Hi researchers,
I'm doing 3D simulation of a single bubble rising in quiescent liquid with a custom body force, for the sharp interface capturing and decreasing CPU time of the simulations I want to use adaptive mesh of the interDyMFoam and get finer mesh near the interface of the gas-liquid.
But Unexpectedly, the CPU time increased dramatically by using interDyMFoam adaptive mesh than the static mesh and moreover the residuals, Courant number and deltaT experienced severe fluctuations which caused very bad results.
For more information I attached my dynamicMeshDict file.
Any help will be appreciated, Thanks in advance.
Regards,
Arsalan.
CPU time increased dramatically by using interDyMFoam adaptive mesh because you should solve the Laplacian mesh adaptation. Furthermore, reducing the mesh size implies lower time steps to prevent an increase in the Co number.
Franco
Yes I know that solving the extra adaptation mesh equations increases the CPU time, but the number of cells greatly decreases than the similar static mesh.
So with this considerations, what is the benefits of adaptive mesh than the finer static mesh?
That you increase significantly the accuracy only where needed and the simulation, for example, need less resources such as ram memory.
Franco
It's strange that you have cpu problem on mesh, where maxCells is 50000.
Is your processor multicore? Did you tryed to run case in parallel mode?
I guess that bubble movement per time step is small. If that's true you don't need to refine mesh in every time step and you can increase refineInterval.
- Badges
- Science topic