What is electromagnetic simulation?

Electromagnetic simulation solves Maxwell’s equations over your geometry to give the electric and magnetic fields, and from those the quantities you actually design against: force and torque, inductance and capacitance, current distribution, losses, and the heat those losses generate.

The single most important thing to establish first is which regime you are in, because it determines the entire class of tool you need.

Compare the wavelength of your operating frequency with the size of your device. At 50 Hz the wavelength is thousands of kilometres; a motor is vanishingly small in comparison. At 5 GHz the wavelength is centimetres, comparable with a connector.

Low frequency, or magnetoquasistatic. The device is much smaller than a wavelength, so wave propagation can be neglected and the equations simplify considerably. This covers motors, generators, transformers, actuators, magnets, sensors, busbars, induction heating and eddy-current effects — the overwhelming majority of electromechanical engineering. Solved with finite elements on a mesh of the device and the air around it.

High frequency, or RF. Wavelength and geometry are comparable, so propagation, reflection and resonance are the whole point. Antennas, waveguides, filters, EMC and signal integrity live here, and the useful methods are different — FDTD, method of moments, or frequency-domain finite elements with absorbing boundaries. The outputs are different too: S-parameters and radiation patterns rather than torque.

These are not settings on one tool; they are different discretisations, different boundary treatments and largely different user communities. Deciding which side you are on is the first design decision, not a detail.

• Magnetostatics. Steady current or permanent magnets, no time variation. Gives field distribution, flux density, force and torque, and shows where the iron is saturating. The starting point for any magnetic device.
• Eddy currents. A changing field induces circulating currents in conductors, which resist the change and dissipate energy. This is loss in transformer cores and motor magnets, the mechanism behind induction heating and magnetic braking, and the reason laminations exist.
• Motor and transformer analysis. Torque against rotor position, back-EMF, cogging, core and copper loss, efficiency maps. Often run as a sequence of positions rather than one solve.
• Electrostatics. Voltage and charge with no current flowing. Field strength for insulation and breakdown assessment, and capacitance extraction.
• Current conduction. How current distributes through a conductor, and the resistive heating that follows — busbars, contacts, PCB copper.

You have to mesh the air. Unlike stress or flow, the field extends outside the device, so the model needs a surrounding air region large enough that truncating it does not distort the answer. Too small and forces are wrong; too large and you are solving mostly empty space.

Materials are nonlinear and often anisotropic. Iron saturates: its permeability collapses once flux density is high enough, which is precisely the regime efficient machines operate in. A linear magnetic material assumption is comfortable and usually optimistic. Hysteresis adds path dependence on top.

Thin conductors need care. At higher frequency, current crowds into a surface layer — the skin depth — which can be far thinner than a sensible mesh cell. Resolving it directly is expensive; ignoring it under-predicts resistance and loss.

Electromagnetic analysis is the most naturally coupled of the disciplines, because its losses are somebody else’s heat source. Copper and core loss feed a thermal analysis; the resulting temperature changes conductivity and can demagnetise magnets, feeding back. Magnetic forces deform structures and excite vibration, which is where motor acoustic noise comes from and a vibro-acoustic analysis begins. Induction heating is the textbook two-way coupled problem.

Elmer is the mature open-source option for low-frequency electromagnetics: a multiphysics finite element code from CSC in Finland, driven by a readable .sif solver input file, with magnetostatics, eddy currents, induction and Joule heating, and internal coupling to its own thermal and elasticity solvers. For RF, openEMS and Palace are the open-source names, and they are a genuinely different kind of tool.