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.

Analytic function of Electric Conductivity doesn't work

Please login with a confirmed email address before reporting spam

Hello,

I've defined following functions as part of my material.

  • Density as a function of temperature
  • Electric Conductivity as a function of temperature
  • Thermal Conductivity as a functio of temperature

The analytic function of Electric Conductivity doesn't work. All other functions work separately or together - i.e. the model works if I type zero for "sigma". I've also tried to add the expression as a local variable and added it to the expression of my material property but it still doesn't work.

Strangely, the "sigma" can only accept the constant between zero and 2.7.

Can anyone help me to solve this problem?

Warm regards, Sungwoo



6 Replies Last Post Mar 23, 2018, 7:10 a.m. EDT
Nicolas Huc COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 6 years ago Mar 21, 2018, 9:47 a.m. EDT

Hi Sungwoo,

Thanks for sharing your experience and the model you have been working on.

I had a look at it. At a glance it seems that the set of material properties that are used corresponds to a solution with a very localized hot spot. And since the thermal conductivity gets lower and lower above a given threshold, the temperature gets hotter and hotter. This very sharp temperature profile cannot be captured using the current mesh.

To check the effect that I mentioned, I recommend to plot the temperature profile while solving, you'll see how the hot point appears. To do so, select the Fully Coupled 1 node in Time-Dependent Solver 1, then in the Result While Solving node enable Plot and select Temperature (ht).

I suggest either to use a (much) finer mesh for this model or to check the material properties to verify that they really correspond to want you want. Of course you are welcome to get in touch with the support for more feedback.

Best regards,

Nicolas

Hi Sungwoo, Thanks for sharing your experience and the model you have been working on. I had a look at it. At a glance it seems that the set of material properties that are used corresponds to a solution with a very localized hot spot. And since the thermal conductivity gets lower and lower above a given threshold, the temperature gets hotter and hotter. This very sharp temperature profile cannot be captured using the current mesh. To check the effect that I mentioned, I recommend to plot the temperature profile while solving, you'll see how the hot point appears. To do so, select the **Fully Coupled 1** node in **Time-Dependent Solver 1**, then in the **Result While Solving** node enable **Plot** and select **Temperature (ht)**. I suggest either to use a (much) finer mesh for this model or to check the material properties to verify that they really correspond to want you want. Of course you are welcome to get in touch with the support for more feedback. Best regards, Nicolas

Please login with a confirmed email address before reporting spam

Posted: 6 years ago Mar 21, 2018, 7:18 p.m. EDT

Hi Nicolas,

Thanks for the help.

I've tried with an 'Extremely Fine' mesh for the material, but still got a singularity error.

I've also limited the time before the max of thermal conductivity (less than 1500K) to avoid the sharp temperature but it still gives me an error. FYI, the current thermal conductivity works fine with a 'Normal' mesh if the Electric Conductivity is zero. In fact, the sigma accepts a value from zero to 2.7 but only zero value shows a proper result.

I double checked the expression (VTF equation) used for Electric Conductivity as a function of temperature and it seems alright.

I've tried the 'Result While Solving' as you suggested but I don't see a temperature plot windows while solves the model...? I've searched the forum and google to see other examples but can't find any examples. Do you have any example or video clip for this?

This question has already been logged in the comsol support and I've had a few responses but I still couldn't solve this problem yet. I've also set the expression as a variable but no success.

Thanks for your help.

Warm regards, Sungwoo

Hi Nicolas, Thanks for the help. I've tried with an 'Extremely Fine' mesh for the material, but still got a singularity error. I've also limited the time before the max of thermal conductivity (less than 1500K) to avoid the sharp temperature but it still gives me an error. FYI, the current thermal conductivity works fine with a 'Normal' mesh if the Electric Conductivity is zero. In fact, the sigma accepts a value from zero to 2.7 but only zero value shows a proper result. I double checked the expression (VTF equation) used for Electric Conductivity as a function of temperature and it seems alright. I've tried the 'Result While Solving' as you suggested but I don't see a temperature plot windows while solves the model...? I've searched the forum and google to see other examples but can't find any examples. Do you have any example or video clip for this? This question has already been logged in the comsol support and I've had a few responses but I still couldn't solve this problem yet. I've also set the expression as a variable but no success. Thanks for your help. Warm regards, Sungwoo

Nicolas Huc COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 6 years ago Mar 22, 2018, 12:20 p.m. EDT

Hi Sungwoo,

For the tests I made I edited the meshing sequence in order to have swepted mesh for the central part. This mesh should be better than the default tetrahedral mesh, and you can still control its size (in particular through the Maximum element size setting).

