Thermoelectric Effect - Adding the physics

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

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hi

I believe you have that pysics in the model builder section, check your doc.
Also use lower case test() for the test function

In the model builder you have access to the del grad ... these have special caracters see the doc

--
Good luck
Ivar

Pheba Cherian

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Thank you Ivar, But its still not working. Do any1 hav any other suggestions. Anyone who worked in thermoelectric Effect please help....

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hi,

I solved the TE problem using the PDE coefficient approach.

If someone still needs halp, I'll be happy to provide it!

bye

Karunakar B

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hi Dario,
I am trying to simulate thermocouple by following the physics builder user guide example.But I am getting an error like failed to evaluate a variable sigma,S...etc.Can you help to solve this problem or please send an example(using PDE) so that I can do it much easier......

Regards
karunakar

Gaston Piller

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hello Dario,

I am having difficulties with the Thermoelectric effect model. Are you still available for help?

Thanks in Advanced,

Gaston Piller

Hi,

I solved the TE problem using the PDE coefficient approach.

If someone still needs halp, I'll be happy to provide it!

bye

Susant Patra

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hi,

I am trying to model transient effect in TEC. Comsol physics builder has a steady state version. Need help in incorporating time dependency.

Gaston Piller

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

I would incorporate the equations for the Seebeck and Peltier Effect as weak contributions instead of using the model provided by COMSOL.

Susant Patra

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hi,

Thanks for your response. The weak contribution is not domain specific. How can I identify which component can be assigned seebek coeff? Is there a way to add seebek coeff in material specification?

Regards,
Susant

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hi everyone,

I'm back.

Someone still needs help?

I have developed the TE model using the PDEs approach, someone succeed in solving the problem using the approach described here:
www.ewp.rpi.edu/hartford/~lewisl2/Engineering%20Project/Other/References/COMSOL%20TE%20Setup.pdf

?

Thanks

Susant Patra

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hi,

I tried to add theromelectric effect in weak form. But the weak form formulation is not domain specific. So how can I provide different seebek coeff associated with different domain?

Regards,
Susant

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hi,

I can't help you with the weak form, but can I ask you to send me the your file? I want to practice with this approach.

I can give you my .mph file with the PDEs form + thermomechanical module, there you can change S coefficient, conductivity (both thermal and electric) and also introduce their temperature dependence

Thanks

Gaston Piller

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hi,

I am not sure why you would say the 'weak contribution' is not domain specific. Here is what I would do: add the conditions and equations as weak form for the phenomenons for the domains needed. Then,
define your coefficients for each domain, Seebeck and Peltier. Note that these can be temperature dependent or simply functions of temperature. This is what I have done to create P-N pairs.

I also created the "Thermoelectric Physics" as Dario suggested, but decided not to use it because it can be quite primitive when creating larger models. Also, why not take advantage of predefined physics interphases already available. As that same PDF suggests, it is not the best way to simulate the Thermoelectric Effect.

Hope this helps,

Gaston

Hi,

I tried to add theromelectric effect in weak form. But the weak form formulation is not domain specific. So how can I provide different seebek coeff associated with different domain?

Regards,
Susant

Susant Patra

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hi,

I am trying to duplicate the form shown in the attached paper. I am not sure how to add the conditions and equations for the specific domains? I would appreciate if you can provide a screen shot of the comsol steps to do it.

Regards,
Susant

Attachments:

crompton_presentation.pdf

Sergei Yushanov Certified Consultant

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Susant,

Attached is example with implementation of the referenced presentation.

Regards,
Sergei

Attachments:

Example Peltier_Seebeck_effect.mph

Susant Patra

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hi Sergei,

Thank you very much for the example. It is very helpful.

Regards,
Susant

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Susant,

Attached is example with implementation of the referenced presentation.

Regards,
Sergei

Dear Sergei, please can you give me the same file compatible with comsol 3.5a?

Thanks

Sergei Yushanov Certified Consultant

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Dario,

Sorry, I don't have files compatible with 3.5a.

Regards,
Sergei

Ying Yun Lim

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Dear Sergei,

I saw in your paper 'Multiphysics Analysis of Thermoelectric Phenomena' that you used COMSOL V4.2 for your simulations. Do you, by any chance, still have the V4.2 of the multiphysics file you sent to Susant?

It would be very much appreciated. Thank you.

Regards,
Ying

Sergei Yushanov Certified Consultant

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Ying,

Attached is 4.2 version of the file.

Regards,
Sergei

Attachments:

Example Peltier_Seebeck_effect_42.mph

Ying Yun Lim

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Thank you very much, Sergei!

