Akademik Veri Yönetim Sistemi
Computers and Industrial Engineering, cilt.217, 2026 (SCI-Expanded, Scopus)
Mini-grids offer an efficient approach for electrifying rural areas that are geographically remote and economically disconnected from larger grids. We address a novel facility location and network design problem for spatial planning of these two-level radial distribution systems. The system integrates primary facilities with renewable generation capabilities and secondary distribution facilities through tree-structured backbone networks, while both facility types serve customers through local tree-structured networks. While backbone networks operate at high voltage with low current, local networks use lower voltages subject to voltage drop constraints that add design complexity. We formulate an optimization model to minimize total distribution costs by determining optimal facility numbers, types, locations, and network configurations at both levels. Our decomposition-based solution method consists of an initialization phase using mixed-integer linear programming to identify demand clusters and initial facility locations under star configurations, followed by iterative refinement that jointly optimizes facility positions in continuous space, tree-structured network topologies, and demand assignments. The algorithm alternates between reassigning demand nodes and adjusting network configurations for given facility locations, and fine-tuning facility locations for current network configurations until convergence. Numerical experiments demonstrate that our method achieves highly conservative cost reductions of 6–19% compared to a computationally simpler hierarchical clustering baseline that relaxes binding physical voltage constraints. These results confirm that rigorous mathematical optimization easily justifies its additional computational expense through massive, physically feasible infrastructure savings.