Network-constrained fixed-horizon resilience scheduling for industrial prosumers under notified heat-outage events

Industrial facilities may protect their own critical loads while creating voltage or loading problems in the distribution feeder that supplies them. This paper presents a deterministic, fixed-horizon resilience scheduler that couples industrial thermal and production states to a radial distribution network during preparation, an announced outage and recovery. The reported simulations solve one 168 h linear program at hourly resolution; they do not use receding-horizon state or forecast updates. Three analytical archetypes and six heat-stress profiles form 18 controlled cases on the IEEE 33-bus feeder, with LinDistFlow constraints followed by nonlinear AC checks. Relative to a cost-only schedule, the event-aware network-constrained schedule reduced unserved critical energy by 62.2%, peak import by 25.5% and recovery time by 44.8% for a 3.3% weekly cost increase. A 300-scenario calibration sample gave violation frequencies of 69.3%, 1.3% and 0% for optimization floors of 0.953, 0.957 and 0.960 p.u., respectively; these are calibration-sample estimates rather than independent holdout results. A separate analysis of the public Øra industrial-grid data used 13,344 fully aligned hours from 45 customers. The measured aggregate peak was 3.84 MW and the coincidence factor was 0.653; calculated incremental import headroom varied from 0.55 to 5.68 MW across customer buses under stated power-factor assumptions. The findings therefore constitute a controlled proof of concept for notified events, not field validation of a real-time controller.