Discussion Forum

Hi

I agree, I haven t seen the exact math equation for reacf() anywhere, probably there are several variants depending on the physics, but normally it's quite precise and works OK.

Now, again depending on your model, I do not see why the integration if ec.nJ and ec.normJ should be always equivalent ? or did I miss a point here ?

Anyhow flux integration is delicate in FEM, you really need to be sure your mesh is resolving the local gradients, and that yuo have no singularities along the integration path (typically sharp corners in ACDC)

Another point, when you plot reacf(V) over a boundary, you will notice its only defined at the nodes, so the integration to be used is the "summation" over nodes and not the "integration" over boundary, any case if you leave "automatic" COMSOL looks after this

This is all related to the fact that most dependent values on boundaries come from "heritage" from the adjacent domain elements

--
Good luck
Ivar

HI

Thanks for the prompt reply. I just wondered what reacf or ec.normJ are. One cannot use anything which she/he does not understand. I agree that numerical calculation of gradients is a challenge, but it is most crucial in electrochemistry. With meshing I can change the value of electric current 0-100%. Which one is correct?

Hence: what is the ec.normJ?

Hi

turn on the equation view (Options preferences) and look at the definition of the COMSOL internal variables, in this cas it is the norm (sqrt(Jx^2+Jy^2+Jz^2)) of the current density, the formula is "complex compliant" by the "realdot()" operator that is just there to correctly get the scalar value for complex current densities (typically AC cases)

Only "support" can give you the internal details, here (and myself) we are all "users" not from COMSOL

--
Good luck
Ivar

Thank you. I have equation view switched on, but I do not see that; never mind.

Lasse

Hi

I find it nicely in my 4.3 ;)

But reacf() is internal programming, that I do not know where is

--
Good luck
Ivar

Integrating the normal current density over an insulating surface should give zero as you expect. However, if you integrate the norm (magnitude) of the current density it should not be zero. That’s because there can be a current parallel to the insulating surface that will affect the integral.

Now reacf() is a tricky operator but very powerful as Ivar mentioned. I wish it was better documented. It provided the nodal “reaction forces” corresponding to fixed degrees of freedom. In solid mechanics, reacf(displacement-x) give the nodal reaction in the x direction, in fluid flow reacf(velocity-x) also gives the reaction in the x-direction and in your case reacf(Voltage) should give the nodal current normal to the boundary. That should be zero on an electrically insulated boundary.

Nagi Elabbasi
Veryst Engineering