Solar irradiance reaching the surface is attenuated by clouds that significantly reduce the production of solar power. Often, improvements in the short-term predictability of clouds and irradiance in numerical weather prediction models can assist grid operators in managing intermittent solar-generated electricity. However, numerical weather models represent sub-grid cloud variability based on cumulus parametrization at coarser resolutions. The cumulus-radiation feedback further influences the amount of solar irradiance reaching the surface. In this study, the performance of the Weather Research and Forecasting (WRF) model in forecasting different components of solar irradiance including Global Horizontal irradiance (GHI) and Direct Normal Irradiance (DNI) will be reported under days of high intermittency at Mildura site located on the border of New South Wales and Victoria, Australia. Initially, four case days were chosen with highly intermittent solar irradiance observed at Mildura site while another four days of low intermittency were taken as control for clear days. A specific configuration and augmentation of WRF model (v 3.6.1) designed for solar energy applications (WRF-SOLAR) was used to simulate solar irradiance with four nested domains nudged to ERA-Interim boundary conditions at grid resolutions (45km, 15km, 5 km, 1.7km) centred over Mildura. The Bureau of Meteorology (BOM) station dataset available at minute timescales and hourly derived satellite irradiance products were used to validate the simulated products. Also, surface variables including winds, air temperature and humidity were initially compared to ensure observed meteorology is preserved in simulations. Results from resolved clouds and unresolved sub-grid clouds from different resolutions and sites will be reported.