Mean Value Modeling of intake and exhaust systems of automotive engines: models identification and related errors

Control and management issues are playing an important role in automotive applications to improve engine and powertrain performance and to lower specific fuel consumption and pollutant emissions. Mean Value Engine Models (MVEM) based on “grey-box” or faster “black-box” algorithms are usually adopted for control applications since they allow for real-time simulation of actual automotive engines. However, “grey“ or “black-box” models usually requires a significant amount of data (usually from experimental investigations) for the identification of sub-models of each component: these data are often not available with the same degree of accuracy and, moreover, are usually gathered in operating conditions that are far from real on-engine conditions. In the paper a brief description of the identification procedures for MVMs of intake and exhaust systems developed by the authors is given. Uncertainties and shifts that affect results given by MVMs is analysed with reference to the accuracy of the identification procedure. To this extent a MVM of an actual turbocharged engine was used to analyse the effects on calculated outputs of defined shifts introduced in the identification parameters of intake/exhaust systems sub-models (with particular reference to sub-models of compressor, turbine and EGR valves). Obtained results are reported in the paper to point out what sub-models have major effects on MVMs outputs and to study the degree of accuracy needed for their identification.