AnsweredAssumed Answered

Symmetry doesnt work correctly

Question asked by Jurgen Zwanziger on Jul 31, 2013
Latest reply on Aug 7, 2013 by Jared Conway

Like • Show 0 Likes0
Comment • 17

Hello,

i have an assembly that you can see at the picture attached to this thread. The problem is, that the rods are connected with a symmetry constraint. The global contact is "bonded" and the result looks very strange. After changing to "no penetration" the result looks more realistic. The rods are connected to the plate with a "bonded" contact.

I have tried serveral things but nothing works. After reducing the plate thickness in several steps, the simulation with bonded contact works. I dont understand the problem. In the picture you can see the same simulation with a difference of 1mm in thickness. In "no penetration" simulation it works always.

Please excuse the bad english. Its not my native language.

thanks

No one else has this question

Outcomes

17 Replies

Bill McEachern Jul 31, 2013 8:26 AM
Mark Correct
Correct Answer
symmetry will not connect parts, it is a boundary condition. I get that maybe language is an issue here. I am not sure I really understand what you are after but here is a rip at it...If you are trying to allow rotation between the vertical rods and the plate you need a spherical connector (I think a bearing connector properly done might work here) or spherical contact surfaces with no penetration. Your bearing can't be just a short cylinder with this approach - it needs to physically mimic the bearing. I suspect this is why when you make the plate thin it gives you what you want since the plates bending stiffness is so low it approximates the flexibility you are after.
Like • Show 1 Like1
Actions
- Jurgen Zwanziger @ Bill McEachern on Jul 31, 2013 9:17 AM
  Mark Correct
  Correct Answer
  Hi,
  
  the vertical rods and the plate are connected. So they have to move like the left picture shows. The vertical rods have a symmetry constraint on the lower end. There is a force on the inside of the plate. On the right picture the rods dont move correctly. Its just like they are bonded to something. I dont know. But when i reduce the thickness of the plate it works.
  
  I have a nearly identical simulation but there the assembly is build in another way. There the bonded global contact works. But the assembly is not the way that i want it to be. So i cant use this. There must be something with the symmetry or the rods and the bonded contact. As I said, using a "no penetration" global contact works fine.
  
  I dont know if it is good to see. But thats what happens between the rods and the plate. The rod deforms, but it doesnt deform like it should be. Its hard to describe. :/
  Like • Show 0 Likes0
  Actions
  - Jared Conway @ Jurgen Zwanziger on Jul 31, 2013 11:46 AM
    Mark Correct
    Correct Answer
    please post a free body diagram or some other diagram explaining your loading case and how the parts should be connected together. i would also suggest posting the full model so we can see what it looks like without symmetry and with symmetry. if you can post the models, even better. if you can't post the models, i would suggest building a simple version to post. this will also help you tell if this is a software, setup or part specific issue.
    
    the reason i suggest this is because i'm really not sure what the issue is. i think you have multiple issues and are trying to draw conclusions which is confusing the matter. it might be helpful to concentrate on just the problem statement and then a list of what you tried and what the outcomes were (Without conclusions drawn).
    Like • Show 0 Likes0
    Actions
    - Jurgen Zwanziger @ Jared Conway on Aug 1, 2013 2:30 AM
      Mark Correct
      Correct Answer
      I build a comparable assembly. It is the same problem. You can switch the global contacts and you see 2 completely different results.
      Attachments
      Assembly.zip149.7 KB
      Like • Show 0 Likes0
      Actions
      - Jared Conway @ Jurgen Zwanziger on Aug 1, 2013 8:43 PM
        Mark Correct
        Correct Answer
        yep. definitely see the problem on your studies.
        
        i created new studies and ran them in 2013 sp4.0 and beta, they both work correctly
        
        also i created a new study and drag/dropped your data over (no mesh) and it worked fine as well
        
        could be version/service pack issue
        could also be a bad study
        
        have you tried creating a brand new study?
        Like • Show 0 Likes0
        Actions
        Jurgen Zwanziger @ Jared Conway on Aug 2, 2013 9:39 AM
        Mark Correct
        Correct Answer
        I created this study in SW 2012 SP 1.0 and SW 2013 SP 3.0. Both show the same results. I create a "New Study" and then i use "Static" thats all. I am downloading SP4.0 at the moment and will try the studies again. Perhaps it is a simulation setting problem.
        
