How do I ensure that my turbulent boundary layer is correctly resolved? |
Unless you are specifically interested in resolving the boundary layer profile in your simulation, CFX-5 will use wall functions to model the near wall region in a turbulent flow... |
How do I set an appropriate Physical timescale size for steady state calculations? |
For steady-state problems, the CFX-Solver uses a robust, fully implicit formulation so that relatively large timesteps can be selected accelerating the convergence to steady-state as fast as... |
How do I specify Boundary Conditions correctly? |
For a given computational domain, boundary conditions can be given that over-specify or under-specify the setup. This usually results in non-physical solutions or failure of the solution to... |
Which fluid velocity do particles feel when they are initiated on a wall? Application is to calculate the flushing out of glass particles with one-way coupled particle tracking. |
The particles see the conservative velocity value, not the zero wall velocity. Therefore they experience drag. If this drag is too low to flush them out of the domain, try to define a small normal... |
When clicking on the button Create/Modify Surface then the following error message appears and ICEM CFD can not be used further on.
The specific GUI entity named "covexcoarsen" does not exist. |
In this case it is required to delete the cache-files created by ICEM CFD. These can be found in the users home-directory in subdirectories with the name .aienv_***_cache, with *** being a version... |
A customer upgraded their PCs to XP with Exceed -3d version 9.0 to connect to
Sun UltraSparc III servers, where they run Ansys 9.0.
Under X11 models all seems okay. When running /show,3d (using OpenGL) we find that when they try to rotate
the model around the x or y screen axis, the axis of rotation is displaced. The displacement is relative to the model, not the screen.
Curiously, rotation around the screen z axis works fine as does translation left, right and in and out.
If the model fills the screen it whizzes off the edge. If you zoom out far enough, the model can be seen to rotate around
the displaced axis.
Is there a fix for this? |
The customer had Exceed 7.1 and it worked fine there. Comparing the settings between the two we found that Exceed 7.1 had hardware acceleration turned off. After turning off hardware acceleration... |
Guidelines for Loss Model setup |
The Loss model in CFX-5 allows the user to define a pressure drop (as an additional source term in the momentum equations) in a certain region of the domain.<br>This is useful when modelling... |
User tries to start ANSYS 9.0 and receives the following error:
'Error in start up script'
'can't find package Itcl while executing
"package require Itcl"
(file "c:documents and settingsuserprogram filesansys incv##commonfileslauncherlaunchermain.itcl" line41.
What is causing this error? |
If the user has TCL_LIBRARY and TK_LIBRARY set to a third party software, this error will occur.<br>Add the following lines to the paths and the launcher will run.<br><br>For... |
Q. What methods for FLuid Structure Interaction utilizing FLOTRAN are/will be available?
There are two ways to run FLOTRAN based. FSI in ANSYS.
The "old way" is referred to as the FSI solver. It utilizes the FS set of commands to interface FLOTRAN with an ANSYS structural solution. The last version this will be supported in is ANSYS 9.0. It will continue to be present in the code, but will not be documented.
The ANSYS replacement for this is the Multifield solver, which has all the capabilities of the FSI solver but generalizes it. There will be support for this in ANSYS with GUI support for the foreseeable future, but it will not be a part of the ANSYS Workbench Solution.
Both of these methods utilize the FSIN flag to identify the FSI/Multifield boundary surfaces. The two sets of menu paths listed below for loads perform the same operation, i.e. set the interface boundary number.
Main Menu>Preprocessor>Loads>Define Loads>Apply>Field
Surface Intr>On Areas
or
Main Menu>Preprocessor>Loads>Define Loads>Apply>Fluid/ANSYS>Field
Surface>On Areas |
Q. What methods for FLuid Structure Interaction utilizing FLOTRAN are/will be available?<br><br>There are two ways to run FLOTRAN based. FSI in ANSYS.<br><br>The... |
Q. I would like to call ANSYS workbench from the DOS prompt and have it automatically run my customized macro. Similar to how ANSYS works.
Can this be done?
A. Yes. Workbench can be started from the DOS prompt via a script file. Take the following code block, save it to a JScript file - RunWB.js, for example - then launch it from the command line with "cscript.exe RunWB.js".
