A new transformer reads 0.3% power factor on its insulation at commissioning. Five years later, the same transformer reads 0.45%. Both numbers are below the IEEE C57.152 threshold of 0.5% that defines healthy insulation. By the absolute limit, the transformer is fine.
It isn’t fine. The insulation has lost a third of its margin in five years. If the trend continues — and power factor degradation often accelerates rather than progressing linearly — the transformer could approach or exceed the acceptance threshold well before the next major maintenance interval. The transformer that “passed” the test is telling you something serious if you read the trend instead of the threshold.
Power factor and tan delta testing — the same test, different names — is the dielectric condition monitor for transformer insulation. It catches moisture ingress, oil degradation, cellulose aging, and contamination at stages where conventional tests still read normal. It’s also one of the most temperature-sensitive tests in transformer work and one where the standard correction tables were explicitly removed because they don’t work reliably. Getting it right means understanding both what the test measures and what it can and can’t tell you about an aging transformer.
This is a practical guide to running and interpreting power factor / tan delta tests on transformers. It assumes familiarity with transformer testing and instruments — the value here is in the interpretive framework, the temperature problem, and the test modes that most online content explains badly.
Table of Contents
Power Factor vs Tan Delta: Same Test, Different Math
The terminology trips people up because two slightly different numbers come from the same measurement. Both describe how much real power is dissipated in the insulation as the AC voltage cycles through it. Perfect insulation would dissipate zero real power — only reactive (capacitive) current would flow. Real insulation always has some real-power loss, and the ratio of that real power to the apparent power is what the test measures.
Power factor (PF) = cos(θ) = Watts / Volt-Amperes, where θ is the angle between voltage and current. For pure capacitance, θ is 90° and PF is zero. For real insulation, θ is slightly less than 90° and PF is a small positive number.
Dissipation factor (DF) = tan delta (tan δ) = tan(90° − θ). The tangent of the complement of the angle. For small angles (which is all the cases we care about — healthy insulation), tan δ and PF are numerically nearly identical. At PF = 0.5%, tan δ is also 0.5%. The numbers diverge only at high loss levels well outside the diagnostic range.
In practice the two terms are used interchangeably. Some standards (IEEE) lean toward “power factor”; others (IEC) lean toward “tan delta” or “dissipation factor.” Modern instruments report both. The threshold values are the same to the precision people work at.
What the Test Actually Reveals
The insulation between transformer windings, between winding and core, and inside bushings is a complex dielectric system: cellulose paper, pressboard, mineral oil (or ester or silicone), with all the interfaces between them. When you apply AC voltage across this system, the current that flows has two components:
A reactive (capacitive) component dominates. The insulation acts like a capacitor. This current is 90° ahead of the voltage.
A small in-phase (resistive) component flows because the dielectric isn’t perfect. Some real power is dissipated. This component is what tan delta measures.
The resistive component is sensitive to several specific degradation mechanisms:
- Moisture in the insulation. Water is highly polar and increases dielectric losses sharply. A small amount of absorbed moisture in the cellulose produces a measurable rise in tan delta.
- Oil degradation. As mineral oil ages and oxidizes, it accumulates polar contaminants (carboxylic acids, alcohols, peroxides) that raise losses. The oil’s own contribution to tan delta increases.
- Cellulose aging. Paper breaks down thermally and chemically over decades, producing polar degradation products and reducing its dielectric quality.
- Contamination. Particles, oxidation by-products, dissolved sludge — anything that introduces conductivity or polarity into the insulation.
- Surface contamination on bushings. Salt, dust, moisture films on bushing porcelain produce surface leakage that shows up as elevated tan delta.
The test is insensitive to mechanical condition (SFRA’s job), turns count (TTR’s job), and active fault gases (DGA’s job). It’s specifically about the dielectric condition — how much real power the insulation is dissipating per unit of stored energy. That’s the bulk insulation aging metric.
