[bitty122]

Dear all,

I am currently working on vehicle door closing simulation to predict maximum pressure build up inside cabin using overset methodology.
My model physics settings are:
constant density, implicit unsteady, K-epsilon turbulence, segregated flow.
I am facing with the below mentioned error while running the unsteady simulation:
A floating point exception has occurred: floating point exception [Overflow]. The specific cause cannot be identified. Please refer to the troubleshooting section of the User's Guide.
Context: star.segregatedflow.SegregatedFlowSolver
Command: RunSimulation
error: Server Error

Can anyone suggest how to resolve this error. I tried using adaptive mesh refinement and under-relaxing the velocity solver and pressure correction solver in segregated flow model.

Any help on this would be highly appreciated.

Thanks and Regards,
Bitty

[koful]

Hi bitty.

This is very general error and star can give this error for various of reasons.

Threshold derive part can be used to show high residual cells or limited veloctiy/pressure cells. Then you can identify the problems.

In my opinion, the usual suspect is always the overset mesh.

I suggest you to disable the fluid solvers and simulate the door motion only. You can observ overset quality with using overset field functions on scalar scenes. You might need to refine background mesh.

[bitty122]

Hi Koful,

Thank you for the reply.
I checked with threshold derive parts for cells with high velocity/pressure zones.It is coming near the hinge area of the door about which the door is rotating.
This zone is in close proximity with the overset region and I had already refined the background zone such that the mesh sizes are same for overset and background region of interest.

I also tried giving a run by disabling the fluid solvers but that run came out with the following error :
Continuum or phase "Physics 1" is missing required models: [Flow]. Model selection may be incomplete or requirements may have changed for newer version. Please add needed models.

Regarding your opinion on viewing overset quality via field functions, I tried it on the simulation that had come out, but nothing is visible on that.
Is there any prerequisite for viewing them (such as saving scenes per time step) .
Also are there any overset mesh quality checks that one needs to check before running the simulation?
I have done the basic volume corrections for the regions.

As for the error that i am still facing .i.e. A non-finite residual (Tke) was added by star.keturb.KeTurbSolver. Typical causes are overflow, underflow, or a division by zero.
I tried to reduce the urfs for K-epsilon turbulence and enabled urf ramp factor (linear ramp ) and also increased the maximum ratio under k-epsilon turbulent viscocity, but still am facing the same issue.

Thanks in advance,
Bitty

[bitty122]

Hi Koful,

Thanks for your reply,
I checked the threshold derive part for checking cells with high velocity/pressure regions. It is near the hinge of the door about which the door rotates.
This zone is in close proximity with the overset mesh and lies in background region. I have also ensured that the mesh sizes of both the overset region and background region (near overset) are of the same sizes.
As per your suggestion , I also tried to disable the fluid solvers (ie. segregated flow) but was facing with the following error:
Continuum or phase "Physics 1" is missing required models: [Flow]. Model selection may be incomplete or requirements may have changed for newer version. Please add needed models.
I tried to observe overset quality of the simulation that had diverged by using over set field functions ,but nothing was visible on that.
Should I save a scalar scene with overset error status/cell type before start of simulation for this purpose?
Also are there any prerequisites or best practices that one must follow while performing overset mesh simulation?

As for the error that i am still facing. i.e. A non-finite residual (Tke) was added by star.keturb.KeTurbSolver. Typical causes are overflow, underflow, or a division by zero.

I have tried many iterations by reducing the urfs of k-epsilon turbulence,k-epsilon turbulent viscosity, increasing maximum ratio of k-epsilon turbulent viscosity to order 2 and using under-relaxation factor linear ramp. But still the error persists.
Any help on this would be highly appreciated.

Thanks in advance,
Bitty

[koful]

Hi Betty.

I think choosing only 2 physics should be sufficient: three dimension and implicit unsteady. with these two you can simulate motion with no flow.

You can create a threshold with overset cell status scalar function. Set below min with the range values of -1.9. During the entire motion, the moving door should always be surrounded by the threshold.

does your simulation crash at the beginning or at a specific position?

I don't know if you have access to these resources:
https://support.sw.siemens.com/knowl...00042153_EN_US
https://support.sw.siemens.com/knowl...00041472_EN_US
https://support.sw.siemens.com/knowl...00040406_EN_US

Good luck.

[bitty122]

Hi Koful,

I tried running the simulation by choosing 3D and implicit unsteady to check overset quality. It seemed to be working fine and I checked for the overset cell status scalar function with the said specifications . The door was within the threshold.

To your question on whether my simulation crashed at the beginning or at a specific position, the answer would be both as I have tried out different setting/physics iterations.
The error msg coming for divergence at the beginning is:

WARNING: Ap = 0 on multigrid level 8, nRows = 214, blockSize = 1
AMG coarsening halted.

Error: AMG solver diverged.

And for other iterations it came out at a specific location with the error i had mentioned in my previous posts .i.e A non-finite residual (Tke) was added by star.keturb.KeTurbSolver. Typical causes are overflow, underflow, or a division by zero.


Thanks and regards,
Bitty

[bitty122]

Hello All,

This is in continuation to the previous query that i had regarding overset mesh in door closing.
Recently, I had tried a case where i only kept the outer panel of the door (representative fig shown in attachment->thickness of 0.5mm). It ran completely, but when i replaced it with the actual door (keeping the rest of the parameters constant), the solution file comes out with the following error:

"Error in metrics computation. The mesh may have non-positive volume cells This may be due to an invalid imported mesh or to errors initializing an interface."

Any leads on how to resolve the following error would be highly appreciated. Thanks in advance.