Hi,

Could anyone advise how to visualise mesh properties (namely aspect ratio) in CFX?

Background;

I'm trying to model flow through one passage of a crossover diffuser (it turns outward radial flow through 180 degrees to the inlet to the next impeller eye in a multistage centrifugal water pump).

I've modelled & meshed geometry with Y+ <2 and 15 inflation layers for the KO SST model. I got a message on meshing that the aspect ratio was excessive and likely to prevent convergence (and to use POST to exam aspect ratio of mesh).

I ploughed on anyway hoping for the best, applied mass flow at inlet, 0 pressure at outlet and set running for ~3 hundred iterations. There's no sign of the solution converging and on examination of the resids I find the maximum (momentum) is two orders of magnitude greater than RMS, indicating a local error. I suspect aspect ratio of course.

The problem is I've spent several hours trying to find a way of displaying the mesh properties in CFX-Mesh, Pre, Solver, Post and in the manual, but apart from the tip in the original error message (use POST)there's no indication how this is done, can anyone enlighten me please?!

Many thanks!

Ianto

I believe mesh statistics gives only the MAX and MIN value of the quality parameters such as Aspect Ratio and angles.

Do you have access to ICEM CFD (companion pre-processor to CFX)? If yes, you can import CFX-pre mesh and analyse the quality parameters and location of bad quality elements.

CFX has no problem with AR < 100 in BL. You may even go up to 200.

One of the suggested solution would be to 1st solve the problem with k-e UPWIND difference scheme (which most often converge). Later use can use the result file as initial condition for your k-omega SST model. This may help you get the convergence!

Hope this solves!

Hi Amod,

Thanks for the quick reply!

The error message from CFX mesh is below -

"Volume elements with aspect ratio(s) > 50 are present in the mesh. It is highly likely that this will give rise to convergence difficulties in the solver. Modify the meshing parameters and mesh controls to reduce the maximum aspect ratio.

Post processing should be used to display high aspect ratio elements"

- I don't have access to ICEM unfortunately. I put a point at the co-ordinates of the max U V & W residuals, but it didn't seem to be in a problematic area of the geometry - far from some very sharp edges I had expected to cause problems.

Good tip re. aspect ration in BL - thanks. The error message would indicate then, that the problem is not in the BL (because the AR it's complaining about is >50). I'll have a play around with the mesh and see if I can improve things and if not I'll try initialising with KE first.

Thanks again!

Ianto

Hi,

Further to this post yesterday, I've tried running with KEpsilon as an easier starting point, but still not achieving convergence. I suspect the problem is still cell aspect ratio, but have spent many hours now playing with settings without managing to fix the problem. Part of the issue is my inability to identify the problematic area to determine whether my changes are having any effect. CFX-Mesh suggests visualising the aspect ratio is possible using POST (see error message below) but I can't figure out how this is done, the help files are far from useful on this front.

Can anyone help with either tips on ensuring suitable aspect ratio, visualising the probelematic cells in the model so I know where the problem is, whether the problem is likely to be in the inflated (boundary) layer or main volume or any other suggestion at all?!

Thanks,

Ianto.

WARNING Volume elements with aspect ratio(s) > 50 are present in the mesh. It is highly likely that this will give rise to convergence difficulties in the solver. Modify the meshing parameters and mesh controls to reduce the maximum aspect ratio. Post processing should be used to display high aspect ratio elements.

To see where your convergence problems are, output the residual fields and postprocess them to see where they are maximum (positive) or minimum (negative). Sometimes convergence is not achieved because of difficult numerics, sometimes because there is a true transient.

you can plot the iso surface of the aspect ratio in post.

as phil said plot the iso surface of residuals and check whether the high aspect ratio and the high residuals are at the same location.

if yes then you have to modify the mesh.

Thanks Phil/Rohit,

The problem was that I didn't know how to plot Aspect Ratio (as it's called in CFX-Mesher). The reason - I was looking for a variable called.... Aspect Ratio. - in Post there IS no variable of this name, instead it's cell volume ratio - fairly obvious with hindsight, but the kind of inconsistant terminology which causes huge amounts of frustration if you're unaware of it!

Thanks again.

Ianto.

To answer my own question to visualise cell aspect ratio-

Go to (in post) tools - mesh calculator - element volume ratio. This is Cell Aspect Ratio elsewhere in the software - same thing. calculate the EVR and it will then appear as a variable which you can plot using isosurface etc.

Regards,

Ianto