Sizing a UPS for a Hospital Critical Branch

Ask three consultants to size a UPS for a hospital critical branch and you will get three different answers, usually because they are quietly answering three different questions. The kVA figure is the easy part. What decides whether the installation is still adequate in five years is a set of assumptions that rarely get written down anywhere, and almost never get revisited.

The Critical Branch Is Not the Whole Hospital

NFPA 99 divides a hospital essential electrical system into three branches, and they do not have the same requirements. The life safety branch covers egress lighting, alarm and alerting systems, emergency communications, generator set lighting and elevator control. The critical branch covers the areas and receptacles associated with maintaining patient treatment. The equipment branch feeds the large plant that patient care depends on but that can tolerate a delay, including central suction, medical-surgical vacuum and the large HVAC serving operating rooms.

Getting this split right is the first sizing decision, and it is made on a drawing rather than in a spreadsheet. A UPS sized against a load schedule that has quietly included equipment-branch plant will be substantially oversized and correspondingly expensive. One sized against a schedule that has left out critical-branch receptacles will be undersized in a way that only shows up during an actual outage.

It is worth saying plainly that the UPS does not replace the generator. The generator carries the outage. The UPS carries the seconds the generator cannot, and it conditions the supply for equipment that will not tolerate the transfer itself.

The 10 Seconds That Set the Floor

Hospital generators are required to restore power to the life safety and critical branches within 10 seconds of losing the utility. That is a demanding window: the system has to detect the outage, crank and start the engine, bring it to rated speed and voltage, and complete the automatic transfer, all inside those 10 seconds.

The naive conclusion is that a UPS with 10 seconds of runtime therefore covers the gap. It does not, and sizing to it is the most common mistake we are asked to correct. Ten seconds is the value the code can hold a generator to under test conditions. It is not a value that survives the case the code cannot legislate for, which is the generator that does not start at all, or starts and then trips on its first block load.

A defensible number for the critical branch is normally somewhere between ten and thirty minutes, and the range matters less than the reasoning. That is long enough for a second start attempt, for a manual transfer to an alternate source, or for a controlled wind-down of whatever procedure is under way. Whatever figure is chosen should be a written decision with a rationale attached, because it is the assumption most likely to be challenged after an incident.

Size in kW, Not kVA Alone

A UPS nameplate carries a kVA rating; the load draws kW. The ratio between them is the power factor, and comparing two quotations on kVA alone compares almost nothing. Older UPS designs were commonly rated at 0.8 power factor, so a 100 kVA unit delivered 80 kW. Many modern units are rated at unity, so a 100 kVA unit delivers 100 kW. Those are not the same machine, and they are not the same price.

On the load side, resist the temptation to add up nameplates. A nameplate is a worst-case figure that assumes the device is doing the most demanding thing it can do, and a ward full of equipment never does that simultaneously. Summing nameplates typically produces a number 40 to 60 percent above anything the branch will actually draw, and hospitals pay for that difference twice: once in capital cost, and again in a UPS running at low load where it is least efficient.

The honest method is to measure. Clamp the existing branch over a full working week, capture the peak rather than the average, and then apply headroom as a deliberate decision on top of a measured figure. Where the branch does not exist yet, build the load schedule from the equipment list and apply a diversity factor you are willing to defend, then measure at commissioning and record the result for the next upgrade.

Leave Room for the Load You Have Not Bought Yet

Hospitals grow into their electrical infrastructure faster than almost any other building type. An imaging suite is added, monitoring is upgraded from one device per bed to three, infusion pumps multiply, and a UPS that was sized precisely against a validated load schedule is at capacity in year three. The schedule was not wrong. It was a snapshot.

We normally recommend sizing with 25 to 30 percent headroom above the measured peak, and then checking that the rest of the installation can actually use it. A UPS with spare capacity feeding a distribution board with no spare ways is not spare capacity; it is a number on a nameplate. The same applies to cable sizing and to the room the batteries live in.

Where the budget allows it, a modular UPS architecture converts this from a forecasting problem into a purchasing one. Power modules are added to the frame as the load grows, so the initial spend is matched to the initial load and the capacity arrives when it is needed. The frame has to be specified for the eventual load from the start, which is where the forecasting reappears, but the consequence of getting it wrong is far smaller.

Runtime Is a Battery Decision

It is worth separating the two questions that get conflated in most conversations. The UPS module determines how much power can be supported. The battery string determines how long. Doubling runtime does not require a larger UPS, it requires more battery, and the constraint is usually floor space and floor loading rather than the electronics.

Battery sizing has to be done at end-of-life capacity, not at the capacity of a new string. A battery is generally considered to have reached end of life at 80 percent of rated capacity, and a system that only meets its runtime target on a brand-new string is a system that misses it for most of its service life. The same calculation has to use the temperature the batteries will actually live at, not the 25 degrees the datasheet assumes.

That temperature point is where most hospital UPS installations in this region quietly fail, and it deserves more attention than it usually gets during design. It is also the subject of a separate article, because the arithmetic behind it explains almost every early battery replacement we are called out to.