FP 6 6 Update

PCL 5.0.1 PIPE STRESS and FEPipe 6.
6
FINITE ELEMENT ANALYSIS USER
DOCUMENTATION
APRIL, 2013
PAULIN RESEARCH GROUP

11211 RICHMOND AVENUE SUITE 109
HOUSTON, TEXAS 77082
Copyright Paulin Research Group Houston, Texas 1

Scope:
This document is intended to describe features of the PCL Gold 5.01 software program and how
they should be used when performing pipe stress analysis and features added to Version 6.6 of
FEPipe.
Software Version:
The Version 6.6 of FE/Pipe contains the following products:
Product Version
FE/Pipe 6.6
PCL Gold 5.0.1
FEBend 1.5
661PRO (replacing FE661) 2.0
FETee 1.2
NozzlePRO 8.6
AxiPRO 3.6
MatPRO 4.0
FE107 2.6
FESif 2.6
StressPlot 4.0

Individual program manuals delivered with Paulin Research Group Software Installations are
described below. CHM files are opened and referenced from the program they apply to or can be
started by double-clicking on the file name. PDF help documents are opened by the program under
the HELP program manus or can be opened by double clicking on the file name in the installation
folder.
PDF Manuals:
FEPipe Ver 6.6 Update Notes (R0) – This manual
2005Doc.pdf – (150pp.) High Temp Guidelines, Plasticity, MeshPRO, RTJ Joints, Thermal Flow Boundary
Conditions, Contoured Center Nozzle Gussets.
2007Doc.pdf - (160pp.) 18dof Beam Elements, FFS-1, Subsection NH, Splash, NP-FFS
AP20.pdf - (168pp.)AxpPRO2 manual – Basic Program Use Descriptions
beamer.pdf – (9pp.)Beam Editor Data File Structure
V6.6-App3.pdf – FEPipe model generation file commands
BF4111.pdf – (349pp) Bos Fluids Version 4.111 2003 Basic Facilities Users Guide
FE5.0App3.pdf – (44pp.) FEPipe V.5.0 Modgen Commands. Superceded by FE6.6App3.pdf
FE50.pdf – (125pp.)First release of comprehensive piping software.
FP4111.pdf – (1004 pp.) Extensive FEPipe Original Manual
License.pdf – Software license
MatPRO_Help.pdf – (61pp.) Original MatPRO Manual (Use for reference only)
MeshPRO_Help – (62 pp.) Original MeshPRO Help Manual – Simplified AutoCAD Modeling tools
NozzlePRO.pdf – (185 pp.)Original NozzlePRO Manual (Use for extensive reference material.)
PCL401.PDF – Replaced with FEPipe Ver 6.6 Update Notes.
Plasplot2.0 – features.pdf – (5pp.) Describes User Interface in PlasPlot 2.0.
PRGVer6.0.pdf – Replaced by Ver. 6.6 Manual
StressPLOT_help.pdf – Stressplot v.2.0 Manual (Use for Reference Material)
CHM Manuals:
661PRO_Manual_16.chm – Interactive API661 FEA Model Program Guide (v.1.6)
FE107_Help.chm – FE107 Interactive Help and Users Guide
FE45newb.chm – Interactive FEPipe Installation and Notes Guide
FE661_Manual_15.chm – Interactive API661 FEA Model Guide (v.1.5) (For Reference)
FEBEND_Manual_15.chm – Interactive Users Guide for Bend w/ Staunchion FEA Models
FETee_Manual_1.chm – Interactive Users Guide for Contoured Tee FEA Models
LoadHistory.chm – Interactive Users Guide for Load History
PipingManual_1.chm – Interactive Help and Users Guide for PCLGold
PlasPlot2.0Help.chm – Brief Description of PlasPlot Functionality
Plastex_Manual.chm – Version 1.x Highly Nonlinear FEA Model Analysis Tool
PRGDB-en.chm – PRGDatabase Introductory Documentation
StressPlot.chm – Interactive Help and Users Guide for StressPlot2013

Contents:
Page Item
2 March 2013 Software Version Numbers
3 Available PRG Software Documentation (pdf and chm files)
5 FEPipe Version 6.6 and NozzlePRO Version 8.6 New Features
9 New Feature Images and Notes
35 PCLGold Pipe Stress Software Features
52 Superelement i- and k- database
54 Superelement Branch Connection and Head FEA Models for Piping
61 Controlling Piping Output
65 Hot Sustained Stress Discussion
69 Piping Thermal Bowing Options and Analysis
73 Path Dependent Friction
85 Bends and Miters
89 Load Combinations
98 Specifics of FEPipe Piping Modeling
129 18+ degree of freedom piping elements
135 P-D Effects in piping and structures
139 Notes applicable to ASME VIII-Division 2 Post 2007 (See also p. 160)
150 New Features added in Version 6.0
181 ASME VIII-2 Fatigue Assessment of Welds
197 Fatigue Basics

FEPipe 6.6/NozzlePRO 8.6 New Features
General:
1) Undeformed view option added in 3D animator
2) Measurement tool added in 3D animator. The user can click on any two points in the model and the
software prints the straight line distance between those two model points. The user can interactively chose
these points while rotating and zooming on the model. Can be used to find model sizes and to verify
element edge lengths.
3) Nonintegral lug attachment option added to NozzlePRO Option form.
4) Automatic cumulative damage report for piping models
5) Material manager user interface update. The material manager user interface has been streamlined and
speed of data lookup significantly improved. A more detailed description of the material manager feature
changes are described below.
6) Expanded SIF and allowable options to address B31.3 2012 Code version.
7) Fatigue contributions due to seismic, wind, or waterhammer type occasional loads for PCLGold piping
analysis.
8) Interactive CAD drawing produced during PCLGold model input.
9) CAESAR II Translator updated to CAESAR II Version 6.1.
10) Brick refractory option added to STRING template head elements.
11) PRGiK spreadsheet available from PCLGold, NozzlePRO, FESIF and FE107 includes stress
calculation spreadsheet and plots allowable and mean failure stresses.
12) Improved Shell Stress Comparisons to Beam Piping Results Module – used to verify that the Code
stress combination methods are adequate for branch connections subject to multiple high loads.
13) FFS “thinned” section capability has been added to the BendPRO and NozzlePRO templates. This
permits the user to locally thin the model and apply the VIII-2 allowable stress rules to the thinned model.
14) Updated i-factor option for sustained stress indices.
15) Updated i-factor option for close proximity of flanges at branch connections.
16) Bolted flange moment boundary conditions are available for flat heads (or any head type) in the
17) 18dof pressure coupling for harmonic loading. AIV added feature to permit shaking of small bore
branch connections due to internal line pressure pulsing. Developed as a research feature. This option is
available in the software currently while testing, evaluations and recommendations for use are ongoing.
18) Long boundary condition option added for NozzlePRO cylinder-cylinder models. Boundary condition
lengths are automatically setup to remove the boundary influence from the junction.
19) Unity load factor control – a general override to increase all model mesh densities. This is a
dangerous factor to apply, but for users that run a continued sequence of analyses and always have to
adjust the NozzlePRO mesh multiplier, this single input can be used to always increase the base mesh
density for NozzlePRO runs.
20) Refractory piping load distribution option – to provide more realistic critical load conditions in
refractory lined pipe analyses. (See PRG Refractory Analysis Recommendations (R0)).
21) Updated CAESAR II Translator – Updated to Version 6.1.
22) Convert CII files from older versions to newer versions and vice-versa.
23) Added option to remove plug from attachments on mitered bends.
24) Included ii18, io18, ki18, and ko18 for 18dof beam elements.
25) Automatic SIF and k-factor adjustments for 18dof mitered bends.

26) Fundamental element descriptions, local orientations, etc. have been modified in the shell element
formulation to accommodate the implementation of the mixed-interpolated tonsorial component element
(MITC), described in Bathe,KJ,Dvorkin,EN,”A formulation of general shell elements – the use of mixed
interpolation of tensorial components.”,Int J. Numerical Methods, Eng, 1986;22:697-722. This element
will support a much improved plastic and large rotation nonlinear solution.
FEPipe/NozzlePRO:
1) Convergence estimation report added based on the high stress differences between averaged and not
averaged stresses. The Convergence report shows the maximum percentage for each load case and the
values that resulted in that difference. Often, convergence in the finite element result is considered
adequate if the difference is less than 5%. Convergence is based on the difference in the top 5% of the
stresses.
2) FLAT option for UFT weld models to compare with older program results that do not have weld
penetration contoured thickness models at branch connections.
3) Unstructured mesh plate alteration control features. (To help users mesh geometries where there is a
strong interaction between multiple plate geometries, or between plate and nozzle geometries.)
4) Mean thickness pressure-area option.
5) Option for weld shrinkage stress distribution estimate in UFT template.
6) Template debug available for more rapid diagnosis for unusual model problems.
7) Bending moment boundary conditions for parent types. (Use to simulate bolt loads or moments
variations over (RT)0.5.
NozzlePRO:
1) Non-integral lug attachment option
2) Operating pressure option. The user can now enter the operating and the design pressure. The
difference is as stated by VIII-2 – where the design pressure includes the operating pressure and any
additional pressure needed to assure that relief devices function, etc.
3) Brick models of cylinder/cylinder intersections for d/D < 0.8 can be constructed with SCLs
4) All node row SCLs – SCLs are generated for each brick nodal row in all NozzlePRO brick models.
Membrane, bending, and membrane+bending stress for each nodal row around the circumference of the
nozzle are automatically generated.
5) Reduced load estimate capability added.
6) Improved pad meshing in the vicinity of boundary conditions
7) Attached pipe estimate entries for axial, inplane and outplane. Calculates load reduction estimate
depending on the length of the attached pipe.
8) Stress Evaluation Sheet
9) More load case combinations added to rigorous fatigue analysis.
10) Default all-model mesh multiplier

AxiPRO:
1) Updated thermal transient capability
FETee:
1) Stress Evaluation Sheet (PRGiK added)
2) Improved contoured tee thickness profiles for all tee size ranges.
3) Link to MATPRO for all pipe sections and contoured tee body
661PRO:
1) Up to 2 horizontal partition plates available
PCLGOLD:
1) PRGCAD – Interactive drawing during model entry
2) Updated sustained stress index options
3) AIV Spreadsheet Update
4) Automated cumulative damage report for piping models with multiple load contributions to fatigue
damage. Seismic, wind, and other occasional loads may contribute to the fatigue evaluation.
5) Expanded UNDO capability. The number of undo steps is recorded for each spreadsheet and the user
can step back through each change to recover information that may have been lost by accident or an errant
keystroke.
6) Expanded harmonic excitation frequencies.
7) 18dof coupling of pressure at intersections to ovalization modes
8) Minimum allowed travel distance added to spring hanger design options
PRGiK:
1) Stress analysis spreadsheet added
i) Mean failure and allowable stress curves plotted with calculated stresses
ii) Various stress combination options available
iii) Cumulative damage, probability of failure and safety factors computed
2) k-factors presented in equivalent elbow basis.
3) Koves and Bildy membrane stress factors added to output reports.
4) Updated to latest version of ST-LLC 07-02.
5) Allowable and mean failure curves for Markl, B31, Hinnant/Paulin and ASME smooth bar curves.

MATPRO 4.0:
1) Single harmonized user interface
2) New "Flat" style menus and forms
3) Up to 7x faster loading
4) Can perform unfiltered list selection of all available materials
5) Faster material and property filtering
6) Revised Temperature display:
a)old display shows "hot" and "cold" values
b) new display shows actual "cold" temperature, and all temperatures used in PCL.
7) Improved Local Thin Area (LTA) grid measurements
STRESSPLOT:
1) Faster data extraction from FEPipe, NozzlePRO or generic solution files
2) Unlimited number of SCLs can be defined (was limited to 7 SCLs in version 3.5)
3) Can mix and match SCL lookup type (version 3.5 limited to one of three lookup methods).
4) Can process SCLs from multiple models in one session (version 3.5 limited to a single model)
5) New: on demand SCL plotting (version 3.5: SCL plots only available during a transient data processing
stage)
6) Unified table presentation
7) Unified plot presentation
8) Rapid filtering
9) Simultaneous processing of Stress Intensity and Equivalent Stress results (for FFS of pre 2007
equipment)
10) New MS Word compatible /compiled HTML report generator.

Version 6.6 Features, Images and Notes
The PRGiK pipe stress spreadsheet is shown below. The user can enter individual loads, individual
stresses, or a single combined stress. For each stress, the safety factor against the selected curve is
displayed along with the damage factor and the probability of failure assuming a Gaussian
distribution about the mean of the failure curves.
Once the stresses are entered they are displayed on the endurance curve and each of four different
allowable curves are displayed as shown below:

The user zooms on the plot using the zoom in button at the bottom of the form giving the user a
good view of the data points relative to failure or to the allowable curve.
There are 10 points in between each 10^x to 10^(x+1). After 10^6, the number of points are reduced
to 1 per 10^x as each curve is an almost straight, slightly sloped line when cycles are high.
To produce ovalization at branch connections due to pressure in 18dof elements check the following
box in the Engineering Details spreadsheet. For large d/D branch connections with high D/T the
user will see pressure produce ovalization at these intersection. For harmonic loadings, pressure
pulsations will cause rocking of the branch connection.
If thermodynamic properties are entered, along with the basic pipe properties, pipe displacements
and strains are computed in this AIV research tool.

AIV Excitation can be produced in local areas around branch connections or other discontinuities in
18dof piping systems. The screen below describes the excitation distribution type, the area focus
nodes around the point of interest and AIV readings, if model excitation is to be taken from
displacement, velocity, acceleration or SPL readings.
For NozzlePRO models the user can enter a default mesh density multiplier. The value entered here
becomes the “unity” mesh density. Global Mesh Default Multiplier Control – The user can enter

the default unity mesh multipler so that meshes will have default slightly higher mesh densities.
This number is recommended to be entered in the 1.1 to 1.3 range. This is for users with faster
machines, or problems that can utilize slightly denser default mesh densities.
The PCL Undo button shows the number of undo steps that are stored in memory for each
spreadsheet in the model file.
The points above show the NozzlePRO shell model points where critical stresses are generated in
the nozzle and shell.
For tank templates, the user may restrain the radial direction at the base of the tank. User’s should
run with, or without this restraint to determine the effect on the solution.
Default SCL locations are shown in the NozzlePRO brick models of cylinder-on-cylinder branch
connections. These connections may have integral or non-integral repads to envelope the
reinforcing pad solutions.

Update FFS “thinning” capability in Nozshell template
The user can enter two types of local thin options in the Nozshell template. The TAPERTHIN
option results in a bowl-type thickness reduction in the area defined by the flaw. The UNIFTHIN
option results in a uniformly thin area in the area defined by the flaw. This feature is most often
used with a standard VIII-2 Part 5 analysis, and sometimes with an FFS-1 calculation included for
additional conservatism.
Local thinned areas and stress results for this nozzle area are shown below. The colored model to
the left shows the thickness distribution. The colored nodes in the middle image show the extent of
the thinned area of interest. The thickness profiles is superimposed over the area of interest.
Displacements and stresses as they are affected by the local thinning are shown in the plot on the
right.
Similar local thin areas are available in the bend with staunchion template. Users can ask that a FFS
evaluation is performed on the nominal pipe wall, or on a tapered or uniform thinned shell section.

Convergence reports based on the difference between averaged and not averaged stresses are
produced automatically and included in the Stress Results – Notes report. Most often stresses that
exceed the given tolerance are for program generated i- and k-factor load cases and can be ignored
when the convergence check is for a user defined load case. An example convergence case is
shown below.
To find where the node number #259 is on the model the user opens the output for the job, enters
File:Keyboard mode, and then enters the keyboard commands:
label, nodes, on
draw, node, 259
To alter the size of plate structures for unstructured meshes, the distance tolerance [2] and [3] values
can be used (see below). [2] adjusts the angular size, and [3] adjusts the meridonal extent of the
plate surface. When these values are less than 1, the mesh around the plates will get closer to the
plate geometries.
Stresses for not averaged, and for Gauss not averated element nodal locations are shown in the
figure below.

NozzlePRO has included Operating and Design pressure inputs. The inputs, and help to for the user
to establish these values is given below.

Undeformed shaded images are superimposed over the deflected models when the user checks the
box next to the slider radio button pointed to by the orange arrow in the figure below.
Shell and piping models with lightly grayed undeformed shapes are shown below.
The piping model also displays comments on the plots. The user enters the comments in the input
spreadsheet, and then under Labels, or Hover in the 3D plotter, the various comments may be
selected and displayed on the plot. The intensity of the undeformed shaded image is controlled on
the options graphic panel using the “Outline Opacity” slider.

NozzlePRO provides a non-integral repad option based on tests conducted at PRG in Houston.
The No Contact option is the most conservative option. In the model below, the separation of the repad
wall from the vessel shell is shown below. This analysis is particularly helpful when relatively thin repads
are used with relatively thick shells, and are welded only at the edge of the repad, and the lugs are highly
loaded.

Photographs of the lug tests conducted in Houston are shown in the photographs below.
Type 1 Lug
Type 2 Lug
Type 3 Lug
Additional example output plots are shown in the figures below.

Stress in Lug Type 1. NozzlePRO – Separate Pad and Parent Shell Model
Stress in Lug Type 2/3. NozzlePRO – Integral Pad and Parent Shell Model

Bolted Flange Moments on Flat Heads
Capability has been added to the Nozshell template to add moment boundary conditions to the edges of flat
or other head types to simulate boltup. (This is done most often with flat heads.)
Characteristics of the model should be taken from ASME VIII-2 Table 4.6.1.
The outside edge of the flat head should be given as the dimension “d” shown in details 7 and 8 from ASME
VIII-1 Table 4.6.1 shown below. For detail 8, there is no operating or gasket seating moment applied.
The moment entered into NozzlePRO is passed to FEPipe as an edge moment per length of circumference.
For the gasket seating load this can be calculated as:
Mg = WghG / (πd)
Mo = Woho / (πd)
Mg = Gasket seating moment entered into NozzlePRO on Line 13 in place of fCrownR.

Mo = Operating load moment entered into NozzlePRO on Line 14 in place of fKnuckR
All other nomenclature are provided in the VIII-2 Para 4.6.5 Nomenclature.
This capability was provided through the LOCAL boundary condition in the Nozzles-Plates-Shells template.
The LOCAL boundary condition can can be applied to any head or cylinder type in FEPipe and provides an
interesting capability to display root(RT) effects for cylinders.

Example model:
Do=100”; T=1”; Len=30”; (RT)1/2 = (49.5)0.5 = 7.03”. The boundary condition at

the top is rigid, and at the bottom is given by the LOCAL directive.
The LOCAL boundary conditions aren’t associated with any particular direction, they are only used to apply
a local circumferential moment to the edge of the parent at the boundary end described. For the entry below
there is a 1000,000 in.lb/in. circumferential moment applied along the bottom edge of the short cylinder
length shown. The displaced shape for this applied load is shown below.
Notice how the circumferential moment alone causes the free end at the bottom of the cylinder to curl under.
If the user specifies the boundary conditon followed by a single unity value as shown below, then the
translational boundary conditons along the specified edge will be zero, but the circumferential moment load
will still be applied. The following is an example:

PCL Model Plots:
A variety of model plots are available from the PCL. Branch connection FEA model locations are shown
along with structural elements in the analysis.
PCL models of multiple reactor vessels and fea branch connections.

PCL CAD window and spreadsheet shown below. When the user is on an active spreadsheet row, the
element is highlighted in the CAD window. Similarly, the user can click on an element in the CAD window
and the cursor will go to that element in the spreadsheet.
Plot of all the stresses for all the nodes for all the load cases in a single interactive 3D plot. The user can
hover over nodes to see node number of stress value.

Multiple models can be shown in the same plot window. The user can step between each model by clicking
on the model to select it. Once the model is selected the user can interact with the chosen model. Models
can be docked together using the Navigation menu shown below:
In the PCL output processor the “settings” menu allows the user to include the shell finite element model
associated with any intersection. The button to activate the settings menu and the settings menu is below.
When shell meshes are displayed, they are drawn at the intersection where the SIFs and flexibility factors are
applied. The rightmost plot below shows this option activated. The Settings menu also contains the
“multiple” model option that permits the user to add multiple models to the same plot. When multiple
models are added to the same graph, they may be oraganized in a grid, or placed randomly on the screen.
This is also controlled from the settings menu.

Gray image no outline
Element Outline Overlays
Beams as Solid Lines
Several model styles may be displayed. The plot above show the same API 661 header box in a piping
system. Thermal bowing is applied to the header box elements along with local flexibilitities computed in the
FE661 program. A variety of color and shading options are available. The user will generally have to select
the one that is most applicable in a given situation.
The 3D model plots give all Code type stresses in a single plot. The selection menu is shown below:
A sustained stress plot with five sustained stress cases is shown below. The allowable surface is based on the
variation in the hot allowable stress in the model. The nodes overstressed can be found by putting the node
numbers on the plot, or by hovering over the nodes. These three dimensional plots can be rotated, zoomed,
clipped, etc. Expansion stresses can be plotted with sustained stresses, displaced model plots, overstressed
plots, etc. This ability to look at many different plot types at one time (on a large second monitor), makes
large model review much easier.

Sustained Stress 3D Plot Cummulative Damage (Eq. 1d)
FAT#1 in the cummulative damage plot uses the B31.3 damage equation (stress ratio exponent = 5), and the
FAT#2 in the cummulative damage plot uses the Hinnant damage equation (stress ratio exponent = 3). The
Hinnant damage equation is considered to more realistically, and conservatively represent welds made after
the 1980’s.
The min allowed travel limit has been added in the latest version (PCLGOLD 5.0). From the General
Options form select “Spring Hanger Controls”.
For backup in PCL. The user enters a number of minutes for backup saving in the Settings window and it
saves a backup with the extension *.dom.bak in a <jobanme>_old folder, if the input file is open, if the input
file is closed, then the folder name changes to <jobname>_bak. This way the PCL knows when an input file
has not closed correctly.

These can then be opened through the “Open backup file” menu item under Files. This is in case something
crashes during operations, the default for backup is 5 minutes. It saves a maximum of 10 backup files,
deleting the earliest in favor of the latest when there are ten in the list.
The undo saves an array of the entire spreadsheet in memory. This is done through a spreadsheet intrinsic
function. It returns all values and settings of the grid. These are then stored in machine memory for a
maximum of 100 changes. As the user keeps entering more data, the 100th value gets erased and everything
is moved back one. For very large models it will probably take a few seconds to load what was on the screen
before. But, the process is the same for small or large models.
Definitive directions for local WRC107 loadings are shown below.

MATPRO 4.0 - NEW FEATURES
The MatPRO user interface has been completely redone into a more modern, harmonized, flat UI.
The MatPRO MainPage, MatPRO Material Master, and MatPRO activeX views have all been
simplified and combined into a single interface.
Performance enhancements include:

* up to 7x faster loading,
* can perform unfiltered list selection of all available materials
* faster filtering
* revised temperature display:
* improved LTA grid measurements: unlimited maximum grid size.
Images from the version of MatPRO are shown below:
Launched from Start menu:
PCL multiple temperature view.

Creep Calculator Updated Interface
Updated Fitness for Service (FFS-1) Interface
STRESSPLOT
The STRESSPlot user interface has also been rewritten. New features include:
* faster data extraction
* unlimited number of SCLs can be defined
* can mix and match SCL lookup type
* can process scls from multiple models in one session
* on demand scl plotting

* unified table presentation
* unified plot presentation
* rapid filtering
* simultaneous processing of Stress Intensity and Equivalent Stress results
* New MS Word compatible /compiled html report generator.
Images:
Start Screen (with Some SCLs defined)
SCL definition screen

Linearization Plot of single SCL/Load Step
Histogram presentation of code stresses for all defined scls
Data Table View Generator

Report Generator
Updated Load History screen editor (NozzlePRO/FE107/FETee)
Right-click editing options make it easier for users to enter multiple load case descriptions.

PRG Equations used in the PRGiK spreadsheet for combination options: The equations provide
different combination approaches for the individual stress components used in pipe stress analysis
and may provide more realistic combinations of the different stress combinations. The combination
method becaomes more important when each stress component is more heavily loaded, (and in
particular pressure). The user should be aware that the lower and upper bound for the stress states
will be: Supper = Σ(|Si|), and Slower = MAX( |S1|,|S2|,…|Si|)
PRG EQ. 5.0 and and B31.3 Eq. P17
PRG EQ. 5.0x for branch

PRG EQ. 5.0x for RUN
Additional documentation is provided in the help for each program, and in some cases, most of the
user documentation is provied in a CHM formated manual linked to the program help. These
manuals can be printed from the CHM processor when activated.

PCL 4.01 Software Features
General Features Note

(12) Thermal Cases
(12) Pressure Cases
Path Dependent Friction
Automatic Seismic and Wind Load Case Development
Fatigue Analysis of Seismic and Wind Loads
Fatigue Analysis of Impact Loads such as Waterhammer
Pressure Fatigue
Incorporation of Finite Element Superelements for i-factors
and k-factors.
Superelement Stress Analysis
FEA Superelement Stresses compared to equivalent B31.1
and B31.3 calculated stresses
Superelement branch connection local membrane and
bending stresses printed for FFS or failure analysis
Cummulative Damage Calculation per B31 Eq. 1.d
Fatigue Cycle Counting per VIII-2 5.B.2 Min-Max Cycle
Counting
18+Dof Refractory Lined Pipe Analysis
B16.9 Welding Tee Industry Model Sizes
(Light/Medium/Heavy)
Welded On Fitting Dimensional Library
a)Fitting Length from MSS SP-97 Table 3
b)Body Profile from “typical” OLET fitting thickness
PRG i-factor and k-factor calculations
Breathing Mode Calculator
Automatic Wind Loads per ASCE-7
Seismic Load Development and Cases per ASCE 7
Rotating Equipment Reports Any or all combination of load cases from
API-610 multiple runs can be analyzed for
API-617 turbines, compressors or pumps.
NEMA SM23 Extensive reporting capability.
Flange Evalution per Any or all combinations of load cases
NC-3658.1 (strength) from multiple runs can be analyzed
NC-3658.3 (leakage) simultaneously. Extensive reporting
VIII-1 Appendix 2 (strength) capability.
Heater Nozzle Analysis per API 560
2-D Plots of Stress, Displacements, Loads and Restraint
History
3-D Plots of Stress Cases
Effect of pressure, torsion and axial loads
Ten i-factors and Eight k-factors for Branch Connection.
are properly evaluated.
Graphics Capture and Management Tool for Report Writing
B31.1 and B31.3 Piping Code
Multiple Instances of Input and Ouput

Thermal Expansion of Structural Members
Thermal bowing of horizontal, vertical and skewed
structural members
Thermal bowing of horizontal, vertical and skewed pipe
Structural input in same spreadsheet as pipe input

Finite Element i-factor and k-factor library
Automatic nodal fatigue case resolution per B31 Eq. 1d
User labeling for elements, restraints, displacements,
forces, and SIFs.
Software determines when integration
point principle stresses exceed tensile
Refractory Failure Analysis minimum or compressive maximum and
remove them fro the stiffness solution for
the pipe.
Rules of Thumb Checklist for Piping Systems program
Allowable Compressive Stress Calculator Combined Load Collapse Analysis
“All Ranges”, “Sequential” and “Unique” Code case
combination options for fatigue.
P-Delta Analysis Effects for One Load Case
i-factor, k-factor library for welding tees, unreinforced
fabricated tees and welded-on outlet fittings.
Default OLET Lengths based on MSS-SP97 Table 3
Extensive Intersection Model Detail
a)Stress Concentration Factor for Welds
b)Local Branch Barrel Thickness
c)Local Branch Barrel Length
d)B16.9 Fitting Length (Internal Database)
e)B16.9 Actual Run OD
f)Spherical/Elliptical/Dished/Conical or Flat Head
i)Diameter
ii) Thickness
iii) Crown Radius
iv) Cone Length
v)Ellipse Ratio
vi) Knuckle Radius
vii)Head Attached Shell Length
g)Cylinder Attached Lengths

h)Contour tee body thickness contours (3)
The user can bring the PCL model up in
NozzlePRO (providing a license of
NozzlePRO is available), and can modify
the model in either NozzlePRO or
Superelement Link from PCL to NozzlePRO FE/Pipe and get the updated results used
in the superelement calculation. (The i-
factor and k-factor database is not
modified. It is protected from user
modification.)
Save As and Open as New Instance
Valve and Flange Database and option to add “half

measures”
CAESAR Importer and Exporter
FEPipe Exporter and Importer
Beam Analysis Piping Example

Input is demonstrated for the following simple example problem:
Enter the jobname: S1502 and select a New Job for FE/Pipe and then use the Beam Models
template:

The following piping/structural input screen will appear:
B31.1 and B31.3 Material Properties can be obtained from MatPRO.
Select 3-Elements: . For the initial input, the user is prompted for the element type,
i.e. pipe, user-defined, or an AISC structural shapes. (For this example, double-click on the
highlighted option.)

The inputs required to construct the geometry in the figure above is shown in the screen shots
below:
Screen #1

Screen #2
Screen #3
Screen #4
Screen #5
Screen #6
The node plot of this model is shown below. Nodes are applied by using the VIEW option in the
window menu or by using the LABEL option in 3D views. Nodes with decimal values are placed at
points around the bend on the centerline of the pipe. For example, the node 35 location is input at
the top bend tangent intersection point. This is a fictitious point in space. For displacements, forces
and stresses, the actual physical points on the piping system are recorded as 35.1 at the near
weldline, 35.2 at the middle, and 35.3 at the far weldline as shown in the graphic below. To apply

restraints or loads on bends, the Bend Extra Data Page should be used to place up to 6 points at
any angle on the bend curvature. Restraints or loads can be applied to these points.
Use VIEWING to see nodes on static plots and Layers on 3D Viewer Plots
When the user presses the TextOUT button, the data in all forms is listed in a semicolon delimited
file which can be imported to Excel, or edited in place. Any data can be edited in this way.
Anchors at 5 and 55 are defined as restraints as shown in the portion of the RESTRAINT form
below.
The +Y support at node 40 is entered as shown below:

The Expansion joint is defined on the screen:
Transverse and rotational stiffnesses are calculated using the axial stiffness, effective diameter and
flexible element length.
The vessel intersection at node 50 is entered as shown below:
With the SIF/Stiffness set to PROGRAM (the default), the user can control how the intersection is
modeled from the GENERAL form. The user can either use the standard Appendix D
intersection model or can use advanced results from an FEA analysis of the tee. For this simple
piping system, the overall stiffness is controlled by the expansion joint, and so the default
standard intersection model is used. For systems where the flexibility of intersections may have
an effect, the user can always just change the default global intersection model from standard to
advanced on the general form (see below), and then compare the results.

Users should not be surprised if stresses in some part of the model go up when FEA flexibilities are
included in the analysis. Advanced models more accurately distribute forces and moments in the
system. For indeterminate systems, when one nozzle is unloaded, the loads on another nozzle may
increase.
When the advanced option is selected, FE/Pipe will check the on-line PRG database and the
user’s local database for the tees in the model. If found, the SIFs and flexibilities will be taken from
the database and inserted into the beam model. If not found, an FEA analysis will be performed and
the SIFs and flexibilities inserted into the database for later retrieval. In this way, FEA models only
have to be analyzed a single time. The online PRG database available to all users current on their
SUA contains over 100,000 models.
When the advanced option is selected, if tees are NOT in any current database, then an FEA
analysis will be automatically performed. A small window will appear in the upper right corner of
the screen indicating that an FEA run is being performed. The total number of tees being analyzed
is also displayed. These FEA calculations may take several minutes or an hour or two depending on
the number of tees in the model. It is important to realize that this time is only required the first
occasion the model is analyzed. For each subsequent run, the SIFs and flexibilities are stored in the
various databases and are retrieved instantly.
For a typical B31 Appendix D intersection (and a 12x36 inch intersection would be modeled by
many piping analysts as an Appendix D intersection), the 12” pipe would be run from the face of
flange shown in the above isometric to the centerline of intersection, with the junction point being
rigid. There are two other ways to evaluate this junction. Each of the three methods is shown
below.
The “rigid-to-surface” model incorporates the fact that the added flexibility provided by the nozzle
pipe connecting the surface to the centerline does not exist. The “rigid-to-surface plus flexibility”

model is considered the most accurate branch model because it includes the fact that there is
negligible flexibility between the centerline of the intersection and the surface of the run pipe, and
the fact that there is a local flexibility at the surface of the nozzle and the run pipe.
Either of these models can be automatically selected by the user. The “rigid-to-surface” model
without flexibility is selected using the RIGID option on the STRAIGHT ELEMENT EXTRA
DATA form:
The “rigid-to-surface plus flexibility model is selected using the FEA option on the STRAIGHT
ELEMENT EXTRA DATA form:
The automatic FEA model also includes run-side flexibilities as described in more detail elsewhere
in this User Guide. Large openings in headers may reduce the strength of the header, increase the
run flexibility and reduce natural frequencies. These changes may have little effect on
displacements and stresses or may affect stresses and displacements by ten times or more. The
degree of effect is determined by the system layout, d/D, D/T and d/t ratios, temperature and
support.
There are four green buttons that control job execution:
Plot – spins off the geometry generation and plot routine and generates listings that should help the
user check the geometry. E-Plot, F-Prepare, G-Submit and I-Submit and Wait. The “checking”
reports generated when any of the green buttons are pressed are:

Input Process – Input listing
Job Title – Text from Title sheet
Input Data Echo – Screen images (not recommended)
Beam Calculated Properties – Echoes of data look ups, SIFs and flexibilities used, etc.
Load Case Report – The load cases the software assembles based on the user’s data input.
The input listing should be checked carefully. The majority of errors in a piping analysis occur
because of errant input.
The Viewing option, (shown below) can be used to draw piping related entities and assist the user
when verifying the model. The “bump” option rotates the model slightly so that objects and node
numbers that appear one-behind-the-other can be seen.
Once the input has been checked and verified, the or

buttons should be pressed to perform the analysis.

When the analysis is completed, the button shows up on the
bottom of the main menu. When the reports button is selected, the following menu appears:
can be used to generate tables and listings of input and/or output, and
brings up the FEA graphical model generator which can be used to view results, or to call the FEA
animation graphic display program.
For Reports, the following are recommended:
The load case report should be reviewed so that the user knows how the load cases were
established. There is a special section on path dependent solutions. The user will likely not
recognize much of a difference in path-dependent and non-path dependent solutions since
displacement and stress output formats do not change, but when friction is a significant part of the
solution, the results from a non path-dependent solution may not be conservative. A typical load
case report (shortened) is shown below.
Load Case Report
This solution is NONLINEAR, and PATH DEPENDANT. Except

for algebraic load cases, all solution cases reflect
an incremental change from the previous case. Path
dependent solutions are considered more accurate than
path independent solutions.
THE 7 LOAD CASES ANALYZED ARE:
1 Weight ONLY
Weight ONLY case …

It is assumed that the weight case accurately

represents the installed stress state.
/-------- Loads in Case 1

Forces due to Weight
2 Sustained
The sustained case is used to satisfy the B31

requirement SL < Sh for piping.

