fireII example with hy2DyMFoam crashes with core dump

[iainsymthe]

When the fire-II example with 11 species is run with

the dynamicMeshDict file show in the "Adaptive mesh refinement" section of https://vincentcasseau.github.io/how-tos-cfd-fleming/how-tos-cfd-fleming-advanced/

crashes with a core dump.

After the specified number of steps = refinementInterval, The log file shows Selected 4748 cells for refinement out of 11682. Refined from 11682 to 44918 cells.

followed by

Selected 0 split points out of a possible 4748. Execution time for mesh.update() = 0.07 s Error in `hy2DyMFoam': corrupted double-linked list: 0x00000000022739c0

[vincentcasseau]

Hi Iain, Could you please run the same simulation using the single-temperature model instead? If I remember well, this is where the problem lies (allocation of two-temperature fields at solver level). Thanks, Vincent

[iainsymthe]

Vincent: Even with the single-temperature settings in constant/thermophysicalProperties I get the same error - case crashes and dumps core Just before crashing the output looks like  Selected 0 split points out of a possible 2295.Execution time for mesh.update() = 0.28 sdiagonal:  Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0 Error in `hy2DyMFoam': free(): invalid size: 0x0000000004eb5620 ======= Backtrace: ========= Thanks Iain On Wednesday, January 13, 2021, 11:52:22 AM CST, Vince notifications@github.com wrote:

Hi Iain, Could you please run the same simulation using the single-temperature model instead? If I remember well, this is where the problem lies (allocation of two-temperature fields at solver level). Thanks, Vincent

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[vincentcasseau]

I couldn't reproduce this bug.

[iainsymthe]

Could you please send me your dynamicMeshDict file?  I have attached the file I used.  I restarted from a solution that had run for 1e-6 sec.Thanks

On Sunday, January 31, 2021, 08:55:29 PM CST, Vince <notifications@github.com> wrote:  

I couldn't reproduce this bug.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[vincentcasseau]

I understand now, I was using AMR for the non-reacting case. I will investigate this later.

• [x] chemistry: corrupted double-linked list for impact ionisation reactions fixed in commit 91ac808b2d76bcef33956519a67664ce6e040272
• [x] temperature jump BC: add keyword uniform after value. For example: value uniform 800.0; instead of value 800.0;
  • website updated: https://vincentcasseau.github.io/how-tos-cfd-fleming/how-tos-cfd-fleming-initial-conditions/#31-trans-rotational-temperature
  • tutorials updated in commit ad5d268eaddf226219adcfc461be7dbf26bb7be9

[iainsymthe]

Thanks, Vince.  I will await your instructions when you get some time to fix the issue. On Wednesday, March 10, 2021, 08:09:12 AM CST, Vince @.***> wrote:

I understand now, I was using AMR for the non-reacting case. Each one of the two features below causes the case to crash. I will investigate this later.

• chemistry
• slip and jump BCs

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[vincentcasseau]

Hi Iain,

It's fixed. That being said

1. I strongly discourage anyone to use the velocity slip and temperature jump boundary conditions in hy2DyMFoam or it will blow up very quickly. Instead, one should use hy2DyMFoam with Dirichlet BCs for temperature(s) and velocity and then use hy2Foam with Robin BCs on the latest grid;
2. I would even discourage anyone to use hy2DyMFoam for high-Mach external aerodynamics. As you know, hy2Foam meshes need to be structured, carefully aligned with the bow shock and refined in its vicinity.

Regards, Vincent