The Test Itself
The setup is conceptually simple. The instrument applies a known AC voltage (typically 10 kV, sometimes higher) at power frequency to a defined section of insulation, measures the resulting current’s magnitude and phase, and calculates the capacitance and the tan delta from those measurements.
Test voltage selection. Standard practice is 10 kV for windings rated 10 kV or higher. For lower-rated windings, test at or slightly below the winding’s rated voltage to avoid overstressing it. Most modern test sets handle this automatically once you tell the instrument the winding rating.
Frequency. Traditional power factor testing runs at power frequency — 50 or 60 Hz depending on region. Modern test sets, however, offer variable frequency capability: typically 1 Hz to 500 Hz on units like the Megger DELTA4000, or 1 Hz to 600 Hz on Omicron equivalents. Running a narrowband sweep across multiple frequencies (often called narrowband DFR or NB DFR) takes only minutes longer than a single-frequency test but provides substantially more diagnostic information — early detection of moisture, more reliable temperature compensation, and the ability to distinguish moisture from oil conductivity as the source of elevated losses. Full DFR/FDS sweeps extending down to microhertz frequencies are covered separately below.
Interference suppression. Modern test sets include advanced noise rejection circuitry — typical specifications handle up to about 15 mA of interference current or signal-to-noise ratios down to 1:20. This matters significantly in substation environments where induced AC from energized adjacent equipment can swamp older test sets. If you’re testing in an active substation with parallel energized circuits, interference suppression capability is what determines whether you get reliable readings.
Variable voltage testing (tip-up / step-up). Optional but valuable. Run the test at multiple voltages (say 2 kV, 5 kV, 10 kV) and compare the tan delta values. A voltage-dependent rise in tan delta — significantly higher loss at higher test voltage — points at voltage-sensitive defects like partial discharge inception in voids or surface contamination. Voltage-independent tan delta is the normal pattern for bulk insulation aging. The tip-up isn’t always reported but it adds diagnostic depth.
The Three Test Modes: GST, UST, GSTg
This is the part of power factor testing that’s most frequently explained badly. The three modes aren’t different tests — they’re different ways of routing the measurement current through the transformer’s insulation system to isolate specific paths.
A two-winding transformer has three insulation paths to think about:
- HV winding to ground (HV-to-ground, includes HV-to-core, HV-to-tank)
- LV winding to ground (LV-to-ground)
- HV winding to LV winding (HV-to-LV)
The three test modes let you measure each path separately or in combination.
GST (Grounded Specimen Test). Measures all current flowing from the high-voltage terminal of the test set to ground, regardless of which insulation path it took. If you apply test voltage to the HV winding with the LV winding grounded, GST measures HV-to-ground AND HV-to-LV currents combined. This is the simplest mode and the default for many quick measurements. It tells you the total insulation behavior but doesn’t separate the paths.
UST (Ungrounded Specimen Test). Measures only current flowing between the HV terminal and a specific low-voltage terminal that is NOT grounded — the rest of the system is effectively guarded. Apply test voltage to the HV winding, with the LV winding connected to the UST measurement input (not grounded), and any current flowing through HV-to-LV insulation is measured; current flowing through HV-to-ground bypasses the measurement. UST isolates the inter-winding insulation.
GSTg (Grounded Specimen Test with Guard). A hybrid. Test voltage to the HV winding, but the LV winding is connected to the instrument’s guard terminal (which is at the same potential as the measurement input but not in the measurement circuit). Current flowing from HV through HV-to-LV insulation goes to the guard and is excluded from the measurement; only current through HV-to-ground is measured. GSTg isolates the HV-to-ground path specifically.
