Evaluation Of Integral Length Scales From Experimental Data In A Flow Through Submerged Vegetation.

Paola Gualtieri, Sergio De Felice, Vittorio Pasquino, Guelfo Pulci Doria

Friday 3 july 2015

11:15 - 11:30h at North America (level 0)

Themes: (T) Hydro-environment, (ST) Ecohydraulics and ecohydrology

Parallel session: 15F. Environment - Ecohydraulic

One of the main concerns in turbulence studies is the estimate of temporal and, if possible, spatial characteristics of the existing eddies in the flow. Possibly most easy is the evaluation of the integral timescale of turbulence defined with the use of the autocorrelation function R(_) derived by the analysis of time series collected in fixed positions. The transformation from temporal characteristics of the signal to spatial characteristics can be made by means of the Taylor hypothesis of the “frozen turbulence” which is applicable if the intensity of the turbulence of the flow is small. Generally, the autocorrelation function of the longitudinal velocities assumes the shape of curve rapidly decaying to its first zero-crossing, after which an alternation of positive and negative values may be evident. In these cases this characteristic point determines the time lag. In other cases the determination of the integration domain could be not straight-forward. In particular, the domain of the autocorrelation function from experimental data is finite, and there is some uncertainty on how best to define the integration domain. Sometimes the integral timescale is assumed corresponding to the time lags at which R(_) drops to 1/e. In this paper the results of experimental studies aimed to the recognition of Eulerian integral length scale in channel with simulated woody vegetation, are discussed. The model of vegetation consisted of regular arrays of stiff vertical cylinders with variation in cylinders density. Two measurements locations were considered. In order to measure instantaneous velocities an LDA system was used. The integral timescales were assumed corresponding to the time lags at which R(_) drops to 1/en, testing different values of the exponent n. Eulerian integral length scales were evaluated through Taylor’s hypothesis. Their distributions suggest that, increasing n, the time lag 1/en corresponds to larger integral length scales and measuring location and cylinders density affect the distributions.