[wuwenbo]

I'm using "velocityLaplacian" to move my mesh. Can you tell me which boundary condition you used instead of "movingWallVelocity"? I can't find a suitable boundary condition in the OF 2.3.0.

[pi06jl6]

Originally Posted by wuwenbo

I'm using "velocityLaplacian" to move my mesh. Can you tell me which boundary condition you used instead of "movingWallVelocity"? I can't find a suitable boundary condition in the OF 2.3.0.

movingWallVelocity works fine for me, I am using displacementLaplacian with inverse distance, I have implemented support for velocityLaplacian but havent used it .

[wuwenbo]

However, I resolved the pressure oscillation: Dont try use movePoints twice in a subiteration, just move the points and assume that the movement is sufficient smaller than the cell size, adjust the time step so that this assumption is valid.

I have review your post about FSI yesterday, you don't using "movePoints" or "mesh.update()" in subiteration, and using "movingWallVelocity" still. So, I'm curious about how the effect of displacement of the solid can be act on fluid in each subiteration. I have done as you mentioned, and observe the "Up" in "movingWallVelocity", I find that "Up" don't change until the mesh is update.

The "movingWallVelocity" is the boundary condition of the interface for fluid computation. So I think the parameters in "movingWallVelocity" don't change mean that the fluid domain is just computed once by itself (not Fsi) in the subiteration. Is that right? So, i hope you can tell me how you act the effect of solid on fluid in each subiteration very much.

[wuwenbo]

if your case can still be computed in "weak couple" (just compute one coupling in each time step) way?

[pi06jl6]

Originally Posted by wuwenbo

I have review your post about FSI yesterday, you don't using "movePoints" or "mesh.update()" in subiteration, and using "movingWallVelocity" still. So, I'm curious about how the effect of displacement of the solid can be act on fluid in each subiteration. I have done as you mentioned, and observe the "Up" in "movingWallVelocity", I find that "Up" don't change until the mesh is update.

The "movingWallVelocity" is the boundary condition of the interface for fluid computation. So I think the parameters in "movingWallVelocity" don't change mean that the fluid domain is just computed once by itself (not Fsi) in the subiteration. Is that right? So, i hope you can tell me how you act the effect of solid on fluid in each subiteration very much.

Correctly, but ... why do you concern with moving the points without update? I wrote that it gave reasonable results only for small time steps and the larger the boundary velocity less added mass where observed, hence a poor choice.

I am little unsure about what the issue is here. To answer your question : Look into the code where does the points enter in the calculation, then you get the answer, it does affect, but it works only for when the changes is small.

Let me remind you that when i first implemented FSI in 2008 everything worked nice, no pressure oscillation, then feb 2014 when i used OF 2.1 everything went wrong, and quickly i realized that it were due to swapping of pointer in movePoints and it were confirmed by others on forum that that version of OF didnt support subiterative FSI (only weak). So, then i did an assumption: for sufficiently small time steps the same assumption as one do for pressure can be done for U for the FSI purpose. Thus, the text above about moving the boundary points without movePoints/update is an approximation (first order, crude) that is used in the version of OF where the swap of the pointers in movePoints prevents you to use the update twice during same time step. I strongly discourage you to use that, partly since not all fields are updated properly as you pointing out yourself and which i also state above. I strongly advice you use movePoints twice instead while you using older version, since I got a letter from one OF user that he used that trick for his work with his thesis.

In current version of OF you can use update in subiterations. I have verified my code by benchmark and i can assure you, just use pimpleDyMFoam, insert your Solver function and couple the solvers. Everything will purr like a cat.

[pi06jl6]

Originally Posted by wuwenbo

if your case can still be computed in "weak couple" (just compute one coupling in each time step) way?

Do mind we take this off-forum instead? Like over email?

[pi06jl6]

For whom that reads this post:

I do receive emails on this subject and questions how to resolve the issue with pressure spikes on strongly coupled FSI, my answer is the following,

0. I have only used pimpleDymFOAM.C from OpenFOAM, i am using DEAL.II package for the solid part. I dont have adequate knowledge to answer properly regarding OpenFOAM, most of my time is adressed to build different FEM solvers and preproccessing/postprocessing tools which is outside OpenFOAM package.

1. I do not give out my research code, perhaps snippet code. The reason is to protect you from cut your finger while using my code.

2. Do use the latest version of OF, this sort most of the issues.

3. Mesh update is the last thing you do per subiteration.

4. Regarding stability issue, use relaxation techniques.

If the administrator could lock/remove this thread, i would appreciate this.