ends (see
Figure 25-19). The plate is modeled with 10 × 2 shell elements and the stiffener is
modeled with 10 beam elements with offsets. Compare the stress results of two
models, an h-element model and a p-element model, with a uniform p-order of p = 2
for shells and beams. The solution is compared with the solution derived from beam
theory.
The nonadaptive h-elements produce very inaccurate stresses with errors around
30%, as shown in Figure 25-20. The adaptive p-elements with a uniform p-order of
p = 2 yield very accurate results—the error rate is only around 4%. In the h-element
model, the normal forces in the shell and beam elements are constant within each
element, which causes the high stress error. In the p-element model, the normal
forces in each element are linear (p = 2), which improves the stress distribution in the
cross-section of the beam.
P = 10,000 lb
Shell p-elements
100”
Beam p-elements with offsets
Cross Section
10 I z = 442.662
P l
.5 M at center = --- ⋅ --- = 250,000 lb-in
2 2
M ( h – yc )
13 = h .5 σ top = – -------------------------- = – 3560 psi
Iz
yc = 6.7 .5 M ( yc )
σ bottom = ---------------- = 3786 psi
Iz
9
Figure 25-19 Plate with Stiffener
CHAPTER 25 803
p-Elements
y (in.)
–3560
-3277
3786
0.0 Stress (psi)
Stress y Stress y Stress y

[psi] [in] [psi] [in] [psi] [in]
3786. 0. 3978. 0. 3834. 0.

–3277. 12.5 –4311. 12.5 –3545. 12.5
–2804. 12.5 –3250. 12.5
–3560. 13.0 –3082. 13.0 –3590. 13.0
p=elements
beam theory h-elements p=2
Figure 25-20 Stress Distribution in the Middle of the Beam
25.5 Solid p-Elements in Linear Dynamic Solution
Sequences
Solid p-elements are available in the dynamic solution sequences SOL 107, to
SOL 112. The p-elements do not adapt; the analysis is done for a fixed p-level.
Displacements, velocities, accelerations, stresses and strains are output in SORT1 or
SORT2 format. The complex output quantities are in rectangular format
(real/imaginary) or in polar format (magnitude/phase).
User Interface in NX Nastran

Output for p-elements is requested with the Case Control command DATAREC,
where SORT1 or SORT2 formats can be chosen. For complex eigenvalues (SOL 107,
SOL 110) and frequency response (SOL 108, SOL 111), the default is SORT1. For
transient response (SOL 109, SOL 112), the default is SORT2.
The output quantities are requested with DISP, VELO, ACCE, STRESS, or STRAIN in
the Case Control, and/or in the OUTPUT Bulk Data entry. In the complex solution
sequences, the formats are rectangular (REAL, IMAG) or polar (MAGNITUDE,
PHASE); the rectangular format is the default. REAL, IMAG or PHASE is requested
in the Case Control command or in the OUTPUT Bulk Data entry.
Limitations and Recommendations

The following limitations and recommendations apply to solid p-element usage in
linear dynamic solution sequences:
• The output request must be all in SORT1 or all in SORT2 format for a single
run.
• PARAM,CURVPLOT,1 must be specified to get SORT1 output in transient
response (SOL 109 and SOL 112).
• If an ADAPT entry is specified with multiple p-levels, the analysis is done
with the first p-level.
Example
The transient response of a cantilever subject to a constant step load is calculated in
SOL 109 (see Figure 25-21). The following is a comparison of the results between the
two different models:
• HEXA8 h-elements, 4 × 4 × 40 element mesh.

• HEXA p-elements, 1 × 1 × 4 element mesh with p = 3.
The plot displacement versus time graph shown in Figure 25-22 illustrates that the
results are in good agreement.
CHAPTER 25 805
p-Elements
y y
(t)
P
z 1.0
x
10.
0.8
E = 3.e+7 P
ν = 0.25
PARAM,W3,9.7 1.0
PARAM,G,0.0647
t
Figure 25-21 Cantilever with Step Load
Figure 25-22 Tip Displacement of Cantilever
25.6 p-Element Thermal Analysis

Introduction
NX Nastran supports p-version elements for linear steady state thermal analysis
(SOL 101). The definitions of model geometry, elements, p-values, and error
estimation methods follow the general terminologies of the p-element structural
analysis. A major difference exists in the specification of thermal loads and boundary
conditions, which requires new or modified Bulk Data entries for thermal analysis.
The following sections provide usage guidelines and examples for the p-version
thermal analysis.
Geometry
The geometry definitions are the same as those defined for the p-element structural
analysis.
Elements
Six kinds of elements—CHEXA, CPENTA, CTETRA, CQUAD4, CTRIA3, and
CBEAM—may be defined as p-elements for thermal analysis.
Materials
The standard isotropic, orthotropic, and anisotropic materials are available for p-
version thermal elements. The only material property required is thermal
conductivity which must be constant for linear steady state analysis.
Loads and Boundary Conditions

