Proper orthogonal decomposition of in-cylinder engine flow into mean component, coherent structures and random Gaussian fluctuations

Abstract

A snapshot proper orthogonal decomposition (POD) is performed from 2D time-resolved PIV measurements obtained in the tumble plane of a Spark Ignition engine flow. Based on this filtering approach, the in-cylinder flow field is decomposed into a mean part, a coherent part and a turbulent incoherent part. The analysis of the one-point statistical moments of orders 3 and 4 (skewness and flatness coefficients) as well as the analysis of the probability density function of the velocity field show that the POD extracts an incoherent random velocity field having Gaussian distribution properties. These small amplitude background fluctuations are also homogeneous. We then demonstrate the ability of the POD procedure, based on an energy criterion, in separating the coherent part and random Gaussian fluctuations. Based on this POD flow partitioning, providing a new triple decomposition of the instantaneous turbulent flow, the flow engine cycle-to-cycle variations associated with each velocity field contribution are then accessed. Thus, one of the first full field quantitative analyses of the cyclic variabilities associated with each part of the in-cylinder flow field is provided from the analysis of the corresponding POD temporal coefficients.

Acknowledgements

The authors wish to thank M. Bonnety for its contribution to the experimental part of this work. The authors would like to gratefully acknowledge RENAULT monitored by M. Dumont and M. Dupont for their financial support and for providing us a Renault transparent car engine.