AnsweredAssumed Answered

Mixed meshing

Question asked by Derek Bishop on Dec 24, 2007
Latest reply on Apr 10, 2012 by Grant Smith

Like • Show 0 Likes0
Comment • 27

I recently tried to do some mixed meshing without success. To investigate it further I made up a model comprising a base plate welded to a equal angle beam. The end of the beam was bonded to the base plate. The sketch line for the beam finished at the plate srface. Couldn't get it to work.

Perhaps the main query is the way in which the contact between the beam end and the plate are defined. Do I make a contact between the plate and the beam end node or beam end face?

Any suggestions or hints would be appreciated.

Regards,

Derek Bishop

No one else has this question

Outcomes

27 Replies

1 2FirstPrevious

Ameer Chilakala Dec 26, 2007 3:20 AM
Mark Correct
Correct Answer
I tried in Cosmos 2008. Yes we have to define contact set between the face where the beam contacting the face i.e it may be solid or shell depending upon the your problem and the beam end joint.

if you are using shell mesh along with beam mesh means, you have to define the contact set between the shell surface which is used during defining the shell thickness and beam end joint.
Like • Show 0 Likes0
Actions
Derek Bishop @ Ameer Chilakala on Jan 2, 2008 5:08 AM
Mark Correct
Correct Answer
Has anyone been able to mesh a beam to a solid or shell?
Like • Show 0 Likes0
Actions
1-59EXF5 @ Derek Bishop on Jan 2, 2008 4:46 PM
Mark Correct
Correct Answer
I have been having alot of difficulty in this area as well, I have an odd shaped frame (rec tube) using sheet metal to 'box' in the corners. I have not been able to sucessfully run a beam/shell mesh. I always get an insufficient restraint error. I have tried using contact set between the members to no avail. Even my VAR was stumpped.
Like • Show 0 Likes0
Actions
1-574SHZ Jan 15, 2008 3:50 AM
Mark Correct
Correct Answer
God this is horrible. The mixed mesh study disappeared from my SolidWorks model when I opened the file. Nothing is bringing the study back, in fact there is no text field under the cosmosworks logo (just beside the property manager)

God I hate this!!! Burned by SolidWorks again!!! I want to send them an Invoice for all the lost data I've suffered through....

Has anyone had this problem? That is, opening up a model with a mixed mesh and it just disappears from the study list!?! It ran fine the first time, now its gone... totally gone

I want to send solidworks a bill for my engineering time @ $175/hr x 10 hrs of lost data. Enough is enough. This has gone on for years.
Like • Show 0 Likes0
Actions
Vince Adams @ 1-574SHZ on Jan 15, 2008 12:34 PM
Mark Correct
Correct Answer
Hi guys, can you send copies of these failing models to me? Especially yours, James.

Mike, I can honestly say that I've never seen a study disappear as you've described. I searched the database and found no reference to this either. Have you talked this thru with your VAR or COSMOS Tech Support? You said you've seen it before but is it repeatable? ...maybe within the same database?

If you want to talk this thru, please drop me a note and I'll give you a call.

-- Vince
Like • Show 0 Likes0
Actions
1-574SHZ @ Vince Adams on Jan 16, 2008 2:19 AM
Mark Correct
Correct Answer
Hello Vince

I was talking to the VAR today, and they say that this problem has been reported. (I cant remember the 2 SPR #s that address this problem)

Here's the kicker - apparently the problem was discovered in SP1.0, but not fixed until SP2.0. The VAR reports that the Cosmos guys said that any Mixed Mesh (Nixed Mesh ha ha ha ha ha) that disappears in SP1.0 or SP1.1 can't be recovered! Ever! The link is permanently broken and can't be restored, even if you update to SP2.0 subsequent to the catastrophe in SP1.0!!! Ain't that a hoot!

So, we have thousands of $$$ in lost time (and a seriously PO'd client) that can never be recovered.

I would be more than happy to send a copy of the model to you. If you can somehow recover the data, we would be extremely grateful. The model is extremely large, over 1Gb even when put into a Zip file. An FTP site would be required to send it to you.

You can reach me at mike@kismetengineering.com

Regards

Mike
Like • Show 0 Likes0
Actions
Vince Adams @ 1-574SHZ on Jan 16, 2008 10:59 AM
Mark Correct
Correct Answer
Mike, I searched the database for these 2 SPRs again this morning with no luck. I'll have a conversation with development to see if there is anything we can do.

Contact me off the forum at my e-mail below and I'll get you an FTP site or work out some other arrangement. Also, can you let me know who your VAR and contact at that VAR is? I'll try to dig a little deeper.

Vince
Like • Show 0 Likes0
Actions
Vince Adams @ Vince Adams on Jan 17, 2008 12:39 AM
Mark Correct
Correct Answer
Mike, our Director of QA has not heard of this either. Are you 100% positive that the original database wasn't defined in a Beta version?

You'll need to touch base with me offline if you'd like me to take this any further.

Your call...