Pressure Case 1
3 Operating
Operating case run to compute the extreme operating

stress state to be used in the shakedown and peak
stress calculations per ASME Section VIII Div 2.

Pressure Case 1
Temperature Case 1
Force Case (Operating)
4 Expansion (Fatigue Calc Performed)
Expansion case used to find the B31 piping stress range.
/-------- Combinations in Expansion Case 4

Plus Stress Results from CASE 3
Minus Stress Results from CASE 1
5 Weight ONLY

6 Expansion (Fatigue Calc Performed)
This case is also used to find the B31 piping stress

range. For this range it is the difference between an
operating case and the return to the weight case.
/-------- Combinations in Expansion Case 6

Plus Stress Results from CASE 3
Minus Stress Results from CASE 5
7 Sustained
Sustained case …
This is a NON-Iterative sustained case run to
determine the hot, or operating case sustained
stresses per B31.3 Appendix S.

Pressure Case 1
For graphical output, the button should be selected. The static viewer will be
displayed where the user can show displacements, stresses and animated results. Two viewers are
available, a static, model construction viewer, and a DirectX animation viewer. The viewers are
designed to be used in a variety of different ways. User’s are encouraged to experiment with each.
The two different viewing panels are shown below:

Each viewer has a method to query the graphic stress. For the static viewer, whenever a colored
plot is shown, the user can access the “pointer” using Controls:Find Contour Value, as shown above
on the left. For the interactive viewer, the “thermometer” tool can be rolled over any colored
image with the left mouse button depressed and the value under the cursor will be displayed in the
bottom left of the screen:
The B31 Results provide output on a case-by-case basis. The calculated stresses are plotted along
with the ratio of the stress over the allowable: Ratio = (Code Stress) / (B31 Code Allowable). The
stresses are calculated according to either B31.1 or B31.3 as the user selects.

The report formats are shown below.
Pipe Restraint Loads
Load Case 1 of 5 Solution Case ID # 1
Node FX FY FZ MX MY MZ
40.0 0. -2626. 0. 0. 0. 0.
5.0 209. -218. 0. 0. 0. -730.
55.0 -209. -1256. 0. 0. 0. -801.
10.0 15.0 -209. 0. 0. 0. 0. 398.
Load Case 2 of 5 Solution Case ID # 2
40.0 0. -2626. 0. 0. 0. 0.
5.0 209. -218. 0. 0. 0. -730.
55.0 -209. -1256. 0. 0. 0. -801.
10.0 15.0 -209. 43804. 43804. 0. 0. 398.
Beam Element Forces & Moments 13:43:16 7 Nov 2008
NODES FX FY FZ MX MY MZ
Load Case 3
5.0 1378. 1313. 0. 0. 0. -4491.

10.0 -1378. -1239. 0. 0. 0. 6405.
10.0 66. 1239. 0. 0. 0. 700.

15.0 -66. -1181. 0. 0. 0. 711.
10.0 -42492. 0. 0. 0. 0. 0.
15.0 42492. 0. 0. 0. 0. 0.

Pipe Properties 13:43:16 7 Nov 2008
Element Output Property Data
From To OD Thk Sect Sc Sh Mat Etype
5.0 10.0 12.750 0.375 43.8 20000. 20000. 1 41

15.0 20.1 12.750 0.375 43.8 20000. 20000. 1 41
20.1 20.2 12.750 0.375 43.8 20000. 20000. 1 42
20.2 20.3 12.750 0.375 43.8 20000. 20000. 1 42
20.3 25.0 12.750 0.375 43.8 20000. 20000. 1 41
25.0 30.0 23.325 3.750 981.9 20000. 20000. 1 41
30.0 35.1 12.750 0.375 43.8 20000. 20000. 1 41
Straight Pipe Stress Intensification Factors
Node i(ax) i(in) i(out) i(tor) i(pr)
45.0 1.00 1.00 1.00 1.00 1.00

50.0 1.00 6.30 6.30 6.30 1.00
50.0 1.00 6.30 6.30 6.30 1.00
55.0 1.00 1.00 1.00 1.00 1.00
Bend Stress and Flexibility Factors
From To i(ax) i(in) i(out) i(tor) i(pr) Ki Ko
20.1 20.2 1.00 2.66 2.22 1.00 0.00 8.826 8.826

20.2 20.3 1.00 2.66 2.22 1.00 0.00 8.826 8.826
35.1 35.2 1.00 2.66 2.22 1.00 0.00 8.826 8.826
35.2 35.3 1.00 2.66 2.22 1.00 0.00 8.826 8.826
Pipe Displacements
Displacement Solution
Load Case: 1 (Rotations in Degrees)
X Y Z RX RY RZ
5.0 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

10.0 0.0000 -0.0002 0.0000 0.0000 0.0000 -0.0009
10.0 0.0000 -0.0002 0.0000 0.0000 0.0000 -0.0009

15.0 0.0000 -0.0321 0.0000 0.0000 0.0000 -0.0009
10.0 0.0000 -0.0002 0.0000 0.0000 0.0000 -0.0009

15.0 0.0000 -0.0321 0.0000 0.0000 0.0000 -0.0009
10.0 0.0000 -0.0002 0.0000 0.0000 0.0000 -0.0009

15.0 0.0000 -0.0321 0.0000 0.0000 0.0000 -0.0009
Beam Element Unbalanced Loads
Load Case: 1
From To FX FY FZ FX/L FY/L FZ/L L
5.0 10.0 0. 74. 0. 0. 4. 0. 18.00

10.0 15.0 0. 58. 0. 0. 58. 0. 1.00
15.0 20.1 0. 21. 0. 0. 4. 0. 5.00
20.1 20.2 0. 58. 0. 0. 4. 0. 14.14
20.2 20.3 0. 58. 0. 0. 4. 0. 14.14
Pipe B31 Code Compliance
CASE 3 Load Type: OPERATING
NODE Axial Sin Sout Stor SL Scode Sall

5.0 20. -1230. 0. 0. 1230. 1250. 51062
10.0 20. 1754. 0. 0. 1754. 1774. 51062
15.0 20. -2141. 0. 0. 2141. 2161. 51062

20.1 20. 2274. 0. 0. 2274. 2294. 51062
20.1 20. 6058. 0. 0. 6058. 6079. 51062

20.2 20. -6493. 0. 0. 6493. 6514. 51062
20.2 20. 6493. 0. 0. 6493. 6514. 51062

20.3 20. -5774. 0. 0. 5774. 5795. 51062
Pipe B31 Overstressed Nodes
CASE 4 Load Type: EXPANSION
The user is reminded that output can be controlled from the GENERAL input screen:
. The Output Control Text button opens the <name>.cmm file

which accepts commands and parameters to print ranges of reports, elements or nodes for the
different beam-type reports. To omit printing of load cases 1, and 5-7, the user would enter:
/XCASE
1
5, 7
The report options are compiled below, and are printed in the initial <name>.cmm file to help user’s
remember the commands that can be used:
# Command Options:
# /NOHEADER /STOPHERE /USEOPTION /OPTION /CASE
# /XCASE /USEFILE
# Report Options:
# /DISP /STRESS /FORCE /PROP /DETAILED /RESTR
# Exclusion/Inclusion Options:
# /NODES /XNODES /ELEM /XELEM
# See documentation for more information.

Automatic FEA Modeling of Branch Connections, Tees and Nozzles in Heads
The automated FEA Modeling of Branch Connections, Tees and Nozzles in Heads can work one of
four ways:
FEA Auto-Modeling Method #1 – The most common method because most users are connected to
the internet, and many tee, head and branch connection geometries already reside in the PRG FEA
Database

FEA Auto-Modeling Method #2
FEA Auto-Modeling Method #3

Automated Beam Intersection Models
(Branch-to-Run, Nozzle-to-Vessel, or Nozzle-to-Head)
There are two types of intersection models:
1) Cylinder-to-cylinder models
2) Cylinder-to-head models
A typical piping analysis uses a flexibility coefficient only for bends. The considerable flexibility that can
exist at intersections is typically ignored and so loads are errantly distributed in the piping system as a result.
These flexibility load redistributions can drop the stresses in one area of the model while increasing it in
others. When the diameter-to-thickness ratios are > 50, thermal expansion stresses are greater than 50% of
the allowable, and there are pipe-to-pipe or pipe-to-vessel intersections in the model, including the effect of
flexibilities in the analysis may have a significant effect on the calculated equipment loads and Code stresses.
The FE/Pipe automated modeler takes the standard pipe intersection model, (shown below), and
automatically adjusts it to include both header and branch SIFs and FLEXIBILITIES.
The automated intersection modeler user has the option of activating updated SIFs alone, FLEXIBILITIES
alone, or SIFs and FLEXIBILITIES together. By default, standard models (without flexibilities or updated
SIFs) are used, but by changing the Intersection Model Options on the General Data screen (shown below)
from “STANDARD” to “ADVANCED”, the more advanced FEA models are used for intersections in
HEADS and CYLINDERS (branches in run pipes). The user may also select the type of model to be used on
the General Data data screen. In the example below, both STIFFNESSES and SIFs are used for the
ADVANCED intersection models.
The typical intersection Model
Considerable additional control is found on the individual intersection data form. The basic transformation
provided by the automatic intersection modeler is shown in the figure below:

The automatic conversion can be made for intersections, flat, elliptical, spherical, dished and conical heads.
Examples are given below. The are some small modeling requirements for offset branch head geometries,
but these are described below.
For intersections, the user should code to the center of the branch and header intersections and enter a single
node at that point. Rigid elements to the surface, etc. should NOT be entered, as the program adjusts the
local model of the intersection in both the header and nozzle directions to include the effect of the opening on
both the header and the branch stiffnesses.
Typical Conical Head Model
The 2nd element attached to the intersection node must be rigid and must define a line that is
perpendicular to the axis of the vessel, and from the nozzle penetration with the shell to the axis of the
vessel. When this rigid element does not exist, the nozzle and the vessel axis should be collinear, and in
this case it will be assumed that the nozzle is a top nozzle in the head.

Flat Head Model Typical Elliptical Head Model
Note that in the models above the offset is defined by an element that is perpendicular to the vessel centerline
and whose length describes the nozzle offset from the centerline.
Typical Centerline Model
When the nozzle and vessel centerlines are collinear there should be no “offset” element.
Inplane and outplane nozzles for orientations of nozzles on heads.
The models shown above are complete models. The input can also be truncated if the user is only interested
in using the SIF/flex FEA model for a single node flexibility, SIF or superelement stress calculation. The
single elements for the head and cylinder are shown in the sketch below:

The same simplification can be used with cylinder-to-cylinder intersections. Only the nozzle or
branch side of the intersection is entered and an additional node is setup in the internal model as
shown above. This capability is used most often when the user wants to stop the model at a vessel
nozzle. In this case the user is recommended to include the local inplane vector orientation so that
the program knows how to orient the inplane and outplane flexibilities and i-factors. A typical
simple nozzle-only vessel tower and nozzle model is shown below.

If the user does not enter the inplane direction when only the single branch pipe is entered, then the
PCL will use the default out-of-plane orientation vector for the element. The default out-of-plane
vector is found by crossing the element axis into the Y direction. If the element axis is in the Y
direction then the out-of-plane vector is taken along the X axis.
Beam Element Output Control
When performing piping analyses, users most often find themselves performing a large number of
runs while varying only a few parameters each run. In most cases, there are a few areas of the
system where results are reviewed, i.e. at pump nozzles, near intersections, etc. It is tedious in these
cases to look through all of the results when only a few numbers are needed.
In some cases, only results for certain supports are needed. In other cases displacements at springs,
or around expansion joints are needed.
The beam element scripting control was developed to help the user in these situations. Using the
scripting control, the user has the option to:
1) Limit the number of load cases output.

2) Limit output to ranges of elements or nodes.
3) Limit output based on a particular report type.
4) Setup various “scenarios” so that the user may select a particular “scenario” to satisfy a particular
need, for example, the support designer only needs loads on certain supports. The rotating
equipment engineer only needs loads on equipment nozzles, and the pipe stress engineer, only needs
results in areas removed from rigid equipment models, steel, vessels, furnace tubes, etc.
As an example, a particular user may only be interested in the operating load on a pump nozzle. If
the operating load case is 3, the user can eliminate all other load cases by entering:
/XCASE
12
4 10
The load cases used for each case can be found in the available after
the piping model is plotted.
The script command file is accessed from the GENERAL form as shown below.

The most common commands limit the load cases printed as shown above and remove the headers
from selected beam reports. (The command to remove the headers is /NOHEADER.)
When the user clicks on the button, notepad should open the output
control text file: <jobname>.CMM as shown below:
To remove the headers and limit the load cases, the user would enter the following text and save the
file:
The notes section can be deleted or allowed to remain. Leaving the notes is convenient, as the notes
section serves as a reminder of the commands that are available.
To provide the limits above for only the elements in the node range between 50 and 75, and
between 125 and 180, the user would enter:

This will result in very readable, compact reports.
The user can also define different ranges for different reports. For example, to include all nodes
between 1 and 100 for restraints, but all nodes for 1000 to 2000 for stresses and forces/moments, the
following inputs could be used:
The second /NODES command replaces the first /NODES command and is used with the /FORCES
reports.
The user can also define multiple output options. For example, say that in one situation, the node
ranges shown in the above example should be used, but for the full report, all nodes should be
printed. The user can establish multiple criteria and options as shown below. The
/USEOPTION “n” command is used to identify which criteria should be used to control printing for
the current run.

Note how arbitrary text can be placed after identified text. There are no formatting requirements,
and most commands can be abbreviated by using only enough letters to identify the command.
All available commands are given in the table below:
Command List1 Parameter Notes

/NOTE NO Start note section. All lines following the /NOTE line will be
considered comments until another / command is encountered
/NOHEADER NO Turn off headers in beam output reports
/OPTION NO Option Command placed to identify block of commands with a unique Option
ID ID. When a /USEOPTION <n> command is encountered, control
jumps immediately to the /OPTION <n> command and continues until
the end of file is encountered, a /STOPHERE is encountered, or
another /OPTION command is encountered. See example above.
/USEOPTION NO Option Used to jump to /OPTION <option ID> command and begin execution
ID of output control scripts from this point. See the example above. The
/USEOPTION line must precede the /OPTION card with the Option ID
to be used.
/USEFILE NO Filename Control jumps to <filename> and begins executing. This option is used
for multiple jobs that have the same node/element references.
/STOPHERE NO Command used to stop continued execution of the script. Can appear
anywhere in the file.
/HEADER NO Used to reactivate printing of headers for beam reports
/XNODES YES A list of single nodes or node ranges to be excluded from the printed
option range of nodes should follow. For element reports like
forces/moments, displacements and stresses, any single node can be in
the exclusion list, and the entire element will be excluded.
/NODES YES A list of single nodes or node ranges to be included in the printed
option range.
/XELEM YES A list of elements that should be excluded from the output list.
/ELEM YES A list of elements that should be included in the element output list.
/NELEM YES A list of element numbers that should be included in the element
output. The user may find the element numbers for the current job in
the Beam Calculated Properties Report
/XNELEM YES A list of element numbers that should be EXCLUDED from the
element output.
/XCASE YES A list of cases, or range of cases that should be excluded from the load
case output list.
/CASE YES A list of cases or range of cases that should be included in the load case
output list. Included cases are checked for the list of excluded cases.
/STRESS NO Exclusion or inclusion commands executed prior to this command will

be used to determine the elements printed in the beam stress output.
/DISP NO Exclusion or inclusion commands executed prior to this command will
be used to determine the elements printed in the beam displacement
output report.
/RESTR NO Nodal exclusion or inclusion commands executed prior to this
command will be used to determine the nodes printed in the beam
restraint report.

/DETAILED NO Exclusion or inclusion commands executed prior to this command will
be used to determine the elements printed in the detailed beam output.
The detailed beam output is used most often with the 18+dof element
because of the additional information printed for each element in the
report.
/FORCES NO Exclusion or inclusion commands executed prior to this command will
be used to determine the elements printed in the beam forces and
moments report.
/PROPERTIES NO Exclusion or inclusion commands executed prior to this command will
be used to determine the elements printed in the beam properties report.
# Any line starting with a # sign is considered a comment line and is
skipped. Comment lines may appear anywhere in the file. The
comment lines can be used to remove items from a list. The list
reading will continue after the comment lines are skipped.
1
When YES appears in the List column, a list should follow the command that describes individual
nodes, groups of nodes, elements or load cases.
Each line preceded by an “#” is treated as a comment, and any number of comments can follow a
/NOTE entry until the next /<command> is found.
Element Print Logic

The element print validation starts by looking for “element number” inclusion or exclusion lists:
(/NELEM or /XNELEM). If an element number inclusion list is found and the element is not in
it, then the element will be excluded from printing. If an element exclusion list is found then the
element may be excluded from printing. If there is no inclusion element number list, and the
element is not excluded from printing, then checking continues, i.e. (/ELEMENT). If the element is
in an element inclusion list, it is printed. If there is an element inclusion list, (i.e. /ELEM) and the
element is not in it, then it is not printed. If the element is in the exclusion list, it is NOT printed.
Next the node inclusion list is checked. If both element nodes are NOT in the node inclusion list,
then the element is not printed. If there is NOT a node inclusion list, then the node exclusion list is
checked. If either of the element nodes are in the node exclusion list, the element is not printed.
Node Print Logic

The node print validation starts by checking for a node inclusion list. If the node is not in the
inclusion list, the node is NOT printed. If there is no inclusion list, then the exclusion list is
checked. If the node is in the exclusion list it is NOT printed.
General Command Notes:
There may be a total of 200 lines per command list, and up to fifty different commands per used
OPTION section.
There may be multiple /USEOPTION commands if necessary. These are convenient for the
different reasons output might be printed. One option, might be used for looking at equipment
loads. Another might be for printing a full report for a client. Another might be for just checking
wind load cases.

All nodes, elements and load cases are printed by default. The user may limit them on a per report
basis. The reports that may be uniquely identified are: /STRESS, /DISP, /RESTR, /DETAILED,
/FORCES, /PROPERTIES. If the user wants one set of node ranges limited for the stress report,
and another limited for the restraint report, but no limits on any of the other reports then the CMM
file might appear:
/NODE
1 200
300 350
/STRESS
/NODE
45 800
/RESTR
/NODE
1 100000
/DISP
/DETAILED
/FORCES
/PROPERTIES
Any command with a list can be set to null by repeating the list name without any data. The above
example could also be repeated as shown below. There is a node list for the STRESS report. All
nodes are printed for the RESTR report, and nodes from 1 to 100000 are printed for the DISP,
DETAILED BEAM, FORCES AND PROPERTIES reports.
/NODE
1 200
300 350
/STRESS
/NODE
/RESTR
/NODE
1 100000
/DISP
/DETAILED
/FORCES
/PROPERTIES
Any command that follows any other command will reset that command. For example, any /NODE
which follows any previous /NODE resets the /NODE list. When a /USEOPTION command is
encountered, the program immediately skips all lines in the datafile until a /OPTION command is
read that identifies the OPTION to be used.
The user should experiment with output control as the inclusion and exclusion commands can be
used in a variety of ways.
FE/Pipe generates extra node numbers for bends and automatic intersection models. The extra node
numbers are identified as decimal points following the standard node number. For example, if the
element 5-to-10 goes to a bend tangent intersection point, the nodes, 10.1, 10.2, and 10.3 will be on
the bend. These decimal designations can be used in the node or element descriptions.

ASME Code overstressed reports will NOT report overstressed areas in parts of the model or load
cases that are excluded from analysis. User’s should be careful when running analyses with
restricted analyses and should always make at least one run when the design is completed to be sure
that loads have not been shifted to other parts of the system that have not been included in the
output reports.
These report controls were designed for beam element models and so only work on the following
reports generally applicable to the beam models:
1) Restraint report
2) Beam Displacement Report
3) Detailed Beam Report (Complete stress state on 6 and 18+ dof beams.)
4) Beam Force/Moment Report
5) Beam element stresses
5) Beam element unbalanced Loads
6) Pipe Properties
8) Pipe B31 Code Compliance
9) Pipe B31 Overstressed Nodes.
By default, when a piping model is run, the ASME tabular stress results are not generated. The
ASME tabular stress results generally provide information about shell element models and are listed
below:
If a shell element model is included, and the user wishes to include these results, a “1” should be
placed on the stress option line in the GENERAL form as shown below:
Piping: Loads applied to restraints are assumed to be due to initial displacements at the base of the
supports and will be subtracted when the restraint report is printed.

Hot Sustained Cases:
The hot sustained case has been discussed in B31.3 Appendix S and in various papers, one of which
is given below:
B31.3 gives the following objectives:
321.1.1(g) addresses “unintentional” disengagement of piping. B31.3 Appendix S Paragraph S302

addresses the potential “intentional” disengagement of piping. S302.6.2 states, “All anticipated
sustained conditions utilizing all possible support scenarios should be considered and either
evaluated or “Approved By Inspection.”
The hot sustained case is run automatically when the FE/Pipe pipe solution processor is used. (The
default processor when beams are used in FE/Pipe 5.0.) There are two user options:
1) The user can turn off the hot sustained case.

2) The user can change the load case that should be used to set the restraints in the hot sustained
case. If the user doesn’t select the case, the program will use the operating case. The user might
selected another case when an occasional case may produce a more critical support condition.
Remember that for high temperature (creep) concerns, the critical hot sustained case support load
should reflect the maximum primary loads PLUS the longest acting primary loads.
These inputs are on the GENERAL form:
The automatically generated hot sustained load case is placed after the operating and first
occasional cases are run. It is identified as the hot sustained load case. Note that if the highest SL
is found in the hot sustained case, then this SL will be used to reduce the expansion case stress
allowable.
In the above geometry, if the initial displacement at node 5 in the vertical direction causes the
support at node 10 to lift off, the operating bending stress at node 5 is due only to the weight of the
cantilever. In the weight only case, the support at node 10 does not lift off and the bending moment
at 5 is reduced.
B31.3 Appendix S discusses this condition and leaves it to the discretion of the analyst.
To be conservative, a hot sustained case can be run where by the weight bending moment at node 5
in the operating case is used to perform a sustained analysis.
This approach is summarized:
1) In the cold condition the support at node 10 does not lift off and the moment at node 5 is some
value less than the operating moment. The stress due to this moment should be less than Sh.
2) In the operating condition the support at node 10 lifts off and the moment at node 5 increases
over the cold condition. The stress due to this moment should also be less than Sh since it is due to
the primary load of weight.
It is clear, however, that any vertical load at the support 10 is due to primary loads in the system,
whether the loads are developed in the operating case, in a weight case, or in a “hot” sustained
case. (This presumes there is no friction at the node 10 support.) As node 5 begins to displace
thermally in the vertical direction, the load at the node 10 support reduces. For the simple system
above, the highest weight moment exists just prior to the liftoff at node 10. Just prior to liftoff,

primary loads are acting to cause the moments at node 5. These moments will not contribute to
collapse of the system since if a plastic hinge forms at node 5 the load at node 10 is increased. This
lack of collapse exists if there is small liftoff at node 10 also. If the limiting criteria in the Code
(SL<Sh) is to prevent collapse by assuring the accuracy of the calculation by limiting primary
displacements, then the above arguments are clear, and it is the analysts responsibility to select
when a reasonable limit is reached. (See B31.3 Appendix S.) If the criteria is based on the limits of
the accuracy of the analysis, the decision is less clear. In this case, the code limit is established to
maintain the displacements within a range where the evaluation of the displacements due to primary
loads is accurate.
There are two ways to develop the weight moments in the operating cases for more complex
systems:
Method 1) Using the support configuration in the operating case run a weight case, and use the
resulting moments in a sustained stress analysis.
Method 2) Take the moments from the operating case and subtract the moments from a thermal
only case. The resulting moments should be used for the sustained stress analysis.
Either of these methods will produce the same hot sustained stress when applied to the above
geometry when liftoff occurs. Both methods will not produce the same hot sustained stress in the
above simple geometry when liftoff does not occur. Reviewing a more general case:
Method 1): Ko dhs = Fw; Ko=operating stiffness matrix, dhs=hot sustained case displacements,
Fw=forces due to weight.
Method 2): Ko do = Ft + Fw: operating case

Kt dt = Ft: thermal only case
do – dt = displacements due to hot sustained.
The method 1 and method 2 displacements must be the same for the methods to produce the same
hot sustained stresses. These two methods are only the same if:
dhs = do-dt
or:
Ko-1 Fs = Ko-1(Ft+Fw) – Kt-1Ft
As can be seen, dhs = do-dt only when Ko = Kt which is often not true as may be seen in the
example below:

It is not clear which method produces the most conservative method at every point in an arbitrary
piping system, although it is clear that in certain situations stresses from each method will be
different.
When gaps exist on vertical supports in the positive direction and these gaps are closed in the
operating case used for the hot sustained case analysis, they will be removed for the non-iterative
primary only case.
Displacements from the hot sustained case should not be used.
A third method is also used, where any gap that exists in the operating case is applied in the hot
sustained case.

Thermal Bowing
Load distributions due to thermal bowing are shown above. Thermal bowing loads will show a
change in the support loads and system deflections. Users should consider adding additional nodes
at span midpoints to detect larger deflections due to thermal bowing.
Thermal bowing can be due to radiation, (furnace, or sun), rain, or separation of the pipe flow. In
either situation the user must determine the distribution of the temperature and the maximum
temperature difference from one side of the pipe to the other. With FE/Pipe the user describes
which parts of the system should be exposed to thermal bowing, and the direction of the thermal
gradient.

Expansion Joints:
To describe a straight pipe element as a singe expansion joint that may optionally be tied, hinged, or
gimbaled the user would enter the form and enter the FROM
and TO node numbers for the expansion joint. The “inplane direction vector” can be used to
describe the axis of the hinge pin. (The vector is in the direction of the hinge pin.) In the
Ki,Ko,Kt,Kax text cell, three values for the expansion joint may be entered. The last value is
optional.
To identify the element as an expansion joint the “Special Element Type” combo-box in the bottom
right of the screen should be set to EXPANSION_JOINT. (Other options include cut short and cut
long cold spring.)
The first entry, “5000” in the example above is the axial stiffness of the expansion joint. The length
of the expansion joint and its axial stiffness are all that’s needed to determine the rotational and
translational transverse spring stiffnesses for the expansion joint.
The second entry, “24” in the example above is the effective diameter of the expansion joint. The
effective diameter is measured to the half-height of the bellows, i.e. the effective diameter is:
Effective Diameter = (Pipe ID) + (2)(convolution half height)
The torsional polar moment of inertia is found by multiplying the effective bending moment of
inertia by 100. Users should always check torsional loads on expansion joints.
The third entry, “4” in the example above is the type of constraint applied to the expansion joint.
The available constraint options are described below:
1-Hinged expansion joint: Axial deflection is restrained and only rotation about the hinge axis is
permitted. There is no friction added to the EJ rotation, and so user’s concerned about large
frictional loads developed due to the restraint of axial displacement should increase the stiffness of
the EJ to compensate, or add equal and opposite moments of an appropriate magnitude to simulate
the effect of the friction. In practice it is often better to increase the EJ stiffness (guess), and then
recalculate the increased moment to see if the assumed increase was sufficient. An iterative
algorithm to evaluate the frictional loads is planned.

If a hinge axis is not specified with a type 1 restrained expansion joint a typical axis will be
assumed. For a non-vertical EJ element axis, the hinge axis is vertical. For a vertical EJ element
axis the default hinge axis is in the X direction.
2- Gimbaled expansion joint: Axial deflection is restrained and rotation about transverse axis is
permitted. See the discussion of friction under the type 1 joint above.
3-Tied expansion joint. This is a simple tied model whereby a single rigid (1e12) restraint is placed
along the axis of the expansion joint. There is no restraint of relative rotational displacement at the
expansion joint ends.
4-Tied expansion joint with rotational restraint. This is a more comprehensive tied expansion joint
model where the axial displacement is restrained along with the relative transverse rotation of the
bellows ends in both transverse directions. Three or more tie-bars removed from the expansion
joint centerline will often provide this rotational restraint providing nuts or stops along the tie bars
NOT provided with gaps. If gaps are provided, the analyst should be sure the system surrounding
the tied expansion joint is sufficiently flexible so that pressure loads are not carried inadvertently by
adjacent, rigid equipment or very stiff piping.
Expansion Joint Pressure Thrust.
The effective diameter is used to multiply the elemental pressure to produce a pressure thrust at the
ends of the bellows to open the bellows if the pressure is positive and to close the bellows if the
pressure is negative. This is an approach taken by most pipe stress programs to simulate pressure
thrust. This does not represent the actual distribution of loads in the piping system however. The
true load distribution is shown below.
Pressure Thrust Distribution. On the left side of the bellows shown above is the approximation of
the expansion joint pressure thrust load. On the right side of the diagram above is the actual
distribution of the load where F1 and F2 combine to equal to the total pressure thrust load tending to
open the bellows.
The above loads show the default distribution of the opening pressure thrust loads on the expansion
joint. This default distribution is applied automatically by the program for any load case including
pressure when the expansion joint element includes pressure. To turn off the pressure thrust, the

user should zero the pressure NOT the effective diameter. The effective diameter, along with the
axial stiffness and expansion joint length is used to develop all other expansion joint stiffnesses.
The above load distribution shows the actual pressure thrust load distribution acting on the
expansion joint ends and surrounding pipe. The above two figures can be compared to evaluate the
differences that may exist. The default model (upper) may induce some compression in the pipe
downstream of the expansion joint that does not exist. The more accurate model (immediately
above) is difficult, if not impossible for pipe stress programs to develop automatically with a 100%
surety.
Expansion Joint Stiffnesses:
The translational relationship for expansion joint stiffnesses is given by the equation below.
Leff = [ (6)(Kax/Ktr) r2 ]1/2
where: Leff = the effective (flexible) length of the expansion joint

Kax = the axial expansion joint stiffness.
Ktr = the transverse expansion joint stiffness
r = the effective expansion joint diameter divided by 2.0.
The transverse stiffness calculated above is often not the lateral stiffness reported by expansion joint
manufacturers since some manufacturer data comes from tests. Tested results are a function of
boundary conditions, and depending on the selected boundary conditions, calculated and measured
values can be different by around four times.
As given in EJMA equations C-5, and C-8 through C-11, lateral and rotational stiffnesses for any
single expansion joint may be related to the axial stiffness as a function of the effective diameter of
the bellows and the flexible length of the joint. It is for this reason that FE/Pipe only requires the
axial stiffness and effective diameter of the expansion joint. Bending and transverse stiffnesses can
be calculated directly, avoiding possible boundary condition confusion.
Applying Pressure Thrust in an FE/Pipe Model
For B31 code compliance purposes pressure thrust is not strictly a weight or an operating load. For
this reason, the applied force/moment data form in FE/Pipe permits the user to identify the
Operating forces and moments specified on the form to be of pressure origin. When this is done,
the thrust force will be applied in both the SUSTAINED and OPERATING cases, but not in the
WEIGHT only case. The checkbox for this option is shown on the bottom of the
FORCE/MOMENT data form below.

Path Dependent Movement and Friction
Path dependent friction effects are illustrated in the figures below.
During startup, as the pipe heats it moves to the red position away from the nozzle. The friction supports
downstream of the vessel nozzle react to the outward sliding and push back on the vessel nozzle putting the
nozzle axially in compression. When the pipe returns to ambient during a shutdown, the direction of the
friction force changes (since friction acts against the direction of motion), and the friction load puts the
nozzle in tension, pulling outward on the vessel as shown in the blue position above. The load on the vessel
nozzle is shown in the sketch below.

The range of loads and stresses on the nozzle is given by the change from c-to-d. If the path dependent
nature of the friction loads was not properly considered the range of stress would only be calculated by the
range of loads a-b, and would be nonconservative.
Another example of path dependent friction is shown in the figure above. The top figure shows the axial line
stop protecting the rotating equipment from the friction loads that would act when the line heated up and
grew to the right, away from the pump. Path dependent friction shows that when the line cools back to
ambient, the sign of the friction forces reverse sign and pull the support away from the line stop and cause
loads on the pump!
The above two situations are relatively easy to recognize. An experienced analyst would often say that they
would always put an axial stop on both sides of the support above protecting the pump, preventing the
reversing sign of the friction load from damaging the pump. In reality, many systems are more geometrically
complex such as the configuration shown below.

In this configuration there are numerous friction restraints resisting movement in multiple directions. As the
line heats and cools the pipe will move out and back and friction forces will reverse sign pulling and pushing
the system in ways that are difficult to estimate without a calculation.
In general, the user should be concerned when the developed, collective friction force is thought to produce
excessive loads at either sensitive rotating equipment or expansion joints. In some cases friction forces may
also develop high stresses in elbows, tees or supports, but this generally only occurs in smaller diameter pipe,
that are overstressed by the incorrect evaluation of much larger diameter pipe that was influenced by friction
loads that were not properly evaluated.
In the example below wind load acts on the system in the operating condition. The displacement at the node
45 in between the gapped guide is shown in the bottom figure. In the operating case, the gap on the guide is
closed in the positive direction. When the wind acts in the opposite direction it moves the pipe to the
opposite side of the guide, closing the gap in the negative direction. In these cases, the difference between an
independently solved operating and operating + or – wind case may not be equivalent to the difference in the
path dependent solutions of the operating+wind then operating case.

