Search

The aim of this paper is to make use of PIV and high-speed PIV in a research engine of moderate tumbling ratio in order to analyze both the spatial structure of the flow and its temporal evolution during series of consecutive cycles. Appropriate analyzing tools are introduced, and four different points are addressed: (1) the chain of events driving the generation of the three-dimensional mean tumbling motion is investigated; (2) a Lagrangian analysis of the roll-up of the tumbling jet in individual cycles demonstrates a strong cycle to cycle variation during the compression phase (the rms of the position of the jet front being approximately 10% of the piston stroke); (3) focussing on the “breakdown” phase, phase invariant proper orthogonal decomposition enables us to distinguish cycles according to their structure near top dead center (TDC). We show that when the coherent energy of the flow is conserved, there is no increase in the fluctuating kinetic energy; (4) finally, the phase-averaged Reynolds stresses is decomposed into a contribution of the in-cycle coherence and the turbulence carried by the flow states. Approximately 30% of the fluctuating kinetic energy is due to cycle to cycle fluctuations in this chamber near TDC.

In this work, snapshot proper orthogonal decomposition (POD) is used to study a pulsed jet in crossflow where the velocity fields are extracted from stereoscopic particle image velocimetry (SPIV) results. The studied pulsed jet is characterized by a frequency f = 1 Hz, a Reynolds number Re j = 500 (based on the mean jet velocity $${\overline{U}_{j}}$$ = 1.67 cm/s and a mean velocity ratio of R = 1). Pulsed jet and continuous jet are compared via mean velocity field trajectory and Q criterion. POD results of instantaneous, phase-averaged and fluctuating velocity fields are presented and compared in this paper. Snapshot POD applied on one plane allows us to distinguish an organization of the first spatial eigenmodes. A distinction between “natural modes” and “pulsed modes” is achieved with the results obtained by the pulsed and unforced jet. Secondly, the correlation tensor is established with four parallel planes (multi-plane snapshot POD) for the evaluation of volume spatial modes. These resulting modes are interpolated and the volume velocity field is reconstructed with a minimal number of modes for all the times of the pulsation period. These reconstructions are compared to orthogonal measurements to the transverse jet in order to validate the obtained three-dimensional velocity fields. Finally, this POD approach for the 3D flow field reconstruction from experimental data issued from planes parallel to the flow seems capable to extract relevant information from a complex three-dimensional flow and can be an alternative to tomo-PIV for large volume of measurement.

We define and illustrate a cluster-based analysis of cycle-to-cycle variations (CCV). The methodology is applied to engine flow but can clearly be valuable for any periodically driven fluid flow at large Reynolds numbers. High-speed particle image velocimetry data acquired during the compression stroke for 161 consecutive engine cycles are used. Clustering is applied to the velocity fields normalised by their kinetic energy. From a phase-averaged analysis of the statistics of cluster content and inter-cluster transitions, we show that CCV can be associated with different sets of trajectories during the second half of the compression phase. Conditional statistics are computed for flow data of each cluster. In particular, we identify a particular subset associated with a loss of large-scale coherence, a very low kinetic energy of the mean flow and a higher fluctuating kinetic energy. This is interpreted as a good indicator of the breakdown of the large-scale coherent tumbling motion. For this particular subset, the cluster analysis confirms the idea of a gradual destabilisation of the in-cylinder flow during the final phase of the compression. Moreover, inter-cycle statistics show that the flow states near TDC and in the measurement zone are statistically independent for consecutive engine cycles. It is important to point out that this approach is generally applicable to very large sets of data, e.g. generated by PIV or LES, and independent of the considered type of information (velocity, concentration, etc.).

A new method employing instantaneous 3D coupled measurements of planar laser-induced fluorescence and particle Image velocimetry is developed to compute 3D fluid fluxes. The three-dimensional fields of concentration and velocity are used for identifying the vortex structures and for the evaluation of the fluxes at the interface between a jet exiting in a crossflow. This approach is helpful for the 3D estimation of the momentum and scalar transport at the turbulent/non-turbulent interface and will help understanding the mixing process.