SYSTUS RefManAnalysisVol3 en PDF

SYSTUS
2019
REFERENCE ANALYSIS MANUAL
(VOL.3)
www.esi-group.com
SYSTUS 2019
REFERENCE ANALYSIS MANUAL
(VOL.3)
© 2019 by ESI Group
Published: January 2019
GL/SYST/18/08/03/A
This documentation is the confidential and proprietary product of ESI Group.
Any unauthorized use, reproduction, or transfer of this documentation is strictly prohibited.
This documentation is subject to limited distribution and restricted disclosure.
All PAM- and SYS- product names as well as other products belonging to ESI's portfolio are
tradenames or trademarks of ESI Group, unless specifically mentioned. All other trademarks are the
property of their respective owners.
Specifications are subject to change without prior notice.
Contents
CHAPTER 9 LINEAR STATIC SOLUTION 613

9.1 Presentation-General Syntax 613
9.2 Static Solution 615
9.2.1 General Operation 615
9.2.2 Choice of Method of Solution 616
9.2.3 Analyze Check- Rerun 617
9.3 Creation of Elementary Matrices 619
9.3.1 Standard Utilization 619
9.3.2 Utilization of Elimination Methods 621
9.4 Harmonic Loading 622
9.5 Thermoelastic Analysis 623
9.5.1 Basis 623
9.5.2 Temperature File 624
9.5.3 Examples 626
9.6 Contacts for Linear Static Analysis 630
9.6.1 Object 630
9.6.2 Files 631
9.6.3 Data Description 632
9.6.4 Example 633
CHAPTER 10 SUBSTRUCTURING 635

10.1 Introduction 635
10.1.1 Terminology 635
10.1.2 Example 636
10.2 Method of Analysis 638
10.3 File Management 640
10.3.1 Description of Files 640
10.3.2 Calculation Steps 642
10.4 Optimization of an Analysis 644
10.4.1 Re-Use of Substructures 644
10.4.2 Change of Axes 646
10.4.3 Re-Use of Files 647
10.4.4 Interfacing of Different Types of Structure 648
10.5 Input Data 650
10.5.1 Zero Level Substructure (superelement) 650
10.5.2 Level n Substructure 653
10.5.3 The Command CALCULATE 656
10.5.4 Automatic Link Generator (A.L.G.) 658
10.6 Procedures associated with Substructure 662
CONTENTS
10.7 Conclusions 663
CHAPTER 11 DYNAMIC 665

11.2 Eigenmodes and Eigenfrequencies - Direct Methods 668
11.2.1 Real Eigenmodes and Eigenfrequencies 668
11.2.2 Real Eigenfrequencies and Eigenmodes with Fluid-Structure Interaction 697
11.2.3 Complex Eigenfrequencies and Eigenmodes (Symmetrical Matrices) 704
11.2.4 Complex Eigenfrequencies and Eigenmodes (Non-Symmetrical Matrices) 709
11.3 Real Eigenmodes and Eigenfrequencies - Modal Synthesis 713
11.3.1 General 713
11.3.2 Description of Methods 714
11.3.3 Description of a Physical Super-Element 719
11.3.4 Description of Computation Steps 720
11.3.5 File Management 722
11.3.6 Input Data 725
11.3.7 Example: Uniform Clamped-Free Beam 736
11.3.8 Experimental Super-Elements 741
11.3.9 Dynamic Automatic Link Generator (D.A.L.G.) 749
11.4 Dynamic Response - Modal Method 751
11.4.1 General 751
11.4.2 Modal Forces - Representativeness of the Modal Base 756
11.4.3 Transient Response (Modal Analysis) 767
11.4.4 Harmonic Response 783
11.4.5 Stochastic Dynamics 790
11.4.6 Seismic Response - Spectral Method 800
11.4.7 Consistant Modal Matrices Method 818
11.5 Harmonic Loads - Direct Method 823
11.5.1 General 823
11.5.3 Results 827
11.6 Sensitivity Analysis 829
11.6.1 General 829
11.6.2 Analytical Re-Analysis in the Modal Base 830
11.6.3 Complete Re-Analysis in the Modal Base 833
11.7 Utility Modules 841
11.7.1 Equivalent Modal Damping 841
11.7.2 Generation of Spectral Response 844
11.7.3 Computation of Effective Masses and Effective Inertia 852
11.7.4 Transient Dynamic 854
CHAPTER 12 TRANSIENT 861

12.1 Overview 861
12.1.1 Introduction 861
12.1.2 Time-Related Scheme 862
12.1.3 Special Features 863
12.1.4 Commands 864
- ii - SYSTUS 2019 REFERENCE ANALYSIS MANUAL

© 2019 ESI Group
CONTENTS
12.2 Linear Thermal Analysis 865

12.2.2 Forced Convection 866
12.2.3 Use of Routine 867
12.2.4 Example of a Quadratic Mesh Treatment 873
12.3 Linear Elasticity 875
12.3.1 General Equation 875
12.3.2 General 876
12.3.4 Relations 883
12.3.5 Calculation of Forces 885
12.4 Non-Linear Dynamic Analysis 888
12.4.2 Detail of Time-Related Scheme 891
12.4.3 Utilization of the Routine 892
12.4.4 Combination between Non-Linearities and Algorithms 913
12.4.6 HISTORY and TRANSIENT Files 920
12.5 Non-Linear Static Analysis 923
12.5.2 Implicit Algorithm 924
12.5.3 Utilization 924
12.5.4 Substructuring 925
12.6 Transient Shell 929
12.6.1 Mechanical 929
12.6.2 Thermal 941
12.7 Utility Routines 944
12.7.2 Transient Thermo-Mechanical Analysis: the TEMPERATURE Command 944
12.7.3 Manipulation of the TRANSIENT File - The SELECT Command 949
12.8 Transfer of Physical Quantities from Mesh to Mesh 952
12.9 Element Activation/Deactivation 962
12.9.2 Activation/Deactivation Function 962
12.9.3 Input Data: Element Properties 963
12.9.4 Input Data: Definition of Initial State 964
12.9.5 Data Storage 964
12.9.6 Practical Tips 964
12.9.7 Examples 965
INDEX i
SYSTUS 2019 REFERENCE ANALYSIS MANUAL - iii -

© 2019 ESI Group
CHAPTER 9 LINEAR STATIC SOLUTION
9.1 Presentation-General Syntax

9.1 PRESENTATION-GENERAL SYNTAX
The SOLVE command (Figure 9-1) provides two main functions:
l Solution of a linear system, and the association of one or more files containing displacements,
reactions and forces (stresses and strains) with a DATA file.

l The association of one or all of the STIFFNESS, MASS, DAMPING and BUCKLING
elementary matrix files with a DATA file, for subsequent utilization with the DYNAMIC or
TRANSIENT routine.
General syntax:
SOLVE (labels)
The labels can be grouped in 4 categories:

l Solution of a linear system
Choice of a solution method
o METHOD
l Computation check
o MATRICES
o RESULTS
o PARTIAL
o FORCES (MODAL) (TRANSIENT)
l Formation of elementary matrices
o STIFFNESS
o MASS
o DAMPING or BUCKLING
o ELIMINATION (FILE or RATIO n ) ELEMENT n FACTOR n

1 2 3
l Printout checks
o EXTRACT i (i ≥1 )
SYSTUS 2019 REFERENCE ANALYSIS MANUAL - 613 -

© 2019 ESI Group
9.1 Presentation-General Syntax
Figure 9-1: The command SOLVE
- 614 - SYSTUS 2019 REFERENCE ANALYSIS MANUAL

© 2019 ESI Group
9.2 Static Solution
9.2 STATIC SOLUTION

9.2.1 General Operation
A linear solution consists of three operations:
a) Execution of routines which, according to the type of computation option, generate from
the file DATA a sequential file named STIFFNESS on which are held:
l for each finite element,
l for each node subject to loading or constraint.
the contribution of each element or node to the left-hand side (stiffnesses Ki) and right-hand
side (forces Fi).
b) Execution of routines which:
l assemble the basic stiffness matrices and load vectors Ki and Fi into the global stiffness
matrix and load vector K and F:

K.U = F
K = Σ Ki
F = Σ Fi
l factorize the left-hand side fully (Gaussian triangulation) or partially, using one of three
methods to form a MATRIX file,
l solve the right-hand side, forming a displacement file containing solution vectors U and a
reaction-residue file
containing vectors, the components of which are infinitely small for the free
degrees of freedom (residues) and reactions for the constrained degrees of freedom.
The largest residue should be negligibly small compared with the resultant of
applied loads, the order of magnitude of residues compared with that of applied
loads being a measure of the precision of simultaneous equation solution.
c) Execution of routines which, according to the type of structure, produce force, stress and
strain files.
The data on results files is stored by load case.

© 2019 ESI Group
9.2 Static Solution
9.2.2 Choice of Method of Solution

This choice is driven by the label associated with METHOD following the command SOLVE.
Sparse Method (Default Method)
SOLVE
The algorithm uses an internal numbering system. This minimizes the number of terms of the
matrices. The working files are stored automatically in core memory or on disk, and this
algorithm is compatible with all types of computer.
The triangulated matrix can be stored for use with other load cases.
If the triangulated matrix is not stored, the optional label WITHOUT is used to optimize the
space on the hard disk.
Choice of the Solver
SOLVE METHOD <label>
For more information about this command and the various solvers available see Chapter 21.

© 2019 ESI Group
9.2 Static Solution
9.2.3 Analyze Check- Rerun

Normal Solution
This corresponds to the following sequence:
SOLVE (STIFFNESS) (MATRIX) (RESULT (PARTIAL))
The STIFFNESS step can be taken separately.

If the sparse method is used, the MATRIX and RESULT steps can be separate.
If the PARTIAL label is present, only displacements and reactions are computed (results are not
printed).
Rerun (Sparse Method)

Initial computation:
SEARCH DATA 1
SOLVE
SAVE MATRIX RESULTS 1
Other applied loads:
SEARCH DATA 2
SEARCH MATRIX 1
SOLVE STIFFNESS RESULTS
SAVE RESULTS 2
Computation of Forces from Displacements
SOLVE FORCES (EXTRACT)
Forces in the elements are calculated from a displacement file. No other operation is required
apart from the calculation of stresses and strains at the Gaussian points, where these exist. If the
EXTRACT label is present, displacements and forces are printed.
The MODAL and TRANSIENT labels (see 11.2.1.3.1 and 12.3.5) are used to integrate
dynamic loads in beams.

© 2019 ESI Group
9.2 Static Solution
Example:
SEARCH DATA DISPLACEMENTS 1 CARD 10

SOLVE FORCES
SAVE DISPLACEMENTS FORCES 2
Results and File

The SOLVE command creates local DISPLACEMENTS, REACTIONS, and FORCES files
for all analysis options.
For continuous media options and the SHELL SPATIAL option (elements 2006 and 2008),
stresses and strains at the Gaussian points are in a STRESSES file, for which an
AVERAGING procedure (see Chapter 13.2) can be used. Local STRESSES at NODES and
STRAINS at NODES files are created for continuous media options, and can be stored in the
RESULTS file. Refer to Chapter 13.2.8 for shell elements.
Example: 3D continuous media
SEARCH DATA 1
SOLVE
SAVE STRESSES 1
AVERAGING STRESSES STRAINS
ELEMENTS DIMENSION 3
RETURN
SAVE RESULTS 1

© 2019 ESI Group
9.3 Creation of Elementary Matrices
9.3 CREATION OF ELEMENTARY MATRICES

The four elementary matrices are stored in sequential files. These files can be made permanent
with the SAVE command, for subsequent use by the SOLVE, DYNAMIC or TRANSIENT
ELASTIC routine. The four matrices are:
l STIFFNESS (file 13) : stiffness K and right-hand sides F
l MASS (file 23) : distributed and concentrated masses M
l DAMPING (file 33) : distributed and concentrated damping C
l BUCKLING (file 33) : geometrical damping Kσ
9.3.1 Standard Utilization

Input Data
SEARCH DATA i
SOLVE (STIFFNESS) (MASS) (DAMPING or BUCKLING)
SAVE (STIFFNESS) (MASS) (DAMPING or BUCKLING) j
The DAMPING and BUCKLING labels cannot coexist.

If element damping is defined in the material properties of the elements, using the ALPHA and
BETA labels (C = α K + βM, see Chap. 4.5), the full command must be used:
SOLVE STIFFNESS MASS DAMPING
Buckling matrices are computed from displacements for 3D continuous media and shell
elements, and from forces for beam elements. In axisymmetric shell option, buckling matrices
are computed from forces contained in the STRESS file.
Example 1: normal utilization
SOLVE
SOLVE BUCKLING

© 2019 ESI Group
Example 2: shell elements or continuous media – displacements obtained by non-

linear computation
SEARCH DATA DISPLACEMENTS 1 CARD j

(MODIFY ; inclusion of large displacements matrix
COORDINATES 1. 1 ; (KL) in Kσ
RETURN)
SOLVE STIFFNESS BUCKLING
DYNAMIC
MODES BUCKLING …
…
Notes:
l If the structure contains at the same time beams and shells elements, it is necessary to do SOLVE
STRESSES before SOLVE STIFFNESS BUCKLING.
l If the imposed load used for non linear computation contains temperatures or imposed strains, the
user must verify the coherence between displacements and the load used to compute buckling
matrices (time function under SOLVE are not taken into account).

© 2019 ESI Group
9.3.2 Utilization of Elimination Methods

If the data contain linear equality relations between displacements or mechanisms which will be
processed using an elimination method (see 4.10.2), the elementary matrix files contain matrices
reflecting the imposed kinematic relations, in addition to the physical element matrices.
These fictitious elements are created automatically, and a certain degree of memory space and
computation time optimization can be achieved by defining the supplementary parameters
described in § 4.10.2.
SOLVE ELIMINATION (FILE or RATIO n1) ELEMENT n2 FACTOR n3
Note
l Elimination methods are more efficient with sparse mode matrix storage. These methods are not
available for non-symmetrical matrices. The EXTRACT 100 label can be used to obtain more fully
detailed printout.

© 2019 ESI Group
9.4 Harmonic Loading
9.4 HARMONIC LOADING

The technique of Fourier series decomposition (see Chapter 8.5) requires the solution of several
linear systems, load cases being arranged by harmonic number. Syntax is as follows:
SOLVE HARMONIC
In this case, it is not possible to execute individually the calculation steps of the solution
process.
l The file STIFFNESS holds for each element as many stiffness matrices and load vectors as
there are harmonics to analyze.

l During equation solution, each harmonic is analyzed in turn. The file MATRIX is therefore
modified for each harmonic, and a further contribution is made to the results file.
l During the calculation of forces, each harmonic is analyzed in turn, by selection of the
relevant data from the displacement file.

When the analysis is complete, data on the results files is arranged by load case (i.e. harmonic).
Notes
l It is also possible to analyze one harmonic at a time using the command SOLVE, provided in each
analysis the type and number of the harmonic is stated in the OPTION command.
l Relations defined by the command RELATION may only be present if harmonics are analyzed one
at a time as described in the preceding note.

© 2019 ESI Group
9.5 Thermoelastic Analysis
9.5 THERMOELASTIC ANALYSIS

9.5.1 Basis
In thermoelastic analysis, it is supposed that temperature cards, which are required during the
solution procedure (in the calculation of stiffnesses and forces) are already specified.
These temperature cards may be defined in two ways:
l in the input data of the problem under the command DEFINITION (under the command
TEMPERATURE). These cards may be used directly by SOLVE, without further
manipulation.
l by the results of a previously executed thermal analysis (steady state or transient, etc ...).
In the latter case, results of the thermal analysis must be transformed into temperature cards
before the routine SOLVE is called.

© 2019 ESI Group
9.5.2 Temperature File

This file is processed in two ways.
a) Automatic processing
If there are no temperature cards in the input data file, the program looks for a file of
temperatures. If no such file is found, then it is created by copying the contents of the
displacements file currently in existence.
b) TEMPERATURE command
The syntax is as follows:
TEMPERATURE (GENERATE (CARD (list)))
This command transforms a displacement file, assumed to hold temperature values, into a file of
temperature cards which may be used in thermoelastic analysis and eventually creates the
corresponding thermoelastic loads. The number of temperature cards is equal to the number of
loads of the initial displacement file.
GENERATE This label is used to chain a thermal transient analysis with a
thermoelastic analysis by generating thermoelastic loads from thermal
transient file. This label can also be used with a result file issued from a
thermal static analysis.
CARD Label CARD defines a list of the thermal computation cards to be taken
into account. Each card generates a load in the DATA file.
The command TEMPERATURE must precede the associated SOLVE command.
It is supposed that a displacement file (temperatures previously calculated by a thermal analysis)
which is compatible with the thermoelastic analysis (i.e. is for the same mesh) is currently in
existence.
Note In automatic processing of a temperature file,
l the file NODAL STRESS must not have been previously accessed or created, as it would be treated
as a file of temperatures,
l the temperature data held on the displacements file must be compatible with the thermoelastic
analysis.
c) TEMPERATURE TRANSIENT SHELL command
The results of a non-linear thermal transient analysis on a structure meshed with thermal shell
elements (code 24XX) with the THERMAL ENTHALPY option (see Heat Transfer
Reference manual - Chapter 6), can be used for a thermomechanical analysis with the
SPATIAL SHELL option.

© 2019 ESI Group
The interface procedure linearizes the temperature field in the thickness of the element
(computation of mean temperature and temperature gradient in the thickness) and computes the
mean temperature at the nodes of the elements. Input data is as follows:
TEMPERATURE TRANSIENT SHELL LINEAR CARD <list>
This procedure requires the DATA input file of the mechanical analysis and the TRANSIENT
input file of the thermal analysis. Thermomechanical loads must be defined by the usual syntax:
THERMAL i.
Label LINEAR obtains linearization of the temperature field in the thickness of the element.
Label CARD defines a list of the thermal computation cards to be taken into account. Each
card generates two loads, the first corresponding to mean temperature, and the second to the
temperature gradient of each node.

© 2019 ESI Group
9.5.3 Examples
Thermal Static Analysis followed by a Thermoelastic Analysis
The two following input data illustrate the use of the command TEMPERATURE. The
difference between the two input data is emphasized by the use of bold characters.
DEFINITION DEFINITION
THERMAL TEST THERMAL TEST
OPTION THERMAL PLANE OPTION THERMAL PLANE
RECTANGLE RECTANGLE
X Y 10 I J 10 X Y 10 I J 10
MATERIAL PROPERTIES MATERIAL PROPERTIES
M / KX KY 10 M / KX KY 10
CONSTRAINTS CONSTRAINTS
LINE / TT LINE / TT
LINE YY 10 / TT LINE YY 10 / TT
LOAD LOAD
1 TEMPERATURE / NOTHING 1 TEMPERATURE / NOTHING
LINE YY 10 / TT = 200 LINE YY 10 / TT = 200
RETURN RETURN
SOLVE SOLVE
SAVE DATA RESU 1 SAVE DATA RESU 1
DEFINITION DEFINITION
THERMOMECHANICAL TEST THERMOMECHANICAL TEST
OPTION TWO-DIME TRANSLATION OPTION TWO-DIME TRANSLATION
THERMO THERMO
RESTART GEOMETRY RESTART GEOMETRY
MATERIAL PROPERTIES MATERIAL PROPERTIES
E / E 210000 NU 0.3 LX LY LZ E / E 210000 NU 0.3 LX LY LZ
0.001 0.001
CONSTRAINTS CONSTRAINTS
LINE/ UX UY LINE/ UX UY
LOAD LOAD
1 FORCE / NOTHING 1 FORCE / NOTHING
LINE YY 10 / FY -10 LINE YY 10 / FY -10
2 THERMOELASTIQUE / THERMAL 1 RETURN
RETURN
SEARCH RESU 1 SEARCH RESU 1

TEMPERATURE TEMPERATURE GENERATE
SOLVE SOLVE
SAVE DATA RESU 3 SAVE DATA RESU 3

© 2019 ESI Group
The thermal input data is followed by a static thermal analysis. In the thermo-mechanical input
data you can notice:
l Two loads in the first case: a mechanical load plus a thermal load on which it is specified that
the thermal distribution will be retrieved from a thermal result file. If this result file has several
load cases it would be necessary to define a line for each load.
l Only the mechanical load is defined in the second case.
The command TEMPERATURE precedes the command SOLVE, and does not have the same
action:
l in the first case, the temperature distribution is retrieved from the result file (RESU 1) and is
associated with the second load.

l in the second case, the label GENERATE specifies to the program to create the second load
as a thermal load. The temperature distribution is also retrieved from the result file.
Thermal Transient Analysis followed by a thermoelastic analysis

The following data input processes 200 thermoelastic loads automatically created from the
thermal results of a transient file. The command TEMPERATURE GENERATE adds 200
loads.
201 load cases (the first is the mechanical load) will be processed by the SOLVE procedure.

© 2019 ESI Group
DEFINITION
THERMAL TEST
OPTION THERMAL PLANE
RECTANGLE
X Y 10 I J 10
MATERIAL PROPERTIES
E / KX KY 10
CONSTRAINTS
LINE / TT
LINE YY 10 / TT
LOAD
1 T VARIABLE / NOTHING
LINE YY 10 / TT = 1 FT 1
TABLE
1 / 1 0 0 1000 1000
RETURN
TRANSIENT NONLINEAR
ALGORITHM OPTIMIZE -1000
PRECISION ABSOLUTE
TIME
200 STEP 1 / STORE 1
RETURN
SAVE DATA TRAN 1

...

© 2019 ESI Group
DEFINITION
THERMOMECHANICAL TEST
OPTION TWO-DIM TRANSLATION THERMOELASTIC
RESTART GEOMETRY
MATERIAL PROPERTIES
E / E 210000 NU 0.3 LX LY LZ 0.001
CONSTRAINTS
LINE / UX UY
LOAD
1 FORCE / NOTHING
LINE YY 10 / FY -10
RETURN
SEARCH TRAN 1
TEMPERATURE GENERATE CARD 1 TO 200
SOLVE
SAVE DATA RESU 3

© 2019 ESI Group
9.6 Contacts for Linear Static Analysis
9.6 CONTACTS FOR LINEAR STATIC ANALYSIS

9.6.1 Object
The CONTACT procedure is used to model contact problems, without friction, only by using
static linear results.
For this purpose, one must add new loads to initial loads of the problem (the last ones being
called "real" loads).
There will be one additional load for each contact point modeled. Contact can happen between
two nodes of a structure, or between one node and the environment.
Contact between one node and the external environment
To model such a contact, the user must define:
l contact axes, specifying local axes on the node,
l one load, which will be a unit force at the node. Its direction is that of the contact force, when
it exists (see Figure 9-2).

Contact between two nodes of the structure
Here, the user must define:
l contact axes, specifying local axes on the two nodes,
l one load, which will be a unit force at each node. Its directions is that of the contact forces,
when it exists (see Figure 9-2).

The SOLVE procedure must be applied to give results which will be used by the CONTACT
procedure.
Effects of any combination of initial loads can be calculated, and results can be stored.
Figure9-2 Figure 9-3

© 2019 ESI Group
9.6.2 Files
Input DATA,
l initial problem loads ("real" loads),
l contact unit loads.
RESULTS,
l file obtained with the previous data (or coming from a "MERGE" procedure).
l output RESULTS, if storage has been required.

© 2019 ESI Group
9.6.3 Data Description

CONTACT ( EXTRACT <no > )
LOAD <list of numbers>
GAP <no> <value> (<no>, <value> ...)
GAP ..
CALCULATE ( PRINT ) (STORE <types of RESULTS> )
<combination of "real" loads>
...
<combination of "real" loads>
RETURN
CONTACT Command
This command initializes the procedure.
EXTRACT controls extraction level:
<no> =0 or 1 extraction of values introduced by the user under CONTACT procedure.2
(default) extraction of RESULTS used by CONTACT procedure.
LOAD Command
<list of numbers> list gives numbers of contact loads.
Note
l When the number of the contact loads is upper than 200, the SIZEn label on the CONTACT
command, with n= number of contact loads, must be used.
GAP
Optional command used to introduce gaps.
<no> is the number of the contact load.
<value> is the value of this contact gap.
CALCULATE Command
Defines combinations of the "REAL" loads whose effect must be calculated.
PRINT prints results during computation
STORE specifies that a local result file will be created at the end of the
CONTACT procedure.
<combination of "real" loads> = <title> / <value> <no> ...
<no> number of the "real" load
<value> load coefficient in the current combination.

© 2019 ESI Group
9.6.4 Example

© 2019 ESI Group

© 2019 ESI Group
CHAPTER 10 SUBSTRUCTURING
10.1 Introduction
10.1 INTRODUCTION
10.1.1 Terminology
The finite element method leads to the subdivision of the structure to be analyzed into
subdomains of simple geometric form (triangle, quadrilateral, tetrahedron etc ...) known as finite
ELEMENTS, which are interconnected by NODES. The purpose of defining a mesh in this
way is to enable the construction of a discrete system of equations describing structural
behavior, by means of integration over regions of simple geometric form.
An alternative representation which corresponds more closely to reality is that in which the
structure is considered as an assembly of components of arbitrary geometric form known as
SUBSTRUCTURES, each of which may itself be an assembly of lower level substructures.
l In the following discussion, by LEVEL ZERO SUBSTRUCTURES, or
SUPERELEMENT, is meant a part of the structure which is modeled in the usual way by
elements, with which are associated nodes, material properties etc ...
l A LEVEL 1 SUBSTRUCTURE is a part of the structure formed by the assembly of several
SUPERELEMENTS. In the same way, a LEVEL N SUBSTRUCTURE consists of several

lower level substructures of which at least one must be of level n - 1.
l A substructure is connected to a higher level substructure by a set of nodes known as the
PRIMARY NODES or INTERFACE of the substructure, the remaining nodes being known
as SECONDARY NODES.
The representation of a structure so defined may be symbolized by a tree in which there are
links between substructures of different levels.
Note that the highest order substructure, which represents the complete structure, has no
primary nodes. In the following text, SE denotes substructure and SE(ij) substructure i, of
level j.

© 2019 ESI Group
10.1 Introduction
10.1.2 Example
In Figure 10.1 are shown two substructures of level zero (types 1 and 2).
+ secondary node
0,+ primary node
The assembly of two substructures of type 1 and of two others of type 2 gives two substructures
of level 1 as shown in Figure 10-2.
The assembly of the substructures SE100 and SE200 so obtained, together with a third
substructure of level zero, forms the level 2 substructure which represents the structure shown in
Figure 10-3.
Figure 10-1
Figure 10-2

© 2019 ESI Group
10.1 Introduction
Figure 10-3

© 2019 ESI Group
10.2 Method of Analysis
10.2 METHOD OF ANALYSIS

Substructuring may only be used in static linear analysis. In this case it is necessary to solve for
each substructure a linear system of the form:
K.U = F (1)
where :
K : stiffness matrix of the substructure
U : nodal displacement vectors
F : load vectors (equivalent nodal forces)
Separation of primary nodes (p) and secondary nodes (s) gives the following system:
Where
Kss . Us + Ksp . Up = Fs (2)
Kps . Us + Kpp . Up = Fp (3)
Elimination of degrees of freedom at secondary nodes gives the following system of equations:
(4)
(5)
(see Figure 10-4)

The matrix K'p so obtained is the stiffness matrix of the substructure with respect to its primary
nodes. A procedure identical with that used to construct the linear system (1) from basic
stiffness matrices and load vectors is used to assemble the matrices and vectors K'p and F'p of
substructures comprising a higher level substructure into a new system of linear equations in
terms of unknowns at primary nodes of the substructures only (Figure 10-5).
Partial eliminations and assemblies are repeated until the linear system for the highest level
substructure is obtained ; this system is then bifactorized.
It is now possible to compute nodal displacements of the highest level substructure, then of all
lower level substructures according to equation (4) (Figure 10-6).

© 2019 ESI Group
10.2 Method of Analysis
This procedure finally results in the calculation of displacements at all nodes of zero level
substructures. Forces and stresses are then computed from these displacements in the same way
as in analysis without substructuring.
Figure 10-4
Figure 10-5
Figure 10-6

© 2019 ESI Group
10.3 File Management
10.3 FILE MANAGEMENT

10.3.1 Description of Files
The solution of a problem by substructuring is subdivided into a number of basic steps. Each of
these steps provides intermediate results which are automatically saved as permanent files for
use in later stages of the analysis. Files are associated with various substructures as shown in the
following table:
TYPE CONTENTS OF FILE SE (°) SE(j) SE(n)
1 Input data X X X
2 Results X
3 Stiffness X
4 Matrix X X X
5 Load vectors (*) X X
6 Displacements X X X
SE(°) : zero level substructures
SE(j) : substructures of level between 1 and n - 1
SE(n) : the highest level substructure (level n)
* : file type 5 exists only if loads are applied to the substructure.
Input data
This file is the result of the interpretation of the user's input data defining the substructure.
Results
This file exists only for a zero level substructure. It holds the results of the analysis and is of a
structure identical with that of a results file created during an analysis without substructuring,
and so may be used by all the postprocessing routines of SYSTUS.
Stiffnesses
This file also exists only for a zero level substructure. Two types of data are stored on the file:
l the basic stiffness matrices Ki, which depend upon the physical properties of the substructure.
l the basic load vectors Fi and initial strains SSi which depend upon applied loads.
Matrix
This file is created after the elimination of unknowns at secondary nodes ; it consists of two
parts :
l the matrix Kss-1.Ksp (see equation 4, paragraph 10.2) required in the calculation of
displacements at secondary nodes from those at primary nodes.

l the stiffness matrix K'p (see equation 5, Chapter 10.2) of the substructure.
Load vectors
The structure of this file is the same as that of the matrix file. On it are held the load vectors of
equations 4 and 5, Chapter 10.2.

© 2019 ESI Group
Displacements
Solution of the system of equations gives displacements of nodes of each substructure ; these
are saved on the displacements file.

© 2019 ESI Group
10.3.2 Calculation Steps

The subdivision of the analysis of a problem corresponds to the stages of the analysis described
in Chapter 10.2. Every calculation step is performed according to the following standard
scheme:
l data necessary to the calculation is supposed to be held on a certain number of files known as
INPUT FILES.
l the procedure consists of the creation, using data read from input files, of one or more files of
results known as OUTPUT FILES (see Figure 10-7).
Figure 10-7
The purposes of this mode of operation are to enable modular programming and also to permit
the execution of the analysis in several stages, with the possibility of stopping and later
restarting the calculation at the beginning of any of the following calculation steps.
Calculation of zero level substructure stiffnesses
Input file : DATA
Output file : STIFFNESS
Calculation of zero level substructure matrix
Input files : DATA and STIFFNESS
Output files : MATRIX and LOAD VECTOR
Calculation of matrix of substructure of level greater than or equal to 1
Input files : DATA for the substructure whose matrix is to be calculated.
DATA
MATRIX
for each associated lower level substructure
LOAD
VECTOR
Output files: MATRIX and LOAD VECTOR
For the highest level substructure, the output files are MATRIX and DISPLACEMENTS.
Calculation of displacements of a substructure
Input files : DATA, MATRIX, LOAD VECTOR.
DATA and DISPLACEMENTS of the associated higher level substructure.
Output files : DISPLACEMENTS

© 2019 ESI Group
Calculation of results for a zero level substructure

Input files : DATA, STIFFNESS, DISPLACEMENTS.
Output files : RESULTS

© 2019 ESI Group
10.4 Optimization of an Analysis
10.4 OPTIMIZATION OF AN ANALYSIS

The method of substructuring described in this chapter is in fact a partitioning of the structure to
be analyzed, resulting in the partitioning of the system of equations to be solved. There is no
change in the number of arithmetic operations to be carried out compared with direct solution of
the entire system, and the advantage of substructuring lies in improved organization of the
calculations.
By a more detailed analysis of the method, it may be seen how in certain circumstances this
improved organization may be used to achieve significant economies in computing cost.
10.4.1 Re-Use of Substructures

It often happens that after the initial analysis of a structure minor modifications of input data and
a second analysis are required, either because modifications of the structure are to be
considered, or simply because there were errors in the input data.
In such a second analysis, the substructuring routines automatically re-use intermediate results
of the initial analysis which are unaffected by the modification.
Figure 10-8
In the example shown in 10.4.1 Re-Use of Substructures, substructure 4 is modified after the
initial analysis. The files STIFFNESS, MATRIX and LOAD VECTOR of all other
substructures are re-used.
It may be seen from this example that by judicious subdivision of the structure, a significant
amount of computer time may be saved. The way in which substructures are assembled is also
of importance, as is shown by the example above in which only substructures 4 and 100 are
affected by the modification.
The alternative means of assembly represented by the tree shown in 10.4.1 Re- Use of
Substructures would have resulted in all substructures on the path linking substructure 4 to the
highest level substructure 100 being affected :

© 2019 ESI Group
Figure 10-9

© 2019 ESI Group
10.4.2 Change of Axes

In the analysis of complex structures, subdivision into substructures simplifies the development
and checking of input data. Specification of the problem may sometimes be further simplified
by defining each substructure with respect to the most appropriate system of axes, and then
moving it into its intended position in the complete structure by one of the following geometric
transformations:
l rotation,
l symmetry with respect to a point,
l symmetry with respect to a straight line,
l symmetry with respect to a plane.
With each geometric transformation is associated a matrix R and its transpose R*, which are
required in the transformation of stiffness matrices K and load vectors F :
K' = R* . K . R
F' = R* . F
Results of the analysis are expressed in terms of the axes used to define each substructure.
Notes
l Translation of axes has no effect on the analysis.
l For axisymmetric structures, geometric transformations are not possible.
l In the analysis of heat flow, geometric transformations have no effect on the analysis.

© 2019 ESI Group
10.4.3 Re-Use of Files

Advantage may be taken of the following factors.
l If two substructures are identical (geometry, material properties, constraints, loads, etc ), their
STIFFNESS files are identical.
l If in addition the arrangements of primary nodes are identical, then their MATRIX and
LOAD VECTOR files are also identical.

If one of these situations exists, then the user may avoid the repetition of identical calculations
by re-using files associated with a substructure.
An appropriate subdivision of the structure often results in substructures which can be obtained
one from another by the geometric transformations of the preceding paragraph, and also by re-
use of files.
Example:
Let us take the structure considered in the example of Chapter 10.4.1 and subdivide it into 5
substructures as shown in Figure 10-9.
Substructures 1, 4 and 5 are identical to within a rigid body rotation, and substructures 2 and 3
are symmetrical with respect to a line coinciding with their interface. Account is also taken of
symmetry in the analysis of substructures 12 and 43 where:
SE12 = SE1 + SE2
SE43 = SE4 + SE3
Substructure 43 is assembled by symmetry with respect to a straight line coinciding with the
interface of substructures 2 and 3, and 5 by a rotation through 90 degrees (see Figure 10.12).
This construction obviates the need to create the STIFFNESS files of substructures 3, 4 and 5
and MATRIX files of substructures 3, 4, 5 and 43.
Figure 10-10 Figure 10- 11 Figure 10-12

Note
l The re-use of files is an option in which it is entirely the user's responsibility to check that
substructures are identical as required by the analysis ; the solution routines are not capable of
carrying out any checks on input files.

© 2019 ESI Group
10.4.4 Interfacing of Different Types of Structure

SYSTUS recognizes two types of structure with 6 and 3 degrees of freedom per node for the
analysis of structures in three dimensions. In the modeling strategies described in earlier
chapters, these relatively costly schemes have been mandatory even when the behavior of
certain parts of the structure could adequately be modeled by a type of structure with fewer
unknowns per node.
The substructuring routines of SYSTUS are capable of analyzing problems in which
substructures are of different structure types. The type of structure at nodes shared by more than
one substructure is chosen by the substructuring routines as indicated in the following table.
OPTIONS 1 2 3 4 7
1 1/3 3 3 3 1/3
2 3 2/3 3 3 3
3 3 3 3 3 3
4 3 3 3 4 4
7 1/3 3 3 4 7/4
where:
Option 1 = all types of structure with d.o.f. UX, UY, RZ
Option 2 = all types of structure with d.o.f. UZ, RX, RY
Option 3 = all types of structure with d.o.f. UX, UY, UZ, RX, RY, RZ
Option 4 = all types of structure with d.o.f. UX, UY, UZ
Option 7 = all types of structure with d.o.f. UX, UY
Combinations of two plane structure types (1,1), (2,2), (7,7) and (1,7) result in nodes of
structure type 3 or 4 if the substructures are not coplanar.
The equations which given displacements of nodes shared between substructures are
constructed from the stiffness matrices of structure types in which account is not taken of all
possible displacements at the node.
Even if the applied loads do not induce displacements unrepresented in the substructures, it is
necessary in order to enable the numerical solution to be carried out to stabilize the structure by
applying fictitious constraints, which have no effect on the results of the analysis.

© 2019 ESI Group
Figure 10-13
Example:
The structure shown in Figure 10- 13 consists of a plate (SE10) loaded in the direction
perpendicular to its plane, connected to another plate (SE20). Under these conditions SE10 is
analyzed as a plate subject to bending (option 2), SE20 as a plate subject to plane stress
(option 7), and the shared nodes as if they were nodes of structure type shell (option 3).
The degrees of freedom of the plate SE10 correspond to the unknowns UZ, RX and RY. The
degrees of freedom of the plate SE20 subject to plane stress are UX and UY, or UY and UZ
after transformation of coordinates.
Equations are therefore obtained at shared nodes in terms of the unknowns UY, UZ, RX and
RY. It is therefore necessary to introduce a third substructure the nodes of which are those
shared by SE10 and SE20, and are subject to fictitious constraints in the directions UX and RZ.
This modeling strategy divides the size of the problem by at least 2 by calculating at most nodes
either 2 or 3 unknowns instead of 6, and results in a significant reduction of computing time. It
should also be noted that the analysis of an assembly of plane membranes in space as a three
dimensional shell requires the stabilization by fictitious constraints of all nodes not located at the
intersection of several membranes.
The possibility of locally representing each membrane as a plate subject to plane stress therefore
results in both a reduction in computing time and simplified data preparation.

© 2019 ESI Group
10.5 Input Data
10.5 INPUT DATA

10.5.1 Zero Level Substructure (superelement)
The input data defining a zero level substructure is identical with those for an analysis without
substructuring, and may be specified in either of two ways.
l Under the command DEFINITION with certain additional parameters. The command is
followed by the input data normally read by the routine DEFINITION.

l Under the command SUPER ELEMENT, in which case subsequent data is read from a local
input data file either previously created by the routine DEFINITION or copied from a
permanent file by the command SEARCH DATA.
DEFINITION SUPER ELEMENT n (FILE) i (TRANS) (R)

SUPER ELEMENT n (FILE) i (TRANS) (R)
In the command DEFINITION, the presence of the label SUPER ELEMENT distinguishes
input data of a substructure from that for a problem without substructuring.
n number of the superelement. (1 < N < 9999)
(FILE) i suppresses the creation of the specified output files and all associated
computation, where existing files are to be re-used. Permissible labels are
STIFFNESS, MATRIX and LOAD VECTOR.
i is the number of the file
(TRANS) (R) geometric transformation to be applied to data held on the files MATRIX
and LOAD VECTOR of the superelement, during assembly of a higher
level substructure.
Possible geometric transformations and associated labels are as follows

l Rotation defined by 3 angles ψ, θ, ϕ :
(TRANS) (R) = ROTATION or AXES PSI = ψ THETA = θ PHI = ϕ

l Symmetry with respect to a plane defined by 3 angles ψ, θ, ϕ (Figure10.14)
(TRANS) (R) = SYMMETRY PLANE PSI = ψ THETA = θ PHI = ϕ

l Symmetry with respect to a straight line defined by 2 angles θ, ϕ (Figure 10-15)
(TRANS) (R) = SYMMETRY LINE THETA = θ PHI = ϕ

l Symmetry with respect to the origin of the system axes (no parameters)
(TRANS) (R) = SYMMETRY POINT

It is possible to specify in addition to these transformations a coordinate translation which has
no effect on the analysis but is referred to by the routine ASSEMBLE (see Chapter 6.3.5), and
by the Automatic link generator (see Chapter 10.5.4).
The three components of the translation are specified as follows:
XX = x YY = y ZZ = z

© 2019 ESI Group
10.5 Input Data
Example:
DEFINITION SUPER ELEMENT 10 STIFFNESS 10
standard input data
RETURN
SUPER ELEMENT 11 STIFFNESS MATRIX LOAD VECTOR 10
SUPER ELEMENT 12 STIFFNESS 10 SYMMETRY LINE THETA =45
SUPER ELEMENT 13 STIFFNESS 10 ROTATION PSI = 90
Figure 10-14

© 2019 ESI Group
10.5 Input Data
Figure 10-15
Note
l for θ rotation round the z axis,
l for ϕ rotation round the y axis,
l This is contrary to common practice.

© 2019 ESI Group
10.5 Input Data
10.5.2 Level n Substructure

The input data for a substructure of a level greater than zero consists essentially of a definition
of links with associated lower level substructures:
SUBSTRUCTURE n (FILE) i (TRANS) (R)
COMPONENTS ei ej ek ...
EQUIVALENCE (ei fi) (ek fk) ...
CONNECTION
n/(ni ei) (nj ej) (nk ek)
...
LOAD
1 TITLE OF LOAD CASE
ci ei cj ej ....
2 ...
RETURN
On the command SUBSTRUCTURE is given the same data as for the definition of a zero level
substructure, that is:
n number of the substructure
(FILE) i suppresses the creation of the specified output files and all associated
computation, where existing files are to be re-used. Permissible labels are
MATRIX and LOAD VECTOR.i is the number of the file.
(TRANS) (R) geometric transformation as described in Chapter 10.5.1.
The command SUBSTRUCTURE is followed by the command COMPONENTS in which are

specified the numbers of associated lower level substructures ei ej ek ...
In the case of using the automatic link generator (see Chapter 10.5.4), it is compulsory,
for a given substructure at any level, to have one "father" and only one.
The manner in which associated substructures are interconnected is defined by the command
CONNECTION followed by groups of data. Each group of data consists of the following
information:
n external number of node (n<32000)
(/) slash
(ni ei) external number ni of node, and number ei of substructure to which it
belongs.

© 2019 ESI Group
10.5 Input Data
Load cases are each specified as the sum of several load cases of associated substructures.
The command LOAD is followed by a series of pairs of lines of input data.
The first line of a pair begins at the first character with the load case number, which must be
sequential. The load case number is followed by the title of the load case.
On the second line of a pair, which must begin with a blank in column 1, are specified pairs of
numbers (ci ei) where ci is the number of the load case and ei that of the substructure, for each
contribution to the total load that is to be taken into account. A substructure number ei may
appear one time only on this line.
Where the input data defining connections and/or load cases of several substructures are
identical apart from the numbers of associated substructures, commands need not be repeated if
preceded by the command EQUIVALENCE.
The command EQUIVALENCE defines a correspondence between the number of an
associated substructure and the number referred to in the data defining connections or load.
In this command therefore are specified pairs of values (ei fi) where ei is the number of the
substructure for which connections or loads are to be defined, and fi is the number by which the
substructure is identified in subsequent data.
The command RETURN terminates the input data defining the substructure.
Notes
l If no error is found by the routine, the input data file of the substructure is automatically saved.
l The permanent input data files of all associated substructures are updated by the addition of data
essential for the execution of later stages of the calculation.
l Each substructure is assigned a status code which has the following signification:
l 0: isolated substructure (not connected to any other)
l 1: connected substructure
l 2: STIFFNESS file created
l 3: MATRIX file created
l 4: LOAD VECTOR file created
l 5: DISPLACEMENTS file created
l 6: RESULTS file created (end of analysis)
l In command CONNECTION a couple (ni, ei) can appear only once.
l There "waiting" connections can exist with only one couple (ni, ei), which will be used to define a
higher level substructure.
l For the higher level substructure loads, the edition labels (Chap.4.8.2) can be used. They will be
taken into account by editing the results of every superelements of the substructure.

© 2019 ESI Group
10.5 Input Data
Example:

© 2019 ESI Group
10.5 Input Data
10.5.3 The Command CALCULATE

Once all the substructures have been defined, the solution may be obtained using the command
CALCULATE. The user need only specify which stage of the calculation it is desired to reach
and the SYSTUS system will automatically carry out all necessary intermediate calculation to
achieve the desired result. A complete analysis by substructuring consists of the following steps:
l calculation of basic stiffness of zero level substructures (superelements),
l assembly and triangularization of stiffness matrices of zero level substructures,
l assembly and reduction of load vectors of zero level substructures,
l assembly and triangularization of stiffness matrices of higher level substructures, in ascending
order of level,
l assembly and reduction of load vectors of higher level substructures, in ascending order of
level. For the highest level substructure, this operation yields nodal displacements,
l calculation of nodal displacements of each substructure, in descending order of level,
l calculation of forces or stresses in zero level substructures.
Caution:
l Before using CALCULATE command data must be load in memory.
The syntax of the command CALCULATE which enables these operations to be carried out is
as follows:
CALCULATE (STEP) n
(STEP) is a label specifying the stage of analysis it is desired to reach for substructure n as
follows:
STIFFNESS calculation of stiffnesses (zero level substructure)
MATRIX calculation of triangularization matrix and reduced load vectors
FORCE or calculation of forces or stresses (zero level substructure)
RESULTS
Each calculation step may be requested individually by including in the command
CALCULATE the label ONLY. In this case, the user must check that all intermediate files
(input files) required during the step are available. Syntax is as follows:
CALCULATE (STEP) n ONLY
(STEP) is a label chosen from one of the following lists.

Zero level substructure
STIFFNESS calculation of basic stiffness matrices
MATRIX calculation of triangularized matrix and reduced load vectors

© 2019 ESI Group
10.5 Input Data
MATRIX calculation of reduced load vector

PARTIAL
DISPLACEMENTS calculation of displacements
FORCE or calculation of forces or stresses
RESULTS
Higher level substructure
FRONT calculation of triangularized matrix and reduced load vectors
FRONT PARTIAL calculation of reduced load vectors
Highest level substructure
FRONT calculation of triangularized matrix and displacements
Notes
l Permanent input data files of substructures are updated upon completion of calculation steps, so
that a record of the status of each substructure is maintained. These files must not be modified by
the user during the solution process.
l If the requested sequence of calculation is terminated prematurely because for example the job
time limit is exceeded or the system goes down, only the calculation step in progress at the time
the job aborted is lost. If the same job is later resubmitted, the calculation is automatically
restarted at the beginning of the step that was aborted.
l Calculation steps that create files that are to be re-used by several substructures must be executed
first.
l If thermal loading is applied to zero level substructures by using a file of temperatures, then the
calculation of stiffnesses and forces or stresses must be requested individually.
l For a zero level substructure, the COMPUTE DISPLA n (ONLY) generates a DISPLA n file
necessary to forces computation. Physic displacements are then contains in the RESU n file.
Examples:
SEARCH TEMPERATURE 1
CALCULATE STIFFNESS 10
SEARCH TEMPERATURE 2
CALCULATE FORCES 20

© 2019 ESI Group
10.5 Input Data
10.5.4 Automatic Link Generator (A.L.G.)

Purpose
One of the main difficulties is using the substructures method is inter-element links generation.
The automatic link generator (A.L.G.) determines connections between superelements of a
structures (passive and active connections) and renumber then in a coherent way. The A.L.G.
operates according to the principle of node geometric coincidence, with a tolerance defined by
the EPSILON parameter.
Connections are generated moving up the substructure from the zero level superelements to the
top of the tree. Geometric transformation of a superelement and a substructure from which it
belongs to, are added.
Remark
l Before using the automatic link generator, the whole substructure must have been defined using
standard SUBSTRUCTURE commands (SUPERELEMENT, SUBSTRUCTURE and their
subcommands).
l This can have been done in a previous session.
File
As with standard substructure commands, input and output files are automatically managed.
The user must never modify these files.
Input file
l DATA: data files of every superelement of the substructure
Output file
l DATA: modified data files: connections are now defined.
l LINK: eventually alphanumerical files where connections are described in the SYSTUS
input data format (command SUBSTRUCTURE, subcommand CONNECTION)
Input Data
CONNECTION <number>
LINK EPSILON <epsilon> FILE
RELEASE
n/p
CHECK (ONLY) <number>
RETURN
CONNECTION command
This is the first level command for the automatic link generation

© 2019 ESI Group
10.5 Input Data
<number > is the number of the highest level substructure on which "connection"
will be applied. If the whole substructure has just been defined, this
number can be omitted.
LINK Command (k)
EPSILON Coupling tolerance. This is the largest distance at which two nodes are
<epsilon>: considered to be connected. Default value is equal to 10 percent of the
minimum of the leading size of the elements, for all the elements. This
leading size is the length of the longest size of an element.
The default value is intended to ensure that no nodes associated with
the same element will be connected.
FILE : specifies that connections will be described on formatted data files as
in the standard "CONNECTION" subcommand of the
"SUBSTRUCTURE" command:
CONNECTION
<NUMBER > / Ni Ei Nj Ej
with:
<number> : external number of connection
(Ni Ei): number of the Ni node of superelement Ei
- One file, whose name is BINDk.TIT, is created for each superelement and can be used in
SUBSTRUCTURE command by ASSIGN and READ command.
- These files, whose name is BINDk.TIT, can of course, be used under SUBSTRUCTURE
command, with $ASSIGN and $READ utilities.
CHECK command (k)
Extracts substructure data for the tree-node k (level, connections, etc., see 10.5.2). If no tree-
node is specified, the complete tree-structure is examined.
(ONLY) : Only one tree-node is examined.
RELEASE command
This command permits the release of a connection generated by the A.L.G.
n : tree-node number (substructure number)
p : absolute number of the connection (given by CHECK)
The named connection is released throughout the tree-structure and renumbering is carried out
by the A.L.G.

© 2019 ESI Group
10.5 Input Data
Example
The following sequence presents a simple use of the A.L.G.
12, 123, 1234 are the tree-structure nodes.

1 and 2 are the sons of 12.
12 et 3 are the sons of 123.
123 et 4 are the sons of 1234.

© 2019 ESI Group
10.5 Input Data
DEFINITION
SIMPLE EXAMPLE OF A.L.G
OPTION TWO DIMENTIONAL
GEOMETRY
NODES
1/00
2/10
3/11
4/01
ELEMENTS
1/1234
RETURN
SUPERELEMENT 1
SUPERELEMENT 2 XX = 1
SUPERELEMENT 3 XX = 1 YY = 1
SUPERELEMENT 4 YY = 1
SUBSTRUCTURE 12
COMPONENTS 1 2
RETURN
SUBSTRUCTURE 123
COMPONENTS 12 3
RETURN
SUBSTRUCTURE 1234
COMPONENTS 123 4
RETURN
CONNECTIONS
LINK
CHECK
RETURN

© 2019 ESI Group
10.6 Procedures associated with Substructure
10.6 PROCEDURES ASSOCIATED WITH

SUBSTRUCTURE
l LIST and VERIFY (see Chapters 7.2.3 and 7.2.4): verification procedures.
l EXTRACT SUBSTRUCTURE (see Chapter 7.2.5) to re-use a reduced matrix in SOLVE,
DYNAMIC or TRANSIENT.
l ASSEMBLE (see Chapter 6.3.5) to create a standard data file with nodes and elements of the
substructure.
l SPLIT (see Chapter 13.11) to chain thermal analysis (TRANSIENT) with mechanical
analysis in substructures.

© 2019 ESI Group
10.7 Conclusions
10.7 CONCLUSIONS
Substructuring gives SYSTUS increased problem solving power and greater flexibility. The
most significant advantages of the method are summarized below.
l There is no absolute limit to the size of a problem that may be analyzed (except memory
limitation, A.L.G.).
l Discretization of the structure is simplified by its subdivision into independent components,
on entry and postprocessing. Freedom to choose the system of axes most appropriate to each
substructure is an important advantage.
l Subdivision of the solution process into calculation steps leads to more efficient use of the
computer, with shorter runs and quicker recovery from incidents such as system failure.
l Economies of computing time can be very significant if advantage is taken of various
opportunities for optimization.

The best use of this facility depends entirely upon the discretization of the structure, in which
the subdivision and assembly sequences should be chosen in such a way as to maximize the
capacity for optimization of the analysis.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.1 Introduction
CHAPTER 11 DYNAMIC
11.1 INTRODUCTION
The DYNAMIC module determines eigenfrequencies and eigenmodes for a physical problem,
and computes dynamic response or sensitivity using a modal superimposition process. The
basic matrices required for the problem (STIFFNESS, MASS, DAMPING and BUCKLING)
must be computed beforehand using the SOLVE command (Chapter 9), as also the static results
which are needed in certain cases.
The computation modules can be classified in four categories:
l Eigenfrequencies and eigenmodes
l Dynamic responses
l Sensitivity analysis
l Utility modules
Chaining of the different analysis procedures is illustrated in the table below.

Eigenfrequencies and eigenmodes
Real eigenmodes and eigenfrequencies can be determined in three different physical domains,
where the characteristics of the system are represented by two matrices:
l Mechanical domain: stiffness and mass matrices.
The stiffness matrix can take account of geometrical stiffness due to a state of static stress, the
mass matrix, and added mass resulting from interaction between the structure and a non-
viscous, incompressible fluid (limited to certain computation options).
Direct analysis methods are suitable for medium size problems, and modal synthesis for large
size problems.
l Thermal domain: conductivity and thermal capacity matrices.
l Buckling domain: stiffness and geometrical stiffness matrices. Critical load factors
correspond to linear stability theory.

Complex eigenmodes and eigenfrequencies can be determined in the case of a mechanical
structure characterized by three symmetrical real matrices (stiffness, mass and damping).
Dynamic responses
Real eigenmodes define a transformation of coordinates, and can be used to write equations of
movement in a modal space, the dimension of which corresponds to the number of selected
eigenmodes. The stiffness and mass matrices are diagonal in this space, and available damping
forms lead to a damping matrix which is also diagonal. Disconnected modal equations are
particularly suitable for problems involving low or medium frequency excitation. The model is
only representative up to a certain frequency level for a given form of space discretization. A
satisfactory degree of accuracy is obtained with modal methods, with an adequate number of
modes to represent excitation and structure.
Excitation can correspond to forces or movements of the support (displacement or acceleration),
defined in time or in terms of frequency (response spectrum and power spectrum density).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.1 Introduction
A special module can be used to study harmonic response with any damping matrix.
Sensitivity analysis
This method can be used to analyze the influence of isolated stiffness, mass or damping
changes, on eigenfrequencies and the form of real eigenmodes.
Utility modules
The 4 utility modules supply relevant data to the computation modules (modal damping and
response spectra), or are used for special processing (effective mass and dynamic transient
values).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.1 Introduction

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.2 Eigenmodes and Eigenfrequencies - Direct Methods
11.2 EIGENMODES AND EIGENFREQUENCIES -

DIRECT METHODS
Direct methods are suitable for small or medium size models (a few thousand DOF).
Distinction is made between three types of problems:
l real eigenmodes,
l real eigenmodes with fluid/structure interaction,
l complex eigenmodes.
11.2.1 Real Eigenmodes and Eigenfrequencies

The problem to be solved is written in the following form:
(K - λM) φ = 0
λ, φ eigenvalues and eigenvectors
K,M symmetrical matrices (with K defined positive), the nature of which
depends on the physical problem (mechanical, thermal or buckling). For
mechanics, eigenmode is the square root of λ. For buckling, the critical
load coefficient is the inverse of λ. For thermics, the physical quantity is
for stable systems a real positive number. Refer to Chapter 11.2.1.Special
problem part for more details.
Note
l Mechanical problem is a second nature problem, λ represents the square of the common physical
size (eigencircular). Thermal and buckling problems are first nature problem. For thermal, results
given will be more physical results by treating the problem like a buckling problem. This is
possible by assignation the calorific capacities file (MASS) to 33 unit (Chapter 11.2.3 General
Part).
Description of Solution Methods

There are six methods for the identification of eigenfrequencies and eigenmodes. This number
is justified by the diversity of the problems to be handled, and as in the case of the SOLVE
command, by the existence of several matrix storage methods (spars, band and compact band
methods).
Table 1 recapitulates the main characteristics of the six methods. The optimum storage method
depends on the model, and the computer used for computation. Methods using the band storage
method create matrices in core storage. Before forming the basic matrices with the SOLVE
command, it is necessary in this case to renumber the nodes in order to minimize assembled
matrix size.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Brief description of algorithms: area of utilization

The six algorithms are classified in three main families: iterative with deflation (INVERSE
POWER), transmutation with no size reduction (GIVENS), and transmutation with
simultaneous size reduction (SUBSPACE methods). Three of the subspace methods employ
the LANCZOS technique.
a. GIVENS algorithm: CHOLESKY factorization of the stiffness matrix makes it possible to
reduce a generalized eigenvalue problem to a problem involving simple eigenvalues. The
eigenvalues of the tridiagonal matrix obtained with the GIVENS method, are computed by QR
type factorization, and the eigenvectors by inverse iterations.
This method is applicable to small systems (less than 500 DOF), and gives all eigenvalues, and
eigenvectors on request. GUYAN's elimination method, with selection on the nodes, extends
the application field for this method. Reduced matrices cannot be obtained.
b. INVERSE POWER algorithm. An iterative process of the type (K'UK+1 = MUK) is used to
obtain the greatest eigenvalue (lowest frequency) and associated eigenvector simultaneously,
from an initial vector. Deflation (matrix reduction by orthogonalization) gives eigenvalues and
eigenvectors in succession.
This algorithm is suitable for very large systems, for identification of a few simple frequencies in
a given frequency band.
c. ITERATIVE SUBSPACE algorithm: an iterative process (K'UK+1 = MUK) is used to obtain
the vector base defining a subspace, from an initial vector base. Eigenvalues and eigenvectors
are determined in this subspace using the JACOBI method. Representing a generalization of the
INVERSE POWER method, this method has much the same field of application, but can be
used to obtain a relatively large number of eigenfrequencies, which can be multiple.
d. LANCZOS SUBSPACE algorithm: the LANCZOS recurrence relations are used to obtain a
vector base, in which the problem matrix is tridiagonal, from an initial vector. This matrix is
processed as for the GIVEN's algorithm.
This method is applicable for large systems, to search for a large number of eigenfrequencies
starting at zero.
e. LANCZOS BAND SUBSPACE algorithm: similar to (d) above, but processed off-line using
the matrix storage technique. This algorithm allows utilization of GUYAN's elimination method
at DOF level, and can be used to obtain reduced mass and stiffness matrices.
f. LANCZOS BLOCK SUBSPACE algorithm: similar to (d) above, but processed off-line
using the matrix storage technique. This algorithm is more computation time greedy than (d)
above, but has three advantages:
l automatic subspace size search,
l possibility for frequency search in a frequency band,
l greater immunity to the presence of multiple frequencies.
l Furthermore, only this algorithm can be used for computation of complex eigenfrequencies
and eigenmodes.