The practical workflow for a complete transformer test:
- GST on HV with LV grounded — measures total HV insulation (HV-to-ground + HV-to-LV)
- UST on HV-to-LV — isolates the inter-winding insulation
- GSTg on HV with LV guarded — isolates HV-to-ground only
- Repeat for LV winding as the energized side
- Bushing C1 and C2 tests on each bushing — separate measurement using the bushing’s test tap
Why bother with three modes when GST gives you the total? Because the three paths can degrade differently. A transformer with moisture in its inter-winding pressboard but clean oil at the bushings shows elevated tan delta on the UST measurement but normal GSTg. A transformer with bushing contamination shows elevated power factor on the bushing tests but normal interwinding values. Knowing where the degradation is helps localize the problem and decide what to do — clean bushings, dry the unit, change oil, or accept and monitor.
Acceptance Criteria: The 0.5% Threshold and Its Reality
IEEE C57.152 gives indicative thresholds for mineral-oil filled power transformers rated above 500 kVA, with all values referenced to 20°C:
- Below 0.5% (0.4% for transformers above 230 kV): Healthy insulation. Typical acceptance level for new and well-maintained transformers.
- 0.5% to 1.0%: Aged or deteriorated insulation. Investigate. May still be acceptable for service depending on trending and other tests.
- Above 1.0%: Should be investigated. Significant insulation deterioration likely. Continued operation depends on a careful assessment.
These values are guidance, not universal acceptance limits. Actual criteria for a given transformer depend on its voltage class, whether it’s new or in-service, the test method used, and the manufacturer’s specific recommendations. The 0.5% figure is the most widely cited reference point in field practice, but acceptance criteria should always be verified against the latest applicable IEEE or IEC guidance, manufacturer documentation, and the transformer’s historical baseline before being treated as pass/fail.
For bushings specifically, IEEE C57.19.01 sets a similar 0.5% reference for power factor and recommends 5-10% deviation from the bushing’s nameplate capacitance as an investigation threshold.
A few practical extensions to the threshold framework:
The doubling rule. A common manufacturer guideline: tan delta should not exceed twice the commissioning baseline value, even if still below 0.5%. A new transformer at 0.2% creeping to 0.4% has lost half its margin and shouldn’t be ignored just because it’s still under 0.5%.
The trend matters more than the threshold on in-service units. A transformer trending from 0.3% to 0.45% in five years is showing accelerated degradation. The current measurement passes the absolute threshold; the trend says something is happening. The absolute threshold catches catastrophically degraded insulation; the trend catches developing problems.
Never act on a single measurement alone. A single elevated power factor result rarely justifies a major maintenance decision by itself. Test conditions vary, measurement noise happens, temperature effects are unpredictable, and surface contamination can produce false elevations. Standard utility practice before any significant intervention is to confirm an elevated result through a combination of: a repeat power factor test under controlled conditions, DGA, oil quality testing (moisture, acidity, IFT, BDV), moisture assessment via DFR/FDS, and review of the transformer’s loading and event history. A consistent finding across multiple diagnostics supports a real decision; a single high reading supports more investigation, not action.
The threshold assumes 20°C. This is where the test gets complicated, and it’s the next section’s whole subject.
The Temperature Correction Problem
Tan delta is strongly temperature dependent — and the dependence isn’t the same across different transformers. This is the source of more confusion in power factor testing than anything else.
For decades, the industry used standardized temperature correction tables to convert measurements taken at field temperatures to a 20°C reference value. IEEE C57.12.90 (the test code) had these tables. They worked roughly. Most online articles and many older procedures still cite them.
Then IEEE C57.12.90-2010 explicitly removed the generic correction tables, with this note:
“Experience has shown that the variation in power factor with temperature is substantial and erratic so that no single correction curve will fit all cases.”
The reason: temperature dependence of tan delta varies with insulation condition. A new, dry, clean transformer has one temperature characteristic. An aged, moisture-contaminated transformer has a different one. A heavily oxidized oil produces a different curve again. A single correction table that’s right for new insulation is wrong for old, and vice versa.
The practical consequence: a measurement taken at 25°C and corrected to 20°C using a generic table gives a number you should treat with skepticism. The correction may be too aggressive or too conservative depending on the transformer’s actual condition.
