Below are Frequently Asked Questions (FAQ’s) about ESRD Software installation, licensing, features, usage and more. The software FAQ page will be periodically updated with fresh FAQ’s, tips & tricks.
Can’t find what you’re looking for? Contact ESRD Support.
Yes, all ESRD software products are compatible with Windows 7 and 8, including StressCheck® Professional and StressCheck®-Powered Apps. StressCheck® Professional has been tested to meet all of the technical requirements to be Windows 10 compatible.
There is nothing preventing you from installing and running both ESRD software products and FLEXnet License Manager on the same computer. You should, however, run the FLEXnet License Manager on a computer that is not shutdown or rebooted often so that the floating license keys are accessible to all users when the user requests them. The FLEXnet License Manager requires very little computer resources but the Service needs to be on a computer that remains running when users expect to use ESRD software products.
The FLEXnet administrator can control access and permissions of specific FLEXnet licensing features based on user, node or group. This is performed by creating a FLEXnet Options file (.opt) and placing this file in the root directory of your FLEXnet installation.
From Chapter 11, “Managing the Options File” in the FLEXnet v11.5 License Administration Guide:
For concurrent (floating) licenses, the license administrator can:
- Allow the use of features
- Deny the use of features
- Reserve licenses
The concurrent licenses can be held either in license files or in fulfillment records within trusted storage.
For activatable licenses, the license administrator can:
- Allow activation of licenses in a specific fulfillment record
- Deny activation of licenses in a specific fulfillment record
For all licenses, the license administrator can:
- Restrict the number of licenses available
- Control the amount of information logged about license usage
- Enable a report log file
Options files allow you, as the license administrator, to be as secure or open with licenses as you like. Lines in the options file are limited to 2048 characters. The \ character is a continuation character in options file lines.
If the path is omitted, the vendor daemon automatically looks for a file according to the following criteria:
The above means that if the name of the vendor daemon (i.e. esrd2.exe) is the same as the options file (esrd2.opt), there is no need to add the path to the options file in the license file. It will be found automatically.
GROUP ESRDEngs matt eric andrew
GROUP ESRDDevs kristen dave meng
RESERVE 2 SC10_WIN_SC_Ref3D GROUP ESRDEngs
RESERVE 1 SC10_WIN_SC_Ref3D GROUP ESRDDevs
EXCLUDE SC10_WIN_SC_Ref3D USER lori
NOLOG QUEUED
This options file restricts the use of concurrent licenses as follows:
The sum total of the licenses reserved must be less than or equal to the number of licenses specified in the FEATURE line. In the example above, there must be a minimum of three licenses on the SC10_WIN_SC_Ref3D FEATURE line. If fewer licenses are available, only the first set of reservations (up to the license limit) is used.
For more details go to: FLEXnet v11.5 License Administration Guide
There are several ways to obtain this information, described in Tip #2 on our Installation and Licensing page.
However, if requesting a Trial Evaluation of ESRD software products for a desktop machine, it is best to open a command line, type “vol” and hit Enter. Copy the 8-digit alphanumeric Disk Volume Serial Number into the “Message” field of Request ESRD Support, fill out the rest of the form and Submit.
ESRD uses Flexera’s FLEXnet v11 License Manager to manage client requests for ESRD software product features via a local network license server. Follow the below steps to get started with setting up a new FLEXnet License Manager for concurrent (floating) license management of ESRD software products.
Note: you must have Administrative privileges in order to properly install FLEXnet.
Note: if you have already sent your FLEXnet server’s HOST ID to ESRD Support, you may skip steps to obtain the server HOST ID.
Flexera’s LMTools Utility.
Note: FLEXnet v11.5 is a legacy version of Flexera license management software, and recent Windows Server operating systems may prevent the LMTools.exe included in FLEXnet v11.5 from launching directly (e.g. via double-click of LMTools.exe or running LMTools.exe from the Start menu). In this case, launching LMTools.exe must be performed via command line by right-clicking cmd.exe and selecting “Run as administrator”, navigating to the FLEXnet installation folder containing LMTools.exe (e.g. “cd C:\Program Files (x86)\FLEXnet”) and running “lmtools.exe” by command line.
Note: if the system administrator prefers to have the client provide a port number and server name, instead of using a copy of the license.dat, this is also supported by setting the environment variable ESRD2_LICENSE_FILE.
For more details: FLEXnet v11.5 License Administration Guide
If a FLEXnet 11 License Manager is already serving floating (concurrent) licenses for other software products, or the version of FLEXnet v11 available on the ESRD Resource Library is not the preferred version of FLEXnet, support for ESRD software product licensing is confirmed through FlexNet Publisher 11.16.2.1 build 245043 (64-bit). Follow the below steps to get started with setting up an existing FLEXnet License Manager for concurrent (floating) license management of ESRD software products.
Note: This operation requires administrative privileges. If you have already sent your FLEXnet server’s HOST ID to ESRD Support, you may skip steps to obtain the server HOST ID.
SERVER 12345abcdef 111222333444 28518
VENDOR esrd
VENDOR esrd2
#Existing features
FEATURE OTHERSOFTWARE companyvendordaemon …
#ESRD software product features
FEATURE SC_WIN_SC_Ref3D esrd2 …
Note: it may be required to update the port number on the SERVER line. By default, ESRD produces license files which use port 29731 but it may be necessary to use a different port (e.g. 28518).
Note: if the system administrator prefers to have the client provide a port number and server name, instead of using a copy of the license.dat, this is also supported by setting the environment variable ESRD2_LICENSE_FILE.
For more details: FLEXnet v11.5 License Administration Guide
Installing ESRD software products is as simple as double-clicking the MSI and following the instructions. Read Tip #1 on our Installation and Licensing page for more information.
Windows does support a silent installation option for all .msi files. Msiexec is the program that “interprets packages and installs products” (see links below) and can be run from the command line. Please note that any command line installation must occur from a command line being run as an administrator.
To run command line as administrator: Click on the start menu. In the search box, type cmd. Wait while Windows searches. In the list of found programs, right click on cmd (or cmd.exe) and choose “Run as administrator.”
Some condensed notes about installation options:
The following example, when executed from an administrative command line, will perform a silent installation of all features (“Complete” option in the GUI) in the StressCheck® 10.1 installation package located on user Kristen’s desktop:
msiexec /i “C:\Users\Kristen\Desktop\StressCheck®PE_10.1_x64.msi” /q INSTALLLEVEL=5
More information about msiexec command-line options:
You will need to create the environment variable “ESRD2_LICENSE_FILE” and set the value of this to be the new license path or PORT@SERVER. See the below:
In System properties, go to Advanced system settings, then “Environment Variables…” Then, click “New…” and set the variable name to “ESRD2_LICENSE_FILE” and the variable value to the license path or PORT@SERVER. Then click OK.
For example, if the new license file is located in “C:\My Documents\license.dat”, you will need to set the value to this. If the license server is “29731@server” (as in the above capture), you will need to set the value to this.
When you start an ESRD product, it should use the “ESRD2_LICENSE_FILE” value.
The quickest way to find the path to the StressCheck® license file or license server is to open a command prompt (cmd.exe) and copy/paste the following text:
REG QUERY "HKCU\Software\FLEXlm License Manager" /v ESRD2_LICENSE_FILE
Note: to paste, right click in the command prompt window, and then click “Paste”
Hit “Enter”, and the path to the license file and/or license server will be returned.
More information:
To change the references to ESRD software product license points, you must change the registry. The registry keys which control where the license points are as follows:
It is best to delete these registry values, and then start StressCheck®. When the license finder appears, simply point to the server.
One can always define the Environment Variable “ESRD2_LICENSE_FILE” and provide the path to the appropriate license file.
Note: An option for floating licenses is to unplug your Ethernet cable, and start StressCheck®. This will also cause the license finder to appear.
In order to replace an ESRD software license file currently in use on a FLEXnet license server, for example if expired or upgraded, the following steps are recommended:
For more information: FLEXnet v11.5 License Administration Guide
If running StressCheck® Professional or another ESRD software product for the first time (or after having removed all ESRD software product licensing references), you will be prompted with the ‘Specify StressCheck® License Path’ dialogue. Here you can specify a local license file or point to a license server. Read Tip #4 on our Installation and Licensing page for more information.
The StressCheck® Core license remains checked out while StressCheck® is open. Once StressCheck® is closed, the license will be released back to the server after a few minutes.
Typically, a solver license is released “in session”, meaning users should not have to close down StressCheck® to release this license once the solver completes or fails.
Once the solver completes or fails, the solver license will be released back to the server after a few minutes.
StressCheck® Advanced modules, such as Fracture Mechanics, StressCheck® Composites, SRS and BRS, may remain checked out once requested during the session. For example, extracting a stress intensity factor (SIF) during a session will result in a Fracture Mechanics module license being checked out. The Fracture Mechanics license will then be returned to the license server once that session is terminated.
The MeshSim Advanced module, which is commonly used to generate boundary layer automeshes around 3D cracks, will be checked in shortly after usage.
Much of the current StressCheck® Professional code base is single-threaded, though over time ESRD is working to improve as much of the code as possible to be multi-threaded. Therefore, the current release of StressCheck® is more multi-threaded than previous releases, and less multi-threaded than it will be in future releases.
Additionally, the current release of StressCheck® does not yet support parallel/multi-core processing.
Regarding the number of cores, the Windows OS handles all the core allocation for StressCheck® processes. It is possible to set processor affinity for specific applications such as StressCheck®: https://www.tekrevue.com/tip/restrict-apps-cpu-cores-processor-affinity/
ESRD has created a Student Version of StressCheck® v9.2 that you are free to download and use, with no expiration date. It is a great way to learn the basics of StressCheck®. The Student Version StressCheck® has the following features and limitations:
Download Software
StressCheck® Professional and StressCheck®-Powered Apps are designed and certified by ESRD to install and run on Windows 7 and Windows 8 64-bit.
StressCheck® Professional v10.3 and v10.4 have been tested to meet all of the technical requirements to be Windows 10 compatible.
Minimum hardware and software configuration requirements can be found here.
Some StressCheck® users may experience decreased model performance and/or integrity in certain situations. The following may be considered as potential reasons:
Is your “Path to Scratch Directory” option pointing locally or to a network location? If your scratch directory (i.e. a temporary directory automatically created for StressCheck®-related session files and solution data) is pointing to a network path, this can drastically slow down model loading and solution times, and even affect the data integrity.
Check the scratch directory path by opening StressCheck® and clicking File > Options, and verify that the “Path to Scratch Directory” value is pointing locally (e.g. C:\Users\…..\AppData\Local\Temp”, which is the default), and is not pointing to a restricted area on your machine (e.g. C:\Program Files\ESRD\StressCheck® PE 10).
It is recommended to keep the scratch directory path relatively clean of old/discarded StressCheck® temporary directories (dataset folders) as well as other non-essential temporary folders to reduce overhead. These temporary directories have the format of “dataset-xxxxxxxxx-xxxxxxxx”, with these directories (and their contents) normally removed after StressCheck® sessions are closed by the user.
However, if StressCheck® sessions exit abnormally, these directories may remain. It is recommended to remove any unwanted temporary directories in the scratch directory path.
It is recommended to read (open) and write (save) StressCheck® files locally instead of relying on a network connection. This ensures the packing/unpacking procedure is allowed to complete uninterrupted. Additionally, the local directory should not be a system or restricted folder requiring additional write permissions.
Opening a StressCheck® file from a mail client, such as Outlook, may cause unpacking issues. Therefore, StressCheck® file attachments should be saved locally before opening.
Adhering to these recommendations will ensure that the integrity of StressCheck®‘s files are preserved, and there are no interruptions during the file read/write that can cause corruptions.
Ideally, the scratch directory path would have plenty of disk space (i.e. > 1 GB), otherwise larger models will almost certainly load and solve at slower rates due to limited disk availability.
Also, there is much file I/O occurring during model loading and solving (especially during stiffness matrix inversion), which can be further limited by your hard drive’s read/write speed.
Ideally, the hard drive would be a steady state hard drive (SSD) or at least a high-RPM mechanical hard disk drive (HDD). If your hard drive is quite fragmented, near capacity and/or has a slower RPM mechanical hard disk drive, model performance may be slowed.
Learn more about ESRD Software Product System Requirements.
Are you running a node-locked or floating license configuration? For node-locked (i.e. machine-specific) licenses, firewalls/network security slowdowns typically are not an issue as there is no network communication involved.
However, for floating (server-based) licenses, communication through a restrictive network firewall to/from the license server’s FLEXnet license manager can cause significant slowdowns in model performance. This communication is required in order to obtain/return StressCheck® license features and check the license server’s “heartbeat” (i.e. determine if the license server is still capable of serving licenses).
Check with your system administrator to determine if your StressCheck® license server communicates through a higher-security, firewall-enabled network that limits/prevents network communication or specific ports. If this is the case, slowdowns may also be observed for other software products which need to communicate with the same license server.
Ideally, the graphics processor for StressCheck® (and all ESRD products) should be set to your video card’s high-performance graphics processing unit (GPU) instead of integrated graphics (e.g. Intel on-board). Some examples of video card manufacturers include Nvidia and AMD, which design dedicated GPU’s for high-performance graphics acceleration and rendering.
For additional information on ensuring you are using your video card’s high-performance GPU (such as Nvidia or AMD) for StressCheck®: https://www.techadvisor.co.uk/how-to/pc-components/how-set-default-graphics-card-3612668/
Has processor affinity (i.e. the specific declaration of CPU cores to be used for an application) been set? Learn more: https://www.tekrevue.com/tip/restrict-apps-cpu-cores-processor-affinity/
Are you connecting to a computer via remote desktop? ESRD products will not run on a physical or virtual machine using a node-locked license when accessed via remote desktop. This is due to the design security of our license manager, FLEXnet. If you would like to run ESRD products on a computer remotely, please contact ESRD Support to upgrade to a floating license.
When attempting to obtain a license from a license server, StressCheck® does not appear to be able to communicate with the license server even though the client machine is on the network. Checking LMTools server status returns error message “lmgrd is not running: Cannot connect to license server system. The license server manager (lmgrd) has not been started yet, the wrong port@host or license file is being used, or the port or hostname in the license file has been changed.”
The firewall on the server machine is blocking incoming requests or the license server has been stopped. Disable the firewall, make sure the license server is running and try again.
To diagnose on the server machine, ensure that lmgrd.exe can run from a command line on the server machine.
The SCW, or StressCheck® Work File, is a binary file containing ONLY StressCheck® model information (e.g. geometry, mesh, boundary conditions, etc.) It does not contain any information about a saved session, or any solution information for post-processing. It is written by clicking File>Export (or the Export icon), and selecting the SCW type (default). It is opened by clicking File>Open, and selecting the SCW file.
The SCP, or StressCheck® Project File, is a binary file containing ALL session information, including model information, log files, solution information, etc. If you were working in a previously unsaved StressCheck® session, and clicked File>Save (or the diskette icon), you would be able to save the entire session to a new SCP file. If you were working in a previously saved StressCheck® session, and clicked File>Save, the previous SCP would be overwritten with the updated session. The SCP file will not be modified unless you click File>Save” to overwrite it. File>Save As may be used to write a NEW SCP file, independent of the previous SCP file.
Element distortion refers to the difference between the shape of a standard element (a perfect cube in the case of a hexahedral element for example) and the shape of the element in the mesh. The transformation between the standard element and the element in the mesh is called mapping. Mapping functions (Q) establish a relationship between the global coordinates of the element in the mesh (x,y,z) and the local coordinates of the standard element (ξ,η,ζ). For example, the image below shows how a tetrahedral element in the mesh (right) is mapped from its standard shape (left):
In traditional FEA packages, where low order (e.g. p=1 or 2) approximations of the displacement field are used, the effect of distorted elements (or more specifically, distorted tetrahedral elements) is discouraged due to the negative impact of distortion on the quality of the finite element solution. Lower-order implementations are very susceptible to element distortions; as a result, meshes are typically dense.
The implementation of the finite element method in StressCheck® is less susceptible to element distortion because the mapping functions and the functions used for the approximation of the displacements are independent. In practice, the elements implemented in StressCheck® can be reasonably distorted (e.g. vertex angles in the range 5 ≤ θ ≤ 175, where θ is the solid angle at each corner of the element) and still be acceptable for detailed analysis. Thus, observation of element distortion should not automatically prompt the user to refine the mesh.
However, if the element is too distorted, such as a tetrahedral element with a vanishing angle, the mapping function Q becomes ill-conditioned and its negative impact on solution quality is only partially compensated by the higher polynomial order of approximation of the displacement functions. Since it is not possible to know a-priori the effect of distortion on the quality of the approximation, StressCheck® provides for the extraction of every engineering quantity of interest as a function of the number of degrees of freedom (DOF) to check for convergence. Please refer to this article for the recommended quality control procedures in StressCheck®.
The examples below demonstrate how auto-meshed tetrahedral elements with reasonable distortion may be used in StressCheck® to perform 3D high-quality detailed stress analyses:
StressCheck Demo: 3D Tie Rod Stress Concentration Factor Study
Note: In 3D the use of hexahedral or pentahedral elements is always preferred over tetrahedral elements, as they are more computationally efficient. A user may hand-mesh any part (or automatically mesh constant thickness parts by extrusion) with geometrically mapped hexahedral or pentahedral elements if auto-mesh generated tetrahedral elements are too distorted and convergence checks following p-extension indicate that the errors are larger than what is considered acceptable. For reference, many 3D examples in the StressCheck® Handbook folders (Edit > Handbook, click Activate Open File Dialogue to browse the Handbook library) are hand-meshed with hexahedral and/or pentahedral elements.
For complex geometry where the use of automatic mesh generation is the only practical option, local mesh refinement must be used to eliminate highly distorted elements in the regions of primary interest.
Although one can connect elements of different references, elements of different references should not be connected (e.g. 1D Beams—>2D Quads or 2D Tris—>3D Pentas) without extreme care. This is because the theory of elasticity requires consistent references (1D, 2D, or 3D) for exact elasticity solutions.
Allowing elements of different references to be connected (via nodes, edges or faces) introduces modeling errors. This practice is known as finite element modeling.
From Dr. Barna Szabó:
Practitioners of finite element modeling are called upon to balance two very large errors: (a) Errors stemming from fact that the finite element model may not have an underlying exact solution and/or the elements chosen my not satisfy the requirements of stability and consistency, and (b) errors of discretization (not enough DOF to represent the exact solution).
In contrast, numerical simulation has a solid scientific foundation. There has to be an underlying exact solution, the finite elements must satisfy the conditions of stability and consistency, and the errors of discretization must be shown to be small. This is why we say that democratization has to be based on numerical simulation, not finite element modelling.
In StressCheck®, mixed mapping refers to the situation where isoparametric/quadratic and geometric (high order blended function) element mappings coexist within the same mesh. Mixed element mapping will result when the user attempts to convert the element mapping from quadratic to geometric via Tools > Convert Element Mapping (All Elements) or Convert Element Mapping (Active Contact Zones Only), and one or more elements fails to convert.
In these cases, neighboring elements of the elements failing to convert will have geometric mapping functions across the shared element boundaries, while the elements failing to convert will remain quadratically mapped.
When one or more elements fails to convert mapping to geometric, an element set _CONVERT_FAIL is automatically generated. This set may then be selected/isolated, and the user can determine where these elements are located in the mesh.
How Do I Use Sets to Select and View Groups of Elements and Other Objects?
In some cases, refining the mesh in these regions may be necessary if the elements failing to convert are in regions of primary interest. Then, the user may attempt to convert the mapping of the refined elements to geometric.
Meshes with mixed element mappings can still be solved; they should cause no errors or prevent the solution from running. In fact, when StressCheck® computes multi-body contact solutions, the solver automatically attempts to convert the mapping of elements in active contact zones to geometric, while leaving all other element mappings quadratic.
Care should be taken on the upper limit of the p-level used for solving meshes with mixed mapping however.
Why Is There a Recommended Maximum P-level for Isopar Automeshes?
When mixed mapping is present, there is a possibility that the issues arising from poor geometry approximation may be exacerbated at the transition between geometric and quadratic mapping (i.e. element boundaries). Elements that fail conversion are often located in areas of interest; the reason for failing is that the mapped element becomes too distorted during conversion.
If extractions are to be performed away from the areas where these elements are present then it is OK to use a mixed mesh and extend the p-level all the way to p = 8. As previously mentioned, after conversion you may receive a warning if some elements cannot be converted to geometric mapping and a set (_CONVERT_FAIL) should be available to verify that those elements are outside the areas of interest.
Helpful Hints and Tips: Case Study on the Influence of Element Mapping and p-Discretization
When compared to a single-part analysis, a multi-body contact analysis of the same degrees of freedom (DOF) is more computationally demanding. This increase in computational time is due to the nonlinear iterations required to accurately depict high-gradient pressure distributions between complex parts, especially if the part count (and by extension the number of contact pairs) is high.
If a multi-body contact analysis is required, there are several approaches to potentially reduce solver time without significantly affecting the data of interest or the load transfer.
Although it is not required to have a 1:1 correspondence of element faces between contacting parts (our contact algorithm is very flexible regarding mesh differences between contacting parts), it is more efficient to have close to a 1:1 correspondence between surface areas in contact.
An additional benefit is that splitting surfaces for contact assignment improves local mesh refinement. This is discussed in the following document:
Helpful Hints and Tips: Optimization of a Simple 3D Contact Problem
By default, multi-body contact solutions require 40 iterations to complete, unless the maximum contact pressure error is <0.01%. In order to change the number of iterations, create the parameter _contact_iter and set this to a value between 10 and 100.
Note: it is not recommended to reduce the number of iterations unless the maximum contact pressure error is <2% and load transfer is >95% at lower iterations.
Note: for some classes of problems, reducing the number of iterations may not affect the multi-body contact solution. For example, close contact problems with very high loads in which the pressure distributions are very smooth. However, some classes of problems may require a high number of iterations in order to properly represent the contact pressures (and therefore load transfer) between parts.
If a part or parts in multi-body contact are relatively simple to manually (hand) mesh, such as simple fasteners and bushings, this approach can potentially reduce computational time:
For example, if a fastener body is to be repeated in multiple holes, it is possible to create/import the solid representing the fastener body, create the mesh/contact zones for the fastener body, and then copy this fastener body to multiple holes.
The below document provides several case studies on reducing the number of contact iterations and hand-meshing/copying fasteners:
Helpful Hints and Tips: Improving Multi-Body Contact Efficiency
In cases where higher accuracy of load transfer and pressures is required, it is useful to increase the polynomial level of the surfaces in contact independently from the rest of the model via p-Discretization. For example, surfaces in contact may be fixed at p=6 (or higher) while all other elements outside the selected surfaces are solved at p=5 (or lower).
Additionally, boundary layer refinement around bores in multi-body contact aids in the full conversion to geometric mapping and allows the polynomial order of the elements in contact to be increased to p=8. This strategy is demonstrated below:
StressCheck Tutorial: Boundary Layer Refinement and P-Discretization in Contact Regions
The model in the tutorial was solved for three cases:
Case 1: Default automesh, 40 contact iterations
Case 2: Boundary layer refinement in holes of middle plate, p-Discretization of p=8 in holes of middle plate, 20 contact iterations
Case 3: Default automesh, p-Discretization of p=8 in holes of middle plate, 20 contact iterations
The time savings between Case 1 and Case 2 was about 19%, and Case 1 and Case 3 about 42%, without significantly affecting the results in the middle plate holes:
Case 1: Default automesh, 40 contact iterations
Case 2: Boundary layer refinement in holes of middle plate, p-Discretization of p=8 in holes of middle plate, 20 contact iterations
Case 3: Default automesh, p-Discretization of p=8 in holes of middle plate, 20 contact iterations
Additionally, general improvements to solution efficiency without significantly affecting the data of interest are described in the below helpful hints document:
Much of the current StressCheck® Professional code base is single-threaded, though over time ESRD is working to improve as much of the code as possible to be multi-threaded. Therefore, the current release of StressCheck® is more multi-threaded than previous releases, and less multi-threaded than it will be in future releases.
Additionally, the current release of StressCheck® does not yet support parallel/multi-core processing.
Regarding the number of cores, the Windows OS handles all the core allocation for StressCheck® processes. It is possible to set processor affinity for specific applications such as StressCheck®: https://www.tekrevue.com/tip/restrict-apps-cpu-cores-processor-affinity/
Some StressCheck® users may experience decreased model performance and/or integrity in certain situations. The following may be considered as potential reasons:
Is your “Path to Scratch Directory” option pointing locally or to a network location? If your scratch directory (i.e. a temporary directory automatically created for StressCheck®-related session files and solution data) is pointing to a network path, this can drastically slow down model loading and solution times, and even affect the data integrity.
Check the scratch directory path by opening StressCheck® and clicking File > Options, and verify that the “Path to Scratch Directory” value is pointing locally (e.g. C:\Users\…..\AppData\Local\Temp”, which is the default), and is not pointing to a restricted area on your machine (e.g. C:\Program Files\ESRD\StressCheck® PE 10).
It is recommended to keep the scratch directory path relatively clean of old/discarded StressCheck® temporary directories (dataset folders) as well as other non-essential temporary folders to reduce overhead. These temporary directories have the format of “dataset-xxxxxxxxx-xxxxxxxx”, with these directories (and their contents) normally removed after StressCheck® sessions are closed by the user.
However, if StressCheck® sessions exit abnormally, these directories may remain. It is recommended to remove any unwanted temporary directories in the scratch directory path.
It is recommended to read (open) and write (save) StressCheck® files locally instead of relying on a network connection. This ensures the packing/unpacking procedure is allowed to complete uninterrupted. Additionally, the local directory should not be a system or restricted folder requiring additional write permissions.
Opening a StressCheck® file from a mail client, such as Outlook, may cause unpacking issues. Therefore, StressCheck® file attachments should be saved locally before opening.
