Fe Safe Tutorials
Fe Safe Tutorials
3DEXPERIENCE
fe-safe 2022
TUTORIALS
©2022 Dassault Systèmes. All rights reserved. 3DEXPERIENCE®, the Compass icon, the 3DS logo, CATIA, BIOVIA, GEOVIA, SOLIDWORKS, 3DVIA, ENOVIA, EXALEAD, NETVIBES, CENTRIC PLM, 3DEXCITE, SIMULIA, DELMIA,
IFWE and MEDIDATA are commercial trademarks or registered trademarks of Dassault Systèmes, a French “société europèenne” (Versailles Commercial Register # B 322 306 440), or its subsidiaries in the United States
and/or other countries. All other trademarks are owned by their respective owners. Use of any Dassault Systèmes or its subsidiaries trademarks is subject to their express written approval.
fe-safe Tutorials
Table of Contents
101 Tutorial 101 : Importing ASCII data into fe-safe
102 Tutorial 102 : Peak picking histories
103 This tutorial has been intentionally removed
104 Tutorial 104 : Using fe-safe with I-DEAS universal .unv files
105 Tutorial 105 : Using fe-safe with Abaqus .fil files
106 Tutorial 106 : Using fe-safe with Abaqus .odb files
107 Tutorial 107 : Using fe-safe with NASTRAN .f06 files
108 Tutorial 108 : Using fe-safe with ANSYS .rst files
109 Tutorial 109 : Analysing the fatigue life of a weld
110 Tutorial 110 : Stress-based thermal fatigue analysis using multiple-temperature S-N data
111 Tutorial 111 : Modal Transient Fatigue Analysis
112 Tutorial 112 : Using fe-safe with NASTRAN .op2 files
113 Tutorial 113 : Analysis of an axially symmetric model using fe-safe/Rotate
114 Tutorial 114 : PSD Fatigue Analysis
115 Tutorial 115: Applying Theory of Critical Distance (TCD) Methods in fe-safe
116 Tutorial 116 : Random Vibration Fatigue in fe-safe
0.0 1000.0
1000.0 -50.0
-700.0 300.0
700.1 800.0
600.1 900.0
500.0 950.0
400.0 1000.0
-1000.0 300.0
150.0 200.0
Start fe-safe by selecting fe-safe from the Windows “Start” menu (Windows) or by running the script fe-
safe (UNIX).
Select an existing project, or create a new one from the welcome page
Select Open Data File from the Data Files section of the File menu and open the created data file
test1.asc. The two signals in the ASCII file will be added to the list of open data files as shown in
Figure 101-1.
Figure 101-1
Select these two signals and then press the Stack Plot(s) button in the main toolbar. This will display
the plots in Figure 101-2 as shown below.
Figure 101-2
Ensure that the data file test1.asc is loaded into the list of open data files as shown in Figure 102-1.
Figure 102-1
Select both of the data signals and from the Amplitude menu item select the Multi-channel Peak Valley
function. This will display the dialog as shown in Figure 102-2.
Figure 102-2
Select a Gate of 10 and enable the Add time information as an additional results signal check box and
press OK to start the peak picking process.
After the analysis is completed the peak picked data signals will be added to the Generated Results
section in the Loaded Data Files window as shown in Figure 102-3.
Figure 102-3
This will have reduced the 2 time histories to 6 points per signal. To view the data select the 3 new
signals and select View >> Stack Plot(s).
Figure 102-4
The X-axis shows the time basis for the new files, based on the originals. To compare the original to the
new file, select each history corresponding to signal #1 as shown in Figure 102-5.
Figure 102-5
Figure 102-6
Figure 102-7
As can be seen samples 4 and 5 from the original data signal was omitted as they were not peaks or
valleys on either of the data signals.
104 Tutorial 104 : Using fe-safe with I-DEAS universal .unv files
104.1 Introduction
The tutorials in this section demonstrate the use of fe-safe with I-DEAS universal (.unv) files, using a
Brown-Miller strain-based fatigue algorithm to evaluate the fatigue life of a uniaxial FE stress solution.
The sample files used for this tutorial are located in the directory <DataDir>\Ideas.
NOTE: The nomenclature used throughout this tutorial when referring to files and directories is as
described in Appendix B.
104.2 Preparation
Start the program by selecting fe-safe from the Windows “Start” menu (Windows) or by running the
script fe-safe (Linux).
Select an existing project, or create a new one from the welcome page
Ensure that the following basic settings are correctly configured. If fe-safe is being run for the first time
then it is possible to skip this preparatory section.
Reset any existing analysis options to their defaults in the Clear Data and Settings dialogue [Tools >>
Clear Data and Settings...], shown in Figure 104-1, below, by selecting the Select all option and clicking
OK.
Figure 104-1
Figure 104-2
The sample UNV file contains stress data at element nodes but this option will not be used as pre-
scanning will be used.
The sample model for this tutorial is one quarter of a notched shaft as shown in Figure 104-3.
Figure 104-3
To open the model, select Open Finite Element Model... from the FEA Solutions section of the File menu.
From the file selection dialog, select the sample file shaft_sae.unv from the directory
<DataDir>\Ideas. A Pre-Scan File dialogue will be displayed as shown in Figure 104-4, select Yes.
Figure 104-4
As fe-safe pre-scans the model, information about the file is written to the file:
<ProjectDir>\Model\scan
This information is also displayed in the Message Log window.
When the pre-scan of the file is complete, the Select Datasets to Read dialogue will be displayed as
shown in Figure 104-5. The file shaft_sae.unv contains only two datasets of elemental stresses, both of
which are required for the tutorial and will be checked. Select OK to load these checked datasets.
Figure 104-5
As fe-safe loads the model, information about the file and the data it contains is written to the file:
<ProjectDir>\Model\reader.log.
This information is also displayed in the Message Log window.
When the model has finished loading, the Loaded FEA Models Properties dialogue box appears, as
shown in Figure 104-6.
Figure 104-6
If the dialogue box does not appear automatically, then it can be displayed by right-clicking on the
icon in the Current FE Models window and selecting Properties.
Figure 104-7
Ensure that the units are as shown in Figure 104-6, then click OK.
A dialogue will show prompting to edit element groups loaded from the model, as shown in Figure
104-8, click No.
Figure 104-8
A summary of the open model appears in the Current FE Models window, showing the loaded datasets
and element group information – see Figure 104-9.
Figure 104-9
Note: if the window does not appear exactly as shown in Figure 104-9, then expand the tree view to
show more details.
The process of loading an FE model into fe-safe involves extracting pertinent data from the FE model,
and writing it to the working FED folder (see Appendix E). Therefore, the Current FE Models window is a
summary of the contents of the working FED folder. In this case there are two stress datasets each
containing elemental data with 1426 elements in each dataset. The source of the dataset (the filename
of the source FE model, the step, increment and timestamp) are also shown. fe-safe only extracts group
information of the same type as the loaded datasets. In this example, the datasets are elemental, so only
When a new model is loaded, the output filename automatically defaults to:
<ResultsDir>\{source_file_name}Results.{source_file_extension}
104.4 Exercise
Objective:
To perform a Brown-Miller strain-based fatigue analysis to evaluate the fatigue life of a uniaxial FE
stress solution. The fatigue load case consists of an elastically calculated unit load FEA stress solution
(i.e. a stress dataset) and some load history data. The unit load and the load history are combined using
a scale-and-combine method, as described in section 13.
Analysis process:
the stress tensors are multiplied by the time history of the applied loading, to produce a time
history of each of the 6 components of the stress tensor;
the time histories of the three principal strains are calculated from the stresses;
a multi-axial cyclic plasticity model is used to convert the elastic stress-strain histories into elastic
plastic stress-strain histories;
a ‘critical plane’ method is used to identify the most damaging plane by calculating the damage on
planes at 10° intervals between 0° and 180° in the surface of the component;
for each of the critical planes, strains are resolved onto the three shear planes (1-2, 2-3 and 1-3);
the time history of the damage parameter (which in this case, using the Brown-Miller algorithm, is
the shear and normal strain) is cycle counted;
individual fatigue cycles are identified using a ‘Rainflow’ cycle algorithm, the fatigue damage for
each cycle is calculated and the total damage is summed;
the plane with the shortest life defines the plane of crack initiation, and this life is written to the
output file.
During this calculation, fe-safe may modify the endurance limit amplitude. If all cycles (on a plane) are
below the endurance limit amplitude, there is no calculated fatigue damage on this plane. If any cycle is
damaging, the endurance limit amplitude is reduced to 25% of the constant amplitude value, and the
damage curve extended to this new endurance limit.
The critical plane method is described and illustrated in detail in the Fatigue Theory Reference Manual.
Method:
The fatigue loading will consist of a unit load stress dataset, combined with a loading history containing
five data points:
3.2
-2.2
2.2
1.3
-3.2
select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading tree;
select Clear all loadings and click Yes from the tree context menu (right mouse click on the tree) to
clear the existing loading definition;
create a loading history file by copying the above text into a text editor;
ensure that the last line is followed by a carriage-return/line-feed character (see Appendix E);
open the file in fe-safe using File >> Data Files >> Open Data File... – this adds the file to the
Loaded Data Files window;
in the Loaded Data Files window, highlight the first channel of data, labelled #1 (next to the channel
icon );
in the Current FE Models window, highlight the first dataset, labelled Dataset 1: (1.0) S : FORWARD
BEND, MODEL_SOLUTION_SOLVE, LOAD SET 1, (next to the stress dataset icon, );
Figure 104-10
The defined loading appears in the loading details list box, as shown below:
Figure 104-11
Note: if the window does not appear exactly as shown in Figure 104-11, then expand the tree view to
show more details.
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
Figure 104-12
Figure 104-13
The component material is SAE 950C (Manten). The material record SAE_950C-Manten from the material
database local.dbase will be defined for the whole component.
Opened material databases are displayed in the Material Databases window as shown in Figure 104-14,
below.
Figure 104-14
If the database local.dbase is not currently loaded in the Material Databases window, select File >>
Materials >> Open Materials Database... and select local.dbase from the directory <UserDir>. Notice
in Figure 104-14 that the database local.dbase is the user’s local copy from the <UserDir> directory.
The other two databases shown are from the <DatabaseDir> directory.
Expand the tree view to show the materials contained in local.dbase. The contents of the database can
be filtered to show only materials of a particular type, by selecting one of the filter icons . To ensure
that all materials in the database are displayed double-click on the All filter icon .
The parameters for a material can be displayed by extending the material name.
To define SAE_950C-Manten for the whole component:
highlight the material SAE_950C-Manten in the local.dbase database;
double-click the Material column header - a Change Material? confirmation dialogue box appears;
Figure 104-15
click YES;
the material name should appear for all groups in the Material column.
Figure 104-16
For SAE 950C (Manten) the default algorithm is Brown-Miller Biaxial Strain Life using Morrow mean-
stress correction.
Figure 104-17
When the FE model was loaded, the output filename automatically defaulted to:
<ResultsDir>\shaft_saeResults.unv
The output filename can be modified either by manually editing the Output File field in the Fatigue from
FEA dialogue, or by clicking the adjacent browse button: .
Figure 104-18
Check that the analysis is configured as shown in Figure 104-18, and then click Continue.
Note that the UTS of the material is used to evaluate Kt from the defined surface finish curve (i.e. surface
finish factor versus UTS). In this case, for a mirror-polished surface, Kt =1 for all values of UTS.
As the analysis is being performed, the following information is written to the analysis log file. The
analysis log file has the same file name as the output file, except that the extension is .log. So, for this
analysis, the analysis log file is:
<ResultsDir>\shaft_saeResults.log
This information is also displayed in the Message Log window and includes:
Summary
=======
Worst Life-Repeats : 3463.065
at Element 677.10
Analysis time : 0:00:08
The analysis log shows that the worst-case life for the whole model is:
3463 repeats of the fatigue loading cycle, at element 677, node 10.
Note:
fe-safe adopts the conventional method of expressing the endurance limit, as the number of reversals,
or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ). In fe-safe, the
design life and the calculated fatigue life always refer to the number of repeats of the complete defined
fatigue loading cycle. An optional conversion factor can be used to convert the fatigue life in repeats to
fatigue life with respect to some other quantity, for example hours or miles (see section 13.2).
Figure 104-20: Critical plane normals at worst element scaled by log1e7 – logNf
105.1 Introduction
This tutorial demonstrates the use of fe-safe with Abaqus .fil files.
The tutorial includes opening the FE model and configuring multi-axial analyses of two superimposed
fatigue load cases and of a dataset sequence, including Factors of Strength (FOS) evaluation.
The sample files used for this tutorial are located in the directory <DataDir>\Abaqus. ASCII and binary
versions of the sample files are available. The ASCII FIL file format is portable between platforms. The
binary FIL file format is not portable between platforms.
This tutorial uses the ASCII version of the sample FIL file.
NOTE: The nomenclature used throughout this tutorial when referring to files and directories is as
described in Appendix B.
105.2 Preparation
Start the program by selecting fe-safe from the Windows “Start” menu (Windows) or by running the
script fe-safe (Linux).
Select an existing project, or create a new one from the welcome page
Ensure that the following basic settings are correctly configured. If fe-safe is being run for the first time
then it is possible to skip this preparatory section.
Reset any existing analysis options to their defaults in the Clear Data and Settings dialogue [Tools >>
Clear Data and Settings...], shown in Figure 105-1 below, by selecting the Check all option and clicking
OK.
Figure 105-1
Figure 105-2
The sample FIL file contains two different types of stress data: nodal-averaged data and stresses at
element nodes. This tutorial uses data for stresses at element nodes. As the model will be pre-scanned
selection of the data type is done in the Select Datasets to Read dialogue described below.
The FIL interface can import and export data from ASCII and binary FIL files. The sample file for this
tutorial is in ASCII format. It is recommended that the Auto Detect format option be used.
The type of variable that fe-safe uses to export lives and FOS results to is configured in the Export
Lives/FOS to FIL section of the dialogue, found by clicking on the Export tab.. The preferred export
variable is UVARM, since this is a variable that most viewers can interpret. If the UVARM variable is
specified, both the fatigue life, or log10 (fatigue life), and the FOS are exported to the same step. If any
other variable is used, a new step is added for each fe-safe export variable.
Patran users should click the Patran As Viewer button. This button automatically changes the following
settings:
the export variable is set to TEMP [variable 2], because Patran does not support the UVARM variable;
the Export logarithmic lives to results file option, Analysis Options dialogue, Export tab, is
automatically selected, since Patran will not handle large values (e.g. 1015), that may exist if the
fatigue life is exported rather than log10(fatigue life).
Figure 105-3
To open the model select Open Finite Element Model... from the File menu. This will display the file
selection dialogue. Select the sample file keyhole.fil from the directory <DataDir>\Abaqus. A Pre-
Scan File dialogue will be displayed as shown in Figure 105-4, select Yes.
Figure 105-4
As fe-safe pre-scans the model, information about the file is written to the file:
<ProjectDir>\Model\scan
This information is also displayed in the Message Log window.
When the pre-scan of the file is complete, the Select Datasets to Read dialogue will be displayed as
shown in Figure 105-5. The file keyhole.fil contains only two datasets of elemental stresses, both of
which are required for the tutorial and will be checked. Select OK to load these checked datasets.
Figure 105-5
As fe-safe loads the model, information about the file and the data it contains is appended to the file:
<ProjectDir>\Model\reader.log.
This information is also displayed in the Message Log window.
When the model has finished loading, the Loaded FEA Models Properties dialogue box appears, as
shown in Figure 105-6.
Figure 105-6
If the dialogue box does not appear automatically, then it can be displayed by right-clicking on the
icon in the Current FE Models window and selecting Properties.
Figure 105-7
Ensure that the units are as shown in Figure 105-6, then click OK.
A dialogue will show prompting to edit element groups loaded from the model, as shown in Figure
105-8, click No.
Figure 105-8
A summary of the open model appears in the Current FE Models window, showing the loaded datasets
and element group information – see Figure 105-9.
Figure 105-9
Note: if the window does not appear exactly as shown in Figure 105-9, then expand the tree view to
show more details.
The process of loading an FE model into fe-safe involves extracting pertinent data from the FE model,
and writing it to the working FED folder (see Appendix E). Therefore, the Current FE Models window is a
summary of the contents of the working FED folder. In this case there are two stress datasets each
containing elemental data with 502 elements in each dataset. The source of the dataset (the filename of
the source FE model, the step, increment and timestamp) are also shown.
fe-safe only extracts group information of the same type as the loaded datasets. In this example, the
datasets are elemental, so only elemental groups are shown.
When a new model is loaded, the output filename automatically defaults to:
<ResultsDir>\{source_file_name}Results.{source_file_extension}
The output filename can be modified either by manually editing the Output File field in the Fatigue from
FEA dialogue, or by clicking the adjacent browse button: .
Objective:
To perform a multiaxial analysis of two superimposed fatigue load cases. Each load case consists of an
elastically calculated unit load FEA stress solution (i.e. a stress dataset) and some load history data. The
unit load and the load history are combined using a scale-and-combine method, as described in
section 13.
Analysis process:
the stress tensors are multiplied by the time history of the applied loading, to produce a time
history of each of the 6 components of the stress tensor;
the time histories of the three principal strains are calculated from the stresses;
a multiaxial cyclic plasticity model is used to convert the elastic stress-strain histories into elastic
plastic stress-strain histories;
a ‘critical plane’ method is used to identify the most damaging plane by calculating the damage on
planes at 10° intervals between 0° and 180° in the surface of the component;
for each of the critical planes, strains are resolved onto the three shear planes (1-2, 2-3 and 1-3);
the time history of the damage parameter (which in this case, using the Brown-Miller algorithm, is
the shear and normal strain) is cycle counted;
individual fatigue cycles are identified using a ‘Rainflow’ cycle algorithm, the fatigue damage for
each cycle is calculated and the total damage is summed;
the plane with the shortest life defines the plane of crack initiation, and this life is written to the
output file.
During this calculation, fe-safe may modify the endurance limit amplitude. If all cycles (on a plane) are
below the endurance limit amplitude, there is no calculated fatigue damage on this plane. If any cycle is
damaging, the endurance limit amplitude is reduced to 25% of the constant amplitude value, and the
The critical plane method is described and illustrated in detail in the Fatigue Theory Reference Manual,
section 7.5.
Method:
For this exercise two fatigue load cases will be superimposed. Each load case will consist of an
elastically calculated unit load stress dataset and some load history data, as follows:
the first dataset (Dataset 1: (1.1) S : Unit Y Load) will be combined with channel 1 of the data file
test_mcg2.amc;
the second dataset (Dataset 2: (2.1) S : Unit X Load) will be combined with channel 2 of the data file
test_mcg2.amc.
To open the loading history data, select Open Data File from the File menu. Use the file selection
dialogue to select the sample file test_mcg2.amc from the <DataDir> directory. The data file is
summarised in the Loaded Data Files window. The sample file contains two channels of load data,
identified in the tree view by the channel icon, .
Figure 105-10
To combine the unit load stress dataset with the loading history for the first case:
in the Current FE Models window, highlight the first dataset, labelled Dataset 1: (1.1) S : Unit Y Load,
(next to the stress dataset icon, );
in the Loaded Data Files window, highlight the first channel of data, labelled fe-safe tutorial scaler
#1, (next to the channel icon );
select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading tree;
select Clear all loadings from the context-sensitive menu (right mouse click) to clear the existing
loading definition;
To combine the unit load stress dataset with the loading history for the second case:
in the Current FE Models window, highlight the second dataset, labelled Dataset 2: (2.1) S : Unit X
Load, (next to the stress dataset icon, );
in the Loaded Data Files window, highlight the second channel of data, labelled fe-safe tutorial
scaler #2, (next to the channel icon );
The defined loading appears in the loading details list box, as shown below:
Figure 105-11
Note: if the window does not appear exactly as shown in Figure 105-11, then expand the tree view to
show more details.
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
Figure 105-12
Figure 105-13
The component material is SAE 950C (Manten). The material record SAE_950C-Manten from the material
database local.dbase will be defined for the whole component.
Opened material databases are displayed in the Material Databases window as shown in Figure 105-14,
below.
Figure 105-14
If the database local.dbase is not currently loaded in the Material Databases window, select File >>
Open Materials Database... and select local.dbase from the directory <UserDir>. Notice in Figure
105-14 that the database local.dbase is the user’s local copy from the <UserDir> directory. The
other two databases shown are from the <DatabaseDir> directory.
Expand the tree view to show the materials contained in local.dbase. The contents of the database can
be filtered to show only materials of a particular type, by selecting one of the filter icons . To ensure
that all materials in the database are displayed double-click on the All filter icon .
The parameters for a material can be displayed by expanding the material name.
To define SAE_950C-Manten for the whole component:
highlight the material SAE_950C-Manten in the local.dbase database;
double-click the Material column header - a Change Material? confirmation dialogue box appears;
Figure 105-15
click YES;
the material name should appear for all groups in the Material column.
Figure 105-16
For SAE 950C (Manten) the default algorithm is Brown-Miller Biaxial Strain Life using Morrow mean-
stress correction.
Figure 105-17
When the FE model was loaded, the output filename automatically defaulted to:
<ResultsDir>\keyholeResults.csv
Before running the analysis, change the output filename to:
<ResultsDir>\keyholeResults_ex01.fil
The output filename can be modified either by manually editing the Output File field in the Fatigue from
FEA dialogue, or by clicking the adjacent browse button: .
Figure 105-18
Check that the analysis is configured as shown in Figure 105-18, and then click Continue.
Note that the UTS of the material is used to evaluate Kt from the defined surface finish curve (i.e. surface
finish factor versus UTS). In this case, for a mirror-polished surface, Kt =1 for all values of UTS.
As the analysis is being performed, the following information is written to the analysis log file. The
analysis log file has the same file name as the output file, except that the extension is .log. So, for this
analysis, the analysis log file is:
<ResultsDir>\keyholeResults_ex01.log
This information is also displayed in the Message Log window and includes.
Summary
=======
Worst Life-Repeats : 30603.336
at Element 144.2
Analysis time : 0:00:03
The analysis log shows that the worst-case life for the whole model is:
30603 repeats of the fatigue loading cycle, at element 144, node 2.
Note:
fe-safe adopts the conventional method of expressing the endurance limit, as the number of reversals,
or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ). In fe-safe, the
design life and the calculated fatigue life always refer to the number of repeats of the complete defined
fatigue loading cycle. An optional conversion factor can be used to convert the fatigue life in repeats to
fatigue life with respect to some other quantity, for example hours or miles (see section 13).
1. In this exercise, the two fatigue load cases are superimposed. Therefore, it does not matter in
which order the loads are defined. For example:
Loading Loading is 1 Repeats
DS#1 * test_mcg2.amc:1
DS#2 * test_mcg2.amc:2
is the same as:
Loading Loading is 1 Repeats
DS#2 * test_mcg2.amc:2
DS#1 * test_mcg2.amc:1
2. In this example, the load history is very short, consisting of only five data points. For short load
histories it is often more convenient to define the loading using a ‘user-defined’ load history,
rather than reading a load history from a data file (see section 13.8.2). This is achieved using
the A user-defined LOAD * dataset option.
To re-define the loading for this exercise using user-defined load histories:
select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading
tree;
select Clear all loadings and click Yes from the context-sensitive menu (right mouse click)
to clear the existing loading definition;
in the Loaded Data Files window, expand the filename test_mcg2.amc, so that both
channels of data are displayed;
highlight both channels of data (to select multiple channels highlight the first channel by
clicking on it with the left mouse button, then hold down the CTRL key on the keyboard
while clicking on any additional files);
display a numerical listing of the two data channels by selecting View >> Numerical Listing
Figure 105-20
highlight the first unit stress dataset (labelled Dataset 1: (1.1) S : Unit Y Load) in the Current
FE Models window;
select Add... >> A user-defined LOAD * dataset to display the Dataset Embedded Load
History dialogue, and copy the five data points from the first column of the numerical listing
display (scaler #1) into the Loading Scale box:
Figure 105-21
click OK;
highlight the second unit stress dataset (labelled Dataset 2: (2.1) S : Unit X Load) in the
Current FE Models window;
define a load sequence for the second case, as described above, using the data from the
second column of the numerical listing display (scaler #2).
