Problems with "Precise Match" while meshing (mesh density not applied)

[maxmarautodesk]

Dear community,

I am trying to mesh a part in Moldflow. I plan to have the part, two mold inserts, and two 3D cooling structures in the model. In order to get good results, I have to apply different mesh densities for all aforementioned elements and I would like to use the "Precise Match" function for meshing.

However, the latter is where the problem arises. Applying the mesh densities works fine. Meshing the part with the default setting of "Ignore contact" also works fine and the mesh looks like intended by the different mesh densities. If I choose "Precise Match" instead of "Ignore contact" for the part mesh, the mesh is completed, but the mesh densities that I have set are not applied. Moreover, the analysis log gives me some information that there were two meshing attempts and something went wrong with the meshing. Hence, an alternative method had to be used. Please find the complete analysis log below. I have no clue why there the software needs two attempts (for "Ignore Contact" only one) and what happens using this alternative meshing method.

Has anybody experience the same? Any suggestions or ideas how to solve the problem?

I appreciate all help, since I have right now no way to proceed with that problem!

----------------------------------------------------------------------------------------------

Copyright Autodesk, Inc. All rights reserved.

(C)2009 2010 2011 2012 2013 2014

Portions of this software are covered by U.S. Patent Numbers 5,287,408 and 6,096,088.

　

Version: ami2015-promethium_compile_windows-x64  (Build 14104-714)

         64-bit build

    Mesh running on host: XXX

        Operating System: Windows 7 Service Pack 1

          Processor type: GenuineIntel Intel64 Family 6 Model 42 Stepping 7 ~3392 MHz

    Number of Processors: 8

   Total Physical Memory: 16340 MBytes

Mesh with the following settings:   

  Mesh Type:                                            3D

  3D Mesher:                                            Advancing Front

  Global edge length:                                        0.5000 mm

  Merge tolerance:                                           0.0100 mm

  Enable chord angle control:                           Yes

  Chord angle:                                                10.0 deg

  Percentage of minimum curvature size vs global size:      20.0000

  Growth rate:                                               0.0000

  Mesh on assembly contact faces:                       Precise Match

  Sliver Removal:                                       No

  Minimum number of elements through the thickness:     10

  Maximum edge length in thickness direction:                0.5000 mm

  Use surface mesh optimization:                        Yes

  Use surface mesh matching:                            Yes

  Automatic tetra aspect ratio control:                 Yes

  Even distribution without biasing:                    No

  Bias ratio:                                                1.2000

  Mesh smoothing:                                       Smooth all nodes

  Extra refinement near gates:                          Yes

  Relative edge length around gates:                        20.0 %

　

This process will not mesh mold blocks or 3D cooling channels.

If you wish to mesh mold blocks or 3D channels, please include Cool(FEM) in analysis

sequence and then click Create 3D Mold Mesh or Create 3D Channel Mesh.

Generating CAD mesh

First attempt:

Second attempt:

Trying to recover from previous error(s).

Retrying the mesh process using a fallback tessellation method.

Mesh density defined on faces will not be used by this approach.

  Start region: 2

    Percent   5 done ....

    Percent  10 done ....

    Percent  20 done ....

    Percent  25 done ....

    Percent  30 done ....

    Percent  35 done ....

    Percent  40 done ....

    Percent  45 done ....

    Percent  50 done ....

Removing disconnected nodes

Removing disconnected nodes

Meshing Stopped

A surface mesh has been created. Please check the mesh by diagnostic tools and

make sure it has no free edges, no non-manifolding edges and no intersections.

A 3D mesh can be generated by launching meshing again.

Meshing complete

Removing disconnected nodes

Removing disconnected nodes

Execution time

   Mesh commenced at        Mon Jul 21 10:19:13 2014

   Mesh completed at        Mon Jul 21 10:19:31 2014

   CPU time used                       11.64 s

   Total elapsed time:                 17.54 s

----------------------------------------------------------------------------------------------

Background information:
OS: Win 7

Moldflow: Synergy/Insight 2015

Analysis type: Cool(FEM)+Fill+Pack+Warp

Elements: not available, since meshing does not work properly

Hardware: i7-2600 3.4 GHz, 16 GB RAM, AMD Radeon HD6450, 500 GB 7200rpm HDD

[maxmarautodesk]

I should add that the meshing fails, when choosing "Fault Tolerant" instead of the other two options. I get the same information in the analysis log about the second attempt of meshing and the fallback tessellation method. The meshing stops afterwards though and the model is not meshed.

