As manufacturers continue to push the boundaries of motor, inverter and battery technologies, test rig suppliers have had to develop innovative solutions and approaches to keep up with the torque, power densities and voltage of the latest electric powertrains.
E-motor testing
To accurately test an electric powertrain, dynamometers are used to conduct comprehensive e-motor testing programmes. In this context, the e-motor (which can also be referred to as an e-axle or e-drive if being tested with its transmission) consists of four elements – the motor, inverter, battery and transmission.
Testing an e-motor involves mounting the motor and inverter on the test rig with the motor shaft connected to a dyno. Alternatively, the vehicle transmission unit can also be incorporated, with driveshaft outputs connected to one or multiple dyno machines as required.
A battery emulator supplies DC energy to the inverter, which converts it into three-phase AC that then drives the motor. The motor generates torque, which is reacted by the dynos. The dynos also act as motor/generator units (MGUs), controlling the rotational energy of the spinning shafts.
“AC dynos are electric MGUs that can apply precise loads to a test piece,” explains Roberts. “Electrical energy either drives the test piece’s rotation, known as motoring, or resists its rotation, known as absorbing. A mechanical driveshaft connects the e-motor to the dyno, and the resultant electrical energy can be recirculated through the inverter drive suite and battery emulator in either direction depending on the test mode.”
Teams will use e-motor rigs to map the efficiency of a motor. Phase angles, torqueto- current scalars and inverter switching frequencies will all be adjusted and tested for different bus voltages and temperatures (Courtesy of Intertek)
The motors used in the dynos are typically designed to different requirements than the motors used in vehicle powertrains. Dyno motors typically have fewer constraints on power density but are required to cover a broad operating range for different test motors and be robust for a long lifetime.
“In principle, dyno motors are no different to the motor being tested,” says Ian de Souza at Ricardo. “If you consider an EV, the motor provides traction to the wheels, and under braking, the motor turns into a generator, converting the kinetic energy of the vehicle into electrical energy. Dynos are effectively mo
tors in ‘regen’ mode, while the test motor is in ‘drive’ mode. However, we also need to validate the test motor in regen mode, so the dyno motors have to operate in ‘drive’ mode as well.”
“Either way, it is about managing the flow of current through the inverter,” explains Chris Hutt at AC inverters supplier Unico. “The torque a motor produces is roughly proportional to the current being provided when operating within its rated speed and torque range. When the dynos are in drive mode, they are taking current from the inverter and putting it into the dynos, whereas when the dyno motors are acting as MGUs and absorbing energy they are actually generating the current, and the inverter controls the torque and speed of the motor.”
The additional energy produced when the dyno motors operate as generators can be fed back into the electrical supply of the building or the battery emulator to be reused in the dyno. “This power is converted back into AC within the dyno drives,” says de Souza. “This allows us to feed the battery emulator, resulting in a closed loop of energy. “No system is 100% efficient though,
so we have to top up the total input power from the grid. Let’s say we are testing a 500 kW motor, and owing to losses we have 450 kW going round the closed-loop system. The test rig itself only draws electricity from the grid to cover the losses, which in this case would be 50 kW.”
That can make electric powertrain testing highly efficient. “The electrical energy circulates within the facility,” says Jonathan Peters from test company Intertek. “At stable speeds, the only energy
wasted is due to losses, and when you consider that a typical Formula E motor operates at over 95% efficiency and the dyno systems are also highly efficient, you’re only talking about tens of kilowatts of required electrical energy. With a conventional engine AC dyno however, any electricity generated cannot be used to power the test engine.”
Dyno motors
In the past, most of the motors used for dynos were of the AC induction type. However, like most motors used in modern vehicle powertrains, synchronous permanent magnet machines are becoming more prevalent. A typical traction motor consists of a stationary stator and a rotating rotor. The stator surrounds the rotor and is made up of multiple ‘teeth’ made from specific grades of electrical steel. Copper coils are wound around these teeth, typically arranged into three phases, and when an alternating current is applied to the phases in sequence, the stator produces a rotating magnetic field.
In an induction motor, this field induces an electric current and a corresponding magnetic field in the rotor, causing it to rotate and generate torque; in a permanent magnet motor, the magnets are located on the rotor. The rotating magnetic fields of the stator effectively attract and repel that of the rotor, generating the rotation.
There are two main types of permanent magnet motor: surface permanent magnet (SPM) and interior permanent magnet (IPM). As their names suggest, the former’s magnets are on the outside of the rotor, the latter’s are on the inside, and are positioned in different orientations. This results in very focused magnetic fields that when carefully aligned can produce greater flux and therefore higher torque.
“IPM motors require very precise angular control, because every degree of rotation will result in a different amount of magnetic flux,” says Hutt. “As a result, they are very complex to control as the orientation of the field vectors are critical to effective control. However, the benefit of IPM motors is that they generate additional reluctance torque.”
