Torque Vectoring Is Now a Core EV Drivetrain Feature
When BMW launched the i4 M50 with independent front and rear electric motors, and when Rivian and BYD built four-motor architectures where each wheel receives individually commanded torque, the test engineer’s challenge changed fundamentally. It was no longer sufficient to test a single motor’s power, 效率, and NVH in isolation. The question became: how do you validate the interaction between two or more motors whose torque commands change thousands of times per second in response to a yaw moment controller?
Torque vectoring — the deliberate asymmetric distribution of torque between left and right wheels, or between front and rear axles — delivers handling benefits that mechanical limited-slip differentials cannot match. But certifying that a torque vectoring eAxle actually delivers the commanded torque at each wheel, with the specified response time and without unexpected coupling between the two channels, requires a test bench setup that goes significantly beyond a conventional e-motor endurance rig.
Anatomy of a Dual-Motor eAxle Test Bench
A torque vectoring eAxle test bench must simultaneously control and measure both output shafts. The most common architecture uses two independent load dynamometers, each coupled to one output shaft of the eAxle under test. The dynos are controlled in real time by a HIL (Hardware-In-the-Loop) system that simulates vehicle dynamics — the wheel speeds fed back to the vehicle’s torque vectoring controller are computed from a seven-degree-of-freedom vehicle model running in real time.
- Independent shaft dynamometers: Both dynos must have matched torque measurement bandwidth (≥ 200 赫兹), and the torque transducers must be calibrated traceably. A calibration drift between the two channels of as little as 0.5% will introduce a spurious yaw moment signal that corrupts the torque distribution validation.
- Wheel speed simulation: The HIL model generates realistic wheel speed signals (能 / Flexray) that replicate the slip conditions of a vehicle in a cornering manoeuvre — a key input for the torque vectoring controller’s slip-ratio observer.
- Driveshaft stiffness representation: Actual vehicle driveshafts have finite torsional stiffness (typically 500–2,000 Nm/rad). The test bench coupling must either replicate this stiffness with purpose-made compliant couplings or compensate for it in the dyno controller’s torque reference loop.
Core Test Items for Torque Vectoring Validation
1. Static Torque Distribution Accuracy
With the vehicle controller commanding a specific left-to-right torque split ratio (例如。, 60:40), the bench measures the actual torque at each output shaft over a 10-second steady-state interval. The test is repeated at 10 speed-torque operating points spanning the full envelope.
关键指标: Absolute torque error at each shaft (纳米); torque split ratio error (% of commanded ratio); 重复性 (σ) over 10 consecutive runs at the same operating point.
2. Dynamic Torque Step Response
The controller is commanded to step from a symmetric 50:50 distribution to an asymmetric 80:20 (or full one-sided) split in minimum time. The bench records the step response of each shaft’s torque, and the yaw moment generated by the torque difference is calculated from the known track width of the target vehicle.
关键指标: Rise time from 10% 到 90% of target torque difference (目标: ≤ 20 ms for performance applications); torque overshoot (% of target step amplitude); coupling — torque change on the non-commanded shaft during the step.
3. Yaw Moment Generation Accuracy
At a defined vehicle speed (例如。, 80 km/h on-axis), the torque vectoring controller is commanded to generate a specific yaw moment (例如。, +500 Nm-m about the vehicle’s centre of gravity). The bench, combined with the vehicle dynamics HIL model, validates whether the actual yaw moment delivered by the torque asymmetry matches the commanded value within the specification tolerance (typically ±5%).
关键指标: Steady-state yaw moment error; yaw moment gradient vs. steering angle over ±20° of simulated lateral acceleration demand; yaw moment consistency under tyre slip variation (simulated by the HIL model).
4. Cross-Coupling and Interference Rejection
When one motor is rapidly changed in torque command, the effect on the other shaft’s torque output must be below the specified interference threshold. Poor decoupling between the two inverter control loops produces a “cross-talk” torque that the driver perceives as an unexpected yaw perturbation. The bench applies a PRBS (Pseudo-Random Binary Sequence) torque perturbation to one shaft and measures the contamination transfer function to the other shaft.
关键指标: Cross-coupling transfer function magnitude at 10 赫兹, 50 赫兹, 100 赫兹; phase angle of coupling; bandwidth within which cross-coupling is below −40 dB (i.e., < 1% contamination).
5. Thermal Symmetry Under Asymmetric Loading
A sustained torque vectoring manoeuvre — for example, a circuit lap with continuous cornering — loads one motor significantly more than the other. The test bench holds a sustained asymmetric torque command (例如。, 80% on the outer wheel motor, 20% on the inner) for a minimum of 10 minutes and monitors winding temperatures on both sides. If the cooling circuit is shared (common coolant rail), hot-side motor temperature must remain within the thermal limit even as cold-side motor generates minimal heat. Coolant flow allocation between the two motors must be characterised under worst-case single-sided loading.
Integration with Vehicle Dynamics Simulation
The highest-value torque vectoring tests are run with the eAxle controller operating in closed loop against a full vehicle dynamics model in real time. The HIL rack simulates tyre forces (via a Pacejka Magic Formula or similar), suspension kinematics, and driver steering input. The controller’s yaw stabilisation responses to simulated oversteer and understeer events are validated against the vehicle dynamics engineering team’s targets — all before the prototype vehicle is built.
ECONOTESTS can supply dual-shaft eAxle test benches with matched dynamometers, HIL integration, and the instrumentation chain needed for torque vectoring validation. Whether your programme involves a luxury performance eAxle or a cost-sensitive SUV dual-motor rear axle, 与我们的工程团队交谈 about the right test architecture.
