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.

Joule Heating Model

Please login with a confirmed email address before reporting spam

Hi,

I'm helping a friend in a simple eletrothermal model (joule heating) - attached file. We are considering conduction heat transfer only. But we are having 2 problems:

- when we run the stationary study, we get an error saying "Returned solution has not converged"

- when we run the time dependent study, the solver works. The voltage distribution seems ok but the temperature distribution is too high (we were expecting max temp around 750K).

It is such a simple a model, we could not find the mistake. Any help is much appreciated.

Thanks,
Andre



15 Replies Last Post 10 mars 2011, 02:31 UTC−5
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 14 janv. 2011, 16:07 UTC−5
Hi

first of all the "strain reference temperature is probably 300 K and not really 0 K, but I do not believe that is your issue.

For me, I do not see where the heat energy can leave the system, so the transient results does only increase and increase (or have I missed a point ?).This explain also, in my opinion, why the the stationary case does not solve succesfully.

Finally, use a tighter mesh, and why not apply the voltage and ground over a boundary, and not a "point" that represents singularities with current densities => Inf, no ?

--
Good luck
Ivar
Hi first of all the "strain reference temperature is probably 300 K and not really 0 K, but I do not believe that is your issue. For me, I do not see where the heat energy can leave the system, so the transient results does only increase and increase (or have I missed a point ?).This explain also, in my opinion, why the the stationary case does not solve succesfully. Finally, use a tighter mesh, and why not apply the voltage and ground over a boundary, and not a "point" that represents singularities with current densities => Inf, no ? -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 14 janv. 2011, 17:03 UTC−5
Hi,
i am not able to open your file .
but i can suggest you certain things.

if it is a 2d model, try increasing the heat transfer coefficient value (in thousands) and solve. by this way you can avoid this error.

if it is a 3d model try using iterative solvers to avoid the error message. (GMRES)

you could temperature in the desired range if you do any of theses depending upon the requirement


sriram
Hi, i am not able to open your file . but i can suggest you certain things. if it is a 2d model, try increasing the heat transfer coefficient value (in thousands) and solve. by this way you can avoid this error. if it is a 3d model try using iterative solvers to avoid the error message. (GMRES) you could temperature in the desired range if you do any of theses depending upon the requirement sriram

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 17 janv. 2011, 14:09 UTC−5
Hi Ivar,

Thanks for your tips! My boundary conditions was wrong ( as you said, "I do not see where the heat energy can leave the system" ). I added a Temperature Node (300K) to work as drain of energy and I got the expected results!

Thanks again,
Andre
Hi Ivar, Thanks for your tips! My boundary conditions was wrong ( as you said, "I do not see where the heat energy can leave the system" ). I added a Temperature Node (300K) to work as drain of energy and I got the expected results! Thanks again, Andre

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23 févr. 2011, 12:52 UTC−5
Hey,

I made a simple Joule heating model. I made it run very well in COMSOL 3.4. However, I move to 4.0a, it does not work. I am really confused.

Could you please look at my model and show me what is wrong with my model?

Thanks.

Tu

Hey, I made a simple Joule heating model. I made it run very well in COMSOL 3.4. However, I move to 4.0a, it does not work. I am really confused. Could you please look at my model and show me what is wrong with my model? Thanks. Tu


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 24 févr. 2011, 08:02 UTC−5
Hi Tu,

I looked at your model and I could not find anything wrong. I tested in v4.1, the solver works well but the temperature distribution is uniform, which is not expected. I draw a equivalent 2d geometry and I got the same results. If you fix it please let us know.

André
Hi Tu, I looked at your model and I could not find anything wrong. I tested in v4.1, the solver works well but the temperature distribution is uniform, which is not expected. I draw a equivalent 2d geometry and I got the same results. If you fix it please let us know. André

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 24 févr. 2011, 09:07 UTC−5
Hi Andre,

Could you please attach your result here? The expected maximum temperature should be about 650oC.
Thanks.

Tu
Hi Andre, Could you please attach your result here? The expected maximum temperature should be about 650oC. Thanks. Tu

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 févr. 2011, 09:13 UTC−5
Hi

have yu checked the resistivity of your device ? for me you are dissipating some 1E-12W not enough to heat it up. Try copper instead it will get hotter ;)

--
Good luck
Ivar
Hi have yu checked the resistivity of your device ? for me you are dissipating some 1E-12W not enough to heat it up. Try copper instead it will get hotter ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 24 févr. 2011, 09:25 UTC−5
Hi Ivar,

I did the same model in COMSOL 3.4 and it work very well. It shows the maximum temperature at the central segment of the microbirdge of 653oC. So, the resistivity may be not the reasons.

