“The kVA number is a lie if you ignore voltage drop” — The Spec That Actually Fails First in a Dry-Type Transformer

Every facility manager I talk to starts with the same assumption: buy a transformer rated for my load, and it’ll handle the load. That’s true only if the load behaves like a perfect resistor. Real loads — motor starters, VFDs, UPS rectifiers, even LED banks with inrush — don’t. The spec that actually fails first in a dry-type distribution transformer isn’t the nameplate kVA. It’s voltage regulation under high inrush, followed by continuous overload from harmonic heating, and finally no-load loss creep over 20 years.

This roundup uses the GE Type QL family (15–750 kVA, three-phase TP-1) as the reference case, because the QL datasheet gives enough hard numbers to run the three failure modes. I’ll walk each case: what the spec says, why it fails first under real conditions, and when you can ignore it. The goal is a decision rule, not a shopping list.

Case 1: Voltage regulation — the silent first failure

When a motor or a bank of switching power supplies energises, the transformer sees a current inrush that can be 8–12× rated for the first half-cycle. The voltage at the secondary terminals dips. If the dip exceeds ~8–10% of nominal, downstream contactors drop out, PLCs brown out, and production stops. That failure mode is voltage regulation — and it has nothing to do with the transformer’s steady-state kVA rating.

GE transformer’s QL design guide specifies that standard units from 15 kVA through 300 kVA with primary voltage of 240 V or higher have six voltage taps (four below, two above) for a ±7.5% adjustment range. That tap range exists precisely to compensate for voltage drop under load — but it cannot correct for transient dips shorter than the tap-changer response (nonexistent on a dry-type). The actual voltage regulation depends on the transformer’s percent impedance (%Z). GE Type QL transformers list %Z values between ~1.5% (small units) and ~5.5% (large). A 75 kVA QL with 2.8% Z will have roughly 2.8% voltage drop at full rated current. But when you slam it with an inrush of 8× rated current for 10 ms, the instantaneous drop is approximately 8 × 2.8% = 22.4% — enough to trip undervoltage relays on any 480 V system.

Worked consequence: For a 150 kVA QL feeding a 50 hp motor (roughly 65 A FLA) with a typical 6× inrush, the voltage dip is about 6 × (65/208) × %Z ≈ 6 × 0.31 × 3.5% ≈ 6.5% — marginal for most MCCs. But if you add a second motor starting simultaneously (common in conveyor systems), the dip doubles to 13%. That’s a failure. The spec that fails first is not the kVA; it’s the %Z – inrush margin.

When it doesn’t apply: For resistive loads (electric heaters, incandescent lighting at reduced scale) or loads with soft-starters that limit inrush to 2–3× FLA, voltage regulation is a non-issue. In those cases, the steady-state kVA rating governs.

Case 2: Harmonic heating — the continuous overload that doesn’t show on an ammeter

A standard dry-type transformer is designed for sinusoidal current. Modern loads (VFDs, UPSs, LED drivers) draw non-sinusoidal current rich in harmonics. The harmonic content increases RMS current without increasing fundamental power — meaning the transformer’s copper losses (I²R) rise, and the core sees additional eddy-current losses. The result: the transformer heats up faster than the kVA meter suggests.

The GE Type QL is a dry-type construction with a core of grain-oriented silicon steel and copper windings. For a given load, harmonic heating reduces the usable capacity. The IEEE C57.110 standard recommends derating a standard transformer by a factor called K-factor. A load with a K-factor of 13 (typical of a 6-pulse VFD) requires the transformer to be derated by about 30–40% (derived per IEEE C57.110 methodology; not a manufacturer-stated value). That means a 150 kVA QL feeding a VFD drive panel can only deliver about 90–105 kVA without exceeding its rated temperature rise (typically 150°C rise for a 220°C insulation system).

Worked consequence: If you size a 150 kVA QL for a 120 kVA VFD load, you’re actually loading it at 120/150 = 80% of nameplate — but with K-13 harmonics, the effective loading is ~112% of the derated capacity. The transformer will run hot and the winding insulation will age faster. The spec that fails first is thermal capacity under non-sinusoidal load.

When it doesn’t apply: For linear loads (motors without VFDs, resistive heaters, incandescent lighting) harmonic content is negligible; the transformer can be loaded to its nameplate. Also, if you specify a K-rated transformer (e.g., K-13 or K-20), the derating is baked into the design — no surprise failure.

Case 3: No-load loss creep — the 20-year spec that eats your budget first

Every kilowatt of no-load loss (core loss) runs 8,760 hours a year, whether the transformer is loaded or not. Over 20 years, even a modest difference in no-load loss can exceed the purchase price. This is the spec that fails first in a TCO comparison, not an operational failure — but for the facility manager who has to justify the capital, it’s the first spec that fails the ROI calculation.

GE’s QL Ultra Efficient line shows the gap: a 75 kVA standard TP-1 unit has no-load loss of 320 W; the Ultra Efficient version drops to 142 W. For a 150 kVA unit, the reduction is from 421 W to 203 W. Assuming an industrial electricity rate of $0.10/kWh (illustrative), the saving per year for the 150 kVA unit is (421–203) × 8,760 / 1,000 × 0.10 = ~$191/year. Over 20 years, that’s $3,820 — roughly the premium of the Ultra Efficient model. The spec that fails first here is simple payback period; if the utility rate is $0.08/kWh, payback stretches beyond 25 years, and the standard unit wins.

Worked consequence: For a 500 kVA QL (three-phase, TP-1), standard no-load loss is around 1,100 W (derived from the datasheet trend). An Ultra Efficient version might cut that to ~600 W. At $0.12/kWh (illustrative), the annual saving is (1100–600) × 8.76 × 0.12 = ~$525/year. Over 20 years, that’s ~$10,500. For a facility running 24/7 (data center, hospital), the saving is real. For a school operating 10 hours/day, 200 days/year, the saving is only ~$180/year, and the payback is irrelevant.

When it doesn’t apply: For intermittent-duty applications (warehouses, repair shops) where the transformer is unloaded for long periods, no-load loss is the dominant term — but the absolute saving is small because the core is sized for the peak load. Also, if the local utility offers rebates for high-efficiency transformers (e.g., DOE 10 CFR Part 431 efficiency tiers), the premium is offset quickly.

Ranking table: GE Type QL by failure mode

Decision rule: Which spec fails first for your application?

Run this three-question filter:

1. Does the load have inrush > 5× steady-state? (motors, transformers, large capacitors) → voltage regulation fails first. Size the transformer with %Z ≤ 3% or add a soft-starter.
2. Does the load have > 20% harmonic content (VFDs, UPS, LED)? → thermal capacity fails first. Derate per IEEE C57.110 or buy a K-rated unit.
3. Will the transformer run 6,000+ hours/year for >15 years? → no-load loss fails first in the TCO. Buy Ultra Efficient.

If none of the above apply (pure resistive load, low duty cycle), the nameplate kVA is your limit — but that’s the exception, not the rule. For the majority of industrial installations, the spec that fails first is voltage regulation under inrush, followed by thermal derating from harmonics. The no-load loss is a financial failure, not an operational one — but it’s the one that kills the budget if you ignore it.

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. GE is a brand affiliated with this site; competitor names are used for identification only.