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 The information provided may be out of date.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.

pulse and heat flow

Please login with a confirmed email address before reporting spam

Dear Sir

I need urgent help !!!

i have to apply pulse train voltage with( 13kv and 1khz) on cable termination to get its effect on temp. after 3 hours.

i can not solve by time dependent solver (take very long time)

i can solve for 10 pluses and get the heat produced Q(W/m3) after 10 pluses then i can get the total energy produced by multiply (Q* (time of 10 pulses =0.01 sec). afterthat , how can i use the value of Q or the value of the energy to get the temperature after 4 hours



12 Replies Last Post 06.03.2013, 01:16 GMT-5
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 14.03.2012, 04:04 GMT-4
Hello

first of all do not forget that here on the Users Forum we are all users of COMSOL with our own work and project to get through, even if we all do what we can for other Comsol users, but we are not from / hired by COMSOL, nor can we garanty any "service" (including me ;)

My way would relly be, as I understand you have a capacitive coupling and not really a joule heating, since your material have a very low conductance, to model in (time) detail one single pulse, then to integrate out the average energy per domain and for one period, and than apply this average value in a constant " heat source way" for a long time (study 2).

This is basically what a frequency-transient or frequency-stationary study does but its less visible and less controllable. But a frequency transient is for a sinus pulse and you have a slightly different one, with a steeper rise time hence probably also a different coupling

And do not forget that you are not seeing any joule heating in the copper the way you have defined the terminal, as you have no real current flowing along the copper line, something that would also heat up your system.

Finally, I would advise to use a much finer mesh as you need to get some better resolution in the "hot region" that looks very much to me as a slight singularity region (upper edge of your domain 3 ) semi screen


--
Good luck
Ivar
Hello first of all do not forget that here on the Users Forum we are all users of COMSOL with our own work and project to get through, even if we all do what we can for other Comsol users, but we are not from / hired by COMSOL, nor can we garanty any "service" (including me ;) My way would relly be, as I understand you have a capacitive coupling and not really a joule heating, since your material have a very low conductance, to model in (time) detail one single pulse, then to integrate out the average energy per domain and for one period, and than apply this average value in a constant " heat source way" for a long time (study 2). This is basically what a frequency-transient or frequency-stationary study does but its less visible and less controllable. But a frequency transient is for a sinus pulse and you have a slightly different one, with a steeper rise time hence probably also a different coupling And do not forget that you are not seeing any joule heating in the copper the way you have defined the terminal, as you have no real current flowing along the copper line, something that would also heat up your system. Finally, I would advise to use a much finer mesh as you need to get some better resolution in the "hot region" that looks very much to me as a slight singularity region (upper edge of your domain 3 ) semi screen -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 15.03.2012, 05:38 GMT-4
Dear Sir

first of all i want to aplogize because of disturbing you (actually i have thought that you are hired in Comsol), i don't think that you are Comsol user beacuse i saw your name in almost all answers of comsol questions. (i am so sorry for this confusion).

i have a question as well!!!

i solved my model in the time rang (0,0.0001,0.1) and get the heat source in the model

i draw the heat (Q) with time in a spesific point in the model, how can i get the average value of Q for this point?

and can i get the average value of Q for all elements in specified domain


thank you , sorry for disturb
Dear Sir first of all i want to aplogize because of disturbing you (actually i have thought that you are hired in Comsol), i don't think that you are Comsol user beacuse i saw your name in almost all answers of comsol questions. (i am so sorry for this confusion). i have a question as well!!! i solved my model in the time rang (0,0.0001,0.1) and get the heat source in the model i draw the heat (Q) with time in a spesific point in the model, how can i get the average value of Q for this point? and can i get the average value of Q for all elements in specified domain thank you , sorry for disturb

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 19.03.2012, 03:39 GMT-4
Hi

no problem, it's just that I'm too busy these days I have no time for the Forum ;)

if you set up a time step over ONE period, with 10-20 points or more then you can use the Results Derived Values and you show the port voltage and current. You can then (as last tab for the derived value definition add an operation on the variable table such as average or sum or ..., at least in my 4.2a version) this will give you the average current and voltage over the ONE period, that you can then apply for a constant time series.

