The idea of adopting a part-time AWD driveline is aimed at enhancing the overall performance of the vehicle by using all the available friction every time it is needed.
The sporting spirit that characterises a high performance car can be observed in certain technical solutions. The power distribution on the rear wheels is the simplest example of that. It is known well that rear wheel drive (RWD) vehicles are fun to drive and faster in their reactions. Unfortunately they are also less intuitive and difficult to control because of their natural oversteering behaviour. The idea of maintaining a RWD driveline in the future is not wise, because it would imply an excessive tyre dimension increase to let the driver use all engine power in many cornering and low friction conditions. The choice of adopting a part-time all wheel drive (AWD) driveline comes from the will to enhance the overall performance by using all the available friction every time it is needed. It has to be kept in mind that a normally aspirated motor of a sports car that supplies 500-600 Hp now, will supply 700-800 Hp in the near future. However, the proposed driveline should not worsen the weight characteristics (mass and load distribution) that make an RWD vehicle better than others.
Taking these constraints into consideration, a new driveline system has been designed. It is based on an RWD driveline with a semi-active differential, to which a controlled wet clutch is added which directly connects the engine to the front differential. This device allows for distribution of drive torque between two axles. It can be understood that in such a device the torque distribution doesn’t depend only on the central clutch action, but also on the engaged gear. This system cannot work in the whole gear range because of thermal problems due to kinematical reasons. So the centre clutch controller has to consider the gear position too. The control algorithms development has to be carried out using a vehicle model which can precisely simulate the handling response, the powertrain dynamics and the actuation system behaviour. Such a modelling precision requires the development of a customised powertrain model library in Matlab / Simulink.
The driveline scheme is an adaptation of an RWD vehicle with rear engine and gearbox (Figure 1) to an all wheel drive car. The free side of the engine crankshaft is directly connected to the front free differential through an electronically controlled wet clutch (Figure 1). This way, it is not necessary to design a new gearbox and a rear differential device to place the controlled clutch after them, which is very useful in terms of saving costs. Such a mechanical solution (patented) can be adopted on an RWD vehicle with the engine placed both on the front axle and on the rear one. From a kinematical and dynamical point of view there is no difference.
Because of the presence of a constant gear ratio TF (front differential) between the front wheels and the center clutch, the center clutch slip is a function of the engaged gear. In fact, rotational speed of the clutch plates connected to the front differential is a function of the vehicle speed only, while the rotational speed of the clutch plates connected to the engine is a function of both the vehicle speed and the engaged gear ratio. Looking at how a clutch works, it can be understood that the plate connected to the engine crankshaft has to rotate faster than the other one while a torque transfer to the front axle is needed. If this condition is required for every engaged gear, the front differential ratio TF has to be designed to assume a value less than the product of the gearbox and final ratio Ti TR even for the latest gear. That means the center clutch can be hardly stressed because of the high slippage when the first gear is engaged, so that thermal and consuming problems would be very critical. Because of those considerations the device has been designed to work in a possible 4WD configuration only when the engaged gear is equal or less than the four. Such a choice can appear limitative in comparison to a common part-time 4WD driveline because it doesn’t allow distribution of the driving torque between the two axles when a gear greater than the fourth one is engaged. Instead, this solution precisely satisfies the requirements of improving the handling performance of a sport RWD car only when the RWD driveline would saturate the rear tyres.
Besides, some words have to be spoken regarding the power-off condition. In fact, it can be understood how any intervention of the centre clutch involves a transfer of driving torque only, whatever the nature (driving or dragging) of the engine torque, every time the engaged gear is equal or less than the fourth one. The torque distribution that would be generated wouldn’t absolutely be acceptable, because a driving torque on the steered wheels would cause an unsuitable oversteering moment, and the car stability in a power off maneuver is controlled with the rear differential.
As mentioned above, the original rear wheel drive vehicle is already equipped with a semi-active differential; the only difference between the two powertrain layouts is the presence of the centre clutch (Figure 2). Because of this, a new logic for the centre clutch as well as for the rear differential was added. Obviously, the two subsystems have to communicate with each other to optimise the improvements of the handling performance (Figure 2). The rear semi-active differential logic has been adopted with regard to the original release to make its intervention a function even as the torque is transferred to the front axle. Instead, the centre clutch logic has been completely designed.
