Discussion Forum

Hi

the volum is simply the domain integration of 1, if you select all you get the total volume, so the quiquest would be to use matlab to run a postint on all items of the integrant "1", no ?
--
Good luck
Ivar

Hi Ivar,

Firstly, thank you for your timely response.

I agree with you that the volume is the domain integration with integrand "1". I have a couple of questions with regards to your solution (which sounds good by implementation is the problem):

Your solution: Use matlab to run a postint on all items of the integrant "1".

Q1. I have connected to MATLAB and then have exported my FEM structure. Is this the correct first step?

Q2. I am assuming postint is short for "post-processing integrate", correct?

Q3. Do you know how COMSOL specifies "all [the subdomains]" in MATLAB?

Q4. I have my workspace (in Matlab) composed of the solved problem (in COMSOL 3.4)... can you help me with the nomenclature for the subdomain integration (with the integrand of "1") in Matlab for each subdomain?

Thanks!!!! (sorry got too caught up in the message)

Hi

Basically I agree its correct what you are sequencing. There is, typically for cases with a few items, in thew GUI the possibility to use an integration of

(dom=dest(dom))

instead of "1". But then you need to define al the destination items (possible even for 500 items).

Typically in V4 this would give you the list of the ID and the different volumes in a table easy to export to i.e. Excel

With matlab you can do it in any way around. The difficulty remains for me: how to identify your individual volumes with the ID numbner of COMSOL that is an increasing value, starting from (0,0,0) and going outwards

--
Good luck
Ivar

Hi Ivar,

Do you know how to specify "subdomain integrate" in the GUI post-processing subdomain tab? If so I can export this as a New Figure then produce an ascii file with all the data.

i.e. in the subdomain tab of the "plot parameters": display "[subdomain integrate](dom)" -----> my goal is for it to display the volumes for each subdomain visually, then export the ASCII file.


Basically I am asking how to write subdomain integration for the GUI.

Thanks,

Paul

Hi Ivar,

Success!!!! I was looking at this problem in the completely wrong context... instead, I let comsol+Matlab do the word.

The theoretical equations I was using was the Upper and Lower Wiener bounds, where:

W_L=postint(fem,'1')/postint(fem,'1/kap_esdl');

and

W_u=postint(fem,'1/kap_esdl')/postint(fem,'1');

This was the theoretical equations I needed to solve (much like the T_avg equation in COMSOL help documentation). This is perfect!

Steps to take for other users:

1. Connect COMSOL to MATLAB

2. Export FEM structure

3. Then type W_L and W_u in the command window in MATLAB


Thanks for all your input,

Paul Selvadurai

Hi Guys,

I have a complex 3D model with 22 subdomains. The model was built and meshed in Comsol 3.5a. It now exists as one complete geometry. I wish to extract one of the subdomains and save this as a separate geometry! Please advise the best way forward.

Thanks

Tiaan

Hi Tiaan,

I don't know if you have processed data that you would like to accompany this subdomain geometry but if you don't then this shouldn't be too hard.

1. I would save your 22 subdomain model as something else, for example "singular_subdomain.mph"

2. In this model I would go to the 'draw' mode. You should be able to modify all subdomain in this mode granted you haven't coerced them into a singular solid.

3. Now, in your newly saved model, physically highlight and substract the domains that you don't want (by pressing backspace when they are selected). Do this until there is only the subdomain you wish to keep.

Regards,

Paul

Hi Paul,

I tried this now, but no luck.

The problem is that although the model has 22 subdomains, comsol recognises these as one geometry (Geom1). It does not give me the option to "Use analyzed geometries" in Draw Mode. I'm not sure why it does not split them into 22 different geometries. The geometries were constructed in matlab, meshed in GMESH then imported int Comsol and saved as a .mph. You think the problem may lie in that comsol recognises GMESH output as one complete geometry?

Thanks

Tiaan

Also, what is your method for extracting the geometry on a domain basis post-processing? Comsol 4.0a gives the option to extract the (x,y,z) co-ordinates for each domain. The reconstruction thereof into an independent geometry, in Matlab, is the tricky part!

Thanks

Tiaan