But when I run you model I see an highly oscillating current while I expected something smoother. This can be due to the extremely large variation of resisitvity you are using. The number theory of binary number say one cannot solve with todays computer any variation much above 1:1E6 to 1:1E8 and you have >1:1E22 (sigma from 6E7 to 1E-16)

This will lead to numerical errors and could explain the strange current oscillations. This is not a COMSOL limitation, but from the binary number theory.
The only way around is to more carefully check how the different matrerial sigma values influences the result, and perhaps try some manual scaling (check the doc), as your model case is far from trivial.
As you screen the copper part with the port definition on the boundary

--
Good luck
Ivar
Hi no problem, it's just that I'm too busy these days I have no time for the Forum ;) if you set up a time step over ONE period, with 10-20 points or more then you can use the Results Derived Values and you show the port voltage and current. You can then (as last tab for the derived value definition add an operation on the variable table such as average or sum or ..., at least in my 4.2a version) this will give you the average current and voltage over the ONE period, that you can then apply for a constant time series. But when I run you model I see an highly oscillating current while I expected something smoother. This can be due to the extremely large variation of resisitvity you are using. The number theory of binary number say one cannot solve with todays computer any variation much above 1:1E6 to 1:1E8 and you have >1:1E22 (sigma from 6E7 to 1E-16) This will lead to numerical errors and could explain the strange current oscillations. This is not a COMSOL limitation, but from the binary number theory. The only way around is to more carefully check how the different matrerial sigma values influences the result, and perhaps try some manual scaling (check the doc), as your model case is far from trivial. As you screen the copper part with the port definition on the boundary -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 22.05.2012, 08:58 GMT-4
Dear Ivar,

I want to define a scalar expresion to be used as my boundary condition. Eventually I want to solve a heat transfer problem with pulsed laser passing on the surface. So I need to apply heat flux in a pulsed manner.

To define my heat flux, I want to somehow use the pulse wave function, that you can find here;
en.wikipedia.org/wiki/Pulse_wave

But I don't know how I can write this equation in Comsol. I am searching all documantation to see how to write Signa (
summation) in Comsol, but no luck.

Will you please help me to write the equation in the wikipedia link in comsol?
Thanks,
Negar
Dear Ivar, I want to define a scalar expresion to be used as my boundary condition. Eventually I want to solve a heat transfer problem with pulsed laser passing on the surface. So I need to apply heat flux in a pulsed manner. To define my heat flux, I want to somehow use the pulse wave function, that you can find here; http://en.wikipedia.org/wiki/Pulse_wave But I don't know how I can write this equation in Comsol. I am searching all documantation to see how to write Signa ( summation) in Comsol, but no luck. Will you please help me to write the equation in the wikipedia link in comsol? Thanks, Negar

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 22.05.2012, 11:25 GMT-4
Hi

there has been several discusssion around pulsing and how to make laser beams move around, there are also example(s) in the model library, try them out and learn from them.

But many short pulses with long waiting time is tedious to set up and to solve, consider carefully how your pulses will behave: cumulative or independent, start solving only one pulse, then two

particularly HT with short pulses can be tedious to mesh and to solve correctly, because of the high gradients and low diffusivity of materials, and oftehn the scaling effects due to the small size of the laser beam

--
Good luck
Ivar
Hi there has been several discusssion around pulsing and how to make laser beams move around, there are also example(s) in the model library, try them out and learn from them. But many short pulses with long waiting time is tedious to set up and to solve, consider carefully how your pulses will behave: cumulative or independent, start solving only one pulse, then two particularly HT with short pulses can be tedious to mesh and to solve correctly, because of the high gradients and low diffusivity of materials, and oftehn the scaling effects due to the small size of the laser beam -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 22.05.2012, 22:58 GMT-4
Hello Ivar,

Thanks for your response.

I’ve seen some of the previous discussions on how to define pulsed laser. I believe people use Heaviside function.
I thought maybe for my design using the mentioned wave equation makes things easier, so I don’t need to use Heaviside function.
Don’t you suggest using this function? Do you thing I will have problem in using this equation?
Will you please help me to write this equation in Comsol anyway? Is there any operation for Sigma in Comsol?