The problems mentioned above also hold for meshing the mold inserts after having the part successfully meshed with the "Ignore Contact" setting.

[mayur_dhumal]

Hi Maxmar,

When you select "Precise Match" that means you advice software to use global edge length same for all parts.
Please use Cool(FEM) for 3D. do not use Cool(BEM).

You can manually remesh the areas where you need to have different mesh densities to compensate for no. of elements.

Which import format you are using? have you tried with CADoctor first for any problem areas to heal?

Some more info you may need:

Mesh on assembly contact faces options

There are three options that can be used for meshing assembly contact faces.

Precise match

When the CAD model assembly has no geometric errors in the contact areas between components, there will be a precise duplication of nodes and elements that are on the contact surfaces.

Fault tolerant

With the Fault tolerant option, small variations in the CAD model are compensated for, and the mesh is aligned where possible.

Ignore contact

The assembly components are meshed without considering surface contact. If there is significant mismatch between the contact area meshes, the contact area nodes may need to be manually aligned.

Regards,
Mayur

[xusho]

When you see "Second attempt", it means the CAD model could not be meshed with the current settings (precise match).

When 'precise match' is selected, the assembly model will be converted into a non-manifolding model, while the contact interface is shared by 2 bodies.

When the model has no problems on the interface, this will give you matched triangles on both bodies. Local mesh density will be used.

However, when the 2 bodies are not matched properly on the interface, this process will fail, and meshing cannot go ahead.

"Second attempt" means the CAD model was actually meshed as an STL model. In this case, only a global edge length will be used, while local mesh densities defined on CAD entities will be ignored.

To avoid this problem, you can go to CAD system and fix the original model with focus on contact interfaces.

Shoudong Xu

Moldflow Meshing Team

[maxmarautodesk]

Hi Mayur,

"When you select "Precise Match" that means you advice software to use global edge length same for all parts.
Please use Cool(FEM) for 3D. do not use Cool(BEM)."

I use Cool(FEM) and have never touched Cool(BEM). I am not sure about the differences and advantages, so I go for the stuff I know.

Anyways, you are saying that "Precise Match" means that the software uses the same global edge length for all parts. That would mean setting different mesh densities (for part, mold, channels or within one of these domains) and the mesh density tool become useless?!

My expectation was that I mesh the part first with the desired mesh densities assigned to the faces of the imported CAD model. Then, I mesh the cooling channels likewise. Finally, I mesh the mold where I do not have to assign mesh densities, because the software will automatically mesh the mold in a way that the mesh will match the one of the part or channels at the contact areas.

"You can manually remesh the areas where you need to have different mesh densities to compensate for no. of elements.

Which import format you are using? have you tried with CADoctor first for any problem areas to heal?"

I believe if I remeshed the areas where I need higher mesh density, I would destroy the mesh match at the contact interfaces again. Or how do you mean that exactly?

Anyhow, I prefer not to remesh, because I made bad experiences with that. Sometimes the remeshing does not give me any effect or I loose geometrical accuracy and edges of my original shape. I think it is also problematic, because starting with a larger global edge length will inevitably lead to a loss in geometrical accuracy, because the mesh will not catch small and round features of the original shape. By starting with a very fine mesh and then reduce the size is usually not an option for me, since the number of elements will shoot through the roof.
I use different file types, usually direct import of SolidWorks (.sldprt, .sldasm), Inventor (.ipt), or Step (.stp) files. In this particular case, I imported a Solidworks assembly (.sldasm) which contains the part, two mold inserts, and two cooling structures. The two inserts are in .sldprt format, derived from .stp. The part and cooling structure I prepared myself by applying boolean operations on the two inserts.
Because of that reason, I assumed a perfect match between the CAD models of the part, mold, and channels, and I did not use CADoctor to check. I cannot see any problems in Solidworks. Where do I get the CADoctor from? If I google it, I find several different versions and providers.

"Some more info you may need:

Mesh on assembly contact faces options

There are three options that can be used for meshing assembly contact faces.

Precise match

When the CAD model assembly has no geometric errors in the contact areas between components, there will be a precise duplication of nodes and elements that are on the contact surfaces.

Fault tolerant

With the Fault tolerant option, small variations in the CAD model are compensated for, and the mesh is aligned where possible.

Ignore contact

The assembly components are meshed without considering surface contact. If there is significant mismatch between the contact area meshes, the contact area nodes may need to be manually aligned."

I am aware of the three different types for meshing on contact interfaces. I think what you write ressembles the help of Autodesk Moldflow. Still, thanks the information. It is good to have it here to make more people aware of it.