Adhering to these recommendations will ensure that the integrity of StressCheck®‘s files are preserved, and there are no interruptions during the file read/write that can cause corruptions.
Ideally, the scratch directory path would have plenty of disk space (i.e. > 1 GB), otherwise larger models will almost certainly load and solve at slower rates due to limited disk availability.
Also, there is much file I/O occurring during model loading and solving (especially during stiffness matrix inversion), which can be further limited by your hard drive’s read/write speed.
Ideally, the hard drive would be a steady state hard drive (SSD) or at least a high-RPM mechanical hard disk drive (HDD). If your hard drive is quite fragmented, near capacity and/or has a slower RPM mechanical hard disk drive, model performance may be slowed.
Learn more about ESRD Software Product System Requirements.
Are you running a node-locked or floating license configuration? For node-locked (i.e. machine-specific) licenses, firewalls/network security slowdowns typically are not an issue as there is no network communication involved.
However, for floating (server-based) licenses, communication through a restrictive network firewall to/from the license server’s FLEXnet license manager can cause significant slowdowns in model performance. This communication is required in order to obtain/return StressCheck® license features and check the license server’s “heartbeat” (i.e. determine if the license server is still capable of serving licenses).
Check with your system administrator to determine if your StressCheck® license server communicates through a higher-security, firewall-enabled network that limits/prevents network communication or specific ports. If this is the case, slowdowns may also be observed for other software products which need to communicate with the same license server.
Ideally, the graphics processor for StressCheck® (and all ESRD products) should be set to your video card’s high-performance graphics processing unit (GPU) instead of integrated graphics (e.g. Intel on-board). Some examples of video card manufacturers include Nvidia and AMD, which design dedicated GPU’s for high-performance graphics acceleration and rendering.
For additional information on ensuring you are using your video card’s high-performance GPU (such as Nvidia or AMD) for StressCheck®: https://www.techadvisor.co.uk/how-to/pc-components/how-set-default-graphics-card-3612668/
Has processor affinity (i.e. the specific declaration of CPU cores to be used for an application) been set? Learn more: https://www.tekrevue.com/tip/restrict-apps-cpu-cores-processor-affinity/
The following error may occur when performing a solution:
This error typically occurs because of the following correctible issues.
If constraints do not prevent all free rigid body motion (zero-strain translations and/or rotations) then the model is underconstrained and cannot be solved.
If the model is self-equilibrated under the applied loads, then the Rigid Body Constraint method may be used (Constraint tab, Select > Node/Point > Rigid Body). No other constraints are required.
Note: For the model to be self-equlibrated under the applied loads, the load check must return relatively small forces/moments (Load tab, Check > All Elements > Selection, enter the load ID and click Accept). The computed total forces/moments may not be exactly zero due to accumulation of small numerical errors.
If the model is NOT self-equilibrated under the applied loads, other constraints must exist that provide reactions for the applied loads (e.g. Symmetry Constraints or Spring Constraints). Such constraints may be used to provide a reaction and prevent rigid body motion in one direction or in all directions. For example, if the applied resultant forces are in equilibrium in the X direction but not the Y direction, then an additional constraint should only be used to react the Y direction force; one or more Nodal Constraints may be used to prevent any remaining rigid by motions (Constraint tab, Select > Node/Point > Node).
Note: in 2D/3D elasticity, nodes and points should NOT react load! They should only be used to prevent rigid body motion, thus setting a reference for the elasticity solution. Weak normal or tangent springs may also be used if a node or point location is not appropriate for cancellation of rigid body motion.
The solution can also fail due to rejected elements, where the volume/Jacobian of the elements cannot be computed and therefore the stiffness is negative or zero. To check the Properties of the mesh:
The distortion of the mesh should not cause the solver to fail.
It can occur that elements are not fully connected at the vertices/nodes, and the mesh is discontinuous.
Typically, disconnected elements occur because of free edges or faces and redundant nodes.
Note: If there is still an issue, it is likely that there is an associativity problem between neighboring nodes on one or more element edges/faces and geometry. This can occur when one node on an element edge or face belongs to one curve or surface, and the other node on the same element edge or face belongs to another curve or surface. Contact ESRD Support if this appears to be the case.
Typically, disconnected elements occur because of a problem with the automesher mesh generation. It is best in this case to delete all elements (Mesh tab, Select > All Elements, click Delete) and all nodes (Mesh tab, Select > Node, marquee select all nodes, click Delete), and export an SCW file. Then, open this SCW file and automesh again.
The term “p-extension” refers to the process of systematically and hierarchically increasing the polynomial order of the element shape functions on a fixed mesh, thereby increasing simulation degrees of freedom (DOF) and representing more complex displacement fields for the mesh. Conversely, traditional FEA implementations require successive mesh refinements on elements of fixed polynomials, or h-extensions, to demonstrate convergence.
While both p-extensions and h-extensions are capable of converging to the exact solution of a mathematical problem solved by the finite element method, p-extensions will achieve convergence with less computational burden. Via the Wikipedia article on p-FEM:
The theoretical foundations of the p-version were established in a paper published Babuška, Szabó and Katz in 1981[2] where it was shown that for a large class of problems the asymptotic rate of convergence of the p-version in energy norm is at least twice that of the h-version, assuming that quasi-uniform meshes are used. Additional computational results and evidence of faster convergence of the p-version were presented by Babuška and Szabó in 1982.[3]
StressCheck® is capable of automatically performing p-extensions during the linear solution process, as shown in the below Linear solver tab:
P-extension from p=2 to 8
Note: it is not always necessary to solve to the maximum polynomial level, p=8. Convergence may be assessed after three (3) hierarchically increasing polynomial levels to determine if continuing the p-extension is necessary to achieve the desired solution quality.
This feature makes it very easy to verify convergence of any data of interest, as all runs of increasing DOF are stored for live dynamic extractions of results:
Key Quality Check #4: Peak Stress Convergence (courtesy StressCheck® Professional)
For more information and examples of p-extensions in practice:
Before we explore the answer to this question, let’s review a two important definitions regarding finite element method implementations:
Traditional isoparametric elements with (b) having quadratic mapping functions
There are several historical reasons for using isoparametric mapping in conventional implementations of the finite element method (FEM), such as the advantage of decoupling the solid model from the mesh. However, advancements in computational power have not only rendered these perceived advantages obsolete, but also shown significant drawbacks of the nodal based approach.
In StressCheck®, however, the order of the element mapping is independent from the polynomial order of the shape functions used for approximating the finite element solution, or solver p-levels. The main reason to separate the two is to allow for the computation of a hierarchic set of solutions used to compute error estimates and convergence on the data of interest.
In this context, the term geometric mapping is used to indicate that the mapping of the elements is performed using high-order polynomials with optimal collocation points so that the representation of the geometry is almost exact. This mapping allows for large, highly-curved elements with high aspect ratios, and is especially useful as fewer elements are needed to obtain a good approximation of the solution when using high-order functions (i.e. p-extension).
Therefore, StressCheck® allows two types of element mapping: second-order, nodal-based (midside-noded elements) referred to as “Quadratic”, and high-order polynomials with optimal collocation points referred to as “Geometric”. Note: in StressCheck® v10.5 and previous releases, “Quadratic” was labeled as “Isoparametric” or “Isopar”.
Geometrically Mapped Elements Generated by Connecting Nodes.
Quadratically Mapped Elements Generated by the MeshSim Automesher.
Geometric mapping is the default for manually generated elements (i.e. elements created by specifying nodes), while Quadratic mapping is the default for automeshed tetrahedral elements. Note: in StressCheck® v10.5 and previous releases, when using the automesher (Create > Mesh > Auto) there is a checkbox option next to the “Isopar” field activated by default. The main reason for the default being set on “Isopar” is that the automesher implemented in StressCheck® is most successful in generating midside-noded, quadratically mapped tetrahedral elements.
However, the mapping for automeshed elements can be converted to geometric (Tools > Convert Element Mapping > Geometric…) provided some conditions are satisfied (e.g., the resulting mapping not having either crossover edges/faces or vanishing (zero) vertex angles). As the automesher optimizes filling the volume with as few elements as possible, the resulting mesh may have some elements with high aspect ratios and/or vanishing angles for which converting the mesh to geometric mapping is not always successful.
Note: for sufficiently refined automeshes, the difference between quadratic and geometric mapping is typically not significant.
Why Is There a Recommended Maximum P-level for Quadratically Mapped Automeshes?
In general, it is recommended not to exceed a polynomial level (p-level) of p = 5 when solving a model with quadratically mapped (i.e. “Isoparametric” or “Isopar”) elements.
The reason for the recommended maximum p-level relates to how well the elements approximate the geometry via mapping functions. For quadratically mapped elements, the approximation is performed using second-order shape functions, while geometrically mapped elements use high-order regional collocation points. For complex geometric details, such as 3D blends, holes and fillets, more quadratically mapped elements may be required for a sufficient geometric approximation.
Why Is There a Recommended Maximum P-level for Quadratically Mapped Automeshes?
In some cases, the automesher may generate too few elements to accurately approximate those geometric details. When that is the case, solving at p-levels that exceed the polynomial level of the mapping may cause mathematical artifacts (in the form of spurious strains/stresses) to appear as the solution will be capturing discontinuities across the elements that are not present in the actual geometry. These artifacts simply mean a lower maximum p-level should be used so the order of the solution approximation is closer to the mapping approximation.
There are several ways to check the quality of the geometric approximation of quadratically mapped elements:
Yes. This conversion step can be done by going to Tools (in the main menu bar at the top of the StressCheck® window) > Convert Element Mapping > Geometric (All Elements). In StressCheck® v10.5 and earlier, quadratically mapped elements were referred to as “Isoparametric” or “Isopar”.
Note: the latter is not recommended for more complex geometries as often the mesh conversion fails for some elements during mesh generation. Also note that the mapping conversion for all elements is a computationally intensive operation.
Converting the entire mesh to geometric mapping will allow for unrestricted p-level extension (i.e. all the way to p = 8). However, as mentioned previously, converting the entire mesh is not always successful. Elements that are too distorted may not be converted to geometric mapping, indicated by a warning message after converting the mesh. An element set will be automatically generated with the elements that failed conversion (_CONVERT_FAIL).
Remark: While the mesh can be modified to eliminate elements that fail to convert, the resulting mesh tends to be good enough to generate good results at low p-levels. Note that when some elements fail to convert to geometric, the model can still be solved with mixed mapping. In fact, when StressCheck® computes multi-body contact solutions, the solver automatically attempts to convert the mapping of elements in active contact zones to geometric. Elsewhere the mapping remains quadratic.
Not necessarily. When mixed mapping is present, there is a possibility that the issues arising from poor geometry approximation may be exacerbated at the transition between geometric and quadratic mapping (i.e. element boundaries). Elements that fail conversion are often located in areas of interest; the reason for failing is that the mapped element becomes too distorted.
If extractions are to be performed away from the areas where these elements are present then it is OK to use a mixed mesh and extend the p-level all the way to p = 8. As previously mentioned, after conversion you may receive a warning if some elements cannot be converted to geometric mapping and a set (_CONVERT_FAIL) should be available to verify that those elements are outside the areas of interest.
Contact ESRD Support if you require additional assistance with the above guidelines.
Element distortion refers to the difference between the shape of a standard element (a perfect cube in the case of a hexahedral element for example) and the shape of the element in the mesh. The transformation between the standard element and the element in the mesh is called mapping. Mapping functions (Q) establish a relationship between the global coordinates of the element in the mesh (x,y,z) and the local coordinates of the standard element (ξ,η,ζ). For example, the image below shows how a tetrahedral element in the mesh (right) is mapped from its standard shape (left):
In traditional FEA packages, where low order (e.g. p=1 or 2) approximations of the displacement field are used, the effect of distorted elements (or more specifically, distorted tetrahedral elements) is discouraged due to the negative impact of distortion on the quality of the finite element solution. Lower-order implementations are very susceptible to element distortions; as a result, meshes are typically dense.
The implementation of the finite element method in StressCheck® is less susceptible to element distortion because the mapping functions and the functions used for the approximation of the displacements are independent. In practice, the elements implemented in StressCheck® can be reasonably distorted (e.g. vertex angles in the range 5 ≤ θ ≤ 175, where θ is the solid angle at each corner of the element) and still be acceptable for detailed analysis. Thus, observation of element distortion should not automatically prompt the user to refine the mesh.
However, if the element is too distorted, such as a tetrahedral element with a vanishing angle, the mapping function Q becomes ill-conditioned and its negative impact on solution quality is only partially compensated by the higher polynomial order of approximation of the displacement functions. Since it is not possible to know a-priori the effect of distortion on the quality of the approximation, StressCheck® provides for the extraction of every engineering quantity of interest as a function of the number of degrees of freedom (DOF) to check for convergence. Please refer to this article for the recommended quality control procedures in StressCheck®.
The examples below demonstrate how auto-meshed tetrahedral elements with reasonable distortion may be used in StressCheck® to perform 3D high-quality detailed stress analyses:
StressCheck Demo: 3D Tie Rod Stress Concentration Factor Study
Note: In 3D the use of hexahedral or pentahedral elements is always preferred over tetrahedral elements, as they are more computationally efficient. A user may hand-mesh any part (or automatically mesh constant thickness parts by extrusion) with geometrically mapped hexahedral or pentahedral elements if auto-mesh generated tetrahedral elements are too distorted and convergence checks following p-extension indicate that the errors are larger than what is considered acceptable. For reference, many 3D examples in the StressCheck® Handbook folders (Edit > Handbook, click Activate Open File Dialogue to browse the Handbook library) are hand-meshed with hexahedral and/or pentahedral elements.
For complex geometry where the use of automatic mesh generation is the only practical option, local mesh refinement must be used to eliminate highly distorted elements in the regions of primary interest.
To delete all automesh objects (elements and nodes) without deleting a global automesh record:
All elements and nodes associated with automeshes will be removed, but all mesh records preserved.
Note: deleting a global mesh record will delete only the elements and nodes associated with the global mesh record.
When compared to a single-part analysis, a multi-body contact analysis of the same degrees of freedom (DOF) is more computationally demanding. This increase in computational time is due to the nonlinear iterations required to accurately depict high-gradient pressure distributions between complex parts, especially if the part count (and by extension the number of contact pairs) is high.
If a multi-body contact analysis is required, there are several approaches to potentially reduce solver time without significantly affecting the data of interest or the load transfer.
Although it is not required to have a 1:1 correspondence of element faces between contacting parts (our contact algorithm is very flexible regarding mesh differences between contacting parts), it is more efficient to have close to a 1:1 correspondence between surface areas in contact.
An additional benefit is that splitting surfaces for contact assignment improves local mesh refinement. This is discussed in the following document:
Helpful Hints and Tips: Optimization of a Simple 3D Contact Problem
By default, multi-body contact solutions require 40 iterations to complete, unless the maximum contact pressure error is <0.01%. In order to change the number of iterations, create the parameter _contact_iter and set this to a value between 10 and 100.
Note: it is not recommended to reduce the number of iterations unless the maximum contact pressure error is <2% and load transfer is >95% at lower iterations.
Note: for some classes of problems, reducing the number of iterations may not affect the multi-body contact solution. For example, close contact problems with very high loads in which the pressure distributions are very smooth. However, some classes of problems may require a high number of iterations in order to properly represent the contact pressures (and therefore load transfer) between parts.
If a part or parts in multi-body contact are relatively simple to manually (hand) mesh, such as simple fasteners and bushings, this approach can potentially reduce computational time:
For example, if a fastener body is to be repeated in multiple holes, it is possible to create/import the solid representing the fastener body, create the mesh/contact zones for the fastener body, and then copy this fastener body to multiple holes.
The below document provides several case studies on reducing the number of contact iterations and hand-meshing/copying fasteners:
Helpful Hints and Tips: Improving Multi-Body Contact Efficiency
In cases where higher accuracy of load transfer and pressures is required, it is useful to increase the polynomial level of the surfaces in contact independently from the rest of the model via p-Discretization. For example, surfaces in contact may be fixed at p=6 (or higher) while all other elements outside the selected surfaces are solved at p=5 (or lower).
Additionally, boundary layer refinement around bores in multi-body contact aids in the full conversion to geometric mapping and allows the polynomial order of the elements in contact to be increased to p=8. This strategy is demonstrated below:
StressCheck Tutorial: Boundary Layer Refinement and P-Discretization in Contact Regions
The model in the tutorial was solved for three cases:
Case 1: Default automesh, 40 contact iterations
Case 2: Boundary layer refinement in holes of middle plate, p-Discretization of p=8 in holes of middle plate, 20 contact iterations
Case 3: Default automesh, p-Discretization of p=8 in holes of middle plate, 20 contact iterations
The time savings between Case 1 and Case 2 was about 19%, and Case 1 and Case 3 about 42%, without significantly affecting the results in the middle plate holes:
Case 1: Default automesh, 40 contact iterations
Case 2: Boundary layer refinement in holes of middle plate, p-Discretization of p=8 in holes of middle plate, 20 contact iterations
Case 3: Default automesh, p-Discretization of p=8 in holes of middle plate, 20 contact iterations
Additionally, general improvements to solution efficiency without significantly affecting the data of interest are described in the below helpful hints document:
In some cases, the MeshSim automesher may fail and provide the following error message (or similar):
“MeshSim meshing error. Problem meshing face 1464095 at (0.059494844831943255, 0.14298635089067505, -0.011241197933960261)”
The face number (in this case 1464095) is an internal designation and will need to be identified by the corresponding location. The location returned is in the Parasolid default (meters), therefore these units may need to be converted to the session units (e.g. inches or millimeters).
Once the location is converted to the session units, it is helpful to create a Locate set to select the problem region based on the location.
Note: if Parasolid’s Workshop .NET product is available, the face number returned by MeshSim may be useful for debugging. Contact ESRD Support for more information.
When CAD files imported into StressCheck® have very small geometric features, these are sometimes suppressed by the MeshSim automesher, resulting in elements not associated with the geometric surfaces in regions near the small features. Some indicators of this are:
1) Loads/constraints applied to geometric surfaces are not applied to element faces
2) MeshSim warnings regarding parasolid error
3) Warning message regarding significant node displacement
To ensure the small geometric features are not suppressed by MeshSim, define the parameter _ms_suppress in the Model Info Parameters tab and set it to less than 5e-5 (relative feature size) and try re-meshing. Conversely, if you wish to suppress small features from automeshing, you may set the value to be the minimum feature size. MeshSim will skip the meshing of these features.
StressCheck®‘s multi-body contact algorithm does not require mesh conformity or a 1:1 ratio between contacting element faces. In fact, the element face distribution across assigned contact zones can be quite dissimilar in most practical applications.
However, the resolution of the gap between bodies for calculation of contact pressures is related to the number of collocation points used for each element face of the assigned contact pairs. By default, 196 points per element face in the contact pair are used for calculating the gap function.
The contact gap/penetration is determined by sampling from one contact zone to the other. If the ratio between the number of element faces in the assigned contact pairs is large (for example, 100:1 meaning that there are about 100 more element faces on one side of the declared contact pair than on the other), then the gap function is evaluated at many more locations on one side than on the other. This means that the approximation of the gap function has a different resolution which can induce oscillation during the nonlinear multi-body contact solution.
Therefore, it is recommended that the ratio between the number of element faces in the assigned contact pairs be kept close to 1:1. Numerical studies have shown that ratios not exceeding 10:1 are in general satisfactory. A comparison between a 100:1 and the more reasonable 10:1 contacting element face ratio is provided below.
In this case, multi-body contact between a lug, clevis and pin was solved. The pin contact zone surrounded 24 element faces, while the number of element faces in the lug was either ~100 or ~10 times more refined. The von Mises stresses were then plotted for the lug in the same stress ranges:
Pin element faces (24 in contact zone)
100:1 ratio between lug and pin contacting element faces
10:1 ratio between lug and pin contacting element faces
As shown in the example, oscillation of the stress distribution is a clear indication that the ratio between contacting element faces is too large.
While regular and similar meshes across contact zones are preferable, there is flexibility to have higher ratios when multiple discretization strategies result in dissimilar element counts across contact pairs. Because the situation is problem-dependent, it is not possible to have a fixed rule to define by how much it can deviate from the optimal 1:1 ratio for all cases, so it is very important to check for oscillations in the results of interest along the contacting surfaces during the post-processing of a contact analysis.
When using Mesh tab, Create > Mesh > Auto, the following parameter inputs are available:
The most important inputs for controlling mesh density and distribution are: Ratio, D/H, MinLen and Trans.
The user may specify the ratio of the length of the longest desired element edge in the body (L1) relative to the longest length of the smallest coordinate aligned bounding box which surrounds the body (L2). MeshSim will attempt to minimize the largest element length based on the given Ratio.
The ratio of the distance from the chord formed by an element edge and its associated curve (D) or surface, relative to the length of the element edge (H). Implies that for low D/H values, MeshSim will attempt to increase refinement along areas of higher curvature. For curvature only, will not affect straight regions.
The minimum length of any element edge in the mesh (in model units). If MinLen is active, MeshSim will attempt to regulate the minimum lengths of the elements. For curvature only, will not affect straight regions.
Rate of transition from small elements in a local region to larger elements in a global region. A value of zero (0.0) indicates that elements will remain small in both the local and global region. A value of one (1.0) indicates that elements will transition rapidly from a small size in the local region to a larger size in the global region. A value in between will indicate a more gradual transition between local and global mesh regions. Default in 2D and 3D is 1.0 and 0.0, respectively.
When the Default option is checked, it will use the latest MeshSim algorithms. When the Legacy option is checked, it will use legacy (pre-v10.3) MeshSim algorithms.
For global mesh records (Create > Mesh > Auto):
Typically, the regions of interest are curved surfaces or lines (holes, fillets, etc.), so modifying ‘D/H’ to control the amount of elements in those areas is usually the first thing to try. In general setting ‘Trans’ to a low value to start (0 – 0.1) and then increasing the value slightly (it’s very sensitive) after an appropriate mesh density is achieved in the areas of interest is a good idea as well. If you notice that ‘D/H’ is not having any affect or the elements stop reducing in size at a certain value of D/H, you likely will need to modify the value of ‘MinLen’ to allow for smaller elements in the mesh. ‘Ratio’ is only used to keep all of the elements in the entire mesh under a certain size. Generally, most uncracked geometries can be meshed adequately using only a single global mesh record.
In cases where refinement for specific features is required:
The most used method for local refinement is Create > Mesh > Curve. Select a surface or curve and set the ‘Ratio’ value to something smaller in order to force all of the elements to stay under a certain size associated to the surface or curve. Again, you may need to adjust the ‘MinLen’ param to allow the elements to be small enough.
The “Size” parameter is typically the absolute size of the elements in the assigned curve/surface after Local automeshing (Create -> Mesh -> Local). The surface type can be checked by using Select -> Any Surface, and clicking on the surface. In the case of splines (B-Splines and P-Splines), the “Size” parameter must be scaled relative to the size of the part itself. Therefore, it will typically be a larger value than the desired element size.
Cylindrical Surface
Spline Surface
Before we explore the answer to this question, let’s review a two important definitions regarding finite element method implementations:
Traditional isoparametric elements with (b) having quadratic mapping functions
There are several historical reasons for using isoparametric mapping in conventional implementations of the finite element method (FEM), such as the advantage of decoupling the solid model from the mesh. However, advancements in computational power have not only rendered these perceived advantages obsolete, but also shown significant drawbacks of the nodal based approach.
In StressCheck®, however, the order of the element mapping is independent from the polynomial order of the shape functions used for approximating the finite element solution, or solver p-levels. The main reason to separate the two is to allow for the computation of a hierarchic set of solutions used to compute error estimates and convergence on the data of interest.
In this context, the term geometric mapping is used to indicate that the mapping of the elements is performed using high-order polynomials with optimal collocation points so that the representation of the geometry is almost exact. This mapping allows for large, highly-curved elements with high aspect ratios, and is especially useful as fewer elements are needed to obtain a good approximation of the solution when using high-order functions (i.e. p-extension).
Therefore, StressCheck® allows two types of element mapping: second-order, nodal-based (midside-noded elements) referred to as “Quadratic”, and high-order polynomials with optimal collocation points referred to as “Geometric”. Note: in StressCheck® v10.5 and previous releases, “Quadratic” was labeled as “Isoparametric” or “Isopar”.
Geometrically Mapped Elements Generated by Connecting Nodes.
Quadratically Mapped Elements Generated by the MeshSim Automesher.
Geometric mapping is the default for manually generated elements (i.e. elements created by specifying nodes), while Quadratic mapping is the default for automeshed tetrahedral elements. Note: in StressCheck® v10.5 and previous releases, when using the automesher (Create > Mesh > Auto) there is a checkbox option next to the “Isopar” field activated by default. The main reason for the default being set on “Isopar” is that the automesher implemented in StressCheck® is most successful in generating midside-noded, quadratically mapped tetrahedral elements.
However, the mapping for automeshed elements can be converted to geometric (Tools > Convert Element Mapping > Geometric…) provided some conditions are satisfied (e.g., the resulting mapping not having either crossover edges/faces or vanishing (zero) vertex angles). As the automesher optimizes filling the volume with as few elements as possible, the resulting mesh may have some elements with high aspect ratios and/or vanishing angles for which converting the mesh to geometric mapping is not always successful.
Note: for sufficiently refined automeshes, the difference between quadratic and geometric mapping is typically not significant.
Why Is There a Recommended Maximum P-level for Quadratically Mapped Automeshes?
When using the “Create” action in the Geometry or Mesh tab, there are multiple ways to create new objects:
It’s possible more than one of these actions was invoked during an object creation. Note: left-click of the mouse in the Model Viewing window will always perform the current action in the “Action” combo (e.g. Create, Select, Edit).
As a best practice, if focus is required in the Model Viewing window, and a new object creation is NOT desired, it is recommended to right-click in the Model Viewing window to establish the window’s focus such that the model can be rotated, zoomed, etc. This can help with reducing duplication of geometry and mesh objects.
