The accuracy of CFD simulations of vertical axis wind turbines (VAWTs) is known to be significantly associated with the computational parameters, such as azimuthal increment, domain size and number of turbine revolutions before reaching a statistically steady state condition (convergence). A detailed review of the literature, however, indicates that there is a lack of extensive parametric studies investigating the impact of the computational parameters. The current study, therefore, intends to systematically investigate the impact of these parameters, on the simulation results to guide the execution of accurate CFD simulations of VAWTs at different tip speed ratios (lambda) and solidities (sigma). The evaluation is based on 110 CFD simulations validated with wind tunnel measurements for two VAWTs. Instantaneous moment coefficient, C-m, and power coefficient, C-p, are studied for each case using unsteady Reynolds-averaged Navier-Stokes (URANS) simulations with the 4-equation transition SST turbulence model. The results show that the azimuthal increment d theta is largely dependent on tip speed ratio. For moderate to high lambda, the minimum requirement for d theta is 0.5 degrees while this decreases to 0.1 degrees at low to moderate lambda. The need for finer time steps is associated to the flow complexities related to dynamic stall on turbine blades and blade-wake interactions at low lambda. In addition, the minimum distance from the turbine center to the domain inlet and outlet is 15 and 10 times the turbine diameter, respectively. It is also shown that 20-30 turbine revolutions are required to ensure statistically converged solutions. The current findings can serve as guidelines towards accurate and reliable CFD simulations of VAWTs at different tip speed ratios and solidities.