Best regards,
maxmar

[maxmarautodesk]

Hi xusho,

I find your post very helpful to get some insight on what happens "behind the curtains" while meshing. According to your explanation, it seems that the problem stems from the CAD model I am having. However, I cannot see any problems in the CAD model in Solidworks (compare to my previous post). Do you have any suggestions how to check the match of the interfaces within SolidWorks or an alternative software? Do you think CADoctor will do the job and where do I get it from?

Best regards,
maxmar

[mayur_dhumal]

Hi Maxmar,

Please refer attached image, when you have multi-body in contact, at that time you should have a same edge length, otherwise how could you expect to match the nodes on both the bodies, this is a general criteria while meshing in contact surfaces, so you will need 'Precise Match' with global edge length same, you cann't alter Edge length for all surfaces, thats why i ask you to remesh the required areas (If your geomtery is simple, no major features, then you may not need fine mesh at this region, it won't be useful, as for e.g. suppose you got some plaine surface at this region,assuming you will not get weld lines, ait traps, no need to fine mesh here.)

CADoctor is a Autodesk tool, to heal geomtery, which creates a *.UDM file to import into synergy.

Regards,

Mayur

[phammer]

Ok , 

meshing of assemblies is and still keeps a gamble. 

I have tried to mesh very often assemblies some with good results and some with no chance. 

1. the error message: Fallig back to tesselation, always comes when there is a problem in the cad mesh and you can not mesh it with precise match. But if you have inserts and you want to simulate core shift thenodes should be matched at contact faces. To use precise match you need to import the assembly

2. How to solve the errors in the cad

Well fist problem is to find the errors. One solution is to open the assembly in fusion and to validate the faces. It shows you where the errors are. So you have perhaps a possibility to correct it. Also good solution is to use caddoctor. Sometimes I use both simulatniously. Try not to do import export the cad to often. You get more  and more errors. And I also had bad experience with importing assemblies of solidworks. 

3. one possibiliy is also to mesh in a external mesher. We did some goog try with hypermesh. Worked very finde. I imported the bdf files from hypermesh in moldflow. 

4. I hope you meshed the cooling channels with channel mesh and the mold with mold mesh. But to use precise match with part and mold you have to use in dual domain the standard mesher. This also works and then use the mold mesher for the mold. 

5. If precise mesh wil absolutely not work you can give the contact face the same mesh density. But if you have a complex part this is much work. 

Ther are always several ways in moldflow. 

Greetings philipp

[harald_goetz]

If the parts are not too complicated I use NX Advanced FEM for meshing.

It handles contact good, gives better results regarding trinagle aspect ration and "deletes" small features by choosing the right mehs length.

What is can´t do is mehs matching, so for DD you´re kind of busted if Moldflow doesn´t work.

The 2nd attempt of meshing is often so bad that i stopp it immidiatley.

Better mesh your part outside Moldflow to an STL, this gives you (in most cases) more control over the process.

Everything else: Good luck.

Regards

Harald

[xusho]

The process to convert assemblies into non-manifloding models is very sensitive to gaps on contact interfaces. That is the reason we do not use it as default option.

When this happens, I would suggest: do not use precise match.

Moldflow solvers have been improved in recent years, and "precise match" does not make big differences as before.

With "Ignore contact", you can define same edge length on both faces. Triangles on interfaces do NOT have to be matched.

Shoudong Xu

[maxmarautodesk]

Dear all,

thank you very much for your lively contribution and feedback! I would like to wrap up and point out some things for all readers and the Autodesk people and come up with some new questions to discuss.

SolidWork assemblies might cause trouble when using "Precise Match" for meshing

It seems to me that the problem was caused by this particular assembly of SolidWorks. If you are having the same problem, try to diagnose thecontact surfaces with:

• Autodesk CAD Doctor
• Autodesk Inventor Fusion

I could not investigate the issues in that way, since we do not have that in house and I do not have access to our Autodesk subscription center. Consequently, I do have neither software to my disposal right now. Nonetheless, I tried the whole meshing process with another assembly that I made in SolidWorks. The meshing worked without any problems and the assembly could be meshed in the first attempt, accepting the assigned mesh density.

"Precise Mesh" can be used if the faces to be matched have the same mesh density

The same local edge length/mesh density must be given to the faces of a CAD assembly where the mesh is to be matched. It is not enough to set only the mesh size on one of the faces. The software will not automatically copy the mesh to the other face (why actually not?).