Combustion engine dynos can be upgraded to test high speed motors with a transmission such as the tGear module from Tectos (middle) which sits between the existing dyno, left, and the unit under test, right (Courtesy of Tectos)
As well as increasing the magnetic flux density in the air gap, IPM motors can also generate additional torque using the reluctance principle. This requires more complex control and the injection of an additional q-axis current offset by 45o to the main phase currents to generate reluctance torque.
The advantage of this over an SPM machine is the ability to generate higher overall rotor magnetic fields with less magnetic material as well as increase and decrease the rotor’s magnetic field by controlling the q axis over the motor’s operating speed range. This reduces high-speed losses and the need for field weakening.
IPM motors can also be developed to achieve higher speed ranges and power, with lower inertia, which allows for better response and control, whereas SPM motors are used mostly in dyno machines as they are much easier to control.
Testing high-speed motors
As dynos have transitioned from testing IC engines to hybrid and electric powertrains, the biggest development challenge has been catering for the high speeds of the motors. A typical IC engine operates at 8000 rpm, whereas some of the latest motorsport motors can reach 30,000 rpm.
“This is definitely one of the toughest challenges when testing electric powertrains,” says Martin Monschein at AVL. “At these higher speeds, any imbalances within the motor or rotating shafts are hugely amplified. You also have to develop a system that can handle the high-power DC loads, which can reach up to 1000 V.”
High motor speeds demand high switching frequencies from the inverter, which causes further challenges for test rig suppliers. An inverter converts the DC electrical power from the battery emulator to an AC output, as the motor requires three-phase AC to generate torque and motion. To transform the constant DC signal into a sine wave AC signal, the DC source is switched directionally across high-frequency, highpower switches.
Pulse width modulation (PWM) can then be used to vary the pulse periods of each switching event, allowing for the continuous variation of the current to generate the AC waveform. With higher
motor speeds, the fundamental electrical frequency increases, as must the frequency of switching events to maintain the quality of the AC waveform.
“The PWM switching frequency is essentially the switching pulse rate required to generate the AC waveform at the electrical frequency required by the motor,” explains Peters. “The higher the switching frequency, the higher the resolution of pulses and therefore the more accurate the resulting AC waveform will be.
“However, beyond typical resistive losses, every switching event induces losses in the inverter. At low switching frequencies, the losses in the inverter are minimal, but those resulting from the motor can increase. On the other hand, at high switching frequencies the motor losses are reduced, but they increase in the inverter.” Hutt adds, “Adjusting the amplitude of the AC waveform allows the inverter to control the torque of the motor, while the frequency of the waveform controls the speed. Those parameters can be adjusted according to how you want to control the system.”
Switching technology has advanced significantly in recent decades. IGBTs (insulated gate bipolar transistors) are common; however, silicon carbide (SiC) has emerged as a new technology owing to its lower losses and higher switching frequencies. The technical challenges of testing faster and more powerful powertrains mean that advanced permanent magnet motors and SiC technology have also been adopted in the latest powertrain dynos.
“SiC is more efficient at switching power and allows higher switching frequencies of around 100 kHz, whereas a typical IGBT is capable of a switching frequency of only about 10 kHz,” explains
Monschein. “The reduced losses of SiC has a secondary benefit because there is less heat rejected, so cooling systems can be smaller.” Back-to-back motor testing As well as upgrading the inverters to SiC, the rest of the dyno also needs upgrading to be able to cope with the substantial loads resulting from the high-speed motors. This can cost millions of dollars, so to try to avoid these costs, one strategy is to test motors against themselves.
Dynos and test powertrains are bidirectional, so can act both as motors and generators. During a test at stable speeds, they operate in the opposite way to each other, so when one is motoring, the other is generating. It is therefore possible for a manufacturer to test their motor design against another, similar motor that acts as the dyno. “In theory, it is possible to test a motor against itself because the motor acting as the dyno generator can achieve the same power and only has to deal with the losses, so it actually has a slightly easier time being a generator rather than a motor,” explains Peters.
“But this back-to-back approach can be challenging during a new product’s development phase. Reliable hardware and control are required on the dyno side to allow for stable data to be generated quickly, which is key during the initial prototype development phase. To achieve that, it is often better to test a prototype with an established dyno.” Back-to-back testing can be more
attractive for later-stage durability testing, where mileage can be accumulated on two test samples.
Another approach is to upgrade existing combustion dynos with a transmission that can amplify the input speed of the dyno. “Many companies have invested heavily in their dyno technology, but they are not suitable for high-speed motor testing,” Christoph Hirt from testbed supplier Tectos points out. “Converting these testbeds to be compatible with e-motors requires a major overhaul, which can cost millions of dollars. That’s why we developed tGear, a specialised gearbox that magnifies the input speed of IC engine testbeds up to 30,000 rpm.”
