Between the utility, the generators, and the most fragile loads in a hospital sit two specialized power layers: uninterruptible power supply (UPS) systems that bridge the gap no generator can cover, and isolated power systems (IPS) that make the most hazardous patient environments electrically forgiving. This article covers the power-conditioning and patient-safety layers that ride on top of the Essential Electrical System — the seconds-long ride-through that the generator transfer cannot provide, and the ungrounded distribution that protects against ground-fault hazards in wet procedure locations.
The Essential Electrical System (EES) restores power after a utility loss, but it does not do so instantaneously. NFPA 110 permits a Type 10 emergency power supply system to begin delivering power within 10 seconds of a normal-power failure — that is the design target for hospital generator plants, and it is the interval during which automatic transfer switches sense the outage, signal the generators to start, allow the engines to reach speed and stable voltage, and then transfer load. Ten seconds is acceptable for lighting, HVAC, and most equipment, but it is catastrophic for any load that cannot tolerate even a momentary interruption.
UPS systems exist to cover that 10-second window — and the much shorter sags, surges, and transfer notches that occur even when generators perform perfectly. A UPS stores energy (almost always in batteries, sometimes in flywheels) and delivers it through power electronics that produce a clean, continuous output regardless of what the upstream source is doing. When utility power fails, the load never sees the interruption; the UPS carries it until the generator picks up the upstream feed, after which the UPS recharges and stands ready again.
The distinction that governs design is between break and no-break power:
The practical sequence in a hospital outage: the lights flicker (or, on life-safety battery units, stay lit), break-power loads drop for up to 10 seconds and then return on generator, while no-break UPS loads never interrupt at all. The UPS is the only layer that delivers true zero-transfer-time power.
Not everything needs no-break power, and UPS capacity is expensive — so the design discipline is to identify the genuinely interruption-intolerant loads and put only those on UPS. Typical UPS-served loads in a contemporary hospital include:
A guiding rule: device-internal batteries protect the device; facility UPS protects the system. A ventilator may ride through a 10-second gap on its own battery, but the central station, the network, and the EHR documenting that ventilator's settings need facility-level no-break power.
UPS units are classified by how the load is normally powered relative to the battery/inverter path. The three classical topologies, in ascending order of protection and cost:
| Topology | How it works | Protection level | Typical healthcare use |
|---|---|---|---|
| Standby (offline) | Load runs on utility; switches to inverter on failure (brief transfer, ~2–10 ms) | Lowest | Rarely used for clinical loads; small desktop/standalone devices only |
| Line-interactive | Load runs on utility through a regulating transformer (buck/boost); inverter engages on failure | Medium — corrects sags/swells without battery | Small IT closets, non-critical edge loads |
| Double-conversion (online) | Utility AC → DC → inverter → AC continuously; load always runs on the inverter | Highest — zero transfer, full conditioning, complete isolation from input disturbances | The standard for critical clinical and data-center loads |