Robustness of a POI-Derived Urban EV Charging-Station Network Under Capacity-Constrained Cascading Failures: A Scenario Analysis of Six Districts in Chengdu, China

Reliable public electric-vehicle charging depends on whether operating stations can absorb demand displaced by local outages. This study develops a reproducible scenario-analysis framework for a charging network derived from a POI snapshot spanning six Chengdu districts. The framework represents potential local substitution through a spatial proximity graph and simulates synchronous, capacity-constrained load redistribution while conserving displaced demand and recording unmet service. Using 648 retained station POIs, we compare random failures with centrality-targeted attacks across attack scales, capacity headroom, topology, load, and redistribution assumptions. The network combines dense, locally clustered central districts with a partially fragmented wider structure. The modelled served-load ratio remains comparatively high under limited random disruption but deteriorates nonlinearly as failures expand and spare capacity is depleted. Targeted outcomes vary across centrality measures and modelling assumptions, showing that structural prominence alone does not identify stations with the greatest service consequences. The framework therefore supports scenario screening for redundancy, substitution pathways, and candidate critical nodes, while station-level planning requires operational and behavioural validation.