If you suspect there are duplicate geometry and/or mesh objects, changing the combos to “Select > Any Object” and clicking on the “Index” subtab within the Geometry or Mesh tabs will display a list of the objects in the model:
If there are duplicates, you may simply select the object(s) from the Index list and click the “Delete” button, or simply click the “DeLast” button if the last object created was a duplicate.
Duplicate nodes may be inadvertently generated when manually meshing models. If nodes are duplicated, clicking the Mesh tab’s “Merge” button will attempt to merge all duplicated nodes within a tolerance of 10e-6 model units. All but one of the coincident nodes are discarded, and all element definitions are updated appropriately. Note: care must be taken for multi-body contact models with coincident parts, as well as fracture models with intentional cracks, as these nodes may be merged.
Alternatively, the user may first determine if any free edges/disconnected elements exist in the mesh by using “Check > Edge > Free Edge” within the Mesh tab. Element edges which are free/not shared will be highlighted in red. Then, the user may select only those nodes which are duplicated (either via marquee selection or Shift + left-click) and click the Merge button to limit the merge to those nodes.
Note: to override the merge tolerance, the user may create the parameter _merge_tol and supply a new merge tolerance.
In general, it is recommended not to exceed a polynomial level (p-level) of p = 5 when solving a model with quadratically mapped (i.e. “Isoparametric” or “Isopar”) elements.
The reason for the recommended maximum p-level relates to how well the elements approximate the geometry via mapping functions. For quadratically mapped elements, the approximation is performed using second-order shape functions, while geometrically mapped elements use high-order regional collocation points. For complex geometric details, such as 3D blends, holes and fillets, more quadratically mapped elements may be required for a sufficient geometric approximation.
Why Is There a Recommended Maximum P-level for Quadratically Mapped Automeshes?
In some cases, the automesher may generate too few elements to accurately approximate those geometric details. When that is the case, solving at p-levels that exceed the polynomial level of the mapping may cause mathematical artifacts (in the form of spurious strains/stresses) to appear as the solution will be capturing discontinuities across the elements that are not present in the actual geometry. These artifacts simply mean a lower maximum p-level should be used so the order of the solution approximation is closer to the mapping approximation.
There are several ways to check the quality of the geometric approximation of quadratically mapped elements:
Yes. This conversion step can be done by going to Tools (in the main menu bar at the top of the StressCheck® window) > Convert Element Mapping > Geometric (All Elements). In StressCheck® v10.5 and earlier, quadratically mapped elements were referred to as “Isoparametric” or “Isopar”.
Note: the latter is not recommended for more complex geometries as often the mesh conversion fails for some elements during mesh generation. Also note that the mapping conversion for all elements is a computationally intensive operation.
Converting the entire mesh to geometric mapping will allow for unrestricted p-level extension (i.e. all the way to p = 8). However, as mentioned previously, converting the entire mesh is not always successful. Elements that are too distorted may not be converted to geometric mapping, indicated by a warning message after converting the mesh. An element set will be automatically generated with the elements that failed conversion (_CONVERT_FAIL).
Remark: While the mesh can be modified to eliminate elements that fail to convert, the resulting mesh tends to be good enough to generate good results at low p-levels. Note that when some elements fail to convert to geometric, the model can still be solved with mixed mapping. In fact, when StressCheck® computes multi-body contact solutions, the solver automatically attempts to convert the mapping of elements in active contact zones to geometric. Elsewhere the mapping remains quadratic.
Not necessarily. When mixed mapping is present, there is a possibility that the issues arising from poor geometry approximation may be exacerbated at the transition between geometric and quadratic mapping (i.e. element boundaries). Elements that fail conversion are often located in areas of interest; the reason for failing is that the mapped element becomes too distorted.
If extractions are to be performed away from the areas where these elements are present then it is OK to use a mixed mesh and extend the p-level all the way to p = 8. As previously mentioned, after conversion you may receive a warning if some elements cannot be converted to geometric mapping and a set (_CONVERT_FAIL) should be available to verify that those elements are outside the areas of interest.
Contact ESRD Support if you require additional assistance with the above guidelines.
With the StressCheck® v10.5 release, there is now a feature to allow assignment of material properties directly to bodies/parts (sheet or solid). This is accomplished by going to the Material tab, Assign subtab, and setting the AOM to Select > Any Body > Selection. Then, select the desired material ID, select a body from the screen or select a body set from the Set: dropdown, and then click Accept.
The Any Body > Selection method assigns the desired material ID to all elements associated with the selected body/part, and supports both manually generated or automatically generated elements. Note: the current functionality supports one body selection per assignment. To assign material properties to a different body, simply repeat the process.
The below tutorial demonstrates an example of using the Any Body > Selection method to assign material properties to one or more bodies:
StressCheck Tutorial: Fitting + I-Beam Multi-Body Contact Analysis
When assigning material properties, boundary conditions or other model attributes, it is best to assign to geometric entities (points, curves or surfaces) instead of mesh entities (nodes, element edges or element faces). This is because StressCheck®‘s elements are designed to inherit assignments to geometry through associativity. Automeshed elements are always geometrically associative, while hand-meshed elements will be geometrically associative if the nodes are explicitly associated to the geometry by the user.
In this way, if a model is remeshed, and the elements are renumbed, any assignment to the geometry will persist and be automatically applied to the new mesh. Here’s an example of using geometric association to persist model assignments:
StressCheck Tutorial: Associative Hand Meshing for a Cylindrical Part
With the StressCheck® v10.5 release, there is now a feature to allow assignment of material properties directly to bodies/parts (sheet or solid). This is accomplished by going to the Material tab, Assign subtab, and setting the AOM to Select > Any Body > Selection. Then, select the desired material ID, select a body from the screen or select a body set from the Set: dropdown, and then click Accept.
The Any Body > Selection method assigns the desired material ID to all elements associated with the selected body/part, and supports both manually generated or automatically generated elements. Note: the current functionality supports one body selection per assignment. To assign material properties to a different body, simply repeat the process.
The below tutorial demonstrates an example of using the Any Body > Selection method to assign material properties to one or more bodies:
StressCheck Tutorial: Fitting + I-Beam Multi-Body Contact Analysis
Sometimes, it may be required to update set lists to reflect changes in the model. This is especially true for sets created by the Locate method, in which the pick coordinates and box tolerance will determine which objects will be added to the set.
When set information must be updated, rather than manually replacing each set it is helpful to define a “dummy” parameter to force all sets and assignments to be updated.
For example, we can create a parameter called “dummy” and enter “1+1” in the Expression field. Click Accept, and all model sets will be automatically updated:
More information on Locate sets:
When assigning material properties, boundary conditions or other model attributes, it is best to assign to geometric entities (points, curves or surfaces) instead of mesh entities (nodes, element edges or element faces). This is because StressCheck®‘s elements are designed to inherit assignments to geometry through associativity. Automeshed elements are always geometrically associative, while hand-meshed elements will be geometrically associative if the nodes are explicitly associated to the geometry by the user.
In this way, if a model is remeshed, and the elements are renumbed, any assignment to the geometry will persist and be automatically applied to the new mesh. Here’s an example of using geometric association to persist model assignments:
StressCheck Tutorial: Associative Hand Meshing for a Cylindrical Part
In order for internal surfaces to be selected when creating Locate Sets (i.e. Sets tab, Select > Any Surface > Locate), the Shaded mode MUST be active (blue box):
In order to better visualize internal surfaces, the Geometry Transparency option may be useful:
Here is an example of using the Locate method to select an internal crack surface:
Additional links:
In StressCheck®, sets are used to group objects by type (such as elements). This is similar to the Group concept used in other FEA. Sets are designed to allow you to view model input and results limited to a group of objects instead of viewing all objects of that particular type.
If a set of elements exists (i.e. Model Input, Sets tab, Select>Any Element, view list items), it can be selected and changed:
To select a set of elements from the model display, hold CTRL+SHIFT and select an element in the set. The Sets Browser will appear, with the Candidate sets including the element set in the list (in this case, the Selection is set to Any Element):
To display only this element set, click Display>Objects… and “Sets” tab. Check the “Elements” box, select the element set from the list, then click “Draw” to display only this element set:
This tutorial video shows how to manipulate element displays using automesh and user-generated sets: https://www.esrd.com/product/stresscheck-tutorial-using-sets/
Using StressCheck® Parts in tandem with Solution Configurations is a great way to analyze different modeling configurations in the same StressCheck® session, as well as quickly switch between different groups of objects for visualization, results extraction and more.
From the StressCheck® Master Guide:
StressCheck® provides a facility for organizing the components of a particular model into parts. Parts are just a generalization of the more familiar set definition in StressCheck®. While the members of a StressCheck® “Set” must be of the same type, for example a set of nodes or a set of boundaries, a “Part” is a set of geometric or mesh objects of any type. This facility is particularly handy when working with complex models where it is helpful to be able to activate and deactivate subsets of the model during construction or during post-processing. Parts are also essential to the definition of Solution Configurations, i.e. the definition of multiple model configurations for the purpose of investigating model variations that are topologically different.
Parts may be defined in the Parts input class. FIGURE 61 illustrates the user interface for the definition of parts. The first step in creating a new part is to make sure the objects of interest are in view in the model display window. Next, select the desired objects using the Select action, combined with an appropriate Object selection. When you are finished selecting objects, you may enter a name to identify this collection of objects as a Part, and supply a description to further identify the part. Click the Accept button when you are ready to save the new part definition. You will see the name appear in the scrolling list in the Parts interface. If you wish to review the definition of any part, simply select the part from the scrolling list. The objects it contains will be highlighted on the screen. You are then free to cancel individual objects that define the part, or to select additional objects that you wish to add to the part. Click the Replace button to change the definition of the selected part, or click Accept to create a new part (after supplying a new part name and description).
Once parts have been defined, they may be chosen for selective viewing using the combo box in the Part Toolbar that may be activated from the View menu in the Main menu bar. The Part combo lists all available parts in the model, and the option All Objects and All Parts. As mentioned earlier, Parts are useful for controlling the contents of the model window during model construction and post-processing. The other use of parts is to define two or more variations of a model that each represent a complete and valid model that may be solved and post-processed by StressCheck®.
After defining the solution configurations, you are ready to solve your model variations. You may choose whether to analyze all available solution configurations, or to select a particular solution configuration for analysis. Since each configuration uses a unique solution ID, you can save solutions for all model variations in the same database if you choose All, and post-process them all without having to solve them separately and post-process them separately. This makes comparison of model variations more convenient.
Demonstrated in the below tutorials and PDF’s are selected tips, tricks and best practices for how StressCheck® Parts may be defined and utilized in collaboration with Solution Configurations:
StressCheck Tutorial: Parts and Solution Configurations Part 1 – Planar + Hand Mesh
StressCheck Tutorial: Parts and Solution Configurations Part 2 – 3D + Automesh
Some key takeaways from the video tutorials:
This allows new objects intended for an existing Part definition to be added without needing to switch to the Parts tab, select the Part name from the list, select the new objects and click Replace.
If it is NOT desired to add new objects to an existing Part definition, then the Parts selector should be set to “All Objects” before creating any new objects. Additionally, if a solid modeling operation is performed on a body referenced in a Part definition, it may be necessary to update the Part definition with the new body number.
The most efficient way to ensure automesh objects are stored with a Part definition is to define the Part using object type “Any Body” before the automesher is executed (see above screen capture and the Part 2 tutorial video). Once the automesher is executed, the automesh objects will automatically be appended to each Part definition with references to the bodies automeshed.
Note: if an automesh already exists before the Part has been defined, defining the Part with “Any Body” and then re-processing the automesh will result in the automesh objects being added to the Part definition.
Selecting the Part in the Parts selector will now display both the body and the automesh objects:
Note: updating parameters associated with a Part’s automeshed body will result in the new automesh objects automatically replacing the current automesh objects in the Part definition.
This tip, along with tips for using Parts to create contact zones, is demonstrated in detail in the below tutorial video:
Any solution may be combined in the StressCheck® Calculator, including those defined in Solution Configurations. Superposition of two solution ID’s is demonstrated in the below tutorial and in the Part 1 tutorial video:
StressCheck Tutorial: Superposition of Axial and Bending Load Case Results
If multiple load and constraint cases need to be solved for the same model definition, multiple solution ID’s may be defined and solved (as shown in Tip #4).
However, if there are significant changes in model configuration, such as comparing a filled hole vs. open hole case, then Solution Configurations are quite powerful.
For more hints and tips on Solution Configurations:
Using the Parts class to organize objects into a single named entity is very useful for pre-processing and post-processing tasks, as well as defining solution configurations and organizing object displays.
However, if a certain Part is selected in the Parts selector at the time of saving a project file (SCP) or workfile (SCW), objects not included in the selected Part will not be displayed when opening the file:
If certain objects are not in the display, first check to ensure each object type is enabled for display.
After opening the file, first check to ensure all object displays are activated (see above). If certain objects are still not displayed, check to see if a Part name is selected in the Parts selector. The Part may have only certain model objects in it, meaning all other model objects will not be displayed.
To remedy the display of all model objects, change the Parts selector from the current Part name to “All Objects”:
All model entities (i.e. elements, nodes, systems, points, etc.) should be displayed. It is then recommended to save the project or workfile in this configuration:
For more information about Parts:
Although one can connect elements of different references, elements of different references should not be connected (e.g. 1D Beams—>2D Quads or 2D Tris—>3D Pentas) without extreme care. This is because the theory of elasticity requires consistent references (1D, 2D, or 3D) for exact elasticity solutions.
Allowing elements of different references to be connected (via nodes, edges or faces) introduces modeling errors. This practice is known as finite element modeling.
From Dr. Barna Szabó:
Practitioners of finite element modeling are called upon to balance two very large errors: (a) Errors stemming from fact that the finite element model may not have an underlying exact solution and/or the elements chosen my not satisfy the requirements of stability and consistency, and (b) errors of discretization (not enough DOF to represent the exact solution).
In contrast, numerical simulation has a solid scientific foundation. There has to be an underlying exact solution, the finite elements must satisfy the conditions of stability and consistency, and the errors of discretization must be shown to be small. This is why we say that democratization has to be based on numerical simulation, not finite element modelling.
The display format is in C-language, which is described in detail here: https://www.cprogramming.com/tutorial/printf-format-strings.html
The model input display format can be accessed and changed by clicking Display>View Controls… and entering a new “Display Format:”
This will be reflected in all model inputs, including parameter values. By default, the display format is %11.4e, which means a maximum length of 11 digits, a precision of 4 digits and exponential. For example, 11,012.54 will be displayed as “1.1013e5”. To change to a floating format, simply replace the “e” with an “f”.
Once a SCW or SCP file is saved, it will store your current model input display format. To change the default model input display format so that every StressCheck® session uses the same display format, click File>Options and modify the Display Format (below). Then click “OK”:
The differences in model input between the default (%11.4e, top) format and floating point (%11.4f, bottom) are below:
The model results display format can be modified by changing the “Format:” input in the Results tabs. The formatting is the same as described above.
The differences in model results between the default (%11.3e) format and floating point (%11.3f) are below:
If all that is required is to simulate load transfer to holes or a compression-only reaction is desired for a single part, thereby reducing the computational time required for multi-body contact, normal springs combined with a Material or General Nonlinear analysis may be an efficient workflow.
This “poor man’s contact” form of load transfer may be represented by applying normal springs to element edges or curves (Planar) or element faces or surfaces (3D), applying the appropriate material properties, and solving a Material or General Nonlinear analysis:
Normal springs may be assigned in the Constraints tab, using Select > Edge/Edge Curve/Any Curve (Planar) or Face/Face Surface/Any Surface (3D) > Spring Coeff:
Note: if a fastener radial spring stiffness is to be approximated, the below expression for Kr may be useful:
Because a General Nonlinear analysis is more computationally expensive than a Material Nonlinear analysis, it may be desired to run an analysis that simulates normal springs in compression only. This requires specifying a nonlinear material with either a representative yield stress (if the linear von Mises stresses call for plasticity to be computed), or an artificial yield stress to force the material nonlinear algorithm to release normal springs that are in tension but otherwise keep the stresses linear.
After a linear analysis is available, a Material Nonlinear analysis (Type: Material (NL Mat)) may be executed to compute plasticity + represent compression-only normal springs, or simply represent compression-only normal springs.
If a General Nonlinear analysis is required to account for large displacements and/or rotations, and equilibrium must be solved in the deformed configuration, then the solution should use Type: General (NL Gen). Normal springs will always be compression-only in a General Nonlinear analysis.
Note: nonlinear material properties do not need to be assigned to any elements in order to solve a General Nonlinear analysis.
For more information:
To display the object numbering for visualization, click on Display>Objects…
Then, check the Label box next to the object numbering you want to visualize and click “Draw”. For example, if we want to visualize the element numbers, we can check the Label box next to “Elements” and click “Draw”:
To turn of the element numbering, uncheck the Label box and click “Draw”.
In StressCheck®, sets are used to group objects by type (such as elements). This is similar to the Group concept used in other FEA. Sets are designed to allow you to view model input and results limited to a group of objects instead of viewing all objects of that particular type.
If a set of elements exists (i.e. Model Input, Sets tab, Select>Any Element, view list items), it can be selected and changed:
To select a set of elements from the model display, hold CTRL+SHIFT and select an element in the set. The Sets Browser will appear, with the Candidate sets including the element set in the list (in this case, the Selection is set to Any Element):
To display only this element set, click Display>Objects… and “Sets” tab. Check the “Elements” box, select the element set from the list, then click “Draw” to display only this element set:
This tutorial video shows how to manipulate element displays using automesh and user-generated sets: https://www.esrd.com/product/stresscheck-tutorial-using-sets/
The Edit Toolbar enables object selection/cancellation (by object type or any object), inverting object selections, object blanking (by object type or any object), object unblanking (by object type or any object), visualizing blanked objects, picking only visible objects and more.
This tutorial video explains the details of the Edit Toolbar:
StressCheck® Tutorial: Using Edit Toolbar for Object Selection and Visualization
StressCheck® Professional can natively import Parasolid files of versions ranging 10.0 to 26.0.151. Other CAD formats require licensed translators.
The following CAD formats and versions are supported in StressCheck® Professional’s 3D Interop R25 translators:
| Formats | File Formats | Product Structure | Graphical | Geometry | Semantic PMI |
| 3DXML | .3dxml | v4.3 | v4.3 | ||
| ACIS | .sat, .sab, .asat, .asab | R1 – R24 | R1 – R24 (Generated from Geometry) | R1 – R24 | N/A |
| CATIA V4 | .model, .exp, .session | 4.1.9 – 4.2.4 | 4.1.9 – 4.2.4 (Generated from Geometry) | 4.1.9 – 4.2.4 | N/A |
| CATIA V5 | .CATPart, .CATProduct, .CGR | R8 – R24 (V5–6R2014) | R8 – R24 (V5–6R2014) | R8 – R24 (V5–6R2014) | R8 – R24 (V5–6R2014) |
| CATIA V6 | .CATPart, .CATProduct, .CGR | V6R2014x | V6R2014x | V6R2014x | V6R2014x |
| DXF/DWG | .dxf, .dwg | 2.5 – 2014 | 2.5 – 2014 (Generated from Geometry) | 2.5 – 2014 | N/A |
| IGES | .igs, .iges | Up to 5.3 | Up to 5.3 (Generated from Geometry) | Up to 5.3 | N/A |
| Inventor | .ipt (V6 – V2015) .iam (V11 – V2015) |
V11 – 2015 | V7 – 2015 V6 (Generated from Geometry) |
V6 – V2015 | N/A |
| NX | .prt | 11 – NX 9 | NX 6 – NX 9 11 – NX 5 (Generated from Geometry) |
11 – NX 9 | 11 – NX 9 |
| NX Direct | .prt | NX 1 – NX 9 | NX 6 – NX 9 NX 1 – NX 5 (Generated from Geometry) |
NX 1 – NX 9 | NX 1 – NX 9 |
| Parasolid | .x_t, .xmt_txt, .x_b, .xmt_bin | 10.0 – 26.0.151 | 10.0 – 26.0.151 (Generated from Geometry) | 10.0 – 26.0.151 | N/A |
| Parasolid Direct | .x_t, .xmt_txt, .x_b, .xmt_bin | 10.0 – 26.01.211 | 10.0 – 26.01.211 (Generated from Geometry) | 10.0 – 26.01.211 | N/A |
| Pro/E / Creo | .prt, .prt.*, .asm, .asm.* | 16 – Creo 2.0 | WF3 – Creo 2.0 16 – WF2 (Generated from Geometry) |
16 – Creo 2.0 | 16 – Creo 2.0 |
| Solid Edge | .par, .asm, .psm | V18 – ST6 | ST– ST6 V18 – V20 (Generated from Geometry) |
V18 – ST6 | N/A |
| Solid Edge Direct | .par, .asm, .psm | V18 – ST6 | ST– ST6 V18 – V20 (Generated from Geometry) |
V18 – ST6 | N/A |
| SolidWorks | .sldprt, .sldasm | 98 – 2014 | 98 – 2014 | 98 – 2014 | N/A |
| SolidWorks Direct | .sldprt, .sldasm | 2003 – 2014 | 2003 – 2014 | 2003 – 2014 | N/A |
| STEP | .stp, .step | 203, 214 | 203, 214 (Generated from Geometry) | 203, 214 | N/A |
| STL | .stl | N/A | All | N/A | N/A |
| VDA-FS | .vda | 1.0 – 2.0 | N/A | 1.0 – 2.0 | N/A |
| XCGM | R2012 – R2014x | R2012 – R2014x | R2012 – R2014x | R2012 – R2014x |
Note: CATPart files created in V5R1 that have been opened and re-saved in a later version of CATIA V5 are not supported for 3D InterOp Graphical and 3D InterOp CGM.
Note: CATIA V6 users should export their database objects as CATIA V5 CATParts, CATProducts, XCGM, or as 3D XML, which can then be imported into applications using InterOp.
Note: NX .prt extensions must be changed to .ug to avoid conflict with ProE files.
StressCheck® uses Parasolid as its native CAD kernel. The native format is .x_t, or the Parasolid Transmit format. StressCheck® integrates multiple CAD translator options to convert from IGES, CATIA, Pro/E and others to Parasolid format. These modules are sold separately.
The Object Resolution input in Display>View Controls… allows you to change the rendering resolution of the displayed surfaces. The default is 21, and the range is 1-100 (with 100 being the maximum Object Resolution).
When Mesh Region is used for assignment of attributes, such as material properties, some elements within the mesh region may not be selected. This may happen because the value of the Object Resolution in the View Controls interface (21 by default), is not large enough to represent the shell accurately enough. From the Main Menu select Display > View Controls and change Object Resolution to 40 or higher (upper bound is 100). Increase this value until all elements have been properly assigned the attributes. To store this value with the file for future StressCheck® sessions, create a parameter with the name _object_res with a value equal to the desired object resolution. Then, save the file to store this object resolution.
The difference between an elliptical surface for an Object Resolution of 21 and 100 is below:
The Midsides input field affects both the resolution of the display of element edges in the model, and the resolution of results extraction. It is also found when performing h-Discretization to describe how many times an element will be “split”.
When in the Plot or Min/Max tab, the Midsides input will control the resolution of the plot grid/search grid:
For plotting purposes, each element is subdivided into a data extraction mesh which depends on the number of “Midsides”, that is, the number of grid points along each side not including the end points. The desired data are then computed in the grid points of the data mesh. The higher the number of midsides, the more accurate the display, and the longer the required computer time. In general, using 5 to 10 midsides is sufficient.
The difference in fringe plots between a Midsides of 2 (upper) and 10 (lower) is shown below:
For Min/Max purposes, specify the data mesh, which is characterized by the number of “Midsides”. The specified function will be computed in the grid points of the data mesh. The data mesh determines the set on which the minimum and maximum will be sought. For a typical p-version mesh, between 8 and 12 midsides should be used. For an automesh, between 5 and 10 midsides should be used.
The difference in maximum stress searches between a Midsides of 3 (upper) and 12 (lower) is shown below:
In the above, more than 3 Midsides were necessary to locate the maximum.
In the display of the model, the Midsides/Edge Resolution input controls the resolution of the element edges/faces to allow for smoother/coarser rendering. It has no influence on the model’s solution. It can be accessed and changed in Display>View Controls… under “Edge Resolution”:
The Edge Resolution/Midsides is set to 5 in the above. By default, the number of element edge midsides is 2. This means each element edge will be displayed as N+1 segments, where N=2 (3 segments in this case). However, when solved the element edge is represented by higher-order polynomial functions.
The default for Edge Resolution/Midsides display can be changed for future sessions under File>Options:
The difference in the element edge display between an Edge Resolution of 2 (upper) and 5 (lower) is shown below:
Note that the number of segments in “Edge Resolution: 2” is 3 for each element edge, while in “Edge Resolution: 5” it is 6 for each element edge. Also, note that the number of symbols for the boundary conditions (i.e. loads/constraints) is equal to the number of segments.
Both DeLast and Undo commands remove previously successful operations from StressCheck® sessions. However, there are important differences as they relate to the model.
DeLast is found in the Geometry/Mesh tabs, and is intended to delete the last object (e.g. node, box, contact zone) created in the model, independent of session. For example, if you load a StressCheck® model and click DeLast, it will delete the last object created in the model.
Undo is also found in under Edit in the main menu bar (or CTRL+Z), and will simply revert the session back to the previous state before the last successful operation (e.g. creation of a mesh, assignment of a boundary condition). For example, if a load was applied to a surface, clicking Undo will reverse the assignment.
Undo only works for the current session. If a file is loaded into a new session, the Undo function will be reset.
Constraining/fixing a point or node should ONLY be used when the model still has a free rigid body mode (i.e. translation or rotation). Equilibrium of the model under the applied loads must be considered before constraining/fixing a point or node.
