How to choose a UPS for hospitals, banks and factories

A UPS that works well in an office may be a poor choice for a hospital or factory.

The reason is simple: the load is different.

A CT scanner cares about power continuity and voltage quality. A bank's server room cares about availability and backup time. A factory may have motors, pumps and compressors that draw several times their normal current during startup.

So don't start with "How many kVA do I need?"

Start with "What does the load need the UPS to protect against?"

1. Hospital: isolation, leakage and continuity

Medical equipment can be sensitive to power interruptions, voltage disturbances and grounding problems.

For critical medical locations, the first question is not simply UPS capacity. It is whether the electrical system requires a medical IT system with an isolation transformer and insulation monitoring under the applicable medical electrical standard such as IEC 60601-1 and local code. In practice this means an IT system that holds leakage current below 0.5 mA.

The UPS and isolation transformer also need to be considered together. The transformer provides galvanic isolation, while the online UPS maintains the load through utility interruptions and continuously regulates the output.

For CT, MRI, DR and other imaging equipment, check:

• Required input and output voltage
• UPS capacity and overload capability
• Transfer performance
• Isolation requirements
• Battery backup time
• Grounding and leakage-current requirements

TAFENG supplies low-frequency online UPS systems with built-in isolation transformers for medical applications. One example is a 120kVA / 96kW UPS for a CT scanner in Cameroon.

The important point is that a medical UPS should be selected from the equipment and room requirements, not simply from the CT's kVA rating.

2. Bank: availability and backup time

For banks and financial facilities, an outage is not only a power problem. It can interrupt transactions, communications and other critical IT services.

The UPS configuration therefore needs to match the required availability.

For larger data centers, N+1 capacity and independent power paths may be used to support maintenance and reduce single points of failure. Uptime Institute's Tier III definition focuses on concurrent maintainability, while Tier IV adds fault tolerance. These are facility-level requirements, not simply UPS specifications.

Battery backup is another important consideration.

Don't choose 30 or 60 minutes simply because it is a common specification. Calculate the required runtime from the actual operating sequence:

Utility failure → UPS battery → generator start → transfer → generator operation

For a smaller bank branch or office, the requirement may be much simpler. A properly sized online UPS with the required battery capacity can be sufficient without building a data-center-level redundant system.

3. Factory: starting current and environment

Industrial loads are where UPS selection often goes wrong.

A 3kW heater and a 3kW motor are not equivalent loads.

A motor can draw several times its rated current during direct-on-line starting. If the UPS has enough capacity for the motor's running load but not enough short-term current capability, the motor may fail to start or the UPS may enter overload or bypass.

Before selecting the UPS, collect:

• Motor power
• Power factor
• Running current
• Starting current
• Starting method
• UPS overload curve
• Other loads on the same output

Don't rely on a rule such as "motor kW × 3." A motor with a soft starter or VFD can behave very differently from a motor using direct-on-line starting. A practical starting point is rated power × 1.5 plus the measured starting inrush, then verify the result against the UPS overload curve rather than a fixed multiplier.

The installation environment also matters. Dust, oil, heat and vibration may require a different enclosure and cooling arrangement than a clean air-conditioned server room. The required IP rating should be selected according to the actual installation conditions rather than simply specifying IP54 or IP65 as a general rule.

For more detail, see our guide on UPS sizing for inductive loads.

Three applications, different priorities

The UPS should be selected from the load

There is no universal "best UPS."

A hospital may need isolation and continuous regulated power. A bank may need redundant power paths and enough battery time to bridge the generator start sequence. A factory may need a UPS capable of handling motor starting current and harsh environmental conditions.

This is why the first step in a UPS project should always be a load list.

List the equipment, power, voltage, power factor, starting characteristics and required backup time before choosing the UPS capacity.

Don't ask only how big the UPS should be. Ask what the load needs the UPS to protect against.