Improving forecasts of tropical cyclones (TCs) undergoing rapid intensification (RI) is important because underprediction of RI can lead to a heavy loss of life and tremendous financial loss, especially if a TC undergoes RI prior to making landfall in a densely populated coastal city. Physical processes involved with RI, and intensification more broadly, involve interactions across multiple spatial and temporal scales. Accurately representing these scale interactions in numerical models is a challenge, and for this reason forecasting RI remains difficult. This study evaluates the impact of model physics on RI prediction using retrospective forecasts of TCs created by the Hurricane Weather Research and Forecasting (HWRF) model. Three components of model physics are investigated here: horizontal diffusion parameterization, cumulus parameterization and boundary-layer parameterization. The performance of RI forecast is fist verified against the Best Track using the standard contingency table. Structural metrics based on observations are then used in model diagnostics. Composite analyses of TC structure at the RI onset for Hit versus Miss cases help determine if a positive impact of the model physics change on RI forecast is due to a right reason or not. Angular momentum budget is also conducted to understand how a certain model physics dynamically affects the RI process.