[panda007]

Hi,

I'm working on a wave-structure interaction problem using overInterDyMFoam. The model works well with small wave height (0.04m) with a constrained rigid body motion.

But after increased the wave height to 0.12m, with larger size of rigid body, (the domain is expanded accordingly), the model suffers instability issue after run for 6-7 wave periods.

I have reduced the accelcerationRelaxation to 0.2, (even tried 0.05) under sixDoFRigidBodyMotion library, it is still the same.

I have also tried to refine the mesh, expand domain size, add turbulence model, adjust turbulence model parameters, reduce max CFL, increase nOuterCorrector from 2 to 4-5, reduce tolerance, none of them worked to solve this instability issues.

And adding damping coefficient did reduce the high angular velocity, but for this case, the maximum angular velocity is much lower after applied damping coefficient, and it still did not help.

The air side velocity is very high, which could be the cause of this instability. I'm out of options, does anyone have any suggestions or ideas how to solve this instability issue? Thank you in advance for any help!!

[David*]

You tried "expand domain size" - in which direction?

I suggest to massively increase the z_max boundary (atmosphere) to avoid/reduce any kind of interaction with the water surface. You can use larger cells so it is computationally cheap. Try 3 times the water depth as total height of the model.

- add turbulence model
You definitely need one. Which one did you try? I suggest k-omega-SST.

-adjust turbulence model parameters
that won't help

- as a last resort, you can use a function object to eliminate velocity in the air phase. But this is a dirty hack and against physics.

Hope that helps!

[panda007]

Originally Posted by David*

You tried "expand domain size" - in which direction?

I suggest to massively increase the z_max boundary (atmosphere) to avoid/reduce any kind of interaction with the water surface. You can use larger cells so it is computationally cheap. Try 3 times the water depth as total height of the model.

- add turbulence model
You definitely need one. Which one did you try? I suggest k-omega-SST.

-adjust turbulence model parameters
that won't help

- as a last resort, you can use a function object to eliminate velocity in the air phase. But this is a dirty hack and against physics.

Hope that helps!

Hi David,

I tried to expand the Zmax to allow the length of z-axis domain be over 3 times of the water depth. Unfortunately, it didn't help, the simulation still stops around the same time (I set a minimum dt threshold to stop the simulation once dt < 1e-7s).

Before I changes z-axis domain, it stopped at 16.53s, after I expand z-axis domain, it stopped at 16.57s.

Anything else I can try? I have tried to refine the mesh, expand y-axis domain before, none of them helped. Reducing CFL doesn't help either. I'm desperate. Thank you so much for helping me!

[David*]

Try writing out the last timesteps before crash and identify the region of highest velocities = region of instability.



Are the air velocity really the cause of the crash or could it be the moving rigid body? Try fixing it for a test.


Unfortunately, these kind of problems take lots of time and many attempts to fix. Often, you need to start again from scratch (new mesh, new initialization). If you are lucky, it works, it not, then not... Best luck!

[JNSN]

Originally Posted by panda007

Hi David,

I tried to expand the Zmax to allow the length of z-axis domain be over 3 times of the water depth. Unfortunately, it didn't help, the simulation still stops around the same time (I set a minimum dt threshold to stop the simulation once dt < 1e-7s).

Before I changes z-axis domain, it stopped at 16.53s, after I expand z-axis domain, it stopped at 16.57s.

Anything else I can try? I have tried to refine the mesh, expand y-axis domain before, none of them helped. Reducing CFL doesn't help either. I'm desperate. Thank you so much for helping me!

Hi Panda,
I suggest increasing the number of outer iterations. 4 is way too low for such a coupled system. Especially with high underrelaxation. Which unfortunately is required.You need to ensure a converged solution of the 6dof solver. Also move the mesh each outer iteration ( if not yet done). Try with at least 8 iterations.
Best, Jan