Adaptive Risk-Aware OpenStreetMap Planning for Scalable Radial Distribution Networks

A street-constrained distribution-network synthesis framework is evaluated using open geographic data, physical medium-voltage (MV) outage consequences, and benchmark electrical checks. A synthetic radial case study in Queens, New York, compares minimum-hop Dijkstra, minimum-spanning-tree connection, distance-weighted Dijkstra, Steiner approximation, conventional OpenStreetMap (OSM) routing, and Adaptive Risk OSM on shared road support for nested population representations of 20–100%. Rather than treating every low-voltage (LV) service segment as an independent contingency, it recomputes primary resilience indicators for aggregated physical MV assets. At full representation and 20 m road mesh, the selected Adaptive Risk OSM solution ($\alpha =4$) has an MV length of 3.2717 km, CVaR₉₅ of 50.42%, critical-asset share of 13.04%, interruption exposure of 0.3014 user-km/user, and targeted-removal robustness AUC of 0.2386. The six-method comparison shows that minimum-hop Dijkstra achieves the lowest CVaR₉₅, critical-asset share, and interruption exposure; conventional OSM routing achieves the shortest MV route; and distance-weighted Dijkstra achieves the highest targeted-removal AUC. The proposed method therefore represents a risk-aware compromise rather than a universally dominant routing strategy. Benchmark three-phase balanced calculations reveal route-dependent voltage, conductor utilization, and loss differences; transformer utilization, by contrast, is governed by the common customer allocation and reaches 108.55% at threefold full-population demand, the only violating loading condition among the 150 convergent electrical cases. The findings are confined to the defined synthetic radial case and benchmark medium-voltage checks; low-voltage electrical performance is outside the validation scope.