# Divergence in parallel running after case reconstruction

 Register Blogs Members List Search Today's Posts Mark Forums Read

September 26, 2022, 04:42
Divergence in parallel running after case reconstruction
#1
New Member

Hung
Join Date: Dec 2021
Posts: 3
Rep Power: 3
Hi all!

I am running a simulation in a very complex geometry (human lung). Due to computer resources, I sometimes have to share clusters to other shared users. I constantly have to change my running processor number to 36, 72, 108, ... That's why I reconstruct and decompose the case again if I want to use different number of processor. My process is:

1. When someone asks for nodes, I stop my case (running in 108 cores) and reconstructPar -latestTime.
2. Then I decomposePar -latestTime the case to 36 cores and restart. This is when I face a divergence problem.

You can refer to the following results. After the second decomposing, my max parallel Courant number is 10.791 at time 0.1424 and keep doubling until divergence.

Quote:
 Time = 0.1421s Courant Number mean: 0.050839 max: 10.791 smoothSolver: Solving for Ux, Initial residual = 0.0138731, Final residual = 1.87886e-05, No Iterations 2 smoothSolver: Solving for Uy, Initial residual = 0.00704013, Final residual = 8.98269e-06, No Iterations 2 smoothSolver: Solving for Uz, Initial residual = 0.00496356, Final residual = 5.72542e-05, No Iterations 1 GAMG: Solving for p, Initial residual = 0.446619, Final residual = 5.04061e-05, No Iterations 13 GAMG: Solving for p, Initial residual = 0.182663, Final residual = 7.16055e-05, No Iterations 9 time step continuity errors : sum local = 6.20123e-08, global = -9.05988e-11, cumulative = -1.4971e-08 Current Inlet Flow Rate = -1.92092e-05 Current Inlet Flow Rate = -1.92092e-05 GAMG: Solving for p, Initial residual = 0.077038, Final residual = 8.26782e-05, No Iterations 8 GAMG: Solving for p, Initial residual = 0.0238152, Final residual = 8.96751e-05, No Iterations 4 time step continuity errors : sum local = 7.20512e-08, global = -1.06162e-09, cumulative = -1.60326e-08 Current Inlet Flow Rate = -1.92092e-05 GAMG: Solving for p, Initial residual = 0.0158456, Final residual = 8.69424e-05, No Iterations 6 GAMG: Solving for p, Initial residual = 0.00734297, Final residual = 8.69807e-05, No Iterations 3 time step continuity errors : sum local = 7.1083e-08, global = -3.9716e-10, cumulative = -1.64297e-08 Current Inlet Flow Rate = -1.92092e-05 GAMG: Solving for p, Initial residual = 0.00314104, Final residual = 5.60797e-05, No Iterations 3 GAMG: Solving for p, Initial residual = 0.00113054, Final residual = 7.12193e-07, No Iterations 7 time step continuity errors : sum local = 5.8211e-10, global = 4.59426e-11, cumulative = -1.63838e-08 Current Inlet Flow Rate = -1.92092e-05 ExecutionTime = 332.33 s ClockTime = 333 s
But when I try to run in one processor, it is only 5.50062 and keep running stably.

Quote:
 Time = 0.1424s Courant Number mean: 0.0509005 max: 5.50062 smoothSolver: Solving for Ux, Initial residual = 0.000300878, Final residual = 5.47137e-07, No Iterations 1 smoothSolver: Solving for Uy, Initial residual = 0.000388301, Final residual = 6.99841e-07, No Iterations 1 smoothSolver: Solving for Uz, Initial residual = 0.000328347, Final residual = 5.75167e-07, No Iterations 1 GAMG: Solving for p, Initial residual = 0.00066149, Final residual = 7.38323e-05, No Iterations 1 GAMG: Solving for p, Initial residual = 0.000248216, Final residual = 2.83435e-05, No Iterations 1 time step continuity errors : sum local = 1.15499e-08, global = 1.48838e-11, cumulative = -1.21772e-09 Current Inlet Flow Rate = -1.92498e-05 Current Inlet Flow Rate = -1.92498e-05 GAMG: Solving for p, Initial residual = 0.000128847, Final residual = 1.38249e-05, No Iterations 1 GAMG: Solving for p, Initial residual = 4.03098e-05, Final residual = 4.03098e-05, No Iterations 0 time step continuity errors : sum local = 1.64264e-08, global = -4.7946e-11, cumulative = -1.26567e-09 Current Inlet Flow Rate = -1.92498e-05 GAMG: Solving for p, Initial residual = 4.08922e-05, Final residual = 4.08922e-05, No Iterations 0 GAMG: Solving for p, Initial residual = 4.08922e-05, Final residual = 4.08922e-05, No Iterations 0 time step continuity errors : sum local = 1.66638e-08, global = -4.7946e-11, cumulative = -1.31361e-09 Current Inlet Flow Rate = -1.92498e-05 GAMG: Solving for p, Initial residual = 4.09558e-05, Final residual = 4.09558e-05, No Iterations 0 GAMG: Solving for p, Initial residual = 4.09558e-05, Final residual = 9.72793e-07, No Iterations 2 time step continuity errors : sum local = 3.96418e-10, global = 6.94878e-12, cumulative = -1.30666e-09 Current Inlet Flow Rate = -1.92498e-05
Note: I tried to reduce the time step to 1e-5 1e-6 (originally 1e-4) and still didn't solve the problem.

Anyone has any idea? Thanks.

 September 8, 2023, 10:01 I got the same #2 New Member   Haitham Osman Join Date: Jan 2020 Posts: 6 Rep Power: 5 I got the same problem like you got. To solve it, I tried using "adjustTimeStep on" if you are using pimpleFoam solver and fix "maxCo 5.5"(as you want in your case). This option would control the sharp increase in the courant number . By monitoring/ observing delta Time during running the simulation, you can see it return to your previous delat time you used before and the max Co number will return to 5.5 you got before. I hope it help any one got the same issue in future