For the prediction of convective heat loss from solar concentrating receiver cavities a number of empirical correlations exist. Geometry and the inclination angle determine the degree to which natural convection can infiltrate the cavity and remove stably stratified hot air out through the aperture. This makes the task of defining characteristic lengths for such Nusselt correlations difficult, neither does their use offer insight as to how one might reduce heat loss through the use of baffles, air curtains or small aperture-to-cavity-area ratios. Computational Fluid Dynamics (CFD) can assist in the design of better cavity receivers as long as the rules upon which it rests are respected. This paper is an exploration of the need for turbulence modelling in cavity receivers using some common linear eddy viscosity closure schemes. Good agreement was obtained with the CFD software OpenFOAMO® 3.0.1 for a deep cavity aperture but it under-predicted a shallow cavity. The experiments used for validation were in the Grashof region Gr ≈ 10 6 , well below the region for transition to turbulence between 10 8 < Gr < 10 9 .