CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)

- Siemens (https://www.cfd-online.com/Forums/siemens/)

- - Help proam (https://www.cfd-online.com/Forums/siemens/55130-help-proam.html)

i am not able to remove mesh check failures like cpmatches is there any systematic help availble for the mesh error removal thanks in advance

I think cpmatch failures are the playthings of the devil. I have been told that one way to reduce them is to increase transition growth in the custom mesh generation template. I have never had that work for me though.

One way to fix them is after you have completed repairing unresolved cells (these are cell type 2, by the way), change your cpmatch errors to cell type two and repair as you would unresolved cells.

If anyone has a better method, I am all ears.

Tom

Yes, God damm cp-errors. You are not alone.

Please be aware there is no automatism. StarCD makes it exactly as you say it, BUT NOT MORE (!!!). That's the key note to keep in mind.

- put failed couples into a couple set.

- generate a cset of the cpset.

- be sure to have ALL SLAVE CELLS to all master cell in your small cset. If there are some slaves missing add it to the cset (choose a vertex in a vset, than cset add vset any). To make it more easier isolate only one master and the 4 slaves.

- a very good help is to number the couples and/or the cells (main menu -> plot -> number -> ...), and see the listing of the cset/cpset, to be sure which couple belongs to which cell and which slave-couples.

- Problems arise when you have NOT ALL slave cells to a master. So, be sure to delete alle slave-couples from a failed couple before you generate a new couple. When you generate a new couple the old possible existing couple of a slave cell is now twice. This causes a new error. (Like a hydra, cut one head, and 2 other heads grow out of the neck)

- Befor you generate a new couple of 4 slaves and 1 master, delete all existing slave and master couples. Check the listing / numbering. But when you delete all, it can happen that you destroy other couples in the cset, which are OK. So complete all other slave couples to cset. It doesn't matter, if the master cell is once more a master cell for 4 other slaves. The problems arise from slave cels.

- So be carefull, be sure what you are doing, think about what have you done. StarCD DOESN'T save exisiting working good couples from unwanted deleting/destroying. That's the problem.

Try this advice. If you need some more infos, tell me.

Thanks for the help i will definitely try is there systematc help manual or help available for surface failure errors removal (in Proam) and mesh error removal (in proam) help.since the userguide provides vague information about only few of the errors to correct in case of surface as well as mesh

If we're only dealing with the number of bad couples fixed, I would be considered a master sifu. That being said, I think that just indicates I'm a sad individual. Below are a couple of observations/techniques that I use.

If you're an ES-Aero or ES-Uhood guy, don't worry about the bad couples. They will be taken care of, read deleted, when the geometry file is written. The added faces/blockage is assumed to be negligible compared to the overall flow field. I agree that it's somewhat weird to say 'some of the mesh that was created is crap but where it's crap it just doesn't matter'. However, I'm a believer. The vehicles I'm working on are somewhat large, ~60m in length. The cells on the surface, where the questionable cells and couples are, are between 1 and 32mm on a side and less than that in height. That's bug splatter. That's road grime. It just doesn't matter.

Now the question is can I live with that philosophy in my other models? Not all of the time. Below is what happens when I'm being picky.

I've been told to collect the bad couples and modify them into a couple type that is defined with partial boundaries. I have found that this technique will eliminate some of the problems but will introduce others.

Some couple errors are caused by the 'fine' side capturing more geometry than the 'coarse' side. Generating your template with curvature based refinement turned on can push the couple interface into flatter regions. Getting beyond the sharp curvature can also be done with a judicious use of cell types and cell type based refinements.

When I'm excising the cancer, I do not initially delete the fluid cells. I turn them to solid. Later I use; cset newset solid cpset newset cset any/all to find the couples that need to be deleted.

I have made some buttons that help me generate new couples. Basically I use the *set command to save the cell numbers of the cells that will be used in the couple. I'll then issue the cp command using the parameters I have just defined. This way there is no question as to which cells are being used in the couple definition. In addition, I don't have to change my cset or plot too much to use the buttons. I have found that if I use the currently in vogue couple create command, I need to be very careful with the cset or couples that I wasn't anticipating will be created.

And finally, vfil vcrs is your friend.

I'm sure that there are other techniques. And as with the other posters, I'm all ears.

Now please excuse me while I go pray to the CFD gods to keep my unresolved cells and bad couples to a minimum.

I have a special ProAM Training Tutorial, but only on real paper. It is a little bit helpfull.

Concerning surface errors. It't a liitle bit ago, since I ave made a big mesh. I I try to remember:

- triangulate the importat surface with the "resurface" option. - improve tetrahedral quality - possibly smooth the surface with vsmooth, but be AWARE after smoothing the geometry has sharinked a little bit, so you have to map the new surface to the original one with vproj. Be aware of the edges, Map this in a separate process.

After this you really got the most perfect surface for meshing. With this strategy I have meshed already really complex geometries.

Couple problems can often be traced back to dodgey bits of surface mesh, so it's often worth getting the bad couples (and their associated cells if you want) into a set, then add the surface mesh and take a close look at what's around the problem areas. If you've got a bad surface mesh (e.g. pieces of geometry that are much smaller than your cell size, that can cause you problems (you're basically trying to mesh something with 1 mm dimensions with an element that's 10 mm in size!).

As already suggested, you can fix this with pro-STAR's surface wrapping/re-meshing.

Look out for meshing in STAR-CCM+ V2 which has the latest and greatest surface wrapping and re-meshing tools.

1. OK, you are right belonging to vehicle aerodynamics. It realy doesn't matter if there is a little wall of about 1mm² area (possibly a slave cell without a couple). The problem becomes more serious when the flow speed arises, so that you have to calculate the problem as compressible (mach > 0.3). I normaly calculate very fast flows (1D3D-coupling for intake and exhaust manifolds) with mach no of around 0.7 (intake), and shockwaves at the exhaust manifold. When the mesh is not proper in such flows, you can see this at the pressure loss. In worst cases the missing couples can cause shockwaves. Be aware: mach 0.8 (border to transonic flow) is NOT 80% of mach 1. The physical behauviour is as 95% to mach 1. Only little disturbances causes a relevant pressure loss.

2. I kown this technique, but I don't agree with it to collect all failed couples, deleting them and make partial boundary couples. You always have to be carfull to delete ALL slave couples of a failed couple. But I do a similar method: In generell I give all failed cells a new cell ID. One seperate cell ID for one kind of failiure. So it is very convinient to type cset news ctype xxx, get get back all failed cells. From this set a take a subset, to make the screen more clearly. When the failed cells / couples are repaired I give to cset the cell ID of the good cells. This technique prevents you from checking and checking the grid to get the next failiure, which takes much time.

The key idea remains: delete all parts of a failed couple (means all (up to 4) slave parts and the master part) AND only this parts, and NOT MORE / OTHER PARTS OF other good working couples. Be sure to have all slave cells in the cset belonging to a master BEFORE generating a new couple. Be sure not to affect good couples as they would be destroyed.