Critical IT Load vs Total Building Load: What Should a UPS Protect?

When a site plans a new UPS, the starting question is often 'How big is the building load?' That can send the project in the wrong direction. A UPS is normally intended to protect a defined group of critical loads, not every socket, fan, heater and mechanical service in the building.

The design task is to identify which services must remain available through a disturbance, which need enough time for an orderly shutdown and which can wait for generator power or normal mains restoration. Only then can capacity, runtime and redundancy be assessed sensibly.

Quick answer: what should a UPS protect?

A commercial UPS should protect the equipment and supporting services required to meet the site's continuity objective. This can include servers, storage, network, communications, security, controls and selected operational technology. It may also include the local cooling, control or auxiliary loads without which the IT equipment cannot remain useful.

The boundary is site-specific. The right answer depends on consequence, shutdown time, generator strategy, equipment architecture and acceptable risk.

Critical IT load is not the same as total connected load

Total building load includes everything drawing power at a point in time or capable of being connected. Critical IT load is the smaller set of technology services that must be supported to achieve the continuity plan. UPS input demand may be different again because it includes conversion losses and battery charging.

1. IT equipment that must remain continuously available
2. systems that require an orderly shutdown rather than an abrupt loss
3. network and communications dependencies
4. security, access or monitoring systems tied to the response plan
5. control systems required to keep a process safe
6. supporting cooling or control loads needed for the intended runtime

A label such as 'server room' is not enough. The load list should identify circuits, equipment, ownership and the reason each item is or is not protected.

Start with business services, not rack nameplates

Begin by listing the services the organisation cannot afford to lose. Map each service to its technology and physical dependencies. This often reveals small but essential items—a network switch, storage controller, authentication service or environmental control—that are easy to omit when attention is focused on servers alone.

For each service, define the required outcome: uninterrupted operation, ride-through until generator transfer, time for controlled shutdown, or no UPS support. This turns a vague criticality discussion into a testable design brief.

Three useful load categories

1. No-break load

Equipment in this category cannot tolerate even a short interruption within the design scenario. It normally remains on the UPS throughout mains disturbances and generator transfer.

2. Controlled-shutdown load

This equipment needs stored energy long enough to save state, complete a safe sequence and shut down. The required runtime should include detection, decision and human response time, not just the software shutdown duration.

3. Delayed or non-UPS load

Some loads can stop briefly and restart when generator or mains power is stable. Keeping high-demand non-critical equipment outside the UPS boundary can reduce cost and avoid unnecessary battery capacity.

Do not forget the supporting dependencies

A server may keep running while the service still fails because a dependency was excluded. Network routes, DNS, storage, firewalls, telecoms and authentication may sit in different rooms or on different distribution boards. Building controls and cooling can also become limiting factors during a longer interruption.

• network switches, routers, firewalls and carrier equipment
• shared storage and backup appliances
• console, monitoring and remote-access systems
• local cooling controls, pumps or fans required for the autonomy period
• generator, ATS and BMS control or communication interfaces
• dual-corded equipment whose two feeds are not genuinely independent

How to establish the real load

Equipment nameplates show maximum or rated values, not necessarily normal demand. A load survey can use UPS data, branch monitoring, power-distribution measurements and equipment records to understand steady demand and peaks. Measurements must be obtained safely by competent people.

Record kW and kVA where available because power factor affects UPS loading. Allow for growth, but make the allowance evidence-led. Excessive oversizing can reduce efficiency, increase cost and produce a design that is difficult to operate near its intended range.

Account for load steps and inrush

Some devices draw a higher current when they start or change state. The UPS must be assessed for both sustained demand and transient behaviour. Printers, motors, compressors and certain power supplies can create a disproportionate peak and may be better supplied outside the critical UPS path.

Decide the runtime from the operating plan

Runtime is not a universal ten- or fifteen-minute number. It should reflect what the site needs to do. If the UPS bridges to a standby generator, the autonomy must cover detection, generator start, stabilisation, transfer and a sensible contingency. If there is no generator, it may need to support a controlled shutdown or a longer business-continuity period.

Battery runtime changes with load, battery condition, temperature and age. The design value and the available value in service are therefore related but not identical.

Coordinate the UPS with generator and bypass supplies

A UPS and generator should be considered as one power path. Check generator capacity, voltage and frequency behaviour, UPS input characteristics, battery recharge demand and the sequence in which loads return. The bypass source must also be suitable for the intended operating and maintenance states.

If the UPS is modular or parallel, define what happens when a module is unavailable. Capacity marked N+1 does not automatically mean the battery, bypass and downstream distribution have the same redundancy.

Common load-boundary mistakes

1. using total building demand as the UPS sizing value
2. protecting servers but not their network or storage dependencies
3. adding comfort cooling without confirming the actual critical cooling requirement
4. relying only on equipment nameplates
5. ignoring power factor, inrush or future load steps
6. assuming dual power cords provide independent paths
7. allowing years of load additions without reviewing runtime
8. specifying runtime without a shutdown or generator-transfer plan

A practical UPS load schedule

For each item or circuit, record the service supported, owner, location, normal kW/kVA, peak behaviour, required runtime, power-path category and shutdown method. Note whether the item is single- or dual-corded and which upstream boards supply it.

Review the schedule during major IT projects, moves and changes, generator upgrades and UPS battery replacement. The most useful load list is a controlled operational record, not a one-off design spreadsheet.

Define the load before choosing the UPS

A clear critical-load schedule links business requirements to circuits, equipment, runtime and redundancy. It also gives engineers a reliable basis for sizing and future change.

Contact P&I Group for a commercial UPS load review, condition assessment or system upgrade plan.