In the above figure, the top diagram shows the loads acting at a single point on the system. The
bottom figure shows how the frictional support at node 45 interacts with the other frictional
supports in the vicinity. The key features to note are that the thermal load (a-b) causes the gap at
the support to close (a-b). Wind loads acting in the other direction can cause the pipe to move from
the positive side of the guide to the negative side, essentially doubling the stress due to the
occasional load. When the pipe cools down after the wind has stopped, the cold condition at “j” is
different from the original ambient condition, resulting in a larger computed stress range. In this
friction case, not following the load history produces a non-conservative (not maximum) evaluation
of the stresses and loads.
The term non-conservative, as used in the above example means that the forces, moments and
stresses are not evaluated in a way to guarantee that they are maximum values that can exist.
Non-conservatism as applied to friction as a characteristic of the numerical solution, means that the
results are dependent on the history of the loading, i.e. that the solution displacements at each step
are dependent on the solution from prior steps, i.e. on the path to get from the startup solution to the
present solution. The term non-conservative, in this case does not apply to the conservatism, or non-
conservatism of the moments or stresses calculated, but rather, to the conservation of energy. The
non-conservative nature of the solution may be seen in the fact that the stiffness matrix is not
symmetric once sliding begins and the normal force changes.
Friction also is directionally dependant. Take an example, where gravity acts in the –Y direction
and a block sits on a frictional plane described by the X and Z coordinates. If the weight of the
block is W, and the friction coefficient µ then the lateral force (in the X-Z plane) to start the block
sliding is µW. If we assume that in the first step of external loading on the block, a force of µW is
applied in the X direction, the block will begin sliding and will be reacting against the applied load,
(1st law). Once the block has begun sliding and reacting against the load to slide in the X direction,

a Z direction load of an infinitesimally small value can act that will both change the direction of
movement and not change the friction load in the X direction. Some rotating clutches utilize this
principle to transfer torque while moving essentially freely along the axis of the shaft. The
theoretical model in FE/Pipe assumes that once the sliding friction plane has become unstable along
a certain direction that movement in the plane but normal to the sliding direction can occur under
zero loading. If any component of the movement during a sliding step has any component along the
direction opposite to the initial sliding direction, then sliding stops and the non-sliding evaluation of
the condition occurs.
Friction can often be a controversial subject. There are kinetic and sliding friction coefficients, and
many analysts believe the effects of friction dissipate over time. For certain situations however
omitting friction can result in significant incorrect estimates of loads on rotating equipment,
structural attachments and nozzle connections. This is particularly true with fiberglass piping
systems where the pipe is relatively light compared to the fluid contents and expands considerably
due to pressure effects. It is believed that when the system is sensitive to friction, that key design
conditions should be analyzed both with and without friction. The checkbox shown below on the 2-
General form can be used to turn off all friction effects.
Some characteristics of a friction solution are shown below with distinctive points enumerated.
Point A shows how typical friction loads change sign in the weight case after going though an
operating cycle. After one or two operating cycles the “friction” loads tend to shake down, just like
the system stresses. The initial weight case however, may not give the full system stress range as
shown at node 5 – point C. In this case the sliding friction loads from the operating case reverse,
acting in the negative direction when the system cools down. Evaluating only the first cold-to-hot
cycle for the stress range misses 60% of the stress range evaluation at point C. (11758 - -
6874)/(11758-95) = 1.6

Friction introduces several complications into a piping solution. The first is the concept of the
resistance cone shown below. As the normal force increases, the resistance to sliding also
increases. Since the normal force frequently changes when moving from cold-to-operating, the
sliding force must change also. It is in this case where the majority of step-wise or non-path
dependent friction algorithms break down. As can be seen by the movement from “a” to “g” and
then to “h”. During the movement from “a” to “g”, a friction force equivalent to µFn will be
developed acting in the minus b direction. If a gap closes, sliding starts at another location, or the
load direction changes, movement can change from the a-g direction to the g-h direction. In this
case, the g-h movement can occur without any additionally developed friction load. This
phenomena is used in some annular clutch designs. The movement will appear to be in the a-h
direction, while the friction load will act along the a-g direction.

The second is the concept that friction tends to move opposite to the relative direction of
displacement, which is what introduces the reversal of load in many thermal cases. As can be seen
in the figure below
In this case, the box begins sliding at point one and moves to point 2, 8mm in the +X direction. The
displacement in this case is in the +X direction and the force acting on the floor due to friction is
also in the +X direction. Due either to a gap closing, another support sliding, or a change in the
load, if the box moves from point 2 to point 3, the direction of sliding changes, and the direction of
the friction force also changes, acting now in the –X direction. Note that even though the
displacement of the box at point 3 is still positive at 6mm, the friction force acts in the negative X
direction. Computer programs that do not follow the path of movement, will show the friction force
incorrectly acting in the positive X direction, because the displacement is positive, even though the
direction of movement is not. This is why the path from hot to cold often results in large friction
loads in the cold condition that did not exist in the initial cold condition before the cycle occurred,
and it is this path dependence and reverse direction of sliding that most pipe stress programs omit.
Going from a not sliding state to a sliding state requires a very small displacement, and so from the
standard restraint output report it is sometimes difficult to determine when a particular event
occurred. The Nonlinear Restraint History Report (see below) tracks each nonlinear restraint
through its path dependent history. Many of these iterations are insignificant, and so FE/Pipe prints
only the restraint state solutions at 10% load intervals.

If each change of state is desired, the user may use the solution output options to print the restraint
condition at each state change as shown below.
Alternately, the user can ask for the restraint state report at some other percentage of the restraint
load case, by entering that percent in place of the -1 above, i.e. 0,0,0,25 will cause the state of the
nonlinear restraints to print as each 25% of the load case is passed.
A typical restraint state report for each load step is shown below. In this straight system, the three
restraints shown begin sliding one after the other, and all restraints are sliding after 0.4% of the
operating load case. The default print step at 10% would show that all three restraints in the
example below were sliding at 10%.

Walking due to Friction
Where a long series of supports interacts, and where parallel systems may heat and cool alternately
systems subject to significant friction loads due either to weight or thermally developed vertical or
horizontal loads may tend to “walk”, or incrementally move relative to their initial condition.
Several illustrations are shown in the figures below.

Friction Miscellaneous:
When there are multiple supports at the same point and they all have friction, the user should be
aware that when the supports are relatively flexible, the friction stiffness may be greater than the
support friction stiffness. The default friction stiffness when not sliding is 1.e10 lb/in. The user
may increase or decrease this value, but if the structure resisting the friction sliding is also flexible,
then base nodes should be used and the local structural stiffnesses provided for the base node. This
is illustrated in the schematic below.
In this sketch, the user recognizes that the support at 20 is flexible and assigns a stiffness of 17,000
lb/in. in the vertical direction with a friction coefficient of 0.3. It is likely that the support at 20 will
have a lateral stiffness equal to or less than 17,000 lb/in. The model shown above however, will
have the default friction stiffness of 1e10 lb/in when not sliding and the normal force, sliding

adjustment applied when sliding. In reality, the support at 20 will likely deflect considerably in the
horizontal direction before the µFn is developed. To simulate this, either:
1) Model the structure.

2) Determine the structural stiffness in the horizontal direction (as well as the vertical direction),
and apply these stiffnesses at a base node that is then connected to node 20 by a rigid vertical
support. Most pipe program friction algorithms behave in this manner, and so all user’s should
be careful of this stiffness interaction effect.
If the horizontal stiffness of the support above is 15,000 lb/in. in both the X and Z directions, the
recommended model construction for this geometry is shown below.
When the user adjusts friction tolerances on the GENERAL screen, the error in the solution
typically increases. It is the user’s responsibility to be sure that load unbalances that result from
very large error tolerances do not affect the usefulness of the solution. When either solution
controls are input or the maximum convergence checkbox is “checked” the user should carefully
review restraint and elemental force loads to be sure a balance exists in the system. Generally the
maximum error will be equal to the maximum friction load developed on the support times the step
tolerance. The default step tolerance is 0.0001. This error estimate is based on experience running
the friction solution and may not be applicable for all situations. Increased error tolerances
generally are needed when sliding occurs in directions normal to one another during a sliding step,
i.e. the support starts sliding in the +X direction and then changes direction during the load step into

the +Z direction. For the typical piping or vessel system, when these instances occur, the actual
magnitude and direction of the friction sliding load is not well defined and so the error introduced
by an increase in the friction tolerance is not critical, however, when the system is very sensitive to
these loads, such as for:
1) Rotating equipment
2) Glass lined pipe
3) Refractory lined pipe
4) Development of torsion about expansion joints,
5) Creep dependent loads,
the user should apply additional caution.
Sliding Direction
In many cases the direction of the friction load is not determined by the direction of displacement.
Take the simple cantilever beam for example, and impose the initial displacement at the wall in an
operating case as shown below:

For this simple model, when the system begins to initially displace, the axial stiffness of the
cantilever will quickly develop µFn at the +Y support in the X direction, while the bending stiffness
of the cantilever, being less than the axial stiffness, will not reach µFn until much later. When
sliding begins at some initial point during the load development in the mostly X direction, there will
be no further load developed in the X direction unless the normal force changes, (which it doesn’t in
this problem). Additionally, there will be no load development for any loads in the Z direction
since the sliding force (µFn) is already developed by the system in the X direction. (The total
sliding force magnitude will never be greater than µFn, i.e. [Fx2 + Fz2]1/2 = µFn.) Since the
operating displacement is predominantly in the Z direction, the sliding will show to be in the mostly
Z direction, while the load acting against friction movement is in the –X direction. Applying
friction loads in this case based on the direction of sliding would produce loads acting along the
weak axis of the cantilever, but also about 90 degrees removed from the actual direction of the
friction forces.
Bends and Miters
Curved bends are uniformly treated in both B31.1 and B31.3. Closely spaced miters are treated like
uniformly curved bends, and so there is little modeling difference. Widely spaced mitered bends
having more than one cut should be modeled as multiple single cut miters. Even though the B31.1
and B31.3 Code suggests this modeling by the heavy lines in their respective mitered bend figures
in Appendix Ds and the notes which accompany those figures, few users follow this guidance since
coding a single bend as multiple, short bends is tedious. An example widely spaced miter is shown
below.
Single, widely spaced miters are simple to enter. The user must only remember to calculated the
effective radius of the miter cut. The spacing (s) is not defined.
The angle theta (θ) is one half of the angle between adjacent miter axes, and is calculated:
θ = β / (2Nc)
Nc is the total number of cuts in the miter bend.

β is the bend angle.

Many miters are made to accommodate long radius bends, so the user should check either the
specification or the drawings. If a radius is provided, the spacing can be calculated from the
number of cuts and the user can determine if the miter is closely or widely spaced. If the miter is
widely spaced, then to satisfy the intent of the B31.1 or B31.3 code, the miter should be modeled as
shown below by finding the dimensions a-through-d, and then entering two single cut bend
elements with tangent intersection points at i’ and i’’. By contrast, the closely spaced miter model
is defined like a standard bend. Each cut does not need to be defined. The user codes to the tangent
intersection point at (k) and enters the number of cuts. Calculations show that widely spaced miter
bends perform like closely spaced bends when s < (1.2)r2(1+tan θ), although this is not a Code
guideline. (The guideline is helpful, because it turns a number of 3 cut miters into closely spaced
miters and greatly simplifies the modeling effort.)
FE/Pipe, and indeed most pipe analysis programs, will make the calculation for closely and widely
spaced miters, but if the number of cuts is greater than 1, will use the uniform bend model for the
bend. To strictly satisfy the Code guidelines, the user must break each widely spaced miter into
single cut miters as shown in the B31.1 and B31.3 Appendix D.
When generating the bend stiffness matrix FE/Pipe will use the radius entered (or 1.5 Dnom if no
radius is entered) to calculate the spacing, and then from the spacing decide if the bend is closely or
widely spaced. From this determination, the flexibility characteristic is determined, and then the
flexibility factor and SIFs.
Stress Intensification Factor Orientations:
The angle associated with an intersection SIF is the angle to rotate from the default element “b” axis
to the inplane axis for the intersection, where the inplane axis for the intersection is defined as the
cross product of the axial branch vector and one of the header vectors framing into the intersection.

Definition of Local orientation angle for intersections.
The local orientation angle is printed for each element for each end that frames into an intersection
where a SIF should be calculated. The orientation angle gives the angle between the local “b” axis
and the intersection inplane bending direction so that SIFs can be properly aligned with the
provided moments. To find the local “b” axis of the element, cross the axial direction of the
element into the Y axis. If the element is along the Y axis, (the axial direction and the Y axis are
parallel), then the local “b” axis is along “x”.
The intersection orientation angle then gives the angle rotated about the axial direction vector to get
to the inplane axis for the element of interest. There will be three intersection orientation angles for
each element that frames into an intersection. In the example above, the intersection orientation

angle is -90 degrees. In general, the user never has to worry about these angles. FE/Pipe detects
them automatically and aligns all stress intensification factors and SIFs accordingly.
Application of Rigid Elements at Intersections
When the d/D ratio at branch connections is greater than 0.5 the rigid element length for the
FE/Pipe automated intersection model is generated by the following curve. Below 0.5, the rigid
element length as shown in the figure below is taken to exist from the centerline of the header to the
outside surface of the header pipe.
Code Compliance Notes
The B31 Code user should note that the full effect of pressure stresses on fatigue is not adequately
addressed in the B31 codes for pressure stresses that cycle significantly. When this occurs, the user
should consider using the advanced intersection models described below. There are a variety of
advanced models that can be used to include the effect of pressure in the load cases. The range of
moments may include displacements due to pressure, but this does not generally include the most
significant contribution of stress due to pressure. This issue is discussed in greater detail in WRC
335.
Weight only cases do not include pressure displacements, but do include the PD/4t longitudinal
stress due to pressure. When the weight cases are used as part of an EXPANSION “range”
calculation, the pressure component is not included in the weight contribution, since it is already
included in the operating contribution.
B31.3 sustained stress calculations do not include the torsional stress component as suggested by
B31.3 302.3.6 (a).
B31 code compliance for occasional cases is found by computing the stress component due to the
addition (or subtraction) of the occasional load. This stress component is summed into the
maximum sustained stress found in the model. Occasional stresses calculated during the path
dependent solution retain the signs on forces and moments. These loads are considered more
accurate, and the user is permitted to evaluate both positive and negative occasional loads in this

manner, although the summation of the stresses will produce conservative results, which may give
artificial predictions for systems where actual loads determine the integrity of the component, i.e.
refractory or glass lined piping.
The B31.3 operating pressure is found from [(Sax + Sb)2 + 4St2]1/2 . Sax is the axial stress in the pipe
including that due to internal pressure. Per B31.3 Appendix P. Sb and St are the bending and
torsional stresses per 319.4.4.
When corrosion is specified, the corroded section modulus is used for all stress cases. The nominal
thickness of the pipe is used for all stiffness calculations. The corrosion of the header is used to
adjust Ts in the effective section modulus calculation.
For B31.3, if the crotch radius is not entered for an extruded welding tee, the rx value in the
equation for the flexibility characteristic for the tee will be tbe B31.3 Code minimum 0.05Db. This
will essentially increase the extruded welding tee SIF such that it is the same as an unreinforced
fabricated tee. For B31.1, the crotch radius or thickness are not used in the SIF calculation. h = T/r,
and i = 0.9/h2/3
For B31.1, when the branch connection geometry requirements for diameter ratios in Table D-1 are
satisfied, the branch connection SIF given in Table D-1 will be used for checking the branch end of
the intersection per 104.8.4(C). This branch connection check will affect all intersection types.
For B31.1, when a crotch radius is provided for a branch weld-on fitting, the SIF is reduced by 1.5
per note 12 in Appendix D.
For thicker bends, the thickness of the bend fitting will be used in the equation for (h) where the
thickness is required, and the thickness of the matching pipe will be used in the equation for (h)
where the mean radius is required.
For bends B31.1 uses the matching pipe for Z and gets t and r for the "i" factor from the fitting.
B31.3 uses the matching pipe for Z and R, but gets "t" from the fitting. When Markl's tests in the
PVP journal are reviewed it is not clear what Markl did, although he almost certainly used Z from
the matching pipe, although it is not clear what he would have used for r and t in the expression for
the flexibility characteristic “h”. The question would be, for Markl's test, which approach yields the
best correlation with the test data. Since Markl tended to use the matching pipe properties, this is
likely what he intended. He would not expect pipe designers to measure bend thicknesses. When
the fitting thickness is available, it seems suitable to use it however, since the equations for bends
match the theory and test results reasonably well. To draw a conclusion, it seems that the B31.1
approach is probably the most correct for defining bends, but that the B31.3 approach is
satisfactory. It is somewhat interpretive however, as the code is not 100% clear. There are
drawings in Appendix D of both B31.1 and B31.3, but then the words refer to pipe and matching
pipe.
Load Combinations:
The FE/Pipe B31 Solver is path-dependent to properly evaluate friction loads, gaps and load case
directionality. The correct development of friction and the various static and kinetic options are

described in a separate section of this manual. Other nonlinear behavior in subsequent releases can
also advantageously utilize an ability to follow the actual loading path for any dependent part of the
cycle. As illustrated below, the proper treatment of friction requires a path-dependent solution.
The FE/Pipe B31 solver automatically establishes the most reasonable load path dependence, but
permits the user to alter those setting to accommodate individual needs.
The typical path dependent case developed for operating and occasional loads is shown below:
A key quality of the solution is the recollection that friction can cause the hot-to-cold expansion
stress range to be larger than the cold-to-hot expansion stress range. Many pipe stress programs
only evaluate the cold-to-hot expansion range and do not follow the “path” of loading, and so miss
the fact that the hot-to-cold expansion stress range is controlling.
This is only an issue when friction loads produce moments that are a significant part of the stress.
When friction can be ignored, the current FE/Pipe solutions will give results that are identical to
current piping programs that utilize +Y restraints and gaps.
One Operating Cycle – This is the automatic default.

Three thermal load cases are shown above. The reason for running multiple operating cases is to
determine if there is any tendency for the system to “walk”. Generally friction shakes down after
the first load case, but systems that are strongly dependent on frictional loads can continue to
“squirm” from case-to-case. A condition of walking is shown in the figure below:
To change from one operating case to three operating cases, a “3” should be placed in the “Number
of Analysis Operating Cycles” field on the General Form. This input is shown below.
The user will see three operating cases separated by three weight cases in the output and the load
case report. These cases are used to develop the range and path of the loadings. A B31 Code
expansion case will be developed between the starting operating case and the first sustained case,
and between the starting operating case and the final sustained case. The difference between these
cases indicates the path dependence of the solution. For FE/Pipe Version 5.0 this path dependence
is due only to friction. If there is no friction in the solution, then the expansion cases between the
first sustained case and the last will be identical. The path dependent nature of the solution will
only show where a stress range has been overlooked in a path independent solution. At the lower
bound the solution will reflect the stress range calculated in a path independent solution providing
the friction directions are correctly calculated. Several examples below show how friction loads in
a piping system can be easily path dependent even on initial loading, and so even though a path-

independent solution has the correct restraint normal force, interaction between supports may alter
the final line of action of the friction restraining forces.
Occasional loads:
Occasional loadings such as wind or earthquake can be applied to the system in a variety of ways.
The default is to apply the occasional load to the sustained loads in a positive sense. The absolute
value of the occasional component stress increment is added into the sustained stress and the result
compared to kSh as per B31.1 and B31.3, where “k” is the occasional load factor and “Sh” is the
hot allowable load.
Occasional loads can also close gaps in the positive direction, and then can move wholly back in the
negative direction, closing the gap again in the negative direction. They can also generally act
when the system is in a cold state or when the system is in a hot state. Often the user can determine
by inspection whether one direction (positive or negative) will result in the worst bending stress
change due to gap closure. In these cases the user can apply the occasional load in either the
positive or negative direction, or he can apply it first in the positive sense, and then followed by a
negative sense. These options are illustrated below, and the FE/Pipe input for these choices is also
shown.
Positive occasional load added to weight (sustained) case. A positive occasional load can also be
added to the operating case.

Negative occasional load added to the weight (sustained) case. A negative load similarly can also
be added to the operating case.
Positive and then negative acting loads are shown above. In the above example figures, the
occasional loads are shown starting from the weight (sustained) case. They can also be started from
the operating case, and the same plus, minus, or plus and then minus options applied.

The above input demonstrates how to apply the loads in first the positive, and then negative
directions after the sustained loads,
Occasional loads can be evaluated as primary or secondary (fatigue causing). When evaluated as
primary, the maximum occurring value of the load is often used as per the design specification, i.e.
the 100 yr wind or earthquake load. For design,
This concept is used in the nuclear code for example, where safe operating bases loads are
developed versus safe shutdown basis, etc.
In refinery or chemical plant service, low level wind loads for overhead lines, or for compressor or
pump piping systems, can produce the equivalent of weight loadings on equipment nozzles and
should be evaluated. These low level loads also act a significant number of times, and so may
contribute to fatigue life damage.
To evaluate occasional loads as “fatigue” loads in fepipe, the user should change the
PRIMARY/FATIGUE setting on the optional form to FATIGUE. In this case, the amplitude of the
occasional load developed will be evaluated using the B31 fatigue criteria.
Occasional Loads Causing Fatigue
Occasional loads can also contribute to fatigue failure. The user can check the effect of the
occasional loads on the fatigue life by changing the fatigue type on the optional form to FATIGUE.
The load case processor will then make the fatigue calculation based on the various combination
methods the user has suggested on the GENERAL form.
Typical Load Path without Friction, change in modulus, or material nonlinearities.

Friction Related load case demonstrating how the operating to cold load case can be larger than the
cold to operating case.
In general this is an oversimplification since the stress range is based on the difference in the signs
of the moments, not the signs of the stresses. The figure above, implies that the error in the
expansion stress case can only be as large as the weight only stresses since the B31 Code stress
values must always have a positive sign. In reality, the difference between the cold-to-operating
case and the operating-to-cold case can be at least as large as the stresses due to friction. This is
illustrated in the figure below.

Combination cases in FE/Pipe for B31.
All combinations of loads for expansion or occasional combination cases are performed at the
force/moment level. Once the occasional loads are developed, then the stresses from the occasional
case are added to the stresses from the sustained case. The path-dependent solver develops the
proper combination cases from the load history followed.
Presently, only a single thermal, pressure and wind load direction can be evaluated at a time.
For wind loads – if a portion of the pipe is inside a building, then use “none” for the wind load
diameter. If 0.0 or 0 is used for the wind load diameter, the default will be to use the pipe OD + two
times the insulation thickness as the wind diameter.
Beam Analysis Editing Notes
New Data for each screen is given on the following pages. Help is available for all text cells. The
user should hit “?” to have the help window popup to explain the cell. Note that for the Excel
interface, help for each column may be accessed by clicking on the header at the top of each
column.
The “TextOUT” and ‘TextIN” features at the top of each input screen:

can be used as an effective editing tool. Click on “TextOUT” to generate a semicolon deliminated
text file that will be displayed. An example from the restraint form is shown below.
The nodes where the restraints act are shown on the left, the directions in the third column, and so
on. The user can change any of the input shown or can add lines if necessary. For example, to
change the design hanger at node 10 in the above example to a Y only support, the user would make
the change shown below:
Note that column widths do not need to be retained. To implement the change the user “saves” the
TEXTIO.TXT file by use of File:Save, and then clicks on the TextIN menu option.

Specifics of FE/Pipe Beam/Pipe Modeling:
The Version 5.0 FE/Pipe beam modeling includes a variety of unique features intended to help users
make more accurate calculations of stresses and displacements in piping systems. A few of the
unique features accessible from the beam/pipe main menu include:
1) Multilayer, nonlinear refractory analysis (including system stiffness changes due to cracked
refractory).
2) Automatic FEA modeling of intersections to include more accurate SIFs and Flexibilities for
both branch and run side piping framing into intersections. Three FEA databases can be accessed:
a) PRG Global Internet FEA Database
b) Company Intranet FEA Database
c) Job Specific Machine FEA Database
3) Path dependent friction at gapped restraints
4) Stress Intensification Factors and Stiffnesses for Nozzles in Heads
5) Automatic Evaluation of Hot Sustained Cases per B31.3 Appendix S
6) Arbitrary thermal gradients for rain, sun, or heater tube radiation loadings
7) Scripting output control
8) Optional Microsoft Excel Interface
9) Through the MatPRO interface, the user can access:

a) Fitness for Service Evaluations per API 579
b) High Temperature Calculations per ASME Section III Subpart NH (Creep-Fatigue)
c) Multi-Code Fatigue Calculator
Key features available in the FE/Pipe Version 5.0 beam/pipe model is described below. (The Excel
interface documentation is separate.)

MAIN FE/PIPE BEAM/PIPE MENU
The FE/Pipe beam/pipe menu has included the GENERAL form for nonlinear and B31 Compliance
control. In addition the MATPRO – Material Wizard button on the bottom left of the main menu to
get access to B31.3 material properties, fatigue calculations from various codes, ASME Section III
Subsection NH Creep-Fatigue interaction rules, and API 579 fitness for service calculations. Users
can make stress calculations using the beam/pipe modeler and then insert those values into the high
temperature, fitness for service or creep calculators. The material database screen is shown below.
Material Database:
The MatPRO material database can be used to collect allowable stresses
and material properties, estimate SIFs, make fatigue calculations according to the latest codes,
perform a basic Fitness for Service evalaluation, and review high temperature allowable stresses per
ASME Section III Subsection NH.

The MatPRO user can enter stresses from the beam analysis to evaluate creep-fatigue interaction
and creep based primary (sustained) loads.
The FE/Pipe beam modeler permits the user to combine structural and piping models. At the top of
the element screen is the “Change Type” button. When pressed a selection window appears that
allows the user to choose the element cross section to be used. A variety of AISC shapes can be
selected, along with a userdefined shape, and/or the pipe cross section. B31.3 or B31.1 Code
Compliance calculations are performed on pipe shapes.

GENERAL
Intersection Model Options:
There are two drop-down menu options next to the FEA Flexibility and SIF Control Label. These
are used to establish which flex/SIF model is used for automatic intersection modeling
"NO_FLEX" - option uses rigid intersections and strict B31 Appendix D SIF'S. This is the
traditional way that pipe stress programs operate. Rigid connections are used at the centerline
intersection of the branch and header. This is known to produce large inaccuracies in certain
solutions in certain D/T and d/D solutions for particular model geometries. When the flexibility of
the intersection is of the same or smaller order than the piping around it, considerable load
redistribution can occur.

"include_FLEX" - option uses a flexible model for intersections and improved SIFs. This model
along with the FEA option should produce the most accurate load distribution in the piping system,
and permits the user to specify head, lateral and offset geometries.
The DEFAULT is NO_FLEX if the user does not specifically make a section.
The second drop down Intersection Model Option lets the user chose which intersection effect he
wishes to use. Appendix D SIFs are geometry and parameter range limited. The user may use FEA
solutions to improve the SIFS only from Appendix D while still using the rigid intersection model.
The three options available provide combinations of improved SIF and FLEXIBILITY models. The
default is to include BOTH improved flexibilities and SIFs.
The three options are:

1-include improved Flexibility model only and use standard SIFs
2-include improved SIF model, and standard (rigid) flexibilities
3-include BOTH improved flexibility and SIF models (DEFAULT).
- Depending on the option selected and the size of the piping system, there
may be a large number of finite element models to be run to support SIFs and flexibilities for the
piping system. The program will always check the PRG global database if there is an open internet
connection, or any user, or company local intranet databases, but if models are not found, a new
FEA analysis of the junction must be performed.
To facilitate the potential large number of runs that need to be made, the user can press the “Start
Asynchronous FEA” button to begin analyzing the finite element models in the FEA que. Models
are placed in the que whenever a geometry is plotted. Repeated plotting clears the que and refills it.
When the Start Asynchronous FEA button is pressed, FE/Pipe starts running jobs not already
analyzed that are in the que. FE/Pipe will find those that have already been run and stored in a
global or local database, and list the geometries that have not been run. The list of intersection
models that are not available will then be queued to run in the background so that the user can
continue to develop the model, or perform some other activity while the needed FEA models from
the piping system are prepared. When all models have been analyzed in the background, the user
can then make any number of piping runs using the stored data without having to make another
FEA run. Other models which use the same intersections can also be analyzed without needed
additional FEA calculations.
- This button is used to access the output control text file. This file can
be used to develop restricted output based on particular beam elements or nodes so that smaller,
more useful reports can be generated. The file accessed by this button is <jobname>cmm.
Commands for the <jobname>.cmm file are found in a separate section. When the user presses the
output control file text button the first time a template shows up that contains the optional
commands that can be used. This template is shown below:
/NOTES
# Command Options:
# /NOHEADER /STOPHERE /USEOPTION /OPTION /CASE
# /XCASE /USEFILE
# Report Options:

# /DISP /STRESS /FORCE /PROP /DETAILED /RESTR
# Exclusion/Inclusion Options:
# /NODES /XNODES /ELEM /XELEM
# See documentation for more information.
A typical input is shown below. This input is used to eliminate headers from the beam reports,
eliminate cases 1,5 and 6 from the beam output, and exclude node numbers 100 to 1000 from the
beam reports.
/NOHEADER
/XCASE
1
5
6
/XNODE
100 1000
- When the user wants to define the refractory properties for pipe
to be used with the 18+ degree of freedom ovalization refractory model this button is pressed. The
user enters the different layers of refractory on the inside or outside of the pipe, and then enters the
properties of each layer using the button. See the separate section on
refractory in this document.
When accurate, nonconservative friction is used in a piping analysis, the system solution becomes
path dependant. Friction can also cause the system to “ratchet” or “walk” as it repeatedly heats and
cools. This input allows the user to direct the solver to “heat and cool” the system repeatedly “x”
times, so that walking of the system can be detected. The last weight and operating cases can be
compared to the first so that the user can see the difference in these solutions.
The first value used here should be 2. The user can then compare the first heating-cooling cycle to
the second to see if there is an appreciable difference. If there is a difference then this value should
be increased until the system reaches equilibrium.
Enter the operating case number (or any case number) where the restraint solution should be used to
evaluate the hot sustained stress case. User’s may want to make multiple runs to evaluate multiple
hot sustained stress cases. See the separate section of this document describing hot sustained stress
cases and what they mean to the analysis. The application of hot sustained stresses has been
discussed heatedly over the last 20 years as many program approach the hot sustained stress case
differently.
If the user does not explicitly enter a load case number the program will select a hot sustained case.
The hot sustained analysis is always performed as the last load case in the set of solution load cases.

The user of the hot sustained case may be important when liftoff at supports is large and when the
system is particularly sensitive to sustained case stresses. i.e. when creep is an important criteria.
User’s should check the box to the right above to TURN OFF the hot sustained load case analysis.
(The main reason to do this is to minimize the number of load cases processed, although the best
way to do this is by using the XCASE command and the CMM file described above. This way, any
Code overstress state encountered can still be checked and printed.)
These inputs describe the methods used for combining occasional loads with sustained or operating
runs. There are three options for each.
1 – The POSITIVE occasional load acts with the specified load case.
2 – The NEGATIVE occasional load acts with the specified load case.
3 - BOTH the POSITIVE and NEGATIVE occasional load acts with the specified load cse.
Option 3 for both the sustained and operating load cases provide the most thorough evaluation of
occasional loads, but also adds the most significant number of load cases to the output reports.
(User’s can minimize output using the CMM file controls described above.)
- Check this option to zero all friction. This is a recommended

option to evaluate the variation in the system solution caused by friction. Where friction may have
a significant effect on stresses, loads or displacements, the user should probably make sure that the
all requirements for solutions WITH and WITHOUT friction are satisfied.
Control item #1:

By default, reports prepared for shell element models are NOT included in beam output results. If a
shell element model is included with the beam model, or for any other reason the user wishes to
include the additional Div 2 results for the beam-type piping model, a “1” should be placed on the
stress option line in the GENERAL form shown below. It is likely that other options will be
included later as needed.
Control item #2 lets the user specify the number of iterations for each nonlinear load case. The
default is 20 times the number of nonlinear restraints. If there is a problem during the output, this
value will be displayed to the user. Item 2 permits large iterations per case to be entered. To be
used, the users entered value must be greater than 10.
Control item #3 – Default is 99.9999%. When the solution reaches this percent of the load case it
will be considered converged. Some solutions slow as they approach the final converged solution.
Solutions very close to the solution are usually satisfactory. As solutions get closer this switch lets

the user take those as converged solutions. This option should be used with caution. Occasionally a
tolerance limit message will appear. This is usually of no concern since in these cases the last valid
increment was just smaller than the 1.0. To verify that this is an artifact, the user can change
control item #3 to 100.0 and make sure the solution does not change.
Control item #4 lets the user control the number of load steps printed in the nonlinear path
dependout output per case. If no value is entered, then the converged state at 10% intervals is
printed. (There will be a maximum of 10 iterations per case, and one if there is no nonlinear path
changes during the load case. A typical value to change this to is 25%, 25.0. The user can enter -
1.0 to have all iterations during the load cases printed. For systems with a large number of
nonlinear supports, using -1.0 can produce a lot of output. When the user is studying friction affects
carefully, on relatively small systems, -1 is a useful value to use.
For example, if the user wants to permit the default number of load case and convergence tolerance,
but print the solution at 25% of each load case, they would enter:
0,0,0, 25
Control item #5 lets the user set a “forced” convergence tolerance step. The default step is 1e-15.
If there are convergence problems, the user can enter a value of 1e-5 to 1e-7. When the user enters
a value, each support during any load step will increase by the given amount of the load step.
Smaller values indicate smaller forced steps and less error, but also give less opportunity to nudge
the solution out of an oscillation. Oscillations occur when the solution steps from one sliding
condition to another without an increment in the load case. When using this option, the user should
investigate the quality of the solution. These tolerances only effect solutions that include friction
and forced steps greater than zero introduce some error in the solution although a forced step of 1e-
15 may introduce 1e-15 of the total system load per iteration step. To check any iterative solution,
the user should look at each friction support status and verify that a sliding support exerts µ times
the normal force in the sliding direction and that a non-sliding support exerts less than µ times the
normal force in a direction perpendicular to the support line of action.
Control item #6 is the friction non-sliding stiffness. The default value used is 1e10. With path
dependent solutions it is difficult to determine if lower or higher values will converge faster or more
effectively. As a first step the user is recommended to increase the default non-sliding stiffness to
1e10. If this does not help a non-converged solution, then the user should try 1e7. Higher values
tend to be more accurate from a displacement point-of-view, but the difference may not be apparent.
The user can estimate the error in the non-sliding friction displacement by dividing the maximum
friction sliding force by the default sliding stiffness.
7-If greater than 0 then activate preferential sliding. Should be some multiple of 10.
8-If greater than 0 then permit overshoot. Values from 1-to-100 are reasonable. 10 is recommended
if used.
9-If 1 then use error tolerance in nonlinear restraint load reports.
10-If greater than 0 then the kinetic friction percent, and also used to activate kinetic friction.
11-If greater than 0.5 and less than 2 then number of allowed changes per support is estimated. If
greater than 2 then this is the number of allowed changes in state per support per load case.

