Torque Vectoring in Electric Vehicles with In-Wheel Motors

In the context of increasing vehicle electrification, a new configuration has come up: the electric vehicle with four independent in-wheel motors. In this configuration, there is more space available for the body interior (because there is no inboard engine), there are less transmission components (because there are no semi-axles) and no need for a mechanical differential (it is implemented electronically). Its biggest advantage is the independence between the motors opens up new possibilities in terms of vehicle dynamics, allowing an easier implementation of torque vectoring strategies. In this respect, this thesis consists in the development of a torque vectoring controller for increased stability and response for an electric vehicle with in-wheel motors. First, a literature review is done, studying different direct-yaw-moment controls and strategies for torque vectoring controllers. Then, a subject vehicle, the laboratory prototype, was modeled in the multi-body dynamics Adams Car software. The electric powertrain was taken into account with its proper inertia and mass in the wheels. The powertrain torque map and the controller were done in Matlab. In this way, the analysis was carried out in a co-simulation environment. The torque vectoring was done with a yaw rate based PI controller. Different saturations were added to account for the maximum motor torque and maximum allowable wheel torque to avoid wheel spin. Different maneuvers were performed to test the controller performance: step steer, ramp steer, double lane change and constant radius cornering. Controller robustness was also tested with different vehicle configurations (forward-wheeldrive and rear-wheel-drive) and in low friction surfaces. The proposed controller was able to enhance vehicle response: it successfully extends the linear region of the steering response and decreases the excessive understeering behaviour of the car, making it almost neutral steer even in conditions with longitudinal acceleration. The controlled vehicle is more agile to turn in and has less overshoots of yaw rate. It works regardless of vehicle configuration and decreases the power under/oversteering, typically present when there is significant longitudinal acceleration in a non all-wheel-drive vehicle. However, the controller depends on the correct estimation of the tire-road friction coefficient μ. In situations where it is overestimated, the controller imposes a too high yaw moment on the vehicle, causing it to drift out. To further develop the controller, an μ estimator can be implemented alongside a secondary vehicle side slip angle controller, that could intervene in critical conditions as a safety device while the yaw rate torque vectoring controller continuously actuates to enhance vehicle response. The overall objective of modelling an electric vehicle with in-wheel motors and implementing a torque vectoring strategy were accomplished. The developed co-simulation tool can even be further expanded to include other vehicle dynamics controllers such as traction control or ABS.