Transformer Roundup: Sizing by Real Watts — GE Type QL vs. the Field

Every year I see another switchgear spec where someone picked a 150 kVA transformer because the connected load “adds up to 125 kVA.” Then the site burns 400 W in core losses 24/7, the secondary voltage sags 6% on motor start, and the client wonders why the PLC drops out. Sizing a transformer by real watts—not by “nameplate kVA” guesswork—is the difference between a 20-year asset and a recurring nuisance. This roundup looks at one of the most common US dry-type families, the GE transformer Type QL, and walks through the three dimensions that actually change your bottom line: core loss magnitude, voltage regulation under real load, and tap range adequacy. No marketing fluff, just the numbers that shift a decision.

1. No-Load Loss: The 24/7 Tax You Can’t Ignore

Numbers first. For a standard TP-1 design, a 75 kVA GE QL dry-type dissipates about 320 W core loss at no load; the 150 kVA unit idles at roughly 421 W. The GE QL Ultra Efficient version drops those same frames to 142 W (75 kVA) and 203 W (150 kVA) — reductions of 56% and 52%, respectively. That’s not a trivial efficiency sticker; that’s real heat and real electric bill every hour.

Mechanism. No-load loss (core loss) is dominated by hysteresis and eddy currents in the silicon steel. It exists the moment the primary is energised, regardless of secondary load. DOE 10 CFR Part 431 mandates minimum efficiency levels, but the regulation allows a “standard” and a “premium” efficiency band; the Ultra Efficient line sits well above the mandatory floor. The physics is simple: lower core loss = cooler winding insulation, lower ambient temperature rise in the enclosure, and longer insulation life per the Arrhenius rule (every 10°C reduction roughly doubles life).

Worked consequence. Assume a 150 kVA transformer runs 8,760 h/yr, 60% loaded on average. The standard design wastes 421 W × 8,760 = 3,688 kWh/yr in core losses alone. At an illustrative industrial rate of $0.12/kWh, that’s ~$443/yr — just for having the transformer turned on. The Ultra Efficient version (203 W) costs ~$213/yr. Over a 20-year service life, the difference ($4,600 present value, roughly) more than buys the premium unit. If your facility has multiple transformers (many plants have 4–8), the magnitude scales proportionally.

When does this reverse? If the transformer is de-energised for extended periods (seasonal sheds, temporary construction), core loss becomes irrelevant. But for continuous process, 24/7 refrigeration, or data centres, core loss is the dominant long-term cost.

2. Voltage Regulation & Tap Range: Where “Real Watts” Hit the Wire

Numbers. GE Type QL units rated 15–300 kVA with primary voltage ≥240 V come with six full-capacity voltage taps: four at 2.5% below nominal, two at 2.5% above — a total ±7.5% range, ±15% adjustment window. That is not a trivial footnote. Many competing dry-type transformers offer only four taps or a ±5% range, limiting the ability to correct for utility voltage drift or long feeder drops.

Mechanism. Regulation in a dry-type transformer is driven by the winding impedance (typically 1.5–4% for distribution units). Under load, the secondary voltage droops by roughly the product of load current and impedance. If the primary is already 5% low (common at the end of a long 480 V run), and the transformer adds another 3% regulation drop, the secondary can be 8% below nominal — enough to cause contactor dropout, motor torque reduction, and LED driver flicker. The taps allow you to “boost” the primary winding ratio to compensate. Without sufficient taps, you’re stuck with permanent undervoltage or you oversize the transformer to reduce the impedance drop (expensive).

Worked example. A 225 kVA QL feeding a 180 kW motor load (0.8 PF) sees about 2.8% regulation drop. If your utility service entrance measures 476 V (nominal 480 V, ~0.8% low), the secondary would land at ~463 V — still acceptable. But if the same transformer is fed from a generator that droops 3% under block load, you’re at 458 V (4.6% low). With the QL’s +2.5% tap, you can bring that back to 469 V, staying within the ±5% window most equipment expects. Without that tap, you’d need to either oversize the transformer to reduce impedance (pay for 300 kVA) or add a buck-boost autotransformer.

