[tareqkh]

Hello Foamers,

I was using simpleFoam on a simple problem and came to an error that I don't understand what it means. The error is as follows ;

#0  Foam::error::printStack(Foam::Ostream&) at ??:?
#1  Foam::sigFpe::sigHandler(int) at ??:?
#2   in "/lib/x86_64-linux-gnu/libc.so.6"
#3  acos in "/lib/x86_64-linux-gnu/libm.so.6"
#4  
 at ??:?
#5  
 at ??:?
#6  
 at ??:?
#7  
 at ??:?
#8  __libc_start_main in "/lib/x86_64-linux-gnu/libc.so.6"
#9  
 at ??:?
Floating point exception (core dumped)

Thanks in advance,

[OlegSutyrin]

It's segmentation fault. There is no way to tell more until you post the case you tried to run (whole case dir is best) here. And don't forget to tell what version of OpenFOAM you use.

[alexeym]

Hi,

This is NOT segmentation fault, this is floating point exception in acos. According to man acos:

SPECIAL VALUES
     acos(1) returns +0.

     acos(x) returns a NAN and raises the "invalid" floating-point exception for
     |x| > 1.

Why you are trying to calculate acos of value greater than 1?

[tareqkh]

I have added a volume source term to the momentum equation to be able to simulate the flow over a HAWT using the blade element theory. I don't really know why am trying to calculate acos of value greater than 1? I am a bit confused on what to do?

Thank you,

[arjun]

Originally Posted by alexeym

Hi,

This is NOT segmentation fault, this is floating point exception in acos. According to man acos:

SPECIAL VALUES
     acos(1) returns +0.

     acos(x) returns a NAN and raises the "invalid" floating-point exception for
     |x| > 1.

Why you are trying to calculate acos of value greater than 1?

Its not clear whether he is doing or openfoam is calculating acos of x greater than 1. The programmer has to be careful about it (who so ever programmed it).

[alexeym]

Originally Posted by tareqkh

... I am a bit confused on what to do?...

You have LOTS of choices:

1. Wait. Maybe in several days (weeks, months) you won't need the simulation.

2. Wait. Till someone with stronger mentalist powers comes to the thread, guesses your set up, guesses your modifications, and posts solution.

3. Go to $FOAM_SRC and run `ack acos`, through this you learn that in general, when using acos OpenFOAM limits argument except:

applications/solvers/multiphase/compressibleMultiphaseInterFoam/multiphaseMixtureThermo/multiphaseMixtureThermo.C
applications/solvers/multiphase/reactingEulerFoam/interfacialModels/virtualMassModels/Lamb/Lamb.C
applications/solvers/multiphase/reactingEulerFoam/reactingMultiphaseEulerFoam/multiphaseSystem/multiphaseSystem.C
applications/solvers/multiphase/twoPhaseEulerFoam/interfacialModels/virtualMassModels/Lamb/Lamb.C
applications/utilities/mesh/conversion/star3ToFoam/createCoupleMatches.C
applications/utilities/mesh/generation/foamyMesh/foamyQuadMesh/insertFeaturePoints.C
applications/utilities/mesh/manipulation/checkMesh/checkGeometry.C
applications/utilities/surface/surfaceCoarsen/bunnylod/vector.C
applications/utilities/surface/surfaceCoarsen/bunnylod/vector.h
src/OpenFOAM/meshes/polyMesh/polyPatches/constraint/cyclic/cyclicPolyPatch.C
src/OpenFOAM/meshes/primitiveMesh/primitiveMeshCheck/primitiveMeshCheck.C
src/OpenFOAM/primitives/Tensor/tensor/tensor.C
src/OpenFOAM/primitives/quaternion/quaternion.C
src/OpenFOAM/primitives/quaternion/quaternionI.H
src/dynamicMesh/motionSmoother/polyMeshGeometry/polyMeshGeometry.C
src/lagrangian/intermediate/submodels/CloudFunctionObjects/ParticleErosion/ParticleErosion.C
src/lagrangian/intermediate/submodels/Thermodynamic/SurfaceFilmModel/ThermoSurfaceFilm/ThermoSurfaceFilm.C
src/lagrangian/spray/submodels/BreakupModel/ETAB/ETAB.C
src/lagrangian/spray/submodels/BreakupModel/TAB/TAB.C
src/mesh/blockMesh/curvedEdges/arcEdge.C
src/meshTools/AMIInterpolation/patches/cyclicAMI/cyclicAMIPolyPatch/cyclicAMIPolyPatch.C
src/meshTools/primitiveMeshGeometry/primitiveMeshGeometry.C

Since you are taking about simpleFoam, guess, we can dismiss lagrangian and multiphase things. So only cyclics are left (more or less). Did you run checkMesh?

