[fxzf]

Hi,

I got one case that GAMG solver is crashed.

I only attached a piece of information here, because they are just repeating.

[325] #1 Foam::sigFpe::sigHandler(int)[38] #1 addr2line failed
[297] #2 ?[182] #1 addr2line failed
[359] #1 Foam::sigFpe::sigHandler(int)[323] #3 Foam::PBiCGStab::solve(Foam::Field<float>&, Foam::Field<float> const&, unsigned char) constFoam::sigFpe::sigHandler(int) addr2line failed
[87] #1 Foam::sigFpe::sigHandler(int)[139] #2 ?[109] #3 Foam::PBiCGStab::solve(Foam::Field<float>&, Foam::Field<float> const&, unsigned char) const addr2line failed
[138] #1 Foam::sigFpe::sigHandler(int)Foam::sigFpe::sigHand ler(int)[261] #1 addr2line failed
[68] #3 Foam::PBiCGStab::solve(Foam::Field<float>&, Foam::Field<float> const&, unsigned char) const addr2line failed
addr2line failed
addr2line failed
[188] #1 Foam::sigFpe::sigHandler(int)[99] #2 ? addr2line failed
[357] #1 Foam::sigFpe::sigHandler(int) addr2line failed
addr2line failed
Foam::sigFpe::sigHandler(int)[236] #1 Foam::sigFpe::sigHandler(int) addr2line failed
addr2line failed
[212] #1 Foam::sigFpe::sigHandler(int)[312] #1 Foam::sigFpe::sigHandler(int) addr2line failed
[42] #1 Foam::sigFpe::sigHandler(int) addr2line failed
addr2line failed
[93] #1 Foam::sigFpe::sigHandler(int) addr2line failed
[224] #2 ? addr2line failed
[328] #1 Foam::sigFpe::sigHandler(int)[102] #2 addr2line failed
addr2line failed
[238] #1 Foam::sigFpe::sigHandler(int) addr2line failed
[132] #2 ?[58] #1 Foam::sigFpe::sigHandler(int) addr2line failed
addr2line failed
[259] #2 ??[77] #1 Foam::sigFpe::sigHandler(int) addr2line failed
[230] #2 ? addr2line failed
[356] #1 addr2line failed
[193] #2 ? addr2line failed
[294] #2 ? addr2line failed
[191] #1 Foam::sigFpe::sigHandler(int)[274] #1 Foam::sigFpe::sigHandler(int) addr2line failed
[145] #1 Foam::sigFpe::sigHandler(int) addr2line failed
[293] #1 Foam::sigFpe::sigHandler(int) addr2line failed
[73] #2 ? addr2line failed
[227] #2 Foam::sigFpe::sigHandler(int)[321] #3 Foam::PBiCGStab::solve(Foam::Field<float>&, Foam::Field<float> const&, unsigned char) const addr2line failed
[220] #1 Foam::sigFpe::sigHandler(int) addr2line failed
[167] #2 ? addr2line failed
addr2line failed
addr2line failed
addr2line failed
addr2line failed
[290] #2 [190] #1 Foam::sigFpe::sigHandler(int) addr2line failed
addr2line failed
? addr2line failed
addr2line failed
[381] #1 Foam::sigFpe::sigHandler(int) addr2line failed
? addr2line failed
[140] #1 Foam::sigFpe::sigHandler(int)[329] #3 Foam::PBiCGStab::solve(Foam::Field<float>&, Foam::Field<float> const&, unsigned char) const[32] #1 Foam::sigFpe::sigHandler(int) addr2line failed
addr2line failed
[76] #1 addr2line failed
[234] #1 Foam::s...........................


I have tried to use directSolveCoarsest, by this setting, it is working and get the results. But problem is my simulation become 3 times longer, which is not affordable for me.

I tried other solvers, but they all unstable.

So does anyone have any suggestions if I can make GAMG working without using directSolveCoarsest option.

[Santiago]

You can only use PCG/CG instead of gauss seidel (directSolve) in GAMG. If you say is unstable you might want to change your settings

[Bloerb]

I'd like to raise the point that nearly always this boils down to improper boundary conditions, discretisation or a low quality mesh. Since you did not provide this information. What is the output does this give you:

checkMesh -allTopology -allGeometry

[fxzf]

Originally Posted by Santiago

You can only use PCG/CG instead of gauss seidel (directSolve) in GAMG. If you say is unstable you might want to change your settings

Thanks very much for reply. I actually tried PCG/CG, but it is unstable and get divergence some points. The GAMG is only solver I found not divergence if it is not crash.

[fxzf]

Originally Posted by Bloerb

I'd like to raise the point that nearly always this boils down to improper boundary conditions, discretisation or a low quality mesh. Since you did not provide this information. What is the output does this give you:

checkMesh -allTopology -allGeometry

Thanks very much for reply. I used checkMesh and all checks are ok. My case is conjugate heat transfer, so in chtMultiRegionSimpleFoam. It is crashed at solid region. Actaully, if I switch off the radiation, GAMG working perfectly. But when radiation is on, GAMG could crash like this without any warning.

Is it possible that qr radiation heat flux passed through coupled boundary and created some bad condition in solid region?

[Bloerb]

Well in that case it is even more likely that something is wrong with your setup. You are only solving a laplace equation in the solid. This shouldn't crash. You might want to lower your relaxationFactor for h or e a bit. That should probably do the trick.

[fxzf]

Originally Posted by Bloerb

Well in that case it is even more likely that something is wrong with your setup. You are only solving a laplace equation in the solid. This shouldn't crash. You might want to lower your relaxationFactor for h or e a bit. That should probably do the trick.

well, my case is tricky. My solid is just a round disc, I added single rotating frame in it to simulation rotating effect like Fluent moving solid. I tried reduce relaxation Factor, but if anything below than 0.99, it will take unaffordable number of iterations to reach converge. So I cannot reduce relaxation factor.