© 2019 ESI Group
CHAPTER 11 DYNAMIC

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Initial modification of stiffness and mass matrices

The stiffness and mass matrices can be modified, prior to the eigenfrequency and eigenmode
search.
- Modification of the stiffness matrix:
In the DYNAMIC module, special commands can be used to replace matrix K by matrix K' =
K + αM. This initial modification has two main purposes:
l Processing free-free structures for which K is singular. The SHIFT label, available with all
algorithms apart from INVERSE POWER, can be used to indicate the value of α
(recommended value: 10 to 20% of where ω1 is the first elastic body circular
eigenfrequency).
l Identification of eigenfrequencies in a given frequency band. FREQUENCY (CIRCULAR
FREQUENCY) labels are used in the ITERATIVE or BLOCK SUBSPACE algorithm, and
the LIMIT label where the INVERSE POWER algorithm, to indicate the position of the
center frequency sought. Matrix K' is not necessarily defined positive (α negative),
explaining its limitation to certain algorithms. Due to their antagonistic function, the SHIFT
and FREQUENCY (CIRCULAR FREQUENCY) labels cannot coexist. In case of
buckling, the shift sign must be the sign of the coefficient, which characterizes the critical
load searched.
Ahead of the DYNAMIC module, the SUM procedure can be used to include geometrical
stiffnesses (BUCKLING file) resulting from a state of static stress, in the STIFFNESS file.
- Modification of the mass matrix:
The mass matrix is normally consistent. Experience shows that eigenvalues are obtained with
sufficient precision for large vibratory problems, if the matrix is concentrated on the diagonal.
Computation time and memory space gains can be considerable. The DIAGONAL label is
available with all algorithms (except GIVENS), and allows the use of a diagonal mass matrix.
For computation options with transnational and rotational degrees of freedom, and with certain
algorithms, the inertia values assigned to rotational degrees of freedom are then ignored (care is
required with certain local modes, such as straight beam torsion).
Utilization of the DIAGONAL label is not relevant for buckling problems.
- Simultaneous modification of stiffness and mass matrices:
GUYAN's elimination method, activated by the SELECT or SUPPRESS label, can be used to
assign stiffness and mass matrices to a reduced geometry. The size of the problem is thus
reduced for search of the eigenvalues. The reduced stiffness is obtained by static condensation,
and reduced mass by limited development of the dynamic equilibrium equation. The degrees of
freedom which should be retained on a priority basis are those which, when blocked, would
increase the eigenfrequencies of the system to the greatest extent. The GUYAN method is
useful with the GIVENS algorithm, which computes all eigenfrequencies, and is also used in
association with the LANCZOS BAND SUBSPACE algorithm, with which the reduced mass
and stiffness matrices can be retained for assignment to special elements (code X9XX).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Note
l In the case of harmonic options, the stiffness and mass matrices depend on the harmonic. There
are consequently as many eigenvalue problems to be solved as there are harmonic taken into
account.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Verification of eigenfrequency and eigenmode search results

Result accuracy is checked in three ways:
T T
l Computation of the modal stiffness (φ Kφ) and mass (φ Mφ) matrices. The stiffness matrix is
also referred to as the K-orthogonality matrix φTKφ. These matrices must be diagonal, and
the diagonal term values depend on the vector norm.
l as for static analysis, computation of residues R = (λM - K) φ which are stored in the reaction
file.
l Computation of Rayleigh quotients. The difference with respect to the eigenfrequency gives
an idea of the accuracy of eigenmode computation.
These operations are always carried out with a consistent mass matrix. The residues depend on
the vector norm. If the structure has rigid constraints, their order of magnitude should be
compared with that of the reactions.
For free-free structures, the number of zero eigenfrequencies must be equal to the number of
eigenmodes of rigid bodies of the option used. If this is not the case with SUBSPACE
methods, it is advisable to increase the size of the computation subspace. In all cases, STURM
sequences can be used to check the number of eigenfrequencies situated in a given frequency
band.
Results and files
The eigenmode search module prints the eigenfrequencies, and in certain cases eigenmodes,
and creates three files which can be saved.
DATA number of loads, equal to the number of computed modes. Circular
eigenfrequencies and generalized masses associated with RESULT file
eigenmodes are contained in this file.
RESULT This standard file contains the types of result effectively computed
(displacement, reaction and force) for each eigenmode. The vector
norm in this file is that selected by the user:
elastic (ELASTIC label) φtKφ = 1
mass (MASS label) φtM φ= 1
maximum (MAX label): the largest component of each eigenvector is
1. This norm is taken by default.
MODE This file is required for modal method dynamic response computation
(XI.4). After computation of eigenmodes, this file contains the same types
of result as the RESULT file. The eigenvector norm is the elastic norm.
The procedures described in paragraph 11.2.1.3 are used to include other
types of result (forces, accelerations, stresses and strains) in the MODE
file, and where appropriate, to align rigid body modes with the MASS
norm.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Note
l Rigid body modes can not be normed with elastic norm. Their number is automatically calculated
from the shift value α and the eigenfrequencies.
If the ELASCTIC norm is selected, rigid body modes are normed (RESULT and MODE files) with
respect to the shifted stiffness matrix (φtMφ =1/α).
If the MASS or MAX norm is selected, rigid body modes are normed (MODE file) with respect to
the mass matrix (φtMφ =1).
Input Data
Commands common to all methods
Input files
- DATA
- STIFFNESSES (K)
- MASSES or BUCKLING (M)
MODE command:
This command is mandatory. It is followed by a label (GIVENS, INVERSE POWER or
SUBSPACE), specifying the selected eigenmode search method (GIVENS by default if the
model size is below 500 d.o.f, if not, use SUBSPACE LANCZOS). Selection of one of the 4
available SUBSPACE methods must be specified.
SHIFT required for free-free structures. An α suitable value is of the order of
0.1 to 0.2 , where ω1 is the first non-zero eigenfrequency of the
structure. When no value of α is specified, an automatic determination
method, based on stiffness and mass matrices trace, is used. In this case
the SHIFT label must be the last one of the line ; the order of size of
α with respect to must be checked.
BUCKLING required for a buckling eigenmodes and critical loads search. If the critical
loads are not all of the same sign, they are sorted by increasing absolute
value order, except with SUBSPACE BAND where they are sorted by
algebraic values. INVERSE POWER, SUBSPACE BLOCK and
ITERATIVE SUBSPACE may be used to search critical loads around a
given value.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
DIAGONAL the mass matrix is diagonalized before computation. This label must
not be used for buckling problems, and is not taken into account with
the GIVENS method.
SIZE t this specifies the subspace size for the SUBSPACE method, but is not
taken into account with the SUBSPACE BLOCK method
By default the subspace SIZE is computed automatically.
For large systems:
. 2nv + 1 for LANCZOS method
. nv (1 + 0.16) for ITERATIVE SUBSPACE (mini nv +8, maxi nv
+32)
VECTORS command
This command is not essential. If it is absent, nv (number of computed eigenfrequencies and
eigenmodes) is taken as at value of 1. All eigenfrequencies are computed with the GIVENS
method.
NORM this specifies the norm for the eigenmodes printed, and stored in the
RESULT file. The MAX norm is used by default. There is a choice
of the following norms: MAX, MASS (φtMφ=1) or ELASTIC
(φtKφ=1).
PRINT i controls printout of eigenmodes.
i: number of eigenmodes printed
by default: no printout.
FORCES this label is used to obtain the forces associated with the eigenmodes
(modal forces). The accelerations are added to the results.
Note
l The eight generalized forces are obtained for the thick shell elements (2006, 2008, 2403, 2404
elements).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Figure 11-3: Input and output files

GIVENS method
DYNAMIC
MODE GIVENS (SHIFT = α)
SUPPRESS ni, nj...
SELECT ni, nj...
(PRINT) VALUES n
VECTORS nv (NORM <LABEL>) (PRINT i) (FORCE)--
(ITERATION niter) (PRECISON ε)
RETURN
SUPPRESS or optional commands using GUYAN's elimination method before the

SELECT eigenfrequency and eigenmode search. These two commands cannot
coexist.
ni, nj, etc. is a list of nodes.
PRINT VALUES optional command limiting eigenfrequency printout to the first n
n eigenfrequencies. By default, the (2nv+1) first eigenfrequencies are
printed.
VECTORS ITERATION and PRECISION labels allow a greater precision in the
(specific labels) search of eigenvectors (by default niter = 5 and ε = 10-2).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example: GIVENS method
SEARCH DATA STIFFNESS MASS 1

DYNAMIC
MODES GIVENS SHIFT 1
SUPPRESS nodes 101 to 1101 STEP 100
PRINT VALUES 20
VECTORS 5 NORM MASS
RETURN
SAVE DATA DISPLACEMENTS REACTIONS 2
Eigenfrequency search for a free- free structure, using GUYAN's method of elimination.
Printout of the first 20 eigenfrequencies, computation of the first 5 eigenmodes with mass norm,
and storage of displacements and reactions in the RESULT file.
INVERSE POWER method
MODE command:
STRUCTURAL The standard sparse method is used. The mass matrix is consistent.
In the lack of STRUCTURAL label, the resolution method can be specified
by the METHOD command.
In the presence of DIAGONAL label mass matrix is diagonalized (the
method can be specified by the SCALING label).
PRECISION ε precision requested for computation of eigenmodes (default value: 10-3).
ITERATION n maximum number of iterations for computation of an eigenfrequency
(default value: 20).
LIMITS command: optional

This command is used to specify the frequency or circular frequency band (fo, f1) in which the
eigenfrequency search is to be made. The FREQUENCY and CIRCULAR FREQUENCY
labels may not appear together. Either the LOWER or UPPER label can be present (by default,
fo = 0 and f1 = f). (fo, f1) define the stiffness matrix shift, but do not constitute imposed bounds
for the eigenfrequency search.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
If fo and f 1 are different, the SHIFT optional label is used to define the number of central
frequencies to be taken into account for the search of nv eigenfrequencies.
For buckling, the command LIMIT LOWER co (UPPER c1) is used to define central critical
loads.
METHOD Command: optional (useless in the presence of the STRUCTURAL label).
It permits to choose a type of solver of the linear equation systems. By default a direct method
of resolution of symmetrical systems is chosen. To search eigenfrequencies in a frequency band
(label LIMIT), GAUSS factorization is necessary. The NON SYMMETRICAL label must be
specified for the iterative method.
FILE Command: optional
This command is used to impose an initial deformation contained in a RESULT file (nload:
load number). An initial deformation suitable to centre frequency leads to the reduction of
computation time.
Note
l The CONDITIONING label on the MODE command is used to obtain a valuation of the initial
numerical conditioning of the linear system associated with the stiffness matrix (static
conditioning).
Example 1: Inverse power method

DYNAMIC
MODEINVERSEPOWER STRUCTURALPRECISION 0.0001 30 ITERATION
LIMITS FREQUENCY LOWER 2000 UPPER 3000
VECTORS 2
RETURN
SAVE DATA DISPLACEMENT 3
Search for 2 eigenfrequencies and eigenmodes round 2500, with special checks for the iterative
algorithm. Eigendisplacements (MAX norm) are stored in the RESULT file.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example 2: Inverse power method : First critical load of a vertical post imposing the
initial deformed shape.
DEFINITION
:
:
LOAD
1 AXIAL FORCE/ NOTHING
N / FZ – 1000
2 TRANSVERSAL FORCE / GX GY 1. NOTHING
RETURN
SOLVE
SOLVE BUCKLING
DYNAMIC
MODE INVERSE POWER BUCKLING
VECTOR 1
FILE 2
METHOD …
RETURN
Example 3: Inverse power method: Condition number of stiffness matrix
SEARCH DATA 1
SOLVE STIFFNESS MASS
DYNAMIC
MODE INVERSE POWER CONDI ITER 5
METHOD…
RETURN
Note:
l If the mass is null, a unity mass matrix is taken into account.
ITERATIVE SUBSPACE method

© 2019 ESI Group
CHAPTER 11 DYNAMIC
MODE command: specific labels

ITERATIVE mandatory for imposing the selection of this SUBSPACE method.
Computation of frequencies round a pivot.
FREQUENCY f
These two labels, which cannot coexist, are used to specify frequency or
or (CIRCULAR circular frequency f, round which frequencies are computed. By default,
FREQUENCY) the lower frequencies are computed.
If the FREQUENCY or CIRCULAR FREQUENCY label is used, the
SHIFT label may not be present.
FACTOR s Definition of scaling factor s. By default, this factor is computed
automatically. It is used to accommodate large differences between the
orders of magnitude of the terms of the K and M matrices.
EXTRACT j j = 10, printing of detailed principal iterations,
j = 20, printing of additional secondary iterations.
VECTORS command: specific labels
PRECISION ε relative precision with which comparison is made in the eigenvalue

convergence test (by default ε = 10–4)
ITERATION n n is the maximum number of iterations.
By default, n = 100 nv.
Note
l This method is not recommended for buckling computation problems. It can be used for very large
structures where a compact storage associated with specific solvers (command METHOD) is
necessary. With the label SHIFT, α defines a central critical load.
METHOD Command: optional
This command is only available when the DIAGONAL label is used.
It permits to choose a type of solver of linear equation systems. By default a direct method of
resolution of symmetrical systems is chosen. To search eigenfrequencies in a frequency band
(label FREQUENCY or CIRCULAR FREQUENCY), the label NON SYMMETRICAL
must be specified for the iterative method.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example 1: Iterative subspace method
SEARCH DATA 1
RENUMBER
1 17 23
RETURN
DYNAMIC
MODE SUBSPACE ITERATIVE SHIFT
VECTOR 10 NORM MAX
RETURN
SAVE DATA DISP 2
The algorithm uses the band storage mode, and the data are reclassified before computation of
the stiffness and mass matrices. The first 10 eigenfrequencies and eigenmodes of a free-free
structure are computed, with automatic shift determination. The reclassified data and results are
stored.
Example 2: Iterative subspace method
SEARCH DATA 1 STIFFNESS 2 MASS 3

DYNAMIC
MODE SUBSPACE ITERATIVE FREQUENCY 50 --
FACTOR 1000 DIAGONAL
VECTOR 5 PRECISION 0.00001 ITERATION 20
RETURN
The 5 frequencies closest to 50 Hz are computed with a scaling factor of 1000, using a diagonal
mass matrix and a special iteration check.
Example 3: Iterative subspace method (specific solver)

DYNAMIC
MODE SUBSPACE ITERATIVE FREQUENCY 50 VECTOR 20
OPTI METHOD
RETURN

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Lanczos .SUBSPACE method

a-This algorithm uses the sparse storage mode, with CHOLESKY factorization of the stiffness
matrix. This is the default algorithm for subspace methods.
DYNAMIC
MODE SUBSPACE (SHIFT = α) (DIAGONAL)(SIZE t)
VECTORS nv (NORM <LABEL> ) (PRINT i) (FORCE) --
(PRECISON ε) (STURM (FREQUENCYf)or (CIRCULARFREQUENCYp))
RETURN
MODE command: specific labels (particularities)

DIAGONAL diagonal mass matrix, inertia associated with rotational degrees of freedom
ignored.
SIZE used to specify subspace size, computed automatically be default. This
label can be useful if a problem is encountered with reorthogonalization. In
this case, a value for t which is below the number of test vector for which
reorthogonalization was unsuccessful, must be imposed.
VECTORS Command: specific label
PRECISION label authorised a greater precision in the search of eigenvectors (by default ε =
10-2).
STURM label is used to compute the number of frequencies lower than f. The initial value of
nv gives a maximum subspace size, jointly to the SIZE label.
Example: Lanczos subspace method

DYNAMIC
MODES SUBSPACE
VECTORS 5 NORM ELASTIC
RETURN
SAVE DATA RESULTS 4
Search for 5 eigenfrequencies and 5 eigenmodes, normalized with respect to stiffness (elastic
norm).
b-An algorithm using the BAND storage mode is also available.
DYNAMIC
MODES SUBSPACE BAND (SHIFT = α) (DIAGONAL(SCALING) (SIZE t) --
VECTORS nv (NORM <LABEL> ) (PRINT i) (FORCE) --
(PRECISON ε) (STURM (FREQUENCY f )or (CIRCULAR FREQUENCY p))
RETURN

© 2019 ESI Group
CHAPTER 11 DYNAMIC
The data are the same as below. The DIAGONAL label can be precised by SCALING.
Example: Lanczos subspace method: BAND
DYNAMIC
MODES SUBSPACE BAND SIZE 15
VECTORS STURM FREQUENCY 50
RETURN
Search for frequency lower than 50 with a maximum subspace size 15 (min. of 101 and 15).
Lanczos SUBSPACE method with GUYAN option
This algorithm is identical to that described in (Chapter. 11.2.1. Input Data Part). The function
of DIAGONAL and SIZE labels is the same. The Lanczos algorithm, for the searching of
eigenvalues and eigenvectors is preceded by a GUYAN reduction of the problem. The
GUYAN's reduced matrices can be stored (REDUCTION label). Eigenfrequencies and
eigenmodes associated with reduced matrices are computed. The METHOD command is used
to choose the employed method for GUYAN's reduction.

GUYAN label for utilization of this algorithm. If REDUCTION label or
SELECT/SUPPRESS command are presents, this label is not essential.
SHIFT α this label must be used for structure free-free for the eigenmode
searching.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
REDUCTION r Extraction of reduced stiffness and mass matrices, computed using

GUYAN elimination, and storage of these matrices in two permanent files
taking the default names of STIGEN and MASGEN respectively. The
STIFFNESS and MASS labels can be used to redefine these permanent file
names.
(r) Used to assign an order number to the matrices in the permanent files.
By default, r = 1. These permanent files contain stiffness and mass
matrices, expressed in the general system of axes of the structure, and
can be used for communication between finite element codes. With
SYSTUS, these matrices can be associated with special code X9XX
elements (Chapter 8.8). In this case, the selected number of nodes is
limited to 20 (instead of 200).
All selected nodes have the same degrees of freedom.
If the STIFFNESS file used for the reduction contains loads, the
reduced file (STIGEN) contains the same number of loads at primary
nodes. It is similar to this obtained by EXTRACT SUBSTRUCTURE
procedure (Chapter 7.2.5).
If the number defined by r is existing in the permanent file, a new
order number is affected to the element matrices. It is noted on the
printing list of the computation.
SELECT/SUPPRESS commands
These two commands, which cannot coexist, define elimination method.
SELECT is used to specify the primary degrees of freedom defined in the following data, while
the SUPPRESS command is used to select the secondary degrees of freedom defined in the
following data.
(nodes) Set of nodes
(U) List of displacement labels, corresponding to the primary (SELECT) or
secondary (SUPPRESS) degrees of freedom. Possible labels are UX, UY,
UZ, RX, RY and RZ.
FREQUENCY f These labels, which cannot coexist, are used to make selection partially
automatic.
PULSATION p On the basis of a user selection, the program searches for primary
degrees of freedom where the ratio of the diagonal terms of K to those
of M is lower than c = (p)2 = (2.π.f)2.
The absence of data lines means that the search is conducted on all
degrees of freedom of the structure.
Note
l The label EXTRACT 10 associated with the command MODE is used to obtain the detailed list of
primary and secondary degrees of freedom.
The label EXTRACT 20 is used to obtain information about eigenmodes of reduced matrices.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
VECTORS Command: specifics labels

PRECISION label authorized a greater precision in the search of eigenvectors (by default ε =
10-2).
STURM label is used to compute the number of frequencies lower than f. The initial value of
nv gives a maximum subspace size jointly to the SIZE label. Otherwise the lower frequencie of
the system associated with secondary degrees of freedom is computed (fr: approximate limit of
the validity in frequency of the GUYAN's reduction). In the presence of INTERRUPTION
label, the computation is stopped if f > fr.
METHOD Command: optional command
This command is used to choose an equation linear system solver type (for the GUYAN's
reduction). For further details, refer to the 21 chapter of the documentation.
Example 1: LANCZOS SUBSPACE with GUYAN option
DYNAMIC
MODE SUBSPACE (GUYAN)
SUPPRESS
10 TO 15 / UX RZ
10 TO 30 STEP 10 INT / UY
VECTORS 20
RETURN
GUYAN's elimination before search for 20 eigenfrequencies and eigenmodes.

Example 2 : LANCZOS SUBSPACE with GUYAN option
DYNAMIC
MODE SUBSPACE DIAGONAL (GUYAN)
SELECT FREQUENCY 100
/ UX UY UZ
VECTOR 10 STURM FREQUENCY 15 INTERRUPTION
RETURN
Guyan's elimination - diagonal mass matrix. Elimination of all rotations, and selection of
translational degrees of freedom for which the ratio of the diagonal terms of K to those of M
exceeds (200.π)2 . Search for frequencies lower than 15 with a maximum subspace size 21. If
the minimum frequency associated with secondary degrees of freedom is lower than 15, the
computation will stop.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example 3 : LANCZOS SUBSPACE with GUYAN option
DYNAMIC
MODE SUBSPACE REDUCTION
SELECT
10 12 15 30 / UX UY UZ
RETURN
Guyan's elimination and creation of STIGEN and MASGEN files, containing reduced
stiffnesses and reduced masses associated with 4 nodes, with selection of translation degrees of
freedom. The first eigenfrequency and the correspondant eigenmodes, associated with the
reduced matrices are computed.
Figure 11-4: Guyan's elimination - Condensation of the dynamic

properties of a structure.
Note
l If the STIFFNESS file contains loads, the STIGEN file contains reduced loads at the primary
nodes.
Lanczos SUBSPACE method with BLOCK option
This LANCZOS type algorithm is particularly suitable for a multiple eigenfrequency search. It
can also be used to search for eigenfrequencies in a frequency band. The band storage mode is
used by default, and is more computer time greedy than the other two LANCZOS type
algorithms. The optional command METHOD can be used for reducing computation times and
size of disk files. Subspace size is computed by the algorithm, according to required
eigenfrequency precision.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
DYNAMIC
MODE SUBSPACE BLOCK nvpb (MEMORY) --
(FREQUENCY (or CIRCULAR FREQUENCY) f )--
(SHIFT α) (DIAGONAL (SCALING)) (EXTRACT j)
VECTOR nv (ITERATION niter ) (PRECISION rtol)--

(NORM <LABEL>) (PRINT i) (FORCE)--
(STURM (FREQUENCY) or (CIRCULAR FREQUENCY)f1)
METHOD …
RETURN
BLOCK nvpb This label is mandatory where the Lanczos subspace method is used with
the block option.
nvpb is the number of vectors per block.
Recommended value: maximum order of multiplicity in the case of
multiple frequencies.
The default value nvpb (1), can be left out if no number appears on the
command after the BLOC label.
MEMORY For little models, the computations executed in the central memory are
quicker.
FREQUENCY f Computation of frequencies round a given frequency.
or CIRCULAR These two labels, which cannot coexist, are used to specify the centre
FREQUENCY frequency round which the frequencies are computed.
By default the lower frequencies are computed.
SHIFT this label cannot be used in association with FREQUENCY or
CIRCULAR FREQUENCY. For buckling α defines a central critical
load.
DIAGONAL this optional label SCALING modifies the mass condensation method
on the diagonal (Chapter 12.3.3).
EXTRACT j j = 10 ; detailed control of eigenvalues convergence.
VECTORS command: specific label
ITERATION niter maximum number of LANCZOS iterations. Subspace size is niter x nvpb
(by default niter=2nv+2).
maximum value of niter: min (niter, ndr /nvpb)
ndr : number of unconstrained degrees of freedom of the structure
nvpb : number of vectors per block

© 2019 ESI Group
CHAPTER 11 DYNAMIC
PRECISION rtol Convergence criterion used for the eigenvalue convergence test.
By default, rtol = 10–4 (rtol = λi - λi - 1/λi -1)
STURM is used to adjust the computed eigenmodes number.
nv begins the number of modes between 0 and f1.
If FREQUENCY label is present on MODE card nv is the number of
modes of [f-f1, f+f1].
initial nv and niter define the maximum subspace size.
Example 1: LANCZOS SUBSPACE method with BLOCK option

DYNAMIC
MODE SUBSPACE BLOCK 2 FREQUENCY 200
VECTOR 5 ITERATION 30 NORM ELASTIC
RETURN
5 eigenfrequencies are computed round 200 Hz. The number of vectors per block is 2, the
maximum number of iterations is 30 (subspace size 60), and the elastic norm is selected.
Example 2: LANCZOS SUBSPACE method with BLOCK option
EARCH DATA STIFFNESS MASS 1

DYNAMIC
MODE SUBSPACE BLOCK
VECTOR 30 STURM FREQUENCY 50
METHOD…
RETURN
The number of eigenfrequencies between 0 and 50 Hz are searched. The eigenmodes obtained
are computed with a subspace maximum size 61. The factorization associated to STURM
sequences increases the computation time.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Determination and Storage of Modal Quantities used for Dynamic Response

Analysis
For mechanical options, the real eigenmode determination modules give displacements, support
reactions and, where appropriate, forces and accelerations (FORCE label). Using
superimposition techniques, the four dynamic response modules (Chapter 11.4) can be used to
handle the quantities indicated in the following table:
MODAL SPECTRAL
Module STOCHASTIC
ANALYSIS ANALYSIS
HARMONIC
Physical quantity
RESPONSE
1 - Displacements ** ** **
2 - Velocities * *
3 - Accelerations * * **
4 - Reactions ** ** **
5 - Forces beam elements * ** **
other elements ** ** **
6 - Nodal stresses **
7 - Nodal strains **
Table 11.3: Mechanical options - Quantities handled by modal

dynamic response analysis
*: quantity which can be obtained
**: quantity obtained if present in the MODES file
Two level 1 commands (SOLVE FORCES MODAL and SEQUENCE) are used to modify
the contents of the MODE file.
SOLVE FORCES MODAL command
This command adds forces in the elements and accelerations at the nodes, to a MODES file
already containing displacements and reactions. This command is particularly suitable for
BEAM options, for which it computes forces taking account of local inertial forces (F=(K-
Mω²)U) and excentrements of neutral fibers. This command is used to select eigenmodes but
not to change their norm.
Input files:
DATA from computation of eigenmodes
MODES containing displacements and reactions
Input data:

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Optional labels:
VECTORS i TO Selection of vectors processed. By default, all vectors are processed.
j STEP k or The selection can be used to reduce the size of the modal base before
(CARDS) dynamic response calculations (Chapter 11.4). For this reason, the
vector number is this of their order in the selection list.
SEARCH nf If the MODES file has been made permanent by a SAVE command, it
can be accessed without the need to copy this file, simply by giving the
file number accompanied by the SEARCH label.
EXTRACT Printout of computed modal forces.
THICK In the presence of thick shells, this label leads to take into account the
shear forces.
The label THICK preserves the 8 components of force in the file
FORCE (in RESU*.TIT) and in the file STRESS (file STRE*.TIT),
instead of 6 usually stored components for the shells. These 8
components of force are, in order of storage:
NX NY NXY MX MY MXY TX TY
The shell elements to which this label is applied are:
- 2403, 2404
- 2406, 2408
- 2006, 2008
It is not taken into account in the following post-processing procedures:
- SHELL (ref Chapter 13.13)
- CONVERT RESULTS STRESS
- COMBINE FORCE/STRESS
Note
l If it exists, the local RESULT file is destroyed by this procedure.
Output files:
MODES containing displacements, reactions, forces and accelerations.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example:
SEARCH DATA 1
SOLVE STIFFNESS MASSES
DYNAMIC
MODES SUBSPACE
VECTORS 8
RETURN
SAVE DATA RESULTS 2 MODES 3
SOLVE FORCES MODAL
SAVE MODES 2
; addition of forces and accelerations with selection
SEARCH DATA 2
SOLVE FORCES MODAL SEARCH 3 CARDS 1 TO 7 STEP 2
SAVE MODES 3
SEQUENCE command
This command converts a local RESULTS file into a MODES file. The MODES file contains
all the quantities of the RESULTS file. The local data file must be the file resulting from
eigenmode computation. If the local displacement file exists, the MODES file can also contain
accelerations (ACCELERATION label).
For all options, this command can be used to align rigid body modes on the MASS norm,
which generally means that modal quantities of a homogeneous order of magnitude can be
obtained for all eigenmodes. (RIGID nvs label)
For the continuous media options (TWO-DIMENSIONAL and THREE-DIMENSIONAL),
elementary forces in the elements, and elementary stresses and strains at the nodes, obtained by
SOLVE FORCES and AVERAGE (cf. 12.2 ) can be replaced by Tresca and von Mises
equivalent stresses (EQUIVALENT STRESS label). This provides an alternative way of
obtaining equivalent stresses from a SPECTRAL ANALYSIS or STOCHASTIC DYNAMIC
study.
Input files:
DATA resulting from eigenmode computation.
RESULTS can contain the 5 normal local files (displacements, reactions, forces, node stresses
and node strains).
Input data:

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Optional labels:
ACCELERATION If the local displacement file exists, the MODES file will contain
accelerations
RIGID nvs If nvs rigid body modes exist, this label is used to align these modes on
the MASS norm .
EQUIVALENT For the continuous media options (TWO-DIMENSIONAL and THREE-
STRESSES DIMENSIONAL), the elementary quantities (forces, node stresses and
node strains) are replaced by the corresponding Tresca and von Mises
criteria in the MODES file.
SHELLS In the SPATIAL SHELL, the RESULT file from the eigenmodes
computation, being treated by SOLVE FORCES and STRESS
(cf.Chapter 13.10), the elementary stresses on the three skins (superior,
average, lower) are replaced by the TRESCA and Von Mises criteria
corresponding, in the MODE file (six components).
BEAMS In the SPATIAL BEAM option and the RESULT file from the eigenmodes
computation being treated by SOLVE FORCES and STRESSES (cf.
Chapter 13.10), the normal stresses at four points of the computation of
each section are obtained at the two extremities of the beam (inertial forces
not taking into account).
VECTORS i TO j Selection of vectors processed. By default, all vectors are processed. The
STEP k selection can be used to reduce the size of the modal base before dynamic
response calculations (Chapter 11.4). For this reason, the vector number is
this of their order in the selection list.
Output files:
VECTORS can contain displacements, reactions, forces, node stresses, node strains
and accelerations.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example 1 - Three-dimensional (two-dimensional) structure.
SEARCH DATA 1; THREE-DIMENSIONAL option

SOLVE STIFFNESS MASSES
DYNAMIC
MODES SUBSPACE
VECTORS 8
RETURN
SAVE DATA RESULTS 100
SOLVE FORCES MODAL
SAVE MODES 100 ; file with DISPL, REACT, FORCE, ACCE
SEARCH DATA RESULTS 100

SOLVE FORCES
AVERAGE STRESSES STRAINS
ELEMENT DIMENSION 3
RETURN
SEQUENCE ACCELERATIONS EQUIVALENT STRESS
SAVE MODES 200; files with DISPL, REACT, ACCEL
; Tresca and von Mises criteria for FORCES, STRESS at NODES,
; STRAIN at NODES
Example 2 - 3D beam structure with 3 rigid body modes

Preliminary alignment of the rigid body modes on the MASS norm.
SEARCH DATA STIFF MASS 1

DYNAMIC
MODES SUBSPACE BLOCK 3 SHIFT
VECTORS 12
RETURN
SAVE DATA 100
SEQUENCE RIGID 3 ; alignment of 3 rigid body modes on the MASS norm

SOLVE FORCES MODAL ; computation of FORCES with local inertial forces
SAVE MODES 100 ; 3 NORM MASS modes
; 9 NORM ELASTIC modes

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Special Problems
Enumeration of eigenvalues (Sturm sequences)
Basically, a number of eigenfrequencies between zero and a given frequency can be obtained.
This method involves triangulation or Gaussian reduction of matrix K-λM where λ is the pivot
associated with the given value. The cost of this reduction is of the order of one static solution.
The number of eigenvalues between two pivots, λi and λj, can be obtained by two successive
computations.
The predetermined number of eigenvalues can differ slightly (in particular if the structure
contains multiple eigenvalues) from the number obtained by eigenvalue computation. This
corresponds to rounding errors.
Input data:
The DIAGONAL label can be used to take account of the effect of a DIAGONAL mass, if an
eigenvalue search method is to be used with this option.
The optional EXTRACT label permits to obtain the localization and the value of negative
pivots.
Vibration modes:
Pivot value is as follows, according to the selected label:
FREQUENCY λi = (2.π.f)2
CIRCULAR FREQUENCY λi = ω2
COEFFICIENT λi = c (c>0)
Buckling modes:
Only the label COEFFICIENT may be used to define the pivot (where c>0 or c<0).
The METHOD label is used to choose the utilized solver (see chapter 21). It must lead to a
total triangulation of GAUSS. Iterative solvers (incomplete triangulation) here are less accurate.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example

DYNAMIC
SEQUENCE FREQUENCY 50 DIAGONAL
SEQUENCE FREQ 100. METHOD
RETURN
Standard solver – DIAGONAL mass

Sky line solver – consistent mass
Prestressed structure
A prestressed structure does not have the same vibration modes, and its stiffness is increased by
a geometric stiffness. The eigenvalue system is then:
((K + Kσ) - λ M) φ = 0
K conventional stiffness
Kσ geometrical stiffness.
Matrix K σ is obtained by the SOLVE BUCKLING command, using the static deflection
associated with the static displacement field. This matrix is contained in the BUCKLING file.
Note
l Quantification of matrices on the deformed geometry makes it possible to integrate the large
displacement matrix. This can have a major impact if static deflection is obtained by non-linear
large displacement computation.
The command:
SUMMATION
is used to include the two stiffnesses in the STIFFNESS file.

Example:
SEARCH DATA STIFFNESS BUCKLING 2

SUMMATION
SEARCH MASS 2
DYNAMIC
MODE
etc.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Figure11-5: SUMMATION procedure - prestressed structure

Thermal eigenvalues
Eigenvalues equation is ( ):
The specific physical quantity is thus the problem eigenvalue of the problem defined by the
calorific conductivity and calorific capacity matrices K and C. For stable systems λ is a real
positive number. Classical output data gives the square root of the absolute value of λ. It is
possible to obtain the real value of λ by processing the problem like a buckling problem.
Example:
SEARCH DATA 1
SAVE STIFFNESS MASS 1
DYNAMIC ; square root
MODES SUBSPACE
VECTORS 10
RETURN
SEARCH DATA STIFFNESS 1
ASSIGN 33 MASS1.TIT BINARY
DYNAMIC ;λ
MODES SUBSPACE BUCKLING
VECTORS 10
RETURN
...

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.2.2 Real Eigenfrequencies and Eigenmodes with Fluid-Structure

Interaction
General
Eigenfrequencies and eigenmodes with interaction
The SHELL FLUID and HARMONIC AXISYMMETRICAL SHELL FLUID options are
used to simulate the behavior of structures modeled with thin shells, and containing non-viscous
incompressible fluids (see Chapter 8.7). The hydrodynamic effect of free surfaces is not taken
into account.
The structure and fluid are modeled with coupling elements at the interface. The eigenmode
search method corresponds to condensing the fluid on the structure, thus fully forming added
mass matrix Ma , which interfaces all the wetted degrees of freedom of the structure with each
other.
The dynamic equilibrium equation is written:
[K - λ (M + Ma)] Φ = 0
The K matrix is stored in band form in core memory. Matrix (M + Ma) is full, and its size is n
(n+1)/2 (where n is the number of degrees of freedom of the structure). This matrix can be
stored on disk. The modified LANCZOS algorithm used gives optimum reclassification of the
fluid and structure nodes, and does not authorize a stiffness matrix shift for computation of rigid
body modes.
Computation of the eigenmodes gives the deflections of the structure and reduced dynamic
pressure over the surface of the structure (interactive dynamic pressure).

Sloshing eigenfrequencies and eigenmodes
THERMAL calculation option is used to calculate sloshing eigenfrequencies and eigenmodes
of a non-viscous compressible or incompressible fluid between stiff walls. Temperature plays
the role of the dynamic pressure. The eigenmode equation is written as follows:
Q1 : effect of the fluid compressibility defined by sound velocity.

Q2 : effect of free surface defined by normal acceleration to free surface.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Eigenfrequencies and Eigenmodes with Interaction

Input data
The stiffness matrices of the structure are assembled from the STIFFNESS file, and the mass
matrices of the structure and fluid, and the interaction matrices are assembled from the MASS
file. These files must first be specified by a SOLVE or SEARCH command.
DYNAMIC
MODEFLUID STRUCTURE(CONDENSATION)(NONE)(VERIFY)(FACTORs)--
(SIZEt)(RESTART)(PRESSURE)(METHOD i)
FRONT FLUID
n1, n2 ... nj
FRONT STRUCTURE
i1, 12 ... ik
VECTORS nv (NORM <LABEL>) (PRINT i) (PRECISION ε) (FORCES)
RETURN
MODE command:
FLUID calls the LANCZOS algorithm adapted for FLUID-STRUCTURE
STRUCTURE problems
CONDENSATION used to stop computation after condensation of the fluid on the structure.
This gives a new mass file, which can be stored under the name MATRIX,
and which contains (M+Ma) . This file is required for SPECTRAL
ANALYSIS (see Chapter 11.4).
NONE is used to minimize the size of necessary files. This label can not
coexist with the CONDENSATION label.
At the end of the eigenmode calculation, the MATRIX file can not be
created.
VERIFY in the presence of this label, there is no computation.
Only
- the number of solid, wet and fluid nodes,
- the size of the fluid matrix,
- the global size for the fluid condensation on the structure (generally the
maximum size of the computation)
are computed.
This command can be used to try different fluid or structure fronts
(optimisation of the bandwidth).
FACTOR s definition of scaling factor s. By default, this factor is computed
automatically. It may be used in the presence of a large number of
constraint elements very stiff (for example: wet wall both quoted).
SIZE t is used to define the size of the subspace (by default this size is computed
automatically).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
RESTART a first computation done and a MATRIX file saved, this label is used to
remake an eigenmodes computation without reform the added mass matrix.
The parameters which can be modify are the followings:
- the number of eigenmodes,
- the numerical parameters (size, factor, precision),
- the element material properties which doesn't have a influence on the
solid mass (method 1), or all element material properties of solid (method
2).
The geometry and the FRONTS (FLUID, STRUCTURE) can't be
modified.
PRESSURE in the presence of this label, the fluid potential UP and the fluid
reactions FP are printed for each fluid node joined.
METHOD i i=1
the added mass matrix is associated with all the solid nodes (method by
default).
i=2
the added mass matrix is associated with all the "wet" solid nodes.
i = 3 (SHELL FLUID option).
Added mass matrix is not explicitly formed. This method minimizes file
size and computation time when a usual number of eigenmodes is
searched. In the presence of the PARTICIPATION label, participation
factors corresponding with a uniform movement of supports are
computed. A mono-spectral analysis is possible (no representative
factor computation, for the reactions: the linked mass effects are not
taken into account).
FRONT command:
This command is used to renumber fluid and structure nodes to optimize the performance of the
band storage method employed for condensation and eigenmode search. The bandwidths
obtained are printed. If independent fluid volumes exist, at least one node of each volume must
appear in the list associated with FRONT FLUID, and likewise for the rigid structure (FRONT
RIGID). This command is similar to RENUMBER (see Chapter 6.3.2).
VECTOR command:
nv number of eigenvalue and eigenvector pairs to be computed.
NORM selection of the norm for the eigenvectors to be printed or stored in the
RESULT file with one of the following labels.
ELASTIC :normalization with respect to the stiffness matrix(φT K φ = 1)
MASS : normalization with respect to mass matrix (φT (M+Ma) φ = 1)
MAX : maximum vector component has the value 1 (default norm).
PRINT i eigenvector print check.
PRECISION ε the label PRECISION permits a larger precision in the eigenmodes
searching (by default ε = 10-2).
FORCE Modal forces computation

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Results
Files created by the DYNAMIC routine are as follows:
l DATA
l RESULTS
l MODES
l MATRIX (required for spectral analysis and calculations with RESTART).
M + Ma : method 1
Ma : method 2
The eigenvectors form a result file with nv loads. Problem data are modified for compatibility
with the result file (number of loads).
The norm in the MODES file is elastic. For spectral analysis purposes, forces and modal
accelerations can be added to this file, using the FORCE label or the SOLVE FORCES
MODAL command (Chapter 11.2.1).
The DATA, RESULTS and MODES files are limited to the shell structure. They consequently
only contain the nodes and elements of the shell, with the nodes given in the renumbering order
obtained with the FRONT STRUCTURE command. Potential fluid displacements are
retained. The number of degrees of freedom for these files is not the same as for conventional
SHELL options (5 for SHELL HARMONIC, 7 for SHELL FLUID).
Note
l The file size much depends on the number of structure nodes.
In SHELL FLUID option
method 1: 5 000 structure nodes MATRIX file 3.6 Go
method 2: 5 000 "wet" structure nodes
The time of computation much depends on the number of fluid nodes and "wet" structure
nodes.
Example: 1- Initial computation

DYNAMIC
MODE FLUID STRUCTURE
FRONT STRUCTURE
1
FRONT FLUID
1 TO 6
VECTOR 10
RETURN
SAVE DATA MODES RESULTS MATRIX 2

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example : 2- Using of the RESTART label
SEARCH DATA 1 MATRIX 2

MODIFY
CHTYPE 2003 NEW 2403
CHTYPE 2004 NEW 2404
RETURN
DYNAMIC
MODE FLUID STRUCTURE RESTART SIZE 80
FRONT STRUCTURE
1
FRONT FLUID
1 TO 6
MODE 20 PRECISION 1.*-4
RETOUR
SAVE DATA RESULTS MODES 3
Figure 11-6: Input and output files

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Eigenfrequencies and Eigenmodes of Sloshing Mode (Thermal Analogy)

The stiff tank is not modelized. The fluid is modelized with 3D elements of thermal
conductivity K and its free surface with elements the thickness of which is H.
Analogy definition:
c : sound velocity in fluid
g : normal acceleration to free surface (gravity)
3D element 2D element
Thermal conductivity K 0
Calorific capacity
Notes
l There's no boundary conditions. Matrix HF is generally a singular matrix and the SHIFT command
has to be used. Only harmonic options with a harmonic number not equal to zero lead to a non-
singular matrix HF.
l In incompressible case (c = ∞), the number of eigenmodes is limited to the number of free surface
nodes.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example: vertical cylindrical tank (g = 9.81 ms-2) containing a compressible fluid

(c=100 ms-1). Eigenmodes of nodal diameter sloshing mode.
DEFINITION
SLOSHING - THERMAL ANALOGY : COMPRESSIBLE FLUID
OPTION THERMAL AXISYM HARMONIC SYMETRIC 1
TWO-DIMENSIONAL
NODES
1 / 0.
11 / 18.288
2 / 0. 12.192
12 / 18.288 12.192
ELEMENT
101 / 1 11 12 2
1001 / 2 12
EDGE
2001 / 1 11 36
2101 / 2 12 36
3001 / 1 2 24
3101 / 11 12 24
MATERIAL PROPERTIES
E 101 TO 964 / KX KY KZ 1.RHO 1. C 1*-4
E 1001 TO 1036 / H 1. RHO 1. C 0.1019
VERIFY
RETURN
DYNAMIC
MODES SUBSPACE
VECTORS 10
RETURN
Analogy:
3D elements : K=1
2D elements : K=0

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.2.3 Complex Eigenfrequencies and Eigenmodes (Symmetrical

Matrices)
General
In the case of a dissipative structure, where sticky damping is characterized by a C matrix, the
eigenmode equation is written: [K + λ C + λ2 M ] φ = 0
Putting:
we obtain a generalized eigenvalue equation, with the same form as that given in Chapter
11.2.1:
(K* + λ M*) Ψ = 0
where K* is not defined positive
λ is a complex number (λ = α + j ω)
Problems can only be processed where the K, M and C physical matrices are symmetrical and
real.
Only one eigenfrequency and eigenmode search algorithm is available, this being the
generalized LANCZOS algorithm for SUBSPACE method with the BLOCK option
(11.2.1.2.6). The K matrice is assembled in memory, according to the choices done using the
METHOD command. By default, it is assembled in band form (renumbering is required). The
matrices M and C are on file. In the presence of the MEMORY label, they are assembled in
memory in band form (renumbering is required).
The M matrix can be condensed on its diagonal.
The matrices (K, M and C) are real, and the eigenfrequencies are real, or conjugated two by
two. Only eigenfrequencies with a positive or zero imaginary part are retained, together with
the corresponding eigenvectors.
Note
l The fact that the K* and M* matrices are double the size of the K, M and C matrices results in
high computation times for complex eigenfrequencies and eigenmodes. An approximate method,
described in Chapter11.6.3, can be used with matrices which can be non-symmetrical. This
method is interesting for large models, for which the real eigenfrequencies have already been
computed.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Input Data
Input files:
DATA
STIFFNESS SEARCH
MASS or SOLVE STIFFNESS MASS DAMPING command
DAMPING
DYNAMIC
MODE SUBSPACE BLOCK nvpb DAMPING (MEMORY)REAL --
((DIAGONAL (SCALING)) (EXTRACT j) (SHIFT (α))
VECTOR nv (ITERATION niter ) (PRECISION rtol)--
(NORM <LABEL>) (PRINT i) (FORCE)
METHOD
RETURN
The input data is very similar to that of a real eigenmode search using the SUBSPACE
algorithm with the BLOCK option (Chapter 11.2.1.input data part). Only the particularities are
described here. The use of STURM sequences is not possible.
MODE command:
DAMPING initiates the complex eigenmode search. In the case of high damping
system, some eigenvalue can be real (ω = 0). The optional label REAL is
used to obtain only the complex eigenmodes. If the number of the real
eigenvalues is important the size of the subspace must be controlled by
using the ITERATION label of the VECTOR command.
The SUBSPACE and BLOCK nvpb labels are required, as only SUBSPACE BLOCK
permits a complex eigenmode search. The frequency search round a given frequency is not
possible. For free-free structures α is approximately 10 to 20% of the first non-zero circular
eigenfrequency. When no value of α is specified, an automatic determination method, based on
stiffness and mass matrices trace, is used. In this case the SHIFT label must be the last one of
the line ; the order of size of α with respect to the first non-zero circular eigenfrequency must
be checked.
EXTRACT j j = 10 ; detailed control of eigenvalues convergence.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
VECTOR command:
ITERATION niter maximum number of LANCZOS iterations. Subspace size is niter x
nvpb.
by default, niter = 4 nv + 2
Maximum value of niter: min (niter, 2 ndr/nvpb).
ndr : number of non-constrained degrees of freedom of the structure
nvpb : number of vectors per block
FORCE This label is used for computation of forces in beams, with the
following formulation:
F = -KU - λCU- λ2 MU
Results
Output files
l DATA: contains the names of the loads corresponding to the eigenmodes
l RESULTS: contains eigenmodes for the selected norm
l MODES: contains elastic norm eigenmodes (ΨT K* Ψ = 1)
Notes
l Eigenmodes Ψ are normalized with respect to matrices K*, M*.
l Only eigenmodes with a positive or zero imaginary part are retained, together with the
corresponding eigenvectors.
l Two cards in the MODES file (or loads in the RESULT file) correspond to each stored complex
eigenmode, such that:
l The real part corresponds to an odd number
l The imaginary part corresponds to the next even number.
Complex eigenfrequencies: λ = α + jω = 2π (α ' + jf)
If the system has damping greater than critical damping on an eigenmode, the corresponding
eigenfrequency is real (ω = 0). All real eigenfrequencies are retained.
If the system has damping less than critical damping, λ and its conjugate are
eigenvalues, and only λ is retained. Eigenfrequencies are stored in ascending value of ω.
A positive value for α corresponds to an unstable mode.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Conventionally, its critical damping fraction is associated with an

eigenfrequency. This γ corresponds to the modal damping used for dynamic response
computation (XI.4). Damping can mix the modes.
Complex eigenmodes:
Modal forces associated with the real and imaginary parts of eigenmodes can be computed by
the command:
SOLVE FORCES MODAL
The appearance of the deflections varies in time:

u = uR cos ωt - uI sin ωt
Rayleigh quotients:
Effective verification of eigenfrequency and eigenmode computation (λ' = λ).
In free-free structures, Rayleigh coefficients have no meaning for rigid body modes.
Example 1 : Complex eigenmodes
SEARCH DATA STIFFNESS MASS DAMPING 1

DYNAMIC
MODE SUBSPACE BLOCK 2 DAMPING
VECTOR 6
RETURN
Six eigenfrequencies and eigenmodes are computed for a damped structure. The number of
vectors per block is 2, as we know that double frequencies exist. The result and eigenvector
files obtained contain 12 loads or cards, with odd numbers corresponding to real parts, and even
numbers to imaginary parts.
Example 2: Complex eigenmodes with integration of geometric stiffness
SEARCH DATA 1
SOLVE
SOLVE BUCKLING
SUMMATION
SAVE STIFFNESS 1
SOLVE STIFFNESS MASS DAMPING
SEARCH STIFFNESS 1
DYNAMIC
MODE SUBSPACE BLOCK 1 DAMPING
VECTOR 5 NORM MASS
RETURN

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Caution must be taken as BUCKLING and DAMPING are on unit 33. Only the SOLVE
STIFFNESS MASS DAMPING command is reliable for damping problems defined in
elements (Chapter 11.2.3).
Figure11-7: Input and output files

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.2.4 Complex Eigenfrequencies and Eigenmodes (Non-

Symmetrical Matrices)
General
A structure with linear behaviour for which the model has n degrees of freedom is characterized
by 3 matrices K, M, C which in this modulus are supposed real and non-symmetrical. The K
matrices must not be singular.
The complex eigenmode equation of the structure is written:
With
The eigenmode equation is solved by factorisation QR. All the eigenvalues are then obtained.
The eigenmode vectors are eventually obtained by inverse iteration algorithm.
Notes
l Unsymmetries
They can be included in the STIFFNESS, MASS, DAMPING files (element having X8XX, X9XX
codes), or can be described directly in the modulus for all types of element.
l Models size
This modulus is particularly suitable to study the model instabilities (positive real eigenvalue) with
weak size (some thousand degrees of freedom) and having large non- symmetries. For tall models,
the modulus PSEUDO MODAL (Chapter 11.6.3)for which the input data is very similar can be
used.
Input Data
Input files:
DATA
STIFFNESS SEARCH
MASS or SOLVE STIFFNESS MASS DAMPING command
DAMPING

© 2019 ESI Group
CHAPTER 11 DYNAMIC
DYNAMIC
MODESCOMPLEX(RATIO r)(ALL)(GYROSCOPIC ω0)
ELEMENTS
nd TO ne (STEP nf) (INTERNAL) / TABLE t3 t2 t1 (ELEMENT n)
ng / TABLE …
TABLE
t1 / (mnij, Kmnij)…
t2 / (mnij, Mmnij)…
t3 / (mnij, Cmnij)…
VECTOR nv (PRINT i) (RAYLEIGH) (THRESHOLD ε)(MODULUS ρ)
RETURN
SAVE DATA RESULTS (MODES) j
COMPLEX MODES command

It initiates the modulus execution.
RATIO r optional label useful to precise the N number of eigenvalues to be
printed.
by default:
no VECTOR command r=10
VECTOR command without PRINT N=2nv+1
ALL optional label useful to save the zero imaginary part eigenvalues.
GYROSCOPIC In option SPATIAL BEAM, the beam being carried by Oy axis, this
ω0 command is used to take into account the anti-symmetrical damping
matrices due to the gyroscopic effects (element of1002 code, ponctual
mass). ω0 is the uniform rotation velocity (rd/s). In output, gyroscopic
damping matrices are stored on DAMPING file which can be used on
HARMONIC RESPONSE modulus (see Chapter.11.5).
ELEMENT command (optional)

This command indicates the elements to which modifications are applied. The modifications
are characterized by stiffness, mass and/or damping matrices which can be symmetrical or not.
nd, ne, nf, ng a list of the numbers of these elements (previously defined with
DEFINITION).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Notes
l the elements concerned do not necessarily have X9XX codes,
l no reference frame can be defined with the ELEMENT command,
l matrices which can be non-symmetrical must be given in full, and are added to those contained in
the STIFFNESS, MASS and DAMPING files.
l non-symmetrical stiffness, mass and damping matrices are given with the TABLE procedure if
t3t2t1 >0 or in files t3t2t1 <0 (Chapter 11.6.3. Stiffness, Mass and Damping Matrices Stored in
Files Part).
TABLE command (optional)
This command defines the stiffness, mass and damping matrices in the way described in the
ELEMENT command.
VECTOR command (optional)
This command is used to release the computation of modal deformed shape.
nv number of complex modal deflections requested (by default nv =1).
PRINT i This label controls result printout.
In the absence of this label, no results are printed.
In the presence of this label:
- i>0, results are printed for the first i modes.
- i=0, results are printed for all modes.
RAYLEIGH In the special case where stiffness, mass and damping matrices are
symmetrical, the Rayleigh quotients give an indication of solution quality.
THRESHOLD is used to erase the stable eigenvalues for which the absolute value of the
imaginary part is lower than ε (choice of the eigenmodes).
MODULUS is used to erase the eigenvalues for which the modulus value is greather
than ρ (choice of the eigenmodes). By default ρ=106.
Results
In the lack of the VECTOR command, only the complex eigenvalues are printed.
If the VECTOR command is present, 3 files are created.
- DATA : contains the titles of the 2nv loads corresponding to the eigenmodes.
- RESULT (MODES): eigenmodes with the norm MAX.
Notes on complex eigenvalues (
l All complex eigenvalues with a positive part are retained because they correspond to a physically
unstable system.
l Only complex eigenvalues with a positive or zero imaginary part are retained. In the lack of the
ALL label, the eigenvalues with a very small imaginary part are not retained.
l The retained eigenvalues are stored in ascending value of size. Their critical damping fraction
is associated with.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example 1: Study of a system stability
SEARCH DATA 1
DYNAMIC
MODE COMPLEX RATIO 5 ALL
RETURN
Automatic formation of the STIFFNESS MASS DAMPING matrices.

Complex eigenvalues with real positive or zero imaginary part edition. The printed complex
eigenvalues number is equal to 20% of the d.o.f. number.
Example 2: Complex eigenmodes with addition of symmetrical stiffness and non-
symmetrical damping.
SEARCH DATA 1
DYNAMIC
MODES COMPLEX
ELEMENTS
1 2 / TABLE 30201
TABLE
1 / 1233 k 2133 k
2/
3/ 1233 c
VECTOR 20 MODULUS 1 * 7
RETURN
SAVE DATA RESULTS (MODES) 2

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.3 Real Eigenmodes and Eigenfrequencies - Modal Synthesis
11.3 REAL EIGENMODES AND EIGENFREQUENCIES -

MODAL SYNTHESIS
11.3.1 General
Modal synthesis methods are used to compute real eigenmodes and eigenfrequencies for a
conservative structure comprising a number of substructures. These methods are suitable for the
study of large models (over 10 000 DOF), or to study local modifications. The eigenmodes
obtained can be used as input data for the dynamic response methods described in Chapter 11.4.
As with all modal methods, the physical displacements (X) of each substructure are expressed
by means of a function base, comprising in this case:
l Substructure eigenmodes ,
l Static displacements U under special loads applied to the constrained degrees of freedom.
Displacements U are also referred to as static modes:
The generalized coordinates (α,β) of each substructure become the unknown factors of the
problem, and are related by reduced mass (M r ) and reduced stiffness (K r ) matrices. These
matrices can be obtained from vibration tests, allowing the integration of experimental super-
elements. The substructures are assembled by writing the equality of the displacements and
physical forces on the constrained degrees of freedom, using Kr and Mr .
Selection of a method means choosing a function base, namely eigenmodes ( ), and static
modes (U). Static modes do not have an orthogonality property with respect to eigenmodes,
thus partially justifying the multiplicity of methods. The seven methods currently available are
described in Chapter 11.3.2.
There is only one substructuring level. Substructures are referred to as level 0 physical or
dynamic super- elements, and the complete structure is referred to as a level 1 dynamic
substructure.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.3.2 Description of Methods

Substructure Modes
Distinction is made between eigenmodes and static modes.
a) Eigenmodes
Eigenmodes are the substructure vibration modes, with certain boundary conditions imposed on
the constrained degrees of freedom. Distinction is made between:
l Eigenmodes with a fixed interface: the constrained degrees of freedom are blocked.
l Eigenmodes with a free interface: the constrained degrees of freedom are free.
l Eigenmodes with a mixed interface: some constrained degrees of freedom of the sub-structure
are blocked, and others free.
b) Static modes
Distinction is made between:
l Constrained static modes: obtained by imposing unit displacements on the constrained
degrees of freedom (associated with fixed links).
l Attachment modes: obtained by imposing unit forces on the constrained degrees of freedom
(associated with free links).
l Residual attachment modes or residual flexibility modes: contribution of eigenmodes not
taken into account in the attachment modes (associated free links) only.
Hypotheses for the Different methods

Seven modal synthesis methods are available in SYSTUS. These comprise one fixed
(interface) method, three free (interface) methods and three mixed (interface) methods.
a) Fixed method
There are no hypotheses or restrictions for this method.
b) Free method n° 1: "Free-ATT" (attachment)
Hypothesis:
The isolated substructure must have sufficient links with the exterior to be statically stable.
c) Free method n° 2: "free-FRF"(residual flexibility - force)
Hypotheses:
1) The isolated substructure must have sufficient links with the exterior to be statically stable.
2) residual eigenmodes are ignored for reduced mass matrix computation.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
d) Free method n° 3 : "free-FRD" (residual flexibility - displacement)

Hypotheses:
1) The isolated substructure must have sufficient links with the exterior to be statically stable.
2) The residual flexibility matrix limited to constrained degrees of freedom, increased by the
constrained flexibilities, must be reversible. In particular, this method cannot be used if all the
eigenmodes of the substructure are used for approximation of displacements.
3) If the residual eigenmodes are not ignored for computation of the reduced mass matrix
(standard case), there must be no constrained stiffness (infinitely stiff connections).
e) Mixed method n° 1: "mixed-ATT" (attachment)
The fixed and free-ATT methods are special cases.
Hypotheses:
The isolated substructure, with fixed constrained degrees of freedom blocked, must be
statistically stable.
f) Mixed method n° 2: "mixed-FRF" (residual flexibility - force)
The fixed and free-FRF methods are special cases.
Hypotheses:
1) The isolated substructure, with fixed constrained degrees of freedom block, must be
statistically stable.
2) The residual eigenmodes are ignored for computation of the reduced mass matrix.
g) Mixed method n° 3: "mixed-FRD" (residual flexibility - displacement)
The fixed and free-FRD methods are special cases.
Hypotheses:
1) The isolated substructure, with fixed constrained degrees of freedom, must be statistically
stable.
2) The residual flexibility matrix, limited to free constrained degrees of freedom, increased by
the constrained flexibilities, must be reversible.
In particular, this method cannot be used if all the eigenmodes of the substructure are used for
approximation of displacements.
3) If the residual eigenmodes are not ignored for computation of the reduced mass matrix
(standard case), there must be no constrained stiffness on the free constrained degrees of
freedom (infinitely stiff connections).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Modes associated with the Different Methods

The types of mode used for each method are given in the following table.
METHOD EIGENMODES (EM) STATIC MODES (SM)

FIXED EM with fixed interface Connected SM
FREE-ATT EM with free interface Attachment SM
FREE-FRF EM with free interface Residual attachment SM
FREE-FRD EM with free interface Residual attachment SM
Connected and attachment
MIXED-ATT EM with mixed interface
SM
Connected and
MIXED-FRF EM with mixed interface
residual attachment SM
Connected and
MIXED-FRD GM with mixed interface
residual attachment SM
Variants
Each free or mixed method has one or more variants, involving a number of supplementary
hypotheses:
l free and mixed ATT methods : 2 variants
l free and mixed FRF methods : 1 variant
l free and mixed FRD methods : 1 variant
The variants are less precise than the basic methods, which are used by default. Three levels of
precision are defined:
Precision 0
Attachment or residual eigenmodes are taken into account for computation of reduced stiffness
and reduced mass matrices. This level of precision concerns the free and mixed ATT and FRD
methods, but does not exist for the free and mixed FRF methods.
Precision 1
Attachment or residual eigenmodes are only taken into account for computation of the reduced
stiffness matrix, and are ignored for computation for the reduced mass matrix.
This level of precision concerns:
l free and mixed FRF methods
l variant no 1 for free and mixed ATT methods
l variant for free and mixed FRD methods

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Precision 2
Attachment or residual eigenmodes are ignored for calculation of the reduced stiffness matrix
and reduced mass matrix.
This level of precision concerns:
l variant no 2 of the free and mixed ATT methods
l variant of the free and mixed FRF methods
This level does not exist for the free and mixed FRD methods.
In this case, static results corresponding to free constrained degrees of freedom are not
necessary. The free and mixed FRF methods allow the use of super-elements with rigid body
modes. The special input data for CALCULATE DYNAMIC MATRIX for these super-
elements, is described in Chapter 11.3.6.
The following table summarizes possible combinations between precision levels and free and
mixed methods.
Method Free and mixed Free and mixed Free and mixed
Precision Level ATT FRF FRD
by default (no
PRECISION = 0 by default
constrained stiffness)
PRECISION = 1 variant 1 by default (**) variant (**)
PRECISION = 2 variant 2(*) variant (*)
Notes
l An experimental type substructure (see Chapter 11.3.8) can use methods and variants marked (*)
or (**). A residual flexibility matrix restricted to free constrained degrees of freedom is not
necessary in cases marked (*), but must be supplied in cases marked (**)
l In all cases, the substructure can be without mass, or with mass and without eigenmodes.
l Precision level is selected using the RESIDUAL label of the CALCULATE DYNAMIC command
(Chapter 11.3.6).
Restrictions
Three methods which use frontier node displacements only as unknown quantities, have no
restrictions:
l fixed method
l free-FRD method
l mixed-FRD method
All other methods have restrictions with respect to inter-substructure connections.

