Divergence occurs during simulation
 

Dear All,

I did simulation in waves involving multi-domain case problem. In the mesh management, I already set to be "no mesh deformation". However, the simulation was crashed in middle of running (pls kindly see the comments and the force monitoring in the attached files). Could you pls share your expert to solve the problem. Thanks in advance.

Regards,

Naid

Did you witness high motion amplitudes ?

Was the wave field initialized in the whole domain ? Your simulation crashing so early on is a bit odd. It might be better for your boat/geometry/vessel to reach a steady state position and then turn the wave maker on.
What kind of time step did you use (relative to your geometry length, and nominal speed) ? How many iterations per time step ?

I have never used multi-domains in FineMarine so I have no experience on the subject...

Hi,

Rombzh are you sure that to reach a steady state in calm water and then turn the waves on is faster and/or more stable than making the boat accelerate slightly (waves should not be initialize in the domain if 2dof are solved)

Afitriadhy can you defined middle of running.
Does the waves have reach the outlet or not ?
Can you post a picture of wave elevation before your simulation diverged?

Hi Rombzh,
Thanks for your reply. So far, the amplitude of motion was not jump yet (only the force). Yes I set the wave to be generated from front (head-sea). I did multiple-domain involving ship to ship interactions. This case, I set for zero speed of ships. Then I do not need to run the ships with speed.
Here, I used time laws as recommended by Numeca (manual book).

I am looking forward to getting your further share in future especially for multiple-domain case.
Tq

Hi RomainRH,
Thanks for the reply.
Actually I set the storage every 100 (output results). However, the simulation has not reached yet to that stipulated time, then i can not open the CFView. Meaning to say, I can post you the wave elevation.


Looking forward to getting your sharing experiences.

Hi, well good question, I don't know if it is faster. The thing is you can run the acceleration ramp in steady mode, maybe with quasi-static + time splitting for the mass fraction, so with a bigger time step and reach nominal velocity quickly. Then make a restart in unsteady mode with wave maker on, with a significant lower time step obviously. I have never run comparisons for this to be honest.
Besides, if problems are to occur, it is more likely they will with a lot of unsteadyness, i.e when waves have reached the boat, so then you do not have to restart from scratch in case of a problem. I don't know if you're saving that much time, I just find it more convenient (and if you are modelling different wave conditions with the same grid, you don't have to redo the acceleration).

Anyway here the boats are at rest so this "method" is useless, but I find it odd that the run crashes after only 2 seconds. I doubt the waves have reached the outlet or even the boats yet.
afitriadhy, when you monitor the residuals, after how many iterations do they start blowing up ?
What kind of minimum orthogonality you have for your grid ?

Like romainRH said, monitor closely your run, and stop it when you see things are getting crazy. You'll then be able to post process your results with cfview (or check after how many time steps you run failed, and restart it with a lower maximum time step number so you're sure then run will stop before it crashes).