ELEMENTS:
- Wind OD
The Wind OD can be used to increase the outside diameter used to develop wind loadings. By
default the diameter of the pipe plus twice any insulation thickness is used. Alternatively the user
can insert any value in this field. The word “NONE” should be used to turn OFF wind.
- Refractory ID.
If refractory exists and the 18+ degree of freedom element option is turned on, the effect of the
refractory weight, thermal expansion and stiffness can be included with the steel pipe weight,
thermal expansion and stiffness. Stresses in the refractory are reported and the nonlinear
characteristics of the refractory can be entered and the solution will determine which areas of the
refractory are likely to fail. Any neutral axis shift due to refractory failure is also included in the
analysis for the most accurate refractory model available. Refractory in ductwork has been known
to increase the stiffness of large elbows by 40-to-50 times. Once the refractory ID is entered, the
user should enter the refractory properties. Buttons to access the refractory property data can be
found on the OPTIONAL form.
Thermal bowing may be

used to consider the effects of liquid layers developed inside the pipe, such as may occur when
filling with LNG, furnace radiation, rain, or sun. Any of these effects can produce a
circumferentially varying temperature gradient along the axial length of the pipe. Depending on the
source of these gradients, they may be found in horizontal or vertical pipe. The figure below shows

how each of the variables are used for thermal bowing. There are seven inputs in the Thermal
Bowing text cell:
Tbot , Ttop , LiqLevel , ProfileType , Vx , Vy , Vz
Tbot – Temperature on the bottom of the pipe

Ttop – Temperature on the top of the the pipe
LiqLevel – The liquid level in the pipe. (Used for Profile Types 2 and 3.)
Profile Type: 1-Linear from bottom to top.
2-Step change in temperature at the liquid level.
3-Constant temperature from bottom of pipe to liquid level and linear from liquid
level to top of pipe.
vx,vy,vz – Orientation vector that points from the bottom of the pipe to the top of the pipe. The
orientation vector can point in any direction that is normal to the pipe axis. (For example, in a
vertical furnace tube, radiation impacts the side of the tube, and the thermal bowing orientation is in
the horizontal direction.)
Profiles and orientations can change from element – to – element. Various profile types are shown
below.

RESTRAINTS
The definition of path dependant, nonlinear, gapped supports in Version 5.0 of FE/Pipe is one of a
number of its most significant benefits. Most programs have particular requirements for nonlinear
restraint definitions and convergence, but the model in FE/Pipe is intended to provide considerable
capability along with input flexibility.
Three entries can be placed in the gap and friction text box for up to three restraints:
<Gap in Minus Direction> , <Gap in Plus Direction> , <Friction Coefficient>
For a typical CAESAR +Y support with a 0.3 friction coefficient the user would enter:
0 , 1000 , 0.3
Note that a total of 20 restraints can be specified on any single restraint data sheet, and up to three
nonlinear directions per restraint. The nonlinear restraint specification will apply to the first three
restraints in the list to facilitate the easy definitions of +Y supports with guides, gaps, and axial limit
stops..
A few example restraint inputs are shown in the figures below:

There is a single +Y restraint with friction at node 5. At node 10, there is a +Y support with friction
and a guide in the Z direction with a gap of 0.25” (on each side of the support), and a friction
coefficient of 0.3. At node 15 there’s a +Y support with friction and weight settlement in along the
negative Y direction and an operating movement in the plus Y direction. (This effect would be used
to simulate a support that was attached to an operating vessel that was subject to some foundation
settlement, and thermal expansion in the operating case.)

Supports at nodes 10 through 55 are shown below. The supports and their descriptions are used to
keep the input clear and organized. Only two input forms are needed since multiple nodes and
directions can be provided at each support. Restraint descriptions (in fact, all user descriptions)
show up in the input listings to help identify parts of the models or to provide reasons why supports
are at particular locations, etc.)

Miscellaneous Restraint Examples:
In the above case there are four restraints at nodes 20, 30, 35 and 40. Each restraint is a +Y with a
friction coefficient of 0.3 and an initial displacement along the Y axis of 1.134”.

FORCES and MOMENTS
The user should check this box if loads in the operating force case are due to pressure.
In this event, the loads will be omitted from the weight case, and added into the sustained and
operating cases.

EXTRA STRAIGHT ELEMENT DATA

FEA generated SIFs and Stiffnesses are stored in a three tiered database so that any single geometry
only has to be analyzed one time. This means, for example, that a piping system with fifteen (15)
different FEA simulated intersections might take thirty minutes to run the first time while the
intersections are analyzed using the FEA method, but that each additional run of the piping system
will only take a few seconds to run as the FEA solutions are extracted from the FEA database.
Global databases are stored on the secure PRG web site, machine, (or company) databases are
stored on company networks or on an individual machine, and job specific databases are
maintained. (The database operations run behind the scenes and are maintained entirely by the
program.)
The user has complete control of where and how SIFs and Stiffnesses from FEA or other methods
are used, and may also include SIFs or Stiffnesses on spherical, elliptical, dished, flat or conical
heads including offsets and laterals.
The STRAIGHT ELEMENT DATA form is also used to define EXPANSION JOINTS and CUT
SHORT or CUT LONG cold spring elements. These features are described in more detail below.
There are two major areas in the STRAIGHT ELEMENT DATA form. The upper area defines
individual nodal SIFs and the lower area describes information that may apply to a single node or to
the entire element.
For example, if the user wants to define the branch SIFs at an intersection, the element that defines
the branch must be identified. An example illustrating this input is shown below:
There will be three elements framing into the intersection node at 20 and only one is the branch.
15-to-20 20-to-25 and 20-to-45
Specifying SIFs and flexibilities at node 20 does not provide enough information to identify which
element should contain the SIF information. For the above intersection, assume the branch SIFs are
2.45 and 3.4, while the header SIFs are 1.8 and 1.4. The lower SIF is the inplane SIF and the higher
SIF is the outplane SIF for the branch leg, while the opposite is true for the header legs. For the
user to manually specify these SIFs, an input screen must be identified for each element end as
shown below:

User Defined Three Segment Intersection Definition
Note that the inplane direction vector is not required since the intersection geometry can be
determined by the program. An example where the inplane direction vector is required is shown in
the figure below.
In this case, the user does not want to enter the 1” drain line because the load on the drain line due
to the weight of the valve is considered negligible, and it’s easier to leave it out. Omitting the
branch, or either run pipe segment framing into the intersection, makes it more difficult for the
program to properly determine and validate the orientation of the branch, although it tries using
diameters and wall thicknesses. If the SIFs for the connection should be entered, (particularly for
MZ loading through the run) the values must be obtained. For this example, the fitting
manufacturer gives the inplane SIF for the run (header) pipe as 1.8, and the outplane SIF for the run
(header) pipe is 1.1. To put these SIFs in for the run pipe elements (since the small branch pipe is
not defined), requires that the user enter the “inplane direction vector”, otherwise there is not

enough geometry included in the model for the program to determine the inplane direction. In the
above example, the inplane direction is described by the Z axis. Moments about the Z axis describe
an in-plane bending of the branch and the header legs in the plane of the intersection. The input for
this geometry is shown below: (Note that there is no third input for the branch SIF because the
branch element is not included because its load is considered small, or simply because it’s easier not
to enter all small branch connections.)
User Defined Two Segment Intersection Definition (Branch Element neglected)
For intersections where the intersection type can be defined by a standard B31 Appendix D SIF,
only the intersection node needs to be entered because the program can determine the header and
branch legs from the geometry. An example of this “standard” intersection type is shown below.
There is no extra data required for this intersection. The program finds all elements framing into
the intersection at node 20, determines which are the header and branch elements and places the
SIFs (and optionally flexibilities) effectively.

SIFs can be placed on any element in the model except for RIGID elements. (For bends or curved
elements use the “EXTRA BEND ELEMENT DATA” form.)
Automated FEA Intersection Model
The beam definition of a typical intersection geometry could be identified using the points below.
Each individual node is identified in the table below. The nodes “20” are associated with the tee
while the ndoes 25, 15 and 45 are associated with nodes in the attached piping system that are
removed from the tee, but connected to tee nodes by the interconnected pipe.
Node Description
15 Remote point on the run portion of the piping system removed from the tee. A straight
piece of pipe connects node 15 to node 20.5. Node 20.5 is located at the weld line of the
tee on the run side.
45 Remote point on the branch portion of the piping system removed from the tee. A
straight piece of pipe connects node 45 to node 20.7. Node 20.7 is located at the weld
line of the tee on the branch side.
25 Remote point on the run portion of the piping system similar to node 15.
20 Tee point physically located at the intersection of the branch and run centerlines.

20.1 Tee point physically located at the intersection of the branch and run centerlines. Points
20, 20.1 and 20.2 occupy the same point in space but are unique points in the stiffness
defintion of the piping system whose degrees of freedom are related.
20.2 Tee point similar to node 20.1.
20.3 Tee point physically located on the centerline of the branch pipe removed from the run
centerline by an amount equal to 1/2D where D is the mean diameter of the run pipe.
20.4 Tee point physically located at the same point on the branch centerline as point 20.3
20.7 Tee point located at the branch weldline. The piping system external pipe attaches to
the welding tee at the weldline identified by the node 20.7.
Note that for unreinforced fabricated tees, pad reinforced tees, and others, some of the nodes in the
above model at weldlines will not exist. For example, there may not be a branch weldline node
#20.7 in an unreinforced or pad reinforced tee intersection.
The most accurate beam model for the intersection would likely involve a superelement developed
from a shell or brick finite element model that interacted with a beam element that supports 30 or
more degrees of freedom per node. For a general application, thirty (30) is likely the smallest
number of degrees of freedom that can be used at a beam node to properly transmit centerline forces
and moments, ovalization, dilation and warping effects from one beam element to another. Most
beam programs in common practice in 2009 do not support more than 6 degrees of freedom per
node. Those programs that do support more than 6 degrees of freedom per node suffer from
problems identified by Bathe that do not make them well suited for general piping applications.
User’s of beam elements having more than 6 degrees of freedom per node should be sure that their
element is suited for general use.
For 6 dof approaches the interaction of branch and header stiffnesses must be considered in the
models for certain loading conditions and D/T ratios. All point spring 6 dof intersection flexibility
models have limitations, and for any particular situation, one model will likely be better than
another. Since these situations rely on the interaction of loads and the boundary conditions acting
on each end of the intersection. It is difficult at the model building stage to determine which model
would be best suited. As a result the following model below is outlined which appears to suffer the
smallest number of difficulties.
The assumptions and requirements in the recommended model are the following:
1) The model must perform in a symmetric manner.

2) All required flexibilities must be provided by interactions between elastic elements of negligible
length interacting between two degrees of freedom on the model.
To evaluate the stiffnesses for each point spring the Markl load configuration will be used. In the
Markl load configuration, one end of the intersection is fixed while load is applied to another end.
The non-fixed and non-loaded third end remains free. The objective of the intersection model is
that a single model is used that results in the proper rotations of the loaded end when a moment is
applied. For the model shown above, one load and support configuration would be to have node 15
free, node 45 loaded and node 25 fixed. When an inplane moment is applied to node 45, there will
be translations and rotations at nodes 45 and 15. The most desireable model will produce the
translations and rotations that compare most favorably with the physical test or the finite element
model duplicating the load and boundary conditions. Given the restrictions of symetry and

negligible length requirement of the stiffness/flexibility model, the configuration shown in the
figure above is considered the most reasonable configuration for the stiffness model.
For each rotational degree of freedom at any free end there will be inplane, outplane and torsional
defintions. Each will be treated similarly, except that the torsional elemental stiffness expression
for a straight section of pipe will be M/φ = EI/(1.3L). (See Wais para 4.2.). The bending elemental
stiffness expression for a straight section of pipe will be M/φ = EI/L. Care must be exercised when
the stiffnesses are calculated independantly but used together. The discussion below attempts to
provide the necessary segregation.
When the moment Mh is applied to the “h” end of the run, with the “f” end of the run fixed, the
expression for rotations at φh will be:
φh = φhg + φgd + φde + φef
The φh is found from an fea solution or test of an intersection component. The equation relating
displacement through the intersection is:
φhFEA = φhg + φgd + φde + φef = M [ Lhg/(EIhg) + 2/kgd + Lef/(EIef) ]
kgd = 2[φhFEA/M - Lhg/(EIhg) - Lef/(EIef) ]-1
For inplane moments applied through the branch, the recommended model has the following
rotation relation when the “h” end is free and the “f” end fixed.
φa = φab + φbc + φde + φef
This equation can be rearranged and substitutions made to find the kbc term:
1/kbc = φbc/M = φa/M – Lab/EIab – 1/kdc – Lef / EIef
kbc = [ φa/M – Lab/EIab – 1/kdc – Lef / EIef ]-1
When the local branch stiffnesses are found as shown in Wais, the kbc term above is not equal to the
Wais bending stiffness term since the kde term is omitted. When Wais stiffnesses are developed, the
model from the point “b” to the point “e” can be constructed and the collective model kbc term that
should be used with the kde term evaluated:
In this case, if the Wais local bending stiffness term is taken as φb/M, where:
φb = φbc + φde
φb/M = φbc/M + φde/M = 1/kbc + 1/kde.
Then the kbc term to be used in the global stiffness model of the intersection can be found from:
kbc = [φb/M - 1/kde ]-1 = [ 1/kbcWAIS + 1/kde ]-1

Example Stiffness/Flexibility Calculations:
Branch = 10.75 x 0.365; Run = 14 x 0.375”
One end freed:
Both ends freed:
Stiffnesses thru the Run
Stiffnesses thru Run (longer lengths) Lb=44, Lr=55 (FEP input values)
The inplane flexibility factor from Wais eq. 4-11 is:
kib = 0.488(D/T)1.279 (d/D)0.391 (d/t)-0.602

= 5.577

The resulting inplane stiffness is 3,879,616 in.lb/deg. This compares to 1.6 and 1.7e6 in.lb./deg.
from the PRG FEA models.
From Widera:
kib = 0.680(d/D)-0.242 (D/T)0.802 (t/T)0.622 [ 3.437(d/D) – 7.414(d/D)2 + 4.766(d/D)3 ]
= 7.05
From PRG:
kib = (1.08(d/D) – 2.27(d/D)2 + 1.39(d/D)3) (R/T)1.17 (t/T)0.46 (d/D)-0.0435
= 3.56
The PRG result produce a slightly stiffer intersection than either Widera or Wais. The Widera
results are generated with longer lengths. (PRG lengths were generated based on default PRG
lengths).
The thru-run flexibility factor from PRG is:
kih = 1.23 (R/T)0.47 (t/T)-0.475 (d/D)5.37

= 1.1325
The stiffness for the thru-run direction is:
Kih = (772.246e6)/1.1325 = 682e6 in.lb/deg
The adjusted kib for the collective intersection model per the approach outlined above is:
Kbc = [φb/M - 1/kde ]-1 = [ 1/kbcWAIS + 1/kde ]-1
= [ 1/(2,123,373) + 1/(682,000,000) ] -1
= [ 4.7095e-7 + 1.466e-9 )-1
= 2,116,782 in.lb./deg.
As can be seen from the above equation, the header stiffness has little effect at this d/D ratio. Note
that flexibility ratios are not comparable, but stiffnesses are what determine the interaction of the
branch and the header. The table below gives an indication of the relationship between the inplane
branch and header stiffness factors that would be used in the above stiffness expression. These
values show that for d/D < 0.5 that the header stiffness does not influence the branch stiffness
regardless of the D/T ratio or t/T ratio. As the d/D ratio approach 1 the header and branch
stiffnesses show some interaction. Ignoring this interaction in the calculation could result in
intersection junctions that are too flexible.

FEA Database Access
The FE/Pipe beam element modeler (pipe28) accesses the global PRG internet FEA-SIF-Stiffness
database to get FEA SIFs and STIFs for intersections if the global database contains those SIFs and
STIFs. If the global database does not contain the appropriate SIFs and STIFs or is not available,
then the FE/Pipe beam element modeler will initiate a finite element run in the background to
calculate the appropriate SIFs and STIFs. When the SIFs and STIFs are calculated, they will be
used in the model, and included in the global PRG database, in a local job database, and in a
machine specific database. The three databases are defines as follows:
“Job” Specific Local Database: This database is in the same folder as other job files and contains
all FEA SIFs and Stiffnesses for intersections in the job. SIF and Stiffness data is lost when the
folder containing the job is deleted.
Machine Specific Local Database: This database is referenced by the Registry on initial program
loading and for efficiency should not be deleted or moved. This database contains all FEA SIFs and

Stiffnesses for any job run on that machine where FEA SIFs and Stiffnesses are used. As the
machine specific local database grows, the need for running FEA analyses is reduced.
PRG Global Database: This database is referenced by the program and is stored on the secure PRG
servers. This database contains all seed FEA model data and results created by PRG and any
additional FEA model data and results submitted to the database by participating users. (To get data
out of the database, user’s must be willing to contribute new results to the database as they are
produced on their own machine.) This global sharing approach will significantly reduce the
number of FEA analysis required as the database grows. This subject is discussed in greater detail
later, but as FEA technology is developed to more accurately address parameters that are generally
not included in SIF and Stiffness calculations, additional runs will be required. See Special Topics
section 6.
Database access is described in the logic diagram below:
FEA SIF and Stiffness Database Access Logic
To Use the FEA Database for SIFs and Stiffnesses, (or any combination of SIFs and Stiffnesses),
the user must activate the “Include_FLEX” option on the general form and select the combination
of updated SIFs and Stiffnesses. (See the second drop down menu below.)

Specify the intersection node number and intersection type:
and a Level 1 full-stiffness/SIF FEA model will be constructed automatically for the intersection.
An example Level 1 model is shown below:
Branch side SIFs are automatically placed at node 20.4 on the element from 20.4-to-45, and header
side SIFs are automatically placed at the node 20.1 on the 15-to-20.1 element, and at node 20.2 on
the 20.2-to-25 element. The nodes 20.1, 20.2, 20.3 and 20.4 are automatically generated by the
program.
See Special Topic Item #7 for simulating nozzles on spherical, elliptical, dished, flat or conical
heads.
Users may control the SIF/Stiffness option for the intersection using the above drop down options.
The defaults for each are PROGRAM and FEA. In the default state, the model of the intersection is
controlled from the GENERAL form shown below.

If the user selects “Include FLEX” and one of the adjacent options including “STIF”, then all valid
intersections will be modeled automatically with FEA SIFs and Stiffnesses as shown in the
Automatic intersection model in the figure on the right above.
Other options for the above drop downs are explained below:
NEVER – Use the classic intersection model – a rigid connection and B31 Appendix D SIFs.
ALWAYS – Always use the intersection model defined in the source drop down. (Ignore input on
the GENERAL form.)
PROGRAM – Use the option in the GENERAL form.
FEA – This is the recommended, and considered the most accurate method.
PRG Manual – Since the first release of FE/Pipe in 1989 PRG has been evaluating data and
equations on intersection flexibilities. A compendium of these methods is included as part of the
PRG Manual method. This is a closed form, single equation solution and provides much faster
responses than the FEA method, but for many parameter ranges will not be as accurate.
API Low Tank – This is an extension of the API 650 low tank stiffness calculation and can only be
used for cylinder on cylinder intersections.
WRC 297 – This method uses Figure 59 and 60 of WRC 297 to develop stiffness factors for
cylinder on cylinder intersections.
NB3600 – The Class 1 Nuclear code in Section 36xx provides estimates of the branch stiffnesses for
cylinder on cylinder intersections.
RIGID – The “rigid” method helps the user evaluate the differences between models that include
“rigid” and realistic, flexible intersections. The “rigid” method uses the multi-element shown in the
example model above, but inserts rigid stiffnesses in between header and branch nodes. In the
sketch above, this is between nodes 20.1 and 20, 20 and 20.2, and between 20.3 and 20.4.
The user may override SIFs or Stiffnesses used at any program identified intersection by entering
the values that should be used in the appropriate fields shown below.

For intersections, the user should recognize that three elements define intersction and SIFs and
Flexibilities, and so if user’s wish to override program computed values for these elements, the user
defined values must be entered on three elements for each intersection.
Expansion Joints.
Five types of expansion joints may be defined on the STRAIGHT ELEMENT DATA FORM:
To describe a straight pipe element as a singe expansion joint that may optionally be tied, hinged, or
gimbaled the user would enter the form and enter the FROM
and TO node numbers for the expansion joint as shown in the example above. The “inplane
direction vector” can be used to describe the axis of the hinge pin. (The vector is in the direction of
the hinge pin.) In the Ki,Ko,Kt,Kax text cell, three values for the expansion joint may be entered.
The last value is optional. These values are defined in more detail below and in the special section
8 on expansion joint modeling.
To identify the element as an expansion joint the “Special Element Type” combo-box in the bottom
right of the screen should be set to EXPANSION_JOINT as shown above. (Other options include
cut short and cut long cold spring.)
The first entry, “5000” in the example above is the axial stiffness of the expansion joint. The length
of the expansion joint, its’ effective diameter and its’ axial stiffness are the only values needed to
determine the rotational and translational transverse spring stiffnesses for the expansion joint.
The second entry, “24” in the example above is the effective diameter of the expansion joint. The
effective diameter is measured to the half-height of the bellows, i.e. the effective diameter is:
Effective Diameter = (Pipe ID) + (2)(convolution half height)
The torsional polar moment of inertia is found by multiplying the effective bending moment of
inertia by 100. Users should always check torsional loads on expansion joints.

The third entry, “4” in the example above is the type of constraint applied to the expansion joint.
The available constraint options are described below:
1-Hinged expansion joint: Axial deflection is restrained and only rotation about the hinge axis is
permitted. There is no friction added to the EJ rotation, and so user’s concerned about large
frictional loads developed due to the restraint of axial displacement should increase the stiffness of
the EJ to compensate, or add equal and opposite moments of an appropriate magnitude to simulate
the effect of the friction. In practice it is often better to increase the EJ stiffness (guess), and then
recalculate the increased moment to see if the assumed increase was sufficient. An iterative
algorithm to evaluate the frictional loads is planned.
If a hinge axis is not specified with a type 1 restrained expansion joint a typical axis will be
assumed. For a non-vertical EJ element axis, the hinge axis is vertical. For a vertical EJ element
axis the default hinge axis is in the X direction.
2- Gimbaled expansion joint: Axial deflection is restrained and rotation about transverse axis is
permitted. See the discussion of friction under the type 1 joint above.
3-Tied expansion joint. This is a simple tied model whereby a single rigid (1e12) restraint is placed
along the axis of the expansion joint. There is no restraint of relative rotational displacement at the
expansion joint ends.
4-Tied expansion joint with rotational restraint. This is a more comprehensive tied expansion joint
model where the axial displacement is restrained along with the relative transverse rotation of the
bellows ends in both transverse directions. Three or more tie-bars removed from the expansion
joint centerline will often provide this rotational restraint providing nuts or stops along the tie bars
NOT provided with gaps. If gaps are provided, the analyst should be sure the system surrounding
the tied expansion joint is sufficiently flexible so that pressure loads are not carried inadvertently by
adjacent, rigid equipment or very stiff piping.
Cold Spring Elements
Cold spring elements are also defined on the STRAIGHT ELEMENT DATA form. The user must
enter the nodes on the element whose length should be used to define either the cut short or cut long
cold spring. Then in the special element type drop down, specify either cut short or cut long. An
example is shown below:

The element from 15-to-20 was added and its length set to the amount of the cold spring. For all
load cases (except dynamic ones), the element length will be removed (thermally contracted to
zero).
Hanger Design Notes
For trapeze hangers the user should be careful that rotation at the centerline might cause a
redistribution of load at the hanger location. The following table is a guideline.
(create a table based on estimated trunion lengths for different diameter of pipe)
(don’t permit more than a 0.1” of movement of the spring .. ?)
The user can specify XZ and rotational freedoms only for the nodes that have Y directions freed.
XZ and rotational freedoms for other nodes will be ignored.
Hot load design can be a much better idea for hot systems since this will minimize the weight loads
when creep may be a part of the system load. The hot load designs result in an increase in load in
the cold case, which can make the hanger stops difficult to remove, although cold load designs
result in a decrease in the cold balanced loads, which can also make the stops difficult to remove,
but they tend to be less difficult.
If using operating (hot load) hanger design, be sure to check the displacements for the weight case
following the hanger design. The vertical displacements at hanger displacements in this case are an
indication of how much movement the system will undergo when the hanger stops are removed.
Some pipe specs permit wide ranges in wall tolerances on the “high” side. For pressure this may be
conservative, but for hanger design when creep is an issue, the varying load may produce a

tendency for the system to creep since the load used in the analysis does not account for the increase
in the wall thickness.
Only the first two letter of the spring manufacturers name must be entered, although the entire name
can be spelled out for clarity.
User’s must recognize that even though the spring hanger report reflects the omission of the fluid
column on the discharge side of the pump, the anchor loads on the pump will still include the fluid
column unless the user adds the operating case fluid column load. See the example included.
When specifying “freedoms” the user can enter: 5,30,50 in the Y freedom section, and 30 only in
the XZ and ROT sections. The processor must deal with these properly. Use this in a hanger
example.
Hanger design does not use friction since the freed anchor restrained weight case may not reflect the
actual load or gap distribution at the nonlinear supports.
18 Degree of Freedom Control
General Discussion
The 18 degree of freedom piping beam element is an industry first implementation for
piping and pressure vessel analysis. The element formulation includes the effects of
ovalization, dilation and warping that are not considered in the 6 degree of freedom classical
elements found in traditional pipe stress programs. Functionality not included in a typical
6dof beam element includes:
1) Modeling of stiffeners (stiffening rings)

2) Simulation of flanges at any cross section in the model (not just at bends)
3) Loads and stiffnesses acting on the surface of the pipe
4) Interaction of ovalization between adjacent elbows
5) Differential thickness or thermal expansion between adjacent elements
6) Shell Stress Formulation to obtain more accurate stress calculations.
7) Inclusion in Dynamic analysis for more accurate shapes and frequencies
8) Correction of bending/torsional shear errors in most 18dof formulations
An example of ovalization and warping modes not explicitly included in typical 6dof beam
elements are shown below:

The strong interaction of ovalization modes between bends can cause one bend to either
augment or retard another’s ovalization modes, resulting in greater or less system stiffness
depending on bend orientation. The distored plot below shows how bend cross sectional
deformations can interact:
Surface loads or supports can also cause ovalization of straight or bend sections:

Local Applied Loads and Stresses can also be simulated. The shell model below shows the type of
local loading and stress supported in the 18dof element, and how straight sections can ovalize when
locally supported.
The 18 degree of freedom piping beam element is located in the “Beam Models” template. The
controls for the element are found in the “Elements” panel grouped in the section labeled
“Ovalization Control”.
Enable Ovalization
Select “YES” to allow ovalization effects for the element defined on this page. “YES” will change
the element type from the standard 6 degree of freedom beam to the 18 degree of freedom element.
NOTE: ONLY A SINGLE ELEMENT SHOULD BE DEFINED PER PAGE WHEN ENABLE
OVALIZATION IS SET TO “YES” AND STIFFNESS FACTORS ARE PROVIDED.
Elemental Stiffness Factor

Four inputs toggle ovalization and warping on or off at four equally spaced points along
the element. Input “1” or “0” in the following order:
<first point>, <last point>, <internal nearest to first>, <internal nearest to last>
where “1” prevents ovalization/warping, and “0” permits ovalization/warping.
For example: the pipe segment shown below would be defined as 1,1,0,0 to describe the
flange restriction to ovalization at both ends of the element.
Element stiffness factors are also used to define stiffening rings. The “1” is replaced by the
radial stiffness of the ring in load per length per length of circumference. For stiffening
rings this stiffness can be estimated as:
K = 4AE/d2 where:
K = stiffness value to input

A = Area of stiffener
d = diameter to centroid of stiffener section

Point Supports on Surface
Stresses and deflections in real pipe are influenced by the location and type of pipe
support. “Point Supports on Surface” controls the support location both along the length
and around the circumference of the element.
Up to four “point supports on surface” can be defined per surface. For each “point
support on surface”, the required inputs are:
Point#, hoop_angle, CosX, CosY, CosZ, Force, Stiff
Point# One of four equally spaced locations along the element. The order of
points is: “1” first, “2” last, “3” middle (closest to first) and “4”
middle (closest to last). If Point# is equal to ”5”, the support is
distributed along all four nodes.
hoop_angle The angle around the circumference of the pipe (see figures below).
Positive angles are defined by the orientation of points 1 to 2 and the
“right hand rule”. For bends, 0 degrees is the intrados. For straight
elements two cases exist: (i) when the element is parallel to the
Global Y axis, 0 degrees is located on the Global +X side of the
element. (ii) for all other orientations, 0 degrees is defined as the cross
product of the element local x axis and the global “Y” axis. The hoop
angle is designated with the symbol Φ in the figure shown below.

CosX/Y/Z The GLOBAL direction cosines of the force or restraint.
Force Spring preload
Stiff Restraint stiffness. Typical values for “rigid” restraints are 1E15 lb/in
[2E14 N/mm]
The “Optional” panel includes an entry for the application of the 18 degree of freedom
elements for the entire model.
Global Ovalization Control
Three options are available. They are GLOBAL_ON, By_ELEMENT, and

GLOBAL_OFF respectively.
GLOBAL_ON Ovalization and warping are activated for all the elements.

By_ELEMENT Ovalization and warping depends on the input specified by “Enable
Ovalization” entry in the Element pages.
GLOBAL_OFF and warping are excluded for all the elements and the input in Enable
Ovalization will be override.
P-∆ Effects, Large Rotation, Stress Stiffening and Stress Softening

P-Delta (large displacement and stress stiffening) effects may have appreciable effects in the
following instances:
1) Pipe overhangs.
2) Vertical Risers
3) Compressive Axial Loads (Softening)
4) GRE Piping
5) Tall building structures supporting heavy column loads
For pipe overhangs, the pipe extends past the last support, turns upward and then runs another
horizontal distance before encountering the next support. In this case, considerable sag of the pipe
may occur if designers do not properly account for the increased flexibility along with the weight.
Condensate can collect in these low, non-drained areas, further increasing the sag, ultimately
resulting in a situation where unexpected slugging occurs in the line.
When vertical risers are permitted to move laterally, developed, compressive axial loads can reduce
the lateral strength of the line, permitting additional lateral displacement and greater offset. These
increased lateral displacements produce greater bending moments at the base of the riser.
When axial loads develop in long lines due to friction, the lateral load bearing capacity is reduced
and large displacement can occur in the lateral direction as the pipe “conforms” to the axial load.
Small displacements in the axial direction (thermal strains), can produce very large lateral
displacements and binding or failure of supports. This geometry effect can be visualized in the
following diagram.
GRE piping is particularly susceptible to the development of axial compressive stresses and so can
be particularly likely to stress soften.

When pipelines or piping systems are likewise subject to tensile loads, such as may be developed
due to pressure, they may experience a lateral stiffening effect and so deflect less when subject to
lateral loads such as wind, weight or developed loads due to attached piping. This is the well
known “guitar-string” effect.
The classical P∆ effect in structures is shown in the figure below. P is the vertical load in the upper
right of the figure and ∆ is the lateral deflection of the structure. With a large enough lateral
deflection, the deflection also induces a moment contribution P∆ at the bottom base of the structure
that is not calculated in a single step linear solution.

∆ Introduction
P-∆
Linear elastic analysis used to determine design forces and moments resulting from loads acting on
a structure is first-order analysis, where small deflection behavior is assumed and equilibrium and
kinematic relationships are based on the undeformed geometry of the structure. Solutions of these
analyses are typically simple and straightforward. However, when lateral loads are applied to the
structure, the configuration can deviate quietly noticeably from its undeformed configuration, which
makes a second-order analysis necessary. In a second-order analysis, the equilibrium and kinematic
relationships are applied to the deformed structure, which includes the effects of large displacement
and stress stiffening.
P-Delta Effect
It can be see from Fig.1 that when the lateral force H act on a frame, the frame will deflect laterally
until an equilibrium position is reached. The corresponding lateral deflection may be calculated on
the basis of the original configuration and is referred to as the first-order deflection and is denoted
by ∆ I . If vertical forces P are acting on the frame, the lateral force H and vertical forces P will
interact with the lateral displacement ∆ I from H to drift the frame further until a new equilibrium
position is reached. The lateral deflection that corresponds to the new equilibrium position is
denoted by ∆ (Fig. 2). The phenomenon by which the vertical forces P interact with the lateral
displacement of the frame is called the P-Delta effect (Chen and Lui (1991)).
P-Delta effect occurs to every structure where elements subject to axial load. The magnitude of the
P-Delta effect is related to the magnitude of axial load P, the stiffness/slenderness of the structure as
a whole, and the slenderness of individual elements.
Figure 1. Frame subject to lateral force only

Figure 2. Frame subject to lateral force and vertical forces
To analyze P-Delta effect accurately, a second-order analysis based on the deformed geometry of
the frame is necessary, which usually entails an iterative process. This is due to the fact that the
deformed geometry of the structure is not known during the formulation of the equilibrium and
kinematic relationships.
Three practical approaches are available to include the P-Delta effect: 1) the beam-column approach
which involves the stability functions, 2) the finite element approach using the energy theorem
which introduces an additional geometric matrix, and 3) the pseudo load approach which accounts
for the geometrical nonlinear effect by updating the load vector instead of the stiffness matrix.
Multi-Cycle Iterative Method
As it is mentioned before, the consideration of P-Delta effect requires nonlinear analysis which is
tedious and time-consuming. However the second-order effect can be considered in an approximate
manner to get an order-of-magnitude estimation of the impact P-Delta, stress stiffening, or stress
softening will have on the solution.
Fe/Pipe employs a multi-cycle iterative method as an approximation to the full nonlinear analysis.
The analysis begins with a first-order analysis on the structure. The stiffness matrix is obtained
without considering any second-order effects. Member axial forces are then calculated based on
these first-order results. Displacements and loads are updated. Load updates are based on the
second order effects. Three iterations through this process are conducted by default, and have found
in trial cases to be sufficient. The number of iterations can be changed on the GENERAL form in
the bottom on the stress-option data line.
References
Chen, W. F. and Lui, E. M. (1991) Stability Design of Steel Frames, CRC Press, Inc.