        Ok. I updated to SP 4.0. Same result. Now I tried some settings in the simulation properties. After switching the "Incompatible bonding options" to "simplified" the deformation result seems to be ok. The Mises-stress has a peak now but over all it looks ok. But i still have an other assembly with almost the same study where everything works fine without this change.
        
        You said that it could be a bad study. What do you mean with that? Is it possible to change the study settings before creating them?
        Like • Show 0 Likes0
        Actions
        Jared Conway @ Jurgen Zwanziger on Aug 2, 2013 11:15 AM
        Mark Correct
        Correct Answer
        hello jurgen, like i said, i opened your file, ran it. duplicated the study, changed to no penetration, i saw the issue.
        
        i created a brand new study in that file, setup the problem again completely, reran it. (all default settings including the incompatible bonding option) it ran fine. so to me, this means there is either a setting in your old study that is causing the problem or it is a bad study. i reviewed all the settings, they are exactly the same. based on your info, its like the software is interpreting some setting incorrectly. you could delete all the setup in that study and set it up 100% again to confirm that it is a bad study but why do that when a brand new study will fix the problem. i'm not clear on whether you have actually tried that or not. if you haven't, please do that. no use trying to guess what is happening until we calibrate that what we're seeing is the same.
        Like • Show 0 Likes0
        Actions
        Jurgen Zwanziger @ Jared Conway on Aug 2, 2013 11:20 AM
        Mark Correct
        Correct Answer
        hi jared,
        
        i rebuild this assembly 2 times. set up a new study and it happens exactly the same. on the other side i tried today to delete the whole setup in the existing study. either way the same result. in solidworks 2012 and 2013.perhaps i should reinstall the whole solidworks installation. could this help ?
        Like • Show 0 Likes0
        Actions
        Jared Conway @ Jurgen Zwanziger on Aug 2, 2013 11:24 AM
        Mark Correct
        Correct Answer
        Hi Jurgen, you're right. You created a brand new test case and it had the same problem.
        
        I wonder if it has something to do with the language version that you're working with vs mine being in english. Or some kind of setting in your simulation options that i'm not seeing. Unfortunately i don't have a way to test your version because I don't have an OS with the language.
        
        you said you have other assemblies that are similar that don't show this problem. have you gone back to them and setup new problems to see if that is still the case?
        
        the onyl thing i can think of is having you put together the steps that you're taking to create the analysis setup and run it to see if there is any difference in the way that we're doing things. like i said, brand new studies or new studies with your info drag/dropped over work fine. only your studies are bad.
        Like • Show 0 Likes0
        Actions
        Jurgen Zwanziger @ Jared Conway on Aug 2, 2013 11:27 AM
        Mark Correct
        Correct Answer
        hi,
        
        unfortunately i m not at work at the moment. i will describe the whole way i am going through on monday and post it here. i will also post my simulation settings. i think this could help perhaps.
        
        so i wish a nice weekend. see you on monday
        Like • Show 0 Likes0
        Actions
Jurgen Zwanziger Aug 5, 2013 5:21 AM
Mark Correct
Correct Answer
Hi,

so i want to show you the way i create the study. The assembly is from the post above. So this should be the same for everyone.

1. I go to the New Study button and choose static.
2. After that i select all 3 parts and "apply/edit material" and take the 1.0577 from the DIN materials.

3. The global contact is set to "bonded" with compatible mesh.

4. I add 2 contact sets and take the type "bonded". In the image you can see as an example one of the two contact sets.
5. For fixtures i add "advanced fixtures" and choose the "Symmetry". Now I take the 2 faces of the rods. The symmetry is shown in the image below.
6. I add another "advanced fixture" and choose "use reference geometry". So the plate can only move in one direction.
7. Now I add a external load to the plate. I choose a pressure of 2,5 N/mm^2 normal to the selected face.

8. After that i "create mesh" and edit the "mesh parameters". I select the "curvature based mesh" and minimize the maximum element size to 4mm. The minimum element size is 4 mm too. For the min number of elements in a circle solidworks advices 8 and for the element size growth ratio it advices 1.5. In the advanced settings i have 4 points for the Jacobian points and the other two settings are disabled. In the options also both settings are disabled.
9. I create the mesh and run the study.

