During a severe accident in a PWR, large quantities of hydrogen can be generated and released into the containment. The generated hydrogen, when mixed with air, can lead to hydrogen combustion. The dynamic pressure loads resulting from hydrogen combustion can be detrimental to the structural integrity of the reactor safety systems and the reactor containment. Therefore, accurate prediction of these pressure loads is an important safety issue. In a previous article, we presented a CFD based method to determine these pressure loads. This CFD method is based on the application of a turbulent flame speed closure combustion model. The validation analyses in our previous paper demonstrated that it is of utmost importance to apply successive mesh and time step refinement in order to get reliable results. In this article, we first determined to what extent the required computational effort required for our CFD approach can be reduced by the application of adaptive mesh refinement, while maintaining the accuracy requirements. Experiments performed within a small fan stirred explosion bomb were used for this purpose. It could be concluded that adaptive grid adaptation is a reliable and efficient method for usage in hydrogen deflagration analyses. For the two-dimensional validation analyses, the application of dynamic grid adaptation resulted in a reduction of the required computational effort by about one order of magnitude. In a second step, the considered CFD approach including adaptive mesh refinement has been further validated against three hydrogen deflagration experiments performed in the ENACCEF facility. For each test, mesh and time step sensitivity analyses have been performed. From the presented validation analyses, it could be concluded that the maximum pressures were predicted within 13% accuracy, while the rate of pressure rise dp/dt was predicted within about 30%. The eigen-frequencies of the residual pressure wave phenomena were predicted within a few %. Therefore, it was overall concluded that the current model predicts the considered ENACCEF experiments very well. (C) 2012 Elsevier B.V. All rights reserved.