Vince
Like • Show 0 Likes0
Actions
Derek Bishop @ Vince Adams on Jan 31, 2008 8:25 AM
Mark Correct
Correct Answer
Vince,

I notice your title. Please advise if there have been problems with mixed meshing facility in Cosmos and if it is now working. Refer to my posts above for details.

Regards,

Derek Bishop
Like • Show 0 Likes0
Actions
Vince Adams @ Derek Bishop on Jan 31, 2008 11:33 AM
Mark Correct
Correct Answer
Hi Derek & everybody...

Mixed meshing in 2008... first, I've gone back & forth with Mike (above) a few times and he's pretty sure the problems he is having are related to imported IGES geometry. It is repeatable with the IGES & doesn't happen with native geometry. Our QA dept will dissect his database and confirm this or share an alternate cause.

We developed a novel way to integrate beam elements and solid models which I firmly believe is the right way to handle this this traditionally complicated mesh type. Being an "old school" analyst, it took me a while to get comfortable with it but I'm on board that this is the way it should be done within SW. While we have been improving beams with each release, we are still seeing some connectivity issues in mixed mesh. We've committed significant development resources to this feature and I'm confident that we'll get there.

In the interim, my advice is this...
(0) get up to the latest SP!!! Big improvements from SP0 to SP1 to SP2
(1) build your 3D sketch cleanly with connecting endpoints so that you aren't over-reliant on pinballs;
(2) Review some of the tips in the COSMOS Companion "SolidWorks for COSMOSWorks Pt2";
(3) Use structural members (vs. extrusions) where ever possible;
(4) split your driving beam sketchs at the points where the connection to a gusset or other non-beam geometry starts (see attached JPG) This helps ensure the nodes are reasonably compatible across the joint... much higher success rate;
(5) If you use extrusions to define beams, make sure you scan your model for small extruded parts that got accidentally turned into beams;
(6) & if... after checking all these... your model is still having problems, please send it to your reseller and request that they fwd it on to COSMOS Tech Support. The more user models and variations we see, the more confident that the changes we're incorporating will be universally sound.

Thanks for the feedback!

Vince
Attachments
- Split Curves for Gusset.JPG12.5 KB
Like • Show 0 Likes0
Actions
Ken Davis @ Derek Bishop on Feb 5, 2008 12:50 PM
Mark Correct
Correct Answer
Vince,

Been a while. Hope all is well. I have been off-line for a while. Related to this topic, I am working on an analysis using mixed-mesh. My reasoning for MM is due to the combination of thick walled aluminum castings and thin sheetmetal parts. It has taken me some time to get the model meshed. Lots of time invested in getting the right controls on the solids (shells are usually no problem).

To the point, once meshed, I get a error message telling me that bonds between shells and solids are invalid. Is that true, if so, what is the purpose of MM. The parts are in contact, and I chose the contacting surface of the sheetmetal part that contacts the casting as my shell surface.

In addition, once I got it meshed without changing the "remesh with incompatible mesh" option, I have bad elements or elements to small. Solver fails.

Any thoughts?

Ken Davis
Mechanical Engineer
General Dynamics
SatCom Technologies
ken.davis@gdsatcom.com
Like • Show 0 Likes0
Actions
Derek Bishop @ Ken Davis on Apr 23, 2008 10:11 AM
Mark Correct
Correct Answer
Mixed meshing still not working for me. I'm running SP 3.1 and still having problems. This time the analysis runs and I get plots for everything except the stress. The stress plot for the solid plate connected to the beam appears but the beam disappears.

Anyone else using this feature and managed to get it working consistently?
Like • Show 0 Likes0
Actions
Ben Floan @ Derek Bishop on Apr 26, 2008 2:10 PM
Mark Correct
Correct Answer
Derek,

I'm also having difficulty under 3.1 w/ mixedm.. model does not seem to recognize the connectivity between solid and shell causing inadequate restraint errors. I thought I might be missing something until I stumbled upon this forum.

-Ben
Like • Show 0 Likes0
Actions
Adam Latour Apr 28, 2008 5:20 PM
Mark Correct
Correct Answer

I'm experiencing the exact same problem as Derek. Ben, I find that you must make source faces coincident with a target joint on a beam in order for a bonding condition to function properly. The mesh appears to work, the analysis runs, but the only matrix with any data in it is the displacement results. In the stress results, the beams disappear, and the strain results in the beam elements are all zero (the beams are visible though).

I'm operating on 2008 x64 SP3.1.

Adam
Like • Show 0 Likes0
Actions
Adam Latour Apr 29, 2008 1:09 PM
Mark Correct
Correct Answer
Derek, I realized that to view stress plots in beams you must create a separate plot. In the plot definition menu you have the option to display stresses for solids and shells OR beams alone. The types of stress you can display (axial, shear, etc.) is different for beams than for shells and solids, hence the need for different plots. This should solve your problem.

Cheers!