No node constraints needed (symmetry on three orthogonal sides)
Fixed Z-direction node constraint needed (rigid body translation for Z-axis)
Rigid body constraints needed (self-equilibrated load)
Note: If constraints do not prevent all free rigid body motion (zero-strain translations and/or rotations) then the model is underconstrained and cannot be solved.
Note: contact constraints are simply normal, grounded distributed springs (in units of F/L/L^2) within the assigned contact zones. Therefore, a single multi-body contact assignment in 3D controls up to 4 rigid body modes per part:
Free rigid body modes must be controlled by tangent springs.
If the model is self-equilibrated under the applied loads, then the Rigid Body Constraint method may be used (Constraint tab, Select > Node/Point > Rigid Body). No other constraints are required.
Note: For the model to be self-equlibrated under the applied loads, the load check must return relatively small forces/moments (Load tab, Check > All Elements > Selection, enter the load ID and click Accept). The computed total forces/moments may not be exactly zero due to accumulation of small numerical errors.
If the model is NOT self-equilibrated under the applied loads, other constraints must exist that provide reactions for the applied loads (e.g. Symmetry Constraints or Spring Constraints). Such constraints may be used to provide a reaction and prevent rigid body motion in one direction or in all directions. For example, if the applied resultant forces are in equilibrium in the X direction but not the Y direction, then an additional constraint should only be used to react the Y direction force; one or more Nodal Constraints may be used to prevent any remaining rigid by motions (Constraint tab, Select > Node/Point > Node).
Note: in 2D/3D elasticity, nodes and points should NOT react load! They should only be used to prevent rigid body motion, thus setting a reference for the elasticity solution. Weak normal or tangent springs may also be used if a node or point location is not appropriate for cancellation of rigid body motion.
When using the “Create” action in the Geometry or Mesh tab, there are multiple ways to create new objects:
It’s possible more than one of these actions was invoked during an object creation. Note: left-click of the mouse in the Model Viewing window will always perform the current action in the “Action” combo (e.g. Create, Select, Edit).
As a best practice, if focus is required in the Model Viewing window, and a new object creation is NOT desired, it is recommended to right-click in the Model Viewing window to establish the window’s focus such that the model can be rotated, zoomed, etc. This can help with reducing duplication of geometry and mesh objects.
If you suspect there are duplicate geometry and/or mesh objects, changing the combos to “Select > Any Object” and clicking on the “Index” subtab within the Geometry or Mesh tabs will display a list of the objects in the model:
If there are duplicates, you may simply select the object(s) from the Index list and click the “Delete” button, or simply click the “DeLast” button if the last object created was a duplicate.
Duplicate nodes may be inadvertently generated when manually meshing models. If nodes are duplicated, clicking the Mesh tab’s “Merge” button will attempt to merge all duplicated nodes within a tolerance of 10e-6 model units. All but one of the coincident nodes are discarded, and all element definitions are updated appropriately. Note: care must be taken for multi-body contact models with coincident parts, as well as fracture models with intentional cracks, as these nodes may be merged.
Alternatively, the user may first determine if any free edges/disconnected elements exist in the mesh by using “Check > Edge > Free Edge” within the Mesh tab. Element edges which are free/not shared will be highlighted in red. Then, the user may select only those nodes which are duplicated (either via marquee selection or Shift + left-click) and click the Merge button to limit the merge to those nodes.
Note: to override the merge tolerance, the user may create the parameter _merge_tol and supply a new merge tolerance.
The error message likely occurred because of an Explicit Associativity in the model. This is a dependency created by the user in which an object definition references another object or objects in the model.
These messages typically occur when attempting to perform a modeling operation such as a Boolean-Union/Subtraction, Clip-Back/Front, Blend Edge, Imprint, Spin or Thicken.
An Explicit Associativity will exist if one or more objects (e.g. points, systems, nodes) are created by an associative method (e.g. Offset, Intersection, Projection) via a user-defined reference to the target geometry. Examples are creating a node by the Offset method on the geometry’s curve, or creating a system by 3-Pt. Plane method using points at the geometry’s intersections (vertices).
There are two types of Associativity: Implict and Explicit. They are described in the following:
Implicit Associativity – The result of a boolean or blend operation is a collection of trimmed surfaces, trimmed curves and points. The trimmed curves represent the intersections between neighboring surfaces. The points represent the intersections between trimmed curves.
Explicit Associativity – It is sometimes necessary to create new explicit associative relationships in a model.
Associativity is extremely valuable because it allows for fully parametric modeling, object inheritance (i.e. surface to element face communication) and highly curved elements (mapping).
The importance of Implicit and Explicity Associativity, and how to determine object associativities and even disassociate points and systems from geometries, is explained in the following tutorial:
Yes, it is very simple to define formulae for live results processing purposes. Formulae may involve expressions of StressCheck®‘s standard engineering functions, such as Sx, Sy, Ex, etc. as well as formula subexpressions (i.e. formula names contained within {} characters) and parameter names.
To perform plots or extractions for a particular formulaic expression, the user simply selects the “Fmla” option under “Functions:” and then selecting the name of the formula:
The following are examples of useful formulaic expressions which may be used for results processing of stresses (credit to Randall Pauley, NAVAIR for submitting many of these to ESRD). Feel free to copy/paste as formula expressions, keeping the formula names exactly as they appear:
A formula file (.PAR) with many of the above stress transformation expressions is available for import:
Helpful Hints and Tips: Stress Transformation Formulae for Results Processing
Here is an example tutorial of plotting with formulaic expressions:
And here is an example of the Tsai-Hill failure criteria for laminated composite analysis:
In 2D/3D elasticity, it is necessary to convert a bending moment into a linear bending stress distribution. From StruCalc:
In StressCheck®, this is usually done via a formula expression using the section properties SPMOMX (moment of inertia about centroidal X-axis) or SPMOMY (moment of inertia about centroidal Y-axis) to represent “I” in the above equation’s denominator, and assigned as a Traction using a coordinate system with a Z-axis normal to a set of element faces or surfaces representing a flat cross section.
Here’s an example of Mx=1000 lbf-in converted to a bending stress (in psi) on a symmetric I-section using the formula expression “BENDING”:
The formula is then assigned to the flat cross sectional surfaces/element faces of the I-section as a normal traction distribution, using the appropriate system and direction:
In the above example, the selected local system (SYS1) is assumed to have an origin at the centroid of the cross section. If the local system is NOT located at the centroid of the cross section but it is aligned with its principal axes, the above equation may simply be modified to be “BENDING = 1000*(y-SPCTRY)/SPMOMX”, where SPCTRY is the y-coordinate of the centroid with respect to the local system.
Below is an example of modifying the bending stress expression for a coordinate system (SYS2) NOT located at the centroid of the symmetric I-beam cross section:
Note: for symmetric cross sections the principal axes are along the symmetry planes
In general, or if you cannot guarantee the local system you select is aligned with the cross section principal directions, the cross moment of inertia may be present. To consider it use:
In StressCheck®, for Mx=1000 lbf-in, the above formula would take the form of the following expression:
BENDING = (-(0*SPMOMX+1000*SPCROSS)*(x-SPCTRX)+(1000*SPMOMY+0*SPCROSS)*(y-SPCTRY))/(SPMOMX*SPMOMY-SPCROSS^2)
The section property intrinsic functions available for use in formula expressions to automatically compute the area (SPAREA) and other section properties of an assigned set of surfaces/element faces are below:
Here is a video tutorial of the definition and assignment of a bending stress:
StressCheck® Tutorial: Defining and Assigning a Bending Stress
StressCheck® supports the use of certain intrinsic functions and mathematical expressions (FORTRAN or Mathematica conventions) when defining parameters and/or formulae. For example, “sin(x)” for sine(x), where “x” is a user input in radians.
The following intrinsic functions and relational expressions, in addition to the intrinsic constant “PI” for π, may be used in parameter and/or formula expression fields:
Note that the asterisked function “if” is intended for use only in special situations in the formula input dialog window. To create an “if” statement, all three arguments, the condition, true value, and false value, must be provided. Also note that when an intrinsic function requires more than one argument, the arguments must be separated by a semi-colon (;) not a comma (,) as you might expect.
Below is an example of defining and assigning a formula shear stress in which intrinsic functions are required:
StressCheck Tutorial: Defining and Assigning a Parabolic Shear Stress Formula Distribution
Below is an example of using a relational expression to define a formula spring coefficient:
StressCheck® also provides a mechanism for evaluating simple expressions and depositing the result in the input field. If you enter an equal sign (=), followed by an expression (FORTRAN or Mathematica conventions), the result will appear after pressing the Tab key.
For example, “=sin(rad(45))^2” can be entered in an input field and pressing the Tab key would convert 45 degrees to radians, evaluate the sine, and then square the result. Conversely, the following input (FORTRAN convention) would also be acceptable and would return the same value: “=sin(pi/4)**2”. All computations are in double precision and are not case sensitive.
In addition, StressCheck® supports intrinsic functions for the computation of 3D section properties when assigning loads to selected elements, element faces, geometric volumes or geometric surfaces. The functions are referenced in a formula without any arguments and without parentheses (e.g. c1*sparea). The following table contains a list of the special intrinsic functions for section property computation:
Below is a FAQ on using 3D section property intrinsic functions to represent the normal stress due to a bending moment:
It may be desired to use a parameter or formula to represent node or point offsets, coordinates or distances. The intrinsic functions described below, interrogate the definitions of existing points and nodes to obtain information about their location in space:
Parameter (or formula) expressions may be defined in terms of other parameters (or formulae) or may be assigned a value directly. Updating the independent parameter value will automatically update the dependent parameter value. Below is an example of an independent parameter “a” and a dependent parameter “b” with expression “2*a”:
If defining a new formula that is dependent on an existing formula (or formulae), curly brackets {} must be used to surround the existing formula name (or names) in the new formula expression. Below is an example of a formula “a” with expression “b+c”, where “b” and “c” are parameters, and a formula “d” with an expression “2*{c}”, where “{c} represents a formula sub-expression:
Yes, it is very simple to define formulae for live results processing purposes. Formulae may involve expressions of StressCheck®‘s standard engineering functions, such as Sx, Sy, Ex, etc. as well as formula subexpressions (i.e. formula names contained within {} characters) and parameter names.
To perform plots or extractions for a particular formulaic expression, the user simply selects the “Fmla” option under “Functions:” and then selecting the name of the formula:
The following are examples of useful formulaic expressions which may be used for results processing of stresses (credit to Randall Pauley, NAVAIR for submitting many of these to ESRD). Feel free to copy/paste as formula expressions, keeping the formula names exactly as they appear:
A formula file (.PAR) with many of the above stress transformation expressions is available for import:
Helpful Hints and Tips: Stress Transformation Formulae for Results Processing
Here is an example tutorial of plotting with formulaic expressions:
And here is an example of the Tsai-Hill failure criteria for laminated composite analysis:
The von Mises yield criterion (also known as the Maximum Distortion Energy Theory of Failure) is a pure shear based criterion (i.e. plasticity occurs as the crystals shear across lattices). Therefore part of the assumption implies that the hydrostatic component of the stress tensor (which cause no shear) are completely ignored.
Hence, you could have an infinite pressure (i.e., S1=S2=S3=+/- infinity, Si being the principal stress components) that would cause no yield whatsoever. Therefore, individual stress components can be really large even without exceeding the input yield stress. This means that plotting and extracting the von Mises stress (Seq) or strain (Eeq) is the best measure of the amount of plasticity in the model.
Note: the input yield stress is the yield value for the pristine material, sometimes chosen as the elastic limit (the stresses are linearly elastic up to this value). However, once the material plasticizes this value changes depending on the hardening criteria, and can be higher (typical case for isotropic hardening theories) or lower than the reference initial value.
The display format is in C-language, which is described in detail here: https://www.cprogramming.com/tutorial/printf-format-strings.html
The model input display format can be accessed and changed by clicking Display>View Controls… and entering a new “Display Format:”
This will be reflected in all model inputs, including parameter values. By default, the display format is %11.4e, which means a maximum length of 11 digits, a precision of 4 digits and exponential. For example, 11,012.54 will be displayed as “1.1013e5”. To change to a floating format, simply replace the “e” with an “f”.
Once a SCW or SCP file is saved, it will store your current model input display format. To change the default model input display format so that every StressCheck® session uses the same display format, click File>Options and modify the Display Format (below). Then click “OK”:
The differences in model input between the default (%11.4e, top) format and floating point (%11.4f, bottom) are below:
The model results display format can be modified by changing the “Format:” input in the Results tabs. The formatting is the same as described above.
The differences in model results between the default (%11.3e) format and floating point (%11.3f) are below:
The StressCheck® Resultant tab allows the user to integrate the tractions over the edges (2D) or faces (3D) of the elements to calculate the resultant forces Fx, Fy, Fz and moments Mx, My, Mz. This can be done over selected edges/faces, or over entire element boundaries.
In the case of Planar elasticity, the Resultant computation is as follows:
Checking Equilibrium in the Resultant Tab
In the equilibrium test, the tractions are computed along the boundaries of elements from the finite element solution and integrated along the perimeter. When no body forces are applied, the contour integral of the tractions should be nearly zero. When body forces are applied, the contour integral of the tractions should be nearly equal and opposite to the applied body forces. You can read more about the feature on page 43 of the Analysis Guide (459 overall) and page 150-151 of the User’s Guide (pages 166-167 overall).
The resultant tab is useful for determining the discretization error of the stresses for elements. If there is high error, or the equilibrium test is not near zero for a particular element or region, then that element has more discretization error. If the element is in your region of interest, then refinement may be necessary, but you may want to extract the maximum value of interest to see if it converges first.
A Note on Nodal Reactions
StressCheck® does not provide nodal reactions, as this information is not conceptually meaningful for higher-order implementations of the finite element method. Also, satisfying equilibrium of nodal forces is a property of the finite element method, and does not give an indication of quality of load transfer/reactions.
Calculating resultant forces and moments by integrating stresses does serve this purpose, and should therefore be employed for an “honest” assessment of load transfer/reactions and equilibrium checks.
Comparing Resultant Load Transfer/Reactions with the Load Check
If there is discretization (numerical) error in the solution’s stresses at or near the region(s) of load transfer/reaction, the resultant forces and moments in these regions may not perfectly match the applied forces and moments as observed in the load check feature. Conducting a load check (Load tab, Check > Any Element or Check > All Elements) simply performs a summation of all the applied loads for the selected elements. A load check is always recommended to make sure what you applied is seen by the model.
Tip: plotting the Error function (Plot tab, Function: Error) is useful to determine where the relative highest error is in your model. The elements that are red (1.0) contribute the most to the overall global error. The elements that are blue (closer to zero) contribute the least.
When the load check and resultant tab (Select > Face, marquee selecting all your elements) give near the same value, then the discretization error is low and the stresses produced in the solution are of good quality. However, just because they are different does not mean your region of interest is not converging. These are merely indicators that you may need more refinement.
In most complex 3D parts the resultant tab will likely return a non-zero check unless your model has a high level of refinement, which isn’t always the most practical approach. If you are happy with the local convergence of your data of interest, then you shouldn’t spend too much time on the equilibrium check.
Resultants and Multi-Body Contact
The integration of the pressures generated on each body due to load transfer must result in a quality resultant force/moment in order to be a reliable contact solution. Therefore, the first check that should be performed after any multi-body contact solution is complete is that neighboring contact zones are in equilibrium. This can be performed in the Resultant tab by Select > Contact Zone, picking neighboring contact zones, and ensuring these are as close to zero as possible.
For a solution or sequence of solutions, errors related to the solution quality may be determined by the following four checks: global (energy norm), deformed shape, fringe continuity and the local data convergence (including how limits may be computed to estimate error).
Additional, but more expensive, checks may include the strain energy error and the error indicators.
When a linear solution of at least three runs of increasing degrees of freedom (N) is completed, the solver window will show an “Estimated Error” percentage to indicate the current estimated relative error in the energy norm (i.e. global error):
The global error value is computed based on the relative error in the energy norm calculator:
We always assume the most conservative algebraic rate convergence when estimating the global error. This means the actual error may be even better than what StressCheck® reports:
This global error value is also what is computed in the Results Error tab, for “Select > All Elements” and the “Estimate” toggle is active:
The global error % vs. DOF in a log-log scale is plotted. A table will show each run #, DOF, and the convergence rate (ideally a value close to or greater than 1.0 for smooth elasticity solutions) and global error % associated with each run #.
For a smooth elasticity solution (i.e. no stress singularities caused by cracks, TLAPs or sharp corners):
Note: Care must be exercised in interpreting this global error measure. In general, it is not sufficient to ensure that the energy norm measure of error is small. It is also necessary to check that the data of interest are converging to their limit values.
Note: In some cases the estimated limit cannot be computed. This happens when the error associated with the available finite element solutions cannot be approximated by a function of the form C/N^2β. Usually there is an error in the input data, or some special symmetry or antisymmetry is present.
The deformed shape is an easy way to spot modeling errors in the solution before checking fringe continuity or local data convergence. For example, if a load was applied in the wrong direction or a rigid body constraint is missing from the solution.
In the Results Plot tab, select “Deform” as the Shape, activate “Overlay”, let the deformation autoscale (i.e. Scale is not checked) and click “Plot”:
The undeformed (red outline) and deformed (elements) shapes will be overlaid, allowing you to determine if the model input resulted in the expected deformation. Checking the “Scale” and inputting a value will allow for custom scaling of deformations.
Note: the deformed shape can be very useful to determine if multi-body contact solutions are transferring load appropriately, but a Scale < 5 is recommended when plotting these results. Each body will be scaled independently of the other for multi-body contact results.
Ideally, for smooth elasticity solutions, the fringes across element boundaries (edges and faces) should be continuous. Fringe continuity can be checked by plotting the fringes of stresses or strains (displacements will always be continuous across element boundaries) for all elements, or a group of elements/element faces.
In the Results Plot tab, activate the “Fringe” toggle, select the function (e.g. S1 for 1st prin stress), input the # of midsides to control the plot grid resolution (8-12 is typically sufficient to assess continuity), and click “Plot”:
Discontinuous Fringes at p=4
Continuous Fringes at p=8
If there are large “jumps” in the fringes across element boundaries, it is an indicator of discretization and/or modeling error.
To check that the data of interest (e.g. local stresses/strains/SIFs) are converging to a limit in the region(s) of interest, select the Results Min/Max tab, and choose between selecting all elements, an element or group elements, an element face or group of element faces, or another available object selection. The selection will indicate where the minimal and/or maximal values will be computed.
When computing minimal and/or maximal values from a sequence of at least 3 solutions, we perform an estimation of the true value of the selected function by projecting the results from the finite element solutions to an infinite number of degrees of freedom.
For a maximal value, the result of this projection is reported as the “Maximum Limit” together with the percent deviation from the value corresponding to the solution chosen having the highest number of degrees of freedom:
In this case, a Maximum Limit was computed for the maximum 1st principal stress within the selected element faces (1.915e5 psi) and thus the Run 8 estimated error is calculated to equal 0.05% (100*(1.915e5-1.914e5)/1.914e5)).
Note: Generally, the solution can be considered to be of good quality if the error measured in energy norm is small and the location and magnitude of the minimal or maximal values of the data of interest clearly converge to a limiting value.
Note: The computation of the estimated limit is performed as follows:
These should be understood as heuristic procedures which tend to work well when the errors in the data are small, but may give a false indication of the limit when the errors are large. The procedure may compute a limit even when the limit does not exist or fail to compute a limit even when one does exist.
The following are additional checks that provide qualitative (error indicator) and qualitative (strain energy error) assessments of the solution but are more expensive to compute in real-time.
If the “Indicator” toggle is selected, an error indicator based on jumps in the stress resultants is computed for an element or group of elements. The error indicator for the “k”th element provides an indication of the relative influence of the element on the global error of the solution.
To compute the error indicator, the program first determines the jumps in the stress resultants across any shared edges (2D) or faces (3D). By definition, the jump (Δ) is the sum of the absolute values of the differences in stress resultants. An element (k) has as many Δ’s as there are shared edges/faces. For example, the jump for shared edge j of element k for a planar (2D) problem is given by:
Note: you can enter a value between 0 and 1 in the Criteria input field, and the program will report only those elements with an error indicator exceeding that value.
To plot the error indicator as outlined above when the Error tab’s “Indicator” toggle is active, select the function “Error” in the Plot tab, select an element or group of elements and click Plot:
This is a visual representation of the same data generated in the Error tab when “Indicator” is activated.
Note: elements which have a high error indicator my indicate a discretization or modeling error, but the local data may still converge so this should be verified independently.
The strain energy error is what is computed in the Results Error tab, for “Select > Any Element”, “Estimate” toggle is active and an element or group of elements is selected. The strain energy (U) is computed from the following expression:
U = ½∫{ε}{σ}dV
Where where {ε} is the mechanical strain tensor, {σ} is the stress tensor and the integral is performed over each element:
For the sequence of solutions selected, the results will include the run #, the unconstrained degrees of freedom (U_DOF), the strain energy, convergence rate and % error associated with each run #, and the estimated limit of the strain energy.
The strain energy error can be useful for identifying an element or elements that most contribute to the global error of the solution, and to identify if the strains or stresses of an element or elements is stable, needs refinement or contains a singularity.
Each Solution ID requires a unique Load ID and Constraint ID. Some users wish to combine the results of two or more load cases together, such as axial/bending or mechanical/thermal load cases, and this may be accomplished via the StressCheck® Calculator (Functions: Calc):
The StressCheck® Calculator can be accessed in the Plot, Min/Max and Points tabs under “Functions: Calc”.
Up to five different solutions may be combined, each “Solution:” representing a Solution ID of a Load ID/Constraint ID. The Operand allows the Solution ID and Run to be converted into an integer specification.
A formula may be entered to represent the superposition of displacements, stresses or strains. For example, the above represents the addition of the stresses in the global X-dir for the 8th runs of Solution ID’s “SOLAXIAL” and “SOLBENDING”.
Here are some examples of combining load case results in the StressCheck® Calculator:
To display the object numbering for visualization, click on Display>Objects…
Then, check the Label box next to the object numbering you want to visualize and click “Draw”. For example, if we want to visualize the element numbers, we can check the Label box next to “Elements” and click “Draw”:
To turn of the element numbering, uncheck the Label box and click “Draw”.
After the solution is complete, the final p-level of each solved element may be displayed/tabulated:
If p-Discretization has been assigned to one or more sets of elements (for example, a fixed p-level of p=2 to the element list in SET24 and p=6 to the other elements in the model), the table will indicate the distribution of mixed p-levels:
The default for plotting/extractions is Cartesian coordinates (X, Y, Z). However, StressCheck® supports the plotting/extraction of results in Cylindrical coordinates (R, T, Z):
Here is a short tutorial of plotting radial and hoop stress: https://www.esrd.com/product/results-in-cylindrical-coords/
The Object Resolution input in Display>View Controls… allows you to change the rendering resolution of the displayed surfaces. The default is 21, and the range is 1-100 (with 100 being the maximum Object Resolution).
When Mesh Region is used for assignment of attributes, such as material properties, some elements within the mesh region may not be selected. This may happen because the value of the Object Resolution in the View Controls interface (21 by default), is not large enough to represent the shell accurately enough. From the Main Menu select Display > View Controls and change Object Resolution to 40 or higher (upper bound is 100). Increase this value until all elements have been properly assigned the attributes. To store this value with the file for future StressCheck® sessions, create a parameter with the name _object_res with a value equal to the desired object resolution. Then, save the file to store this object resolution.
The difference between an elliptical surface for an Object Resolution of 21 and 100 is below:
The Midsides input field affects both the resolution of the display of element edges in the model, and the resolution of results extraction. It is also found when performing h-Discretization to describe how many times an element will be “split”.
When in the Plot or Min/Max tab, the Midsides input will control the resolution of the plot grid/search grid:
For plotting purposes, each element is subdivided into a data extraction mesh which depends on the number of “Midsides”, that is, the number of grid points along each side not including the end points. The desired data are then computed in the grid points of the data mesh. The higher the number of midsides, the more accurate the display, and the longer the required computer time. In general, using 5 to 10 midsides is sufficient.
The difference in fringe plots between a Midsides of 2 (upper) and 10 (lower) is shown below:
For Min/Max purposes, specify the data mesh, which is characterized by the number of “Midsides”. The specified function will be computed in the grid points of the data mesh. The data mesh determines the set on which the minimum and maximum will be sought. For a typical p-version mesh, between 8 and 12 midsides should be used. For an automesh, between 5 and 10 midsides should be used.
The difference in maximum stress searches between a Midsides of 3 (upper) and 12 (lower) is shown below:
In the above, more than 3 Midsides were necessary to locate the maximum.
In the display of the model, the Midsides/Edge Resolution input controls the resolution of the element edges/faces to allow for smoother/coarser rendering. It has no influence on the model’s solution. It can be accessed and changed in Display>View Controls… under “Edge Resolution”:
The Edge Resolution/Midsides is set to 5 in the above. By default, the number of element edge midsides is 2. This means each element edge will be displayed as N+1 segments, where N=2 (3 segments in this case). However, when solved the element edge is represented by higher-order polynomial functions.
The default for Edge Resolution/Midsides display can be changed for future sessions under File>Options:
The difference in the element edge display between an Edge Resolution of 2 (upper) and 5 (lower) is shown below:
Note that the number of segments in “Edge Resolution: 2” is 3 for each element edge, while in “Edge Resolution: 5” it is 6 for each element edge. Also, note that the number of symbols for the boundary conditions (i.e. loads/constraints) is equal to the number of segments.
