New Fault-Tolerant Strategy Cuts Torque Ripple in Demagnetized Doubly Salient Motors
In the high-stakes world of electric mobility and aerospace propulsion, reliability isn’t just a desirable feature—it’s non-negotiable. Electric vehicles (EVs) and aircraft demand drive systems that can withstand component failures without catastrophic loss of function. Among emerging motor technologies, the Doubly Salient Electromagnetic Machine (DSEM) has long promised a compelling mix of simplicity, cost-effectiveness, and robustness. However, one Achilles’ heel has persisted: its vulnerability to excitation failure. When the field winding loses current—due to a broken wire, a failed power switch, or thermal degradation—the motor’s primary torque mechanism collapses, threatening to bring the entire system to a halt.
Now, a team of researchers from Nanjing University of Aeronautics and Astronautics (NUAA) and Hohai University has unveiled a breakthrough control strategy that not only enables DSEM to keep running after such a failure but does so with dramatically reduced torque ripple and minimized copper losses. Published in the Proceedings of the CSEE, their work introduces a novel vector control approach specifically tailored for the post-demagnetization operating regime, transforming what was once a limp-home scenario into a viable, high-performance fault-tolerant mode.
The core of the innovation lies in a fundamental rethinking of how to command the motor’s armature currents once the field is gone. In normal operation, a DSEM generates torque primarily through the interaction between the field current and the armature current—a principle similar to a traditional DC motor. But when the field current vanishes, this “excitation torque” disappears. The only remaining source of torque is the reluctance torque, which arises from the machine’s doubly salient geometry—the fact that both the stator and rotor have protruding poles that create a position-dependent magnetic reluctance.
This reluctance torque is notoriously difficult to control smoothly. It is proportional to the square of the phase current and is highly sensitive to the nonlinearities in the machine’s self-inductance profile. Traditional fault-tolerant strategies, such as the three-phase, six-state method referenced in the paper, often resort to crude, square-wave current injection. While this can produce a net positive torque, it comes at a steep price: massive torque ripple during commutation and inefficient use of copper, leading to excessive heat and wasted energy.
The NUAA team, led by Lei Xiong, Hongjuan Ge, and Bo Zhou, recognized that to tame this beast, they needed a more sophisticated and continuous control framework. Their solution was to adapt the powerful principles of vector control—a technique perfected in induction and permanent magnet synchronous motors—for the unique physics of a demagnetized DSEM.
The first major hurdle was the lack of a suitable mathematical model. Vector control relies on a clear relationship between a “current vector” in a rotating reference frame and the resulting torque. For a demagnetized DSEM, no such model existed. The team’s key insight was to leverage the principle of instantaneous power invariance. By carefully analyzing the power flow in the machine and applying a coordinate transformation that preserves this power, they were able to derive a torque equation in a polar coordinate system. This new model elegantly links the reluctance torque to the amplitude and, crucially, the phase angle of a synthesized current vector.
With this model in hand, the researchers could now formulate a two-pronged optimization strategy. The first goal was to minimize copper losses. They demonstrated mathematically that for a given torque output, copper loss is minimized when a specific cosine term in their torque equation is equal to one. This condition directly dictates the optimal phase angle for the current vector, which is a simple function of the back-EMF vector’s angle—a quantity that can be derived from the machine’s known geometry and the measured rotor position.
However, a naive implementation of this optimal angle leads to a practical problem: discontinuities. The required phase angle would jump abruptly by 180 degrees at certain rotor positions, which is physically impossible for an inductive winding. The team’s elegant fix was to implement a state-machine-based logic that alternates between two equivalent solutions (k=0 and k=-1 in their formulation). This simple yet brilliant trick doubles the electrical period of the current waveform, ensuring that the commanded phase currents are smooth and continuous, respecting the fundamental physics of the inductor.
