Discussion Forum

Marouane,

You can evaluate voltage as a line integral of the electric field between two conductors.

See “Finding the Impedance of a Parallel-Wire Transmission Line” from Comsol Model Library for details.

Regards,
Sergei Yushanov

Thanks for the reply Mr Sergei. It's indeed a good idea that I should study further.

In the meanwhile, I found that the Electrostatics physic (es) does support the V voltage as an entity and a simulation result.
Can one use both physics at the same time ? I mean having a conductor excited with the magnetic field
physic and calculate the V voltage using the electrostatic physic in another conductor subdued to this same magnetic field ? If it's possible, how can I couple the 2 physics, please ?



Thanks again

Marouane,

You might want to consider Magnetic and Electric Fields (mef) interface. This interface solves simultaneously Ampere’s Law and Current Conservation law using magnetic vector potential A and scalar electric potential V. Thus, electric potential is available for imposing boundary conditions at the exterior boundaries and for post-processing in the entire computational domain.

Regards,
Sergei

Thanks for the reply. I thought about using (mef) but since it's not compatible with the time dependent study that I'm needing, I just can't use it...

Anyway, I started the line integral method that you provided but i have a problem, and I'll be grateful if you help me further.
I drew the line of the intergral in my model using the bezier polygon node according to the parallel wires model. But when I want to select the bezier as a boundary source selection, it's not considered as a domain and I can't assign it.

I noted that while in 2D, bezier polygon can be set as a "solid", in 3D, you only have to chose between open curve and closed curve which I think causes the bezier to not being recognised as a solid. Any idea how to fix that ? Or any different idea about how to draw the line of the integral in 3D ?

Marouane,

If you are in 3D, then locate the Source Selection section and from the Geometric entity level list, choose” Edge” to select your Bezier curve.

Regards,
Sergei

Hi

to master COMSOL better it's important to classify correctly Geometry Objects and FEM Entities, Domains and Boundaries (v4 naming), and differences for 3D and 2D dimensions etc (check the doc).

Geometry obects are volumes, surfaces, lines and points for respectively 3D to 0D objects, these geometry objects are "analysed", => made unique, ordered etc by COMSOL and transformed into FEM Entities such as Domains and Boundaries (for the two highest dimensions considered) and edges and points if these do not overlap with the two formers. Hence "3D volumes" objects becomes "3D domain" entities, their respective "surfaces" becomes "boundaries", and the edges surrounding a surface becomes lines, and the lines are defined between points. In 2D it's the surfaces that becomes domains, and boundaries are the surface edges, while points are one level lower and are mostly ignored for the model buildup. For 1D its the edges that are the domains and the points the boundaries.

Now physics (or their PDE definitions) always applies to domains (whatever dimension) and the boundary conditions to the respective boundaries. There are the exceptions of "thin layer physics, where domain type equations are simplified and attached to boundaries to fake very thin layers of materials, thes are then a type of domains with a 1d reduction

Therefore open bezier lines are at most edges (hence at most boundaries in 2D, and hardly used to define physics in 3D), while closed bezeir lines entangle a surface and might be the edges = boundaries of the enclosed surface = domain in 2D, or its the surface defined by the closed bezier lines that is the boundary of a set of such surfaces that have been grouped to for a 3D "solid".

So far you have not really told us the dimensions you are working in, so the wording you use is already confusing for me.

Now to come back to the pysics:, I do not really understand which time behaviour you want,
You talk about a voltage to drive a magnetic field, but it's the current and it's geometrical path that generates the magnetic field. The voltage is there to force the current to flow in a resistive material = your conductor.
So if you have a voltage and a geometrically defined conductor, you can use EC physics (dependent variable V) to give you the current flow (ignoring any magnetic field effects) then you could use the current to define a magnetic field with MF (dependent variable A).
If you do not apply any time changes (consider the magnetic field is made "instantaneous" w.r.t. the current flow) you can also use MEF (dependent variable A and V) that combines EC and MF physics (with some limitations on time and frequency domain solving), or you run time solving in EC and a parametric sweep in MF based on the current domain mapping. The latter is correct, apart you have no back-EMF effects: the magnetic field generating an opposed voltage back into in your conductor. Back EMF effects become important if you have coils or motors. For that you can use frequency domain sweeps of MF at low frequencies too. The ACDC hypothesis: wavelength large compared to the model sizes means also that the frequency domain is such that a magnetic field change is seen "instantaneously by the full model.

If this ACDC hypothesis is no longer respected, one enter the RF field and the electromagnetic laws are adapted differently in RF than in ACDC. Take a look at the equations used, and see the changes for the different solvers

To add a comment to Sergei's reply, if you use one of the RF solvers, or solve for E, then indeed you can integrate the electric field, but do not forget that this is integration is independent of the path chosen (if your model is homogeneous) and you can define a straight or a curved line to integrate along. The only issue is the starting and ending points, these should be representative of your average field value at the end of your wire. This is also the difference of a Electric field BC node and a lumped port, or a potential BC and a Terminal BC node. The electric field or voltage BC expect a field value E(x,y,z), but you might define a scalar, that is then applied evenly over the full selected boundary. But the latter case is exactly what the lumped port ot Terminal BC condition assumes, but these two last BC nodes also adds then some interesting variables for you such as voltage, current, impedance, power etc

--
Good luck
Ivar

Thanks for your advices, and sorry for the late reply ;)

The 3D model's dimensions are a 0.02m radius copper torus, another 0.003m copper torus in its middle and 0.1 air sphere.
The point : I'm exciting the first big torus with a current to generate a magnetic field. And I want to know the effect of this magnetic field on the little unexcited torus by calculating its value and the value of the V in its borders. This why I need the magnetic field and the voltage. What I've done is to open the revolution angle of the little torus to 350 and create a besier polygon between its borders which will serve as a line integral of the Electric field, which means the V. But so far I have a probleme meshing the model since I have an "out of memory"error with a normal mesh and an "edge is much shorter than mesh element" error with a coarse mesh...

Marouane,

Seems, your geometry is axisymmetric . Can’t you take advantage of this? This would allow you to get rid of “out of memory” error. (Don’t know if your physics is axisymmetric as well?)

Regards,
Sergei

Hi
Thanks for the reply.

Actually, I solved that issue. I chose a custom element size parameters right between "normal" and "coarse".And I've got no longer mesh errors.

The problem is that when I run the simulation and I pot the Electric field, the inner little torus doesn't seem affected by the outer torus' fields since the electric field is null there. Like it's shown in the attached file.

When I simulate the V voltage it's even a worst problem since no torus seem to have a voltage. Also attached.

What's wrong in my approach please ?

Thanks very much for you help

Marouane,

It is hard to tell what is wrong with model setup without seeing it. I am attaching test model2s (2D axisymmetric and 3D) where time-dependent current in the large torus creates induced field in the small torus.

Regards,
Sergei

Thanks very much for your help. I really appreciate it ;)

Unfortunately, I'm using a 30 trial version of Comsol 4.3 (but I'm urging my university to buy a licence) and can't open your attached files, I got an error message.

Can you please attach the report or precise screenshots of the 3D model at least.

Thanks again and sorry for the inconvenience

Hi I solve my problem using RF module , i get the result as electric field.

I make line from boundary to the point i want and i use line integral to solve it but i still confuse which value to choose if my line not tangential to Ex or Ey or EZ?


How can i use line integral of electric field to get the voltage at certain point because they have Ex,Ey and Ez


Any help really appriciate , thanks