[sibo]

Hi All,

I was using simpleFoam to run an airfoil flow case in parallel, and I run into this error:

mpirun noticed that process rank 3 with PID 76036 on node compute-2-28 exited on signal 8 (Floating point exception).

At the beginning, it was running well, this error come out at 500 iterations.

After searching around in the forum, I see some people have similar problems and some people say it might be caused by mesh. So I tried to use a coarse mesh, and the simulation is running well. Therefore, I think this error is caused by my "finer mesh", but I have no idea why finer mesh can cause this problem!

Can anybody give me any suggestions about how to rebuilt the finer mesh to solve this problem?

Thanks in advance!

Sibo

[piu58]

It is a fine idea to observe the Courant number. If this number rises and rises, then your time step is too long.
A fine mesh usually need a shorter time step. That explains the behavior you see.

[agustinvo]

Hello,

if you check the log file, you should see what was going wrong at that point. If you post it we can take a look on it, and try to help.

[sibo]

Hi Uwe,

Thanks for your kind reply. But I was thinking since simpleFoam is a steady state solver, which means du/dt is 0. So the dt in simpleFoam solver is just something like the number of iterations. There should not be Courant number setting.
Correct me if i am wrong.
Thanks!

[sibo]

Hi Agustín,

Thanks for the advice. I adjusted my controlDict to make it output more values. After it's done, I will post it here. Thanks!!

[agustinvo]

Originally Posted by sibo

Hi Uwe,

Thanks for your kind reply. But I was thinking since simpleFoam is a steady state solver, which means du/dt is 0. So the dt in simpleFoam solver is just something like the number of iterations. There should not be Courant number setting.
Correct me if i am wrong.
Thanks!

Indeed, when you run a steady solver the unsteady settings in controlDict are not taken into account, you only have number of iterations.
Take a look to your boundary conditions. Did you use wall functions? How much resolved is your mesh? Do you know the yPlus fo your mesh?

Originally Posted by sibo

Hi Agustín,

Thanks for the advice. I adjusted my controlDict to make it output more values. After it's done, I will post it here. Thanks!!

More than the outputs, take a look to the log file, and there you should see the evolution of the residuals and mass continuity.

[sibo]

HI Agustín,
I used the nutUSpaldingWallFunction in nut file. And the y plus is expected to be 0.
This is the log file:

Time = 526

smoothSolver:  Solving for Ux, Initial residual = 0.2037803, Final residual = 0.01951528, No Iterations 6
smoothSolver:  Solving for Uy, Initial residual = 0.2312924, Final residual = 0.0148174, No Iterations 6
GAMG:  Solving for p, Initial residual = 0.9926779, Final residual = 0.09628591, No Iterations 14
time step continuity errors : sum local = 2.523865e+30, global = 2.128306e+14, cumulative = 2.192435e+15
--------------------------------------------------------------------------
mpirun noticed that process rank 3 with PID 19972 on node compute-4-2 exited on signal 8 (Floating point exception).

I noticed there is an time step continuity error, and this error start from the beginning!

Also, in my postprocessing folder, The forceCoefficient increase greatly.

n# Time Cd      Cl      Cm
30      1.369018        -3.405345       2.043749
60      -1.856338       -3.3926 -2.006583
90      6.168214        45.1218 1.185081
120     51.43253        252.5338        33.43611
150     205.9458        724.2902        316.1817
180     11340.94        35250   17279.78
210     -10434.01       -34111.99       -17401.88
240     -921074.1       -5595259        1141612
270     1.502871e+07    9.104924e+07    -2.001005e+07
300     -356938.5       -1372324        -382784.5
330     -3105626        -1.76059e+07    3843736
360     1.640823e+07    1.004517e+08    -2.257931e+07
390     -265929.5       1016040 -2509366
420     1243717 1.338678e+07    -6188400
450     1229633 9770905 -1939033
480     1.297626e+10    1.919775e+10    -1.671089e+09
510     2.239133e+38    6.596274e+38    5.292967e+37

[sibo]

Sorry my mistake, the y plus are expected to be 1.