On frontier points where two substructures are connected
For these points, there is no restriction on the choice of type (fixed or free) for the constrained
degrees of freedom of the substructures, nor regarding the method used for each substructure.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
On the points of the structure to which more than two substructures are connected
For these points, a restriction is imposed on the type (fixed or free) of the constrained degrees of
freedom of the associated substructures:
l if free or mixed ATT or FRF methods are used, only one free constrained degree of freedom
can exist for each displacement component at this point.

l there is no restriction if the fixed method, or the free or mixed FRD method is used.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.3.3 Description of a Physical Super-Element

The geometry of each physical or level 0 super-element is defined conventionally (nodes,
elements and physical properties). Particularities concern boundary conditions and loads.
Boundary Conditions
Connections with the exterior are described conventionally.
For constrained degrees of freedom between super-elements, there is a choice between fixed
interfaces, which stabilize the super-element and allow static modes, and free interfaces.
The degrees of freedom of fixed interfaces are constrained:
l stiffly
l elastically if constrained stiffnesses exist between super-elements.
The degrees of freedom of free interfaces are left free in all cases, and constrained stiffnesses,
where these exist, are defined in the physical properties of the substructure.
Static Loads
Normally only exterior loads for which dynamic response computation is to be executed for the
complete structure are defined. Loads included in the STIFFNESS file for the super-element
are integrated by the ASSEMBLE DYNAMIC STIFFNESS routine (Chapter 11.3.6).
Static loads corresponding to static modes are generated automatically by the program
(CALCULATE DYNAMIC RESULT command), and the RESULT file has the number of
the STIFFNESS file for the super-element.
Nevertheless, computation of static modes can be initiated manually by including loads in the
data file, and applying the following rules:
l unit displacement loads imposed on fixed interface degrees of freedom are defined before the
unit force loads imposed on the free interface degrees of freedom.

l in each category (fixed or free), the order of definition of loads imposed on the interface
degrees of freedom is the same as the order of appearance of these degrees of freedom in the
definition of the substructure links.
In this case, no external load must be defined.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.3.4 Description of Computation Steps

With the structure subdivided into physical or level 0 super-elements, complete modal synthesis
computation involves a preliminary step, followed by eight separate computation steps.
Experimental super-elements represent special cases.
Preliminary step:
Description of level 0 super-elements. The dynamic response computation is envisaged, and
the corresponding loads must be described.
Creation of DATA, STIFFNESS, MASS and MODES files for each super-element.
Step 1
Creation of data files for the level 0 super-element, and assignment of its files to this super-
element.
A level 0 super-element has at least three associated files:
l STIFFNESS
l MASS
l MODES containing the eigenmodes.
If rigid body modes exist, they must be aligned with the MASS norm (see SEQUENCE
command: Chapter 11.2.1).
If computation is restarted, three other pre-existent files can be associated with the super-
element (these files are not recreated):
l RESULT: containing static modes
l MATRIX: 2 separate files containing reduced mass and stiffness matrices.
Step 2
Creation of data file for level 1 substructure. This step is essentially used to describe
connections between super-elements (position and type of link), but also:
l Type of static modes associated with free links
l Number of eigenmodes effectively integrated for each level 0 super-element
l Connection stiffness which can be associated with the free links.
Step 3
Automatic computation of static modes for each level 0 super-element.
Step 4
Computation, for each level 0 super-element, of reduced mass and reduced stiffness matrices
(Kr, Mr).
Step 5
Assembly of Kr and Mr matrices to obtain assembled reduced stiffness matrix KR and
assembled reduced mass matrix MR.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Step 6
The eigenvalue equation is solved: (KR - ω2 MR) VR = 0
This gives circular eigenfrequencies ω and eigenmodes VR.
The eigenmodes VR contain the generalized coordinates (α, β).
Circular eigenfrequencies ω are the circular eigenfrequencies of the assembled structure.
Step 7
The physical eigenmodes for each substructure are reconstituted from VR, eigenmodes and
static modes.
Step 8
The physical eigenmodes of the substructures are assembled, to obtain the eigenmodes of the
complete structure. It is also possible to assemble the stiffness and mass matrices, for dynamic
response computation on the complete structure.
Steps 3, 4, 5 and 7 must be executed for each super-element.
The different steps are initiated by four specific commands, described in detail in Chapter
11.3.6.
Command Step
SUPERELEMENT DYNAMIC 1
SUBSTRUCTURE DYNAMIC 2
CALCULATE DYNAMIC 3 to 7
ASSEMBLE DYNAMIC 8

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.3.5 File Management

Data required for computation are assumed to be present in a certain number of INPUT FILES.
The program generates specific files for the procedure, and result files referred to as OUTPUT
FILES, from the INPUT FILES.
DATA files associated with each super-element are created and modified by the program,
during the various steps. The user is not advised to modify these data files.
This operating mode allows fractionalized solution of the problem, with the possibility of
restarting computation at any of the steps described above.
INPUT Files
Each level 0 super- element is assumed to possess the following files before step 1 is
commenced:
DATA
MODE eigenmodes
MASS elementary masses
STIFFNESS elementary stiffnesses and loads
The number of the MODE file must be different from the number of the super-element, in order
to avoid erasing this file on completion of step 7.
Specific Files for Execution of the Procedure

The files created for each level 0 super-element on computation are as follows:
RESU Step 3
RIGD Step 4
MASD Step 4
The following files are created for the level 1 substructure:
DONN Step 2
RIGD Step 5
MASD Step 5
VECD Step 6
Files RIGD, MASD and VECD have the same number as the level 0 super-element, or the
level 1 substructure.
File RESU
This file has the same number and as the STIFFNESS file, and contains the static modes. The
number of this file must be different from the number of the super-element, to avoid erasing the
file on completion of step 7.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
File RIGD
This file contains the reduced stiffness matrix for a level 0 super-element (Kr). For a level 1
substructure, the file contains the assembled reduced stiffness matrix (KR).
File MASD
This file contains the reduced mass matrix for a level 0 super-element (Mr). For the level 1
substructure, the file contains the assembled reduced mass matrix (MR).
FileVECD
This file is created for the level 1 substructure only, and contains circular frequencies ω and
eigenmodes VR in generalized coordinates.
OUTPUT Files
Two OUTPUT files are created for each level 0 super-element on completion of step 7. These
are:
MODE
RESULT
These files have the same number as the level 0 super- element concerned, and contain
restriction of level 1 substructure modes to the level 0 super-element. These are standard
SYSTUS files, which can be handled by post-processing procedures.
Furthermore, the DATA file for the super-element is compatible with the MODE and RESULT
files.
These files are assembled, together with the STIFFNESS and MASS files where appropriated,
in step 8. The eigenmodes are normalized on the elastic norm for the complete structure, in the
MODE file.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
MODAL SYNTHESIS - FILE MANAGEMENT
8
Step 1 2 3 4 5 6 7
DATAP DATA1
DATA1
STIF0 DATA1
VECD1 DATA1
MASS0 DATA0
DATA1 DATA1 DATA0
MODE0 DATA0 RIGD0
Input DATA0 MODE0,S
files DATA0 STIF0 MASD0
DATA0 RIGD1 STIF0
(RESU0)* MASS0
STIF0 MASD1 MASS0 STIF0
(RIGD0)* RESU0
RESU0 MASS0
(MASD0)* MODE0
MODE0
DATA1 DATAS
DATA0 DATA0
DATA0 DATA0 RIGD1 DATA1 RESUS
Output DATA0 MASD1 MODES
files RIGD0 RESU0,S
DATA1 RESU0 VECD1
MASD0 MODE0,S
STIFS
MASS
(*) : computation rerun

Indices:
l P: physical properties of the elements
l 0: level 0 super-element
l 1: level 1 substructure
l 0,S : level 0 super-element geometry, results for complete structure
l S: complete structure
Note
l DATA0 files are modified throughout computation.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.3.6 Input Data

The structure is defined in the same way as for the substructure method, with the DYNAMIC
label used to distinguish modal synthesis commands from static SUBSTRUCTURE
commands.
The four main commands are described in succession:
l SUPERELEMENT DYNAMIC
l SUBSTRUCTURE DYNAMIC
l CALCULATE DYNAMIC
l ASSEMBLE DYNAMIC
SUPERELEMENT DYNAMIC Command

This command executes computation step 1.
Data for a level 0 super-element are defined by one of the following routines:
DEFINITION SUPERELEMENT DYNAMIC n MODE i --

(EXPERIMENTAL iexp) (SUPPRESS nvsupp) (RESULT j) --
MASS k STIFFNESS l (MATRIX m)
RETURN
or
SEARCH DATA ig
.
SUPERELEMENT DYNAM n MODE i MASS k STIFFNESS l --
(RESULT j) (MATRIX m) (SUPPRESS nvsupp) --
(EXPERIMENTAL iexp)(XX x 0 YY y 0 ZZ z0)
SUPERELEMENT DYNAMIC
If the program detects no error, the DATA file for the level 0 super- element is created
automatically.
XX, YY, ZZ define translations taking into account at the ASSEMBLE DYNAMIC
level.
SUBSTRUCTURE DYNAMIC Command

This command executes computation step 2.
The physical properties of the level 1 substructure are used principally for defining the links of
the various level 0 super-elements.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
SUBSTRUCTURE DYNAMIC n (ATTACH) (DISPLACEMENT)(LOAD) (PRINT)

COMPONENTS (ei nvi)... (ej nvj)
CONNECTIONS
m/ (ei ni dofi type).... (ej nj dofj type)
(CONNECTION STIFFNESS
ei / (ni dofi ki)(nj dofj kj ))
(MODIFY ei ej )
RETURN
SUBSTRUCTURE DYNAMIC Command

n substructure identification number.
ATTACH or Selection of static mode associated with the free links.
DISPLACEMENT
A method is selected for all level 0 super-elements having free links. The ATTACH label
indicates selection of the free-ATT method, and the DISPLACEMENT label indicates selection
of the free-FRD method. The free-FRF method is used by default.
LOAD in the presence of this label, the computation generates only .ASCII
files including the static loads necessary for the manual computation of
the static modes with eventually an optimized solver.
PRINT optional label allowing the complete data printing of the substructure
and the level 0 super elements.
COMPONENTS Command
This command describes the level 0 super-elements composing the level 1 substructure.
ei level 0 super-element number introduced
nvi quantity of eigenmodes selected from the level 0 super-element
eigenmodes.
nvi can be zero.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
CONNECTIONS Command
This command describes the connections between the level 0 super-elements.
Each connection defines a link between two or more level 0 super-elements.
A level 0 super-element is involved in a connection by one degree of freedom only.
A degree of freedom for a level 0 super-element node must not be involved in more than one
connection.
The fixed method, and the free- FRD and mixed- FRD methods have no restrictions.
Restrictions for the other methods are described in Chapter 11.3.2.
M : connection number introduced
ei : level 0 super-element number introduced
ni : level 0 super-element node number introduced
dofi : constrained degree of freedom label.
The labels used are UX, UY, UZ, RX, RY and RZ
Type : type of link. The label is FIXED or FREE.
STIFFNESS Command
This optional command is used to introduce constrained stiffnesses on the free interface degrees
of freedom of level 0 super-elements, where the connection is not infinitely stiff.
ei : level 0 super-element number introduced
ni : level 0 super-element node number introduced
dofi : connected degree of freedom label
The labels used are UX, UY, UZ, RX, RY and RZ
ki : constrained stiffness value.
Note
l Constrained stiffnesses are introduced on definition of the substructure, and are taken into account
in step 4 (computation of RIGD for the level 0 super-element).
MODIFYCommand
Where the physical properties for certain super-elements are modified, and in order to avoid
repeat computation for unchanged super-elements, this command can be used to recompute
only those super-elements concerned.
ei, ej number introduced for a level 0 super-element for which the physical
properties have been modified, and for which computation must be
repeated.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Notes
l If the program detects no error, the DATA file for the level 1 substructure is created
automatically.
l The permanent data files for all level 0 super-elements are modified by the inclusion of information
essential for subsequent computation phases.
l Each level 0 super-element, and the level 1 substructure, has a status code which is interpreted as
follows:
l 0 : isolated super-element or substructure
l 1 : super-element constrained
l 3 : super-element or substructure RIGD or MASD files created
l 5 : level 0 super-element MOD and RESULT files created.
l Level 1 substructure VECD file created.
l (the code value can be obtained by the VERIFY DYNAMIC n command – Chap 7.2.4)
CALCULATE DYNAMIC command

Having created the level 1 substructure data, the CALCULATE DYNAMIC command is used
to obtain the solution. This command executes computation steps 3 to 7.
The program generates the computation sequence required for execution of the specified step
automatically. Non-specified steps are executed with the parameter default values.
CALCULATE DYNAMIC <LABEL> n (VECTORS n1) --

(EXTRACT ) (PRINT i) (NOCHECK n3) (OTHER) --
(RESIDUAL j) (BLOCK n4) (ITERATIONS n5) (PRECISION ε) --
(SHIFT α) ( FREQUENCY or (CIRCULAR FREQUENCY) f)

© 2019 ESI Group
CHAPTER 11 DYNAMIC
<LABEL >n This label indicates the computation step required for super-element
number n
The labels used are:
l RESULT
l MATRIX
l MODE
RESULT n indicates that only static results for level 0 super-element
number n (step 3), or all elements of level 1 substructure number n, are
to be computed.
MATRIX n indicates that files RIGD and MASD for level 0 super-
element number n (step 4), or level 1 super-element number n (step 5)
are to be created.
MODE n indicates that MODE and RESULT files for level 0 super-
element number n (step 7), or file VECD for level 1 substructure
number n (step 6) are to be created. In this last case and in the presence
of the WHOLE label, the step 7 will be done for all levels 0 super
element of the substructure number n.
VECTOR n1 n1 is the number of modes required.
For the level 1 substructure, by default (n1)1 = 1
For the level 0 super-elements: : (n1)0 ≤ (n1)1 by default : with
VECTOR label (n1)0 = 1
without VECTOR label (n1)0 = (n1)1
(with MODE label)
EXTRACT in the presence of this label, print RIGD and MASD matrices computed
during the step.
(with MATRIX label)
PRINT i This label is used to control printout of the following:
- static results for level 0 super-elements
- modes for level 0 super-elements or the level 1 substructure.
If the label does not exist, no modes are printed.
If i > 0, the first i modes are printed
If i is null or absent, all modes are printed.
It is advisable to use this label, in particular for computation of static
results, in order to avoid overloading the listing.
NOCHECK n3 Having computed the modes for the level 0 super-element (step 7) or
the level 1 substructure (step 6), this label authorizes recomputation of
the modes, changing n1, n4, n5, ε or i.
n3 = 1 authorizes recomputation of the modes for the level 0 super-element
(step 7), without repeating the preceding steps.
n3 = 10 authorizes recomputation of the modes for the level 1 substructure (step 6),
without repeating the preceding steps
(with MODE label).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
OTHER If the free-ATT (or free-FRF) method is selected for computation of all
super-elements composing the substructure, the OTHER label
authorizes computation of level 0 super-element n using the free-FRF
(or free-ATT) method (with MATRIX label - step 4).
RESIDUAL j authorizes utilization of variants for the 3 free methods (XI.3.2.4):
j = 0 integration of residual stiffness and mass
j = 1 integration of residual stiffness only
j = 2 residual stiffness and mass ignored.
(With MATRIX label - step 4)
the following table indicates possible values according to selected free
method:
Free method ATT FRF FRD
j=0 j=0
by default by default
j=1 j=1 j=1
RESIDUAL j
by default
j=2 j=2
BLOCK n4 n4 is the number of vectors per block

n4 = 1 by default.
ITERATIONS n5 n5 is the number of Lanczos iterations
n5 is computed by default.
PRECISION ε ε is the convergence criterion
ε = 10-4 by default.
SHIFT α for computation of structures for which assembled reduced stiffness
matrix KR is singular (this is the most frequent case with free-free
structures).
FREQUENCY or Computation of eigenfrequencies around a given frequency.
CIRCULAR These two labels, which cannot coexist, neither exist conjointly with
FREQUENCY SHIFT, are used to specify the central frequency around which the
frequencies are computed. By default the lower frequencies are
computed.
The 5 BLOCK, INTERACTION, PRECISION, SHIFT and FREQUENCY or (CIRCULAR
FREQUENCY) labels can accompany the MODE label (step 6). See Chapter 11.2.1. for fuller
details.
Note
l Level 0 super-elements with rigid body modes.
These super-elements cannot have static modes. Only the FRF method can be used. The
RESIDUAL 2 label must appear with the CALCULATE DYNAMIC MATRIX command for
step 4 corresponding to this super-element.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Special syntax must be used:
CALCULATE DYNAMIC MATRIX i RIGID nvs (MGEN m1,... mnvs)
Nvs number of rigid body modes for the super-element (nvs ≤ 6).
m1,... m solid body mode generalized masses as supplied by the
nvs
PARTICIPATION module (cf. 11.4.2), if not set at 1 by using
SEQUENCE RIGID nvs (cf. 11.2.1.).
ASSEMBLE DYNAMIC Command

This command is used to create the following from level 0 super- element files and the
substructure physical properties:
l Physical properties and eigenmodes for the complete structure (Chapter. 11.3.6.)
l Physical properties and stiffness and mass matrices for the complete structure (Chapter.
11.3.6.).
The fixed constraints of nodes forming part of a fixed link of the level 1 substructure are deleted
from the physical properties, and the associated constrained stiffnesses are cancelled, as these
have no physical reality in the assembled structure.
Following modal synthesis, the files obtained are equivalent to those created following
computation of modes for the complete structure.
However, constraints between super- elements corresponding to the links of the dynamic
substructure are not included in the STIFFNESS file, making conventional static computation
of the complete structure impossible, and restricting certain possibilities of the dynamic response
module (see Chapter 11.4). Unless LINK command can be used (Chapter 11.3.6.)
Assembly of data and modes
Input files:
DATA for level 1 substructure
DATA for level 0 super-elements
MODE for level 0 super- elements, with each MODE file containing the modes of the
structure, restricted to the degrees of freedom of the level 0 super-element nodes.
These are the result files for modal synthesis step 7.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Output files:
DATA for the complete structure, reduced to the following entities:
l Geometry
l Physical properties
l Constraints (fixed constraints forming part of a link are deleted)
l Load titles (corresponding to computation of eigenmodes)
l Generalized masses and circular eigenfrequencies
RESULT eigenmodes for the complete structure (specified norm).

MODE eigenmodes for the complete structure (elastic norm).
Input data:
ASSEMBLE DYNAMIC n (FILE i ) (LINK (TRANS kt ROTATION kr))-- (NORM

<label>) (PRINT j)
n number of level 1 substructure.

FILE i used to define the number of the output files (DATA, RESULT and
MODE). By default, the output files take the number of substructure n.
Caution:
l if i is not defined, the level 1 substructure data file is destroyed.
LINK is used to generate 1602 connections elements by replacing the links of the
substructure. The stiffness, between the translation degrees of freedom can
be precised by TRANS kt, between rotation degrees of freedom by
ROTATION kr (by default kt = kr=1012). Static computation can be done
on the data file obtained.
Note
l This label is actually available only with THREE DIMENSIONAL and 3D SHELL options.
NORM <label> defines the norm for the eigenmodes in the RESULT file. The
associated label can be MAX, MASS or ELASTIC. The MAX norm
is used by default.
PRINT j edition of the j first eigenmodes. In the lack of PRINT, no printing. In
the presence of PRINT without j value , edition of all eigenmodes.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example: Static results generation for a spectral analysis with static correction.
ASSEMBLE DYNAMIC 1000 FILE 2000 --

LINK TRANS 1.*11
DEFINITION RESTART
LOAD
1 ACCEL. X / GX 1. NOTHING
2 ACCEL. Y/ GY 1. NOTHING
3 ACCEL. Z/ GZ 1. NOTHING
VERIFY
RETURN
SOLVE
SAVE DATA RESULTS 2001.
Assembly of data and rigidity and mass matrices

Input files:
DATA for level 1 substructure
DATA for level 0 super-elements
STIFFNESS for level 0 super-elements
MASS for level 0 super-elements
Output files:
DATA for the complete structure, reduced to the following entities:
l geometry
l physical properties
l connections (fixed connections forming part of a link are deleted)
l static load titles.
STIFFNESS stiffness matrix for the complete structure, containing the specified second
members.
MASS mass matrix for the complete structure.
The MASS file for the complete structure contains the elementary mass matrices. The
STIFFNESS file for the complete structure contains the elementary stiffness matrices, and the
applied load vectors specified by the LOAD command.
Input data:
ASSEMBLE DYNAMIC STIFFNESS (MASS) n(FILE i)

LOAD
j1 Title/
(ek nk ck) (ei ni ci)...
j2 Title/
.
© 2019 ESI Group
CHAPTER 11 DYNAMIC
.
.
RETURN
ASSEMBLE DYNAMIC command

STIFFNESS assembly of stiffness matrices,
MASS assembly of mass matrices. These two labels can coexist.
N number of level 1 substructure.
FILE i used to define the number of the output files (DATA, STIFFNESS
and/or MASS). By default, the output files take the number of
substructure n.
Caution: if i is not defined, the level 1 substructure data file is destroyed.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
LOAD command
This command is used to create the applied load vectors contained in the STIFFNESS file for
the complete structure:
j1 Title/
(ek nk ck) (ei ni ci)...
j1 number of the complete structure vector to be created

ek, ei, .. numbers of level 0 super-elements
nk, ni, .. numbers of load vectors for super-elements ek, ei... in the level 0 super-
element STIFFNESS file
ck, ci, .. multiplying factors associated, respectively, with vectors nk, ni, etc.
Notes
l a maximum of 32 loads can be defined in this way
l creation of the STIFFNESS file for the complete structure with no load:
ASSEMBLE DYNAMIC STIFFNESS n FILE i

RETURN
l creation of the MASS file for the complete structure only:
ASSEMBLE DYNAMIC MASS n FILE i

RETURN
Dynamic response computation

Dynamic response computation for a structure analyzed by modal synthesis is conducted using
the modal methods described in Chapter 11.4.
l Modal analysis
l Modal harmonic response
l Spectral analysis
l Stochastic analysis
The only difference occurs with computation of modal forces (PARTICIPATION module –
Chapter 11.4.2), and stems from the fact that it is not possible to create the static result file.
Multi-excitation computation is not possible. For computation with forces on the free nodes, the
PARTICIPATION command must be followed by the LOAD command (Chapter. 11.4.2).
If the command LINK can be used, (Chapter 11.3.6), the restrictions of the previous paragraph
doesn't exist any more.
Sensitivity computation (Chapter 11.6) is also possible.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.3.7 Example: Uniform Clamped-Free Beam

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example 1: seismic excitation defined by its pseudo-acceleration spectrum.

Computation of eigenmodes by the FIXED-FIXED method (detail of successive steps).
DEFINITION
MODAL SYNTHESIS * FIXED-FIXED METHOD * SUPER1 *
OPTION PLANE BEAM
ONE-DIMENSIONAL
NODE
1/ 0. 0.
6/ 0. 1.
ELEMENTS
1/ 1 6 5
MATERIAL PROPERTIES
M 1 TO 5 /E 2.*11 NU 0.30 RHO 7800 AX 0.0005 IZ 2.6*-8
CONSTRAINTS
N 1 6 / UX UY RZ
RETURN
SAVE DATA 1
DYNAMIC
MODE
VECTOR 6
RETURN
SAVE MODES 11
DEFINITION
MODAL SYNTHESIS * FIXED-FIXED METHOD * SUPER2 *
OPTION PLANE BEAM
ONE-DIMENSIONAL
NODE
6/ 0. 1.
11/ 0. 2.
ELEMENTS
6/ 6 11 5
MATERIAL PROPERTIES
M 6 TO 10 /E 2.*11 NU 0.30 RHO 7800 AX .0005 IZ 2.6*-8
CONSTRAINTS
N 6/ UX UY RZ
RETURN
SAVE DATA 2 ...

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example 1: continued

DYNAMIC
MODE
VECTOR 6
RETURN
SAVE MODE 21
SEARCH DATA 1
SUPERELEMENT DYNAMIC 100 STIFFNESS MASS 1 MODES 11
SEARCH DATA 2
SUPERELEMENT DYNAMIC 200 STIFFNESS MASS 2 MODES 21
COMPONENTS 100 5 200 5
CONNECTIONS
1 / 100 6 UX FIXED 200 6 UX FIXED
2 / 100 6 UY FIXED 200 6 UY FIXED
3 / 100 6 RZ FIXED 200 6 RZ FIXED
RETURN
CALCULATE DYNAMIC RESULT 100
CALCULATE DYNAMIC MATRIX 100
CALCULATE DYNAMIC RESULT 200
CALCULATE DYNAMIC MODE 1000 VECTORS 5 TOUT
ASSEMBLE DYNAMIC 1000 FILE 1100
ASSEMBLE DYNAMIC MASS 1000 FILE 1100
RETURN
SOLVE FORCES MODAL SEARCH 1100

DYNAMIC
PARTICIPATION
SPECTRAL ANALYSIS GX
COMBINE DISPLA REACTION FORCE ACCEL
MODE 2 SRSS
SPECTRUM 1
TABLE
1 / 1 0.1 2. 2. 2. 3. 10. 30. 10. 33. 1.5 100. 1.5
RETURN
END

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example 2: Time-domain response to excitation of the support and a force applied to

the end of the beam. Computation of eigenmodes by the mixed-FRF method,
rerunning the matrices for element 2 (example 1).
SEARCH DATA 1
DEFINITION
MODAL SYNTHESIS * METHOD MIXED F.R.F. * SUPER1 *
OPTION PLANE BEAM
REPEAT
CONSTRAINTS
N 1/ UX UY RZ
RETURN
SAVE DATA 3
DYNAMIC
MODE
VECTOR 6
RETURN
SAVE MODES 31
SEARCH DATA 2
DEFINITION
MODAL SYNTHESIS * METHOD MIXED F.R.F.* SUPER2 *
OPTION PLANE BEAM
REPEAT
CONSTRAINTS
N 6/ UX UY RZ
LOAD
1 FORCE AT END /
N 11 / FX -2
RETURN
SAVE DATA STIFFNESS MASS 4
SEARCH DATA 3
SUPERELEMENT DYNAMIC 300 STIFFNESS MASS 3 MODE 31
SEARCH DATA 4
SUPERELEMENT DYNAM 400 STIFF MASS 4 MODE 21 RESULT 2 MATR 200
SUBSTRUCTURE DYNAMIC 2000 ; (METHOD F.R.F.)
CONNECTIONS
1 / 300 6 UX FREE 400 6 UX FIXED
2 / 300 6 UY FREE 400 6 UY FIXED
3 / 300 6 RZ FREE 400 6 RZ FIXED
RETURN ...

© 2019 ESI Group
CHAPTER 11 DYNAMIC
• Example 2: continued
CALCULATE DYNAMIC MODE 2000 VECTORS 5

ASSEMBLE DYNAMIC STIFFNESS MASS 2000 FILE 2200
LOAD
1 FORCE AT END FX /
400 1 1.
RETURN
SEARCH DATA STIFFNESS MASS MODE 2200

DYNAMIC
PARTICIPATION 5 FORCE 1
LOAD FORCE 1
MODAL ANALYSIS 5 DISPL 1 FORCE 1
FUNCTION 1 2
TABLE
1 / 3 9. 0. 0. 100.
2 / 3 18. 0. 0. 100.
TIME
1. STEP 0.001 / DISPL 100 STORE 1
RETURN
DYNAMIC
COMBINATION 5 RELATIVE
PRINT MAXI MODULUS
DISPLACEMENT 11
REACTION 1
FORCE 1
CARD 0 TO 1000 STEP 100
CARD 1 TO 1000 STEP 1 STORE
RETURN
SAVE DATA TRANSIENT 2201
END

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.3.8 Experimental Super-Elements

General
Experimental super-elements must have free links, and can be used with the free and mixed
(ATT, FRF and FRD) methods.
Data required to create reduced stiffness (K r) and reduced mass (Mr) matrices are in sequential
binary file EXPEiexp.TIT (iexp is the super-element number).
This file contains:
l Circular eigenfrequencies ω
t
l Generalized stiffnesses (K G = φ Kφ)
l Generalized masses (MG = φt Mφ)

l Experimental eigenmodes z, reduced to constrained degrees of freedom (z is part of φ). The
eigenmode norm is not imposed, but (z , KG, MG, ω) must be coherent (KG = ω2 MG).
l Residual flexibility matrix GL, restricted to free constrained degrees of freedom, where
appropriate. This matrix is essential with the mixed-FRD method, and can also be used with
the mixed-FRF method.
GL = -PU' = restriction to constrained degrees of freedom of residual flexibility modes U'.
Experimental super-elements cannot be loaded.
FREE and MIXED
ATT FRF FRD
methods
With residual
flexibility matrix
(RESIDUAL=1) X X
No residual flexibility
matrix
(RESIDUAL=2) X X
Utilization of experimental super-elements

Note
l The FRF and FRD methods allow the use of super-elements with rigid body modes. With the FRF
mode, a low displacement value is necessary with the CALCULATE DYNAMIC MODE command
for the level 1 substructure in certain cases.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Experimental Super-Element Data File

l Name: EXPEiexp.TIT iexp = super-element number
l Type: sequential binary
All real values are double precision
l Record n° 1: Like any other SYSTUS file, this record contains 21 words: WORD(21)
o WORD(1) = $SEV
o WORD(3), WORD(4) = super-element creation date and time
o WORD(5), WORD(6) = super-element creation date and time
o WORD(12) = number of degrees of freedom per node
o WORD(16:21) = problem title
l Record n° 2: NV, NLINKL, IRESI

o NV = number of modes measured
o NLINKL = number of FREE links
o IRESI = 1 if residual flexibility matrix given, otherwise 0
l Record n° 3: NV circular eigenfrequencies

l Record n° 4: NV generalized stiffnesses
l Record n° 5: NV generalized masses
l Record n° 6: NV records of this type, with an eigenmode reduced to constrained degrees of
freedom (NLINKL terms) in each record.
l Record n° 7: if IRESI=1 NLINKL records with a column of the residual flexibility matrix
(NLINKL terms) in each record.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
File creation program example
Simple Practical Example: Addition of a Load at the End of a Beam

A rigid mass is supported by the end of a beam.
Added mass effects are taken into account by the two rigid body translation modes and rigid
body rotation mode. These modes are aligned with respect to the MASS mode.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
DEFINITION
Beam with load at end - EL.1: manual computation of static modes
OPTION TWO-DIMENSIONAL
TWO-DIMENSIONAL
NODES
1/
2 / 10.
3 / 12.
11 / 0. 1.5
12 / 10. 1.5
13 / 12. 1.5
ELEMENTS
101 / 1 2 12 11
501 / 2 3 13 12
EDGES
1001 / 1 11 3
1011 / 1 2 20
1031 / 2 3 4
MATERIAL PROPERTIES
E 101 TO 160 501 TO 512 / E 2.*11 NU 0.3 RHO 7800. H 1.
CONSTRAINTS
N 1 11 1001 1002 / UX UY
LOADS
1 N 12 FX / NOTHING
N 12 / FX 1.
2 N 12 FY / NOTHING
N 12 / FY 1.
3 N 13 FX / NOTHING
N 13 / FX 1.
4 N 13 FY / NOTHING
N 13 / FY 1.
VERIFY
RETURN

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example: experimental super-element (contd.)
SAVE DATA 1
SOLVE
SAVE RESULT 1
DYNAMIC
MODE SUBSPACE
VECTORS 10
RETURN
SAVE STIFFNESS MASS MODES 1
DEFINITION
BEAM WITH LOAD AT ONE END - EL.2 : EXPERIMENTAL ELEMENT -
OPTION TWO-DIMENSIONAL
GEOMETRY
NODES
12 / 10. 1.5
13 / 12. 1.5
VERIFY
RETURN
SAVE DATA 20
SEARCH DATA 1
SUPERELEMENT DYNAMIC 100 STIFF MASS 1 MODE 1 RESULT 1
SEARCH DATA 20
SUPERELEMENT DYNAMIC 220 EXPERIMENTAL 222

CONNECTIONS
1 / 100 12 UX FREE 220 12 UX FREE
2 / 100 12 UY FREE 220 12 UY FREE
3 / 100 13 UX FREE 220 13 UX FREE
4 / 100 13 UY FREE 220 13 UY FREE
RETURN

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example: experimental super-element (contd.)
CALCULATE DYNAMIC MATRIX 220 RESIDUAL 1

CALCULATE DYNAMIC MODES 1000 VECTORS 5 SHIFT 1.*-6
CALCULATE DYNAMIC MODES 100 VECTORS 5
CALCULATE DYNAMIC MODES 220 VECTORS 5
END
Note
l With some fixed links, practically necessary with an experimental super-element, the eigenmodes
order of magnitude, (normed with the MASS norm), and static modes can be different. This leads to
an eigenvalue problem not correctly solved and necessitates the use of a low offset which does not
change the circular and eigenmodes values.

© 2019 ESI Group
CHAPTER 11 DYNAMIC

© 2019 ESI Group
CHAPTER 11 DYNAMIC

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.3.9 Dynamic Automatic Link Generator (D.A.L.G.)

The links description of the SUBSTRUCTURE DYNAMIC command is one of the step the
more heavy in the modal synthesis setting up. Beforehand, it is possible to generate a file
containing links with the form of SUBSTRUCTURE DYNAMIC using the static ALG
(Chapter 10.5.4), for the most normal cases.
The different physic elements being defined, static elements (SUPER ELEMENT command)
and a static substructure (SUBSTRUCTURE command) must be created and the procedure
CONNECTION DYNAMIC procedure can be applied.
CONNECTION DYNAMIC (FREE) (TRANSLATION) num

LINK FILE
RETURN
CONNECTION command
DYNAMIC activate the D.A.L.G.
FREE the generated links are supposed to be free (by default FIXED).
TRANSLATION
the links are generated only for the translation degrees of freedom of
the considered option (the fluid structure options are not allowed).
num number of the static substructure
LINK command
FILE compulsory label to generate the ASCI LINKnum.TIT file including
the dynamic links.
Limitations:
l Maximum number of substructures connected into one point : 6,
l The generation of mixed links is not yet available.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Simple example: Generation of SUBSTRUCTURE DYNAMIC data

Vertical cylinder embeded-embeded with 15 m length, divided into 3 identical elements (file
DATA50.TIT, SPATIAL SHELL option).
Computation by a method with steady interfaces by connecting only the translation degrees of
freedom (reduction from the problem size to the border).
SEARCH DATA 50
SUPER ELEMENT 1
SUPER ELEMENT 2 ZZ 5.
SUPERELEMENT 3 ZZ 10.
SUBSTRUCTURE 100
COMPONENTS 1 2 3
RETURN
CONNECTION DYNAMIC TRANSLATION 100
LINK FILE ; file generation LINK 100.TIT
RETURN
SEARCH DATA 50
DYNAMIC
MODES …
VECTORS nv
RETURN
SAVE STIFFNESS MASS MODES 50
SEARCH DATA 50
SUPER ELEMENT DYNAMIC 101 STIFNESS MASS MODES 50
SUPER ELEMENT DYNAM I C 102 STIFNESS MASS MODES 50
SUPERELEMENT DYNAMIC 103 STIFNESS MASS MODES 50
COMPONENTS 101 nv 102 nv 103 nv
LINK
$READ LINK 100.TIT
RETURN
END

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.4 Dynamic Response - Modal Method
11.4 DYNAMIC RESPONSE - MODAL METHOD

11.4.1 General
General Equations - Driving Movement and Relative Movement
With index 1 designating free degrees of freedom, and index 2 constrained degrees of freedom,
the dynamic equilibrium equation for a structure is written as follows, with customary notations:
The unknown quantities of the problem are U 1 , absolute displacements of free nodes, and R 2 ,
reactions on the constrained nodes.
Driving movement Ue corresponds to the static deflection imposed by the supports
(Ue = - K11-1 K12 U2). Relative movement Ur is defined with respect to this static deflection
(U1 = Ue + Ur). This verifies the equation:
The relative movement is projected into eigenmode base ϕ , computed with the constrained
degrees of freedom 2 (U 2 = 0), U r = ϕ q . Generalized coordinates q checks the dynamic
equilibrium equation:
As we know q, we can deduce Ur, and hence U1 and R2.

For a problem where only imposed forces exist, relative and absolute movements are the same.
If the displacements of all supports are identical, the analysis is said to be of the MONO-
EXCITATION type. In this case, the driving movement is the common movement of the
supports, and generates neither stresses nor reactions.
If the displacements of the supports differ, analysis is said to be of the MULTI-EXCITATION
type, and the driving movement must be determined by static computation.
Note
l These methods are employed in particular in the mechanical field, which explains the vocabulary
employed. However they can also be used in any context where eigenmodes have to be computed
(e.g. thermics).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Modal Analysis Principal Hypotheses

T T
The stiffness modal matrix (ϕ K 11 ϕ) and mass modal matrix (ϕ M11 ϕ) are diagonal, due to
the orthogonal properties of the eigenmodes. It is also assumed:
T
l that the modal damping matrix (ϕ C 11 ϕ) is diagonal,
l and that the velocity-related modal forces are negligible (Fmv = 0)
In this case:
l generalized coordinates q verify independent equations,
l relative movement, reflecting dynamic behavior, depends on the imposed forces and
acceleration of the supports,

l driving movement depends on the displacements of the supports.
Generally, the behavior of the structure is linear, thus allowing linear accumulation of the
different excitations. A number of non-linearities (stops, friction) can be taken into account,
where the excitation is defined in temporal terms.
Solver limitations:
The number of eigenvectors of the modal base (before a possible selection) cannot exceed 640.
Definition of Excitations
Four different computation modules are available, corresponding to four different types of
excitation definition.
MODAL ANALYSIS module (Chapter 11.4.3)
Excitation is defined in terms of time. The modal transient response, followed by the physical
transient response are obtained in the relative or absolute reference frame. A number of non-
linearities (stops, friction) can be taken into account, by introducing pseudo-forces in the second
member of the equations.
HARMONIC RESPONSE module (Chapter 11.4.4)
Excitation amplitude is defined in terms of frequency. The stationary response to a harmonic
excitation (amplitude, phase) is obtained. Variation of the excitation frequency gives the
stationary or transient transfer function.
STOCHASTIC module (random vibrations - Chapter 11.4.5)
Associating its power spectrum density (PSD) with each ergodic zero mean value stationary
type excitation, the following are computed using transfer functions:
l response PSD.
l mathematical expectation of the response, and various statistical parameters.
If cross spectra exist between a number excitations, only the real part of these cross spectra can
be taken into account.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
SPECTRAL ANALYSIS module (seismic studies - Chapter 11.4.6)

Excitations are exclusively movements of the supports, defined by their pseudo-acceleration
spectrum (Chapter 11.7.2). If the movements of the supports are different (multi-excitation
analysis), the maximum displacements of the supports are required. This module is designed in
particular for seismic studies, and combines responses in accordance with the RG 1.92 rules of
the NRC (US Nuclear Regulatory Commission).
Definition of Damping - Modal Equations

Physical hypotheses
Damping is viscous, and is defined differently for elastic and rigid body eigenmodes.
Elastic body eigenmodes (ω ≠ 0) :
Damping is defined for each eigenmode by critical damping fraction γ, representing the ratio of
total damping to critical damping. For a system with one degree of freedom and using
customary notations, this gives:
If damping is introduced for the complete structure, in BASILE form (C = αK + βM), the α
and β coefficients are transformed, for each circular eigenfrequency ω, into a critical damping
fraction:
Rigid body eigenmodes (ω = 0) :

A special damping form is used: C = βM
Modal equations
Elastic body modes (ω ≠ 0):
Eigenvectors ϕ are normalized on the elastic norm
Rigid body modes (ω = 0):

Generalized mass m G , must be computed for each eigenmode (label RIGID of
PARTICIPATION see Chap. 11.4.2.3).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
DAMPING command
This command is included in the MODAL ANALYSIS module ( Chapter 11.4.3 ) and
HARMONIC RESPONSE module ( Chapter 11.4.4 ). Critical damping fractions γ ,
corresponding to a given physical damping factor, can be obtained for elastic eigenmodes using
the MODAL DAMPING routine (Chapter 11.7.1 ), or by complex frequency computation
(Chapter 11.2.3).
DAMPING (ALPHA α) (BETA β) (GAMMA γ) (MODAL) (RIGID)

i1 TO i2 STEP i3 / (BETA or GAMMA) γi
Elastic body modes:

ALPHA, BETA
GAMMA all modes have damping γ

or
MODAL damping is given for each mode on the following cards ( γi = 0 for
modes not mentioned)
In the absence of RIGID label, MODAL label and its data can be
replaced by FILE nƒ if the DAMPING process (Chapter 11.7.1) has
created a DAMPnf.DAT type file.
Three optional labels can make the label FILE more precise:
METHOD i i = 1 weighting by strain energies (default)
i ≥ 2 weighting by kinetic energies
LOWER inf γ minimal value
UPPER sup γ maximal value
Notes
l Various forms of damping are added:
l The labels GAMMA and MODAL or FILE cannot coexist.

Rigid body modes (label RIGID)
BETA:all rigid body modes have damping β
MODAL: damping is given for each mode on the following cards (modes not mentioned βi =
0)

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Notes
l The number of rigid body modes is defined by the command RIGID of the computation command.
For HARMONIC RESPONSE MODAL this label is not always necessary. Without this label, only
BETA type damping is taken into account.
l If the BETA and MODAL labels coexist, the damping of elastic body modes is defined with BETA
and the damping of rigid body modes is defined with MODAL.
l In case a variable damping is used (label MODAL) conjointly with a selection of modes, the selection command
must precede the command DAMPING. Modes are given by their serial number in the selected list.
Examples:
1 - Elastic modes with damping, rigid body modes without damping:
MODAL ANALYSIS 20 FORCE 1 RIGID 6

SELECT 1 TO 6 10 TO 20
DAMPING ALPHA 0.002 BETA 2 MODAL
7 TO 15 / GAMMA 0.02
16 17 / GAMMA 0.04
modes 1 to 6 : β = 0
modes 10 to 18 :
modes 19 and 20 :
2 - Elastic and rigid body modes with damping:
MODAL ANALYSIS 20 FORCE 1RIGID 6

DAMPING BETA 4 MODAL RIGID
1 TO 3 / BETA 1
4 TO 6 / BETA 3
7 TO 20 / GAMMA 0.02
modes 7 to 20 :

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.4.2 Modal Forces - Representativeness of the Modal Base

The modal forces associated for each load with the real eigenmodes, are computed by the
PARTICIPATION module, and stored in the DATA file.
This module also computes the modal base representativeness criteria, by analyzing the
reproduction quality of a static deflection representing the excitation.
Modal Forces or Participation Factors

l Participation factors pk, associated with mode k are defined differently in each of the
following cases:
l Force
l Imposed acceleration
o for a mono-excitation study
o for a multi-excitation study, must be determined by static computation (RESULT file

with results corresponding to imposed U2 values).
l The modal base is incomplete, and can be completed, in the case of seismic studies, with the
inclusion of pseudo-modes or residual modes UR, orthogonal to the selected modes, with
which participation factors similar to those for the m selected eigenmodes, are also associated.
US : static deflection associated with an uniform acceleration type load (GX, GY or GZ).
In a multi-excitation context, the approximation of an uniform acceleration is most of the
time sufficient.
φk : eigenmode
There is a pseudo-mode for each seismic direction, and the existence of these pseudo-modes
explains why the PARTICIPATION module associated with the SPECTRAL ANALYSIS
module (Chapter 11.4.6 ) is different from that associated with the MODAL ANALYSIS
module (Chapter 11.4.3) and the HARMONIC RESPONSE module (Chapter 11.4.4).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Representativeness Criteria
Associated with analysis of a static deflection, U S , these criteria require the existence of a
RESULT file containing this deflection, except in a MONO-EXCITATION context where this
deflection is generated by the program.
"Support movement" type load
US : static deflection associated to the U2 movement of the supports.
Taking account of the orthogonality of the eigenmodes and residual modes:
we can write:
We put: effective mass of mode k :
total effective mass:

mek characterizes the importance of mode k
me is compared with the unsupported mass, to assess the quality of the modal base.
"Force on free nodes" type loads
US : static deflection associated to force type load.
We define:
- Mode k strain energy percentage
- Modal base strain energy
- Mode k displaced mass percentage
- And the mass displaced by the modal base
r and r' must be close to 1.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Input Data
* Participation factors for MODAL ANALYSIS or HARMONIC RESPONSE module
Input files:
DATA from static computation
STIFFNESS from static computation (if FORCE type loads)
RESULTS from static computation, required if MULTI is present or FORCE present without
RIGID label or LOAD command
MASSES
MODES
Input data:
PARTICIPATION command:

© 2019 ESI Group
CHAPTER 11 DYNAMIC
nv computation of participation factors corresponding to the first nv

eigenmodes (including rigid body modes). In the presence of the
DOUBLE label, computation are done in double precision
nv can be replaced by FREQUENCY fo or CIRCULAR
FREQUENCY ω0.
The eigenmodes of which the frequency is lower than fo (or the
circular frequency is lower than ω0) will be taken into account.
The minimal number is the number of rigid body modes (nvs) or 1 if
nvs is null.
In the lack of nv, FREQUENCY or CIRCULAR FREQUENCY, the
number of eigenmodes taken into account is that of the MODE file.
In the absence of the MULTI label, participation factors corresponding
to uniform unit accelerations for all supports are computed. By default,
the number of participation factors is equal to the number of
translational degrees of freedom for the option concerned (1, 2 or 3). In
the presence of DIRECTION label, only one excitation direction is
considered (c1, c2, c3 : director cosines). They are used for MONO-
EXCITATION type studies. These factors are reused with the
DISPLA label, together with those corresponding to the FORCE label
where appropriate.
In the presence of DIRECTION label, only results in the relative
reference frame can be given.
MULTI nd computation of participation factors corresponding to the first nd loads
of the static data imposed displacement type (movement of the support).
This label is not authorized if the eigenmodes have been computed by
the modal synthesis method (cf. Chapter 11.3).
FORCE nf computation of participation factors corresponding to the first nf loads of
the static data force type.
RIGID nvs recomputation of generalized masses for the first nvs eigenmodes which
are rigid body modes (eigenfrequency zero).
LOAD command (optional)
If this command is used, a static RESULT file is not required in the case of FORCE type loads.
Only the participation factors are obtained in this case.
This command must be used after computation of eigenmodes by the modal synthesis method
(ASSEMBLE DYNAMIC routine (cf. 11.3.6.4)) and FORCE type loads.
DISPLA ed1, ed2 number of imposed displacement type loads (movement of the support)
FORCE if1, if2 number of force type loads.
Results
DATA file with generalized masses and participation factors.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
The printout list also contains the modal base representativeness criteria: effective masses,
displaced masses and strain energies.
Example 1: Mono-excitation of the supports (PLANE BEAM option)
SEARCH DATA STIFFNESSES MASSES 1 MODES 2

DYNAMIC
PARTICIPATION 10
RETURN
SAVE DATA 100
The participation factors corresponding to uniform movements of the supports on the x and y
axes are computed. DATA 100 contains 2 modal loads.
Example 2: Mono-excitation of the supports and force (SHELL option)
SEARCH DATA STIFFN MASSES RESULT 1 MODES 2

DYNAMIC
PARTICIPATION 10 FORCE 2 RIGID 1
RETURN
SAVE DATA 100
The participation factors corresponding to uniform movements of the supports on x, y and z

axes, and the first two force type loads, are computed. There is a rigid body mode. DATA 100
contains 5 modal loads.
Example 3: Multi-excitation of supports and force
SEARCH DATA STIFF MASSES RESULT 1 MODES 2

DYNAMIC
PARTICIPATION 10 MULTI 4 FORCE 2
RETURN
SAVE DATA 100
The participation factors corresponding to the first form imposed displacement type loads and
the first two force type loads are computed.
DATA 100 contains 6 modal loads.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
* Participation factors for SPECTRAL ANALYSIS

In this type of analysis, the only force type loads authorized are unit accelerations in the seismic
directions studied. These loads are used to determine the residual modes (static corrections). In
a mono-excitation context (mono-spectral analysis), these are the only loads to be defined,
where appropriate. In a multi-excitation context, they must follow the imposed displacement
type loads describing the movements of the different supports.
No provision is made for the presence of rigid body modes.
Input files:
DATA from static computation
RESULTS from static computation if MULTI or RESIDUAL-TERMS
MASS (if fluid-structure interaction, use MATRIX with the method 1, and MASS
and MATRIX with the method 2)
MODES
Input data
PARTICIPATION command
nv computation of MONO-SPECTRAL participation factors for the first nv
eigenmodes. These correspond to uniform unit accelerations of all
supports, following the translational degree of freedom for the option
concerned. In the presence of the DOUBLE label, computation are done in
double precision.
nv can be replaced by FREQUENCY fo or CIRCULAR FREQUENCY
ω0.
The eigenmodes of which the frequency is lower than fo (or the circular
frequency is lower than ω0) will be taken into account.
In the lack of nv, FREQUENCY or CIRCULAR FREQUENCY, the
number of eigenmodes taken into account is that of the MODE file.
DIRECTION optional label used to define, for a mono-spectral study, a seismic
direction. Only this direction is taken into account subsequently (GN
label).
CX CY CZ c1, c2, c3 are the director cosines for this direction.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
RESIDUAL- TERMS command : if residual modes or static corrections are taken into
account.
Computation of residual terms, and inclusion of these modes in the MODE file. This MODE
file can only be used by a consistent SPECTRAL ANALYSIS, about level seismic directions
taken into account, with the RESIDUAL-TERMS command.
GN or GX, GY, Seismic directions
GZ
if the GN label is present, redefinition of director cosines c1, c2, c3 is
mandatory.
n or nx, ny, nz numbers of imposed acceleration loads defining static deflections.
MODE mc label and mandatory value
number of dynamic eigenmodes taken into account.
FREQUENCY fc optional label. The value of mc is recomputed, integrating all eigenmodes
for which the eigenfrequency is not greater than frequency of cutting fc.
<LABEL> at least one of the following result LABELS must be specified:
DISPLACEMENTS, REACTIONS, FORCES, STRESSES,
STRAINS or ACCELERATIONS. These results are printed if the
PRINT label is present.
In the presence of ACCELERATION label, the keyword
QUADRATIC starts up the calculation of residual modes in
acceleration with a quadratic summation of the contribution of
eigenmodes taken into account.
EQUIVALENT if the MODE file has been obtained by SEQUENCE (STRESS) (SHELL)
(BEAM) EQUIVALENT command (see section 11.2.1.3.2), these same
labels must be present to process FORCES, STRESSES, STRAINS
(contains transformation of the file static RESULT).
PARTICIPATION MULTI command: required in the case of a MULTI-SPECTRAL study.
nv computation of multispectral participation factors corresponding to the
first nv eigenmodes, and residual modes where appropriate.
m number of static data imposed displacement loads (movements of the
support) taken into account.
Notes
l Computation can be fractionized.
l If fluid-structure interaction is taken into account and if modes have been computed with the
method 1, the MASS file must be replaced by the MATRIX file. If modes have been computed with
the method 2, the MASS file must be recomputed. As a static RESULT file is not possible, the
RESIDUAL-TERMS and MULTI labels are not authorized.
l The RESIDUAL-TERMS and MULTI labels are not authorized after computation of eigenmodes by
the modal synthesis method (no static RESULT file).
l In the case of a MULTI-SPECTRAL study, the data file imposed displacement values are taken into
account for computation of participation factors, and pseudo-acceleration spectra must
consequently be defined.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
l In a fluid-structure interaction computation, the effective mass of the rigidly constrained degrees
of freedom is not accounted for. Nevertheless, this is compatible with a spectral dynamic analysis.
Results
Files:
DATA with generalized masses and participation factors.
MODES including residual modes (if RESIDUAL-TERMS label present).
The printout list also contains the representativeness criteria associated with imposed
displacement type loads: effective masses.
Example 1: Mono-spectral study with static correction (PLANE BEAM or BIDIM option)
DEFINITION
:
LOAD
1DIRECTION X / GX 1. NONE
2 DIRECTION Y / GY 1. NONE
RETURN
:
SEARCH DATA MASS RESULT 1 MODES 2
DYNAMIC
PARTICIPATION FREQUENCY 33
RESIDUAL-TERMS GX 1 GY 2 FREQ 33 DISPLA FORCE
RETURN
SAVE DATA 100 MODES 2

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example 2: Multi-spectral study without static correction (SHELL options) in one

direction
DEFINITION
:
CONSTRAINT
N 1 101 / UX (UY UZ RX RY RZ)
LOAD
1 NODE 1 X / NONE
N 1 / UX 1.
2 NODE 101 X / NONE
N 101 / UX 1.
RETURN
:
DYNAMIC
PARTICIPATION 10 MULTI 2
RETURN
SAVE DATA 100

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example 3: Multi-spectral study with static correction in two directions.
DEFINITION
:
CONSTRAINT
N 1 101 / UX UY
LOAD
1NODE1 Y / NONE
N 1 / UY 1.
2NODE1 Z / NONE
N 1 / UZ 1.
3NODE101 Y / NONE
N 101 / UY 1.
4NODE101 Y / NONE
N 101 / UZ 1.
5DIRECTION Y / GY 1. NONE
6DIRECTION Z / GZ 1. NONE
RETURN

DYNAMIC
PARTICIPATION 10
RESIDUAL-TERMS GY 5 GZ 6 MODES 6 DISPLA --
REACT ACCEL
RETURN
Computation can be done in three separate steps:

. Mono-spectral analysis without correction (6 modes).
SEARCH DATA MASS 1 MODES 2

DYNAMIC
PARTICIPATION 10
RETURN
SAVE DATA 11

© 2019 ESI Group
CHAPTER 11 DYNAMIC
. Mono-spectral analysis with correction (8 modes : 6 + 2)
SEARCH DATA 11 MASS RESULT 1 MODES 2

DYNAMIC
RESIDUAL-TERMS GY 5 GZ 6 MODES 6 DISPLA --
REACT ACCEL
RETURN
SAVE DATA MODES 11
. Multi-spectral analysis with correction (8 modes : 6 + 2)

DYNAMIC
RETURN
SAVE DATA 100

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.4.3 Transient Response (Modal Analysis)

General
For a linear system with a diagonal modal damping matrix, generalized coordinates q describing
the relative movement verify the following differential equations:
l elastic mode:
l rigid body mode:

F1(t) : force loads on free nodes
Fmγ(t) :modal forces due to movements of the supports
mG : eigenmode generalized mass.
This module allows time- related integration of these equations, and recombination of the
generalized coordinates, to obtain the relative movement physical response. Absolute physical
response can also be obtained.
DUHAMEL's semi-analytical method is used. This method is exact if the exciting forces vary
linearly during a time increment. Time increment dt must be selected in relation to period T min
for the highest frequency eigenmode (dt ≤ Tmin/50) .
The number of modes must be sufficient to represent the following correctly:
l evolution of the excitation in time,
l initial physical conditions: displacements Uo and velocities .
Representativeness criteria similar to those already described (cf. Chapter 11.4.2) are
computed. Initial conditions are those of the relative movement.
Force type excitations on the degrees of freedom can be accumulated with forces due to
movements of the supports.
Non linearities:
Stop or friction stop type non-linearities can be taken into account by adding the modal forces
corresponding to the physical pseudo-forces generated by the non-linear elements, to the second
member of the modal equations.
The selected modal base must allow correct representation of non-linearities. Generally:
l eigenmodes must be computed with minimum constrained stiffnesses (stops not included),
l the number of eigenmodes to be included in the modal base is greater than for linear studies,
so that modal superimposition can take account of the movement correctly.

