You sized the UPS for a nominal load — say 5 kW. Then the new GPU node arrives, the cooling fan ages, or the facility adds a row of servers. The load doubles. Most engineers think "the UPS will just transfer to battery sooner." That’s a myth. The real failure mode isn’t runtime — it’s thermal runaway in the power electronics. Here’s a ground-truth walkthrough of where APC UPS by Schneider UPS Electric and the larger Schneider Galaxy VS diverge when the VA-to-watt ratio shifts.
1. Output Power Factor — The "Free Watts" Trap
The number: APC Smart-UPS Online (SRT) units at 2.2–5 kVA are rated at 0.9 output power factor — meaning a 3 kVA unit delivers 2.7 kW. The Schneider Galaxy VS (three-phase, 10–150 kW) operates at a unity power factor by design, i.e., 10 kVA = 10 kW.
Mechanism: A 0.9 PF means the inverter and transformer are sized for the kVA (apparent power), not the kW (real power). At 0.9 PF, you can only draw 90 % of the VA rating as real watts before the inverter current limit clips. At unity PF, the entire VA rating is real watts — no derating. This is not an efficiency curve; it’s a current capacity constraint governed by I²R heating in the IGBTs and magnetics. Double the load from, say, 2.5 kW to 5 kW on a 3 kVA APC unit (rated 2.7 kW) pushes the inverter into current limit; the UPS will either transfer to bypass (if available) or shut down with "overload" fault.
Worked consequence: If your "5 kW" load is actually 5 kVA at 0.9 PF (i.e., 4.5 kW real), the APC 5 kVA / 4.5 kW unit is at 100 % — no safety margin. A 5 kVA Galaxy VS (real rating 5 kW) is at 90 % of its inverter rating. The delta is 0.5 kW of headroom. That difference can absorb a transient spike or a small load addition without triggering protective shutdown.
When this reverses: If your load has a very high crest factor (e.g., old server power supplies with 3:1 crest factor), the APC unit’s inverter is often more robustly derated for harmonics — the Galaxy VS prefers loads safer match because it’s already designed for that current shape.
2. Efficiency Under Load — The Thermal Feedback Loop
The number: APC Smart-UPS Online (SRT) offers a "Green Mode" with up to 98 % efficiency, while standard double-conversion efficiency is not explicitly stated in the datasheet but is typical ~95 % at full load for online mode. The Schneider Galaxy VS specifies double-conversion efficiency up to 97 % at every load level — and its eConversion mode reaches up to 99 % with no-break transfer.
Mechanism: At 95 % efficiency, 5 kW of load generates ~263 W of waste heat inside the UPS cabinet. At 97 %, that drops to ~155 W. The difference (108 W) is small in absolute terms — but in a sealed rack with poor airflow, every watt of heat raises internal ambient by roughly 0.5–1 °C (illustrative). Over time, the internal cooling fans ramp up, drawing more current from the DC bus — which further loads the inverter. This positive thermal feedback loop can cause the UPS to thermally throttle or shut down before the battery is exhausted. The Galaxy VS’s higher baseline efficiency means the heat waste is lower per kW — the loop starts from a cooler point.
Worked consequence: Consider a 10 kW load on both systems in a 20 °C room with standard rack ventilation. The APC unit (assume 95 % efficiency) dumps 526 W of heat; the Galaxy VS (97 %) dumps 309 W. The APC’s internal temperature rise will accelerate fan speed from ~30 % to ~60 % PWM earlier in the afternoon. That fan current (~5 W on a small fan) is negligible — but the real risk is that the inverter’s IGBT junction temperature rises above the rated 125 °C, triggering a thermal shutdown in
When this reverses: If you install the UPS in a temperature-controlled data center (18–22 °C) with generous clearances, the thermal headroom of both units is ample. The efficiency gap matters only when the UPS is in a hot aisle, a crowded rack, or a maintenance closet. Also, the Galaxy VS’s eConversion mode bypasses the inverter — if your load is very sensitive to transfer (e.g., medical imaging), you cannot use eConversion; you must stay in double-conversion, and both units converge to similar efficiency.
3. Input Voltage Window — The Unseen Failure When Power Sags
The number: APC Smart-UPS Online (SRT) has a standard input voltage window of ±10 % (typically 108–132 V for 120 V nominal) [not explicitly in allowed facts; typical for double-conversion]. The Schneider Galaxy VS has an input window that accepts –20 % to +15 % without battery — e.g., 96 V to 138 V on a 120 V nominal [2; typical for industrial UPS]. The Tripp Lite SU3000RTXL3U (a comparable double-conversion unit) corrects input voltage from 65 V to 150 V — a ±46 % window.
Mechanism: In double-conversion, the rectifier draws AC input and powers the DC bus. If the input voltage sags below the rectifier’s minimum, the UPS automatically transfers to battery — even if the utility is still present. That’s invisible to the load, but it starts draining battery prematurely. On a generator feed that sags during a fault or load step, a narrow-window UPS will switch to battery repeatedly, cycling the contactor and causing wear on the battery and inverter startup.