“Hands-off” automation of StressCheck® Professional pre-processing, solutions and live dynamic processing is available via StressCheck®‘s COM API. The StressCheck® COM API includes access to the toolkit of StressCheck®‘s objects, methods and properties to allow for solutions to repetitious and/or challenging FEA models. Below is an example of a fracture mechanics Sim App:
Customized functions and subroutines may be written and executed either via scripting or a distributable Sim App product which uses StressCheck® as its FEA engine. In order to get started with the StressCheck® COM API, the following are required:
The following demonstrates in several programming languages (Excel VBA, MATLAB and Python) how to get started with the StressCheck® COM API for scripting. Note: in the below examples, the ® character is not required:
In Excel VBA, it is very easy to automate StressCheck® because all that is required is an installation of Microsoft Office Excel.
The following video demonstrates how to set a reference to the latest StressCheck® Application Object library in VBA’s References list, and then initiate a new StressCheck® Application via VBA scripting:
More examples of Excel VBA scripting with StressCheck®:
https://www.esrd.com/product/helpful-hints-tips-excel-vba-3d-example/
In MATLAB, it is simple enough to create a new ActiveX server StressCheck®.Application process and save the commands as an M-file (.m):
% Launch StressCheck® application as a COM ActiveX Server
SC = actxserver(‘StressCheck®.Application’);
Below is an example of MATLAB scripting with StressCheck®:
Before using the StressCheck® COM API with Python, it is necessary to install the latest Python for Windows and PyWin32 extensions:
Once these are installed, you may write a Python script (.py) and execute via command line or Pythonwin. Import win32com in the Python script to open a new StressCheck®.Application:
import win32com.client as win32
sc = win32.gencache.EnsureDispatch(‘StressCheck®.Application’)
No, the Modal/Buckling solver cannot currently be combined with multi-body contact analysis. The Modal/Buckling solver is for eigenvalue/eigenmode analysis for single parts.
Modal analysis is available for planar, plate bending and three-dimensional problems, while Buckling analysis is available only for three-dimensional problems.
In eigenvalue buckling analysis, the goal is to compute the buckling load factor (BLF) and the corresponding buckling mode shape. The BLF is the multiplier of the applied loads to the component to produce the first buckling mode. Note: checking the values of strains/stresses from the eigenvalue buckling solution is not meaningful because the buckling mode shape is known only to an arbitrary constant.
Therefore the linear solution multiplied by the BLF is the reference for when buckling will initiate. That means the stresses corresponding to the linear solution multiplied by the BLF are the stresses that will be developing at the time instability occurs (paired with the corresponding mode shape determined by the buckling analysis). Note: a linear analysis of the model is computed during an eigenvalue buckling analysis, and is available for results processing once the eigenvalue buckling solution is available.
No, the Modal/Buckling solver cannot currently be combined with multi-body contact analysis. The Modal/Buckling solver is for eigenvalue/eigenmode analysis for single parts.
Modal analysis is available for planar, plate bending and three-dimensional problems, while Buckling analysis is available only for three-dimensional problems.
When elastic-plastic effects do not need to be considered in a multi-body contact analysis (i.e. the converged maximum von Mises stress in the parts is below the yield stress of the materials), only the Linear Elasticity solver is required.
This is because for each multi-body contact iteration, StressCheck® only needs to update the right-hand side of the equation (i.e. load vector/pressures) and not the left-hand side of the equation (i.e. stiffness matrix). StressCheck® currently accounts for small-strain, small-deformation solutions (linear or elastic-plastic materials) in multi-body contact analysis.
The Nonlinear Solver is required when the stiffness matrix needs to be updated due to a change in stress-strain relationship or a change in geometric configuration. Additionally, if normal springs have been assigned to element faces which contain elastic-plastic material assignments, the normal springs will release in tension.
StressCheck® currently does not account for changes in geometric configuration for multi-body contact analysis.
More Resources:
When compared to a single-part analysis, a multi-body contact analysis of the same degrees of freedom (DOF) is more computationally demanding. This increase in computational time is due to the nonlinear iterations required to accurately depict high-gradient pressure distributions between complex parts, especially if the part count (and by extension the number of contact pairs) is high.
If a multi-body contact analysis is required, there are several approaches to potentially reduce solver time without significantly affecting the data of interest or the load transfer.
Although it is not required to have a 1:1 correspondence of element faces between contacting parts (our contact algorithm is very flexible regarding mesh differences between contacting parts), it is more efficient to have close to a 1:1 correspondence between surface areas in contact.
An additional benefit is that splitting surfaces for contact assignment improves local mesh refinement. This is discussed in the following document:
Helpful Hints and Tips: Optimization of a Simple 3D Contact Problem
By default, multi-body contact solutions require 40 iterations to complete, unless the maximum contact pressure error is <0.01%. In order to change the number of iterations, create the parameter _contact_iter and set this to a value between 10 and 100.
Note: it is not recommended to reduce the number of iterations unless the maximum contact pressure error is <2% and load transfer is >95% at lower iterations.
Note: for some classes of problems, reducing the number of iterations may not affect the multi-body contact solution. For example, close contact problems with very high loads in which the pressure distributions are very smooth. However, some classes of problems may require a high number of iterations in order to properly represent the contact pressures (and therefore load transfer) between parts.
If a part or parts in multi-body contact are relatively simple to manually (hand) mesh, such as simple fasteners and bushings, this approach can potentially reduce computational time:
For example, if a fastener body is to be repeated in multiple holes, it is possible to create/import the solid representing the fastener body, create the mesh/contact zones for the fastener body, and then copy this fastener body to multiple holes.
The below document provides several case studies on reducing the number of contact iterations and hand-meshing/copying fasteners:
Helpful Hints and Tips: Improving Multi-Body Contact Efficiency
In cases where higher accuracy of load transfer and pressures is required, it is useful to increase the polynomial level of the surfaces in contact independently from the rest of the model via p-Discretization. For example, surfaces in contact may be fixed at p=6 (or higher) while all other elements outside the selected surfaces are solved at p=5 (or lower).
Additionally, boundary layer refinement around bores in multi-body contact aids in the full conversion to geometric mapping and allows the polynomial order of the elements in contact to be increased to p=8. This strategy is demonstrated below:
StressCheck Tutorial: Boundary Layer Refinement and P-Discretization in Contact Regions
The model in the tutorial was solved for three cases:
Case 1: Default automesh, 40 contact iterations
Case 2: Boundary layer refinement in holes of middle plate, p-Discretization of p=8 in holes of middle plate, 20 contact iterations
Case 3: Default automesh, p-Discretization of p=8 in holes of middle plate, 20 contact iterations
The time savings between Case 1 and Case 2 was about 19%, and Case 1 and Case 3 about 42%, without significantly affecting the results in the middle plate holes:
Case 1: Default automesh, 40 contact iterations
Case 2: Boundary layer refinement in holes of middle plate, p-Discretization of p=8 in holes of middle plate, 20 contact iterations
Case 3: Default automesh, p-Discretization of p=8 in holes of middle plate, 20 contact iterations
Additionally, general improvements to solution efficiency without significantly affecting the data of interest are described in the below helpful hints document:
In general, the contact constant should be selected based on the moduli of elasticity of the materials in contact, in such a way that its value is between 0.1 and 1.0 times the value of the smallest modulus of elasticity (0.1Emin < Kc < Emin) when using US units (psi), and between 0.004 and 0.04 times the value of the smallest modulus of elasticity (0.004Emin < Kc < 0.04Emin) when using SI units (MPa). The units for the contact constant are [F/L/L^2].
For aluminum or steel, this is usually 1-10 million psi/in or so, assuming significant operational loads.
Helpful Hints and Tips: Multi-Body Contact Overview and General Recommendations
Some other guidelines:
It’s best to use a parameter for the contact constant, so it can be updated based on the applied load magnitude. The idea is that if the applied loads produce significant stress, the contact constant should be increased relative to the load. If the applied loads don’t produce significant stress, the contact constant should be lowered.
When using the Frictionless Augmented Lagrangian Method (StressCheck®‘s contact algorithm), the contact constant is multiplied by the gap function to determine the change in contact pressures, which are then added to the previous iteration’s pressures, so it should not be a small value (relative to the loads/material). It would then take many iterations to stabilize the contact pressures.
From the blog article: https://www.simscale.com/blog/2016/09/contact-mechanics-friction/
Augmented Lagrange Method was formulated to bring the good of both penalty and Lagrange multiplier. It does not demonstrate an oscillatory behavior and has minimal interpenetration. The change in energy contribution for a small change in gap can be given as:
(previous pressure + contact coefficient * gap) x change in gap
StressCheck®‘s multi-body contact algorithm does not require mesh conformity or a 1:1 ratio between contacting element faces. In fact, the element face distribution across assigned contact zones can be quite dissimilar in most practical applications.
However, the resolution of the gap between bodies for calculation of contact pressures is related to the number of collocation points used for each element face of the assigned contact pairs. By default, 196 points per element face in the contact pair are used for calculating the gap function.
The contact gap/penetration is determined by sampling from one contact zone to the other. If the ratio between the number of element faces in the assigned contact pairs is large (for example, 100:1 meaning that there are about 100 more element faces on one side of the declared contact pair than on the other), then the gap function is evaluated at many more locations on one side than on the other. This means that the approximation of the gap function has a different resolution which can induce oscillation during the nonlinear multi-body contact solution.
Therefore, it is recommended that the ratio between the number of element faces in the assigned contact pairs be kept close to 1:1. Numerical studies have shown that ratios not exceeding 10:1 are in general satisfactory. A comparison between a 100:1 and the more reasonable 10:1 contacting element face ratio is provided below.
In this case, multi-body contact between a lug, clevis and pin was solved. The pin contact zone surrounded 24 element faces, while the number of element faces in the lug was either ~100 or ~10 times more refined. The von Mises stresses were then plotted for the lug in the same stress ranges:
Pin element faces (24 in contact zone)
100:1 ratio between lug and pin contacting element faces
10:1 ratio between lug and pin contacting element faces
As shown in the example, oscillation of the stress distribution is a clear indication that the ratio between contacting element faces is too large.
While regular and similar meshes across contact zones are preferable, there is flexibility to have higher ratios when multiple discretization strategies result in dissimilar element counts across contact pairs. Because the situation is problem-dependent, it is not possible to have a fixed rule to define by how much it can deviate from the optimal 1:1 ratio for all cases, so it is very important to check for oscillations in the results of interest along the contacting surfaces during the post-processing of a contact analysis.
Stress resultant computation at selected element faces.
When performing a multi-body contact analysis, unexpected part deformations, contact stress distributions and/or contact pair load transfer checks may be caused by one or more of the following common issues:
Significant penetration between contact zones.
“Knife edge” effects at lug termination.
For cases where the initial gap is numerically zero, the projected ray from one zone to another may not find the proper location.
If singularities are causing the poor multi-body contact load transfer, “boundary layers” of refinement may be added in the contact regions to improve the errors. The stresses will still have jumps due to the discontinuous pressure distribution, but the influence is lessened on the overall resultant check.
Also, the Resultant check should include element faces surrounding the contact zone in order to capture the “bleed” of contact pressure:
StressCheck Tutorial: Influence of Stress Singularities On Multi-Body Contact Resultant Calculations
More Resources:
Currently, there is not a capability to combine incremental plasticity (Type: Material (NL Mat) with Technique: Incremental) or general/geometric nonlinear (Type: General (NL Gen)) solutions with multi-body contact.
Multi-body contact may be combined with deformation theory of plasticity, however, to represent elastic-plastic material nonlinearities (Type: Material (NL Mat) with Technique: Deformation) in the case of limit load analysis.
Here is the appropriate Nonlinear Solver window for deformation theory of plasticity with multi-body contact:
Here is an example of how to combine multi-body contact with deformation theory of plasticity: https://www.esrd.com/product/helpful-hints-tips-converting-linear-elastic-contact-elastic-plastic-contact/
If a 3D cold-working operation is desired, the effects of mandrel insertion/removal may be represented with the following processes.
The von Mises yield criterion (also known as the Maximum Distortion Energy Theory of Failure) is a pure shear based criterion (i.e. plasticity occurs as the crystals shear across lattices). Therefore part of the assumption implies that the hydrostatic component of the stress tensor (which cause no shear) are completely ignored.
Hence, you could have an infinite pressure (i.e., S1=S2=S3=+/- infinity, Si being the principal stress components) that would cause no yield whatsoever. Therefore, individual stress components can be really large even without exceeding the input yield stress. This means that plotting and extracting the von Mises stress (Seq) or strain (Eeq) is the best measure of the amount of plasticity in the model.
Note: the input yield stress is the yield value for the pristine material, sometimes chosen as the elastic limit (the stresses are linearly elastic up to this value). However, once the material plasticizes this value changes depending on the hardening criteria, and can be higher (typical case for isotropic hardening theories) or lower than the reference initial value.
When elastic-plastic effects do not need to be considered in a multi-body contact analysis (i.e. the converged maximum von Mises stress in the parts is below the yield stress of the materials), only the Linear Elasticity solver is required.
This is because for each multi-body contact iteration, StressCheck® only needs to update the right-hand side of the equation (i.e. load vector/pressures) and not the left-hand side of the equation (i.e. stiffness matrix). StressCheck® currently accounts for small-strain, small-deformation solutions (linear or elastic-plastic materials) in multi-body contact analysis.
The Nonlinear Solver is required when the stiffness matrix needs to be updated due to a change in stress-strain relationship or a change in geometric configuration. Additionally, if normal springs have been assigned to element faces which contain elastic-plastic material assignments, the normal springs will release in tension.
StressCheck® currently does not account for changes in geometric configuration for multi-body contact analysis.
More Resources:
If all that is required is to simulate load transfer to holes or a compression-only reaction is desired for a single part, thereby reducing the computational time required for multi-body contact, normal springs combined with a Material or General Nonlinear analysis may be an efficient workflow.
This “poor man’s contact” form of load transfer may be represented by applying normal springs to element edges or curves (Planar) or element faces or surfaces (3D), applying the appropriate material properties, and solving a Material or General Nonlinear analysis:
Normal springs may be assigned in the Constraints tab, using Select > Edge/Edge Curve/Any Curve (Planar) or Face/Face Surface/Any Surface (3D) > Spring Coeff:
Note: if a fastener radial spring stiffness is to be approximated, the below expression for Kr may be useful:
Because a General Nonlinear analysis is more computationally expensive than a Material Nonlinear analysis, it may be desired to run an analysis that simulates normal springs in compression only. This requires specifying a nonlinear material with either a representative yield stress (if the linear von Mises stresses call for plasticity to be computed), or an artificial yield stress to force the material nonlinear algorithm to release normal springs that are in tension but otherwise keep the stresses linear.
After a linear analysis is available, a Material Nonlinear analysis (Type: Material (NL Mat)) may be executed to compute plasticity + represent compression-only normal springs, or simply represent compression-only normal springs.
If a General Nonlinear analysis is required to account for large displacements and/or rotations, and equilibrium must be solved in the deformed configuration, then the solution should use Type: General (NL Gen). Normal springs will always be compression-only in a General Nonlinear analysis.
Note: nonlinear material properties do not need to be assigned to any elements in order to solve a General Nonlinear analysis.
For more information:
The stress-strain curve can be plotted for nonlinear materials in the Material tab, Assign sub-tab by clicking Plot:
By default, 24 points are displayed for the stress-strain curve fitting, which may not be sufficient for some nonlinear material curves.
You can increase the points generated in the stress-strain curve by defining two underscore parameters:
For example, set _NL_COUNT to 100 and _NL_MAX to 0.1, and the curve will be plotted using 100 points in the strain range from 0 to 0.1. However, note that these parameters have no effect on the solution itself.
Note: This technique also increases the # points displayed in the property vs temperature material curve for Temperature-Dependent materials.
Currently, there is not a capability to combine incremental plasticity (Type: Material (NL Mat) with Technique: Incremental) or general/geometric nonlinear (Type: General (NL Gen)) solutions with multi-body contact.
Multi-body contact may be combined with deformation theory of plasticity, however, to represent elastic-plastic material nonlinearities (Type: Material (NL Mat) with Technique: Deformation) in the case of limit load analysis.
Here is the appropriate Nonlinear Solver window for deformation theory of plasticity with multi-body contact:
Here is an example of how to combine multi-body contact with deformation theory of plasticity: https://www.esrd.com/product/helpful-hints-tips-converting-linear-elastic-contact-elastic-plastic-contact/
If a 3D cold-working operation is desired, the effects of mandrel insertion/removal may be represented with the following processes.
From the StressCheck® Master Guide:
A Fastener is a special element that can be attached to circular boundaries. It consists of a rigid core with two degrees of freedom connected to the planar body by a distributed radial (normal) spring. The rigid core can be connected to the rigid core of other fasteners directly or using a link element, it can be fixed in one or both directions, or it can be loaded by a force or an imposed displacement.
The spring coefficient (Kr) is determined from the elastic solution of a disc under radial compression (plane-strain condition), as the ratio between the applied pressure and the radial displacement. In order to properly account for the effects of the Fastener element, it is necessary to run both a Linear AND a Nonlinear solution. This is because the normal springs must release in tension, and this requires a Nonlinear solution. The type of Nonlinear solution required depends on the following:
If no nonlinear materials have been assigned to any elements in the model, then it is necessary to run a Type: General (NL Gen):
If nonlinear materials have been assigned to at least one element in the model, and geometric nonlinearities are not important, then a Type: Material (NL Mat) is recommended:
If the effects of material AND geometric nonlinearities are important, then a Type: General (NL Gen) is recommended.
StressCheck® Tutorial: Fastener Element and General Nonlinear Analysis
In order to determine the applicability and viability of representing loading/unloading events via Incremental plasticity theory (IPT), it is important to conduct pre-solution checks before performing potentially time-consuming nonlinear analyses.
Note: these guidelines and recommendations to determine if IPT is an option for the current model inputs are a complement to the technical documentation on StressCheck®‘s implementation of IPT found here:
After performing a p-extension of at least three linear runs (e.g. p=2 to 4 or p=6 to 8), we should explore the model’s linear solution to assess the following characteristics:
StressCheck Tutorial: Directional vs Follower Loads in Geometric Nonlinear Analysis
StressCheck Tutorial: Linear vs Nonlinear Results for a Single Lap Joint
StressCheck Tutorial: Checking ITP Viability for Solving Nonlinear Events
The Nonlinear solver enables two types: Material (NL Mat) or General (NL Gen). The Nonlinear solver interface is below:
Note: to solve a nonlinear problem, a linear problem must be solved first and selected as the initial step in the nonlinear analysis. Additionally a Fastener Element requires a Nonlinear solution.
StressCheck Tutorial: Linear vs Nonlinear Results for a Single Lap Joint
Material Nonlinear (NL Mat) accounts for nonlinearities associated with changes in material properties. Stresses and strains are related by a strain-dependent matrix, and strains are not known in advance.
An elastic-plastic stress-strain relationship (e.g. Ramberg-Osgood or elastoplastic) must be defined and assigned to at least one element in the model for the Material Nonlinear algorithm to succeed.
There are two techniques/plasticity theories associated with Material Nonlinear: Deformation and Incremental. The capabilities are described below:
Note: Details of both plasticity theories can be found in the Master Guide. A note on the hardening models for incremental plasticity can be found here. A note on the implementation of Deformation theory can be found here.
The scope of Material Nonlinear analysis is as follows:
Small-strain, small-displacement, with elastic-plastic and nonlinear-elastic material characterization for plane-stress, plane strain, axisymmetric and three-dimensional problems. Springs will remain attached to the boundary only under compression if defined in the normal direction.
Here are some examples of Material Nonlinear analysis:
General Nonlinear (NL Gen) accounts for nonlinearities associated with changes in configuration, as in large deflections of a slender elastic beam, as well as accounting for elastic-plastic material properties (if assigned to any element).
The General Nonlinear implementation is based on the Spatial/Eulerian formulation satisfying equilibrium in the deformed configuration (accounts for changes in geometry). It includes nonlinear strains (Almansi) and large deformations with the material description being linear-elastic or elastic-plastic. For elastic-plastic materials the von Mises yield criterion is used together with the Deformation theory of plasticity. A note on the General Nonlinear implementation can be found here.
There are two integration techniques associated with General Nonlinear: Direct and Newton-Raphson.
If a General Nonlinear analysis fails for Direct integration, Newton-Raphson integration is recommended.
The scope of General Nonlinear analysis is as follows:
StressCheck Tutorial: Directional vs Follower Loads in Geometric Nonlinear Analysis
Here are some examples of General Nonlinear analysis:
Ideally, the error in the linear solution should be small before running a nonlinear (Material or General) solution. From the Master Guide:
Having entered the input data, execute a Linear analysis. It is recommended that you obtain a sequence of solutions and estimate the relative error in the energy norm. At least one of the solutions in the sequence should have a small relative error (typically under 1% for 2D-problems and under 5% for 3D-problems). If this were not the case, then either refine the mesh or switch to the product space (or both) and repeat the Linear analysis. In general, you should expect the accuracy of the nonlinear solution to be somewhat lower than the accuracy of the linear solution. In the case of plane strain, axisymmetric, and 3D problems, for example, the plastic strain is subject to the incompressibility constraint, which is not present in the case of the linear solution.
Assuming the discretization error was small in the linear solution, the following may cause the Material or General Nonlinear algorithm to fail, producing a LAPACK, element stiffness generation, or some other error:
In some cases, the applied load/displacement causes such high stresses/strains in the linear solution that the nonlinear algorithm has difficulty projecting from the linear solution to the elastic-plastic material stress/strain curve. This can result in the algorithm exiting prematurely or failing.
Possible solution: Use load stepping to increment the applied load/displacement from a smaller % of the target value to the target value:
If the final value is still too large for convergence to be achieved in the nonlinear algorithm, the desired analysis may be outside the scope of implementation. From the Master Guide:
In general, it is good practice to use the linear solution to estimate the level of loading which will not cause extensive plastic flow, the treatment of which would lie outside of the scope of small strain models.
In general, the largest strain of the linear solution will become even larger for the nonlinear solution. The spread of the plastic zone can be anticipated by examining the equivalent stress (Seq) contour plot of the linear solution. The size of the plastic zone will be larger than the region for which Seq > Syield.
If there are “sliver” elements in the mesh, meaning the solid angles of some elements are extremely small (<1 degree) or extremely large (>179 degrees), these elements may fail during General Nonlinear analysis as the volume of the element cannot be computed in the deformed configuration. The curvature and/or aspect ratios of the elements must be improved in order to address this issue.
Possible solution: Additional mesh refinement may be required, or the geometry may need to be modified in order to accommodate elements with improved solid angles and mapping (i.e. curvature).
Normal springs will always release in tension for a General Nonlinear analysis, and for a Material Nonlinear analysis if the normal springs are attached to elements with elastic-plastic materials. This means that if normal springs were used to keep the model in equilibrium, additional constraints or directional springs may be necessary.
Possible solution: Check to determine if the normal springs can be replaced by different boundary conditions, or by directional (XYZ) springs (which will not release in tension).
Additionally, determine if the model can be self-equilibrated by applying balancing loads instead of normal springs.
In order for the Material Nonlinear algorithm to proceed, there must be enough elastic material (i.e. no permanent plastic strains) surrounding regions of plastified material (i.e. permanent plastic strains). From the Master Guide, on Deformation Theory of Plasticity:
The effects of a single overload event on structures made of ductile materials are of substantial practical importance. Such effects can be well represented by mathematical models based on the deformation theory of plasticity provided that the plastic flow is contained, that is, the plastic zone is surrounded by elastic material.
Make sure that the applied load or imposed displacement will not cause excessive plastic strains, except in the immediate vicinity of singular points. Remember that the small strain, small displacement theory is used. The plastic zone should be completely confined by an elastic zone.
Possible solution: Incorporate load stepping to incrementally plastify the material, and ensure that the region becoming plastified has a low discretization error in the linear solution and is confined by an elastic zone.
An error occurs similar to “no nonlinear material properties were assigned to any elements”.
Possible solution: Assign at least one element elastic-plastic material properties in order to solve a Material Nonlinear analysis.
The linear solution’s maximum von Mises stress never exceeded the yield stress of the elastic-plastic material assignment, and the Material Nonlinear algorithm exits.
Possible solution: If it is expected that plasticity occur based on the model inputs, check the model and make sure the inputs are appropriate. Otherwise, a Material Nonlinear analysis is not required.
The Material Nonlinear algorithm is sensitive to the elastic-plastic material curve inputs, so it is very important that the yield stress, tangent moduli and other material curve inputs are consistent.
Possible solution: Check the elastic-plastic material curve inputs and ensure the inputs are correct and in consistent units.
Additional information:
The display format is in C-language, which is described in detail here: https://www.cprogramming.com/tutorial/printf-format-strings.html
The model input display format can be accessed and changed by clicking Display>View Controls… and entering a new “Display Format:”
This will be reflected in all model inputs, including parameter values. By default, the display format is %11.4e, which means a maximum length of 11 digits, a precision of 4 digits and exponential. For example, 11,012.54 will be displayed as “1.1013e5”. To change to a floating format, simply replace the “e” with an “f”.
Once a SCW or SCP file is saved, it will store your current model input display format. To change the default model input display format so that every StressCheck® session uses the same display format, click File>Options and modify the Display Format (below). Then click “OK”:
The differences in model input between the default (%11.4e, top) format and floating point (%11.4f, bottom) are below:
The model results display format can be modified by changing the “Format:” input in the Results tabs. The formatting is the same as described above.
The differences in model results between the default (%11.3e) format and floating point (%11.3f) are below:
By default, displaying the load arrows for a selected load ID will display the load arrows for every load record within the load ID. Additionally, the load arrows will be scaled automatically to appropriate sizes based on the current model view, and the relative lengths of the load arrows will be with respect to the applied load magnitudes in the load records.
The below provides tips and tricks for displaying load arrows for a selected load ID.
If multiple load ID’s are defined in the model, and we wish to view a particular load ID, we can use the “Display Loads/Flux” drop down (above image). The current load ID will have a blue circle.
This is the easiest way to quickly cycle between the load arrow displays of each load ID in the model.
The relative sizes of the load arrow displays for the selected load ID records are controlled via the “Scale” box in the Load tab. The “Scale” value can be automatically generated (default) or user-defined.