Impact phenomena associated with creation of a contact are only represented in a frequency
domain below the maximum eigenfrequency of the modal base. This validity limit is more
restrictive, but also clearer than with the direct integration methods described in Chapter 12.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Analysis
Time-related modal analysis assumes execution of a number of modules.
In the most comprehensive case (absolute response for a multi-excitation context), five main
steps can be distinguished:
A - Static analysis (SOLVE).
B - Modal base (DYNAMIC MODES).
C - Participation factors (PARTICIPATION).
D - Evolution of generalized coordinates (MODAL ANALYSIS).
E - Evolution of physical quantities (COMBINATION).
Step A: Although this step is only absolutely necessary for the supports in a MULTI-
EXCITATION context, it is recommended in the case of force excitations, in order to be able to
compute the representativeness criteria of the selected modal base. Supports with imposed
movement form part of the constrained nodes.
Step B: This step computes the eigenmodes of the structure. Constraint conditions are
identical to those for step A.
Step C: This step computes modal forces, and various criteria characterizing the
representativeness of the modal base. The PARTICIPATION module is described in
paragraph 11.4.2.
Step D: This step computes the time-domain evolution of the generalized coordinates, taking
account of non-linearities. These modal acceleration velocity and displacement responses, as
also forces induced by non- linearities, are stored in the MODAL file. The MODAL
ANALYSIS module is described in paragraph 11.4.3.
Step E: This step computes the physical components for the desired results on the selected
elements and nodes. Where appropriate, these are stored in a TRANSIENT file, reduced to the
selected nodes and elements. A DATA file (also reduced) is associated with the TRANSIENT
file. The COMBINATION module is described in section 11.4.3.4.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
MODAL ANALYSIS - Description of Data

Evolution of the generalized coordinates, and of their first and second time-domain derivatives
(modal acceleration, velocity and displacement) are computed, and stored in the MODAL file
where appropriate.
Input files:
DATA resulting from the PARTICIPATION module
MODES resulting from computation of eigenmodes
RESULTS resulting from static computation (multi-excitation study)
if the initial modal conditions are rerun:
MODAL
if initial physical conditions:
STIFFNESS used for computation of eigenmodes
MASS
RESULT* in the case of rerun
TRANSIENT
* For this application, a direct search for this file is made by the INITIAL CONDITION
command.
Output files:
DATA
MODAL containing the generalized coordinates and their derivatives, and forces
induced by non-linearities.
The listing also contains maximum values for the generalized coordinates, and their time of
occurrence.
Input data

© 2019 ESI Group
CHAPTER 11 DYNAMIC
MODAL ANALYSIS command: initiates execution of the transient response module
nv number of the highest order eigenmode taken into account for computation.
DISPLA nd the first nd mono-excitation type support participation factors are used:
nd ≤ 3 (cannot coexist with MULTI).
MULTI nd the first nd multi-excitation type support participation factors are used
(cannot coexist with DISPLA).
FORCE nf the first nf imposed force type participation factors are used.
RIGID nvs there are nvs rigid body modes (eigenfrequency zero) for which generalized
mass has been recomputed (PARTICIPATION procedure).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
FLUID this label is optional if the structure-coupling (MODE FLUID

STRUCTURE) has been taken into account when calculating eigenmodes.
Only excitations of mono-excitation type support and forces are available
(labels DISPLA and FORCE). Only modal initial conditions are possible.
STIFFNESS and MASS matrices must be recalculated after the
eigenmodes calculation to take into account the inertia forces of the solid.
TITLE command optional
This command is used to associate a title (40 characters) with the combination of loads defined
by FUNCTION command.
By default, the title of load 1, if it is existing, is kept.
SELECT command (list) optional
Selection of eigenmodes effectively taken into account (see 11.3.4 for list syntax). If this
command is omitted, the first nv eigenmodes are taken into account.
FUNCTION command mandatory.
This command accumulates loads, taking account of the time functions defined in TABLE.
Transient computation is done for the resultant cumulative load. Two forms of syntax are
possible.
If no non-linearity is involved, and if response is only required in the relative reference frame,
only support acceleration functions, and functions associated with imposed force type loads are
necessary. The following syntax is used in this case:
FUNCTION i1,i2,...,ind, j1,j2,...,jnf
ik number of the table defining the acceleration function associated with the
kth support movement type load (1 ≤ k ≤ nd).
(jl , j2,...,jnf) numbers of tables associated with the first nf force type loads defined in
the data.
In the more general case, a triplet is required for each of the nd support movement type loads.
(ik1,ik2,ik3) numbers of displacements, velocity and acceleration tables associated with

the kth support movement type load (1 ≤ k ≤ nd).
(j1, j2, ..., jnf) numbers of tables associated with the first nf force type loads defined in
the data.
Notes
l nd and nf values cannot exceed 1024.
l If the simplified form of syntax is used, the FUNCTION command is followed by numbers only.
© 2019 ESI Group
CHAPTER 11 DYNAMIC
l For a multi-excitation study, the complete syntax form is recommended, only response in the
absolute reference frame having a physical significance.
TABLE command
All tables must be defined under this command, or under TABLE in the DEFINITION
procedure. In the latter case, the TABLE command must not appear at this point.
Sequential file type tables (FILE) are not authorized for modal analysis, and direct access files
must be used in their place (see Chapter 4.15.4 - tabulated file functions).
DAMPING command (optional)
A precise physical description of the parameters for this command is given in paragraph 11.4.1.
No damping is taken into account if this command is not present.
STOP command (PRINT n) (RELEASE) (optional)
If this command is present, one or more of the five types of non-linearity described below
(unidirectional or cylindrical stop, unidirectional or plane frictional stop and material stop) is
taken into account.
PRINT n non-linearity status and temporal non-linearities evolution are printed every
n computation steps
RELEASE contact terminates in the stop when the contact force is zero (F=(ku+c )).
By default, the contact terminates if differential displacement u is zero.
MODAL The prediction of the solution is made on the generalized coordinates.
METHOD m if the ALGORITHM VARIABLE STEP is used, m> 1 changes the
prediction method of the impact time.
Notes
l In the case of force loads only, or in a mono-excitation context, the gap criteria are expressed in
the relative reference frame.
l For multi-excitation, the gap criteria are in the absolute reference frame. In the case of contact
with the exterior, the "MULTI fd, fv, fγ" label specifies the numbers of the tables describing the
displacement, velocity and acceleration of the external medium in each direction where relevant.
l When a local reference frame is defined, the angles PSI, THETA and PHI are expressed in
degrees.
a - Unidirectional stop
n1 (n2) (INTERNAL) / (U) (R) MIN min MAX max --

STIFFNESS k DAMPING c (MULTI (fd, fv, fγ))
n1 only stop between node n1 and external medium

n1 and n2 stop between nodes n1 and n2
(U) displacement label, selected from UX UY UZ RX RY and RZ, indicating
impact direction, where appropriate, in a local reference frame (R) (PSI
THETA PHI),

© 2019 ESI Group
CHAPTER 11 DYNAMIC
min min. and max. gap values, counted algebraically on the axis defined by (U)
max
k stop stiffness
c stop damping
Note
l Node n1 is stopped by the external medium, or node n2 if
b - Cylindrical stop
Friction-free contact be two concentric cylinders with local axis z, the gap being measured
radially .
n1 (n2) (INTERNAL) / CYLINDER (R) MAX max STIFFNESS k --

DAMPING c (MULTI (fdx, fvx, fγx ) ,(fdy, fvy, fγy ))
c - Unidirectional frictional stop

This frictional stop acts in direction x only, when the d.o.f. of the associated stop are in contact.
The stop acts in direction z, except for PLANE BEAM and TWO-DIMENSIONAL options
where the stop is acting in direction y. These directions can be with respect to a local reference
frame. Frictional force is computed by the ODEN method.
n1 (n2) (INTERNAL) / FRICTION UNIDIRECTIONAL (FORCE f) --

(R) MIN min MAX max STIFFNESS k DAMPING c --
EPSI ε NUST ns NUDY nd (MULTI (fdx, fvx, fγx ),(fdz, fvz, fγz ))
UNIDIRECTIONAL
The frictional stop is considered to be unidirectional
n1, n2 same definition as before
min, max, k , c same definition as for a unidirectional stop (acting in z or y direction with
respect to the option)
FORCE f if this label is present, the frictional stop is said to be decoupled. The
two gap parameters of the stop must be defined. Tangential force is
calculated from f, NUST, NUDY, according to the value of EPSI.
Only tangential force is taken into account in the computation.
Is this label is not present, the frictional stop is said to be coupled. The
four parameters of the stop must be defined. Tangential force is
calculated from stop normal force and static (NUST) and dynamic
(NUDY) friction coefficients according to the value of EPSI.
Tangential and normal forces are taken into account in the computation.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
EPSI ε frictional threshold (ODEN method).

If relative tangential velocity is less than EPSI, there is static friction,
otherwise sliding (EPSI must be taken as a value between 1 and 10%
of predictable relative velocity).
NUST ns coefficient of static friction (dynamic friction occurs if tangential force
exceeds NUST times normal force) (default value ns = 0).
NUDY nd coefficient of dynamic friction (reversion to static friction occurs if
tangential force is less than NUDY times normal force N) (default value nd
= 0).
d - Plane frictional stop

This frictional stop acts in the plane (x,y), when the d.o.f. of the associated stop are in contact.
The stop acts in direction z. These directions can be with respect to a local reference frame.
The modulus of the frictional force is computed by the ODEN method.
n1 (n2) (INTERNAL) / FRICTION PLANE (FORCE f) (R)--

MIN min MAX max STIFFNESS k DAMPING c EPSI ε--
NUST ns NUDY nd (MULTI (fdx, fvx, fγx ),(fdy,fvy,fγy ),(fdz,fvz,fγz ))
PLANE The frictional stop is considered to be plane. Data are identical to those for
a unidirectional frictional stop (MULTI label followed by 3 triplets of
values).
e - Material stop
The material stop is used to take into account a force-displacement relation of plastic material
type.
This stop acts either in traction or in compression depending on the sign of the gap. If the gap is
zero, it acts in traction and compression, it is equivalent to an unidirectional stop.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Due to the history effect, restart calculations are not possible when the material behavior is
becoming plastic.
n1, n2 Number of the stop nodes. The number of nodes is compulsorily 2.

MATERIAL Label specifying this type of stop.
PLASTIC Optional label: with a cyclic behavior the stop acts like a plastic material.
(U) Degrees of freedom concerned by the stop (UX UY UZ RX RY
RZ).
(R) Stop reference frame (PSI, THETA and PHI expressed in degrees)
MIN min: Threshold parameter related to u1 – u2.
If MIN and MAX are zero the stop is linear (the stop acts in traction
and compression).
or MIN stop in 'compression' (min < 0, MAX ≅ + ∞).
MAX max Stop in "traction"( max > 0, MIN ≅ -∞)
This label acts in this way if the local axis of the stop is oriented from
node 2 toward node 1.
STIFFNESS k1 Elastic stiffness.
DAMPING c Constant damping.
ALPHA α Damping proportional to the tangent stiffness. If the stop is active a viscous
force is taken into account.
RAK2 k2 Plastic stiffness.
RAK3 k3 Unloading stiffness.
EPSI Non linearity threshold based on u1-u2 (always positive).
HISTORY ik ik ≥ 0 the stop takes into account the history effect (restart calculation is
impossible when the stop turns non-linear).
ik < 0 the stop is bilinear.
COEF xc xc > 1 the threshold variable MIN or MAX is divided by this parameter
when the stop turns non-linear.
ALGORITHM command (optional)
This command is followed by one or two labels (which can coexist), and is used to modify
certain parameters of the integration algorithm:

© 2019 ESI Group
CHAPTER 11 DYNAMIC
ZERO initial modal accelerations have imposed zero values, except if the initial
conditions are picked up from the MODAL file.
VARIABLE stop impact time between two consecutive computation times is
STEP nR recomputed. The minimum step is defined by
nR ( , by default nR = 5)
if this command is not present:
l Initial accelerations are computed by balancing modal forces.
l Impact times are not computed. Numerical integration using constant step is quicker but can
lead to numerical instability.

INITIAL CONDITIONS command (optional)
In the absence of this command, physical and modal initial conditions are taken with zero
values (displacement and velocity). Initial modal accelerations are computed in accordance
with balanced modal forces.
Initial modal or physical conditions can be imposed.
Initial modal conditions (MODAL file):
INITIAL CONDITIONS (RESTART) MODAL (CARD i)
Modal displacements and velocities are picked up from card i of the MODAL file, and modal
accelerations are computed from the equilibrium equation. This then becomes card (i+1) of the
MODAL file.
In the lack of the CARD label, the last card of the MODAL file is restarted. By default (lack of
the INITIAL label on the TIME command), the initial time of the computation is the one
corresponding to the chosen card.
Initial physical conditions (TRANSIENT or RESULT file)
nodes n1 ... nj / (U), (V)
Relative movement displacements and physical velocities can be defined from the following:
l TRANSIENT file (i : card number).
l RESULT file:no: file number. Velocity values are zero. Load 1 is taken into account.
l list syntax, taking precedence over the other definitions for the nodes listed.
(U) : UX, UY, UZ, RX, RY, RZ and corresponding values.

(V) : VX, VY, VZ, WX, WY, WZ and corresponding values.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Physical velocities and displacements are used to compute MODAL velocities and
displacements. MODAL accelerations are deduced from the equilibrium equation, or taken to
have zero values (ALGORITHM command).
Note
l If a TRANSIENT file generated by COMBINATION (Chapter 11.4.3.) is picked up by the
RESTART CARD command, displacement and velocity values for all nodes must have already
been computed and stored.
TIME command
Definition of computation times
TIME (INITIAL to)

ti STEP dti / DISPLACEMENT ie STORE is
...
to initial computation time (default value 0)

The ti STEP dti card subdivides time interval [t(i-1) ,ti] into time steps with constant length dti.
Modal coordinates are printed every ie interval, and stored in the MODAL file every is
intervals (ie and is not superiors with 1000). By default, printout occurs at times t(i-1), ti and
the data are stored at each time step. This command is followed by as many DATA commands
as required. The number of the card in the MODAL file is the number of the storage step.
The Modal Superimposition Nonlinear procedure is currently limited to 65536 instants (2 16
cards) by COMBINATIONS post-processor.
COMBINATION - Description of Data

The module is an essential postprocessor for the MODAL ANALYSIS command, and is used
to obtain physical results for selected nodes or elements only, and for selected cards.
Recombination is the most expensive operation in modal analysis.
For a mono-excitation study, results can be recombined:
l in the relative reference frame,
l in the absolute reference frame, introducing displacement and velocity functions, in addition
to acceleration functions, for each support movement type load.

For a multi-excitation study, results can be recombined:
l in the relative reference frame, although the results do not have a real physical significance,
l in the absolute reference frame.
Notes:
l Where any force type loads are considered, the relative and absolute reference frames are the
same.
l Where participation factors has been computed in a given direction (DIRECTION label), only
results in the relative reference frame can be obtained.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Input files :
Output files:
DATA
TRANSIENT
These files correspond to a geometry reduced to the selected nodes and elements (the DATA
file only contains geometry and material properties required for analysis of the TRANSIENT
file).
Listing:
Optional printout of requested results for nodes and elements, and selected cards. Printout of
maximum values and occurrence times, where appropriate.
Input data
DYNAMIC
COMBINATIONS nv (RELATIVE) (FLUID)
TITLE
title
SELECT i1 TO i2 STEP i3
(FUNCTION)((i11, i12, i13) ...(ind1, ind2, ind3))
(TABLE)
iT/ ...
PRINT (MAXI) (MODULUS) (COMPONENT)
(RESULT) i1 TO i2 STEP i3 (INTERNAL)
CARD n1 TO n2 STEP n3 (STORE )
...
RETURN
COMBINATION nv (RELATIVE) command

nv integration of first nv eigenmodes for recombination.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
RELATIVE used to obtain recombination in the relative reference frame (by default,
recombination in the absolute reference frame).
FLUID this label is optional if the fluid-structure coupling (MODE FLUID
STRUCTURE) has been taken into account when calculating eigenmodes.
Reduced pressures of the fluid are not recombined.
TITLE command
This optional command is used to associate a title (at most 40 characters) with the sum of loads
defined by FUNCTION command. By default the title of load 1 is kept.
SELECT command
Optional mode selection command.
Only modes specified in the list are used for recombination. If only one mode is selected, the
response on this mode is obtained.
The list must be a subset of the list defined in the MODAL ANALYSIS module.
FUNCTION command (optional)
If a MODAL ANALYSIS has been executed defining only the support acceleration functions,
displacement and velocity functions can be defined with this command, to obtain results in the
absolute reference frame. All TABLES must be redefined.
This command is present if the FUNCTION command is also present (see Chap. 11.4.3.3).
PRINT command
No printout or storage occurs if this command is absent.
If the MAXI label is present, maximum values for the results requested for any of the selected
cards are printed, with their corresponding occurrence times.
The COMPONENT and MODULUS labels can coexist, and one of these labels must at least
exist.
If the COMPONENT label is present, maximum values are computed component by
component.
If the MODULUS label is present, maximum modulus values are computed.
RESULTS command
These commands specify list of nodes or element used to define a reduced geometry. The
requested results are computed for a reduced geometry. The type of results requested must exist
in the RESULTS and MODES files. A number of RESULT commands can exist
simultaneously. No computation is done in the absence of a RESULTS command.
The types of result are selected by means of the following labels:
(RESULTS) = DISPLACEMENT list of nodes
REACTION list of nodes connected to the external medium,or belonging to a
STOP

© 2019 ESI Group
CHAPTER 11 DYNAMIC
The REACTION label is used to obtain, in addition, the force or forces exerted by the nodes on
the non-linearities (action of n1 on the exterior, or n1 on n2).
RESULTS can also be replaced by NODES, synonymous with DISPLACEMENTS and
REACTIONS (if DISPLACEMENTS is used, VELOCITIES and ACCELERATIONS are
automatically computed).
(RESULTS) = FORCE list of elements
(if FORCES the reduced geometry contains the nodes of the selected elements).
Forces in beams are computed in the element local axes. The local damping matrices are
neglected. If local reference frames are defined in the material properties, they are ignored for
this computation. .
For elements other than beams, forces are computed by combination of modal forces. In these
cases, it is supposed that the relation between displacement and acceleration is still available
mode by mode . Modal forces must exist in the MODES file (SOLVE FORCES
MODAL - 11.2.1.3.3). For all elements, using SOLVE FORCES TRANSIENT (12.3.5) after
DYNAMIC COMBINATION is used to obtain forces precisely, by neglecting however for
beams the loads of span, but taking into account concentrated dampings possibly defined in the
characteristics .
CARD command
Card selection command for result printout and storage.
n1 TO n2 STEP list of card numbers for printout or storage.
n3
STORAGE this label indicates that the command selects cards to be stored.
There can be one or two CARD commands.
No storage or printout is executed in the absence of a CARD command.
If the CARD command does not have the STORE label, printout and storage are done
simultaneously.
If the STORE label exists for one of the two CARD commands:
l printout is executed in accordance with the command not containing the STORE label.
l storage is executed in accordance with the command containing the STORE label.
Note
l In the case of variable step computations (see 11.4.3.3 ALGORITHM command), if n3 is too high,
the computation will stop at the last card for which the generalized coordinates were computed.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example: Transient response
DEFINITION
MODAL ANALYSIS NON LINEAR * STOP IN MULTI-EXCITATION CASE
OPTION PLANE BEAMS
ONE-DIMENSION
NODES
1/ 1. 0.
21/ 1. 0.5081
ELEMENTS
1/ 1 21 20
MATERIALS
/E=2.07*11 NU=0.3 IZ=8.32*-7 S=3.87*-3 RO=7483
CONSTRAINT
1/UX UY RZ
LOAD
1 EXCITATION HORIZONTAL /
NODE 1/UX=1
RETURN
SAVE STIFFNESS MASS DATA 2
SOLVE
SAVE RESULT 2
DYNAMIC
MODES GIVENS
VECTORS 10
RETURN
SAVE MODES 2
SEARCH DATA STIFFNESS MASS MODES RESULT 2
DYNAMIC
MODAL ANALYSIS 10 MULTI 1
FUNCTION MULTI 1 29
TABLE
1/2, (0.001,0.,0.) (0.011,0.423333*-5,0.127*-2) (0.016, 1.164167*-5, 0.127*-2)
2/2, (0.001,0.,0.) (0.011,0.127*-2,0.254) (0.016,0.127*-2,-0.254)
9 / 1, (.001,0), (.011 0.254) (0.016,-0.254)
STOP
21 / UX MIN=-1.27*-7 MAX=1.27*-7 STIFFNESS 2.50*5 MULTI 1 2 9
INITIAL TIME 0.001
0.015 STEP 0.000125/DISPLACEMENT 200 STORE 1
RETURN
SAVE DATA MODAL 22
...

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example: (continued)
SEARCH DATA MODAL 22 MODES 2

DYNAMIC
COMBINE 10 RELATIVE
PRINT
DISPLACEMENT 1,11,21
RETURN
SAVE DATA TRANSI 23
SEARCH DATA MODAL 22

SEARCH STIFFNESS MASS MODES RESU LT 2
DYNAMIC
COMBINE 10
PRINT
DISPLACEMENT 1,11,21
FORCE 1
REACTION1,21
CARD 0 TO 112 STEP 8
RETURN
SAVE DATA TRANSI 24
END

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.4.4 Harmonic Response

General
The response of a linear structure, subjected to loads with sinusoidal variation in time, is
studied. All loads vary at the same frequency, and are in phase. The dynamic equilibrium
equation for the structure is written:
This problem is considered as a physical variable problem in Chapter 11.5 (absolute movement,
any damping matrix). With modal methods, generalized coordinates q k , associated with each
eigenmode, and describing relative movement with respect to the static deflection imposed by
the supports, verify the analytical equation:
The physical solution, in amplitude-phase form, is obtained by combining the contribution of

each eigenmode.
Transfer Functions
Transfer functions H(ω) are basic tools for the study of random vibrations. They describe the
dynamic amplification of the response for each excitation frequency. These functions are used
for stochastic dynamic studies in particular (cf. Chapter 11.4.5 ). There are two equivalent
mathematical definitions:
l H(ω) is the ratio of the Fourier transform of the response to the Fourier transform of the
excitation:
o Computation of harmonic response combined with a frequency scan, provides the transfer
function. The modal method is consequently particularly suitable for their determination.
l H(ω) is also the Fourier transform of unit impulse response h(t) to a Dirac delta function at
t=0.
This definition can be generalized to a transient transfer function:
which can be computed analytically for each eigenmode, and is obtained in the physical space
by combining the different modal contributions.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Input Data
Input files:
DATA obtained from PARTICIPATION, containing participation factors pk
MODES obtained from computation of modes, containing modal forces if this
type of result is requested (SOLVE FORCES MODAL )

© 2019 ESI Group
CHAPTER 11 DYNAMIC
HARMONIC RESPONSE MODAL command

nv number of highest order eigenmode taken into account for computation.
DISPLA nd the first nd mono-excitation type support participation factors are used
(cannot coexist with MULTI).
MULTI nd the first nd multi-excitation type support participation factors are used
(cannot coexist with DISPLA).
OMEGA where the DISPLA or MULTI label (support movements) is present,
excitations correspond to acceleration Γ, constant according to circular
frequency (Γ = 1 for DISPLA, but for MULTI, the value of Γ is the
value specified in the corresponding load). The OMEGA label makes
it possible to have a constant displacement load, according to circular
frequency (U = 1 for DISPLA, or U = load value for MULTI).
If only FORCE label exists, the amplitude of excitation forces is
multiplied by the squared circular frequency (forces of unbalance type).
FORCE nf the first nf imposed force type participation factors are used.
RIGID nvs if rigid body modes exist, this label is used for the computation of
harmonic responses only:
l to set to zero their circular frequencies (excitation frequency can not be
equal to zero)
l to define a variable damping mode by mode (the RIGID label must be
present with DAMPING command)
FLUID this label is optional if the fluid-structure coupling (MODE FLUID
STRUCTURE) has been taken into account when calculating
eigenmodes. Only excitations of mono-excitation type support and
force are available (labels DISPLA and FORCE).
SELECT command (list) (optional)
Selection of modes effectively taken into account (Chapter 1.3.4 for list syntax details). If this
command is omitted, the first nv modes are taken into account.
DAMPING command (optional)
A precise physical description of the parameters for this command is given in Chapter 11.4.1. If
this command is absent, no damping is taken into account.
FUNCTION command (optional)
In the presence of this command, the amplitude of nd loadings of type displacement of support
and of nf loadings in force, is multiplied by the value of the corresponding table. Table
abscissae are frequencies or circular frequencies according as frequency scanning is defined by
FREQUENCY label or CIRCULAR FREQUENCY label. A table number, negative or equal
to zero, amounts to define an unit function. The effect of the multiplication factor thus defined
eventually adds to the effect associated with OMEGA label.

© 2019 ESI Group
CHAPTER 11 DYNAMIC

All tables have to be defined under this command or under TABLE in the DEFINITION
procedure. If the last one is used, the TABLE command does not have to exist here.
FREQUENCY (or CIRCULAR FREQUENCY) command (mandatory)
Definition of the minimum frequency or circular frequency scan step (the two labels cannot
coexist).
INITIAL fo scan start frequency (by default fo = 0)
There is no computation for this frequency.
fp STEP dfp definition of computation frequencies in the interval (fp, fp-1). All
results are stored in the TRANSIENT file. If DISPLA label is present,
the results are printed every np computation steps (no printout if
DISPLA label is not present; if np is not present results are printed for
fp only)
MODE command (optional)
Definition of the scan step in the vicinity of the eigenmodes contained in the minimum scan
window defined by the FREQUENCY command. Only the chosen frequencies (SELECT) and
not rigid body (RIGID) are treated. If there is no damping, this command is ineffective.
If fm is the eigenfrequency computed, the (2nm+1) frequencies obtained by dividing the interval
[ fm(1-r), fm(1+r) ] into 2 nm equal intervals, are added to the minimum scan window. If r = 0,
only eigenfrequencies processed are added. PRECISION command is used to suppress
additional frequencies which are too close (by default ε = 10-5).
TRANSFER command (mandatory)
Labels :
STATIONARY computation of stationary transfer functions or harmonic response (default
label).
TIME ttr computation of transient transfer functions for time ttr.
ABSOLUTE Response is computed in the absolute reference frame.
By default, response is computed in the relative reference frame. If
only FORCE type loads exist, the absolute and relative reference
frames are identical.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
RESULT label:
At least one or more labels, with one per data line, selected from the following:
DISPLACEMENT VELOCITY ACCELERATION REACTION FORCE
A list of nodes (ni) or elements (mi) is associated with each type of result.
A new reduced geometry is defined from the various lists of nodes and elements, comprising:
l Nodes for which at least one result is explicitly requested,
l Nodes belonging to an element for which forces are requested,
l Elements for which forces are requested.
Computation is then executed for each type of result explicitly requested, for all nodes or
elements of the reduced geometry.
Results
Files:
DATA reduced geometry at nodes and elements, selected by
DISPLACEMENT, VELOCITY, ACCELERATION, REACTIONS
or FORCE commands.
TRANSIENT result cards numbered sequentially according to ascending order of
frequencies, in modulus and phase form:
Modulus (positive) odd card
Phase (between -π and π) even card
Notes
l In contrast to HARMONIC RESPONSE computation (11.5), results are in the relative reference
frame, except in a case where the ABSOLUTE label can be used with the TRANSFER command.
l In the DATA and TRANSIENT files, the order of loads is redefined by the computation modules.
The loads corresponding to movements of the support are followed by force type loads.
l TRANSIENT DYNAMIC computation (11.7.4) can be used to combine phase-offset excitations.
Results are given in time-related form.
SOLVE FORCES TRANSIENT (12.3.5) after TRANSIENT DYNAMIC is used to obtain coherent
forces with displacement, velocity and acceleration fields. This forces can locally be different from
those directly obtained by combination of modal forces, especially in the proximity of concentrated
loads.
l Transfer function units depend on the type of result and the nature of load (response/excitation
ratio).
l The conversion of a transient file, outcoming from multi-excitation in modal dynamic response, by
using the command CONVERT is not possible directly. It is necessary to copy the cards of a local
TRANSIENT file in a RESULT file with the command MERGE CARD.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example: Harmonic response - Comparison of direct and modal methods in the

absolute reference frame.
Direct method - absolute displacements of the complete geometry
DEFINITION
COMPARE RHD RHM ABSOLUTE REFERENCE FRAME
OPTION PLANE BEAMS
ONE-DIMENSION
NODES
1 / 1. 0.
21 / 1. 0.5081
ELEMENTS
1 / 1 21 20
MATERIAL
/ E = 2.07 * 11 NU = 0.3 IZ = 8 . 32 * - 7 S = 3 . 87 * - 3 RHO = 7483
CONSTRAINT
1 / UX UY RZ
LOAD
1 /
NODE 1 / UX = 1
RETURN
SOLVE STIFFNESS mass
SAVE STIFFNESS mass DATA 2
DYNAMIC
HARMONIC RESPONSE
FREQUENCY
30 STEP 10 / DISPLA 1
RETURN
Note
l The direct method is described in Chapter 11.5.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Modal method - Displacements, velocities and forces on a selected geometry
SEARCH DATA STIFFNESS mass 2

DYNAMIC
MODE SUBSPACE
VECTOR 10 PRINT -1 FORCE
RETURN
SAVE MODES 2
DYNAMIC
PARTICIPATION 10
HARMONIC RESPONSE MODAL 10 DISPLA 1 omega
FREQUENCY
30 STEP 10 / DISPLA 1
TRANSFER STATIONARY ABSOLUTE
DISPLA 1 11 21
VELOCITY 1 11 21
FORCE 1 10 20
RETURN
OMEGA required to obtain an imposed displacement amplitude which is

constant with frequency. The load effectively taken into account is that
generated by PARTICIPATION. Amplitude is 1.
ABSOLUTE response in absolute reference frame.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.4.5 Stochastic Dynamics

General
This module is used to compute and analyze the statistical response parameters for a structure
subjected to random loads. The loads are assumed stationary, ergodic and with zero mean
value, and known by their power spectral density or PSD (Fourier transform of the signal
autocorrelation function). The real parts of the cross-PSD associated with the intercorrelation
functions between two loads, can also be taken into account. The PSD S(ω) is defined by the
following usual theoretical formula:
dimension : X2 / (rd/s) where X is the physical value considered (force, velocity,…)

it is to within 2π the Fourier transform of the autocorrelation R(τ) of temporal signal. This
formula is also known as the Wiener-Khintchine relation.
Complete computation is in three steps:
l determination of the PSD of response Sx , from the PSD values for the excitations and
transfer functions H(ω), corresponding to generalization of the formula used for a system
with one degree of freedom (Sx = H2 PSDexcitation).
l determination of statistical moments Mi of the response, by integration of Sx with respect to
the circular frequency: . The zero order moment is used to evaluate the
mean value of the response.
l in the case of Gaussian signals, analysis of statistical moments and computation of material
property parameters: bandwidth, mean number of times a given level is exceeded, statistical
data on absolute and local maximum values, and fatigue analysis.
Before this module is used, the transfer functions must be computed by HARMONIC
RESPONSE MODAL (Chapter 11.4.4). Each of the three steps is covered by a separate
submodule.
Response Power Spectral Density

Input files (created by HARMONIC RESPONSE MODAL)
DATA this file contains the definition of the reduced geometry
TRANSIENT this file contains the modulus and phase of the transfer functions

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Input data
DYNAMIC
PSD (CROSS) (LOG) (EDIT)
FUNCTION n1,...,nk
CALCULATE (list of result labels)
TABLE (FREQUENCY)
n /1,(........)
CARD na TO nb ... STEP nc
RETURN
SAVE DATA TRANSIENT n
PSD command
This command initiates computation of the response PSD for the structure.
The CROSS PSD label indicates integration of the real part of the cross power spectral densities
between the different loads, in the computation.
The LOG label selects bi- logarithmic interpolation in the load spectrum tables (linear
interpolation by default). In the presence of this label, tables are necessary type 1. (Chapter
4.15.3).
The EDIT label is used to edit the response power spectral density (PSD).
FUNCTION n1, ... ,nk command (mandatory)
Omitting cross PSD values, k is equal to j, the number of loads for which transfer functions
have been computed.
If cross PSD values are taken into account, k is equal to j+j.(j-1)/2
List n1,..,nk is used to give a list of the load spectrum tables (classed in succession by imposed
displacement and force on the free DOF, and then in ascending order of load numbers
introduced), followed by a list of cross PSD tables, classified coherently by load pair with the
load classification.
Example with 4 loads
FUNCTION sp1 sp2 sp3 sp4isp12 isp13 isp14 isp23 isp24 isp34
with
spi = PSD for load i
ispij = cross PSD between loads i and j

© 2019 ESI Group
CHAPTER 11 DYNAMIC
CALCULATE command (mandatory)

This command defines the types of result to be computed (DISPLACEMENT, VELOCITY,
ACCELERATION, REACTION, FORCE). Transfer functions for the results requested must
have been computed beforehand.
TABLE command
All tables must be specified simultaneously with this command, or DEFINITION (see Chapter
4.15).
The FREQUENCY label with the TABLE command indicates that the table abscissas are
frequencies, in which case all tables must be defined under PSD. The excitation PSD is not
modified by the label FREQUENCY, it remains expressed by rd/s.
In the presence of the LOG label, under PSD, if the first number of the table is different of 1,
PSD values are multiplicated by this number mandatory positive.
CARD command (mandatory)
This command is used to define frequencies or circular frequencies for which the PSD is
computed. These frequencies must be selected from the cards stored for the transfer functions
computed with HARMONIC RESPONSE MODAL, and stored in the TRANSIENT file.
Only cards containing amplitudes are mentioned, identified by their order numbers in the
TRANSIENT file (order number ≤ 32 000).
Results:
Output files:
DATA
TRANSIENT this file contains the response PSD for the reduced structure. Each card
corresponds to a requested frequency. PSD values for the requested
results are computed under each card.
Statistical Moments for Power Spectral Densities

Input files (created by PSD)
DATA
TRANSIENT This file contains response PSD values.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Input data
DYNAMIC
INTEGRATION
CARD n1 TO n2 STEP n3
MOMENT( m1,..,mj) (PRINT)
RETURN
SAVE DATA DISPLA VELOCITY ACCELERATION REACTION FORCE n
INTEGRATION command (mandatory)

This command initiates computation of statistical moments by DSP integration.
CARD command (mandatory)
List of cards to be taken into account for integration.
This card indicates the type of result to be computed. This must exist in the TRANSIENT file.
(possible labels: DISPLACEMENT, VELOCITY, ACCELERATION, REACTION,
FORCE)
MOMENT command (mandatory)
m1,...mj list of moment orders to be computed by increasing order. Odd order
moments are not significant. The essential moments for computation of
statistical parameters are 0, 2 and 4. They are computed by default (empty
list). In the presence of the PRINT label all computed moments are printed.
By default, only the first computed moment is printed (the square root of
M0 has a physical sense).
Results
Output files:
DATA this file contains as many loads as there are computed moments – mj ≤ 30.
RESULT the PSD integration moments for each result are saved to this file. The
0 order moment is a special case. It is its square root which is stored
and printed.
SAVE RESULT only saves DISPLACEMENT, REACTION and
FORCE. SAVE DISPLACEMENT VELOCITY ACCELERATION
REACTION FORCE must be used to save velocity and acceleration
moments. The 0 order moment represents the good approximation of
the mean value of the response.
Notes
l Statistical moments Mi are computed by PSD integration for response Sx:

© 2019 ESI Group
CHAPTER 11 DYNAMIC
l The trapeze integration method is used. Frequency scan must be sufficiently fine in the vicinity of
a relatively undamped eigenfrequency.
l Integration is done between 0 and +∞, which explains why the odd order moments are non-zero.
Statistical Parameters for a Centered Gaussian Stationary Ergodic Signal

After execution of the INTEGRATION module (cf. 11.5.4), this module can be used to obtain
conventional statistical parameters corresponding to a very common signal, by means of
analytical formulae. Mathematical definition of these parameters can be found in the theoretical
manual. Input moments are 0, 2 and 4.
Input files: (created by INTEGRATION)
DATA
RESULT this file must be called by the command:
SEARCH DISPLACEMENT VELOCITY ACCELERATION
REACTION FORCE if velocity and acceleration results are to be
handled.
Input data:
DYNAMIC
STOCHASTIC n1 n2 n3
BANDWIDTH
LEVEL CROSSING j
u1,...,uj
MAXIMA k
r1,...,rk
FATIGUE COEF c EXPONENT a
RETURN
SAVE DATA DISPLA VELOCITY ACCEL REACTION FORCE n
STOCHASTIC command
This command initiates computation of statistical parameters.
n1, n2, n3 are load numbers corresponding to order 0, 2 and 4 moments for integration
computation. By default of every list, ni = i
This command specifies the type of result computed (list of possible labels:
DISPLACEMENT, VELOCITY, ACCELERATION, REACTION, FORCE).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
BANDWIDTH command (optional)

This command computes available response bandwidths. Bandwidth is denoted by ε ; this
parameter ε is between 0 and 1.
0: pure harmonic signal
1: any signal
LEVEL CROSSING command (optional)
This command computes the mean number of crossings of level u, in the up and down
directions, by the response concerned. This number is obtained by time unit.
j number of values on u
u1,...,uj values of u
MAXIMA command (optional)

This command computes:
l mean number of local maxima by time unit,
l mean number of positive local maxima per time unit,
l mean maximum value for a signal of length t,
l standard maximum deviation for a signal with length t:
o k number of signal lengths
o r1,...,rk values of t.
FATIGUE command (optional)

This command computes the mean damage to a structure by random fatigue analysis. The
fatigue law is of the WOEHLER type, and makes it possible to express the admissible number
of cycles N as a function of physical parameter S considered. The command CALCULATE
must be followed by one result label only, usually DISPLACEMENT or FORCE.
COEFFICIENT fatigue curve multiplier coefficient c

EXPONENT fatigue curve exponent a (0 < a < 66).
Computed damage D corresponds to a mean damage value per time unit, from which we can
deduce:
l time at fracture: T = 1/D
l number of cycles to fracture = ν(0) + * T

where ν(0) + = number of positive slope crossings of level 0 (ν(0) + = 0.5 * ν(0))
ν(0) is obtained with the LEVEL CROSSING command.
The value of the damage calculated by the command FATIGUE is even more accurate than the
bandwidth ε of the signal analyzed is small (ε < 0.2).
Results
Output file:

© 2019 ESI Group
CHAPTER 11 DYNAMIC
DATA
RESULT
These two files contain as many loads as there are computed parameters, in the same order as
for printout.
The listing contains all requested results.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example: stochastic dynamic computation of a beam under bending load
DEFINITION
STOCHASTIC - POUTRE CRANDALL ET MARK
OPTION PLANE BEAM
GEOMETRY
NODE
1/
2 / 0. 4.
ELEMENT
1/12
MATERIAL CARACTERISTICS
ELEM/E10.3*6AX 0.0625IZ3.23*-4VZVPZ0.125
CONSTRAINTS
N 1 / UX UY RZ
N 2 / UY
MASS
N 2 / AX 7.28*-4
RETURN
; eigen modes
DYNAMIC
MODES
VECTOR 1 FORCE NORM ELASTIC
RETURN
; introduction of stresses in MODE file
STRESSES
SEQUENCY BEAM
; transfer function in DISPLA. and FORCES
DYNAMIC
PARTI
HARMONIC RESPONSE MODAL DISPLA
DAMPING GAMMA 0.01
FREQUENCY
2000. STEP 10.
MODE 20 RATIO 0.1
TRANSFER STATIONNARY
DISPLA 2
FORCE 1
RETURN

© 2019 ESI Group
CHAPTER 11 DYNAMIC
; P.S.D. response to accel.x (P.S.D. response)
DYNAMIC
PSD
FUNCTION 1 2 2
CALCULATE DISPLA FORCES
TABLE FREQUENCY
1/1(0.0.)(19.90.)(20.5930.)(2000.5930.)(2000.10.)
2 / 1 (0. 0.) (2000.1 0.)
CARDS 1 TO 241 STEP 1
RETURN
; statistical moments
DYNAMIC
INTEGRATION
CARDS 1 TO 241 STEP 1
CALCULATE DISPLA FORCES
MOMENT
RETURN
SAVE DATA RESULTS 3
; displacement analysis
DYNAMIC
STOCHASTIC
CALCULATE DISPLA
BANDWITH
LEVEL 2
0. 0.001
MAXIMA 3
10. 100. 1000.
FATIGUECOEFFICIENT0.31EXPONENT6.09
RETURN
SAVE DATA RESULTS 4
; stresses analysis ( fatigue sigma=241431.*u )

COMBINATION STORED FORCES; unit changement forstresses(coeff=1.*-3)
STRESSES USUAL UNITS / 1.*-3 1

© 2019 ESI Group
CHAPTER 11 DYNAMIC

RETURN
DYNAMIC
STOCHASTIC
CALCULATE FORCES
LEVEL1;onlyvalidfor0.levelbecausechangeofunits
0.
FATIGUECOEFFICIENT1.006*14EXPONENT6.09
RETURN
SAVE DATA RESULTS 5
END

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.4.6 Seismic Response - Spectral Method

General
This module describes a special method for the study of linear structures subjected to the
movement of their supports. It is designed for seismic studies:
l Translational support movements are described principally by pseudo-acceleration response
spectra (translational movements).

l Results are combined in accordance with statistical rules, taking account of the particularities
of seismic excitation, essentially according to the RG 1.92 rev. 1 rules of the US Nuclear
Regulatory Commission. The ZPA method defined in RG 1.92 rev. 3 is not available. For
SYSTUS, amplified modes are combined with the usual combinations available (SRSS,
CQC etc.) up to the frequency cutting of the analysis, i.e. up to F point defined on RG 1.92
curves. For these modes, no distinction is performed between periodic and rigid components.
All modes with a frequency higher than the frequency cutting are taken into account by using
a pseudo-mode. The contribution of these rigid modes is evaluated by using the structure
static deflection when it is subjected to a uniform acceleration load. The overall response is
then derived by taking the square root of the sum of the square of amplified response and
rigid response.
Absolute movement U1 of the free nodes is broken down into two parts:
U1 = Ue + Ur
Ue : static driving movement resulting from displacement U2 of the supports.
Ur : relative movement with respect to the static deflection imposed by the supports.
U r depends on acceleration U 2 of the supports. This movement is studied by projection in the
modal base corresponding to U 2 = 0. This modal base can also include the residual modes
(static corrections).
Two types of analysis are possible:
l If all the supports have the same movement, the analysis is said to be MONO-SPECTRAL.
All nodes constrained by the external medium are assumed to possess this movement. This
concerns the non-modeled end of an elastic constraint. Only the relative movement is
studied, according to the pseudo-acceleration spectra.
l If the supports have different movements, the analysis is said to be MULTI-SPECTRAL.
Imposed displacement type loads are used to identify the different groups of supports. These
loads are the first of the static computation. Relative movement and driving movement are
studied according to the pseudo-acceleration spectra, and maximum support displacement
data.
Analysis Procedure
Spectral analysis assumes execution of a number of modules. Chaining of the modules differs
according to whether the excitation is mono- or multi-spectral.
Distinction is made between four main steps:
A - Static analysis (SOLVE)

© 2019 ESI Group
CHAPTER 11 DYNAMIC
B - Modal analysis (DYNAMIC MODES)

C - Participation factors (PARTICIPATION)
D - Spectral analysis proper, or combination of the different types of response.
A- Static analysis
The structure is described using the DEFINITION routines. All degrees of freedom subject to
excitation must be declared in the CONSTRAINT command. Description of loads differs
according to the type of analysis:
a) Mono-spectral analysis
Static computation is only required if it is desired to take static corrections into account later in
the analysis. In this case, the loads are unit accelerations in the seismic directions studied.
b) Multi-spectral analysis
As many loads (imposed displacements) must be specified as there are groups of supports with
different spectra. These are followed by acceleration loads if static corrections are to be taken
into account. Acceleration loads must be defined last.
B - Modal analysis
This step consists in computing the eigenmodes of the structure, with constraint conditions
identical to those for the static analysis.
C - Computation of participation factors
This step is used:
l To compute the contribution of each eigenmode to the response, in the form of participation
coefficients for each excitation.
l To determine whether the selected eigenmode base can suitably describe the response, by
printing a certain number of criteria (effective modal masses, the sum of which is compared
with the effective total mass with unconstrained degrees of freedom, for each excitation
analyzed).
l To compute static corrections which allow for modes omitted from the analysis, to ensure
correct representation of the static component of the seismic response (not amplified by
inertial effects).
D - Analysis by spectral combination
This final step provides the seismic response physical results. This response is not obtained in a
fully deterministic manner.
It is the result of the following operations:
l Computation of the response of each mode for a given direction or excitation (value read in
the spectrum, multiplied by the modal participation factor).
l Combination of modal responses for a given direction or excitation. These combinations are
based on statistical considerations.

We have a choice between the conventional methods of the RG1.92 code of practice of the
US Nuclear Regulatory Commission, or the C.Q.C. (Complete Quadratic Combination)
method.
l Integration of the contribution of residual modes.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
l For multi-spectral analysis only, combination of responses to excitations in one direction, to

obtain the response for this direction.
l Combination of responses for each direction.
l For multi-spectral analysis only, combination of static responses to obtain the driving
response, and combination with the modal responses.
Special Options
There are two special options for this type of analysis. The SHELL AXISYMMETRICAL
HARMONIC FLUID option embodies the particularities of both.
FLUID-STRUCTURE options
The mass required for computation of modal participation factors must take account of the
added mass due to the fluid. For execution of the PARTICIPATION module (Chapter 11.4.2),
the MASS file must be replaced by the MATRIX file, created after computation of eigenmodes,
and taking account of fluid-structure interaction.
As a static RESULT file is impossible, only mono-spectral analysis is possible without static
corrections.
Only the combination of fluid reduced pressures has a physical meaning. For the fluid the
reduced pressures associated with the walls movement are always obtained (processing of
displacements or accelerations).
AXISYMMETRICAL HARMONIC options (SHELL or THREE DIMENSIONAL)
Multi-spectral analysis is not possible.
Support excitations are axisymmetrical. Axis of symmetry OY must be vertical, but the position
of the horizontal seismic directions is not imposed.
A vertical translational movement, parallel to OY, is axisymmetrical, and corresponds to
HARMONIC 0. For a unit translation, the components at each point are as follows:
UX = 0 UY = 1 UZ = 0 (RZ = 0).
Only the axisymmetrical modes are loaded, or more precisely those which are symmetrical with
respect to any meridian plane, such that UZ = 0 at any point. Antisymmetrical torsion modes,
where only UZ is non-zero, can only be loaded by rotation round the axis of symmetry.
A unit translation in direction X corresponds to a radial displacement equal to cos θ, and a
circumferential displacement equal to -sin θ, namely HARMONIC 1 SYMMETRICAL, with
the following support excitation components:
UX = 1 UY = 0 UZ = - 1 (RZ = 0)
Only the modes of this harmonic are loaded.
A unit translation in direction Z, deduced from the previous translation by rotating through
+π/2, obviously corresponds to HARMONIC 1 ANTISYMMETRICAL, and only loads the
modes of this harmonic, deduced from those of the symmetrical harmonic (rotation through
π/2). Radial displacement is equal to sin θ, and circumferential displacement to cos θ, giving
the components:

© 2019 ESI Group
CHAPTER 11 DYNAMIC
UX = 1 UY = 0 UZ = 1 (RZ = 0).

Decomposition of excitation is consequently particularly easy, as it involves only the first two
circumferential orders: with only one harmonic corresponding to each seismic direction. Each
of the three unidirectional seismic excitations therefore only loads one set of modes, and a given
mode only responds to one excitation.
The user must select a symmetry for computation of modes and participation factors. The
problem of integration of two symmetries for effective response computation consequently
disappears in the case of spectral analysis.
The spectral analysis module can only compute one harmonic at a time, and therefore for one
direction at a time. It is left to the user to combine the directions externally.
Although the program can compute modes and participation factors for harmonic 1, both
symmetrical and antisymmetrical, the identity of a rotation close to the two sets of modes
corresponds, in practice, to determining one only.
Harmonic 0 gives the response to excitation in the direction OY (GY).
Harmonic 1 symmetrical gives the response excitation in the direction OX (GX).
Harmonic 1 antisymmetrical gives the response excitation in the direction OZ (GZ). If the x
and z spectra are identical, one or other of the last two responses is sufficient (isotropy).
MONO-SPECTRAL Analysis
All supports have identical movements. Loads corresponding to support movements (unit
accelerations) are generated by the PARTICIPATION module.
Preliminary steps
l Step 0: static DATA file
constraints with the external medium
unit acceleration loads (GX, GY, GZ) if static corrections.
l Step 1: static results (SOLVE) if static corrections
l Step 2: modal analysis (Chapter 11.2)
DYNAMIC MODES
(SOLVE FORCES MODAL: if forces and accelerations).
l Step 3: participation factors (cf 11.4.2.3)

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Note
l If DIRECTION and RESIDUAL-TERMS GN are used, the static load for step 0 must be in this
direction, by combination of the elementary accelerations defined by GX, GY, GZ
(for example GX 0.707 GY - 0.707).
Input data
Input files:
DATA from PARTICIPATION
MODES from MODES computation without static corrections
from PARTICIPATION with static corrections.
Input data

© 2019 ESI Group
CHAPTER 11 DYNAMIC
SPECTRAL ANALYSIS command

GN or describe seismic directions, with GN corresponding to a direction
GX GY GZ described on computation of the participation factors.
Static corrections are taken into account if the RESIDUAL-TERMS label is used. The pseudo-
acceleration associated with each residual mode is the last table value describing the excitation
spectrum in its direction. This value is also used to correct reactions at supports.
The complete list of results is printed if the EDIT label is used.
COMBINE command (optional command)
In the lack of this command, all types of result included in the MODES file are treated.
Otherwise, the labels of the COMBINE command describe the types of result to be analyzed,
and the corresponding result must exist in the MODES file.
Labels:
DISPLACEMENTS
REACTIONS If a local reference frame has been defined for certain nodes, reactions are
expressed in this reference frame.
FORCES
STRESSES
STRAINS
ACCELERATIONS Combinations concerning this last type can be saved with the following
command:
SAVE DATA RESULTS ACCELERATION i
MODE command
In the MODE command, n is the number of dynamic modes combined. It can be defined by the
frequency of cutting fc. If static corrections are used, this number must be coherent with the
number of modes used for their computation in PARTICIPATION.
The following labels describe modal response combination methods. At least one of these
labels must be given; otherwise the label by default is taken into account.
CQCS or CQCA labels can be followed by:

© 2019 ESI Group
CHAPTER 11 DYNAMIC
PRINT simplified analysis of correlation coefficients existing between modes.

LIMIT lim threshold on correlation coefficients.
in the presence of this label, only correlation coefficients superior to lim
are taken into account.
by default lim = 0.001 (lim ≤ 0.5)
DURATION command
This command is required in the presence of the DOUBLE SUMMATION label in the MODE
command.
t is the duration of the seismic loading.
DAMPING command
This command is required in the presence of the DOUBLE SUMMATION, CQCS or CQCA
label in the MODE command, or if the label FILE is specified with the command
SPECTRUM.
The critical damping fraction γ associated with each mode, in coherence with the damping
values used to calculate the excitation spectra, can be defined by three ways:
l GAMMA γ: damping identical to all modes
l MODAL: allows to define a value per mode (unsaid modes γ = 0)
l FILE nf: dampings are searched in DAMPnf.DAT file created by DAMPING (11.7.1).
Additional labels can exist :

o METHOD i: i = 1 weighting by strain energies (default)
i ≥ 2 weighting by kinetic energies

oLOWER inf: γ minimal value
l UPPER sup: γ maximal value
GAMMA, MODAL, FILE labels can not coexist.
SPECTRUM command
This command associates table numbers in each excitation direction, in direction order.
In the absence of FILE label, these numbers are table numbers defined by the TABLE
command. In the presence of RESIDUAL-TERMS label, the acceleration associated with static
correction is the last value of the table.
In the presence of FILE label, these numbers define SPECix.DAT files.
These files are of the same type of those created by GENERATION SPECTRA (Chapter
11.7.2).
Ground maximal accelerations have to be defined with the help of the ACCELERATION
label. They are used to correct reactions at support levels and in the presence of RESIDUAL-
TERMS label are associated with static correction. Dampings have to be defined for each
mode. PRINT label brings about the printing of spectra actually taken into account.
DIRECTION command

© 2019 ESI Group
CHAPTER 11 DYNAMIC
The labels of the DIRECTION command describe the methods for response combination in
each excitation direction. At least one of the following labels must be given; otherwise the label
by default is taken into account.
SRSS (by default)
ABSOLUTE
SEMI-SRSS(in this case, there must be 3 directions for the label to be meaningful).
TABLE command
Tables ix, iy and iz defining the pseudo-acceleration spectra:
l are type 1 (Chapter 4.15.3). If the first number of the table is different from 1, the pseudo-
acceleration values are multiplicated by this number mandatory positive (case of the table iy).
l comprise pairs (frequency in Hz, acceleration in units of acceleration),
l must not contain zero values (bi-logorithmic interpolation).
It is necessary to use this command in the absence of FILE label on SPECTRUM command.
Results
Files:
DATA
the number of loads depends on selected combination methods
RESULTS
In the presence of the EDIT label, the listing contains all results, if not it contains only the load
titles generated by the analysis.
The quantities obtained are relative movement amplitudes.
Accelerations represent a special case. With the inclusion of static corrections, a good
approximation of absolute accelerations is obtained. Otherwise these are relative accelerations.
The acceleration storage must be explicitly requested (SAVE DATA RESULTS
ACCELERATION i commands).
The reactions at nodes linked to a local frame are also a particular case (CONSTRAINT
command), they are defined in the local frame of the node.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example: MONO-SPECTRAL analysis with static corrections
DEFINITION
MONO-SPECTRAL ANALYSIS WITH STATIC CORRECTIONS
OPTION SPATIAL BEAMS
ONE-DIMENSION
NODE
1/
11/ 10. 0. 0.
ELEMENT
1/ 1 11 10
MATERIALS
/E=2*11NU=0.3 RHO 7800 S 0.025 IX IY IZ 0.0002
CONSTRAINT
N 1 11 / UX UY UZ RX RY
LOAD
1 ACCELERATION GY / GY 1.
2 ACCELERATION GZ / GZ 1.
RETURN
SOLVE STIFFNESS MASS MATRIX RESULTS

SAVE DATA RESULTS STIFFNESS MASS 1000
DYNAMIC
MODES
VECTORS 6
RETURN
SOLVE FORCES MODAL
SAVE MODES 1000
SEARCH DATA STIFFNESS MASS RESULT MODES 1000
DYNAMIC
PARTICIPATION 6
RESIDUAL-TERMS GY 1 GZ 2 MODES3 DISPLA REAC FORCE ACCEL PRINT
RETURN
SEARCH DATA 1001 MODES 1000

DYNAMIC
SPECTRAL ANALYSIS GY GZ RESIDUAL-TERMS
COMBINE DISPLACEMENT REACTION FORCE ACCELERATION
MODES 3SRSS DOUBLE SUMMATION
DURATION 20
DAMPING GAMMA 0.02
SPECTRUM 1 2
DIRECTION SRSS
TABLES

© 2019 ESI Group
CHAPTER 11 DYNAMIC
1 / 1 0. 1 1. 3. 1. 5. 10. 25. 10. 30. 1. 100. 1.
2 / 1 0. 1 2. 3. 2. 6. 15. 30. 15. 33. 1.5 100. 1.5
RETURN
END

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Table 11.4 - Mono-spectral analysis (with static corrections)

© 2019 ESI Group
CHAPTER 11 DYNAMIC
MULTI-SPECTRAL Analysis
The different sets of supports have different movements, and are identified by means of
imposed displacement type loads.
The main steps of the analysis are summarized in Table 11. 5.
Preliminary steps
l Step 0: static DATA file
o Constraints with the external medium. If local reference frames are defined for the nodes,
support movements are expressed in these reference frames. Generally, only identical
reference frames for all nodes are significant in a seismic context, and static corrections are
only strictly valid in this case.
o Imposed displacement loads defining the amplitude and direction of the excitation (in the
local reference frame for the nodes).

o Unit acceleration loads for static correction (in the local reference frame where appropriate,
by combination of GX, GY and GZ).

l Step 1 : Static results (SOLVE)
l Step 2 : Modal analysis (Chapter 11.2)
DYNAMIC MODES
(SOLVE FORCES MODAL: if forces and accelerations)
l Step 4 : Participation factors (cf. 11.4.2.3)
DYNAMIC
PARTICIPATION nv
RESIDUAL-TERMSGX nx GY ny GZnz MODES mc--
(FREQUENCY fc) <LABEL> (EQUIVALENT) (PRINT)
PARTICIPATION nv MULTI m
RETURN
Note
l If static corrections are not included, only the PARTICIPATION nv MULTI m command is
necessary.
Input data
Input files:
DATA from PARTICIPATION MULTI
RESULTS from SOLVE (static computation)
MODES from MODES no static corrections
from PARTICIPATION with static corrections

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Input data
DYNAMIC
SPECTRAL ANALYSIS MULTI GX n1x ... nix GY n1y ... niy GZ n1z ... niz--
(RESIDUAL-TERMS) (EDIT)
COMBINE (LABELS)
STATIC (LABELS)
DISPLACEMENTS U1, U2, ..., Um
MODE n (FREQUENCY fc) (LABELS)
DURATION t
DAMPING (GAMMA γ ) (MODAL) (FILE nf)
j1 TO j2 STEP j3 / (GAMMA) γj
SPECTRUM i1 i2, ..., im(FILE ACCEL g1 g2 ... gm)
DIRECTION (LABELS)
TABLES
i1/1 (fn, San)
:
im/1 "
RETURN
The COMBINE, MODE, DURATION, DAMPING and TABLE commands are identical to
those for MONO-SPECTRAL analysis.
SPECTRAL ANALYSIS MULTIPLE command
GX n1x ... nix, GY n1y ... niy GZ n1z ... niz
nix, niy, niz are static load numbers (imposed displacements associated
with each direction X, Y, Z).
RESIDUAL- inclusion of static corrections
TERMS
The pseudo-acceleration associated with each residual mode is the last maximal table value
describing the excitation spectrum in its direction. This value is also used to correct reactions at
supports.
EDIT Complete printing of the results.
STATIC command:
This command is used to describe methods for combination of static results with the following
labels:
l SRSS (by default)
l ABSOLUTE
l MIXED
At least one of these labels must be given; otherwise the label by default is taken into account.
In the lack of this command, only the relative movement with respect to the static deflection
imposed by the supports is determinated.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
DISPLACEMENT command:
U1, U2, ..., Um are multiplier factors, used to obtain maximum values for support physical
displacements from the displacement values specified in the m imposed displacement type static
loads (determination of driving movement due to the supports).
If the STATIC command is existed, it must be immediately followed by the DISPLACEMENT
command.
SPECTRUM command:
This command associates the table numbers with each imposed displacement type static load in
load order.
In the absence of FILE label, these numbers are table numbers defined by the TABLE
command. In the presence of RESIDUAL-TERMS label, the acceleration linked with static
correction is the last value of the table.
In the presence of FILE label, these numbers define SPECix.DAT files.
These files are of the same type of those created by GENERATION SPECTRA (Chapter
11.7.2).
Ground maximal accelerations have to be defined with the help of the ACCELERATION
label. They are used to correct reactions at support levels and in the presence of RESIDUAL-
TERMS label are associated with static correction. Dampings have to be defined for each
mode. PRINT label brings about the printing of spectra actually taken into account.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
DIRECTION command
This command enables the user to manage the combination of the excitations due to different
supports in one direction and the combination of the basic directions (X, Y, Z)
Groups of supports in one direction. Two methods exist:
l Root-sum squares: QUADRATIC and SEMI-QUADRATIC labels,
l Simple summation of the response of each support: ABSOLUTE, MIX labels. At the end of
this stage, we have the dynamic responses for each direction (X, Y, Z). In the presence of the
CORRECTION label, the unit static correction (calculated in the CORRECTION module)
for the considered direction, is summed in a quadratic manner to the dynamic response. The
acceleration taken into account when calculating the static correction is the maximum of the
last values of the pseudo-acceleration spectra in each of the basic directions (X, Y, Z). If
pseudo-acceleration spectra are given on file, the value of the acceleration taken into account
in each direction must be given.
Summation of the basic directions
The end of the procedure is carried out by taking into account the response in each elementary
direction and taking into account one of the four possible labels: QUADRATIC, SEMI-
QUADRATIC, ABSOLUTE, MIX
At this stage of the calculation, we have relative responses compared to the static deformation
imposed by the absolute displacements of the various groups of supports. The quadratic sum
(mandatory) of the static deformation due to the supports displacements (Commands STATICS,
DISPLACEMENT) and of the relative dynamic responses due to the supports accelerations
enable to the obtain the final result.
Notes:
l If a stress analysis (or forces in continuous media with the bending stresses method, especially
used in the nuclear domain, has been carried out :
l The driving movement creates secondary stresses (Q),
l The relative movement with respect to the static deflection creates primary stresses (P).
l This remark justifies the distinction between different kinds of results.
l It is obvious that total stresses must be considered as primary stresses.
Results
Files:
DATA
the number of loads depends on the selected combination method
RESULTS
In the presence of the EDIT label, the listing contains all results, if not it contains only the load
titles generated by the analysis.
Quantity amplitudes are obtained for the following:
l driving movement (static),
l relative movement with respect to the static deflection imposed by the supports,
l absolute movement.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Accelerations represent a special case (no driving acceleration). With the inclusion of static
corrections, a good approximation of absolute accelerations is obtained, otherwise these are
relative accelerations. The accelerations storage must be explicitly requested (command SAVE
DATA RESULTS ACCELERATION i).
The reactions at nodes linked to a local frame are also a particular case (CONSTRAINT
command), they are defined in the local frame of the node.
Example: Multi-spectral analysis with static corrections
DEFINITION
MULTI-SPECTRAL ANALYSIS WITH STATIC CORRECTIONS
OPTION SPATIAL BEAMS
ONE-DIMENSION
NODE
1/
11/10.0.0.
ELEMENTS
1/1 1110
MATERIALS
/E=2*11NU=0.3 R0 7800 S 0.025 IX IY IZ 0.0002
CONSTRAINT
N 111 / UX UYUZRXRY
LOAD
1 NODE 1 - UY=1/ NONE
N 1 / UY 1.
2 NODE 1 - UZ=1/ NONE
N 1 / UZ 1.
3 NODE 11 - UY=1/ NONE
N 11 / UY 1.
4 NODE 11 - UZ=1/ NONE
N 11 / UZ 1.
5 ACCELERATION GY / GY 1. NONE
6 ACCELERATION GZ / GZ 1. NONE
RETURN
SOLVE
SOLVE MASS
DYNAMIC
MODES GIVENS
VECTORS 6 PRINT -1
RETURN

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example: (continued)
SOLVE FORCES MODAL

SAVE MODES 9000
SEARCH DATA MASS RESULT MODES 9000

DYNAMIC
PARTICIPATION 6
RESIDUAL-TERMS GY 5 GZ 6 MODE 3 DISPL REAC FORCE --
ACCEL PRINT
RETURN
DYNAMIC
RETURN
SAVE DATA 9002
SEARCH DATA 9002 RESULTS MODES 9000

DYNAMIC
SPECTRALANALYSISMULTI GY 1 3 GZ 2 4 RESIDUAL-TERMS
COMBINEDISPLACEMENTS REACTION FORCE ACCELERATION
STATIC SRSS ABSOLUTE
DISPLACEMENTS 0.01 0.01 0.01 0.01
MODES 3 SRSS
SPECTRA 1 2 3 4
DIRECTION SRSS ABSOLUTE
TABLE
1/ 1 0.3 2. 0.7 22 1 22. 1.75 10. 10 10. 35 3. 1000. 3.
2/ 1 0.3 1. 2.6 8. 9 12. 12. 12. 26. 3.5 1000. 3.5
3/ 1 0.3 2. 0.7 16. 1. 16. 2.5 14. 5. 20. 21. 20. 50 15. --
90. 9. 81 1000 9.81
4/ 1 0.3 1. 0.7 8. 9 15. 12. 15. 28. 4.5 1000. 4.5
RETURN
END

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Table 11.5: Multi-spectral analysis (with static corrections)

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.4.7 Consistant Modal Matrices Method

Generalities
The classical modal method described in the Chapter 11.4.6 supposes the modal damping
matrices can be diagonalized. Considering the orthogonal properties of the eigenmodes, beside
the stiffness and mass matrices, lead to disconnected modal equations. Critical damping fraction
associated with each real eigenmode is often an experimental data. However in the lack of
testing, the damping fraction must be determined by the followings:
l The computation of complex eigenmodes of dissipatif system,
l Energetic weighting method.
Damping can invert the eigenfrequency order.