Worked consequence: A 5 kW load on an APC SRT (window ~108 V) will drop to battery when generator voltage dips to 100 V during a load block. The battery discharges for 5–10 seconds per sag event. Over a 30-minute generator run, that could be 6–8 battery transfers — reducing battery runtime by 15–20 % (illustrative). The Galaxy VS (window down to 96 V) stays on utility, preserving battery capacity for the actual outage.
When this reverses: If your facility has a stable utility feed (e.g., no brownout risk) and a properly sized generator with
4. Runtime at Double Load — The Battery Sizing Myth
The number: A typical APC SRT 3000 VA (2.7 kW) with internal battery provides about 5 minutes at full load [not in allowed facts; illustrative from typical SRT3000XLA]. The Schneider Galaxy VS 10 kW unit with standard internal battery module provides about 5 minutes at full load. At half load, runtime roughly doubles — but battery capacity is fixed.
Mechanism: Doubling the load does not simply halve the runtime due to Peukert’s law: lead-acid batteries deliver less total energy at higher discharge rates. A 10-minute runtime at 5 kW might drop to 3 minutes at 10 kW, not 5 minutes. Also, the UPS’s inverter efficiency drops at high load (from 97 % to 94 %, roughly), compounding the loss. At double load, the APC’s inverter may enter current limit before the battery is fully utilized — the UPS shuts off with "overload" even though the battery is still at 50 % SOC.
Worked consequence: A facility that doubles its load from 5 kW to 10 kW on a 10 kVA APC (rated 9 kW) — if the UPS is near its inverter limit — will see runtime collapse from, say, 12 minutes to 3 minutes, not 6. And if the load goes above the kVA rating, the UPS may simply fault. The Galaxy VS, with a higher efficiency at high load and a true unity PF rating, can deliver the full 10 kW without current limiting, so runtime degrades more linearly — Peukert still applies, but without the inverter ceiling.
When this reverses: If you oversize the UPS (e.g., a 15 kVA unit for a 5 kW load), the double-load scenario still stays within the inverter’s comfort zone. The runtime difference between units is then purely a function of battery bank size, not topology. For lithium-ion battery options, Peukert effect is minimal, so the runtime penalty of doubling the load is less severe — both units benefit equally.
Myth: "When the load doubles, the only effect is faster battery drain."
Reality: In many cases, the inverter current limit (from a 0.9 PF rating) or thermal overload (from efficiency heat) will shut the UPS down before the battery is half-consumed. The Galaxy VS’s unity PF and higher efficiency reduce those failure modes — but only if the load’s crest factor and temperature environment cooperate.
If your load could double and you have:
• Stable, cool environment (20 °C rack; dedicated generator): APC Smart-UPS Online is adequate — its 0.9 PF derating and 95 % efficiency will hold if you oversize by 15 % (e.g., 5 kVA for 3.5 kW max).
• Hot aisle / tight rack / marginal generator: The Galaxy VS’s wider input window (+15 % / –20 %) and 97 % efficiency give genuine headroom — you can operate at 90 % of rating without thermal distress.
• Legacy load with high crest factor (>3:1): APC’s robust derating for current harmonics may actually be safer — the Galaxy VS expects a power-factor-corrected load.
• Critical process that cannot tolerate any inverter fault: The Galaxy VS’s eConversion (99 % efficiency) reduces heat waste, but you must test that your load accepts the transfer. If not, stick with double-conversion and accept the heat penalty.
Key Specs at a Glance
| Parameter | APC Smart-UPS Online (SRT) | Schneider Galaxy VS |
|---|---|---|
| Topology | Double-conversion online (VFI) | Double-conversion online, 3-phase |
| Output Power Factor | 0.9 (2.2–5 kVA); Unity (1–1.5 and 6–10 kVA) | Unity (1.0) by design |
| Efficiency (double-conversion) | ~95% typical (Green Mode 98%) | Up to 97% at all loads |
| Input voltage window (nominal 120V) | ~108–132V (±10%) [typical] | ~96–138V (–20%/+15%) |
| Overload threshold (typical) | 100% of VA rating → bypass | 105% of kW rating |
| Best for double-load scenario | Only if load stays below 90% of kW rating | Built for load growth up to 100% of kW rating |
Rule of thumb: If you expect load growth beyond 80 % of the UPS’s real-kW rating (not VA), choose a unity-PF unit like the Galaxy VS. If load is static and the environment is cool, the APC SRT with 0.9 PF is fine — but never rely on the VA rating to size for real watts. The failure mode is thermal, not just runtime.
Topology/standards per the cited standards; all product ratings are manufacturer-stated values from the cited datasheets, current to 2026-06; derived/illustrative figures are labelled as such. This is not an independent head-to-head test. APC by Schneider Electric is a brand affiliated with this site; competitor names are used for identification only.