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 The information provided may be out of date.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.

non uniform heat flux

Anastassios Mavrokefalos

Please login with a confirmed email address before reporting spam

Hello all,

I'm trying to model simple 2D conduction model in a slab. What I want to do is use a non-uniform heat flux for a boundary condition. I already have the heat transfer coefficient distribution along that boundary, which is a gaussian type distribution. Instead of using a gaussian distribution I would prefer to actually call a file that will assign the values of the heat transfer coefficient along the boundary. Do you know if that is doable?

Thanks,
Anastassis

8 Replies Last Post 28 feb 2013, 02:35 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 8 mag 2012, 01:45 GMT-4
Hi

In matlab certainly, else you need to load a parametric function with your data and run it like that

--
Good luck
Ivar
Hi In matlab certainly, else you need to load a parametric function with your data and run it like that -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2 nov 2012, 07:10 GMT-4
Hello sir,

How can apply Gaussian heat flux distribution to my model? whether MATLAB required for it? Please help...


Thank You.

Sagar
Hello sir, How can apply Gaussian heat flux distribution to my model? whether MATLAB required for it? Please help... Thank You. Sagar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2 nov 2012, 08:53 GMT-4
Hello, Sagar,

I think that you can use simply the Gaussian Pulse function available in COMSOL (in Global Defin. > Functions, or in Model > Defin. > Functions).
1. Define the function as you need (location, standard dev.)
2. Apply it where you need it, for instance 1000[W/m^2]*gp1(x[1/m]) as a Heat Flux condition.

Jesus.
Hello, Sagar, I think that you can use simply the Gaussian Pulse function available in COMSOL (in Global Defin. > Functions, or in Model > Defin. > Functions). 1. Define the function as you need (location, standard dev.) 2. Apply it where you need it, for instance 1000[W/m^2]*gp1(x[1/m]) as a Heat Flux condition. Jesus.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 3 nov 2012, 02:10 GMT-4
Hi Jesus,

Thanks a lot for your reply. I applied Gaussian heat flux as per your suggestion.

I am very new at COMSOL. Will you please tell me significance of 'x' in gp1(x[1/m]), because my out put is changing with change in 'x'.

Thank You.


Sagar
Hi Jesus, Thanks a lot for your reply. I applied Gaussian heat flux as per your suggestion. I am very new at COMSOL. Will you please tell me significance of 'x' in gp1(x[1/m]), because my out put is changing with change in 'x'. Thank You. Sagar

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 3 nov 2012, 08:22 GMT-4
Hi

check the help regularly it's also for that ;)
In general when you define a fuction you give a name for the argument (somtimes COMSOL fixes it i.e. to "t") BUT this name is arbitrary and only there for the definition, the variable you pass is the one you use in the expression.

Then now in V4.3 often you can define unts for the expression and the arguments. If these are NOT defined you should always send a "unitless" value to the function.

so if you define a gaussian function gp1(my_arg), when you call it you could wrtite gp1(x[1/m]) where the x[1/m] is the value of the coordinate "x" expressed unitless, in values of 1[m] x[1/mm] is also unitless but the value is 1000 =1[m/mm] times greater than the first example.

Now where is "x" taken from ? THat is another default behaviour of COMSOL, not necesarily obvious or trivial, when you are a newcomer: x is taken from the selection of entities you have choosen, it states that for each position along the entity (Domain or boundary) you have selected the variable at the position "x" will take the value of gp(x) in the desired units.

Hope I'm clear, and it can be understood ;)
Try it ou on some simple geometry, easy to interprete, first

--
Good luck
Ivar
Hi check the help regularly it's also for that ;) In general when you define a fuction you give a name for the argument (somtimes COMSOL fixes it i.e. to "t") BUT this name is arbitrary and only there for the definition, the variable you pass is the one you use in the expression. Then now in V4.3 often you can define unts for the expression and the arguments. If these are NOT defined you should always send a "unitless" value to the function. so if you define a gaussian function gp1(my_arg), when you call it you could wrtite gp1(x[1/m]) where the x[1/m] is the value of the coordinate "x" expressed unitless, in values of 1[m] x[1/mm] is also unitless but the value is 1000 =1[m/mm] times greater than the first example. Now where is "x" taken from ? THat is another default behaviour of COMSOL, not necesarily obvious or trivial, when you are a newcomer: x is taken from the selection of entities you have choosen, it states that for each position along the entity (Domain or boundary) you have selected the variable at the position "x" will take the value of gp(x) in the desired units. Hope I'm clear, and it can be understood ;) Try it ou on some simple geometry, easy to interprete, first -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 4 nov 2012, 09:20 GMT-5
thanks alot
thanks alot

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 28 feb 2013, 02:28 GMT-5
Sir, we are using comsol 4 version for simulation and we want to perform thermal stress analysis but it gives following problems
1. Temperature plot does give consistent results [ less than ambient temp. or negative values]
2. We want to incorporate temperature dependent thermal properties but for silicon material some properties are missing like youngs modulus, possions ratio and thermal expansion coefficient
3. Is silicon material has built in temperature dependent properties or it has to insert manually, if manually it has to do then give us detail procedure for that.
4. We are also facing problem in meshing [ as we decrease the mesh size we get errors]

Sir, we are using comsol 4 version for simulation and we want to perform thermal stress analysis but it gives following problems 1. Temperature plot does give consistent results [ less than ambient temp. or negative values] 2. We want to incorporate temperature dependent thermal properties but for silicon material some properties are missing like youngs modulus, possions ratio and thermal expansion coefficient 3. Is silicon material has built in temperature dependent properties or it has to insert manually, if manually it has to do then give us detail procedure for that. 4. We are also facing problem in meshing [ as we decrease the mesh size we get errors]


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 feb 2013, 02:35 GMT-5
Hi

you need to study further your basic homework first !
1) your mesh is too coarse, look at your gradients and compare it to the mesh you are not resolving anything like that
2) what is your alpha[mm^2/s) check the forum, the heat diffusivity, the mesh size and the step time for transient diffusion problems are linked and some relation MUST be respected, else you will get all wrong results
3) take care when you plot high gradients on a low resolution mesh, COMSOL interpolates the results, but this can, in such step wise cases give overshoots (hence our negative T), this is only postprocessing numerics issues, still you have also solving numerical issues with that mesh discretization, the mesh values can be obtained if you check the mesh grid and turn off the plot refinement

Consider the mesh as the digital sampling of music, try sampling at 50Hz, your nice concert will become as noisy as a mains transformer on the corner of the street, it's the same for FEM ...

But I'm sure you will fix this rapidly ;)

--
Good luck
Ivar
Hi you need to study further your basic homework first ! 1) your mesh is too coarse, look at your gradients and compare it to the mesh you are not resolving anything like that 2) what is your alpha[mm^2/s) check the forum, the heat diffusivity, the mesh size and the step time for transient diffusion problems are linked and some relation MUST be respected, else you will get all wrong results 3) take care when you plot high gradients on a low resolution mesh, COMSOL interpolates the results, but this can, in such step wise cases give overshoots (hence our negative T), this is only postprocessing numerics issues, still you have also solving numerical issues with that mesh discretization, the mesh values can be obtained if you check the mesh grid and turn off the plot refinement Consider the mesh as the digital sampling of music, try sampling at 50Hz, your nice concert will become as noisy as a mains transformer on the corner of the street, it's the same for FEM ... But I'm sure you will fix this rapidly ;) -- 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.