Dynamic modeling and model predictive control of an RCCI engine

Reactivity controlled compression ignition (RCCI) is a low temperature combustion strategy that offers one of the highest reported indicated thermal efficiencies for internal combustion engines, while having ultra-low nitrogen oxides (NOx) and soot emissions. The complex nature of RCCI makes it challenging to control combustion for an optimum heat release shape at broad engine operation with low cyclic variability and without exceeding maximum allowable in-cylinder pressure rise rate. This study aims at developing a control oriented model (COM) and a model predictive controller (MPC) to adjust combustion phasing, including crank angle by which 50% of fuel mass is burnt (CA50) and load, including indicated mean effective pressure (IMEP) during both steady-state and transient RCCI operations. A new COM is developed using a combination of physics-based and empirical models. An MPC with a 5-cycle prediction horizon is developed and implemented on an experimental RCCI engine setup. The IMEP and CA50 are controlled by adjusting injected fuel quantity, dual-fuel premixed ratio (PR), and start of injection (SOI) timing. To extend the controller operating range, switched MPCs are developed using PR as the scheduling variable. In addition, a sensitivity-based control strategy is developed to select between PR and SOI as the control variable. Experimental validation results show that the designed controller can track desired CA50 and IMEP with less than 1.4 CAD and 15 kPa tracking errors on a range of RCCI operation.