Electrical efficiency of a standard silicon-based photovoltaic cell declines with increasing its surface temperature exceeds 25⁰ C; this makes the PV cooling technique to be essential for solar power generation, especially in hot climates. This work aims to first analyse the convection behaviour at the surface of building integrated photovoltaic (BIPV) systems, and then develop new correlations to calculate their convective heat transfer. Computational fluid dynamics (CFD) technique is used to simulate a 3D BIPV model placed in an atmospheric wind tunnel. The validated model offers good results in resolving velocity, temperature and turbulence fields. New correlations of convection performance derived from the simulated data in format of Nusselt number with a determination coefficient of about 0.995. The average uncertainty of the final local Nu regression to present real case is also approximately 3%. Hence, this study suggest more accurate correlations for both local and average Nusselt numbers comparing to the traditional correlations.