This work takes place in the context of potential hazards in the use of hydrogen in fuel cells. The present article describes comparisons between Ply measurements performed on a two vented cavity with an helium injection and Large Eddy Simulation of the same configuration. A two vented cavity is chosen because a quasi state is reached rapidly and it facilitates both CFD calculations by reducing the CPU costs and also enables statistical treatment of the data, the temporal averaging being possible at steady state. At the same time, this configuration is close to fuel cell designs, except for the set-up reduced size. We also describe the experimental set-up and the care which has to be taken to produce Particle Image Velocimetry velocity fields. The final goal of the paper is to validate a L.E.S approach as a good replacement to experiments, since access to both velocity and concentration fields is required to improve existing simplified models. Indeed, most of the 2 vents models rely on simplified assumptions such as a constant entrainment coefficient, a bi-layer formation which is not always the case in real situations. (C) 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.