Hinweise zum Import von: / Hints for importing:
STEP, STL, AutoCAD-DXF, NASTRAN, ABAQUS, ANSYS, COSMOS, Z88

Moderatoren: ccad, mz15, auroraIco, Lehrstuhl

vom Christophe
#1132
Hi,

I am working on a project that has me stresstesting a frame for a compressor, well, I'm not going to bore you with the details, but I was trying to accomplish this with Aurora. The frame is designed using Inventor. It isn't extremely complex, it basically consists of a central square platform, with extra support on the bottom to reïnforce it and make it more rigid while trying to keep its weight to a minimum. I'll try to add an image on the bottom to illustrate the exact layout.

The problem is as follows: whenever I add complexity to the frame - for example, by making holes, adding more strutts,... - I am well aware that this adds to the complexity of the eventual imported step/stl file. What bothers me, is that I seem to be unable to make a relatively "clean" structure. What I mean by this, is that whenever I import the frame, the basic form that is created is allready quite a mess. I believe this to be the fault of Inventor, as I can see on the screenshots that it is perfectly possible to make very clean objects, but although not extremely complex, it is still too complicated to build in another way than with a solid CAD program. Is there a "proper" way to designing a structure, in such a way that the resulting .stp/.slt is the cleanest way possible, or do I just have to bite trough and work with what inventor/meshlab hands me?

Added below is how the resulting Netgen Mesh looks. As you can see, because of the roundness of the supporting struts, all detail is lost and it just appears as a black "blur" which brings me to the second part of my question. The frame, as you see it, is nearly complete, but in fact, almost every corner of it has to have some kind of "fillet" to it because of the construction process of the steal. (it is basically a steal plate, molded into this form) Adding this detail to the frame, however, makes it impossible to work with. The complexity suddenly skyrockets and all I am left with, is a black square and a nearly frozen computer, which makes it impossible to make any kind of stress assessment. The initial part of that sentence isn't really the problem, as I can perfectly let my workstation run for a long time to come up with an answer, but it is the handling that makes this an impossible venture. Is there a way that makes me able to mesh the structure, and switch over to a sort of low-poly model to make it handle better? Because when all detail is added, I can't even reliably rotate the frame, let alone select nodes on which I want forces to act. All the while demanding 3+GB Memory.

I ask these questions, because I can see in your examples that it is perfectly possible to make clean meshes out of complex items, if I am missing some vital point, I apologize, I've been working with Z88 for a good week now, so when giving answers, keep that in mind. Also, if a similar question has already been asked, I'm afraid I haven't been able to go through all the threads yet. I understand some German, but it takes me a long time to do so. :)

Bild
http://nlt.dommel.be/Shrimp/Bach/framemesh.png
Benutzeravatar
vom flonue
#1136
Hi Christophe,

it is possible that the Netgen-Meshersuppresses some details of components. My first suggestion would be to use the Tetgen-Mesher.

Best wishes

flonue
#1137
Thanks for your answer, but I think you misunderstood. What I meant with loss of detail, is actually the opposite. The mesher adds so much triangles, that I can almost no longer see the underlying structure. So the detail is still there, it gets simply enormously fine.

Also, Netgen seems to be the only one of the two with which I can mesh the structure. Tetgen just makes the entire frame one black collision of points. In the meantime I've gotten Netgen to Mesh the entire structure - with roundness in the corners. It handles sluggishly, but that is a bit expected. But still, I find it odd that, even with the maximum length of 10000 inserted in my mesh, that it creates relatively small triangles on the big expanses on top of the frame.
Benutzeravatar
vom flonue
#1144
Hi Christophe,

I think the mesher tried to mesh the part with all its details. That is why the mesh is quiet fine. Are all these details important for your analysis? Perhaps you can model your part without all details. Unfortunately the only parameters we can set for the mesher is the element size and coarse factor.

Best regards

flonue
vom Tim
#1171
flonue hat geschrieben:Hi Christophe,

I think the mesher tried to mesh the part with all its details. That is why the mesh is quiet fine. Are all these details important for your analysis? Perhaps you can model your part without all details. Unfortunately the only parameters we can set for the mesher is the element size and coarse factor.

Best regards

flonue

Hi - I have had quite a few similar problems with corner fillets and other 'radii' or 'spheres' or 'cones' where the auto mesher over refines the mesh in my opinion way too much. Perhaps as above you could delete these details HOWEVER, in your example above, i think you have intentionally 'RIBBED' the part as your interested in (maybe) comparative deflections with or without the ribs AND therefore deleting these isn't an option. I have found in the stress studies that i have done, an answer is to replace the curves/ radii with a linear interpretation. So, replace the semi circle rib section with an appropriately sized triangle section or to refine this further, add another side so to flatten off the triangle rib top edge.
I, like you are a new user. I can't promise you that this is an approved technique - all i can say is that it has overcome similar frustrating issues i was experiancing. The meshing is much faster and the solving runs much , much faster with no crashing!

hope this helps you.

Regards.
Tim ( very much a new user - there's alot of learning to be done!)

ps - i have a non-history based modeller, so i can easily manipulate my geometry to suit or just make a new file just for CAE purposes. Appreciate that with a history system it can be a problem, but with 'Fusion' ( if you have it), then changes can/could be easily made (in theory at least).

Hallo, vielen Dank, das werde ich ausprobieren mi[…]

Start

Hallo Schlorchi00, leider konnten wir den Fehler[…]

Hallo liebes Z88 Team, gibt es bereits Neuigkeite[…]

UPDATE Nachdem ich es nach weiteren Tests leide[…]