The new loading definition appears in the Loading Details area of the Fatigue from FEA dialogue,
as shown in Figure 105-22, below.
Figure 105-22
Note: if the window does not appear exactly as shown in Figure 105-22, then expand the tree
view to show more details.
3. The load definition (LDF) file can be used to define simple and complex loading scenarios.
Section 13 includes a full description of the LDF file, including syntax.
The loading for this exercise can be seen by saving the loading to a load definition file:
select File >> Loadings >> Save Current FEA Loadings As...;
open the file in a text editor to display the contents (comments may vary):
INIT
END
# Block number 1
BLOCK
lh=0.0099465 -0.00696255 0.00696255 0.0039786 -0.0099465 , ds=1
lh=-0.000369029 0.00221417 0.00590447 0.00738059 0.00221417 , ds=2
END
4. In this exercise the surface finish was defined by selecting a surface type (in this case “mirror-
polished”) from a surface finish database. If the actual surface finish factor, Kt, for the
component is known (for “mirror-polished”, the Kt factor has a value of 1) then it can be
specified as a value. To define a user-defined value for Kt, select Define Kt as a value, in the
Surface Finish Definition dialogue box:
Figure 105-23
5. The results data can be exported to an alternative file format at the end of the analysis, by
changing the extension of the output file (Fatigue from FEA >> Output File) to a recognised
type.
For example:
to export to an ASCII CSV results file, change the extension to .txt, .csv or .asc (this is
Alternatively, after the analysis is complete, and the FIL file has been created, the results data
can be written to another output file in addition to the original output FIL file, using the Save
FEA Fatigue Results As... option in the File menu. This option displays the Save FEA Fatigue
Results As... dialogue box, as shown in Figure 105-24, below:
Figure 105-24
The FER (.fer) file (see Appendix E) is an intermediate file containing the results of the analysis.
This new output file type is determined by the extension of the file entered in the Output File
box. A copy of the original output FIL file can be created if the extension .fil is used.
6. To analyse individual elements or nodes, list the required items in the Exports and Outputs
dialogue, and select the Only analyse listed items dialogue. Additional outputs and diagnostics
options can also be specified. Diagnostics and additional outputs are discussed in detail in
section 22.
7. To save the current analysis configuration, select Save FEA Fatigue Definition File..., and enter a
filename ending with the extension .stlx. The analysis can be reloaded at a later date by
restoring the keyword file, using Open FEA Fatigue Definition File....
105.5 Exercise 2 : Multiaxial analysis of a data sets sequence, including Factor Of Strength evaluation
Objective:
To perform a multiaxial analysis of a series of events – in this case a sequence of two elastically
calculated FEA stress solution (i.e. stress datasets).
The analysis will also include an evaluation of the Factor of Strength (FOS). This is the factor by which
the loading must be scaled to achieve a specified design life. The FOS method is described in detail in
section 17.
Analysis process:
- if the calculated life is lower than the design life, the elastic stresses at the node are scaled by a
factor less than 1.0;
- if the calculated life is greater than the design life, the elastic stresses at the node are scaled by
a factor greater than 1.0.
the elastic stress history is recalculated using the re-scaled nodal stresses;
for local strain analysis, the cyclic plasticity model is used to recalculate the time history of elastic-
plastic stress-strains - the fatigue life is then recalculated;
(ii) the step change of 0.01 or 0.1 in the FOS value causes the design life to be bracketed, or
(iii) the FOS exceeds the max factor (default 2.0) or is less than the min factor (default 0.5).
Method:
The series of events being analysed in this exercise consists of a sequence of two elastically calculated
FEA stress solution (i.e. stress datasets) scaled by a defined scale factor.
the first dataset (Dataset 1: (1.1) S : Unit Y Load) multiplied by a scale factor of 0.01,
followed by:
the second dataset (Dataset 2: (2.1) S : Unit X Load) multiplied by a scale factor of 0.01.
The loading for this loading sequence can be defined in a single loading block:
select Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading tree;
select Clear all loadings and click Yes from the tree context menu (right mouse click on the tree) to
clear the existing loading definition;
highlight the first unit stress dataset in the Current FE Models window;
highlight the second unit stress dataset in the Current FE Models window;
highlight the dataset item in the loading tree of the Fatigue from FEA dialogue;
select Scale from the tree context menu (right mouse click on the tree) and enter the value 0.01;
INIT
END
# Block number 1
BLOCK
ds=1, scale=0.01
ds=2, scale=0.01
END
The new loading definition appears in the Loading Details area of the Fatigue from FEA dialogue, as
shown in Figure 105-25.
Figure 105-25
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
click on the Factor of Strength... button in the Fatigue from FEA window to open the Factors of
Strength Calculations dialogue box:
Figure 105-26
select the Perform Factor of Strength (FOS) Calculations? option to enable the FOS analysis.;
select Infinite design life (use material’s Endurance Limit) – this uses the constant amplitude
endurance limit for the component material, which for SAE 950C (Manten) is: 2.0 × 107 reversals;
Note: fe-safe adopts the conventional method of expressing the endurance limit, as the number of
reversals, or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ).
click OK;
specify the FOS band limits in the Analysis Options dialogue [FEA Fatigue >> Analysis Options...]
Safety Factors tab as shown in Figure 105-27, to determine the limits and precision of the FOS
analysis:
- between the ‘fine’ limits, the FOS is calculated to a resolution of approximately 0.01 repeats;
- between maximum and minimum, the FOS is calculated to a resolution of approximately 0.1
repeats;
- if the evaluated FOS is greater than 2, then a FOS of 2 is exported for the node;
- if the evaluated FOS is less than 0.5, then a FOS of 0.5 is exported for the node.
Figure 105-27
Figure 105-28
Check that the analysis is configured as shown in Figure 105-28, and then click Continue.
Note that the UTS of the material is used to evaluate Kt from the defined surface finish curve (i.e. surface
finish factor versus UTS). In this case, for a mirror-polished surface, Kt =1 for all values of UTS.
As the analysis is being performed, the following information is written to the analysis log file. The
analysis log file has the same file name as the output file, except that the extension is .log. So, for this
analysis, the analysis log file is:
<ResultsDir>\keyholeResults_ex02.log
This information is also displayed in the Message Log window and includes the following:
Summary
=======
Worst Life-Repeats : 555040.5
at Element 67.2
Worst FOS@Life=Infinite : 0.669
at Element 65.2
Analysis time : 0:00:05
In the configuration of the Factor of Strength analysis, the design life specified was the ‘Infinite design
life’, which is assumed to be the materials constant amplitude endurance limit, which for SAE 950C
(Manten) is: 2.0 × 107 reversals.
The worst-case fatigue lives are of the order 5.8 × 105 repeats of the fatigue loading cycle, which is less
than the design life (1 × 107 repeats of the loading). This is why the worst-case strength factor (the
factor by which the loading must be scaled to achieve the design life) is somewhat less than 1. The FOS
method is described in detail in section 17.
Note:
fe-safe adopts the conventional method of expressing the endurance limit, as the number of reversals,
or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ). In fe-safe, the
design life and the calculated fatigue life always refer to the number of repeats of the complete defined
fatigue loading cycle. An optional conversion factor can be used to convert the fatigue life in repeats to
fatigue life with respect to some other quantity, for example hours or miles (see section 13).
The first set of exported fatigue results in the file contains the fatigue lives, which should look similar to
Figure 105-29, the viewer colour scheme may need to be set to "Reversed Rainbow" and any averaging
of values at nodes should be disabled.
Figure 105-29
The second set of exported fatigue results in the file contains the strength factors (FOS), which should
look similar to Figure 105-30, the viewer colour scheme may need to be set to "Reversed Rainbow" and
any averaging of values at nodes should be disabled.
Figure 105-30
106.1 Introduction
This tutorial demonstrates the use of fe-safe with Abaqus .odb files.
The tutorial includes opening the FE model and configuring multi-axial analyses of two superimposed
fatigue load cases and of a dataset sequence, including Factors of Strength (FOS) evaluation.
The sample files used for this tutorial are located in the directory <DataDir>\Abaqus. The ODB file
format is portable between platforms.
NOTE: The nomenclature used throughout this tutorial when referring to files and directories is as
described in Appendix B, section 202.2.
106.2 Preparation
Start the program by selecting fe-safe from the Windows “Start” menu (Windows) or by running the
script fe-safe (Linux) – (see section 5.2).
Select an existing project, or create a new one from the welcome page
Ensure that the following basic settings are correctly configured. If fe-safe is being run for the first time
then it is possible to skip this preparatory section.
Reset any existing analysis options to their defaults in the Clear Data and Settings dialogue [Tools >>
Clear Data and Settings...], shown in Figure 106-1, below, by selecting the Check all option and clicking
OK.
Figure 106-1
Figure 106-2
To open the model, select Open Finite Element Model... from the FEA Solutions section of the File menu.
This will display the file selection dialogue. Select the sample file keyhole_xxx.odb from the directory
<DataDir>\Abaqus.xxx
A Pre-Scan File dialogue will be displayed as shown in Figure 106-3, select Yes.
Figure 106-3
As fe-safe loads the model, information about the file and the data it contains is written to the file:
<ProjectDir>\Model\scan
This information is also displayed in the Message Log window.
When the pre-scan of the file is complete, the Select Datasets to Read dialogue will be displayed as
shown in Figure 106-4. The file keyhole_xxx.odb contains four datasets of elemental stresses. The
strain and temperature datasets as well as stress datasets for the initial increment do not need to be
loaded. Check only the Stresses and Last increment only boxes, click Apply to Dataset List to apply the
selection and select OK to load the remaining stress datasets.
Figure 106-4
As fe-safe loads the model, information about the file and the data it contains is appended to the file:
<ProjectDir>\Model\reader.log.
This information is also displayed in the Message Log window.
When the model has finished loading, the Loaded FEA Models Properties dialogue box appears, as
shown in Figure 106-5.
Figure 106-5
If the dialogue box does not appear automatically, then it can be displayed by right-clicking on the
icon in the Current FE Models window and selecting Properties.
Figure 106-6
Ensure that the units are as shown in Figure 106-5, then click OK.
A dialogue will show prompting to edit element groups loaded from the model, as shown in Figure
106-7, click No.
Figure 106-7
A summary of the open model appears in the Current FE Models window, showing the loaded datasets
and element group information – see Figure 106-8.
Figure 106-8
Note: if the window does not appear exactly as shown in Figure 106.3-6, then expand the tree view to
show more details.
The process of loading an FE model into fe-safe involves extracting pertinent data from the FE model,
and writing it to the working FED folder (see Appendix E). Therefore, the Current FE Models window is a
summary of the contents of the working FED folder. In this case there are two stress datasets each
containing elemental data with 502 elements in each dataset. The source of the dataset (the filename of
the source FE model, the step, increment and timestamp) are also shown
fe-safe also extracts group information of the same type as the loaded datasets. In this example, the
datasets are elemental, so only elemental groups are shown.
When a new model is loaded, the output filename automatically defaults to:
<ResultsDir>\{source_file_name}Results.{source_file_extension}
The output filename can be modified either by manually editing the Output File field in the Fatigue from
FEA dialogue, or by clicking the adjacent browse button: .
To perform a multiaxial analysis of two superimposed fatigue load cases. Each load case consists of an
elastically calculated unit load FEA stress solution (i.e. a stress dataset) and some load history data. The
unit load and the load history are combined using a scale-and-combine method, as described in
section 13.
Analysis process:
the stress tensors are multiplied by the time history of the applied loading, to produce a time
history of each of the 6 components of the stress tensor;
the time histories of the three principal strains are calculated from the stresses;
a multiaxial cyclic plasticity model is used to convert the elastic stress-strain histories into elastic
plastic stress-strain histories;
a ‘critical plane’ method is used to identify the most damaging plane by calculating the damage on
planes at 10° intervals between 0° and 180° in the surface of the component;
for each of the critical planes, strains are resolved onto the three shear planes (1-2, 2-3 and 1-3);
the time history of the damage parameter (which in this case, using the Brown-Miller algorithm, is
the shear and normal strain) is cycle counted;
individual fatigue cycles are identified using a ‘Rainflow’ cycle algorithm, the fatigue damage for
each cycle is calculated and the total damage is summed;
the plane with the shortest life defines the plane of crack initiation, and this life is written to the
output file.
During this calculation, fe-safe may modify the endurance limit amplitude. If all cycles (on a plane) are
below the endurance limit amplitude, there is no calculated fatigue damage on this plane. If any cycle is
damaging, the endurance limit amplitude is reduced to 25% of the constant amplitude value, and the
damage curve extended to this new endurance limit.
The critical plane method is described and illustrated in detail in the Fatigue Theory Reference Manual,
section 7.5.
Method:
For this exercise two fatigue load cases will be superimposed. Each load case will consist of an
elastically calculated unit load stress dataset and some load history data, as follows:
the first dataset (Dataset 1: (1.1) S : Unit Y Load) will be combined with channel 1 of the data file
test_mcg2.amc;
the second dataset (Dataset 2: (2.1) S : Unit X Load) will be combined with channel 2 of the data file
test_mcg2.amc.
To open the loading history data, select Open Data File from the Data Files section of the File menu. Use
the file selection dialogue to select the sample file test_mcg2.amc from the <DataDir> directory. The
data file is summarised in the Loaded Data Files window. The sample file contains two channels of load
data, identified in the tree view by the channel icon, .
Figure 106-9
To combine the unit load stress dataset with the loading history for the first case:
in the Current FE Models window, highlight the first dataset, labelled Dataset 1: (1.1) S : Unit Y Load,
(next to the stress dataset icon, );
in the Loaded Data Files window, highlight the first channel of data, labelled fe-safe tutorial scaler
#1, (next to the channel icon );
select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading tree;
select Clear all loadings from the context-sensitive menu (right click) to clear the existing loading
definition;
To combine the unit load stress dataset with the loading history for the second case:
in the Current FE Models window, highlight the second dataset, labelled Dataset 2: (2.1) S : Unit X
Load, (next to the stress dataset icon, );
in the Loaded Data Files window, highlight the second channel of data, labelled fe-safe tutorial
scaler #2, (next to the channel icon ).
The defined loading appears in the loading details list box, as shown below:
Figure 106-10
Note: if the window does not appear exactly as shown in Figure 106-10, then expand the tree view to
show more details.
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
Figure 106-11
Figure 106-12
The component material is SAE 950C (Manten). The material record SAE_950C-Manten from the material
database local.dbase will be defined for the whole component.
Opened material databases are displayed in the Material Databases window as shown in Figure 106-13,
below.
Figure 106-13
If the database local.dbase is not currently loaded in the Material Databases window, select File >>
Materials >> Open Materials Database... and select local.dbase from the directory <UserDir>. Notice
in Figure 106-13 that the database local.dbase is the user’s local copy from the <UserDir> directory.
The other two databases shown are from the <DatabaseDir> directory.
Expand the tree view to show the materials contained in local.dbase. The contents of the database can
be filtered to show only materials of a particular type, by selecting one of the filter icons . To ensure
that all materials in the database are displayed double-click on the All filter icon .
The parameters for a material can be displayed by expanding the material name.
To define SAE_950C-Manten for the whole component:
highlight the material SAE_950C-Manten in the local.dbase database;
double-click the Material column header - a Change Material? confirmation dialogue box appears;
Figure 106-14
click YES;
the material name should appear for all groups in the Material column.
Figure 106-15
For SAE 950C (Manten) the default algorithm is Brown-Miller Biaxial Strain Life using Morrow mean-
stress correction.
Figure 106-16
When the FE model was loaded, the output filename automatically defaulted to:
<ResultsDir>\keyhole_xxxResults.odb
Before running the analysis, change the output filename to:
<ResultsDir>\keyhole_xxxResults_ex01.odb
The output filename can be modified either by manually editing the Output File field in the Fatigue from
FEA dialogue, or by clicking the adjacent browse button: .
Figure 106-17
Check that the analysis is configured as shown in Figure 106-17, and then click Continue.
Note that the UTS of the material is used to evaluate Kt from the defined surface finish curve (i.e. surface
finish factor versus UTS). In this case, for a mirror-polished surface, Kt =1 for all values of UTS.
As the analysis is being performed, the following information is written to the analysis log file. The
analysis log file has the same file name as the output file, except that the extension is .log. So, for this
analysis, the analysis log file is:
<ResultsDir>\keyhole_xxxResults_ex01.log
This information is also displayed in the Message Log window and includes:
Summary
=======
Worst Life-Repeats : 30603.336
at Element [0]144.2
Analysis time : 0:00:00
Note:
fe-safe adopts the conventional method of expressing the endurance limit, as the number of reversals,
or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ). In fe-safe, the
design life and the calculated fatigue life always refer to the number of repeats of the complete defined
fatigue loading cycle. An optional conversion factor can be used to convert the fatigue life in repeats to
fatigue life with respect to some other quantity, for example hours or miles (see section 13).
In the last step of the file <ResultsDir>/keyhole_xxxResults_ex01.odb, the results for the exported
variable should look similar to Figure 106-18, the viewer colour scheme may need to be set to
"Reversed Rainbow" and any averaging of values at nodes should be disabled.
1. In this exercise, the two fatigue load cases are superimposed. Therefore, it does not matter in
which order the loads are defined. For example:
Loading Loading is 1 Repeats
DS#1 * test_mcg2.amc:1
DS#2 * test_mcg2.amc:2
is the same as:
Loading Loading is 1 Repeats
DS#2 * test_mcg2.amc:2
DS#1 * test_mcg2.amc:1
2. In this example, the load history is very short, consisting of only five data points. For short load
histories it is often more convenient to define the loading using a ‘user-defined’ load history,
rather than reading a load history from a data file (see section 13.8.2). This is achieved using
the A user-defined LOAD * dataset option.
To re-define the loading for this exercise using user-defined load histories:
select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading
tree;
select Clear all loadings from the context sensitive menu (right-click) to clear the existing
loading definition;
in the Loaded Data Files window expand the filename test_mcg2.amc, so that both
channels of data are displayed;
highlight both channels of data (to select multiple channels highlight the first channel by
clicking on it with the left mouse button, then hold down the CTRL key on the keyboard
while clicking on any additional files);
display a numerical listing of the two data channels by selecting View >> Numerical Listing
Figure 106-19
highlight the first unit stress dataset (labelled Dataset 1: (1.1) S : Unit Y Load) in the Current
FE Models window;
select Add... >> A user-defined LOAD * dataset to display the Dataset Embedded Load
History dialogue, and copy the five data points from the first column of the numerical listing
display (scaler #1) into the Loading Scale box:
Figure 106-20
click OK;
highlight the second unit stress dataset (labelled Dataset 2: (2.1) S : Unit X Load) in the
Currrent FE Models window;
define a load sequence for the second case, as described above, using the data from the
second column of the numerical listing display (scaler #2).
The new loading definition appears in the Loading Details area of the Fatigue from FEA dialogue,
as shown in Figure 106-21, below.
Figure 106-21
Note: if the window does not appear exactly as shown in Figure 106-21, then expand the tree
view to show more details.
3. The load definition (LDF) file can be used to define simple and complex loading scenarios.
Section 13 includes a full description of the LDF file, including syntax.
The loading for this exercise can be seen by saving the loading to a load definition file:
select File >> Loadings >> Save Current FEA Loadings As...;
open the file in a text editor to display the contents (comments may vary):
INIT
END
# Block number 1
BLOCK
lh=0.0099465 -0.00696255 0.00696255 0.0039786 -0.0099465 , ds=1
lh=-0.000369029 0.00221417 0.00590447 0.00738059 0.00221417 , ds=2
END
4. In this exercise the surface finish was defined by selecting a surface type (in this case “mirror-
polished”) from a surface finish database. If the actual surface finish factor, Kt, for the
component is known (for “mirror-polished”, the Kt factor has a value of 1) then it can be
specified as a value. To define a user-defined value for Kt, select Define Kt as a value, in the
Surface Finish Definition dialogue box:
Figure 106-22
5. The results data can be exported to an alternative file format at the end of the analysis, by
changing the extension of the output file (Fatigue from FEA >> Output File) to a recognised
type.
For example:
to export to an ASCII CSV results file, change the extension to .txt, .csv or .asc (this is
useful for viewing a table of results in an editor or spreadsheet).
Alternatively, after the analysis is complete, and the ODB file has been created, the results data
can be written to another output file in addition to the original output ODB file, using the Save
FEA Fatigue Results As... option in the File menu. This option displays the Save FEA Fatigue
Results As... dialogue box, as shown in Figure 106-23, below:
Figure 106-23
The FER (.fer) file (see Appendix E) is an intermediate file containing the results of the analysis.
This new output file type is determined by the extension of the file entered in the Output File
box. A copy of the original output ODB file can be created if the extension .odb is used.
6. To analyse individual elements or nodes, list the required items in the Exports and Outputs
dialogue, and select the Only analyse listed items dialogue. Additional outputs and diagnostics
options can also be specified. Diagnostics and additional outputs are discussed in detail in
section 22.
7. To save the current analysis configuration, select Save FEA Fatigue Definition File..., and enter a
filename ending with the extension .stlx. The analysis can be reloaded at a later date by
restoring the keyword file, using Open FEA Fatigue Definition File....
106.5 Exercise 2 : Multiaxial analysis of a data sets sequence, including Factor Of Strength
evaluation
Objective:
To perform a multiaxial analysis of a series of events – in this case a sequence of two elastically
calculated FEA stress solution (i.e. stress datasets).
The analysis will also include an evaluation of the Factor of Strength (FOS). This is the factor by which
the loading must be scaled to achieve a specified design life. The FOS method is described in detail in
section 17.
Analysis process:
- if the calculated life is lower than the design life, the elastic stresses at the node are scaled by a
factor less than 1.0;
- if the calculated life is greater than the design life, the elastic stresses at the node are scaled by
a factor greater than 1.0.
the elastic stress history is recalculated using the re-scaled nodal stresses;
for local strain analysis, the cyclic plasticity model is used to recalculate the time history of elastic-
plastic stress-strains - the fatigue life is then recalculated;
(ii) the step change of 0.01 or 0.1 in the FOS value causes the design life to be bracketed, or
(iii) the FOS exceeds the max factor (default 2.0) or is less than the min factor (default 0.5).
Method:
The series of events being analysed in this exercise consists of a sequence of two elastically calculated
FEA stress solution (i.e. stress datasets) scaled by a defined scale factor.
the first dataset (Dataset 1: (1.1) S : Unit Y Load) multiplied by a scale factor of 0.01,
followed by:
the second dataset (Dataset 2: (2.1) S : Unit X Load) multiplied by a scale factor of 0.01.
The loading for this loading sequence can be defined in a single loading block:
select Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading tree;
select Clear all loadings and click Yes from the tree context menu (right mouse click on the tree) to
clear the existing loading definition;
highlight the first unit stress dataset in the Current FE Models window;
highlight the second unit stress dataset in the Current FE Models window;
highlight the dataset item in the loading tree of the Fatigue from FEA dialogue;
select Scale from the tree context menu (right mouse click on the tree) and enter the value 0.01;
INIT
END
# Block number 1
BLOCK n=1
ds=1, scale=0.01
ds=2, scale=0.01
END
The new loading definition appears in the Loading Details area of the Fatigue from FEA dialogue, as
shown in Figure 106-24.
