Hi all,
I'm doing multiphase FSI coupling using OpenFOAM and an in-house solid solver, connected by a coupling library preCICE. I'm trying to study the interaction between a caisson breakwater and waves. The fluid domain is calculated using olaFlow and the mesh deformation is calculated by the solid solver.
After the caisson is hit by waves, it deformed and there was some overtopping water on the top on the caisson. Suddenly, the simulation crashed due to the divergency of fluid domain (OpenFOAM). I checked the log file and found after the overtopping water occured, the phase fraction(alpha.water) began to get larger than 1 and kept increasing till the simulation stopped. At first, I assumed the problem is caused by dynamic mesh which affected the calculation of phase fraction. So I increased the Young's modulus of the caisson to reduce its deformation. However, the simulation crashed as before.
Therefore, I wonder if the problem is caused by the instability of the phase fraction (or maybe the turbulence model?). Has anyone faced with this problem? Is there any method to stabilize phase fraction calculation in multiphase FSI coupling with dynamic mesh?
Any advice or suggestions will be appreciated! Thanks in advance!
Hi,Tomasz Waclawczyk
Thanks for your reply!
Of course, I've uploaded some figures to show the deformation of the grid, along with part of the running log (I cannot upload the whole log file since it is too large).
As for Co
A first idea could be to reduce the timestep of FSI coupling, thereby allowing both solvers to get to a tighter convergence.
Secondly it appears, as you are using meshes with jumps in mesh volume / resolution. This is not a very favourit situation for a CFD solver and it can cause divergence issues as well (e.g. on top lid of your deformed obstacle the mesh seems to be substantially coarser and this is the location, where divergence in volume fraction seems to occure). So the requirement would be to make the mesh more isotropic and homogeneous around the obstacle.
Regards,
Dr. Th. Frank.
Hi Thomas Frank and Tomasz Waclawczyk , many thanks for your suggestions!
I'll improve the mesh quality and adopt a smaller timestep to get a better convergence.
Thanks again!
I would suggest trying a single phase version. It's hard to discern what is the culprit here, the FSI or interface tracking
Hi Wojciech Aniszewski ,
It sounds like a good idea. However, I'm not sure how to generate waves using single phase OpenFOAM solver. Could you please share more experience on this? Thanks!
Dear Haoran Zhou , "single phase waves" is not what I meant. Just a single phase flow (liquid or gas only). At least in the former case it could lead to the caisson deformation, thus, the similar physics but w/o the interface. It probably would not solve your issue, only help to debug.
Dear Wojciech Aniszewski , I realized your advice. Thanks again for your help!
- Badges
- Science topic