The following thermal loads and boundary conditions are supported in p-version
thermal analysis.
• Prescribed temperatures applied on point, curve, 2D edge, surface, or 3D
surface.
• Normal or directional heat fluxes applied on point, curve, 2D edge, surface,
3D surface, or solid.
• Volumetric heat generation within conduction elements.
• Free convection applied on point, curve, 2D edge, surface, or 3D surface.
CHAPTER 25 807
p-Elements
Table 25-9 summarizes the Bulk Data entries for p-version thermal loads and
boundary conditions. A full description of these entries may be found in the NX
Table 25-9 Bulk Data Entries for p-Version Thermal Loads and Boundary
Conditions
Load or Boundary Condition Type Bulk Data

Prescribed Temperature GMBC
Normal or Directional Heat Flux GMLOAD
Volumetric Heat Generation GMQVOL
Free Convection GMCONV
p-Version Adaptivity
p-values and error estimation methods are the same as those for p-element
structural analysis.
Data Recovery
Similar to structural output, the DATAREC and OUTRCV Case Control commands
as well as the OUTPUT and OUTRCV Bulk Data entries define the output of p-
version thermal analysis. Temperatures, temperature gradients, heat fluxes,
polynomial values, and element errors may be printed, plotted, or punched. The
TEMP output command of the OUTPUT Bulk Data entries requests for calculating
temperatures while the FLUX output command of the OUTPUT Bulk Data entries
requests for recovering temperature gradients and heat fluxes for conduction
elements.
Thermal Stress Analysis

The interface to input temperature results as temperature loads for thermal stress
analysis is described in the Case Control command “TSTRU” on page 419 of the NX
Limitations and Recommendations

• Only Solution Sequence 101 (linear steady state) performs p-element
thermal analysis.
• Temperature boundary conditions can be spatially varying, but must be
independent or time.
• Heat fluxes and volumetric heat generation can be spatially varying, but
must be independent of temperature or time.
• The convection heat transfer coefficients and ambient temperatures of free

convection boundary conditions can be spatially varying, but must be
independent of temperature or time.
• It is important to select the appropriate entity type while applying loads
and boundary conditions. Concentrated loads or boundary conditions will
cause singularities or inaccurate results and should be avoided. For
example, the entity type FEFACE or GMSURF should be used to prescribe
temperatures on a surface, rather than the entity type FEEDGE, GMCURVE,
or GRID.

NX Nastran Users Guid

Uploaded by

Copyright:

Available Formats

NX Nastran Users Guid

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

NX Nastran Users Guid

Uploaded by

Copyright:

Available Formats

NX Nastran

NX Nastran Reference ManualNX

■ Overview of the NX Nastran Analysis Process, 3

■ Understanding NX Nastran Input Records, 10

■ Specifying Continuations for Bulk Data, 14

■ Replicating Bulk Data Entries, 17

■ Inserting External Files with INCLUDE, 20

■ Including Comments in the Input File, 21

■ The NASTRAN Statement (Optional), 27

■ Case Control Section (Required), 32

■ Bulk Data Section, 40

■ Understanding the .f04 File, 59

■ Understanding the .log File, 68

■ Specifying Grid Point Properties, 79

■ Understanding Scalar Points, 81

■ Defining a Local Coordinate System, 87

■ Understanding the Element and Material Coordinate System, 101

■ Other Special Coordinate System Entries, 102

■Isotropic Materials, 107

■ Anisotropic Materials, 119

■ Material Properties for Heat Transfer, 133

■ Material Properties for Fluid Analysis, 136

■ Material Properties for Nonlinear Analysis, 136

■ Defining Static Loads, 148

■ Distributed Loads, 156

■ Inertia Loads (Acceleration Loads), 172

■ Thermal Loads, 182

■ Loads Due to Enforced Motion, 194

■ Automatically Applying Single-Point Constraints, 216

■ Enforced Displacements at Grid Points (SPCD, SPC), 220

■ Rigid Body Supports, 229

■ Static Condensation (Guyan Reduction), 248

■ Sets for Dynamic Reduction, 257

■ DMIG Examples, 266

■ Mesh Stress Discontinuities at Grid Points, 294

■ Data Processing and Matrix Operations by Functional Module, 308

■ Determining Mesh Density, 399

■ Creating Mesh Transitions, 407

■ Reviewing Grid Point Stresses, 432

■ Stress Error Estimators, 494

■ Postprocessor Checks, 506

■ Read-Write Restarts, 516

■ Read-Only Restarts, 526

■ Using the File Management Section with Databases, 541

■ Guidelines for Large Problems, 550

■ Superelement Plotting, 589

■ Linear Buckling Assumptions and Limitations, 606

■ Understanding Classical Lamination Theory, 641

■ Calculating Interlaminar Shear Stress and Strains, 656

■ Acoustic Cavity Modeling, 683

■ Using Cyclic Symmetry, 729

■ Performing Axisymmetric Analysis, 745

■ p-Version Capabilities, 756

■ p-Element Loads and Constraints, 757

■ p-Element Example Problems, 763

■ p-Element Thermal Analysis, 806

■ General Overview of NX Nastran

■ Overview of the NX Nastran Analysis Process

1.1 General Overview of NX Nastran

• Linear statics (including inertia relief)