Figure 106-24
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
click on the Factor of Strength... button in the Fatigue from FEA window to open the Design Life
(Factors of Strength Calculations) dialogue box:
Figure 106-25
select the Perform Factor of Strength (FOS) Calculations? option to enable the FOS analysis.;
select Infinite design life (use material’s Endurance Limit) – this uses the constant amplitude
endurance limit for the component material, which for SAE 950C (Manten) is: 2.0 × 107 reversals;
Note: fe-safe adopts the conventional method of expressing the endurance limit, as the number of
reversals, or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ).
click OK;
specify the FOS band limits in the Analysis Options dialogue [FEA Fatigue >> Analysis Options...]
Safety Fators tab as shown in Figure 106-26, to determine the limits and precision of the FOS
analysis:
- between the ‘fine’ limits, the FOS is calculated to a resolution of approximately 0.01 repeats;
- between maximum and minimum, the FOS is calculated to a resolution of approximately 0.1
repeats;
- if the evaluated FOS is greater than 2, then a FOS of 2 is exported for the node;
- if the evaluated FOS is less than 0.5, then a FOS of 0.5 is exported for the node.
Figure 106-26
Figure 106-27
Check that the analysis is configured as shown in Figure 106-27, and then click Continue.
Note that the UTS of the material is used to evaluate Kt from the defined surface finish curve (i.e. surface
finish factor versus UTS). In this case, for a mirror-polished surface, Kt =1 for all values of UTS.
As the analysis is being performed, the following information is written to the analysis log file. The
analysis log file has the same file name as the output file, except that the extension is .log. So, for this
analysis, the analysis log file is:
<ResultsDir>\keyhole_xxxResults_ex02.log
This information is also displayed in the Message Log window and includes:
Summary
=======
Worst Life-Repeats : 555040.5
at Element [0]67.2
Worst FOS@Life=Infinite : 0.669
at Element [0]65.2
Analysis time : 0:00:00
In the configuration of the Factor of Strength analysis, the design life specified was the ‘Infinite design
life’, which is assumed to be the materials constant amplitude endurance limit, which for SAE 950C
(Manten) is: 2.0 × 107 reversals.
The worst-case fatigue lives are of the order 5.8 × 105 repeats of the fatigue loading cycle, which is less
than the design life (1 × 107 repeats of the loading). This is why the worst-case strength factor (the
factor by which the loading must be scaled to achieve the design life) is somewhat less than 1. The FOS
method is described in detail in section 17.
Note:
fe-safe adopts the conventional method of expressing the endurance limit, as the number of reversals,
or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ). In fe-safe, the
design life and the calculated fatigue life always refer to the number of repeats of the complete defined
fatigue loading cycle. An optional conversion factor can be used to convert the fatigue life in repeats to
fatigue life with respect to some other quantity, for example hours or miles (see section 13).
The first set of exported fatigue results in the file contains the fatigue lives, which should look similar to
Figure 106-28, the viewer colour scheme may need to be set to "Reversed Rainbow" and any averaging
of values at nodes should be disabled.
Figure 106-28
The second set of exported fatigue results in the file contains the strength factors (FOS), which should
look similar to Figure 106-29, the viewer colour scheme may need to be set to "Reversed Rainbow" and
any averaging of values at nodes should be disabled.
Figure 106-29
107.1 Introduction
This tutorial demonstrates the use of fe-safe with NASTRAN results (*.f06) files.
The tutorial includes opening the FE model and configuring multi-axial analyses of two superimposed
fatigue load cases and of a dataset sequence, including Factors of Strength (FOS) evaluation.
The sample files used for this tutorial are located in the directory <DataDir>\Nastran.
NOTE: The nomenclature used throughout this tutorial when referring to files and directories is as
described in Appendix B.
107.2 Preparation
Start the program by selecting fe-safe from the Windows “Start” menu (Windows) or by running the
script fe-safe (Linux).
Select an existing project, or create a new one from the welcome page
Ensure that the following basic settings are correctly configured. If fe-safe is being run for the first time
then it is possible to skip this preparatory section.
Reset any existing analysis options to their defaults in the Clear Data and Settings dialogue [Tools >>
Clear Data and Settings...], shown in Figure 107-1, below, by selecting the Check all option and clicking
OK.
Figure 107-1
Figure 107-2
To open the model select Open Finite Element Model... from the File menu. This will display the file
selection dialogue. Select the sample file keyhole_01.f06 from the directory <DataDir>\Nastran. The
Pre-Scan File dialogue will be displayed as shown in Figure 107-3, select Yes.
Figure 107-3
As fe-safe pre-scans the model, information about the file is written to the file:
<ProjectDir>\Model\scan
This information is also displayed in the Message Log window.
When the pre-scan of the file is complete, the Select Datasets to Read dialogue will be displayed as
shown in Figure 107-4. The file keyhole_01.f06 contains four datasets, two of which are stress and
two are strain datasets. This tutorial only requires the stress datasets, the strain datasets do not need to
be loaded. Check the Stresses and Select last increment only boxes, click Apply to Dataset List to apply
the selection. Change the Available Positions combo box to Elemental & Centroidal and select OK to load
the stress datasets.
As fe-safe loads the model, information about the file and the data it contains is appended to the file:
<ProjectDir>\Model\reader.log.
This information is also displayed in the Message Log window:
Figure 107-4
When the model has finished loading, the Loaded FEA Models Properties dialogue box appears, as
shown in Figure 107-5.
Figure 107-5
If the dialogue box does not appear automatically, then it can be displayed by right-clicking on the
icon in the Current FE Models window and selecting Properties.
Figure 107-6
Ensure that the units are as shown in Figure 107-5, then click OK.
A dialogue will show prompting to edit element groups loaded from the model, as shown in Figure
107-7, click No.
Figure 107-7
A summary of the open model appears in the Current FE Models window, showing the loaded datasets
and element group information – see Figure 107-8.
Figure 107-8
Note: if the window does not appear exactly as shown in Figure 107-8, then expand the tree view to
show more details.
The process of loading an FE model into fe-safe involves extracting pertinent data from the FE model,
and writing it to the working FED folder (see Appendix E). Therefore, the Current FE Models window is a
summary of the contents of the working FED folder. In this case there are two stress datasets each
containing elemental and centroidal data with 1004 elements (502 on the top shell and 502 on the
bottom shell) in each dataset. The source of the datasets (the filename of the source FE model, the step,
increment and timestamp) are also shown.
fe-safe also extracts group information from the ‘SETS’ defined in the Case Control deck of the .f06
file.
When a new model is loaded, the output filename automatically defaults to:
<ResultsDir>\{source_file_name}Results.{default_output_file_extension}
For NASTRAN .f06 files, the default output file type is .csv, so in this case, the output filename
automatically defaults to:
<ResultsDir>\ keyhole_01Results.csv
The output filename can be modified either by manually editing the Output File field in the Fatigue from
FEA dialogue, or by clicking the adjacent browse button: .
To perform a multiaxial analysis of two superimposed fatigue load cases. Each load case consists of an
elastically calculated unit load FEA stress solution (i.e. a stress dataset) and some load history data. The
unit load and the load history are combined using a scale-and-combine method, as described in
section 13.
Analysis process:
the stress tensors are multiplied by the time history of the applied loading, to produce a time
history of each of the 6 components of the stress tensor;
the time histories of the three principal strains are calculated from the stresses;
a multiaxial cyclic plasticity model is used to convert the elastic stress-strain histories into elastic
plastic stress-strain histories;
a ‘critical plane’ method is used to identify the most damaging plane by calculating the damage on
planes at 10° intervals between 0° and 180° in the surface of the component;
for each of the critical planes, strains are resolved onto the three shear planes (1-2, 2-3 and 1-3);
the time history of the damage parameter (which in this case, using the Brown-Miller algorithm, is
the shear and normal strain) is cycle counted;
individual fatigue cycles are identified using a ‘Rainflow’ cycle algorithm, the fatigue damage for
each cycle is calculated and the total damage is summed;
the plane with the shortest life defines the plane of crack initiation, and this life is written to the
output file.
During this calculation, fe-safe may modify the endurance limit amplitude. If all cycles (on a plane) are
below the endurance limit amplitude, there is no calculated fatigue damage on this plane. If any cycle is
damaging, the endurance limit amplitude is reduced to 25% of the constant amplitude value, and the
damage curve extended to this new endurance limit.
The critical plane method is described and illustrated in detail in the Fatigue Theory Reference Manual,
section 7.5.
Method:
For this exercise two fatigue load cases will be superimposed. Each load case will consist of an
elastically calculated unit load stress dataset and some load history data, as follows:
the first dataset (Dataset 1: (1.1) S : KEYHOLE TUTORIAL LS1 – CS1) will be combined with channel 1
of the data file test_mcg2.amc;
the second dataset (Dataset 2: (2.1) S : KEYHOLE TUTORIAL LS2 – CS2) will be combined with
channel 2 of the data file test_mcg2.amc.
To open the loading history data, select Open Data File from the File menu. Use the file selection
dialogue to select the sample file test_mcg2.amc from the <DataDir> directory. The data file is
summarised in the Loaded Data Files window. The sample file contains two channels of load data,
identified in the tree view by the channel icon, .
Figure 107-9
To combine the unit load stress dataset with the loading history for the first case:
select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading tree;
select Clear all loadings and click Yes from the tree context menu (right mouse click on the tree) to
clear the existing loading definition;
in the Current FE Models window, highlight the first dataset, labelled Dataset 1: (1.1) S : KEYHOLE
TUTORIAL LS1 – CS1, (next to the stress dataset icon, );
in the Loaded Data Files window, highlight the first channel of data, labelled fe-safe tutorial scaler
#1, (next to the channel icon );
To combine the unit load stress dataset with the loading history for the second case:
in the Current FE Models window, highlight the second dataset, labelled Dataset 2: (2.1) S : KEYHOLE
TUTORIAL LS2 – CS2, (next to the stress dataset icon, );
in the Loaded Data Files window, highlight the second channel of data, labelled fe-safe tutorial
scaler #2, (next to the channel icon ).
The defined loading appears in the loading details list box, as shown below:
Figure 107-10
Note: if the window does not appear exactly as shown in Figure 107-10, then expand the tree view to
show more details.
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
Figure 107-11
Figure 107-12
The component material is SAE 950C (Manten). The material record SAE_950C-Manten from the material
database local.dbase will be defined for the whole component.
Opened material databases are displayed in the Material Databases window as shown in Figure 107-13,
below.
Figure 107-13
If the database local.dbase is not currently loaded in the Material Databases window, select File >>
Materials >> Open Materials Database... and select local.dbase from the directory <UserDir>. Notice
in Figure 107-13 that the database local.dbase is the user’s local copy from the <UserDir> directory.
The other two databases shown are from the <DatabaseDir> directory.
Expand the tree view to show the materials contained in local.dbase. The contents of the database can
be filtered to show only materials of a particular type, by selecting one of the filter icons . To ensure
that all materials in the database are displayed double-click on the All filter icon .
The parameters for a material can be displayed by clicking the symbol next to the material name.
To define SAE_950C-Manten for the whole component:
highlight the material SAE_950C-Manten in the local.dbase database;
double-click the Material column header - a Change Material? confirmation dialogue box appears;
Figure 107-14
click YES;
the material name should appear for all groups in the Material column.
Step 5: Define the analysis algorithm:
The material database identifies a default analysis algorithm for each material. The default analysis
algorithm is normally selected automatically. To confirm that the default algorithm will be used for the
whole model (i.e. all element groups):
double-click the Algorithm column header to open the in Group Algorithm Selection dialogue box
for all groups;
Figure 107-15
For SAE 950C (Manten) the default algorithm is Brown-Miller Biaxial Strain Life using Morrow mean-
stress correction.
Figure 107-16
When the FE model was loaded, the output filename automatically defaulted to:
<ResultsDir>\keyhole_01Results.csv
The output filename can be modified either by manually editing the Output File field in the Fatigue from
FEA dialogue, or by clicking the adjacent browse button: .
Figure 107-17
Check that the analysis is configured as shown in Figure 107-17, and then click Continue.
Note that the UTS of the material is used to evaluate Kt from the defined surface finish curve (i.e. surface
finish factor versus UTS). In this case, for a mirror-polished surface, Kt =1 for all values of UTS.
As the analysis is being performed, the following information is written to the analysis log file. The
analysis log file has the same file name as the output file, except that the extension is .log. So, for this
analysis, the analysis log file is:
<ResultsDir>\keyhole_01Results.log
This information is also displayed in the Message Log window and includes
Summary
=======
Worst Life-Repeats : 42191.711
at Element 1.4:1
Analysis time : 0:00:03
Note:
fe-safe adopts the conventional method of expressing the endurance limit, as the number of reversals,
or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ). In fe-safe, the
design life and the calculated fatigue life always refer to the number of repeats of the complete defined
fatigue loading cycle. An optional conversion factor can be used to convert the fatigue life in repeats to
fatigue life with respect to some other quantity, for example hours or miles (see section 13).
1. In this exercise, the two fatigue load cases are superimposed. Therefore, it does not matter in
which order the loads are defined. For example:
Loading Loading is 1 Repeats
DS#1 * test_mcg2.amc:1
DS#2 * test_mcg2.amc:2
is the same as:
Loading Loading is 1 Repeats
DS#2 * test_mcg2.amc:2
DS#1 * test_mcg2.amc:1
2. In this example, the load history is very short, consisting of only five data points. For short load
histories it is often more convenient to define the loading using a ‘user-defined’ load history,
rather than reading a load history from a data file (see section 13.8.2). This is achieved using
the A user-defined LOAD * dataset option.
To re-define the loading for this exercise using user-defined load histories:
select Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading tree;
select Clear all loadings and click Yes from the tree context menu (right mouse click on the
tree) to clear the existing loading definition;
in the Loaded Data Files window expand the filename test_mcg2.amc, so that both
channels of data are displayed;
highlight both channels of data (to select multiple channels highlight the first channel by
clicking on it with the left mouse button, then hold down the CTRL key on the keyboard
while clicking on any additional files);
display a numerical listing of the two data channels by selecting View >> Numerical Listing
Figure 107-19
highlight the first unit stress dataset (labelled Dataset 1: (1.1) S : KEYHOLE TUTORIAL LS1 –
CS1) in the Current FE Models window;
select Add... >> A user-defined LOAD * dataset to display the Dataset Embedded Load
History dialogue, and copy the five data points from the first column of the numerical listing
display (scaler #1) into the Loading Scale box:
Figure 107-20
click OK;
highlight the second unit stress dataset (labelled Dataset 2: (2.1) S : KEYHOLE TUTORIAL LS2
– CS2) in the Current FE Models window;
define a load sequence for the second case, as described above, using the data from the
second column of the numerical listing display (scaler #2).
The new loading definition appears in the Loading Details area of the Fatigue from FEA dialogue,
as shown in Figure 107-21, below.
Figure 107-21
Note: if the window does not appear exactly as shown in Figure 107-21, then expand the tree
view to show more details.
3. The load definition (LDF) file can be used to define simple and complex loading scenarios.
Section 13 includes a full description of the LDF file, including syntax.
The loading for this exercise can be seen by saving the loading to a load definition file:
select File >> Loadings >> Save Current FEA Loadings As...;
open the file in a text editor to display the contents (comments may vary):
INIT
END
# Block number 1
BLOCK n=1
lh=0.0099465 -0.00696255 0.00696255 0.0039786 -0.0099465 , ds=1
lh=-0.000369029 0.00221417 0.00590447 0.00738059 0.00221417 , ds=2
END
4. In this exercise the surface finish was defined by selecting a surface type (in this case “mirror-
polished”) from a surface finish database. If the actual surface finish factor, Kt, for the
component is known (for “mirror-polished”, the Kt factor has a value of 1) then it can be
specified as a value. To define a user-defined value for Kt, select Define Kt as a value, in the
Surface Finish Definition dialogue box:
Figure 107-22
5. The results data can be exported to an alternative file format at the end of the analysis, by
changing the extension of the output file (Fatigue from FEA >> Output File) to a recognised
type.
For example:
to export to an ASCII CSV results file, change the extension to .txt, .csv or .asc (this is
useful for viewing a table of results in an editor or spreadsheet).
Alternatively, after the analysis is complete, and the .csv file has been created, the results data
can be written to another output file in addition to the original output .csv file, using the Save
FEA Fatigue Results As... option in the File menu. This option displays the Save FEA Fatigue
Results As... dialogue box, as shown in Figure 107-23, below:
Figure 107-23
The FER (.fer) file (see Appendix E) is an intermediate file containing the results of the analysis.
This new output file type is determined by the extension of the file entered in the Output File
box. A copy of the original output .csv file can be created if the extension .csv is used.
6. To analyse individual elements or nodes, list the required items in the Exports and Outputs
dialogue, and select the Only analyse listed items dialogue. Additional outputs and diagnostics
options can also be specified. Diagnostics and additional outputs are discussed in detail in
section 22.
7. To save the current analysis configuration, select Save FEA Fatigue Definition File..., and enter a
filename ending with the extension .stlx. The analysis can be reloaded at a later date by
restoring the keyword file, using Open FEA Fatigue Definition File..., .
107.5 Exercise 2 : Multiaxial analysis of a data sets sequence, including Factor Of Strength
evaluation
Objective:
To perform a multiaxial analysis of a series of events – in this case a sequence of two elastically
calculated FEA stress solution (i.e. stress datasets).
The analysis will also include an evaluation of the Factor of Strength (FOS). This is the factor by which
the loading must be scaled to achieve a specified design life. The FOS method is described in detail in
section 17.
Analysis process:
- if the calculated life is lower than the design life, the elastic stresses at the node are scaled by a
factor less than 1.0;
- if the calculated life is greater than the design life, the elastic stresses at the node are scaled by
a factor greater than 1.0.
the elastic stress history is recalculated using the re-scaled nodal stresses;
for local strain analysis, the cyclic plasticity model is used to recalculate the time history of elastic-
plastic stress-strains - the fatigue life is then recalculated;
(ii) the step change of 0.01 or 0.1 in the FOS value causes the design life to be bracketed, or
(iii) the FOS exceeds the max factor (default 2.0) or is less than the min factor (default 0.5).
Method:
The series of events being analysed in this exercise consists of a sequence of two elastically calculated
FEA stress solution (i.e. stress datasets) scaled by a defined scale factor.
the first dataset (Dataset 1: (1.1) S : KEYHOLE TUTORIAL LS1 – CS1) multiplied by a scale factor of
0.02,
followed by:
the second dataset (Dataset 2: (2.1) S : KEYHOLE TUTORIAL LS2 – CS2) multiplied by a scale factor of
0.02.
The loading for this loading sequence can be defined in a single loading block:
select Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading tree;
select Clear all loadings and click Yes from the tree context menu (right mouse click on the tree) to
clear the existing loading definition;
highlight the first unit stress dataset in the Current FE Models window;
highlight the second unit stress dataset in the Current FE Models window;
highlight the dataset item in the loading tree of the Fatigue from FEA dialogue;
select Scale from the tree context menu (right mouse click on the tree) and enter the value 0.02;
INIT
END
# Block number 1
BLOCK
ds=1, scale=0.02
ds=2, scale=0.02
END
The new loading definition appears in the Loading Details area of the Fatigue from FEA dialogue, as
shown in Figure 107-24.
Figure 107-24
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
click on the Factor of Strength... button in the Fatigue from FEA window to open the Factors of
Strength Calculations dialogue box:
Figure 107-25
select the Perform Factor of Strength (FOS) Calculations? option to enable the FOS analysis.;
select Infinite design life (use material’s Endurance Limit) – this uses the constant amplitude
endurance limit for the component material, which for SAE 950C (Manten) is: 2.0 × 107 reversals;
Note: fe-safe adopts the conventional method of expressing the endurance limit, as the number of
reversals, or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ).
click OK;
specify the FOS band limits in the Analysis Options dialogue [FEA Fatigue >> Analysis Options...]
Safety Factors tab as shown in Figure 107-26, to determine the limits and precision of the FOS
analysis:
- between the ‘fine’ limits, the FOS is calculated to a resolution of approximately 0.01 repeats;
- between maximum and minimum, the FOS is calculated to a resolution of approximately 0.1
repeats;
- if the evaluated FOS is greater than 2, then a FOS of 2 is exported for the node;
- if the evaluated FOS is less than 0.5, then a FOS of 0.5 is exported for the node.
Figure 107-26
Figure 107-27
Check that the analysis is configured as shown in Figure 107-27, and then click Continue.
Note that the UTS of the material is used to evaluate Kt from the defined surface finish curve (i.e. surface
finish factor versus UTS). In this case, for a mirror-polished surface, Kt =1 for all values of UTS.
As the analysis is being performed, the following information is written to the analysis log file. The
analysis log file has the same file name as the output file, except that the extension is .log. So, for this
analysis, the analysis log file is:
<ResultsDir>\keyhole_01Results_ex02.log
This information displayed in the Message Log window includes the following
Summary
=======
Worst Life-Repeats : 433610.125
at Element 1.4:1
Worst FOS@Life=Infinite : 0.631
at Element 1.4:1
Analysis time : 0:00:09
Step 10: Reviewing the results
The analysis log shows that the worst-case life for the whole model is:
433620 repeats of the loading, at element 1, node 4, shell section 1.
In the configuration of the Factor of Strength analysis, the design life specified was the ‘Infinite design
life’, which is assumed to be the materials constant amplitude endurance limit, which for SAE 950C
(Manten) is: 2.0 × 107 reversals.
The worst-case fatigue lives are of the order 1.067 × 106 repeats of the fatigue loading cycle, which is
less than the design life (1 × 107 repeats of the loading). This is why the worst-case strength factor (the
factor by which the loading must be scaled to achieve the design life) is somewhat less than 1. The FOS
method is described in detail in section 17.
Note:
fe-safe adopts the conventional method of expressing the endurance limit, as the number of reversals,
or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ). In fe-safe, the
design life and the calculated fatigue life always refer to the number of repeats of the complete defined
fatigue loading cycle. An optional conversion factor can be used to convert the fatigue life in repeats to
fatigue life with respect to some other quantity, for example hours or miles (see section 13).
Figure 107-28
The strength factors (FOS), should look similar to Figure 107-29, the viewer colour scheme may need to
be set to "Reversed Rainbow" and any averaging of values at nodes should be disabled.
Figure 107-29
108.1 Introduction
This tutorial demonstrates the use of fe-safe with the ANSYS results (.rst) file.
The use of ANSYS results .rst files in fe-safe is discussed in detail in Appendix G.
The tutorial includes opening the FE model and configuring multi-axial analyses of two superimposed
fatigue load cases and of a dataset sequence, including Factors of Strength (FOS) evaluation.
The sample files used for this tutorial are located in the directory <DataDir>\Ansys.
NOTE: The nomenclature used throughout this tutorial when referring to files and directories is as
described in Appendix B.
108.2 Preparation
Start the program by selecting fe-safe from the Windows “Start” menu (Windows) or by running the
script fe-safe (Linux).
Select an existing project, or create a new one from the welcome page
Ensure that the following basic settings are correctly configured. If fe-safe is being run for the first time
then it is possible to skip this preparatory section.
Reset any existing analysis options to their defaults in the Clear Data and Settings dialogue [Tools >>
Clear Data and Settings...], shown in Figure 108.2-1, below, by selecting the Check all option and
clicking OK.
Figure 108.2-1
Figure 108.3-1
To open the model, select Open Finite Element Model... from the File menu. This will display the file
selection dialogue. Select the sample file keyhole_01.rst from the directory <DataDir>\Ansys. The
Pre-Scan File dialogue will be displayed as shown in Figure 108.3-2, select Yes.
Figure 108.3-2
As fe-safe pre-scans the model, information about the file is written to the file:
<ProjectDir>\Model\scan
This information is also displayed in the Message Log window.
When the pre-scan of the file is complete, the Select Datasets to Read dialogue will be displayed as
shown in Figure 108.3-3. The file keyhole_01.rst contains two datasets of elemental stresses that are
required for this tutorial. The strain, temperature and forces datasets do not need to be loaded. Check
only the Stresses and Select last increment only boxes, click Apply to Dataset List to apply the selection
and select OK to load the remaining stress datasets.