Particularly in the presence of punctual damping, the energetic methods can prove very rought.
This paragraph presents a variant of the modal methods which treats the consistant damping
matrix projected on the subspace defined by the real eigenmodes of the conservative structure.
The modal stiffness and mass matrices can also be consitant, which allows to take into account
physical stiffness and mass modifications without recomputed the real eigenmodes.
Notes:
l Both computation modules work only with the new data structure.
l The generalized forces or the participation coefficients must be computed in double precision
(PARTICIPATION DOUBLE modulus).
l In the presence of rigid body eigenmodes, mass modifications can not be defined.
Transient Response
This modulus solves in time the modal equation of an elastic structure:
by a Newmark’s temporal scheme with or without numerical damping. It is used to obtain

physical displacements, velocities and accelerations.
Input files:
DATA resulting from the PARTICIPATION DOUBLE modulus.

MODES resulting from DYNAMIC MODES (displacements in double precision).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
SEARCH DATA MODES i

DYNAMIC
PSEUDO MODAL TRANSIENT
ELEMENTS
TABLES*
ALGORITHM (ALPHA α) (BETA β GAMMA γ) (ZERO)
TITLES
Title
ELEMENTS
M1 TO Mn
FUNCTION i1…ind j1…jf
TABLES*
i1 /
:
:
jf /
INITIAL CONDITIONS (RESTART) (CARD nc)
(list nodes) / UX…VX…
TIME (INITIAL t0)
ti STEP dti / STORE n1
tj STEP dtj / STORE n2
RETURN
SAVE DATA TRANSIENT n
* The TABLE command must appear only once and defines all tables useful for the modulus.
PSEUDO MODAL TRANSIENT
initiates execution of the modulus.
ELEMENTS, optional commands of structure modification (§ Chapter 11.6.3).
TABLES
ALGORITHM Newmark’s parameter definition.
or α (default 0.05)
ZERO : initial null modal accelerations.

This command follows necessary structure modification commands if
they are existing.
TITLE Possibility to define a title for the transient studied.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
ELEMENTS Selection of the geometry by the elements (default: no selection). The

reduced geometry includes all nodes belonging to the elements selected.
FUNCTION Number of the tables affected to the different loads defined by
PARTICIPATION modulus.
Remark
l In the first one, there are generally three support movement type loads
TABLES* Defined here or in DEFINITION.

INITIAL Restart of physical initial conditions. Usual Syntaxis.
CONDITIONS
TIME (INITIAL Definition of time data.
t0)
Output files:
DATA with one load

TRANSIENT including displacements, velocities and accelerations
Note
l As TRANSIENT ELASTIC, forces in the elements can be obtained with the SOLVE FORCE
TRANSIENT command.
Harmonic Response
This modulus is used to solve, for a circular frequency or frequency given, the modal equation
of an elastic structure:
Input files:
DATA resulting from PARTICIPATION DOUBLE

MODE resulting from DYNAMIC MODES (nodal values in double precision).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
OMEGA label
PSEUDO MODAL HARMONIC
initiates the module execution.
ELEMENTS, optional commands of the structure modification (Chapter 11.6.3).
TABLES
FREQUENCY (INITIAL f0)
defines the frequency scan.
This command follows necessary the commands of the structure
modification if they are existing.
MODE Command
Definition of the scan step in the vicinity of the eigenmodes contained in the minimum scan
window defined by the FREQUENCY command.
If fm is the eigenfrequency computed, the (2nm+1) frequencies obtained by dividing the interval
[ fm(1-r), fm(1+r) ] into 2 nm equal intervals, are added to the minimum scan window. If r = 0,
only eigenfrequencies processed are added. PRECISION command is used to suppress
additional frequencies which are too close (by default ε = 10-5).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
ELEMENTS Selection of the geometry by the elements (default: no selection). The

reduced geometry includes all nodes belonging to the chosen elements.
DISPLACEMENT, VELOCITY, ACCELERATION, REACTION, FORCE
Choice of the physical result type. These one must exist in the MODE file
(Particularly case : VELOCITY, ACCELERATION).
Remark:
l For the continuous media options, if the MODE file proceeds from the SOLVE FORCE command
at the end of the eigenmodes computation
SOLVE FORCE
AVERAGING STRESS STRAINS
:
:
RETURN
SEQUENCE (EQUIVALENT STRESS)
l Labels which can exist are FORCE, STRESS (NODES), STRAINS (NODES).
Output files:
DATA Number of loads equal to this generated by the PARTICIPATION

modulus.
TRANSENT So many pair of cards as computation frequencies.
Odd cards: amplitude,
Even cards: phase (radian).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.5 Harmonic Loads - Direct Method
11.5 HARMONIC LOADS - DIRECT METHOD

11.5.1 General
Structure behavior is linear. Forces and imposed displacements vary sinusoidally at the same
frequency, with no mutual phase offset. We look for a special solution, which reflects the
permanent state of the system, without the influence of initial conditions.
The dynamic equilibrium of the structure is written:
A solution in the form is obtained by solving a linear system for each value
of circular frequency ω.
No damping (K - Mω2) X = F Y=0
The solution is infinite for the circular eigenfrequencies of the system. A real component, with
its sign, is obtained.
With damping
The moduli and phases of the solution are obtained.

Damping can be viscous or structural (hysteretical).
Matrix C V corresponding to viscous damping is constituted from Basile coefficients α and β.
These coefficients can be defined globally or element by element. In this case the damping file
(C0) must be created by the SOLVE command:
Matrix C H corresponding to structural damping is constituted from a structural damping

coefficient or a loss factor η. This factor can be defined either globally (GAMMA label), or
element per element (ALPHA label in the material properties):
Coefficients γ and α are equal to half of the usual loss factor η:

E*=E(1+jη)

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Notes
l Only one damping file exists as input file of this procedure. C0 is of viscous type (GENERALIZED
label) or structural (STRUCTURAL label).
l With usual notations for one degree of freedom systems:
l Definition of γ guarantees homogeneity with modal methods.

l In case of a structural damping, if a β label is defined in the material properties, it is taken into
account without having any precise mechanical definition.
l The treatment of harmonic response problems is possible in thermal .

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.5.2 Input Data

Input files:
DATA file created by DEFINITION, which must contain the load or loads for
which harmonic response is computed.
STIFFNESS files created from the DATA file by the command. SOLVE
MASS STIFFNESS MASS (DAMPING).
DAMPING This file is only required if GENERALIZED or STRUCTURAL type
damping is applied.
MODE This file is only required in the presence of the MODE command.
Input data
DYNAMIC
HARMONIC RESPONSE (VELOCITY) (ACCELERATION)(REACTION)
DAMPING (ALPHA α) (BETA β) (GAMMA γ) ((GENERALIZED) (STRUCTURAL))
(METHOD <label> ( COMPLEX))
CIRCULAR FREQUENCY INITIALw0
FREQUENCY
ω1 STEP dω1 / DISPLA n
ω2 STEP dω2/ DISPLA n
...
MODES nm (RATIO r) (PRECISION ε) (COMPLEX)
RETURN
RESPONSE command
If the VELOCITY and ACCELERATION labels are present, velocities and accelerations are
computed in amplitude and phase form as displacements.
In the presence of the REACTION label, reactions and residual in forces are computed. The
maximal residual amplitude force is printed.
DAMPING command
If the GENERALIZED or STRUCTURAL labels are present, damping Co is extracted from
the DAMPING file. The various damping forms are added. If STRUCTURAL label is present
Co is of structural type.
This command is used to choose solver type. In the presence of damping, a complex solver is
required.
<label>
without damping SKYLINE, OPTIMISED
with damping SKYLINE, SPARSE, ITERATIV (COMPLEX)

© 2019 ESI Group
CHAPTER 11 DYNAMIC
For more information about this command and the properties of the various solvers available see
Chapter 21.
By default, standard solvers are used.
CIRCULAR FREQUENCY or FREQUENCY command
Computation circular frequencies or frequencies are introduced after the command of the same
name, followed by the initial bound of the interval (default value 0). Each interval (ωi-1 , ωi ) is
then subdivided into an integer number of subintervals, each with length dωi.
The DISPLA n parameter forces printout of moduli and phases every n intervals. In the
absence of this label there is no printout. Cards are stored on TRANSIENT file at each
subinterval.
MODE Command: optional
In the presence of a MODE file, definition of the step of balayage near the eigenfrequencies
included in the minimum balayage window defined by the FREQUENCY command. In the
lack of damping, this command is ineffective.
If fm is the treated eigenfrequencies, the (2nm+1) frequencies obtained by dividing the interval
[fm(1-r), fm(1+r)] into 2 nm equal intervals are added to the minimum balayage0 If r is null, only
the treated eigenfrequencies are added. The PRECISION label is used to suppress the
additional frequencies too close (by default ε = 10-5).
If the MODE file contains complex eigenmodes (modulus SUBSPACE BLOC DAMP or
PSEUDO MODAL DAMP), the COMPLEX label is required. If the system is enough
damped, the insertion of complex eigenfrequencies is better.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.5.3 Results
Results (displacement, velocity, acceleration) are obtained in the absolute reference frame.
Computation of reactions and forces is not available under complex form.
Output files:
DATA unmodified
TRANSIENT only complex results are stored in this file, in the form:
. with damping: modulus and phase,
. no damping: real component.
MODULUS (positive) or real component odd card
PHASE (between - π and π) even card
Notes
l TRANSIENT DYNAMIC (Chapter 11.7.4) can be used to obtain displacements, velocities and
accelerations under time-related form, with eventually combination of out of phase loads .
l SOLVE FORCE TRANSIENT can be used to associate dynamic forces to time-related response.
l This command is also used to associate with the dynamic forces (real or complex) the harmonic
response (real or complex).
Example 1:
Viscous damping
SEARCH DATA STIFFNESS MASS DAMP 1

DYNAMIC
HARMONIC RESPONSE
DAMPING GENERALIZED ALPHA 0.001
FREQUENCY INITIAL 100.
110 STEP 5. / DISPLA 1
200 STEP 9. / DISPLA 5
RETURN
SAVE TRANSIENT 1
Example 2:
Viscous and structural damping
SEARCH DATA STIFFNESS DAMPING 1

DYNAMIC
HARMONIC RESPONSE
DAMPING ALPHA 0.001 BETA 2 GAMMA 0.02 STRUCTURAL
FREQUENCY
10 STEP 5. / DISPL 1
RETURN

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example 3:
Generalized damping using iterative method and force calculation in a magnetodynamic
problem.
SEARCH DATA STIFF DAMP 1

DYNAMIC
HARMONIC RESPONSE
DAMPING GENERALIZED
METHOD ITERATIVE COMPLEX
FREQUENCY
50 /
RETURN
SAVE DATA TRAN1
SOLVE FORCE TRANSIENT COMPLEX CARD 1
SAVE DATA TRAN 2
SEARCH DATA TRAN 1
SOLVE FORCE TRANSIENT COMPLEX REAL CARD 1
SAVE DATA TRAN 3
The TRANSIENT file number 2 contains the results stored in amplitude and phase while
TRANSIENT file number 3 contains results stored in real or imaginary part.
Example 4:
Harmonic response in a thermal problem.
SEARCH DATA 1
DYNAMIC
HARMONIC RESPONSE
DAMPING GENERALIZED
(METHOD COMPLEX <LABEL>)
CIRCULAR FREQUENCY
:
RETURN
<LABEL> SKYLINE
DIRECT (SPARSE)
ITERATIVE (SPARSE)
(for more precision, see chapter 21).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.6 Sensitivity Analysis
11.6 SENSITIVITY ANALYSIS

11.6.1 General
This module computes the eigenfrequency and eigenmode variation of an initially conservative
structure, in which damping or isolated stiffness or mass modifications are introduced.
Modifications applied to the structure are projected in the computed real eigenmode sub-space.
The validity of the two methods available is limited to the capacity of the real modal base to
represent the modifications introduced.
The first method conducts an analytical re-analysis in the modal base. This is adapted for rapid
examination of isolated stiffness, mass or damping modifications. All results appear in the
printout list.
The second method involves complete re-analysis in the modal base. This method is suitable
for computing complex eigenfrequencies resulting from distributed damping, or isolated
stiffness, mass or damping modifications defined by matrices which can be non-symmetrical.
Corresponding eigen deflections can be obtained.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.6.2 Analytical Re-Analysis in the Modal Base

Modifications applied to the structure can take any of the following forms:
l addition of a mass, damping or stiffness value at a given point,
l addition of a damping or stiffness value between two points of the structure which can belong
to the same element or not.

The reference frame in which the modifications are applied must be indicated (implicitly the
general reference frame).
Each modification corresponds to a unit variation of the quantity concerned.
An eigenfrequency and eigenmode complex number formulation is used. This is essential for
damping modifications. The sensitivity of the eigenmodes (modal deflections) can only be
obtained if the structure does not have multiple eigenfrequencies.
Input Data
Input files:
DATA
for the unmodified model
MODES
Input data
DYNAMIC
SENSITIVITY nv1 VECTOR nv2
MODIFICATION (list of modifications)
NODE (list of nodes)
ELEMENT (list of elements)
COUPLE (list of pairs of nodes)
RESPONSE
DISPLACEMENTS (list of nodes)
AXES ref (RESPONSE)(MODIFICATION)
RETURN
SENSITIVITY command
nv1 number of modes for which sensitivities are computed,
nv2 number of modes of the modal base in which computation is performed
(nv1 ≤ nv2).
MODIFICATION command
(list of list of labels indicating the type of modification applied to the structure
modifications)
AX, AY, AZ addition of mass in directions x, y, z
IX, IY, IZ addition of rotational inertia about x, y, z axes

© 2019 ESI Group
CHAPTER 11 DYNAMIC
CX, CY, CZ addition of damping in directions x, y, z

CRX, CRY, CRZ addition of damping in rotational directions x, y, z
KX, KY, KZ addition of stiffness in directions x, y, z
HX, HY, HZ addition of stiffness in rotational directions x, y, z
This command must be followed by one or more of the NODE, ELEMENT or COUPLE
subcommands (COUPLE subcommands must appear after NODE or ELEMENT
subcommands).
NODE subcommand
List of nodes on which modifications are applied
Example:
NODE 1, 5, 10 TO 15 INTERNAL
ELEMENT subcommand
List of elements on which modifications are applied.
These are two-node elements.
Addition of a mass value between two nodes has no meaning.
Example:
ELEMENT 8, 4, 12 TO 18
COUPLE subcommand
List of pairs of nodes (not connected by existing elements), between which changes are applied.
Addition of a mass value between two nodes has no meaning.
Example:
COUPLE (4, 5), (10, 20)
RESPONSE command
This command initiates computation of sensitivities on mode shape. The list of nodes for the
DISPLACEMENT subcommand specifies where the eigendisplacement modifications are
computed.
AXES command
This command is used to define the reference frame, by means of its Bryan angles (Chapter
4.4.4), in which:
l modifications are defined (MODIFICATION label),
l responses are computed (RESPONSE label).
The RESPONSE and MODIFICATION labels can coexist.
Results
All results appear in the listing. No file is created.
Sensitivity of circular eigenfrequencies (or eigenfrequencies)

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Complex circular eigenfrequency s = α + jω corresponds to real circular eigenfrequency ω of

the initial conservative structure. Modification of stiffness and mass introduces modification
∆ω of ω. The program then gives ∆ω and ∆ω/ω.
A damping modification induces modification of α (initially zero). The program gives ∆α=α.
Sensitivity of mode shape
Displacement of a degree of freedom is represented by its modulus and phase.
A stiffness or mass modification induces modification of the modulus of the displacement.
A damping modification induces a phase modification of the displacement. By convention, the
initial phase is taken at -π/4 if the real eigendisplacement is positive, or 3π/4 if the real
eigendisplacement is negative.
Example:
l Sensitivity on circular eigenfrequencies

DYNAMIC
SENSITIVITY 8 VECTORS 8
MODIFS KX HZ AX IZ
NODE 11 18
RETURN
l Sensitivity on circular eigenfrequencies and mode shape

DYNAMIC
SENSITIVITY 4 VECTORS 8
MODIFS KX KY KZ CX CY CZ
NODE 11
RESPONSE
DISPLACEMENT 11 18
AXES MODIF PSI 90.
RETURN

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.6.3 Complete Re-Analysis in the Modal Base

This method is used to compute complex eigenfrequencies and eigenmodes where:
l Symmetrical damping values are defined by a damping file,
l Individual stiffness, mass and damping modifications, whether symmetrical or not, are
defined in the existing elements.
Input Data
Input data files required:
DATA
MODES (real eigendisplacements)
DAMPING
STIFFNESS required for computation of modal matrices, reactions, forces in beams and
MASS Rayleigh quotients.
Input data
DYNAMIC
PSEUDO MODALnv(DAMPING ONLY)(ORTHOGONALITY)--
(NORM <LABEL>) (EXTRACT)
SELECT na TO nb STEP nc
ELEMENT
ndTO neSTEPnf(INTERNAL)/TABLEt3t2t1(ELEMENTn)
ng / TABLE ...
TABLE
t1 / (mnij, Kmnij), etc.
t2 / (mnij, Mmnij), etc.
t3 / (mnij, Cmnij), etc.
VECTORS ns (LABEL) (PRINT i) (RAYLEIGH)
RETURN
PSEUDO MODAL command

nv number of real eigenvectors taken into account (nv ≤ 640).
nv can be defined by a frequency or a circular frequency. By default,
nv is equal to the eigenmodes number of the MODE file.
DAMPING This label (optional) indicates that the complex eigenfrequencies and
eigenmodes of the structure are computed. Damping is defined either in the
DAMPING file, using the ELEMENT command, or both. The ONLY label
avoids the DAMPING file is automatically recomputed.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
ORTHOGONALITY
The stiffness and mass modal matrices are stored, and used by default.
This label (optional) provides for recomputation from the STIFFNESS,
MASS and MODE files.
NORM This label (optional) indicates the MODE file real eigenmode norm to
be used: MAX, MASS or ELASTIC. The ELASTIC norm is used by
default.
EXTRACT This label (optional) provides for printout of the following:
-stiffness mass and damping modal matrices,
-global matrix of the system to be analyzed,
-eigenmodes obtained in modal space.
These data are printed in addition to circular eigenfrequencies, critical
damping fractions, logarithmic decrements and physical modal
deflections called for by the VECTOR command.
SELECT command (optional)
na TO nb STEP nc
selects the modes effectively taken into account. If omitted, the first nv modes are taken into
account.
ELEMENT command (optional)
This command indicates the elements to which modifications are applied. The modifications
are characterized by stiffness, mass and/or damping matrices which can be symmetrical or not.
nd, ne, nf, ng create a list of the numbers of these elements (previously defined with
DEFINITION).
Notes
l the elements concerned do not necessarily have X9XX codes,
l no reference frame can be defined with the ELEMENT command,
l matrices which can be non-symmetrical must be given in full, and are added to those contained in
the STIFFNESS, MASS and DAMPING files.
l non-symmetrical stiffness, mass and damping matrices are given with the TABLE procedure if
t3t2t1 >0 (cf. 7.8.1.2 a), or in files t3t2t1 <0 (Chapter 11.6.3).
This command defines the stiffness, mass and damping matrices in the way described in the
ELEMENT command.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
VECTOR command (optional)

This command is use number of modal deflections requested (ns ≤ 2nv).
ns is automatically limited to the number of eigenfrequencies stored.d to release the computation
of modal deformed shape.
ns number of modal deflections requested (ns ≤ 2nv). ns is automatically
limited to the number of eigenfrequencies stored.
<LABEL> The optional labels can be DISPLACEMENT or FORCE. Both labels
cannot be used simultaneously. Reactions are computed by default. Use
the DISPLACEMENT label to compute displacements only, avoiding the
need for the STIFFNESS and MASS files. The FORCE label is used for
computation of forces in beam type super-elements.
PRINT i This label controls result printout.
In the absence of this label, no results are printed.
In the presence of this label:
- i>0, results are printed for the first i modes.
- i=0, results are printed for all modes.
RAYLEIGH In the special case where stiffness, mass and damping matrices are
symmetrical, the Rayleigh quotients give an indication of solution
quality. These quotients have no physical reality in other cases. The
STIFFNESS, MASS (DAMPING) files must be present if this label is
used.
Results
Output files:
DATA contains the titles of loads corresponding to the eigenmodes
RESULTS contains MAX norm eigenmodes
MODE contains MAX norm eigenmodes
Notes
l Only eigenfrequencies with a positive or zero imaginary part are retained, together with the
corresponding eigenvectors.
l For each complex eigenmode stored there are two cards in the MODE file (or two loads in the
RESULT file), such that:
l the real part corresponds to an odd number
l the imaginary part corresponds to the next even number.
l In the special case where the structure is not damped, and there are no elements which are non-
symmetrical in terms of stiffness or mass, the modes are real and stored by card.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Stiffness, Mass and Damping Matrices stored in Files

It is not necessary to store a matrix for each element, as only a few elements are generally
concerned, and a number of elements can have the same additional stiffness mass or damping
matrix.
An additional matrix is associated with each ELEMENT, and each element concerned refers to
a super-element.
t3t2t1 < 0 6-digit negative number giving the numbers of logical units t1, t2 and t3
(preferably >60) to which the unformatted files containing the
descriptions of the matrices are assigned by the following commands:
ASSIGN t1 (file name) BINARY
t1 number formed by the units and tens digits of t3t2t1, designating the
number of the logical unit assigned to the file to which the stiffness
matrix is stored.
t2 number formed by the hundreds and thousands digits of t3t2t1, designating
the number of the logical unit assigned to the file to which the stiffness
matrix is stored.
t3 number formed by the tens and hundreds digits of t3t2t1, designating the
number of the logical unit assigned to the file to which the stiffness matrix
is stored.
Structure of a file of the sequential unformatted type, in which k matrices for k super-elements
are stored:
l first entry (free),
l k entries, each comprising 7 integers followed by matrix s in double precision form:
l1, ne, n0, n0, ls, n0, n0, (s(i), i=1,ls)

with
o l1: number of terms in the entry
l1 = 7 + 2*ls
o ne: super-element number
n0=0
o ls: dimension of matrix s
o s: matrix stored by block
l the stiffness matrix associated with an element with nb nodes comprises *nb blocks, each
with kp*kp terms, where kp is the number of dof. Storage is by column, both for blocks and
the terms of each block.
Storage example for a matrix with (2 nodes and 3 dof):
ls=36
1 4 7 19 22 25
2 5 8 20 23 26

© 2019 ESI Group
CHAPTER 11 DYNAMIC
3 6 9 21 24 27
10 13 16 28 31 34
11 14 17 29 32 35
12 15 18 30 33 36
An identity matrix (2 nodes, 3 dof) can be written to the IDENTITY.DAT file in the following
form:
.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Examples
Example 1:
Computation of complex circular eigenfrequencies and eigenmodes for a plane beam. Damping
is symmetrical, and can be contained in the DAMPING file or entered directly using the
PSEUDO MODAL procedure.
DEFINITION
COMPLEX FREQUENCIES OF A DAMPED BEAM
OPTION PLANE BEAM
ONE-DIMENSIONAL
NODES
1 / 1. 0.
21 / 1. 0.5081
ELEMENTS
1 / 1 21 20
MATERIAL PROPERTIES
/ E 2.07*11 NU 0.3 AX 3.87*-3 IZ 8.32*-7 RHO 7483.
CONSTRAINTS
N 1 / UX UY RZ
LOAD
1
DAMPING
1 21 / AX AY IZ 1.
2 TO 20 / AX AY IZ 2.
VERIFY
RETURN
SAVE DATA 2
; direct computation
SEARCH DATA 2
DYNAMIC
MODE SUBSPACE BLOCK DAMPING
VECTORS 4
RETURN

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example 1 (contd.)
; pseudo-modal computation
SEARCH DATA 2
DYNAMIC
MODE SUBSPACE
VECTORS 100 STURM FREQUENCY 20000
RETURN
SAVE MODE 2
SEARCH DATA STIFFNESS MASS DAMPING MODE 2

DYNAMIC
PSEUDO MODAL FREQUENCY 10000 DAMPING
VECTORS 2
RETURN
SAVE DATA MODE 20
DEASSIGN 33 ;necessarytoavoidthesumofthetwotypesofdamping
SEARCH DATA MODE 2

DYNAMIC
PSEUDO MODAL DAMPING ONLY
ELEMENT
1 TO 20 / TABLE 10000
TABLE
1 / 1111 1. 1122 1. 1133 1. 2211 1. 2222 1. 2233 1.
VECTORS 2 DISPLACEMENTS
RETURN

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example 2
Computation of complex frequencies for a structure with non-symmetrical stiffness matrix.
DEFINITION
Complexfrequencies ofa structure withnon-symmetricalstiffness matrix
OPTION PLANE-BEAM
GEOMETRY
NODES
1/
2 / 1.
ELEMENTS
1 / 1 2 TYPE 1902
MATERIAL PROPERTIES
E 1 / TABLE 1
MASS
N 1 2 / AX 1.
TABLE
1 / 1111 2. 2211 2. 1211 1.
CONSTRAINTS
N 1 2 / UY RZ
LOAD
1
VERIFY
RETURN
SAVE DATA 5

DYNAMIC
MODE GIVENS
VECTORS 2
RETURN
SAVE VECTORS 5
SEARCH DATA MODE 5

DYNAMIC
PSEUDO MODAL 2 EXTRACT
ELEMENT
1 / TABLE 2
TABLE
2 / 2111 -2.
VECTORS 2 DISPLACEMENTS
RETURN
SAVE DATA DISPLACEMENTS 50

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.7 Utility Modules
11.7 UTILITY MODULES

This paragraph describes four utility modules, designed for data preparation or specific analysis
tasks.
11.7.1 Equivalent Modal Damping

General
The dynamic response modules (cf. Chapter 11.4 ) require definition of modal damping
associated with each eigenmode taken into account. Where the structure comprises a number of
components for which the damping is different, reference 1 proposes a method for computing
equivalent modal damping for the complete structure. This method is particularly suitable for
different materials (concrete, steel, etc.) with relatively moderate damping. The method is less
suitable for local damping phenomena. The results obtained can be compared, where
appropriate, with modal damping deduced from complex eigenfrequency computation (cf.
Chapters 11.2.3 or 11.6.3).
The principle involves computing a weighted strain energy mean (E = 1/2 φ t Kφ), or kinetic
energies (T=1/2 φtMφ) for the different components.
Damping of the components can be introduced by one of two methods:
l Method 1: This method concerns viscous physical damping. Each of the NC components is
characterized by its critical damping ratio ξi (ratio of damping to critical damping). Two
equivalent dampings, associated with E or T, are then computed for each mode k:
o If some dampings are defined in element material properties by ALPHA, BETA, EPSX
labels, the resultant modal damping ( is added to the damping

defined in the procedure described above.
l Method 2: This method concerns hysteretic physical damping (complex stiffness matrix j η
K), previously transformed to viscous damping form (j C ω). The critical damping ratio ξ
corresponding to C is assumed to vary linearly with circular frequency.
o
o ωk : mode k circular frequency

o ωR : reference circular frequency for each component
A single equivalent damping, associated with strain energy E, is then computed:

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Input Data
Output files:
STIFFNESS, used for computation of eigenmodes
MASS
DATA, MODES obtained from computation of eigenmodes
Input data:
DYNAMIC
DAMPING nv (FILE nf (LOWER inf)(UPPER sup))
(SELECT i1 TO i2 STEP i3)
ELEMENTS j1 TO j2 STEP j3 (INTERNAL) / XSIE xsie
ELEMENTS k1 TO k2 STEP k3 (INTERNAL) / XSIE xsie XSIH xsih--
XSIV xsiv CIRCULAR p
ELEMENTS ...
...
RETURN
DAMPING Command nv
The critical damping ratios corresponding to the first nv eigenmodes are computed (nv ≤ 640).
If nv is absent, it is taken equal to the number of eigenmodes of the MODE file.
nv can although be defined by FREQUENCY fo (or CIRCULAR FREQUENCY ωo)
In the presence of FILE label, weighted dampings are saved in DAMPnf.DAT file (XSIE and
XSIT quantities). LOWER-UPPER optional labels can impose lower and upper values of
weighted equivalent dampings.
SELECT i1 TO i2 STEP i3 command optional
Mode selection. Computation is performed only for the modes selected from the first nv modes.
The eigenmodes of rigid bodies (null frequency) must in priori be suppressed.
ELEMENTS i1 TO i2 STEP i3 / XSIE xsie XSIH xsih XSIV xsiv CIRCULAR p
command
This command is used to define a component and its damping parameters. A component can be
defined with a number of ELEMENT commands, if necessary:
XSIE xsie critical damping ratio for component i (method 1)
XSIH xsih coefficient ξH - equation 3 (method 2)

© 2019 ESI Group
CHAPTER 11 DYNAMIC
XSIV xsiv coefficient ξV - equation 3 (method 2)

CIRCULAR p coefficient ωR - equation 3 (method 2)
Notes:
l Damping data are substitutive for the same element.
l The stiffness of elastic constraints is not taken into account for computation of strain energies (use
equivalent stiffness elements). Local masses are not taken into account for computation of kinetic
energies (use equivalent mass elements).
Results
All results appear in the listing. The critical damping ratio value is obtained for each
eigenmode.
Damping values defined by method 1 give XSIE1 (eq.1) and XSIT (eq.2). The user must
select weighting of strain energies or kinetic energies. Damping values defined by method 2
(XSIH, XSIV) give XSIVH (eq.4). XSIE column contains the summation of dampings
weighted by strain energies (XSIE1+XSIVH).
Example:
SEARCH DATA STIFFNESS MASS MODES 2

DYNAMIC
DAMPING8
SELECT 1 2 6
ELEMENT 1 TO 10/ XSIE=0.001; damping 0.1%
ELEMENT 11 TO 20 / XSIE=0.005; damping 0.5%
ELEMENT21 /XSIE=0.005XSIH=0.003XSIV=0.002CIRCULAR=1000
RETURN
Reference:
1 - B.V. WINKEL and L.J. JULIK
Experience with non-uniform damping in the seismic analysis of nuclear plant components.
ASME Pressure Vessel and Piping Conference - June 19-23, 1983

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.7.2 Generation of Spectral Response

General
This module is used to determine response spectra for an oscillator with one degree of freedom,
subjected to a driving acceleration transmitted by its base.
Pseudo-acceleration spectra constitute normal input data for the SPECTRAL ANALYSIS
module (see Chapter 11.4.6).
The dynamic equilibrium equation for the oscillator is written as follows:
relative displacement velocity and acceleration with respect to the base.
γ critical damping ratio
ω non-damped system circular eigenfrequency

driving acceleration of the base.
The spectra of oscillator (ω, γ) are noted as follows:
l displacement
l velocity
l acceleration
Integration of equation (1) is obtained using the DUHAMEL coefficient or CLOUGH-
PENZIEN methods. The time step selected does not introduce a numerical error insofar as it
represents the accelerogram correctly. The second member varies linearly with the selected
time step.
The accelerogram is defined by a TABLE or by a TRANSIENT file created by the preliminary
computation, and is taken into account over [ti, tf] with a zero value outside this range.
Maximum response is studied over the range [ti, +∞ [.
Input Data
Accelerogramm defining by a table
Input file:
DATA A data file must exist and can contain the TABLES defining the
accelerograms (Chapter 4.15).
Input data

© 2019 ESI Group
CHAPTER 11 DYNAMIC
GENERATION command
the PRINT -1 label cancels printout.
ACCELEROGRAM command
i1 / TABLE nt TIME (INITIAL ti) FINAL tf STEP step (FILE nf)
i1 accelerogram number
nt1 number of TABLE containing the load 1.
ti initial time (0 by default).
tf final time.
step time step defined over [ti,tf]
The time step must be small enough for linear description of the
accelerogram by segments (mandatory parameter).
nf results are stored in SPECnf.DAT (20 by default).
i2 /.. other accelerogram (nf different)
DAMPING γ1, γ2,... γn command
Critical damping ratio values are introduced in sequence, with n < 10.
METHOD command
Selection of the integration method (DUHAMEL by default).
FREQUENCY INITIAL f0 command
This command defines frequency or circular frequency scan ranges
f0 : scan start frequency (f0 > 0)
One or more fp STEP dfp commands are used to define computation steps. The fp STEP dfp
card breaks down the interval into (fp - (fp - 1))/dfp intervals.
The DISPLA np label forces printout every np subintervals. By default, printout occurs at the
end of each interval. Card storage occurs on each subinterval.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Accelerogramm defining by a TRANSIENT file

Input file:
TRANSIENT DATA
All spectra corresponding to the degrees of freedom of nodes translation (number of nodes ≤
13; number of spectra ≤ 39) are computed.
Input data
ACCELEROGRAM TRANSIENT command

ntabl intermediate table number which will contain the accelerations (by default
99).
i1 the only one command line defining the parameters of computation for all
the spectra.
nt: by default 1.
ti initial time (by default, time of the 1st card of the TRANSIENT file).
tf final time (by default, time of the last card of the TRANSIENT file).
step: mandatory parameter.
nf: number of the 1st spectra (61 by default).
Results
Output files
SPECnf.DAT (one file per accelerogram)
This sequential binary file contains the following for each critical damping ratio value,
according to frequency:
l pseudo-acceleration spectrum,
l pseudo-velocity spectrum,
l acceleration spectrum,
l times corresponding to maximum displacement (or pseudo-acceleration) response.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Results can be plotted using the CURVE FUNCTION SPECTRUM routine (see 14.8). The
abscissas of the curves are frequencies.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
SPECnf.DAT File
The SPECnf.DAT file is a function tabulated on a sequential binary type file (see 4.15.4.1);
It can be directly used by SPECTRAL ANALYSIS (cf 11.4.6 SPECTRUM FILE command).
Only pseudo-accelerations are used. Critical damping fraction values γ 1 , γ 2 , γ n must at this
moment form a rising monotonous series.
It is organized as follows:
l 1 record containing NGAMMA + 3 words:
NFREQ, L, NGAMMA, GAMMA (1:NGAMMA)

NFREQ : number of frequencies studied
L: length of current record (8*NGAMMA)
NGAMMA : number of critical damping ratio values
GAMMA : table containing critical damping ratio values.
l NFREQ records containing L values
For each critical damping ratio value, four cards are stored ( SD, SV, SA and Tmax ), with
each card containing a pair of values:
Card1 to NGAMMA: frequency and associated pseudo-acceleration spectrum
Card NGAMMA+1 to 2 NGAMMA : frequency and velocity spectrum
Card 2 NGAMMA + 1 to 3 NGAMMA : frequency and acceleration spectrum
Card 3 NGAMMA +1 to 4 NGAMMA : frequency and time corresponding to maximum
displacement value.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
The SPECnf.DAT file can be reread or created by a program of the following type:

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example 1: Accelerogram on TABLE

Creation of a response spectrum
;
; RESPONSE SPECTRUM AND GRAPHIC VERIFICATION
;
DEFINITION
HORIZONTAL AND VERTICAL RESPONSE SPECTRUM
OPTION TWO-DIMENSIONNAL
GEOMETRY
NODE
1/
TABLE
; TEMPORAL ABSOLUTE ACCELERATION
1 / 6 (2. 11) (5. 12) (3. 13) ; HORIZONTAL ACCELERATION
2 / 6 (1. 11) (2.2 12) (1.8 13); VERTICAL ACCELERATION
11/ 3 31.416 -1.5708 0. 20. ; SINUS - FREQUENCY 5 HERTZ
12 / 3 94.248 -1.5708 0. 20. ; SINUS - FREQUENCY 15 HERTZ
13 / 3 188.496 -1.5708 0. 20. ; SINUS - FREQUENCY 30 HERTZ
; GRAPHIC VERIFICATION OF RESPONSE SPECTRUM
31 / FILE 81 CARD 1 ; HORIZ.- GAMMA=0.02
41 / FILE 82 CARD 1 ; VERTI.- GAMMA=0.02
VERIFY
RETURN
SAVE DATA 1
; RESPONSE SPECTRUM CALCULATION
DYNAMIC
GENERATION SPECTRUM
ACCELEROGRAM
1 / TABLE 1 TIME INIT 0. FINAL 20. STEP 0.0005 FILE 81
2 / TABLE 2 TIME INIT 0. FINAL 20. STEP 0.0005 FILE 82
DAMPING 0.02 0.04 0.06
FREQUENCY INITIAL 1.
50. STEP 1.
RETURN
;GRAPHIC PLOTTING OFPSEUDO-ACCELERATION RESPONSESPECTRUM

SEARCH DATA 1

© 2019 ESI Group
CHAPTER 11 DYNAMIC
CURVE FUNCTION GRID SPECTRUM

FUNCTION 313233XMIN 1.XMAX 50.NSTEP500LOGX LOGY MAPPING
FUNCTION 414243XMIN 1.XMAX 50.NSTEP500LOGX LOGY MAPPING
RETURN
Example 2: Accelerogram on TRANSIENT FILE
SEARCH DATA TRANSIENT 1001

DYNAMIC
GENERATION SPECTRA
ACCELEROGRAM TRANSIENT
1 / STEP 5 *-4
DAMPING 0.02 0.04
INITIAL FREQUENCY 1
33 STEP 0.5 / DISPL 5
RETURN
The DATA and TRANSIENT files are taken from MODAL ANALYSIS. They contain 3
nodes in PLANE BEAM option. 6 spectra files will be generated by the computation (2
translation ddl by node). The files will have the number 61 to 66 (furthermore, a file
TABL99.DAT will be created).

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.7.3 Computation of Effective Masses and Effective Inertia

General
This procedure is used to compute:
l Generalized masses,
l Participation factors corresponding to uniform movement of supports,
l Effective masses and inertia for each mode,
l Unsupported mass,
l Effective global masses and inertia.
Participation factors, and effective masses and inertia are computed for all degrees of freedom of
the option.
This procedure is a complement to the PARTICIPATION module, computing effective inertia
with respect to a given center of rotation. Modal response computation is not possible
following EFFECTIVE MASS.
Input Data
Input files:
DATA
MASS
MODES
There is no output file. Results appear in the listing.
DYNAMIC
EFFECTIVE MASS nv (NODE I1) (POINT x y z)
RETURN
EFFECTIVE MASS nv (NODE i1 (INTERNAL)) (POINT x y z) command

The nv first modes are computed
The centre of rotation, which is by default the origin point, can be defined:
l by a node number : NODE i1 (INTERNAL)
l or by its coordinates : POINT x y z.
In the lack of centre of rotation definition, nv is by default, equal to the eigenmodes number of
the MODE file.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Example
SEARCH DATA MASS 2 MODES 2

DYNAMIC
EFFECTIVE MASS
EFFECTIVE MASS 10 NODE 1
RETURN

© 2019 ESI Group
CHAPTER 11 DYNAMIC
11.7.4 Transient Dynamic

General
HARMONIC RESPONSE computation creates a transient file. When the results are obtained
under amplitude/phase form, TRANSIENT DYNAMIC procedure can be used to obtain the
evolution of time-related response by addition of:
l Phase shifted loads
l Contribution of the different computed frequencies.
We obtain displacements (deformed mesh), velocities and accelerations. Forces in elements can
be obtained by SOLVE FORCE TRANSIENT procedure.
Mathematical formulation
. n elementary loads Fi (Ui) (DEFINITION)
. m circular frequencies (HARMONIC RESPONSE)
We obtain the response to a load of form
: phase shifts between n loads.

In many cases, such as plotting deformed mesh, it is useful to reconstitute evolution of
displacement with time.
The TRANSIENT DYNAMIC procedure provides this function.
Input Data
Input files:
DATA a file contains:
-geometry (reduced)
-number of loads of the harmonic response computation
-(tables describing the evolution of loads amplitude and phase with circular
frequency or frequency)
TRANSIENT this file contains complex displacements stored in modulus and phase
form. It is created from:
HARMONIC RESPONSE with (see Chapter 11.5)
HARMONIC RESPONSE MODAL (see Chapter 11.4.4)

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Input data:
TRANSIENT DYNAMIC
TITLE
…
PHASE(list)(DEGREES)(VARIABLE(FREQUENCY))
AMPLITUDE (list) (VARIABLE (FREQUENCY))
TABLE
t1/
:
tn/
CARD (list) (COMPLEX)
TIME INITIAL t 0
t 1 STEP dt 1
-----
or PERIOD p NSTEP n

RETURN
TITLE optional command

This command is used to define a title (40 characters maximum) for frequencies and loads
combination considered.
PHASE (list) list of phase shifts between loads. The list must contain as many values as
there are loads. By default, all phase shifts are zero. If the DEGREES
label is included, phase shifts are expressed in degrees (by default in
radians).
AMPLITUDE list of amplitudes by which the loads are multiplied (one value per load).
(list) By default, all amplitudes are 1.
VARIABLE phase shifts and/or amplitudes vary according to circular frequency or
frequency if the FREQUENCY label is present. Values entered in the
lists are then the numbers of TABLES defining circular frequency or
frequency functions.
If TABLES are not included in the data file, the optional command
TABLE is used to define them.
CARD (list) list of input transient file cards to be analyzed. Only cards
corresponding to modulus values (odd cards) are declared. If the
COMPLEX label is present, phase cards (even cards) are taken into
account.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
Definition of computation times:
TIME INITIAL t0
t1 STEP dt1
----
tn STEP dtn
PERIOD p NSTEP n
p period number or ratio to be reconstituted. The period concerned is always

the period corresponding to the first card in the list specified by the CARD
command.
n number of computation steps (for p periods).
All cards computed are stored in the transient output file.
Results
Output files:
DATA modified file containing one load only,
TRANSIENT file containing displacements, velocities and accelerations for all times
requested.
Using the DATA and TRANSIENT files supplied by TRANSIENT DYNAMIC, the
procedure can be used to compute forces at each time (see Chapter 12.3.5).
SOLVE FORCES TRANSIENT (CARDS i TO j STEP k ) (EXTRACT)
Examples
Embedded-free beam under two harmonic forces.

© 2019 ESI Group
CHAPTER 11 DYNAMIC
l Transfer functions for a reduced geometry comprising nodes 1, 2, 6 and 11 and element 1.
DEFINITION
MULTIPLES AND PHASE-SHIFTED EXCITATIONS
OPTION PLANE BEAM
ONE-DIMENSIONAL
NODES
1/
11 / 2.
ELEMENTS
1 / 1 11 10
MATERIAL PROPERTIES
E 1 TO 10 / E 2.*11 NU 0.3 RHO 7800. AX 0.0005 IZ 2.6*-8
CONSTRAINTS
N 1 / UX UY RZ
LOADS
1 FORCE AT NODE 6 /
N 6 / FY 1000.
2 FORCE AT NODE 11 /
N 11 / FY 1000.
VERIFY
RETURN
SAVE DATA 1
SOLVE
SAVE RESULTS 1

© 2019 ESI Group
CHAPTER 11 DYNAMIC
DYNAMIC
MODES GIVENS
VECTORS 10 FORCES
RETURN
SAVE MODES 1
DYNAMIC
PARTICIPATION 10 FORCE 2
HARMONIC RESPONSE MODAL 10 FORCE 2
DAMPING GAMMA 0.04
FREQUENCY
5. STEP 5. / DISPL 1
30. STEP 25. / DISPL 1
TRANSFER STATIONARY
DISPL 6 11
FORCE 1
RETURN
F6 = 1000(cos 10 πt + cos 60 πt)

TRANSIENT DYNAMIC
PHASE 0. 45. DEGREES
AMPLITUDE 1 2
CARD 1 3 COMPLEX
PERIOD 1 NSTEP 200
RETURN
SOLVE FORCES TRANSIENT

© 2019 ESI Group
CHAPTER 11 DYNAMIC
F6 = 1000 cos 10 πt

TRANSIENT DYNAMIC
TITLE
F6 AND F11 VARIABLES
PHASE 0. 45. DEGREES
AMPLITUDE 1 2 VARIABLE FREQUENCY
TABLE
1 / 1 0. 1. 6. 1. 7. 0. 100. 0.
2 / 1 0. 0. 6. 0. 7. 2. 100 2.
CARD 1 3 COMPLEX
PERIOD 1 NSTEP 200
RETURN
Note
l When TABLES have been defined in HARMONIC RESPONSE MODAL, TABLES associated to
VARIABLE label must be compulsorily defined on the reduced geometry.

© 2019 ESI Group
CHAPTER 12 TRANSIENT
12.1 Overview
12.1 OVERVIEW
12.1.1 Introduction
In this chapter the linear and non-linear analyses of the response of structures to time-dependent
loads is described. The general equations considered are those for heat conduction,
K. θ + C . = Q (t)
where θ and are temperature and the first derivative of temperature with respect to time.
and for the dynamic analysis of structures,
KU+C. +M. = F (t))
where U, and are displacement, velocity and acceleration
Modal method (cf. Chapter 11.4.3) has some limitations concerning damping forms and non-
linearity types which are taken into account.
The direct method is stepwise integration of the physical ordinary differential equations. This
method is used for all viscous damping forms and for all non-linearities (material, geometry,
contact..). The time-related discretization of general equations uses a time-related scheme.

© 2019 ESI Group
12.1 Overview
12.1.2 Time-Related Scheme

The time-related scheme is different whether it is heat conduction or dynamic analysis of
structures.
Heat conduction
The generalized trapezoid algorithm is used
Authorized values for parameter ν vary from 0 to 1. If ν=0, we obtain the Euler algorithm. If
ν=1/2, we obtain the trapezoid algorithm with a velocity linear approximation between t and
t+dt. For ν ≥ 1/2 the time related scheme is unconditionally stable.
Dynamic analysis of structures
The time-related discretization is obtained from Newmark time-related scheme:
Authorized values for parameters guarantee an unconditional stability of

the displacements computation.
Current algorithm is stable. Using higher values of β and introduces a

numerical damping which restricts the fluctuations of velocities and accelerations. For
convenience sake, this numerical damping can be introduced by a unique parameter α from
which β and coefficients are defined.
By default, a numerical damping of is taken into account; the corresponding energy

dissipation is low. Values of α smaller than 0.3 are recommended.

© 2019 ESI Group
12.1 Overview
12.1.3 Special Features

The direct solution method has two special characteristics, which have an effect on the way in
which a problem is modeled.
l A single load case may be analyzed. Only one load case may thus be specified under
DEFINITION. For linear structural behavior however, several load cases may be specified
which will then be combined during the calculation into a single load case (FUNCTION
command).
l The method of solution (by default) is suited to band equations.In this case, it is necessary to
ensure that nodes are numbered in such a way as to minimize the bandwidth. This may be
done by means of the routine RENUMBER.
The METHOD command (see Chapter 21) is used to choose others equation solvers.
Computed results are essentially displacement or temperature cards, velocities, accelerations and
possibly reactions and forces.
These cards are written to the file TRANSIENT if requested.

© 2019 ESI Group
12.1 Overview
12.1.4 Commands
The routines in which the direct method of calculation of dynamic response is implemented are
called by the command
TRANSIENT (parameters)
The parameters indicate the type of analysis to be carried out. Distinction is made between:
l Analysis of linear thermal or elastic behavior,
l Analysis of non-linear dynamic behavior,
l Analysis of non-linear static behavior.
The command TRANSIENT is followed by commands and data relevant to the chosen type of
analysis.
All of this data is described in detail in the rest of this chapter.
The command
RETURN
concludes the set of commands and data read under TRANSIENT, and initiates the
corresponding calculation of response.

© 2019 ESI Group
12.2 Linear Thermal Analysis
12.2 LINEAR THERMAL ANALYSIS

12.2.1 Introduction
The problem of transient linear analysis of heat conduction may be reduced to the solution of
the system of first order differential equations:
K.θ +C. , = Q (t)(1)
where :
θ is the nodal temperature vector.
= δθ / δt is the vector of nodal rates of temperature change.
K is the thermal conductivity matrix, taken to be independent of temperature.
C is the thermal capacity matrix, also independent of temperature.
Q(t) is the equivalent nodal heat source vector, which is a time function.
Equation (1) is solved by a time-stepping procedure in which, given values of θ and , at time t
satisfying equation (1), values of θ and , at time t + ∆t are sought such that the equation is still
satisfied.
For this purpose a recursive algorithm is used, in which at each time step a system of equations
whose matrix depends on ∆t is solved.
Provided the time step ∆t does not change, it is not necessary to reconstruct and triangularize the
matrix at every step ; given the high cost of this operation, considerable reductions in computing
time may be achieved by adopting a constant time step. In addition, in order to optimize the
performance of the routine, the thermal capacity matrix C is diagonalized.
Limitations:
This kind of calculation does not fit well with quadratic elements. If a 2D mesh already exists,
two solutions can be used to avoid the problem:
l transform the quadratic mesh to a linear mesh with an internal algorithm when computing the
matrices (subdivision of elements without creation of nodes) - cf. Chapter 12.2.3,

l use the LINEARIZE command to transform the element type (Chapter 6.3.6),
Do not use RELATIONS in LINEAR THERMAL.

© 2019 ESI Group
12.2.2 Forced Convection

In the presence of fluid flow, heat flow is modified by the phenomenon of convection.
If there is steady-state fluid flow, assumed to be independent from thermal phenomena, it is said
that forced convection exists.
In this case the equations to be solved include an additional term which depends on the flow
velocity field. It is then necessary to model both the structure and the fluid.
The parameter CONVECTION (see 8.6) must be included in the command OPTION
THERMAL and flow velocity at each fluid node must be specified by the command
VELOCITY (Chapter 4.13).
The matrix of equation coefficients is now non-symmetric and solution cost is therefore higher.

© 2019 ESI Group
12.2.3 Use of Routine

The thermal conductivity and capacity matrices K and C are assembled from the STIFFNESS
and MASS files.
It is necessary to define these files, either by the command:
SOLVE STIFFNESS MASS (LINEARIZE)
or by the command :
SEARCH STIFFNESS i MASS j
LINEARIZE this label is used to transform quadratic elements (type 1003, 2006,
2008, 3010, 3015 and 3020) into linear elements without creation of
nodes. This transformation is internally performed. The stiffness
(conductivity) matrices calculation is carried out by an internal
assembly of subdivision of quadratic elements into linear elements.
Forces and stresses are computed on the original elements in order to
perform standard post-processing.
The routine for transient analysis is called by the command
TRANSIENT THERMAL (SCALING)
or
TRANSIENT LINEAR (SCALING)
The label SCALING specifies the use of an alternative method of diagonalization of the thermal
capacity matrix C. If it is omitted, the leading diagonal term is taken to be the sum of all the
terms of a line ; this procedure sometimes results in negative leading diagonal terms. If the label
SCALING is present, the diagonal matrix is obtained by multiplying the initial leading diagonal
by a scale factor, chosen so that there is no change in global thermal capacity.
Note
l If the STIFFNESS and MASS files does not exist during the input in the TRANSIENT module, they
are automatically computed. In the presence of quadratic elements (type 1003, 2006, 2008, 3010,
3015, 3020), the LINEARIZE label is automatically taken into account.

© 2019 ESI Group
Other commands, possibly followed by data, provide details of the analysis.
ALGORITHME IMPLICIT ν
½ ≤ ν ≤ 1 : parameter of the time-related scheme (cf. 12.1.2)

If ALGORITHM command is not present, ν=0.5 is used.
In transient analysis only one load case may be analyzed, but several load cases may be defined
in the input data of the problem. The data is therefore modified at the end of the analysis, to
ensure compatibility with the computed results for a single load case.
The specification of several load cases allows more than one variation of load with time to be
considered. All loads with the same time-related variation are grouped together as one load
case.
The superimposition of load cases is specified by the command
FUNCTION i1 i2 .... ij
where :
j is the number of load cases specified in the input data,
ik is the number of a table, defined by the routine TABLE, in which is
given the function of time fik by which the load cases k is to be
multiplied.
If Qk is the load defined by load case k of the input data, the total load Q(t) at any time is given
by :
If in the command FUNCTION the value of ik is given as zero, then fik (t) = 1.
The absence of the command FUNCTION is equivalent to the unity function.
The functions fik(t) must be defined by the routine TABLE, called in the usual way by a
command followed by groups of data :
TABLE
i / type (parameters)
i / FORTRAN
i / PROG
i / FILE n CARD j
For a description of the routine TABLE, refer to Chapter 4.15.

© 2019 ESI Group
The command
METHOD ...
is used to chose the solver for the linear equations system. The chapter 21 gives a detailed
description of this command.
The recursive algorithm requires the definition of initial temperatures.
By default, these temperatures are taken to be zero at all nodes.
Otherwise, initial temperatures may be specified by means of a routine called by the command
INITIAL CONDITIONS (RESTART (CARD n or LAST))
RESTART Indicates that initial temperatures are to be taken from the file
TEMPERATURE (or DISPLACEMENT), which must previously
have been defined.
CARD It is also possible to take initial temperatures from a card of a previously
created TRANSIENT file. In this case, the label RESTART must be
followed by CARD n, n being the number of the card non equal to 0 to
be selected. The results are then stored on the same TRANSIENT file,
after the n- card.
LAST Indicates that the last stored card is taken without specifying its number.
In this case, the label LAST must follow the label CARD in place of
the number n.
This procedure enables an analysis to be subdivided into several stages, with each stage
continuing from the point reached by the last. It also allows the possibility of repeating a stage
of the calculation, with a shorter time step for example.
Finally, initial temperatures having been set as described above (and in particular having been
set to zero if the label RESTART is omitted), it is possible to selectively reset nodal
temperatures to any specified values by following the command INITIAL CONDITIONS by
groups of data of the form
(nodes) / TT=θ
where :
(nodes) is a list of nodes
θ is the initial temperature of those nodes.

