Cycle-to-cycle combustion variability modelling in spark ignited engines for control purposes

Abstract

A control-oriented model of spark ignition combustion is presented. The model makes use of avaliable signals, such as spark advance, air mass, intake pressure, and lambda, to characterize not only the average combustion evolution but also the cycle-to-cycle variability. The conventional turbulent flame propagation model with two states, namely entrained mass and burnt mass, is improved by look-up tables at some parameters, and the cycle-to-cycle variability is estimated by propagation of an exogenous noise with a normal probabilistic distribution at the turbulent and laminar flame speed, which intends to simulate the unknowns at turbulent flow, temperature distribution, or initial kernel distribution. The model is able to estimate which is the expected variability during the combustion evolution and might be used online for characterizing the time response of closed-loop control actions or it can be used offline to improve the control strategies without large experimental test campaigns. Experimental data from a four-stroke commercial engine was used for calibration and validation purposes, demonstrating the capabilities of the model in steady and transient conditions.