As fe-safe loads the model, information about the file and the data it contains is appended to the file:
<ProjectDir>\Model\reader.log.
This information is also displayed in the Message Log window.
Figure 108.3-3
When the model has finished loading, the Loaded FEA Models Properties dialogue box appears, as
shown in Figure 108.3-4.
Figure 108.3-4
If the dialogue box does not appear automatically, then it can be displayed by right-clicking on the
icon in the Open FE Models window and selecting Properties.
Figure 108.3-5
Ensure that the units are as shown in Figure 108.3-4, then click OK.
A dialogue will show prompting to edit element groups loaded from the model, as shown in Figure
108.3-6, click No.
Figure 108.3-6
A summary of the open model appears in the Open FE Models window, showing the loaded datasets and
element group information – see Figure 108.3-7.
Figure 108.3-7
Note: if the window does not appear exactly as shown in Figure 108.3-7, then expand the tree view to
show more details.
The process of loading an FE model into fe-safe involves extracting pertinent data from the FE model,
and writing it to the working FED folder (see Appendix E). Therefore, the Open FE Models window is a
summary of the contents of the working FED folder. In this case there are two stress datasets each
containing elemental data with 502 elements in each dataset. The source of the dataset (the filename of
the source FE model, the step, increment and timestamp) are also shown.
For ANSYS models, element groups can be defined by setting different regions in the model to have
different material numbers (although the material data can be the same). fe-safe then extracts group
information from the RST file by grouping elements that have the same ANSYS material number.
When a new model is loaded, the output filename automatically defaults to:
<ResultsDir>\{source_file_name}Results.{source_file_extension}
To perform a multiaxial analysis of two superimposed fatigue load cases. Each load case consists of an
elastically calculated unit load FEA stress solution (i.e. a stress dataset) and some load history data. The
unit load and the load history are combined using a scale-and-combine method, as described in
section 13.
Analysis process:
the stress tensors are multiplied by the time history of the applied loading, to produce a time
history of each of the 6 components of the stress tensor;
the time histories of the three principal strains are calculated from the stresses;
a multiaxial cyclic plasticity model is used to convert the elastic stress-strain histories into elastic
plastic stress-strain histories;
a ‘critical plane’ method is used to identify the most damaging plane by calculating the damage on
planes at 10° intervals between 0° and 180° in the surface of the component;
for each of the critical planes, strains are resolved onto the three shear planes (1-2, 2-3 and 1-3);
the time history of the damage parameter (which in this case, using the Brown-Miller algorithm, is
the shear and normal strain) is cycle counted;
individual fatigue cycles are identified using a ‘Rainflow’ cycle algorithm, the fatigue damage for
each cycle is calculated and the total damage is summed;
the plane with the shortest life defines the plane of crack initiation, and this life is written to the
output file.
During this calculation, fe-safe may modify the endurance limit amplitude. If all cycles (on a plane) are
below the endurance limit amplitude, there is no calculated fatigue damage on this plane. If any cycle is
damaging, the endurance limit amplitude is reduced to 25% of the constant amplitude value, and the
damage curve extended to this new endurance limit.
The critical plane method is described and illustrated in detail in the Fatigue Theory Reference Manual,
section 7.5.
Method:
For this exercise two fatigue load cases will be superimposed. Each load case will consist of an
elastically calculated unit load stress dataset and some load history data, as follows:
the first dataset (Dataset 1: (1.1) S : Keyhole) will be combined with channel 1 of the data file
test_mcg2.amc;
the second dataset (Dataset 2: (2.1) S : Keyhole) will be combined with channel 2 of the data file
test_mcg2.amc.
To open the loading history data, select Open Data File from the File menu. Use the file selection
dialogue to select the sample file test_mcg2.amc from the <DataDir> directory. The data file is
summarised in the Loaded Data Files window. The sample file contains two channels of load data,
identified in the tree view by the channel icon .
Figure 108.4-1
To combine the unit load stress dataset with the loading history for the first case:
select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading tree;
select Clear all loadings and click Yes from the tree context menu (right mouse click on the tree) to
clear the existing loading definition
in the Current FE Models window, highlight the first dataset, labelled Dataset 1: (1.1) S : Keyhole,
(next to the stress dataset icon, );
in the Loaded Data Files window, highlight the first channel of data, labelled fe-safe tutorial scaler
#1, (next to the channel icon );
To combine the unit load stress dataset with the loading history for the second case:
in the Current FE Models window, highlight the second dataset, labelled Dataset 2: (2.1) S : Keyhole,
(next to the stress dataset icon, );
in the Loaded Data Files window, highlight the second channel of data, labelled fe-safe tutorial
scaler #2, (next to the channel icon ).
The defined loading appears in the loading details list box, as shown below:
Figure 108.4-2
Note: if the window does not appear exactly as shown in Figure 108.4-2, then expand the tree view to
show more details.
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
Figure 108.4-3
Figure 108.4-4
The component material is SAE 950C (Manten). The material record SAE_950C-Manten from the material
database local.dbase will be defined for the whole component.
Opened material databases are displayed in the Material Databases window as shown in Figure 108.4-5,
below.
Figure 108.4-5
If the database local.dbase is not currently loaded in the Material Databases window, select File >>
Materials >> Open Materials Database... and select local.dbase from the directory <UserDir>. Notice
in Figure 108.4-5 that the database local.dbase is the user’s local copy from the <UserDir> directory.
The other two databases shown are from the <DatabaseDir> directory.
Expand the tree view to show the materials contained in local.dbase. The contents of the database can
be filtered to show only materials of a particular type, by selecting one of the filter icons . To ensure
that all materials in the database are displayed double-click on the All filter icon .
The parameters for a material can be displayed by extending the material name.
To define SAE_950C-Manten for the whole component:
highlight the material SAE_950C-Manten in the local.dbase database;
double-click the Material column header - a Change Material? confirmation dialogue box appears;
check the confirmation dialogue – it should say: Are you sure you want to change the material for
the groups ‘Default, Material 1, Material 2’ to ‘SAE_950C-Manten’ in database ‘local.dbase’? ;
click YES;
the material name should appear for all groups in the Material column.
Figure 108.4-6
For SAE 950C (Manten) the default algorithm is Brown-Miller Biaxial Strain Life using Morrow mean-
stress correction.
Figure 108.4-7
When the FE model was loaded, the output filename automatically defaulted to:
<ResultsDir>\keyhole_01Results.rst
Before running the analysis, change the output filename to:
<ResultsDir>\keyhole_01Results_ex01.rst
The output filename can be modified either by manually editing the Output File field in the Fatigue from
FEA dialogue, or by clicking the adjacent browse button: .
Figure 108.4-8
Check that the analysis is configured as shown in Figure 108.4-8, and then click Continue.
Note that the UTS of the material is used to evaluate Kt from the defined surface finish curve (i.e. surface
finish factor versus UTS). In this case, for a mirror-polished surface, Kt =1 for all values of UTS.
As the analysis is being performed, the following information is written to the analysis log file. The
analysis log file has the same file name as the output file, except that the extension is .log. So, for this
analysis, the analysis log file is:
<ResultsDir>\keyhole_01Results_ex01.log
This information is also displayed in the Message Log window and includes
Summary
=======
Worst Life-Repeats : 122466.609
at Element 1.3
Analysis time : 0:00:05
Note:
fe-safe adopts the conventional method of expressing the endurance limit, as the number of reversals,
or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ). In fe-safe, the
design life and the calculated fatigue life always refer to the number of repeats of the complete defined
fatigue loading cycle. An optional conversion factor can be used to convert the fatigue life in repeats to
fatigue life with respect to some other quantity, for example hours or miles (see section 13).
1. In this exercise, the two fatigue load cases are superimposed. Therefore, it does not matter in
which order the loads are defined. For example:
Loading Loading is 1 Repeats
DS#1 * test_mcg2.amc:1
DS#2 * test_mcg2.amc:2
is the same as:
Loading Loading is 1 Repeats
DS#2 * test_mcg2.amc:2
DS#1 * test_mcg2.amc:1
2. In this example, the load history is very short, consisting of only five data points. For short load
histories it is often more convenient to define the loading using a ‘user-defined’ load history,
rather than reading a load history from a data file (see section 13.8.2). This is achieved using
the A user-defined LOAD * dataset option.
To re-define the loading for this exercise using user-defined load histories:
select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading
tree;
select Clear all loadings and click Yes from the tree context menu (right mouse click on the
tree) to clear the existing loading definition;
in the Loaded Data Files window expand the filename test_mcg2.amc, so that both
channels of data are displayed;
highlight both channels of data (to select multiple channels highlight the first channel by
clicking on it with the left mouse button, then hold down the CTRL key on the keyboard
while clicking on any additional files);
display a numerical listing of the two data channels by selecting View >> Numerical Listing
Figure 108.4-10
highlight the first unit stress dataset (labelled Dataset 1: (1.1) S : Keyhole) in the Current FE
Models window;
select Add... >> A user-defined LOAD * dataset to display the Dataset Embedded Load
History dialogue, and copy the five data points from the first column of the numerical listing
display (scaler #1) into the Loading Scale box:
Figure 108.4-11
click OK;
highlight the second unit stress dataset (labelled Dataset 2: (2.1) S : Keyhole) in the Current
FE Models window;
define a load sequence for the second case, as described above, using the data from the
second column of the numerical listing display (scaler #2).
The new loading definition appears in the Loading Details area of the Fatigue from FEA dialogue,
as shown in Figure 108.4-12, below.
Figure 108.4-12
Note: if the window does not appear exactly as shown in Figure 108.4-12, then expand the tree
view to show more details.
3. The load definition (LDF) file can be used to define simple and complex loading scenarios.
Section 13 includes a full description of the LDF file, including syntax.
The loading for this exercise can be seen by saving the loading to a load definition file:
select File >> Loadings >> Save Current FEA Loadings As...;
open the file in a text editor to display the contents (comments may vary):
INIT
END
# Block number 1
BLOCK
lh=0.0099465 -0.00696255 0.00696255 0.0039786 -0.0099465 , ds=1
lh=-0.000369029 0.00221417 0.00590447 0.00738059 0.00221417 , ds=2
END
4. In this exercise the surface finish was defined by selecting a surface type (in this case “mirror-
polished”) from a surface finish database. If the actual surface finish factor, Kt, for the
component is known (for “mirror-polished”, the Kt factor has a value of 1) then it can be
specified as a value. To define a user-defined value for Kt, select Define Kt as a value, in the
Surface Finish Definition dialogue box:
Figure 108.4-13
5. The results data can be exported to an alternative file format at the end of the analysis, by
changing the extension of the output file (Fatigue from FEA >> Output File) to a recognised
type.
For example:
to export to an ASCII CSV results file, change the extension to .txt, .csv or .asc (this is
useful for viewing a table of results in an editor or spreadsheet).
Alternatively, after the analysis is complete, and the RST file has been created, the results data
can be written to another output file in addition to the original output RST file, using the Save
FEA Fatigue Results As... option in the File menu. This option displays the Save FEA Fatigue
Results As... dialogue box, as shown in Figure 108.4-14, below:
Figure 108.4-14
The FER (.fer) file (see Appendix E) is an intermediate file containing the results of the analysis.
This new output file type is determined by the extension of the file entered in the Output File
box. A copy of the original output RST file can be created if the extension .rst is used.
6. To analyse individual elements or nodes, list the required items in the Exports and Outputs
dialogue, and select the Only analyse listed items dialogue. Additional outputs and diagnostics
options can also be specified. Diagnostics and additional outputs are discussed in detail in
section 22.
7. To save the current analysis configuration, select Save FEA Fatigue Definition File..., and enter a
filename ending with the extension .stlx. The analysis can be reloaded at a later date by
restoring the keyword file, using Open FEA Fatigue Definition File..., .
108.5 Exercise 2 : Multiaxial analysis of a data sets sequence, including Factor Of Strength
evaluation
Objective:
To perform a multiaxial analysis of a series of events – in this case a sequence of two elastically
calculated FEA stress solution (i.e. stress datasets).
The analysis will also include an evaluation of the Factor of Strength (FOS). This is the factor by which
the loading must be scaled to achieve a specified design life. The FOS method is described in detail in
section 17.
Analysis process:
- if the calculated life is lower than the design life, the elastic stresses at the node are scaled by a
factor less than 1.0;
- if the calculated life is greater than the design life, the elastic stresses at the node are scaled by
a factor greater than 1.0.
the elastic stress history is recalculated using the re-scaled nodal stresses;
for local strain analysis, the cyclic plasticity model is used to recalculate the time history of elastic-
plastic stress-strains - the fatigue life is then recalculated;
(ii) the step change of 0.01 or 0.1 in the FOS value causes the design life to be bracketed, or
(iii) the FOS exceeds the max factor (default 2.0) or is less than the min factor (default 0.5).
Method:
The series of events being analysed in this exercise consists of a sequence of two elastically calculated
FEA stress solution (i.e. stress datasets) scaled by a defined scale factor.
the first dataset (Dataset 1: (1.1) S : Keyhole) multiplied by a scale factor of 0.03,
followed by:
the second dataset (Dataset 2: (2.1) S : Keyhole) multiplied by a scale factor of 0.03.
The loading for this loading sequence can be defined in a single loading block:
select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading tree;
select Clear all loadings and click Yes from the tree context menu (right mouse click on the tree) to
clear the existing loading definition;
highlight the first unit stress dataset in the Current FE Models window;
highlight the second unit stress dataset in the Current FE Models window;
highlight the dataset item in the loading tree of the Fatigue from FEA dialogue;
select Scale from the tree context menu (right mouse click on the tree) and enter the value 30000;
INIT
END
# Block number 1
BLOCK
ds=1, scale=0.03
ds=2, scale=0.03
END
The new loading definition appears in the Loading Details area of the Fatigue from FEA dialogue, as
shown in Figure 108.5-1.
Figure 108.5-1
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
As in Exercise 1, the component is assumed to be made from SAE 950C (Manten) – see 108.4
(Exercise 1), Step 4.
click on the Factor of Strength... button in the Fatigue from FEA window to open the Factor of
Strength Calculations dialogue box:
Figure 108.5-2
select the Perform Factor of Strength (FOS) Calculations? option to enable the FOS analysis.;
select Infinite design life (use material’s Endurance Limit) – this uses the constant amplitude
endurance limit for the component material, which for SAE 950C (Manten) is: 2.0 × 107 reversals;
Note: fe-safe adopts the conventional method of expressing the endurance limit, as the number of
reversals, or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ).
click OK;
specify the FOS band limits in the Analysis Options dialogue [FEA Fatigue >> Analysis Options...]
Safety Factors tab as shown in Figure 108.5-3, to determine the limits and precision of the FOS
analysis:
- between the ‘fine’ limits, the FOS is calculated to a resolution of approximately 0.01 repeats;
- between maximum and minimum, the FOS is calculated to a resolution of approximately 0.1
repeats;
- if the evaluated FOS is greater than 2, then a FOS of 2 is exported for the node;
- if the evaluated FOS is less than 0.5, then a FOS of 0.5 is exported for the node.
Figure 108.5-3
Figure 108.5-4
Check that the analysis is configured as shown in Figure 108.5-4, and then click Continue.
Note that the UTS of the material is used to evaluate Kt from the defined surface finish curve (i.e. surface
finish factor versus UTS). In this case, for a mirror-polished surface, Kt =1 for all values of UTS.
As the analysis is being performed, the following information is written to the analysis log file. The
analysis log file has the same file name as the output file, except that the extension is .log. So, for this
analysis, the analysis log file is:
<ResultsDir>\keyhole_01Results_ex02.log
This information is also displayed in the Message Log window and includes
Summary
=======
Worst Life-Repeats : 1047287.625
at Element 71.3
Worst FOS@Life=Infinite : 0.725
at Element 71.3
Analysis time : 0:00:07
Step 9: Reviewing the results
The analysis log shows that the worst-case life for the whole model is:
In the configuration of the Factor of Strength analysis, the design life specified was the ‘Infinite design
life’, which is assumed to be the materials constant amplitude endurance limit, which for SAE 950C
(Manten) is: 2.0 × 107 reversals.
The worst-case fatigue lives are of the order 1.1 × 106 repeats of the fatigue loading cycle, which is less
than the design life (1 × 107 repeats of the loading). This is why the worst-case strength factor (the
factor by which the loading must be scaled to achieve the design life) is somewhat less than 1. The FOS
method is described in detail in section 17.
Note:
fe-safe adopts the conventional method of expressing the endurance limit, as the number of reversals,
or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ). In fe-safe, the
design life and the calculated fatigue life always refer to the number of repeats of the complete defined
fatigue loading cycle. An optional conversion factor can be used to convert the fatigue life in repeats to
fatigue life with respect to some other quantity, for example hours or miles (see section 13.2).
Figure 108.5-5
The SY variable in the last set in the file <ResultsDir>/keyhole_01Results_ex02.rst should look
similar to Figure 108.5-6, the viewer colour scheme may need to be set to "Reversed Rainbow" and any
averaging of values at nodes should be disabled.
Figure 108.5-6
109.1 Introduction
This tutorial demonstrates how to perform an analysis of a welded joint and should be read in
conjunction with section 16,‘Fatigue Analysis of Welded Steel Joints’.
The sample files used for this tutorial are located in the directory <DataDir>\ weld_tutorial\.
109.2 Preparation
The tutorial uses an ANSYS .rst model, however, the same techniques can be applied to all FE formats.
Sample models and stress result files are provided in the following formats:
The results shown below are for illustration only. The actual stresses (and therefore the fatigue results)
are different for each model format. In addition to the analysis of the welded joints, a further analysis of
the entire model would normally be performed, since the shortest lives are not necessarily at the welds.
This tutorial assumes that the user is familiar with the basic program operation. Before attempting this
tutorial, it is strongly recommended that one of the introductory tutorials is followed:
Start the program by selecting fe-safe from the Windows “Start” menu (Windows) or by running the
script fe-safe (Linux) – (see section 5.2).
Select an existing project, or create a new one from the welcome page
Figure 109-1
Ensure that the following basic settings are correctly configured. If fe-safe is being run for the first time
then it is possible to skip this preparatory section.
Reset any existing analysis options to their defaults in the Clear Data and Settings dialogue [Tools >>
Clear Data and Settings...], shown in Figure 109-2, below, by selecting the Check all option and clicking
OK.
Figure 109-2
Figure 109-3
The weld is simulated on the FE model as a solid chamfer. For this analysis, the nominal stresses at the
toe of the weld are of interest. However, high stress concentrations will occur at the toe of the weld, as
shown in Figure 109-4 below (showing Principal Stress):
Figure 109-4
To avoid the high stress concentration at the toe of each weld, and give the best approximation of the
actual stresses at the weld, the stresses within the first 2 to 3 mm of the chamfer/surface junction
should be ignored. The stresses in the band of elements adjacent to this gap provide the best
approximation of the nominal stresses at the weld itself. Depending on the dimensions of the model, it
may be necessary to reduce the size of the elements in the analysis region.
A set of stress results were obtained by constraining the base of the component through one hole, and
applying a load to the other hole, in a direction away from the constrained hole and in the plane of the
plate. The stress results file contains stresses for the whole model, as well as stresses for a defined
group of elements in two narrow bands (a few mm wide), close to the weld.
The sample model and stress results are contained in the following files:
Note: in ANSYS, groups are supported by using a different material number for different groups of
elements. fe-safe recognises these groups, and lists them in the Current FE Models window. In this
example, the different material numbers have identical properties.
Figure 109-5
Ensure the model units are MPa, uE and deg.C. To change the units, right click within Current FE Models
window and select properties. Then change the values to those shown in Figure 109-6:
Figure 109-6
In the Fatigue from FEA window, Loading Settings tab, configure a user defined load and dataset using
data set 1 and a load sequence of [1, 0]:
Select the Loading Settings from the Fatigue from FEA dialogue to switch to the loading tree;
select Clear all loadings and click Yes from the tree context menu (right mouse click on the tree) to
clear the existing loading definition;
in the Current FE Models window, highlight the stress dataset (next to the stress dataset icon, );
enter 1 and 0 in the Dataset Embedded Load History dialogue (i.e. one repeat of the stress load
case);
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
It is not necessary to define the surface finish, material or in-plane residual stresses for this type of
analysis.
In this tutorial only the welded joints will be analysed, a further analysis of the entire model would
normally be performed, since the shortest lives may not necessarily be at the welds.
To define the analysis algorithm for the welded joints:
double-click the Algorithm column header to open the in Group Algorithm Selection dialogue box
for all groups;
choose the Do not analyse option;
double-click the Algorithm cell for Material3;
choose the Select an algorithm to be used option;
click the User algorithm browse button, , and select BS5400 Weld Finite Life (CP) from the drop-
down menu, as in Figure 109-7;
Figure 109-7
in the BS5400 Weld Definition section of the Group Algorithm Selection dialogue select Weld Class F
and enter a design criteria of –2, as in Figure 109-8;
Figure 109-8
click OK.
The Fatigue from FE dialogue should now look similar to Figure 109-9:
Figure 109-9
Summary
=======
Worst Life-Repeats : 646835.375
at Element 4056.4
Analysis time : 0:00:01
This fatigue results from this exercise were written to the file:
<ResultsDir>/weld_ansResults_01.rst.
The exported results contain log10 of lives, which can be viewed in ANSYS and should look similar to
Figure 109-10, the viewer colour scheme may need to be set to "Reversed Rainbow" and any averaging
of values at nodes should be disabled.
Regions immediately next to the toe of the welds were NOT included in the analysed region to avoid the
areas of high stress concentration at the toe, and to give the best estimate of the fatigue life at the weld.
110 Tutorial 110 : Stress-based thermal fatigue analysis using multiple-temperature S-N
data
110.1 Introduction
This tutorial demonstrates a stress-based thermal (temperature-dependent) fatigue analysis of an
exhaust manifold assemblage, using multiple-temperature S-N data.
This tutorial should be read in conjunction with section 18,‘Conventional High Temperature Fatigue’.
The sample files used for this tutorial are located in the directory <DataDir>\Abaqus.
110.2 Preparation
The tutorial uses an Abaqus .odb model. However, the same techniques can be applied to all FE formats
for which temperatures are supported, which currently include:
This tutorial assumes that the user is familiar with the basic program operation. Before attempting this
tutorial, it is strongly recommended that one of the introductory tutorials is followed:
Start the program by selecting fe-safe from the Windows “Start” menu (Windows) or by running the
script fe-safe (Linux).
Select an existing project, or create a new one from the welcome page
Ensure that the following basic settings are correctly configured. If fe-safe is being run for the first time
then it is possible to skip this preparatory section.
Reset any existing analysis options to their defaults in the Clear Data and Settings dialogue [Tools >>
Clear Data and Settings...], shown in Figure 110.2-1, below, by selecting the Check all option and
clicking OK.
Figure 110.2-1
Figure 110.3-1
To open the model, select Open Finite Element Model... from the File menu. From the file selection
dialog, select the sample file manifold_xx.odb from the directory <DataDir>\Abaqus.
A prompt to pre-scan the file will be displayed if the file has not yet been pre-scanned. The stress and
temperature datasets for the initial step increment do not need to be loaded. Check the Stresses,
Temperatures and Last increment only boxes, click Apply to Dataset List to apply the selections and
select OK to load the remaining datasets, as shown in Figure 110.3-2:
Figure 110.3-2
As fe-safe loads the model, information about the file and the data it contains is written to the file:
<ProjectDir>\Model\reader.log.
This information is also displayed in the Message Log window.
When the model has finished loading, the Loaded FEA Models Properties dialogue box appears, as
shown in Figure 110.3-3.
Figure 110.3-3
If the dialogue box does not appear automatically, then it can be displayed by right-clicking on the
icon in the Current FE Models window and selecting Properties.