External meshers might help you out

Any automatic second attempt will give you most likely bad meshing results. You can try external software instead:

• Altair HyperMesh
• Siemens NX Advanced FEM (not suitable for Dual Domain, since mesh match is not available

Going with the default "Ignore contact" might give you decent results, since the Moldflow solver has evolved a lot and the difference between the two settings decreased.

"Precise Match" option seems to conflict with mesh refinement at injection location

Since Moldflow 2015, it is possible to let the software refine the mesh at the injection location (given that the injection location is placed on the CAD bodies prior to the meshing step). I think this feature is very handy. In a recent study, I have figured out though that the mesh refinement at the injection location does not work, if the "Precise Match" option is selected. 

It does not always happen. I have run the meshing with another part (similar size and meshing settings). The mesh refinement was carried out both with "Precise Match" and "Ignore Contact".

Please have a look at the attached pictures where I made a comparison between "Precise Match" and "Ignore Contact" for both studies.

Anyone seen the same? Any ideas where this could come from?

Best regards,
Max

[mtessier]

Hi Max,

Good job pulling together the relevant information.

Here are some additional considerations from an Autodesk support case that I went through in June 2013 regarding achieving perfect match with assemblies for Cool(fem); hopefully they are of some use.

Firstly, as you have noted, precise matching between CAD bodies is ideal, but often not obtainable.  The solvers are said to be tolerant of imperfect matching between bodies.

Next, the "precise match" option refers only to mating CAD geometry, and it ignores previously created elements.  One of the primary recommendations was to mesh all components simultaneously with the default mesher (as if they were part (3D) components).  After meshing, the element definition could be changed to mold block (3D) or mold insert (3D) as desired.  If you mesh the mold components after the plastic components, the "precise match" option is ignored for those interfaces.

Also, I learned some history on the Cool(fem) mesher.  Prior to Design Link being included in the normal software release, users needed a way to achieve good mesh matching between CAD bodies when they were unable to import assemblies.  As a general rule, when possible, the regular mesher should be used to mesh all CAD geometries in a single step (except 3d cooling channels).  This Cool(fem) mesher accounts for previously created elements (and ignores CAD geometry on those surfaces).

Apart from the support case, I also have the following tips from experience:

• Usually, these meshing problems relate to interfaces where the contact is part of a larger surface.  For example, the tip of a pin geometry contacts a large planar geometry.  In these cases, I have had some success with surface splitting or part splitting.  Once the surface mesh is generated, you can stitch the desired components together manually.
• I'm not sure that your "same mesh density" requirement is true.  Usually I define a fine mesh density on the plastic body, a medium mesh density on the inserts, and a coarse mesh density on the mold blocks.  Using this set-up, the plastic-insert interfaces are meshed at the fine mesh density and the insert-mold interfaces are meshed at the medium mesh density.  Maybe you have been seeing this problem because you have been meshing in separate steps.

Hope this helps.

Matt

[harald_goetz]

I could add another tipp to this topic:

Meshing with Open Source.

I didn´t believe myself at first.

I had some difficulties to mesh a large part with a special cold runner with 40(!) gates. Gate shape is a half cone with a radii of 0.3mm.

The whole part is about Ø 200 mm.

Whether Moldflow 2015 nor NX in that case serves me well.

I could have tried Altair HM, but I´m lazy. And since I´ve used Cadmould a lot I know that they use Netgen internally.

So did I!

I had to use GMESH also to get the mesh as surface mesh from Netgen to Moldflow (as *.bdf/nas).

And imagine my surprise when looking for the quality. Of course no matching!

But since I´ll do 3D in most cases anyway...

Meshratio was 4:1 on the bad side!

Netgen and Gmesh do not mesh touching parts, - so no help for that. But with the tolerant behavior of the solver we should get away meshing every single part.

Though I would expect some of the characteristics of this open source mesher in the next Moldflow release.

Cheers

Harald

---

[maxmarautodesk]

Hi Harald,

thank you very much for your insight into few alternatives for meshing. According to your pics, GMESH seems to be indeed a quite handy and powerful tool. I have downloaded it and would like to try it out. However, there are hardly any instructions available and I am having trouble to get started. Can you recommend some tutorials or manuals?

Best regards,
Max

[maxmarautodesk]

Hi Matt,

thanks for your additional information. I think you point out some very precious details which are not obvious:

• "If you mesh the mold components after the plastic components, the "precise match" option is ignored for those interfaces."
• "The regular mesher should be used to mesh all CAD geometries in a single step (except 3D cooling channels). After meshing, the element definition could be changed to mold block (3D) or mold insert (3D) as desired."

