The intermittent property of a photovoltaic (PV) system requires supplementary energy such as the utility grid or batteries to meet load demand. However, when large scale PV systems are connected to the utility grid, they might affect the grid stability if the overall system is not properly designed. Hence, an accurate model for forecasting the PV system output would be useful in enhancing the system stability and reliability. The dynamic modelling of PV systems is thus crucial to the rapidly developing technologies and integrated sources in the smart grid application. This paper presents different approaches to model PV systems and identifies their pros and cons in modelling. The paper then explains the importance of a dynamic model, followed by the methodology in building up such a dynamic model. A three-vertex representation of a nearby building casting a shadow onto the PV array is also proposed as a novel approach in shadow analysis. The implementation of the dynamic model for PV systems was demonstrated in a case study in Hong Kong.