The unique gear design of the tGear box achieves transmission ratios of 1:2 to 1:7 with a maximum torque of 1800 Nm. Unlike industrial testbed gearboxes, the tGear solution has no axle offset between the input and output shafts, resulting in a compact design measuring 474 mm long x 590 mm wide x 1105 mm high and weighing only 450 kg. “Customers can also use tGear to conduct powertrain testing,” Hirt says. “With no axle offset, the unit can be easily switched around and operated in reverse, shifting the high speeds of the e-motor down to lower speeds that can simulate the rotation of the wheels. This can then be used to test the main gearbox, differential and engine assembly in a variety of configurations.”
This comes with its own challenges however, as it can add significant inertia to the system, which is undesirable particularly when transient speed changes are required during a test. “Motors for hybrid applications experience many transient speed changes owing to gearshifts. And in electric series like Formula E there are also rapid changes in speed caused by wheel slip events, bumps and kerb strikes,” explains Peters. “A high-inertia system will have to do a lot of work to accelerate and decelerate the motor, which in some cases can cause lag when simulating
those transient responses.”
Formula One teams are now limited to a maximum number of test benches, occupancy hours and operation hours for both engine dynos (top) and ERS dynos (above) as specified by the Sporting Regulations
Motor speeds have reached a peak due to the technical limitations of the inverter switching technology and motor mechanical systems. Developing even higher speed motors is increasingly costly and subject to highly diminishing returns. “I think we have seen the peak of motor speeds,” says Monschein. “Producing motors with these high speeds requires special bearings and balancing techniques, which are expensive, so the tendency at the moment is for motor speeds to come back down again and therefore increase in power. “Even the likes of Formula One don’t fully use the speed range they are allowed to by the regulations. I think they have found a sweet spot between speed, performance and cost.”
Battery emulator
Using a real battery to test the motor and inverter is possible and sometimes desirable, but often not practical due to the need to operate extended test patterns, battery recharge times, thermal management of the battery and battery safety considerations. Therefore, to replicate the behaviour of the battery, battery emulators are used instead. These are high-voltage, bidirectional AC-DC power converters where the DC output voltage and current can be carefully controlled to replicate the behaviour of a physical battery; they can also absorb current from the device under test. They can typically be programmed to simulate the characteristics of different battery chemistries, sizes and manufacturing processes.
“Ultimately, the battery model results in a series of voltage and/or current setpoints that need to be supplied throughout a drive cycle,” says Roberts. “That information can come from real test data or an offline model; our CADET control system then sends those setpoints to the battery emulator. “For example, a team might want to simulate that a certain proportion of cells has gone into thermal runaway, so only 87% of them are working. That could translate to 820 V instead of 1000 V, so that will be the new setpoint the emulator needs to provide to the e-motor on the dyno.”
Graph of how the DC internal resistance varies with the state of charge and temperature for a high-performance battery (Courtesy of Bold)
During a race, the voltage and power of the battery varies significantly. On the start line, the battery is typically charged to less than 100% so that it can accept power during regen at the first braking points, otherwise drivers would not be able to brake with the electric recovery system at all. “When the battery is at a high state of charge [SoC], there is more power available,” says Bernat Carreras at Bold. “As the race goes on and the voltage falls with lower SoC, there is a lower power capability. This is due to the battery getting closer to the minimum voltage limit of the cell and increased current to satisfy the power demand at system level.
“At less than about 20% SoC, the internal resistance increases significantly, reducing power even more. On top of that, a motorsport cell has to provide aggressive high-power peaks in charge and discharge through decelerations and accelerations respectively, which lead to a relatively quick loss of capacity and power capability, also known as the state of health [SoH]. “The available discharge power is limited or derated to manage the SoH through the championship events for the different participants so that they all get the same performance. That reduces the race pace but avoids degrading the performance of the battery, which is the constant trade-off in championships like Formula E.” Once the test conditions for the e-motor under test has been defined, they are compiled into a test script that controls the battery emulator to test the e-motor through the required duty cycles. As well as e-motor testing, a battery emulator can also be used to test battery packs, acting as a DC source and sink to replicate the effect of current flowing in and out of the e-motor system to the battery pack.
“We work with the customer to develop the series of tests that will be used to stress-test the battery pack over a wide range of operating conditions,” says James Miller at Mahle Powertrain. “That could include varying the ambient conditions, the amount of cooling from the thermal management system and the severity of the charge and discharge cycles. “Typically, the battery cells will have recommended operating limits specified by the cell manufacturer, but they can vary depending on the application, so customers often want to understand those limits and stresstest them a bit to unlock potential performance.”
