Hi! I could use some help on my model
As shown in below picture the there are extra curves at the supposed intersection of geometry in my non-manifold sheet solid. I've tried feature removal, Geometry Editing - Project/Move Point but it wont work.
There are 4 surfaces joining at that intersection.
Solved! Go to Solution.
A couple of other general pieces of advice...
(a) The Femap modeller is VERY precise, so if geometry starts off a bit untidy, the consequences can be what you are seeing.
(b) If you are using mm dimensioned geometry for a "normal" sized model (eg. anything from 1,000mm to 40,000mm, say), then from many years' experience with non-manifold add for complex models, the best starting tolerance is about 1e-4, not 1e-6 (too tight, too precise). When you use the command you should always inspect the yellow free edge curves to see if there are problem areas. If so, then maybe Undo and retry with 5e-4 or perhaps 1e-3. These choices can make a difference to what is stitched and what is missed if the geometry is a little untidy, and some tolerances do a better job than others.
(c) If you need to re-check for free edge curves later, then the 2nd top left icon in the Meshing Toolbox allows you to Toggle Entity Locator (not sure why it isn't on all the time). Anyway... Search For: Curves -> Free Edges (make sure your highlighter mode includes Show Labels, eg. by Window | Show Entities... any entity type, but Label with ID = on). And notice that you can use the Entity Locator to find small curves and small surfaces.
In the way you are showing curves with labels on, you should also do the same with surfaces (with curve lables off), so you can see whether those little problems are holes (they will have free edges), or sliver surfaces without free edges (those edges are also used for the big surfaces), or offcut excess surfaces (will also have free edges).
(d) If none of the collection of items I mentioned in the other post work, then you should also familiarise yourself with Meshing Toolbox -> Feature Suppression. Like Combined Curves and Surfaces, Feature Suppresion is also a "fudge" mesh method. In this case small curves or surfaces are meshed as a single node.
Femap does also have some useful cut-and-shut geometry creation, so if those problems are too hard to repair with the scalpel, sometimes I just delete the local mesh and geometry and create some simple new curves & surfaces and add those to the bigger non-manifold sheet solid using the non-manifold add command.
You have a problem with gap tolerance WITH SURFACE EDGES when performing the NonManifold-Add command, when the command finish execution FEMAP will color in YELLOW the free edges informing the user visually if the command was performed properly with success. Only if the gap distance between surfaces is quite big, in this case FEMAP will give error. Please start using the default tolerance 1e-6, only if failed try 1e-5, and so on.
In your case, the only solution I see is to go back to the previous state using command "Geometry > Surface > Recover Manifold Geometry...": the command will take the selected “general body” in your model and separate them into component “Manifold” sheet solids (FEMAP Surfaces). Once you have the surfaces exploded, run GEOMETRY > SOLID > CLEANUP command to remove any extraneous small features which are not part of the original surface.
I suggest to use GEOMETRY > SOLID > STITCH first with planar continuos surfaces, and when most of the surfaces are stitched next use GEOMETRY > SURFACE > NONMANIFOLD-ADD to solve the "T-Joint" problem.
Simply isolating a portion of your geometry:
After doing RECOVERY you have surfaces with reversed orientation of normals (wrong!):
If I STITCH surfaces together the inner edges disappeared (take note of free edges): this is the way, not do perform Nonmanifold of the FULL model, go by regions, OK?.
Thank you for your help!
You mentioned my surfaces have reversed orientation of normals, what is the consequence of this? How may I avoid getting surface of reversed normals in the first place?
While using solid clean up function after I recovered my manifold sheet, some of my geometry disappear completely, any idea what might be causing it?
You have a problem with the SOLID GEOMETRY SCALE FACTOR in your FEMAP preferences: you are using INCHES (scale factor 39.37) instead MILIMETERS (scale factor 1000), you need to be consistent ...
List Solids 187 Solid(s) Selected... Solid 1 - Manifold Sheet Address: 79 Scale: 1000. Solid 5 - Manifold Sheet Solid Address: 232 Scale: 1000. Solid 6 - Manifold Sheet Solid Address: 657 Scale: 1000. Solid 74 - Manifold Sheet Solid Address: 4466 Scale: 1000. Solid 75 - Manifold Sheet Solid Address: 4802 Scale: 1000. Solid 166 - Stitched Body Address: 5138 Scale: 1000. Solid 185 - Stitched Body Address: 5249 Scale: 1000. Solid 267 - Ruled Surface Address: 6216 Scale: 1000. Solid 268 - Ruled Surface Address: 6284 Scale: 1000. Solid 447 - Manifold Sheet Address: 36950 Scale: 1000. Solid 454 - Manifold Sheet Address: 39837 Scale: 1000. Solid 456 - Coons Surface Address: 40384 Scale: 1000. Solid 457 - Coons Surface Address: 40452 Scale: 1000. Solid 460 - Stitched Body Address: 40520 Scale: 39.37
If you have surfaces with reversed normals, then when you mesh surfaces your Shell elements will heritage the surface normals as well, so normals compatibility will not be achieved. Before creation of NonManifoldAdd geometry all surface normals should be oriented in the same direction, if not use MODIFY > UPDATE OTHER > SURFACE NORMAL.
Thank you so much for your help!
Apparently there are many information I am not aware of using femap, could you kindly point me to any source from which I could learn all these stuff?
There's an old joke that says engineers are brilliant with numbers, except if there's a dollar sign in front!
Unless you are a student / researcher with much time and not much money... wherever you are in the world there will be domain experts in Femap (not always the resellers BTW), who can provide you with proper training. Your employer can either spend (waste) $xx dollars an hour while you take weeks or months googling, youtubing, researching, testing and teaching yourself (even with help from this forum) - or you could spend less than a tenth of all that time learning the best way to do things face-to-face from people who have done it all before. If you were in south western Europe or North Africa, Blas could run such training, and whatever he charges would be value-for-money. That's because it doesn't matter how quickly you can teach yourself, you will take far more time (= more cost) than the cost and time spent with an expert, to reach any comparable level of productivity.
That's not directed at you specifically - it's relevant to any (commercial) beginner/intermediate users here.
Siemens and their predecessors have spent decades developing this technology - so there is much to learn, always.