***********
// Start and run the Workbench
//
var WB = new ActiveXObject("AnsysWB.WB.90");
WB.StartApplet = "DSApplet";
WB.Run(false);
// Get the Simulation applet pointer
//
var DS = WB.AppletList.Applet("DSApplet").App;
// Put your JScript code here
//
// Shut down the executable
//
WB.FileExit();
***************** |
Q. I would like to call ANSYS workbench from the DOS prompt and have it automatically run my customized macro. Similar to how ANSYS works.<br>Can this be done?<br><br>A. Yes. ... |
How do I solve for fully developed flow with a short length of pipe in CFX-5.7.1? |
Fully developed flow implies that the momentum equations are periodic and the level of difficulty of the problem depends on how many other variables are not periodic.<br><br>The... |
My ANSYS Workbench Simulation 9.0 model of a flat plate in tension does not
finish when a Convergence object is added to a Solution item like equivalent
stress. In 10.0, it gives me the error message shown below, but I do not see
the indicated problem. Can anyone provide more information about this?
ANSYS - Error
There are no candidate nodes or elements available for adaptive refinement.
Check that the results with convergence are not scoped to singular boundary
conditions. |
For the attached 9.0 database, the code is behaving as designed. Simulation<br> does not allow nodes to be refined at "singular supports". Fixed Supports and<br> ... |
Are there any examples of using birth and death in Workbench Simulation? |
Yes, please see the attached Workbench Simulation 9.0 SP1 example.<br><br>APDL commands are used to incorporate the birth and death feature into<br>Workbench Simulation. A solid... |
After installation of ICEM CFD 5.1 with Service Pack 1 and update of the license manager to ANSYS License Manager (ansyslmd) the direct interface to UG NX2 might not work anymore. A license error is obtained. |
Check that the following environment variables are set on your operating system.<br><br>For using the direct interface from UG the following environment variables are... |
Q. How many particles are required for a good solution.
A. Each CFX particle represents a sample of real particles that all follow an identical path and thus represents an averaged track for the mass associated with that particle. Typically the mass associated with each CFX particle will represent many thousands or millions of actual particles. The basic idea is to choose enough particles to get a representative sample of all possible tracks. For example, as an extreme case, if all your real particles followed a single path by falling straight down under gravity from a single point then you would only need a single CFX particle track to adequately simulate this. Typically most people set between a few tens to a few hundreds of particles. |
Q. How many particles are required for a good solution.<br><br>A. Each CFX particle represents a sample of real particles that all follow an identical path and thus represents an... |
######## WORK NOTES COPIED FROM INCIDENT 422713 ########
QUESTION:
The /inquire command is documented to return the date/time stamp as follows: yyyymmdd.hhmmss
When I test that command, it returns the proper date string, but it usually returns only 1 value for the time string, for example 20050727.1
Is the command functioning properly? Is there a way to retrieve the time string?
ANSWER:
There is a limitation of # of significant figures reported in the lister window (utility menu => parameters => scalar parameters').
Have a look at the attached readme file. This strategy seems to work around the issue satisfactorily.
Instructions:
1) Copy all files to a working directory
2) Start ANSYS in the working directory where the files reside
3) Invoke macro file date01.mac
4) Output: file date_time.txt containing listing of date and time for file scrap.txt.
Contects of date01.mac:
/inquire,scrap,date,scrap.txt
date_=nint(scrap)
time_=nint(1e6*(scrap-nint(scrap)))
*cfo,date_time,txt
*vwr,date_
DATE: %I
*vwr,time_
TIME: %I
*cfc
Take care,
Bill |
######## WORK NOTES COPIED FROM INCIDENT 422713 ########<br><br><br>QUESTION:<br><br>The /inquire command is documented to return the date/time stamp as follows: ... |
Class3 Error 2005-13 states that if new nodes are defined after an IC command, and those new node numbers are larger than the largest node number that was active when the IC command was issued, the initial conditions input by the second and any subsequent IC commands are not correctly defined for all those nodes with larger node numbers. This Class3 report goes on to give two workarounds that are helpful for cases where this error has not already occurred. However, no workaround is given to help someone who already has this error in their database. What can I do if my database already contains this error? |
Two problems occur due to this error:<br>¿ The displacements and velocities of these new nodes are not initialized to zero<br>¿ The record size of these new nodes are... |
Why does the user receive the following error while running ICEM 5.1 on Unix HP UX 11i 11.11?
Product: ANSYS ICEM CFD Output Interface for Patran (feature 'aioutpat')
Checkout failed for the above product.
Problem encountered:
The following license file either does not exist or cannot be read:
/ansys_inc/shared_files/licensing/license.dat |
The problem is that user is running patran, which could run it if has either aioutput or aioutpat or some other features, and it can't find any of them.<br>In this case the user had... |
Q) Why does the solid being produced after completing a loft/spline operation look really twisted?
