Simultaneous Allocation, Sizing, and Quantity Optimization of Distributed Generation in a Real Campus Grid: A Comparative Study of GA and GWO

Abstract

The integration of Distributed Generation (DG) into Radial Distribution Networks (RDNs) is crucial for reducing technical losses and enhancing voltage stability. Unlike traditional studies utilizing idealized test systems without variable DG counts, this study proposes a simulationbased multi-objective optimization method to simultaneously determine the optimal location, capacity, and quantity of DG units in an actual 17-bus, 34.5/0.4 kV campus distribution grid (Trakya University Balkan Campus). We couple a Genetic Algorithm (GA) and a Grey Wolf Optimizer (GWO) with MATPOWER's analytical load-flow equations to optimize active power loss (Ploss), reactive power loss (Qloss), and the Voltage Deviation Index (VDI). A novel dynamic decoding strategy prevents duplicate bus allocations, while static penalty functions enforce operational constraints. Baseline analysis revealed severe voltage constraints due to overloaded transformers (>80%). Both GA and GWO independently converged to identical optimal placements of two DG units (totaling 4.82–4.84 MW) at Bus 4 and Bus 6. Under balanced multiobjective conditions, active losses decreased by ~15% (from 0.0318 MW to 0.0270 MW) with simultaneous improvements in Qloss and VDI. Conversely, single-objective voltage optimization yielded >5.3 MW over-generation and >10% increased active losses, demonstrating that balanced multi-objective frameworks are essential for effective RDN planning.