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 The information provided may be out of date.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.
negative temperature -at different solver time range
Posted 3 nov. 2011, 08:26 UTC−4 Heat Transfer & Phase Change Version 4.1 3 Replies
Please login with a confirmed email address before reporting spam
Dear everyone,
I’m working on a laser-material interaction only using the “heat transfer mode”, but facing a tough problem relative to the negative temperature effect that the solved temperature is small than the initial temperature (the heat flux on one boundary is positive, the other boundary conditions are thermal insulation). I have searched in this web and found there will be three probably wrong settings, i.e. the boundary condition, mesh size/density and the time range or something esle.
In my model (heat transfer mode), there are simply only two kinds of BCs - thermal insulation and heat flux. The mesh size is “physical-controlled mesh” and “extremely fine”, and the average element quality is 0.9962. According the underlying probably wrong mistake, at first, I checked the BCs and then refined the mesh. Finally, I setting the “Study 1 --> Step 1: Time Dependent --> Study Settings --> Times” to “range (0, 1e-10, 45e-9)” , the resultant temperature is becoming smaller than the initiate temperature (293.15 K) as the solver time goes on, which is illustrated in the attachment “negative temperature 1” and “negative temperature 1 (enlarged)”. Afterward, I try to change the time step into “range (0, 1e-9, 45e-9)”, the negative temperature effect also exists. While, when the time step is “range (0, 1, 45)”, or “range (0, 1, 3)”, this problem disappears.
I’m quite puzzle about this problem. Could anyone else do me a favor to the mistake? Thanks sincerely in advance.
Ps. the “.mph” file (negative temperature.mph) is in the attachment.
Yours
FM Huang
Nov. 3rd, 2011
I’m working on a laser-material interaction only using the “heat transfer mode”, but facing a tough problem relative to the negative temperature effect that the solved temperature is small than the initial temperature (the heat flux on one boundary is positive, the other boundary conditions are thermal insulation). I have searched in this web and found there will be three probably wrong settings, i.e. the boundary condition, mesh size/density and the time range or something esle.
In my model (heat transfer mode), there are simply only two kinds of BCs - thermal insulation and heat flux. The mesh size is “physical-controlled mesh” and “extremely fine”, and the average element quality is 0.9962. According the underlying probably wrong mistake, at first, I checked the BCs and then refined the mesh. Finally, I setting the “Study 1 --> Step 1: Time Dependent --> Study Settings --> Times” to “range (0, 1e-10, 45e-9)” , the resultant temperature is becoming smaller than the initiate temperature (293.15 K) as the solver time goes on, which is illustrated in the attachment “negative temperature 1” and “negative temperature 1 (enlarged)”. Afterward, I try to change the time step into “range (0, 1e-9, 45e-9)”, the negative temperature effect also exists. While, when the time step is “range (0, 1, 45)”, or “range (0, 1, 3)”, this problem disappears.
I’m quite puzzle about this problem. Could anyone else do me a favor to the mistake? Thanks sincerely in advance.
Ps. the “.mph” file (negative temperature.mph) is in the attachment.
Yours
FM Huang
Nov. 3rd, 2011
Attachments:
3 Replies Last Post 8 nov. 2011, 03:14 UTC−5
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Using a very small time step gives the heat less time to penetrate through the structure. This results in the temperature rise localizing in a very thin outer “skin” region. If the depth of this region is smaller than that of the outer element layer you will get this numerical (non-physical) drop in surface temperatures. It should go away with time, but if you absolutely need to avoid it at the very small time steps, reduce the element size at the surface according to this equation: L=sqrt(k*dt/(rho*Cp)) where dt is the time step size. You will find more details on this topic in an older thread: www.comsol.com/community/forums/general/thread/16808/.
Nagi Elabbasi
Veryst Engineering
Nagi Elabbasi
Veryst Engineering
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Using a very small time step gives the heat less time to penetrate through the structure. This results in the temperature rise localizing in a very thin outer “skin” region. If the depth of this region is smaller than that of the outer element layer you will get this numerical (non-physical) drop in surface temperatures. It should go away with time, but if you absolutely need to avoid it at the very small time steps, reduce the element size at the surface according to this equation: L=sqrt(k*dt/(rho*Cp)) where dt is the time step size. You will find more details on this topic in an older thread: www.comsol.com/community/forums/general/thread/16808/.
Nagi Elabbasi
Veryst Engineering
Thanks for your immediate reply sincerely. According to your suggestion, the problem have been solved.
Best wishes to you and your family.
Yours
FM Huang
Nov. 8th, 2011
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
I made a simple example to play with mesh sizes and HT material values to see the relations with the mesh size and time stepping. All numerical values are just to play with, nothing "real" but the numerical effects YES
I calculate the decay times and the theoretical mesh size limits from the material parameters and the derived values estimates the average mesh size and corresponding decay times td = h^2/alpha (alpha heat diffusion k/Cp/rho)
It's also worth to study closely the Biot number and the theory there around
One clearly see (cut line graphs that for time steps shorter than about "td" we can easily get "negative temperatures"
the model is a v4.2a version
--
Good luck
Ivar
I made a simple example to play with mesh sizes and HT material values to see the relations with the mesh size and time stepping. All numerical values are just to play with, nothing "real" but the numerical effects YES
I calculate the decay times and the theoretical mesh size limits from the material parameters and the derived values estimates the average mesh size and corresponding decay times td = h^2/alpha (alpha heat diffusion k/Cp/rho)
It's also worth to study closely the Biot number and the theory there around
One clearly see (cut line graphs that for time steps shorter than about "td" we can easily get "negative temperatures"
the model is a v4.2a version
--
Good luck
Ivar
Attachments: