Statistical analysis of the flow characteristics and cyclic variability using Proper Orthogonal Decomposition of highly resolved LES in internal combustion engines

•Cyclic variability in IC-engines: recent studies and phenomenology.•Classic statistical analysis and rigorous modal POD investigation of highly-resolved LES data.•Discussion of findings.

The results of a highly resolved Large Eddy Simulation (LES) of the flow field in a motored engine have been analysed by means of Proper Orthogonal Decomposition (POD) with the objective of deriving a statistical and modal characterisation of the dominating velocity structures. The analysis has been conducted in a subset of the computational domain, deemed of particular relevance for the development of the instabilities leading to cyclic variability, located in the vicinity of the spark plug. This area has been demonstrated to be sufficiently representative of the dynamics of the flow field and has been examined by means of a fully three-dimensional, phase-independent POD.Investigations based on the analysis of the velocity’s variability depending on the prevailing crank angle position also show that the strongest variations appear during the compression and expansion strokes. The POD weighting coefficients display, not surprisingly, an identical behaviour. On one side, it can be argued that the reduced system obtained via POD in the considered control volume can be conveniently used to derive possibly criteria for the identification of anomalous cycles on a predictive basis. On the other side, however, it has been observed that weaker modes, neglected in the POD analysis on the basis of the global energetic convergence of the eigenmodes, play an important role in promoting and supporting cyclic variability, as also observed in independent experimental studies.