Mesh for race car
 

Hi!

First of all - great forums! I've learned so much from just reading some of the threads. Really helpful. This is my first post here, so hello to everyone :)

Anyway, I have some problems creating mesh for calculating Cx factor for an electrical race car. I'm doing this as part of my m.sc project but my tutor doesn't know icem at all, just ansys mesher, so he kinda can;t help me much with this.

The main idea of mesh is that - I've divided domain into 4 parts: front, top, bottom and back. Front, top and back are ment to be hex, and bottom, where the actual car is, tetra and prism.

The biggest problem I encounter is bad qualit mesh after creating prism. I've read some of advices about creating prism on this forums, and also some other sites on the inetrnet, and I know it's crucial to have good quality mesh of tetra before creating prism.
So the qualit of tetra is acceptable i think - there are no errors, and quality doesn;t drop below 0.2. Still after creating prisms, some elements have really bad quality, especially near the last layer of prism, and also near the edges of the cockpit (cockpit, wheels, and drivers head are not set to prism).
I trided smoothing bad elements, with Laplace and then without in few cycles it doesn't help much. Refining bad elements and then smoothing is even worse. Any advice how to deal with it?

Second problem is that i don;t know how to merge hexa mesh, created with blocking with the tetra-prism part. For the time being i used hexa-core instead.

I would apprecieate any help with this.

Here are some more pics.

Keep in mind that the quality criteria of prism is not realistic/harsh. So prism with quality 0.01 should be Ok (see post #15 in this thread http://www.cfd-online.com/Forums/ans...rism-mesh.html)

By the way, whats the min quality of prism in your case?


For merging see this post : http://www.cfd-online.com/Forums/ans...-geometry.html. Search other post, where Simon describe the procedure in detail.

Thanks for quick answer :) Unfortunately I can't try your suggestions, cause university cluster is down right now. I'll check it later.

Anyway, I brought the quality issue up becacuse, whenever i try to run calculations in fluent they don't converge and eventually float point error occurs. Mesh quality was my first suspect :) Maybe quality isn't the matter.

Well, finally I've been able to merge hex mesh with tetrahedral, but still I can't make tetrahedral elements on prism boundary with good quality.
And another thing, why does the prism handles one edge of geometry differently than other (look at the file attached).
On the left side, where the cockpit edge is, the prism is very steep. While on the right, it goes down very smoothly. Do I have any control on how does the prism handle this edges?

I've dealt with prism edge handling but another problem came out.
This is the region with the worst tet quality now. You can clearly see why. Any ideas how to set paramters so the prism will fill the whole curve region and wont stop? I've attached my current settings.

Simon puts online some documents which explain how to deal with prism.
So try to do some research on this forum for the link, or ask Simon about it (I can't attach those documents because of the size).

OK, thanks. I'll look into that.