Switched reluctance motors advantages challenges and simulation

Key Takeaways

• Switched reluctance machines trade magnet supply exposure for control complexity, which makes them a programme-level decision rather than a component substitution.
• Torque ripple behaves as an operating-point property set by the machine, converter and controller as one loop, so it cannot be specified away at the machine level.
• Closed-loop validation against production firmware separates SRM programmes that reach volume from those that stall after the prototype.

A switched reluctance motor delivers high torque density without a single gram of rare earth material, and that property alone explains why the topology keeps returning to serious evaluation for traction and industrial drives.

Interest rises whenever magnet supply looks fragile. United States net import reliance for rare-earth compounds and metals sits at 80%, with China supplying 70% of those imports between 2020 and 2023. That exposure makes a magnet-free machine strategically attractive long before it becomes technically straightforward.

The hard part isn’t the machine. Torque quality, acoustic behaviour and efficiency in an SRM emerge from machine, converter and controller acting as a single loop. Assess any one of them alone and you’ll reach conclusions the assembled drive won’t support. Teams that treat SRM evaluation as a machine design exercise usually find the gap late, on a dynamometer, when control code meets real magnetics for the first time.

How a switched reluctance motor produces torque

An SRM produces torque through magnetic reluctance rather than magnetic attraction. Both rotor and stator carry salient poles, the rotor holds no windings and no magnets, and torque appears as the rotor moves toward alignment with an energised stator phase. Phase currents are switched in sequence against rotor position.

A common 8/6 configuration makes the sequence concrete. Eight stator poles form four phases, six rotor poles pass beneath them, and the controller energises each phase during the rising inductance region before commutating to the next. Torque production stops entirely once inductance begins to fall, so the switch-off angle matters as much as the current magnitude.

This construction is what gives the machine its reputation for ruggedness. No magnets means no demagnetisation risk at high temperature and no open-circuit voltage when a fault occurs. It also means torque exists only while the controller is actively doing something correct, which shifts a large share of performance responsibility away from the machine and onto the drive.

Where magnet free construction lowers total motor cost

Removing magnets removes the most volatile line in the bill of materials. Rotor manufacture reduces to a stack of laminations with no bonding, no magnetisation step and no handling precautions, and the stator carries simple concentrated windings. The savings are structural rather than incremental.

Consider a mid-volume industrial pump drive qualifying two suppliers. A permanent magnet design ties the programme to neodymium and dysprosium pricing and to an export licensing regime the buyer doesn’t control. The reluctance alternative moves that risk into steel and copper, both of which trade on deeper markets with more suppliers. Cost advantage isn’t free. Higher engineering effort moves into the converter and the control software, and the asymmetric half-bridge topology an SRM needs uses more switching devices per phase than a standard three phase electric motor inverter. The honest framing is that SRM relocates cost from procurement into development, which suits high-volume programmes far better than one-off builds.

Why torque ripple dominates switched reluctance motor performance

Torque ripple is the constraint that decides most SRM programmes. Doubly salient geometry and deep magnetic saturation make torque a strongly nonlinear function of both current and rotor position, so torque contribution rises and falls sharply as each phase hands over to the next. Audible noise and vibration follow directly.

The commutation overlap region shows where it originates. As one phase demagnetises and the next builds current, total torque dips unless the two overlap precisely. At low speed with limited back electromotive force the controller has authority to shape that handover. At high speed it does not, and ripple that measured acceptably on a bench at 500 rpm can become unacceptable at 8,000 rpm.

This is why ripple resists a general solution. Every mitigation strategy trades against something else, usually efficiency, torque density or switching losses, and the balance depends on the operating point. Treating ripple as a fixed machine property rather than an operating-point behaviour produces specifications the finished drive quietly fails to meet.

“Treating ripple as a fixed machine property rather than an operating-point behaviour produces specifications the finished drive quietly fails to meet.”

What effective control of an SRM drive requires

SRM control requires accurate rotor position, current regulation fast enough to track a nonlinear inductance profile, and firing angles tuned per operating point. Simple hysteresis current control works at low speed. Anything approaching production torque quality needs a scheme that shapes the torque contribution of each phase directly.

Torque sharing functions illustrate the approach well. Instead of commanding phase currents, the controller commands a torque profile for each phase and lets an inverse torque model derive the current reference, so the handover between phases stays smooth. Direct instantaneous torque control and model predictive schemes pursue the same outcome through different mathematics.

Practical implementations converge on a short set of requirements.

• Rotor position resolution fine enough to place turn-on and turn-off angles accurately
• A flux linkage or inverse torque model characterised across current and position
• Current loop bandwidth well above the phase commutation rate at maximum speed
• Firing angle scheduling that adapts across the speed and load range
• Fault handling that accounts for single-phase operation without magnet back electromotive force

Each of those items depends on the machine’s measured magnetics, which is where paper specifications stop being useful.

Why offline models misrepresent switched reluctance motor behaviour

Offline simulation understates SRM difficulty because it removes the two things that cause most failures, real timing and real control code. A fixed-step model with an idealised inductance profile will produce clean torque waveforms that the physical drive never reproduces. Sampling delay is the clearest example. A controller running at 20 kHz sees phase current at discrete instants while inductance changes continuously with rotor position, so the current it regulates is already stale by the time the duty cycle updates. Offline models often ignore that latency, along with dead time, sensor noise and finite computation time in the control loop.

Magnetic representation compounds the problem. Linear inductance assumptions break down exactly where SRMs operate, deep in saturation, and a model that misses saturation will report torque and ripple figures that flatter the design. The result is a validation record that looks complete and predicts almost nothing about the assembled drive.

“The result is a validation record that looks complete and predicts almost nothing about the assembled drive.”

How real time simulation validates SRM drive designs

Real-time simulation closes the gap by running the machine and converter model at hardware speed against the actual controller. The control unit executes production code, reads emulated sensor signals and drives emulated switching devices, so timing, quantisation and computation delay all participate in the result.

Fidelity requirements are strict. SRM converter switching happens in microseconds, which puts the power electronics beyond what a CPU-based solver handles inside a real-time step. Running the machine model with a saturation-accurate flux linkage map on FPGA, with the asymmetric half-bridge solved alongside it at sub-microsecond resolution, is what makes the loop representative rather than approximate. OPAL-RT users typically pair an FPGA machine model with an eHS-solved converter for exactly this reason.

The payoff is coverage that a dynamometer can’t reach safely or cheaply. Position sensor failure, phase open circuit, converter device faults and extreme thermal operating points can all be exercised repeatedly against production firmware, weeks before any hardware exists.

Choosing between switched reluctance and permanent magnet machines

The main difference between switched reluctance and permanent magnet machines is where a programme carries its risk. Permanent magnet designs deliver higher torque density and simpler control while holding exposure to magnet supply and cost. Switched reluctance designs remove that exposure and charge for it in control complexity and validation effort.

Volume shifts the arithmetic considerably. Global electric car sales exceeded 20 million in 2025, reaching a 25% share of the overall car market. At that scale a per-unit magnet cost becomes a programme-level liability, and the engineering investment an SRM requires amortises across enough units to pay back.

Programmes that succeed with switched reluctance treat it as a drive development effort from the first week and put the controller in a closed loop with an accurate machine model long before hardware arrives. That discipline, more than any single control algorithm, separates the SRM projects that reach production from the ones that stall at the prototype stage. It’s the reason OPAL-RT‘s simulation work with electric machine developers concentrates on the loop rather than the machine in isolation.