© 2019 ESI Group
Examples:
INITIAL CONDITION RESTART CARD 8

10 20 / TT = 150
The initial temperatures are read from card 8 except for nodes 10 and 20 where prescribed
temperature is 150 degrees.
INITIAL CONDITIONS
TT = 50
This sets the initial temperatures of all the nodes of the structure to 50 degrees.
The recursive solution algorithm also requires a definition of the time interval. This is done by
means of the command
TIME (INITIAL to) (COMPACT) (RESTART) (SKIP)
where
INITIAL to to is the initial value of time. By default, to = 0.
COMPACT This label activates the compact numbering of the transient file. This
label can be used when some of the cards are not stored.
RESTART The initial time is taken from the specified card of the transient file, in
the case of a restart calculation.
SKIP This label avoids re-defining time steps in the case of a restart
calculation.
Time steps are specified by groups of data of the following form :
ti STEP ∆ti / TEMPERATURE s STORE n
where :
l ti is the upper limit of the time interval [t(i-1), ti]
l ∆ti is the time step to be taken in the interval [(ti-1), ti]
If necessary the routine adjusts ∆ti to be an exact sub-multiple of [ti - t(i-1)].

By default, ∆ti = ti - t(i-1).
If the duration of a time interval is specified to be zero (i.e ti = t(i-1)) then a steady state analysis
is carried out.
Printout and storage of results are performed according to information given in the input data
field.

© 2019 ESI Group
(E) Selective printout labels selected from:

TEMPERATURE or DISPLACEMENT, VELOCITY
(e) Printout edition periodicity: printout of results all "e" cards.
In the absence of value or if e = 0, results are only printed at end of interval
(time ti).
No printout occurs in the absence of any (E) labels.
STORE n: Storage of results all "n" cards.
In the absence of value or if n = 0, all card results are stored in the
TRANSIENT file.
By default of the label STORE, no results are stored.
(S) Selective storage labels selected from:
TEMPERATURE or DISPLACEMENT, VELOCITY
(s) Storage of results all "s" cards .
In the absence of value or if s = 0, results are only printed at end of interval
(time ti).
No selective storage occurs in the absence of any (S) labels.
Note
l The last card where the convergence is obtained is always saved.
Example:
TIME
4 STEP 2 / TEMPERATURE 1
10 STEP 3 / TEMPERATURE
15 /TEMPERATURE
Temperatures are computed for times t = 2, 4, 7, 10, 15.

Temperature cards are printed for times t = 2, 4, 10 and 15, and are written to file for these
times.
End of input data is indicated by the command
RETURN
This initiates the transient analysis, resulting in the calculation, printing and storage on file of the
requested temperature cards.

© 2019 ESI Group
Key : optional

mandatory
Table 12.1 - Input and output files

© 2019 ESI Group
12.2.4 Example of a Quadratic Mesh Treatment

The following input data represents a thermal shock (0° to 100°) applied on the upper surface of
a cube meshed with quadratic hexaedra.
This input data is executed in two runs: the first one without the LINEARIZE (written in bold
characters) label and the second one with this command.
DEFINITION
THERMAL SHOCK
OPTION THERMAL SPATIAL
PARALLELEPIPED
X Y Z 0.1 I J K 5
ADD 1
MATERIAL PROPERTIES
E / KX = KY = KZ = 0.75 RHO = 1602 C = 840
CONSTRAINTS
PLANE ZZ 0 / TT
PLANE ZZ 0.1 / TT
LOAD
1 TEMPERATURE 1 / NOTHING
PLANE ZZ 0.1 / TT 1
RETURN
SOLVE STIFFNESS MASS LINEARIZE
TRANSIENT LINEAR
FONCTION 1
TABLE
1 / 1 0 0 0.1 100
TIME
1000 STEP 100 / STORE 1 ; Transient analysis
1000 / STORE 1 ; Steady-state calculation
RETURN
SOLVE FORCE TRANSIENT
The results shown in figures 12.1 and 12.2 below represent the temperature distribution at time
step 1000 for respectively transient analysis and steady-state.

© 2019 ESI Group
Figure 12-1 Figure12-2

Figures 12.3 and 12.4 represent the same temperature distribution with an internal
transformation of the quadratic mesh, i.e. with use of the LINEARIZE label.
Figure 12-3 Figure 12-4

Results for steady-states with or without transformation of the mesh are identical (figures 12.2
and 12.4).
For transient analysis the results are non-physical when the mesh is quadratic. On the opposite
the results with an internal mesh transformation are close to those expected.

© 2019 ESI Group
12.3 Linear Elasticity
12.3 LINEAR ELASTICITY

12.3.1 General Equation
The problem of transient linear elastic analysis may be reduced to the solution of the system of
second order differential equations:
K.U+C. +M. = F (t) (1)
Where
U is the nodal displacement vector.
= δU / δt is the nodal velocity vector
= δU2 / δt2 is the nodal acceleration vector
K is the stiffness matrix, taken to be independent of temperature
C is the damping matrix, taken to be independent of temperature
M is the mass matrix, taken to be independent of temperature
F(t) is the equivalent nodal force vector, which is a time function.

© 2019 ESI Group
12.3.2 General
Equation (1) is solved by a time-stepping procedure.
Given the values of U, et at time t satisfying equation (1), values of U, and at
time t + ∆t, are sought such that the equation is still satisfied.
To this end a recursive algorithm is used (see Chapter 12.1.2), in which at each time step a
system of equations whose matrix depends upon ∆t is solved.
Provided the time step ∆t does not change, it is not necessary to reconstruct and triangularize the
matrix at every step ; given the high cost of this operation, considerable reductions in computing
time may be achieved by adopting a constant time step.
In addition, to optimize the performance of the computer routine, the damping and mass
matrices C and M are diagonalized. Particularly in the case of damping matrices with the form
C = αK, the user must verify that the physical properties of the model are not too much
modified by the diagonalization operation; The diagonalisation does not take into account the α
K damping introduced on the material properties of elements or on the DAMPING command of
TRANSIENT ELASTIC.

© 2019 ESI Group
12.3.3 Use of Routine

The stiffness, mass and damping (where appropriate) matrices K, M and C are assembled from
the STIFFNESS, MASS and (DAMPING) files. It is therefore necessary first to define these
files, either by the command
SOLVE STIFFNESS MASS (DAMPING)
or by the command
SEARCH STIFFNESS i MASS j (DAMPING k)
The routine for transient analysis is called by the command
TRANSIENT ELASTIC (SCALING)
or
TRANSIENT LINEAR (SCALING)
The label SCALING specifies the use of an alternative method of diagonalisation of the mass
and damping matrices. If it is omitted, the leading diagonal term is taken to be the sum of all the
terms of a line separately for translations and rotations. Rotational diagonal dampings are set to
zero. If the label SCALING is present, the diagonal matrix is obtained by multiplying the initial
leading diagonal by a scale factor, chosen so that there is no change in global mass or damping.
The label SCALING is required with quadratic elements.
Other commands, possibly followed by data, provide details of the analysis.
In the absence of any indication to the contrary, the structure is taken to be undamped (C = 0).
ALGORITHM (IMPL or ALPHA α )

(and optionally)
(ALGORITHM (BETA β) (GAMMA γ)
The optional ALGORITHM command is used to specify the values of the two parameters (β,
γ) of NEWMARK's time-related scheme (cf. Chapter 12.1.2.). Giving α parameter is a way of
automatically defining the βand γ parameters. If no label is present, a low numerical damping is
set (α = 0,05).
If α is given (with α ≥0, then and γ = 1/2 + α. If α ≤ 0 : no numerical damping
or
If β is given (β≥ 1/4), then γ = 2β if γ is not given

© 2019 ESI Group
If γ is given (γ ≥ 1/2), then β = γ/2 if β is not given

By default of any particular information, the structure is supposed without physical damping
(C=0).
If the structure is damped, then it is necessary to define the type of damping by means of the
command
DAMPING ALPHA = α BETA = β GENERALIZED
The damping matrix C is taken to be of the form :

C = α K + βM + Co
Where α and β define a uniform damping for all the elements of the structure. The label
GENERALIZED is used to add the nodal or elements damping previously defined under
DEFINITION procedure (cf. chapter 4) to uniform damping. Matrix Co is assembled from
DAMPING file obtained by command SOLVE (cf. Chapter 9.2).
In transient analysis only one load case may be analyzed, but several load cases may be defined
in the input data of the problem. The data is therefore modified at the end of the analysis, to
ensure compatibility with the computed results for a single load case.
The specification of several load cases allows more than one variation of load with time to be
considered. All loads with the same time-related variation are grouped together as one load
case.
The superimposition of load cases is specified by the command
FUNCTION i1, i2 ..... ij
where:
j is the number of load cases specified in the input data.
Ik is the number of a table, defined by the routine TABLE, in which is given
the function of time fik by which the load case k is to be multiplied.
If Fk is the load defined by load case k of the input data, the total load F (t) at any time is given
by:
If in the command FUNCTION the value of ik is given as zero, then

fik (t) = 1.
The functions fik(t) must be defined by the routine TABLE, called in the usual way by a
command followed by groups of data :
TABLE
i / type (parameters)

© 2019 ESI Group
i / FORTRAN
i / PROG
i / FILE n CARD j
For a description of the routine TABLE, refer to Chapter 4.15.

The command
METHOD ...
is used to choose the solver for linear equations solver. The Chapter 21 gives a detailed
description of this command.
The recursive algorithm requires the definition of initial displacements and velocities.
By default, these displacements and velocities are taken to be zero at all nodes.
Otherwise, initial displacements and velocities may be specified by means of a routine called by
the command
INITIAL CONDITIONS (RESTART (CARD n or LAST))
RESTART The label RESTART indicates that initial conditions are issued from a
previous analysis. If the CARD label is not present, only the
displacements are to be taken from the local DISPLACEMENTS file.
In this case, they are supposed coming from a static computation; initial
velocities and accelerations are equal to zero.
CARD Initial displacements and velocities are to be taken from a card number
n of a TRANSIENT file previously computed. In this case, the label
RESTART must be followed by CARD n, n being the normal number
of the card to be selected. The results are then stored on the same
TRANSIENT file, after the n-th card.
If one of the PLASTIC, VISCOELASTIC or MODIFY labels is
present after the BEHAVIOR command, FORCES corresponding to
card number n are also retrieved.
If the number of recomputed cards is inferior to the initial number of
cards of the file, the last cards of the initial file are saved. To erase
these last cards, the SELECT TRANSIENT command must be used.
LAST Indicates that the last stored card is taken without specifying its number. In
this case, the label LAST must follow the label CARD in place of the
number n.

© 2019 ESI Group
Finally, initial displacements and velocities having been set as described above (and in
particular having been set to zero if the label RESTART is omitted), it is possible selectively to
reset nodal displacements and velocities by following the command INITIAL CONDITIONS
by groups of data of the form:
(nodes) / (U = u), (V = v)
where :
(nodes) is a list of nodes.
(U = u) are displacement labels chosen according to the type of computation option,
each followed by an initial value.
(V = v) are velocity labels chosen according to the type of computation option, each
followed by an initial value.
Example:
INITIAL CONDITIONS RESTART CARD 24

101 TO 107 / UX = 0.01
201, 203 / UX = 0.01 UY = 0.2 VY = 3
The recursive solution algorithm also requires a definition of the time interval. This is done by
means of the command
TIME (INITIAL to) (COMPACT) (RESTART) (SKIP)
where
INITIAL to to is the initial value of time. By default, to = 0.
COMPACT This label activates the compact numbering of the transient file. This label
can be used when some of the cards are not stored.
RESTART The initial time is taken from the specified card of the transient file, in the
case of a restart calculation.
SKIP This label avoids re-defining time steps in the case of a restart calculation.
All the macro time steps for which the final time (ti) is lower to the initial
restart time step (to) are automatically skipped.
Time steps are specified by groups of data of the following form:
ti STEP ∆ti / (E = e) STORE n (S = s)
Where:
ti is the upper limit of the time interval ((ti-1), ti)
∆ti is the time step to be taken in the interval ((ti-1), ti).If necessary the routine
adjusts ∆ti to be an exact sub multiple of (ti - t(i-1)).

© 2019 ESI Group
By default: ∆ti = ti - t(i-1).

If the duration of a time interval is specified to be zero (i.e ti = t(i-1)), then a static analysis is
carried out (except RELATIONS).
Computed displacements, velocities and accelerations (and forces in beam elements) are printed
and written to file selectively according to the contents of the instruction field of each group of
data.
field.
DISPLACEMENT, VELOCITY, ACCELERATION, REACTION,
(FORCE)
In the absence of value or if e = 0, results are only printed at end of interval
(time ti).
STORE n Storage of results all "n" cards.
In the absence of value or if n = 0, results are only stored at the end of
interval in the TRANSIENT file.
(FORCE)
(s): Storage of results all "s" cards .
In the absence of value or if s = 0, results are only stored at end of interval
(time ti).
FORCES special case: if the structure is made of beams only, the label FORCE is used to
obtain the forces by taking into account the inertia forces ( : diagonal mass) damping forces
( ). Forces are expressed in the geometrical local reference frame of the beam. When
eccentricity is specified or a principal inertia frame is defined by a third point (BEAMS option),
forces cannot be transformed into stresses by the usual procedure (STRESS 13.10). The
command SOLVE TRANSIENT FORCE (Chapter 12.3.5) can be used in every case.
Notes
l The last card where the convergence is obtained is always stored.
Example:
TIME
2 STEP 0.1 / DISPLA 4 STORE 1 REACT 2
10 STEP 0.5 / DISPLA 2 VELO REAC FORCE 3
End of input data is indicated by the command

© 2019 ESI Group
RETURN
This initiates the transient analysis, resulting in the calculation, printing and storage on file of the
requested results.

© 2019 ESI Group
12.3.4 Relations
This option, available under the command TRANSIENT ELASTIC is used for modeling
contacts either internal or external to the structure.
It is not allowed for static analysis.
During the transient response:
l Initially unconnected nodes may make contact, or vice versa;
l Initially unconstrained nodes may make contact with a support, or vice versa.
The linkage between two nodes of the structure, or with a support is represented by a linear
spring of finite stiffness normal to the surfaces which make contact. This structural component
comes into play when a limit of travel is exceeded, and ceases to be active if subsequently the
relative displacement falls below that limit.
Interpolation within a time step is automatically carried out, in order to determine exactly the
instant at which contact is made or broken.
Consequently, each change of state of a contact introduces shorter time steps, chosen so that the
change coincides with the end of a step.
Each change produces new cards of results. As a safeguard, if the number of time steps
considered exceeds twice the number requested, the analysis is terminated.
Relations must previously be defined under the command DEFINITION, by means of the
command RELATION (cf. 4.10).
This command is followed by groups of data as shown below:
RELATION
ni nj / (U) MIN = α MAX = β ABS = γ
ni / (U) MIN = α MAX = β ABS = γ
where :
ni, nj are the numbers of nodes which may make contact ;
(U) is a displacement label, chosen according to the type of computation option.
The first group of data specifies a contact interval within the structure, for which
α< U (ni) - U (nj) < β
The second group of data specifies a contact between the structure and a support, for which
α < U (ni) < β
The value γ of the stiffness of the spring which comes into play when contact is made must
always be given.

© 2019 ESI Group
This stiffness acts in a direction defined by the local axes of the nodes as specified under the
command CONSTRAINT.
The analysis remains numerically stable, provided the stiffnesses of springs are of the same
order of magnitude as structural stiffnesses associated with the degrees of freedom involved.

© 2019 ESI Group
12.3.5 Calculation of Forces

As has already been noted in Chapter 12.3.3, it is possible to compute forces in beam elements.
SOLVE FORCE TRANSIENT (SEARCH f) (COMPLEX (REAL)) --

(CARD i TO j STEP k) (EXTRACT) ((FILTER) (RESIDUE))
can be used. This command, starting from the TRANSIENT file created in TRANSIENT
analysis, computes FORCES associated with the selected cards.
Files
Input
DATA
TRANSIENT including DISPLACEMENTS, VELOCITIES, ACCELERATIONS and
REACTIONS
Output
DATA
TRANSIENT including DISPLACEMENTS, VELOCITIES, ACCELERATIONS,
REACTIONS and FORCES
Optional labels
SEARCH f By default the input TRANSIENT file is copied on unit 30. The output
TRANSIENT file is created on unit 19 as usual (copy on unit 30 may
consume CPU time for large files). If the file TRANSIENT has previously
been created on a permanent medium by a SAVE command, the copy
operation may be omitted using the SEARCH label associated with the
TRANSIENT file identification number used on SAVE command.
COMPLEX and REAL
These labels are used to associate the dynamic forces (real or complex)
with the harmonic response (real or complex).
CARDS i TO j STEP k
The selected cards are copied to the output TRANSIENT file.(the number
of cards that may be selected is limited to 65 536, they conserve their initial
number).By default all cards are selected.
EXTRACT Control printout of results for each card.
FILTER The accelerations are recomputed by a natural physical formula
(derivation) from velocities. The forces in beam elements are computed
from filtered accelerations. This lead to acceleration and force levels
more physical than those obtained directly with accelerations of
Newmark's method.

© 2019 ESI Group
RESIDUE This label results in the calculation of the quantity

for each d.o.f.
M : Mass matrix
: Newmark's time related schema acceleration.
: Filtered accelerations (recomputed from velocities).
R (homogeneous to a force) is stored in the PRESSURE file
associated with the TRANSIENT file. If R is high with respect to
applied forces, this indicates a limit of the implicit methods for
phenomena of rapid dynamic.
Example (no card selection)
If input TRANSIENT file has not been SAVED
If input TRANSIENTfile has been SAVED with number 100
SOLVE FORCES TRANSIENT SEARCH 100
Note
l After a calculation with the TRANSIENT NON LINEAR command, the size of the FORCE file
being reduced by SELECT TRANSIENT, the command SOLVE FORCES TRANSIENT (RESIDUE)
can be used.

© 2019 ESI Group
Key :
optional
mandatory
Table 12.2 - Input and output files

© 2019 ESI Group
12.4 Non-Linear Dynamic Analysis
12.4 NON-LINEAR DYNAMIC ANALYSIS

For the sake of brevity, non-linear dynamic analysis of mechanical, thermal and electromagnetic
problems is described in this paragraph.
Analysis is similar in all cases. For heat flow for example, it is merely necessary to delete all
reference to mass matrices, initial forces and acceleration, and replace displacements and
velocities by temperatures and time-related temperature derivatives.
12.4.1 Introduction
For non-linear dynamic problems, finite element analysis involves repeated solution of the
following type of equation:
ψ = F - φ - fa - fi = 0 (1)
Where
ψ = residual nodal force vector
F = vector of external forces applied to the structure
φ= vector of internal forces due to displacements of the structure
fa = vector of damping forces due to velocities of the structure
fi = vector of inertial forces due to accelerations of the structure.
All these vectors are functions of the state of the structure at time t, the state of the structure
being defined by:
U : nodal displacement vector
nodal velocity vector.
nodal acceleration vector.

We can also write:
φ = K.U
fa = C.
fi = M.
where:
K = stiffness matrix
C = damping matrix
M = mass matrix.
For linear behavior, matrices K, C and M are constant.
For non-linear analysis, equation (1) must be solved repeatedly using an iterative method.

© 2019 ESI Group
Various options are possible (ALGORITHM command).

With Ψi the residue resulting from the ith iteration, obtained with displacement field Ui, a better
solution can be obtained by writing:
where:
represents the tangent matrix resulting from the ith iteration.

Solution using a direct or iterative solver (METHOD command) of the matrix equation:
gives a new approximation:
and a new residue Ψ i+1 , corresponding to ui+1, and so on until the residue is less than the
precision factor required by the user (PRECISION command):
Matrix , representing the left-hand side of the equations to be solved, is only used
to supply a convergence operator. It is not always necessary to recompute and retriangulate on
each iteration (OPTIMISE label of the ALGORITHM command). Where the BFGS method
(BFGS label of ALGORITHM command) is selected, the left-hand side matrix corresponds to
the cutting matrix between two successive iterations. In certain cases where convergence is
particularly difficult to obtain, in particular in the presence of large strains, it can be necessary to
compute tangent matrix with precision. A perturbation method can be used for
this purpose (PERTURBATION label of ALGORITHM command):
with
where n and m refer respectively to the nth and mth components of vectors ψ and U.
In all cases, the essential requirement is that residual forces decrease on each iteration, and tend
to cancel each other out.
Furthermore, to facilitate convergence, a prediction of the solution based on known quantities at
previous instants is made systematically for dynamic problems, for the first iteration of the
iterative process for solution of equation (1). For static problems, it is also possible to activate a
solution prediction algorithm (PREDICTION label of ALGORITHM command). This
optional algorithm should only be used for quasi-proportional and monotonic load cases, and
involves estimating the solution at the current time step, on the basis of solutions for the two
preceding time steps:

© 2019 ESI Group
In contrast to linear computation, mass matrix M and damping matrix C are processed integrally
in this case (consistent, non-diagonal matrices).

© 2019 ESI Group
12.4.2 Detail of Time-Related Scheme

Heat Conduction
The value of the time-related scheme parameter ν is given by means of IMPL label after
ALGORITM command (see Chapter 12.1.2: ALGORITHME (IMPLICIT ν)). By default ,
ν= 0.5 is used.
Initial state
Theoretically only the initial temperature is necessary. To avoid utilization off small time steps
at the beginning of transient analysis, a initial thermal velocity is computed :
: diagonalized heat capacity matrix (Chapter 12.2.3).
Dynamic Analysis of Structures

Initial state
In theory only the initial displacement and initial velocity are necessary to define the
problem. With Newmark algorithms γ≠2β, initial acceleration is necessary. In all case, an initial
acceleration which respects the equilibrium equation at initial state, is computed:
: diagonalized mass matrix (Chapter 12.3.3).

Time step
For harmonic type excitations a time step lower or equal to 1/50th of the period gives a good
precision. Higher time steps lead to a period extension.
Dynamic stiffness matrix
The left handside of the equation 1 (see 12.4.1) which is also called dynamic stiffness matrix
KD, is of type:
For the classical algorithm, we find:

© 2019 ESI Group
12.4.3 Utilization of the Routine

Introduction
By contrast with linear analysis, it is not necessary to define the STIFFNESS, MASS and
DAMPING files prior to computation, as the corresponding matrices can be non-linear, and are
recomputed at each time step.
The results of the analysis are stored in a TRANSIENT file. For non-linear problems, these
files are generally very large, obliging the user to limit storage times. To overcome this
difficulty, it is possible to store minimum results in a HISTORY file. These results can then be
used to reconstruct any card of the TRANSIENT file. This procedure is described in Chapter
12.4.6.
The transient state computation procedure is called by the following command:
TRANSIENT NON-LINEAR (STATIC) (SCALING) (ARCHIVATE) --

(REBUILD) (VERIFY) (JOURNAL) (EXTRACT i)
NON-LINEAR Mandatory label.

STATIC Label indicating non-linear computation without dynamic behavior.
Damping and initial forces are not taken into account. Velocities and
accelerations are not computed (see Chapter 12.5).
By default = dynamic behavior.
SCALING If this label is present, the matrix is diagonalized following the
procedures described in Chapter 12.2 and 12.3.
ARCHIVATE This label is used to store full result history in a HISTORY file (see
12.4.6).
REBUILT This label is used to recreate a TRANSIENT file from the HISTORY
file (see 12.4.6).
VERIFY Requests verification of input data and static and dynamic memory
requirements, without execution of the analysis (see Installation
Manual).
JOURNAL This label creates an alphanumeric journal file which can be
interrogated by the user during execution of a long computation. The
characteristics of this file depend on the computer used.

© 2019 ESI Group
Code specifying whether matrices and load vectors are to be printed for
EXTRACT i
checking purposes.
i = 0 : no printout
i = 1 : printout of the left-hand side matrix of each element KD (cf.
12.4.2.2)
i = 2 : printout of right-hand side matrix for each element
i = 3 : printout of both matrices for each element
i = 4 : printout of force data for each element, the material of which is non-
linear (plasticity, creep, etc.), before and after correction of stresses.
Level 1 command TRANSIENT NON-LINEAR is followed by a set of level 2 commands
defining the following:
l Solution algorithm,
l Structural behavior (non-linearities, generalized plane strain),
l Time functions to be applied to special elements,
l Initial conditions,
l Time step.
TRANSIENT NON-LINEAR ...

ALGORITHM ...
PRECISION ...
METHOD ...
BEHAVIOUR ...
FUNCTION
DAMPING
STEP
INITIAL CONDITIONS ...
TIME INITIAL t 0
TI STEP DTI / . ...
...
SIL sil_processing p1 p2 p3
RETURN
End of computation data definition is indicated by the command:
RETURN
which starts computation of transient conditions, and printout of the required results.

© 2019 ESI Group
Solution Algorithm
The command:
ALGORITHM (OPTIMIZE i1 (i2) (i3)) (RELAXATION c r) (PERTURBATION) --

(BFGS ibf) (REFORM r) (ITERATION iter) (PREDICTION) (LINEAR) (SAVE or SEARCH (n)) --
(RIKS (p)) (MEITERATION(q)) (BERGAN (n)) (GROUP j) (STEP k) --
(IMPL α or ALPHA α) or (BETA β) (GAMMA γ) (PENALTY a)(LSEARCH i) --
(CONTACT(nc))
defines the solution algorithm.

OPTIMIZEi1(i2)(i3) Tangent matrix triangularization and computation frequency.
Triangularization of the matrix every i1 iterations between iterations i2 and
i3.
By default, i1 = i3 = iter + 1, i2 = 0, which means that triangularization
occurs once only at the start of each increment.
By default, i2 = 0.
By default, i3 = iter + 1.
If i1 is negative (with i2 = i3 = 0), triangularization occurs at the start of

each (i1) time step, and at the start of the first time step of the command
TRANSIENT NON-LINEAR in the case of dynamic analysis.
RELAXATION c r When the tangential matrix is computed, a global sub-relaxation method is
applied.
c is the sub-relaxation coefficient (by default, 0.25),
r is the relative precision (forces) over which the method is no longer apply
(by default, 0.1).
PERTURBATION Computation of tangent matrices by the numerical perturbation method
for shell elements (SHELL SPATIAL option), 2-D elements (TWO-
DIMENSIONAL option) or 3-D elements (THREE-DIMENSIONAL
option).
BFGSorGRADibf Application of the BFGS (quasi-Newton) method. It is advised to activate
simultaneously the Line-Search algorithm. If ibf is greater than 3, the
storage is limited to ibf iterations. This is useful for huge models. It is
recommended to enter values greater than 10 to keep a good convergence.
REFORM r This label is associated with the BFGS method to improve the global
convergence. If the force residual increases, the tangent matrix is computed
(reformed) ; once if r=1 (default value) or twice in a row if r=2.
ITERATIONSiteror Specifies the maximum number of iterations for each time step.
ITMAXIMALiter (by default iter = 20).
Iter<0 activates, from iteration iter/2, the detection of the no-
convergence of each time step. In case of no-convergence, the
computation of the time step is stopped. It is recommended to use the
automatic time step in this case.

© 2019 ESI Group
PREDICTION Activates the solution prediction algorithm for STATIC problems.

LINEAR This label indicates that the problem is assumed to be linear. The
program executes one solution operation for each computation time
step.
SAVE (n) Automatic save of the triangulated left-hand side matrix in the FMAT.DAT
file. Storage is total if number n is omitted, or partial (up to line n) if value
n is given.
SEARCH (n) Total or partial search for the previously saved left-hand side matrix.
Search is total if number n is omitted, or partial (as far as line n) if value
n is defined.
RIKS (p) For STATIC mechanical problems, application of the RIKS automatic
incrementation algorithm (see description in Chapter 19). The four variants
of the constant arc incrementation method are available. If we combine the
RIKS method with the BERGAN method, or the CRISFIELD method
(MEITERATION label), the variable arc-length incrementation method,
which updates arc-length at each incrementation, is used.
The value p is the method number of the constant arc-length incrementation
scheme (value between 1 and 4). By default, p = 1.
MEITERATION (q) For STATIC mechanical analysis, this label specifies that the
CRISFIELD method of automatic incrementation is used. At each new
increment, the method determines time step size according to the
number of iterations obtained at the previous increment, and the desired
number of iterations, defined by the value q. By default, q = 10.
BERGAN (r) For static mechanical analysis, this label specifies that the BERGAN
method of automatic incrementation is used. At each new increment, the
method determines time step size according to variation of the BERGAN
parameter between the two previous increments.
BERGAN parameter S is used to characterize global structure stiffening (S
>1) or global structure softening (S <1) under increasing loads.
Value r defines the desired variation of the BERGAN parameter between
two consecutive increments.
By default, r = 0.05.
GROUP j Defines the group of time step data for which an automatic incrementation
method is used (RIKS, BERGAN or CRISFIELD).
j is the time step data number. By default, j = 1.
STEP k Specifies the maximum number of increments inside a time step data group
associated with an automatic incrementation method (RIKS, BERGAN or
CRISFIELD).
k is the maximum number of increments.
i.IMPL; (or ALPHA) α or ( (BETA β) (GAMMA γ))

© 2019 ESI Group
These parameters are used to give NEWMARK time-related scheme

parameters (β, γ). Giving α parameter is way of automatically defining
β and γparameters.
By default of all labels, a low numerical damping is given (α = 0.05).
If α is given (α ≥ 0), then and γ = 1/2 +α (α ≤ 0 :
computation without numerical damping).
If β is given (β≥ 1/4), then γ = 2β if γ is not given.
If γ is given, (γ≥ 1/2), then β = γ/2 if β is not given.
For electromagnetism and thermal options IMPL α is used. By default,
α=0.5
This label refers to penalty factor applied in 2D surface contact medium. If
PENALTY a
this label is not present, penalty factor is given by:
P = K.b
where K is the maximum diagonal term of the left handside matrix,
and where the value of b is dependent of the application.
b = 104 for a rigid target body with static analysis
b = 10 for a deformable target body with static analysis
b = 103 for a rigid target body with dynamic analysis
b = 1 for a deformable target body with dynamic analysis.
If a value is given following the PENALTY label, the previous definition
is:
P = K.b.a a is a "relative" penalty factor.
LSEARCH i This algorithm allows in most cases a quickly convergence of the iterative
process to the solution.
i number of "Line Search" algorithm.
i = 1 Method 1 energy type algorithm (default)
i = 2 Residue type algorithm
i = 3 Variant of method 1
By default, the Method 1 Energy type Algorithm is used.
CONTACTi This label allows the use of specific iterations level for each increment
of load with ’Slide Plane’ contact element (see Chap. 8.8.1)
Level 1: de-activation of material behavior non linearities
Level 2: re-activation of non linearities with fixed contact status
Level 3: re-activation of the possibility of changes in the contacts status
The parameter nc ≠0 allows the control of certain level.
nc = 1: with frictional contact element (model ≥ 4), level 1 iterations is
splitted into two sub-level: a first is done by turning the contacts into
sliding one (model 3), then a second with the initial contact model.
nc = 2: de-activation of level 3 iterations.
nc = 3: initiate actions of nc values 1 and 2.

© 2019 ESI Group
Notes:
l If the BFGS or GRAD label is present, it is sufficient to construct and triangularize the right-hand
side matrix once only, by specifying OPTIMIZE - n (where n is very large).
l However in certain cases, it can be appropriate to recompute (at start of increment) the matrix
during the analysis, if convergence cannot be obtained.
l In certain cases, it may be necessary to reconstruct and triangularize the left-hand side matrix
during the iterative process (total Lagrange formulation in large displacements, for example).
However it may be sufficient to execute this operation for the firs t few iterations only.
Example: ALGORITHM OPTIMISE 1 0 2
Triangularization is done on iteration 0 (start of increment), and then on iterations 1 and 2.
l If the PERTURBATION label is present, certain behavior models and modified updated Lagrange
formulation ( EULER label associated with BEHAVIOR command: see below) produce non-
symmetrical tangent matrices. It is therefore advisable to associate a non-symmetrical solver
(NON-SYMMETRICAL label associated with METHOD command: see below) with the
computation of tangent matrices by the PERTURBATION method.
l The PERTURBATION method leads to cumbersome computation, and should consequently be
reserved for cases where convergence cannot be obtained with conventional methods.
l The solution prediction method (PREDICTION label) is particularly efficient in the case of quasi-
proportional loads, namely where the direction of strain at any point varies little in time, and is
monotonic. However this method should be avoided in other cases.
l The RIKS method can be combined with the BFGS or OPTIMISE label on the one hand, and the
BERGAN or MEITERATION label on the other.
l In the case of the BERGAN or CRISFIELD method, time step ∆t or arc increment ∆s (if one of
these methods is combined with the RIKS method) are updated from the second increment.
Examples:
1/ RIKS method
TRANSIENT STATIC NON-LINEAR

BEHAVIOR EULER
ALGORITHM BFGS RIKS 2 GROUP 2 STEP 30
TIME INITIAL 0
1 STEP 0.5 / STORE 1
...
The second method of constant arc-length incrementation is used between times 1 and 10.
Computation stops if the time exceeds 10, or if the number of increments performed since time
1 exceeds 30.
Note that time step DT = 0.5 in the second group of time step data determines the value of the
reference load, noted P in Chapter 19 (initial amplitude, noted , consequently has a value of
1).

© 2019 ESI Group
2/ BERGAN method
TRANSIENT STATIC NON-LINEAR

BEHAVIOR EULER
ALGORITHM BFGS BERGAN 0.1 GROUP 2 STEP 30
TIME INITIAL 0
Between times 1 and 10, time step DT (which had an initial value of 0.5), is automatically
updated at each increment according to variation of the BERGAN parameter.
For the CRISFIELD method, interaction between the data of the ALGORITHM command and
the time step data follows the same principle.
The command:
PRECISION (ABSOLUTE) (STRESS s) (FORCE f) (DISPLACEMENT d) --

(ENERGY e) (TOLERANCE t) (NORM nor) (LARGE rmax) (UNIT) (MAXF fmax) - -
(LARDIS dmax)
defines the convergence criteria for solution of equilibrium equations.

ψ = F - φ - fa - fi = 0.
ABSOLUTEor Convergence criteria are expressed as absolute values. By default, they
ITMAXIMALiter are expressed as relative values.
STRESS s Required precision for stresses
ψ (σ)≤ s> for absolute precision
or ψ (σ)≤ s for relative precision
where E(θ) is the Young modulus at temperature θ.
In that case, only the stress threshold is changed but the convergence
criteria is still done on absolute value comparison.
FORCE f Required precision for residual forces
< f or < f.
where reference residual force r is the force noted on iteration 0 for an
imposed force load, or iteration 1 for an imposed displacement load.
By default:
f = 10-3 for absolute precision
f = 5.10-3 for relative precision.
UNIT In two-dimensional axisymmetric option, the precision in force is given
for a unit hoop length and not in a radian unit.

© 2019 ESI Group
DISPLACEMENTd Required precision for displacement differences between two successive

iterations:
<d
or
< d
where reference displacement difference δUr is the initial displacement
difference noted on iteration 0.
By default: d = 10-5 for absolute precision
d = 10-4for relative precision.
ENERGY e Required precision for internal energy variations between two
successive iterations, combining the two types of precision defined
above
δW < e or δW < e. δWr
where
By default:
e = 10-6 for absolute precision
e = 10-5 for relative precision.
TOLERANCE t Required precision for relational equations, which must be expressed in
absolute values (even if the ABSOLUTE label is absent).
By default, t = 10-3.
NORM nor nor = 0 (default value) indicates the maximum norm:
nor = 1 represents the quadratic norm:
LARGE rmax Maximum value of residual forces accepted. The computation is

stopped when the residual forces exceed rmax after the fourth iteration.
MAXF fmax This label is used to relax the convergence in force. When the
maximum number of iteration is reached, then the required precision for
the force residual is changed into MAX(f, fmax). Warning, this label is
valid only for mechanical computation and for absolute precision.
LARDIS dmax Maximum value of displacement differences accepted. The
computation is stopped when the displacement increment exceeds
dmax after the third iteration.

© 2019 ESI Group
Notes
l The convergence is dependent on the quantities specified by the user. If no criterion is defined,
the convergence is obtained with respect to the default force, displacement and energy precision
values.
l Convergence is obtained when the first criterion is met.
l The internal energy criterion is only available for non-linear mechanical problems.
l If the convergence is not obtained after the n iterations specified in the ALGORITHM command,
computation is stopped and an appropriate message is printed. However the partial results
obtained are printed and stored, for possible restart of the computation.
l When a convergence error occurs, the last card where the convergence is obtained is
automatically saved.
The command
METHOD (DIRECTor ITERATIVE)(NON-SYMMETRICAL (PRECISION (ABSOLUTE)p)
is used to select the type of solver for linear equation systems. By default, a direct method of
solving symmetrical systems is selected. In some cases (contact media with friction, for
example), a non-symmetrical solver must be activated.
Iterative solution methods are generally suitable for processing systems with a large number of
equations.
Refer to Chapter 21 for a detailed description of the METHOD command and the different
solvers.
The rotational degrees of freedom are not used for convergence criteria for the shell spatial and
shell spatial multi options.
Structural Behavior
Material behavior and geometrical non-linearities
The command
BEHAVIOR...
is used to specify the material and geometrical non-linearities taken into account in a mechanical
analysis. If this command is absent, the materials are assumed to behave in accordance with
Hooke's law (linear or non-linear) with small displacements and small strains.
Behavior models, different geometrical non-linearity formulations and the contents of the
TRANSIENT file, are described in detail in Chapter 17.

© 2019 ESI Group
a/ PLANE BEAM and SHELL AXISYMMETRICAL options

BEHAVIOUR command is not required, the only material non- linearity being non- linear
elasticity.
b/ BEAM SPATIAL and SHELL SPATIAL options
BEHAVIOR (PLASTIC) (EULER) - -(TOTLAGRANGE) (UPDLAGRANGE)
Material non-linearities
PLASTIC Used to calculate a structure in which certain materials have:
- elastoplastic behavior
- elasto-visco-plastic behavior.
The label is used to activate the selected PLASTIC and/or CREEP
model type for each element.
Geometrical non-linearities
EULER Specifies integration of large displacement and large rotation
phenomena using eulerian variables (Cauchy stress tensor). This
formulation is compatible with material nonlinearities.
TOTLAGRANGE Specifies integration of large displacement and large strain phenomena
using the TOTAL LAGRANGE formulation.
UPDLAGRANGE Specifies integration of large displacement and large rotation
phenomena using an updated lagrangian formulation.
c/ TWO-DIMENSIONAL and THREE-DIMENSIONAL options
BEHAVIOR (PLASTIC) (VISCOELASTIC) (HYPERELASTIC) --

(TOTLAGRANGE) (EULER) (JAUMAN) (UPDLAGRANGE)
Material non-linearities
PLASTIC Used to compute a structure in which certain materials have:
- elastoplastic behavior
- elasto-visco-plastic behavior
- or viscoplastic behavior.
The label is used to activate the selected PLASTIC and/or CREEP model
type for each element.
VISCOELASTIC Used to assign a linear viscoelastic behavior to the structure.
HYPERELASTIC Indicates that certain materials have hyperelastic behavior.
Geometrical non-linearities
TOTLAGRANGE Specifies integration of large displacement and large strain phenomena
using the TOTAL LAGRANGE formulation.

© 2019 ESI Group
EULER Specifies integration of large displacement and large strain phenomena

using eulerian variables (Cauchy stress tensor). This formulation is
compatible with material nonlinearities.
JAUMAN This label refers to the formulation of stress variation rate (objective
derivation) for the quasi-EULER formulation. The Green-Naghdi
derivative is taken by default.
UPDLAGRANGE Specifies integration of large displacement and large rotation
phenomena using an updated lagrangian formulation. Compared to the
quasi-EULER formulation, it allows using consistent tangent matrixes,
thus ensuring a better convergence scheme in difficult cases.
WARNING, this formulation is compatible with material
nonlinearities only with three-dimensional quadratic elements.
Note:
l For elastoplastic calculations, it is possible to activate a technique called "sub-incrementation" by
using the label FORCE associated with a number n:
BEHAVIOR PLASTIC… (FORCE n)
This technique consists in dividing the imposed load at Gauss points in the elements, in several
load stages in order to compute the stress correction. The number of sub- increments is
calculated from the ratio: threshold/yield stress. It is limited to the value n entered by the user
(by default, 50).
Generalized plane strain
The command:
STRAIN GENERALISED DISPL ROTX ROTY BOTTOM b FT g
This command is used to impose a global behavior for a generalized plane strain formulation.
This option can be used to analyze a structure with translation geometry, subjected for example
to a heterogeneous temperature field in the cross- section of the structure, using a two-
dimensional cross-section model. The structure is then assumed to be situated between two
infinitely rigid planes, with mutual displacement of the planes in the form of a translation and
two rotations. One, two or three degrees of freedom can thus be added to the problem:
l translation on the z axis (DISPL label)
l rotation round the Ox axis (ROTX label)
l rotation round the Oy axis (ROTY label).
If none of the labels, (DISPL, ROTX or ROTY) are specified, the DISPL label is taken by
default. Furthermore, the ROTX and ROTY labels are only taken into account if the
translational degree of freedom is activated (DISPL label).
The user can also introduce a "bottom effect" in direction z, specifying total force f after the
BOTTOM label. A space function (defined by table g) can be associated with the bottom effect.

© 2019 ESI Group
This function is only available with the TWO-DIMENSIONAL TRANSLATION option,

which must be declared in the DEFINITION procedure. The structure can have small
displacements and small strains, and also large displacements and large strains. In the latter
case, only the modified updated Lagrange formulation can be used (UPDATE label on
BEHAVIOR board).
In addition to usual output, the program prints the resultant translations and rotations round the
Ox and Oy axes.
FUNCTION command
In contrast to linear analysis, only one load case may be specified in the input data for the
problem. The FUNCTION command is therefore generally omitted.
This command is still required however if a type 9 element is used (see 8.8 SPECIAL
ELEMENTS), the stiffness matrix and load vector of which are the result of static compression
using the sub-structuring method (see 7.2.5 EXTRACT BINARY).
FUNCTION i
i is used to assign the time function described in table i to the load of this element.
The various corresponding time functions are assigned directly to the loads concerned by the
DEFINITION procedure.
These time functions, as also the tables required for integration of the various non-linearities,
must be defined by a TABLE routine, called by the usual command, followed by its groups of
data:
TABLE
i / type, (parameters)
i / FORTRAN
i / PROG
i / FILEn CARDj
Refer to Chapter 4.15 (DEFINITION - TABLES) for a description of this procedure. It is

illegal to define part of the tables under DEFINITION, and the other part under
TRANSIENT.
Initial Conditions
Recursive solution algorithms require definition of initial displacements and velocities.
By default, these displacements and velocities are taken to be zero at all nodes.
Otherwise, it is possible to specify initial displacements and velocities, using a routine called by
the command:
INITIAL CONDITIONS (RESTART (CARD n or LAST) (RIKS))

© 2019 ESI Group
RESTART The label RESTART indicates that initial conditions are issued from a
previous analysis. If the CARD label is not present, only the
displacements are to be taken from the local DISPLACEMENTS file.
In this case, they are supposed coming from a static computation; initial
velocities and accelerations are equal to zero.
CARD Initial displacements and velocities are to be taken from a card number
n of a TRANSIENT file previously computed. In this case, the label
RESTART must be followed by CARD n, n being the number of the
card non equal to 0 to be selected. The results are then stored on the
same TRANSIENT file, after the n-th card.
If one of the PLASTIC, VISCOELASTIC or MODIFY labels is
present after the BEHAVIOR command, FORCES corresponding to
card number n are also retrieved.
If the number of recomputed cards is inferior to the initial number of
cards of the file, the last cards of the initial file are saved. To erase
these last cards, the SELECT TRANSIENT command must be used
In this case, the label LAST must follow the label CARD and in place
of the number n.
RIKS During a preliminary calculation performed with the RIKS method (cf.
12.4.3.2), the reference load (cf. 19.1) has been computed and saved in
the RIKS.TIT file. It is advisable to restart the calculation with this file
by adding the label RIKS on the command INITIAL CONDITIONS.
Finally, initial displacements and velocities having been set as described above (and in
particular having been set to zero if the label RESTART is omitted), it is possible selectively to
reset nodal displacements and velocities by following the command INITIAL CONDITIONS
by groups of data of the form
(nodes) / (U = u), (V = v),
where:
(nodes): list of nodes
(U = u) displacement labels selected according to the computation option, followed
by initial value at the nodes.
(V = v) velocity labels selected according to the computation option (elasticity
formulation only), followed by initial value at the nodes.

© 2019 ESI Group
Example:
INITIAL CONDITIONS RESTART CARD 24

101 to 107 / UX = 0.01
201 , 203 / UX = 0.01 UY = -0.2 VY = 3
Incremental release of rigid constraints

Rigid constraints defined under DEFINITION can be released at the restart of a transient
computation.
The label CONSTRAINT REMOVE modifies internally given rigid constraints at nodes into
decreasing concentrated load. Values of the concentrated loads are equal initially in opposition
to reaction vectors coming from the restart card and are multiplied by a linear time function
decreasing from 1 to 0 for the time required.
Ad hoc load vectors (one per node to release) and linear tables (type 1) are automatically
created when using this constraint release feature.
INITIAL CONDITIONS RESTART

CONSTRAINT REMOVE GROUP list (DURATION dt1) (EXTRACT)
CONSTRAINT REMOVE GROUP list (DURATION dt2)
...
CONSTRAINT Activates the incremental release of rigid constraints. A blank character ‘ ‘
REMOVE must be preceded by the word CONSTRAINT.
Up to 10 lines can be defined.
GROUP list Group(s) of nodes to release.
The list can be a list of group names (between $ characters) and/or group
numbers.
Example: GROUP $grp1$ num2 num3 $grp4$ …
DURATION dt DURATION sets the duration of the time-dependent function to which is
multiplied the concentrated loads.
The linear table starts from value 1 at time tnc (time of the restart card) and
reaches the value 0 at time tnc + dt.
By default dt is equal to the difference between ending time of the transient
computation and restart time.
EXTRACT Lists information on the released nodes.
Note:
l The keyword CONSTRAINT REMOVE is available only for mechanical computations.
l Not all rigid constraints should be released to avoid rigid body motion.
l A node listed several times in the lists of nodes to release will be treated (constraint removal and
load assignment) only the first time encountered by the process.

© 2019 ESI Group
Time Step
General
The recursive solution algorithm also requires definition of computation times. This definition
is made using a procedure called by the command:
TIME (INITIAL to) (COMPACT) (RESTART) (SKIP) (DTLAST)
where
INITIAL to is the initial value of time. By default, to = 0.
COMPACT This label activates the compact numbering of the transient file. This
label can be used when some of the cards are not stored.
RESTART The initial time is taken from the specified card of the transient file, in
the case of a restart calculation.
SKIP This label avoids re-defining time steps in the case of a restart calculation.
All the macro time steps for which the final time (ti) is lower to the initial
restart time step (to) are automatically skipped.
DTLAST In case of restart of a computation, this label overwrites the initial time
step with the previous one to reach first macro time.
The previous time step is computed from the difference between the
time of the restart card (defined in INITIAL CONDITION section) and
the one of the card that precedes it.
The overwriting of the time step is available only with automatic time
step, provided that the computed step is greater than the initial one.
Time steps are defined by the following groups of data:
ti STEP ∆ti / (E = e) STORE n (S =s)
where
ti = the upper limit of the time interval [t(i-1), ti]
∆ti = time step taken for computation in the interval [t(i-1), ti]. The value of
ti is adjusted by the routine, to a sub-multiple of [ti - t(i-1)].
By default, ∆ti = ti - t(i-1)
By convention, a time step equal to zero (ti = t(i-1)) leads to the computation of a static steady
state.
field.

© 2019 ESI Group

FORCE, PRESSURE and VALUE
In the absence of value or if e = 0, results are only printed at end of
interval (time ti).
STORE n Storage of results all "n" cards.
In the absence of value or if n = 0, all card results are stored in the
TRANSIENT file.
FORCE, PRESSURE and VALUE
(s) Storage of results all "s" cards.
In the absence of value or if s = 0, results are only printed at end of interval
(time ti).
Obviously, printout and storage of accelerations are only possible for an elasticity formulation
(and not for a thermal formulation).
Note
l The last card where the convergence is obtained is always saved.
Example:
INITIAL CONDITION RESTART CARD LAST

TIME INITIAL RESTART
1. STEP 0.01 / STORE
TIME
2 STEP 0.1 / DISPLACEMENT 4 STORE 1
10 STEP 0.5 / DISPL 2 VELO 2 STORE 2 DISPLA 2
Automatic time step adjustment

Determination of time steps for a dynamic problem, or load increments for a static problem, is
always difficult, and in many cases intrinsically impossible, as both time step and load
increment depend on the loads imposed on the model, and the constitutive laws for the
materials. These considerations force the user to make selections prior to computation, with the
consequent need to restart if convergence is not obtained, or if the degree of precision achieved
is considered insufficient. This leads to considerable analysis time requirements.
Algorithm principle

© 2019 ESI Group
Starting from a time step ∆t which is already known (time step previously predicted or test time
step suggested by the user in the TIME data of the procedure TRANSIENT NON LINEAR),
the program tries to compute a solution at t+ ∆ t. If this solution cannot be obtained (non
convergence of computation), then a new time step is suggested:
∆tp = α∆t
where α is a user-defined parameter (by default, α=0.5). This operation is repeated until the
computation converges. ∆ t min being the minimum time step authorized by the user,
computation stops as soon as ∆tp < ∆tmin.
Having a solution at ( being the time step where convergence has been reached), the
general rules apply to the automatic time step adjustment are the following:
Checking the solution in time interval [t, t+∆t].
The solution obtained at t+∆t assumes predefined evolution of the displacement field between t
and t+∆t :
l linear variation for a static mechanical analysis,
l linear variation for a thermal or electromagnetic analysis (phenomena governed by a first
order, time-related differential equation),

l quadratic variation for a transient mechanical problem,
and there is nothing to indicate that this evolution is compatible with load evolution.
The solution on the interval is checked by computing the residue at taking

into account the appropriate evolution of the displacement field.
If the residue at this time exceeds an acceptable value (user-supplied precision ε), a new time
step is assessed according to ε , and the solution previously obtained at time t+ ∆ t is not
validated.
Prediction of a new time step
When the residue at is acceptable, the program keeps the solution obtained at
and thus tries to take the next time step.
The prediction of the new time step is based on two estimations. The minimum value of these
two estimations is kept. The first estimation is based on the previous time step (between t and
) and is used to estimate the time step which could have been chosen (in place of )
while keeping an acceptable solution at the end of the time step. The second estimation is based
on an attempt to predict the future. To do this, the program computes a residue at

assuming the appropriate evolution of the displacement field. From this residue and the one
requested by the user (ε), a new time step is estimated. The interest of this second estimation is
to make sure that loads do not change abruptly in the time steps which follow .
Notes
l The precision given by the user is a priori a residue in force (force unbalance). For thermal
options, it is generally easier to define a precision in temperature. Thus, an additional possibility is

© 2019 ESI Group
given for thermal, thermo-metallurgical and electromagnetism options to define the precision in
term of "displacement tolerable gap".
l By adding the label FORCE in the input data, this algorithm can be made predictive only. It
suppresses the control of the solution in the interval .
Input data
This algorithm is incompatible with the RIKS, BERGAN and CRISFIELD automatic
incrementation method.
TRANSIENT NON-LINEAIR ...

...
STEP DTMINI dtmin DTMAXI dtmax (LABEL) ε --
FACC kc FACP kp fp NRES nres RESM resm FORCE --
WATCh DEACtivate NOEXtract
...
TIME ...
ti STEP ∆ ti / ...
...
RETURN
with:
dtmin minimum time step authorized by the user (∆t < dtmin --> computation
stopped) - The DTMINI label must be given.
dtmax maximum time step authorized by the user (optional).
LABEL ε
acceptable precision for the control of the solution at and for the
prediction of the solution.
LABEL:
EPSD, if the acceptable precision is expressed in "DISPLACEMENT"
for the thermal and thermo-metallurgical options , ε = 5 degrees by
default.
EPSF, if the acceptable is expressed in "FORCE". By default,
where f is the precision in force, indicated on the command
PRECISION (ABSOLUTE or RELATIVE precision).
fc time step multiplication factor if convergence not obtained.
By default, fc=0.5.
fp maximum authorized time step multiplication factor. By default, fp=2.
nres number of successive time that the residue can increase during a card.
resm limiting value of the maximal residue.
The FORCE label avoids to control the solution at , thus the model becomes predictive
only.

© 2019 ESI Group
The label WATCH allows to deactivate the automatic time stepping algorithm until a
convergence issue occurs. The time step is given by the user.
Note that the label FORCE is automatically activated in that case because it switches off the
control at .
The label DESACTIVATE allows to deactivate the controls at and at . Without

control, the automatic time stepping algorithm increases the time step until ∆ t = dtmax
(DTMAXI label is mandatory). When the convergence does not occur, the algorithm decreases
the time step. Warning, as this option deactivates all controls, the quality of results is not
guaranteed anymore.
The NOEXtract label is used to suppress any printout related to the algorithm execution.
The user indicates the macro time steps after the TIME command. Macro time steps ti are
always respected. It might happen that the time step increase can be higher than fc to avoid
unneeded small time steps in that case. The routine automatically adjusts the time step between
two macro time steps. The first time step selected at the beginning of an analysis, or on restart,
corresponds to the time step indicated at the STEP label.
Example :
l With 0 as an initial time
10 step 5
The time step that is considered is 5.
l With 0 as an initial time
10
The time step that is considered is 10.
Recommendations
l The algorithm searches for problem discontinuities, and reduces the time step for this purpose.
On the other hand, the algorithm extends the time step for regular responses. In the presence of a
number of successive discontinuities (contact problem for example), it is advisable to limit the
time step extension factor (FACP) to a reasonable value (of the order of 1.1), and to avoid
excessive demands concerning half-step precision.
DAMPING Command
DAMPING
This command is required to specify concentrated damping at the nodes (cf. 4.12), and damping
for each element (labels ALPHA and BETA for most elements - cf. 4.5) described in the
physical properties.
SIL Command
SIL processing_sil P1 P2 P3
This command executes the « sil_processing.cmd » process before and after each card with
three parameters:
P1: 0: call before the card, 1: call after the card.

© 2019 ESI Group
P2: time computing.

P3: number of the card.
Example: test.cmd
//
// Example
//
ibeg = p1
time = p2
ncard = p3
if (ibeg = = 0) {
printf(« Execution before the card %d\n », ncard) ;
else {
printf(« Execution after the card %d\n », ncard) ;
}
RUNEND Command
The end of run can be controlled either by a final time (Cf. Time Step Section), or by stop
criteria. These stop criteria are managed by the command RUNEND.
"Card number" stopping criterion (thermal computation)
The "card number" criterion lets the user stop a transient thermal computation after a given
number of cards. The functionality is activated with keywords RUNEND CARD:
TRANSIENT NON-LINEAR
...
RUNEND CARD (nc) (INITALIZE) (EXTRACT)
...
RETURN
RUNEND Activates condition criteria for ending a computation

CARD (nc) The stop criterion is managed by a number of cards.
By default, nc=10.
INITIALIZE Initializes the counter on cards number.
EXTRACT Prints information during computation.
Remark:
l Stop criteria "Interpass temperature" (Cf. SYSWELD Reference Manual, §A.2.2.5) and "Number of
cards" can be used at the same time by writing two distinct lines beginning by keyword RUNEND
in a TRANSIENT NON-LINEAR block. In such usage, conditions are checked one after the other
and the first satisfied criterion stops the computation.

© 2019 ESI Group
"Thermal" stopping criterion (thermo-mechanical computation)

The ‘thermal’ criterion lets the user stop a transient thermo-mechanical computation at the last
computed thermal time (i.e. time of the last thermal card loaded by command
TEMPERATURE, see Reference Analysis Manual, §4.13):
...
RUNEND THERMAL (GOTO n) (EXTRACT)
...
RETURN
RUNEND Activates condition criteria for ending a computation

THERMAL The stop criterion is managed by the time of the last loaded thermal card.
GOTO n Jumping condition when ending time of the transient block is reached.
With a stop criterion, the computation is most of time stopped before
reaching the ending time of the transient block. But in case the
computation reaches this ending time, the jumping condition can be
activated to jump forward to another part of the input deck. This feature
is usually to be used in addition to the level 1 ‘GOTO n REWIND
command.
Note that a command line "POINT n" with same number n must be
present in the remaining input deck file.
EXTRACT Prints information during computation.
Remark:
l Stop criteria "Interpass temperature" (Cf. SYSWELD Reference Manual, §A.2.2.5) and "Number of
cards" can be used at the same time by writing two distinct lines beginning by keyword RUNEND
in a TRANSIENT NON-LINEAR block. In such usage, conditions are checked one after the other
and the first satisfied criterion stops the computation.

© 2019 ESI Group
12.4.4 Combination between Non-Linearities and Algorithms

The BEHAVIOUR command defines geometrical non- linearities (large displacements and
large strains), and non-linearities associated with the behavior of the materials:
l Geometrical non-linearities
l Total Lagrange, updated Lagrange and modified updated Lagrange formulations
l Material non-linearities
l Hyperelasticity, plasticity, viscoplasticity or creep, viscoelasticity and non-linear elasticity.
Other non-linearities can be combined, such as those associated with the different MEDIA
(Chapter 8.11), and contact media in particular, and those resulting from the introduction of
special RELATIONS (Chapter 12.4.5).
The ALGORITHM command defines:
l BFGS,
l NEWTON (OPTIMISE 1 label),
l NEWTON initial (OPTIMISE - n label with n very large)
l Modified NEWTON method (no BFGS or OPTIMISE labels, with the tangent matrix
computed only on iteration 0 of each card)

and automatic incrementation methods for STATIC problems:
l RIKS
l BERGAN
l CRISFIELD
Table 12.3 indicates elements for selection of the most efficient algorithm according to the non-
linearity (or non-linearities) concerned.
The algorithms are assigned a priority index for each non-linearity.
Automatic incrementation methods can be associated with all algorithms defined above (BFGS,
Newton, etc).
Table 12.4 describes possible combinations of non-linearities and incrementation methods.

