[noelanish]

I am analyzing blocked outlet cases. So I made like 8 outlets. I need to block 7 outlets (assigning them as wall boundaries) and give the remaining single outlet as pressure-outlet. I will then have to repeat this process for all the pre-conceived outlets.

In order to considerably save 3D mesh generation time so that I don't have to mesh again and again (8 times), I have decided to give an inflation layer on all the pre-conceived outlet boundaries as well. And for each case, I shall give a pressure-outlet for one of those pre-conceived outlet boundaries.

Is this a wise thing to do or will there be side effects?

I am using K-Epsilon. My mesh consists mainly of polyhedral mesh elements.




A simple algorithm to illustrate my case better:

P=Pressure-Outlet at boundary
W=Wall at boundary

Case 1: P,W,W,W,W,W,W,W
Case 2: W,P,W,W,W,W,W,W
Case 3: W,W,P,W,W,W,W,W
Case 4: W,W,W,P,W,W,W,W
Case 5: W,W,W,W,P,W,W,W
Case 6: W,W,W,W,W,P,W,W
Case 7: W,W,W,W,W,W,P,W
Case 8: W,W,W,W,W,W,W,P

[LuckyTran]

I get what you're trying to do and it makes sense. Sort of. There will be some not so nice influences of using inflation layers in flow perpendicular direction, but presumably you can take care of it with running more iterations.

This is why you do a grid dependence study before doing such calculations. However, you are trying to cheat and not do any work so suggesting to do more work also does not answer your question.

Is your geometry symmetric? Might I suggest you make 1 grid with 1 outlet, and simply rotate the mesh?

[noelanish]

Thanks a lot. Yes, you're right, I gave it a run with a pressure surface monitor at the blocked outlet and it was of order 1e+3. I might as well "not cheat". I shall try to rotate the mesh as well.

[noelanish]

May I also ask, how I can perform a grid independence study?

I'm aware of grid convergence index from CFD class where the grid in contention must lie in the asymptotic region. But that is for 2D structured meshes where the grid spacing never changes in space throughout the model.

My present application is using polyhedral cells for a very intricate internal pipe flow.

Thanks.

[LuckyTran]

Originally Posted by noelanish

May I also ask, how I can perform a grid independence study?

I'm aware of grid convergence index from CFD class where the grid in contention must lie in the asymptotic region. But that is for 2D structured meshes where the grid spacing never changes in space throughout the model.

My present application is using polyhedral cells for a very intricate internal pipe flow.

Thanks.

It is the same idea being applied to real 3D meshes. A structured grid does not have to be uniform, you can have a structured non-uniform grid where the spacing does change. Structure or unstructure does not really mean anything. What's important is that from grid to grid you have the same overall grid topology and you systematically refine/coarsen the grid.

There was a recent gridpro blog on grid convergence. It is a cute article. http://blog.gridpro.com/generating-s...ergence-study/