Learning Forum

[Ahmed_Aissa]

Dear all,

I am running a two-way FSi simulation of complex topology. I want to close a gap to prevent leakage. To do this I need to run a standalone Fluent case and use the results as an initial guess to my FSI model.

I know that I should be using Patching data, or connect Fluent standalone solution cell to Fluent FSI solution cell. However, I am not using Workbench in my simulation (System coupling GUI). Does anyone have an idea on how to read previous results? I noticed that there is a way to initialize the solution so that SC does not override Fluent data. Can you guide me through this, please? If you have better suggestions please let me know!

[Stephen Orlando]

nIt is possible to have a System Coupling FSI handle a closing gap automatically. Please see this tutorial: https://ansyshelp.ansys.com/account/secured?returnurl=/Views/Secured/corp/v211/en/sysc_tut/sysc_tut_reedvalve_fluent.html. The method involves using the Contact Offset in Mechanical and Contact Marks in Fluent.nPlease have a look at my comment on this discussion for the initialization. You can specify a .jou journal file as the input file to System Coupling instead of the .cas, then load the .cas and .dat in the journal file. https://forum.ansys.com/discussion/23485/edit-input-and-getting-results-from-a-file-for-system-coupling-from-command-linenAlso see this discussion for how to use the patch method with Case Modification instead. https://forum.ansys.com/discussion/22971/fsi-volume-of-fluid-calculations-not-patching-the-water-phase#latestnThis sounds like a cool project. When you get it running, you should post a video of the results.nSteven

[Ahmed_Aissa]

Dear Steve, nThank you for your reply, and sorry for not getting back to you earlier. nI tried the option of journaling Fluent which means in my opinion to read the case and data of the previous job and assign it to be the actual initial guess. I prefer this method over patching since it allows for reading different output variables that may have influence over the current run. nWhat I did is the following: I run a standalone CFD analysis and wrote the case and data in the same Fluent-to-FSI folder (To prevent dealing with path issues). Then I wrote the following journal :nfile/confirm-overwrite? nonfile/start-transcript FLUENT-1_SC_002.trnfile set-batch-options yes yes yes ,nfile/read-case/steadystate.cas.h5nfile/read-data/steadystate.dat.h5n(sc-solve)nfile/stop-transcriptnexitnoknHowever, I don't see SC able to read the file and I receive an error after waiting a while. nHow can I solve this issue? nOn the other hand, I was thinking of the following, I am running a transient case where my intention is to investigate the dynamic vibration of a soft structure. The idea is to start from a closed valve using gap features and to put the fluid system at an equilibrium state. The best thing is to run a steady-state simulation and extract the solution and put it as an initial condition to my FSI problem. Does making a steady-transient FSI run in the first-time step and then switch to transient-transient FSI simulation afterwards make sense?nWhat is more accurate in your opinion? Knowing that I am using a Quasi-Newton method for System coupling, nAnd sure once it is done I will definitely make a video to show the results!nThank you, nAhmed,

[Stephen Orlando]

Hi ArraynWhen launching from the command line, you'll need to redirect the stderr to a file with something like this:n$SYSC_ROOT/bin/systemcoupling --guiserver -R run.py &> alloutput.txtnThe best thing is to run a steady-state simulation and extract the solution and put it as an initial condition to my FSI problem. Does making a steady-transient FSI run in the first-time step and then switch to transient-transient FSI simulation afterwards make sense?nI'm not sure exactly what you mean here. What do you mean by steady-transient and transient-transient?n

[Ahmed_Aissa]

Thank you for your reply! nI did not understand what you mean here. In fact, I am using System coupling GUI. nDoes making a steady-transient FSI run in the first-time step and then switch to transient-transient FSI simulation afterward make sense?nI mean here combining a steady-state Fluent run coupled with a transient mechanical, then activate time-integration within Fluent by changing Time from steady to transient, and continue the run.nThank you, nAhmed,n

[Ahmed_Aissa]

Dear Steve, nI followed another alternative to initialize my simulation. It consists of running restarted FSI analysis. nI set up Fluent with a very high courant number, disabled equations solving so that convergence is not evaluated, and added additional constraints (BCs). I then run a 2-way fsi with only displacement data-transfers. I do this for a few time-steps (I believe only one time-step could also be sufficient). Then when I interrupt the run I modify my fluid set-up as intended and continue. nI saw also another method where it is possible to make the initial state of the run, people mimic the steady-state scenario by turning off time integration within mechanical APDL. Whereas within Fluent they set high Courant number and disable convergence evaluation. nI believe this is an approach to consider. However, I had an issue within this approach. In fact, since I am not doing FSI in the Workbench environment I did not succeed to turn on time-integration when I wanted to. Can you guide me through this? nThank you, n