Ensure that the stress, strain and temperature units are MPa, µE and deg.C, respectively, as shown in
Figure 110.3-3, then click OK.
A summary of the open model appears in the Current FE Models window, showing the loaded datasets
and element group information – see Figure 110.3-4.
Figure 110.3-4
The dataset and group details in the tree view can be expanded to show more details.
The model contains three stress datasets, three temperature datasets. fe-safe also extracts element
group information from the ODB file.
When a new model is loaded, the output filename automatically defaults to:
<ResultsDir>\{source_file_name}Results.{source_file_extension}
Objective:
Each load case consists of an elastically calculated FEA stress solution (i.e. a stress dataset). Each stress
dataset has a corresponding temperature dataset containing the temperature of each node.
The analysis will also include an evaluation of the Factor of Strength (FOS). This is the factor by which
the loading must be scaled to achieve a specified design life. The FOS method is described in section 17.
Analysis process:
the maximum temperature from all temperature results sets in the FEA model is determined;
the material properties are modified so that they apply at the maximum temperature of the node;
the modified material properties are used to calculate the fatigue life for the node.
It is implicit in this analysis that each node uses materials data specifically modified for the maximum
temperature of that node.
The temperature-corrected material properties will only be used if the following requirements are
satisfied:
the option to disable temperature analysis [FEA Fatigue >> Analysis Options..., General tab, Disable
temperature-based analysis option] is not set.
Method:
The loading consists of a single fatigue loading block, cycling between the second dataset (the
mechanical load plus the thermal load) and the third dataset (the mechanical load at ambient
temperature).
Note: The definition of fatigue loading for varying temperature, as discussed in section 13, applies
only to thermo-mechanical fatigue analyses, as described in section 20. This definition is not
required for conventional high temperature fatigue, as in this tutorial.
select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading tree;
select Clear all loadings and click Yes from the tree context menu (right mouse click on the tree) to
clear the existing loading definition;
in the Current FE Models window, highlight the fourth dataset, labelled Dataset 2: (2.6) S : Step 2.)
APPLY PRESCRIBED THERMAL LOAD, (next to the stress dataset icon, );
in the Current FE Models window, highlight the sixth dataset, labelled Dataset 3: (3.4) S : Step 3.)
RETURN TO AMBIENT TEMPERATURES, (next to the stress dataset icon, );
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
Figure 110.4-1
Since this is a thermal analysis, we need temperature-dependent material data. For this exercise a copy
of the material SAE_950C-Manten from the material database local.dbase will be created, and then
modified to give it some sample temperature-dependent data. This tutorial uses a stress-based
algorithm using S-N data, so it is the S-N curve parameters that will be modified.
in the Material Databases window, select (highlight) the material SAE_950C_Manten from the list of
available materials in the local.dbase material database;
select Material >> Copy Material to create a new material called CopyOfSAE_95OC-Manten, which
will appear at the bottom of the local.dbase database;
double-click the newly created material to edit it – change the name to Manifold.
select the constant amplitude endurance limit field: gen : Const Amp Endurance Limit (2nf);
double-click on the value field to edit it – a temperature list dialogue will open;
modify the list to contain the two temperatures 300°C and 800°C, as shown in Figure 110.4-2:
Figure 110.4-2
click OK.
select one of the S-N curve fields (either sn curve : S Values (MPa) or sn curve : N Values (nf));
double-click on the value field to edit it - a SN Values Table (MPa) dialogue will open;
modify the table so that it contains S-N curve data for two temperatures, as shown in Figure
110.4-3:
Figure 110.4-3
double-click the Material header in the Groups Parameter box in the Fatigue from FEA dialogue - a
Change Material? confirmation dialogue box appears;
check the confirmation dialogue – it should say: Are you sure you want to change the material for
the groups ‘all’ to ‘Manifold’ in database ‘local.dbase’? ;
click YES;
the material name should appear for all groups in the Material column.
open the Material Plot dialogue by selecting Material >> Generate Material Plot Data...;
click Apply to create the S-N curve – the plot is added to the Loaded Data Files window;
enter another temperature, click Apply, and repeat as needed. Click Cancel to exit the dialogue
highlight the plot file in the Loaded Data Files window and plot it by selecting View >> Plot;
multiple plots can be plotted on the same axis using View >> Overlay Plot(s) – see Figure 110.4-4.
Figure 110.4-4
Definition of multiple-temperature S-N parameters as well as the process used for interpolating and
extrapolating material data is described in detail in section 8.
click the User algorithm browse button, , and select Normal Stress (CP) >> Goodman from the
drop-down menu:
Figure 110.4-5
click OK.
Figure 110.4-6
click on the Factor of Strength... button in the Fatigue from FEA window to open the Factor of
Strength Calculations dialogue box:
Figure 110.4-7
select the Perform Factor of Strength (FOS) Calculations? option to enable the FOS analysis.;
click OK;
specify the FOS band limits in the Analysis Options dialogue [FEA Fatigue >> Analysis Options...]
Safety Factors tab as shown in Figure 110.4-8 to determine the limits and precision of the FOS
analysis:
- between the ‘fine’ limits, the FOS is calculated to a resolution of approximately 0.01 repeats;
- between maximum and minimum, the FOS is calculated to a resolution of approximately 0.1
repeats;
- if the evaluated FOS is greater than 5, then a FOS of 5 is exported for the node;
- if the evaluated FOS is less than 0.5, then a FOS of 0.5 is exported for the node.
Figure 110.4-8
When the FE model was loaded, the output filename automatically defaulted to:
<ResultsDir>\manifold_xxxResults.odb
The output filename can be modified either by manually editing the Output File field in the Fatigue from
FEA dialogue, or by clicking the adjacent browse button: .
Figure 110.4-9
Check that the analysis is configured as shown in Figure 110.4-9, and then click Continue.
Note that the UTS of the material is used to evaluate Kt from the defined surface finish curve (i.e. surface
finish factor versus UTS). In this case, for a mirror-polished surface, Kt =1 for all values of UTS.
As the analysis is being performed, the following information is written to the analysis log file. The
analysis log file has the same file name as the output file, except that the extension is .log:
<ResultsDir>\manifold_xxxResults.log
This information is also displayed in the Message Log window and includes
Summary
=======
Worst Life-Repeats : 81420.734
at Element 7273.3
Worst FOS@Life=1E7-Repeats : 0.669
at Element 7273.3
Analysis time : 0:00:05
Note: Due to changes in the Abaqus solution, when using a manifold model from 2018 or earlier, you
will see the following fe-safe result:
79252 repeats of the loading, at element 7273, node 3.
Also due to corrections to previous approximations in critical plane axis systems and tensor projection
during fe-safe 2020, when using fe-safe version 2019 or earlier you will see a life result of
79397 repeats of the loading at , at element 7273, node 3 (and FOS 0.65)
In the configuration of the Factor of Strength analysis, the design life specified was 107 repeats.
The worst-case fatigue lives are of the order 7.9 × 105 repeats of the fatigue loading cycle, which is less
than the design life (1 × 107 repeats of the loading). This is why the worst-case strength factor (the
factor by which the loading must be scaled to achieve the design life) is somewhat less than 1. The FOS
method is described in detail in section 17.
Note:
fe-safe adopts the conventional method of expressing the endurance limit, as the number of reversals,
or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ). In fe-safe, the
design life and the calculated fatigue life always refer to the number of repeats of the complete defined
fatigue loading cycle. An optional conversion factor can be used to convert the fatigue life in repeats to
fatigue life with respect to some other quantity, for example hours or miles (see section 13.2).
The first set of exported fatigue results in the file contains the fatigue lives, which should look similar to
Figure 110.4-10, the viewer colour scheme may need to be set to "Reversed Rainbow" and any averaging
of values at nodes should be disabled.
The second set of exported fatigue results in the file contains the strength factors (FOS), which should
look similar to Figure 110.4-11, the viewer colour scheme may need to be set to "Reversed Rainbow"
and any averaging of values at nodes should be disabled.
Figure 110.4-11
111.1 Introduction
This tutorial demonstrates modal transient dynamic fatigue analysis of a simple bracket.
The tutorial includes opening the FE model and configuring the analysis. Two different analysis methods
are explored:
Exercise 2: Complete a Factors of Strength (FOS) evaluation of the modal fatigue analysis.
Exercise 3: A sensitivity analysis to determine the most damaging mode at worst life location.
The ODB file format is portable between platforms. The use of Abaqus .odb files in fe-safe is discussed
in detail in Appendix G.
The sample files used for this tutorial are located in the directory <DataDir>\Abaqus.
NOTE: The nomenclature used throughout this tutorial when referring to files and directories is as
described in Appendix B.
111.2 Preparation
This tutorial assumes that the user is familiar with the basic program operation. Before attempting this
tutorial, it is strongly recommended that one of the introductory tutorials is followed:
Start the program by selecting fe-safe from the Windows “Start” menu (Windows) or by running the
script fe-safe (Linux).
Select an existing project, or create a new one from the welcome page
Ensure that the following basic settings are correctly configured. If fe-safe is being run for the first time
then it is possible to skip this preparatory section.
Reset any existing analysis options to their defaults in the Clear Data and Settings dialogue [Tools >>
Clear Data and Settings...], shown in Figure 111.2-1 below, by selecting the Select all option and
clicking OK
Figure 111.2-1
Mass M
a
Base acceleration s
(translation x, and s
rotations about y, and z)
Figure 111.3-1
To open the model, select Open Finite Element Model... from the FEA Solutions section of the File menu.
This will display the file selection dialogue. Select the sample file modal_bracketxxx.odb from the
directory <DataDir>\Abaqus. The Pre-Scan File dialogue will be displayed as shown in Figure 111.3-2,
select Yes.
Figure 111.3-2
As fe-safe pre-scans the model information about the file is written to the folder:
<ProjectDir>\Model\scan
This information is also displayed in the Message Log window.
When the pre-scan of the file is complete, the Select Datasets to Read dialogue will be displayed as shown in
Figure 111.3-3. The file modal_bracket_xxx.odb contains 10 datasets of elemental stresses from normal mode
analysis and generalized displacements. Check the Stresses and Others boxes, click Apply to Dataset List to
apply the selection and select OK to load the stress datasets and histories. A message will be shown to notify
that the Histories that have been loaded from the model will be added to the Loaded Data Files window.
Figure 111.3-3
As fe-safe loads the model, information about the file and the data it contains is appended to the file:
<ProjectDir>\Model\reader.log.
This information is also displayed in the Message Log window.
When the model has finished loading, the Loaded FEA Models Properties dialogue box appears, as
shown in Figure 111.3-4.
Figure 111.3-4
If the dialogue box does not appear automatically, then it can be displayed by right-clicking on the
icon in the Current FE Models window and selecting Properties as shown in Figure 111.3-5
Figure 111.3-5
Ensure that the units are as shown in Figure 111.3-4, then click OK.
A dialogue will show prompting to edit element groups loaded from the model, as shown in Figure
111.3-6, click No.
Figure 111.3-6
A message will be shown informing that Dataset 1 contains all zeros, this can be safely ignored
A summary of the open model appears in the Current FE Models window, showing the loaded datasets
and element group information – see below Figure 111.3-7.
Figure 111.3-7
Note: if the window does not appear exactly as shown in Figure 111.3-7, then expand the tree view to
show more details.
The process of loading an FE model into fe-safe involves extracting pertinent data from the FE model,
and writing it to the working FED folder (see Appendix E, section 205.4.2). Therefore, the Current FE
Models window is a summary of the contents of the working FED folder. In this case there are 11 normal
modes stress datasets each containing elemental data with 10384 elements in each dataset. The source
of the dataset (the filename of the source FE model, the step, increment and timestamp) are also shown.
fe-safe also extracts element group information from the ODB file.
The Abaqus odb file also included the modal participation factors. The data is summarised in the Loaded
Data Files window. The sample file contains 11 participation factors, identified in the tree view by the
channel icon, as shown in Figure 111.3-8.
Figure 111.3-8
When a new model is loaded, the output filename automatically defaults to:
<ResultsDir>\{source_file_name}Results.{source_file_extension}
The output filename can be modified either by manually editing the Output File field in the Fatigue from
FEA dialogue, or by clicking the adjacent browse button: .
To perform a dynamic analysis combining the 10 modal stress conditions and the modal participation
factors corresponding to a base excitation. The excitation at the base consists of a described vertical
displacement in the z-direction, and two rotations about the x and y-axes for a duration of 1 second. A
modal transient finite element analysis is performed using Abaqus in two steps. First, the normal
modes analysis is done, and the modal stresses are written out to the output database (odb). In the
following step, the forced response is performed and ONLY the generalized displacements (a.k.a. modal
participation factors) are written to the output database. The odb file is read into fe-safe for fatigue
analysis based on these two steps. Each fatigue load case consists of an elastic modal stress (a.k.a.
stress dataset) and the corresponding generalized displacement (a.k.a. modal participation factor). The
stress conditions and modal participation factors are combined in a typical scale-and-combine method.
Analysis process:
the stress tensors due to each mode are multiplied by the generalised displacements due to the
base excitation, to produce a time history of each of the 6 components of the stress tensor per
mode;
contributions from all modes are superimposed to produce a time history of each of the 6
components of the stress tensor for all modes;
the time histories of the three principal strains are calculated from the stresses;
a multiaxial cyclic plasticity model is used to convert the elastic stress-strain histories into elastic
plastic stress-strain histories;
a ‘critical plane’ method is used to identify the most damaging plane by calculating the damage on
planes at 10° intervals between 0° and 180° in the surface of the component;
for each of the critical planes, strains are resolved onto the three shear planes (1-2, 2-3 and 1-3);
the time history of the damage parameter (which in this case, using the Brown-Miller algorithm, is
the shear and normal strain) is cycle counted;
individual fatigue cycles are identified using a ‘Rainflow’ cycle algorithm, the fatigue damage for
each cycle is calculated and the total damage is summed;
the plane with the shortest life defines the plane of crack initiation, and this life is written to the
output file.
During this calculation, fe-safe may modify the endurance limit amplitude. If all cycles (on a plane) are
below the endurance limit amplitude, there is no calculated fatigue damage on this plane. If any cycle is
damaging, the endurance limit amplitude is reduced to 25% of the constant amplitude value, and the
damage curve extended to this new endurance limit.
The critical plane method is described and illustrated in detail in the Fatigue Theory Reference Manual,
section 7.5.
Method:
For this exercise 10 modes and participation factors will be superimposed. Each load case will consist of
an elastically calculated modal stress dataset and a corresponding generalised displacement, as follows:
the second dataset (Dataset 2 (1.0) mS: Normal Modes Extraction, mode=1) will be combined with
generalized displacement 2 from the Abaqus output file modal_bracketxxx.odb in the Loaded Data
Files window: Modal_bracketxxx.odb[step_2].amc,1;
the third dataset (Dataset 3 (1.0) mS: Normal Modes Extraction, mode=2) will be combined with
generalized displacement 3 from the Abaqus output file modal_bracketxxx.odb in the Loaded Data
Files window: Modal_bracketxxx.odb[step_2].amc,2;
Note: The first Stress Dataset shown in Figure 111.3-7 represents a zero solution. The direct and shear
stress ranges, and mode number are zero. Accordingly, the first modal participation factor shown in the
Loaded Data Files window does not correspond to a mode. Modes 1-10 correspond to stress data sets
and generalized displacements 2-11 in each window.
Figure 111.4-1
in the Settings block right click on Loading is equivalent to 1 Repeats and select Edit from the drop
down menu
in the Conversion factors dialogue click the LDF Time button and then OK.
Figure 111.4-2
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
Figure 111.4-3
select Whole group;
click OK.
Figure 111.4-4
The component material is SAE 1008. The material record SAE-1008 from the material database
system.dbase will be defined for the whole component.
Opened material databases are displayed in the Material Databases window as shown in Figure 111.4-5,
below. If this window is not currently displayed, select Material >> Display Materials Database Window
to display it.
Figure 111.4-5
If the database system.dbase is not currently loaded in the Material Databases window, select File >>
Materials >> Open Materials Database... and select system.dbase from the directory <DatabaseDir>.
Notice in Figure 111.4-5 that the database system.dbase is the System Database from the
<DatabaseDir> directory.
Expand the tree view to show the materials contained in system.dbase. The contents of the database
can be filtered to show only materials of a particular type, by selecting one of the filter icons . To
ensure that all materials in the database are displayed double-click on the All filter icon .
The parameters for a material can be displayed by expanding the tree view of the material name.
To define SAE-1008 for the whole component:
highlight the material SAE-1008 in the system.dbase database;
select all groups by clicking on the Group column header in the Group Parameters box in the
Fatigue from FEA dialogue;
double-click the Material column header - a Change Material? confirmation dialogue box appears;
check the confirmation dialogue – it should say: Are you sure you want to change the material for
the groups ‘Default, WanrElemDistorted, EBRACKET, EMASS’ to ‘SAE-1008’ in database
‘system.dbase’? ;
click YES;
the material name should appear for all groups in the Material column.
Figure 111.4-6
For SAE 1008 the default algorithm is Brown-Miller Biaxial Strain Life using Morrow mean-stress
correction.
Step 6: Define the in-plane residual stress:
For this exercise it is assumed that there are no in-plane residual stresses. To set the residual stress to
0 (zero) for the whole component (i.e. all element groups):
in the Group Parameters table, use the scroll bar to the right, in order to double-click the In-plane
residual stress column header to open the In-plane Residual Stress dialogue for all groups:
Figure 111.4-7
When the FE model was loaded, the output filename automatically defaulted to:
<ResultsDir>\modal_bracketxxxResults.odb
Before running the analysis, change the output filename to:
<ResultsDir>\modal_bracketxxxResults_ex01.odb
The output filename can be modified either by manually editing the Output File field in the Fatigue from
FEA dialogue, or by clicking the adjacent browse button: .
After steps 1 through 6 are complete the Fatigue from FEA dialogue appears as shown in Figure 111.4-8
Figure 111.4-8
Step 8: Run the analysis:
fe-safe should now be configured to run the analysis.
Press the Analyse! button. A summary of analysis parameters is displayed:
Figure 111.4-9
Check that the analysis is configured as shown in Figure 111.4-9, and then click Continue.
As the analysis is being performed, information is written to the analysis log file. The analysis log file
has the same file name as the output file, except that the extension is .log. So, for this analysis, the
analysis log file is:
<ResultsDir>\modal_bracketxxxResults_ex01.log
This information is also displayed in the Message Log window and includes:
Summary
=======
Worst Life-Hours : 15.288
at Element [0]5939.1
Analysis time : 0:00:25
To save the current analysis configuration, select Save FEA Fatigue Definition File..., and enter a filename
ending with the extension .stlx. The analysis can be reloaded at a later date by restoring the keyword
file, using Open FEA Fatigue Definition File....
Note: The contour plot above is a reverse-contour. It shows the lowest Log Life as red and the highest
Log Life as blue. Depending on which post-processor you use, the method to reverse the contour
colours will vary and any averaging of values at nodes should be disabled.
1. In the case that the modal superposition loading does not automatically populate as indicated in
Step 1, the following procedure can be done to effect the same result:
in the Current FE Models window, highlight the second stress dataset, labelled Dataset 2
(1.1) mS: Normal Modes Extraction, Mode=1, (next to the stress dataset icon, );
in the Loaded Data Files window, highlight all but the first channel of data, labelled
Modal_bracket_xxx.odb[step_2].amc:GU1 through GU10 (next to the channel icon );
Figure 111.4-11
select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading
tree;
select Clear all loadings from the tree context menu (right mouse click on the tree) to clear
the existing loading definition; and click Yes in the delete loadings dialogue.
Figure 111.4-12
Figure 111.4-13
in the Loaded Data Files window, highlight the first channel of data, labelled
Modal_bracketxxx.odb[step2].amc:Time
In the Loading Details dialogue right click on the Elastic Block and choose Add Time History.
Figure 111.4-14
The time base record is loaded into the elastic block with a time history symbol as shown in
Figure 111.4-14.
in the Settings block right click on Loading is equivalent to 1 Repeats and select Edit from the drop
down menu
in the Conversion factors dialogue click the LDF Time button and then OK.
Select Save Current FEA Loadings as... from the Loadings section of the File menu
Use the Save Loading as .LDF file dialogue to browse to a file location where you want to save the
loadings, and name the ldf file. Click Save.
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
In this exercise, the many modal load cases are superimposed. Section 13 includes a full description of
the LDF file, including syntax. The file is as follows:
INIT
transitions=Yes
END
# Block number 1
BLOCK
lhtime=modal_bracketxxx.odb[step_2].amc, signum=1
lh=modal_bracketxxx.odb[step_2].amc, signum=2, ds=2
lh=modal_bracketxxx.odb[step_2].amc, signum=3, ds=3
lh=modal_bracketxxx.odb[step_2].amc, signum=4, ds=4
lh=modal_bracketxxx.odb[step_2].amc, signum=5, ds=5
lh=modal_bracketxxx.odb[step_2].amc, signum=6, ds=6
lh=modal_bracketxxx.odb[step_2].amc, signum=7, ds=7
lh=modal_bracketxxx.odb[step_2].amc, signum=8, ds=8
2. In this exercise the surface finish was indicated by a user-defined value for Kt. For an example
of selecting a surface type (in this case “mirror-polished”) from a surface finish database, see
Tutorial 106.
3. The results data can be exported to an alternative file format at the end of the analysis, by
changing the extension of the output file (Fatigue from FEA >> Output File) to a recognised
type.
For example:
to export to an ASCII CSV results file, change the extension to .txt, .csv or .asc (this is
useful for viewing a table of results in an editor or spreadsheet).
Alternatively, after the analysis is complete, and the ODB file has been created, the results data
can be written to another output file in addition to the original output ODB file, using the Save
FEA Fatigue Results As... option in the File menu. This option displays the Save FEA Fatigue
Results As... dialogue box, as shown in Figure 111.4-15 below:
Figure 111.4-15
The FER (.fer) file (see Appendix E) is an intermediate file containing the results of the analysis.
This new output file type is determined by the extension of the file entered in the Output File
box. A copy of the original output ODB file can be created if the extension .odb is used.
111.5 Exercise 2: Factors of Strength (FOS) evaluation of the modal fatigue problem
Objective:
Add an evaluation of the Factor of Strength (FOS) to the previously completed Example 1. To be
determined is the factor by which the loading must be scaled to achieve a specified Factor of Strength.
The FOS method is described in detail in section 17.
Analysis process:
- if the calculated life is lower than the Factor of Strength, the elastic stresses at the node are
scaled by a factor less than 1.0;
- if the calculated life is greater than the Factor of Strength, the elastic stresses at the node are
scaled by a factor greater than 1.0.
the elastic stress history is recalculated using the re-scaled nodal stresses;
for local strain analysis, the cyclic plasticity model is used to recalculate the time history of elastic-
plastic stress-strains - the fatigue life is then recalculated;
(ii) the step change of 0.01 or 0.1 in the FOS value causes the Factor of Strength to be
bracketed, or
(iii) the FOS exceeds the max factor (default 2.0) or is less than the min factor (default 0.5).
Method:
Step 1: This analysis makes use of the previous loading and group definitions set up in section 111.4
Choose File >> Open FEA Fatigue Definition File… Use the Open FEA Fatigue Definition dialogue to
choose the .stlx file created in Section 111.2: Exercise 1, Step 8. Select Open
click on the Factor of Strength... button in the Fatigue from FEA window to open the Factor of
Strength Calculations dialogue box:
Figure 111.5-1
select the Perform Factor of Strength (FOS) Calculations? option to enable the FOS analysis.;
select User-defined Factor of Strength and enter 20 in the box next to Hours;
click OK;
specify the FOS band limits in the Analysis Options dialogue [FEA Fatigue >> Analysis Options...]
choose the Safety Factors tab as shown in Figure 111.5-3 to determine the limits and precision of
the FOS analysis:
- between the ‘fine’ limits, the FOS is calculated to a resolution of approximately 0.01 repeats;
- between maximum and minimum, the FOS is calculated to a resolution of approximately 0.1
repeats;
- if the evaluated FOS is greater than 2, then a FOS of 2 is exported for the node;
- if the evaluated FOS is less than 0.5, then a FOS of 0.5 is exported for the node.