For the plot while solving have you edited the option from the Fully Coupled 1 node ? I just tryed it again and it was working ok.

You can send this question to the support and address it to me, I'll keep looking at it.

Best regards,

Nicolas

Hi Sungwoo, For the tests I made I edited the meshing sequence in order to have swepted mesh for the central part. This mesh should be better than the default tetrahedral mesh, and you can still control its size (in particular through the **Maximum element size** setting). For the plot while solving have you edited the option from the **Fully Coupled 1 node** ? I just tryed it again and it was working ok. You can send this question to the support and address it to me, I'll keep looking at it. Best regards, Nicolas

Please login with a confirmed email address before reporting spam

Posted: 6 years ago Mar 23, 2018, 3:58 a.m. EDT

Hi Nicolas,

Thanks for the tip. I guess you've edited my model but I don't see the attachment of your edition...?

I need to learn about the swept function as it has so many options, and will get back to you. I've already opened a case for this problem (case number is 2880501), so I will also update the progress in there.

Yes, I checked the plot option in the Fully Coupled 1 node. I've also checked other options in 'Step 2: Time Dependant' but nothing still no plot. I can only see the Convergence Plot 1 and 2. Strange... I used the model attached in this thread. Could you show me a screenshot how the plot window comes wihile solving please?

Warm regards, Sungwoo

Hi Nicolas, Thanks for the tip. I guess you've edited my model but I don't see the attachment of your edition...? I need to learn about the swept function as it has so many options, and will get back to you. I've already opened a case for this problem (case number is 2880501), so I will also update the progress in there. Yes, I checked the plot option in the Fully Coupled 1 node. I've also checked other options in 'Step 2: Time Dependant' but nothing still no plot. I can only see the Convergence Plot 1 and 2. Strange... I used the model attached in this thread. Could you show me a screenshot how the plot window comes wihile solving please? Warm regards, Sungwoo

Nicolas Huc COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 6 years ago Mar 23, 2018, 4:54 a.m. EDT

Dear Sungwoo,

I see, you need to make sure that the temperature plot refers to the solution you are computing (sol1 here) as shown on attached screenshot temperature_plot.PNG. Then in the Graphics tab I get the temperature plot (see plot_while_solving.PNG).

Best regards,

Nicolas

Dear Sungwoo, I see, you need to make sure that the temperature plot refers to the solution you are computing (sol1 here) as shown on attached screenshot temperature_plot.PNG. Then in the Graphics tab I get the temperature plot (see plot_while_solving.PNG). Best regards, Nicolas


Please login with a confirmed email address before reporting spam

Posted: 6 years ago Mar 23, 2018, 7:10 a.m. EDT

Dear Nicolas,

Thanks for the screenshot. Now I know what I should look. I was expecting to see a new plot graph window appears, and that's why I couldn't see the 'Result while solving' is working. ;)

I've monitored what happens while solving the model, and have found interesting things. The temperature sparks occasionally with below cases.

  1. Material is getting really hot with a hotspot - up to 4.056E81 x1.0E81 degC.
  2. The air domain is getting extremely hot - up to 3.456E250 x 1.0E250 degC. In this case, the hotspot in the material becomes coolest point. The temperature of the hotspot is minus here (see the screenshot).
  3. When the time reaches 209 seconds, the convergence value continuously grow as high as 2.6E12, then stop with a singularity error.

I've test a simple swept mesh with the material but it only increase the computation time and get the same error. I will try more options for swept now.

BTW, I see you are using the full simulation model but I use the half simulation model. I believe that there shouldn't be any differences between them.

Warm regards, Sungwoo

Dear Nicolas, Thanks for the screenshot. Now I know what I should look. I was expecting to see a new plot graph window appears, and that's why I couldn't see the 'Result while solving' is working. ;) I've monitored what happens while solving the model, and have found interesting things. The temperature sparks occasionally with below cases. 1. Material is getting really hot with a hotspot - up to 4.056E81 x1.0E81 degC. 2. The air domain is getting extremely hot - up to 3.456E250 x 1.0E250 degC. In this case, the hotspot in the material becomes coolest point. The temperature of the hotspot is minus here (see the screenshot). 3. When the time reaches 209 seconds, the convergence value continuously grow as high as 2.6E12, then stop with a singularity error. I've test a simple swept mesh with the material but it only increase the computation time and get the same error. I will try more options for swept now. BTW, I see you are using the full simulation model but I use the half simulation model. I believe that there shouldn't be any differences between them. Warm regards, Sungwoo

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.