Discussion Forum

Hi

what are you modelling ?
a ferrite magnet and need to define this in COMSOl ?,

then add an ampere law node for the magnet material domain, select the BH to Magnetisation and enter the magnetisation value of the Hc for your magnet. Note: this is a simplification and only truely valid for the latest high quality rare earth magnets, as else mu_r of the magnet is not constant with the field and you need to use full BH curves to give the true magnet behaviour.

Chek the web with google on Ferrite magnets, you get most info from there today ;)
And if you are starting with COMSOL, then go carefully over the library mdels for the related field you want to study, this is the best way to learn

--
Good luck
Ivar

thanks for the reply sir!!

no i am not modelling a ferrite magnet as such.actually,i am trying to model a ferrite phase shifter.the modelling steps are like that first i need to find the DC magnetic state of the ferrite in a magnetostatic solver.By DCmagnetic state i mean that i need to find the internal field and magnetization of the ferrite.

I am copying the words of the research paper( i am going to implement) as it is.may be this makes my words clear:

"To evaluate the tensor entries for the ferrite accurately, the internal field and magnetisation within the ferrite must be known. The relationship between these values and the applied bias field is complicated by the effect of the demagnetising field due to the surface divergence of the magnetisation vector. A magnetostatic finite element solution is used here to calculate the M and H variables directly. A mesh is constructed over the full geometry including the ferrite loaded waveguide, magnetic material and the bias circuitry. A magnetisation curve is used to characterise the nonlinear ferrite material. A standard two-dimensional FE magnetostatic solver is used to solve the nonlinear Poisson equation in terms of the axial component of the magnetic vector potential (φ )
∇.(μ ∇φ)=-J
where J is the bias current density and 
is described by the nonlinear magnetisation curve. From the internal flux density (B) and magnetic field (H) are derived before using the constitutive relation,
B=μH+M
to find the magnetisation (M) of the ferrite.

i hope i made the query clear.

CORRECTION:


∇.(μ ∇φ)=-J
where J is the bias current density and 
is described by the nonlinear magnetisation curve.

From the φ internal flux density (B) and magnetic field (H) are derived before using the constitutive relation,
B=μH+M
to find the magnetisation (M) of the ferrite.

Hi

if you look into the ACDC doc (pdf files) you will find how COMSOL handles magnetism, basically it solves for Vm or A=(ax,Ay,Az) the scalar respectively vector magnetic potential in the MFNC, MF or MEF physics (check the related dependent variables) I assume you will use one of the two first physics.

ACDC has always been "a chicen and an egg" issue for me, what comes first ? ;)

Once you have A you deduce H or B, but then the local flux is influenced by the material proprieties that you must provide to COMSOL. Be it mu_r or the full B-H or H-B relation or the known magnetisation (check the way the soft iron material is implemented in COMSOL) hence you can get H, B and or M.
But if you have magnetisation, which is a hysteresis remannece of the material behaviour as subject to magnetic fields (histoey), you really have to have on one side correct material data, including hysteresis curve info, plus relative permeability.

You need to add several "Ampere laws" "nodes" and set the respective" Magnetic field - Constitutive relations" setting according to your knowledge of the material

If you have experimental data, and knowing B - H relation in a given field you could consider optimising for M to get the least square value of measured M

--
Good luck
Ivar