Optimal distributed generation allocation and sizing in a distribution system using heuristic algorithms

This thesis presents an optimization framework for the allocation and sizing of Distributed Generators (DGs) with reactive power capabilities in unbalanced three-phase distribution systems. The aim is to reduce active power losses and improve voltage profiles while also maintaining phase balance. To achieve the aim, a modified IEEE 37-bus unbalanced distribution feeder is used as the test system, and four metaheuristic algorithms are applied to determine optimal DG placement, capacity, and power factor. These algorithms include Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Grey Wolf Optimization (GWO), and Pattern Search (PS). The simulations are carried out using MATLAB linked with OpenDSS. This setup allows the algorithms to interact directly with the distribution system model during the optimization process. The study examines the performance of each algorithm under two operating conditions: unity power factor, where DGs supply only real power, and optimal power factor, where DGs can supply both real and reactive power within allowed limits. In the unity power factor case, GA achieved the highest reduction in active power loss at 73.85%, followed by GWO (73.78%), PSO (73.32%), and PS (72.98%). In the optimal power factor scenario, GA again performed best, reducing losses by 90.39%. PSO, GWO, and PS followed with reductions of 90.32%, 90.02%, and 88.63%, respectively. These results show the clear benefit of including reactive power support in DG operation. Voltage profiles were also analyzed for all three phases. Without DGs, voltages dropped significantly, especially at the end of the feeder. With DGs operating at optimal power factor, voltage levels were much more consistent and stayed within acceptable limits. Among all the algorithms, GA consistently delivered the best overall results in terms of both loss reduction and voltage regulation. The findings suggest that optimizing power factor alongside DG location and size can make a meaningful difference in system performance, especially in networks that are unbalanced or weak. Beyond the improvements in loss reduction and voltage control, the approach used in this study also shows promise for real-world implementation. In addition, the findings also suggest that DG placement should not be based solely on network layout, but also take into account how each unit operates under real system conditions.