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 The information provided may be out of date.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.

Integrating a step response in a time-dependent heat transfer study

Please login with a confirmed email address before reporting spam

Hello,
Currently I have a functioning stationary 2D heat transfer model which simulates the radial thermal distribution of a cylindrical conductor in light of all the relevant boundary heat sources.

I am now trying to simulate a transient response to a current step increase in order to assess the impact a short circuit could have on the radial thermal distribution of my conductor. I did this by adding a time-dependent study to my model and incorporating a step function. However, I get an error running it. The error message is "failed to evaluate variable jacobean".

I think I am having trouble incorporating the current step increase into my model but I also lack general familiarity with running time dependent studies on COMSOL. Do you have any suggestions? How do I link a step function with one of the parameters? I think I am getting an error because I did not do this properly.

Please let me know if you would like me to clarify some points.

Thanks,
Sonya

7 Replies Last Post 12 mars 2012, 11:34 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 28 nov. 2011, 06:25 UTC−5
Hi

Jacobian error comes mostly from a non derivable BC condition, are you using Bollean values ? or an abs() max() min () etc

--
Good luck
Ivar
Hi Jacobian error comes mostly from a non derivable BC condition, are you using Bollean values ? or an abs() max() min () etc -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 28 nov. 2011, 06:29 UTC−5
Thanks for your response.

All the boundary conditions are differentiable except for the one where I integrate the step function. I must be doing that wrongly. My step function is called step1 and I simply added it to my boundary condition. How else can I link my step function to a parameter in my model?
Thanks for your response. All the boundary conditions are differentiable except for the one where I integrate the step function. I must be doing that wrongly. My step function is called step1 and I simply added it to my boundary condition. How else can I link my step function to a parameter in my model?

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 28 nov. 2011, 11:13 UTC−5
Hi

your step function has a transition region, of a certain width, you should ensure that COMSOl is considering this width and that you force it to take a few steps along this transition, otherwise it appears like a Dirac step and the derivative estimator diverges. Check your stepping (time or Paramatric and see if you hit the transition region(s), and correct accordingly, by adding a few points. Note time stepping if in "automatic" might skip your values, then an "intermediate" stepping sequence would be more appropriate (or a strict)

--
Good luck
Ivar
Hi your step function has a transition region, of a certain width, you should ensure that COMSOl is considering this width and that you force it to take a few steps along this transition, otherwise it appears like a Dirac step and the derivative estimator diverges. Check your stepping (time or Paramatric and see if you hit the transition region(s), and correct accordingly, by adding a few points. Note time stepping if in "automatic" might skip your values, then an "intermediate" stepping sequence would be more appropriate (or a strict) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 16 févr. 2012, 09:01 UTC−5
Hi,
I think I have the same problem : I want to have one parameter (the temperature, in Global definitions) evolving during the time (as a step, or sinusoidal and with a step)... since it works with variables (for example : Name Text, Expression Tmoy+deltaT*sin(w*t), with Tmoy, deltaT and w defined in Parameters).
i tried with the function Step, and I copied the function name in Convective cooling>External temperature...or in the previous expression to replace Tmoy...
it appears the same error as the previous user !
however i didn't really understand your last explanation ...
Hi, I think I have the same problem : I want to have one parameter (the temperature, in Global definitions) evolving during the time (as a step, or sinusoidal and with a step)... since it works with variables (for example : Name Text, Expression Tmoy+deltaT*sin(w*t), with Tmoy, deltaT and w defined in Parameters). i tried with the function Step, and I copied the function name in Convective cooling>External temperature...or in the previous expression to replace Tmoy... it appears the same error as the previous user ! however i didn't really understand your last explanation ...

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 16 févr. 2012, 17:38 UTC−5
Hi

if you define one or more steps in the middle of your stepping range, you should reduce the steps you define (and ask for intermediate or strict stepping, NOT free) to help the solver "see" the steps, otherwise the internal stepping algorith might skip the step and ignore it. The defaut solver is set up for typically exp type assymptotical models

I often use 2^{range(-3,1,4)} type time sequences for thermal cases, but with oscillating BCs a linear variable step is better

Do not foret to check the mesh density, material heat diffusivity and time minimum steps, these must match to avoid numerical issues (check the Forum)

--
Good luck
Ivar
Hi if you define one or more steps in the middle of your stepping range, you should reduce the steps you define (and ask for intermediate or strict stepping, NOT free) to help the solver "see" the steps, otherwise the internal stepping algorith might skip the step and ignore it. The defaut solver is set up for typically exp type assymptotical models I often use 2^{range(-3,1,4)} type time sequences for thermal cases, but with oscillating BCs a linear variable step is better Do not foret to check the mesh density, material heat diffusivity and time minimum steps, these must match to avoid numerical issues (check the Forum) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 12 mars 2012, 06:28 UTC−4
Hi,
I also have the same problem, I'm trying to simulate a 2D system with varying BC. I define a function step 1 that i try to associate to a heat source but i get this error
Syntax error in expression
-Expression: (0)+([W/m^3]E1)
-Subexpression:([W/m^3]E1)
-Position:5
Error in multiphysics compilation.
Error in automatic sequence generation.




Thanks
Kamelia
Hi, I also have the same problem, I'm trying to simulate a 2D system with varying BC. I define a function step 1 that i try to associate to a heat source but i get this error Syntax error in expression -Expression: (0)+([W/m^3]E1) -Subexpression:([W/m^3]E1) -Position:5 Error in multiphysics compilation. Error in automatic sequence generation. Thanks Kamelia

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 12 mars 2012, 11:34 UTC−4
Hi

that looks like a typo, you should use "number (exponent) [units]" in this order, and abuse a little with the () 's to be sure COMSOL is getting the equations right.

I often copy my formulas into a temporary parameter to check the resulting value AND units, as forsome older version 1[Hz] = 1/1[s] is not the same as 1[Hz] = 1/(1/[s])

--
Good luck
Ivar
Hi that looks like a typo, you should use "number (exponent) [units]" in this order, and abuse a little with the () 's to be sure COMSOL is getting the equations right. I often copy my formulas into a temporary parameter to check the resulting value AND units, as forsome older version 1[Hz] = 1/1[s] is not the same as 1[Hz] = 1/(1/[s]) -- 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.