If you're specifying an ABB UPS for a substation or industrial process, stop sizing for 'worst-case' simultaneous full load. It's the single most expensive mistake I made repeatedly—costing clients roughly $12,000 in unnecessary hardware and installation over three projects before I learned what the field data was actually telling me. The right approach, which I'll explain below, saves money and increases system reliability, but it feels deeply counterintuitive at first.
When I first started handling power protection orders in 2018, I assumed the safest route was the biggest UPS I could fit in the budget. The reasoning seemed bulletproof: 'Industrial loads spike, and ABB builds rock-solid gear, so let's put in a 500 kVA unit for a load we calculated at 350 kW.' Five years and several painful conversations later, I realize I was burning capital on unused capacity and missing the real risk: how a partially loaded UPS actually performs.
The Initial Mistake: My 'More Is Better' Rule
My initial approach to UPS sizing was completely wrong. I thought a larger ABB UPS meant more safety margin, longer battery runtime, and happier clients. The first major error was in 2020 on a water treatment plant upgrade. We spec'd an ABB PCS100 UPS rated for 250 kVA to protect a SCADA system whose measured peak draw was never above 120 kVA.
It looked fine on my spreadsheet. The result? A $4,200 overpay on the UPS itself, plus another $1,800 in unnecessary cabling and floor space. The client paid for cooling capacity that was never needed. Three budget overruns later, I learned about something called 'total cost of ownership'—and the fact that running a large UPS at under 40% load degrades its efficiency and can shorten component life (Source: ABB Technical Application Papers on UPS Efficiency Curves, 2019).
(Should mention: we'd also assumed the load would grow over 5 years. It didn't. The SCADA upgrade stayed within the original specs.)
What the Data Actually Says About ABB UPS Sizing
Here's what I wish someone had told me in 2018: ABB UPS systems—particularly the PCS100, PVS series for solar integration, and the DPA 250—achieve their highest efficiency (often 96-97%) between 50% and 80% load. Below 30-40% load, efficiency can drop to 88-92%.
The numbers said size for actual measured peak load plus a 20-25% growth buffer. My gut said that was too aggressive. Went with my gut on that first project. Turns out the 'expert' who told me to oversize was giving advice based on 1990s transformer technology, not modern IGBT-based UPS design.
According to ABB's official UPS sizing guide (available at new.abb.com/ups), the recommended practice for industrial applications is to size based on 'actual load profile data, not nameplate ratings of connected equipment.' I ignored that guidance for 18 months and paid for it.
A Better Method: The 24-Hour Load Log
What I do now:
- Measure, don't guess. Run a 24-hour load log on the actual switchboard. I once found that a 'critical 200 kVA load' was actually two 30 kVA pumps running alternately, with a peak of 65 kVA.
- Size for the peak measured load plus 25%. Not the theoretical worst case. Real loads seldom coincide at full rated power.
- Add parallel redundancy, not oversizing. For a 160 kVA measured load, I now spec two 100 kVA ABB DPA UPS units in N+1 configuration. Total capacity is higher than needed, but each unit runs at 80% load—optimal efficiency—and I get true redundancy.
This method reduces upfront cost by 20-30% compared to my old 'big single unit' approach. (Note to self: should write this up as a formal checklist for the team.)
When Oversizing Still Makes Sense (The Exception)
This approach has limits, and I'd be dishonest not to mention them. I can only speak to my experience with continuous process loads, data centers, and utility substations. If you're dealing with high-inrush loads like large motor starts, welding equipment, or pulsed power applications, the calculus is different. In those cases, the inrush current can briefly exceed 3x the steady-state load, and a moderately oversized UPS might be the safer call.
For example, on a pipeline compressor station where the motor starting current hit 450% of running load, we had to spec a unit 50% above running load just to handle the 2-second transient. That's valid oversizing—not the 'fear-based' kind I was doing.
I'd also offer a caveat about battery sizing. Even with a correctly sized UPS, if your client needs 30 minutes of runtime to execute a controlled shutdown for a critical transformer, oversize the battery bank, not the inverter. ABB's modular battery cabinets make this easy to adjust without upsizing the power module.
Lessons Applied: A Solar Farm Example
In Q2 2024, we quoted an ABB UPS for a 15 MW solar farm's control room. The original design specified a 100 kVA unit based on 'total panel power plus a 50% safety margin.' I ran the 24-hour load log.
The actual peak: 22 kVA (SCADA, inverters, lighting, HVAC for the small control room).
We installed a 40 kVA ABB PCS100 with a parallel battery cabinet for 45 minutes of runtime. The client saved approximately $9,000 compared to the original 100 kVA quote. The system runs at 55% load—right in the efficiency sweet spot.
Oh, and we added a note in the commissioning report: 'UPS sized per measured load profile. Verify annually as site expands.' (I really should add this to our standard proposal template.)
Key Takeaways for Specifiers
This worked for us, but our situation was typical industrial process control—predictable loads, non-pulsing, with room for battery expansion. If you're dealing with critical infrastructure where any voltage dip causes a million-dollar process interruption, you might need more margin than I'm suggesting.
The value of correct sizing isn't just the upfront cost saving—it's the running efficiency, longer component life, and—counterintuitively—higher reliability under normal operating conditions.
Prices as of Q4 2024 for ABB UPS systems in the 20-200 kVA range run approximately $400-700 per kVA for the base unit, excluding batteries and installation (verify current rates with your ABB channel partner). The cost of getting it wrong isn't just the hardware premium—it's the wasted floor space, cooling load, and the efficiency penalty you'll pay for the next 10-15 years of operation.
Honestly? I still get nervous on every sizing call. The data says one thing; the fear of brownouts says another. But I've learned to trust the load log. In 47 projects over the past 18 months, we've caught 4 potential oversizing errors using this method. Each one would have cost $2,000-5,000 unnecessarily—and that's a record I'm happy to stand behind.
Jane Smith
I’m Jane Smith, a senior content writer with over 15 years of experience in the packaging and printing industry. I specialize in writing about the latest trends, technologies, and best practices in packaging design, sustainability, and printing techniques. My goal is to help businesses understand complex printing processes and design solutions that enhance both product packaging and brand visibility.