-
Notifications
You must be signed in to change notification settings - Fork 24
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Slow nonlinear convergence in cmm/test_pipe_3d #191
Comments
@wgyang, can you try running a "real-world" |
Yes, will do. |
Pulsatile flow poses challenges for CMM simulations. The required number of iterations is influenced by instantaneous flow. Even under steady flow conditions, achieving convergence, defined by a threshold of 1e-12, typically necessitates approximately 55 to 100 time steps, with around 7 Newton iterations needed. Note that the solver advances to the next time step when either Ri/R1 or Ri/R0 becomes smaller than the threshold. The convergence order for the relative residual at each time step is approximately 1. Achieving quadratic convergence is unlikely for the solver. In patient-specific cases, I found convergence more difficult compared to a simple cylinder. The threshold of 1e-12 is too strict. Here is the output for steady flow in a cylinder using cmm,
Using the same input files (a cylinder with pulsatile flow), svFSI and svFSIplus have similar output.
|
Thank you for looking into it, @wgyang!! We can close this issue. Even though the convergence is not great, it looks like it's "normal" for CMM. Maybe someone wants to dig that out in the future and check if the linearization is correct. |
Description
While adding missing results to test cases (#175), I noticed that the test
cmm/test_pipe_3d
(steps3a-inflate-cmm
and3b-prestress-cmm
) have a very slow nonlinear convergence.Reproduction
In
tests
, runpytest -vrx -k pipe_3d
. The second simulation output corresponds to step3a-inflate-cmm
, the second to last to3b-prestress-cmm
.This is the output I get for
3a-inflate-cmm
:I increased
Max_iterations
and setTolerance = 1e-12
.Expected behavior
Should converge to this tolerance in ~5 Newton iterations.
Additional context
@ktbolt, @wgyang, I don't understand how
CMM
works. Is it normal to converge this slowly?Before, we had a very coarse tolerance of
1e-4
, which might explain why we got different results on different machines (#137).Code of Conduct
The text was updated successfully, but these errors were encountered: