Search Results

Currently, the development of higher specific output and higher efficiency S.I. engines requires better control and knowledge of knock mechanisms. As it is not easily possible to instrument an engine to determine the beginning of fuel auto-ignition, knock modeling by means of 3D CFD simulation, can be a powerful tool to understand and try to avoid this phenomenon [1, 2, 3]. The objectives of the work described in this paper are to develop and validate a simple model of auto-ignition. This model, developed at IFP, is implemented in the 3D CFD code KMB [4, 5]. It is based on an AnB model [6, 7] which creates a ‘precursor’ species transported with the flow in the combustion chamber. When its concentration reaches a limiting value, the auto-ignition phenomenon occurs.

In order to determine the area where knock occurs in a single cylinder engine, an acoustic methodology needs a minimum of four simultaneous pressure measurements in the combustion chamber. A specific cylinder head gasket integrating 12 pressure sensors has been developed and tested. The gasket is based on a bonded multilayer technology including high temperature piezoelectric cells, metallic and insulating sheets and printed circuit films. The total thickness is close to 1.25 mm (1/20 inch) and allows a straight forward substitution of the original gasket without modification. The sensors have large frequency bandwidth (typically 3-100 kHz) and withstand severe conditions (heat, combustion, pressure, vibrations, static pre-stress, electromagnetic fields and shocks). Signal processing adaptation of the dedicated exploitation software has brought good success for the single cylinder prototype, which remains operational after 100 hours of extreme conditions running (high knock).

A methodology which uses simultaneously 3D calculations and advanced experimental tools has been developed in order to characterize air-fuel mixing and combustion of gasoline direct injection engines at every stage of development. The analysis of Mitsubishi GDI engine has been carried out in order to validate this approach. The experimental tools used in this analysis underline the great cycle-to-cycle variability and show that the air-fuel ratio variations at spark plug correlate closely with the fluctuations of combustion starting and development. Despite this variability, average measurements are reproducible and in good agreement with 3D computational results obtained with KIVA-MB code. The common use of both kinds of tools allows to get a very fine understanding of Mitsubishi wall-guided concept.