Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Problems with heat continuity over an identity pair (+moving mesh)

Please login with a confirmed email address before reporting spam

I'm having a lot of problems modeling a rotating machine heat transfer.

It's working nicely if i Disable the Prescribed deformation node in the Moving mesh physical description (ie. the machine is not moving at all, works also if i set rotation speed parameter omega to zero), but once i rotate the mesh the boundary between two assembly parts becomes insulated, or so it seems.

Could anyone please help me get the settings right so the temperature is continuous over the assembly pair?

Gorazd


12 Replies Last Post 30 mars 2011, 15:02 UTC−4
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 24 mars 2011, 02:57 UTC−4
Hi

I'm not sure what is happening, but you are first of all observing the heating up from the shaft temperature.

I would propose that you
1) turn on the plotting while solving to better follow what is happening select in the Fully Coupled Solver node
2) set the initial temperature to 315 and not the default 20°C

Still it's strange, indeed, sorry haven't found anything else obvious, apart that you might need to wait longer, perhaps support might help

I tried to exchange source and destination, no effect, sometimes one sees the heat pass, then it isolates again.

By the way you have done a good job for the geometry, but still, wouldn't it have been easier to work in 2D with a work plane and then extrude it. One thing in COMSOL, there is no need to make the system as 1 part, you can cut it up the parts with interior boundaries without risk, you have the array and rotate functions to repeat the shapes, when they are nice ans symmetric as yours

--
Good luck
Ivar
Hi I'm not sure what is happening, but you are first of all observing the heating up from the shaft temperature. I would propose that you 1) turn on the plotting while solving to better follow what is happening select in the Fully Coupled Solver node 2) set the initial temperature to 315 and not the default 20°C Still it's strange, indeed, sorry haven't found anything else obvious, apart that you might need to wait longer, perhaps support might help I tried to exchange source and destination, no effect, sometimes one sees the heat pass, then it isolates again. By the way you have done a good job for the geometry, but still, wouldn't it have been easier to work in 2D with a work plane and then extrude it. One thing in COMSOL, there is no need to make the system as 1 part, you can cut it up the parts with interior boundaries without risk, you have the array and rotate functions to repeat the shapes, when they are nice ans symmetric as yours -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 24 mars 2011, 11:21 UTC−4
I got exactly the same problem.
any way out?
Songi
I got exactly the same problem. any way out? Songi

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 25 mars 2011, 04:06 UTC−4
Hello guys,

i think i got it on my own. Seems to me that, when i set the Frame of the identity pair to Material (X,Y,Z), things work out nicely. Still thanks for the help Ivar!

BR, Gorazd
Hello guys, i think i got it on my own. Seems to me that, when i set the Frame of the identity pair to Material (X,Y,Z), things work out nicely. Still thanks for the help Ivar! BR, Gorazd

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 25 mars 2011, 10:49 UTC−4
Hi Gorazd

could you please let me know what do you mean by 'when i set the Frame of the identity pair to Material (X,Y,Z),'

I mean, I have done the moving mesh in 3D and when I go to the prescribed deformation there is just an option to input dx or dy or dz. . could you please let me know where should I find the frame of pair? have you changed it from heat transfer setting? I did that but again did not answered.


another problem is I can define just free deformation, when I define prescribed mesh velocity or mesh displacement there is no object selectable and all surfaces became not applicable!!

could you say where is my problem.

Navid
Hi Gorazd could you please let me know what do you mean by 'when i set the Frame of the identity pair to Material (X,Y,Z),' I mean, I have done the moving mesh in 3D and when I go to the prescribed deformation there is just an option to input dx or dy or dz. . could you please let me know where should I find the frame of pair? have you changed it from heat transfer setting? I did that but again did not answered. another problem is I can define just free deformation, when I define prescribed mesh velocity or mesh displacement there is no object selectable and all surfaces became not applicable!! could you say where is my problem. Navid

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 26 mars 2011, 03:53 UTC−4
Hi

should it be the Spatial Material or Mesh Frame, in ALE they differ !?

