This thesis intents to make a review on the actual state of art in modelling turbulent gas explosions and how to do improvements on current CFD models comparing to experimental data. All the cases studied are large scale explosions of a gas such like hydrogen or natural gas. The gas is confined and premixed with air to simplify. The main objective is to achieve a model which allow to estimate correctly the pressure wave and flame spread in different cases. For this reason the CFD codes are used to model an initial case and then try to find the most suitable combustion and turbulent model. Then it is been compared the real data cases with the modified models in this project and it is evaluated how exact is the new approach. The data used flame spread and pressure against time. Then a sensitivity analysis has been performed to study variations of parameters such as turbulence, models and boundary conditions.