UPS Systems: Sizing According to Your Business Criticality

Calibrating a tertiary UPS rests on three quantifiable decisions, which determine most of the total cost of ownership.

• Runtime versus investment. A 15-minute runtime represents a noticeable additional cost compared with a 10-minute target, but covers the majority of micro-outages observed in densely populated urban areas. Beyond 20 minutes, the cost/benefit curve becomes unfavourable outside critical sites.
• Rated load versus scalability. Sizing at 70% of the maximum load preserves headroom for growth while keeping efficiency at the peak of the manufacturer’s curve, generally located between 50 and 75% load.
• N+1 redundancy versus budget. The N+1 configuration increases the initial investment but very significantly reduces the risk of downtime. It is justified once the cost of a business interruption is high.

For the CFO and asset manager: convert risk into EUR before signing. The UPS trade-off is not an engineering question, it is a cash-flow question. A trading floor with a 50,000 EUR/h interruption cost justifies N+1 redundancy; a back office at 800 EUR/h does not. Our reading differs from common practice in electrical engineering firms: contrary to the implicit rule « N+1 everywhere above 30 kVA », Kytom recommends a single UPS backed by an on-demand rented generator below 5,000 EUR/h of documented business interruption cost. On the simple sites we have supported, none retrospectively justified a move to an N+1 architecture: the avoided additional cost represents several tens of thousands of euros per site, equivalent to more than a year’s rent for a 200 m² open-plan office.

When this framework does not apply. Below 10 kVA of effective load, or for a site whose business interruption cost remains under 500 EUR per hour, the N+1 trade-off is counterproductive. Likewise, a runtime target above 30 minutes falls outside the scope of battery UPS: you need to switch to generator coupling with an automatic transfer switch, otherwise you risk investing in an oversized battery bank whose ROI exceeds 10 years.

Four common mistakes: undersizing, thermal, ageing, testing

Across the audits carried out since 2006, four mistakes recur regularly in existing installations.

1. Underestimating the actual load. Most installations are set to the rated power of the equipment rather than to the effective consumption, which is significantly lower than the rated value. Working on effective consumption rather than rated power avoids costly oversizing.
2. Neglecting temperature rise. Above 25 °C, battery runtime degrades noticeably: an unconditioned technical room constitutes a risk factor to anticipate from the design stage. A room without dedicated air conditioning significantly accelerates battery ageing; thermal control of tertiary electrical rooms remains a systematic checkpoint in the Kytom method.
3. Ignoring battery ageing. The capacity of lead-acid or lithium batteries degrades progressively after a few years of standard use. Kytom practice incorporates a derating coefficient from the design stage and plans a progressive battery replacement to smooth the operating budget.
4. Omitting load tests. Load tests at handover detect faults that are invisible without real stress: switchover, harmonics or voltage drift.

When these corrections are not a priority. On a site whose effective load remains below 5 kVA and which already has a dual-feed power supply, investing in advanced monitoring has no measurable ROI: an annual check is enough. Systematically correcting these four points significantly reduces continuity incidents from the first year; it is justified above all beyond 20 kVA and for businesses with a documented interruption cost.

The Kytom method in 4 steps: audit, 72 h measurement, corrected calculation, load test

The sizing procedure applied since 2006 is carried out in 4 sequential steps.

• Step 1: business criticality audit. Mapping of equipment into three levels (vital, important, standard) and costing of the interruption cost per minute. This step involves the IT department, the office manager and finance management.
• Step 2: measuring actual loads. 72 h monitoring of effective consumption, with readings of peaks, averages and harmonics. For a 600 m² floor accommodating 70 workstations, the measured load is generally between 18 and 28 kVA effective, whereas the contracted power often exceeds 50 kVA: a common gap that justifies real measurement before any sizing.
• Step 3: corrected runtime calculation. Application of temperature coefficients (derating in hot ambient conditions), ageing (× 0.8) and actual efficiency (85 to 90% depending on the technology, online or line-interactive).
• Step 4: validation through load testing. Simulation of a mains failure, measurement of switchover time (target < 10 ms in category VFI-SS-111) and verification of effective runtime.

For the interior architect and the project owner: the 72 h measurement determines the electrical heat load of the UPS room. An underestimated sizing may require reworking the ventilation of the UPS room, generally not budgeted. The 4 steps mobilise 6 to 10 man-days on a project of 500 to 1,000 m², i.e. 1.5 to 2% of the high-current works budget.