10. I got an not realistic result. The rods dont move like in the study with "no penetration". It looks like the picture in the first post.

11. my settings are:

And here are the study properties:

At the Flow/Thermal Effects is only Input temperature activated.

I hope this can help to solve my problem.
Like • Show 0 Likes0
Actions
- Jared Conway @ Jurgen Zwanziger on Aug 5, 2013 11:30 AM
  Mark Correct
  Correct Answer
  hi jurgen, thanks for the information, i can reproduce your issue, the problem is related to curvature based mesh. take your model, change it to standard mesher (the default), run the analysis, works great. switch back to curvature, no good. that is the difference between the studies that i created that worked fine and yours that didn't work. the second peice of this is the contact set, so it is almost like under curvature based, it adds a rigid between the bottom symmetry restraints that isn't real. report this to your reseller.
  Like • Show 1 Like1
  Actions
  - Jared Conway @ Jared Conway on Aug 5, 2013 11:31 AM
    Mark Correct
    Correct Answer
    but actually. switching back to curvature based, and just using the defaults, the problem starts working again. it is almost like the problem is "stuck" in a bad state and just needs the mesh method swapped to "fix it".
    Like • Show 1 Like1
    Actions
    - Jurgen Zwanziger @ Jared Conway on Aug 6, 2013 2:36 AM
      Mark Correct
      Correct Answer
      so you think this is a software problem? i still have a study from another assembly that is almost the same. the assembly is just build with a skeleton model. there the symmetry works fine. i'll go with the standard mesher but i m not very happy with that. i think reporting this to my reseller is a good idea. did you try this in beta too? perhaps the problem is solved in this version. if not. can you please try it.
      
      Thanks for your help. It helps me a lot At least i know now where the issue is coming from. So i can avoid it.
      Like • Show 0 Likes0
      Actions
      - Jared Conway @ Jurgen Zwanziger on Aug 6, 2013 11:04 AM
        Mark Correct
        Correct Answer
        hi jurgen, if you look at my last responses. when i take your study and switch to standard, mesh and then switch back to curvature the problem goes away. so it seems as though the problem is study specific (corrupt study). BUT, the fact that you can reproduce the problem to me means there may be a software issue that causes it. but it is very specific to your creation workflow since you have been able to reproduce it on multiple parts. i haven't tried your exact steps but your VAR can to confirm the software issue.
        
        I did check in 2014. Same there. it is really built on 2013 sp3.0/4.0 so if the problem exists there, it will likely exist in the beta.
        
        regarding standard mesher, i use standard mesher for 85%+ of the analysis consulting projects.i wouldn't discriminate against it to hard. it gets the job done and efficiently enough. the curvature based mesher is only about 15-30% faster and sometimes only in certain circumstances. and even then, sometimes the standard mesher is better because it creates a more uniform mesh.
        Like • Show 0 Likes0
        Actions
        Jurgen Zwanziger @ Jared Conway on Aug 7, 2013 2:44 AM
        Mark Correct
        Correct Answer
        Hi,
        
        i tried again a couple of things and i think i got the problem. like you said, it works after switching back to curvature based mesh. did you change any settings of the curvature based mesh? taking the standard settings works fine for me. but after editing those and getting a more fine meshing i can reproduce my problem. i think something is going wrong with a fine meshing and symmetry constraint.
        
        i took the curvature based mesh because i have in my real assembly some parts, that are in relation to the assembly very small. but i didnt want to use mesh controls because i want to copy the assembly including the study and switch only some of the parts. i dont know if this will work with mesh controls too. for the moment i can live with the problem, but i didnt know why this is happening. it is definitly an issue with the meshing. perhaps my reseller can report this problem and it will change in future versions.
        
        i really thank you for your help jared.
        Like • Show 0 Likes0
        Actions
        Jared Conway @ Jurgen Zwanziger on Aug 7, 2013 11:24 AM
        Mark Correct
        Correct Answer
        when i created new studies on my end, i didn't make any changes to the settings so that may be the factor here. to note, the reason i didn't change them is because the defaults matched the values that you mentioned. but you're correct that your reseller can get this reported to solidworks so that it can get fixed in a future version.
        Like • Show 0 Likes0
        Actions

Symmetry doesnt work correctly

Outcomes

Attachments

Related Content

Recommended Content