Tu
Hi Ivar, I did the same model in COMSOL 3.4 and it work very well. It shows the maximum temperature at the central segment of the microbirdge of 653oC. So, the resistivity may be not the reasons. Tu

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 24 févr. 2011, 09:51 UTC−5

oh, many thanks to Ivar.

I try with copper and it work well. However, I still dont know why it still work in COMSOL 3.4 with silicon.

Tu
oh, many thanks to Ivar. I try with copper and it work well. However, I still dont know why it still work in COMSOL 3.4 with silicon. Tu

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 févr. 2011, 10:03 UTC−5
Hi

check the conductivity values, careful with material data, it's often safer to type in the values yourself

--
Good luck
Ivar
Hi check the conductivity values, careful with material data, it's often safer to type in the values yourself -- Good luck Ivar

Magnus Ringh COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 24 févr. 2011, 11:14 UTC−5
Hi,

My guess is that you weren't using the electric conductivity or resistivity of silicon in the 3.4 version. The Silicon material did not include values for those properties in earlier versions, and COMSOL Multiphysics then (pre-4.0) retained the default values for properties undefined by the material. Only material property values shown in a bold font were taken from the material. I think the resisitivity of silicon should be something like 6*10^2 ohm*m whereas the default value in the pre-4.0 COMSOL Multiphysics for Joule heating was something like 2*10^(-8) ohm*m. That is a huge difference that will affect the results a lot.

Best regards,
Magnus Ringh, COMSOL
Hi, My guess is that you weren't using the electric conductivity or resistivity of silicon in the 3.4 version. The Silicon material did not include values for those properties in earlier versions, and COMSOL Multiphysics then (pre-4.0) retained the default values for properties undefined by the material. Only material property values shown in a bold font were taken from the material. I think the resisitivity of silicon should be something like 6*10^2 ohm*m whereas the default value in the pre-4.0 COMSOL Multiphysics for Joule heating was something like 2*10^(-8) ohm*m. That is a huge difference that will affect the results a lot. Best regards, Magnus Ringh, COMSOL

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2 mars 2011, 11:43 UTC−5

Hi,

My guess is that you weren't using the electric conductivity or resistivity of silicon in the 3.4 version. The Silicon material did not include values for those properties in earlier versions, and COMSOL Multiphysics then (pre-4.0) retained the default values for properties undefined by the material. Only material property values shown in a bold font were taken from the material. I think the resisitivity of silicon should be something like 6*10^2 ohm*m whereas the default value in the pre-4.0 COMSOL Multiphysics for Joule heating was something like 2*10^(-8) ohm*m. That is a huge difference that will affect the results a lot.

Best regards,
Magnus Ringh, COMSOL


Many thanks to Ivar and Magnus.

You are right. I fixed it and the simple model worked well now.

I wanted to go to 2nd step. I need to know the temperature surrounding the microbridge. To do that I add a hydrogen block and see the heat transfer in solid and fluid.

However, I got error like so

Failed to evaluate variable Jacobian.
- Variable: mod1.T
- Geometry: 1
- Domain: 1

Can you suggest me a way to fix it?

Any suggestion is appreciated.
Thanks

Tu
[QUOTE] Hi, My guess is that you weren't using the electric conductivity or resistivity of silicon in the 3.4 version. The Silicon material did not include values for those properties in earlier versions, and COMSOL Multiphysics then (pre-4.0) retained the default values for properties undefined by the material. Only material property values shown in a bold font were taken from the material. I think the resisitivity of silicon should be something like 6*10^2 ohm*m whereas the default value in the pre-4.0 COMSOL Multiphysics for Joule heating was something like 2*10^(-8) ohm*m. That is a huge difference that will affect the results a lot. Best regards, Magnus Ringh, COMSOL [/QUOTE] Many thanks to Ivar and Magnus. You are right. I fixed it and the simple model worked well now. I wanted to go to 2nd step. I need to know the temperature surrounding the microbridge. To do that I add a hydrogen block and see the heat transfer in solid and fluid. However, I got error like so Failed to evaluate variable Jacobian. - Variable: mod1.T - Geometry: 1 - Domain: 1 Can you suggest me a way to fix it? Any suggestion is appreciated. Thanks Tu


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 2 mars 2011, 14:56 UTC−5
Hi

have you carefully checked your model ? I see at least a k_Si parameter or variable not fully defined for the solid thermal domain (it's orange so you have an units issue) I would write something as:

k_Si = (33 + 5500/(T[1/K]) - 423000/(T[1/K])^2 + ...)*1[W/m/K]

In your HT module the heat sorce conditions englobes both gas and solid, while the physics you are invoquing is only defined for the solid.

Then I do not see how the gas is cooling the resistior, in the JH module your resistor is "thermally isolated", which is NOT the true cases since the gas is cooling it, or ?

