Discussion Forum

Posted: 1 decade ago Apr 21, 2011, 11:28 a.m. EDT

Hi I cannot find back the example I believe existed in 3.5 but you could study , by analogy, the model library case: Comsol Multiphysics - Electromagnetics - Thin film resistance its a electric resistive and not thermal case, but very similar For the basic physics without the contact though. It's somewhat more complex when you add the contact as the contact area varies (as well as the pressure hence normally the thermal conductivity of the contact too, but that you must model yourself I believe, via your own equations) The PAirs as you say. Is linked to the COMSOL Assembly mode (nothing to do with CAD assembly) you set that in the Geometry Finish node. But before, group (via a "Geometry - Boolean Operations - Compose") your geometries where you want to apply "Continuity" by default and leave only (if possible) the contact boundaries as separate items. In assembly mode, if two independent Domains (=volumes in 3D, surfaces in 2D) are adjacent with a common boundary this boundary will be de-doubled, one attached to the "left domain" the other to the "right domain". While in "union" mode the common boundary will be replaced by a shared boundary with full continuity of the physics applied to it. This is a fundamental concept in COMSOL that must be carefully studied and understood to take all the advantages of this tool For boundaries that are de-doubled in assembly mode, and for which you want a "continuity" of the physics, you need to define an identity pair under "Definitions" and then apply a "Continuity Pair - Boundary Condition, specifically -- Good luck Ivar

Posted: 1 decade ago Apr 22, 2011, 10:56 a.m. EDT

Ivar, I attempted to do what I think you are saying, but when I do I get the error: Fail to evaluate variable Jacobian in addition a heat flux I am trying to model is locally reaching 10^4 K while the surrounding areas are remaining unaffected. If you could take a look at my model I would be very appreciative. Thank you, John Haase

Attachments:

• Real_Geometry2.mph

Posted: 1 decade ago Apr 22, 2011, 4:10 p.m. EDT

Everyone, I'm sorry about the triple post. I figured out that I needed to apply a conductive boundary and a thin thermally resistive layer. However there is one more issue which I am having trouble with. I am trying to model a laser with a simple heat flux, however the laser interact beneath the surface and the problem I'm having is that when the area I want to affect is in a union there is only the top of the shape to select and when there is no union (and thereby an assembly) there is no heat transfer from that region. If someone could take a look at my model I would appreciate it. Thank you, John Haase

Attachments:

• Real_Geometry4.mph

Posted: 1 decade ago Apr 23, 2011, 8:16 a.m. EDT

Hi its linked to the way you select the options in your union. If you do not keep the interior boundaries the two intersecting volumes will not coexist later. If you need just the surface shape, you can forget the interior boundaries, if you want to use a volumetric heat source penetrating into the material add the interior boundaries, you can do it the way I believe you are trying, but cutting out a small volume from the main one onto which you can apply a body heat source, not just a surface heat flux. Check the number of domains you end up with it varies from 2 to 3, Inspect the same way your boundaries. In "assembly mode" you have two overlapping boundaries for the adjacent interfaces (be aware volumes that penetrate are nut cut !) This allows to apply specific physics on the de-doubled boundaries, or identify pairs and decide yourself how you couple them (to be done for all physics applicable !) -- Good luck Ivar

Posted: 1 decade ago Jun 1, 2012, 11:03 a.m. EDT

Hi Ivar, I have a geometry with multiple interfaces but I only want one interface to be de-doubled and the rest 'unionized'. Can this be done? Thank you. EH [QUOTE] Hi I cannot find back the example I believe existed in 3.5 but you could study , by analogy, the model library case: Comsol Multiphysics - Electromagnetics - Thin film resistance its a electric resistive and not thermal case, but very similar For the basic physics without the contact though. It's somewhat more complex when you add the contact as the contact area varies (as well as the pressure hence normally the thermal conductivity of the contact too, but that you must model yourself I believe, via your own equations) The PAirs as you say. Is linked to the COMSOL Assembly mode (nothing to do with CAD assembly) you set that in the Geometry Finish node. But before, group (via a "Geometry - Boolean Operations - Compose") your geometries where you want to apply "Continuity" by default and leave only (if possible) the contact boundaries as separate items. In assembly mode, if two independent Domains (=volumes in 3D, surfaces in 2D) are adjacent with a common boundary this boundary will be de-doubled, one attached to the "left domain" the other to the "right domain". While in "union" mode the common boundary will be replaced by a shared boundary with full continuity of the physics applied to it. This is a fundamental concept in COMSOL that must be carefully studied and understood to take all the advantages of this tool For boundaries that are de-doubled in assembly mode, and for which you want a "continuity" of the physics, you need to define an identity pair under "Definitions" and then apply a "Continuity Pair - Boundary Condition, specifically -- Good luck Ivar [/QUOTE]

Posted: 1 decade ago Jun 1, 2012, 2:05 p.m. EDT

Hi as often yes and no, what you can do is to group a series of geometrical objects and make an (or several) explicit union(s) between these, then when you exit in Assembly mode the dedoubled boundaries will only be applie between the last full objects. So if you end up with two union onjects, each conaining many sub onbjects, and these two objects have one common boundary, then in assembly mode this common border will be dedoubled with an "up" and "down" side. You will notice this if you click on it you first catch one, then the other -- Good luck Ivar