Discussion Forum

suppose you are in 2D. A domain is 2D, an edge is 1D (a point is 0D).
A source in a domain is distributed over the elements belonging to this domain.
A source in an edge is distributed over the elements belonging to the edge only.
Mathematically (related to the weak form of the equation and the finite element method), the integral of the source times the test function is done over 2D for a domain source, over 1D for an edge source, or added to a single point (0D) if this is a point source.
So it is essentially a matter of location of the source : which elements are impacted by the source?

I hope this helps,
Eric Favre
Comsol France

Hello,

thank you for your help M. Fabre.
I try to simulate a 3D cylinder source with the Helmholtz equation.
When I define an edge source on only one edge of this cylinder, the results of my study seem to be correct.
However, when I define the whole cylinder as a source (domain source) and with the same power, the results are very underestimated.
Am I wrong when I guess that the results are done over 3D for a domain source but also in 3D for an edge source? how can I explain this results difference?

Clément Dupont.

I cannot be sure I understand fully your concern.
If this is about electromagnetics and magnetic field induced by a current flowing along a wire in a frequency domain computation, if you model the wire as en edge with a current as input, you should find the analytic solution without any problem (with the possible exception of the location of the wire, I did not check this out). Because doing so you impose the total current and prevent any induced current.
If you model the thin wire with a current density flowing through a (thick) cylinder with some non-zero electrical conductivity (as it is in reality), then you will get induced currents in your thick copper wire.
In that case, either you compensate for this "reactive power" by adjusting your input so that it gives the total current that you want (that is usually the job done by your generator), either you simply put a zero conductivity for the conductive inductor so that your current density creates the current that you want to impose.
Otherwise, the AC/DC module gives acceess to using terminals, ports, or coils (single turn or multi-turn) and solves this kind of issues and let you access in a straightforward way the impedance or inductance of your system, which is often wanted.
www.comsol.fr/support/knowledgebase/916/ is somehow related to this question.

I hope I correctly understood the question.

Good luck,
Eric Favre
COMSOL France

Thank you for your specifications M. Favre.

I am sorry, I didn't mention that I tried to simulate photons propagation in a turbid media with the Helmholtz PDE.
My light source was an optical fiber.
I attempted to declare the whole cylinder as the light source.
I resolved my problem defining a wire inside the cylinder (in its center) and defined this as the source.
I compared my results with a Monte-Carlo simulation and it seemed to be correct.

Clément DUPONT.

Well, I also have the same problem. Anybody knows what the mistake is in this case?