© 2019 ESI Group
Optimize 1 Optimize –n NEWTON

BFGS
NEWTON NEWTON Initial Modified
Non-linear elasticity 1 2
Plasticity 1 2 2
Viscoplastic creep 1 2
Viscoelasticity 1 2
Total Lagrange 1 2
Updated Lagrange 1 2 2
Modified updated
1 2
Lagrange
Relation incompatibility 2 1
Contact media 1
Table 12.3 - Combinations between non-linearities and algorithms

(1 indicates better than 2)
RIKS BERGAN CRISFIELD

Non-linear elasticity + + +
Plasticity + + +
Creep incompatibility +
Viscoelasticity incompatibility +
Total Lagrange + + +
Modified Lagrange + + +
Relation incompatibility +
Contact media incompatibility +
Table 12.4 - Combinations between non-linearities and

automatic incrementation algorithms

© 2019 ESI Group
12.4.5 Relations
It is possible to integrate additional relations or inequalities between degrees of freedom
previously defined with RELATION commands (see 4.10) under the DEFINITION procedure.
Equations
Degrees of freedom at a number of nodes can be related by an equation with the form:
bij Uj = gi
An unknown Lagrange multiplier, interpreted as a force, is associated with this equation.
This equation must be described using the RELATION command under DEFINITION.
Two types of relation can be defined:
a) General relation
RELATION
(nodes) / TABLE (t)
where (nodes) = list of nodes.

Table t describes a relation of the most general type, between components of displacements at
the nodes:
where n is the node number, and j the number of the degree of freedom. The table comprises
pairs of values:
where k is an integer associated with node n and degree of freedom j by the relation k =
10.no+j, no being the position in which node n appears in the list of nodes in the RELATION
command.
Where appropriate, value g gives the right-hand side of the equation.
Example
RELATION
1 8 5 7 / TABLE 5
TABLE
5 / (12,1) (22,.5) (31, -7) (43,2.3) 10.
These data define the following relation:

© 2019 ESI Group
For a three-dimensional structure:
Note
l If all the nodes indicated are assigned the same system of axes by a CONSTRAINT command (see
4.6), the relation is expressed in the corresponding axes.
b) Displacement equalities
RELATION
(nodes) / (U)
(nodes) list of nodes

U displacement label according to the computation option.
The displacement label indicates the degree of freedom which must be identical for all nodes
described in the list of nodes. The nodes have a system of axes assigned by a CONSTRAINT
command (see 4.7), the degree of freedom is defined with respect to these axes.
Inequalities
a) Contact type
The TRANSIENT NON- LINEAR procedure can integrate contact phenomena, whether
internal or external to the structure.
During the dynamic response:
l Initially separate nodes can come into contact with each other, or vice versa,
l Initially connected nodes can come into contact with a support, or vice versa.
The link (structure - structure or structure - support) is provided by additional equations,

supplying contact forces, which appear when a gap criterion is violated, and which disappears
when this criterion is again met, as unknown quantities.
The solution of relations of this type is obtained by introducing Lagrange multipliers. To
achieve convergence of the iterative process, the structures, taken separately, must be statically
in a state of equilibrium. If a structure is not physically constrained, the user can only drive
contact problems with displacement (and not force) formulations.
In contrast to the TRANSIENT ELASTIC procedure:
l there is no time interpolation for precise entry of the moment of appearance or disappearance
of the contact. Contact status is taken into account at the moment of computation, during the
time progression.

© 2019 ESI Group
Nevertheless the iterative procedure for a given time value ensures contact status stability.
l No particular user action is required.
RELATIONS
ni nj / (U) MIN = α MAX = β (1)
ni / (U) MIN = α MAX = β (2)
where:
ni, nj numbers of nodes which can come into contact.
(U) displacement label according to the computation option.
The first group of data defines a contact internal to the structure:
α <U(ni) - U(nj) < β
The second group of data describes a contact between the structure and the support:
α <U(ni)<β
The inequality is expressed in the local system of axes of the nodes (defined by the
CONSTRAINT procedure).
b) Friction
Decoupled friction
There is no gap to be closed, and contact is assumed to be already established.
A conventional relation of equality between one or two degrees of freedom is introduced.
However the associated force exceeds a specified value, a quality is not longer maintained, the
force remains constant, and sliding occurs.
RELATION
ni n j / (U) ABS = - val
ni / (U) ABS = - val
where ni , nj: numbers of two nodes between which friction occurs

(U) displacement label
-val maximum force value with minus sign.
Coupled friction

© 2019 ESI Group
In this case, the relation is only active if a contact relation acting on the same node or nodes and
another degree of freedom, is itself active.
Friction without sliding occurs as long as the associated force does not exceed a limit value,
dependent on contact force Fn.
RELATION
ni nj / (U) ABS = α
ni / (U) ABS = α
where :
ni nj numbers of two nodes between which friction can occur
(U) displacement label
α coefficient of friction (α > 0)
Sliding occurs when Ft ≥ α . Fn (Fn = normal contact force)
Example
RELATION
45 26 / UY MIN -.3 MAX 3.
45 26 / UX ABS = .5
Nodes 45 and 26 can shift freely inside a gap of +0.3 in direction Y.

(-.3 < Uy 45 - Uy 26 < +3.)
When this gap is closed on one side or the other, frictional force FX, limited to 0.5 . FY (where
FY is the contact force) appears.
Sliding occurs if FX > 0.5 FY (intermittent sliding friction stop).
The algorithm is only available for 2-D friction problems. The following relations cannot
therefore be defined:
RELATION
10 15 / UX MAX 4.
10 15 / UY ABS = .3
10 15 / UZ ABS = .3
c) Axes
Contact and friction relations operate in axes associated with the nodes by the CONSTRAINT
command in the DEFINITION procedure (see 4.7).

© 2019 ESI Group
d) Output results dependent on relation input

** Number of primary and secondary iterations (i) (j)
Equilibrium iteration number (k) residue for relations :
**
where
represents the number of iterations required to determine normal contacts (Lagrange
(i)
multipliers).
represents the number of iterations required to determine the Lagrange multipliers associated with
(j)
tangential displacement during sliding.
(N) number of relations.
(k) equilibrium iteration number for a given time card.
represents sliding displacement of a node between two equilibrium iterations i and i-1. The
∆bi
value (residue) must tend towards zero to obtain equilibrium for an iteration.
Relation status
Relation Stress Limit Code Force Node-component
1 0.000E+0 0.000+E0 1 0.321 52 2 63 2
2 -100. 0.000E+0 0 0.00E+0 52 2 63 2
3 -0.279E-01 0.600 3 0.192 52 1 63 1
Internal node
number where
Relation Imposed gap Relation Force at
ABS value component to
number or computed status nodes
which relation
applied
Interpretation of relation status codes:
0 Inactive
1 Active for coupled contact-friction relation
2 Intermittent friction stop without sliding (FT < Fn)
3 Intermittent sliding contact ( FT = Fn)
-1 Active for decoupled contact-friction or displacement equality relation
-2 Permanent friction stop without sliding
-3 Permanent sliding contact

© 2019 ESI Group
12.4.6 HISTORY and TRANSIENT Files

TRANSIENT files created by the TRANSIENT NON-LINEAR procedure are generally very
large, in particular following an elasto-plastic analysis. The user is then obliged to select storage
times in advance.
Rather than maintaining a TRANSIENT file, SYSTUS enables the user to store minimum
results for subsequent reconstruction of any card of the TRANSIENT file, in a HISTORY file
(mass storage file).
The HISTORY file contains all displacements, in double precision, for all computation steps.
Furthermore, in the presence of "contact" type macro elements (CONTACT, GAP,
INTERFACE and PLANE), the HISTORY file contains the PRESSURE file, for all computed
time steps, in addition to displacements.

© 2019 ESI Group
To store results in the HISTORY file, it is merely necessary to add the ARCHIVATE label to
the TRANSIENT NON-LINEAR command.
TRANSIENT NON-LINEAR ... ARCHIVATE

...
RETURN
The HISTORY file can be manipulated with the usual SAVE, DELETE, SEARCH and
ASSIGN commands (unit 25).
SAVE HISTORY i
DELETE HISTORY i
SEARCH HISTORY i
ASSIGN 25 HISTi.TIT BINARY
The computation cards stored in the HISTORY file can be listed by the following command
sequence:
ASSIGN 19 HISTi.TIT BINARY

CARD
The user can then request storage of any precomputed card by searching the HISTORY file,
and specifying the REBUILD label on the TRANSIENT NON-LINEAR command. The
routine then reintegrates the material behavior law from the displacement history. This
recreation phase delivers a conventional SYSTUS TRANSIENT file.
The syntax for the recreation procedure is as follows:

© 2019 ESI Group
TRANSIENT NON-LINEAR ... REBUILD

BEHAVIOUR ...
FUNCTION
CARDS <list>
RETURN
The BEHAVIOUR and FUNCTION commands must contain the same information as for the
initial analysis.
The INITIAL CONDITIONS command defines either initial conditions (defined in initial
analysis) for integration of the behavior laws, or the card from which integration of the behavior
laws is made (recreation procedure restart). In this case, the TRANSIENT file containing the
card mentioned in the command is required at input. By default, recreation starts at card 0.
The CARD label defines the list of computation cards for which results must be stored.
This function is available for all computation options recognized by the TRANSIENT NON-
LINEAR procedure (Except for generalized plane strain option).

© 2019 ESI Group
12.5 Non-Linear Static Analysis
12.5 NON-LINEAR STATIC ANALYSIS

The TRANSIENT NON-LINEAR procedure, designed primarily for the analysis of dynamic
non-linear behavior, can also be used to analyze the static non-linear behavior of structures.
Formulation is the same. Certain terms (mass, damping, velocity and acceleration) are omitted.
This paragraph is restricted to a review of Chapter 12 indicating differences of formulation,
algorithms and utilization.
12.5.1 General Equation

The problem involves minimization of residual nodal forces:
ψ = F - φ(1)
Notations are the same as for Section 12.4.1.

© 2019 ESI Group
12.5.2 Implicit Algorithm

Equation (1) is solved in the same way as described in Chapter 12.4.2, using a double iteration
process:
l Time-stepping
l Iteration for each time value, in which residual forces are minimized.
Time steps are used only to define the quasi-static application of loads to the structure. Velocity
and acceleration are not involved.
The procedure for minimization of residual forces is the same as described in Chapter 12.4.2.
Where the quasi-Newton is selected, convergence is substantially accelerated by modification of
the iterative procedure.
The essential factor is that residual forces decrease on each iteration, tending to zero.
12.5.3 Utilization
As mentioned above (Chapter 12.4.3), the STATIC label must appear in the TRANSIENT
command.
All other parameters and commands have the same meaning as before.
The recursive solution algorithm only requires definition of initial displacements in this case.
For mechanical problems, the RIKS, Bergan or Crisfield automatic incrementation methods can
be used. Furthermore, the RIKS method can also be used for structural instability problems.
Only the labels (U = u) appear in the groups of data following the INITIAL CONDITIONS
command.
In the same way, the only results printed and stored are DISPLACEMENTS, REACTIONS
and FORCES (no velocities or accelerations).

© 2019 ESI Group
12.5.4 Substructuring
Introduction
The substructuring method has numerous advantages for dealing with large- scale linear
problems, including the sub-division of computation, possibility of elementary modifications,
optimum integration of structure symmetries, etc.
In a large number of non-linear mechanical problems, the structure has non-linear behavior
(plasticity, large strains, etc.) in a restricted area, which is generally clearly demarcated. It is
then of interest to isolate this area, and compute, once and for all, the elastic matrices and right-
hand side of equations for the other areas, which have a linear behavior (elasticity, small strains
and small displacements).
To take another example in the field of electromagnetism, there is a special medium with linear
behavior. It can be relevant to consider air as a special sub-structure, in order to concentrate
non-linear analysis on the fellow-magnetic areas.
Other examples can also be given, but in general terms, the use of sub-structuring in a non-
linear analysis leads both to a reduction of computation time, and an increase in the size of the
problems which can be handled.
It is obvious that the advantage which can be obtained from this method, depends directly on
the type of problem and the quality of sub-division into sub-structures (super elements).
Creation of the Tree Structure

The tree structure for super elements is created in the standard form described in Chapter 10.
A single sub-structure can have a non-linear behavior.
This non-linear sub-structure must belong to level 0, and be connected directly to a higher level
sub-structure.
An example is given in the following diagram:
All conventional operations (symmetries, etc.) described in Chapter 10 are applicable to

standard (linear) sub-structures.

© 2019 ESI Group
On the other hand, all areas showing non-linear behavior must be grouped in a single, level 0
sub-structure.
Input Data for a Non-Linear Sub-Structure

Input data for a non-linear sub-structure (level 0) are entered using a standard DEFINITION
procedure:
DEFINITION
---
---
RETURN
followed by a SUPER ELEMENT procedure:
SUPER ELEMENT nl
where nl is the number of the non-linear sub-structure (super element).

The following procedure can also be used:
DEFINITION SUPER ELEMENT nl

---
---
RETURN
Computation Procedure
The conventional TRANSIENT NON-LINEAR procedure is used. Only the SUPER
ELEMENT nl label must be added.
TRANSIENT NON-LINEAR SUPER nl

or ELEMENT nl
---
FUNCTION i 1 , i 2 ..., i j
---
RENUMBER (ITERATION n)
<renumbering front>
---
---
RETURN
No DATA file is needed at the start of the procedure, as the latter works directly on the data of
sub-structure n1.

© 2019 ESI Group
In addition to non-linear computation on sub-structure n1, the procedure computes, where

necessary, the matrices and right-hand sides of the equations for standard sub-structures. These
can also be pre-computed by the CALCULATE procedure described in Chapter 10.
The FUNCTION command can be used to modify the loads applied to standard sub-structures
in time. The time functions defined by tables i1, i2 ... i j, concern the loads applied to the
highest level SUB-STRUCTURE (which has j tables). These time functions have no effect on
the non-linear super element (n1), which time-related multiplication of the various parameters is
described directly in the data (n1). Standard sub-structures can naturally include several loads,
whereas only one load must be defined for the non-linear element (n1).
The RENUMBER command is used to minimize the bandwidth of the problem. This
command is similar to the standard RENUMBER command (Chapter 6), except that it also
integrates the connections of the standard SUB-STRUCTURES, and those of some media.
Although this command is optional, it should be activated in most cases to obtain optimum
efficiency for the method.
At output, the procedure delivers a DATA file, and a conventional TRANSIENT file which
can be saved by the command:
SAVE DATA TRANSIENT p
The DATA file delivered at output is compatible with the TRANSIENT file, and contains:
l all data for sub-structure n1
l connecting nodes between the standard sub-structures which do not belong to sub-structure nl.
Example:
X nodes added
Furthermore, codes are modified according to the nature of the problem.
At the present time, only results concerning the non-linear geometry are accessible, and the
DATA and TRANSIENT files produced by the procedure must be used together for post-
processing.
For a restart, only the TRANSIENT file is required as input data for the procedure, as the latter
works directly on the permanent data file for SUPER ELEMENT n1. On the other hand, the
RENUMBER and FUNCTION commands

© 2019 ESI Group
Example:
SEARCH TRANSIENT p
or
ASSIGN 19 TRANp.TIT BINARY
TRANSIENT NON-LINEAR SUPER nl
----
FUNCTION i1, i2 ..., ij
----
RENUMBER
(renumbering front )
----
INITIAL CONDITION RESTART CARD c
----
RETURN
SAVE DATA TRANSIENT p
Limitations
These limitations are particular to the method. All areas considered as "standard" sub-structures
are assumed to have linear static behavior. In particular, this means that mass or inertial effects
are ignored in standard super structures. Furthermore for mechanical problems, standard super
structures are created on the assumption that the parts of the structures to which they
correspond, have a linear elastic behavior (small stresses and small displacements).

© 2019 ESI Group
12.6 Transient Shell
12.6 TRANSIENT SHELL

12.6.1 Mechanical
This particular transient feature allows buckling and nonlinear phenomena to be analyzed. For
this kind of computation the structure has to be represented by 8-node 3D isoparametric shell
(SHE8) or 2-node axisymmetric shell (SHED) elements.
The transient analysis for SHE8 and SHED elements is activated by the following command:
TRANSIENT SHELL (VERIFY) (FILE S)
SHELL Label required for SHE8 and SHED elements

VERIFY Specifies a check-run without analysis.
FILE S Indicates input of a thermal load from a file unit :
l if S = NODE, nodal temperatures and gradients.
l if S = MESH, temperatures at cross thickness integration points
of elements (not available).
This file is created by the command:
TEMPERATURE TRANSIENT SHELL
and includes nodal temperatures and gradients from a TRANSIENT SHELL THERMAL
calculation.
The command:
DIMENSION MEMORY mtot TABLE maxtab
provides for specification of dimension parameters.

MEMORY mtot Specifies allocation of mtot memory-words for the execution of the
transient calculation. The default value is 100 000.
TABLE maxtab Specifies the value of the maximum load table index. The default value
is 1000.
The command:
BEHAVIOUR (PLASTIC S or CREEP i) (LAGRANGE S)
defines the type of non-linearity to be used. By default, the computation is linear elastic.

© 2019 ESI Group
PLASTIC S Activates a plasticity model, according to the value of label S :

l if S = KINEMATIC, plasticity with kinematic hardening. (See Note 1.)
CREEP i Activates a creep model. Each type corresponds to a creep rate
integration scheme :
l if i = 1, one-step implicit scheme.
l if i = 2, implicit iterative scheme.
l if i = 3, mid-point scheme.
l if i = 4, Eulerian scheme. (See Notes 1. and 2.)
LAGRANGE S Initiates analysis with large displacement formulation, (cf. Note 1) :
l if S = UPDATE, the formulation used is the updated Lagrangian
formulation with geometry updating at each iteration. (mandatory with
SHE8).
l if S is omitted, total Lagrangian formulation is used (mandatory with
SHED). (See Note 1.)
Notes
l In the models adopted, plastic strains can develop in conjunction with creep strains and in the
presence of large displacements.
The labels PLASTIC and CREEP have both to be selected. With cinematic strain-hardening, the
curve σ - ε may have two slopes only : the elastic slope followed by the slope due to strain
hardening. This curve is thus defined by two points, the first corresponding to the initial elastic
limit.
l The creep model integrated is a strain hardening type. Four creep integration schemes are used:
l Implicit, one-step scheme: non-iterative, good stability and accuracy, more economical than
schemes 2 and 3, recommended for relatively small time steps.
l Implicit iterative scheme: Iterative, even for elastic calculations, unconditionally stable, large
time steps are allowed.
l Mid-point scheme: Iterative, unconditionally stable, large time steps are allowed, recommended
in the presence of plasticity.
l Eulerian scheme: Non-iterative, but with severe time step limitations because of conditional
stability, useful for manually introduced creep laws.
The following command concerns the selected algorithm and the increment parameters.
ALGORITHM (SAVE) (RIKS igroup) (AMPLITUDE iampl) (ITERATION n) --

(RESIDUE r) (RATE m) (ITAN itan) (IT1 it1) (IT2 it2) (IT3 it3)

© 2019 ESI Group
LAGRANGE S Initiates analysis with large displacement formulation, (cf. Note 1) :

l if S = UPDATE, the formulation used is the updated Lagrangian
formulation with geometry updating at each iteration. (mandatory with
SHE8).
l if S is omitted, total Lagrangian formulation is used (mandatory with
SHED). (See Note 1.)
SAVE Within an elastic analysis, the computation of several load cases with
the same stiffness matrix may be performed. (See Note 1.)
RIKS igroup Initiates the Riks-Wempner algorithm with automatic load
incrementation
- igroup = increment group number defined in the command TIME for
which the Riks algorithm will be used. (See Note 2.)
AMPLITUDE Initial amplitude as a percentage of the unit load associated with the
iampl Riks increment group.
ITERATION n n is the maximum number of iterations (10 by default).
RESIDUE r r is the relative convergence criterion on the normalized equilibrium
forces (r = 0.005 by default).
ITAN itan Governs the type of stiffness matrix used at the first iteration of an
increment (see the card TIME to define increments).
if ITAN = 0, elastic stiffness matrix.
if ITAN = 1, stiffness matrix defined by it1, it2, it3.
IT1 it1 Reforms stiffness at the it1-th iteration.
IT2 it2 then at the it2-th iteration.
IT3 it3 Then every it3 iterations (by default, itan=0, it1=2, it2=3, it3=1). (See
Note 2.)
Notes
l 1. The advantage of commands ITAN, IT1, IT2, IT3 is to set up an efficient and economical
strategy for the solution of nonlinear problems. The default values are perfectly suitable for
conventional nonlinear analyses. Their activation may prove necessary for the implementation of
the Riks method.
l 2. Riks method:
The method is described in Chapter 19.
l 2.1. Types of load incrementation:
l Constant arc-length incrementation : the arc length is constant along the initial tangent (first
method described in Chapter 19) ; this method is used with the command STATIC.
l Variable arc-length incrementation combined with a buckling calculation : this method has the
advantage of obtaining a good approximation of the limit load in a few steps. The calculation
continues with a constant arc-length incrementation.
l This incrementation is performed with the command BUCKLING.

© 2019 ESI Group
BUCKLING TYPE ityp = 13, 14, 15 or 16
l In this case, the above values of ityp correspond respectively to values 3, 4, 5 or 6 of ityp defined
in the table 12.6.1.
Example:
TIME
1.5 STEP 0.5
4. STEP 0.5
ALGORITHM RIKS 2 AMPL 20 ITAN 1 IT1 1 IT2 2 IT3 1
BUCK TYPE 15 INCR 8 EIGE 2
BEHAVIOUR LAGRANGE
COND REST CARD 3
The increment number to be specified necessarily corresponds to the last increment of the Riks
method.
General Notes on the algorithm.
If all the eigenvalues obtained are negative, the algorithm employs the constant arc-length
incrementation.
The type of incrementation can be modified during a restart procedure.
For elasto-plastic materials and when using the present method, it is advisable to define ITAN
1IT1 1IT2 2.
A check parameter S (Bergan's parameter) is provided, which characterizes the global stiffness
behavior under increasing loads.
If S > 1, the structure becomes stiffer, and if S < 1, the structure becomes softer.
At a limit point type of buckling, S = 0.
The parameter S to be considered is the one printed at the end of the first iteration of each
increment.
Definition of the reference load with
TIME
t1 STEP dt1
t2 STEP dt2
....
tn STEP dtn
ALGORITHM RIKS igroup AMPL iampl
If lambda is the multiplication factor, the time is given by :

© 2019 ESI Group
and any load function is given by the following expression :
constant reference unit

load load
l The reference unit load cannot be modified during a restart, using the Riks method.
l To increase the number of increments required, modify the data STEP dt (igroup).
l After the Riks method, it is possible to come back to the conventional algorithms by replacing
t(igroup) by t(λ (igroup)) :
Definition of Reference Load

The following command specifies the type of analysis required :
STATIC
or
BUCKLING (INCREMENT incr) (TYPE itype) --

(CENTR lamb) (RANGE lamb1 lamb2) (CRITERION epsi) --
(MAXI maxiter) (EIGENVALUE nbval) (LANCZOS nlanc) --
(CHANG ichang)
STATIC Command: static analysis

BUCKLING Command: buckling analysis
INCREMENT Increment at which the buckling mode will be sought. This data is
incr necessary for incr > 2.
TYPE ityp Type of bifurcation (0 by default). (see Note 1.)
CENTR lamb Central value around which the eigenvalue will be sought (0 by
default). (See Note 2.)

© 2019 ESI Group
CRITERION Convergence criterion of iterative method (0.00001 by default). (See

epsi Note 3.)
MAXI maxiter Maximum number of iterations (16 by default).
EIGENVALUE Number of eigenvalues requested (2 by default).
nbval
RANGE lamb1 Range for the eigenvalue analysis lamb2(- ∞ and + ∞ by default).
LANCZOS Dimension of the blocks of the Lanczos method. (See Note 3.)
nlanc:
CHANGE Dimension of the orthogonal bifurcation space (0 by default). (See
ichang Note 4.)
Notes
l 1. Types of bifurcation problem
There are two approaches to the bifurcation problem:
l 1.1. Stability analysis of equilibrium point on the load-displacement curve (fundamental path),
corresponding to the solution at the load increment incr. This involves computation of a buckling
mode, alone or preceded by a stress analysis, incr being the load increment at which the
computation of buckling modes will be performed.
l 1.2. Approximation of the critical load along the tangent to the load-displacement curve,
computed at the equilibrium point corresponding to the increment incr. By contrast with the
previous case, loading at the increment (incr+1) must be supplied by the user so that the tangent
at the equilibrium point on the load-displacement curve can be computed.
Linearized Bifurcation Analysis

l 2.Conditioning of the eigenvalue problem
Correct processing is checked a posteriori by computing the error bounds, which indicate
bracketings for each of the eigenvalues calculated and an error measurement coefficient,
characterizing the accuracy of the computation. This coefficient must not exceed a few percent.
To improve the conditioning and accelerate the convergence, the label CENTRE can be used.
The data (lamb) chosen should be close to the eigenvalue sought, since convergence takes place
first around the center.

© 2019 ESI Group
l 3.>Eigenvalue problem solution methods

Two solution methods are available, the subspace-iteration method and the Lanczos block method,
the latter being generally more efficient but possibly less stable than the former.
By default, the subspace-iteration method will be used.
The data (nlanc) define the dimension of the vector blocks, on which the simultaneous iteration will
be performed. The value by default is:
nlanc = min (nbval, 4)
l 4.Bifurcation in an orthogonal space
The label CHANGE allows the displacement-space of the bifurcation problem to be changed with
respect to the fundamental path. It can be used during a restart under the following conditions.
node and element data cannot be modified.
For a SHED mesh, this command can be used to enter additional components (sine and/or
cosine) ; the data (ichang) are then equal to the number of added modes.
The types of problem accepted are then as follows :
ityp = 0, 1, 7, 8, 9, 10
Moreover, if the boundary conditions are not modified for the Fourier modes of the fundamental
path, the types ityp = 5 and 6 are also allowed.
The table below summarizes the different types of bifurcation analysis and their conditions of
use.

© 2019 ESI Group
Table 12.5: Types of bifurcation analysis

© 2019 ESI Group
The following command is allowed only for the SHED elements:
DEFECT TRANSIENT ntran (mi, ci)
DEFECT Specifies the initial defect of the SHED model.

TRANSIENT TRANSIENT file number stemming from a previous computation
ntran based on the same model. Generally this TRANSIENT file will be
stemming from a buckling static analysis on the structure without
default.
mi Card number on TRANSIENT file including the modal displacement
di.
ci Multiplication coefficient of the modal displacement.
The defect will be of the following type:
c1d1 + c2d2 + .... + cndn (n ≤ 10)
The following command is used to define initial conditions and process restart procedures.
INITIAL CONDITIONS (RESTART CARD n)
RESTART The program restarts at the increment (n+1).

CARD n
The increments are performed using the command:
TIME
ti step dti / (S=s) STORE n CREEP dtm RESTART nr
ti step dti Increment group i is defined by t(i)-t(i-1).

The number of increments in the group i is given by (t (i)-t(i-1))/dti.
S=s S label to be chosen from DISPLACEMENTS, FORCES,
REACTIONS.
s print-out periodicity. If s=0, print-out at time ti.
n same principle as s for result storage on TRANSIENT file. By default,
n=s if the label STORE exists. If not the last card of each macrotime
step is stored.

© 2019 ESI Group
dtm If negative, the present data indicates the table (-dtm) of the creep
function defined under the command DEFINITION.
If positive, (dtm) defines the dwell time during each increment of the
group number (i). If (tflu) is the dwell time during increment j, then the
dwell time during increment (j+1) is (tflu+dtm). Reminder: Creep
analysis is activated, only if keyword CREEP is declared under the
command BEHAVIOR.
RESTART nr Storage of data useful for restart on file RESTART.DAT, every nr
times steps.
By default nr is the minimum of s values.
It is not possible to chain a first run with a restart run in the same
session.
It is necessary to perform the two computations separately.
The following command governs the print-out of the results:
PRINT COMPLETE
COMPLETE:Activates a full print-out at each iteration.

By default partial results are printed at the end of each analysis increment. We recommend
using the default value. For more details, see chapters 8.9.1.4 and 8.9.2.4.
Precaution with files:
1) Restart files
Restart of the transient calculation needs:
l saving of file RESTART.DAT at end of 1st calculation.
l access to file RESTART.DAT at restart.
Note:
l file RESTART.DAT is used in read and write modes.
2) DATA and TRANSIENT files
In the case of the DATA file activated by TRANSIENT SHELL has several loads (for
example: STATIC study), the DATA and TRANSIENT files created from the computation are
not completely compatible (number of loads, number of forces components) and certain
precautions must be taken before using certain standard procedure of SYSTUS.

© 2019 ESI Group
Example

© 2019 ESI Group

© 2019 ESI Group
12.6.2 Thermal
This particular transient feature allows thermal analysis (linear and nonlinear) of structures
modeled with the thermal shell elements (SHET).
The thermal transient analysis for SHET elements is activated by the command:
TRANSIENT SHELL THERMAL (VERIFY)
SHELL labels required for SHET elements

THERMAL
VERIFY see the chapter 12.6.1
The command :
DIMENSION MEMORY mtot TABLE maxtab
provides for specification of the dimension parameters (see Chapter 12.6.1).

The command :
BEHAVIOUR (LINEAR or NONLINEAR)
specifies whether the analysis will be linear or not (linear by default).

In nonlinear analysis, material properties dependent on the temperature are recalculated at each
iteration.
The command:
REGIME (STATIONARY or TRANSIENT)
specifies whether the analysis will be STATIONARY or TRANSIENT.

In transient analysis, material properties dependent on the temperature are recalculated at each
increment.
The following command is used to define the restart procedures (see 12.6.1)
INITIAL CONDITIONS (RESTART CARD n)
The increments are defined by the command:
TIME
ti STEP dti / (S=s) STORE n RESTART m

© 2019 ESI Group
Notes
l In stationary analysis: use ti = dti = 0.
l A transient analysis must be initiated by a stationary time step (see also 12.6.1).
The following command governs the printout of results:
PRINT (COMPLETE)
COMPLETE Activate additional printout of mean temperatures at each Gauss points

in the thickness of the shell.

© 2019 ESI Group
Example : Cold shock on the external face of a tube.

© 2019 ESI Group
12.7 Utility Routines
12.7 UTILITY ROUTINES

12.7.1 Introduction
Two utility routines are described here. The first is used to link a transient thermal analysis to a
transient mechanical analysis, and the second is used to transform TRANSIENT files in such a
way as to make them more manageable or compatible with other modules of SYSTUS, and to
carry out certain postprocessing operations.
12.7.2 Transient Thermo-Mechanical Analysis: the

TEMPERATURE Command
Transient thermomechanical analysis, linear or nonlinear, may be performed by the
TRANSIENT NON-LINEAR routine. This analysis requires the use of temperature cards
previously written to a TRANSIENT file in the course of a linear or nonlinear transient thermal
analysis.
Prior to the execution of mechanical analysis, it is necessary to create 1 or 2 temporary files
including thermal information (at nodes and/ or elements) at selected times.
For the execution of this routine, it is necessary to enter DATA and TRANSIENT files
corresponding to the thermal analysis.
The input data differ according to whether the analysis uses:
l "continuous media" elements only in the following options
o THERMAL
o THERMAL METALLURGY
l thermal shell elements (code 240x) which are associated or not with "continuous media"
elements in the following options:

o THERMAL ENTHALPY,
o THERMAL ENTHALPY MULTI,
o THERMAL METALLURGY MULTI
l thermal shell elements (code 240x) followed by a computation using the option spatial
SHELL SHE8
l thermal shell elements from SHET thermal shell option.
a) Thermo-mechanical analysis for "continuous media" type structures with THERMAL

option (without metallurgy)
To use results from a linear or non linear thermal analysis (TRANSIENT THERMAL or
TRANSIENT NON- LINEAR procedures) for a non- linear transient thermomechanical
analysis, the following command is used:
TEMPERATURE TRANSIENT CARD <list> or LAST DT tshift

or
TEMPERATURE TRANSIENT CARD TIME tinit tfinal DT tshift

© 2019 ESI Group
This routine creates a file including temperatures at nodes. Provided there is no abnormal
behavior, the time value for each card selected is printed.
It is possible to specify, either a list of cards, or a range of cards including the two instants tinit
and tfinal using label TIME or the last card only with the label LAST.
It is also possible to shift the time range computed during the thermal analysis by using the label
DT. For example, if the thermal results have been computed from 0 to 100 seconds, by
specifying DT 50, the instants sent to the thermomechanical analysis will be from 50 to 150
seconds. This can be useful for mechanical structures submitted to a cyclic thermal loading
saving the useless duplication of thermal computation.
Since the temporary file is lost at the end of a run, the TEMPERATURE command must be
executed before any thermo-mechanical analysis by the TRANSIENT NON-LINEAR routine.
In the mechanical analysis, the times specified in groups of data following the TIME command
need not correspond to those of the thermal analysis. If necessary, temperature fields are
interpolated linearly over a time step of the thermal analysis to obtain values required by the
mechanical analysis.
Rules used for the interpolation are:
l If temperatures are required for times less than the lowest value of time for which a card has
been calculated, or greater than the highest value of time for which a card has been
calculated, they are taken to be the temperatures at those lowest or highest values of time,
respectively.
l otherwise, the interpolation is made between values of time of the thermal analysis which
"frames" the value of time of the mechanical analysis.

Note:
l The transient analysis of the mechanical problem may also be carried out with thermal input data
in the form of a single temperature card created by the TEMPERATURE command as described in
Section 4.12. In this case, the temperature field is taken to be independent of time.
Example:

TEMPERATURE TRANSIENT CARD 0 TO 2
TRANSIENT NONLINEAR
BEHAVIOR PLASTIC
TIME
1.5 STEP 0.5 / DISPLACEMENT FORCE 1
RETURN

© 2019 ESI Group
b) Thermo-mechanical analysis for a "continuous media"-type structure with

metallurgy from THERMAL METALLURGY option
To use results from a thermal analysis with metallurgy, the following command is used:
TEMPERATURE TRANSIENT METALLURGY CARD <list> or LAST DT tshift

or
TEMPERATURE TRANSIENT METALLURGY CARD TIME tinit tfinal DT tshift
This routine creates 2 files:

l a file including temperatures at nodes,
l a file including metallurgic information at elements having metallurgic phases
c) Thermo-mechanical analysis for a structure totally or partially modeled with shell

thermal elements (type 240X) from THERMAL ENTHALPY (MULTI) option
For a computation in spatial SHELL option with code 24XX elements and through the medium
of the NON LINEAR TRANSIENT solver, the routine provides transfer of all information
necessary to compute the temperature at each station of mechanical element. Thus, the input
data is:
TEMPERATURE TRANSIENT SHELL CARD <list> or LAST DT tshift

or
TEMPERATURE TRANSIENT SHELL CARD TIME tinit tfinal DT tshift
The family of thermal shell elements (code 24XX) of THERMAL ENTHALPY computation
option (Heat Transfer Reference manual - Chapter 6) is compatible with mechanical elements
(code 24XX) of SHELL spatial option.
This routine creates 2 files:
l a file including thermal information at shell elements.
d) Thermo-mechanical analysis for a structure partially modeled with enthalpy thermal
shell elements (type 240X) and with "continuous media with metallurgy" elements from
THERMAL METALLURGY MULTI option
TEMPERATURE TRANSIENT SHELL METALLURGY CARD <list> or LAST DT tshift

or
TEMPERATURE TRANSIENT SHELL METALLURGY CARD TIME tinit tfinal DT tshift

© 2019 ESI Group
This routine creates two files:

l a file including thermal information at shell elements and metallurgic information at
"continuous media" elements having metallurgic phases.

e) Thermo-mechanical analysis for a structure partially modeled with thermal shell
elements (type 240X) with metallurgy from THERMAL METALLURGY option
TEMPERATURE TRANSIENT SHEM METALLURGY CARD <list> or LAST DT tshift

or
TEMPERATURE TRANSIENT SHEM METALLURGY CARD TIME tinit tfinal DT tshift
This routine creates two files:

l a file including thermal information at shell elements and metallurgic information at shell
elements having metallurgic phases.

f) Thermo-mechanical analysis for a structure modeled with enthalpy thermal shell
elements followed by a SHELL SHE8 computation option
For a spatial SHELL SHE8 computation option the input data are as follows:
TEMPERATURE TRANSIENT SHELL SHE8 (GRADIENT) CARD <list> or LAST
The GRADIENT label is used to transfer thermal gradients in addition of mean temperatures at
model nodes.
This routine creates a file including thermal information at nodes of the model.
g) Thermo-mechanical analysis for a structure modeled with shell thermal elements
SHET followed by a spatial SHELL SHE8 computation option
To use results from a transient thermal analysis using SHET shell elements (TRANSIENT
THERMAL SHET procedure) for a thermo-mechanical analysis with SHE8 sell elements
(TRANSIENT SHELL procedure), the input data are as follows:
TEMPERATURE TRANSIENT SHELL (GRADIENT) CARD <list> or LAST
GRADIENT label (optional) is used if the TRANSIENT file created in thermal analysis
contains temperature gradients through the thickness of the shell besides temperatures.
This routine creates a file including thermal information at nodes of the model.

© 2019 ESI Group
Example:

TEMPERATURE TRANSIENT SHELL CARD 1 GRADIENT
SEARCH DATA 130
TRANSIENT SHELL FILE NODES
DIMENSION MEMORY 20000 TABLE 1
STATIC
TIME
0. STE P 0 .
RETURN

© 2019 ESI Group
12.7.3 Manipulation of the TRANSIENT File - The SELECT

Command
The TRANSIENT files created by transient analyses are often very large.
Two situations may arise:
l 1) The dynamic analysis has been carried out for a structure modeled by a mesh with a small
number of degrees of freedom, but with a very large number of time steps, for these had to be
very short. As a result a very large number of cards are written to the TRANSIENT file.
l 2) A static non-linear analysis with the labels PLASTIC or EULER LAGRANGIAN has
been carried out for a structure modeled by a mesh with a large number of degrees of
freedom. The amount of data in each card is greatly augmented, in particular by force or
stress history data.
Furthermore, in the second of these situations the volume of force or stress data can render
impossible the use of certain modules (POSTPROCESSING for example).
The SELECT command may be used to get around certain of these difficulties.
The OUTPUT command described in Section 3.3 may also be indirectly used to achieve
processing selectivity.
Input files:
DATA (use SEARCH to attach this file: see also notes)
TRANSIENT (automatically attached by the program: SEARCH is irrelevant).
Output files:
DATA (modified)
TRANSIENT (if the label STORE is present)
Syntax
SELECT TRANSIENT n
CARDS <list> or LAST or LAS2 (labels) COMPONENTS i1 i2 --
(NOSEQUENCE) (STORE) (PRINT) (ENVELOPE)
CARDS (numbers) (labels) COMPONENTS j1 j2 --
(STORE) (PRINT) (ENVELOPE)(TIME (t0))
RETURN
SELECT Command
TRANSIENT n Number of the TRANSIENT permanent file in which the cards will be
selected.

© 2019 ESI Group
CARD Command
(Numbers) or Specifies which cards are to be selected: given by their numbers or by
LAST or LAS2 keywords LAST for the last card or LAS2 for the last two cards.
Several CARD commands can coexist in the same SELECT
command.
(Labels) DISPLACEMENT
VELOCITY
ACCELERATION
REACTION
FORCE
PRESSURE
VALUE
These labels specify which results are to be processed. Several of these labels can coexist in the
same CARD command.
COMPONENT i1 Specifies which components of force (or stress) are to be selected. If i2 neq
i2 0, components i1 to i2. If i2 = 0, components 1 to i1. By default (or if
i1=i2=0), the first KPC components will be selected. This selection is
useful in restoring the force data to their standard size (see further the
relation between KPC and the option).
STORE If present, creates a new TRANSIENT file which can be made permanent
by means of the SAVE command.
NOSEQUENCE If present, the original card numbers are retained on the new
TRANSIENT file. This label is only effective on the first CARD
command, in order to prevent erasing records of the new TRANSIENT
file. By default, if this label is not used, the cards written on the new
TRANSIENT file are renumbered.
PRINT If present, the selected results are printed out.
ENVELOPE If present, the extreme values of the selected results and the time at
which they occur are printed out.
TIME (t0) If present, the time of the selected cards are translated: the first card takes
for time value t0 and the next cards, the values t0+Δt with Δt equal to the
initially existing difference of time between first selected card and the next
ones (default value: t0 = 0).

© 2019 ESI Group
Notes
l Employ SEARCH DATA before SELECT, if you previously used another SELECT command
containing a STORE command.
l If in the same SELECT you use several CARD commands in which force components are stored.
The number of such components must remain constant.
l If not, the program uses by default the number of components in the first such command.
option 1 2 3 4 5 6 7
KPC 4 3 6 6 3 6 4
Default values for number of stresses in linear elasticity analysis.
where :
Option 1 = all types of structure with d.o.f. UX, UY, RZ
Option 2 = all types of structure with d.o.f. UZ, RX, RY
Option 3 = all types of structure with d.o.f. UX, UY, UZ, RX, RY, RZ
Option 4 = all types of structure with d.o.f. UX, UY, UZ
Option 5 = thermal and electromagnetic options
Option 6 = all types of structure with d.o.f. UX, UY, UZ, RZ
Option 7 = all types of structure with d.o.f. UX, UY.
Example:
SEARCH DATA 101

SELECT TRANSIENT 101
CARDS 8 TO 10 DISPL FORCE COMPO 1 TO 4 PRINT --
STORE ENVELOPE NOSEQUENCE
CARDS 9 FORCE COMPO 6 TO 9 PRINT STORE ENVELOPE
RETURN

© 2019 ESI Group
12.8 Transfer of Physical Quantities from Mesh to Mesh
12.8 TRANSFER OF PHYSICAL QUANTITIES FROM

MESH TO MESH
This problem concerns the transfer of all known physical quantities for discretization elements
(nodes and elements) from one mesh to another. The initial mesh (source) and new mesh
(target) are assumed to be known, and have necessarily the same geometrical support.
Applications concern:
l Large strain problems (fracture, forming, etc.). In this case, the transfer takes place from initial
deformed mesh to the target mesh.
l Multi-physical problems, such as thermal computation and mechanical computation on
separate meshes in connection with a thermal mechanical analysis.

Transfer of known quantities at the nodes (displacements, temperatures, post-averaging stresses,
etc.) raises no particular problem. Each node of the new mesh is projected onto the initial mesh.
We then know the element of the initial mesh which contains the node, and its reduced
coordinates (ξ,η,ζ) and the quantities assigned to the node through the shape functions of the
element.
Transfer of known quantities at the integration points of the elements is a more delicate matter.
In SYSWORLD two mapping tools are available: the first one uses two procedures proper to
SYSWORLD dealing either with results averaging at nodes or with history file rebuilding to
map fields defined at integration points, the second one uses the library external to
SYSWORLD "Mapping Engine Component" which contains its own techniques to map
computed results.
The two procedures proper to SYSWORLD for map values defined at integration points are
described hereafter.
The first procedure is simple and fast, although generally giving limited precision. The
quantities at the nodes of the initial mesh are averaged. The integration points of the new mesh
are then projected onto the initial mesh, and finally the new values are computed, using the
shape functions of the initial mesh. In no case does this method provide for compliance with
the behavior law of the material.

© 2019 ESI Group
Input data for this procedure is as follows:
SEARCH DATA ns (source data)

PROJECT TRANSIENT (or result labels) nts STRESSES ncs DEFORMED CARD nc
SELECT ELEMENTS <list> (or MATERIALS m or DIMENSIONS d)
DATA nc (REMESH inc METHOD met)
PARAMETERS INTERIOR EXTERIOR EPSILON e
CARDS <list> or LAST
RETURN
SAVE DATA TRANSIENT nc
PROJECT Procedure call command.

TRANSIENT Number of transient file containing source transient analysis results.
nts All results are projected if this label is present. It is also possible to
project only part of these results, using one or more of the result labels
(DISPLACEMENTS, VELOCITIES, ACCELERATIONS,
REACTIONS, FORCES, NODE STRESSES, NODE STRAINS,
PRESSURES OR VALUES) in place of TRANSIENT.
STRESSES ncs Number of transient file containing node stresses resulting from an
averaging procedure (AVERAGING TRANSIENT) on transient file
number nts. This labels is only required for projection of the FORCES
file.
DEFORMED This label must be used when the target data mesh is obtained from the
CARD nc deformed source geometry at card nc.

© 2019 ESI Group
SELECT Selection of source mesh elements concerned by the projection. This

label is particularly useful in a case where the source and target meshes
only differ in a restricted zone.
DATA nc Number of target data file. This file must contain a full description of
the problem, namely material properties, loads, boundary conditions,
etc.
REMESH inc This label indicates that source and target meshes have been built from
the same initial mesh using the command REMESH (cf. 6.3.9).
Projection is made onto the whole mesh (the command SELECT is not
valid) by using the information from the common initial mesh before
cutting.
Limitations for this label, which are added to those existing on the
mesh, are the following:
l average quantities at nodes (stresses and strains) and reactions are not
projected,
l projection cannot be applied to the deformed geometry.
When the activation/deactivation option is used (cf. 12.9), the element
state (activated or deactivated) can also be transferred by using the label
STATE as well as the results labels (DISPLACEMENTS,
VELOCITIES...). When using the label TRANSIENT, the state(as
well as other results) is transferred.
An increment inc must be specified, if during remeshing, an increment
different from the default one has been used for the numbering of
elements.

© 2019 ESI Group
METHOD met The method used is the following:

In case of refinement, all Gauss points of an element take the value of
the Gauss point belonging to the nearest source element.
When de-refining, if met = 1 (by default), then the value assigned to the
target Gauss point is equal to the value of the Gauss point of the nearest
source mesh.
If met = 2, then the value assigned to the target Gauss point is equal to
the mean value of the Gauss points of the nearest source mesh.
PARAMETER If the point to project fall exterior to the element but within a distance
EXTERIOR EPSILON e (relative to the element size) it is consider as a surface
point.

© 2019 ESI Group
PARAMETER If the point to project fall interior to the element but within a distance
INTERIOR EPSILON e (relative to the element size) it is consider as a surface
point.
PARAMETER Relative distance of the boundaries (default e=0)
EPSILON e
CARDS <list> List of cards to be projected and stored in the target transient file.
If no card is specified, then all cards are selected.
the number n.
This procedure delivers a transient output file which can be used for a new analysis. However
it is impossible to recompute mean values of the FORCES file.
With the second procedure, the complete known result history for the initial mesh, contained in
a HISTORY, file is projected onto the new mesh, and a coherent TRANSIENT file for the new
mesh is then reconstituted (REBUILD label with TRANSIENT non-linear command). This
method is more cumbersome, but does allow full compliance with the material behavior law
(total coherence between nodal displacements and results for element integration points).

© 2019 ESI Group
Input data for this procedure is as follows:
SEARCH DATA ns (source data)

PROJECT HISTORY ns
SELECT ELEMENTS <list> (or MATERIALS m or DIMENSION d)
DATA nc
CARDS 0 TO n
RETURN
STORE DATA HISTORY nc
TRANSIENT NON-LINEAR REBUILD

...
RETURN
SAVE DATA TRANSIENT nc
Interpretation of the labels is the same as for the first procedure, apart from label CARDS,
which must appear in the form indicated above (0 TO n). The projection procedure delivers a
HISTORY output file concerning the target data. A TRANSIENT file can then be
reconstituted using the TRANSIENT NON-LINEAR REBUILD procedure (Chapter 12.4).
Limitations
The procedure is limited to the following computation option elements:
l OPTION TWO-DIMENSIONAL (all sub-options except generalized plane strain option)
l OPTION THREE-DIMENSIONAL
l OPTION THERMICS (all sub-options)
l OPTION ELECTROSTATIC (all sub-options)
l OPTION ELECTROKINETIC (all sub-options)
l OPTION MAGNETIC VECTOR PLANE or AXISYMMETRIC
l OPTION MAGNETIC VECTOR SPATIAL
l OPTION ELECTRODYNAMIC MAGNETIC FIELD) (all sub-options)
l OPTION MAGNETIC SCALAR (all sub-options)
l OPTION ELECTROMAGNETIC AV
l OPTION ELECTROMAGNETIC APSI

© 2019 ESI Group
Input data for using the "Mapping Engine Component" tool is as follows:
PROJECT MULTI (METHOD met) (CRITERIA crit DISTANCE d) (UPDATE) (EXTRACT)

MODEL ids1 DATA nds1 (NAME names1_)
FIELD ids1 TRAN nts1 <list> CARD nc or LAST
…
MODEL idsn DATA ndsn (NAME namesn_)
FIELD idsn TRAN ntsn <list> CARD nc or LAST
MODEL idc DATA ndc (NAME namec_)
FIELD idc INITIALIZE
(entities) / <field1> v11, …, v1n
(entities) / <field2> v21, …, v2n
…
SOURCE ids1 (NODE ns1 ns2 ns3 (INTERNAL) (layer)) TARGET idc (NODE nc1 nc2 nc3) (PART idp1)
…
SOURCE idsn (NODE nsn nsn nsn) TARGET idc (NODE ncn ncn ncn) (PART idpn)
RETURN
SAVE DATA TRANSIENT ndc
PROJECT Procedure call command.

MULTI Activates the use of ‘Mapping Engine Component’ library.
METHOD met Mapping algorithm :
met=1 : Interpolation (default value)
For this algorithm, nodal values at a target node are interpolated from
the nodal values of the source element in which the target node is
projected and, values at a target integration point are copied from the
value of the nearest source integration point.
CRITERIA crit Criteria to use if a target point is outside the source discretization.
crit = 1 : Mapping job is cancelled (default value).
crit = 2 : Mapping job continues and a default value is assigned to the
problematic target points. The default field is either a zero value, a set of
zero values, a nil vector or a nil tensor depending of the nature of the
field.
crit = 3 : Mapping job continues and the target point is given the value
of the nearest source point around, at a given distance. The distance is
given by the keyword DISTANCE.
UPDATE If selected with mechanical models, the geometries of the source models
are updated through the displacement fields at the models’ nodes before
doing the mapping.
DISTANCE d Radius of the sphere inside which the nearest source entity of a target
entity is seeking.
This keyword is required only if crit=3.

© 2019 ESI Group
EXTRACT Prints information during the procedure progress.

MODEL id This label defines a model for the ‘Mapping Engine Component’ library. In
the component, each model is defined with an identifier id.
DATA nd This label associates the data structure nd to the model.
NAME NAME_ This label defines the name of the data structure given by file NAME_
DATAnd.TIT.
FIELD id This label defines the field(s) associate to the previously defined model
ids.
TRAN nt <list> Number of the transient file containing the results to map. In addition to
this label it is required to list the results file to map: DISPLACEMENT,
VELOCITY, ACCELERATION, REACTION and/or FORCE.
CARD nc Number of the card to map.
the number nc.
INITIALIZE Initialization of a selection of fields on the target model only. The line
(entities) / … FIELD idc INIT must be followed by one or several lines describing
the fields and their values, each line beginning with a blank character ‘
‘, with:
l (entities): list of nodes or elements given either by their numbers or by the
groups they belong to.
l <field> v1, .., vn: field name, and values:
l TT θ: initial temperature
l P p1 p2 .. pn: initial values of the proportions of phases 1 to n.
SOURCE ids These keywords identify the source and target models by their identifier
TARGET idc for the mapping procedure.
(PART idp) With the label PART following the label TARGET, we limit the target
zone to the idpth PART of the target model.

© 2019 ESI Group
(NODE n1 n2 n3 In case of source and target models are not at the same position and/or
(INTERNAL) not in the same orientation, the labels NODE following both SOURCE
(layer)) and TARGET are necessary to compute the translation and/or rotation
operations of the source model to fit geometrical position of the target
one. Nodes m1, m2 and m3 of the target are then the result of the
transformation leading nodes n1, n2 and n3 from source geometry to the
position of the target one.
Remarks:
l Nodes Ids can be set by their user numbers (default) or their internal one
(INTERNAL must be specified then).
l Layer stands for LOWER, MIDDLE (default) or UPPER:
In case of mapping of a shell model onto a solid one or vice versa, the
layer the nodes of the shell model belong to must be set with respect to
their position in the solid model.
Example for a local frame defined at the lower surface:
RETURN Initiate the mapping of each defined couples SOURCE/TARGET.

© 2019 ESI Group
Limitations
The procedure is limited to the following computation option:
l TWO-DIMENSIONAL (all sub-options)
l OPTION THREE-DIMENSIONAL
l OPTION SHELL SPATIAL
l OPTION SHELL SPATIAL MULTI
l OPTION THERMAL (all sub-options)
l OPTION THERMAL MULTI (all sub-options)

© 2019 ESI Group
12.9 Element Activation/Deactivation
12.9 ELEMENT ACTIVATION/DEACTIVATION

12.9.1 Introduction
Modeling of reparation processes which involve material requires activation of finite elements
during simulation. Conversely, cutting, drilling and machining operations require deactivation
of certain elements.
SYSTUS provides for activation/deactivation of finite elements to reproduce the
appearance/disappearance of material during a thermal or mechanical simulation. The finite
elements concerned are defined using a space/time coordinate function.
This procedure is available with the TRANSIENT NON-LINEAR solver for the following
options:
l THERMAL ENTHALPY
l TWO-DIMENSIONAL
l THREE-DIMENSIONAL
The activation/deactivation of elements on Gauss points is limited to linear and quadratic

(reduced integration) elements.
12.9.2 Activation/Deactivation Function

The activation/deactivation of elements is defined by a space/time co-ordinate function, entered
by means of a function written in simplified FORTRAN, or an USFONC user function
(TABLE routine in data).
This function can take the following values
Function Value Effect
0 None
1 Activate
-1 Deactivate
1000 Return to initial state
The initial state of elements can be defined in the initial conditions for the TRANSIENT NON-
LINEAR routine.
Elements can be activated or deactivated globally or partially at each integration point.
For a mechanical problem, the activation/deactivation function applies by default to the initial
mesh (non-deformed). It can however be applied to the deformed mesh if required.
Permanent transformation is obtained with a function which takes the value 1 or –1 at the
instant corresponding to the change of state of the element, and which either keeps this value or
takes the value 0 subsequently.
Temporary transformation is obtained with a function which takes the value 1 or –1 for the time
interval corresponding to the change of state of the element, and the value 1000 subsequently.

© 2019 ESI Group
12.9.3 Input Data: Element Properties

DEFINITION
…
MATERIAL PROPERTIES
…
ELEMENT (list) / STATE -f GAUSS val DEFORMED d REDUCTION xk
…
RETURN
Four labels are associated with the activation/deactivation of elements. These labels complete
the normal properties describing the behavior of the material.
STATE -f specifies the number (f) of the table giving the activation/deactivation
function.
GAUSS val if val = 1, changes of state are assessed and applied at element
integration points.
if val = 0, changes of state are assessed at the centre of gravity of the
element, and applied globally to the complete element.
DEFORMED d if d = 1, changes of state are taken on the deformed geometry.
REDUCTION factor used to multiply the properties of an element when the latter is
xk deactivated.
When this value is negative (for the mechanical options only), the
Young modulus of the deactivated elements is equal to –xk, note that
the Poisson ratio is multiplied by 10-6.
Deactivated Elements
Mechanical problems
The behavior of a deactivated element is assumed to be elastic, and the Young’s modulus (E)
and Poisson’s ratio values are multiplied by reduction factor xk.
Forces are cancelled and the element is assumed to have zero mass (RHO = 0).
Thermal enthalpy
The thermal conductivity (KX,KY,KZ) of a deactivated element is multiplied by factor xk.
Heat flux is cancelled, and the element is assumed to have zero mass (ρC = 0).
Activated Elements
Mechanical problems
Thermal strain is cancelled on activation.
Stresses and strains are computed from the moment when the element is activated. The initial
strain of the element at the moment of activation is ignored.

© 2019 ESI Group
12.9.4 Input Data: Definition of Initial State

…
INITIAL CONDITIONS
ELEMENTS (list) / IS i
…
RETURN
IS i If i = 1 the element is inactive.

Otherwise, and by default, the element is active.
12.9.5 Data Storage

The state of an element, or its integration points, is/are stored in the FORCES file (blocks of
mean values):
SS(4) for THERMAL ENTHALPY approach,
SS(40) for TWO-DIMENSIONAL option,
SS(60) for THREE-DIMENSIONAL option.
Values in the elements that could be activated or deactivated can not be post-processed.
12.9.6 Practical Tips

An element which could be deactivated should not be loaded, or serve as a bearing element
(CONSTRAINT) for the rest of the structure.
Numerical instabilities can occur when elements are activated/deactivated, in particular if the
time step is too great with respect to the change of state. One way of restricting these
instabilities is to define the change of state for a time interval, and to ensure the adequation of
the selected step with this interval.
For a thermal problem, it is preferable to use 1 st order elements, and when an element is
deactivated, to execute a global change of state and not at the integration points.

