Page 1 of 1
Very large velocities and pressures
Posted: Thu Sep 17, 2020 7:53 am
by smatsumoto
Hello,
I have been running simulations for the aorta and have been getting unreasonably large values for pressure and velocity. The simulation indicates that the residual was somewhat minimized (~e-1) but the results are not great. Would it be possible for me to share my simulations to see what the issue is?
Specs:
-inlet flow rate: 50 cc/s
-outlet BCs (LCA/LSA/RSA/RCA): 20000
-outlet BCs (DescA): 2000
Thanks,
Shion
Re: Very large velocities and pressures
Posted: Thu Sep 17, 2020 10:29 am
by davep
Hi Shion,
Upload your entire SV project someplace I can download it and I'll have a look.
Cheers,
Dave
Re: Very large velocities and pressures
Posted: Mon Sep 21, 2020 12:45 pm
by davep
Hi Shion,
I looked at the simulation
histor.dat file and see
Code: Select all
46 2.130E+02 8.409E+78 ( 796) 6.970E+03 1.626E+04 < 8806- 1| 16> [ 0 - 0]
47 2.140E+02 5.326E+83 ( 844) 3.506E+01 1.178E+04 < 8788- 1| 18> [ 0 - 0]
47 2.170E+02 3.367E+87 ( 882) 5.904E+02 7.928E+04 < 8806- 1| 16> [ 0 - 0]
48 2.180E+02 8.374E+91 ( 926) 5.507E+00 2.714E+04 < 8818- 1| 17> [ 0 - 0]
48 2.200E+02 1.346E+96 ( 968) 6.873E+03 2.192E+03 < 8806- 1| 16> [ 0 - 0]
49 2.220E+02 7.008+100 (1015) NaN NaN < 8818- 1| 18> [ 0 - 0]
49 2.310E+02 NaN (****) NaN NaN < 1- 1|****> [ 0 - 0]
50 2.410E+02 NaN (****) NaN NaN < 1- 1|****> [ 0 - 0]
50 2.510E+02 NaN (****) NaN NaN < 1- 1|****> [ 0 - 0]
I reran the simulation using MPI and four processes and it ran fine. I think the problem is running the solver without MPI. I thought I fixed this problem but it seems the fix did not get into the current svSolver release, I'll open an Issue on this.
Try to rerun your simulation using MPI, maybe use four processes, should not melt your laptop.
Cheers,
Dave
Re: Very large velocities and pressures
Posted: Fri Sep 25, 2020 12:28 pm
by kizil015
I also have this issue, where running my simulation without MPI leads to arbitrarily large pressures (though never NaN). With MPI, it solves normally.
Is there an updated webpage where these major bugs are listed so that I can quickly check whether my problem would be fixed by downloading a newer version of svSolver?
Re: Very large velocities and pressures
Posted: Mon Sep 28, 2020 11:46 am
by davep
Hi Shannen,
Each SimVascular application (SV, svSolver, svFSI) has a GitHub Issues page used to manage bugs and feature requests. The no-mpi svSolver Issue is here
https://github.com/SimVascular/svSolver/issues/45
Cheers,
Dave