Abstract
The analysis of turbulent complex flows requires the evaluation of turbulence length scales (in addition to other turbulent features). The already published experimental results reveal that the measurement of the Taylor length scale by two-point laser Doppler velocimetry (LDV) necessitates some care. The experimental uncertainties are mainly governed by the finite probe volume length and the portion of the correlation curve used to fit a matching parabola (H. Belmabrouk, M. Michard, Experiments in Fluids 25 (1998) 69–76). The importance of these parameters is investigated theoretically in the present paper first to any flow and then to the in-cylinder flow in a motored engine. In the first case, the approach consists on simulating the effect of the two parameters given a modelled correlation function. It appears clearly that the finite probe volume length has no significant effect on the measured integral scale but has an important effect on the Taylor scale. The results reveal also that the relative uncertainty on the Taylor scale is governed not only by the probe volume length and the procedure of the data processing but also by the ratio of the integral to the Taylor scales, that is to say, by the Reynolds number. The general investigation is then applied to the flow in a reciprocating engine. To achieve this study, a second-order model of compressed turbulence is used to predict the theoretical evolution of the integral and Taylor length scales during the compression and expansion strokes. This prediction coupled to the previous analysis permits to estimate the uncertainty on the measured length scales by simulating the finite length of the probe volumes. Finally, the application of the theoretical analysis in a real motored engine is presented and an attempt to remove bias from experimental Taylor scale has been made.