One of the main concerns for modular Very High Temperature Gas-Cooled Reactors (VHTR) is the design of passive heat removal systems from the reactor vessel cavity. The Reactor Cavity Cooling System (RCCS) is an important heat removal system during normal and up-normal conditions. The design and validation of the RCCS is necessary to demonstrate that HTGRs can survive the postulated accidents. Here we investigate this using the Computational Fluid Dynamics (CFD) STAR-CCM+ V3.06.006 code to simulate the Pressurized Conduction Cooling (PCC) and Depressurized Conduction Cooling (DCC) accident scenarios. Heat is transported by radiation and free convection from the Reactor Pressure Vessel surface to the cooling panels or standpipes. The standpipes are cooled by natural circulation of air or forced circulation of water flowing through the pipes. A representative VHTR RCCS configuration was considered, represented experimentally by a 180 scaled model facility that was used to measure temperature and velocity distributions inside the cavity. The CFD model constructed incorporated the features of the experimental facility. Using the vessel temperature profile obtained from the experimental facility as boundary conditions in the CFD simulations, different tests were performed increasing the vessel average wall temperature progressively. Grid independence was achieved and different turbulence models and near-wall treatments were tested. For the standpipes, simulations with both natural circulation of air and forced circulation of water were performed. A reasonable agreement between the experimental results and the CFD simulations was achieved for the temperature distributions in the RCCS cavity. Also the standpipes external wall temperature was close to the experimental data. The fraction of heat exchange due to radiation determined by STAR-CCM+ code was in reasonable agreement with the experimental results. The k- turbulence models results were compared against the other turbulence models (i.e., the k-, Reynolds Stress Transport, and Spalart-Allmaras). Some differences were found between the turbulence models used. The k- turbulence models showed in general better performance than the k- and Spalart-Allmaras models if compared with the Reynolds Stress Transport (RST) results and experimental data. Among the k- turbulence models, the Realizable k- turbulence models with two-layer all-y+ near wall treatment performed better than the standard and the Abe-Kondoh-Nagano (AKN) k- models with Low-Reynolds Number low-y+ and all-y+ near wall treatments, if compared to both the RST and experimental results. The RST model was expected to perform better than the other models considering the strong anisotropy of the Reynolds stress tensor close to the vessel wall. The discrepancy between the experimental data with the RST model predictions may be due to the need for finer computational mesh model. A scaling analysis was developed to address the distortion introduced by the experimental facility and CFD model in simulating the physics inside the RCCS system with respect to the real plant configuration. The scaling analysis demonstrated that both the experimental facility and CFD model give a satisfactory reproduction of the main flow characteristics inside the RCCS cavity region, with convection and radiation heat exchange phenomena being properly scaled from the real plant to the model analyzed.