Discussion Forum

Hi

pls add at least the version and if possible the physics tag to allow for us to better understand your case, it's difficult to answer just like that

--
Good luck
Ivar

Hi Ivar.

It's COMSOL 4.3 (updated, original file in 4.2a) and the Heat Transfer Module.
Thanks a lot for getting back to me!!

best,
Gustav

Hi

indeed I'm discovering too that these values seem to have changed, there is something in the release notes too. If you check in the Equation field, you will see that you have two options in V4.3

either you can use "material.k11" or you use the "ht.kxx" variable name.

Before "k" scalar existed, today it's mapped to the diagonal of the tensor k11 k22 k33, but is is then again mapped into ht.kxx once you launche HT

My turn for a question: have you discovered if thehet diffusivity is defined in the new 4.3 ?, if so it should be defined as a tensor too now

--
Good luck
Ivar?

Hey.

Perhaps I did not formulate my question well enough.
I do not have a problem with the tensorial nature, but with the plotting of two thermal conductivity functions as a single graph. I thought I had the right command/approach, but certainly do not.

I have an isotropic case (so kscalar = k11=k22=k33, where 1=x, 2=y, 3=z directions) and two materials, A & B, i.e. matA and matB (in COMSOL A & B are numbers, A=1 & B=2, but I will use letters to avoid confusion) with the corresponding thermal conductivity functions kA(T) & kB(T). kA(T) & kB(T) differ by a factor ~10.

So when plotting k(T) across my sample, I know there must be a step of approx. 10 at the interface between the two materials (when switching from kA(T) to kB(T)).

Now the two commands, “material.k11” and “ht.kxx”, should (if I understand the user guide correctly) give me the k(T), i.e. kA(T) when the material is A, and kB(T) if the material is B. However, it certainly does not. I am missing the step of approx. 10, and the values are 2 orders of magnitude too low.

The commands “matA.def.k11” & “matB.def.k11”, giving me kA(T) and kB(T) respectively, regardless of the material assigned to the different intervals of my line graph (i.e. “matA.def.k11” plots me kA(T), even if the material is B), work fine and yield the right values and qualitative behavior. (So apparently the functions are not ill defined)

Now my question is: How can I plot a single k(T) graph, that is kA(T) on intervals where the material is A, and kB(T) on intervals where material B is assigned to? And since the commands “material.k11” and “ht.kxx” do not yield this, what do the yield? Can anyone give me a physical or mathematical interpretation of the numbers the program throws at me?

Of course I could do line plots for each material interval and then use photoshop and imageJ to scale and add them together, but as I do a parameter sweep, this implies an exploding workload and I am sure COMSOL has thought about this before (e.g. user guide p. 392).

I hope this made my problem a bit clearer, and that I find someone who has some good advice or input.

I have not found the heat diffusivity, but also I am not quite sure why I should look at it. I would have to define Cp(T) and roh(T) and multiply it to the diffusivity. So instead of defining two functions and plotting them, I have to define four functions and multiply my graphical result by them, that is, if the same problem does not occur again. So even if I find it, the workload will probably be more than that of the Photoshop approach (I do not have that many sweep steps and material intervals).

Nonetheless, where did you find it? It is not in the drop down list for “expression” (I assume you mean that by equation field). Or is there a “ht.diffusivity” command (this spelling is not recognized)?

Thanks a lot for your help,
Kind regards,
Gustav

Hi

carefull, are you in 4.2 or 4.3, I fear the naming has changed (not sure, I have only 4.3 up just now)

Try to use the COMSOL convention it would be easier to follow you. And turn on the equation view to check the COMSOL equationns

You can also define a single variable that changes arguments in the different domains to combine my_k for domains A and B.

but the use of material.k11 and ht.kxx, as this did work perfectly for me in 4.3.

the matA.def.k11 in your notation is not giving you the mapping vi the domain, but you access material A and B, but do not know where they apply, therefore I prefer the two above, if you change the domain id mapping the two above remain correct, while the matA.def.k11 will be static

--
Good luck
Ivar

is it possible to perform this task in 3.5a? or at any rate, is it possible to extract the values (with respect to another variable, time or distance) into a text format?

Hi

I would say normally yes, there is always a way to dump results (or material data extracted as "results" into a txt file, but I cannot remember the quick way any longer for 3.5, sorry that is getting to vague in my mind, have been using v4 to much I understand ...

But there are still quite a few 3.5a users out there they should be able to help

--
Good luck
Ivar

Hello.

Perhaps you can represent something like "nitf.kxxht", where nitf is the physics tag.

Good luck,

Jesus.

Hello, Nishant,

In COMSOL 3.5 there is something similar: kmean_htgh, named "mean effective thermal conductivity".

Bye,

Jesus.