Jonatan Persson

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Dear Sergei, have you tried plotting the current distribution? I think you will find the results interesting. I have attached a figure of the current distribution of the Jaegel/comsol guide thermoleg simulated with your physics.

I'm interested in how one should choose testfunctions for the weak equations in one physics interface with dependent variables. Using the method with weak contributions the one in the heat transfer interface tests for its own variable T and in electric currents use V. But in the physics guide I cant find any reason for that one variable should test one and not the other weak equation.

q·test(∇T)+Q·test(T)
-J·test(∇V)

Why not:
q·test(∇V)+Q·test(V)
-J·test(∇T)

The same goes for boundary conditions
qin*test(T)
Jin*test(V)

Attachments:

Currentdensity.png

Sergei Yushanov Certified Consultant

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Jonatan,

Attached contains derivation of the thermoelectric weak contributions. Hopefully, this will answer your question.

Regards,
Sergei

Attachments:

Thermoelectricity.pdf

Jonatan Persson

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

"Multiply energy balance equation by test function Ttest and integrate over
computational domain :"
and
"Multiply current balance equation by test function Vtest and integrate over
computational domain :"

This is like the guide more of a statement rather than an explanation. Both equations depends on both variables to a simmilar degree, so I still dont see why this is the obvious way to do this. I might have to go with the coefficient form where I dont have to account for this step.

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Jonatan,

Attached contains derivation of the thermoelectric weak contributions. Hopefully, this will answer your question.

Regards,
Sergei

Dear Sergei,

can you help me in implementing this in comsol 3.5a?

It is impossible to modify the weak term in the heat transfer mode. How this can be done?

Thanks,

Dario

Sergei Yushanov Certified Consultant

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Dario,

Attached example shows Peltier weak contribution implementation in Comsol 3.5a.

Regards,
Sergei

Attachments:

peltier_test-35a.mph

Jonatan Persson

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

I figured the reason to be the coefficient values (low S high sigma ) which makes the proposed use of testfunctions to make good aproximations.

Sergei Yushanov Certified Consultant

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Jonatan,

Thermoelectricity is now available in 4.4 version.

Regards,
Sergei

Ying Yun Lim

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Dear Sergei,

A couple of weeks ago you sent me a copy of the Seebeck effect file for comsol v4.2a, and I am trying to make it work in a 3D environment.

may I know how did you add the 'weak equation' under the Electric Current tab or how should I add a 'weak equation' under that tab? Because right clicking the tab, I couldn't find 'weak equation' under the list.

Thank you in advance!

Ying

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Thank you so much Sergei, yesterday I managed to do it by myself in 2D axyal symmetric and 3D thanks to the pdf file you have attached. With the 3D model I make a comparison with my previous model in PDE form obtaining exactly the same result. Now I can also check if i did it correctly the 2D model.

Ying Yun Lim

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Thank you so much Sergei, yesterday I managed to do it by myself in 2D axyal symmetric and 3D thanks to the pdf file you have attached. With the 3D model I make a comparison with my previous model in PDE form obtaining exactly the same result. Now I can also check if i did it correctly the 2D model.

Dear Jonathan,

Could you tell me how did you add the weak equation to your 3D model? or How did you make it work in a 3D model?
Thank you!

Ying

Sergei Yushanov Certified Consultant

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Ying,

To add ‘weak equation’ under the Electric Current node:

Right-click Electric Currents (ec) node and select More>Weak Contribution under domain features list (top section above the line)

Regards,
Sergei

Ying Yun Lim

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Dear Sergei,

It seems that there is no 'more' option when a 3D domain is used. Is there any other way to add the weak equations?
Thank you!

Ying

Sergei Yushanov Certified Consultant

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Ying,

To see "More/Weak Contribution" you have to activate “Advanced Physics Options” under “Show” button – see attached images.

Regards,
Sergei

Attachments:

Ying Yun Lim

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Dear Sergei,

Thank you very much! It's working now :D

Ying

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Dear Sergei,

I'd like to ask you some questions:

1)why in the 2D model you gave me you write:
(-k_ht*Tx+P*Jx_dc)*test(Tx)+(-k_ht*Ty+P*Jy_dc)*test(Ty) without volume integration?

2)where you have declared "P" ?

Attached you find my 2D axial symmetric model, there I can find only a heat release at one junction but no heat absorption at the other. Can you explain me where I am wrong and why?

Note that my file is more than 10 MB so I was forced to split it in 3 parts with winrar. Also i was forced to rename *.rar files as *.mph files so you have to rename the files in *.rar (otherwise the forum does not allow me to attach)

Hope this is not too boring for you...

Thank you soo much!

