Discussion Forum

Hi. Your .jpg file provides a picture but no additional information. However, it looks fairly straightforward.

First question: Did you include an overall enclosing area and set its exterior boundary conditions? You need to model not just the ferrite, coil and steel plate, but also at least some of the space surrounding these parts (model it as vacuum) and include a boundary condition on the magnetic field on the resulting bounding edge. For the latter, I recommend that your problem boundary extend to at least +/- 60 in z and to at least 60 in rho, so that you don't perturb the fields at the parts inside very much. Set that outer boundary condition such that the normal component of B is zero there, since all B field line have to wrap back (so this is a reasonably ok condition).

2nd question: You wish to model 1000 turns. I think your approach in that regard is probably ok. As a first approximation, you can just model the coil region as a constant current density normal to the page, such that the current density X the area = 1000 X single wire current. And, at least initially, don't bother with the magnetic or skin depth properties of the coil itself. That's what you are doing, right?

3rd question: There are many kinds of steel, some with large permeabilities, some not. And some with a lot of hysteresis, and some not. What kind of steel are you studying and do you know its magnetic properties? The conductivity of the steel also matters and should be included. Are you doing that?

4th question: You'll also need to include the proper permeability for the ferrite. Have you done that?

5th question: Depending on the frequency applied and the conductivity, and permeability of the steel, there may be very sharp fall-offs in field strength when penetrating the steel. Make sure you estimate the skin depth first analytically (you can find formulas for skin depth on the web). The make sure you mesh the steel plate finely enough to capture those field gradient details, especially near the top of the plate. Have you done that?

I hope the above questions and comments help.

Thanks for the attention Robert. First of all I'll answer your questions and then I'll rise some new doubts for you or anyone else that could help me. This time, I'm attaching my model.

Question 1: Yes, I included an overall enclosing area and inserted the boundary conditions.

Question 2: That is exactly what I'm doing. I have 142 mA in one single wire and the coil area is about 30[mm^2]. Considering 1000 turns, my current density is close to 4.7e6[A/m^2]. Is that ok?

Question 3 and 4: I don't know exactly the magnetic and electrical properties of my steel, but I think it is close to Steel AISI 4340, so I could load the properties from the COMSOL library. For the ferrite I used the Iron properties from the library. These approaches are good?

Now I'll first model the probe (without the plate), just the ferrite and the coil, when I'm sure that my probe is working, I'll start to make the problem more difficult (with the plate). I did some measurements of the magnetic flux with a gaussmeter in a real probe and I found some values around 0.5 mT closer to the probe. I'm trying to reproduce this experimental test with COMSOL but the results are not very similar. Do you think these differents results are because my approaches are not good or might have some construction mistake in my model?

Thanks for helping me!

Cesar

Hi Cesar,

I am actually trying to do a very similar thing. Like you i want to model a ferrite cored probe but i am interested in getting a relationship between the inductance of the coil and the lift-off between the coil and a block of aluminium. I am using d=0.22mm diameter wire and approx 400 turns driving the coil at Io = 170mA.

One difference that i think i have with your system is in the way we calculate the current density J0. I assumed that if i calculate the current density within one wire (Io/(pi*(d/2)^2)) then as this is a density (i.e. per unit area) i can just apply this to the sub-domain which approximates the total area of the wire conductor (ie Num_Turns * area_of_wire). Therefore i dont actually use the number of turns in the calculation for current density, like you seem to be doing, but instead use it only to specify the area of this sub-domain. It doesn't make sense to me to use it in to calculate the density (although i am very new to all this so possibly doing something stupid).

As it happens my calculations for coil inductance are not coming out correct (based on my experimental results) and i am not really sure why! I have attached my model for you to have a look at. If you spot any flaws please let me know.

Regards,

Liam

EDIT: Its the second model "Eddy Current2.mph" that is the most up to date - cant figure out a way to delete the original.