The combustion of zirconium potassium perchlorate (ZPP) in a closed vessel is modelled and validated through a numerical simulation. Because the extremely high pressure oscillation occurs in less than a millisecond, an in-house computational fluid dynamics (CFD) code is used to observe the detailed flow structures and determine the adequate burning characteristics, including the burning rate. A hybrid RANS/LES scheme with a 5th order upwind weighted essentially non-oscillatory (WENO) is implemented to capture complex, strong shock waves in highly turbulent combustion. A two-way coupled Eulerian-Lagrangian scheme tracks the combusting ZPP granules reasonably well. Monodisperse and Rosin-Rammler assumptions are applied to the ZPP granule distribution. The monodisperse assumption reveals that the diameter of the ZPP (17 mu m) achieves marginal agreement with the measurements. However, the Rosin-Rammler distribution improves the transient and dynamic combustion characteristics in that the small granules contribute to a faster burning rate and stronger shock waves. These results are more analogous to the closed vessel tests used as validation data.