FE/Pipe Version 6.0 NozzlePRO Version 8.0 PCL Version 3.1
The first piping program to support FEA models

automatically has just released Version 2.0!
ASME Section VIII Division 2 – 2008 Update to FE/Pipe,

NozzlePRO, PCL and other PRG products. This document principally
addresses required ASME Section VIII Division 2 changes.
Contents
New Features in FE/Pipe 6.0 (all PRG products)

A08 ASME Section VIII Division 2
ASME Welded Curve Fatigue Method Using Elastic Analysis (5.5.5)
Fatigue Nomenclature:
Fatigue Basics:
ASME Polished Bar Fatigue Method (for Welded and Non-Welded Specimens) 5.5.4
PRG Welded Methods for Tees and Girth Welds
User Defined Fatigue Curves

Summary
The 6.0 Version of FE/Pipe contains revisions for the mandatory A08 version of ASME Section VIII
Division 2. These changes are incorporated into each of the PRG products.
The Excel interface for the piping checklist has also been completely rewritten and a significant number of
features added.
The 3D viewer has also been updated and permits multiple instances, labels, new color maps, etc.
All modifications to the program are described in the new features list below.
The ASME Code update may have a significant effect on results. New stress calculations are often 10% or
more lower than A06 and earlier calculations, welded fatigue curves are available, and the use of “k” for
occasional loads has been replaced with the AISC 7 load factor method. Load factor guidance is provided
below in this document, but it remains the user’s responsibility to select appropriate occasional loads from
AISC 7 or similar documents. The 3Sm allowable has been replaced by SPS, and local stress checks have
been added to protect against non-ductile failure in hydrostatic stress states. Extensions of the new fatigue
curves are also provided based on fatigue tests of welded components performed at PRG and on compilations
of all other available fatigue results. When comparing post 06 VIII-2 calculations to pre 06 VIII-2
calculations, one should not be surprised if the newer calculation gives results that are 10% lower.
Stress classification lines (SCLs) have been added automatically for brick intersection models so that
comparisons can be made with the shell solutions and results compared to Code allowables. Both shell and
brick models have unique analytical difficulties, which are discussed below. Having SCLs for both model
types, allows the user to compare them and decide which result is more conservative. The post 06 VIII-2
Code makes more refernece to finite element analysis and gives guidelines for modeling and evaluation of
finite element models. Methods of stress classification in volumetric and shell elements has received special
treatment. (See VIII-2 Annex 5.A.) Direction from the Code and supporting WRC documents has been
added to FE/Pipe and can be found in the approaches taken for load case development and stress
classification definition and use. There is a considerable amount of technical information and discussion in
the pages below. User’s are encouraged to read through it briefly to determine what is applicable to their
problem types.
The FE/Pipe input processor now contains “Files”, “Units” and “database” options so that users can change
any of these items from the input making transitioning between different jobs, or parent and child models
easier. A variety of options have also been added to the tabular output processor.
Multiple models may be sent to the 3d viewer, or multiple outputs from the same model so that user’s can
interrogate their model more easily and can compare different displacement or stress profiles by clicking in
between images in the same view. A variety of tools has been added to support stepping between the
multiple models. An example is shown in the image below. User’s are encouraged to experiment with these
features to determine which ones are most well suited to their output interrogation appraoch.

Since the maximum calculated stresses in PVP geometries may drop from between 10 and 15% using
equivalent stress instead of stress intensity, this can materially change allowable loads that are based on
The Piping Checklist (PCL) has also undergone considerable change and allows more interaction with the
FEA model associated with the piping system. Model duplication and rotation, Excel interrogation of
results, output viewing with input, etc. are all features that have been added to the program. A complete list
of new features in the PCL is given below.
Chapter 38 Example Model

March 13, 2008

Nodes 25-26 are at same point in space. Local flexibilities are provided between these points.

Original and Deflected Thermal Shape
Thermal Loads with Rigid Intersection Model, and Thermal Loads with Flexible Intersection Model
Loads: lb.
Moments: in.lb.
FE/Pipe
25.0 -392. -2050. 16. 3205. 1571. 157005. NOFLX
25.0 -305. -725. 15. 3259. 1453. 32175. FLX
ADLPipe
25.0 -370. -1934. 16. 2964. 1512. 148044. NOFLX
25.0 -288. -698. 15. 3024. 1404. 31728. FLX

No Flexibility Included (p24)
Selected Comparisons
Case Node Direction ADLPipe FE/Pipe CAESAR

Weight 10beg FY 3250 3246
50end FY 4619 4624
80end FY 801 801
Weight 10beg MX -11949 -11897

50end MX -4790 -4890
80end MX 1 0.5
Thermal 10beg FY 13304 14185

50end FY -15237 -16236
80end FY 1934 2050
10beg FZ 67287 71649

50end FZ -67271 -71633
80end FZ -16 -16
10beg MX -150422 -160426

50end MX 485393 517416
80end MX -570 -601
25 DZ 0.232 0.23198
40 DZ 0.234 0.3581
30 DZ 0.372 0.372
70 DY 0.483 0.483
1
Looks like CAESAR/FEPipe bends are a little stiffer than ADLPipe bends

No Flexibilities (Compare to P24) 6dof Elements
FE/Pipe Ver. 4.5b
Loads: lb. Moments: in.lb.
10.0 -10. 3246. -986. -142769. -1060. -2472.

20.0 10. -366. 986. -73951. -127. 2472.
20.0 -3. 189. -986. 73915. 141. 813.

30.0 3. 1539. 986. -25349. -362. -813.
30.0 -3. -1539. -986. 25349. 362. 813.

40.0 3. 2896. 986. 12300. -472. -702.
40.0 -3. -2896. -986. -12300. 472. 702.

50.0 3. 4624. 986. -58685. -472. -481.
20.0 -7. 177. 0. 35. -14. -3285.

25.0 7. -144. 0. -35. 12. 1361.
26.0 -7. 144. 0. 35. -12. -1361.

60.0 7. 115. 0. -35. -4. -33.
60.0 -7. -115. 0. 35. 4. 33.

70.0 7. 153. 0. -34. -6. 1066.
70.0 -7. -153. 0. 34. 6. -1066.

80.0 7. 801. 0. 6. -6. -570.
25.0 -7. 144. 0. 35. -12. -1361.

26.0 7. -144. 0. -35. 12. 1361.
Load Case 2
10.0 46. 14185. 71649. -1925112. 22367. 130831.

20.0 -46. -14185. -71649. 222878. -16901. -130831.
20.0 438. 16236. 71633. -226083. 15136. -50780.

30.0 -438. -16236. -71633. -942894. 16376. 50780.
30.0 438. 16236. 71633. 942894. -16376. -50780.

40.0 -438. -16236. -71633. 1051409. 32132. 35024.
40.0 438. 16236. 71633. -1051409. -32132. -35024.

50.0 -438. -16236. -71633. 6208993. 32132. 3512.
20.0 -392. -2051. 16. 3205. 1765. 181611.

25.0 392. 2051. -16. -3205. -1571. -157005.
26.0 -392. -2051. 16. 3205. 1571. 157005.

60.0 392. 2051. -16. -3205. -26. 39844.
60.0 -392. -2051. 16. 3205. 26. -39844.

70.0 392. 2051. -16. -3350. 118. 54770.
70.0 -392. -2050. 16. 3350. -118. -54770.

80.0 392. 2050. -16. -7213. 118. -39335.
25.0 -392. -2050. 16. 3205. 1571. 157005.

26.0 392. 2050. -16. -3205. -1571. -157005.

Flexibility Included (bknn07)
Selected Comparisons
Case Node Direction ADLPipe FE/Pipe CAESAR

Weight 10beg FY 3243 3239
50end FY 4615 4619
80end FY 813 813
Weight 10beg MX -11915 -11862

50end MX -4779 -4878
80end MX 0 0
Thermal 10beg FY 14035 14970

50end FY -14734 -15694
80end FY 698 725
10beg FZ 67032 71372

50end FZ -67017 -71357
80end FZ -15 -15
10beg MX -154206 -164699

50end MX 484652 516009
80end MX -562 -591
25 DZ 0.233 0.23245
40 DZ 0.234 0.23517
30 DZ 0.372 0.37211
70 DY 0.485 0.48494

Flexibilities (Compare to bknn07) 6dof Elements
FE/Pipe Ver. 4.5b
Loads: lb. Moments: in.lb.
NODES FX FY FZ MX MY MZ
Load Case 1
10.0 -8. 3239. -984. -142348. -821. -1580.

20.0 8. -359. 984. -73556. -177. 1580.
20.0 -1. 194. -984. 73521. 190. 485.

30.0 1. 1534. 984. -25293. -244. -485.
30.0 -1. -1534. -984. 25293. 244. 485.

40.0 1. 2891. 984. 12274. -270. -458.
40.0 -1. -2891. -984. -12274. 270. 458.

50.0 1. 4619. 984. -58538. -270. -405.
20.0 -8. 165. 0. 35. -13. -2065.

25.0 8. -133. 0. -35. 11. 279.
26.0 -8. 133. 0. 35. -11. -279.

60.0 8. 127. 0. -35. -5. -12.
60.0 -8. -127. 0. 35. 5. 12.

70.0 8. 165. 0. -33. -6. 1184.
70.0 -8. -165. 0. 33. 6. -1184.

80.0 8. 813. 0. 5. -6. -635.
25.0 -8. 133. 0. 35. -11. -279.

26.0 8. -133. 0. -35. 11. 279.
Load Case 2
10.0 -135. 14970. 71372. -1973694. -5175. 27889.

20.0 135. -14970. -71372. 177328. -11061. -27889.
20.0 169. 15694. 71357. -180587. 9423. -12983.

30.0 -169. -15694. -71357. -949414. 2762. 12983.
30.0 169. 15694. 71357. 949414. -2762. -12983.

40.0 -169. -15694. -71357. 1054426. 8854. 6891.
40.0 169. 15694. 71357. -1054426. -8854. -6891.

50.0 -169. -15694. -71357. 6192108. 8854. -5294.
20.0 -305. -725. 15. 3259. 1639. 40872.

25.0 305. 725. -15. -3259. -1453. -32175.
26.0 -305. -725. 15. 3259. 1453. 32175.

60.0 305. 725. -15. -3259. 26. 37400.
60.0 -305. -725. 15. 3259. -26. -37400.

70.0 305. 725. -15. -3398. 164. 41182.
70.0 -305. -725. 15. 3398. -164. -41182.

80.0 305. 725. -15. -7095. 164. -31905.
25.0 -305. -725. 15. 3259. 1453. 32175.

26.0 305. 725. -15. -3259. -1453. -32175.

Flexibilities and SIF Evaluation
Flex NB-3686.5-1 FESIF (Lw=0) FESIF (Lw=0.125)

Ki 376,467 279,046 310,065
Ko 94,265 65,824 70,378
Units: in.lb./deg
FESIF Input
FESIF Results Lw = 0
FESIF Results Lw=0.125

Model Weights Used:
Pipe 24 x 0.375 Insulation Thickness = 1.5 inch.

Pipe Density = 0.283 lb/cu.in.
Insulation Density = 0.006597 lb./cu.in
Fluid Density = 0.03611 lb/cu.in.
Element Density = 24 lb/in.
Pipe 6.625 x 0.280 Insulation Thickness = 0.5 inch.

Insulation Density = 0.0069023 lb/cu.in
Fluid Density = 0.03611 lb/cu.in.
Element Density = 2.7 lb/in.
Rigid Pipe 6.625 x 2.00 Insulation Thickness = 0.0

Insulation Density = 0.0
Fluid Density = 0.0
Element Density = 2.7 lb/in.

New Features FE/Pipe 6.0
Summary of New Features
1) A08 ASME Section VIII Division 2 Update for all PRG Programs
2) ActiveX Programmatic Interface for NozzlePRO
3) Welded Fatigue Curves in FE/Pipe
4) Welded Fatigue Curve Calculator
5) User Interface Updates
6) Piping Checklist Version 3.1
7) Fatigue Damage Report
Complete Feature List
1) New A08 ASME Section VIII Division 2 Implementation for all PRG Programs
a) Rigorous Fatigue Option
b) Welded Fatigue Curves
c) Code Control Form for all Programs and Templates
d) Code Updates for WRC 107/297 Calculations
e) SPS Replacing 3Sm
f) PRG welded fatigue curves
g) Updated Fatigue Report Formats
h) Welded Fatigue Calculator added to Stresplot and MatPRO.
2) Update PCL
a) Speed and Excel Interface Improvement
b) Output Interaction in Excel
c) Updated Rotating Equipment Module
d) Updated Wind Load Module
e) Updated Seismic Load Module
3) Automated Stress Classification Lines in Brick Template
4) Screen Format Updating
a) Users can change configuration parameters from input.
a) Units
b) Database block size parameters
c) ASME Code defaults
b) Output processor updated to permit tabular or graphical report generation
c) Pipe Size and MatPRO button from input
d) Ability to start and change jobs from the input file menu.
5) Updated Output Report Writer
6) StressPlot and Plastex Updates
7) Miscellaneous
a) Nominal Diameter and Pipe Wall Thickness Lookups.
b) Option to change jobnames or select new templates from the input processor.
c) User may change database parameters from the input processor.
d) User may change units from the input processor.
e) AxiPRO improvements for large diameter metric flanges
f) FE/Pipe Batch Job Processor (V2.0a)
g) Pad on bends may be entered assuming bend wall removed due to errosion.
h) Updated Batch Processor
i) Unstructured mesher available for flat heads with nozzles.

j) Improved welding tee shell geometries for SIF evaluation of actual thicknesses
8) Piping Checklist (PCL) Modifications (v3.1)

1. Rotate Model
o Rotate models up to 360 degrees about the X, Y and Z axis.
2. Copy/ Mirror Model
o Copy section of a model and mirror them in XY, XZ and YZ plane.
3. Automatic Unit Conversion
o Units are automatically converted from English to Metric and vice-versa by a click of a
button.
4. Use FE/Pipe Editor
o Export piping files as child models for later use in FE/Pipe.
o Import FE/Pipe models as child models for structures.
5. Output Control
o Select Load Cases to exclude/include in reports
o Select Nodes to exclude/include in reports
6. Excel Results
o Results can be displayed in an Excel Sheet.
o Any number of load cases can be selected to view in the Excel Sheet.
o Displacements, Restraint Loads, Forces and Moments and Code Stresses can be seen.
o The Excel Results can be saved as *.xls for later opening.
o The user can have full manipulation of the Excel Sheet for post processing.
7. Mat/PRO Link
o Import Material Properties for B31 and ASME Tables.
o High Temperature Calculations
o Fatigue Calculations
8. Nozzle/PRO Link
o Open and modify Finite Element models in Nozzle/PRO.
o Heads and intersections.
9. FEBend Link
o Open and modify Finite Element bend Models in FEBend.
o Bends only and Bends with supports.
10. Quick Link to Program Manual
o Right Click option to link to manual
o Pressing F1 on the Excel Sheet
o Pressing the Help button
11. Email Support
o Right click option to directly email [email protected]
12. Error Messages
o Error messages link to cell with error.
13. Comments
o Comments can be entered in the Nodes Cell of multi-row sheets.
o Comments can be displayed in the 3D viewer.
14. Input Duplication

o Columns with blue font are duplicated; the value from above will be carried down for
simplicity.
o Disabling the feature will leave only the changed values; the rest will remain as blank.
15. Autofit and borders
o Autofitting the cells can be done by pressing the Auto Fit button or right click and autofit.
o Border can be redrawn after inserting or deleting a row.
16. Hide/Show Columns
o The user may dynamically hide and show columns with the right click options.
17. Data Lists can be seen by right clicking, this will show data from the multi row sheets, this is helpful
when large number of elements and restraints are found in the model.
18. Checkboxes to disable Pressure, Temperature, Densities, External Forces and movement for run
purposes.
19. NPS and DN Outer diameters are allowed. Tables for NPS and DN can be accessed by pressing
Ctrl+Shift+O.
20. Schedule thicknesses are allowed by entering “STD” or “s40” (for Schedule 40), if it is found, then
the actual thickness will be entered. Tables for thicknesses can be accessed based on the actual
diameter entered by the user by pressing Ctrl+T.
21. Files may be opened by double clicking the input file; a new instance of the piping software will be
opened.
22. Recently used files feature in the “Open” button.
23. New ASME Section VIII Div 2 preferences added.
9) Nozzle/PRO (v8.0)
• New ASME preferences menu options to control ASME VIII-2 solutions.
• Various updates and corrections.
10) Axi/PRO (v3.0)

• Batch processing capability
• New option to define width of pressure distribution across gasket surface.
• Incremental bolt load analysis options to evaluate several bolt loads for a given flange geometry.
• Added new ASME preferences menu options to control code compliance reports and solutions.
11) FE-107 (v2.0)

12) FE-SIF (v2.0)

13) Stress/PLOT (v3.0)

• Various upgrades to support new ASME VIII-2 Part 5 rules.
• Cycle counting tool to evaluate stress results with time-history solutions. Includes ASME VIII-2,
Annex 5.2 methodologies for cycle counting.
• Report generating tool to construct tables of linearized stress results.
• General stress input option to allow stress linearization of stress results from general purpose FEA
tools.
14) Mesh/PRO (v3.5)

• Upgraded for AutoCAD 2010 support (2000 and 2005 editions still supported).

• Stress and joining region capabilities for shell models.
• 3D hexahedral volumetric meshing with surface guiding.
• 3D hexahedral volumetric sweep meshing tool.
• 3D element and mesh improvement tools.
15) Viewer Updates

a) Multiple viewer instances allowed
b) Translucent option on brightness control
c) Multiple color gradient selections from File:Options
d) Pipe Labeling
e) Control of legend labeling
f) Control of trifecta labeling
g) Viewer Overloading
16 MatPRO (v3.0)
a) Update to A08 Code Addenda
FE-Bend (v1.2)
1. Long file names are allowed, including spaces.
2. Double Clicking input file will open and populate FE-Bend Screen.
3. Added Density field entry for material properties.
FE-661 (v1.2)
1. Long file names are allowed, including spaces.
2. Double Clicking input file will open and populate FE-661 Screen.
3. Added Density field for header box and nozzles.
Updated Batch Processor Screen

A08 ASME Section VIII Division 2
In July of 2007 ASME Section VIII Division 2 was significantly changed.
In July of 2009, the 2007 changes became mandatory.
Along with restatements and clarifications of existing rules a new fatigue evaluation method was added and
the stress basis changed from Tresca to Von Mises. The occasional load factor (k=1.2) that had been part of
the allowable in the A06 and earlier versions of VIII-2 was removed and the AISC 7 load factor approach
was incorporated to replace the occasional load factor (k=1.2) method.
The change from a stress intensity basis to a Von Mises basis will reduce many calculated stresses by about
10% and was intended to put the Code on a foundation that is more accurate and accomodating to elastic-
plastic (nonlinear) analysis.
A fatigue procedure was added that specifically addresses welded components. This method goes beyond
the fatigue methods in BS5500, AASHTO, AISC and the EN Code and is based on WRC 474, “Master S-N
Curve Method for Fatigue Evaluation of Welded Components.” The method allows the user to control a
number of the variables that determine the suitability of a component to resist fatigue loadings but may only
be used by owner approval (5.5.5.1 (c)). Considerable documentation is included below to explain the new
method and to help the user understand how to control the method so that more accurate fatigue results may
be obtained. Additional options are provided to permit the user to extent the new method to take advantage
of the BS5500, AASHTO, AISC, EN, new ASME methods and recent testing at PRG. It is believed that
results based on the recent PRG test methods will give the most accurate results since they are based on
fatigue tests of full sized components. The PRG methods, though more exacting, are limited to girth butt
welds and welds in tee joints. PRG methods will likely give better results for failure prediction in the low
cycle range. The standard deviations of each method are provided so that user’s can make estimates of the
probability of failure.
The Master S-N Curve method from WRC 474 is intended to reduce the weld failure standard deviation by
using an equivalent structural stress developed from the membrane and bending stress at the weld acting
normal to the plane of crack growth. With the reduced standard deviation, less shift is required from the
mean curve to gain the same level of design dependability. Additionally, parameters not quantitatively
separated in earlier methods have been identified and included in the structral stress definition or in the
development of the allowed number of cycles so that the Code user has more control over the accuracy of the
method.
To address VIII-2 updates a new form was added to each FE/Pipe program menu and the output reformatted.
The new output format contains life and load fractions to help user’s satisfy Eq. 5.36 for the polished bar
approach and 5.66 for the welded fatigue approach. The updated fatigue output appears below:
New Fatigue Report Format Old Fatigue Report Format

If the old report format appears for a fatigue analysis, (i.e. Pl+Pb+Q+F) stresses, then the user does NOT
have the new Div 2 options turned on. All Code options activated are described in the “Stress Results-
Notes” output report.
SPS replaces 3Sm as the membrane + bending secondary stress allowable. SPS is the larger of three times Savg
or two times the average value of the yield stress. Operating rather than design pressure may be used for
occasional load cases. This is not conservative, and is NOT the default. User’s may request this option
however and develop smaller occasional loads. See checkbox #9 in the discussion below.
Various combinations of the operating load components are evaluated for a rigorous fatigue analysis. This
approach and the life fraction output is discussed in more detail below.
The new ASME form and options are described below.
Non-Code Membrane and Bending Stress Printout Control
Membrane and bending stresses are printed for different load cases so that user’s can extract values used for
fracture mechanics or other evaluations. In some cases, a user may prefer a less cluttered report, and would
like to deactivate “non-code” stress outputs. There are two files in the installation folder that control how
these stress components are printed: HOPPER.BI2 and HOPPER.BIN. The 14th entry on the 3rd line of this
file should be a 1 to cause the stress components to be printed and a 0 to deactivate printing of the non-code
membrane and bending stress components.
Rigorous Load Combination Fatigue Analysis
The rigorous load combination fatigue analysis prepares load cases and evaluations for any possible
combination of operating loads. This process logic is new in Version 6.0 of FE/Pipe-NozzlePRO and must be
activated from the ASME Option screen described below.
Rigorous load combination fatigue analysis is used when the operating state of
Weight+Pressure+Temperature+Forces (W+P+T+F) is not obtained in a monotonic increase of the final load
from the starting weight case. If the system may be pressurized for any length of time prior to heating, or
heated for any length of time when not pressurized, then the change in loads from weight (W) to W+P+T+F
is not monotonic, and it is possible that some intermediate load state might produce the worst case fatigue
damage.
When the user asks to perform a Rigorous Load Combination Fatigue Analysis, several options are activated:
1) All permutations of intermediate load conditions for startup are evaluated. The discussion below provides
a thorough discussion of the non-proportional cycling associated with PVP load conditions and how the
rigorous load case control should be used. Simply, the largest possible largest damage factor is extracted
from all possible permutations of the load conditions. This damage factor should be combined with any
fatigue damage from occasional loads, or separately applied single loads. Several examples are given to help
the user understand the approach. The resulting damage factor should be less than unity. (See VIII-2 Eq.
5.37).
2) If there are any intermediate cycle ranges that may occur, the user must add the unity factors from those
cases.
3) This cummulative damage approach is conservative since the maximum stress from different cases might
occur at different local regions. Maximum damage factors are reported on a region-by-region basis.

4) Load cases are setup that demonstrate the reduced allowable that occurs when cycling is between two hot
conditions as may exist when the cycle is between W+P+T and W+T. Pressure stresses can be significant,
since bending pressure stresses are combined in this limit.
5) Pressure stresses are also included in the fatigue evaluation in cycle cases: W+P to W. In this case, the
peak stress due to pressure cycling is included in the evaluation. In the non-rigorous approach, the only
cycle evaluated is W+P+T to W.
Standard brick model from NozzlePRO:

Speak = 24044 psi at the inside corner in the longitudinal plane.
To find the regions shown, the user can plot the geometry. When the model generation window appears,

Region 339:
Point_radius,4
Draw, region, 339
meshmult=1.5

Model
Model Peak Membrane M+B Type
Run002a – brick 24044 19264 20526 Non-integral, mesh multiplier =1
Run002a-brick 23812 19067 20918 Non-integral, mesh multiplier = 1.5
Shell 14937x2=29874 15137 22129 Shell solutions give the membrane stress
at the junction as the membrane stress for
the sum of the plates. This is different
from the brick model. The shell solution
uses an SCF=1.35. 0.9 operating stress is
NOT used for secondary stress analysis.

Welded Fatigue Curve Calculator
The welded fatigue curve calculator has also been added in Version 6.0 of FE/Pipe-NozzlePRO so that users
can enter membrane and bending stress values from any source or at any location in the model for welded
fatigue curve evaluation. The weld fatigue curve calculator can be found in StressPlot, MatPRO and Plastex.
The welded fatigue curve calculator is shown below:
Welded Fatigue Curve Calculator

New A08 ASME Section VIII Division 2 Part 5 Implementation
Program changes required to implement the ASME VIII-2 A08 update are listed below.
1) ASME Program Control Form added in FE/Pipe-NozzlePRO, BendPRO, PCL, etc.

2) Equivalent Stress (Von Mises) replaced Stress Intensity (Tresca)
3) 3Sm allowable replaced by SPS.
4) Welded fatigue curve method in addition to older polished bar fatigue method.
5) Damage ratio replaced allowable stress for fatigue assesment.
6) Occasional load factor (k) replaced by AISC load factor method.
7) Option added to evaluate operating nozzle loads as primary (elastic followup).
8) Difference considered for operating and design pressure load cases.
9) Inclusion of standalone weight case for Code stress evaluation.
10) Local stress check (σ1+σ2+σ3 < 4S).
11) Rigorous Load Case Fatigue Damage Analysis
12) Thru-thickness stress component manipulation for volumetric integration.
The fatigue life of welded components has been studied extensively and adjustments for plasticity correction
factors that may be needed for polished bar specimens are likely not needed for most welded constructions in
the PVP industry for conservative designs. Potential increases in strain accounted for by Ke or Neuber
adjustments are offset by the change in stress vs. cycle slope and the reduction in Kf in the low cycle range.
Ke and Neuber corrections are provided because of the possible underestimate of strains from an elastic
analysis when the secondary non-peak stress range exceeds 2Sy. Cantilever tests show that these increases
are offset by the reduction in Kf and transition to the plastic stress vs. cycle failure slope. Options exist to
omit these plasticity corrections. Supporting documentation is included. Readers should review the notes
carefully before applying these options. Generally, a more accurate analysis is desired when a failure study
is being performed, otherwise, the letter of the Code should be followed. Plasticity correction is used in the
pre-07 VIII-2 by use of the Ke factor, and is provided in the post-07 VIII-2 Code by use of the Neuber
Correction. Each of these methods, corrections and references are discussed below. The defaults employ
conservative Code recommendations.
The new ASME Program Control form is available from many places. Most prominent is in the input
processor for all FE/Pipe templates, in NozzlePRO, BendPRO, etc. The form and its’ access point from the
FE/Pipe input processor menu is shown below.
FE/Pipe Access to ASME Program Control Form

ASME Program Control Form
Once entered, the ASME settings may be stored locally, globally, and/or saved with the input. Any changes
to the ASME controls are stored with the input when the user exits the form by clicking on the green OK
button in the bottom right had corner of the form. Local and Global options are provided so that user’s can
establish a Code environment. With local storage all new jobs started in the local folder will have those
Code settings. New jobs anywhere in the system will use global settings if there are no alternate local
settings. Current settings are stored with the model input files. Options selected are printed in the output
report in Stress Results – Notes.
Each numbered item in the ASME Program Control form is discussed in detail below:
1-Use to select the Code year and addenda. This option will become a dropdown box when subsequent
Division 2 updates for Version 6.0 are released.

2-Use to include the protection against local failure in VIII-2 para 5.3 for operating cases. The S1+S2+S3 <
4S check in Eq. 5.5 is used for primary membrane plus bending principal stresses. If the user feels that this
check should also be made for operating stresses then this box should be checked.
3-The use of VIII-2 para 5.3 is not required (per 5.3.1.1) if the component design is in accordance with the
standard details of Part 4. If this is the case, the user can omit the S1+S2+S3 local check from the Code
evaluation by checking this box.
4-For stress classification lines (SCLs) along brick element edges adjacent to singularities where there are
rapid changes in thickness, (such as at the toe of a fillet weld or square corner), there can be large through-
thickness stress components that do not reflect bending and membrane states along the SCL. In these casees
the thru thickness components and shears for the 3D linearization can be ignored to obtain more realistic
linearization results. This option only applies to volumetric (non-shell, and non-beam) elements. See
“Stress Classification Lines Straight Through Singularities”, Dr. Arturs Kalnins, Professor Emeritus of
Mechanics, Lehigh University; and “3D Stress Criteria Guidelines for Application”, J.L. Hechmer & G.L.
Hollinger, WRC Bulletin 429 – Februray 1998.
5-The stress selection radio buttons 5,6 and 7 permit the user to force the stress calculation method. By
default, the stress calcualtion method is determined by the Code year used. For 06 and previous versions of
VIII-2 the stress intensity (Tresca) method is the default. For 07 and later versions the equivalent stress (Von
Mises) method is used. The Von Mises, equivalent stress method is considered more accurate, and will often
result in a 10% or slightly larger reduction in calculated stress since many PVP geometries have biaxial stress
states that are sensitive to the stress calculation method.
6-To override the default and force use of Tresca click in the #6 radio button. This is provided so that user’s
can produce conservative results similar to those used in the 06 and earlier versions of VIII-2. The override
will work for any version or year of the Code!
7-To override the default and force use of the equivalent stress click in the #7 radio button. This is provided
so that user’s can see the effect of the equivalent stress reduction in an A06 or earlier analysis. The override
will work for any version or year of the Code!
8-Sometime pressure stresses will subtract from weight stresses such that pressure or weight analyzed alone
will produce higher stresses than both analyzed together. Although this is not common for most nozzle or
lug type geometries, the condition may be significant for saddles or large D/T vessels or pipes. To consider
this possibility, an independent Code calculation can be performed using the weight case. (See VIII-2
5.1.3.2).
9-The load case combinations in VIII-2 Table 5.3 show the wind and seismic factored loads combined with
operating rather than design pressure loads. This is not conservative but is a part of the Code evaluation
procedure. The default is to use the design pressure input with the occasional loads entered when performing
the Code compliance. Per Table 5.3 only the operating pressures need to be combined with the factored
occasional loads. The user may force this non-conservative requirement to be used by clicking in Checkbox
#9.
10- Guidelines for elastic stress classifications are given in VIII-2 Table 5.6. For nozzle necks, supplemental
requirements are provided in para 5.6. These classifications require that the user determine if loads due to
the free end displacement of the attached pipe are primary or secondary. These loads are usually considered
secondary but have been shown to sometimes have a primary character when the attached pipe is long and
flexible and where elastic followup can produce a load constrolled condition. The load controlled condition
can be thought of as a constant effort spring load. The displacement controlled condition can be thought of as
a variable spring load. The variable spring changes load significantly during displacement, and the load
controlled condition does not. The user can assume the free end displacements produce primary loads and
produce a conservative result. This might be recommended when the attached pipe is long, or when the D/T
ratio is high and the system sensitive to overload, for example if the junction is refractory lined or a prone-to-
leak flange joint is attached to the nozzle.

11-Establish Rigorous Fatigue Load Cases – When intermediate states exist in the operating cycle, the
maximum stress range and cycle count may be based on an intermediate state. The typical single operating
cycle involves the stress change from Weight to Weight+Pressure+Temperature (W+P+T) and then a return
to Weight (W). If the intermediate cases W+P, and W+T may exist, i.e. W W+P W+P+T W+T
then the rigorous load case option should be investigated to be sure that maximum fatigue stresses are
calculated. For a single pressure and temperature condition there is a maximum of two cycles that can be
developed from the permutations and when this box is checked. An analysis of the worst permutation is
calculated and the maximum fatigue damage ratio conservatively calculated and printed for each region in
the model. This checkbox should be selected when pressure can exist in an operating state without
temperature and/or temperature can exist in an operating state without pressure and the fatigue analysis is a
key part of the design.
12-In VIII-2 Table 5.3 the operating pressure is assumed to be 90% of the design pressure. This may not be
the case and the user may enter the actual pressure range used in the text cell identified by number 12. The
user may enter 1.0 in this cell to have the design pressure used in all cases where pressure is included in the
analysis. 0.9 is the default. This cell interacts with checkbox #9.
13-The A07 and later versions of VIII-2 have two fatigue methods. The smooth bar method has been in the
Code for many years. The welded fatigue curve method was introduced in 07. A detailed discussion of the
welded fatigue curve method is given below and the user can choose between them. To use the welded
fatigue curve method the analyst must get the approval of the owner. There are several other options which
are extensions of the Code welded fatigue method. These are PRG girth weld and tee welded curves. They
were developed to be used with other parameters in the welded fatigue curve method. Additionally, the user
can provide their own welded fatigue curve parameters. These options are described in more detail below.
14-18-The #14 radio button should be selected to use the new welded fatigue curve method. The welded
fatigue curve method can be conservative in the low cycle range (10,000 cycles), but non-conservative in the
high cycle range, (>100,000 cycles). User’s should employ the welded fatigue curve method with caution
until more experience is gained with its’ use. The welded fatigue curve provides 3 bases for developing the
allowable stresses from the mean failure curve. Each of these methods provide for 1, 2 or 3 standard
deviation shifts from the mean curve. The typical standard deviation on Log(N) is about 0.25 for the welded
fatigue curve tables according to the standard deviation differences given. The mean failure line is not
recommended for design purposes but can be used to get an estimate of a components mean life to failure.
19-The Ramberg Osgood material model is used with the welded fatigue curve to provide the Neuber
correction described by Eqs. 5.53 and 5.54. The coefficients recommended for the Neuber correction are
given in Annex 3.D.2 and can be entered by the user in text cells 19 and 20. When using the welded fatigue
curve the user is recommended to verify the defaults entered for carbon steel. The units for the strain
hardening coefficient should always be ksi. The use of the Neuber correction is discussed in some detail
below. The user can deactivate this conservative modification to the calculated strain. PRG and numerous
other tests of welded components do require this correction. The Neuber correction can be used with any
welded fatigue curve. It is thought that the increase in strain predicted by the Neuber correction is offset by
the change in slope of the cycles vs stress range curve in the low cycle range. The change in the slope occurs
due to the level of plasticity and its’ effect on fatigue behavior. (See Tagart and Manson papers.)
20-The environmetal factor of 4 on fatigue life has been historically used. Fatigue tests in air have factors of
around 1.1. Water environments can be 1.3 or more if corrosion is allowed to form on crack tips in the
presence of high oxygen concentrations in the liquid. Unless other information is available, the factor of 4
on life is recommended.
21-The strain hardening exponent is a component in the Ramberg Osgood material property used with the
welded fatigue curves. The Ramber Osgood equation is:
e = σ/E + 2[σ/(2xK)]css
where E is the material modulus of elasticity, σ is the stress, xK is the strain hardening coefficient, and css is
the strain hardening exponent.

