Time-dependent heat transfer problem - no solution

Please login with a confirmed email address before reporting spam

Hello everyone,

I have a time-dependent problem in which there is a wall with a uniform temperature at time zero equl to 16 °C, after that, I apply a temperature pulse following a triangular temperature profile at one side, while the opposite side is kept at a constant temperature (equal to 16 °C). The triangular profile is defined with a piecewise linear function in the Definition section. The aim is to determine the heat flux (expressed in W/m2) at the opposite side of the solicitation. Usually, I get a solution in which the heat flux at the opposite side is zero at the beginning, then it follows a function approximately as e^(-ln(t^2)). The solver I am adopting is the Runge-Kutta one. Absolute tolerance is equal to 10^-5 and relative tolerance 10^-4.

So far there is no problem. I always get a solution. However, when I change temperature profile, that is, no more piecewise function but a function with a noise, I do not get any solution anymore. Instead of having a heat flux following e^(-ln(t^2)) I get weird trends.

Is there anyone that can help me? Should I configure the solver in a aspecific way since the boundary conditions is noisy?

I attach the COMSOL file at the external .txt file with the noisy triangular profile of the temperature.

Thanks for the help.



3 Replies Last Post Feb 26, 2024, 9:21 a.m. EST
Edgar J. Kaiser Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 2 months ago Feb 21, 2024, 1:20 p.m. EST

Maja,

noise can be tricky in a time dependent study. You may need to use manual time stepping that resolves the temporal characteristics of the noise function.

Cheers Edgar

-------------------
Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
Maja, noise can be tricky in a time dependent study. You may need to use manual time stepping that resolves the temporal characteristics of the noise function. Cheers Edgar

Dave Greve Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 2 months ago Feb 22, 2024, 7:05 p.m. EST
Updated: 2 months ago Feb 22, 2024, 6:59 p.m. EST

I would expect difficulty with such a calculation unless (1) the noise was band-limited and (2) fixed time stepping was used with a time step small compared to the reciprocal of the roll-off frequency.

I would expect difficulty with such a calculation unless (1) the noise was band-limited and (2) fixed time stepping was used with a time step small compared to the reciprocal of the roll-off frequency.

Please login with a confirmed email address before reporting spam

Posted: 2 months ago Feb 26, 2024, 9:21 a.m. EST

Thanks a lot to both of you. I appreciate your answers, they helped me to understand better the problem.

Do you know by chance if it is possible to generate a random signal with variable amplitude that I can sum to the "clean" triangular temperature profile in a sweep analysis?

Thanks a lot to both of you. I appreciate your answers, they helped me to understand better the problem. Do you know by chance if it is possible to generate a random signal with variable amplitude that I can sum to the "clean" triangular temperature profile in a sweep analysis?

Reply

Please read the discussion forum rules before posting.

Please log in to post a reply.

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.