[nickdaish]

Hi all

The problem I would like to solve with FLUENT is the wind-driven flow inside a cuboidal room with two or more openings: flow enters at one or more openings and leaves at one or more other openings (see attached geometry snapshot for a typical example, with 2 inflow windows, and 1 outflow window on the opposite wall). I am a novice FLUENT user (via Workbench) and would greatly appreciate help in setting up a grid for this problem.

My main requirement is to include regions of fine grid (fine division in normal direction) adjacent to all walls to resolve boundary layers. Given the rectilinear nature of the boundaries I was expecting it to be straightforward (if fiddly), probably using a hexahedral mesh (possibly structured). The inflation feature seemed ideal for creating such fine gridding. However I've been unable to work out how to achieve this with the ANSYS Meshing program. I've had very limited success so far: tried various combinations of local controls, but they either aren't allowed or they don't produce what I was expecting.

Sweeping + inflation isn't sufficient, as I need the fine mesh next to all the walls, not just on the walls parallel to the sweep direction. Breaking down the volume into different parts might help in some way, I guess (e.g. corresponding roughly to the jet-like regions inside the room downstream of the windows), but this would be tedious to do for lots of different arrangements of openings.

The nearest I have come to a promising mesh is with no local controls at all! I just used the global settings, and (a) Automatic Inflation over a named selection comprising all the wall surfaces, and (b) Post Inflation. This gives a tetrahedral mesh with a “crust” of finer cells around the walls. Maybe this is good enough, but I can't believe it's so difficult to put together a hexahedral mesh for this problem! Since ANSYS Meshing comes from a CAD background, I guess complex geometries are the norm, not simple ones like mine.

I don't know whether to persist with Meshing, try ICEM or TGrid, or simply abandon all 3 and try Gambit (although note I have not used ICEM/TGrid/Gambit before).

If anybody could outline suggestions for how to mesh this problem I would be extremely grateful – with ANSYS Meshing if this is straightforward or (distant second choice) with one of the other meshing programs.

Many thanks!

[max3.2]

try icem and block mesh. shouldnt be a problem with this geom. 3d block around, ogrind, done.
maybe i can do it 2morrow.
only thing then is you will have the BL on the inlet outlet. there should be a way around that. maybe someone else can jump in here

[nickdaish]

Hi max3.2

Thanks for replying. However, I'd prefer to stick with ANSYS Meshing if possible...maybe you're a Meshing wizard too

[PSYMN]

I suggest trying "assembly Meshing" with "CutCel". Just set smaller sizes on the windows to increase refinement there. You can easily set up inflation layers (boundary layers), etc...

[nickdaish]

Hi PSYMN

Many thanks for the suggestion. I haven't fully explored this yet, but my initial attempts haven't been too successful when there is inflation, although the main mesh they produced was nicely hexahedral.

As a first stage I have been trying to simulate a symmetric arrangement with one inlet window and one outlet window both on the centreline of the room. I am not interested in any processes in the walls (heat transfer), so these have no substance, nor do I want to include any details of the windows - they are just rectangular holes. So I just have a rectangular parallelepiped of fluid with some parts of its boundary as named selections. In what sense is this an assembly? Why did you recommend using cutcell assembly meshing - because of its tendency to produce hexahedral cells? Are there other options to produce a hexahedral, essentially structured, mesh whose cell size distribution I can control?

Thanks!

[PSYMN]

"Assembly Meshing" is just a very unfortunate name that I argued about until I was becoming unpopular internally and decided to drop it.

It was given that name because it can be used to mesh the fluid volume inside or around an assembly without requiring you to extract a fluid volume. However, it can also mesh inside your box.

The only true structured hexa meshing we have at ANSYS is in ICEM CFD. In this case, structured hexa is defined as hexa in terms of i,j,k space.

Now if you are actually just looking for a pretty looking hexa mesh in an unstructured (xyz) format, then you could try multizone in ANSYS Meshing. You may find it works automatically, or you may find that you need to split up some surfaces, etc. to break up the volume better.

[ayoxon]

My geometry is almost identical, other than I'm measuring the temperature in the room from a radiator. I've used cutcell and have used face sizing on the window and radiator. But I can't seem to get the inflation to work, comes up with a warning saying 'inflation created stairstep mech at some locations' and 'assembly meshing inflation meshaing has ignored inflation controls...'

Is there a specific method when doing inflation?