Discussion Forum

Posted: 1 decade ago 29 juil. 2010, 16:01 UTC−4

I encounter almost the same problem when I tried to drawn two bodies while there is contacted surface between them, what I need was two boundary condition in this surface, but I got only one. There was a solution for this is we can displace one surface by a certain distance, but I don't know whether this is OK or there are other way to solve this problem.

Posted: 1 decade ago 19 août 2010, 04:46 UTC−4

Try selecting "Use Assembly" from the draw menu.... Thanks and Regards

Posted: 1 decade ago 20 août 2010, 01:41 UTC−4

Hi you should learn the diferences of asembled (two bodies in contact but with each it's own boundary overalapping) and union mode where both bodies have a common boundary. The latter is mostly the default, as it means you have continuity of flux over this boundary, once you have "geomanalysed" your geometry, you have two bodies but one common boundary with "continuity". If you want to add specific physics on the boundary depends on te physics or application modes mostly you must use assembly mode to define the two boundary surfaces, each with opposed normals. In particular if you are working with contact surfaces, with our without sliding, then you should use assembly mode, if not normal union mode is easier and the model is simpler. Assembly mode means also that you will mesh each of the two boundary surfaces independently Hope this helps -- Good luck Ivar

Posted: 1 decade ago 29 mars 2011, 11:15 UTC−4

As a followup to this thread, I wonder if you could resolve a very odd problem I'm having, which perhaps has something to do with incorrect assembly of parts (??) I should mention that I brought in these 2 parts from Mimics, where I first smoothed and remeshed the parts from original stl files. I brought these new files into Comsol as stl files and assigned the properties of the two subdomains. The attached images give an idea of the example problem I'm running. I'm doing an example stress analysis of a small screw inside of a larger structure (mainly just to make sure this problem is indeed running OK). I did not apply any contact pairs or identity pairs where the threads meet inside. The puzzle is that the model appears to run fine when I am applying a distributed positively-directed z load on the top surface of the little screw; I fix the boundaries of most of the bottom of the outside structure into which the small screw is placed. On the one hand, the ex strains look reasonable for the bending of the small screw's shaft as expected, e.g., tension on the one side and compression on the other. HOWEVER, here is the very odd result: note that the deformed shape and the z displacement output seem to be saying that the screw bends toward the negative z direction! And note the scale for the displacements -- in the balpark of 1 e12 !! What is wrong? Am I missing something very basic in the model setup, leading to this absurd result? Thanks very much. [QUOTE] Hi you should learn the diferences of asembled (two bodies in contact but with each it's own boundary overalapping) and union mode where both bodies have a common boundary. The latter is mostly the default, as it means you have continuity of flux over this boundary, once you have "geomanalysed" your geometry, you have two bodies but one common boundary with "continuity". If you want to add specific physics on the boundary depends on te physics or application modes mostly you must use assembly mode to define the two boundary surfaces, each with opposed normals. In particular if you are working with contact surfaces, with our without sliding, then you should use assembly mode, if not normal union mode is easier and the model is simpler. Assembly mode means also that you will mesh each of the two boundary surfaces independently Hope this helps -- Good luck Ivar [/QUOTE]

Attachments:

• ex strains.jpg
• setup of boundary loading.jpg
• deformed shape plot.jpg
• z displacement.jpg

Posted: 1 decade ago 29 mars 2011, 16:10 UTC−4

Hi well its difficult to undertand your volume, from the smll items, are they really one volume or many assembled ? if you have such a displacement, are you sure youre part is "fixed" or sufficient constrined somewhere, it's just as if it hase translated freely by 1E12 m -- Good luck Ivar

Posted: 1 decade ago 29 mars 2011, 21:48 UTC−4

Thanks for the help. I would have attached the Comsol file itself but it's 5 MB and wasn't accepted. I figure you're right -- there must be some sort of missing constraint -- but why would the strain values be plausible if the small screw was not constrained where its threads exist in the other part? And why would the translation of the small screw (if that's what it is) be occurring in a direction opposite to the applied loading? There are two subdomains in this problem, one for the internal screw and one for the other body. My suspicion is that the mating between the two subdomains is the origin of the problem somehow. I'll keep looking into it. [QUOTE] Hi well its difficult to undertand your volume, from the smll items, are they really one volume or many assembled ? if you have such a displacement, are you sure youre part is "fixed" or sufficient constrined somewhere, it's just as if it hase translated freely by 1E12 m -- Good luck Ivar [/QUOTE]