22 – The welded fatigue curve method permits the user to identify methods by which the fatigue life is
improved. There are three methods the user may select from the Code options, or the user may specify their
own fatigue improvement constant override. The fatigue improvement is a factor on life. The Code defined
methods available are: burr grinding, tig dressing and hammer peening. Each is described in VIII-2 Annex 3.
28-29 – There are two material curves available for the VIII-2 welded fatigue method. The default is for
ferritic and stainless steels. Alunimum may also be chosen. The Master S-N curve method gives the same
welded fatigue properties for stainless and ferritic steels.
30 – This checkbox is used to exclude the Neuber correction. The Neuber correction only effects stresses in
the cycle range lower than about 6000 cycles. When the Neuber correction is omitted, equations 5.53 and
5.54 are left out of the structural stress computation process and the elastic strain is used for equation 5.55.
31 – This checkbox causes a welded fatigue curve algorithm to be used with the welded PRG tee fatigue
curve. The basis of this curve is a four parameters fit to PRG’s best estimate of welded tee fatigue data that
includes both piping and tubular joint data. The four parameters fit used is:
N = [(S/C) (Tref/T)k] A1
This is the allowable equation used in FE/Pipe for girth butt welds and unreinforced fabricated tees.
The “S” to be used in the correlation is M+B stress at the toe of the weld. The M+B stress can be
modified by Neuber based on the ASME optional screen. Poisson’s correction per VIII-2 5.55 is
omitted.
UFTs excluding UFT All Tubular

Item Butt Welds
Wilde & Markl and Piping Data
Cmean 2755.1 2461.6 1999.755
C (-1 stdev) 2053.17 2024.7 1530.915
C (-2 stdev) 1530.02 1665.4 1171.994
C (-3 stdev) 1140.36 1370. 897.221
Tref 0.7874 0.26 0.21727
k 0.25 0.23452 0.27727
A1 -2.994 -3.1763 -3.4364
Membrane and bending corections (Iβ) using VIII-2 5.61 and 5.62 are not included when PRG methods are
used included. Mean stress corrections, fM, environmental factors fE and improvement factors fI are used
with the PRG methods as recommended in VIII-2.
32 – This checkbox causes a welded fatigue curve algorithm to be used with the PRG girth butt welded
curve. The basis for this curve also involves four parameters in the fit. Girth butt welds are not as apparently
strong as tee welds when subject to the same maximum membrane+bending stress since the stress gradients
along the girth butt weld are weaker, i.e. in a girth butt weld, the high stress does not dissipate as quickly
along the weld as it does in the typical tee joint. The use of girth butt weld curves is more conservative than
tee curves.
33 – This checkbox causes a welded fatigue curve algorthm to be used with user’s entered parameters. The
user may only enter C and h, such that the fatigue calculation is based on the equation:
N = fI/fE[ (Eavg/Ecold) C ] (1/h)
A thickness correction is not used for the user defined curve. There are no adjustments for standard deviation
when a user defined curve is entered. (The standard deviation is not known.) The user must enter a mean

curve if a failure analysis is being performed, or a standard deviation or “safety factor” shifted curve if a
design analyis is being performed.
Note that ASME Section VIII Division 1 uses the hot allowable stress for SPS, while Division 2 uses the
average of the hot and cold material properties for SPS.
Occasional Loads
The 06 and earlier VIII-2 used the local membrane stress allowable 1.5(k)Sm and k=1.2 for loads developed
from primary and occasional cases such as wind, wave and earthquake. FE/Pipe permits the user to apply
wind and seismic loads to most templates and/or to define occasional loads which act on vessel nozzles. In
previous versions of the Code the allowable for the local primary membrane stresses developed from these
loadings was 1.5(1.2)Sm. The 07 and later versions of VIII-2 utilize the load factor approach to primary
loads outlined in ASCE 7. The load cases to be used are defined by Table 5.3 of VIII-2 Part 5 reproduced
below.
Inputs for Elastic Analysis of “Occasional” Loads

The following is a comparison of “occasional” load evaluations for codes commonly addressed by
NozzlePRO and FEPipe users, and load case 5 of the 2007 ASME Section VIII Division 2.
ASME B31.3: [302.2.4(d)] D+Pd+Occ < 1.2S

ASME VIII Div 1: [UG-23(d)] D+Pd+Occ < 1.2S
ASME VIII Div 2: [Table 5.3] D+0.9Pd+Occ < S
Where: Pd = MAWP or “Design” Presure

D = dead loads (weight)

Occ is discussed below
S = Sm for general primary membrane and 1.5Sm for local primary membrane
For a given basic allowable stress, the ASME VIII-2 load combination is between 10 and 20% more
conservative than ASME VIII Div1 and ASME B31.3 depending on which load component is dominant:
eg: if stress is dominated by Pd, then
D+(0.9Pd+0.1Pd)+Occ = S + 0.1(Pd) = 1.1S < 1.2S (10% lower)
otherwise if stress is dominated by D or Occ, then
D+0.9Pd + Occ = S < 1.2S (20% lower)
In applying FEPipe and NozzlePRO to the detailed analysis of ASME B31.3 components or ASME Section
VIII Division 1 components subjected to “occasional” loads the user may either use the 06 version of Div 2,
or select the occasional load case k=1.2 option with the A07 and later Division 2.
Clarifying the “Occasional” Load components.
Two load components in the 2007 VIII-2 that are traditionally classified as “Occasional” loads are “W”
(wind) and “E” (seismic). The load component definitions in the 2007 edition of ASME Section VIII
Division 2 are strongly tied to the 2005 Edition of ASCE-7.
This is an important point for seismic loads for two reasons: (1) because ASME Section VIII Division 1 and
ASME B31.3, and earlier editions of ASME Section VIII Division 2 are tied to earlier (pre 1998) editions of
ASCE-7; and (2) there was a change in methodology in 1998 to seismic loads using the ASCE-7 LRFD
structural design that resulted in larger calculated load components. The seismic load calculated in the 1998
and newer editions of ASCE-7 is not directly compatible with allowable stress design. The equivalence
between seismic loads is as follows:
Occ = E(ASCE-7,1995 and earlier) = 0.7E(ASCE-7,1998 and newer)
“Occ” is the FEPipe and NozzlePRO input load. This equivalence applies to distributed loads (g-loads) and
nozzle or clip reaction loads. Where other seismic codes are used, the user should review the ASD load
combinations for the load case closest to load case 5 in table 5.3 of ASME VIII Division 2 (shown above). If
the load combination uses “D + E” then input the seismic load “E” as calculated by the code. However, if
the load combination uses “D+0.7E” or similar combination, it is implied that the FEPipe and NozzlePRO
occasional input is “0.7E”.
For wind loads “W”, this distinction is less significant than for seismic loads. In ASCE-7, there is an
optional load reduction factor “Fd” that can be applied to“W” when used in the load combinations above.
The intent is to provide a general load for evaluating the overall load distribution in a structure. For local
evaluations, say the strength of an individual panel, or for extreme events and load combinations not required
by ASME Section VIII Division, the code’s intent is to set Fd=1. The windloads calculator available from
the FEPipe main screen defaults to Fd=1, but allows user input and provides a selection of ASCE factors for
different structures.

Equivalent Stress
The 07 and later VIII-2 uses the equivalent stress instead of the stress intensity which was used in all prior
versions. The equivalent stress in many cases will be lower than the stress intensity by 10% or more. The
two are the same for the uniaxial tension test but vary by up to a maximum of 15% when more than a single
stress component is relatively large, for example, in the case of a biaxial stress state in a cylindrical shell.
The cylindrical calculation below demonstrates:

Stress State due to Pressure in a Cylinder
Sx = PD/2t, Sy=PD/4t, Sz=0;

SI = (2)(PD/2t – 0)/2 = PD/2t
SE = (1/2)0.5 [(PD/2t-PD/4t)2 + (PD/4t)2 + (PD/2t)2]1/2
= (1/2)0.5 [ (PD/4t)2 (1+1+22) ]1/2
= (1/2)0.5 (6)1/2 (PD/4t) (2/2)
= (1/2)0.5 (6)1/2 (1/2) (PD/2t)
= 0.866 (PD/2t) … Equivalent stress is 13.4% lower than stress intensity for pressure stress in cylinder.
The stress intensity is twice the maximum shear stress:

Stress Intensity = SI = (σ3 – σ1)(2/2) = (σ3 – σ1)
The equivalent stress is the octahedral shear stress, or Von Mises stress and is expressed as.
Equivalent Stress = (1/2)0.5 [(σ1-σ2)2 + (σ2-σ3)2 + (σ3-σ1)2 ]0.5.
The fact that a 10% reduction in stress is obtained when using the 06 and later versions of VIII-2 is
significant from a weight or occasional external load point-of-view. Since sustained loads are limited by
1.5Sm, and since pressure often consumes the majority of the allowable, very small external loads are
allowed. In a case where the design pressure stresses are 95% of the allowable for a particular nozzle, 5% of
the allowable remains for external sustained loads. If the pressure stresses are (0.9)(0.95) when the
equivalent stress is used for the calculation, then the allowable sustained or occasional external loads
increase by (1.0-(0.9)(0.95))/0.05 = 2.9 times!
Fatigue Analysis and Cycle Counting
Most PVP load conditions that involve changes from the installed case to more than a single load condition
involve non-proportional loading. Non-proportional loading cycles should not be counted using rainflow
methods.
FE/Pipe, NozzlePRO, etc. provide straight forward methods to perform a rigorous fatigue analysis of
multiple combinations of a cycling single pressure and temperature load. Where multiple pressure and/or
temperature conditions cycle some alternate simplification must be employed.
A rigorous analysis is generally required when the system may cycle from the installed condition to the
pressure only case, without thermal loads, or to the thermal only case, without pressure loads, when the
major operating load case is installed+pressure+temperature (W+P+T). When these intermediate load
conditions exist, it is possible that cycling from W+P to W+P+T produces a larger stress cycle than cycling
from W to W+P+T. This concept is illustrated in the figures below.

Looking at the above figure, cycling from b-to-c certainly results in a larger stress range than cycling from a-
to-c. A number of analysts run the operating case only and assume that it produces the largest stress range.
As can be seen from the diagram above, when there is an intermediate stress condition (b), this only occurs
when the stress at “a” is approximately equal to the stress at “b”.
In the figure below, the stress tensor component in the X direction in the weight case (point b) is of the
opposite sign than the stress tensor component in the X direction in the weight+pressure case (points c and
e). In the weight case the tensor value is negative, while in the W+P case the value is positive. Since the
operating stress tensor component in the X direction is negative, the largest range is from the W+P case and
not either the W case or the zero load case.
It is clear that this condition is only possible when:
1) The weight+pressure condition can exist independantly from the weight+pressure+temperature condition.
2) The weight+pressure tensor value is relatively high, (which is not uncommon).
3) The sign of the W+P tensor must be different from the W+P+T tensor.
The pressure stresses can easily be large when compared to the operating stresses since only the membrane
stress due to pressure is limited to 1.5S. As a minimum, the membrane stress tensor could be 1.5S in the
opposite sense of the W+P+T tensor. (The bending stress due to pressure at intersections or other
discontinuities is classified as a secondary stress category and are often inadvertantly combined with
temperature, ignoring the possibility that the pressure and temperature bending stresses may subtract from
each other, which can produce a not-conservative range calculation.)
A possible worst case scenario is described below:
Worst Case Rigorous Fatigue Load Case
In the figure below, the W+P case is shown to have a greater stress absolute magnitude than the W case
(|c|>|b|), and the distance from c-to-d is greater than the distance from b-to-c. In this case, if W+P is a step
along the possible load history to W+P+T, the worst stress range occurs from W+P to W+P+T rather than
from W to W+P+T.

For the worst case scenario, assume that the pressurized ambient condition is a part of the start-up procedure.
If the analyst assumes that only the operating condition alone produces the stress range for fatigue, then only
the stress range from a-to-d would be evaluated.
If the analyst assumes that only the operating to weight condition produces the stress range for fatigue, then
only the stress range from b-to-d would be evaluated.
As can be seen in the above figure, each of these assumptions would miss the maximum cycle range from c-
to-d.
To evaluate the maximum error in the cycle range evaluation, assume that the stresses at the point of interest
have only a single stress tensor component so that visualization of the effect is easier.
Assume that the weight stresses are negligible and so about 0, i.e. σW = σb = 0. Assume that the pressure
stress is equal to 1.5S, which is the membrane sustained allowable, so σW+P = σc = -1.5S. Since the stress at
the operating case can equal 3S, the temperature contribution can be equal to a maximum of 3S, i.e. σD = 3S.
The error that occurs in the stress range calculation by ignoring the separate W+P case is:
(1.5S+3S) / (3S) = 1.5 = 50%.
If there is no separate weight+pressure condition, and the pressure is increased during the startup along with
the temperature, then points “c” and “e” in the above load history are removed and the conditions that
produce the increased stress range are removed, i.e. 1) The W+P case is not a separate stress case, and 2) The
W+P case major stress tensor components have a sign different than the W+P+T components.
In the b-c-d-e-f cycle there are two ranges to be calculated per VIII-2 5.B.5.2: c-d and e-f. Since they each
occur at the same number of cycles, the damage factors for each would be added together and the result
should be less than 1.

Separate Load Conditions
The diagram below shows the largest number of differences that can occur during any single W+P+T
condition:
The possible stress states that can exist during the stress range are:
1) Weight (W)
2) Weight + Pressure (W+P)
3) Weight + Pressure + Temperature (W+P+T)
4) Weight + Temperature (W+T)

Because there can be three random stress tensors the possibility for non-proportional loading is high, since
the weight, pressure and temperature loads all produce unrelated stresses on most PVP geometries, i.e. the
temperature stress tensor is not equal to some multiple of the pressure or weight stress tensors.
Two valleys and one peak, or two peaks and one valleys are the worst possibilities of the above combinations
of load case during a single startup. (During startup and shutdown, the system must pass through a W, W+P,
W+P+T and W+T stress state for this number of peaks and valleys to occur.
Note that there can be independent load cases that are W-to-W+P that are independent of the startup and
shutdown cases, and these will be discussed later.

Example
Startup-Shutdown Load Cycle (700 cycles)

W W+P W+P+T W+T W
Operating Load Cycle for Batch Operation (29,200 cycles)

W+P W+P+T W+P
To evaluate these cycles, the user would check on the Rigorous Load Case Cycle Development in the ASME
options spreadsheet.
Select any A07 or later Div 2 Code.
The output will have the following form and is developed for the 700 startup and shutdown cycles.
Rigorous Fatigue Analysis Output Report Example
For the startup-shutdown cycles, the highest two unity factors are selected that may exist and added together.
This represents the two largest possible cycle arrangements for the possible ranges that could exist in the
startup-shutdown, i.e. W+P, W+P+T, W+T. The damage is calculated for the number of cycles entered,
which for this example is 700.
The W+P to W+P+T cycle occurs 29,200 times. For the inner point, the damage for this range is:
0.196703 x 29,200 / 700 = 8.2
This damage factor is much too high and would certainly cause a problem.

Note that the welded fatigue development method is based on the membrane+bending stress at a section and
so there is no differentiation between inner and outer stresses.
For the above example, if the wall thickness of the vessel was increased and the following report was
regenerated:
Rigorous Fatigue Analysis Output Report Example
The total damage factor for the 700 operating cycles and the 29,200 W+P W+P+T cycles for the inside of
the vessel in the Header Next to the Nozzle is:
0.179715 + 0.062535 x (29,200/700) = 0.179715 + 2.6086 = 2.788
The batch stress cycle is still too high.
Proportional and Non-proportional cycle counting.
Most cycles that must be counted in a pressure vessel or piping application are non-proportional and so
cannot use rainflow or similar counting approaches. In these cases, very simple, but very conservative
methods may be used, or the Max-Min Cycle Counting method can be used per VIII-2 Annex 5.B. For the
simple case, the equivalent stress from all cycles are added to each other. The allowed cycles is calculated
from the result equivalent stress summation and compared to the stress with the largest number of cycles. An
example is given below.
Annex 5.B gives the following definitions for Proportional and Non-Proportional Loading.

“5.B.2(c) Proportional Loading – During constant amplitude loading, as the magnitudes of the applied
stresses vary with time, the size of Mohr’s circle of stress also varies with time. In some cases, even though
the size of Mohr’s circle varies during cyclic loading, if the orientation of the principal axes remains fixed,
the loading is called proportional.”
An example of proportional loading is the load cycle that occurs when the thermal+pressure loads are added
simultaneously to the weight loads. In this case the pressure+thermal loads increase from ambient to their
design values simultaneously and then return to the weight loads. This is illustrated in the following figure
which shows the maximum stress in the X direction for this condition. From the weight only condition to the
W+P+T condition the principal axis range does not change condition.
At point “a” the pipe is in a state of zero stress – still in the truck, or supported uniformly by temporary
supports. A nozzle at point “a” would be essentially unbolted so that stresses are zero or negligible. Point
“b” is the weighted stress condition. At the point being evaluated, the weight stress is negative in the X
direction. Since the loading is proportional, the final condition is weight+pressure+temperature, but to get
from the weight condition to the W+P+T condition a stress tensor is added that at each point in between W
and W+P+T the added tensor components are scalar multipliers of each other.
If it is not clear that proportional loading may be used, then non-proportional loading should be used.
When it is possible that weight+pressure cases can exist without temperature or that temperature can exist
without weight+pressure, then there may be a stress in between the weight+pressure+temperature case that
constitutes a larger stress range during the startup. This is demonstrated below. In each of the figures the
W+P+T stress is the same, but the intermediate case introduces a greater stress state. In the past, when WRC
107 was the only tool available and pressure stress components were not well defined, the thorough
evaluation of intermediate stress states was difficult.

A08 VIII-2 Code Applicable Sections
The A08 VIII-2 is divided into nine (9) Parts and associated Annexes. Some of the annexes are normative
and mandatory, others are informative and considered good practice. The Parts of most interest are Part 4 –
Design By Rule Requirements, which provides requirements for design of vessels and components using
rules, and Part 5 – Design by Analysis Requirements, which provides requirements for design of vessels and
components using analytical methods.
A summary of important items in the now mandatory A08 VIII-2 are collected from throughout this
document and listed below.
1) For VIII-2 stamped equipment, Part 5 can be used to design thinner components than by Part 4.
2) VIII-2 Part 5 rules are not permitted in the creep regime unless previous experience with similar
equipment is available as mandated by the applicable fatigue exemption rules. “Creep regime” is where
Section II-D material allowable stresses indicate that they are governed by time dependent properties.
3) New VIII-2 provides allowable stresses in the creep regime.
2Sy is only greater than 3Sm in lower temperature cases (ie closer to ambient) when the margin against UTS
governs. The benefit of 2Sy instead of 3Sm is greatest at ambient and begins to decrease as the temperature
increases.
Difference in allowable for new VIII-2 is only that the margin against UTS is 2.4 instead of 3.0 in previous
editions. Also, temperatures exceed those in previous editions (ie they extend into the creep regime for many
materials).
Notes from the paragraphs of Part 5 are included below.
Part 5 – “design-by-analysis methodology” provides procedures for design using the results from an
experimental stress analysis and for fracture mechanics evaluations. Since wording from the A06 VIII-2 has
been changed, some interpretations are different.
5.1.1.2 … If multiple assessment procedures are provided for a failure mode, only one of these procedures
must be satisfied to qualify the design of a component.
Failure modes:
a) All pressure vessels … irrespective of size or pressure shall be provided with protection against
overpressure …
b) Protection against Plastic Collapse ..
c) Protection Against Local Failure. … It is not necessary to evaluate the local strain limit criterion if the
component design is in accordance with Part 4.
d) Protection Against Collapse from Buckling …
e) Protection Agaisnt Failrue from Cyclic Loading …
5.1.1.3 The design-by-analysis procedures in Part 5 may only be used if the allowable stress in Annex 3.A at
the design temperature is governed by time-independent properties unless otherwise noted…. If the
allowable stress from Annex 3.A. at the design temperature is governed by time-dependent properties and the

fatigue screening criteria of paragraph 5.5.2.2 is satisfied, the elastic analysis procedures in paragraphs 5.2.2,
5.3.2, 5.6, 5.7.1, 5.7.2 and 5.8 may be used.
5.1.2.3 Recommendations on a stress analysis method, modeling of a component, and validation of analyis
results are not provided. … However, an accurate stress analysis including validation of all results shall be
provided as part of the design.
5.1.2.4 … material properties … shall be determined using data and models in Part 3 for:
a) E, alpha, conductivity, diffusivity, density, Poisson’s ratio,

b) allowable stress, yield and tensile strength,
c) monotonic stress-strain curve,
d) cyclic stress-strain curve – stabilized true stress-strain amplitude curve.
5.1.3.1 … An overview of the supplemental loads and loading conditions that shall be considered in a design
are shown in Table 5.1. Table 5.1 includes wind loads and pressure testing loads as well as piping loads (but
these are “to be considered”). Additionally, 5.1 notes, “Normal Operation plus Occasional … note:
occasional loads are usually governed by wind and earthquake, however other load types such as snow and
ice loads may govern, see ASCE-7).
5.1.3.2 … Typical load descriptions are provided in Table 5.2. Load case combinations for elastic analysis
(Table 5.3) limit loads (5.4) and plastic analysis (5.5) … In evaluating load cases involving the pressure
term P – the effects of the pressure being equal to zero shall be considered.
5.1.3.3 If any of the loads vary with time … (b) When creating the histogram, the history to be used in the
assessment shall be based on the anticipated sequence of operation. (c) Applicable loadings such as pressure,
temperature, … nozzle reaction loadings, … [shall be considered]. (d) The relationship between the applied
loadings during the time history …[shall be considered].
5.2 Protection Against Plastic Collapse.
Three methods are given: Eastic, Limit Load, Elastic-Plastic.
5.2.1.2 For components with a complex geometry and/or complex loadings, stress classification may be
difficult. In these cases limit loads or elastic-plastic analysis methods in para 5.2.3 and 5.2.4 are
recommended.
Pre-2007 versions of VIII-2 dealt with occasional loads by permitting the primary load case that included
them to have a higher allowable by 20%. Post 2007 versions of VIII-II utilize the load factor approach.
The program user must prescribe the occasional component of the load to be combined with the weight and
pressure loads. Using Table 5.3 Load Case Combinations:
5) 0.9P + Ps + D + (W or 0.7E) < S

6) 0.9P + Ps + D + 0.75L + 0.75Ss
7) 0.9P + Ps + D + 0.75(W or 0.7E) + 0.75L + 0.75Ss
0.9P = Operating load. (Program user can enter factor)

Ps = Static head
D = Dead Weight
W = Wind Load
E = Earthquake Load

L = Live Load
Ss = Snow Load
The user should see ASCE 7 for precise definitions of wind, earthquake, live and snow loads. For use with
the program, the user should enter the occasional load as any component of the load case that does not
include 0.9P+Ps+D. For example, for case 5 above, the user should enter either W or 0.7E for the occasional
loadings.
Sections of this manual above give more guidance for use of AISC 7 and VIII-2.
5.2.1.3. The user of elastic stress analysis combined with stress classification procedures to demonstrate
structural ntegrity for heavy-wall (R/T<=4) components (especially around discontinuities) may produce
non-conservative results and is not recommended.)
5.2.2.1 (b) The maximum distortion energy yield criterion shall be used to establish the equivalent stress:
Se = (1/2)0.5 [(S1-S2)2 + (S2-S3)2 + (S3-S1)2 ]0.5.
There are common instances where the equivalent stress is different than the maximum shear stress intensity
used previously. The difference can be a maximum of about 15%. The equivalent stress will be lower than
the stress intensity. When comparing post 06 VIII-2 designs to pre 06 VIII-2 designs, one would possibly
expect stresses in the new designs to be up to 15% lower. The stresses are of different value when there is
more than one high stress component.
5.2.2.2 … Stress limits for the pressure test condition are covered in paragraph 4.1.6.2.
5.2.2.4 (c)(3) Kf = Fatigue strength reduction factor. (Use is noted in this paragraph.) There is considerable
discussion of the Kf term later in this document.
5.3 – Protection Against Local Failure
The evaluation of 5.3 is provided for a hydrostatic stress state where the material will not behave in a ductile
manner due to hydrostatic stresses. These checks are provided as directed by the Code for primary stresses
only. The user may wish to perform this evaluation also for operating cases although the Code does not
require it. Options are available in the ASME Control form to add these additional checks. These checks
might be considered when high strength steels are used, load conditions might exist at cold temperatures, the
material may otherwise behave in a brittle manner, or stresses are particularly high.
5.3.1.1 … The local strain limit criterion does not need to be checked if the component design is in
accordance with the standard details of Part 4.
5.3.2 Protection Against Local Failure using Elastic Analysis. “The sum of the local primary membrane
plus bending principal stresses shall be less than 4S for each point in the component.”
5.4.1 Protection Against Collapse From Buckling. These requirements apply to all components where the
thickness and configuration is established using design-by-analysis rules and the applied loads result in a
compressive stress field.
The program user must determine if a compressive stress field is present. External pressure and primary
loads or “load controlled” loads on large D/T nozzles can introduce conditions where compressive stress
fields that can cause buckling may occur. Elastic bifurcation buckling calculations are permitted by VIII-II
and the required design factors (or load factors) are given in 5.4.1.3. PRG experience with large diameter

ducts or vessels subject to external loads and pressure are that design factors should exceed 6 or more, and
are often greater than 9.
5.5 Protection Against Failure From Cyclic Loading
5.5.1.3 (a) Smooth bar fatigue curves may be used for components with or without welds. The welded joint
curves shall only be used for welded joints.
5.5.1.3 (b) The smooth bar fatigue curves are applicable up to the maximum number of cycles given on the
curves. The welded joint fatigue curves do not exhibit an endurance limit and are acceptable for all cycles.
5.5.1.6 Protection against ratcheting shall be considered for all operating loads listed in the User’s Design
Specification and shall be performed even if the fatigue screening criteria are satisfied. [Even if a fatigue
analysis is not required, Pl+Pb+Q should still be calculated and compared to SPS to be sure that ratcheting of
the component does not occur.]
5.5.3 Fatigue Assessment – Elastic Stress Analysis.
5.5.3.1 (a) The controlling stress for the fatigue evaluation is the effective total equivalent stress amplitude,
defined as one-half of the effective total stress range (Pl+Pb+Q+F) calculated for each cycle. Where shell
finite element are used, the approach in 5.A.7 will be used where:
Pl+Pb+Q+F = Salt,k = (Pl+Pb+Q)(Ke)(Kf)/2
Pl+Pb+Q as calculated will include all thermal effects defined for the load case, (and/or) the range of all
thermal effects defined for the load case, including through thickness effects.

General Discussion of ASME VIII-2 5.5.5 “Fatigue Assessment of Welds”
There are two elastic fatigue evaluation methods in the A07 and later VIII-2:
1) Polished bar (5.5.3)

2) Welded Fatigue Curves (5.5.5)
A significant part of this document is dedicated toward explaining the use of welded fatigue curves for both
the vessel and piping engineer since the piping engineer is directed to ASME Section VIII Division 2 Part 5
in paragraphs X04.7.2 in B31 Code sections.
Ev Rodabaugh in NUREG 3243 showed how the polished bar method compared favorably with the Markl
fatigue curve using the plastic correction factor Ke. Ke is used when the secondary membrane + bending
stress (Sn) exceeds 2Sy and adjusts the polished bar curve in the range below below about 10,000 cycles.
Rodabaugh showed that with the use of Ke and a Kf=2 for welds, the polished bar curves match the Markl
girth weld failure prediction. Use of Ke with elastic fatigue analysis has been proven to be conservative
throughout the years as emphasized most recently by Hinnant in PVP 2008-61871. The Hinnant curve for
girth butt welds is shown on the Fig. 4 NUREG 3243 curve in Figure 1 below and shows that the Ke
adjustment is conservative in the low cycle range below about 2000 cycles. (The production of the welded
fatigue curve is demonstrated in the “Case 1 Comparison” below.)
The ASME VIII-2 5.5.5 welded fatigue curve is included in red in Figure 1 below and can be seen to be
reasonably similar to the polished bar curves when the polished bar curve is modified by Ke in the region
below about 3000 cycles. There are a number of parameters which affect the design use of the welded
fatigue curves and which may result in different predicted fatigue lives, but for the mean failure prediction of
girth butt welds, the Markl, polished bar (adjusted by Ke, and with Kf=2), and the welded fatigue curve
methods appear to agree. Hinnant and others have shown that for many welded geometries the Ke adjustment
for girth butt welds and tees is not needed. (See BS7806 and Hinnant-Paulin curves below.)
Figure 1 – Comparison of Fatigue Methods (Modification of Fig. 4 in NUREG 3243)

If Hinnant-Paulin tests, BS6806, IIW and others had all agreed that a change in the slope of the welded
fatigue curves should occur at between 2000 and 10,000 cycles, or if there was an agreement on the general
slope of the curve, (-0.3 instead of -0.2), there would be little reason for designers to delve into the
construction or meaning of the above curves. Since there is a difference in the slopes and/or possible slope
changes, design guidance may be considerably different below 2000 cycles and/or above 100,000 cycles. As
can be inferred from Figure 1 above, below 100 cycles, or above about 3 million cycles, these errors will
increase the design probability of failure from about 0.5% to 50%. Between 100 cycles and 2000 cycles, and
between 100,000 cycles and 3 million cycles, the design probability of failure will vary logarithmically
between the intended value of 0.5% to a real value of 50%.
For the construction of Figure 1 above, the mean stress range to failure for the polished bar curve is given by
the expression:
S(mean alternating stress to failure in polished bar) = 8664N-0.5 + 21.6 ksi
This is based on the ASME tests and the equation form: (E/4)Loge(100/100-A)N-0.5 + B
where the area reduction in the tensile test (A) is 69% and E is 29e6 psi.
The allowable alternating stress on the polished bar (Sa) is established by applying the greatest shift in the
mean curve using 20 on cycles or 2 on stress. As can be seen in the plot below, 20 on life governs below
about 50,000 cycles, at which point 2 on stress governs. The lowest value from either of the red or black
lines in the figure below is for the range of allowed stress (Sa).
2Sa is compared in the figure below to 490,000N-0.2. This is the basis for the B31.3 and VIII-2 Polished Bar
curve comparisons. When polished bar curves are used with welded specimens a fatigue strength reduction
factor (FSRF or Kf) is applied. Developing relations between notched and unnotched specimens using
FSRF’s (K2 ≅ Kf factors), has been demonstrated by Neuber, Topper, Wetzel and Morrow since the 1960’s.
K2 is the ASME III NB terminology and Kf is the ASME VIII-2 terminology. There are potential differences
between K2 and Kf, but for the discussion here they can be considered equal.

2Sa is the range of allowed peak stress from the polished bar curve (Ke adjustment is not considered in 2Sa.).
490,000N-0.2 is Markl’s determined allowed nominal stress range in a girth butt weld. Since the nominal
stress in the polished bar should be multiplied by a K2 = Kf factor of 2.0 to reach the range of peak stress at
failure in a notched specimen, a comparison between 2Sa and 490,000N-0.2 is appropriate. The K2xSn curve
includes the Ke adjustment! The rationale behind the comparison is summarized below.
Allowable Alternating Stress for Polished Bar = Sa.

Allowable Stress Range for Polished Bar ≅ 2Sa.
Markl Allowable Stress Range: (M/Z)(i) < 245,000N-0.2
Section III Allowable Stress Range (approx): C2K2(M/Z) < 2Sa
Since i = C2K2/2
(M/Z)( C2K2/2) = 245,000N-0.2

(M/Z)( C2K2) = (2)(245,000N-0.2) = 490,000N-0.2
The comparison is 2Sa vs. 490,000N-0.2
The comparison of 490,000N-0.2 and 2Sa was thought favorable (NUREG 3243) until PVP 2008-61871 (HP)
showed that the Markl curve was likely too low in the cycle range below about 10,000 cycles. The
comparable Hinnant Paulin Eq. 3 stress range is added to the above figure in the sketch below and the cycle
range cutoff at 10 cycles. As can be seen, the Ke adjustment and the 490,000N-0.2 curve apppear to produce
overly conservative results. Prior to the PVP 2008-61871 paper it was thought that the Markl curve should
not match BS, EN, AASHTO, IIW or other welded fatigue codes because the other welded fatigue codes did
not include the strain concentration effect that exists in a piping cantilever low cycle test. (See Scavuzzo-
Rodabaugh.) When the Hinnant-Paulin test results (cantilever low cycle), matched the BS, EN, AASHTO,
and IIW tests instead of the Markl results, it was realized that:

1) The low cycle Markl results are likely too low because of either test details or poorer quality
1940-50’s circa welding when high plastic strains are involved.
2) The Ke factor is likely not needed for welded geometries because at the same time an increase in
strain occurs due to plasticity (Neuber, Ke) because an increase in fatigue strength around notches
also occurs (Kf 1) and these effects tend to cancel for certain geometries. There will be more
provided on this discussion and the figure shown above will be expanded on considerably.
Much information surrounding this topic was presented in PVP 2008-61871. A Ke or similar adjustment is
not made in the BS5500, EN13445, AASHTO, IIW, or BS7608 welded fatigue Codes. The Neuber
correction used with the welded fatigue curve in A08 VIII-2 is seen to produce an adjustment in certain cases
that is identical to Ke, and therefore is likely also not needed for PVP type welded fatigue curve evaluations.
The VIII-2 5.5.5 welded fatigue curve is added in the figure below showing that both the polished bar and the
welded fatigue curve linear methods in the ASME VIII-2 Code approach agree when evaluating mean failure
of girth butt welds, but as seen above, both are conservative in the low cycle range for at least girth butt
welds and likely for other components. (Hinnant and Paulin suggest that the approach is conservative for
Markl style loading of tees also, except that fatigue failure of tees is far more complex than the failure of
girth welds in straight pipe.) The conservatism can be seen to start at around 2000 cycles for the Ke
adjustment and at around 10,000 cycles for the Markl curve although the error is not too significant below
about 5000 cycles. Since most B31 piping is not designed below 3125 cycles the conservatism is not thought
significant from a design point-of-view. Problems arise however, when SIFs are developed in the low cycle
range because of their reliance on the accuracy of the girth butt weld curve. (The SIF is the ratio of the stress
to cause failure in a girth butt weld at “N” cycles to the stress to cause failure in the component of interest
also at “N” cycles.) Additional problems exist in the high cycle range because of the difference in slopes and
the divergence of the predicted mean failure lines by the different methods. The high cycle (>1e6 cycles) is
not discussed in detail here.
The Hinnant-Paulin curve is given by Eq. 3 in PVP 2008-61871 by:
S = 1895N-0.3363 ; nominal stress range from M/Z in ksi.
NUREG 3243 contains the same comparison in its Fig. 4.

