How a hospital keeps its elevators moving — or at least keeps them from trapping people — when normal utility power fails. This Article covers the standby-power side of vertical transport: which elevator loads ride the generator, how cars are sequenced when the bus transfers, and how the whole assembly is engineered to survive the events a hospital is actually expected to survive. The sibling Article on firefighters' emergency operation and occupant-evacuation elevators covers the fire-service recall sequence and evacuation-elevator design; here the focus is the electrical and resilience architecture behind it.

Why elevator power is a life-safety system in a hospital

In most building types an elevator power loss is an inconvenience. In an acute-care hospital it is a patient-safety event. Inpatients are moved vertically on stretchers and in beds for surgery, imaging, dialysis, and intensive care; a code-blue team responding to a cardiac arrest on an upper floor depends on a car arriving in seconds; sterile supply, pharmacy (including USP 797/800 compounded preparations), blood bank, and food service all flow vertically. A multi-hour outage that stalls the elevators effectively partitions the hospital into stranded floors.

For this reason the elevator system is not treated as ordinary building equipment when standby power is engineered. Under NFPA 99 (Health Care Facilities Code) and NFPA 110 (Standard for Emergency and Standby Power Systems), selected elevator functions are part of the facility's Essential Electrical System (EES) and must be restored within a defined time after a normal-power failure. The governing intent is twofold: keep enough vertical transport running to operate the hospital, and guarantee that no occupant is left trapped in a stalled car.

The essential electrical system and where elevators sit in it

A hospital's standby power is organized by NFPA 99 / NFPA 110 / NEC Article 517 into a structured Essential Electrical System, which for a Type 1 EES (required for facilities providing critical care) is divided into branches:

Elevators are generally fed from the Equipment Branch because elevator motors are large, inrush-heavy loads that should not be slammed onto the generator at the same instant as the 10-second life-safety and critical loads. The exception is the small but critical elevator life-safety interface — car lighting, the communication/phone link, and fire-service signaling — which is typically carried so that those functions persist even when the drive itself is not running.

The practical design question on every project is therefore not "are the elevators on emergency power?" but how many cars, in what sequence, and with what controls, the standby source can run.

Standby capacity: running all cars vs. one-at-a-time

A hospital rarely has enough generator capacity to start and run every elevator simultaneously. Generator and feeder sizing for the full simultaneous starting current of an entire elevator bank is usually uneconomical and unnecessary. The codes resolve this with two related provisions:

• Selective / sequential operation. When normal power is lost, an emergency-power elevator controller (sometimes called an automatic selection / sequencing panel) brings cars onto the standby source in a controlled order rather than all at once. A common arrangement is to land all cars first (so no one is trapped), then keep a defined subset running on standby — for example, one car per bank, or the cars serving critical departments — selectable manually from a lobby or fire-command-center panel.
• Manual selection. During an extended outage, staff or the fire service can transfer standby power from one car to another so that any given car in a group can be operated, even if not all can run at once.

The owner's program drives the number that must run continuously. A facility that performs surgery, deliveries, and critical-care transport around the clock will typically require that at least one stretcher-capable car per major bank runs continuously on standby, plus the ability to bring additional cars on as needed. This is a decision to make explicitly during design and to document in the basis-of-design — it is not a code default.

The recall–transfer interaction (and the "regeneration" trap)

The most failure-prone moment in elevator emergency power is the transfer itself — the few seconds when the bus swings from utility to generator (and later back). Two interacting sequences must be coordinated:

1. Power-loss sequencing. On loss of normal power, cars in motion should be brought to the nearest floor and the doors opened so passengers can exit, before the system decides which cars continue on standby. ASME A17.1 / CSA B44 addresses this behavior; ANSI/ASME A17.1 Rule sets covering "emergency or standby power operation" require that cars are returned to a landing in a controlled manner.
2. Open-transfer "regeneration." Modern elevators with regenerative or variable-frequency drives can push energy back onto the bus when a loaded car descends. During an open transition between sources — or when a car is moving as the transfer occurs — that regenerated energy can over-voltage or trip the generator. The mitigation is coordinated controls: hold cars during the transfer window, use closed-transition transfer switches where appropriate, and provide drive-side load banks or regeneration management so the generator is not destabilized.

A correct design treats the transfer switch, the generator controls, and the elevator controller as one coordinated system, not three independently specified components. This coordination is a frequent gap when the elevator, electrical, and generator packages are bid separately, and it is a primary target of integrated systems commissioning.

The trapped-passenger problem and the 2-hour/4-hour resilience question