Thank you so much!!
Negar
Hello Ivar, Thanks for your response. I’ve seen some of the previous discussions on how to define pulsed laser. I believe people use Heaviside function. I thought maybe for my design using the mentioned wave equation makes things easier, so I don’t need to use Heaviside function. Don’t you suggest using this function? Do you thing I will have problem in using this equation? Will you please help me to write this equation in Comsol anyway? Is there any operation for Sigma in Comsol? Thank you so much!! Negar

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 23.05.2012, 00:56 GMT-4
Hi

if you use a square or dirac type, the solver will never manage to converge, as for most physical math cases, the numerical solvers operate successfully mainly on derivable functions if possible derivable twice, so choose rather gaussain or the internal smoothed "step" functions (v4) or the heavyside functions in 3.5

--
Good luck
Ivar
Hi if you use a square or dirac type, the solver will never manage to converge, as for most physical math cases, the numerical solvers operate successfully mainly on derivable functions if possible derivable twice, so choose rather gaussain or the internal smoothed "step" functions (v4) or the heavyside functions in 3.5 -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23.05.2012, 00:59 GMT-4
Thank you so much Ivar. I will try Heavyside function, and let you know if I had any question!
Thanks!
Thank you so much Ivar. I will try Heavyside function, and let you know if I had any question! Thanks!

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23.05.2012, 10:26 GMT-4
Hello Ivar,
Since the pulse duration is much smaller than the time between two pulses, is there any way to skip the time between each pulses, or uses a kind of adaptive time stepping?
Thanks,
Negar
Hello Ivar, Since the pulse duration is much smaller than the time between two pulses, is there any way to skip the time between each pulses, or uses a kind of adaptive time stepping? Thanks, Negar

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 23.05.2012, 10:54 GMT-4
Hi

the time stepping must be set up manually by a series and for the type of pulses you have it's tedious. But indeed you should have many steps during the rise and fall time and fewer inbetween, even following a power rule if you have decay phenomena in between your pulses to reduce the number of time steps.
Furthermore, deep down in the time solver nodes, you should adjust the time stepping from "free" to "strict" or "intermediate", otherwise the solver will definitively skip your 2nd and next pulses. Sorry that was for v4, in v3.5 there is too the time stepping settings of the solver, but I have forgotten where they hide by now, the terms was perhaps "automatic" instead of "free", but still change to "strict" or "interemediate"

--
Good luck
Ivar
Hi the time stepping must be set up manually by a series and for the type of pulses you have it's tedious. But indeed you should have many steps during the rise and fall time and fewer inbetween, even following a power rule if you have decay phenomena in between your pulses to reduce the number of time steps. Furthermore, deep down in the time solver nodes, you should adjust the time stepping from "free" to "strict" or "intermediate", otherwise the solver will definitively skip your 2nd and next pulses. Sorry that was for v4, in v3.5 there is too the time stepping settings of the solver, but I have forgotten where they hide by now, the terms was perhaps "automatic" instead of "free", but still change to "strict" or "interemediate" -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 05.03.2013, 12:07 GMT-5
haii

can you give me sample how to make analys for current in underground cable because an effect of temperature ground (earth)
i get confused how to start it
haii can you give me sample how to make analys for current in underground cable because an effect of temperature ground (earth) i get confused how to start it

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 06.03.2013, 01:16 GMT-5
HI

You need to be a bit more verbose to describe your model.
such as: What geometry (a concentric cylinder in a large block ?)
what physics, or rather first which are the (driven) dependent variables (T and ..)
which are the driving variables (electric current , ...?)
which are the initial conditions (currents, temperatures ...)
solver type ? time dependent or steady state (frequency domain ?
...

--
Good luck
Ivar
HI You need to be a bit more verbose to describe your model. such as: What geometry (a concentric cylinder in a large block ?) what physics, or rather first which are the (driven) dependent variables (T and ..) which are the driving variables (electric current , ...?) which are the initial conditions (currents, temperatures ...) solver type ? time dependent or steady state (frequency domain ? ... -- 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.