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.

short-pulsed heat source resolution

Please login with a confirmed email address before reporting spam

Hi everyone!

I need to model a short pulsed heat source. The heating-time is around 0.5ms. After that there is a delay time of around 0.1s until the next heating puls should occur. After that the delay is around 0.2s until the next heating puls should occur.
I use the piecewise definition: (start stop value)
0 0.0005 1
0.0005 0.1005 0
0.1005 0.1010 1
0.1010 0.3010 0
By using the periodic setting i get a continuous pulsing with two different alternating delay times.
No smoothing is used!!
For the solver to recognize the short pulses between the long delay times i defined 4 periodic events.
Solving steps are set to free, event-tolerance to 0.0001! As time steps the solver steps are used.

Now the problem: The solver "catches" the pulses very good and doesn't jump over any. But the pulses don't have the exact same length but vary in the range of several 0.0XXms!
As I use extreme high power for the heat source, the achieved temperature of my heated material varies in a range of several X0K :(

Any ideas of how can I get a better resolution of my pulses?

3 Replies Last Post Jun 8, 2015, 11:32 a.m. EDT
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: 9 years ago Jun 7, 2015, 12:41 p.m. EDT
Hi

I'll guess that with so few points over the cycle you are having severe aliasing effects. Just like for music sampling, if you need to catch the high frequency component of some nice music, you must use a 4-8 samples per highest frequency to get a reasonable recording.

With heat dissipation you must have a good resolution (I would use intermediate or strict stepping) over the pulse width (and use smooth pulses from the "Definition - Functions" section.
And then the time stepping at least 10 time steps over twice the pulse duration something like a power 2^*range(-1.0.5,7)*1E-3 decay until next pulse

And then I would follow carefully the solver log to see when the "out" marks come to ensure at least 10 of these per pulse and perhaps as many if not more between the pulses.

And as important check your mesh resolution around the region of the heat pulse, can you resolve reasonably the T gradient at these speeds in these regions, you will probably need a quite tight mesh (check the heat dissipation coefficient "ht.alphaTxx" = ht.kxx/ht.rho/ht.Cp for an isotropic material

--
Good luck
Ivar
Hi I'll guess that with so few points over the cycle you are having severe aliasing effects. Just like for music sampling, if you need to catch the high frequency component of some nice music, you must use a 4-8 samples per highest frequency to get a reasonable recording. With heat dissipation you must have a good resolution (I would use intermediate or strict stepping) over the pulse width (and use smooth pulses from the "Definition - Functions" section. And then the time stepping at least 10 time steps over twice the pulse duration something like a power 2^*range(-1.0.5,7)*1E-3 decay until next pulse And then I would follow carefully the solver log to see when the "out" marks come to ensure at least 10 of these per pulse and perhaps as many if not more between the pulses. And as important check your mesh resolution around the region of the heat pulse, can you resolve reasonably the T gradient at these speeds in these regions, you will probably need a quite tight mesh (check the heat dissipation coefficient "ht.alphaTxx" = ht.kxx/ht.rho/ht.Cp for an isotropic material -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jun 7, 2015, 1:58 p.m. EDT
Thx so far,
i'll test your advice and give a feedback.
Thx so far, i'll test your advice and give a feedback.

Walter Frei COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jun 8, 2015, 11:32 a.m. EDT
Dear Michael,

It sounds like you are following this blog: www.comsol.com/blogs/modeling-a-periodic-heat-load/
which would be the appropriate way to model what you are describing.

I would agree that, if you are correctly using the explicit events, you should see that the solver is taking a step exactly at the time of the events you specify.

Please post your model, or if you do not wish to post it publicly, send your model to the COMSOL Support Team so that we can investigate this, as it is not expected behavior, based upon your description. The other suggestions in this thread about mesh size, timestepping, and pulse smoothing are likely not the resolution of the issue that you are seeing.
Dear Michael, It sounds like you are following this blog: http://www.comsol.com/blogs/modeling-a-periodic-heat-load/ which would be the appropriate way to model what you are describing. I would agree that, if you are correctly using the explicit events, you should see that the solver is taking a step exactly at the time of the events you specify. Please post your model, or if you do not wish to post it publicly, send your model to the COMSOL Support Team so that we can investigate this, as it is not expected behavior, based upon your description. The other suggestions in this thread about mesh size, timestepping, and pulse smoothing are likely not the resolution of the issue that you are seeing.

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.