PhD defense : Etude numérique et expérimentale du champ de vitesse en canaux composés

Two approaches are followed in this study. The first one is an experimental study that aims to fulfill the lack of data concerning velocity distribution in narrow and deep compound channels which can mainly be found in sewer systems. The experiments were carried out in a 16m long and 0.6m wide compound channel located in the experimental hall of ICube laboratory. A series of measurements were performed in order to observe the velocity distribution under different hydraulic parameters. The second approach is a numerical one. The objective of this study is to develop a methodology for numerical modelling of velocity distribution in compound channels. To do the experimental data of (Proust et al., 2013) were modelled and three turbulence models (k-epsilon, k-omega sst and RSM) were compared in order to determine the most suitable model to reproduce velocity distribution and Reynolds stress tensors in compound channels. The results showed that the isotropic turbulence model k-omega sst is perfectly able to reproduce velocity distribution and even Reynolds stress tensors unlike RSM turbulence model that showed a very instable behavior and need to be very well calibrated in order to obtain good results. In addition to this comparison, the large eddy simulation method were also utilized for two experiments of (Proust et al., 2013) and (Nezu, Onitsuka, Sagara, & Ikenita, 1990). Two specific boundary conditions were implemented in order to compare the efficiency of each method.