What to do instead:
Test at consistent temperatures when possible. If your commissioning measurement was at 22°C and your annual measurement is at 24°C, the temperature difference is small enough to ignore. Trending across small temperature ranges is reasonable. Don’t try to compare a 15°C measurement to a 40°C measurement and trust the correction.
Wait for thermal stability. Test a de-energized transformer after enough hours that top and bottom oil temperatures have equalized and the windings are at oil temperature. This eliminates internal temperature gradients that distort the measurement.
Record temperature with every measurement. Without the temperature record, no later correction or comparison is possible. The temperature is part of the result.
Use DFR/FDS for accurate temperature compensation when it matters. Dielectric frequency response — sweeping across a frequency range from millihertz to kilohertz — captures the full dielectric behavior of the insulation. The frequency response curve and the temperature response are related; DFR allows accurate temperature compensation by modeling the insulation’s actual behavior rather than applying a generic table. For diagnostic decisions on a transformer near the 0.5% threshold, DFR is more reliable than power factor temperature correction. It’s a separate test that uses similar equipment in modern test sets.
Treat power factor results near the threshold cautiously. A measurement of 0.52% at 30°C, corrected by table to 0.45% at 20°C, might be a real 0.55% if the correction is wrong for this transformer. Don’t make critical decisions on a single measurement near the threshold without supporting evidence.
The Environmental Conditions That Distort the Test
Beyond temperature, several environmental factors affect the measurement directly:
Humidity and weather. Don’t test in rain, fog, or relative humidity above 75%. Surface moisture on bushings creates leakage paths that produce falsely elevated tan delta. IEEE C57.152 explicitly recommends postponing the test under these conditions. If you’re committed to testing in marginal weather, accept that the bushing results may be contaminated.
Surface cleanliness. Bushings should be cleaned before testing. Dust, salt deposits, and surface contamination from years in service produce elevated readings that look like internal degradation. Clean the porcelain (and the test tap area) before the measurement.
Discharge before testing. After the transformer is de-energized, allow time for capacitive discharge before connecting. The stored energy in a large transformer is real. Modern test sets include discharge resistors but the discipline of waiting and using a hot stick is non-negotiable.
Reliable grounding. A poor connection to ground can produce erratic readings. Single-point ground at the test set, solid connections, clean connection surfaces.
Lead arrangement. Test leads should be arranged to minimize mutual capacitance and inductive coupling. Bunched leads or leads close to grounded structures introduce stray effects.
Reading the Results: Pulling It Together
A complete power factor / tan delta test produces a lot of numbers — capacitance and tan delta for each of the test modes on each winding, plus C1 and C2 values for each bushing. Reading the result well means looking at three things together:
Absolute values against the threshold framework. Each measurement falls in one of the bands: well below 0.5% (healthy), 0.5-1.0% (investigate), above 1.0% (serious). Note which mode and which path the value comes from.
Comparison against commissioning baseline. The commissioning measurement is the reference for that specific transformer with that specific design. Trends are the diagnostic — a doubling from baseline is significant even within the threshold band.
Pattern across modes and bushings. A transformer with one bushing showing elevated power factor and everything else clean has a bushing problem, not a transformer problem. A transformer with elevated GSTg (HV-to-ground) but clean UST (HV-to-LV) has a ground insulation issue, not an inter-winding issue. The pattern localizes the problem.
The diagnostic patterns:
All values elevated uniformly. Bulk insulation aging or moisture contamination throughout. The whole insulation system is degrading. Consider oil reclamation or replacement, drying procedures, or end-of-life assessment.
Inter-winding (UST) elevated, ground paths clean. Problem in the pressboard barriers between windings. Could be moisture, partial discharge damage, or aging of specific components.
One winding-to-ground elevated, the other clean. Issue specific to one winding’s ground insulation. May correlate with thermal stress history of that winding.
One bushing elevated, others clean. Bushing problem. Could be surface contamination (cleanable), moisture ingress into the bushing (developing fault), or internal degradation of the bushing’s condenser insulation. Bushing replacement may be the right call rather than transformer-level intervention.