However, racing teams often prefer not to simulate the behaviour of the battery, and instead use the battery emulator to supply DC current at a fixed voltage setpoint. That ensures stable conditions, allowing them to more accurately determine the influence of other parameters on the powertrain.
This engine and powertrain test cell at Xtrac was designed and built by Sierra CP (Courtesy of Sierra CP)
“The SoC model is dependent on where you are during a test cycle, so if you are testing a particular device, you want that to be linked to a specific point within the test cycle,” explains Hutt. “You don’t want to be testing something at 40 V because the battery model says so, when you actually need to be testing at 60 V. So I always think the test model should be held within the actual control automation platform rather than something that is running autonomously within the battery emulator.”
Setting up a dyno
Before testing can begin, the dyno needs to be set up and calibrated. This will include inputting base parameters such as the phase resistances and electromotive force constant into the inverter’s current control model, as well as establishing the relevant CAN protocols. The rig will then conduct some shakedown testing, similar to a racecar on track, cycling the rig to ensure that all the flows are at the correct pressure and that the various sensors, alarms and shutdown procedures are working before the rig is operated in anger. “There’s much more of an emphasis on front-loading the test programme now,” says Monschein. “Dyno testing is expensive and this, together with the new restrictions for the likes of Formula One, mean you have to make sure everything is working properly to get results fast during a test.”
Test programmes
The test programmes for the e-motor will vary considerably depending on the category, the regulations and the available budget. For series that use spec components, teams might conduct some validation testing, but otherwise testing is relatively minimal. If though they have a choice between several types of motors or inverters, they will probably test all the options and select the one with the highest performance, even if that is only by a few percent. For championships that allow development, the r&d testing programmes can be comprehensive and consist of a mixture of performance and durability testing.
For a new powertrain, the first test is commonly a spin check. This is where the motor is spun up through different speed ranges and the mechanical behaviour of the system is analysed. For example, the functionality of the bearing pack can be measured through drag and temperature measurements, and rotor balance can be assessed through vibration measurement. Stability of the position feedback to the inverter can also be recorded before powered operation begins. Another key activity is mapping the motor within the inverter, which is the EV equivalent of an engine map. “The goal is to optimise system efficiency across the operating range, and the highest measured torque accuracy,” Peters says. “It is common to tune a range of parameters such as phase angles, torque-to-current scalars and inverter switching frequencies, and then repeat and adjust them for different bus voltages and temperatures.
An e-motor testbed at AVL (Courtesy of AVL List)
“For a high-performance powertrain operating at lower loads, it can be a real challenge to achieve high levels of efficiency, because of losses such as drag making a larger proportion of the power. But in the higher power ranges, that’s where the system can achieve those headline efficiency figures. “Running efficiently doesn’t just increase the range of the battery, it also means the powertrain components are not as hot and require less cooling, so it is a double win. However, overall the most sensible focus is on optimising efficiency at a system level, so it’s about establishing the best blend of efficiencies across the entire powertrain, rather than focusing on maximising the efficiency of a single component.”
This motor and efficiency mapping can now be completed in simulation using electromagnetic software tools, which can predict motor efficiency and provide initial motor control parameters that are key for the inverter. Simulation engineers can generate a map as a baseline for the inverter, which can reduce the originally large matrix of physical testing to a smaller activity. However, the feedback loop from testing is still highly valuable for honing simulation tools and future phases of motor design. Once the most efficient system has been defined, the test programme shifts to durability testing. That can involve simply running the e-motor for long durations of time or lap simulations where the transient response during challenging tracks is tested.
Condensing testing
Since 2021, Formula One has been restricted in the amount of power unit bench testing teams can complete. For engine as well as energy recovery system dynos, teams are limited to a maximum number of test benches, a maximum number of occupancy hours and a maximum number of operation hours. In the past, powertrain assemblies could be dyno-tested 24/7. “Teams can run unlimited testing on individual components, but any testing of a powertrain assembly comes under the restrictions defined by the sporting regulations,” Monschein says. “Teams are trying to reduce the amount of validation testing to avoid using up too much of their allocated testing time. This has shifted the focus slightly to achieving more realistic component testing, so we have had to get more creative with our component testing capabilities to make it as realistic as possible to front-load the r&d at the component level, before testing the powertrain assembly.”
Another tactic has been to condense testing methodologies. This is where the required speed, load and power profiles that need to be tested are classified into areas and then the equivalent runtimes are defined. “Rather than running certain points individually, you can establish a set of tests that optimise the testing of all these combinations with the least amount of runtime on the dyno,” says Monschein. “Let’s say you run a speed profile between 10,000 and 14,000 rpm at full load and there are some braking zones. I could then run at 12,000 rpm for a certain amount of time at a particular load, which could be equivalent to running the original speed profile. “These testing methods have been around for a long time, but their importance has grown recently because of the drive to conduct more efficient testing.”