Note: the “Scale” value has no effect on the magnitudes of the applied loads.
A model may contain multiple load records within a load ID, and we may want to display only one load record within the load ID to display its load arrow magnitudes and directions.
To only display the load arrows for a single load record, we recommend the following:
This process ensures that the load arrow displays for an assignment can be isolated for easier viewing.
To display the object numbering for visualization, click on Display>Objects…
Then, check the Label box next to the object numbering you want to visualize and click “Draw”. For example, if we want to visualize the element numbers, we can check the Label box next to “Elements” and click “Draw”:
To turn of the element numbering, uncheck the Label box and click “Draw”.
In StressCheck®, sets are used to group objects by type (such as elements). This is similar to the Group concept used in other FEA. Sets are designed to allow you to view model input and results limited to a group of objects instead of viewing all objects of that particular type.
If a set of elements exists (i.e. Model Input, Sets tab, Select>Any Element, view list items), it can be selected and changed:
To select a set of elements from the model display, hold CTRL+SHIFT and select an element in the set. The Sets Browser will appear, with the Candidate sets including the element set in the list (in this case, the Selection is set to Any Element):
To display only this element set, click Display>Objects… and “Sets” tab. Check the “Elements” box, select the element set from the list, then click “Draw” to display only this element set:
This tutorial video shows how to manipulate element displays using automesh and user-generated sets: https://www.esrd.com/product/stresscheck-tutorial-using-sets/
The Edit Toolbar enables object selection/cancellation (by object type or any object), inverting object selections, object blanking (by object type or any object), object unblanking (by object type or any object), visualizing blanked objects, picking only visible objects and more.
This tutorial video explains the details of the Edit Toolbar:
StressCheck® Tutorial: Using Edit Toolbar for Object Selection and Visualization
There are a number of variables that may affect fringe plot rendering rates, especially if the cutting planes option is enabled. Though, models with large numbers of elements and/or solution degrees of freedom (DOF) will cause the rendering to perform slower in general.
Fringe plot rendering time can be drastically altered depending on user inputs
Here are a few suggestions that might help reduce the fringe plot rendering time when using the Plot tab with the Contour: Fringe option enabled:
More information on plotting:
Some StressCheck® users may experience decreased model performance and/or integrity in certain situations. The following may be considered as potential reasons:
Is your “Path to Scratch Directory” option pointing locally or to a network location? If your scratch directory (i.e. a temporary directory automatically created for StressCheck®-related session files and solution data) is pointing to a network path, this can drastically slow down model loading and solution times, and even affect the data integrity.
Check the scratch directory path by opening StressCheck® and clicking File > Options, and verify that the “Path to Scratch Directory” value is pointing locally (e.g. C:\Users\…..\AppData\Local\Temp”, which is the default), and is not pointing to a restricted area on your machine (e.g. C:\Program Files\ESRD\StressCheck® PE 10).
It is recommended to keep the scratch directory path relatively clean of old/discarded StressCheck® temporary directories (dataset folders) as well as other non-essential temporary folders to reduce overhead. These temporary directories have the format of “dataset-xxxxxxxxx-xxxxxxxx”, with these directories (and their contents) normally removed after StressCheck® sessions are closed by the user.
However, if StressCheck® sessions exit abnormally, these directories may remain. It is recommended to remove any unwanted temporary directories in the scratch directory path.
It is recommended to read (open) and write (save) StressCheck® files locally instead of relying on a network connection. This ensures the packing/unpacking procedure is allowed to complete uninterrupted. Additionally, the local directory should not be a system or restricted folder requiring additional write permissions.
Opening a StressCheck® file from a mail client, such as Outlook, may cause unpacking issues. Therefore, StressCheck® file attachments should be saved locally before opening.
Adhering to these recommendations will ensure that the integrity of StressCheck®‘s files are preserved, and there are no interruptions during the file read/write that can cause corruptions.
Ideally, the scratch directory path would have plenty of disk space (i.e. > 1 GB), otherwise larger models will almost certainly load and solve at slower rates due to limited disk availability.
Also, there is much file I/O occurring during model loading and solving (especially during stiffness matrix inversion), which can be further limited by your hard drive’s read/write speed.
Ideally, the hard drive would be a steady state hard drive (SSD) or at least a high-RPM mechanical hard disk drive (HDD). If your hard drive is quite fragmented, near capacity and/or has a slower RPM mechanical hard disk drive, model performance may be slowed.
Learn more about ESRD Software Product System Requirements.
Are you running a node-locked or floating license configuration? For node-locked (i.e. machine-specific) licenses, firewalls/network security slowdowns typically are not an issue as there is no network communication involved.
However, for floating (server-based) licenses, communication through a restrictive network firewall to/from the license server’s FLEXnet license manager can cause significant slowdowns in model performance. This communication is required in order to obtain/return StressCheck® license features and check the license server’s “heartbeat” (i.e. determine if the license server is still capable of serving licenses).
Check with your system administrator to determine if your StressCheck® license server communicates through a higher-security, firewall-enabled network that limits/prevents network communication or specific ports. If this is the case, slowdowns may also be observed for other software products which need to communicate with the same license server.
Ideally, the graphics processor for StressCheck® (and all ESRD products) should be set to your video card’s high-performance graphics processing unit (GPU) instead of integrated graphics (e.g. Intel on-board). Some examples of video card manufacturers include Nvidia and AMD, which design dedicated GPU’s for high-performance graphics acceleration and rendering.
For additional information on ensuring you are using your video card’s high-performance GPU (such as Nvidia or AMD) for StressCheck®: https://www.techadvisor.co.uk/how-to/pc-components/how-set-default-graphics-card-3612668/
Has processor affinity (i.e. the specific declaration of CPU cores to be used for an application) been set? Learn more: https://www.tekrevue.com/tip/restrict-apps-cpu-cores-processor-affinity/
The Object Resolution input in Display>View Controls… allows you to change the rendering resolution of the displayed surfaces. The default is 21, and the range is 1-100 (with 100 being the maximum Object Resolution).
When Mesh Region is used for assignment of attributes, such as material properties, some elements within the mesh region may not be selected. This may happen because the value of the Object Resolution in the View Controls interface (21 by default), is not large enough to represent the shell accurately enough. From the Main Menu select Display > View Controls and change Object Resolution to 40 or higher (upper bound is 100). Increase this value until all elements have been properly assigned the attributes. To store this value with the file for future StressCheck® sessions, create a parameter with the name _object_res with a value equal to the desired object resolution. Then, save the file to store this object resolution.
The difference between an elliptical surface for an Object Resolution of 21 and 100 is below:
The Midsides input field affects both the resolution of the display of element edges in the model, and the resolution of results extraction. It is also found when performing h-Discretization to describe how many times an element will be “split”.
When in the Plot or Min/Max tab, the Midsides input will control the resolution of the plot grid/search grid:
For plotting purposes, each element is subdivided into a data extraction mesh which depends on the number of “Midsides”, that is, the number of grid points along each side not including the end points. The desired data are then computed in the grid points of the data mesh. The higher the number of midsides, the more accurate the display, and the longer the required computer time. In general, using 5 to 10 midsides is sufficient.
The difference in fringe plots between a Midsides of 2 (upper) and 10 (lower) is shown below:
For Min/Max purposes, specify the data mesh, which is characterized by the number of “Midsides”. The specified function will be computed in the grid points of the data mesh. The data mesh determines the set on which the minimum and maximum will be sought. For a typical p-version mesh, between 8 and 12 midsides should be used. For an automesh, between 5 and 10 midsides should be used.
The difference in maximum stress searches between a Midsides of 3 (upper) and 12 (lower) is shown below:
In the above, more than 3 Midsides were necessary to locate the maximum.
In the display of the model, the Midsides/Edge Resolution input controls the resolution of the element edges/faces to allow for smoother/coarser rendering. It has no influence on the model’s solution. It can be accessed and changed in Display>View Controls… under “Edge Resolution”:
The Edge Resolution/Midsides is set to 5 in the above. By default, the number of element edge midsides is 2. This means each element edge will be displayed as N+1 segments, where N=2 (3 segments in this case). However, when solved the element edge is represented by higher-order polynomial functions.
The default for Edge Resolution/Midsides display can be changed for future sessions under File>Options:
The difference in the element edge display between an Edge Resolution of 2 (upper) and 5 (lower) is shown below:
Note that the number of segments in “Edge Resolution: 2” is 3 for each element edge, while in “Edge Resolution: 5” it is 6 for each element edge. Also, note that the number of symbols for the boundary conditions (i.e. loads/constraints) is equal to the number of segments.
Using the Parts class to organize objects into a single named entity is very useful for pre-processing and post-processing tasks, as well as defining solution configurations and organizing object displays.
However, if a certain Part is selected in the Parts selector at the time of saving a project file (SCP) or workfile (SCW), objects not included in the selected Part will not be displayed when opening the file:
If certain objects are not in the display, first check to ensure each object type is enabled for display.
After opening the file, first check to ensure all object displays are activated (see above). If certain objects are still not displayed, check to see if a Part name is selected in the Parts selector. The Part may have only certain model objects in it, meaning all other model objects will not be displayed.
To remedy the display of all model objects, change the Parts selector from the current Part name to “All Objects”:
All model entities (i.e. elements, nodes, systems, points, etc.) should be displayed. It is then recommended to save the project or workfile in this configuration:
For more information about Parts:
When using the “Create” action in the Geometry or Mesh tab, there are multiple ways to create new objects:
It’s possible more than one of these actions was invoked during an object creation. Note: left-click of the mouse in the Model Viewing window will always perform the current action in the “Action” combo (e.g. Create, Select, Edit).
As a best practice, if focus is required in the Model Viewing window, and a new object creation is NOT desired, it is recommended to right-click in the Model Viewing window to establish the window’s focus such that the model can be rotated, zoomed, etc. This can help with reducing duplication of geometry and mesh objects.
If you suspect there are duplicate geometry and/or mesh objects, changing the combos to “Select > Any Object” and clicking on the “Index” subtab within the Geometry or Mesh tabs will display a list of the objects in the model:
If there are duplicates, you may simply select the object(s) from the Index list and click the “Delete” button, or simply click the “DeLast” button if the last object created was a duplicate.
Duplicate nodes may be inadvertently generated when manually meshing models. If nodes are duplicated, clicking the Mesh tab’s “Merge” button will attempt to merge all duplicated nodes within a tolerance of 10e-6 model units. All but one of the coincident nodes are discarded, and all element definitions are updated appropriately. Note: care must be taken for multi-body contact models with coincident parts, as well as fracture models with intentional cracks, as these nodes may be merged.
Alternatively, the user may first determine if any free edges/disconnected elements exist in the mesh by using “Check > Edge > Free Edge” within the Mesh tab. Element edges which are free/not shared will be highlighted in red. Then, the user may select only those nodes which are duplicated (either via marquee selection or Shift + left-click) and click the Merge button to limit the merge to those nodes.
Note: to override the merge tolerance, the user may create the parameter _merge_tol and supply a new merge tolerance.
For models located many units away from the Global origin (0,0,0), or in general when multiple length scales (e.g. 0.01->1000 units) in the model display are rendered, small features such as Boolean-unioned crack surface meshes may appear to be distorted or “wavy”. This small feature distortion is simply an artifact of single-precision object rendering by StressCheck®‘s visualization utility (HOOPS by TechSoft 3D), and is not related to the double-precision computation of element stiffnesses, mappings, solution data, etc. Therefore, results will not be affected.
In the below example, a plate with a hole and 0.01″ corner crack is re-located from 0,0,0 in. to 1000,1000,1000 in. This represents a significant change in feature rendering scale (i.e. 0.01″ to 1000″) in the same view. The data of interest is the stress intensity factor (SIF) distribution along the crack:
While the display of the crack becomes distorted as the model is moved far away from the Global origin, the SIF distribution is unaffected (all three curves are coincident). Contact ESRD Support if you suspect results are influenced by the model’s location.
The StressCheck® Resultant tab allows the user to integrate the tractions over the edges (2D) or faces (3D) of the elements to calculate the resultant forces Fx, Fy, Fz and moments Mx, My, Mz. This can be done over selected edges/faces, or over entire element boundaries.
In the case of Planar elasticity, the Resultant computation is as follows:
Checking Equilibrium in the Resultant Tab
In the equilibrium test, the tractions are computed along the boundaries of elements from the finite element solution and integrated along the perimeter. When no body forces are applied, the contour integral of the tractions should be nearly zero. When body forces are applied, the contour integral of the tractions should be nearly equal and opposite to the applied body forces. You can read more about the feature on page 43 of the Analysis Guide (459 overall) and page 150-151 of the User’s Guide (pages 166-167 overall).
The resultant tab is useful for determining the discretization error of the stresses for elements. If there is high error, or the equilibrium test is not near zero for a particular element or region, then that element has more discretization error. If the element is in your region of interest, then refinement may be necessary, but you may want to extract the maximum value of interest to see if it converges first.
A Note on Nodal Reactions
StressCheck® does not provide nodal reactions, as this information is not conceptually meaningful for higher-order implementations of the finite element method. Also, satisfying equilibrium of nodal forces is a property of the finite element method, and does not give an indication of quality of load transfer/reactions.
Calculating resultant forces and moments by integrating stresses does serve this purpose, and should therefore be employed for an “honest” assessment of load transfer/reactions and equilibrium checks.
Comparing Resultant Load Transfer/Reactions with the Load Check
If there is discretization (numerical) error in the solution’s stresses at or near the region(s) of load transfer/reaction, the resultant forces and moments in these regions may not perfectly match the applied forces and moments as observed in the load check feature. Conducting a load check (Load tab, Check > Any Element or Check > All Elements) simply performs a summation of all the applied loads for the selected elements. A load check is always recommended to make sure what you applied is seen by the model.
Tip: plotting the Error function (Plot tab, Function: Error) is useful to determine where the relative highest error is in your model. The elements that are red (1.0) contribute the most to the overall global error. The elements that are blue (closer to zero) contribute the least.
When the load check and resultant tab (Select > Face, marquee selecting all your elements) give near the same value, then the discretization error is low and the stresses produced in the solution are of good quality. However, just because they are different does not mean your region of interest is not converging. These are merely indicators that you may need more refinement.
In most complex 3D parts the resultant tab will likely return a non-zero check unless your model has a high level of refinement, which isn’t always the most practical approach. If you are happy with the local convergence of your data of interest, then you shouldn’t spend too much time on the equilibrium check.
Resultants and Multi-Body Contact
The integration of the pressures generated on each body due to load transfer must result in a quality resultant force/moment in order to be a reliable contact solution. Therefore, the first check that should be performed after any multi-body contact solution is complete is that neighboring contact zones are in equilibrium. This can be performed in the Resultant tab by Select > Contact Zone, picking neighboring contact zones, and ensuring these are as close to zero as possible.
Sometimes, it may be required to update set lists to reflect changes in the model. This is especially true for sets created by the Locate method, in which the pick coordinates and box tolerance will determine which objects will be added to the set.
When set information must be updated, rather than manually replacing each set it is helpful to define a “dummy” parameter to force all sets and assignments to be updated.
For example, we can create a parameter called “dummy” and enter “1+1” in the Expression field. Click Accept, and all model sets will be automatically updated:
More information on Locate sets:
The stress-strain curve can be plotted for nonlinear materials in the Material tab, Assign sub-tab by clicking Plot:
By default, 24 points are displayed for the stress-strain curve fitting, which may not be sufficient for some nonlinear material curves.
You can increase the points generated in the stress-strain curve by defining two underscore parameters:
For example, set _NL_COUNT to 100 and _NL_MAX to 0.1, and the curve will be plotted using 100 points in the strain range from 0 to 0.1. However, note that these parameters have no effect on the solution itself.
Note: This technique also increases the # points displayed in the property vs temperature material curve for Temperature-Dependent materials.
Use the intrinsic function “mod(x;y)” to round to the nearest integer value.
For a parameter A=3.14159, create a parameter expression B=A-mod(A;1). The resulting value for B=3.0. If B=4.0 is required, use B=A-mod(A;1)+1.
“Mod(x;y)” may be used in formula expressions.
When CAD files imported into StressCheck® have very small geometric features, these are sometimes suppressed by the MeshSim automesher, resulting in elements not associated with the geometric surfaces in regions near the small features. Some indicators of this are:
1) Loads/constraints applied to geometric surfaces are not applied to element faces
2) MeshSim warnings regarding parasolid error
3) Warning message regarding significant node displacement
To ensure the small geometric features are not suppressed by MeshSim, define the parameter _ms_suppress in the Model Info Parameters tab and set it to less than 5e-5 (relative feature size) and try re-meshing. Conversely, if you wish to suppress small features from automeshing, you may set the value to be the minimum feature size. MeshSim will skip the meshing of these features.
StressCheck® supports the use of certain intrinsic functions and mathematical expressions (FORTRAN or Mathematica conventions) when defining parameters and/or formulae. For example, “sin(x)” for sine(x), where “x” is a user input in radians.
The following intrinsic functions and relational expressions, in addition to the intrinsic constant “PI” for π, may be used in parameter and/or formula expression fields:
Note that the asterisked function “if” is intended for use only in special situations in the formula input dialog window. To create an “if” statement, all three arguments, the condition, true value, and false value, must be provided. Also note that when an intrinsic function requires more than one argument, the arguments must be separated by a semi-colon (;) not a comma (,) as you might expect.
Below is an example of defining and assigning a formula shear stress in which intrinsic functions are required:
StressCheck Tutorial: Defining and Assigning a Parabolic Shear Stress Formula Distribution
Below is an example of using a relational expression to define a formula spring coefficient:
StressCheck® also provides a mechanism for evaluating simple expressions and depositing the result in the input field. If you enter an equal sign (=), followed by an expression (FORTRAN or Mathematica conventions), the result will appear after pressing the Tab key.
For example, “=sin(rad(45))^2” can be entered in an input field and pressing the Tab key would convert 45 degrees to radians, evaluate the sine, and then square the result. Conversely, the following input (FORTRAN convention) would also be acceptable and would return the same value: “=sin(pi/4)**2”. All computations are in double precision and are not case sensitive.
In addition, StressCheck® supports intrinsic functions for the computation of 3D section properties when assigning loads to selected elements, element faces, geometric volumes or geometric surfaces. The functions are referenced in a formula without any arguments and without parentheses (e.g. c1*sparea). The following table contains a list of the special intrinsic functions for section property computation:
Below is a FAQ on using 3D section property intrinsic functions to represent the normal stress due to a bending moment:
It may be desired to use a parameter or formula to represent node or point offsets, coordinates or distances. The intrinsic functions described below, interrogate the definitions of existing points and nodes to obtain information about their location in space:
Parameter (or formula) expressions may be defined in terms of other parameters (or formulae) or may be assigned a value directly. Updating the independent parameter value will automatically update the dependent parameter value. Below is an example of an independent parameter “a” and a dependent parameter “b” with expression “2*a”:
If defining a new formula that is dependent on an existing formula (or formulae), curly brackets {} must be used to surround the existing formula name (or names) in the new formula expression. Below is an example of a formula “a” with expression “b+c”, where “b” and “c” are parameters, and a formula “d” with an expression “2*{c}”, where “{c} represents a formula sub-expression:
For a solution or sequence of solutions, errors related to the solution quality may be determined by the following four checks: global (energy norm), deformed shape, fringe continuity and the local data convergence (including how limits may be computed to estimate error).
Additional, but more expensive, checks may include the strain energy error and the error indicators.
When a linear solution of at least three runs of increasing degrees of freedom (N) is completed, the solver window will show an “Estimated Error” percentage to indicate the current estimated relative error in the energy norm (i.e. global error):
The global error value is computed based on the relative error in the energy norm calculator:
We always assume the most conservative algebraic rate convergence when estimating the global error. This means the actual error may be even better than what StressCheck® reports:
This global error value is also what is computed in the Results Error tab, for “Select > All Elements” and the “Estimate” toggle is active:
The global error % vs. DOF in a log-log scale is plotted. A table will show each run #, DOF, and the convergence rate (ideally a value close to or greater than 1.0 for smooth elasticity solutions) and global error % associated with each run #.
For a smooth elasticity solution (i.e. no stress singularities caused by cracks, TLAPs or sharp corners):
Note: Care must be exercised in interpreting this global error measure. In general, it is not sufficient to ensure that the energy norm measure of error is small. It is also necessary to check that the data of interest are converging to their limit values.
Note: In some cases the estimated limit cannot be computed. This happens when the error associated with the available finite element solutions cannot be approximated by a function of the form C/N^2β. Usually there is an error in the input data, or some special symmetry or antisymmetry is present.
The deformed shape is an easy way to spot modeling errors in the solution before checking fringe continuity or local data convergence. For example, if a load was applied in the wrong direction or a rigid body constraint is missing from the solution.
In the Results Plot tab, select “Deform” as the Shape, activate “Overlay”, let the deformation autoscale (i.e. Scale is not checked) and click “Plot”:
The undeformed (red outline) and deformed (elements) shapes will be overlaid, allowing you to determine if the model input resulted in the expected deformation. Checking the “Scale” and inputting a value will allow for custom scaling of deformations.
Note: the deformed shape can be very useful to determine if multi-body contact solutions are transferring load appropriately, but a Scale < 5 is recommended when plotting these results. Each body will be scaled independently of the other for multi-body contact results.
Ideally, for smooth elasticity solutions, the fringes across element boundaries (edges and faces) should be continuous. Fringe continuity can be checked by plotting the fringes of stresses or strains (displacements will always be continuous across element boundaries) for all elements, or a group of elements/element faces.
In the Results Plot tab, activate the “Fringe” toggle, select the function (e.g. S1 for 1st prin stress), input the # of midsides to control the plot grid resolution (8-12 is typically sufficient to assess continuity), and click “Plot”:
Discontinuous Fringes at p=4
Continuous Fringes at p=8
If there are large “jumps” in the fringes across element boundaries, it is an indicator of discretization and/or modeling error.
To check that the data of interest (e.g. local stresses/strains/SIFs) are converging to a limit in the region(s) of interest, select the Results Min/Max tab, and choose between selecting all elements, an element or group elements, an element face or group of element faces, or another available object selection. The selection will indicate where the minimal and/or maximal values will be computed.
When computing minimal and/or maximal values from a sequence of at least 3 solutions, we perform an estimation of the true value of the selected function by projecting the results from the finite element solutions to an infinite number of degrees of freedom.
For a maximal value, the result of this projection is reported as the “Maximum Limit” together with the percent deviation from the value corresponding to the solution chosen having the highest number of degrees of freedom:
In this case, a Maximum Limit was computed for the maximum 1st principal stress within the selected element faces (1.915e5 psi) and thus the Run 8 estimated error is calculated to equal 0.05% (100*(1.915e5-1.914e5)/1.914e5)).
Note: Generally, the solution can be considered to be of good quality if the error measured in energy norm is small and the location and magnitude of the minimal or maximal values of the data of interest clearly converge to a limiting value.
Note: The computation of the estimated limit is performed as follows:
These should be understood as heuristic procedures which tend to work well when the errors in the data are small, but may give a false indication of the limit when the errors are large. The procedure may compute a limit even when the limit does not exist or fail to compute a limit even when one does exist.
The following are additional checks that provide qualitative (error indicator) and qualitative (strain energy error) assessments of the solution but are more expensive to compute in real-time.
If the “Indicator” toggle is selected, an error indicator based on jumps in the stress resultants is computed for an element or group of elements. The error indicator for the “k”th element provides an indication of the relative influence of the element on the global error of the solution.
To compute the error indicator, the program first determines the jumps in the stress resultants across any shared edges (2D) or faces (3D). By definition, the jump (Δ) is the sum of the absolute values of the differences in stress resultants. An element (k) has as many Δ’s as there are shared edges/faces. For example, the jump for shared edge j of element k for a planar (2D) problem is given by:
Note: you can enter a value between 0 and 1 in the Criteria input field, and the program will report only those elements with an error indicator exceeding that value.
To plot the error indicator as outlined above when the Error tab’s “Indicator” toggle is active, select the function “Error” in the Plot tab, select an element or group of elements and click Plot:
This is a visual representation of the same data generated in the Error tab when “Indicator” is activated.
Note: elements which have a high error indicator my indicate a discretization or modeling error, but the local data may still converge so this should be verified independently.
The strain energy error is what is computed in the Results Error tab, for “Select > Any Element”, “Estimate” toggle is active and an element or group of elements is selected. The strain energy (U) is computed from the following expression:
U = ½∫{ε}{σ}dV
Where where {ε} is the mechanical strain tensor, {σ} is the stress tensor and the integral is performed over each element:
For the sequence of solutions selected, the results will include the run #, the unconstrained degrees of freedom (U_DOF), the strain energy, convergence rate and % error associated with each run #, and the estimated limit of the strain energy.
The strain energy error can be useful for identifying an element or elements that most contribute to the global error of the solution, and to identify if the strains or stresses of an element or elements is stable, needs refinement or contains a singularity.
In eigenvalue buckling analysis, the goal is to compute the buckling load factor (BLF) and the corresponding buckling mode shape. The BLF is the multiplier of the applied loads to the component to produce the first buckling mode. Note: checking the values of strains/stresses from the eigenvalue buckling solution is not meaningful because the buckling mode shape is known only to an arbitrary constant.
Therefore the linear solution multiplied by the BLF is the reference for when buckling will initiate. That means the stresses corresponding to the linear solution multiplied by the BLF are the stresses that will be developing at the time instability occurs (paired with the corresponding mode shape determined by the buckling analysis). Note: a linear analysis of the model is computed during an eigenvalue buckling analysis, and is available for results processing once the eigenvalue buckling solution is available.
For a solution or sequence of solutions, errors related to the solution quality may be determined by the following four checks: global (energy norm), deformed shape, fringe continuity and the local data convergence (including how limits may be computed to estimate error).
Additional, but more expensive, checks may include the strain energy error and the error indicators.
