Abstract
In this paper, Model Predictive Control (MPC) is applied to control a Waste Heat Recovery system for a highly dynamic automotive application. As a benchmark, a commonly applied control strategy is used that consists of a feedforward based on engine conditions and of two PI controllers that individually control the post-EGR and post-exhaust evaporator temperature. By controlling the temperatures, this strategy deals with the constraints on working fluid temperature and aims to guarantee vapor state at the evaporator outlet. To design the local PI controllers, a frequency domain based loop shaping method is applied. Using a high-fidelity Waste Heat Recovery system model, the performance of the MPC and PI controller is compared in simulations over a World Harmonized Transient Cycle for both cold-start and hot-start conditions. Due to the availability of a model and knowledge of the engine disturbance, the MPC strategy is shown to outperform the reference PI controller, it gives higher energy recovery and guarantees operation within safety margins with respect to physical capabilities of the plant.
KEYWORDS – Energy recovery systems, Rankine Cycle, Diesel engine, PI control, Model Predictive Control