Transient Control of SIDI-Engine with VGT and VVT

Abstract: Today's development of spark ignited (SI) engines is to a large extent focused on downsizing, i.e where a large naturally aspirated engine is replaced with a smaller turbocharged engine with the same output power. This gives a fuel consumption benefit but worsen the transient response since the torque build-up gets dependent on the turbocharger dynamics. To mitigate this problem, technology like variable geometry turbine (VGT), which has been widely used on diesel engines, and variable valve timing (VVT) can be used. In this thesis, it has been investigated how VVT and VGT should be controlled during a torque transient in order to achieve fastest possible torque build-up. The engine that has been considered is a 2 liter, direct injected, SI-engine equipped with a turbocharger with VGT. This has been investigated by applying an optimization routine from Matlab to a GT-Power model of the engine and from measurements on the actual engine. The optimization in GT-Power showed that measures which takes both the naturally aspirated part and the boosted part of the transient into account increases with the torque, and that the optimal transients had an approximately linear relationship between intake and exhaust pressure. Two control strategies based on these results were tested on the engine. In the first control strategy, the VVT and VGT were controlled such that the product of volumetric efficiency and exhaust pressure was maximized. This resulted in quite moderate torque build-up and no improvement compared to using a stationary calibration in transients. In the second control strategy the VGT was used to closed loop control the exhaust pressure towards an exhaust pressure reference, dependent on the current intake pressure. This, combined with controlling the VVT according to the stationary calibration, resulted in an apparent improvement in the torque build-up. This strategy was found to give good performance also at slightly lower oil and water temperature than nominal but the sensitivity towards gain errors in the intake or exhaust pressure measurements were quite high. This sensitivity can be handled by limit how much the control actions from the exhaust pressure controller are allowed to deviate from the stationary calibration.