When a linear solution of at least three runs of increasing degrees of freedom (N) is completed, the solver window will show an “Estimated Error” percentage to indicate the current estimated relative error in the energy norm (i.e. global error):
The global error value is computed based on the relative error in the energy norm calculator:
We always assume the most conservative algebraic rate convergence when estimating the global error. This means the actual error may be even better than what StressCheck® reports:
This global error value is also what is computed in the Results Error tab, for “Select > All Elements” and the “Estimate” toggle is active:
The global error % vs. DOF in a log-log scale is plotted. A table will show each run #, DOF, and the convergence rate (ideally a value close to or greater than 1.0 for smooth elasticity solutions) and global error % associated with each run #.
For a smooth elasticity solution (i.e. no stress singularities caused by cracks, TLAPs or sharp corners):
Note: Care must be exercised in interpreting this global error measure. In general, it is not sufficient to ensure that the energy norm measure of error is small. It is also necessary to check that the data of interest are converging to their limit values.
Note: In some cases the estimated limit cannot be computed. This happens when the error associated with the available finite element solutions cannot be approximated by a function of the form C/N^2β. Usually there is an error in the input data, or some special symmetry or antisymmetry is present.
The deformed shape is an easy way to spot modeling errors in the solution before checking fringe continuity or local data convergence. For example, if a load was applied in the wrong direction or a rigid body constraint is missing from the solution.
In the Results Plot tab, select “Deform” as the Shape, activate “Overlay”, let the deformation autoscale (i.e. Scale is not checked) and click “Plot”:
The undeformed (red outline) and deformed (elements) shapes will be overlaid, allowing you to determine if the model input resulted in the expected deformation. Checking the “Scale” and inputting a value will allow for custom scaling of deformations.
Note: the deformed shape can be very useful to determine if multi-body contact solutions are transferring load appropriately, but a Scale < 5 is recommended when plotting these results. Each body will be scaled independently of the other for multi-body contact results.
Ideally, for smooth elasticity solutions, the fringes across element boundaries (edges and faces) should be continuous. Fringe continuity can be checked by plotting the fringes of stresses or strains (displacements will always be continuous across element boundaries) for all elements, or a group of elements/element faces.
In the Results Plot tab, activate the “Fringe” toggle, select the function (e.g. S1 for 1st prin stress), input the # of midsides to control the plot grid resolution (8-12 is typically sufficient to assess continuity), and click “Plot”:
Discontinuous Fringes at p=4
Continuous Fringes at p=8
If there are large “jumps” in the fringes across element boundaries, it is an indicator of discretization and/or modeling error.
To check that the data of interest (e.g. local stresses/strains/SIFs) are converging to a limit in the region(s) of interest, select the Results Min/Max tab, and choose between selecting all elements, an element or group elements, an element face or group of element faces, or another available object selection. The selection will indicate where the minimal and/or maximal values will be computed.
When computing minimal and/or maximal values from a sequence of at least 3 solutions, we perform an estimation of the true value of the selected function by projecting the results from the finite element solutions to an infinite number of degrees of freedom.
For a maximal value, the result of this projection is reported as the “Maximum Limit” together with the percent deviation from the value corresponding to the solution chosen having the highest number of degrees of freedom:
In this case, a Maximum Limit was computed for the maximum 1st principal stress within the selected element faces (1.915e5 psi) and thus the Run 8 estimated error is calculated to equal 0.05% (100*(1.915e5-1.914e5)/1.914e5)).
Note: Generally, the solution can be considered to be of good quality if the error measured in energy norm is small and the location and magnitude of the minimal or maximal values of the data of interest clearly converge to a limiting value.
Note: The computation of the estimated limit is performed as follows:
These should be understood as heuristic procedures which tend to work well when the errors in the data are small, but may give a false indication of the limit when the errors are large. The procedure may compute a limit even when the limit does not exist or fail to compute a limit even when one does exist.
The following are additional checks that provide qualitative (error indicator) and qualitative (strain energy error) assessments of the solution but are more expensive to compute in real-time.
If the “Indicator” toggle is selected, an error indicator based on jumps in the stress resultants is computed for an element or group of elements. The error indicator for the “k”th element provides an indication of the relative influence of the element on the global error of the solution.
To compute the error indicator, the program first determines the jumps in the stress resultants across any shared edges (2D) or faces (3D). By definition, the jump (Δ) is the sum of the absolute values of the differences in stress resultants. An element (k) has as many Δ’s as there are shared edges/faces. For example, the jump for shared edge j of element k for a planar (2D) problem is given by:
Note: you can enter a value between 0 and 1 in the Criteria input field, and the program will report only those elements with an error indicator exceeding that value.
To plot the error indicator as outlined above when the Error tab’s “Indicator” toggle is active, select the function “Error” in the Plot tab, select an element or group of elements and click Plot:
This is a visual representation of the same data generated in the Error tab when “Indicator” is activated.
Note: elements which have a high error indicator my indicate a discretization or modeling error, but the local data may still converge so this should be verified independently.
The strain energy error is what is computed in the Results Error tab, for “Select > Any Element”, “Estimate” toggle is active and an element or group of elements is selected. The strain energy (U) is computed from the following expression:
U = ½∫{ε}{σ}dV
Where where {ε} is the mechanical strain tensor, {σ} is the stress tensor and the integral is performed over each element:
For the sequence of solutions selected, the results will include the run #, the unconstrained degrees of freedom (U_DOF), the strain energy, convergence rate and % error associated with each run #, and the estimated limit of the strain energy.
The strain energy error can be useful for identifying an element or elements that most contribute to the global error of the solution, and to identify if the strains or stresses of an element or elements is stable, needs refinement or contains a singularity.
Each Solution ID requires a unique Load ID and Constraint ID. Some users wish to combine the results of two or more load cases together, such as axial/bending or mechanical/thermal load cases, and this may be accomplished via the StressCheck® Calculator (Functions: Calc):
The StressCheck® Calculator can be accessed in the Plot, Min/Max and Points tabs under “Functions: Calc”.
Up to five different solutions may be combined, each “Solution:” representing a Solution ID of a Load ID/Constraint ID. The Operand allows the Solution ID and Run to be converted into an integer specification.
A formula may be entered to represent the superposition of displacements, stresses or strains. For example, the above represents the addition of the stresses in the global X-dir for the 8th runs of Solution ID’s “SOLAXIAL” and “SOLBENDING”.
Here are some examples of combining load case results in the StressCheck® Calculator:
By default, displaying the load arrows for a selected load ID will display the load arrows for every load record within the load ID. Additionally, the load arrows will be scaled automatically to appropriate sizes based on the current model view, and the relative lengths of the load arrows will be with respect to the applied load magnitudes in the load records.
The below provides tips and tricks for displaying load arrows for a selected load ID.
If multiple load ID’s are defined in the model, and we wish to view a particular load ID, we can use the “Display Loads/Flux” drop down (above image). The current load ID will have a blue circle.
This is the easiest way to quickly cycle between the load arrow displays of each load ID in the model.
The relative sizes of the load arrow displays for the selected load ID records are controlled via the “Scale” box in the Load tab. The “Scale” value can be automatically generated (default) or user-defined.
Note: the “Scale” value has no effect on the magnitudes of the applied loads.
A model may contain multiple load records within a load ID, and we may want to display only one load record within the load ID to display its load arrow magnitudes and directions.
To only display the load arrows for a single load record, we recommend the following:
This process ensures that the load arrow displays for an assignment can be isolated for easier viewing.
In 2D/3D elasticity, it is necessary to convert a bending moment into a linear bending stress distribution. From StruCalc:
In StressCheck®, this is usually done via a formula expression using the section properties SPMOMX (moment of inertia about centroidal X-axis) or SPMOMY (moment of inertia about centroidal Y-axis) to represent “I” in the above equation’s denominator, and assigned as a Traction using a coordinate system with a Z-axis normal to a set of element faces or surfaces representing a flat cross section.
Here’s an example of Mx=1000 lbf-in converted to a bending stress (in psi) on a symmetric I-section using the formula expression “BENDING”:
The formula is then assigned to the flat cross sectional surfaces/element faces of the I-section as a normal traction distribution, using the appropriate system and direction:
In the above example, the selected local system (SYS1) is assumed to have an origin at the centroid of the cross section. If the local system is NOT located at the centroid of the cross section but it is aligned with its principal axes, the above equation may simply be modified to be “BENDING = 1000*(y-SPCTRY)/SPMOMX”, where SPCTRY is the y-coordinate of the centroid with respect to the local system.
Below is an example of modifying the bending stress expression for a coordinate system (SYS2) NOT located at the centroid of the symmetric I-beam cross section:
Note: for symmetric cross sections the principal axes are along the symmetry planes
In general, or if you cannot guarantee the local system you select is aligned with the cross section principal directions, the cross moment of inertia may be present. To consider it use:
In StressCheck®, for Mx=1000 lbf-in, the above formula would take the form of the following expression:
BENDING = (-(0*SPMOMX+1000*SPCROSS)*(x-SPCTRX)+(1000*SPMOMY+0*SPCROSS)*(y-SPCTRY))/(SPMOMX*SPMOMY-SPCROSS^2)
The section property intrinsic functions available for use in formula expressions to automatically compute the area (SPAREA) and other section properties of an assigned set of surfaces/element faces are below:
Here is a video tutorial of the definition and assignment of a bending stress:
StressCheck® Tutorial: Defining and Assigning a Bending Stress
When assigning material properties, boundary conditions or other model attributes, it is best to assign to geometric entities (points, curves or surfaces) instead of mesh entities (nodes, element edges or element faces). This is because StressCheck®‘s elements are designed to inherit assignments to geometry through associativity. Automeshed elements are always geometrically associative, while hand-meshed elements will be geometrically associative if the nodes are explicitly associated to the geometry by the user.
In this way, if a model is remeshed, and the elements are renumbed, any assignment to the geometry will persist and be automatically applied to the new mesh. Here’s an example of using geometric association to persist model assignments:
StressCheck Tutorial: Associative Hand Meshing for a Cylindrical Part
The moment summation values for each point load case definition (Edit > Point Load Info…, select a case from the Current Case drop down) will include the summation of moment magnitudes for each moment vector (Mx, My, Mz) as well as moments of the forces (Fx, Fy, Fz) with respect to (0, 0, 0).
For example, instead of simply computing Mx = ΣMx0, the summation value will report Mx = Σ(Mx0 + Fz × Y – Fy × Z), where the summation is performed for all the data points in the table, Mx0 is the value of the moment-X for each entry in the table, Fy and Fz are the values of the Y- and Z-components of the forces, and Y and Z are the coordinates of the points. Analogously expressions are used for the other moment components.
The below case definition shows that the sum of forces is Fx=0, Fy=-16, Fz=0, and the sum of the moments is Mx=0, My=2, Mz=0:
Sample case definition table.
The force summation value for Fy: ΣFy = -1 + -2 + -3 + -4 + -3 + -2 + -1 = -16
The moment summation value for Mx(@ 0, 0 ,0): Σ(Mx0 + Fz × Y – Fy × Z) = (0 – 0*7.5 + 1*0.125) + (0 – 0*7.5 + 2*0.125) + (0 – 0*7.5 + 3*0.125) + (0 – 0*7.5 + 4*0.125) + (0 – 0*7.5 + 3*0.125) + (0 – 0*7.5 + 2*0.125) + (0 – 0*7.5 + 1*0.125) = 2
Bearing loads are defined as sinusoidal traction distributions applied to cylindrical surfaces as force vectors (Fx and Fy), or as a force magnitude (F) and angle (α). Bearing loads are defined in the Load tab via Select > Any Surface/Face/Face Surface > Bearing.
In order to assign a 3D bearing load to a cylindrical surface (i.e. hole), a local system in the center of the hole with the Z-axis directed along the center of the cylinder must be available and specified in the System: combo-box or by the System Picker during the bearing load assignment.
The below animated example shows a local system acceptable for the assignment of a 3D bearing load to a cylindrical surface. The System Picker 
was used to first select the local system, then the cylindrical surface was selected. Finally, the Accept button was clicked:
The following are additional expectations and requirements for applying 3D bearing loads to cylindrical surfaces:
StressCheck Tutorial: Autocorrect for Bearing and TLAP-Bearing Loads
If a local system meeting the above requirements does not yet exist, there are several ways to generate this system such that it is located in the center of the cylindrical surface. After creation of a local system, the system location and orientation can always be edited (Geometry tab > Edit > System, select the system).
The user may indirectly generate a local system by creating a wireframe circle by the Sample method (Geometry tab > Curve selector 
> Create > Circle > Sample).
> Select > Circle > Locate, left click on the wireframe circle) and clicking the Delete button.
The user may indirectly general a local system by creating a cylindrical surface by the Sample method (Geometry tab > Surface selector 
> Create > Circle > Sample, Surface toggle 
).
> Select > Cylinder > Locate, left click on the cylindrical surface) and clicking the Delete button.
Note that the bearing load must be assigned to the cylindrical feature (e.g. hole) instead of the sampled cylindrical surface. Additionally, the cylindrical feature can be edited (e.g. modify the Radius: input field to be larger than the cylindrical feature) and used in a subsequent solid modeling operation.
For more information on 3D bearing loads:
When is TLAP-Traction applicable to my model? How can the results be affected if I’m too close to my data of interest?
When using TLAP-Traction, there must be sufficient distance between the area of load application and the region of interest such that the data in the region of interest is unaffected by this distance.
Here are some resources on TLAP-Traction for Global-Local Analysis:
As of StressCheck® v10.4, users can select from two different TLAP-Bearing options: Default and IMO (Ignore Moments & Offsets).
The main difference between the two options is that the Default option will include traction distributions due to input moments as well as offset corrections (due to moving the TLAP to the center of the hole), whereas the IMO option will ignore these traction distributions.
This option ensures that all forces and moments are kept by generating a statically equivalent bearing traction distribution. On assignment the force and moment components are rotated and translated to a hole centered system:
TLAP-Bearing Default Option.
This option removes input moments and moments generated from offsets (Ignore Moments and Offsets “IMO”). It will generate a TLAP-Bearing distribution that ignores the original point load moment components and the moment contribution caused by the offset distance between the location of the point load and the center of the hole. Note that this option will NOT generate a statically equivalent system.
TLAP-Bearing IMO Option.
More information on the TLAP-Bearing options:
In StressCheck®, there are two TLAP-Traction options for applying one or more Total Loads at a Point (TLAPs) on selected element faces or geometric surfaces to generate a statically equivalent set of tractions: Near Faces and All Faces.
For either TLAP-Traction option, each selected TLAP (representing point force/moment vectors at global X, Y, Z locations) is converted into tractions distributed over the faces of the solid elements such that they are statically equivalent to the corresponding TLAP.
The contribution of each TLAP to element faces is controlled by the selected TLAP-Traction option, and very different yet statically equivalent traction distributions can result. Note: regardless of TLAP-Traction option, the tractions in the region of TLAP-Traction application will be discontinuous as there is no requirement for continuity in stresses.
All selected TLAPs will be distributed to all selected element faces. From the Master Guide Global-Local Analysis section, the following details the algorithm used to produce the statically equivalent tractions when TLAP-Traction All Faces option is selected:
In the case of TLAP-Traction All Faces option, each selected element faces gets an area-proportionate “share” or contribution from the selected TLAPs. As a result, large corrective tractions can occur if a TLAP is far from the selected element face.
This option is typically recommended when the number of TLAPs is much less than the number of selected element faces, as it will ensure each element faces receives a proportionate contribution of each TLAP.
Each TLAP applied to only the element face closest to that TLAP. That is, the algorithm detailed above will assume that the element face closest to each TLAP will get 100% of the “share” from that TLAP.
This option is typically recommended when the number of TLAPs is greater than or similar to the number of selected element faces.
More Information on TLAP-Traction Options:
If the load check (Check->Any Record, Check->Any Element or Check->All Elements in the Load tab) returns an incorrect summation for a TLAP-Bearing load, what are some causes and how to fix?
The most likely cause in this case is a hole that passes the taper tolerance (default of 0.05), but is not a constant enough height to allow for the proper representation of the bearing distribution. TLAP-Bearing and Bearing loads require a constant thickness hole for proper representation.
The load summation in this case may depend on the clocking of the bearing load with respect to the hole thickness. The best solution is to either a) split the surface by imprinting curves in such a way as to produce a constant height cylindrical surface, and use this surface for the TLAP-Bearing or b) define a formula bearing load.
Another possible cause is that the mesh is not refined enough to represent the sinusoidal stress distribution. Adding local refinement to the hole will usually fix this problem.
When applying a Bearing or TLAP-Bearing load, some formulae are automatically generated to define a normal traction distribution on a cylindrical hole. The normal traction distribution superimposes (1) normal traction terms representing in-plane loads and (2) normal traction terms representing bending moments. The formulae that are generated include a conditional statement that checks the value of the superimposed traction components (1) + (2) that are distributed around the bore of the hole. When the value of the superposition is negative (indicating a compressive pressure load on the hole bore), the traction is applied to the hole bore. When it is positive (indicating a positive or tensile normal traction), then no traction is applied. This is because the Bearing load implemented in StressCheck® is based on representing the contact conditions of a neat-fit pin bearing against the bore of a hole.
In practice, the in-plane force due to pin is often much larger in magnitude than the bending moment, and the conditional statement in the bearing traction formulae works as expected. Whether using the “Default” or “IMO” TLAP-Bearing option, the in-plane force components (Fx and Fy) are unaffected.
However, when the applied moments are much larger than the in-plane force magnitudes (and particularly when the loaded hole has a short height compared to its diameter), there is a possibility that the in-plane force components can become skewed. In these cases, it is recommended to manually define a formulaic bearing load + bending traction distribution, or if the area of interest is of sufficient distance from the loaded hole, to apply the force/moments as a TLAP-Traction.
For more details read the TLAP Bearing Technical Brief.
As of the StressCheck® v10.5 release, an autocorrect feature is now available to ensure the force/moment magnitudes are resolved. This feature overcomes the limitations of many of the above cases:
StressCheck Tutorial: Autocorrect for Bearing and TLAP-Bearing Loads
There are several possibilities why the TLAP-Bearing load cannot be applied:
The hole contains features that cause it to no longer be classified as cylindrical (notches, intersecting holes, pressure release, etc.).
Note: introducing the tolerance parameters is no guarantee to fix the problem, as the geometry may not be valid for proper application of TLAP-Bearing or Bearing loads. For more information, consult the Master Guide section on “TLAP-Bearing” and read the TLAP Bearing Technical Brief.
For certain objects, the integral average of one or more engineering functions (e.g. Sz in the Global coordinate system) over the length/area of the selected object(s) may be computed by enabling the Average button in the Point extraction tab:
The integral average of the stress in the global Z (Sz) is computed over the selected element faces.
The user may select a solution ID, specify the run numbers, select one or more engineering functions, enable the Average button, and click Accept. A graph window will appear with the integral average computation for the engineering function (or functions) over the selected path/area.
The integral average of any supported engineering function can be computed for element edges, element faces, elements volumes, and for geometric curves. Other object types (points, nodes, surfaces) are not enabled for Average extractions.
To compute an engineering function’s arithmetic mean value for a selected geometric surface (or surfaces), the following is recommended:
Cylindrical Surface with P1 Min/Max and P2 Min/Max inputs.
In general the range when entering Offset, P1 (Min/Max) and P2 (Min/Max) inputs for object generation, or object boundary extractions in the Points tab, is typically normalized to [0,1]. However for some geometries, such as spheres, cylinders and cones, the P1 (Min/Max) and/or P2 (Min/Max) inputs may be conveniently entered in degrees. The values in the Offset, P1 and P2 inputs can always be edited and updated (either with constants, parameters or formulae) after object generation.
Curve selector enabled (yellow filled circle) for generation of wireframe geometry.
Wireframe geometries are characterized by a single parameter, P1. Here are the expected ranges (and defaults) when using Offset or P1 for select wireframe geometries:
For additional wireframe geometries, consult the Master Guide Chapter 3: 2D Geometry Creation.
Surface selector enabled (green wave) for generation of surface/solid geometry.
Surface geometries are characterized by two parameters, P1 and P2. By modifying the values in the P1 and P2 inputs, surface geometries can be limited to “patches”:
Representation of a spherical surface patch (courtesy StressCheck® Master Guide).
Here are the expected ranges (and defaults) when using P1 and P2 for select surface geometries:
For additional surface geometries, consult the Master Guide Chapter 4: 3D Geometry Creation.
When placing explicitly associated objects on surface boundaries via Offset (e.g. Create > Node > Offset), or checking a surface boundary Offset via Check > Any Boundary > Offset, the Offset ranges will normalized to 0≤Offset≤1.
The below demonstrates how an Offset=0.25 on a surface boundary represents +90 degrees in the circumferential direction of the cylindrical surface:
Creation of an offset local system at Offset=0.25 on cylindrical surface curve.
Note: A range of 0≤Offset≤1 is anticipated even for surfaces generated with P1 and/or P2 ranges in degrees. This is because these surfaces are internally represented by Parasolid with normalized parameter spaces of [0,1]. However, editing of these objects will return the original inputs for P1 and/or P2 in degrees.
Extractions using “Select > Any Boundary” in the Points tab of Results can be limited to the ranges specified by P1 & P2. For example, if we wish to extract in the range [0,0.5] for P1 and P2 then only 1/4 of the surface will be considered for the point extraction:
Point extraction on a surface for a limited P1 and P2.
Similarly, if a wireframe geometry or surface curve is selected, then P1 will control the range of the extraction. For more information, consult the Master Guide Chapter 5: Post-Processing Operations.
When assigning material properties, boundary conditions or other model attributes, it is best to assign to geometric entities (points, curves or surfaces) instead of mesh entities (nodes, element edges or element faces). This is because StressCheck®‘s elements are designed to inherit assignments to geometry through associativity. Automeshed elements are always geometrically associative, while hand-meshed elements will be geometrically associative if the nodes are explicitly associated to the geometry by the user.
In this way, if a model is remeshed, and the elements are renumbed, any assignment to the geometry will persist and be automatically applied to the new mesh. Here’s an example of using geometric association to persist model assignments:
StressCheck Tutorial: Associative Hand Meshing for a Cylindrical Part
In general, the contact constant should be selected based on the moduli of elasticity of the materials in contact, in such a way that its value is between 0.1 and 1.0 times the value of the smallest modulus of elasticity (0.1Emin < Kc < Emin) when using US units (psi), and between 0.004 and 0.04 times the value of the smallest modulus of elasticity (0.004Emin < Kc < 0.04Emin) when using SI units (MPa). The units for the contact constant are [F/L/L^2].
For aluminum or steel, this is usually 1-10 million psi/in or so, assuming significant operational loads.
Helpful Hints and Tips: Multi-Body Contact Overview and General Recommendations
Some other guidelines:
It’s best to use a parameter for the contact constant, so it can be updated based on the applied load magnitude. The idea is that if the applied loads produce significant stress, the contact constant should be increased relative to the load. If the applied loads don’t produce significant stress, the contact constant should be lowered.
When using the Frictionless Augmented Lagrangian Method (StressCheck®‘s contact algorithm), the contact constant is multiplied by the gap function to determine the change in contact pressures, which are then added to the previous iteration’s pressures, so it should not be a small value (relative to the loads/material). It would then take many iterations to stabilize the contact pressures.
From the blog article: https://www.simscale.com/blog/2016/09/contact-mechanics-friction/
Augmented Lagrange Method was formulated to bring the good of both penalty and Lagrange multiplier. It does not demonstrate an oscillatory behavior and has minimal interpenetration. The change in energy contribution for a small change in gap can be given as:
(previous pressure + contact coefficient * gap) x change in gap
Constraining/fixing a point or node should ONLY be used when the model still has a free rigid body mode (i.e. translation or rotation). Equilibrium of the model under the applied loads must be considered before constraining/fixing a point or node.
No node constraints needed (symmetry on three orthogonal sides)
Fixed Z-direction node constraint needed (rigid body translation for Z-axis)
Rigid body constraints needed (self-equilibrated load)
Note: If constraints do not prevent all free rigid body motion (zero-strain translations and/or rotations) then the model is underconstrained and cannot be solved.
Note: contact constraints are simply normal, grounded distributed springs (in units of F/L/L^2) within the assigned contact zones. Therefore, a single multi-body contact assignment in 3D controls up to 4 rigid body modes per part:
Free rigid body modes must be controlled by tangent springs.
If the model is self-equilibrated under the applied loads, then the Rigid Body Constraint method may be used (Constraint tab, Select > Node/Point > Rigid Body). No other constraints are required.
Note: For the model to be self-equlibrated under the applied loads, the load check must return relatively small forces/moments (Load tab, Check > All Elements > Selection, enter the load ID and click Accept). The computed total forces/moments may not be exactly zero due to accumulation of small numerical errors.
If the model is NOT self-equilibrated under the applied loads, other constraints must exist that provide reactions for the applied loads (e.g. Symmetry Constraints or Spring Constraints). Such constraints may be used to provide a reaction and prevent rigid body motion in one direction or in all directions. For example, if the applied resultant forces are in equilibrium in the X direction but not the Y direction, then an additional constraint should only be used to react the Y direction force; one or more Nodal Constraints may be used to prevent any remaining rigid by motions (Constraint tab, Select > Node/Point > Node).
Note: in 2D/3D elasticity, nodes and points should NOT react load! They should only be used to prevent rigid body motion, thus setting a reference for the elasticity solution. Weak normal or tangent springs may also be used if a node or point location is not appropriate for cancellation of rigid body motion.
The material property database (accessed by clicking the “Browse…” button in the Material tab, Define subtab) may be manually expanded by modifying the material.dat file found in the StressCheck® installation directory (i.e. C:\Program Files\ESRD\StressCheck®
PE 10\UI\material.dat). Note: administrative privileges may be required to update this file.