The Welded fatigue curve with the Neuber correction is included above and can be seen to follow the 2Sa
curve with the Ke adjustment. The grey curve in the above plot is the HP girth butt weld test plot. The
Neuber correction drops the curve below the test data starting at around 5000 cycles. Markl’s girth butt
welds, the Ke adjustment and the Neuber adjustment agree, but do not match the Hinnant-Paulin test data –
or other welded fatigue curve test data. The stress range acting on a girth butt weld is plotted along the
ordinate in the above plots.
In general the stress range or the amplitude is included on fatigue plot ordinates. Readers must be careful
when evaluating these stresses. For the VIII-2 (5.5.5) welded fatigue curves there is a “structural stress”
definition that is different from the applied nominal stress definitions. In the low cycle range there are
linearized stress definitions and there are εE definitions, where εE is the strain measured in the test times the
modulus of elasticity. The strain (ε) can also be estimated from the test using calculations.
The welded fatigue curve without the Neuber correction follows the Hinnant test data very closely. Options
are provided in the PRG A08 VIII-2 implmentation to leave the Neuber correction out of the welded fatigue
curve method. This is not a Code recommendation, but can be used when studying actual fatigue failures
and artifical modifications are not needed.
The Ke factor equation is given in VIII-2 Eqs. 5.31, 5.32 and 5.33. This expression is thought to be
conservative for most geometries. (See Tagart.) The Code establishes a maximum value for Ke of 1/n,
where for ferritic materials n=0.2, so 1/n=5. For the cantilever, the maximum Ke effect due to bending only
is (1+2n)/(3n) = 2.33 although the actual value of the strain concentration factor Ke is a function of load, the
nonlinear material model and geometry.
Ke also includes the Poisson’s effect that occurs when a material undergoes plastic deformation. In this case,
the volume of the material doesn’t change and Poisson’s ratio goes to a value of 0.5. The maximum effect of
Poisson’s effect at the surface for a biaxial stress state is given by Steve Adams as 1.4. For a uniaxial stress
state (longitudinal stress due to bending) there is no Poisson’s effect causing additional strain concentration.
The above comparisons are intended to extend Figure 4 in NUREG 3243 and are not intended to be a
discussion of the welded fatigue method. The welded fatigue curve above is developed from the mean curve
coefficients and adjustments necessary to compare it to the cantilever fatigue test of girth butt welds. The
strength of the welded fatigue method is that it provides a number of modifications for the user so that the

fatigue assessment can be made more accurately. A more accurate assessment reduces the scatter (standard
deviation is smaller), and the 3STDEV shift from the mean produces a higher allowable.
Details of Welded Fatigue Curve Use in A08 VIII-2.
A more detailed discussion of the A08 VIII-2 welded fatigue curve approach (5.5.5) is given below.
The maximum membrane and bending stresses are found at all points in shell models in FE/Pipe and
NozzlePRO and are grouped in regions specific to the model, for example, “nozzle adjacent to junction” or
“header adjacent to junction.” The highest values of the membrane and bending stresses are almost always
at weld lines in models of welded construction. Membrane and bending stresses can also be extracted from
2D and 3D volumetric models using stress classification lines (SCL’s). The membrane and bending stresses
from shell, 2D or 3D volumetric models of an identical geometry under identical loads should theoretically
be the same. In many situations this comparison provides a useful validation of the model and/or an
understanding of the inaccuracy of the method. Brick or “volumetric” models are often thought superior to
shell models of similar geometries, but volumetric models may themselves have difficulties with stress
linearization along lines where differing thicknessese exist on either side of the classification line. In these
cases, lower bound limit loads or elastic-plastic analysis with large rotation and large strain are thought to be
more directly applicable to the load capacity prediction.
Per VIII-2 paragraph 5.5.5.2(c) the membrane and bending stresses should be normal to the hypothetical
crack plane. Since the maximum magnitude of the membrane and bending stresses are found in FE/Pipe, the
method applied in FE/Pipe is conservative insofar as the evaluation of the membrane and bending stresses at
welds is concerned. User’s may generate plots that show the principal stress directions on the surface of the
shell component in both the inside and outside to determine if the approach described above is thought to be
too conservative:

Pressure Mem+Bend Outplane Mem+Bend
Once the membrane and bending stress at the point to be evaluated is determined, the steps below are
followed by FE/Pipe. The user may follow this simplified development in VIII-2 Part 5 section 5.5.5. The
nomenclature used is similar to that used in the A08 VIII-2 document.
5.5.5.2(d): ∆SM+B = ∆SMembrane + ∆SBending = ∆σElastic III-2 Eq. 5.51 Eq. 1

5.5.5.2(e): ∆εElastic = ∆SM+B / E = ∆σElastic / E Eq. 2
Paras 5.5.5.2(d) and (e) compute the elastic stress and strain at the point of interest, defined as ∆σElastic and
∆εElastic.
The structural stress and strain ranges are determined by simultaneously solving two equations using
Neuber’s Rule and a model for the material hysterisis loop stress-strain curve. Note that the Eq. 4 form of
the Ramberg-Osgood equation below begins to produce significant plastic strain after twice yield has been
exceeded. The two equations to be satisfied simultaneously are given below:
∆σElasticPlastic ∆εElasticPlastic = ∆σElastic ∆εElastic Eq. 3
∆εElasticPlastic = ∆σElasticPlastic/E + 2 [∆σElasticPlastic / (2Kcss) ] (1/ncss) Eq. 4

Figure 2 – Neuber Correction
The solution of the simultaneous equations can be visualized as solving for ∆εElasticPlastic in the figures above
where the grey area in the left plot is equal to the grey area in the right plot. These are VIII-2 equations 5.53
and 5.54.
When the equations are solved simultaneously the structural stress range (∆σ) is found from:
∆σ' = (E)(∆εElasticPlastic) Eq. 5
and ∆σ = ∆σ’ / (1-υ2) Eq. 6
Equations 5 and 6 above are equivalent to the Code equation 5.55.
The Neuber correction is similar to the Ke factor used in VIII-2 Code equations 5.31 through 5.33 in that:
1) The Neuber correction becomes significant when the sum of membrane and bending stresses is
greater than about two times the yield stress.
2) The correction results in an increase of the calculated stress due to a plasticity correction.
3) The value (1-υ2) is a plane strain constant, which is independent of material or stress state and so
can be considered a part of the constant C and not a valid part of the stress evaluation procedure
since it is used whether the stress state is axial, biaxial or triaxial.
The Neuber correction given by the simultaneous solution of equations 5.53 and 5.54 and the Ke factor given
by equations 5.31 5.32 and 5.33 are shown in the graph below. The Neuber correction does not have the
same basis as Ke but does serve to enhance the effect of plastic deformation once the secondary stress
exceeds 2Sy. Each is an approximation of this strain concentration effect, and can be shown to effect the
allowable stress in a similar way.
Figure 3 – Neuber Correction Compared to Ke

The Neuber correction shown in Figure 3 is based on the simplified model described below using the elastic-
perfectly plastic evaluation. The 55 ksi comparison is most appropriate when Kcss=105ksi and Ncss = 0.12.
The stress correction factor used in VIII-2 A08 when Kcss=105ksi and Ncss=0.12 is lower than the stress
correction factor implied by Ke of Eq. 5.31-5.33. It is not meant to imply that the Code authors intended to
develop the same factors, only that both factors modify the calculated stress used in the Code fatigue
evaluation when the stress is greater than 2Sy. As can be seen in the curves included at the start of this
chapter, the Neuber correction matches very well the polished bar with Kf=2 and the Ke correction.
The Neuber correction in Equations 5.53 and 5.54 can be simplified by using an elastic, perfectly plastic
material model. For the Ramberg Osgood equation coefficients recommended in the A08 Div 2, the yield
stress for the elastic perfectly plastic model should be 55 ksi. This comparison is shown in the Figure 4
below where the black line is approximating the grey line developed using the Ramberg Osgood equation.
The simplification is not employed in FE/Pipe, but is used here since it can be used to make comparisons of
the method simplier.
Figure 4 – Neuber Correction Comparison to Elastic-Perfectly Plastic Material Model
The Neuber correction and Eq. 5.55 simplified using an elastic-perfectly plastic material model can be
expressed as ∆σ = (Ke’) ( ∆σElastic ), if Ke’ = 1/(1-υ2) when ∆σElastic < 2Sy. and Ke’ = ∆σElastic / [(2Sy)(1-υ2)]
when ∆σElastic > 2Sy. The modified strain from the elastic perfectly plastic Neuber correction becomes:
∆εElasticPlastic = ∆σElastic ∆εElastic / ∆σElasticPlastic Eq. 7

and ∆σ is:
∆σ = [ E/1-υ2) ] ∆εElasticPlastic Eq. 8
To summarize the elastic perfectly plastic simplification:
If ∆σElastic > 2Sy then ∆σ = (∆σElastic)2 / [(1-υ2)(2Sy)] Eq. 9
If ∆σElastic < 2Sy then ∆σ = ∆σElastic / (1-υ2) Eq. 10

When Kcss is 105 ksi and ncss, the strain hardening exponent is 0.12, Sy=55,000 psi gives a good correlation
for the Neuber correction as shown in Figure 4 above.
Once the modified stress (∆σ) is found from the Neuber correction using either the simplification or the
solution of the simultaneous equations, other factors that affect fatigue life are included:
1) Thickness adjustment factor: (Teff/Tb)-0.2222

2) Bending adjustment factor: Iβ
3) Mean stress adjustment factor fM
The thickness adjustment factor is well documented in the literature and reflects the fact that as the thickness
increases, the fatigue life decreases. The base thickness Tb used in A08 VIII-2 is 0.03937 in. (1mm). The
thickness correction factor is shown in Figure 5 below:
Figure 5 – Welded Fatigue Curve Thickness Correction Factor
The adjustment factor (Teff/Tb)0.222 factor is a multiplier on the stress.
A mean stress adjustment is provided in the Code for the welded fatigue curves based on the R factor, where
R is the ratio of minimum to maximum stress: R=σmin/σmax. The adjustment is given in Eq. 5.63 as fM = (1-
R)1/3.6 = (1-R)0.2778. For a fully reversing stress state, fM = 1.0, and for a stress state varying from 0 to a
maximum tensile value, fM = 1. For R values that cycle from positive minimum to positive maximum values
there is an increase in the stress due to the adjustment, i.e. fM < 1, since fM appears in the stress denominator.
The adjustment is only included when σmean > 0.5Sy and when σm+σb < 2Sy. For comparisons with most PVP
test results and for loads that cycle from a very small load to the maximum value, fM = 1.
The bending adjustment in the A08 VIII-2 welded fatigue curve method is given as a function of the bending
ratio Rb = SBending/(SMembrane+SBending). The function is given in Eq. 5.61 and is
Iβ = (1.23 – 0.364Rb – 0.17Rb2 ) / (1.007 – 0.306Rb – 0.178 Rb2).
Rb can vary from 0 to 1, and the function in Eq. 5.61 is given in Figure 6 below.

Figure 6 – Welded Fatigue Curve Bending Stress Factor ( Iβ )
Membrane stresses reduce fatigue life more than bending stress. If the stress at a section is all bending, the
stress that will cause failure increases by about 9% as can be seen in Figure 6.
The stress range calculation for the welded fatigue curve method can now be expressed as shown in Eq. 11
below:
∆σ = (SMembrane+SBending)(Ke’) [ (Teff/Tb)0.2222 ] / [(Iβ)(fM)] Eq. 11
Once the sum of the membrane and bending stress is adjusted as shown in Eq. 11, it can be used to determine
the allowable number of fatigue cycles by Eq. 12 from Annex 3 in A08 VIII-2:
Nf = fI/fE [ (Ea/E) (C/∆σ) ] (1/h) Eq. 12
fI is a fatigue improvement which can be increased by performing grinding, shot peening, etc. fE is an
environmental and surface factor, which is 4 unless some other value can be substantiated.
Ea/E is the average elastic modulus for the cycle under study divided by the modulus used to develop the
fatigue curves.
C and h are constants for the material given in the Code as a function of the number of standard deviations
shifted from the mean of the failure data, and ∆σ is the stress calculated from Eq. 11.
C for the mean of the failure curve for carbon and stainless steels is 2,890,702 psi. and h is 0.3195. Note that
these constants are not found in Annex 3 but can be calculated knowing the standard deviation, given as the
difference between the log of the stress differences for 1 to 2 or 2 to 3 standard deviations and the adjustment
due to the base thickness, which is included in the Code evaluation of the constant “C”, but taken out here so
that the effect of thickness can be clearly demonstrated and unit clarity maintained. To include the thickness
correction in the estimate of C, for example, to compare the VIII-2 method with the Markl recommended
values for C, the thickness correction can be included: Ctadj = (C)(Tb/Teff)0.2222 = (C)(0.03937 in./Teff)0.2222.
For example, the C to be used with Markl’s 4 in. std wall tests would be:
Cmarkl_adj = (C)(0.03937 in./0.625 in.)0.2222. Note that even though Markl used a 0.237 wall pipe, the
thickness correction is constant for tmaterial < 0.625 in. (See excerpt from VIII-2 below, para 5.5.5.2 Step 6.)

So for any Markl style fatigue test where the thickness of the test component is less than or equal to 0.625 in.
(16mm), the C value can be:
Cmarkl_adj = (C)(0.03937/0.625)0.222 = (C)(0.5453) = (2,890,702)(0.5453) = 1,576,317 psi.
The nominal membrane + bending stress can be releated to the mean failure prediction by:
N = (C/S)(1/h). where h=0.3195. This can be adjusted for comparison to the Markl equation for mean failure
in a girth butt weld: S(range) = CN-h = 1,576,317N-0.3195. psi, range.
With this development, the welded fatigue curves can now be compared to other fatigue tests.
For the three standard deviations and mean failure lines the constants “C” and “h” are:
Shift C (psi) h
Mean 2,890,702 0.3195
-1 STDEV 2,411.947 0.3195
-2 STDEV 2,012,000 0.3195
-3 STDEV 1,679,061 0.3195
Case 1 Comparison:

The characteristics of the Case 1 comparison are:
1) Fatigue stress is fully reversing and a membrane stress only, i.e. Sb = 0.

2) Sy = 55,000 psi. for use with the simplified Neuber correction. (This is not the yield stress of the material,
but only the yield stress that best matches the Neuber correction when Kcss=105 ksi and Ncss=0.12.)
3) The plate thickness is 0.625” and so the thickness adjustment factor on stress is 1.9.
4) Since there is no bending stress, the bending adjustment factor on stress is 0.82 per Fig 6.
5) For the fully reversing stress fM = 1.
6) The effective stress is ∆σ = (Sm)(Ke’)(1.9)(0.82)(1.0) = (Sm)(Ke’)(1.56)
7) The number of average cycles to failure for the test condition where fI=1 and fE=1 is:
Nf = { 2,890,702 / [(Sm)(Ke’)(1.56)] }3.13. Eq. 13
The number of cycles where Ke’ > 1 can be found by setting Sm = 2Sy = (2)(55,000) and Ke’ = 1/(1-υ2)
Nf = { 2,890,702 / [(2)(55,000)(1.1)(1.56)] }3.13 = 5121 cycles Eq. 14
Equation 14 can be manipulated to yield Sm for the two parts of the cycle range that are of interest, i.e. (1-to-
5121) and (5121 –to- infinity).
When Nf is greater than 5121 cycles:
Nf = { 2,890,702 / [(Sm)(1.1)(1.56)] }3.13. Eq. 15
When Nf is less than 5121 cycles:
Nf = { 2,890,702 / [(Sm)2(1.1)(1.56)/(2)/(55,000)] }3.13. Eq. 16
The two equations above can be rewritten in terms of the membrane stress Sm:
Sm = 2,890,702 / [(1.1)(1.56)(Nf0.3195)] … for Nf > 5121 cycles
Sm = { (2,890,702)(2)(55,000) / [(Nf0.3195)(1.1)(1.56)] }1/2 … for Nf < 5121 cycles
These equations can be further simplified and compared to the plots from NUREG 3243 as shown above:
Sm = 1,684,558 / Nf0.3195 …. for Nf > 5121 cycles
Sm = [185.3e9 / Nf0.3195 ] 1/2 … for Nf < 5121 cycles

= 430,466 / Nf0.1597 … for Nf < 5121 cycles
Case 2 Comparison

The elastic-perfectly plastic Neuber correction is compared to the exact Neuber correction in the plot below.
Sy = 55,000 psi
Kcss = 105 ksi
Ncss = 0.12
The black and light blue lines are the exact Neuber correction and the 2Sy approximation. The red line is a
linear only approximation added for comparison.
Case 3 Comparison
If the Neuber correction (in the document as Ke’) can be compared to Ke from VIII-2 Eqs. 5.31, 5.32 and
5.33 then …
If 3Sm < Sn < 3mSm then

Ke = 1 + (1-n) / [n(m-1)] (Sn/3Sm-1) = 1 + (1-n)/[n(m-1)] (Sn/2Sy-1)
n = 0.2 m=2.
Ke = 1 + (1-0.2)/[0.2(1)] (Sn/(2Sy) – 1)
else if Sn > 3mSm then
Ke = 1/n
else
Ke = 1
end if
The Ke’ (which is the simplification of the Neuber correction) can be plotted as shown below. The logic is
if s < 2Sy then

Ke’ = 1
else
Ke’ = s / (2Sy)
end if

Computing Stress Intensification Factors
The i-factor is the ratio of the nominal (M/Z) stress to cause failure in the component of interest with respect
to the same nominal stress that would cause failure in a girth butt weld. This relationship permits girth butt
welds to be placed at any location in a piping system without concern for the stress increase that
accompanies a typical girth butt weld.
This should not be the case for welded on fittings. Their tests are performed without girth butt welds in the
vicinity of the welded on fitting weld, and so welded on fittings, or contoured fittings should likely not be
placed on or in girth butt welds unless some other fatigue penalty is considered.
It is commonly shown that i=C2K2/2, when K2=1, i.e. for elbows and bends. (ref xxx).
Rodabaugh as early as 1967 (Ref … Phase 5), noted that K2=1 for i-factors found from strain gage results
extrapolated to welds for certain tests. In EPRI 19xxxx Rodabaugh again noted that K2=1 when finding i-
factors from elastic finite element data. Paulin continued the practice in the 07-02 project and produced
additional shell and brick finite element data using the stress classification line approach outlined in VIII-2
Annex 5?. In WRC 329 para ??? Rodabaugh noted again that K2=1.
This is not to suggest that K2=1, for the component under study, only that when developing stress
intensification factors that give the relationship to a girth butt weld that K2 = 1. When the component under
study has a stress concentration (as defind in VIII-2 XXX) that is expected to be greater than a typical Markl
girth butt weld, it is only then that a K2>1 should be considered for developing i-factors from finite element
data.
Hinnant has shown in 61871 that even when Ke>1 that for cantilever type loading in a piping system, the i-
factor can be used to predict low cycle fatigue providing the appropriate allowable is used. (Note in Fig. 1
below how the VIII-2 3STDEV curve for girth butt welds is higher than the Markl allowable, which suggests
that for low cycle design (below 4000 cycles, the Markl design appraoch is too conservative, although most
piping systems are not desgined in this low cycle range. )
It is likely that intersection i-factors developed using elastic analysis may also be used down to at least the
4000 cycle design range. Some evaluation of Ke for these components is recommended however. Ke as a
function of load is expected to vary from component to component, and so should likely be evaluated for
some intersections if the design into the lower cycle ranges is taken to the calculated design limit.
i-factor = Sf(girth butt weld) / Sf(component)

Fatigue Basics
There are three basic fatigue regimes: low, medium and high cycle. The definitions are approximate and
vary from document-to-document so readers should be sure what definition is being used. In some cases low
and medium cycle fatigue is considered any fatigue below 1e6 cycles. For the discussions here, the low
cycle range is less than 10,000 cycles and the high cycle range is above 1,000,000 cycles. In some cases a
very low cycle range less than 100 cycles, and a very high cycle range above 1e8 cycles will be mentioned.
There are many fatigue factors and constant multipliers. Some are applied to life and some to stress. Since
life and stress are related by a power relationship, i.e. N = S3, factors on stress have a much greater affect on
life than factors on life have on stress. When looking at graphs or tables of multipliers always remember to
differentiate if the factor is on life or stress.
When employing this document the user must understand his objective:
1) Apply a particular ASME Code Method?

2) Predict the failure of a component.
3) Compare one method to another.
The A07 VIII-2 Code welded fatigue method is based on a probability of failure. Adjustments are made to
the anticipated mean of the failure curve. Generally a three standard deviation shift from the mean is used.
When a three standard deviation shift from the mean of the failure data is used there is a roughly 0.5%
chance that a failure will occur in the design life of the vessel. A two standard deviation shift gives about a
2.5% chance of failure and a one standard deviation shift gives about a 16% chance of failure. No shift from
the mean means that 50% of the specimens would be anticipated to fail during the design life.
The original polished bar method does not use a probability of failure approach, rather variables that are not
accounted for are estimated and some shift from the mean of the failure data provided. A comparison of both
methods is included below.
The comparison in the table above suggests that the welded and smooth bar curves are quite compatible in
terms of shifts from the mean of the failure data, but this is not the case. When fatigue concerns are a critical
part of the design, it is not uncommon to compare the results from each, and address any difference that
affects the project.
There are five principle fatigue regimes:

0) Very low cycle (<100 cycles)
1) Low cycle (< 10,000 cycles)
2) Medium cycle (10,000 > N < 1e6 )
3) High cycle (>1e6 cycles)
4) Very high cycles (>1e8 cycles)

The cycle limits are arbitrary but intended to demarcate:
1) Low cycle areas where significant plastic zones surround the crack tip and minimize the effect of the
geometric concentration. Also in the low cycle range, there is a transition where the majority of life (80%) is
spent developing only small (less than 0.03 in.) cracks, and where the majority of life is spent propogating
cracks though a more significant part of the geometry.
2) Medium cycle where relatively small plastic zones surround the crack tip. These small plastic zones are
typically a fraction of the thickness in depth through the material, and some fraction of the diameter in the
circumferential or meridonal direction.
3) High cycle areas where little or no plastic zones surrounding the crack tip, or where the plastic zone
around the crack tip is very small and any crack growth is very slow or terminates after only small progress.
Fatigue life can also generally be broken down into propogation and initiation portions of lfie. Generally,
fatigue cracks in non-welded, or base metal parts are usually dominated by a large initiation life, i.e. most of
the specimen life is spent developing a detectable crack. Fatigue life in welded parts are usually dominiated
by propogation of a pre-esisting, althoguh very small crack from a weld imperfection of some type. Because
the mechanisms are different, welded and base metal curves have different characteristics. The use of
Fatigue Strength Reduction Factors are problematic when propogation issues become significant.
Characteristics that affect fatigue in one regime may not have the same significance in another regime.
Fatigue tests may be conducted several ways. The methods that affect the presentation of the data are
described below:
1) Strain controlled, unnotched

2) Strain controlled, notched
3) Load controlled, unnotched
4) Load controlled, notched
For most piping and pressure vessel components strain controlled tests are generally the most suited and are
the most often run.
Load controlled and strain controlled tests are essentially identical when the gross section that controls
displacement is cycled within the elastic range. This is usually attained when the gross stress range is within
Sy. In these cases, load control strain cycling is identical to strain controlled cycling.
When the gross stress range exceeds Sy in a load controlled situation there is likely some significant
plasticity at each end of the cycle range although perhaps only at concentrations, discontinuities and/or
notches. In this case, the variation of strain from cycle to cycle may not stabilize.
There can be cyclic plasticity also in the cycle range below 2Sy when notches have concentration factors that
create plastic zones adjacent to the notch. In this case, the general stress state shows to be elastic during
cycling, but adjacent to the concentration there is plasticity during cycling. In these cases, the elastic
material controls the strain and elastic predictions of plastic strains are reasonable (i.e. Ke=1, and Neuber
corrections = 1). When the membrane + bending stresses in the general section exceed 2Sy, outer fiber
stresses removed from discontinuities may undergo cyclic plasticity and may interact with cracks starting
from the surface that further increase strains.
A typical fatigue test result is shown below. These plots are most often in terms of Log10 on cycles and
Log10 on stress.

Unnotched fatigue test. Eεt and σ plotted. Notice that the actual total strain modulus (εtE) product will be
higher that the applied stress (σ). How this occurs is shown in the sketch below.
Stress-strain curve showing artificial elastic stress for the actual stress at point 3 on the stress-strain curve.
In the top figure point 4 reflects tha same failure at N8 as point 1, where N8 is the number of cycles at point
8 in the figure, since the stress σ at point 4 corresponds to the actual load on the specimen, whereas point 1
corresponds to the actual strain on the specimen multiplied by the elastic modulus. In the top figure, at N5,
the plastic and elastic curves come together at point 2. At this point, the gross stresses in the section are
elastic.
If a notch is added to the top figure, a downward adjustment of the curve is expected in the medium to higher
cycle ranges as shown below:

At the very low cycle end, (around N8), the effect of the notch is smaller. The effect of the notch increases
until N5, when the notch has a uniform effect on both curves, whether they are reported in terms of actual
stress or linearized stress (εtE).
For the same test, two different values are displayed in the low cycle range depending on whether the
linearized (εtE) or actual stresses (σ) are used. This can affect the allowable stress, as in the low cycle range,
the allowable stress calculated using the linearized stress (εtE), may be higher than the allowable stress
calculated for the stress computed using the actual stress (σ) to provide the same level of safety.
When the geometry gets more complicated than an axial tensile specimen, the situation becomes more
complicated. The bimetallic bar is an example.
As long as F/A is below the yield strength of the aluminum, a load on the left hand bar in the figure above
would be in the elastic range for all materials and the strains will be apportioned based on the length and
elastic modulus of each part. As soon as F exceeds the yield strength of the aluminum, the strain will be
concentrated in the aluminum and incremental load will tend to concentrate strain in the aluminum as a
function of its’ length with respect to the total bar length. If the aluminum stress-strain curve is elastic,
perfectly plastic, all the added strain will go only into the aluminum length and collapse occurs once the yield
point is exceeded.
This concentration of strain after yielding is termed elastic-followup. In the all-steel bar on the right, once
yield in the steel is exceeded, there will be plastic strain, but the plastic straining will be proportioned over
the entire test piece not focused into a single piece and elastic followup does not occur. Taggart provides a
thorough evaluation of elastic followup. The situation in the right hand bar can also become complicated
when the bar is installed in parallel with another system. In this case elastic followup can occur again.
Strain can also be concentrated at notches. In the low cycle range, strain at notches or concentrations is
lightly plastic and self limiting while the strain surrounding the notch and through the majority of the cross
section of the shape is elastic. In the medium cycle range, the nominal strain surrounding the notch
transitions from purely elastic to twice yield. Certainly if the general membrane stress on a section
approaches or exceeds twice yield for many materials and is not strain limited tensile failure will result. This
can be seen in the following diagram. For collapse to occur, the external load applied must be at stress level
#8 and a corresponding displacement to support a strain level of εf. If an elastically calculated thermal stress
approaches or exceeds twice yield as shown in the diagram below at point #8 the actual strain state will likely
be closer to εt since thermal stresses are strain limited and generally do not support unbounded or the large
displacements necessary to attain εf level strains. Occasionally these conditions can, however develop in
materials where a weak section is adjacent to a strong section. The weakness can occur due to temperature,

thickness, diameter or material. They can also develop where a long, slender pipe or vessel section is
adjacent to a short, stiff section. In this case, plastic deformation developing in between the weak and strong
sections will have some load controlled character.
The strain concentration factor (Ke) for the steel bar on the right in the figure above is equal to 1 for thermal
or secondary loads (which is the only loads that Ke is used for), since even after plastic strain develops, strain
controlled elastic displacements will overpredict the stress, (as typical), and accurately predict the strain.
The strain concentration factor for the steel-aluminum-steel bar on the left in the figure above is similar to
the condition found in an axial loaded welded tensile specimen where
Ke = (L1+L2)/(L1+(A/E)(εpn-1)(L2) where εp is the plastic strain in the aluminum and:
L1 = Length of the weld area undergoing plastic deformation

L2 = Length of the tensile specimen not undergoing plasticity
ε = Aσn. A=145 ksi if E=29e3 ksi. (approx values)

This shows that Ke can increase very quickly as plastic deformation in a local portion of the test occurs. The
above numbers were generated for L1=0.25, and L2=8. The elastic yield strain used to compute Sy is 51 ksi.
For these properties the Ke value in an axial loaded test into the plastic zone where the distribution of strain
is not evenly distributed throughout the weld zone, would produce a considerable Ke factor.
For a cantilever steel pipe, as yielding develops through the entire section at the wall as shown in the figure
below, the total strains at the wall are no longer proportional to the elastic strains. Elastic followup occurs
and the plastic strains at the wall are greater than would be predicted by a linear analysis.
Once a plastic hinge forms, the maximum increased strain over that predicted by elastic analysis is given by
Taggart as (1+2n)/(3n) where n is the exponent in the equation ε = Aσn. This maximum ratio varies as a
function of the deflection and applied nominal stress. The variation is shown below. The equation below is
developed using E.Rodabaugh’s elastic-plastic cantilever approach used in WRC 433. The difference in the
maximum Ke from WRC 433 and Taggart is thought to be the material model used.
??? Range?
If the fatigue test is based on a bar of constant material and area the plastic strain in the test can be found as a
product of the elastically calculated strain and Ke, when Ke is computed as a function of the applied stress or
displacement.

If the allowable curve is based on 1-2-3-7 (εtE)then an accurate estimate of the strain must be made from an
elastic analysis or a plastic analysis must be run to compute the fatigue life if the nominal strains are above
2Sy.
If the allowable curve is based on 4-2-3-7 (σlinear) which is a linear extrapolation of the stress due to elastic
displacement (M/Z in the Markl test), then an accurate estimate of only the elastic strain in an elastic model
must be made to compute the fatigue life.
If the allowable curve is based on 9-2-3-7 (σact) which is the actual stress corresponding with the actual strain
(εt) or found from Fact x L/Z, then an elastic-plastic analysis of the cantilever must be performed and the
actual stress computed, but as can be seen from the above figure, even in an accurate elastic-plastic analysis,
when the stress significantly exceeds yield, small changes in stress result in big changes in life (and strain),
and so predictions must be made carefully, and in these low cycle ranges are most often always made with
strain, which is why actual stress is not used to report fatigue test results.
If an identical axially loaded steel bar and cantilever loaded steel bar were tested, one in bending and the
other in an axial load to the same membrane+bending stress level, one would expect some difference in the
fatigue curves in the low cycle range due to the geometric variation in the strain concentration in the two
geometries if σlinear was used to report the fatigue results, and essentially no difference if Eεt was used to
report the fatigue results.

To accurately evaluate the degree of elastic followup the ratio of plastic strain to elastic strain for a given
load must be determined from an elastic-plastic analysis.
The VIII-2 A08 Code permits an actual Ke to be used with the polished bar tests, although for welded
components there is only the Neuber adjustment, and it is not related to Ke in the Code. The amount of
strain concentration developed in a complex component in the low cycle range is difficult to determine using
an elastic analysis, although tee and girth butt weld tests plot as straight lines on a log-log plot of cycles vs.
elastic strain, implying that linear elastic analysis will successfully predict failures or accurate design lives
down to 100 cycles or so.
A number of collected low cycle welded fatigue curves are shown below and suggest that for the strain
modulus product (εE), a straight line log-log plot is obtained. There must be some disconnect between the
collected tests shown below and the cantilever tests of PRG since PRG shows that the elastically calculated
stress (σlinear) plots along a straight line, while for the tests below, the strain modulus (Eεt) plot shows as
linear. It is likely that the slight deviations in slope and limited range of the tests allow PRG to show
log10(σlin) vs. log10 (N) as linear, while the investigators below show log10(Eεt) vs. log10(N) as linear. The
PRG method results in standard deviations and mean values over a large range of test data. The fact that
linear results can be obtained using either of the recommended low cycle fatigue approaches, implies that a
variety of fatigue curves and constants can be provided over applicable ranges and used effectively. The
variety of approaches, i.e. σlin vs. Eεt support the different analysis methods, i.e. elastic, or elastic-plastic.
For certain test geometries, relationships can be developed between them.
Note how the Delft low cycle tests show approximately 100-to-1000 cycles for stress levels whose tests also
show only a single cycle of life. This suggests that designing for cyclic stress levels in the 500,000 psi (3450
Mpa) or 2% strain ranges is fraught with potential difficulty, and so that additional caution should be
exercised when working in these design ranges. This is an important note when it is considered that some
operating companies, and some codes eliminate concern for fatigue below a certain number of cycles. This
would imply that very high local strains are acceptable providing collapse or ratcheting is not immenent. The

test below suggests that single cycle failure is possible. These tests should be studied more closely to see if
the specifics of the test coincides with the speculation above.

Since many fatigue evaluations are performed using an elastic analysis, the effect of the strain increase due to
the loading may be included in the allowable curve. If the allowable curve is based on an axially loaded

welded bar test then the possibility of strain concentration must be added to the calculated elastic strains or
subtracted from the allowable if the geometry under study is not an axially loaded bar. If the allowable is
based on a cantilevered bar type test, then the allowable can be compared directly to the elastic analysis of a
cantilevered component because the effect of the strain concentration has already been considered. For other
geometries, the ASME code provides a conservative estimate of Ke. The Neuber approach in the welded
fatigue curve method also provides a less conservative (although possibly not as all-encompassing), method
for including the effect of low cycle stress/strain prediction from an elastic analysis.
There are three phenomena in the medium to low cycle fatigue range that affect the transition between
polished bar and notched, or welded bar behavior. One is the Ke – or strain concentration effect that occurs
in a geometry with a stress or strength gradient, the second is a concentration at the base of notch in the
nominal stress field, and the third is the Poisson’s effect that may equal 1.4 on stress in a biaxial stress state.
Generally for externally applied loads, Poisson’s effect is not significant, although in cases where externally
applied loads and pressure produce plasticity the elastic prediction of stresses may be underestimated since
Poisson’s ratio goes to 0.5 when the point under study is plastic.
The low cycle plasticity described in the illustrations above occurs throughout a large part of the component
section, i.e. typically plastic zones whose size are larger than 2t, or some percentage of the diameter, (if not
all of the diameter).
In the medium cycle range there may still be plasticity at notches, but the surrounding material is usually
elastic, (at least certainly when Sn < 2Sy) and so the plastic strain is proportional to the surrounding elastic
strain, and a geometric concentration factor (Ke) is not needed.
In the bar and cantilever example above, the plastic strain concentration effect drew downward the cycles—
to--failure curve because of the increased strain. In the sketches shown above, the line 1-2-3 and 4-2-3 are
almost the same, and essentially have the same slope. This is often the case when Ke factors are involved in
steel geometries. The geometric strain concentration tends to pull the stress vs. failure curve downward such
that the low and medium cycle ranges share the same slope.
The collection of girth butt weld and flat plate data point failures compiled by Chris Hinnant at PRG is
shown below. Note that the plot is in terms of linear stress and linear cycles.

