The sensitivity of transient response prediction of a turbocharged diesel engine to turbine map extrapolation

Abstract

Mandated pollutant emission levels are shifting light-duty vehicles towards hybrid and electric powertrains. Heavy-duty applications, on the other hand, will continue to rely on internal combustion engines for the foreseeable future. Hence there remain clear environmental and economic reasons to further decrease IC engine emissions. Turbocharged diesels are the mainstay prime mover for heavy-duty vehicles and industrial machines, and transient performance is integral to maximizing productivity, while minimizing work cycle fuel consumption and CO2 emissions. 1D engine simulation tools are commonplace for "virtual" performance development, saving time and cost, and enabling product and emissions legislation cycles to be met. A known limitation however, is the predictive capability of the turbocharger turbine sub-model in these tools. One specific concern is accurate extrapolation of turbine performance beyond the narrow region populated by supplier-measured data to simulate non-steady conditions, be it either to capture pulsating exhaust flow or, as is the focus here, engine transient events. Extrapolation may be achieved mathematically or by using physics-based correlations, sometimes in combination. Often these extrapolation rules are the result of experience. Due to air system dynamic imbalance, engine transients force instantaneous turbine mass flow and pressure ratio into regions well away from the hot gas bench test data, necessitating great trust in the extrapolation routine. In this study, a 1D heavy-duty turbocharged diesel engine model was used to simulate four transient events, employing a series of performance maps representing the same turbine but with increasing levels of extrapolation, using commonly-adopted methodologies. The comparison was enabled by measuring real turbine performance on the dynamometer at Imperial College London. This testing generated a wide baseline dataset which was used to produce corresponding transient response predictions, and against which cases of increasing degrees of extrapolation could be compared. This paper studies the sensitivity of response time to the degree and technique of the extrapolation applied, demonstrating its importance for reliable transient engine simulations.