[Bou Med]

Hi dear researchers

I'm working on the IBM method, do you have any idea about the best document to learn this method ? and which method of IBM is the best to simulate a flow with a rigid cylinder inside ( Ghost cell, Cut cell, Forcing Function etc...) ?

Thank you

[FMDenaro]

Originally Posted by Bou Med

Hi dear researchers

I'm working on the IBM method, do you have any idea about the best document to learn this method ? and which method of IBM is the best to simulate a flow with a rigid cylinder inside ( Ghost cell, Cut cell, Forcing Function etc...) ?

Thank you

http://www.annualreviews.org/doi/abs....061903.175743

[Bou Med]

Originally Posted by FMDenaro

http://www.annualreviews.org/doi/abs....061903.175743

Thanks a lot

[arjun]

i once made a post about ghost cell method on cfd online. Try to search it, it was good summary of this method.

What are you solving with immersed boundary method?

[sbaffini]

Originally Posted by Bou Med

which method of IBM is the best to simulate a flow with a rigid cylinder inside ( Ghost cell, Cut cell, Forcing Function etc...) ?

It may seem stupid, but it pretty much depends from the way you build your grid and the effort you are going to put on it. The numerical method you are going to use will also play a role (finite volume, finite difference, spectral). In contrast, i'm unsure about the existence of a best, overall, method in terms of accuracy, stability and complexity.

Cut cell basically means performing a series of geometric operations very similar to the actual body fitted gridding. Considering the effort, i would just make a grid generator instead of spending time on the cut cell.

The forcing function is very easy to implement on any grid and has some advantages for moving objects on static grids, but you have serious limitations both in terms of stability (the forcing function tends to be stiff) and flexibility (how do you implement, say, a wall function with a forcing function?)

Similar approaches based on field functions (level set/vof) also exist (e.g., gerris and flow-3d), but they have the same pros and cons (except, probably, for the stability).

The Ghost cell approach seems to be among the most preferred, especially for finite volume methods. Basically, you end-up handling immersed boundaries pretty much the same as classical, body fitted, ones.

Ghost cell means that, trough interpolation/EXTRAPOLATION, you provide values for cell centers outside your fluid domain. The extrapolation is due to the fact that, by construction, your cell center is outside the fluid domain. Some variations of the method, instead, fix the values either on the domain boundaries (faces of the cells) or in the first fluid cell inside the domain. I tend to prefer the latter, which avoid the extrapolation part.

In the end, the overall method is full of small details which can make the difference. You need to do it by yourself to really understand them all.