Firefly algorithm for optimal sizing of Standalone Photovoltaic System / Nurizzati Abdul Aziz

Although Stand-Alone Photovoltaic (SAPV) systems are frequently used as a mode of electrification in rural areas which are deprived of conventional grid electricity, a common issue of such systems is the system sizing. If the system is poorly designed, the system operation would be interrupted, thus reducing the overall reliability of the system as a power supply entity. In addition, as there are numerous models of system components in the market, selection of the optimal model for each component has always become a tedious and time consuming for system designers. Therefore, optimization methods are often used in the sizing algorithms for such systems. This study presents the development of Firefly Algorithm-based Sizing Algorithm, known as FASA for sizing optimization of SAPV systems. The sizing algorithm utilized Firefly Algorithm (FA) to optimally select the model of each system component such that a system technical performance indicator is consequently optimized. FA was incorporated in two sizing approaches, i.e. the intuitive method and the hybrid intuitive-deterministic method with the technical performance indicator set as Performance Ratio (PR) and Loss of Power Supply Probability (LPSP) respectively. Besides that, two design cases of PV-battery system, i.e. system with standard charge controller and system with MPPT-based charge controller were investigated. Apart from that, Iterative-based Sizing Algorithms (ISA) for each design case with the two sizing approaches were developed to determine the optimal solutions which were used as benchmark for FASA. The results showed that FASA had successfully found the optimal PR and LPSP in all design cases using both intuitive and hybrid intuitive-deterministic methods. In addition, sizing algorithm with FA was also discovered to outperform sizing algorithm with selected computational intelligence, i.e. Genetic Algorithm, Evolutionary Programming and Particle Swarm Optimization in producing the lowest computation time in the sizing optimization.