ABSTRACT
Optimal location and sizing of Distribution Generation (DG) units in radial distribution networks is critical to achieving an improvement in stability, reliability and reduction of power losses in the network. This dissertation presents a hybridized solution which is a combination of the analytical method and the firefly algorithm for optimal placement and sizing of DGs. The conventional analytical method for optimal DG placement and sizing was first modelled. The standard firefly algorithm was then modelled and integrated with the analytical method to form the proposed hybridized solution. The conventional analytical method and the proposed method were applied on standard IEEE 33 and 69 test buses for optimal DG location and sizing and the results obtained were compared to the base case scenario without DG. For the 33-bus, the analytical method found the optimal location and size of the DG to be bus 30 and 548kW respectively while the proposed method found the optimal location and size to be bus 2 and 133kW respectively. These results obtained for 33-buscaused 26.87% reduction in system losses and16.44% improvement in voltage profile for the analytical method while the proposed model?s result caused 28.36% reduction in losses and 33.54% improvement in voltage profile when compared with the base case. For the 69-bus network, the analytical method and the hybrid model obtained the same results which was bus 63 for the optimal location and 590kW for the optimal size thereby resulting in 10.29% reduction in losses and 45.35% improvement in voltage profile. Furthermore, the proposed method had a 90.06% reduction in simulation time for the 33-bus and 76.28% for the 69-bus as compared with the analytical method. The proposed hybrid model was validated by comparing the results obtained for the 33-bus system with the published results by Viral and Khatod, (2015).This comparison showed that the proposed hybrid model is valid for solving optimal DG allocation problems as the voltage profile follows the same trend with a better performance than results published by Viral and Khatod, (2015). This is an indication that hybridization of two methods would provide an optimal result faster than stand-alone methods.