Discussion Forum

Dear all,

Hi, I'd like to calculate torque of an electric motor model in 2D. The model consists of "Rotating Machine, Magnetic(rmm)" physics and "stationary" study (in quasi-static sens). For special reasons, it must be done without "Force Calculation" node. I know the fundamentals such that;

Maxwell's stress tensor,

Torque calculating formular,

In the integrand of the torque calculating formular, the term itself can be expressed with comsol expressions;

rmm.nTx_name and rmm.nTy_name

And also, according to the Maxwell's stress tensor, it can be expressed in comsol like;

rmm.Bx*(rmm.Bx*nx + 2*rmm.By*ny)/(2*mu0_const) and rmm.By*(2*rmm.Bx*nx + rmm.By*ny)/(2*mu0_const).

But, there is a difference between the two. The first attached figure tells it is.

And after integrating them through the edges of the rotor the difference gets larger. In the second figure, blue line is the torque vs angle curve obtained by the "Force Calculation" node and green one by manual.

I've considered replacing with ( that is rmm.normB/rmm.normH ) but there was no big difference.

What is the reason of this difference? Does it just due to the numerical error? If it shows us significant difference to calculate torque with Maxwell's stress tensor, how does comsol do?