Attachments:

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Thank you so much Sergei, yesterday I managed to do it by myself in 2D axyal symmetric and 3D thanks to the pdf file you have attached. With the 3D model I make a comparison with my previous model in PDE form obtaining exactly the same result. Now I can also check if i did it correctly the 2D model.

Dear Jonathan,

Could you tell me how did you add the weak equation to your 3D model? or How did you make it work in a 3D model?
Thank you!

Ying

I'm Dario not Jonathan :D

This is the equation I put in my 3D model:

dVol_htgh*(S*T)*(-Jx_emdc*test(TSx_htgh)-Jy_emdc*test(TSy_htgh)-Jz_emdc*test(TSz_htgh))

please wait for Sergei validation

Sergei Yushanov Certified Consultant

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Dear Sergei,

I'd like to ask you some questions:

1)why in the 2D model you gave me you write:
(-k_ht*Tx+P*Jx_dc)*test(Tx)+(-k_ht*Ty+P*Jy_dc)*test(Ty) without volume integration?

2)where you have declared "P" ?

Attached you find my 2D axial symmetric model, there I can find only a heat release at one junction but no heat absorption at the other. Can you explain me where I am wrong and why?

Note that my file is more than 10 MB so I was forced to split it in 3 parts with winrar. Also i was forced to rename *.rar files as *.mph files so you have to rename the files in *.rar (otherwise the forum does not allow me to attach)

Hope this is not too boring for you...

Thank you soo much!

Dario,

1. I didn’t include volume integration because out-of-plane thickness was assumed to be one in that particular 2D example.

2. Peltier coefficient “P” is defined under Model Variables on the domains level.

3. I was not able to open your *.rar files.

Regards,
Sergei

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

ok thank you so much!
Therefore also in 2D axial symmetry I don't need to do so?

You believe that this equation is good as weak term for a 3D model?
dVol_htgh*(S*T)*(-Jx_emdc*test(TSx_htgh)-Jy_emdc*test(TSy_htgh)-Jz_emdc*test(TSz_htgh))
0

Sergei Yushanov Certified Consultant

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

ok thank you so much!
Therefore also in 2D axial symmetry I don't need to do so?

You believe that this equation is good as weak term for a 3D model?
dVol_htgh*(S*T)*(-Jx_emdc*test(TSx_htgh)-Jy_emdc*test(TSy_htgh)-Jz_emdc*test(TSz_htgh))
0

Dario,

1. No, 2D plane and 2D axial symmetry are different. Multiplier dVol is Jacobian of coordinate transformation. In 2D plane geometry it reduces to 1 and in 2D axial symmetry case it reduces to 2*pi*r.

I think, confusion here is due to different notations in 3.5 and 4.3. As far as I remember, in 3.5 multiplier dVol is always present in weak form notation (it was a while ago I use 3.5, though).

2. I believe, your weak term is OK for 3D model in version 3.5a.

Regards,
Sergei

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Thank you again Sergei, can you tell me if this equation is good for a 2d axial symmetry domain?

dVol_htgh*(S*T)*(Jr_emdc*test(TSr_htgh)+Jz_emdc*test(TSz_htgh))

dario

Sergei Yushanov Certified Consultant

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Thank you again Sergei, can you tell me if this equation is good for a 2d axial symmetry domain?

dVol_htgh*(S*T)*(Jr_emdc*test(TSr_htgh)+Jz_emdc*test(TSz_htgh))

dario

Dario,

Equation seems OK for 2D axial symmetry.

Regards,
Sergei

Susant Patra

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hi Sergei,

This is with reference to your model, Is there any modification to be done to the model if I want to do time dependent analysis instead of steady state?

Regards,
Susant

Sergei Yushanov Certified Consultant

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Susant,

It should work without any modifications in the weak contribution node. Just replace Stationary Step by Time Dependent Step. Then click Heat Transfer (ht) root node and make sure that Equation form is selected as “Time dependent”. Then click Electric Currents (ht) root node and make sure that Equation form is selected as “Stationary” (assuming that you are solving AC case).

Regards,
Sergei

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Thank you sergei.

Now i'll try to introduce also the generation effect in the 3D model

Susant Patra

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hi Sergei,

Thanks. In my application, I want to run heat transfer model in steady state and TEC model in transient mode to see change in temperature field after TEC is activated.