Adam
Like • Show 0 Likes0
Actions
Derek Bishop @ Adam Latour on Apr 30, 2008 12:21 PM
Mark Correct
Correct Answer
You might be onto something there Adam. I'll check it out.
Like • Show 0 Likes0
Actions
Bill McEachern @ Vince Adams on May 15, 2008 8:33 PM
Mark Correct
Correct Answer
I have had it happen once in like 10 years and it was last year. Hasn't happened since. I could not get it to repeat but I was pretty ticked when it happended.
Like • Show 0 Likes0
Actions
Bill McEachern @ Derek Bishop on May 15, 2008 8:37 PM
Mark Correct
Correct Answer
I have not done a lot in 2008 but some. I did not have any trouble on the ones I tried - beams to shells and shell to solids. It pays to remeber that you can not have have big gaps between the entities - they need to be in the order of an element of shell thickness - I just make them coincident or near enough. Obviously you have to manually specify the contacts you want bonded in a mixed mesh.
Like • Show 0 Likes0
Actions
Derek Bishop @ Bill McEachern on Jul 14, 2008 1:18 PM
Mark Correct
Correct Answer
Am pleased to be able to say that I am having much more success now with the mixed meshing problems. One thing that has helped is to be sure to define contacts carefully. For beam to shell meshing I select the beam component and then the face to which it bonds. Have not yet tried again the case mentioned above where the end of a beam connects to a solid.
Like • Show 0 Likes0
Actions
Derek Bishop @ Derek Bishop on Jul 27, 2008 2:38 PM
Mark Correct
Correct Answer
Have tried some mixed meshing in 2009 and was annoyed to see after just managing to get this thing going in 2008, the format has been changed. It seems harder to select the mesh type you want.

The program automatically assigns the mesh type to the model. I cannot now work out how to shell mesh items. The models I need to run are assigned to be solids where they should be shells. I cannot see how to change to shells from the solid or define reference surfaces.

Anyone had success in this matter?
Like • Show 0 Likes0
Actions
Derek Bishop @ Derek Bishop on Aug 8, 2008 1:55 PM
Mark Correct
Correct Answer
Model parts can be defined as shells by methods other than using modelling components as sheet metal parts.
Like • Show 0 Likes0
Actions
F. g @ Derek Bishop on Sep 18, 2008 10:48 AM
Mark Correct
Correct Answer
I'd like to begin modeling mixed-meshes, but I'm facing the same problems that al lot of users have found, regarding mesh and analysis...
After some attempts with different combination of contact constraints, loads and so on, I have finally come to some sort of result.
Now the problem is that this result doesn't really represent what I expected: infact I wanted to create a connection between the external frame and the sheel, but it seems like only two vertex of the sheel are connected to the beams.
The shell vertex at the corner btween the two frames is floating away (see the displacement plot) instead of remaining linked to the beams.

Any idea for this problem?
ps: it seems that uploading html reports is not allowed, so I'm uploading a couple of jpg...

Thank you and CIAO!
Attachments
- part1-studio 1-studio 1-1-immagine-1.jpg40.1 KB
- part1-studio 1-risultati-spostamento1.jpg46.2 KB
Like • Show 0 Likes0
Actions
Vince Adams @ F. g on Sep 18, 2008 11:14 AM
Mark Correct
Correct Answer
I would send this problem into tech support for a more thorough investigation but I am curious as to why you would want to use a beam for this stiffener at all. A good rule of thumb is that if the analytical representation doesn't at least suggest the actual geometry, it may not be appropriate. Your stiffneners are nearly the same aspect ratio as the plate. Why not use all solids or all shells?

-- Vince
Like • Show 0 Likes0
Actions
F. g @ Vince Adams on Sep 18, 2008 11:53 AM
Mark Correct
Correct Answer
Vince, I agree with you, but you should consider this problem as a "first attempt"; it is not related to any particular "real" study, and dimensions, stiffener sections and loads are "random".

Infact I just wanted to set-up and run a mixed-mesh study in order to face the difficulties and understand the limits of mixed mesh models...
While I was testing this problem, this strange result came out.

I thought that maybe I had to connect the beam vertex entity with the surface, but this cannot be made: if I specify this connection, the study does not run at all.
I will try to send the problem to the tech support.

Thank you
Like • Show 0 Likes0
Actions
Vince Adams @ F. g on Sep 18, 2008 12:05 PM
Mark Correct
Correct Answer
Gotcha! Makes sense. When you bond the beam to the plate make sure you are bonding the beam, not the joints. It is also possible that the depth of the beam (or more specifically, the distance from the neutral axis to the plate) is proving problematic for COSMOSWorks. (Now called SolidWorks Simulation... in case you haven't heard.)

If you did bond the beam to the plate, try adjusting your beam height to about a 1:1 aspect ratio & see if that works. While not intentional, the less beam-like a solid being treated as a beam is, the more difficult it is to get bonding right.

Vince
Like • Show 0 Likes0
Actions

Mixed meshing

Outcomes

Attachments

Attachments

Related Content

Recommended Content