Improve meshing quality
 

Greetings!

I have reached a point where my poor meshing holds me back (multiphase analysis demands high mesh quality). Let us assume a simple water tank with 1 inlet and 1 outlet (1m height 1m diameter). What must i do to improve it?

Here's a picture of the the fluid domain in question:

http://i59.tinypic.com/2vwv1i9.png

Thank you in advance!

Hmm... it's seems your mesh consists of 0 elements. LOL

Yes, zero elements appears to be a significant meshing problem.

You are really asking for a mesh quality tutorial and you are unlikely to get that on the forum. I don't have time to write that. I can suggest you do the tutorials, even better get some training, try lots of options to see what happens and read the literature to see what others have done. That is how we sorted it out and I suggest you do likewise.

What i was really looking for was pointers. Here you need inflation, there body sizing etc. I keep "meshing" up...:)

Appropriate mesh settings depends on many many factors. You need to know what regime the flow is in (Laminar, turbulent), how accurate you want to be, any complex physics (multiphase, radiation), what the basic flow pattern is likely to be and many more - and that is just before you can make general comments like whether inflation is required, if tet meshing will be OK, body element size etc.

It's about a multiphase turbulent flow. Water starts pouring into an air filled tank. Tank's floor is being steadily heated. Mesh is, at least i suspect so, creating numerical instabilities and, therefore, must improve its quality somehow.

Is surface tension important?

Free surface models have a much higher mesh quality requirement than other models. They also benefit greatly from high quality hex meshes in the region where the free surface occurs.

So this geometry may be best meshed with a high quality hex mesh done as an extrusion from the top to bottom, with the two ports added as GGIs.

I see your point and thank for your answers. Evidently, it's much more complex than i thought. Could you propose any viewing reading material? I tried ANSYS tutorials but i find them quite convoluted to say the least.

I know the free surface stuff from personal experience. A really good way to see how things go is to simplify the concept and try different mesh types. So in this case maybe filling a rectangular air-filled box with water pouring in from the top. You can easily mesh this tets, prisms (triangles extruded up from the bottom) and hexas (and with a variety of aspect ratios).

Run them all and see for your self how they go. And don't forget to have a look at the free surface modelling parameters as some of them make important differences too. A bit of time spent working out the best way to model something will mean you can model the real configuration with confidence.

And is surface tension important? That is the model with the highest mesh quality requirements of them all.

For the sake of simplicity, surface tension is not important.

Here's an overview of the automatic mesh:
http://oi60.tinypic.com/29zpsi8.jpg

and here's a closer look on the outlet:
http://oi59.tinypic.com/2hwdyt4.jpg

How could i possibly improve this?

From what i understand, i must use inflation on the walls and at the same time try to "align" interior nodes with the flow. Exterior seems OK but, so far i haven't found a way to shape up the interior.

Inflation on the walls is recommended for turbulent flows. Your flow will not have a major turbulence component (from what I understand), your major complexity will be a free surface. Free surface models benefit greatly from hex meshes with good aspect ratioes.

Also your interior mesh is way too coarse for any useful modelling.

So if you are meshing in workbench:
1) Split this body into its three bodies for each cylindrical component. Do not make them a multi body part, make them separate parts.
2) For the main region use a sweep mesh from bottom to top using hex elements. Make them aspect ratio approximately 1.
3) For the inlet and outlet ports do a inflation mesh with a tet core.
4) Connect them up with GGIs.

You are enthralling! Thank you so much!

I do try to please. But I rarely enthral :)