École Nationale Supérieure de
Techniques Avancées

Abstract

This internship aimed at identifying meshing requirements that would guarantee the dynamic and thermal validity of numerical simulations. Simulation campaigns were led with the CFD solver elsA on an infinite plane channel with or without ribs. Obtained results were compared to reference DNS or experimental data to estimate the relative influence of the first cell thickness y+0 , the progress ratio r and the turbulence model on the accuracy of the results. General flow dynamics is correctly reproduced for progress ratio under 1.1. Nevertheless, the dynamic and thermal fields close to the wall need a normalized first mesh thickness y+ 0 lower than 1. Among turbulence models, the k-omega SST Menter model give the best results. Advanced models appear to be necessary to gain in representativity.