4. You can post your case description and case files, so people can check if you made some obvious error in case description, or just run your case and look for error. Though this could be problematic if your are under NDA.

5. Run simulation with debug version of OpenFOAM, so instead of your truncated printStack (and looking at you printStack output: do you run OpenFOAM 2.2.x?), you get output with method names and locations in the code. So you can learn, which method is calling acos.

I stop here, 5 variants are enough to start.

[alexeym]

Originally Posted by arjun

Its not clear whether he is doing or openfoam is calculating acos of x greater than 1. The programmer has to be careful about it (who so ever programmed it).

Thank you, Captain!

[CFD-Lover]

Hello Tareqkh,

I think the issue could be relevant to the mesh.

Best,

[tareqkh]

Dear Alexy,


What this error is telling from your experience?

#0  Foam::error::printStack(Foam::Ostream&) at ??:?
#1  Foam::sigFpe::sigHandler(int) at ??:?
#2   in "/lib/x86_64-linux-gnu/libc.so.6"
#3   in "/lib/x86_64-linux-gnu/libm.so.6"
#4  exp in "/lib/x86_64-linux-gnu/libm.so.6"
#5  
 at ??:?
#6  
 at ??:?
#7  
 at ??:?
#8  
 at ??:?
#9  __libc_start_main in "/lib/x86_64-linux-gnu/libc.so.6"
#10  
 at ??:?
Floating point exception (core dumped)

Best,

[alexeym]

Is it some kind of joke?

[tareqkh]

You have LOTS of choices:
1. Wait. Maybe in several days (weeks, months) you won't need the simulation.

Well, I am not type of person who gives up quickly so please fix your sentence .

You have LOTS of choices:
2. Wait. Till someone with stronger mentalist powers comes to the  thread, guesses your set up, guesses your modifications, and posts  solution.

Well, are you familiar with BEM theory used to simulate wind turbines. If so, I can post the case file.

Thank you,

[tareqkh]

Originally Posted by alexeym

Is it some kind of joke?

What do you mean? I want to know the main reason of the error because when I ran the same case using different rotational speed, it worked.

The main simulation is using the actuator disk model to simulate the flow over a wind turbine. I am using fixed wind speed with different rotational speed to obtain different tip speed so that I can build the power curve of the wind turbine. I am experiencing this issue when I increase the rotation speed, which corresponds to values above 4. That's why I posted the error to know what could be the reason and am thankful for your help.

Best,

[alexeym]

Originally Posted by tareqkh

Well, are you familiar with BEM theory used to simulate wind turbines. If so, I can post the case file.

I am not. And in fact I do not have to. I will just run set of standard tests and then try to figure out, where you call acos and why absolute value of argument is greater then 1. But if familiarity with BEM theory is a requirement for posting case files, then no, do not post them.

[alexeym]

Originally Posted by tareqkh

What do you mean? I want to know the main reason of the error because when I ran the same case using different rotational speed, it worked.

To learn why you get FPE in exp call, you can just run `man exp`. On Darwin you get

SPECIAL VALUES
     ...

     For all these functions, a range error occurs if the magnitude of x is too
     large.

on Linux

Errors

See math_error(7) for information on how to determine whether an error has occurred when calling these functions.

The following errors can occur:

Range error, overflow
errno is set to ERANGE. An overflow floating-point exception (FE_OVERFLOW) is raised.
Range error, underflow
errno is set to ERANGE. An underflow floating-point exception (FE_UNDERFLOW) is raised.

I do not know why you are trying to exponentiate very large value.

[tareqkh]

Thanks again for your help and hints.

I do not know why you are trying to exponentiate very large value.

The tip speed = (rotational speed)*Radius/Wind speed.

I am trying to do a validation study and in the paper that am trying to validate with, they used fixed wind speed to create the power curve. Therefore, to get different tip speed ratios, I have to change the rotational speed to match their results.

Best,

[arjun]

First it was cos then it is exp so from it, it seems that the actual problem happenmuch earlier than these places where the values are blown up.

So the good news is that you calculation has diverged which is something could be fixed and not a programming bug. Bad news is that simulation is diverged and you don't know why.

My advise try playing with under-relaxation, courant criteria etc etc.

Good luck.

[tareqkh]

Dear Arjun,

Many thanks for the hint. It might solve the issue. I will give it a try and let you know how it goes. Another issue you might know is about running in parallel and monitor the results. When I run the code (Actuator disk + simpleFoam) in single core and monitor the results i.e. forces using "log.simpleFoam", the code outputs the value. However, when I run the solver in parallel, all forces in the "log.simpleFoam" become zero even after use reconstruct. What could be the issue?

Best,