Heat and mass transfer modelling in building facades with ventilated cavities requires information on the cavity air change rates, which can be a complex function of the building geometry and meteorological conditions. This paper applies Reynolds-averaged Navier-Stokes (RANS) CFD to study wind-induced airflow in the narrow (23 mm) ventilated facade cavities of a low-rise building by coupled and decoupled simulations. In the coupled simulations, the wind-flow pattern around the building and the resulting air flow in the cavities are calculated simultaneously and within the same computational domain. In the decoupled simulations, two separate CFD simulations are conducted: a simulation of the wind flow around the building (with closed cavities) to determine the surface pressures at the cavity inlet and outlet openings, and a simulation of the cavity airflow, driven by these surface pressures. CFD validation is performed for the external and internal (cavity) flows. It shows that while both laminar and turbulent cavity airflow can be accurately reproduced with low-Reynolds number modelling, this method fails in the transitional regime. The valid CFD results (outside the transitional regime) are analysed in terms of cavity air flow patterns and air change rates per hour (ACH) for different cavity positions, wind speeds and wind directions. The CFD results of cavity air speed and ACH compare favourably with values from previous experimental studies. The coupled and decoupled simulation results provide an indication of the local losses. Future work should focus on adapting RANS CFD low-Re-number models to accurately model cavity flow in the transitional regime.