Figure 111.5-2
Figure 111.5-3
Check that the analysis is configured as shown in Figure 111.6-3, and then click Continue.
As the analysis is being performed, the following information is written to the analysis log file. The
analysis log file has the same file name as the output file, except that the extension is .log. So, for this
analysis, the analysis log file is:
<ResultsDir>\modal_bracketxxxResults_ex02.log
This information is also displayed in the Message Log window and includes:
Summary
=======
Worst Life-Hours : 15.288
at Element 5939.1
Worst FOS@Life=20-Hours : 0.925
at Element 5939.1
Analysis time : 0:01:53
A copy of the original .odb file was created, onto which a new step containing the fatigue results was
appended. The type of variable that the results are exported to will depend on the configuration in the
Abaqus ODB Interface Options dialogue (see 111.2, above).
Open the .odb file with your preferred post-processor and plot a contour of the results.
In the last step of the file <ResultsDir>/modal_bracketxxxResults_ex02.odb, the results for the
exported variable should look similar to Figure 111.5-4:
Note: The contour plot above is a reverse-contour. It shows the lowest Log Life as red and the highest
Log Life as blue. Depending on which post-processor you use, the method to reverse the contour
colours will vary and any averaging of values at nodes should be disabled.
Analysis process:
Tabulate the Load Sensitivity factors as life and as percentage change in life
Method:
Step 1: This analysis makes use of the previous loading and group definitions set up in section 111.4
Choose File >> Open FEA Fatigue Definition File… Use the Open FEA Fatigue Definition dialogue to
choose the .stlx file created in Section 111.2: Exercise 1, Step 8. Select Open
This analysis makes use of the location where the lowest life was recorded in Exercise 1.
click on the Factor of Strength... button in the Fatigue from FEA window to open the Factors of
Strength Calculations dialogue box.
deselect the box next to Perform Factor of Strength Calculations?. Select OK.
The Fatigue from FEA window displays FOS disabled next to the Factor of Strength… button
Click on the Exports… button in the Fatigue from FEA window to open the Exports and Outputs
dialogue box.
Select the List of Items tab and enter the Element and Node number: 5939.1 in the item box.
Select the check box next to Only analyse listed items. as shown in Figure 111.6-1
Figure 111.6-1
Select the check box next to Enable Load Sensitivity Analysis. Click OK.
Figure 111.6-2
Figure 111.6-3
Check that the analysis is configured as shown in Figure 111.6-3, and then click Continue.
As the analysis is being performed, the sensitivity analysis information is written to the analysis log file.
The analysis log file has the same file name as the output file, except that the extension is .log. So, for
this analysis, the analysis log file is:
<ResultsDir>\modal_bracketxxxResults_ex03.log
Fatigue analysis for the selected node is displayed in the Message Log window and includes:
Summary
=======
Worst Life-Hours : 15.288
at Element 5939.1
Analysis time : 0:00:01
SENSITIVITY ANALYSIS for Element [0]5939.1 (The life is for 1 repeat of the block
(i.e n=1), it does not consider the n Value if this is an LDF analysis)
The percentage change in Life and the percentage life change corresponding to the removal of each data
set one-by one are tabulated in the log file. Which modes contribute most of the damage?
112.1 Introduction
This tutorial demonstrates the use of fe-safe with NASTRAN .op2 files.
The tutorial includes opening the FE model and configuring multi-axial analyses of two superimposed
fatigue load cases and of a dataset sequence, including Factors of Strength (FOS) evaluation.
The sample files used for this tutorial are located in the directory <DataDir>\Nastran.
NOTE: The nomenclature used throughout this tutorial when referring to files and directories is as
described in Appendix B.
112.2 Preparation
Start the program by selecting fe-safe from the Windows “Start” menu (Windows) or by running the
script fe-safe (Linux).
Select an existing project, or create a new one from the welcome page
Figure 112-1
Ensure that the following basic settings are correctly configured. If fe-safe is being run for the first time
then it is possible to skip this preparatory section.
Reset any existing analysis options to their defaults in the Clear Data and Settings dialogue [Tools >>
Clear Data and Settings...], shown in Figure 112-2 below, by selecting the Select all option and clicking
OK.
Figure 112-2
Figure 112-3
To open the model, select Open Finite Element Model... from the File menu. From the file selection
dialog, select the sample file keyhole_01.op2 from the directory <DataDir>\Nastran. The Pre-Scan
File dialogue will be displayed as shown in Figure 112-4, select Yes.
Figure 112-4
As fe-safe pre-scans the model information about the file is written to the folder:
<ProjectDir>\Model\scan
This information is also displayed in the Message Log window.
When the pre-scan of the file is complete, the Select Datasets to Read dialogue will be displayed as
shown in Figure 112-5. The file keyhole_01.op2 contains four datasets, two of which are stress and
two are strain datasets. This tutorial only requires the stress datasets, the strain datasets do not need to
be loaded. Check the Stresses and Last increment only boxes, click Apply to Dataset List to apply the
selection. Change the Available Positions combo box to Elemental & Centroidal and select OK to load the
stress datasets.
As fe-safe loads the model, information about the file and the data it contains is appended to the file:
<ProjectDir>\Model\reader.log.
This information is also displayed in the Message Log window.
Figure 112-5
When the model has finished loading, the Loaded FEA Models Properties dialogue box appears, as
shown in Figure 112-6.
Figure 112-6
If the dialogue box does not appear automatically, then it can be displayed by right-clicking on the
icon in the Current FE Models window and selecting Properties.
Figure 112-7
Ensure that the units are as shown in Figure 112-6, then click OK.
A dialogue will show prompting to edit element groups loaded from the model, as shown in Figure
112-8 click No.
Figure 112-8
A summary of the open model appears in the Current FE Models window, showing the loaded datasets
and element group information – see Figure 112-9.
Figure 112-9
Note: if the window does not appear exactly as shown in Figure 112-9, then expand the tree view to
show more details.
The process of loading an FE model into fe-safe involves extracting pertinent data from the FE model,
and writing it to the working FED folder (see Appendix E). Therefore, the Current FE Models window is a
summary of the contents of the working FED folder. In this case there are two stress datasets each
containing elemental and centroidal data with 1004 elements (502 on the top shell and 502 on the
bottom shell) in each dataset. The source of the datasets (the filename of the source FE model, the step,
increment and timestamp) are also shown.
fe-safe also extracts group information from the ‘SETS’ defined in the first case control record in the
case control data block in the .op2 file.
When a new model is loaded, the output filename automatically defaults to:
<ResultsDir>\{source_file_name}Results.{source_file_extension}
To perform a multiaxial analysis of two superimposed fatigue load cases. Each load case consists of an
elastically calculated unit load FEA stress solution (i.e. a stress dataset) and some load history data. The
unit load and the load history are combined using a scale-and-combine method, as described in section
13.
Analysis process:
the stress tensors are multiplied by the time history of the applied loading, to produce a time
history of each of the 6 components of the stress tensor;
the time histories of the three principal strains are calculated from the stresses;
a multi-axial cyclic plasticity model is used to convert the elastic stress-strain histories into elastic
plastic stress-strain histories;
a ‘critical plane’ method is used to identify the most damaging plane by calculating the damage on
planes at 10° intervals between 0° and 180° in the surface of the component;
for each of the critical planes, strains are resolved onto the three shear planes (1-2, 2-3 and 1-3);
the time history of the damage parameter (which in this case, using the Brown-Miller algorithm, is
the shear and normal strain) is cycle counted;
individual fatigue cycles are identified using a ‘Rainflow’ cycle algorithm, the fatigue damage for
each cycle is calculated and the total damage is summed;
the plane with the shortest life defines the plane of crack initiation, and this life is written to the
output file.
During this calculation, fe-safe may modify the endurance limit amplitude. If all cycles (on a plane) are
below the endurance limit amplitude, there is no calculated fatigue damage on this plane. If any cycle is
damaging, the endurance limit amplitude is reduced to 25% of the constant amplitude value, and the
damage curve extended to this new endurance limit.
The critical plane method is described and illustrated in detail in the Fatigue Theory Reference Manual,
section 7.5.
Method:
For this exercise two fatigue load cases will be superimposed. Each load case will consist of an elastically
calculated unit load stress dataset and some load history data, as follows:
the first dataset (Dataset 1: (1.1) S : LS1 – CS1, SUBCASE 1) will be combined with channel 1 of the
data file test_mcg2.amc;
the second dataset (Dataset 2: (2.2) S : LS2 – CS2, SUBCASE 2) will be combined with channel 2 of
the data file test_mcg2.amc.
To open the loading history data, select Open Data File from the File menu. Use the file selection
dialogue to select the sample file test_mcg2.amc from the <DataDir> directory. The data file is
summarised in the Loaded Data Files window. The sample file contains two channels of load data,
identified in the tree view by the channel icon, .
Figure 112-10
To combine the unit load stress dataset with the loading history for the first case:
select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading tree;
select Clear all loadings and click Yes from the tree context menu (right mouse click on the tree) to
clear the existing loading definition;
in the Current FE Models window, highlight the first dataset, labelled Dataset 1: (1.1) S : LS1 – CS1,
SUBCASE 1, (next to the stress dataset icon, );
in the Loaded Data Files window, highlight the first channel of data, labelled fe-safe tutorial scaler
#1, (next to the channel icon );
To combine the unit load stress dataset with the loading history for the second case:
in the Current FE Models window, highlight the second dataset, labelled Dataset 2: (2.1) S : LS2 –
CS2, SUBCASE 2, (next to the stress dataset icon, );
in the Loaded Data Files window, highlight the second channel of data, labelled fe-safe tutorial
scaler #2, (next to the channel icon ).
The defined loading appears in the loading details list box, as shown below:
Figure 112-11
Note: if the window does not appear exactly as shown in Figure 112-11, then expand the tree view to
show more details.
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
Figure 112-12
select Whole group;
click OK.
Figure 112-13
The component material is SAE 950C (Manten). The material record SAE_950C-Manten from the material
database local.dbase will be defined for the whole component.
Opened material databases are displayed in the Material Databases window as shown in Figure 112-14,
below.
Figure 112-14
If the database local.dbase is not currently loaded in the Material Databases window, select File >>
Materials >> Open Materials Database... and select local.dbase from the directory <UserDir>. Notice
in Figure 112-14 that the database local.dbase is the user’s local copy from the <UserDir> directory.
The other two databases shown are from the <DatabaseDir> directory.
Expand the tree view to show the materials contained in local.dbase. The contents of the database can
be filtered to show only materials of a particular type, by selecting one of the filter icons . To ensure
that all materials in the database are displayed double-click on the All filter icon .
The parameters for a material can be displayed by extending the material name.
To define SAE_950C-Manten for the whole component:
highlight the material SAE_950C-Manten in the local.dbase database;
double-click the Material column header - a Change Material? confirmation dialogue box appears;
Figure 112-15
click YES;
the material name should appear for all groups in the Material column.
Step 5: Define the analysis algorithm:
The material database identifies a default analysis algorithm for each material. The default analysis
algorithm is normally selected automatically. To confirm that the default algorithm will be used for the
whole model (i.e. all element groups):
double-click the Algorithm column header to open the in Group Algorithm Selection dialogue box
for all groups;
Figure 112-16
For SAE 950C (Manten) the default algorithm is Brown-Miller Biaxial Strain Life using Morrow mean-
stress correction.
Figure 112-17
When the FE model was loaded, the output filename automatically defaulted to:
<ResultsDir>\keyhole_01Results.op2
Figure 112-18
Check that the analysis is configured as shown in Figure 112-18, and then click Continue.
Note that the UTS of the material is used to evaluate Kt from the defined surface finish curve (i.e. surface
finish factor versus UTS). In this case, for a mirror-polished surface, Kt =1 for all values of UTS.
As the analysis is being performed, the following information is written to the analysis log file. The
analysis log file has the same file name as the output file, except that the extension is .log. So, for this
analysis, the analysis log file is:
<ResultsDir>\keyhole_01Results_ex01.log
This information is also displayed in the Message Log window and includes:
Summary
=======
Worst Life-Repeats : 42191.711
at Element 1.4:1
Analysis time : 0:00:05
Note:
fe-safe adopts the conventional method of expressing the endurance limit, as the number of reversals,
or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ). In fe-safe, the
design life and the calculated fatigue life always refer to the number of repeats of the complete defined
fatigue loading cycle. An optional conversion factor can be used to convert the fatigue life in repeats to
fatigue life with respect to some other quantity, for example hours or miles (see section 13).
1. In this exercise, the two fatigue load cases are superimposed. Therefore, it does not matter in
which order the loads are defined. For example:
Loading Loading is 1 Repeats
DS#1 * test_mcg2.amc:1
DS#2 * test_mcg2.amc:2
is the same as:
Loading Loading is 1 Repeats
DS#2 * test_mcg2.amc:2
DS#1 * test_mcg2.amc:1
2. In this example, the load history is very short, consisting of only five data points. For short load
histories it is often more convenient to define the loading using a ‘user-defined’ load history,
rather than reading a load history from a data file (see section 13.8.2). This is achieved using
the A user-defined LOAD * dataset option.
To re-define the loading for this exercise using user-defined load histories:
Select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading
tree;
select Clear all loadings and click Yes from the tree context menu (right mouse click on the
tree) to clear the existing loading definition;
in the Loaded Data Files window expand the filename test_mcg2.amc, so that both
channels of data are displayed;
highlight both channels of data (to select multiple channels highlight the first channel by
clicking on it with the left mouse button, then hold down the CTRL key on the keyboard
while clicking on any additional files);
display a numerical listing of the two data channels by selecting View >> Numerical Listing
Figure 112-20
highlight the first unit stress dataset (labelled Dataset 1: (1.1) S : LS1 – CS1, SUBCASE 1) in
the Current FE Models window;
select Add... >> A user-defined LOAD * dataset to display the Dataset Embedded Load
History dialogue, and copy the five data points from the first column of the numerical listing
display (scaler #1) into the Loading Scale box:
Figure 112-21
click OK;
highlight the second unit stress dataset (labelled Dataset 2: (2.1) S : LS2 – CS2, SUBCASE 2)
in the Current FE Models window;
define a load sequence for the second case, as described above, using the data from the
second column of the numerical listing display (scaler #2).
The new loading definition appears in the Loading Details area of the Fatigue from FEA dialogue,
as shown in Figure 112-22, below.
Figure 112-22
Note: if the window does not appear exactly as shown in Figure 112-22, then expand the tree
view to show more details.
3. The load definition (LDF) file can be used to define simple and complex loading scenarios.
Section 13 includes a full description of the LDF file, including syntax.
The loading for this exercise can be seen by saving the loading to a load definition file:
open the file in a text editor to display the contents (comments may vary):
INIT
END
# Block number 1
BLOCK
lh=0.0099465 -0.00696255 0.00696255 0.0039786 -0.0099465 , ds=1
lh=-0.000369029 0.00221417 0.00590447 0.00738059 0.00221417 , ds=2
END
4. In this exercise the surface finish was defined by selecting a surface type (in this case “mirror-
polished”) from a surface finish database. If the actual surface finish factor, Kt, for the
component is known (for “mirror-polished”, the Kt factor has a value of 1) then it can be
specified as a value. To define a user-defined value for Kt, select Define Kt as a value, in the
Surface Finish Definition dialogue box:
Figure 112-23
5. The results data can be exported to an alternative file format at the end of the analysis, by
changing the extension of the output file (Fatigue from FEA >> Output File) to a recognised
type.
For example:
to export to an ASCII CSV results file, change the extension to .txt, .csv or .asc (this is
useful for viewing a table of results in an editor or spreadsheet).
Alternatively, after the analysis is complete, and the OP2 file has been created, the results data
can be written to another output file in addition to the original output OP2 file, using the Save
FEA Fatigue Results As... option in the File menu. This option displays the Save FEA Fatigue
Results As... dialogue box, as shown in Figure 112-24, below:
Figure 112-24
The FER (.fer) file (see Appendix E) is an intermediate file containing the results of the analysis.
This new output file type is determined by the extension of the file entered in the Output File
box. A copy of the original output OP2 file can be created if the extension .op2 is used.
6. To analyse individual elements or nodes, list the required items in the Exports and Outputs
dialogue, and select the Only analyse listed items dialogue. Additional outputs and diagnostics
options can also be specified. Diagnostics and additional outputs are discussed in detail in
section 22.
7. To save the current analysis configuration, select Save FEA Fatigue Definition File..., and enter a
filename ending with the extension .stlx. The analysis can be reloaded at a later date by
restoring the project definition file, using Open FEA Fatigue Definition File..., .
112.5 Exercise 2 : Multiaxial analysis of a data sets sequence, including Factor Of Strength
evaluation
Objective:
To perform a multiaxial analysis of a series of events – in this case a sequence of two elastically
calculated FEA stress solution (i.e. stress datasets).
The analysis will also include an evaluation of the Factor of Strength (FOS). This is the factor by which
the loading must be scaled to achieve a specified design life. The FOS method is described in detail in
section 17.
Analysis process:
- if the calculated life is lower than the design life, the elastic stresses at the node are scaled by a
factor less than 1.0;
- if the calculated life is greater than the design life, the elastic stresses at the node are scaled by
a factor greater than 1.0.
the elastic stress history is recalculated using the re-scaled nodal stresses;
for local strain analysis, the cyclic plasticity model is used to recalculate the time history of elastic-
plastic stress-strains - the fatigue life is then recalculated;
(ii) the step change of 0.01 or 0.1 in the FOS value causes the design life to be bracketed, or
(iii) the FOS exceeds the max factor (default 2.0) or is less than the min factor (default 0.5).
Method:
The series of events being analysed in this exercise consists of a sequence of two elastically calculated
FEA stress solution (i.e. stress datasets) scaled by a defined scale factor.
the first dataset (Dataset 1: (1.1) S : LS1 – CS1, SUBCASE 1) multiplied by a scale factor of 0.02,
followed by:
the second dataset (Dataset 2: (2.1) S : LS2 – CS2, SUBCASE 2) multiplied by a scale factor of 0.02.
The loading for this loading sequence can be defined in a single loading block:
select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading tree;
select Clear all loadings and click Yes from the tree context menu (right mouse click on the tree) to
clear the existing loading definition;
highlight the first unit stress dataset in the Current FE Models window;
highlight the second unit stress dataset in the Current FE Models window;
highlight the dataset item in the loading tree of the Fatigue from FEA dialogue;
select Scale from the tree context menu (right mouse click on the tree) and enter the value 0.02;
INIT
END
# Block number 1
BLOCK
ds=1, scale=0.02
ds=2, scale=0.02
END
The new loading definition appears in the Loading Details area of the Fatigue from FEA dialogue, as
shown in Figure 112-25.
Figure 112-25
select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
As in Exercise 1, the component is assumed to be made from SAE 950C (Manten) – see 112.4
(Exercise 1), Step 4.
click on the Factor of Strength button in the Fatigue from FEA window to open the Factor of Strength
Calculations dialogue box:
Figure 112-26
select the Perform Factor of Strength (FOS) Calculations? option to enable the FOS analysis.;
select Infinite design life (use material’s Endurance Limit) – this uses the constant amplitude
endurance limit for the component material, which for SAE 950C (Manten) is: 2.0 × 107 reversals;
Note: fe-safe adopts the conventional method of expressing the endurance limit, as the number of
reversals, or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ).
click OK;
specify the FOS band limits in the Analysis Options dialogue [FEA Fatigue >> Analysis Options...]
Safety Factors tab as shown in Figure 112-27, to determine the limits and precision of the FOS
analysis:
- between the ‘fine’ limits, the FOS is calculated to a resolution of approximately 0.01 repeats;
- between maximum and minimum, the FOS is calculated to a resolution of approximately 0.1
repeats;
- if the evaluated FOS is greater than 2, then a FOS of 2 is exported for the node;
- if the evaluated FOS is less than 0.5, then a FOS of 0.5 is exported for the node.
Figure 112-27
Figure 112-28
Check that the analysis is configured as shown in Figure 112-28, and then click Continue.
Note that the UTS of the material is used to evaluate Kt from the defined surface finish curve (i.e. surface
finish factor versus UTS). In this case, for a mirror-polished surface, Kt =1 for all values of UTS.
As the analysis is being performed, the following information is written to the analysis log file. The
analysis log file has the same file name as the output file, except that the extension is .log. So, for this
analysis, the analysis log file is:
<ResultsDir>\keyhole_01Results_ex02.log
This information displayed in the Message Log window includes the following
Summary
=======
Worst Life-Repeats : 433610.125
at Element 1.4:1
Worst FOS@Life=Infinite : 0.631
at Element 1.4:1
Analysis time : 0:00:09
In the configuration of the Factor of Strength analysis, the target life specified was the ‘Infinite design
life’, which is assumed to be the materials constant amplitude endurance limit, which for SAE 950C
(Manten) is: 2.0 × 107 reversals.
The worst-case fatigue lives are of the order 1.067 × 106 repeats of the fatigue loading cycle, which is
less than the design life (1 × 107 repeats of the loading). This is why the worst-case strength factor (the
factor by which the loading must be scaled to achieve the design life) is somewhat less than 1. The FOS
method is described in detail in section 17.
Note:
fe-safe adopts the conventional method of expressing the endurance limit, as the number of reversals,
or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ). In fe-safe, the
design life and the calculated fatigue life always refer to the number of repeats of the complete defined
fatigue loading cycle. An optional conversion factor can be used to convert the fatigue life in repeats to
fatigue life with respect to some other quantity, for example hours or miles (see section 13).
Figure 112-29
Figure 112-30
113.1 Introduction
This tutorial demonstrates the application of fe-safe/Rotate to models of rotating components that
exhibit axial symmetry.
This tutorial should be read in conjunction with section 21, ‘Analysis of Axially Symmetric Models Using
fe-safe/Rotate’.
The following sample files used for this tutorial are located in the directory <DataDir>\fe-
safe_rotate:
wheel_01.db : Ansys FE model file.
wheel_01.s01 : Load step file for load step 1.
wheel_01.s02 : Load step file for load step 2.
wheel_01.s03 : Load step file for load step 3.
wheel_01.s04 : Load step file for load step 4.
wheel_a_01.rst : Results of FE stress analysis for load steps 1, 2, 3 and 4.
113.2 Preparation
The tutorial uses an Ansys RST model. However, the same techniques can be applied to all FE formats
supported by fe-safe/Rotate, which currently include:
This tutorial assumes that the user is familiar with the basic program operation. Before attempting this
tutorial, it is strongly recommended that one of the introductory tutorials is followed:
Start the program by selecting fe-safe from the Windows “Start” menu (Windows) or by running the
script fe-safe (Linux) – see section 5.
Select an existing project, or create a new one from the welcome page
Figure 113-1
Ensure that the following basic settings are correctly configured. If fe-safe is being run for the first time
then it is possible to skip this preparatory section.
Reset any existing analysis options to their defaults in the Clear Data and Settings dialogue [Tools >>
Clear Data and Settings...], shown in Figure 113-2, below, by selecting the Select all option and clicking
OK.
Figure 113-2
Figure 113-3
The elements in the model exhibit axial symmetry in 64 orientations. For this tutorial, we will take
advantage of this symmetry to divide the model into four 90° axisymmetric segments. The master
segment needs to be defined by listing the group/material numbers that make up the segment (Ansys
uses material numbers to define groups of elements). The complete sample model is actually defined by
seven different material numbers, as shown in Figure 113-4:
Figure 113-4
We can define a 90° master segment by specifying that it consists of materials 1, 2, 3, 4 and 5.
