Discussion Forum

Hi

I do not really understand your question, outside you "boundary" you cannot applly anything.

But you do not only have Dirichlet boudaries (fixed values), there are also the Neumann conditions that is you can act on the derivatives, which means that you are doing something like an extension of your boundaries.

If what you want to do is equivalent to define the slope ofthe variables along the boundary, then you must play with the full boundary constraints settings (including the boundary normals ...)

It is mostly essential to have a reasonnable mathematical expression of what one want to do, at least on paper side by side with the COMSOL equations

Have fun Comsoling
Ivar

Hi Ivar,

The model that i talked about is a DC model. For that, i apply Neumann conditions and Dirichlet conditions at boundaries.
For the conditions (V=0 when x=(+or-) infini), if my model is too small, the response is affected. So, i stretch the model and extend boundaries to apply these conditions. you can see an exemple in attached (i extend the model in x direction).

My question was: Can i apply these conditions at boundaries without stretch the model?

Because, i have another situation. It is a 3D model. When i choise to stretch the model, it increases the number of elements in the mesh. And it is time-consuming to run.
The model can run one day without a response.

Please Do you have a suggestion?


Linda.

Hi

unfortunatey I do not have the earth science modle so I cannot open your model in COMSOL, and its rather complex to read 500kbytes of raw Comsol instructions

If you are looking for electric fields in ACDC and you want to extend to infinity take a look at teh PLM (Perfectly matched layer elements, this reduces drastically the extend of yourmodel) There are already a couple of discussion about this subject on the forum

Have fun Comsoling
Ivar

Hi Ivar,

i look RF document which talks about PML in pages 43-48 and RF model document (pages 8-13).

On 'Physics', 'Subdomain settings', 'Infinite elements' , I check box 'stretched in x direction' in the cartesian type. because i want to extend model on this direction.

When i run the model after changing these, i have an error message 'Error:7043 Initial guess leads to undefined function value'.

What is the meaning of Sdx_guess_emdc (for DC modul) , S0x_guess_emdc (for AC modul), Sdy_guess_emdc (for AC modul), S0y_guess_emdc (for AC modul),? It is center point of.....?

What can be the meaning of this error message?






Thanks,

Linda.

Hi

indeed you should select the direction, as you state, I have never botherd about the varaibles and left COMSOL look after them, they are normally optimised internally to homogenise the solution.

one thing though read ccarefully the way to apply boundary condtions before and after the PML, something wrong there might overconstrain the PML, and give such errors

Hope this helps
Ivar