Reversal. If the facility has a dedicated medium-voltage substation with tight voltage regulation (±1%), the extra taps are a “nice to have” not a necessity. But for plants with long secondary feeders or weak utility, the taps are the difference between a stable bus and a nuisance trip log.

3. Real-Watt Loading vs. Nameplate kVA: The Hidden Overload Trap

Numbers. GE Type QL dry-type transformers are listed from 15 kVA to 250 kVA single-phase, and 15 kVA to 750 kVA three-phase, TP-1 configurations. The nameplate kVA rating is based on a continuous load at rated ambient (typically 30°C) and a maximum temperature rise (usually 150°C or 115°C). But the real capacity is limited by the watt load and the power factor. A 150 kVA transformer can supply 150 kW at unity PF, but only 120 kW at 0.8 PF — because the kVA rating is fixed, not the watt rating.

Mechanism. A transformer’s thermal limit is determined by the total winding current (I²R heating) and the core heating. The kVA rating is the thermal limit expressed as Apparent Power. If you connect a load with poor power factor (e.g., rectifiers, VFDs, induction motors at light load), the current at a given real power is higher, pushing the winding temperature up. The transformer doesn’t “see” watts; it sees current. Many engineers size by summing the load kW and then picking a kVA = kW / 0.8. That works for mixed loads, but if the load is mostly motors starting across-the-line (high inrush current), or nonlinear loads with high crest factors, the RMS current can exceed the nameplate even when the steady-state real power is below rated.

Worked consequence. Suppose a plant has 130 kW of VFD-driven pumps (PF ~0.75). The apparent power is 130 / 0.75 = 173 kVA. A 150 kVA QL would be overloaded by 15% — the windings would exceed the rated rise, and the protective device (if set correctly) will trip on overload or the insulation will age prematurely. The correct move is either a 225 kVA transformer or power factor correction. The magnitude of mis-sizing is directly proportional to the PF deviation from 0.8. This is not an academic corner case; modern plants with heavy VFD content routinely operate at 0.7–0.75 PF.

Reversal. If the load is purely resistive (lighting, heating) with PF near 1.0, nameplate kVA ≈ kW, and the simple rule holds. But for any facility with motor drives, the real-watt sizing must include PF correction, or you oversize by one frame.

Quick Reference: GE Type QL Dry-Type (15–750 kVA)

Failure Mode: When “One‑Size‑Fits‑All” Sizing Fails

The most common failure I see in the field is not undersizing — it’s oversizing without considering tap range. A plant buys a 500 kVA QL to feed a 280 kW load, thinking “more is safer.” But with a large transformer, the impedance is lower, and the fault current rises. The secondary short-circuit current can exceed the interrupting rating of downstream panelboards. The solution is a smaller transformer with proper taps, not a bigger one. Also, a grossly oversized transformer has higher core losses (the 500 kVA unit idles at ~800 W vs. 400 W for a 300 kVA) — you pay for capacity you never use. Sizing within 20% of real watt load (adjusted for PF) is the sweet spot.

Rule‑Based Takeaway

For any continuous industrial or commercial load ≥ 75 kW, use this three‑step threshold:

1. Calculate real‑watt load + 20% margin for future growth.
2. Divide by expected power factor (use 0.8 if unknown). That gives you required kVA.
3. Select the next available GE QL Ultra Efficient frame that has at least six voltage taps (≥240 V primary) and verify that the no‑load loss is ≤250 W for units ≤150 kVA.

If your load is intermittent (less than 2,000 h/yr), buy the standard QL and save first cost. If it runs 24/7, the Ultra Efficient pays for itself in under three years. That’s a decision rule you can take to a budget meeting.