New York City Power Outages: What an Underground Grid Cannot Prevent

Reconstruct the event before naming the lesson

New York’s Public Service Commission summarized the Manhattan event in its July 15, 2021 order. Approximately 73,000 customers lost service on July 13, 2019 after a distribution-feeder fault near the West 65th Street substation. Several networks were affected, and the account reports restoration to all customers in less than five hours. Those figures describe this historical event, not today’s New York outage status. [1]

The same order summarizes the Department of Public Service investigation’s identification of omitted ground or neutral wiring for protective relays as the direct failure mechanism. Keeping the cable fault separate from that protective-system problem is essential to understanding the documented explanation. The order also records the company’s dispute of departmental allegations and the subsequent settlement process. [1]

Why cable location is an incomplete reliability metric

Con Edison’s description of electricity delivery follows a path through transmission, substations, distribution equipment, and transformers before the customer connection. It reports both underground and overhead infrastructure in its broader service area. That system description supports a basic distinction: undergrounding concerns where some conductors are located, while service continuity depends on the functioning delivery chain. [2]

Our interpretation of the 2019 case is consequently narrow. It demonstrates that the presence of underground equipment does not establish immunity from a significant interruption. It does not quantify whether underground or overhead service is better for every location. Such a comparison would need matched exposure, equipment condition, network design, and observation periods.

Read improvement plans as plans

In May 2026, Con Edison announced $3.9 billion in electric-delivery investment across New York City and Westchester. Its description includes transformers, cables, technology, and location-specific projects. The announced program is evidence of the company’s intended work and stated priorities. An independent assessment would still need actual completion information and subsequent operating results. [3]

An evaluation should ask what failure mechanism each project addresses. Added capacity, faster fault detection, replacement of aging equipment, and flood protection are not interchangeable interventions. Even when several contribute to reliability, the strongest explanation connects a defined project to an observed reduction in a relevant kind of interruption.

How to research a neighborhood interruption

Begin with a dated utility notice and establish the affected equipment or network if the provider identifies it. Then distinguish a local delivery interruption from a wider system emergency and record how restoration estimates changed. A borough or city total alone cannot determine the cause at an individual building.

For historical comparisons, retain the original customer unit. A customer account can represent a building rather than one person, so the 2019 account count should not be converted into a population estimate without separate evidence. The useful lesson is that reliability resides in the complete system and its operation, while a map’s boundary and a cable’s location describe only parts of that system.

A meaningful comparison also distinguishes the reach of an event from the detail available about it. A network-level disruption may be reported quickly, while the effect on individual accounts becomes clearer later. Changing estimates during that process are not automatically evidence of an incorrect map. The appropriate test is whether each estimate retains its timestamp, geographic scope, and stated uncertainty, and whether later evidence supports a correction rather than silently rewriting the earlier observation.