A new methodology for uncertainties characterization in combustion diagnosis and thermodynamic modelling

Abstract

Combustion diagnosis based on in-cylinder pressure signals as well as 0D thermodynamic modelling, are widely used to study and optimize the combustion in reciprocating engines. Both approaches share some uncertainties regarding the sub-models and the experimental installation that, for the sake of accuracy, must be reduced as much as possible in order to obtain reliable results. A methodology, based on the sensitivity effect of such uncertainties on heat release and simulated pressure, is proposed to adjust their values. The methodology is capable of identifying the separate influence of each parameter and to provide a set of values thanks to the Multi-Variable linear regression (MLR) in motoring conditions. The method is flexible enough to deal with different number of uncertainties and can be applied to different engines and thermodynamic models. The final results of the adjustment are validated in combustion conditions, showing an improvement of the apparent combustion efficiency of about 7% with respect to the reference values.