[blaise]

Hello all,

I am stumbling across a weird issue when running a certain parallel case in OFv1712. The simulation stops writing any output at a certain time step (in time step 1.165 s of simulation time), but keeps running on the involved CPUs. Looking at what the processes are doing I get the following output:

poll([{fd=6, events=POLLIN}, {fd=14, events=POLLIN}], 2, 0) = 0 (Timeout)

The solver endlessly repeats these poll system calls forever and does not return.
The case is run with an overWaveDyMFoam solver which I fiddled together from the waves2Foam (of Niels Gjoel Jacobsen) package and overInterDyMFoam. This solver already run successfully on several different test cases I set up before the one at hand.

What I did try until now:
- Running the case locally parallel to exclude any nfs related issues
- Running the case on a single CPU
- Running the case on other computers
- Changing fixed time steps, changing to dynamic time step
- Changing the number of partitions

Everything to no avail: The simulation keeps stopping to write output every time at the very same situation.

The usual output of the last time step - i.e. the time step this issue occurs in - looks like:

Time = 1.165

PIMPLE: iteration 1
forces forces:
    Not including porosity effects

6-DoF rigid body motion
    Centre of rotation: (-2.335 0 0.194664)
    Centre of mass: (-2.335 0 0.194664)
    Orientation: (0.999663 0 0.0259474 0 1 0 -0.0259474 0 0.999663)
    Linear velocity: (0 0 -0.0109389)
    Angular velocity: (0 -0.111191 0)
    Euler angles: (0 1.48684 0)

inverseDistance : detected 2 mesh regions
    zone:0 nCells:755040  voxels:(53 53 53) bb:(-30 -4.50004 -4.24004) (3.27277 4.50004 4.00004)
    zone:1 nCells:435686  voxels:(53 53 53) bb:(-150 -22.5001 -20.7201) (-89.9999 -13.4999 -12.4799)
Overset analysis : nCells : 1190726
    calculated   : 1154246
    interpolated : 24794 (interpolated from local:0  mixed local/remote:0  remote:24794)
    hole         : 11686

Execution time for mesh.update() = 10.34 s
smoothSolver:  Solving for alpha.water, Initial residual = 2.38881e-05, Final residual = 1.53679e-09, No Iterations 1
Phase-1 volume fraction = 0.533664  Min(alpha.water) = -5.76508e-09  Max(alpha.water) = 1.00002
MULES: Correcting alpha.water
MULES: Correcting alpha.water
MULES: Correcting alpha.water
Phase-1 volume fraction = 0.533664  Min(alpha.water) = -1.93627e-06  Max(alpha.water) = 1.00002
smoothSolver:  Solving for alpha.water, Initial residual = 2.388e-05, Final residual = 1.54164e-09, No Iterations 1
Phase-1 volume fraction = 0.533664  Min(alpha.water) = -6.33039e-09  Max(alpha.water) = 1.00002
MULES: Correcting alpha.water
MULES: Correcting alpha.water
MULES: Correcting alpha.water
Phase-1 volume fraction = 0.533664  Min(alpha.water) = -2.05242e-06  Max(alpha.water) = 1.00002

I have attached a text file containing the output when I turn on several debug switches, but actually I am not sure what switches to turn on for an effective analysis of the problem...

Any hint is appreciated!

Best regards,

Pascal.

[gfilip]

Hi Pascal,

This is an old post but I'm curious if you ever resolved this mysterious overset stopping? I too see this behavior with a waves2Foam + overInterDyMFoam solver or just a pure interDyMFoam solver. I saw in another post (overInterDyMFoam stops without error message) a similar problem that was resolved by making the bottom boundary a slip condition, which seems odd.

Anyway, please let me know if you ever found the source of the problem or a solution.

Thanks

[tuncm18]

I am also facing with the same problem. My dsmcFoam+ solver is also giving same treatment, analysis running but there is no output. I am trying to solve my problem with changing boundary conditions, I will update here.

[tuncm18]

Originally Posted by tuncm18

I am also facing with the same problem. My dsmcFoam+ solver is also giving same treatment, analysis running but there is no output. I am trying to solve my problem with changing boundary conditions, I will update here.

Apparently there is something wrong with my initialize dictionary. At first I was setting initial velocity and initial particle number to zero. After I change initial values to my supposedly final values it worked. But before that, serial analysis was working without any problem, only parallel analysis was stuck. I think it might be related to some kind of weird bug.
I will check mean Courant number, max Courant number and min Courant number to see anything wrong with it.

[Yiny]

Hi, did you succeed in coupling the slovers of waves2foam and overinterdymfoam? I am also trying this work and have encountered an unsolvable problem:
I finished compiling overwavedymfoam and no error was reported. But when I run a case, I find the case diverged after 35s which is close to stability. I tried various modifications but couldn't change the divergence phenomenon and the version of OpenFoam I used is OF2206. Can you give me some advice? Thanks.
I will attach my case and solver below.

[ashkum2527]

Originally Posted by tuncm18

I am also facing with the same problem. My dsmcFoam+ solver is also giving same treatment, analysis running but there is no output. I am trying to solve my problem with changing boundary conditions, I will update here.

I face the same problem now. my simulation shows it is running but no new values are printed. Did you debug this error? I use dsmcFoam+ solver to simulate flow through micronozzle operating at exit vacuum condition.

[tuncm18]

Originally Posted by ashkum2527

I face the same problem now. my simulation shows it is running but no new values are printed. Did you debug this error? I use dsmcFoam+ solver to simulate flow through micronozzle operating at exit vacuum condition.

Hi,

Unfortunately not. I discovered dsmcFoam+ just waits before writing new positions in processor files after some time. As a crude solution, I send my analyzes with time limit and delete last written writing interval before sending them again.

Hope it helps.