The points are labeled M+B since there are membrane and bending stresses when the girth welds were
attached to flanges in the tests. The flange increases the maximum surface stress by approximately 20% due
to bending. Where the girth weld failures are in flat plate tests, M+B is equal to the membrane stress only.
This difference is significant since fatigue failure due to a purely membrane stress occurs sooner than fatigue
failure due to a purely bending stress. The interaction between membrane and bending stresses are thought
to be more complex when the stress state varies along the weld line, but correlations based on this variation
are not present.
These basic stress states described are visualized in the sketches below.
The top two figures show the basic geometries used in a majority of welded specimen fatigue tests. The
bottom left figure shows the type of welded geometry used in pipe tests where the straight pipe is welded to a
tapered flange. The bottom right geometry shows the type of geometry that introduces the stress state
experienced in tees and supports where the membrane and bending stresses vary along the weld.
Fatigue tests based on these specimens are conducted using constant strain tests, constant displacement tests,
and constant load tests. As described at the top of this section, the type of test, and the data recorded from
each test can influence how the allowable is developed and what stress should be compared against the
allowable to get an accurate evaluate of the fatigue life.
In the medium to high cycle ranges, the largest volume of the critical geometry is elastic and the range of
membrane+bending secondary stresses are below twice yield. In this case, the elastic strain predictions
above yield are relatively valid, and no adjustments beyond the consideration of a proper Kf for any notch or
weld concentration are required. In the low cycle range, the type of test and how it is reported becomes more
important. Unfortunately welded speciments are difficult to fabricate in a perfectly symmetric manner, and
the smaller the specimen, the more error associated with the lack of symetry. Where high plasticity is
developed grossly in a small or non-symmetric model, plastic strains will tend to be concentrated on one side
of the weld. Since the load to continue the plastic straining increases by only a small amount, the
disproportionate distribution of plasticity continues through the cycling, even though strain hardening
increases the local strength of the material. This behavior also tends to introduce strain concentrations (Ke)
even in straight bar tests of welded geometries so that some manner of Ke modification is required even
when low cycle tests are conducted on straight bars. This is born out by inspection of welded low cycle

specimen tests, and by the observation that tee and girth butt weld tests do not exhibit the increase in strength
at low cycles (high plasticity) observed in polished bar tests. It is thought this reduction is on the order of
two times. It is significant only in that adjustments for Ke or Neuber corrections for welded specimens are
not needed since the nature of flat bar welded specimen tests includes the Ke already, just as parent metal
cantilever tests already include the effect of Ke.
The welded fatigue curve method in ASME Section VIII Division 2 A07 and later versions is based on the
WRC 474 “Master S-N Curve Method for Fatigue Evaluation of Welded Components”, by Pingsha Dong,
Jeong K. Hong, David A. Osage, and Martin Prager.
The mean curve for girth butt weld points from application of the method in VIII-2 A07 is shown below
when plotted against the failure data compiled by Hinnant.
The red points are produced using the mean VIII-2 welded curve and the blue points along the bottom of the
curve are developed using the VIII-2 welded curve shifted by three standard deviations. Note that the red
data points don’t seem to fit the center of the points particularly well at the high cycle end of the curve but
there is also very little data at this end of the curve and the red points show to be conservatively at the low
side of the failure data. To introduce the standard deviation shift, the blue points are moved to the left by 3
standard deviations on log10 of cycles.
Note that the center of the blue collection of points is almost twice removed from the center of the red
collection of points on stress down to the low cycle ranges where this shift is difficult to apply.
Hinnant and Paulin in PVP 2008-61871 give Eq. 3 for girth butt weld failures. Their equation 3 plotted
against the failure data is shown below. Note that this equation 3 is very similar to the red points above from
the A07/08 VIII-2 welded fatigue method. There is no scatter in the Hinnant data since there is only a single
variable used in the 2008-61871 Eq. 3 - the membrane stress. The welded fatigue curve method of VIII-2
accounts for membrane and bending stresses, which vary somewhat in the test data base sited by Hinnant. In
both cases, the mean curve tends to fall to the bottom of the scatter of the failure points at the high cycle end
of the curve.

\
The test data points for girth butt welds in a smaller, but lower focused cycle range are shown below. It is
interesting to note that four points seem to fall below the “center” of the data, and well below the expected
scatter in the cycle range between 700 and 2100 cycles. These four points were generated by Markl in an
attempt to produce large D/T tests. Markl himself suggested that the test points were disappointing. This is
likely true since Markl had to bore 4” pipe to 0.05 and 0.01 inch thicknesses for the tests.
The VIII-2 mean curve for girth butt welds and the Hinnant-Paulin PVP 2008-61871 Eq. 3 are plotted
together in the plot below. The membrane plus bending stress is used as the ordinate in each case. The
Hinnant Eq. 3 in PVP 2008-61871 matches the welded fatigue curve method very favorably.

If the welded fatigue and Hinnant curves are viewed in log-log space and focus drawn to the low cycle end of
the curve a difference begins to appear.
In a linear space, the difference also appears when the cycles are limited to 10,000. In both plots, the master
curve (A08 welded fatigue curve) method falls below the Hinnant girth weld test data at around 4000 cycles.

One of the differences in the development of the two curves is the Neuber correction that is included in the
A08 VIII-2 calculation. The Neuber correction is an adjustment for plasticity that is typically not needed in
cantilever or other welded fatigue tests where the test data is developed on the basis of an elastic calculation.
With the Neuber correction removed from the A08 VIII-2 curve the agreement between the actual Hinnant
test data is improved as shown below. Approximately 66% of the difference between the test method and the
A08 VIII-2 curve can be accounted for by removing the Neuber correction.
The Neuber correction only effects data below about 5200 cycles.

The PVP 2008-61871 paper only uses the membrane stress and does not include thickness, mean stress or
other corrections. Options that are available in the ASME method were selected for the above comparisons
so that the test conditions could be replicated.
(67) Anand and Parks point out, that from a sensitivity point-of-view fatigue life is proportional to plastic
strain amplitude to about the -1.67 power, a relatively insensitive amount so that small differences in the
plastic strain lead to small disfferences in the resulting predictions of low cycle fatigue life. In the high cycle
range the cycle life is often proportional to the inverse of a very low number, and so the number of cycles are
influenced very strongly by variations in the predicted stress, i.e.
The master curve has a standard deviation of 0.246.
Neuber’s Rule and Ke.
Neuber’s Rule is used to relate the stress concentration at a notch to the nominal stress in the part. Neuber ‘s
rule is often presented in the following form:
Kt2σnomεnom = σε = σ/E + (σ/K)1/n
where Kt is the elastic stress concentration factor for the notch. Neuber’s rule can also be expressed:
(Ktσnom)(Ktεnom) = σε
In this form the elastic stress concentration factor applies to both the nominal stress and strain. This equation
implies that the product of the nominal stress and strain and the factor Kt is equal to the product of the
maximum stress and strain at the root of the notch. Kt is only considered valid when both the stress at the
notch, and the nominal stress in the part are elastic.
Kf is often called the fatigue strength reduction factor or FSRF and replaces Kt as the stress increases and
plastic effects at the notch (discontinuity) effect the fatigue life. Kf is less than or equal to Kt, and Kf
generally replaces Kt in the medium cycle range when N < 106 cycles.

Throughout the medium cycle range (10,000 >N < 106 ) Kf generally remains constant., but below about
10,000 cycles, the effect of the stress concentration decreases, eventually becoming 1 at a single cycle since
the stress concentration at the root of a notch does not affect the single load collapse of the part. This effect
can be observed by comparing the Hinnant test data to the polished bar curve. In the plot below the strain
modulus (εE) product increases while the elastically predicted stress (M/Z) remains linear. A Kf value of 2
represents the difference between the polished bar curve and the welded fatigue curve before plasticity
effects (Sn>2Sy) begin to affect the behavior.
For the end loaded cantilever beam a strain concentration factor shifts the εE curve away from the linear
elastic strain modulus curve. In the cantilever case linear elastic strain predictions underpredict the actual
strain at the fixed boundary with increasing error once the membrane+bending nominal stress in the pipe
exceeds 2Sy. This underprediction is independent of the notch at the boundary due to the weld and tapered
flange hub. This effect can be seen in the above plot, where εE is the measured strain, and the blue lines and
points show elastically calculated strain. The Hinnant εE test data does not fall on the HP GBW test data
line. As long as the test data line (in blue) is used in design, further adjustments are not needed with
elastically predicted stresses.
Ke or strain concentration effects may also be seen by plotting the linear predicted strain and the actual
measured strain along with a variety of estimates of the strain from various elastic-plastic models. This is
shown in the plot below.

The Hinnant test data points reflect strains that are removed from the elastic strain straight line. As the
elastic strain increases linearly, the actual strains move further away from the straight line. There are two Ke
plots included in the above figure. The blue line to the far left is the Ke equation representing strain
increases presently in the Code. The hook at the top of the blue line as it leaves the plot at 400,000 psi
alternating stress is the point where Sn = 3mSm and Ke = 1/n. The orange line with solid orange triangles is
the Ke equation developed for cantilever beams, where Kemax = 2n/(1+3n). This expression is not in the
Code, but is referenced in a number of Code reference documents. The smooth lines are from the program
Ev Rodabaugh introduced for WRC 433. It can be seen that the Ke adjustment using the cantilever
developed equation matches the measured test points relatively well, and that the WRC 433 equation using
an Sy=68,000 psi also matches the test data points fairly well. Probably the best match comes using the
Neuber correction (the dashed orange line). This corroboration is thought to be circumstantial.
The plot below shows how the εE curve might continue into the very low cycle range for the Hinnant-Paulin
and Markl cantilever test. Since crack growth mechanisms should begin to influence behavior in the lower
cycle failure range (<250 cycles), it is unknown whether the εE curve will continue to follow the straight line
or, even if the blue Hinnant test data points will continue to follow the straight line. The curve, as it is
however, shows that for a 4” standard wall carbon steel cantilever pipe welded to a flange, a linear elastic
analysis using beam elements would accurately predict a fatigue failure down to at least 250 cycles, a
somewhat surprising result. Additional fatigue tests are planned in the lower cycle ranges to determine
where the behavior deviates from linearity.
Welded Fatigue Curves
Major features of the basic welded fatigue curves are shown in the figure below.

Surface strains at failure follow the heavy blue line in the figure below.
The heavy blue line represents the product of the elastic modulus and the non-peak surface strains at the
point of failure.
The heavy red line represents the product of the elastic modulus and the elastic non-peak surface strains at
the point of failure. Often the red line will be straight for failures greater than 200 cycles. The shape of the
red line depends on the shape of the Ke curve for the geometry and loading when the secondary stresses are
greater than 2Sy. When (Ke)(Eεel) ≅ E(εel+εpl), the elastic straight line can be used to predict fatigue life
throughout the majority of the low cycle range.
When (Ke)(Eεel) > E(εel+εpl) then

In some cases, an elastic analysis will predict the correct elastic and plastic strains through the entire plastic
range. For example, the straight bar with a short, weak axial section can have a Ke of any value that’s equal
to:
Predicted elastic strain is distributed evenly and is D/(L1+L2), where D is the displacement at the end of the
bar. When the weak section goes perfectly plastic, the strain is D/L1. The plastic over elastic strain would
then be: e(plastic) / e(elastic) = (D/L1) / (D/(L1+L2)) = L1+L2 / L1 and when L1 is small with respect to
L2, the Ke = L2/L1.
, for example, those along the green line in the plot below. In other cases, the cantilever, or axially loaded
taperd beam, for example, the elastic analysis will underpredict the correct strains.
More on Neuber’s Rule, Kf , and Low Cycle Plasticity
Neuber’s rule for the low cycle range is shown in the equation below.
Kf(ε'nomσ'nom) = εσ
Nominal stress and strain due to elastic followup is expressed by the following equation, where ε’nomσ’nom is
the actual stress and strain adjacent to, but not at, the toe of the weld or bottom of the notch.
ε’nomσ’nom = Keεnomσnom

In test conditions for reversed loading, Kf is less than Kt, and as Kt increases, the ratio of Kf/Kt gets smaller.
Very sharp notches (high Kt) have less effect on fatigue strength than would be expected from the high value
of Kt.
Below 106 cycles, Kf is appreciably lower than Kt. Neuber’s approximation is that KεKσ = Kt2. where Kε is a
multiplier on nominal strain and Kσ is a multiplier on nominal stress when plasticity becomes significant. As
the nominal strain surrounding the concentration or notch exceeds yield, Kσ goes to 1, and Kε increases until
the product then begins to drop, i.e. KσKε goes to 1 as the mean cycles to failure goes to 1.
Per Dieter (p.425) Another way of interpreting this correlation is that a notched specimen and a smooth
specimen will form detectable cracks at the same life provided Kf(σnomεnomE)1/2 for the notched specimen is
equal to (σmaxεmaxE)1/2 for the smooth specimen, and this assertion forms the basis for using the polished bar
curves for fatigue of welded and/or notched specimens.
Ke and Neuber Adjustments.
Ke and Neuber adjustments are not equal, and are not intended to be equal, but perform similar functions in
that they both adjust the allowable fatigue curve for the ASME VIII-2 Code when elastic analyses are
performed and when the sum of membrane and bending stress at the section exceeds 2Sy. The sum of the
membrane and bending stresses are computed directly from beam or shell analyses and can be extracted from
a volumetric analysis by use of stress classification lines through the thickness (SCL’s).
When the membrane and bending stress at the section exceeds 2Sy, in a simpified sense, the stresses are high
and there is nominal plasticity at each end of the loading cycle. The nominal membrane + bending stress at a
section does not include concentrations or the effect of weld toes, etc. and so gives a generally nominal outer
fiber stress at the section. The presence of concentrations, notches, welds, or corners adds a further increase
to an already high stress state.
The fatigue curve is broken down into three regions:
1) High stress, low cycle (less than 10,000 cycles)

2) Medium stress, medium cycle (between 10,000 and 1e6 cycles)
3) Low stress, high cycle (greater than 1e6 cycles)
There are two basic types of welded stress states:
1) No geometric concentration
2) With geometric concentrations
The no geometric concentration welded state is analogous to the ground smooth radius on a nozzle or plate
attachment. The with geometric concentration welded state is analogous to the as welded nozzle weld.
Welds with or without geometric concentrations have metallurgical concentrations, and strain concentration
factors (Ke) can be due to geometry and loading only, and do not depend on whether or not a weld,
concentration or other notch is present.
Welds can further be broken down into

1) non-full penetration fillets
2) fillets that provide cover to a full penetration weld.
Fillets that are non-full penetration often fail through the throat although may fail starting at the toe of the
fillet. Depending on the geometry and loading one location will be more sensitive to stress than the other,

and when a partial penetration weld is evaluated, both throat and toe failures should be examined. It may be
that fillet welded geometries in a high strain concentration area result in lower strengths than the full
penetration girth and tee welds tested by Hinnant and Markl, and for this reason, the Ke and Neuber
corrections should be applied as intended. In situation where further corrections are not required, (girth butt
welds and tees for example), the Neuber and Ke corrections are likely not required.
In the low cycle range, Neuber’s rule and Ke become a part of the Code calculation. In this cyclic stress
range, the growing crack from the surface or near the surface grows into an existing surrounding plastic zone.
Plastic zones in the low cycle range are generally large compared to the thickness of the section. Results
from load controlled tests and strain controlled tests vary significantly in this zone when the load controlled
tests is measured as εtE. If the load controlled test and the strain controlled test are reported in terms of εtE,
they are approximately equal.
In the medium cycle range, Neuber’s rule and Ke do not become a significant part of the Code calculation.
In this cycle stress range, the growing crack from the suface or near the surface grows into an existing
surrounding elastic stress. The zone immediately surrounding any crack tip at the section is plastic. The
stress around a discontinuity or notch may also be plastic, but the major characteristic of cracks in the
medium cycle range is that they grow into a surrounding area of generally elastic material. The majority of
the fatigue life in this range is spent establishing a crack (initiation), or in establishing a crack that grows
progressively. In this situation, the straining is generally controlled by the elastic material surrounding the
plastic zone, and at the end of each cycle the strain returns to its state at the start of the cycle as the elastic
material forces the plastic material back where it began at the start of the cycle. 90% of the life is spent
developing a crack that’s less than 0.01 inches thick.
In the high cycle range the Kt factor applies since the stress state is low. The secondary outer fiber stress is
generally 5-to-10 times below yield stress on amplitude, such that stresses at the tip of cracks, toes of fillets,
or at the bottom of notches remain elastic, i.e. σnomKt < 2Sy. In this case, where there are some poorly
oriented grains, alterations in the chemistry or weld geometry such that some plastic straining on one end of
the cycle exists, the crack may or may not grow.
Since the state surrounding the crack tip is considerably different in each of the ranges, the variables that
effect crack growth differ for each range. Since the low cycle range is dependent on the yield stress and
strain hardening, one would expect larger variations between materials when the fatigue test is conducted in
this range. For example stainless and carbon steel tests vary in this range.
Generally Kf=Kt in the very high cycle failure range. Kf < Kt starts in the medium cycle range. Neuber’s rule
involves adjustments between Kt and Kf, where Kt is the elastic stress concentration factor and Kf is the
elastic/plastic stress concentration factor at welds or other concentrations or notches.
The effect of stress raisers on fatigue are:

1) there is an increase or concentration of stress at the root of the notch,
2) a stress gradient is setup from the root of the notch in toward the center of the specimen,
3) a triaxial atate of stress is produced in the vicinity of the notch root (once the maximum stress at the notch
root is above the yield stress of the material.)
Unlike elastic stress concentrations which can reach Kt values in excess of 10 as the notch is mades sharper
and deeper, Orowan has shown that the plastic-constraint factor cannot exceed a value of 2.57.
Kt = elastic stress concentration factor = σmax_elastic / σnom_elastic

Kε = plastic strain concentration factor = εmax / εnom
Kσ = plastic stress concentration factor = σmax / σnom

Neuber suggested that KεKσ = Kt2
Notches increse the tendency for brittle fracture in four ways:

1) By producing high local stresses
2) By introducing a triaxial tensile state of stress
3) By producing high local strain hardening and cracking
4) By producing a local magnification to the strain rate
Most of the data for notched specimens are found by comparing the S-N curves of notched and unnotched
specimens. The effect of the notch to decrease the fatigue limit is expressed by the fatigue strength reduction
factor or fatigue-notch factor, Kf. Kf is the ratio of the fatigue limit of unnotched specimens to the fatigue
limit of notched specimens. For materials which do not exhibit a fatigue limit, some agreed-upon ratio
between the notched and unnotched mean failure curves is esatablished and used in the determination
process.
Kf is less than Kt, and the ratio of Kf/Kt decreases as Kt increases, thus very sharp notches (high Kt) have less
effect on fatigue strength than would be expected from the high value of Kt.
Neuber proposed that Kf = 1 + Kt-1 / (1+(1-ω)(a/r)1/2)
In high stress, low cycle fatigue Neubers KεKσ = Kt2 governs.
Kε = εmax/εnom
Kσ = σmax/σnom ~ (1 + (Kt-1)(Es/E)
Es = secant modulus.
When plastic flow starts at the root of a notch, (as determined by Kt), the maximum local strain increases
rapidly and Kε increases.) The limit Kσ = 1 when the plastic flow spreads over the entire specimen for an
elastic perfectly plastic material. In this case Kε = Kt2.
KεKσ = Kt2 can be rewritten:
Kt = Kf = (εmax/εnom x σmax/σnom ) 1/2
or Kf(σnomεnomE)1/2 = (σmaxεmaxE)1/2.
This shows that a function of nominal stress and strain (σnomεnomE)1/2 need only be multiplied by a consant
concentration factor to obtain the values of the true stress and strain at the notch root.
A notched specimen and a smooth specimen will form detectable cracks at the same life provided
Kf(σnomεnomE)1/2 for the notched specimen is equal to (σmaxεmaxE)1/2 for the smooth specimen. Thus
unnotched fatigue results for completely reversed stress cycles can be used to estimate the fatigue life of
notched members if Kf is known. Aditional simplification results when the nominal stress and strain are both
elastic. Since σavg = εavgE, Kfσnom = (σmaxεmaxE)1/2. This occurs only in the high cycle range.
“If the plastic zone is highly localized, the surrounding elastic material controls the strain in the plastic
material and no strain concentration occurs. When the plastic zone is lare enough to become a significant
factor in the stress distribution the strains in the plastic zone become larger than those calculated by elastic
theory and a strain concentration must be considered to properly estimate the maximum strain when needed.
The question is, why don’t we need the maximum strain in the cantilever to predict the proper failure? The
answer is that the elastically calculated stress plots in a straight line on a log-log plot of elastically calculated

stress and cycle life and so it is simpler to use the elastically calculated stresses. The maximum plastic strain
could be calculated and used for evaluation, but the allowable curve would be nonlinear and more difficult to
develop. Most fatigue curves in use are straight lines on log-log plots. Multiple fatigue tests tend to plot on
a straight line with scatter within an accepted standard deviation. The scatter is better described by a Weibul
distribution, but normal distributions are acceptable. As the number of cycles to failure gets smaller, the
effect of concentrations and geometric irregularities gets smaller since large plastic zones obscure details that
are otherwise significant in a higher cycle range.
Results in linear space from the Hinnant cantilever tests are shown below. The red line is the true strain
modulus product curve, and the black and blue lines are the linear elastic stress line.
A closeup of the stress-to-failure prediction for girth butt welds given in PVP 2008-61871 and the test data is
shown below. The true strain modulus product is shown as the red line on top. This plot is generated in log-
log space.

There are two tensile tests reported for HP tests in PVP 2008-61872 for the heat:
Sy, Su = 54500, 78500
Sy, Su = 51000, 77000
The PRG tests G1 thru G4, G9 and G10 are flux core and the yield is likely around 70ksi for the weld rod.
for the tests G5 through G8 were stick welded using Markls Fleetweld No5 rod still available from Lincoln,
and the yield for the weld material is probably 61 through 64 ksi, which is 20% greater than the parent metal
yield strength. The mismatch of weld metal and base metal yield strengths is common and serves to
minimize the effect of the geometric concentration, although makes the effective volumetric modeling of the
plastic behavior around cracks difficult.
The ASME Code fatigue approachs using polished bars and the welded fatigue method is an attempt to
predict strain cycling failures. In many cases, especially where higher cycles are involved it is also stress
cycling, but in low cycles it is certainly not, since only in rare cases (axial loaded tensile test with perfect
balancing between both sides of the weld) does an elastic prediction of the strain equal an elastic/plastic
prediction of the strain. Strain ranges of interest are around 10-to-15% when one is making cyclic load
failure predictions in the very low cycle range. Manson points out that a strain range of 1% in parent metal
will typically survive 10,000 cycles, while a strain range of 2% will typically withstand 1000 cycles.
In the “Criteria of the ASME Boiler and Pressure Vessel Code for Design by Analysis” Fig. 4 shows two
curves that reflect stress-controlled tests plotted with strain controlled tests. Even where linear elastic
analysis is concerned curves B and C are not generated from an elastic analysis, and so these curves are only
curiosities. Stress controlled comparisons should only be made when an elastic-plastic analysis is conducted,
in which case allowable strains would be extracted from the polished bar curves.
The fact that the Neuber or Ke corrections can be omitted does not mean that the strain concentration does
not exist in these regions, only that it is difficult in tests where there is a significant amount of plasticity to
produce a test where the extent of plasticy is uniform along the piece and on both sides of the weld. In this
case, the elastic followup occurs even in the simple, small sample test, and the strain concentration causes the
test to have an appearance of following a straight line. When tE is plotted in tests where it is more likely
that a uniform plasticity is setup through the high stress area on both sides of the weld, (i.e. the 4 point bend

test – See Scavuzzo WRC 433), the strains do not fall on a straight line extended from the medium cycle
range.
In general, the fact that the yield stress is exceeded by the nominal stress state surrounding a crack does not
have too much effect on crack growth below the unstable crack growth regime since all cracks growth
through yield level stress zones regardless of the far field stress level. The secondary nature of the crack may
limit growth in this case, or elastic followup may promote growth, but these are specific geometry, material
and stress dependant items that cannot be determined a priori. This is one reason for exhibiting care when
extrapolating tests on one geometry and stress state to other geometries and stress states, since the degree of
growth, and/or stress distribution in high yield cases is unknown without a J integral style of crack growth
evaluation. It is reasonable however, to estimate the degree of crack growth and ultimate damage when
crack growth is significant by turning to fracture mechanics approaches. For stress levels where the majority
of life is spent with the crack below a depth of 0.01 inches. welds perform in a manner that is similar to an
unwelded geometry of the same size and Kf factor can be used to accurately predict fatigue life.
Nonlinearities in behavior and variations in Kf begin to occur when the nominal stress removed from the
crack tip exceeds twice yield and plasticity occurs in the vicinity of the crack tip each cycle, i.e. the stress in
the vicinity of the crack tip does not shake down to elastic action. In this case, the crack propogates through
a surrounding field of sliding dislocations and the slope of the endurance curve changes as a function of the
size of the plastic zone and the effect of surrounding dislocations on the progress of the crack. When the
majority of life is spent with cracks beyond this somewhat arbitrary threshold, fracture mechanics should
likely govern. One would also expect a transition between the Sn < 2Sy region and the Sn>2Sy region. If
the damage mechanism proceeds at the same rate as a function of the applied stress (or more likely strain
range), then the slopes would be expected to be the same. For ferritic and austenitic steels in air or other
non-aggressive service environments at temperatures of to 212F, API 579 gives an expression for da/dN =
8.61x10-10 ∆K3.0 ksi.(in)1/2. This equation can be separated and integrated to produce the result that some
function of stress raised to the -3 power is equal to a constant times the number of cycles to failure:
(Coσ)-3 = C1 Nf.
This can also be rearranged:
C2 σ = N -0.3.
This relation suggests that in the very low cycle range, where significant crack growth controls life, a slope
equal to the HPGBW curve is expected from the calculated strain modulus product.
For the cantilever test, the linearized M/Z stress vs. cycles appears to be on a straight line plot in log N and
log σ space, but the εE stress vs. cycles seems to turns up. These curves are shown below.

The εE curve is certainly moving upward suggesting that in the lower cycle ranges, welded curves are
stronger than would be suggested by the linearized M/Z straight line curve. The Markl curve by contrast
drops below the HPGW linearized M/Z curve.
If the crack propogation relation is applicable, one would have expected the εE line to either drop or stay
along the HPGBW line. The slope of the εE line from the PRG tests is -0.593 which is the same as the plastic
slope of the universal slopes line. (See Fig. 23 Manson, “Fatigue: A Complex Subject – Some Simple
Approximations”.
The question to ask, is where the crack propogation portion of the curve begins to govern the behavior. When
this happens, the slope of the curve would remain -0.3. Ke and Neuber adjustments support folliwng the
Markl curve shown above, which seem appropriate if the stress range entering the curve is calculated by a
solely linear elastic analysis. The fact that the HPGBW curve does not follow the Markl, adjusted Ke, or
adjusted Neuber curves indicates that at least the increased strains in the tested cantilever welds do not result
in reduced life. There is no question that cantilever tests suffer additional strains beyond those calculated by
linear methods when the M/Z stresses exceed yield. There is also no question that some geometries when
loaded beyond the yield stress suffer strains greater than those predicted by a linear analysis, (just like there
is no question that some geometries do not show greater strains.) When greater strains are experienced, the
Ke and Neuber adjustments come closer to predicting those strains, but the Ke and Neuber methods are
based on assumptions of elastic followup that may or may not be valid for the geometry and loads of interest.
As can be seen in the above figure the εE curve begins to rise away from the linearized M/Z curve and moves
closer to the polished bar curve for the material. This is expected since the Kf factor for low cycle failures
tends to 1.

In summary, there are several mechanisms that come into play once the nominal stress removed from the
crack tip exceeds 2Sy over a significant part of the thickness and/or diameter:
1) A reduction of Kf due to increased plasticity. As the size of the nominal plastic zone around the crack tip
increases in size the effect of the geometric or mechanical notch is minimized since since the rate of strain at
the notch root diminishes with increased load until Kf=1 at 1/4 cycle.
2) For certain geometries, (biaxial bars, notched plates, cantilever loaded beams) a Ke multiplier is needed to
properly estimate the true strain at the notch location.
3) Once the crack reaches a certain size a change in load distribution due to crack growth and a change in the
crack growth rate due to the changing size of the crack results in differing crack growth rates and a change in
the slope of the fatigue curve. It is shown above that once crack propogation is the governing mechanism for
points on the endurance curve, for ferritic and austenitic materials, the slope of the crack growth rate curve is
the same as the slope of the crack development curve for welded materials.
4) Manson shows that for low cycle fatigue failures in polished bars, the slope of the cycle-to-stress to failure
curve becomes steeper, and where the universal slopes approach is valid transitions from about -0.12 in the
elastic regime to about -0.6 in the plastic regime.
To determine which of the methods begins to control fatigue life in these low cycle ranges, low cycle tests
can be conducted and the method causing failure observed. In the case of the cantilever, the point where
crack propogation begins must be noted, and the speed of crack growth, and possibly depth measured during
the cycle life. In addition, the strain must be measured carefully. Tests at 10 and 100 cycles are
recommended. These are shown at points A and B in the above plot if the εE line is followed. If the
HPGBW line is followed

Strain Softening or Hardening
Cyclic stress-strain curves for low carbon steel materials are shown below. Note that Annex 3 gives static
and cyclic stress strain curves so that effects of cyclic softening or hardening can be considered where the

material displays these characteristics. (As can be seen from the plot below, carbon steel does not display
significant strain hardening or softening.)
A common rule-of-thumb for cyclic softening and hardening is given below from “Fatigue of Metallic
Materials”, by Mirko Klesnil, and Petr Lukas, “Extensive low-cycle fatigue data enabled the formulation of
an empirical rule, predicting, on the basis of the conventional tensile diagram, whether the material will
cyclically soften or harden. If the ratio fo the tensile strength to the yield stress is higher than 1.4, the
material will harden under cyclic loading. If this ratio is lower than 1.2, the material will soften.”
The tensile to yield ratio for A106 is higher than 1.4, however, and the material is not found to significantly
harden, and so for other materials the user is encouraged to compare the various curves from the Code.
Two On Stress
It is not uncommon to use two or three times the standard deviation for a shift outside of the range of the
desired probability of failure. As it turns out, the three standard deviation shift is very close to a two times
multiplier on stress in the medium to higher cycle ranges. For the unreinforced tee data base these shifts
from the mean of the failure curve are shown in the figure below. Several other plots are included below so
that users can draw their own conclusions about the nature of the allowable stress being used.
User’s can also take the calculated membrane+bending stress output from a calculation and compare it to the
values shown below.

The above plot shows the ASME Curve without Ke adjustment using Kf=2. 20 on cycles or 2 on stress.
Solid blue line is 2 on stress using the Hinnant mean curve.
Girth Butt Weld points, structural stress and predicted mean and -3stdev with fen=4 are also shown on the
above plot.
The reduction in fatigue life caused by surface roughness increases as the strength of the steel increases.
This behavior is related to a reduction in the stress concentration caused by plastic deformation which occurs
more easily as strength decreases. Similarly the effect of surface roughness is more pronounced at low
stress, high cycle life than at high stress, low cycle life.
For welds, “Small differences in the size of the initial crack or cracklike discontinuity result in large
differences in the total fatigue life.” This observation tends to support the overall effect of inspection on
fatigue life on a population basis. For critical sections however, the weld quality is generally good, and more
attention is paid to weld quality, and so the inspection effects will be less significant.
This tends to be very true in the medium to high cycle ranges, but tends not to be so true in the low cycle
range where imperfections are not as important due to the high degree of plasticity.
Plots from WRC 432 are shown below.

The significance of the Master curve is that it looks at the geometry and characteristics of each test and
reduces the spread in the standard deviation for all of the above class curves except for W down to 0.246.


FP 6 6 Update

Uploaded by

Copyright:

Available Formats

FP 6 6 Update

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

FP 6 6 Update

Uploaded by

Copyright:

Available Formats

PCL 5.0.1 PIPE STRESS and FEPipe 6.

PAULIN RESEARCH GROUP

Copyright Paulin Research Group Houston, Texas 1

The Version 6.6 of FE/Pipe contains the following products:

Copyright Paulin Research Group Houston, Texas 2

Copyright Paulin Research Group Houston, Texas 3

Copyright Paulin Research Group Houston, Texas 4

Copyright Paulin Research Group Houston, Texas 5

Copyright Paulin Research Group Houston, Texas 6

Copyright Paulin Research Group Houston, Texas 7

Copyright Paulin Research Group Houston, Texas 8

Copyright Paulin Research Group Houston, Texas 9

Copyright Paulin Research Group Houston, Texas 10

Copyright Paulin Research Group Houston, Texas 11

Copyright Paulin Research Group Houston, Texas 12

Copyright Paulin Research Group Houston, Texas 13

Copyright Paulin Research Group Houston, Texas 14

Copyright Paulin Research Group Houston, Texas 15

Copyright Paulin Research Group Houston, Texas 16

Copyright Paulin Research Group Houston, Texas 17

Additional example output plots are shown in the figures below.

Copyright Paulin Research Group Houston, Texas 18

Copyright Paulin Research Group Houston, Texas 19

Mg = Gasket seating moment entered into NozzlePRO on Line 13 in place of fCrownR.

Copyright Paulin Research Group Houston, Texas 20

Do=100”; T=1”; Len=30”; (RT)1/2 = (49.5)0.5 = 7.03”. The boundary condition at

Copyright Paulin Research Group Houston, Texas 21

PCL models of multiple reactor vessels and fea branch connections.

Copyright Paulin Research Group Houston, Texas 22

Copyright Paulin Research Group Houston, Texas 23

Copyright Paulin Research Group Houston, Texas 24

Element Outline Overlays

Beams as Solid Lines

Copyright Paulin Research Group Houston, Texas 25

Copyright Paulin Research Group Houston, Texas 26

Definitive directions for local WRC107 loadings are shown below.

Copyright Paulin Research Group Houston, Texas 27

Performance enhancements include:

Images from the version of MatPRO are shown below:

Launched from Start menu:

PCL multiple temperature view.

Copyright Paulin Research Group Houston, Texas 28

Updated Fitness for Service (FFS-1) Interface

Copyright Paulin Research Group Houston, Texas 29

Start Screen (with Some SCLs defined)

SCL definition screen

Copyright Paulin Research Group Houston, Texas 30

Histogram presentation of code stresses for all defined scls

Data Table View Generator

Copyright Paulin Research Group Houston, Texas 31

Updated Load History screen editor (NozzlePRO/FE107/FETee)

Copyright Paulin Research Group Houston, Texas 32

PRG EQ. 5.0 and and B31.3 Eq. P17

PRG EQ. 5.0x for branch

Copyright Paulin Research Group Houston, Texas 33

Copyright Paulin Research Group Houston, Texas 34

General Features Note

Copyright Paulin Research Group Houston, Texas 35

Thermal bowing of horizontal, vertical and skewed pipe