Grid-Impact-Aware Planning of Electric Vehicle Charging Infrastructure for a Tourism-Intensive Island Grid Using Mixed-Integer Optimization and AC Power Flow Assessment

Electric vehicle charging station (EVCS) deployment is conventionally framed as a siting-and-sizing problem, yet in tourism-intensive island grids the binding question is whether an optimized layout remains acceptable once embedded in the distribution network. This paper develops a grid-impact-aware planning framework that links simultaneous charging factor (SCF)-based demand conversion, multi-objective mixed-integer programming (MOMIP), bus-level load mapping, AC power-flow assessment, and source-side N-1 contingency screening in one traceable pipeline. A Hybrid strategy, in which Binary Integer Particle Swarm Optimization with Greedy Repair (BIPSO-GR) provides a feasible incumbent for Branch-and-Cut (B&C) refinement, is benchmarked against standalone B&C and BIPSO-GR on a 16-bus, 22 kV feeder under base, source-side N-1, and seasonal/event-driven demand-growth scenarios. Voltage magnitude remains non-discriminating across all cases. The binding factor is the post-contingency hosting-capacity margin (HCM) of the transfer corridor, governed by charger allocation and spatial load concentration. Under source-side N-1 operation, the distributed B&C and Hybrid layouts yield contingency severity index (CSI) values of 18.6% and 16.2%, whereas the concentrated BIPSO-GR layout reaches 57.3%, about 3.1 times the B&C benchmark and 3.5 times the Hybrid value. Under the $1.5\times $ demand-stress case, the Hybrid layout preserves 35.3% HCM, compared with only 5.3% for BIPSO-GR. A ±20% SCF sensitivity confirms that the lower-severity ranking of the distributed layouts is preserved within the tested simultaneity range. These results show that EVCS layouts must be screened for bus-level grid impact and post-contingency thermal security, not for optimization feasibility or voltage compliance alone.