A description of the format of the material properties listed in the material.dat file is as follows:
Six nonlinear coefficients: v, E1, E2, A1, A2, A3. Where v is the Poisson’s ratio, E1 is the modulus of elasticity and the other coefficients depend on the stress-strain law:
NewAluminum, ALUMINUM, ISOTROPIC, Ramberg-Osgood, US <-Record 1
0.107e8, 0.395, 0.128e-4, 0.101. 0.262e-3 <- Record 2
0.395, 0.107e8, 0.60e5, 18, 0.0, 0.0 <- Record 3
User-defined material <- Record 4
In the above example, the elastic modulus is 10.7e6 psi, Poisson’s ratio 0.395, S70E is 60 ksi, and the exponent “n” is 18. Watch below for how the above example is added to the material database:
The stress-strain curve can be plotted for nonlinear materials in the Material tab, Assign sub-tab by clicking Plot:
By default, 24 points are displayed for the stress-strain curve fitting, which may not be sufficient for some nonlinear material curves.
You can increase the points generated in the stress-strain curve by defining two underscore parameters:
For example, set _NL_COUNT to 100 and _NL_MAX to 0.1, and the curve will be plotted using 100 points in the strain range from 0 to 0.1. However, note that these parameters have no effect on the solution itself.
Note: This technique also increases the # points displayed in the property vs temperature material curve for Temperature-Dependent materials.
In many cases, CAD geometry may be imported in an undesired location/orientation, and the geometry needs to be moved to a new location/orientation for your StressCheck® analysis. In this case, there is a relatively simple workflow for doing so:
Now, only the “copy” CAD Geometry remains, in the desired location/orientation for the analysis.
For more information on this process:
StressCheck® Professional can natively import Parasolid files of versions ranging 10.0 to 26.0.151. Other CAD formats require licensed translators.
The following CAD formats and versions are supported in StressCheck® Professional’s 3D Interop R25 translators:
| Formats | File Formats | Product Structure | Graphical | Geometry | Semantic PMI |
| 3DXML | .3dxml | v4.3 | v4.3 | ||
| ACIS | .sat, .sab, .asat, .asab | R1 – R24 | R1 – R24 (Generated from Geometry) | R1 – R24 | N/A |
| CATIA V4 | .model, .exp, .session | 4.1.9 – 4.2.4 | 4.1.9 – 4.2.4 (Generated from Geometry) | 4.1.9 – 4.2.4 | N/A |
| CATIA V5 | .CATPart, .CATProduct, .CGR | R8 – R24 (V5–6R2014) | R8 – R24 (V5–6R2014) | R8 – R24 (V5–6R2014) | R8 – R24 (V5–6R2014) |
| CATIA V6 | .CATPart, .CATProduct, .CGR | V6R2014x | V6R2014x | V6R2014x | V6R2014x |
| DXF/DWG | .dxf, .dwg | 2.5 – 2014 | 2.5 – 2014 (Generated from Geometry) | 2.5 – 2014 | N/A |
| IGES | .igs, .iges | Up to 5.3 | Up to 5.3 (Generated from Geometry) | Up to 5.3 | N/A |
| Inventor | .ipt (V6 – V2015) .iam (V11 – V2015) |
V11 – 2015 | V7 – 2015 V6 (Generated from Geometry) |
V6 – V2015 | N/A |
| NX | .prt | 11 – NX 9 | NX 6 – NX 9 11 – NX 5 (Generated from Geometry) |
11 – NX 9 | 11 – NX 9 |
| NX Direct | .prt | NX 1 – NX 9 | NX 6 – NX 9 NX 1 – NX 5 (Generated from Geometry) |
NX 1 – NX 9 | NX 1 – NX 9 |
| Parasolid | .x_t, .xmt_txt, .x_b, .xmt_bin | 10.0 – 26.0.151 | 10.0 – 26.0.151 (Generated from Geometry) | 10.0 – 26.0.151 | N/A |
| Parasolid Direct | .x_t, .xmt_txt, .x_b, .xmt_bin | 10.0 – 26.01.211 | 10.0 – 26.01.211 (Generated from Geometry) | 10.0 – 26.01.211 | N/A |
| Pro/E / Creo | .prt, .prt.*, .asm, .asm.* | 16 – Creo 2.0 | WF3 – Creo 2.0 16 – WF2 (Generated from Geometry) |
16 – Creo 2.0 | 16 – Creo 2.0 |
| Solid Edge | .par, .asm, .psm | V18 – ST6 | ST– ST6 V18 – V20 (Generated from Geometry) |
V18 – ST6 | N/A |
| Solid Edge Direct | .par, .asm, .psm | V18 – ST6 | ST– ST6 V18 – V20 (Generated from Geometry) |
V18 – ST6 | N/A |
| SolidWorks | .sldprt, .sldasm | 98 – 2014 | 98 – 2014 | 98 – 2014 | N/A |
| SolidWorks Direct | .sldprt, .sldasm | 2003 – 2014 | 2003 – 2014 | 2003 – 2014 | N/A |
| STEP | .stp, .step | 203, 214 | 203, 214 (Generated from Geometry) | 203, 214 | N/A |
| STL | .stl | N/A | All | N/A | N/A |
| VDA-FS | .vda | 1.0 – 2.0 | N/A | 1.0 – 2.0 | N/A |
| XCGM | R2012 – R2014x | R2012 – R2014x | R2012 – R2014x | R2012 – R2014x |
Note: CATPart files created in V5R1 that have been opened and re-saved in a later version of CATIA V5 are not supported for 3D InterOp Graphical and 3D InterOp CGM.
Note: CATIA V6 users should export their database objects as CATIA V5 CATParts, CATProducts, XCGM, or as 3D XML, which can then be imported into applications using InterOp.
Note: NX .prt extensions must be changed to .ug to avoid conflict with ProE files.
StressCheck® uses Parasolid as its native CAD kernel. The native format is .x_t, or the Parasolid Transmit format. StressCheck® integrates multiple CAD translator options to convert from IGES, CATIA, Pro/E and others to Parasolid format. These modules are sold separately.
Cylindrical Surface with P1 Min/Max and P2 Min/Max inputs.
In general the range when entering Offset, P1 (Min/Max) and P2 (Min/Max) inputs for object generation, or object boundary extractions in the Points tab, is typically normalized to [0,1]. However for some geometries, such as spheres, cylinders and cones, the P1 (Min/Max) and/or P2 (Min/Max) inputs may be conveniently entered in degrees. The values in the Offset, P1 and P2 inputs can always be edited and updated (either with constants, parameters or formulae) after object generation.
Curve selector enabled (yellow filled circle) for generation of wireframe geometry.
Wireframe geometries are characterized by a single parameter, P1. Here are the expected ranges (and defaults) when using Offset or P1 for select wireframe geometries:
For additional wireframe geometries, consult the Master Guide Chapter 3: 2D Geometry Creation.
Surface selector enabled (green wave) for generation of surface/solid geometry.
Surface geometries are characterized by two parameters, P1 and P2. By modifying the values in the P1 and P2 inputs, surface geometries can be limited to “patches”:
Representation of a spherical surface patch (courtesy StressCheck® Master Guide).
Here are the expected ranges (and defaults) when using P1 and P2 for select surface geometries:
For additional surface geometries, consult the Master Guide Chapter 4: 3D Geometry Creation.
When placing explicitly associated objects on surface boundaries via Offset (e.g. Create > Node > Offset), or checking a surface boundary Offset via Check > Any Boundary > Offset, the Offset ranges will normalized to 0≤Offset≤1.
The below demonstrates how an Offset=0.25 on a surface boundary represents +90 degrees in the circumferential direction of the cylindrical surface:
Creation of an offset local system at Offset=0.25 on cylindrical surface curve.
Note: A range of 0≤Offset≤1 is anticipated even for surfaces generated with P1 and/or P2 ranges in degrees. This is because these surfaces are internally represented by Parasolid with normalized parameter spaces of [0,1]. However, editing of these objects will return the original inputs for P1 and/or P2 in degrees.
Extractions using “Select > Any Boundary” in the Points tab of Results can be limited to the ranges specified by P1 & P2. For example, if we wish to extract in the range [0,0.5] for P1 and P2 then only 1/4 of the surface will be considered for the point extraction:
Point extraction on a surface for a limited P1 and P2.
Similarly, if a wireframe geometry or surface curve is selected, then P1 will control the range of the extraction. For more information, consult the Master Guide Chapter 5: Post-Processing Operations.
When using the “Create” action in the Geometry or Mesh tab, there are multiple ways to create new objects:
It’s possible more than one of these actions was invoked during an object creation. Note: left-click of the mouse in the Model Viewing window will always perform the current action in the “Action” combo (e.g. Create, Select, Edit).
As a best practice, if focus is required in the Model Viewing window, and a new object creation is NOT desired, it is recommended to right-click in the Model Viewing window to establish the window’s focus such that the model can be rotated, zoomed, etc. This can help with reducing duplication of geometry and mesh objects.
If you suspect there are duplicate geometry and/or mesh objects, changing the combos to “Select > Any Object” and clicking on the “Index” subtab within the Geometry or Mesh tabs will display a list of the objects in the model:
If there are duplicates, you may simply select the object(s) from the Index list and click the “Delete” button, or simply click the “DeLast” button if the last object created was a duplicate.
Duplicate nodes may be inadvertently generated when manually meshing models. If nodes are duplicated, clicking the Mesh tab’s “Merge” button will attempt to merge all duplicated nodes within a tolerance of 10e-6 model units. All but one of the coincident nodes are discarded, and all element definitions are updated appropriately. Note: care must be taken for multi-body contact models with coincident parts, as well as fracture models with intentional cracks, as these nodes may be merged.
Alternatively, the user may first determine if any free edges/disconnected elements exist in the mesh by using “Check > Edge > Free Edge” within the Mesh tab. Element edges which are free/not shared will be highlighted in red. Then, the user may select only those nodes which are duplicated (either via marquee selection or Shift + left-click) and click the Merge button to limit the merge to those nodes.
Note: to override the merge tolerance, the user may create the parameter _merge_tol and supply a new merge tolerance.
The error message likely occurred because of an Explicit Associativity in the model. This is a dependency created by the user in which an object definition references another object or objects in the model.
These messages typically occur when attempting to perform a modeling operation such as a Boolean-Union/Subtraction, Clip-Back/Front, Blend Edge, Imprint, Spin or Thicken.
An Explicit Associativity will exist if one or more objects (e.g. points, systems, nodes) are created by an associative method (e.g. Offset, Intersection, Projection) via a user-defined reference to the target geometry. Examples are creating a node by the Offset method on the geometry’s curve, or creating a system by 3-Pt. Plane method using points at the geometry’s intersections (vertices).
There are two types of Associativity: Implict and Explicit. They are described in the following:
Implicit Associativity – The result of a boolean or blend operation is a collection of trimmed surfaces, trimmed curves and points. The trimmed curves represent the intersections between neighboring surfaces. The points represent the intersections between trimmed curves.
Explicit Associativity – It is sometimes necessary to create new explicit associative relationships in a model.
Associativity is extremely valuable because it allows for fully parametric modeling, object inheritance (i.e. surface to element face communication) and highly curved elements (mapping).
The importance of Implicit and Explicity Associativity, and how to determine object associativities and even disassociate points and systems from geometries, is explained in the following tutorial:
In many cases, CAD geometry may be imported in an undesired location/orientation, and the geometry needs to be moved to a new location/orientation for your StressCheck® analysis. In this case, there is a relatively simple workflow for doing so:
Now, only the “copy” CAD Geometry remains, in the desired location/orientation for the analysis.
For more information on this process:
The moment summation values for each point load case definition (Edit > Point Load Info…, select a case from the Current Case drop down) will include the summation of moment magnitudes for each moment vector (Mx, My, Mz) as well as moments of the forces (Fx, Fy, Fz) with respect to (0, 0, 0).
For example, instead of simply computing Mx = ΣMx0, the summation value will report Mx = Σ(Mx0 + Fz × Y – Fy × Z), where the summation is performed for all the data points in the table, Mx0 is the value of the moment-X for each entry in the table, Fy and Fz are the values of the Y- and Z-components of the forces, and Y and Z are the coordinates of the points. Analogously expressions are used for the other moment components.
The below case definition shows that the sum of forces is Fx=0, Fy=-16, Fz=0, and the sum of the moments is Mx=0, My=2, Mz=0:
Sample case definition table.
The force summation value for Fy: ΣFy = -1 + -2 + -3 + -4 + -3 + -2 + -1 = -16
The moment summation value for Mx(@ 0, 0 ,0): Σ(Mx0 + Fz × Y – Fy × Z) = (0 – 0*7.5 + 1*0.125) + (0 – 0*7.5 + 2*0.125) + (0 – 0*7.5 + 3*0.125) + (0 – 0*7.5 + 4*0.125) + (0 – 0*7.5 + 3*0.125) + (0 – 0*7.5 + 2*0.125) + (0 – 0*7.5 + 1*0.125) = 2
When is TLAP-Traction applicable to my model? How can the results be affected if I’m too close to my data of interest?
When using TLAP-Traction, there must be sufficient distance between the area of load application and the region of interest such that the data in the region of interest is unaffected by this distance.
Here are some resources on TLAP-Traction for Global-Local Analysis:
As of StressCheck® v10.4, users can select from two different TLAP-Bearing options: Default and IMO (Ignore Moments & Offsets).
The main difference between the two options is that the Default option will include traction distributions due to input moments as well as offset corrections (due to moving the TLAP to the center of the hole), whereas the IMO option will ignore these traction distributions.
This option ensures that all forces and moments are kept by generating a statically equivalent bearing traction distribution. On assignment the force and moment components are rotated and translated to a hole centered system:
TLAP-Bearing Default Option.
This option removes input moments and moments generated from offsets (Ignore Moments and Offsets “IMO”). It will generate a TLAP-Bearing distribution that ignores the original point load moment components and the moment contribution caused by the offset distance between the location of the point load and the center of the hole. Note that this option will NOT generate a statically equivalent system.
TLAP-Bearing IMO Option.
More information on the TLAP-Bearing options:
In StressCheck®, there are two TLAP-Traction options for applying one or more Total Loads at a Point (TLAPs) on selected element faces or geometric surfaces to generate a statically equivalent set of tractions: Near Faces and All Faces.
For either TLAP-Traction option, each selected TLAP (representing point force/moment vectors at global X, Y, Z locations) is converted into tractions distributed over the faces of the solid elements such that they are statically equivalent to the corresponding TLAP.
The contribution of each TLAP to element faces is controlled by the selected TLAP-Traction option, and very different yet statically equivalent traction distributions can result. Note: regardless of TLAP-Traction option, the tractions in the region of TLAP-Traction application will be discontinuous as there is no requirement for continuity in stresses.
All selected TLAPs will be distributed to all selected element faces. From the Master Guide Global-Local Analysis section, the following details the algorithm used to produce the statically equivalent tractions when TLAP-Traction All Faces option is selected:
In the case of TLAP-Traction All Faces option, each selected element faces gets an area-proportionate “share” or contribution from the selected TLAPs. As a result, large corrective tractions can occur if a TLAP is far from the selected element face.
This option is typically recommended when the number of TLAPs is much less than the number of selected element faces, as it will ensure each element faces receives a proportionate contribution of each TLAP.
Each TLAP applied to only the element face closest to that TLAP. That is, the algorithm detailed above will assume that the element face closest to each TLAP will get 100% of the “share” from that TLAP.
This option is typically recommended when the number of TLAPs is greater than or similar to the number of selected element faces.
More Information on TLAP-Traction Options:
If the load check (Check->Any Record, Check->Any Element or Check->All Elements in the Load tab) returns an incorrect summation for a TLAP-Bearing load, what are some causes and how to fix?
The most likely cause in this case is a hole that passes the taper tolerance (default of 0.05), but is not a constant enough height to allow for the proper representation of the bearing distribution. TLAP-Bearing and Bearing loads require a constant thickness hole for proper representation.
The load summation in this case may depend on the clocking of the bearing load with respect to the hole thickness. The best solution is to either a) split the surface by imprinting curves in such a way as to produce a constant height cylindrical surface, and use this surface for the TLAP-Bearing or b) define a formula bearing load.
Another possible cause is that the mesh is not refined enough to represent the sinusoidal stress distribution. Adding local refinement to the hole will usually fix this problem.
When applying a Bearing or TLAP-Bearing load, some formulae are automatically generated to define a normal traction distribution on a cylindrical hole. The normal traction distribution superimposes (1) normal traction terms representing in-plane loads and (2) normal traction terms representing bending moments. The formulae that are generated include a conditional statement that checks the value of the superimposed traction components (1) + (2) that are distributed around the bore of the hole. When the value of the superposition is negative (indicating a compressive pressure load on the hole bore), the traction is applied to the hole bore. When it is positive (indicating a positive or tensile normal traction), then no traction is applied. This is because the Bearing load implemented in StressCheck® is based on representing the contact conditions of a neat-fit pin bearing against the bore of a hole.
In practice, the in-plane force due to pin is often much larger in magnitude than the bending moment, and the conditional statement in the bearing traction formulae works as expected. Whether using the “Default” or “IMO” TLAP-Bearing option, the in-plane force components (Fx and Fy) are unaffected.
However, when the applied moments are much larger than the in-plane force magnitudes (and particularly when the loaded hole has a short height compared to its diameter), there is a possibility that the in-plane force components can become skewed. In these cases, it is recommended to manually define a formulaic bearing load + bending traction distribution, or if the area of interest is of sufficient distance from the loaded hole, to apply the force/moments as a TLAP-Traction.
For more details read the TLAP Bearing Technical Brief.
As of the StressCheck® v10.5 release, an autocorrect feature is now available to ensure the force/moment magnitudes are resolved. This feature overcomes the limitations of many of the above cases:
StressCheck Tutorial: Autocorrect for Bearing and TLAP-Bearing Loads
Model courtesy of Altair
Why are there stress “spikes” in my results where I applied TLAP-Traction? Do I need to be concerned?
TLAP Traction always introduces some error into the model, due to the traction fitting process of distributing the TLAPs proportionately (by area) onto the element faces to represent a statically equivalent system.
In some cases the combination of a relatively significant distance between the TLAP and element face centroids, and a significant number of element faces pointing to a single TLAP, attribute to large “jumps” or “spikes” in tractions within a TLAP application surface.
For these reasons, in TLAP-Traction analyses typically the global error is not the best measurement of the quality of the data of interest, as TLAP-Traction may introduce discontinuities in the traction fittings (i.e. traction “jumps” between element faces).
Even if the global error is relatively high, as long as the TLAP Traction application surface is not in your area of interest, the data of interest can be analyzed and fringe plots can be meaningful.
Another important thing to note is you can also plot the error indicator (Function: Error) to view the errorassociated with the TLAP elements.
The Global-Local section of the ESRD Resource Library covers this in more detail:
Please read pages 714-717/1151 in the Master Guide for a more comprehensive description of how TLAP Traction is applied and how StressCheck® enforces the requirements.
There are several possibilities why the TLAP-Bearing load cannot be applied:
The hole contains features that cause it to no longer be classified as cylindrical (notches, intersecting holes, pressure release, etc.).
Note: introducing the tolerance parameters is no guarantee to fix the problem, as the geometry may not be valid for proper application of TLAP-Bearing or Bearing loads. For more information, consult the Master Guide section on “TLAP-Bearing” and read the TLAP Bearing Technical Brief.
Valid extractions of stress intensity factors (SIF) via the Fracture or Points tabs are as follows:
For example, if the material is linear isotropic, and the solution type is General Nonlinear, it is ok to extract the SIF. The stress intensity factor is a linear elastic fracture mechanics (LEFM) quantity.
The nature of the singularity changes as you approach a free edge – then there are both the edge and the free face corner singularities. The drop in SIF near a free face is well-known; for example, see NASA Technical Note D-8414 “Three-Dimensional Finite-Element Analysis of Finite-Thickness Fracture Specimens”, Raju and Newman, May 1977.
Consult the Fracture Mechanics section of the ESRD Resource Library for more information.
Valid extractions of stress intensity factors (SIF) via the Fracture or Points tabs are as follows:
For example, if the material is linear isotropic, and the solution type is General Nonlinear, it is ok to extract the SIF. The stress intensity factor is a linear elastic fracture mechanics (LEFM) quantity.
The nature of the singularity changes as you approach a free edge – then there are both the edge and the free face corner singularities. The drop in SIF near a free face is well-known; for example, see NASA Technical Note D-8414 “Three-Dimensional Finite-Element Analysis of Finite-Thickness Fracture Specimens”, Raju and Newman, May 1977.
Consult the Fracture Mechanics section of the ESRD Resource Library for more information.
There are a number of variables that may affect fringe plot rendering rates, especially if the cutting planes option is enabled. Though, models with large numbers of elements and/or solution degrees of freedom (DOF) will cause the rendering to perform slower in general.
Fringe plot rendering time can be drastically altered depending on user inputs
Here are a few suggestions that might help reduce the fringe plot rendering time when using the Plot tab with the Contour: Fringe option enabled:
More information on plotting:
In eigenvalue buckling analysis, the goal is to compute the buckling load factor (BLF) and the corresponding buckling mode shape. The BLF is the multiplier of the applied loads to the component to produce the first buckling mode. Note: checking the values of strains/stresses from the eigenvalue buckling solution is not meaningful because the buckling mode shape is known only to an arbitrary constant.
Therefore the linear solution multiplied by the BLF is the reference for when buckling will initiate. That means the stresses corresponding to the linear solution multiplied by the BLF are the stresses that will be developing at the time instability occurs (paired with the corresponding mode shape determined by the buckling analysis). Note: a linear analysis of the model is computed during an eigenvalue buckling analysis, and is available for results processing once the eigenvalue buckling solution is available.
Model courtesy of Altair
Why are there stress “spikes” in my results where I applied TLAP-Traction? Do I need to be concerned?
TLAP Traction always introduces some error into the model, due to the traction fitting process of distributing the TLAPs proportionately (by area) onto the element faces to represent a statically equivalent system.
In some cases the combination of a relatively significant distance between the TLAP and element face centroids, and a significant number of element faces pointing to a single TLAP, attribute to large “jumps” or “spikes” in tractions within a TLAP application surface.
For these reasons, in TLAP-Traction analyses typically the global error is not the best measurement of the quality of the data of interest, as TLAP-Traction may introduce discontinuities in the traction fittings (i.e. traction “jumps” between element faces).
Even if the global error is relatively high, as long as the TLAP Traction application surface is not in your area of interest, the data of interest can be analyzed and fringe plots can be meaningful.
Another important thing to note is you can also plot the error indicator (Function: Error) to view the errorassociated with the TLAP elements.
The Global-Local section of the ESRD Resource Library covers this in more detail:
Please read pages 714-717/1151 in the Master Guide for a more comprehensive description of how TLAP Traction is applied and how StressCheck® enforces the requirements.
The term “p-extension” refers to the process of systematically and hierarchically increasing the polynomial order of the element shape functions on a fixed mesh, thereby increasing simulation degrees of freedom (DOF) and representing more complex displacement fields for the mesh. Conversely, traditional FEA implementations require successive mesh refinements on elements of fixed polynomials, or h-extensions, to demonstrate convergence.
While both p-extensions and h-extensions are capable of converging to the exact solution of a mathematical problem solved by the finite element method, p-extensions will achieve convergence with less computational burden. Via the Wikipedia article on p-FEM:
The theoretical foundations of the p-version were established in a paper published Babuška, Szabó and Katz in 1981[2] where it was shown that for a large class of problems the asymptotic rate of convergence of the p-version in energy norm is at least twice that of the h-version, assuming that quasi-uniform meshes are used. Additional computational results and evidence of faster convergence of the p-version were presented by Babuška and Szabó in 1982.[3]
StressCheck® is capable of automatically performing p-extensions during the linear solution process, as shown in the below Linear solver tab:
P-extension from p=2 to 8
Note: it is not always necessary to solve to the maximum polynomial level, p=8. Convergence may be assessed after three (3) hierarchically increasing polynomial levels to determine if continuing the p-extension is necessary to achieve the desired solution quality.
This feature makes it very easy to verify convergence of any data of interest, as all runs of increasing DOF are stored for live dynamic extractions of results:
Key Quality Check #4: Peak Stress Convergence (courtesy StressCheck® Professional)
For more information and examples of p-extensions in practice:
In order to determine the applicability and viability of representing loading/unloading events via Incremental plasticity theory (IPT), it is important to conduct pre-solution checks before performing potentially time-consuming nonlinear analyses.
Note: these guidelines and recommendations to determine if IPT is an option for the current model inputs are a complement to the technical documentation on StressCheck®‘s implementation of IPT found here:
After performing a p-extension of at least three linear runs (e.g. p=2 to 4 or p=6 to 8), we should explore the model’s linear solution to assess the following characteristics:
StressCheck Tutorial: Directional vs Follower Loads in Geometric Nonlinear Analysis
StressCheck Tutorial: Linear vs Nonlinear Results for a Single Lap Joint
StressCheck Tutorial: Checking ITP Viability for Solving Nonlinear Events
In order to determine the applicability and viability of representing loading/unloading events via Incremental plasticity theory (IPT), it is important to conduct pre-solution checks before performing potentially time-consuming nonlinear analyses.
Note: these guidelines and recommendations to determine if IPT is an option for the current model inputs are a complement to the technical documentation on StressCheck®‘s implementation of IPT found here:
After performing a p-extension of at least three linear runs (e.g. p=2 to 4 or p=6 to 8), we should explore the model’s linear solution to assess the following characteristics:
StressCheck Tutorial: Directional vs Follower Loads in Geometric Nonlinear Analysis
StressCheck Tutorial: Linear vs Nonlinear Results for a Single Lap Joint
StressCheck Tutorial: Checking ITP Viability for Solving Nonlinear Events
© 2020 · Engineering Software Research & Development, Inc. | Terms & Conditions | Privacy & Cookie Policy | Software License Agreement | Software Maintenance and Technical Support Policy