I would like to comment moreover on your thoughts at the end of your post:

• I agree that it sometimes makes sense to divide bigger planar surfaces artificially in the CAD model. I think it also helps with regard to defining local mesh densities. However, it is additional manual work to be done by the user.
• I disagree with you. So far, I always had to assign the same mesh density to both surfaces in contact. If they were of different size, the produced mesh did not match. Can you perhaps tell me what are the values of the mesh sizes you used for the entities (fine=part, medium=insert, coarse=mold)? My feeling is that we are both kind of correct and that I experience the trouble, because the range of mesh sizes I need is very large. I deal with micro plastic parts which have molds still of small conventional size. I typically need mesh sizes in the range of 0.05-5.00 mm. Consequently, a typical scenario for me looks like: fine=part=0.04-0.20 mm, medium=insert=feed system=0.50-1.00 mm, coarse=mold=5.00 mm. My impression is that Moldflow has a lot of trouble with such large differences in mesh size and that is the reason why the mesh match does not work. It does not work either, if the mesh ratio is smaller than 10, e.g. at about 4 or 5. Therefore, I usually have to assign manually mesh densities to the insert and mold, too, which I choose equal to the mesh size of the touching face of part or mold, respectively. This large mesh ratio also seem to give the smooth mesh transition a hard time. According to Autodesk, the mesh transition from smaller to larger mesh size (or vice versa of course) was improved in the 2015 version. Honestly speaking, it is a first step in the right direction and many transitions look better. Yet, I think this feature should be further improved, especially to be able to handle large mesh differences better. 

Best regards,
Max

[harald_goetz]

Gmesh is a powerful tool, and I use(d) to to analyse meshes.

In this case I use Netgen to mesh, and it was somewhat of a journey to find the right settings.

I could have used Gmesh directly, but I didn´t find the parameter in there. At least I found it more difficult.

For 3D Meshes Gmesh uses Netgen anyway (There are to famous 3D Netgenerators out there, beside Netgen it´s Tetgen)

So to make it easy if you´re going to test it I attach a PNG that might be helpful for first tests.

May the force be with you.

Harald

[MFI-MaF]

Hi,

Gmsh is a very nice tool 🙂 thanks for the tip.

After a while of testing and learning, its very good to handle. I've meshed directly in gmsh, there is a good option to remesh the geometry by defining number of nodes (Mesh--> Define--> Transfinite--> Line) But one think I'm searching:

Is it possible to define the minimum number of layers?

It would be very nice, if its possible to directly import the 3d-mesh without remeshing (refining) in Moldflow.

 Markus

[harald_goetz]

Markus,

I'm not sure what Gmesh can do (or Netgen, which is what Gmesh uses at last).
I got pretty good meshes the way I described above. So taking the surface mesh and let Moldflow do the rest seems better for me,

since I'm thinking the Moldflow can care about the mesh it needs (or maybe a biased mesh).

Maybe Netgen will do the trick, but I don't think so.

If you get the meshing done in Gmesh alone, you might tell us (me?) how to? So we have something up our sleeves.

On the other hand I should retry it with the current release as well as with upcoming ones, there are always slight or bigger
improvements found in every new release of  Moldflow.

Harald

[yannick.moret]

Moldflow has specific requirements for it's 3D mesh.

I don't think a standard mesher can give us the nice tetra inside the thickness we requires.

If you get a good surface mesh, Moldflow shouldn't have any issues to generate your 3D mesh.

Remember to pay attention at the number of layer you generate. If you want a higher number of layer, it is better to have a small enough edge length on the surface to avoid high aspect ratio on tetra.

[MFI-MaF]

Hi,

thats the way i've gone to generate a mesh:

1. Open CAD-Model

2. Review a first 3D- Mesh

• Go to: Modules -> Mesh -->3D (you will see, your mesh is not perfect 😉 )

3. Optimize the 3D-Mesh

• Make nodes visible: Go to: Tools -> Options --> Mesh --> activate Nodes

4. Place more nodes

• Mesh --> Define --> Transfinite --> Line  
• Select one or more line(s) and choose a number of points
• Start with pressing "e"
• Go to: Modules --> Mesh --> 1D to see your new nodes
• Go to: Modules --> Mesh --> 3D to see the new mesh

Do this for every line you want to optimize and you'll get a nice mesh 🙂

There are also other ways to refine but i didn't test it till now.

Hope this helps.

Markus