Voltage-dependent rise (tip-up). Voltage-sensitive defects — partial discharge, surface tracking. Combine with PD testing if available.
Capacitance increase paired with tan delta increase. Often indicates moisture absorption — the moisture both adds capacitance and increases losses.
Capacitance decrease without tan delta change. Unusual; sometimes indicates a partial open or shorted section of a multi-section insulation system.
How Power Factor Fits With the Other Tests
The full low-voltage diagnostic suite for a transformer is TTR, winding resistance, SFRA, and power factor / tan delta. Each catches different things:
- TTR catches turns problems, connection errors, and gross winding faults
- Winding resistance catches loose connections, broken strands, and tap-changer contact problems
- SFRA catches mechanical changes — winding movement, core shifts
- Power factor catches dielectric degradation — moisture, oil aging, contamination
A complete assessment runs all four. A transformer that passes all four is in good shape across the dimensions accessible from its terminals without applying high voltage. A transformer that fails one and passes the others has a specific, localized problem. The combination is more powerful than any single test.
Power factor is the one of the four that’s most sensitive to gradual degradation. The others catch step changes — something that broke or moved. Power factor watches the slow drift of insulation condition over years. That makes it the primary trending tool for in-service transformer aging, and it’s why the commissioning baseline matters so much — that number is the reference for everything that comes after.
Moisture Assessment: Why Power Factor Is Only One Input
Moisture in transformer insulation is one of the most common reasons power factor rises, but power factor isn’t the only — or even the best — way to assess moisture. A proper moisture assessment combines several tests:
| Test | What it reveals about moisture |
|---|---|
| Power factor / tan delta | Increased dielectric losses; moisture is one possible cause among several |
| DFR / FDS | Direct estimation of moisture content in cellulose insulation, separated from oil conductivity |
| Karl Fischer titration (oil ppm) | Water dissolved in the oil — moisture in transit, not necessarily in cellulose |
| Equilibrium moisture curves | Estimate cellulose moisture from oil moisture using temperature and equilibrium data (less direct) |
| DGA | Active faults causing moisture generation; not a direct moisture indicator but relevant context |
The relationships matter. Most moisture in a transformer is held in the cellulose, not the oil — typically more than 99%. Oil moisture (ppm) shifts quickly with temperature; cellulose moisture changes slowly. A high oil moisture reading on a warm transformer can drop when the unit cools, simply because the cellulose absorbs the water back. DFR is the modern preferred method for cellulose moisture assessment because it estimates the cellulose moisture directly rather than inferring it from oil readings.
When power factor flags possible moisture, the follow-up isn’t “drain the transformer.” It’s the combination above — DFR for cellulose moisture, oil sample for ppm, DGA for context, and a review of the transformer’s history (recent oil work, seal integrity, breather condition). Moisture is a manageable condition if caught early; the diagnostic question is how much, where, and how fast it’s changing.
The Takeaway
Power factor and tan delta testing reveals the dielectric condition of transformer insulation. Done well, it catches moisture ingress, oil degradation, and cellulose aging early enough to plan interventions instead of responding to failures. Done badly — without temperature awareness, without trending against baseline, without understanding the three test modes — it produces numbers that may pass the 0.5% threshold while real degradation goes unnoticed.
The discipline is concrete: test in stable thermal conditions, record temperature with every measurement, don’t trust generic temperature corrections near the threshold, use all three test modes to localize problems, trend against the commissioning baseline rather than just checking the absolute threshold, and combine the results with TTR, winding resistance, and SFRA for a complete picture.
The test is straightforward enough that anyone with a power factor instrument can produce a number. Producing a number that means something — and reading it correctly — is what separates a useful diagnostic from a check-the-box exercise. On a transformer that will run for forty years, the difference is the difference between catching a problem in year fifteen and dealing with a failure in year twenty.