To be more precise Material and Mesh Frame dissociates with "Deformed Geometry, not with ALE)

In my present simple 2D example (v4.1.0.185) its the Spatial one that gives the most "physical" correct results

Remains why its not working correctly in the "Rotary" example

--
Good luck
Ivar
Hi should it be the Spatial Material or Mesh Frame, in ALE they differ !? To be more precise Material and Mesh Frame dissociates with "Deformed Geometry, not with ALE) In my present simple 2D example (v4.1.0.185) its the Spatial one that gives the most "physical" correct results Remains why its not working correctly in the "Rotary" example -- Good luck Ivar


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 26 mars 2011, 08:38 UTC−4
Hello,

I'll answer the questions with what i did. It seems to me it works correctly but i do not guarantee it.


"could you please let me know what do you mean by 'when i set the Frame of the identity pair to Material (X,Y,Z),'"
Follow the nodes:
Model 1 (mod 1) >>Definitions>> Identity pair 1 (2, 3,4 depends on how many you have)
In the settings window you can find a "Frame" sub-section and there you can set the frame to MAterial(X,Y,Z)

Now, this will only work if you have created an assembly not a union at the end of geometry modeling.


"I mean, I have done the moving mesh in 3D and when I go to the prescribed deformation there is just an option to input dx or dy or dz. . could you please let me know where should I find the frame of pair? have you changed it from heat transfer setting? I did that but again did not answered."
I got the mesh to rotate by prescribing the deformation. It was odd to me, and i also tried with the mesh velocity with the same problems...but it seems to be working with current settings.
Look at the model I posted for the correct geometrical definition of mesh deformation for rotation.


Hope it helps

Gorazd
Hello, I'll answer the questions with what i did. It seems to me it works correctly but i do not guarantee it. "could you please let me know what do you mean by 'when i set the Frame of the identity pair to Material (X,Y,Z),'" Follow the nodes: Model 1 (mod 1) >>Definitions>> Identity pair 1 (2, 3,4 depends on how many you have) In the settings window you can find a "Frame" sub-section and there you can set the frame to MAterial(X,Y,Z) Now, this will only work if you have created an assembly not a union at the end of geometry modeling. "I mean, I have done the moving mesh in 3D and when I go to the prescribed deformation there is just an option to input dx or dy or dz. . could you please let me know where should I find the frame of pair? have you changed it from heat transfer setting? I did that but again did not answered." I got the mesh to rotate by prescribing the deformation. It was odd to me, and i also tried with the mesh velocity with the same problems...but it seems to be working with current settings. Look at the model I posted for the correct geometrical definition of mesh deformation for rotation. Hope it helps Gorazd

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 26 mars 2011, 08:46 UTC−4
Hello Ivar,

I'll check your model on monday, i don't have comsol at home.

But as far as i can tell from setting all materials to have the same heat conductivity in my model, the heat flow is smooth and the temperature continuous. Now i haven't done a simple model that would confirm this with analitical calculation as i thought that it should be correct since it's the only one that provides continuity over the pair. I checked it on a couple of more demanding models and the results are the same...with rotation only Material frame works for continuity ower the pair...don't know why.

To me either one of the others would be more logical...


... but the problems do not end here... if i do the same model with the magnetic fields module, none of the frames provides magnetic field continuity with rotation but works for a stand -still model. I'll send the model on Monday as well if you can help :)

I'll also try and do them in 2D just to see if it's conceptually easier to understand what works and why...

Again, I'm using 4.0a ...maybe this is different for 4.1


Gorazd
Hello Ivar, I'll check your model on monday, i don't have comsol at home. But as far as i can tell from setting all materials to have the same heat conductivity in my model, the heat flow is smooth and the temperature continuous. Now i haven't done a simple model that would confirm this with analitical calculation as i thought that it should be correct since it's the only one that provides continuity over the pair. I checked it on a couple of more demanding models and the results are the same...with rotation only Material frame works for continuity ower the pair...don't know why. To me either one of the others would be more logical... ... but the problems do not end here... if i do the same model with the magnetic fields module, none of the frames provides magnetic field continuity with rotation but works for a stand -still model. I'll send the model on Monday as well if you can help :) I'll also try and do them in 2D just to see if it's conceptually easier to understand what works and why... Again, I'm using 4.0a ...maybe this is different for 4.1 Gorazd

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 26 mars 2011, 11:58 UTC−4
Hi

these issues are best sent to "support", as we out here do not have insights to the internals of the software, and indeed there is something not working in the rotating cases,
I just crashed 3x my simple rotary model and now COMSOL.exe seems to be stuck too, it stops while flashing the splash ;)

Pls keep us informed, so we may share all work-arounds

--
Good luck
Ivar
Hi these issues are best sent to "support", as we out here do not have insights to the internals of the software, and indeed there is something not working in the rotating cases, I just crashed 3x my simple rotary model and now COMSOL.exe seems to be stuck too, it stops while flashing the splash ;) Pls keep us informed, so we may share all work-arounds -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 29 mars 2011, 04:45 UTC−4

Hi

should it be the Spatial Material or Mesh Frame, in ALE they differ !?