The second goal was to suppress torque ripple. Even with the optimal phase angle, the torque would still ripple because the machine’s inductance profile is not perfectly sinusoidal. The back-EMF coefficients in their model fluctuate with rotor position. To counteract this, the team proposed a dynamic adjustment of the current vector’s amplitude. By calculating the exact amplitude needed at each rotor position to produce a constant, ripple-free torque output, they effectively “pre-distort” the current command to cancel out the machine’s inherent nonlinearities.
The result is a complete, closed-loop control system. A speed controller provides a torque reference. This reference, combined with the pre-measured machine parameters (the back-EMF coefficients), is fed into an amplitude calculator. Simultaneously, the rotor position is used by a phase angle controller to determine the optimal current vector direction. These two components—the amplitude and the angle—are then combined to synthesize three smooth, sinusoidal reference currents for the phases. A standard current hysteresis controller then drives the standard three-phase bridge inverter to make the actual currents track these references.
The experimental validation of this strategy is where its superiority becomes undeniable. The team tested their method on an 18/12-pole DSEM prototype, simulating a complete loss of excitation. They pitted their new vector control approach against the established square-wave method from prior literature.
The results were stark. At a modest 200 rpm and a 3 N·m load, the square-wave strategy produced a staggering torque ripple of 416.99%—a level that would be utterly unacceptable in any real-world application, causing severe vibration, acoustic noise, and mechanical stress. In contrast, the new vector control strategy slashed this ripple to just 46.1%, an 88.9% reduction. This is a transformation from a violently jerking motor to one that runs with a smoothness approaching that of a healthy machine.
The efficiency gains were equally impressive. By ensuring the cosine term was always at its optimal value of one, the new strategy achieved true minimum-copper-loss operation. Measurements showed the armature copper loss was reduced by 27%, from 132.3 W down to 96.6 W. This isn’t just about saving a few watts; in a fault-tolerant scenario where every joule of energy is precious, this efficiency directly translates to a longer limp-home range for an EV or a safer glide path for an aircraft.
The team also investigated performance across a speed range, from 200 rpm to 600 rpm. They found that while the square-wave method’s torque ripple and copper losses both increased with speed, their new strategy maintained its copper loss at a near-constant level. Its torque ripple did increase slightly at higher speeds, a consequence of the inverter’s limited bandwidth and the challenge of tracking faster current commands, but it remained far lower than the square-wave alternative. An added benefit was that the fundamental frequency of the phase currents in the new strategy was only half that of the square-wave method, which is a significant advantage for extending the motor’s usable speed range and reducing switching losses in the inverter.
Finally, the researchers tested the system’s dynamic response, subjecting it to sudden speed changes and step changes in load torque. The motor, operating in its fault-tolerant mode, demonstrated robust and stable performance, quickly settling to the new command without instability or excessive overshoot. This confirms that the strategy is not just a steady-state curiosity but a practical, dynamic control solution ready for real-world deployment.
This work represents a significant leap forward in the reliability of electric drive systems. By providing a method for a DSEM to not just survive, but to thrive, after a catastrophic field failure, the NUAA team has addressed a critical gap in the technology’s readiness for safety-critical applications. Their approach is elegant in its theory, practical in its implementation (relying on a standard inverter topology), and transformative in its results.
For the automotive industry, this means a potential path to more affordable, robust EV motors that don’t rely on expensive rare-earth magnets and can handle a major failure mode gracefully. For aerospace, it offers a new level of assurance for electric and hybrid-electric propulsion systems, where redundancy and fault tolerance are paramount. The strategy effectively turns the DSEM’s post-fault operation from a liability into a feature, showcasing the power of advanced control theory to unlock new capabilities from existing hardware.
By Xiong Lei, Ge Hongjuan, Zhou Bo, Jiang Siyuan, Wei Jiadan from Jiangsu Key Laboratory of New Energy Generation and Power Conversion (Nanjing University of Aeronautics and Astronautics), and Zhou Xingwei from Hohai University, published in Proceedings of the CSEE, DOI: 10.13334/j.0258-8013.pcsee.230036.