**** Entered By: rlange @ 12/12/2007 01:14 PM ****
######## WORK NOTES COPIED FROM INCIDENT 692434 ########
Q.
I performed an MFX simulation (2-way FSI between CFX and ANSYS) from t=0[s] to 2[s] and also performed a restart run at t=1[s] using the same model as a separate run. The results just after restart at t=1.1[s] are not the same for both models. Why is this?
A.
This is a known issue when the values transferred from ANSYS to CFX (displacements in this case) have not been fully converged within a timestep. It is only a problem when ANSYS is solving second (last). When performing a restart, in the first timestep ANSYS will ask CFX for the last displacement values that it sent from the previous run, since ANSYS does not store these values itself. When ANSYS solves last it will not have sent the most recent displacement values to CFX in the last timestep of the previous run, therefore it will not be restarting from the most recent displacements. This is not a problem when the displacements have been fully converged within the last timestep of the previous run, but if the displacements were still changing a restarted run will show different behaviour than a continuous run. QA Defect 31747 is associated with this issue.
|
**** Entered By: rlange @ 12/12/2007 01:14 PM **** ######## WORK NOTES COPIED FROM INCIDENT 692434 ######## Q. I performed an MFX simulation (2-way FSI between CFX and ANSYS) from t=0[s] to 2[s] and also performed a restart run at t=1[s] using the same model as a separate run. The results just after restart at t=1.1[s] are not the same for both models. Why is this? A. This is a known issue when the values transferred from ANSYS to CFX (displacements in this case) have not been fully converged within a timestep. It is only a problem when ANSYS is solving second (last). When performing a restart, in the first timestep ANSYS will ask CFX for the last displacement values that it sent from the previous run, since ANSYS does not store these values itself. When ANSYS solves last it will not have sent the most recent displacement values to CFX in the last timestep of the previous run, therefore it will not be restarting from the most recent displacements. This is not a problem when the displacements have been fully converged within the last timestep of the previous run, but if the displacements were still changing a restarted run will show different behaviour than a continuous run. QA Defect 31747 is associated with this issue. |
||
|