To be more precise Material and Mesh Frame dissociates with "Deformed Geometry, not with ALE)

In my present simple 2D example (v4.1.0.185) its the Spatial one that gives the most "physical" correct results

Remains why its not working correctly in the "Rotary" example

--
Good luck
Ivar


Hi, Ivar

It's my daily work to check your new threads, and i found really insightful and illuminating for me to sort out my project, which concerns the mobile induction heating.

what I have ever learned about ALE is something like "remesh" and "stop condition", sorry i can't remember clearly. But i find there is nothing to do wiht your posted model, which looks wonderful. Does this mean that the new version COMSOL can deal with the remeshing problem more powerful? In the post result, how to display the deformation of the mesh?

Thanks for your kindness.
[QUOTE] Hi should it be the Spatial Material or Mesh Frame, in ALE they differ !? To be more precise Material and Mesh Frame dissociates with "Deformed Geometry, not with ALE) In my present simple 2D example (v4.1.0.185) its the Spatial one that gives the most "physical" correct results Remains why its not working correctly in the "Rotary" example -- Good luck Ivar [/QUOTE] Hi, Ivar It's my daily work to check your new threads, and i found really insightful and illuminating for me to sort out my project, which concerns the mobile induction heating. what I have ever learned about ALE is something like "remesh" and "stop condition", sorry i can't remember clearly. But i find there is nothing to do wiht your posted model, which looks wonderful. Does this mean that the new version COMSOL can deal with the remeshing problem more powerful? In the post result, how to display the deformation of the mesh? Thanks for your kindness.

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 29 mars 2011, 15:31 UTC−4
Hi

in V4 "ask for" a "mesh plot". But what I find more useful is the plot while solving , see the main solver node. Then you can see how the solver seeks its solution and you will rapidly detect what is going wrong.

Indeed in none of these models there are remeshing, only moving mesh a la ALE. Which works fine for me in linear, I have still some issues (as others) with the rotary model, still working on it, but I have little access to COMSOl these days, mostly on the move

--
Good luck
Ivar
Hi in V4 "ask for" a "mesh plot". But what I find more useful is the plot while solving , see the main solver node. Then you can see how the solver seeks its solution and you will rapidly detect what is going wrong. Indeed in none of these models there are remeshing, only moving mesh a la ALE. Which works fine for me in linear, I have still some issues (as others) with the rotary model, still working on it, but I have little access to COMSOl these days, mostly on the move -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 30 mars 2011, 03:37 UTC−4
Hello Ivar,
The sliding block model works pretty well and how can one set up a pulsing heating? Let say in the sliding block? When you say you have issues with rotary model, is the rotary model such that instead of having rectangle sliding on top of the other, you have a circle (2D) rotating as it moves on top of a fixed rectangle?
thanks
song
Hello Ivar, The sliding block model works pretty well and how can one set up a pulsing heating? Let say in the sliding block? When you say you have issues with rotary model, is the rotary model such that instead of having rectangle sliding on top of the other, you have a circle (2D) rotating as it moves on top of a fixed rectangle? thanks song

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 30 mars 2011, 15:02 UTC−4
Hi

if you want to do a transient analysis, <ou select such a study, which will define by default "t" for you as the time. + adding in the time derivative elements of the general PDE equations. It's up to you then to add some time dependence on a particular boundary condition, i.e. a heat source Q that you multiply by a step function, or a Gaussian or ...

Only advice use a smooth derivable (at least once, better twice) time amplitude varying function.

I believe my rotary ALE issue is with the HT BCs and the coordinate transforms, I have pushed the issue further to "support". If you are not careful with nthe frames used, you might get circular references, with anoying side effects ;)

--
Good luck
Ivar
Hi if you want to do a transient analysis,

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.