Which are the best meshers?

[pedrorb]

Hello everyone

I am at present making employ of PATRAN as pre and post processor of NASTRAN, both from MSC. Although the software enviroment is friendly and there are are many possibilities for loads and BC conditions, I am not very happy with the mesher, as the meshing options are not very wide. Specially, I don't find ways to re-mesh certain areas with smaller elements after a first mesh is done.

We at our firm are considering some other software to mesh and then export to PATRAN toget the model finished.
Which are the best meshers in the market?
Best: versatile, 2 order tets, quick...

Thank you all

PedroRB

 

[nikolas]

try altair's hypermesh

cheers

 

[gillesD]

Hi pedrorb,

I've been told Hypermesh is very broad efficient but not so good for meshing from a user. I am actually working in UG/structures v18 and I've been impressed by their latest mesh technology for auto tet10. I also beta tested V19 (announced for July) and that's even better for 3D and 2D. Plus, you get all the good stuff from parasolid function for tweaking the geometry, midsurface and simplification of small blends (that are not in Hypermesh). But Hypermesh seems to be more capable for PHD users (CFD,...).

I am now looking for more advanced stuffs (non linear and dynamics) and I am wondering about FEMAP for Nastran. I had a demo and it seems easy even for hexa (brick) meshing and it is cheap. Any other feedback ?

Thanx,
Gilles

 

[zuardy]

I have used FEMAP version 7.0 for NASTRAN (NASTRAN for Windows) too and it has a good mesh options for 3D (Tetmesh and Hexmesh) like what Gilles said. But if you want to find more about mesh algorithms or mesh programs, please have a look at

http://www.aue.auc.dk/education/m/mesh-links.html

good luck !

regards

 

[brad]

Like nikolas, I've also been very pleased with Hypermesh.
Brad

 

[ISEAG]

Just a quick note on Hyper and FEmap. I haven't looked a the latest version of either......I am an ANSYS grinder..........But each, as most have at one time or another had difficulty with the tet meshing in thin field volumes. They had a tendancy (again, last time I tried them) to create very long elements to fill a void. These elements often stretched "inches" across a span when it tried to close the volume. You wound up with what I called Ice pick elements that were, of course, useless.

 

[racecareng]

ANSA is absolutely the best pre-processing tool in the world. It has the fastest mesher I,ve ever seen. You can easily build midsurface models, futhermore it has a really good tet-mesher.
For further information contact the software vendor

http://www.lasso.de

 

[feajob]

Pedrorb, we had the same problem last year. We use MSC Soft wares too. I am completely agree with you. Patran Mesher is not perfect, at least, it doesn't reflect the latest technology in FEA. We got HyperMesh last year. Based on my experience, HyperMesh is very powerful for clean up of the geometry (comparing with Patran). But, we need to apply sin (or cos) pressure distribution on lugs, so, we have to import our mesh into Patran again. Our main problem, is that, we don't have any more association between initial geometry and our mesh (prepared in HyperMesh). In other words, clean up and meshing is easier for us, but applying sin pressure distribution in Patran on the faces of elements is a time consuming task. In HyperMesh, I don't see any similar sin pressure distribution (PCL Function in Patran). I think that personally, if MSC guys buy a good product like HyperMesh and integrate it into their software, we can save time.

AAY

http://www.geocities.com/fea_tek/asd/

 

[johnhors]

feajob

A general purpose program that creates a quality mesh (tri and quad on surfaces, tets in solids) BUT also has simple to use tools for applying sin or cos pressure distributions on lugs/pins (or better still the empirical formula invented by Gencoz at Boeing in 1980 that better describes the pressure distribution between a pin and a lug which is the default) is available at

http://www.roshaz.com

Nastran (or Abaqus) meshes can also be imported, have the lug loading applied and then be exported.

 

[feajob]

Thank you Johnhors, I am going to try roshaz software.
AAY

 

[brad]

johnjors--
I've never heard of roshaz before, although I have observed that there do seem to be many pre-processors in Europe that aren't widely used in the US.

How widely-used is roshaz over there? They don't mention customers or too many details on their site. Just curious...

Regards,
Brad

 

[johnhors]

Brad

It's been going for just one year and has a handful of users.

 

[nodepicker]

Hi,
I have tested Patran, I-deas, Hypermesh, Ansa and a little Femap. In my opinion Ansa is by far the best mesher apart from when you mesh a lot of solid models. Shell meshing is the strongest part in Ansa. If you have used Ansa, you don't want to use anything else. It is the only program I've ever used where it is evident that there has been a deeper thought behind the layout of the program, location of buttons etc. This might, however, be a problem too as you work too fast and "clicks" too much in a day resulting in pain in your neck and elbow. Believe me, this is not a joke ! Still Ansa would be the choice for me.

And finally, I do NOT sell Ansa !

 

[sean01]

I think you are going to have as many answers for this as there are pre and post processors !!

Have to say Ansa is the choice for me - its incredible and in the benchmarks I did last year, between similar trained people on hypermesh and Ansa - the Ansa tool delivered results in between 30-50% of the time the hypermesh guys did. It seems to have good connectivity managers, reliable representation of finite elements that read through in the solver, good penetration checking tools and good topology repair tools. In the new releases, there are some pretty cute functions to do with mesh morphing (parameterising) which when they get fully developed will be on a par with SFE concept. Beta systems post processor metapost is a pretty reasonable freebie as well - I personally found it easier to get on with than animator.

I understand though that IDEAs and unigraphics both have good meshing and section tools in them. In my opinion the mesher in abaqusCAE is pretty poor by comparison

Sean

 

[Drej]

Also check out AI*Environment. Advanced level pre- and post-processing tool from ANSYS with links to industry solvers (NASTRAN, ABAQUS, ANSYS...), originally developed for the CFD market and marketed as ICEM - now part of the ICEM CFD software range. Quite expensive but very, very good.

-- drej --