In Pursuit of the Silky-Smooth EV
Researchers aim to conquer the cogging-torque effect in electric machines that stands in the way of more EV refinement.
For next-generation electric vehicles (EVs), absolute smoothness in every operating mode will be an essential criterion. It will be a product differentiator, along with greater range, while raising the bar for NVH engineers. A vital detail of that smoothness will be achieving total absence of an electric motor effect called cogging torque, explains Dr. Richard Burke at the University of Bath’s Institute for Advanced Automotive Propulsion Systems (IAAPS), in the U.K.
Cogging torque is all about a mild (to the purist, bad) vibration felt through an EV’s structure at very slow speeds. The effects of cogging torque potentially could tarnish the perceived quality of a vehicle or a brand. The vibes would certainly be sufficient for NVH specialists to detect and identify the type of motor architecture causing it.
“If you feel a ripple of torque creating an uncomfortable oscillation, it’s a permanent magnet (PM) machine - chosen for its high efficiency and light weight but bringing a torque ripple penalty,” Burke explained. “Easing into a parking space or creeping forward at a junction with the motor rotating at around 25 – 100 rpm can create an uncomfortable vibration. It is at a very low amplitude but at a frequency, around 5 Hz, to which the human body is particularly sensitive.”
The effect is caused by mild asymmetries in the magnetic field generated by the motor windings setting up variations in the reluctance, he noted. This is specifically due to the geometry of the stator’s “teeth” and the variation in rotor flux. The effect can be combated, to a degree, by adding some ‘skew’ to the stator so that the windings are not straight – meaning that the rotor transitions more gently from one tooth to the next.
“The analogy would be the difference between straight-cut transmission gears and helical gears,” Burke said. “However, this comes at a cost of reduced efficiency, because the stator is no longer targeting its flux optimally but is spreading it out more - sort of hedging its bets.”
A relatively esoteric effect such as cogging is acceptable for first-generation, relatively low-volume EVs that may exhibit it, he said, but needs to be resolved as electric propulsion becomes the norm, with more quality-demanding buyers. Burke noted what is needed is “a low-cost technique that eliminates the source of the vibration so that mitigation systems such as additional damping – which add cost, weight and imprecision – could be avoided.”
That was the goal of a project led by Burke in collaboration with a so far un-named EV manufacturer. Working with the university’s NVH team, a vehicle was equipped with accelerometers to characterize the powertrain-induced vibration profiles that most affect occupant comfort at the ultra-low speeds involved. Building on this data, a software model was developed to test and refine possible solutions. “A multidisciplinary approach, calling on specialists in driveline, electric motors, motor control and power electronics, ensured that all relevant aspects of the complete driveline system and their critical interactions were considered,” Burke said.
The conclusion was that the optimum solution is to reduce the torque ripple at source by amplifying the control signal at low torque points. This was found to work well at motor speeds above 30 rpm but below that speed, where the effect of torque ripple is felt most by the vehicle occupants, the study revealed that vehicle inertia greatly reduced the authority of control systems. As stated, “Proportional integral (PI) control technology, most commonly specified for electric vehicles, was found to be unable to deliver sufficient headroom to mitigate this effect.”
The solution developed added a resonant controller to the system, operating in parallel with the conventional PI controller to increase signal gain in the critical ultra-low-speed frequency range. This novel application of a proven technique allows very fast modulation of the control signal, requesting either increased or decreased torque in real-time to cancel the ripple by around 90%.
To validate the technique on the engineering vehicle, the University developed a new set of control algorithms and a new inverter. Sensing was uprated and the sampling frequency increased to improve linearity of response, providing further improvements in vehicle refinement and drivability, Burke noted. He is confident that the new technique for resonance control of PM motors gives the team what he terms “a unique ability to quickly and affordably mitigate refinement issues related to motor-generated torque ripple,” to the significant benefit of next-generation EV motors.
Most significant for passenger cars, he said, will be the application of the technique to switched reluctance machines (SRMs). The team’s research implies it will be one of the techniques that make the next-generation motor suitable for volume applications. Burke said it will allow powertrain engineers to choose a technology that eliminates rare earth magnets and hence the environmental, social welfare, cost and stability of supply issues that continue to go along with these materials.
With an SRM, there is still a moving rotor, but it does not have magnets on it – it is made entirely of soft magnetic steel. “The stator creates an electromagnetic field but the torque is generated entirely by ‘reluctance’ – the desire of the rotor (of a ferromagnetic material) to align itself as best as possible with the magnetic field of the stator, in the same way that any steel object will try to align itself with a magnetic field – e.g., you can make a compass by floating a pin on water; the pin will then align itself with the Earth’s magnetic field,” Burke explained.
But although SRMs will prove to have lower unit costs and be easier to recycle, they also will create those unwanted cogging resonances, said Burke’s colleague, Dr. Chris Vagg. “Designing PM machines for end-of-life recovery of rare earth materials is proving very difficult and will become even more challenging as shaft speeds increase,” Vagg said. “The architecture of SRMs means this issue doesn’t arise. However, the flip side is that they are less efficient, so must spin even faster to deliver the same power density.”
The “extreme” shaft speeds Vagg noted (around 35,000 rpm) present a further SRM liability: lack of refinement. At high speeds, their complex electromagnetic harmonics can create mechanical resonances across the entire vehicle, along with louder and higher frequency noise from the reduction gearbox. “Existing design tools are excellent for optimizing individual systems like motors and transmissions,” Vagg said, “but to control resonances we must know how all the associated systems interact under different conditions.”
Development of a new generation of design tools and processes that treat the entire driveline and the vehicle interfaces as one system would be essential. To do so, the University has embarked on a program to meld specialist expertise in rotating machines, electro magnetics, thermal management and NVH to enable the entire EV driveline ecosystem to be modelled.
“Our goal is the development of a modelling process that will quickly and efficiently provide the data needed to optimise the NVH characteristics of the driveline with minimum additional damping,” Vagg said. “As with torque ripple, the lightest, cheapest, most elegant solution is to design-out the problem. We believe our research will make that possible.”
Team leader Burke stressed the importance of getting costs down now, enabling the timely implementation of required developments in design and production for the next generation of EVs. “What we emphatically do not want is to find in 15 years’ time that we have hundreds of thousands of end-of-life vehicles containing rare earth materials we desperately need back in the supply chain but cannot economically recover,” Burke said. “Designing out the barriers to SRM adoption will take us a long way to solving both challenges.”