Regards,
Susant

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Dear Sergei I'm afraid that I need again your help.
I'm trying to introduce the the Seebeck effect in the 3D model, my equation is the following one:
dVol_emdc*(S*sigma)*((TSx_htgh+TSy_htgh+TSz_htgh)*test(Ex_emdc)+ (TSx_htgh+TSy_htgh+TSz_htgh)*test(Ey_emdc)+(TSx_htgh+TSy_htgh+TSz_htgh)*test(Ez_emdc))

this equation is accepted by Comsol (not in red)

I put at 273 K one end of the leg and at 473 the other end, one end is grounded, the other is at a floating potential V.

When solving there is no voltage generation and all the leg is at 0V

for the Peltier effect the model works if I put the respective equation in the field "dweack"

where It is wrong?
where is better to put the equation for the Seebeck?

Thank you

Dario

Sergei Yushanov Certified Consultant

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Dario,

Seebeck weak contribution is placed for current balance, "emdc" interface, in your case.
This term is shown in the attached image.

There are couple issues with your weak contribution expression:
1. According to Comsol sign convention, Seebeck weak term should be with “minus” sign.
2. Seems you are modeling material with isotropic electric conductivity: sigma_xx=sigma_yy=sigma_zz=sigma. Then, all off-diagonal terms for electric conductivity should be zero. Yet, those terms (containing sigma_xy, sigma_xz, etc.) are non-zero in your expression.

Regards,
Sergei

Attachments:

Seebeck_Weak_Contribution.png

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Thank you,

Yes I'm making the assumption that it is isotropic. So the right equation should simply be the one in the attached file?

Attachments:

Seebeck_Weak_ContributionMod.png

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

ok it works perfectly now

forgive me for my ingenuity

Sergei Yushanov Certified Consultant

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Dario,

Yes, this is the right equation. (Please correct my typo for the very last factor d(Vtest)/dy->d(Vtest/dz).

Regards,
Sergei

Dario Mastrolillo

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Yes, thank you

Can I ask you what sw you use to write down equation in a fast way?

Junce Zhang

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Dear Sergei,
I have a question for the peltier test model: So you will get T and V for the result. How can I get the current (I) in the p and n legs to calculate the power?
Thanks a lot!
Junce

Junce Zhang

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Dear Jonatan,
How did you get the current density? Now what I have is voltage but I need current density to calculate the power generation.
Thanks!
Junce

Jonatan Persson

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Are you using the physics builder guide method?

To meassure at boundrary, create a boundrary probe and use the insert or replace button in the expression section and find your identifier.Jx Jy or Jz from the drop down tree, choose your used physics then current density. To get current, change the type in the probe setting section to Integral. I write the expressions in variables with aproriate names then export the results to tables through global variable probes to get tables with nice headers.

Junce Zhang

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

Hi Jonatan,
Thanks for answering the question. I was using the example model, just combined T and V. But now I'm trying to build up the physics builder. But the guide is for 4.3 and I'm using 4.3b. There are several difference and it's not working. Could you share your physics builder file?
Thanks a lot.
Junce

Are you using the physics builder guide method?

To meassure at boundrary, create a boundrary probe and use the insert or replace button in the expression section and find your identifier.Jx Jy or Jz from the drop down tree, choose your used physics then current density. To get current, change the type in the probe setting section to Integral. I write the expressions in variables with aproriate names then export the results to tables through global variable probes to get tables with nice headers.

Jonatan Persson

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago

I think the only difference is that in 3.3b we use the features called generic feature, domain conidtion or boundrary condition instead of just feature as in the 3.3guide.
I used domain condition for the domain contition but genereic feature for the other condtions, where i choosed boundary in the supported geometric entity levels but you could use boundrary condtition directly, it should be the same thing.

Since the seebeck effect does not readily exist as a material parameter I have the material parameters deactivated and use user input instead as in the guide. Do not use material parameters in model mode but instead variables and for selected domains.

The deactivated weak form equations 3 and 4 is never used since for real materials the seebeck effeckt is always much lower then the conductivity.

I think the attached file will work in your version. It can be tricky to successfully get the interface to the physics area but i think its good to locate the physics builder manager from the view dropdown and press the register development files button in the upper right corner of the manager.

Attachments:

ThermoelectricEffect3.zip

Aria Abbasi

Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 7 years ago

Hi,

I know it is an old post but I was hoping you could help me out.
I'm new with Comsol and I'm trying to test the Seebeck coefficient in Copper. I simulated a of copper wire on a FR4 and set one side of the wire to 10 K and the other end of the wire at 400 K, one end is grounded, the other is not connected anywhere as I want to measure the open circuit voltage. I added the Seebeck coefficient of the copper to its list of the coefficient and put the right number for it. The problem is that I can't see any voltage. I wonder if you could help me with that.

Thanks!

Discussion Forum

Thermoelectric Effect - Adding the physics

Suggested Content