The approach adopted to define the reference functions of the logic components was based on an intensive numerical simulation exercise. Due to this fact, the performance of the feed-forward components largely depends on the reliability of the vehicle model. For this reason, a validated 14 degrees of freedom model (IPG CarMaker®) has been used, and it has been integrated with the actuators model and with the control logic. The model has been validated by comparing its outputs to experimental data of a passive demo 4WD vehicle with three differentials.
The control logic structure reflects the primacy of the semi-active differential on the centre clutch. As we have already seen, new control logic for the rear axle management was not designed, but the main structure and philosophy of the original algorithm was retained by adapting it to the presence of another controlled device. Because of this planning choice, the ideal two-way communication between the two device controllers (Figure 2) becomes one-way communication. In other words, only the semi-active differential control logic has to know what the centre clutch is doing. The vice-versa is not provided in order to prevent any algebraic loop. Such a strategy is particularly profitable because it reflects the goals fixed at the beginning of the work itself. Thus AWD is designed to be a simple option to make the vehicle easier to be driven and let the tyres transmit the whole engine power to the ground every time a RWD wouldn’t be able to do that. The aim is not to transform a sportive rear wheel drive car into an all time four wheel drive one. The vehicle must continue to be a RWD as much as possible, and it has to be an AWD only when otherwise the stability would be a compromise. The target quantity for the actuator is constituted, both in the semi-active differential logic and in the center clutch logic (Tref,F in Figure 3), as the sum of a feed-forward contribute and a feed-back contribute, whose parameters depend on the vehicle state and the driving condition.
The center clutch controller is based on an inner loop consisting of a feed-forward subsystem and a feed-back subsystem (Figure 3), and an outer closed loop. The feed-forward (FF) contribute confers a fast reaction to the controller and the feed-back (FB) contribute guarantees precision when the steady-state is reached. The outer closed loop (Oversteering Management in particular) has to guarantee a proper behaviour in some particular. The output quantity of both FF and FB is a torque. These torques are added and the result represents the engine torque to be sent to the front axle.
1.a The Feed-Forward
The FF has to reduce the system response time and it constitutes a map which is a non-linear function of the vehicle state. The torque to be transferred to the front axle (Tf,FF in Figure 3) is calculated as the difference between the instant engine torque reduced to the rear axle and the torque level that saturates the rear tyres. This last quantity is estimated as a triple dependence on the longitudinal speed square (aerodynamics effect), the longitudinal acceleration (pitch load transfer) and the lateral acceleration (tyre combined use).
1.b The Feed-Back
The FB has to mainly assist the FF every time its intervention would be not so precise with respect to the instant external condition. The algorithm is based on some vehicle dynamics principle describing the overall system targets. As mentioned earlier, the RWD configuration is considered the best every time the instant drive torque does not exceed its maximum transmittable value. In fact, if that happens, it would cause a dangerous oversteer. Considering the typical tyre characteristics, a simple way to detect the closeness of the tyre to its limit is to monitor the actual longitudinal slip. When the tyre combined use becomes extreme, this quantity increases and the tyre in-plane force increases too, until a maximum value is reached.
Because of such a physics of the contact force, the FB component compares the rear actual longitudinal slip to two threshold values to keep the actual slip as close as possible to the maximum value, and thus the rear axle close to its limit. The two thresholds are necessary to create relay behaviour of the FB sub-algorithm in order to manage all the uncertainty in the longitudinal slip calculation and in its best value evaluation. The actual longitudinal slip is calculated as a weight average of the inner and outer one.
The control system has been firstly tested simulating both ISO and non-ISO maneuvers. Besides, many laps on a high speed circuit and on a mountain track with different road friction level have been simulated, in order to predict the center clutch stress in every driving condition and to evaluate the performance gain respect to the RWD car. The simulated gain (in term of lap time) is 2.5 percent on dry asphalt and 5 percent on wet asphalt.
