|
[Sponsors] |
Restart causes FGMRES orthogonalization failed |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
March 13, 2020, 09:07 |
Restart causes FGMRES orthogonalization failed
|
#1 |
New Member
David Stevens
Join Date: Nov 2019
Posts: 11
Rep Power: 7 |
Hello all,
I am currently running a SS and transient model with and inlet profile and a rotating wall using a 3D hex mesh. I receive the following error whenever I attempt to restart the solution in SS or the time domain. This is the first time I have run into this and I have successfully performed similar operations in v7.0.0 in the past. I get the same error in v7.0.0 and 7.0.2. Error in "void CSysSolve<ScalarType>::ModGramSchmidt(int, std::vector<std::vector<ScalarType> >&, std::vector<CSysVector<ScalarType> >&) const [with ScalarType = double]": ------------------------------------------------------------------------- FGMRES orthogonalization failed, linear solver diverged. ------------------------------ Error Exit ------------------------------- Thank you! David W. Stevens Staff Engineer Peregrine Turbine Technologies |
|
March 13, 2020, 10:50 |
|
#2 |
Senior Member
Pedro Gomes
Join Date: Dec 2017
Posts: 466
Rep Power: 14 |
Hello David,
Can you create a minimal case for which this happens and share it? Does SU2 also diverge with other linear solvers? If not, do the residuals shoot up upon restart? Are you using adaptive CFL? If so can you check if using a fixed (and maybe lower) CFL helps. Cheers, Pedro |
|
March 13, 2020, 11:23 |
|
#3 |
New Member
David Stevens
Join Date: Nov 2019
Posts: 11
Rep Power: 7 |
Pedro,
I turned off the rotating wall and the issue went away. My CFL is already quite low but I am going to try to lower it further. If I cant resolve it I will certainly send you a case. Thank you, David |
|
March 23, 2020, 07:23 |
|
#4 |
Senior Member
Pedro Gomes
Join Date: Dec 2017
Posts: 466
Rep Power: 14 |
I replicated the problem also on simpler cases, the status of this issue can be tracked via the SU2 GitHub https://github.com/su2code/SU2/issues/915
|
|
March 23, 2020, 08:53 |
|
#5 |
New Member
David Stevens
Join Date: Nov 2019
Posts: 11
Rep Power: 7 |
Thank you Pedro!
|
|
April 1, 2020, 08:44 |
|
#6 |
Senior Member
Pedro Gomes
Join Date: Dec 2017
Posts: 466
Rep Power: 14 |
Hi David,
The restart should work if the case was first run with the following option in the config file: VOLUME_OUTPUT= PRIMITIVE, GRID_VELOCITY (primitive variables are a default output, we need the grid velocities as extra for the restart to work) We will make this simpler on a future version (7.0.4 probably). Cheers, Pedro |
|
April 1, 2020, 10:30 |
|
#7 |
New Member
David Stevens
Join Date: Nov 2019
Posts: 11
Rep Power: 7 |
Pedro,
That works! Thank you for your help! David |
|
March 3, 2021, 05:42 |
I am trying to use restart for unstaedy case but getting error .
|
#8 |
New Member
Vivek kumar
Join Date: Mar 2021
Location: India
Posts: 8
Rep Power: 5 |
I am getting error,
Error in "void CSysSolve<ScalarType>::ModGramSchmidt(int, std::vector<std::vector<ScalarType> >&, std::vector<CSysVector<ScalarType> >&) const [with ScalarType = double]": ------------------------------------------------------------------------- FGM Last edited by vivekdixit194; March 3, 2021 at 06:16. Reason: image is not uploaded |
|
March 3, 2021, 06:27 |
|
#9 |
New Member
Vivek kumar
Join Date: Mar 2021
Location: India
Posts: 8
Rep Power: 5 |
Hello all,
I was trying to use RESTART for my next iteration in my unsteady case. So these changes I did, 1. WRT_BINARY_RESTART= YES READ_BINARY_RESTART= YES 2. RESTART_SOL= YES RESTART_ITER = 785 3. I renamed restart_flow.dat to solution_flow.dat after execution ------------------------------ Begin Solver ----------------------------- Simulation Run using the Multizone Driver The simulation will run for 3000 time steps. CSysSolve::ConjugateGradient(): system solved by initial guess. +-----------------------------------------------------------------------------+ | Multizone Summary | +-----------------------------------------------------------------------------+ | Time_Iter| Inner_Iter| Outer_Iter| avg[bgs][1]|MinVolume[0]|DeformIter[0| +-----------------------------------------------------------------------------+ | 785| 0| 0| -21.333333| 5.2486e-10| 0| Error in "void CSysSolve<ScalarType>::ModGramSchmidt(int, std::vector<std::vector<ScalarType> >&, std::vector<CSysVector<ScalarType> >&) const [with ScalarType = double]": ------------------------------------------------------------------------- FGMRES orthogonalization failed, linear solver diverged. ------------------------------ Error Exit ------------------------------- -------------------------------------------------------------------------- MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD with errorcode 1. NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes. You may or may not see output from other processes, depending on exactly when Open MPI kills them. PLEASE HELP ME |
|
Tags |
diverged, error, fgmres, restart |
Thread Tools | Search this Thread |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Foam::error::printStack(Foam::Ostream&) with simpleFoam -parallel | U.Golling | OpenFOAM Running, Solving & CFD | 52 | September 23, 2023 04:35 |
Initial conditions for uniform flow | andreas | OpenFOAM | 5 | November 16, 2012 16:00 |
[OpenFOAM] ParaView/Parafoam error when making animation | Disco_Caine | ParaView | 6 | September 28, 2010 10:54 |
user subroutine error | CFDUSER | CFX | 2 | December 9, 2006 07:31 |
user defined function | cfduser | CFX | 0 | April 29, 2006 11:58 |