In Ansys, four load steps (wheel_01.s01, wheel_01.s02, wheel_01.s03 and wheel_01.s04) were
created. The first load step was generated for the model in its original orientation. The second load step
was generated as if the model had been rotated through 22.5°. The third and fourth load steps were
generated as if the model had been rotated through 45° and 67.5°, respectively. An FE analysis of static
stresses was performed by solving the four load steps, and the results were written to the Ansys results
file wheel_a_01.rst. The stresses produced can be visualised by plotting element solutions for the four
results sets in wheel_a_01.rst. These are illustrated in Figure 113-5, which shows the von Mises
stresses for each solution.
Figure 113-5
Figure 113-6
fe-safe/Rotate is now configured, and ready to process the FE results file. Press the OK button.
A prompt to pre-scan the file will be displayed if the file has not yet been pre-scanned. When the pre-
scan of the file is complete, the Select Datasets to Read dialogue will be displayed. The file
wheel_a_01.rst contains four stress datasets corresponding to the four solutions identified earlier and
corresponding temperature datasets. Only the stress datasets are required for the tutorial, the
temperature datasets do not need to be loaded. Check the Stresses and Last increment only boxes, click
Apply to Dataset List to apply the selection and select OK to load the remaining datasets, as shown in
Figure 113-7.
Figure 113-7
When the model has finished loading, ensure that the stress, strain and temperature units are set to
MPa, µE and deg.C, respectively. If the Units dialogue is not shown automatically, right click within
Current FE Models window and select ‘Properties’. Then change the values to those shown in Figure
113-8 and click OK.
Figure 113-8
As fe-safe loads the model, information about the file and the data it contains is appended to the file:
<ProjectDir>\Model\reader.log.
This information is also displayed in the Message Log window.
A summary of the open model appears in the Current FE Models window, showing the loaded datasets
and element group information, as in Figure 113-9.
Figure 113-9
Figure 113-10
select Whole group;
click OK.
Figure 113-11
The component is assumed to be made from SAE 950C (Manten), set the Material option to SAE_950C-
Manten.
The material record SAE_950C-Manten from the material database local.dbase will be defined for the
whole component.
Opened material databases are displayed in the Material Databases window as shown in Figure 113-12,
below.
Figure 113-12
If the database local.dbase is not currently loaded in the Material Databases window, select File >>
Materials >> Open Materials Database... and select local.dbase from the directory <UserDir>. Notice
in Figure 113-12 that the database local.dbase is the user’s local copy from the <UserDir> directory.
The other two databases shown are from the <DatabaseDir> directory.
Expand the tree view to show the materials contained in local.dbase. The contents of the database can
be filtered to show only materials of a particular type, by selecting one of the filter icons . To ensure
that all materials in the database are displayed double-click on the All filter icon .
The parameters for a material can be displayed by expanding the material name.
To define SAE_950C-Manten for the whole component:
highlight the material SAE_950C-Manten in the local.dbase database;
double-click the Material column header - a Change Material? confirmation dialogue box appears;
Figure 113-13
click YES;
the material name should appear for all groups in the Material column.
fe-safe is now ready to perform the analysis. Press the Analyse! button and accept the pre-analysis
summary. The following text appears in the message log:
Summary
=======
Worst Life-Repeats : 262006.219
at Element 886.2
Analysis time : 0:00:17
This fatigue results from this exercise were written to the file:
<ResultsDir>/wheel_a_01Results.rst
The fatigue results can now be viewed in Ansys by loading the results file and selecting the last result
set in the file. The fatigue life is saved in place of the SX (stresses in X direction) parameter, plotting SX
displays a contour plot of the log of the fatigue life, as shown in Figure 113-14. The viewer colour
scheme may need to be set to "Reversed Rainbow" and any averaging of values at nodes should be
disabled.
Figure 113-14
114.1 Introduction
This tutorial demonstrates PSD fatigue analysis of a flat plate.
The tutorial includes opening the FE model and configuring the analysis, followed by calculating life
using the Dirlik algorithm with Von Mises stress. A later section demonstrates the use of the alternative
Tovo-Benasciutti algorithm using an appended residual stress dataset to provide a non-zero cycle
mean.
The ODB, RST and OP2 file formats are portable between platforms. The use of Abaqus .odb files in fe-
safe is discussed in detail in Appendix G.
The sample *.psd files used for this tutorial are located in the directory <DataDir>.
The sample Abaqus files used for this tutorial are located in the directory <DataDir>\Abaqus.
The sample ANSYS files used for this tutorial are located in the directory <DataDir>\ANSYS.
The sample NASTRAN files used for this tutorial are located in the directory <DataDir>\Nastran.
NOTE: The nomenclature used throughout this tutorial when referring to files and directories is as
described in Appendix B.
Figure 114-1
The FE model under investigation is the steel plate illustrated above. The actual FE mesh is comprised of
second order hexahedral elements. Boundary conditions ensure that points A and B are fixed in DOF 1-
6. A natural frequency analysis was run to retrieve the first 10 natural frequencies and their
corresponding stress states, resulting in abaqus_plate_single_R20xx.odb, ansys_plate.rst or
nastran_plate.op2 depending on the selected solver. Two Frequency sweep analyses were then run
on each model. These are called Steady State Dynamic (Abaqus), Harmonic (ANSYS) or Frequency
Response (NASTRAN) analyses depending on the solver. Each response analysis is characterised by a 1N
force applied in the Z direction at one of the pilot nodes displayed as points C (channel 1) or D (channel
2). In the case of Abaqus, the Steady State Dynamic Analysis steps are also in file
abaqus_plate_single_R20xx.odb for both steps. In the case of ANSYS two .mcf files were generated
from the Harmonic Analysis corresponding to the two channels. The ANSYS .mcf files were renamed
with respect to channel number (resulting in plate_1.mcf and plate_2.mcf) in order to satisfy the file
name convention detailed in section 27 of the fe-safe User Guide. In the case of NASTRAN two .pch files
were generated from the Frequency Response Analysis corresponding to the two channels. The
NASTRAN .pch files were renamed with respect to channel number (resulting in plate_1.pch and
plate_2.pch) in order to satisfy the file name convention detailed in section 27 of the fe-safe User Guide.
The PSD spectrum shown in Figure 114-2 will be used in the loading, and is repeated twice in the file
plate.psd so that there are two PSD channels, one corresponding to each channel in the Generalized
Displacement data (also called Modal Participation Factors in ANSYS).
Figure 114-2
Note: Due to differences in implementation of damping, frequency sweep responses from Abaqus,
ANSYS and NASTRAN models are different, but the same process flow can be applied for each solver for
the purposes of this tutorial.
For this tutorial a pre-existing PSD file has been supplied. For information on the PSD file format see
Chapter 27 of the User Guide (section 27.3.2). Also note that PSD files in the fe-safe PSD format
including cross-channel correlation may be generated within fe-safe from time-domain signals using
the Cross-Spectral Density Matrix File option in the Frequency menu. Note that this can be also be used
for the simpler case of the PSD of a single signal. See also section 27.3.2 of the User Guide. More
information on fe-safe signal processing options is given in Chapter 10 of the User Guide.
114.3 Preparation
This tutorial uses a single Abaqus ODB model, an ANSYS RST model with associated .mcf files, or a
NASTRAN OP2 model with associated .pch files. It is assumed that the user is familiar with the basic
program operation. Before attempting this tutorial, it is strongly recommended that one of the
introductory tutorials is followed:
Tutorial 106: Using fe-safe with Abaqus .odb files
Tutorial 108: Using fe-safe with ANSYS .rst files
Select an existing project, or create a new one from the welcome page
Figure 114-3
If an existing project directory is chosen, existing data in that project will be retained. This includes FE
data, load history files, analysis settings and more.
Ensure that the following basic settings are correctly configured. If fe-safe is being run for the first time
then it is possible to skip this preparatory section.
Reset any existing analysis options to their defaults in the Clear Data and Settings dialogue [Tools >>
Clear Data and Settings...], shown in Figure 114-4 below, by selecting the Select all option and clicking
OK.
Figure 114-4
For Opening the sample FE model (Abaqus Model) continue to section 114.4.
For Opening the sample FE model (ANSYS Model) continue to section 114.5.
For Opening the sample FE model (NASTRAN Model) continue to section 114.6
Figure 114-5
Note: The complex Generalized Displacement Data being imported into fe-safe can be expressed in
either polar or rectangular form. By default, a Steady State Dynamic Analysis in Abaqus exports such
data in polar form, i.e. with modulus and argument components (where the angles are expressed in
degrees), hence the Complex number notation field setting above.
The Select Datasets to Read dialogue will appear as shown in Figure 114-6. Check the Quick-select
boxes for Stresses and Other, and if it is selected, deselect the box for Last Increment Only. Click Apply
to Dataset List.
Deselect the dataset marked Mode 0 (this Initial State is not used in PSD loading) as shown in Figure
114-10.
Click OK.
Figure 114-6
Scroll down and make sure that Steps 2 and 3 (the Generaized Displacements and Phase Angle Histories)
are selected.
As fe-safe loads the model, information about the file and the data it contains is appended to the file:
<ProjectDir>\Model\reader.log.
This information is also displayed in the Message Log window.
When the model has finished loading, the Loaded FEA Models Properties dialogue box appears, as
shown in Figure 114-7 Set the units to MPa, N, mm.
Figure 114-7
If the dialogue box does not appear automatically, then it can be displayed by double-clicking on the
icon in the Current FE Models window.
Ensure that the units are as shown in Figure 114-7, then click OK.
A dialogue will appear prompting to edit element groups loaded from the model, click Yes.
Use the button to move all groups to the list of Unused Groups. Click OK.
A summary of the open model appears in the Current FE Models window, showing the loaded datasets
and element group information – see below.
Figure 114-8
Note: If the window does not appear exactly as shown in Figure 114-8, then expand the tree view to
show more details.
The process of loading an FE model into fe-safe involves extracting pertinent data from the FE model,
and writing it to the /model subdirectory of the project directory (see Appendix E, section 205.4.2).
Therefore, the Current FE Models window is a summary of the contents of the working FED folder. In this
case there are 10 modal stress datasets each containing elemental data with 925 elements in each
dataset. The source of the dataset (the filename of the source FE model, the step, increment and
timestamp) are also shown. fe-safe also extracts element group information from the FE model file.
The Generalized Displacements (GDs) were loaded from Steps 2 and 3 of the FE model. These define
weightings for each node for each mode within each channel, and are used to convert the input channel
PSDs to the nodal response PSDs which are the basis of the fatigue calculations. The data is
summarised in the Generalized Displacements item in the Current FE Models window. The sample file
contains 10 modes, and two channels of GDs from 0 to 300 Hz as shown above.
When a new model is loaded, the output filename automatically defaults to:
<ResultsDir>\{source_file_name}Results.{source_file_extension}
The output filename can be modified by manually editing the Output File field in the Fatigue from FEA
dialogue, or by clicking the adjacent browse button: .
Continue to section 114.7 - Exercise 1: Finite life using the Dirlik algorithm with Von Mises stress.
Figure 114-9
Note: The default settings for a Harmonic Analysis in ANSYS result in complex-valued Generalized
Displacement data (also called Modal Participation Factors in ANSYS) which are exported in rectangular
form, i.e. with real and imaginary components, hence the Complex number notation field setting above.
A dialogue will be shown prompting the user Do you want to Pre-scan…? Click Yes.
The Select Datasets to Read dialogue will appear as shown in Figure 114-10. Check the Quick-select
box for Stresses , and if it is selected, deselect the box for Last Increment Only. Click Apply to Dataset
List.
The Generalized Displacements (GDs) will not appear in the list. This is expected, as these are loaded
from separate .mcf text files for ANSYS models.
Click OK.
Figure 114-10
As fe-safe loads the model, information about the file and the data it contains is appended to the file:
<ProjectDir>\Model\reader.log.
This information is also displayed in the Message Log window.
When the model has finished loading, the Loaded FEA Models Properties dialogue box appears, as
shown in Figure 114-11. Set the units to Pa, N, m.
Figure 114-11
If the dialogue box does not appear automatically, then it can be displayed by double-clicking on the
icon in the Current FE Models window.
Ensure that the units are as shown in Figure 114-11, then click OK.
A dialogue will appear prompting to edit element groups loaded from the model, click Yes.
Use the button to move all groups to the list of Unused Groups. Click OK.
A summary of the open model appears in the Current FE Models window, showing the loaded datasets
and element group information – see below Figure 114-12.
Figure 114-12
Note: If the window does not appear exactly as shown in Figure 114-12, then expand the tree view to
show more details.
The process of loading an FE model into fe-safe involves extracting pertinent data from the FE model,
and writing it to the /model subdirectory of the project directory (see Appendix E, section 205.4.2).
Therefore, the Current FE Models window is a summary of the contents of the working FED folder. In this
case there are 10 modal stress datasets each containing elemental data with 925 elements in each
dataset. The source of the dataset (the filename of the source FE model, the step, increment and
timestamp) are also shown. fe-safe also extracts element group information from the FE model file.
The FE analysis files (.mcf files for ANSYS) containing the Generalized Displacements were loaded. The
data is summarised in the Generalized Displacements item in the Current FE Models window. The
sample file contains 10 modes, and two channels of GDs from 0 to 100 Hz as shown above.
When a new model is loaded, the output filename automatically defaults to:
<ResultsDir>\{source_file_name}Results.{source_file_extension}
The output filename can be modified by manually editing the Output File field in the Fatigue from FEA
dialogue, or by clicking the adjacent browse button: .
Continue to section 114.7 - Exercise 1: Finite life using the Dirlik algorithm with Von Mises stress.
Figure 114-13
A dialogue will be shown prompting the user Do you want to Pre-scan…? Click Yes.
The Select Datasets to Read dialogue will appear as shown in Figure 114-14. Check the Quick-select
box for Stresses , and if it is selected, deselect the box for Last Increment Only. Click Apply to Dataset
List.
The Generalized Displacements (GDs) will not appear in the list. This is expected.
Click OK.
Figure 114-14
As fe-safe loads the model, information about the file and the data it contains is appended to the file:
<ProjectDir>\Model\reader.log.
This information is also displayed in the Message Log window.
When the model has finished loading, the Loaded FEA Models Properties dialogue box appears, as
shown in Figure 114-15. Set the units to MPa, N, m.
Figure 114-15
If the dialogue box does not appear automatically, then it can be displayed by double-clicking on the
icon in the Current FE Models window.
Ensure that the units are as shown in Figure 114-15, then click OK.
A dialogue will appear prompting to edit element groups loaded from the model, click Yes.
Use the button to move all groups to the list of Unused Groups. Click OK.
A summary of the open model appears in the Current FE Models window, showing the loaded datasets
and element group information – see below Figure 114-16.
Figure 114-16
Note: If the window does not appear exactly as shown in Figure 114-16, then expand the tree view to
show more details.
The process of loading an FE model into fe-safe involves extracting pertinent data from the FE model,
and writing it to the /model subdirectory of the project directory (see Appendix E, section 205.4.2).
Therefore, the Current FE Models window is a summary of the contents of the working FED folder. In this
case there are 10 modal stress datasets each containing elemental data with 925 elements in each
dataset. The source of the dataset (the filename of the source FE model, the step, increment and
timestamp) are also shown. fe-safe also extracts element group information from the FE model file.
The FE analysis files (.pch files for NASTRAN) containing the Generalized Displacements were loaded.
The data is summarised in the Generalized Displacements item in the Current FE Models window. The
sample file contains 10 modes, and two channels of GDs from 0 to 100 Hz as shown above.
When a new model is loaded, the output filename automatically defaults to:
<ResultsDir>\{source_file_name}Results.{source_file_extension}
The output filename can be modified by manually editing the Output File field in the Fatigue from FEA
dialogue, or by clicking the adjacent browse button: .
114.7 Exercise 1: Finite life using the Dirlik algorithm with Von Mises stress
Objective:
Calculate fatigue life of the plate using the loaded Generalized Displacements and 2-channel PSD signal.
This procedure is applicable to Abaqus, ANSYS and NASTRAN-derived input files (as summarised in
sections 114.4, 114.5 and 114.6); however, the screenshots will focus on the ANSYS-related procedure
and results.
Analysis process:
The finite life for each node is calculated as follows:
Modal stresses and GDs are collected and used to calculate frequency response functions.
The frequency response functions and the PSD data are used to calculate the Von Mises
equivalent stress.
The Dirlik method makes use of the Von Mises equivalent stress to provide a closed form
solution to estimate the Probability Density Function (PDF) of stress range and produce a cycle
histogram of stress ranges.
Fatigue damage per block is calculated from the cycle histogram:
o The number of ‘bins’ in the cycle histogram is configured by setting the Number of
stress range intervals.
o The maximum stress range in the histogram is configured by setting the RMS stress
cut-off multiple (see section 27 of the fe-safe User Guide).
Method:
Step 1: Automatic loading definition:
For this exercise 10 modes and GDs for two channels will be combined with a two channel PSD file to
create the PSD loading. The necessary loading definition was generated automatically. To view the
automatic loading definitions:
Select the Loading Settings tab from the Fatigue from FEA dialogue to switch to the loading tree. The
defined loading appears in the loading details list box, as shown in Figure 114-17.
Note: If the window does not appear exactly as shown in Figure 114-17, then expand the tree view to
show more details. Instructions on how to complete the same loading definitions manually are in the
notes at the end of this exercise.
Figure 114-17
The signal time length of 1 seconds is specified in the plate.psd file header. If a psd file is used without
this optional header line, then the time would display as zero, and you would need to double click on the
Length per repeat field to set the time length of the signal.
Select the Analysis Settings tab from the Fatigue from FEA dialogue to switch to the main settings.
Figure 114-18
Figure 114-19
select Select Surface Finish from list;
select the surface finish database file uni7670-1988-Steel.sfprop from the Definition File drop-
down list;
from the drop-down Surface finish list, select Mirror Polished – Ra <= 0.25µm;
click OK.
Figure 114-20
If the database system.dbase is not currently loaded in the Material Databases window, select File >>
Materials >> Open Materials Database... and select system.dbase from the directory <DatabaseDir>.
Notice in Figure 114-20 that the database system.dbase is the System Database from the
<DatabaseDir> directory.
The parameters for a material can be displayed by expanding the tree view of the material name.
To define SAE-1045-6 for the whole component:
In the Material Databases window, highlight the material SAE-1045-6 in the system.dbase
database.
In the Fatigue from FEA dialogue (group parameters table), double-click the Material column header
- a Change Material? confirmation dialogue box appears; click YES.
The material name should appear for all groups in the Material column.
Figure 114-21
Select the default Dirlik option then click OK.
There is a further option, selectable by checkbox, to bound the Dirlik damage integral at the Ultimate
Tensile Strength (UTS) of the configured material as shown in Figure 114-22. If the UTS is lower than the
SN curve intercept point, then the effect is to use the UTS in place of the SN curve intercept as the upper
limit to the integral, after which the tail is treated as having damage of 1. Use of this option is usually
over-conservative at low life, as for most materials the UTS is lower than the SN curve intercept, but is
provided for backwards compatibility with earlier versions of fe-safe, or for when a material’s SN curve
is not regarded as valid beyond the UTS.
Figure 114-22
After steps 1 through 7 are complete the Fatigue from FEA dialogue appears as shown in Figure 114-23.
Figure 114-23
Step 8: Run the analysis:
Figure 114-24
Check that the analysis is configured as shown in Figure 114-24, and then click Continue.
As the analysis is being performed, information is written to the analysis log file. The analysis log file
has the same file name as the output file, except that the extension is .log. So, for this analysis, the
analysis log file is:
<ResultsDir>\*_plateResults_ex01.log
This information is also displayed in the Message Log window and includes:
Summary
=======
Worst Life-Repeats : 3.199
at Element [0]25.7
Analysis time : 0:00:10.846
Note: these are figures taken from the log of the analysis on the Abaqus model, continue reading for the
results from both Abaqus and Ansys examples.
To save the current analysis configuration, select Save FEA Fatigue Definition File..., and enter a filename
ending with the extension .stlx. The analysis can be reloaded at a later date by restoring the project
definition file, using Open FEA Fatigue Definition File....
The analysis log shows that the worst-case life for the whole model is:
Abaqus: 3.199 repeats of the 1 second loading block, at element 25, node 7.
ANSYS: 37.435 repeats of the 1 second loading block, at element 925, node 3.
NASTRAN: 9169.305 repeats of the 1 second loading block, at element 25, node 7.
Note: fe-safe adopts the conventional method of expressing the endurance limit, as the number of
reversals, or half cycles ( 2 N f ), whereas fatigue lives are expressed in terms of cycles ( N f ). In fe-safe,
the calculated fatigue life always refers to the number of repeats of the complete defined fatigue loading
cycle. If necessary, a conversion factor can be used to convert the fatigue life in repeats to fatigue life
with respect to hours or miles (see section 13 of the fe-safe User Guide).
Note: The contour plot above is a reverse-contour. It shows the lowest Log Life as red and the highest
Log Life as blue. Depending on which post-processor you use, the method to reverse the contour
colours will vary.
Additional notes for Exercise 1:
1. Recall that the RMS stress cut-off multiple was set to 4. Note that a higher value will increase
the stress range under investigation. To illustrate this point the same analysis using ANSYS files
was run for a number of different RMS stress cut-off multiple values. The greatest damage was
always at element 925 (similar to Figure 114-25), leading to the results (in log base 10 notation)
displayed in Figure 114-26. Such results will vary on a case by case basis; however, the general
trend is always the same. This illustrates why, for a conservative worst life estimate, a large RMS
stress cut-off multiple is recommended (which is why the default value is set to 10).
log10(worst life)
4
0
0 5 10 15 20
RMS stress cut-off multiple
Figure 114-26
d. The Log for Items tab allows additional information about the PSD analysis for the nodes on the
List of Items to be exported to the results .log file by checking the box to Export PSD Items.
The PSD diagnostics table in the log file appears as follows:
e. For every PSD analysis there is an implicit assumption that all loaded modal stress dataset / GD
pairs contribute to the PSD loading block definition(s). The Log for Items tab allows a sensitivity
analysis to be performed on such data. To run, check the Enable load sensitivity analysis box.
For each item on the list of items, a series of analyses will be run where each analysis is
characterised by the omission of a single modal stress dataset and its associated set of GDs, per
PSD block, i.e. in this example there would be ten additional analyses where the first analysis
omits Dataset 1 and its associated GDs, the second analysis omits Dataset 2 and its associated
GDs, etc… The results are then exported to the projects .log file. The sensitivity analysis for the
ANSYS analysis is shown below:
SENSITIVITY ANALYSIS for Element 925.3 (The life is for 1 repeat of the block
(i.e n=1), it does not consider the n Value if this is an LDF analysis)
Excluding Dataset 1 (Natural Frequency 8.14741 Hz) and its associated GDs increases the life
over 3000 times.
3. In Figure 114-22 the Implement Von Mises-based nodal filtering check box is a critical plane
search-related option, i.e. it will only be enabled when a critical plane option is selected. If
checked, the box indicates that fe-safe will implement 'nodal filtering'. More precisely, response
PSD moments will be initially calculated (for all nodes) by using the Von Mises stress, and only
nodes with lives below the constant amplitude endurance limit will be further processed using a
critical plane search. In models where most of the lives are infinite this allows faster processing
of the majority of nodes which undergo no damage. A 20% error margin is applied to the Von
Mises stress in this filtering, and for speed the life being checked in the filtering process is
estimated using a conservative narrowband approximation.
114.8 Exercise 2. Using the Tovo-Benasciutti algorithm with residual stress effects (ANSYS Model)
The second example uses a notched beam model taken from experiments described in [1,2]. This is
illustrated below in Figure 114-7.1. In the experiments the region labelled as “Restrained Nodes” in
Figure 114-7.1 was attached to a vertical rod (Z direction), which was the source of the vibration. The
Large Mass Method was use with Ansys Workbench to perform a modal analysis with 10 modes and the
frequency range limited to 0.1Hx - 1E8 Hz. The node associated with the Large Mass had all its degrees
of freedom removed, apart from movement in the Z direction (i.e. the input oscillation direction). An
illustration of the first 4 modes, and their resonant frequencies are shown in Figure 114-28 Modal
Analysis.
Next a Modal Superposition Harmonic Analysis was performed in Ansys to generate Generalized
Displacements (Modal Participation Factors in ANSYS terminology), which are used in fe-safe to weight
the contribution of each mode shape when calculating the response PSD.
The Generalized Displacements are shown in Figure 114-29 (Mode 3 has a negligible Generalized
Displacement in this case). Thirdly the input PSD was defined (see [1] or PSD #1 in [2]), representing the
driving acceleration of the vertical rod, as in Figure 114-30.
In addition a forming analysis was carried out to calculate residual stresses present in the beam. These
will be used to provide the background mean stress for the PSD analysis.
Figure 114-31
A dialogue will be shown prompting the user Do you want to Pre-scan…? Click Yes.
The Select Datasets to Read dialogue will appear as shown in Figure 114-32. Check the Quick-select
box for Stresses , and if it is selected, deselect the box for Last Increment Only. Click Apply to Dataset
List.
Figure 114-32
The Generalized Displacements (GDs) will not appear in the list. This is expected as they are read in
from separate (.mcf) text files.
Click OK.
When the Loaded FEA Models Properties dialogue appears to conclude the model loading, select MPa as
the stress units; other units are left at their defaults (Figure 114-33).
Figure 114-33
A dialogue will be shown prompting to edit the group list or not, choose No
Figure 114-34
A warning will be shown listing some unsupported element types, this can be safely ignored
Next the residual stress dataset output from the Forming Analysis must be loaded to provide the
background mean stress. Use Append Finite Element Model from the File >> FEA Solutions menu. Select
file notched_beam_residual_stress.rst from the <DataDir>/Ansys directory. A warning message
will appear saying that in PSD analyses only residual stress or temperature datasets may be appended.
This is to be expected, click OK on the warning, and then click Yes on the subsequent pre-scan option
message to bring up the pre-scan dialogue, and then select stress and strain datasets (Figure 114-35).
Note that the associated strain dataset must be loaded as well as the stress dataset. This is not used in
the PSD analysis, but the general fe-safe residual stress interface for Loadings expects both to be
present, due to its original use in Elastic-Plastic analyses.
Figure 114-35
Next the loading must be defined. There will be an automatic loading defined by the PSD model open
process, but it is necessary to define the signal length, and add the residual. Switch to the Loading
Settings tab. Double-click on the Length per repeat in seconds field of the PSD Block and set the signal
length to 1 (second).
To add the residual stress dataset to the loading, first click on it under the appended model on the
Current FE models area to the right (Dataset 11).Then click on the Settings field near the top of the
loading page. Click the Add… button above, then select Residual dataset pair from the menu that
appears.
Finally double click on the associated Residual strain dataset which must also be added (even though
this will not be used in the PSD analysis) and set it to Dataset 12. The loading should look like Figure
114-36.
Figure 114-36
Then switch back to the Analysis Settings tab to select the material and select Tovo-Benasciutti as the
PSD algorithm. Expand the system database of the Material Databases, and select SAE-1045 as the
material. Then double-click on the Material column in the Analysis Settings to apply this material. Then
double-click on the Algorithm column (which will have defaulted to Dirlik). This will bring up the PSD
algorithm selection dialog as shown in Figure 114-37. Select Tovo-Benasciutti using the Stochastic
mean centred on residual stress sub-option, and leave the mean stress correction method at the default
Morrow. The stochastic mean sub-option will account for the effect on the mean-stress correction of
random variability about the overall mean defined by the residual stress.
Figure 114-37
Figure 114-38
Summary
=======
Worst Life-Repeats : 86537.422
at Element 1224.2
Analysis time : 0:00:00.644
The expected life results will be written to stress component Sx in the Ansys results file. This can be
viewed back in Ansys and should look like Figure 114-39. As might be expected the fatigue hotspots
are around one of the notches.
Figure 114-39
Figure 114-40 shows a zoomed in view around the notch containing element 1224 where most of the
damage occurs.
Figure 114-40
A more detailed investigation of the response PSD can be carried out for the hotspot node. On the
Exports dialog select List of Items tab and enter 1224.2 as the item. Then click on the Histories for
Items tab and select Export PSD frequency response function plots. This will produce a plottable text file
of the reponse PSD against frequency in the project fe-results directory. Select File >> Data Files >>
Open Data File(s) to open file:
notched_beam_modalResults_psd_frf_plot-e1224.2_block_1.txt in the jobs/job_01/fe-results sub-
directory of the project.
This will appear in the Loaded Data Files column. Expand the data tree below this file in the Loaded Data
Files column, to reveal the Frequency and PSD Response fields. Select both, then right click and select
Cross Plot from the popup menu. Right click on the displayed plot, and select Properties on the popup
menu, then change the Vertical Axis scaling to Logarithmic using the Logarithmic Scale checkbox.
The response PSD plot should look like Figure 114-41.
References
1. V. K. Nagulapalli (2005). Fatigue Life of Notched Aluminium Beams at Elevated Stress Levels,
MsC Thesis, 2005, Northern Illinois University.
2. V. K. Nagulapalli, A. Gupta, S. Fan (2007). Estimation of Fatigue life of Aluminium Beams
subjected to Random Vibration, Department of Mechanical Engineering, Northern Illinois University, in
Conference Proceedings: 2007 IMAC-XXV: Conference & Exposition on Structural Dynamics.
115 Tutorial 115: Applying Theory of Critical Distance (TCD) Methods in fe-safe
115.1 Introduction
This tutorial demonstrates how to run Factor of Strength (FOS) and Fatigue Reserve Factor
(FRF) analyses using Theory of Critical Distance methods in fe-safe. The sample model is a
connecting joint of a crankshaft subjected to torsional and bending loads.
In general, the fatigue life to crack initiation is determined by the sequence of stress/strain
cycles on the surface of a component. At stress concentration regions (such as sharp notches,
grooves, corners, etc.) there may exist stress gradients with sub-surface stresses significantly
lower than those at the surface. Whether or not the crack will propagate depends on the
stresses at some distance below the surface. The critical distance, L, is a material property,
and the difference between the two stresses is an indication of the material “notch sensitivity.”
The larger the critical distance L, the lower the “notch sensitivity” of the material. Fig.1 shows
the stress sampling point at distance rc below the surface. For the simplest critical distance
method, the Point Method, rc = L/2.
The benefit of using critical distance calculations in fatigue analysis is that stresses at the
surface of model may be increased to a level where cracks initiate at the surface, but those
cracks will not propagate internally. If FOS or FRF values close to 1.0 are observed at areas of
stress concentrations, it may be worthwhile running TCD calculations to check if cracks that
may develop on the surface of your model will propagate and cause fatigue failure.
fe-safe also includes the Line Method for TCD calculations. For the Line Method, the mean
stress over a distance of 2L below the surface is used.
At the end of the tutorial a quick reference topic summarizes the key points to remember
when running TCD calculations with either FOS or FRF analysis in fe-safe.
The sample files used for this tutorial are located in the directory <DataDir>\Abaqus.
These Abaqus CAE output files are labelled:
• crankshaft_TCD_R20xx.odb
Where xx denotes the version of the Abaqus solver used to generate the files.
A second set of sample files is used to rerun the analysis with the nodal-averaged stress
values:
o crankshaft_TCD_nodal_averaged_R20xx.odb
Before you proceed with the tutorial steps, it is recommended to review chapter 26 in the fe-
safe User Guide: Theory of Critical Distance Methods (TCD).
NOTE: The nomenclature used throughout this tutorial when referring to files and directories
is as described in Appendix B, Section 202.2.
The model for this tutorial is a connecting joint of a crankshaft. A bending moment and a
torque of 1E+006 N.mm each are applied to the joint as shown in Fig.2. The mesh is
generated with a global element size set to 4 mm, but it is refined in the fillet regions that
have the highest stresses (Fig.3). Second-order tetrahedral elements are used over most of
the model, and second-order wedge elements form two boundary layers at the surface.
To ensure accurate interpolation of stresses for TCD calculations, you should mesh the model
with at least four elements over a critical distance 2L for the Line method and with at least
two elements over a critical distance L/2 for the Point method.
Select an existing project directory, or create a new one from the welcome page In the Welcome to
fe-safe screen.
1. Click File, and select FEA Solutions, Open Finite Element Model.
2. Browse to the directory <DataDir>\Abaqus, and select one of the sample files:
crankshaft_TCD_R20XX.odb
3. A Pre-Scan File Check dialog box prompts you to pre-scan the file. Click Yes.
NOTE: As fe-safe loads the model, it appends information about the solution and the data it
contains to the file:
<ProjectDir>\Model\reader.log.
This information is also displayed in the Message Log window.
5. When the model is loaded, use the Loaded FEA Models Properties dialog box to set
Stress Units to MPa and Distance Units to mm, and click OK.
6. In the Edit Group List dialog box, click Yes to access the Manage Groups dialog box.
The following steps will be used to remove element groups from the analysis.
a. Move all groups listed under Analysis Groups (right side) to Unused Groups
(left side) using the left-arrow button marked All.
b. Click OK.
A summary of the loaded odb file appears in the Current FE Models window, showing the
loaded data sets and element group information, as shown in Fig.7.
NOTE: If the window does not appear exactly as shown in Fig.7, expand the tree view to show
more details. The Current FE Models window shows a summary of the contents of the
loaded odb file. There are two stress data sets each containing elemental data for
88,226 elements.
To run TCD calculations, you need to define one of the two TCD material parameters:
taylor: L (mm) for the critical distance, or taylor: Kthreshold@R (MPa m^1/2) for the fatigue
crack growth stress threshold.
In this example you apply the material SAE-1045 (from the system material database) and set
the critical distance value L = 0.154 mm for material parameter taylor: L (mm).
First, search for the material that will be copied.
1. In the Material Databases window, use the <New Filter> field to enter the text: 1045
and expand the system folder to find the material labelled SAE-1045.
Next, you will copy the material SAE-1045 to the local material database to edit the critical distance.
3. Navigate to the local material folder, and expand the SAE-1045 entry to set the value
of the taylor: L (mm) parameter to 0.154 mm.
NOTE: The value of the parameter L is taken from Ref [1]. The Kthreshold@R value of 5
is a material default but is only used as a fallback method to calculate L if it is not
specified, and is not used here as L is defined.
5. From the Fatigue from FEA -> Analysis Settings tab, double-click the Material column
header. When prompted to change the material for the Default group, click Yes.
The material SAE-1045 appears for the Default group in the Material column.
4. Click OK.
NormalStress:-Goodman appears for the Default group under the Algorithm column.
Note: The default element group is used for the FOS calculations. Surface is selected for
Subgroup to limit the fatigue analysis to elements with exterior faces.
2. In the Current FE Models window, expand Increment 1 and select (highlight) Dataset
1:(1.1) S: Stress.
3. Right-click Block, and select Add dataset.
The Stress Dataset 1 icon appears under Elastic Block.
4. Right-click the Stress Dataset 1 in the Elastic Block, and select Scale to enter the
factor 0.4.
5. Right-click Stress Dataset 1, and select Add user loading to dataset. Type 0 in the first
row and 1 in the second row of the load history table.
Follow Steps 2 to 5 to add Dataset 2:(1.2) S: Stress in the Elastic Block. Scale Stress
Dataset 2 by a factor of 0.17.
When a new model is loaded, the output filename automatically defaults to:
<ResultsDir>\{source_file_name}Results.{source_file_extension}
which for this example is:
<ResultsDir>\ cranskshaft_TCD_R20xxResults.odb
The critical distance implementation is primarily intended for infinite life calculations. It is possible
to specify a finite design life for either FOS or FRF, but note that the critical distance L should then be
modified to be appropriate to the target life. fe-safe does not model the variation of L with life which
has been observed with elastic FEA analyses, and it is the user’s responsibility to specify a value
appropriate to the design life. See also notes in section 115.12. If the infinite life value of L is simply
retained, then the results will generally be conservative (at least for factors close to 1), as L will have
been under-estimated. Some research indicates that a constant L may be used for finite life
calculations if a full elastic-plastic FEA is performed. In this example, an infinite design life is used.
NOTE: fe-safe adopts the conventional method of expressing the endurance limit as the
number of reversals, or half cycles (2 𝑁𝑁𝑓𝑓 ), whereas fatigue lives are expressed in terms of
cycles ( 𝑁𝑁 𝑓𝑓 ).
cycle (legacy) option is only meaningful in a legacy mode, see Section 26.7 in
the fe-safe User Guide for details).
b. Under Log, select Critical Distance summary.
Note: For this small tutorial model, you can leave the limits for the FOS calculation unchanged.
When running TCD on larger models, computational time could be an issue. It may be worth
reducing the FOS or FRF maximum threshold to effectively limit TCD calculations to hot spot
elements on the surface; for example, for large models reduce the max FOS or FRF value from
10.0 to 1.50 for FOS or 2.0 for FRF. Critical distance calculations are, by default, performed
with the Point method. It is also possible to change to the Line Method as follows. The FOS
bands are specified on the FOS subtab of the FEA Fatigue->Analysis Options->Algorithms
dialog, and further FOS/FRF limits for TCD calculation and Point vs Line method choices are
made via the FEA Fatigue->Analysis Options->Algorithms->TCD subtab.
The results show that for this particular fatigue loading event, the calculated worst FOS value
for infinite life is less than 1.0 (0.938 at element 3646/node 1), and a crack initiation is
possible after 3,922,102 repeats of the fatigue loading. The worst calculated FOS at critical
distance shows a marginal improvement (1.074 at element 577/ node 8 located on the fillet).
Depending on the assessment of the consequences of failure for the particular component,
the user might require higher values of the FOS at critical distance to ensure the surface
cracks will not propagate to the interior.
Three fatigue result sets ---the strength factor for infinite life at critical distance, the
strength factor at the surface for infinite life, and the fatigue life--- are appended to a new
step in the output file <ResultsDir>/crankshaft_TCD_R20xxResults.odb.
For a list of all the Critical Distance warning and error codes see Section 26.9 Diagnostic
Codes and viewing the diagnostic contour in the fe-safe User Guide.
The figures below show contour plots of the crankshaft model for FOS at critical distance and
FOS for infinite life.
1. From the Fatigue from FEA -> Analysis Settings tab, double-click the Algorithm
column header to open the Group Algorithm Selection dialog box.
2. In the Group Algorithm Selection dialog box, for the User algorithm, select Goodman
under Normal Stress (CP) -> Infinite Life – Fatigue Reserve Factor from Envelopes.
3. Select Infinite design life (use material’s endurance limit), and select the Run TCD
checkbox to enable critical distance calculations. Click OK.
4. To disable the FOS calculations, in the Analysis Settings tab click Factor of
Strength, and deselect Perform Factor of Strength (FOS) calculations.
5. In the Analysis Settings tab, click Exports. In the Contours tab, select the Worst
cycle mean stress and damage parameter.
6. To save the results from the FRF analysis in a separate odb file: in the Analysis
Settings tab use the Output File field, to modify the filename, for example:
<ResultsDir>/crankshaft_TCD_R20xxResultsFRF.odb.
7. Click Analyse and click OK to proceed.
When the analysis completes, fe-safe reports the FRF results, as shown below.
The FRF results are comparable to the FOS results and confirm a marginal improvement for
critical distance Worst FRFs. The largest cycle amplitude reduces from 198 MPa to 173 MPa,
corresponding to the increase in radial FRF from 0.937 to 1.071 with TCD
115.10 Re-running FOS and FRF Analyses with Different Surface Algorithms
For the material SAE-1045, the default algorithm for the surface calculation, which is the
strain-based BrownMiller – Morrow. This strain-based method is really intended for medium
or low cycle fatigue, but nevertheless can be used with the TCD. However, the TCD is primarily
intended for stress based methods, and care must be taken in interpreting TCD results for
strain methods. In a medium to low cycle fatigue regime, the Critical Distance (L) can be
increased to be more applicable to the target life. Some research [2] concluded that a constant
value of L may be assumed with a full elastic-plastic FEA, but it is not possible to use FOS
methods on elastic-plastic loadings. For further discussion, see Quick Reference Notes for
TCD Analysis in this tutorial.
1. From the Fatigue from FEA -> Analysis Settings tab, double-click the Algorithm
column header to open the Group Algorithm Selection dialog box.
2. In the Group Algorithm Selection dialog box, for the User algorithm, select Morrow
under Brown-Miller (CP) -> Finite Life – Mean Stress Corrections.
3. To enable FOS, in the Analysis Settings tab, click Factor of Strength.
a. Select Perform Factor of Strength (FOS) Calculations?.
b. Select Infinite design life (use material’s Endurance Limit).
c. Select the Run TCD checkbox to enable critical distance calculations.
d. Click OK.
4. To save the results from the Brown-Miller FOS analysis in a separate odb file, in the
Analysis Settings tab use the Output File field, to modify the filename, for example:
<ResultsDir>/crankshaft_TCD_R20xxResultsBrownMiller.odb.
5. Click Analyse, and click OK to proceed.
When the analysis completes, the worst life results indicate no damage for the particular
fatigue loading. A strain based algorithm automatically employs a plasticity correction, which
could make the finite life higher as compared to the finite life associated with stress based
algorithm used previously. Crack initiation on the surface is not a probable event, so critical
distance calculations aren’t strictly necessary in this case. Comparing the worst FOS for
infinite life target to the critical distance FOS indicates that the factor improved for the critical
distance vs the FOS at the surface which was very close to 1.
Figure 20: FOS results (with material default algorithm Brown Miller – Morrow).
When the analysis completes, the reported worst FRF values verify that that design meets
the infinite life target at the surface. The critical distance FRF improves the margin of the
FRF-Radial.
115.11 Re-running the FOS Analysis with a Nodal-Averaged Stress Data Set
You re-run the FOS analysis with a second data set of nodal-averaged stress values, and
compare the results of the elemental and nodal-averaged stress data sets. The fatigue
loading settings and material definition remain the same.
1. Click File, and select FEA Solutions, Open Finite Element Model.
2. Browse to the directory <DataDir>\Abaqus, and select the sample file
cranskshaft_TCD_nodal_averaged.odb.
Set the Stress Units to MPa and the Distance Units to mm.
3. In the Opening FEA Models dialog box, a list of errors appears. These errors are due
to scanning of duplicate nodal values from the .odb file. These stress values are
reported at the common nodes of the C3D10 tetrahedron and C3D15 wedge elements
on the exterior surface. In an odb file with nodal-averaged stress values, nodal
averaging is not performed at nodes shared by different element types. Click OK to
proceed.
4. For the Material algorithm, select: NormalStress:-Goodman.
5. To enable FOS, in the Analysis Settings tab, click Factor of Strength.
a. Select Perform Factor of Strength (FOS) Calculations?.
b. Select Infinite design life (use material’s Endurance Limit).
c. Select the Run TCD checkbox to enable critical distance calculations.
d. Click OK.
6. Click Analyse, and click OK to proceed.
When the analysis completes, fe-safe displays the summary of results, as shown.
Figure 22: Summary of Results for Nodal-Averaged Stress Data (with material
algorithm NormalStress-Goodman).
Fatigue analysis based on the nodal-averaged stress data predicts less conservative
results. The table summarizes fatigue analysis results for the two data sets (common
material algorithm is NormalStress-Goodman). The largest difference in the results
(~24%) between the two data sets is observed for the Worst Life Repeats prediction.
The section summarizes key points to keep in mind when running TCD calculations with
Meshing
For accurate interpolation of stresses for TCD calculations, it is recommended to mesh the
model with at least four elements over critical distance 2L for the Line method and with at
least two elements over critical distance L/2 for the Point method.
Material data
• Specify the critical distance L for the material property taylor: L (mm), or the value of
the cyclic stress intensity threshold for fatigue crack propagation for the material
property taylor: Kthreshold@R:-1 (Mpa m1/2).
• The reference Multiaxial Notch Fatigue by Susmel L., Woodhead Publishing, 2009 is a
valuable source of critical distance values determined experimentally by specimens of
different engineering materials.
constant value of L can be used in the medium to high cycle fatigue realm for
estimating life if an elastic-plastic FEA is used, but FOS analyses cannot be performed
on elastic-plastic loadings.
Configuring Settings
In FEA Fatigue -> Analysis Options -> Algorithms -> FOS, and adjust FOS band settings and
iteration count as needed. The bands at which FOS (or FRF) result in a TCD calculation can be
defined under FEA Fatigue -> Analysis Options -> Algorithms ->TCD; this dialog also
permits switching between Point and Line methods for the analysis.
Note that running TCD on the whole model can result in a significant slowdown and sensible
use of the FRF/FOS bands for TCD is encouraged, so that the computational effort is
expended around significant hotspots. When running an element-nodal analysis with reduced
bands for TCD (e.g. 0.1<=FRF<=1.5) there can be elements on the border of hotspot regions
where some surface nodes have a TCD calculation and others do not. Generally the nodes
without a value will have the exported contour value filled in based on the calculated nodes.
This defaults to using the worst of the calculated nodes, but this can be changed (e.g., to the
average) using the radio buttons on FEA Fatigue -> Analysis Options ->Exports.
• Under the Contours tab, select Critical Distance for success and diagnostics contours
output (CritDist-Success and CritDist-Diagnostics).
• Under the Log tab, select Critical Distance summary to get a list of problem nodes with
the corresponding numeric diagnostic code and a short text explanation written to the
log file.
• Under the List of Items tab, specify the nodes or elements for which you want further
details to be written in the log file (critical plane search, interpolated stress
calculation, and critical path).
• All diagnostics and history data under Log for Items and History for Items will be
duplicated for surface and TCD data, with the secondary TCD history file having
“_TCD” appended to the end of its name. See section 26.7 of the User Guide.
• Under the Histories for Items tab, select Critical Distance, Critical distance stress
tensors to get critical distance tensors for each dataset in the loading for the selected
nodes or elements under List of Items. For each node in the List of Items, a text file is
created (.txt) in the results directory. Alternatively, use the Load Histories export,
which will produce output of the TCD tensor data in a similar form to surface data in
the _TCD version of the history file.Note that essentially the same information can be
output to the main Log using Dataset Stresses under Log for Items. In the latter case
there will be an entry at the surface and another for the TCD labelled as such with L.
• Under Log for Items, select Critical Distance (Critical Distance items) to get a log of the
critical distance path through the mesh in the log file for the selected nodes or
elements under List of Items. This can be helpful in checking in which element the L/2
point is located (or 2L line for Line method) .
1. Analysis Algorithms
Choose one of the finite life HCF algorithms for the surface fatigue life calculation.
2. Configuring Settings
• Select Perform factor of strength (FOS) calculations with Infinite design life.
1. Analysis Algorithms
• Choose one of the infinite life algorithms for the surface fatigue life calculation.
• Note these include Susmel-Lazzarin and the Prismatic Hull, as well as all available
Normal-Stress FRF methods. It is intended to add Dang Van for TCD at a future
release.
2. Configuring Settings
• For FRF calculations, clear Perform factor of strength (FOS) calculations with Infinite
design life.
References
1. The Theory of Critical Distances as an alternative experimental strategy of the
determination of 𝐾𝐾𝐼𝐼𝐼𝐼 and ΔKth by L. Susmel, and D. Taylor, Engineering Fracture Mechanics,
Vol.77, pp 1492-1501, 2010.
2. Susmel, L., & Taylor, D. (2010). An elasto-plastic reformulation of the theory of critical
distances to estimate lifetime of notched components failing in the low/medium-cycle
fatigue regime. Journal of engineering materials and technology, 132(2).