© 2019 ESI Group
12.9.7 Examples
Thermal Problem
Elements are activated in this example. A band with a width of two elements is initially in the
inactive state (initial conditions). A heat source is moved across the band, activating the
individual elements in the process. The heat source and the function activating the elements are
superimposed in this example. The elements are activated permanently (F = 0 outside the
activated zone). The part is allowed to cool after 2.5 seconds.
DEFINITION
2D WELDING
OPTION THERMAL ENTHALPY PLANE
TWO-DIMENSIONAL
NODES
1/
2 / 10
3 / 10 10
4 / 0 10
ELEMENTS
1001 / 1 2 3 4
2001 / 2 3
EDGES
100 / 1 2 20
200 / 2 3 20
300 / 3 4 20
400 / 4 1 20
MATERIAL PROPERTIES
E / RHO 1 KX=KY=KZ 0.05 C 0.006 STATE -10
CONSTRAINT
ELEM 2001 TO 2020 / KT 0.01
LOAD
1 / NOTHING
ELEM / QR 35 VARI -20
ELEM 2001 TO 2020 / TT 100.
…

© 2019 ESI Group
TABLE
1 / 1 0 0 5 10
10 / FORTRAN
FUNCTION F(X)
DIMENSION X(6)
XX = X(1)
YY = X(2)
ZZ = X(3)
TT = X(4)
VT = 2*TT
AA = XX
BB = YY - VT
AA = AA * AA
BB = BB * BB
CC = AA + BB
CC = SQRT(CC)
F = 0.
IF (CC.GT.2. ) RETURN
IF (TT.GT.4.) RETURN
F = 1.
RETURN
END
20 / FORTRAN
FUNCTION F(X)
DIMENSION X(6)
XX = X(1)
YY = X(2)
ZZ = X(3)
TT = X(4)
VT = 2*TT
AA = XX
BB = YY - VT
AA = AA * AA
BB = BB * BB
CC = AA + BB
CC = SQRT(CC)
F = 0.
CC = CC/2.
CC = - CC
F = EXP (CC)
RETURN
END

© 2019 ESI Group
RETURN
RENUMBER ITERATION 30
1 INTERN
RETURN
TRANSIENT NON LINEAR

ALGORITHM BFGS ITERATION 20 IMPLICIT 1
PRECISION ABSOLUTE FORCE 1*-3 DISPLACEMENT 2. ENERGY 1*-30
INITIAL CONDITIONS
N / TT 100
E 1001A1381P20 1002A1382P20 / IS -1
TIME INITIAL 0.
5. STEP 0.1 / STORE 1
10. STEP 0.5 / STORE 1
20. STEP 2. / STORE 1
RETURN
SAVE DATA TRAN 1
SEARCH DATA 1
ASSI 19 TRAN1.TIT
TEMPERATURE TRANSIENT CARD
DEASSIGN 19
Mechanical Problem
The thermal analysis for the previous example is used again in this case. The elements are
consequently activated in the same way. Note that the structure should be anchored to a node
not belonging to a deactivated element, where this is possible. On the other hand, symmetry is
naturally applied to ad hoc elements, including deactivated elements.
SEARCH DATA 1
DEFINITION
THERMO-MECHANICAL COMPUTATION
OPTION TWO-DIMENSIONAL PLANE THERMO
RESTART GEOMETRY
CONSTRAINT
NODE LINE 1 4 / UX
NODE 2 / UY
MATERIAL PROPERTIES
ELEMENTS / E=-11 H = 0.2 LX=LY=LZ=15*-6 --
LIMIT=-15 MODEL=1 NU=0.3 STATE -10

© 2019 ESI Group
LOAD
1 / NOTHING
TABLE
1 / 1 0 0 5 10
10 / FORTRAN
FUNCTION F(X)
DIMENSION X(6)
XX = X(1)
YY = X(2)
ZZ = X(3)
TT = X(4)
VT = 2*TT
AA = XX
BB = YY - VT
AA = AA * AA
BB = BB * BB
CC = AA + BB
CC = SQRT(CC)
F = 0.
F = 1.
RETURN
END
11 / 1 20 200000 300 190000 500 165000 750 70000 1000 20000 1200 50
15 / 1 20 700 200 625 400 550 550 400 750 100 850 50 1000 25 1250 1
RETURN
…
TRANSIENT NON-LINEAR STATIC

BEHAVIOR PLASTIC
ALGORITHM BFGS ITERATION 30
PRECISION ABSOLUTE FORCE 1. DISPLACEMENT 1*-6 ENERGY 1*-30
INITIAL CONDITIONS
E 1001 TO 1381 STEP 20 1002 TO 1382 STEP 20 / IS -1
TIME INITIAL 0.
0.5 STEP 0.05 /STORE 1
1. STEP 0.1
10 STEP 1. / STORE 1
20 STEP 2. / STORE 1
RETURN

© 2019 ESI Group
SAVE DATA TRAN 2

© 2019 ESI Group
Contents
CHAPTER 1 INTRODUCTION 1
1.1 Overview 1
1.1.1 Program Execution 1
1.1.2 SYSTUS 3
1.1.3 Data Structure 3
1.1.4 Preprocessing 4
1.1.6 Postprocessing 4
1.1.7 Field of Application 5
1.1.9 Command Compatibility with New Data 7
1.1.10 Environment Variables 10
1.2 Basic Concepts 18
1.2.1 Theoretical Basis 18
1.2.2 Discretization 18
1.2.3 Degrees of Freedom 18
1.2.4 Coordinate Systems 19
1.2.5 Boundary Conditions 19
1.2.6 Loads 20
1.2.7 Node and Element Numbering Systems 20
1.2.8 Units 20
1.3 Command Language 21
1.3.1 Overall Operation of SYSTUS 21
1.3.2 Language Characteristics 22
1.3.3 Language Structure 26
1.3.4 Automatic Number Generation 29
1.3.5 Node and Element Groups 30
1.3.6 Comments 31
1.3.7 Batch and Interactive Processing 32
1.4 Notation 35
1.4.1 Boxes 35
1.4.2 Labels and Numbers 36
1.4.3 Prefixes 36
1.4.4 Choice of Labels 37
1.4.5 Coordinate System 38
1.4.6 Node and Element Groups 40
1.5 Groups of Nodes or Elements 41
1.5.1 Creation of a Group 42
CONTENTS
1.5.2 Modification of a Group 48

1.5.3 Printing Information of a Group 49
1.5.4 Suppression of a Group 49
1.5.5 List of Existing Groups 49
1.5.6 Suppression of Temporary Groups 49
1.5.7 Packing of Existing Groups 49
1.5.8 Suppression of Groups 50
1.5.9 Updating Groups with Respect to the Creation Method 50
1.5.10 Suppression of Name Repetition 50
1.5.11 Utilization of Groups 51
1.5.12 Particular Groups: PARTS 51
CHAPTER 2 UTILITY PROCEDURES 53

2.1 Introduction 53
2.2 Operation Sequence Control 54
2.3 Alternative Input File 56
2.3.1 The Command ASSIGN 56
2.3.2 The Command DEASSIGN 57
2.3.3 The Command READ 57
2.3.4 The Command STOP 58
2.3.5 The Command REWIND 58
2.3.6 Comments 58
2.3.7 Examples 59
2.4 Accounting Information 62
2.4.1 The Command TIME 62
2.4.2 The Command END 62
2.5 Modification of Execution Mode 63
2.5.1 The Command MODE 63
2.5.2 The Command PAGING 63
2.6.1 Executing Operating System Commands 64
2.6.2 The Command HISTORY 64
2.7 Stop the Execution (ONLY IN TRANSIENT COMPUTATION) 65
CHAPTER 3 FILE MANAGEMENT 67

3.1 Introduction 67
3.1.1 The Role of the Files 67
3.1.2 Glossary 68
3.1.3 Brief Description of Files 69
3.2 Permanent Files 71
3.2.1 Identification of a Permanent File 71
3.2.2 The Command SAVE 72
3.2.3 The Command SEARCH 75
3.2.4 The Command DELETE 77
3.2.5 File Labels 78
3.2.6 ASCII File Structure 79
3.2.7 The Command CARD 79

© 2019 ESI Group
CONTENTS
3.3 Communication With Other Programs 80

3.3.1 INPUT/OUTPUT Commands 80
3.3.2 IMPORT/EXPORT Commands 97
3.3.3 PAM-STAMP Interface 99
3.3.4 Convert SYSTUS Results into erf File with PAM-CSM Format 103
3.3.5 Convert SYSTUS/SYSWELD Results into erf File with CATGEN Format 104
CHAPTER 4 DEFINITION 107

4.1 Overview 107
4.1.2 Commands 108
4.1.3 Command DEFINITION RESTART and Label CONTINUE 110
4.2 Title of Problem 113
4.3 The Command OPTION 114
4.3.2 Types of Analysis 115
4.3.3 Thermoelastic Analysis 116
4.3.4 Periodic Structures Subjected to Fourier Decomposed Loads 117
4.4 The Command GEOMETRY 118
4.4.2 Definition of Nodes 119
4.4.3 Coordinate System 120
4.4.4 Translation and Rotation of Coordinate Systems 122
4.4.5 Format of Node Data 123
4.4.6 Definition of Elements 125
4.4.7 The Element Code 126
4.4.8 Format of Element Data 127
4.4.9 Generation of Sequences of Nodes and Elements 128
4.5 Material Properties 132
4.5.3 Global Definition of Material Properties 134
4.6 Media 136
4.6.3 Comments 137
4.7 Constraints 138
4.7.2 Rigid Constraints 139
4.7.3 Elastic Constraints at Nodes 143
4.7.4 Elastic Constraints over Elements 145
4.7.5 Comments 145
4.8 Loading 146
4.8.2 Definition of Load Case 147
4.8.3 Concentrated Loads 150
4.8.4 Distributed Loads 152

© 2019 ESI Group
CONTENTS
4.8.5 Loads Applied to Beams 154

4.8.6 Format of Load Data 156
4.8.7 Comments 157
4.8.8 Example 157
4.8.9 Loads Applied to Media 157
4.9 Releases 158
4.10 The Command RELATION 161
4.10.1 Decoupled Lagrange Multiplier Relations 162
4.10.2 Elimination Relations 164
4.10.3 Example: Structure Supported by another Structure 166
4.11 The Command MASS 169
4.11.2 Conventional Lumped Masses 170
4.11.3 Consistent Masses 170
4.11.4 Comments 171
4.11.5 Thermal Analysis in LINEAR TRANSIENT 171
4.12 The Command DAMPING 172
4.12.2 Conventional Lumped Damping 173
4.12.3 Consistent Damping 173
4.12.4 Comments 174
4.12.5 Thermal Analysis in NON-LINEAR TRANSIENT 174
4.13 The Command TEMPERATURE 175
4.14 The Command VELOCITY 177
4.15 The Command TABLE 178
4.15.2 Arrays 179
4.15.3 Functions Presented as Input Data 180
4.15.4 Functions on File 182
4.15.5 Generation of Functions using Simplified Fortran 185
4.15.6 Generation of Functions using Routines External to SYSTUS 188
4.15.7 Vectorial Functions of Several Variables 189
4.15.8 Text Type Tables 193
4.16 The Command LEVELSETS 195
4.17 The Command CRACKS 197
CHAPTER 5 AUTOMATIC MESH GENERATION 199

5.2 The Routine RECTANGLE 200
5.3 The Routine PARALLELEPIPED 202
5.4 The Routine ONE-DIMENSIONAL 204
5.5 The Routine SECTOR 207
5.5.2 Principles of Operation 207
5.5.4 Node and Element Numbering 210
5.5.5 Program Size Limits 210
- iv - SYSTUS 2019 REFERENCE ANALYSIS MANUAL

© 2019 ESI Group
CONTENTS
5.5.6 Example 211

5.6 The Routine TWO-DIMENSIONAL 212
5.6.4 Generation of Interface Nodes 217
5.6.5 Generation of Macro-Elements 218
5.6.6 Generation of Singularities 220
5.6.7 Problem Size Limits 221
5.6.8 Examples 222
5.6.9 Solids of Revolution or Translation 225
5.7 The Routine THREE-DIMENSIONAL 228
5.7.4 Example 232
5.8 User-Supplied Routines 234
5.8.2 Programming Considerations 235
5.8.3 Example 236
CHAPTER 6 MODIFICATION OF DATA 237

6.2 Routines Called by DEFINITION 238
6.2.1 The Routine MODIFY 238
6.2.2 The Routine ADD 241
6.2.3 The Routine RENUMBER 243
6.2.4 The Routine TRIANGLE 244
6.2.5 The Routine RBE 244
6.2.6 The routine VERIFY 249
6.2.7 The Routine RESTART 250
6.3 Routines Called Outside DEFINITION 251
6.3.1 The Routine MODIFY 251
6.3.2 The Routine RENUMBER 261
6.3.3 The Routine CORRECT 262
6.3.4 The Routine SURFACE 263
6.3.5 The Routine ASSEMBLE 264
6.3.6 The Routine LINEARISE 266
6.3.7 The Routine SECTION 270
6.3.8 The Routine ASSOCIATE 272
6.3.9 The Routine REMESH 297
6.3.10 The Routine RETRO 309
6.3.11 The Routine MULTI 310
6.4 Programmed Meshes 311
6.4.1 Principles 311
6.4.2 Types of Element Generated 312
SYSTUS 2019 REFERENCE ANALYSIS MANUAL -v-

© 2019 ESI Group
CONTENTS
6.4.3 Specification of Trajectories 314

6.4.5 Example 317
6.5 The Routine MEDIA 319
CHAPTER 7 VERIFICATION OF DATA 321

7.2 Verification Routines 322
7.2.1 The EXTRACT Command 322
7.2.2 The CHECK Command 324
7.2.3 The LIST Command 331
7.2.4 The VERIFY Command 332
7.2.5 Extraction of the Condensed Matrix of a Substructure 333
7.3 Equilibrium 334
7.3.1 Routine EQUILIBRIUM FORCE 334
7.3.2 Routine EQUILIBRIUM STIFFNESS 335
7.3.3 Routine EQUILIBRIUM MASS 336
7.3.4 Routine EQUILIBRIUM ENERGY 337
INDEX i
- vi - SYSTUS 2019 REFERENCE ANALYSIS MANUAL

© 2019 ESI Group
Contents
CHAPTER 8 TYPES OF ANALYSIS 339

8.1.1 Features Common to Analysis of All Types of Structure 340
8.1.2 Residues 341
8.1.3 Interpretation of Computed Results 341
8.1.4 Printing of Results 342
8.2 Beams and Pin-Jointed Frames 346
8.2.2 Continuous Plane Frames 347
8.2.3 Grids 355
8.2.4 Continuous Three Dimensional Frames 363
8.2.5 Plane Pin-Jointed Frames 378
8.2.6 Three Dimensional Pin-Jointed Frames 383
8.3 Plates and Shells 389
8.3.1 Plates 389
8.3.2 Axisymmetric Shells 395
8.3.3 General Shells 402
8.4 Continuous Media 434
8.4.2 Two Dimensional Analysis 435
8.4.3 Three Dimensional Analysis 451
8.5.1 Harmonic Loading of Shells 465
8.5.2 Harmonic; Loading of Three Dimensional Solids 477
8.6 Thermal and Electromagnetic Analysis 487
8.6.1 Thermal Analysis 487
8.6.2 Electromagnetic Analyses 487
8.6.3 Coupled Analyses 488
8.7 Fluid-Structure Interaction 489
8.7.2 Harmonic Loading of Axisymmetric Shells 490
8.7.3 General Shells 495
8.7.4 Coaxial Cylinders 500
8.8 Special Elements 505
8.8.1 Element Type 9 (X9XX) 505
8.8.2 Element Defined by User Supplied Subroutine (type 8) 524
8.8.3 Special Constraint Element (type 6) 527
8.9 Shells Specially Designed for Non-Linear Analysis 533
CONTENTS
8.9.1 SHE8 Spatial Shell Option 533

8.9.2 SHED Axisymmetric Harmonic Shell Option 548
8.9.3 SHET Thermal Shell Option 555
8.10 Mixing Option (MULTI Command) 559
8.10.2 Two Dimensional Structures 560
8.10.3 Three Dimensional Structures 562
8.10.4 Routine of Transition Element Creation 565
8.10.5 Limitation on the Use of Post Processing 566
8.10.6 Example 1: Mixing Options for 2D/Beam Elements 567
8.10.7 Example 2: 3D/Space Shell Mixed Problem 569
8.11 Different Types of Media 576
8.11.1 General 576
8.11.2 Regulation-Adjustment Medium 578
8.11.3 Contacts 582
8.11.4 Trajectory Loading Medium 608
INDEX i

© 2019 ESI Group
Contents
CHAPTER 9 LINEAR STATIC SOLUTION 613

9.1 Presentation-General Syntax 613
9.2 Static Solution 615
9.2.1 General Operation 615
9.2.2 Choice of Method of Solution 616
9.2.3 Analyze Check- Rerun 617
9.3 Creation of Elementary Matrices 619
9.3.1 Standard Utilization 619
9.3.2 Utilization of Elimination Methods 621
9.5 Thermoelastic Analysis 623
9.5.1 Basis 623
9.5.2 Temperature File 624
9.5.3 Examples 626
9.6 Contacts for Linear Static Analysis 630
9.6.1 Object 630
9.6.2 Files 631
9.6.3 Data Description 632
9.6.4 Example 633
CHAPTER 10 SUBSTRUCTURING 635

10.1.1 Terminology 635
10.1.2 Example 636
10.2 Method of Analysis 638
10.3 File Management 640
10.3.1 Description of Files 640
10.3.2 Calculation Steps 642
10.4 Optimization of an Analysis 644
10.4.1 Re-Use of Substructures 644
10.4.2 Change of Axes 646
10.4.3 Re-Use of Files 647
10.4.4 Interfacing of Different Types of Structure 648
10.5 Input Data 650
10.5.1 Zero Level Substructure (superelement) 650
10.5.2 Level n Substructure 653
10.5.3 The Command CALCULATE 656
10.5.4 Automatic Link Generator (A.L.G.) 658
10.6 Procedures associated with Substructure 662
CONTENTS
10.7 Conclusions 663
CHAPTER 11 DYNAMIC 665

11.2 Eigenmodes and Eigenfrequencies - Direct Methods 668
11.2.1 Real Eigenmodes and Eigenfrequencies 668
11.2.2 Real Eigenfrequencies and Eigenmodes with Fluid-Structure Interaction 697
11.2.3 Complex Eigenfrequencies and Eigenmodes (Symmetrical Matrices) 704
11.2.4 Complex Eigenfrequencies and Eigenmodes (Non-Symmetrical Matrices) 709
11.3 Real Eigenmodes and Eigenfrequencies - Modal Synthesis 713
11.3.1 General 713
11.3.2 Description of Methods 714
11.3.3 Description of a Physical Super-Element 719
11.3.4 Description of Computation Steps 720
11.3.7 Example: Uniform Clamped-Free Beam 736
11.3.8 Experimental Super-Elements 741
11.3.9 Dynamic Automatic Link Generator (D.A.L.G.) 749
11.4 Dynamic Response - Modal Method 751
11.4.1 General 751
11.4.2 Modal Forces - Representativeness of the Modal Base 756
11.4.3 Transient Response (Modal Analysis) 767
11.4.4 Harmonic Response 783
11.4.5 Stochastic Dynamics 790
11.4.6 Seismic Response - Spectral Method 800
11.4.7 Consistant Modal Matrices Method 818
11.5 Harmonic Loads - Direct Method 823
11.5.1 General 823
11.5.3 Results 827
11.6 Sensitivity Analysis 829
11.6.1 General 829
11.6.2 Analytical Re-Analysis in the Modal Base 830
11.6.3 Complete Re-Analysis in the Modal Base 833
11.7 Utility Modules 841
11.7.1 Equivalent Modal Damping 841
11.7.2 Generation of Spectral Response 844
11.7.3 Computation of Effective Masses and Effective Inertia 852
11.7.4 Transient Dynamic 854
CHAPTER 12 TRANSIENT 861

12.1 Overview 861
12.1.2 Time-Related Scheme 862
12.1.3 Special Features 863
12.1.4 Commands 864

© 2019 ESI Group
CONTENTS
12.2 Linear Thermal Analysis 865

12.2.2 Forced Convection 866
12.2.4 Example of a Quadratic Mesh Treatment 873
12.3 Linear Elasticity 875
12.3.2 General 876
12.3.5 Calculation of Forces 885
12.4 Non-Linear Dynamic Analysis 888
12.4.2 Detail of Time-Related Scheme 891
12.4.3 Utilization of the Routine 892
12.4.4 Combination between Non-Linearities and Algorithms 913
12.4.6 HISTORY and TRANSIENT Files 920
12.5 Non-Linear Static Analysis 923
12.5.2 Implicit Algorithm 924
12.5.3 Utilization 924
12.6 Transient Shell 929
12.6.1 Mechanical 929
12.6.2 Thermal 941
12.7.2 Transient Thermo-Mechanical Analysis: the TEMPERATURE Command 944
12.7.3 Manipulation of the TRANSIENT File - The SELECT Command 949
12.8 Transfer of Physical Quantities from Mesh to Mesh 952
12.9 Element Activation/Deactivation 962
12.9.2 Activation/Deactivation Function 962
12.9.3 Input Data: Element Properties 963
12.9.4 Input Data: Definition of Initial State 964
12.9.5 Data Storage 964
12.9.6 Practical Tips 964
12.9.7 Examples 965
INDEX i

© 2019 ESI Group
Contents
CHAPTER 13 POST PROCESSING 971

13.2 Average Node Stress and Strain 977
13.2.3 Files 979
13.2.4 Method of Calculation 980
13.2.6 Example 982
13.2.7 Location of Gaussian Points for Calculation of Stress 983
13.2.8 Average for Thick Shells 988
13.2.9 Average Transient 990
13.3 Combination of Results 993
13.3.2 Linear Combinations 994
13.3.3 Quadratic and Absolute Combination 996
13.3.4 Storage on File of Combinations 997
13.3.5 Combination of Fourier Series Terms (harmonics) 999
13.3.6 Complex Combination 1001
13.4 Merging of Results 1003
13.4.2 The Command MERGE 1004
13.4.3 Merging of Posts 1006
13.4.4 Merging of Cards 1007
13.4.5 Merging of Transient Files 1008
13.5 Cross Sections 1009
13.5.2 Files 1009
13.5.3 Method of Calculation 1010
13.5.5 Input data 1011
13.5.6 Example 1017
13.6 Postprocessing Module 1018
13.6.2 Files 1019
13.6.5 Example 1029
CONTENTS
13.7 Printing of Results 1033

13.7.1 The PRINT Command 1033
13.7.2 The Command PRINT TRANSIENT or MODE 1035
13.7.3 Command PRINT CONSTRAINTS 1037
13.8 The Command REARRANGE 1039
13.9 The Command FORCE 1040
13.10 The Command STRESS 1042
13.10.1 General Case 1043
13.10.2 Stresses in Laminated Shells 1046
13.10.3 Stresses in Beams using Section Database 1047
13.11 The command SPLIT 1054
13.12 The Command MINMAX 1056
13.13 The Command SHELL 1057
13.14 Three-Dimensional 1061
13.15 Local 1065
13.15.2 Files 1065
13.15.3 Input Data 1066
13.15.4 Example 1071
13.16 A Posteriori Error Estimator 1076
13.16.1 General 1076
13.16.2 Application Fields 1077
13.16.3 Input Data 1077
13.16.4 Error Analysis 1077
CHAPTER 14 POSTPROCESSING AND GRAPHIC OUTPUT 1079

14.2 Open - Close 1081
14.3 Post-Processing File Result Conversion 1082
14.3.1 CONVERT Procedure 1082
14.3.2 Post-Processing File Conversion - Printout 1084
14.3.3 Dataset 1085
14.4 Averaging Post-Processing 1098
14.5 Visualization of Results 1101
14.6 The Command CURVE POSTPROCESSING 1112
14.6.3 Example 1115
14.7 The Command CURVE TRANSIENT 1116
14.8 The Command CURVE FUNCTION 1121
14.9 The Command CURVE SATURATION 1122
14.10 The Command CURVE TABULATION 1123
14.11 The Command CURVE AVERAGE 1124
14.12 Specific Post-Processing for Nonlinear Dynamic Analysis 1127

© 2019 ESI Group
CONTENTS
CHAPTER 15 TOPOLOGY OF ELEMENTS 1131
CHAPTER 16 SYSTEMS OF AXES 1133

16.1 Global Axes (X, Y, Z) 1133
16.2 Local Axes (x, y, z) 1134
16.2.1 One-Dimensional Elements 1134
16.2.2 Two-Dimensional Elements 1136
16.2.3 Three-Dimensional Elements 1143
CHAPTER 17 MATERIAL BEHAVIOR LAWS 1145

17.1 Notations 1145
17.2 Elasticity 1147
17.2.1 General 1147
17.2.2 Isotropic Linear Elasticity 1149
17.2.3 Anisotropic Linear Elasticity 1151
17.2.4 Non-Linear Elasticity 1154
17.3 Elastoplasticity 1159
17.3.1 Theory 1160
17.3.2 User-Programmed Criterion 1168
17.3.3 Description of Models and Input Data 1172
17.4 Viscoplasticity 1209
17.4.2 Non-Unified Elasto-Viscoplastic Models 1210
17.4.3 Unified Elasto-Viscoplastic Models (Model = 21) 1213
17.4.4 Chaboche Model 1216
17.4.5 Thermo-Elasto-Viscoplastic Behavior Models (Model = 51) 1218
17.4.6 Rigid Viscoplastic Model (Model = 61) 1219
17.4.7 Elastoviscoplastic Model Coupled with Creep Damage (Model 71) 1221
17.4.8 Creep Models Applicable to Tube Elements 1223
17.5 Viscoelasticity 1226
17.5.2 Generalized MAXWELL Model (Linear Viscoelasticity) (Model = 11) 1227
17.5.3 Thermo-Viscoelastic Model (Model = 12) 1229
17.5.4 Non-Linear Viscoelasticity Model (Model = 13) 1234
17.6 Hyperelasticity 1236
17.6.1 Theory 1236
17.7 User-Programmed Behavior Law 1240
17.7.1 Principle 1240
17.7.2 Programming and Interpretation of Arguments 1241
17.8 Large Displacements and Large Strains 1247
17.8.1 Spatial Beam and Shell Options 1247
17.8.2 Two-Dimensional and Three-Dimensional Options 1249
17.9 Transient File Force Contents 1251

© 2019 ESI Group
CONTENTS
17.9.2 General Structure of the Force File 1252

17.9.3 Spatial SHELL - Spatial BEAM Option 1254
17.9.4 Two-Dimensional Option 1262
17.9.5 Three-Dimensional Option 1272
17.9.6 COQ8 Three-Dimensional Shell Option 1282
17.9.7 COQD Axisymmetrical Harmonic Shell Option 1283
17.9.8 COQT Thermal Shell Option 1284
17.10 References 1285
CHAPTER 18 COMMON BLOCKS 1289

18.1 Blank Common 1289
18.2 Common COMAIL 1293
18.3 Common COMMAT and COMMIL 1294
18.4 Common COMTRA 1304
CHAPTER 19 RIK'S METHOD 1305

19.1 General 1305
19.2 Constant Arc-Length Incrementation 1306
19.3 Variable Arc-Length Incrementation: Arc-Length Updated at Each Increment 1309
19.4 Variable Arc-Length Incrementation associated with a Buckling Calculation 1310
19.5 Note 1311
CHAPTER 20 SECTION MANAGEMENT DATABASE 1313

20.1 Simple Database Section 1314
20.2 Standard Database Section 1315
CHAPTER 21 LINEAR EQUATION SOLVERS 1319

21.2 Characteristics 1321
21.2.1 Standard Solvers 1321
21.2.2 External Solvers 1323
21.2.3 Summary Table of Solvers Properties 1323
21.3 Input Data 1324
21.3.1 Type 1324
21.3.2 Properties 1324
21.3.3 Parameters 1325
21.3.4 Extract Labels for the Treatment 1326
21.4 Notes 1327
CHAPTER 22 SPACE TRANSFORMATIONS 1329

22.2 Infinite Unidirectional Transformation (type 1) 1331
22.3 Ring or Sphere Transformation (Type 2) 1332
22.4 Square Ring or Square Sphere Transformation (Type 4) 1333
22.5 User Transformation (Type 8 or 9) 1334
22.6 Post-Processing 1335

© 2019 ESI Group
CONTENTS
CHAPTER 23 DYNAMIC ELEMENTS OF ROTORS 1337

23.1 Generalities - Modelization 1337
23.2 Complex Eigenfrequencies 1338
23.3 Harmonic Stationnary Response 1339
23.4 Time History Excitation Response 1340
23.5 Example 1341
INDEX i
SYSTUS 2019 REFERENCE ANALYSIS MANUAL -v-

© 2019 ESI Group
INDEX
EXPORT 97
INDEX EXTRACT
FILE
322, 333
300
FRACTURE 300
FUNCTION 299
(VOL.1) GEOMETRY 118
GOTO 54
GROUP 42
A HISTORY 64
HYDROSTATIC 299
ASCII 72 IMPORT 97
Automatic Number Generation 29 INCREMENTATION 301
INPUT 83
INPUT/OUTPUT 80
B LEVEL 299
LEVELSETS 195
Batch Mode 32 LINEARISE 266
Boundary Conditions 19 LIST 331
LOADS 146
MASS 169
C MASTER 293
MATERIAL PROPERTIES 132
Command MEDIA 136, 319
$FORMAT 123 MISES 299
ADD 241 MODE 63
ASEMBLE 264 MODIFY 251
ASSIGN 56 MULTI 310
ASSOCIATE 273, 280, 290-292, 295 NODES 214
AXES 122, 238 OFFSET 259
CHECK 324 ONE-DIMENSIONAL 204
CONSTRAINTS 138 OPTION 114, 301
CONTACTOR 283, 290 OUTPUT 80
COORDINATES 120, 238 PAGING 63
COORDINATES CARTESIAN 120 PARALLELEPIPED 202
COORDINATES CYLINDRICAL 120 POINTS 54
COORDINATES POLAR 120 RBE 293
COORDINATES SPHERICAL 121 READ 57
CORRECT 239, 262 RECTANGLE 200
Crack 300 RELATION 161
CRACK 260 RELATION ELIMINATION 164
CRACKS 197 RELEASE 158
CRITERION 274, 281, 290-291, 295 REMESH 298
DAMPING 172 RENUMBER 243, 261
DEASSIGN 57 RESTART 244, 250
DEFAULT 134 RETRO 309
DEFINITION 108, 238 RETURN 108
DEFINITION RESTART 110 REWIND 58
DELETE 77 SAVE 72
EDGE 215 SEARCH 75
ELEMENTS 125, 215 SECTION 270
END 62 SECTOR 207
EQUILIBRIUM ENERGY 337 SINGULARITY 215
EQUILIBRIUM FORCE 334 SLAVE 293
EQUILIBRIUM MASS 336 SLIDING 281, 290
EQUILIBRIUM STIFFNESS 335 SOURCE 274, 290-291
ERROR 54 STOP 58
SYSTUS 2019 REFERENCE ANALYSIS MANUAL -i-

© 2019 ESI Group
INDEX
SUPPRESS 216 Stiffnesses 70

SURA 296 Transient 69
SURB 296 TRANSIENT 94
SURFACE 263 File Management 4
TABLE 178
TARGET 274, 283, 290, 292
TEMPERATURE 175, 299
G
THERMAL 295
TIME 62 Generation of Sequences of Nodes and
TRIANGLE 244 Elements 128
VELOCITY 177
VERIFY 249, 332 I
Concentrated Loads 150
Consistent Masses 170 Interactive Mode 32
Coordinate System 38, 120 Interface Nodes 217
Coordinate Systems 19
Creation of a Group 42 L
D Label
ALPHA 208, 225
Definition of Nodes 119 AXES 208
CARTESIAN 120
Degrees of Freedom 18
CLOSED 225
Discretization 18 COORD 43
Distributed Loads 152 CYLINDRICAL 120
DAMPING 337
E ELEMENT 43
FORCE 274
Elastic Constraints at Nodes 143 GENERATE 204, 208
GROUP 329
Elastic Constraints over Elements 145 LAST 300
LENGHT 208
F LENGTH 225
LOWER 43
File MEDIUM 261
Direct Access 68, 183 MIDDLE 221
Displacements-Reactions-Forces-Nodal NODE 43
Stresses-Strains 69 POLAR 120
Eigenvectors 70 PRECISION 44
History 70 PROGRESSION 208
Input Data 69 RELATIVE 154
INTERFACE 84 SPHERICAL 120
Labels 78 STEP 225
Local 68 UNTIL 155
Mass-Buckling-Damping 70 UPPER 43
Matrix 70 Labels and Numbers 36
Media 69 Loads 20
Modal 70 Loads Applied to Beams 154
Permanent 68, 71
Postprocessing 70 Loads Applied to Media 157
Random Access 68
RESULT 94 M
Sequential 68, 182
Sequential Stress File 69 Macro Element 218

© 2019 ESI Group
INDEX
Modify Three-Dimensional 228

Coincident nodes 253 TIED Elements 291
Constraints 252 Types of Analysis 115
Coordinates 252
Isolated nodes 254
Offset 259 U
Orientation 254
Vectors 252 Units 20
N V
Node and Element Numbering Systems 20 Vectorial Functions of Several Variables 189
O
Operating System Commands 64
P
PAMSYS 101
Periodic Structures Subjected to Fourier
Decomposed Loads 117
Postprocessing 4
Prefixes 36
Preprocessing 4
Programmed Meshes
Specification of Trajectories 314
Types of Element Generated 312
R
RBE Elements 292
REFERENTIEL
DEGRES 38
GRADES 38
RADIANS 38
TANGENTE 38
S
Simplified Fortran 185
Singularities 220
Slide Plane 277
Substructuring 6
T
Thermal Analysis in Linear Transient 171
Thermal Analysis in Non-Linear Transient 174
Thermoelastic Analysis 116

© 2019 ESI Group
INDEX
SHELL 565
INDEX SOLID
SOLVE HARMONIC
565
464
Computed Results
(VOL.2) Axisymmetric or Translation Shell Option
Beams Option
401
377
Grids Option 361
Plane Truss Option 382
A Plate Option 394
SHED axisymmetric harmonic shell option 553
Applied Loads Shell Axisymmetric Fluid Harmonic Option 493
Axisymmetric or Translation Shell Option 399 Shell Fluid Option 498
Beams Option 373 Shell Harmonic Option 475
Grids Option 359 Shell Option 431
Plane Beam Option 351 Shell SHE8 Option 545
Plane Truss Option 380 SHET Thermal Shell Option 558
Plate Option 393 Three Dimensional Harmonic Option 485
SHED Axisymmetric Harmonic Shell Option 551 Three Dimensional Option 462
Shell Axisymmetric Fluid Harmonic Option 492 Truss Option 388
Shell Fluid Option 498 Two Dimensional Option 450
Shell Harmonic Option 473 Constraints
Shell Option 422 Axisymmetric or Translation Shell Option 398
Shell SHE8 Option 541 Beams Option 371
SHET Thermal Shell Option 557 Grids Option 358
Three Dimensional Harmonic Option 483 Plane Beam Option 350
Three Dimensional Option 459 Plane Truss Option 379
Truss Option 386 Plate Option 392
Two Dimensional Option 447 SHED Axisymmetric Harmonic Shell Option 550
Axisymmetric of Translation Shell Option Shell Axisymmetric Fluid Harmonic Option 492
Applied Loads 399 Shell Fluid Option 497
Computed Results 401 Shell Harmonic Option 472
Constraints 398 Shell Option 421
Material Properties 396 Shell SHE8 Option 538
SHET Thermal Shell Option 557
B Three Dimensional Harmonic Option 482
Three Dimensional Option 458
Truss Option 384
Beam/Solid Transition Element 510 Two Dimensional Option 446
Beams Option 363 Contacts 582
Applied Loads 373
Computed Results 377 Continuous Three Dimensional Frames 363
Constraints 350, 371 Coupled Analyses 488
Geometry of Section 366 Criterion
List of Elements 364 Hill 417
Material Properties 365 Tsai 418
C E
Coaxial Cylinders 500 Eccentricity
Command Shell option 411
BEAM 565 Electromagnetic Analyses 487
COMPATIBILITY 565
DEFINITION MEDIUM 609
MATERIAL PROPERTIES 515
MIXING 565

© 2019 ESI Group
INDEX
F SC3D 601
SL2D 592, 601
SPHERE 597
Family TORUS 597
2003-2004 405
2006-2008 407
Material Properties
2203-2204 405 Axisymmetric or Translation Shell Option 396
2403-2404 405, 407 Beams Option 365
Grids Option 356
Fluid-Structure Interaction 489 Plane Beam Option 348
Plane Truss Option 379
G Plate Option 391
SHED Axisymmetric Harmonic Shell Option 549
Generalized Contact 597 Shell Axisymmetric Fluid Harmonic Option 491
Grids Option 355 Shell Fluid Option 497
Applied Loads 359 Shell Harmonic Option 467
Computed Results 361 Shell Option 410
Constraints 358 Shell SHE8 Option 535
Material Properties 356 SHET Thermal Shell Option 556
Three Dimensional Harmonic Option 479
I Truss Option 384
Two Dimensional Option 440
Interpretation of Computed Results 341 Mixing Option 559
L O
Label Option
FAST 584 Plane Beam 347
MULTI 560 Plane Truss 378
PENALTY 584 Plate 389
List of Elements SHE8 Spatial Shell 533
Beams Option 364 SHED Axisymmetric Harmonic Shell 548
Plane Truss Option 379 Shell 402
SHED Axisymmetric Harmonic Shell Option 549 Shell Axisymmetric Fluid Harmonic 490
Shell Axisymmetric Fluid Harmonic Option 491 Shell Fluid 495
Shell Fluid Option 496 Shell Harmonic 465
Shell Option Option 403 SHET Thermal Shell Option 555
Shell SHE8 Option 535 Three Dimensional 451
SHET Thermal Shell Option 556 Three Dimensional Harmonic 477
Three Dimensional Harmonic Option 479 Two Dimensional 435
Truss Option 384
Two Dimensional Option 438
P
Plane Beam Option
M Applied Loads 351
Constraints 350
Macro Elements Material Properties 348
CONE 597 Plane Truss Option
CYLINDER 597 Applied Loads 380
GAP 583 Computed Results 382
INTERFACE 587 Constraints 379
NIT3D 605 List of Elements 379
PLANE 589, 597 Material Properties 379
RECTANGLE 597

© 2019 ESI Group
INDEX
Plate Option Thermo-Elastic Properties of Anisotropic

Applied Loads 393 Shell Elements 412
Computed Results 394 Thermo-Elastic Properties of Isotropic Shell
Constraints 392 Elements 412
Material Properties 391 Shell SHE8 Option
Printing of Results 342 Applied Loads 541
Default Element Type Codes 345 Computed Results 545
Displacements 342 Constraints 538
Forces 343 Material Properties 535
Reactions 342 Shell/Solid Transition Element 509
Special Case of Beams 343 Shells specially Designed for Non-Linear
Type pf Structure (Number) 345
Analysis 533
SHET Thermal Shell Option
R Applied Loads 557
Computed Results 558
Regulation-Adjustment Medium 578 Constraints 557
Residues 341 List of Elements 556
Material Properties 556
S Special Constraint Element 527
Special Contact Element 513
Sandwich Materials 420 Special Elements 505
SHED Axisymmetric Harmonic Shell
Option T
Applied Loads 551
Computed Results 553 Tabulated Element 506
Constraints 550 Thermal Analysis 487
Three Dimensional Harmonic Option
Shell Axisymmetric Fluid Harmonic Option Applied Loads 483
Applied Loads 492 Computed Results 485
Computed Results 493 Constraints 482
Constraints 492 List of Elements 479
List of Elements 491 Material Properties 479
Three Dimensional Option
Shell Fluid Option Applied Loads 459
Applied Loads 498 Computed Results 462
Computed Results 498 Constraints 458
Constraints 497 List of Elements 454
List of Elements 496 Material Properties 454
Shell Harmonic Option Reduced Integration Rule for an Element 457
Applied Loads 473 Trajectory Loading Medium 608
Computed results 475
Constraints 472
Transition Element 511
Material Properties 467 Truss Option 383
Applied Loads 386
Shell Option
Applied Loads 422
Constraints 384
List of Elements 384
Constraints 421
Definition of Pressure 425
Excentricity 411 Two Dimensional Option
Geometrical Properties 410 170X Elements 439
List of Elements 403 Applied Loads 447
Material Properties 410 Computed Results 450
Constraints 446

© 2019 ESI Group
INDEX
List of Elements 438

Overall Axisymmetric Behavior 437
Overall Plane Strain Behavior 437
Reduced Integration Rule for an Element 445

© 2019 ESI Group
INDEX
INITIAL
INDEX CONDITION
S 776, 869, 879, 903, 937, 941
INITIAL CONDITIONS RESTART 905
(VOL.3) INTEGRATION 793
LEVEL CROSSING 795
LIMITS 677
A LINK 659, 749
LOAD 632, 735, 759
MAXIMA 795
Automatic Link Generator 658
METHOD 845, 900
MODAL ANALYSIS 770
C MODE 674, 677, 680, 682-683, 687,
698, 705, 786, 805, 826
Combination between Non-Linearities and MODIFICATION 830
Algorithms 913 MODIFY 727
Command MOMENT 793
ACCELEROGRAM 845 PARTICIPATION 758, 761
ACCELEROGRAM TRANSIENT 846 PARTICIPATION MULTI 762
ALGORITHM 775, 894, 930 PRECISION 898
ALGORITME IMPLICIT 868 PRINT 779, 942
ASSEMBLE DYNAMIC 731, 734 PRINT COMPLETE 938
AXES 831 PRINT VALUES 676
BANDWITH 795 PROJECT 953
BEHAVIOUR 900, 929, 941 PSD 791
CALCULATE 632, 656, 792-794 PSEUDO MODAL 833
CALCULATE DYNAMIC 728 REGIME 941
CARD 780, 792-793, 950 RELATION 883
CHECK 659 RELEASE 659
CIRCULAR FREQUENCY 786, 826 RESIDUAL TERMS 762
COMBINATION 778 RESPONSE 825, 831
COMBINE 805 RESULTS 779
COMPLEX MODES 710 RUNEND 911
COMPONENTS 726 SELECT 676, 771, 779, 785, 834, 842, 949
CONNECTION 658, 727, 749 SENSITIVITY 830
CONTACT 632 SEQUENCE 691
DAMPING754, 772, 785, 806, 825, 842, 845, 910 SIL 910
DEFECT TRANSIENT 937 SOLVE 613, 616, 668, 877
DIMENSION 941 SOLVE BUCKLING 695
DIMENSION MEMORY 929 SOLVE FORCE TRANSIENT 885
DIRECTION 806, 814 SOLVE FORCES MODAL 689, 707
DISPLACEMENT 813 SOLVE HARMONIC 622
DURATION 806 SPECTRAL ANALYSIS 805
EFFECTIVE MASS 852 SPECTRAL ANALYSIS MULTIPLE 812
ELEMENT 710, 834 SPECTRUM 806, 813
FATIGUE 795 STATIC 812, 933
FREQUENCY 786, 826 STIFFNESS 727
FREQUENCY INITIAL 845 STOCHASTIC 794
FRONT 699 STOP 772
FUNCTION 771, 779, 785, 791, 878, 903 STRAIN GENERALISED 902
GAP 632 SUBSTRUCTURE 653
GENERATION 845 SUBSTRUCTURE DYNAMIC 725-726
HARMONIC RESPONSE MODAL 785 SUMMATION 695
SUPER ELEMENT 650
SUPERELEMENT DYNAMIC 725

© 2019 ESI Group
INDEX
SUPPRESS 676
TABLE 711, 772, 779, 786, 792, 807, 834, 878, G
903
TEMPERATURE 624, 944 Generation of Spectral Response 844
TEMPERATURE TRANSIENT 944
TEMPERATURE TRANSIENT SHELL 929, 947 H
TIME 777, 941
TITLE 771, 779 Harmonic Loads - Direct Method 823
TRANSFER 786
TRANSIENT 864, 877-878
Harmonic Response 783
TRANSIENT NON LINEAR 892, 921
UNIDIRECTIONAL 773 L
VECTOR 680, 687, 699, 706, 711, 835
Complex Eigenfrequencies and Eigenmodes Label
(Non-Symmetrical Matrices 709 ABSOLUTE 814, 898
ACCELERATION 692
Complex Eigenfrequencies and Eigenmodes ALL 710
(Symmetrical Matrices) 704 AMPLITUDE 855
Computation of Effective Masses and ARCHIVATE 892
Effective Inertia 852 BEAMS 692
Coupled Friction 917 BERGAN 895
BFGS 894
Cylindrical Stop 773 BINARY 921
BLOCK 687, 730
D BUCKLING 674
CARD 624, 855, 869, 879, 904, 956
Decoupled Friction 917 CARDS 885
Detail of Time-Related Scheme 891 CIRCULAR FREQUENCY 677, 680, 687
COMPACT 870, 880, 906
DYNAMIC 687, 698, 793-794, 812, 825
COMPLEX 885
Dynamic Responses 665 CONDENSATION 698
CONSTRAINT REMOVE 905
E CONTACT 896
DATA 954
Eigenfrequencies and Eigenmodes 665 DEFORMED CARD 953
Enumeration of Eigenvalues (Sturm DESACTIVATE 910
DIAGONAL 675, 682, 687
Sequences) 694 DIRECTION 761
Equivalent Modal Damping 841 DISPLA 770
DISPLACEMENT 899
F DT tshift 944
DTLAST 906
File DURATION dt 905
FMDAT.DAT 895 EDIT 791
HISTORY 920 ENERGY 899
MASD 723 EQUIVALENT STRESSES 692
RESU 722 EULER 901
RIGD 723 EXTRACT 729, 885, 893, 905
SPECnf.DAT 848 FACTOR 680
TRANSIENT 920 FILE 732
VECD 723 FILTER 885
FLUID 771, 779
FLUID STRUCTURE 698
FORCE 706, 759, 770, 898
FORCES 675
FREQUENCY 677, 680, 684, 687

© 2019 ESI Group
INDEX
GENERATE 624 RIGID 759, 770

GIVENS 674 RIKS 895, 904
GRAD 894 SAVE 895
GROUP 895 SCALING 867, 877, 892
GUYAN 683 SEARCH 885, 895
GYROSCOPIC 710 SELECT 954
HYPERELASTIC 901 SEMI-QUADRATIC 814
INITIAL 870, 906 SHELL 692
INVERSE POWER 674 SHELL THERMAL 941
ITERATION 680, 687, 706 SHIFT 674, 683, 687, 730
ITERATIONS 730, 894 SIZE 675, 682
ITERATIVE 680 SKIP 870, 880, 906
ITMAXIMAL 894 STATIC 892
JAUMAN 902 STEP 895
JOURNAL 892 STRESS 898, 953
LARDIS 899 STURM 685, 688
LARGE 899 SUPER ELEMENT 926
LAST 869, 879, 904, 956 SUSPACE 674
LINEAR 895 TOLERANCE 899
LINEARIZE 867 TOTLAGRANGE 901
LINK 732 UNIT 898
LOG 791 UPDLAGRANGE 901-902
LOWER 677 UPPER 677
LSEARCH 896 VARIABLE 855
MAXF 899 VECTOR 729
MEITERATION 895 VERIFY 892
METHOD 955 VISCOELASTIC 901
MODULUS 711 WATCH 910
MULTI 759, 770 Linear Elasticity 875
NOCHECK 729 Linear Thermal Analysis 865
NON LINEAR 892
NORM 675, 699, 899
OPTIMIZE 894 M
OTHER 730
PARAMETER EPSILON 956 Material Stop 774
PARAMETER EXTERIOR 955 Method
PARAMETER INTERIOR 956 GIVENS 676
PENALTY 896 INVERSE POWER 677
PERTURBATION 894 ITERATIVE SUBSPACE 679
PHASE 855 Lanczos subspace 682
PLASTIC 901 Lanczos SUBSPACE Method with BLOCK
PRECISION 680, 688, 730 Option 686
PREDICTION 895 Lanczos SUBSPACE Method with GUYAN
PRINT 675, 699, 729, 732 Option 683
PSD 791 Modal Forces - Representativeness of the
PULSATION 684
RATIO 710 Modal Base 756
REAL 885 Modal Method 751
REBUILT 892 Modal Synthesis 713
REDUCTION 684 Modification of the Mass Matrix 671
REFORM 894 Modification of the Stiffness Matrix 671
RELAXATION 894
REMESH 954
Mono-Spectral analysis 803
RESIDUAL 730 Multi-Spectral Analysis 811
RESIDUE 886
RESTART 869-870, 879-880, 904, 906

© 2019 ESI Group
INDEX
N Z
Non-linear Dynamic Analysis 888 Zero Level Substructure 650
Non-Linear Static Analysis 923
O
Option
Axisymmetrical Harmonic 802
Fluid-Structure 802
P
Participation Factors 756
Plane Frictional Stop 774
R
Real Eigenfrequencies and Eigenmodes with
Fluid-Structure Interaction 697
Response Power Spectral Density 790
S
Seismic Response - Spectral Method 800
Sensitivity Analysis 666, 829
Simultaneous Modification of Stiffness and
Mass Matrices 671
Stochastic Dynamics 790
Substructure Modes 714
Substructuring 925
T
Thermoelastic Analysis 623
Transfer Functions 783
Transient Dynamic 854
Transient Response (Modal Analysis) 767
Transient Shell 929
U
Unidirectional Frictional Stop 773
Unidirectional Stop 772
Utility Modules 666

© 2019 ESI Group
INDEX
MEDIUM 1117
INDEX MERGE
METHOD
1004
1324
MINMAX 1056
(VOL.4) MODE
NODES
1035
1013, 1025, 1117
NUMBER 1084
A OPEN 1081
POINTS 1013
POSTPROCESSING 1021
Anisotropic Linear Elasticity 1151 PRINT 1033, 1035
Average for Thick Shells 988 PRINT CONSTRAINTS 1037
PRINT POINT 1034
C REARRANGE 1039
RETURN 992, 1028, 1066
Code SECTION 1048
COMMON PAGE 1289 SELECT 1066
KDEBT 1289 SELECT ELEMENTS 1048
LCODE 1289 SET 1113
NCODE 1289 SHELL 1057
SOLVE 978
Command
SPLIT 1054
AVERAGE 981, 988
STRESS 1042
AVERAGE TRANSIENT 991
STRESS SECTION 1048
AVERAGING POST-PROCESSING 1099
THREE-DIMENSIONAL 1062-1063
AXES 1026
TRANSF 1027
CARD 991, 1116
TRANSIENT 1035
CLOSE 1081
UNITS 1048
COMBINE 994, 1028, 1063
USER 1048
COMBINE FOURIER 999
VISUALIZATION 1101
COMPONENTS 992
CORRECTION 981 Criterion
CROSS SECTION 1011 Drucker-Prager 1162, 1173
CURVE AVERAGE 1124 Stassi 1163, 1174
CURVE FUNCTION 1121 Tsai-Hill 1163, 1174
CURVE POST PROCESSING 1112 User 1164
CURVE SATURATION 1122 User-programmed 1174
CURVE TABULATION 1123 Von Mises 1161, 1172
CURVE TRANSIENT 1116 Cross Sections 1009
DAMPING GYROSCOPIC 1340
DEASSIGN33 1338 D
DISTRIBUTION 1062, 1064
EDGE 1014 Dataset 1085
ELEMENT 981, 1117
ELEMENTS 989, 991, 1014, 1024
EXTRACT CODES 1289 E
FORCE 1040
FORMAT 1113, 1117 Elasticity 1147
GENERATE 1062, 1064 Elastoplasticity 1159
GRID 1014
HEADER 1014, 1026
LINE 1014
I
LOAD 1048, 1063, 1066
LOCAL 1066 Isotropic Linear Elasticity 1149
MATERIALS 1025
MATRIX 1026

© 2019 ESI Group
INDEX
TUBE 1022
L UNBALANCE 1339
UPPER 989
Label VARIABLE 989, 1002
ALL 1040 VMISE 1023
AMPLITUDE 1001 Large Displacements and Large Strains 1247
ASYNCHRONOUS 1339
Local Axes
AXES 989
One-Dimensional Elements 1134
AXISYMMETRIC 1023
Three-Dimensional Elements 1143
BEAM 1024
Two-Dimensional Elements 1136
CARD 1035
COARSE 1066 Location of Gaussian Points for Calculation
COMPONENTS 1034 of Stress 983
COMPOSITE 1059
CRITERION HILL 1059 M
DATASET 1084, 1100
DELTA 991 Merging of Cards 1007
DISPLACEMENT 991
ENERGY 1040 Merging of Posts 1006
EXTRACT 989, 1046 Merging of Results 1003
EXTRAPOLATION 1099 Merging of Transient Files 1008
FINE 1066 Model
FORCE COMPOSITE 1024 Chaboche 1216
GYROSCOPIC 1338-1339 Generalized Maxwell Hardening (Linear
HILL 1060 Viscoelasticity 1227
INPLA 1023 Model=1, Perfect Plasticity 1172
INPUT 1100 Model=10, Ductile Damage Model with
INREV 1023 Isotropic Strain Hardening - Leblond
INTEGRAL 991 – Perrin Model 1195
INTRI 1023 Model=11, Plasticity with combined Strain-
LAST 991, 1035
Hardening - isotropic strain
LAYER 989, 1046, 1059
hardening and PRAGER kinematic
LOWER 989
strain hardening 1184
MASS 1040
Model=12, Plasticity with combined strain
MISE 1022
MODE 1035 hardening and RAMBERG-OSGOOD
MULTIPLY 1022 strain hardening law 1184
NORM 1119 Model=13, Ductile Damage Model with
ONLY 989 Combined Isotropic and Kinematic
OUTPUT 1100 Strain Hardening - Leblond – Perrin
PART 1100 Model 1196
PHASE 1001 Model=14 Gurson - Tvergaard Model 1197
PRESSURE 1099 Model=2, Plasticity with Isotropic Strain-
PRINCIPAL 1022 Hardening 1175
PRINT 1084 Model=2, Plasticity with Prager Kinematic
SHELL 989 Strain-Hardening 1179
SHELL COMPOSITE 989 Model=21 Global Perfect Elastoplasticity
SPATIAL 1012 Model Applicable to Thin Shell
STRAIN 1046 Elements - Simplified Model 1197
STRESS 991 Model=22 Global Perfect Elastoplasticity
SUM 1118 Model Applicable to Thin Shell
THIN SHELL 1023 Elements 1198
THREE DIMENSIONAL 1022
THRESHOLD 991, 1099
TRANSF 1024
TSAI 1059

© 2019 ESI Group
INDEX
Model=23 Global Elastoplasticity Model

with Kinematic Strain Hardening, R
Applicable to Thin Shell Elements -
Simplified Model 1198 Rik's Method 1305
with Kinematic Strain Hardening, S
Applicable to Thin Shell Element 1198
Model=25 Global Elastoplasticity Model Sky Line Solvers 1321
with Isotropic Strain Hardening, Standard Sparse Solvers 1322
Applicable to Thin Shell Elements - Strain-hardening 1165
Simplified Model 1199
Stresses in Laminated Shells 1046
with Isotropic Strain Hardening,
Applicable to Thin Shell Elements 1199 T
Model=4, Plasticity with Ductile Damage -
ROUSSELIER Model 1192 Three-Dimensional 1061
Model=5, Cyclic Plasticity -Von Mises
Criterion, EISENBERG Model 1184 U
Model=6, Plasticity with Kinematic Strain-
Hardening – Ramberg – Osgood User-Programmed Criterion 1168
Strain Hardening Law 1180
Model=7, Plasticity with isotropic Strain-
V
Hardening and Sensitivity to,load
Rapidity – Johnson Cook Law 1178
Model=7, Plasticity with Kinematic Strain-
Viscoelasticity 1226
Hardening RAMBERG-OSGOOD Viscoplasticity 1209
Strain Hardening Law 1177
Model=8, Cyclic Plasticity - Chaboche-
Marquis Model 1187
Model=8, Non-Linear Kinematic Strain-
Hardening Armstrong - Frederick 1180
Model=9, Ductile Damage Model with
kinematic Strain-Hardening -
Leblond – Perrin Model 1193
Non-Linear Viscoelasticity 1234
Rigid Viscoplastic 1219
Strain Hardening 1210
Thermo-Elasto-Viscoplastic Behavior 1218
Thermo-Viscoelastic 1229
Time Hardening 1211
Unified Elasto-Viscoplastic 1213
Viscous 1211
N
Non-Linear Elasticity 1154
Non-Unified Elasto-Viscoplastic Models 1210
P
Postprocessing 1018

© 2019 ESI Group
ABOUT ESI GROUP
ESI Group is a leading innovator in Virtual Prototyping software and services. Specialist in material physics, ESI
has developed a unique proficiency in helping industrial manufacturers replace physical prototypes by virtual
prototypes, allowing them to virtually manufacture, assemble, test and pre- certify their future products.
Coupled with the latest technologies, Virtual Prototyping is now anchored in the wider concept of the Product
Performance Lifecycle™, which addresses the operational performance of a product during its entire lifecycle,
from launch to disposal. The creation of a Hybrid Twin™, leveraging simulation, physics and data analytics,
Copyright © ESI Group, 2019
enables manufacturers to deliver smarter and connected products, to predict product performance and to
anticipate maintenance needs.
ESI is a French company listed in compartment B of NYSE Euronext Paris. Present in more than 40 countries,
and addressing every major industrial sector, ESI Group employs about 1200 high-level specialists around the
world and reported annual sales of €141 million in 2016. For more information, please visit www.esi-group.com.
All PAM- and SYS- product names as well as other products belonging to ESI's portfolio are tradenames or trademarks of ESI Group, unless specifically mentioned.
All other trademarks are the property of their respective owners - Specifications are subject to change without prior notice.
www.esi-group.com

SYSTUS RefManAnalysisVol3 en PDF

Uploaded by

Copyright:

Available Formats

SYSTUS RefManAnalysisVol3 en PDF

Uploaded by

Document Information

Original Title

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

SYSTUS RefManAnalysisVol3 en PDF

Uploaded by

Copyright:

Available Formats

SYSTUS

REFERENCE ANALYSIS MANUAL

CHAPTER 9 LINEAR STATIC SOLUTION 613

CHAPTER 10 SUBSTRUCTURING 635

10.7 Conclusions 663

CHAPTER 11 DYNAMIC 665

CHAPTER 12 TRANSIENT 861

- ii - SYSTUS 2019 REFERENCE ANALYSIS MANUAL

12.2 Linear Thermal Analysis 865

SYSTUS 2019 REFERENCE ANALYSIS MANUAL - iii -

CHAPTER 9 LINEAR STATIC SOLUTION

reactions and forces (stresses and strains) with a DATA file.

The labels can be grouped in 4 categories:

o FORCES (MODAL) (TRANSIENT)

l Formation of elementary matrices

o ELIMINATION (FILE or RATIO n ) ELEMENT n FACTOR n

SYSTUS 2019 REFERENCE ANALYSIS MANUAL - 613 -

Figure 9-1: The command SOLVE

- 614 - SYSTUS 2019 REFERENCE ANALYSIS MANUAL

9.2 STATIC SOLUTION

l for each node subject to loading or constraint.

matrix and load vector K and F:

SYSTUS 2019 REFERENCE ANALYSIS MANUAL - 615 -

9.2.2 Choice of Method of Solution

Sparse Method (Default Method)

Choice of the Solver

SOLVE METHOD <label>

- 616 - SYSTUS 2019 REFERENCE ANALYSIS MANUAL

9.2.3 Analyze Check- Rerun

SOLVE (STIFFNESS) (MATRIX) (RESULT (PARTIAL))

The STIFFNESS step can be taken separately.

Rerun (Sparse Method)

Other applied loads:

Computation of Forces from Displacements

SOLVE FORCES (EXTRACT)

SYSTUS 2019 REFERENCE ANALYSIS MANUAL - 617 -

SEARCH DATA DISPLACEMENTS 1 CARD 10

Results and File

- 618 - SYSTUS 2019 REFERENCE ANALYSIS MANUAL

9.3 CREATION OF ELEMENTARY MATRICES

l MASS (file 23) : distributed and concentrated masses M

l DAMPING (file 33) : distributed and concentrated damping C

l BUCKLING (file 33) : geometrical damping Kσ

9.3.1 Standard Utilization

The DAMPING and BUCKLING labels cannot coexist.

SOLVE STIFFNESS MASS DAMPING

SYSTUS 2019 REFERENCE ANALYSIS MANUAL - 619 -

Example 2: shell elements or continuous media – displacements obtained by non-

SEARCH DATA DISPLACEMENTS 1 CARD j

- 620 - SYSTUS 2019 REFERENCE ANALYSIS MANUAL

9.3.2 Utilization of Elimination Methods

SOLVE ELIMINATION (FILE or RATIO n1) ELEMENT n2 FACTOR n3

SYSTUS 2019 REFERENCE ANALYSIS MANUAL - 621 -

9.4 HARMONIC LOADING

there are harmonics to analyze.

relevant data from the displacement file.

- 622 - SYSTUS 2019 REFERENCE ANALYSIS MANUAL

9.5 THERMOELASTIC ANALYSIS

SYSTUS 2019 REFERENCE ANALYSIS MANUAL - 623 -