I hope you have noticed that you have 2 temperatures as you have defined your physics a T from the JH and a T2 from the HT. But you define all your material properties only with T hence your error message.
Then both physics are calculating different temperatures since there is no heat exchange in the JH

Then why two transient solvers, one is plenty, specially since each physics are active in both solvers

I believe you have not chosen an optimum combination of physics, as things are done in double and are not really coupled. With your non-linear T dependent materials properties you need to couple to a T calculation i.e. from HT but with solid to fluid exchange set-up. And, for the current voltage and finally heat dissipaion its enough with an EC which solves the current density from the material shape, resisitivity and applied voltage and estimates the dissipated power (provided an external temperature is used for the resistivity T dependence).

So what do you think? I propose tha you sketch down the exchages, and then choose your physics with a minimum of overlap (and to check that HT does effectively couple already heat exchange between the fluid and the solid, and that you do not need a Conjugate Heat Transfer module)

--
Good luck
Ivar
Hi have you carefully checked your model ? I see at least a k_Si parameter or variable not fully defined for the solid thermal domain (it's orange so you have an units issue) I would write something as: k_Si = (33 + 5500/(T[1/K]) - 423000/(T[1/K])^2 + ...)*1[W/m/K] In your HT module the heat sorce conditions englobes both gas and solid, while the physics you are invoquing is only defined for the solid. Then I do not see how the gas is cooling the resistior, in the JH module your resistor is "thermally isolated", which is NOT the true cases since the gas is cooling it, or ? I hope you have noticed that you have 2 temperatures as you have defined your physics a T from the JH and a T2 from the HT. But you define all your material properties only with T hence your error message. Then both physics are calculating different temperatures since there is no heat exchange in the JH Then why two transient solvers, one is plenty, specially since each physics are active in both solvers I believe you have not chosen an optimum combination of physics, as things are done in double and are not really coupled. With your non-linear T dependent materials properties you need to couple to a T calculation i.e. from HT but with solid to fluid exchange set-up. And, for the current voltage and finally heat dissipaion its enough with an EC which solves the current density from the material shape, resisitivity and applied voltage and estimates the dissipated power (provided an external temperature is used for the resistivity T dependence). So what do you think? I propose tha you sketch down the exchages, and then choose your physics with a minimum of overlap (and to check that HT does effectively couple already heat exchange between the fluid and the solid, and that you do not need a Conjugate Heat Transfer module) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 9 mars 2011, 18:43 UTC−5
Hello,

My model is pretty good now. Thank you Ivar. Your suggestion really helps me a lot.

The only one thing that I concerned now is the difference between Steady state model and Transient model.

For the transient model. I cannot apply a higher voltage to get higher temperature. For example, my model work well with the voltage is no greater than 3.5 V. At a higher voltage, however, it does not work and shows an error of "inconsistent initial values".

For the steady state model. It work very well with any value of resistivity or voltage.

Any one can give me some explanation for those things?

Thanks.

Best regards,
Tu
Hello, My model is pretty good now. Thank you Ivar. Your suggestion really helps me a lot. The only one thing that I concerned now is the difference between Steady state model and Transient model. For the transient model. I cannot apply a higher voltage to get higher temperature. For example, my model work well with the voltage is no greater than 3.5 V. At a higher voltage, however, it does not work and shows an error of "inconsistent initial values". For the steady state model. It work very well with any value of resistivity or voltage. Any one can give me some explanation for those things? Thanks. Best regards, Tu


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 10 mars 2011, 02:31 UTC−5
Hi

not directly, probably its because your default initial temperature is too far off for the solver to find a solution. one way is to add a specific temperature profile on your heating part, that is closer to the solution.

Still one thing is bothering me: I see a symmetric heat flux into the He, but is is not "transported by the flow, and I see no convection (convecting energy loss is also "0") on the surfaces. There are a few subtlety with the JH I'm missing, you might need to use a NS physics for the fluid, need to check the doc there

And are you sure your bridge structure has a sigma of 1E-12 S/m ? for the base I understand but not for your resistor

And you have an unit issue with Ux but this might be that you have been lazy with the units in the constants for its definition

--
Good luck
Ivar
Hi not directly, probably its because your default initial temperature is too far off for the solver to find a solution. one way is to add a specific temperature profile on your heating part, that is closer to the solution. Still one thing is bothering me: I see a symmetric heat flux into the He, but is is not "transported by the flow, and I see no convection (convecting energy loss is also "0") on the surfaces. There are a few subtlety with the JH I'm missing, you might need to use a NS physics for the fluid, need to check the doc there And are you sure your bridge structure has a sigma of 1E-12 S/m ? for the base I understand but not for your resistor And you have an unit issue with Ux but this might be that you have been lazy with the units in the constants for its definition -- Good luck Ivar

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.