A) The problem here is that the line defining how the two sketches join up (the guide line) is not correctly set. If you right click on the skin/loft in the modeling tree and choose edit/selection. In the viewport you should see a grey line (the guide line) connecting the two sketches. The line needs to join corresponding points on the two sketches. To change which point on a sketch the guideline used, right click on the viewport and choose fix guide line. All you need to do is to select the point you want the guideline to use. |
Q) Why does the solid being produced after completing a loft/spline operation look really twisted?<br>A) The problem here is that the line defining how the two sketches join up (the guide... |
How to restart a run, allowing the solver to read the values from a previous results file, but giving own values for certain variables and/or domains?
Example1:
Simulation with two domains D1 and D2.
First simulation carried out, then restart reading the simulation1.res for D1 and assigning a new value for D2 (not read from simulation1.res).
Example2:
First simulation carried out, then restart reading the simulation1.res for all variables, but overriding some of the variables with new values (p, T, etc.) |
Initialisation with "Automatic" or "Automatic with Value" will not work for the custom restart, as it will read the variables from the results... |
How to specify an initial Mesh Deformation at t=0?
Original mesh is ready. Start a new run (with or without mesh deformation), for which some boundary is already shifted at t=0 with a delta_X, for example.
We need then to shift the boundary from the original mesh "before" starting the run. |
In order to get an initial mesh deformation at t=0, a first case must be run in advance where the desired mesh deformation will be achieved, and then use this to set-up the final case. This can be... |
My solver run stops with the message 'RMS res. and GLOBAL imbalances are below their target criteria', yet the imbalances on some variables are still large. How do I ensure that the solver only stops when a tight convergence is achieved? |
On the Solver Control form/Basic Settings/Convergence Criteria, tick the box 'Conservation Target' and set the imbalance target that you want to achieve (a value of 0.01 stands for a... |
Q. I am using the "HP MPI Distributed Parallel for x86 64" start method for my CFX distributed parallel runs on a IBM machine equiped with Myrinet interconnect. But the CFX solver seems to using standard ethernet, although the documentation says the HP MPI method will automatically use the fastest available interconnect. How can I get CFX Solver to use the Myrinet interconnect?
A. The problem is where the GM protocol shared libraries are located on IBM machines. By default HP MPI loads the driver from here:
/opt/gm/lib/libgm.so (32 bit driver)
/opt/gm/lib64/libgm.so (64 bit driver)
But on IBM machines the libraries are stored in a different location. You can fix this by adding this environment variable to override
the default path:
setenv MPI_ICLIB_GM /opt/gm-some-path/lib64/libgm.so
you need to find the correct path from your IBM rep. Usually the path is:
/opt/gm/2.0.18-2.6.5-7.139-smp/lib64/libgm.so
To get the CFX Solver to use this environment variable you may have to do the following:
1) Create a text file called cfx5rc (no extensions), which contains the following line:
setenv MPI_ICLIB_GM /opt/gm-some-path/lib64/libgm.so
2) Put the file in the following folder:
~/.cfx/10.0 |
Q. I am using the "HP MPI Distributed Parallel for x86 64" start method for my CFX distributed parallel runs on a IBM machine equiped with Myrinet interconnect. But the CFX solver... |
Q. In viscoelastic curve fitting routine, when user reads in bulk relaxation data only, curve fitter fails to produce a curve and generate coefficients correctly.
When he adds same data for both shear and bulk relaxation, curve fitting is successful. What is wrong?
A. This was a bug in ANSYS 9.0 and 10.0. At 9.0 the curve fitter fails completely. At 10.0, it is worse because the curve fitter generates a curve, calculates the coefficients, but then saves the coefficents incorrectly with the TB command. (Refer to Defect 33341). This has been corrected at 11.0beta. This is not a class 3 error at 10.0 because it is identified in errata report. |
Q. In viscoelastic curve fitting routine, when user reads in bulk relaxation data only, curve fitter fails to produce a curve and generate coefficients correctly.<br>When he adds same data... |
I ran a transient simulation and saved transient timetsep data. My run finished normally and I was able to load my final results file into Post. Whenever I tried to load any of the intermediate timesteps, however, I got the following error:
"Error Reading Number of Domains (G/NZN).
The timestep list in the Timestep Selector in Post was also incorrect- many of the timesteps were shown as full backups when they were partial and there were several which were listed as -1.
There were no indications of problems in the output file. My filename had the _full_ string in it. |
<br>The ¿_full_¿ string in the filename has been observed to cause exactly this problem in CFX-5.7 and ANSYS CFX 10.0. If you edit the filename so that the string... |