How do I get a proper inflated mesh on a "short chord length" thin surface?

Problem is this: inflation layers on thin surfaces increase in "steps" from the leading leading edge, and then decrease again in steps to the trailing edge. But if the chord length of the surface is "short" (I currently have about 10 surface elements along the chord) there's not enough elements for the inflation to increase to the specified 10 layers before it starts decreasing again, causing me to end up with a maximum of 4 layers in the middle of the thin surface!

How do I correct this without resorting to an (unnecessary) fine mesh on the thin surface? Easy way out would probably be to refine the mesh only at the leading and trailing edges, but it still doesn't feel like an elegant solution to me!

Louwrens

Hey Louwrens, How about GGI'ing a structured mesh onto the Thin Surface. Again it will take a bit of doing, nad I'm not sure what the flow will be like at the structured / unstructured interface, but it would enable you to have a constant thickness Layer of cells. Bob

That would mean meshing the main flow region with a cutout solid in which the thin surface's mesh would exactly fit, right?

In terms of total effort, I think this would only be useful if I were to do a number of simulations, where the (I presume) shorter solution time for the combined mesh would be shorter than for a more complex unstructured mesh. If I only ran one or two cases creating a smaller unstructured mesh on the thin surface would probably be my best bet. But then again seeing that I've got time to play with the software... ;-)

Seems I'll have to follow your suggestion... even when refining the grid quite severely the inflation still doesn't look nice, and of course the grid explodes out of my processing capability...

Louwrens, I'd try a simple test first, just to check there aren't any unforseen problems, but it should work. If you use a constant mesh spacing, then it should allow you to match the unstructured mesh with that of the structured mesh, reducing any chance of problems at the interface. I have seen some kciks in the boundary layer when looking at atmospheric boundary layers. There we were going from a structured mesh in the far field to an unstructured (with inflation) int he near field. The the interface we saw a bit of a kick in the velocity profiles, which we reduced by matching the mesh more closely between the two sections. I'm not sure if this will give your model any problems, hence the need to test it. Just thinking of your problem, how will you define the thin surface ? I've never had to do this before using a structued mesh. I'm sure you can do this, but how ? Would you need to defing two solids ? Actually, in build if you do not equivalence your nodes between solids when using the structured mesher, then pre will see the junction between the two solids as an interface, and will define a wall I think. This may well be all you would need for a thin surfcae. This would definitely need some testing. I'll give it a try myself, let me know if you get any where with it, Bob

Thanx for the tip about not equivalencing the nodes. That works for making the thin surface nodes available if you create only one 3D region for both top and bottom of the tin surface. What also works is to create two 3D regions (one on top and one at the bottom). For this case you can equivalene the nodes and the 2D region of the thin surface is still available in Pre. Incedently you can't create subregions when using patran volume meshing...

I'll make sure to try and get the node lined up between the structured and unsructured grids.

Louwrens,

I have created inflation on thin surfaces with succes. What you should do is create an additional surface that extends your thin surface further into the flow domain. Then, you have to inflate that surface too. Of course you should not define a wall there ;-) What you end up with is a decreasing inflation in your domain (with probably bad elements). That is the price for a nice inflation layer on your thin surface. It would even be better to extend the additional surface towards an opposing wall. Then you won't have bad elements at all. However, the price you have to pay is an inflation layer perpendicular to the wall which will result in local large elements and thus high Y-plus values. You can compensate this with a mesh control....

Well, there are many solutions if you know the tricks ....... If you need help, send me your .db file.

Astrid

Thanx Astrid, I can see how this will work. Am currently first trying Bob's GGI solution, one advantage of that is a much smaller final mesh, which currently appeals to me... (and the fact that I've never played with GGI before so it gives me a chance to learn!)

I'll definitely try your solution as well though.

L

Hi, CFD guys, has someone involved into the simulation for the duct erosion problem caused by the solid particles(for example iron particles) in the flu gas? By CFX5.6 it seems that the particle tracking model can handle this problem through wall stress. But I don't know if it says this stress can reflect the effect of erosion, and what's meaning by its unit Pa.?

Thanks for every replying here.

hannah