Development of a control-oriented model to optimise fuel consumption and NOX emissions in a DI Diesel engine

This paper describes a predictive NOX and consumption model, which is oriented to control and optimisation of DI Diesel engines. The model applies the Response Surface Methodology following a two-step process: firstly, the relationship between engine inputs (intake charge conditions and injection settings) and some combustion parameters (peak pressure, indicated mean effective pressure and burn angles) is determined; secondly, engine outputs (NOX and consumption) are predicted from the combustion parameters using NOX and mechanical losses models. Splitting the model into two parts allows using either experimental or modelled combustion parameters, thus enhancing the model flexibility. If experimental in-cylinder pressure is used to obtain combustion parameters, the mean error of predicted NOX and consumption are 2% and 6% respectively, with a calculation time of 5.5 ms. Using modelled parameters reduces the calculation time to 1.5 ms, with a penalty in the accuracy. The model performs well in a multi-objective optimisation, reducing NOX and consumption in different amounts depending on the objective of the optimisation.