2026-11-19
A ratio of 1.002 is not a verdict. Neither is an insulation resistance of 4 GΩ. Results only mean something against a baseline, a temperature and a comparison.
Routine maintenance produces numbers: ratio deviations, resistance values, insulation resistance, oil breakdown voltage. The hard part is not measuring them. It is deciding whether a number describes a healthy transformer, a measurement problem, or a developing fault. That decision depends on four things - the reference the result is compared against, the temperature it was taken at, the pattern across phases and taps, and whether another test supports the same conclusion.
Transformer tests are comparative by design. A ratio test compares a measured ratio with a nameplate value. A resistance test compares phases with each other and with factory data. An insulation test compares today with last year. Remove the comparison and the number loses most of its meaning.
This is why two identical readings can mean different things. An insulation resistance of 4 GΩ is unremarkable on a 10 MVA unit that read 5 GΩ last year, and alarming on a small unit that read 40 GΩ. The absolute value is not the signal; the direction and the size of the change are.
All four collapse if the test conditions are not recorded. Temperature, tap position, test voltage and instrument are part of the result, not optional notes.
The ratio test energises one winding and measures the induced voltage on another, then compares the result with the nameplate ratio for the tap in use. Modern three-phase testers also determine the vector group and the interphase angle, which is often more valuable than the ratio itself: a wrong vector group will trip a differential relay even when every ratio is correct.
Deviation is normally expressed as a percentage of the nameplate ratio:
Ratio deviation
Deviation (%) = (Measured ratio − Nameplate ratio) / Nameplate ratio × 100
Measured ratio: tester reading at the tap position under test
Nameplate ratio: marked ratio for the same tap position
A small consistent deviation across all phases and all taps is usually a measurement or nameplate issue: the nameplate ratio may be quoted at a different tap, the tester may be set to the wrong vector group, or the winding may be connected differently from the drawing. A deviation that changes with tap position points at the tap changer. A deviation on one phase only points at that phase's winding, its leads or its tap connection.
Worked example
Nameplate ratio at tap 5: 33.00 kV / 11.00 kV = 3.000
Measured ratio: 3.006
Deviation = (3.006 − 3.000) / 3.000 × 100 = +0.20 %
Whether 0.20% is acceptable is not decided by the size of the number alone. Under IEC 60076-1, the permitted voltage-ratio deviation at the principal tapping is the lower of ±0.5% of the specified ratio or one-tenth of the actual percentage impedance. Requirements for other tapping positions and for additional winding pairs also apply, and national standards or purchase specifications may add their own limits. Check the criterion against the edition of the standard in force for the unit, not against a remembered figure.
| Pattern | First hypothesis | Next step |
|---|---|---|
| All phases deviate equally at one tap only | Tap changer position or tap contact | Repeat at adjacent taps, inspect tap changer |
| One phase deviates at every tap | Winding turns, lead connection or tap lead | Resistance test on the same phase; check leads |
| Deviation grows towards one end of the tap range | Tap winding or tap selector fault | Full tap sweep, compare with factory tap data |
| All phases deviate in the same direction by a similar amount | Nameplate basis, tester setting, vector group | Verify nameplate, test setup and connections before condemning the unit |
| Ratio correct but angle wrong | Vector group or phase identification | Check nameplate vector group and lead marking |
Ratio alone cannot distinguish a shorted turn from a tap problem. It narrows the search; resistance and, where justified, excitation current or SFRA complete it.
DC resistance proves winding continuity, joint quality and tap contact condition. It is also the test most often misread, because resistance changes with temperature at roughly 0.4% per degree Celsius for copper. A reading taken in a cold substation in winter and compared with a summer reading will show a difference that has nothing to do with the transformer.
Results are compared at a reference temperature, usually 75 °C for distribution and power transformers, using the standard correction:
Temperature correction
R2 = R1 × (K + t2) / (K + t1)
R1: measured resistance at temperature t1
R2: resistance corrected to the reference temperature t2
t1: winding temperature during measurement (°C)
t2: reference temperature, usually 75 °C
K: 235 for copper windings, 225 for aluminium windings
The constant follows the winding material: 235 for copper and 225 for aluminium. Some standards use 234.5 for copper, which changes the corrected value by well under 0.1% at normal test temperatures. What matters is that one convention is used consistently across the fleet and stated in the report.
Worked example
Measured: 0.4215 Ω at 18 °C, copper winding
R75 = 0.4215 × (235 + 75) / (235 + 18)
R75 = 0.4215 × 310 / 253 = 0.5165 Ω
For three phases: 0.5165, 0.5148 and 0.5201 ohm. Average 0.5171 ohm, spread 0.0053 ohm, imbalance 1.0%. A 1% spread on a distribution transformer would not normally trigger action on its own; many utilities use around 2% between phases as an investigation trigger, with tighter limits for larger units and for specifications that call for them. The absolute values still have to be compared with factory data at the same reference temperature.
Insulation resistance is measured between windings and between each winding and earth, at a test voltage appropriate to the winding rating. The reading reflects moisture, contamination and insulation condition, and it is strongly temperature dependent: insulation resistance roughly halves for every 10 °C rise in oil temperature, which is why a hot reading and a cold reading are not comparable.
For oil-immersed transformers, the polarization index compares the resistance at ten minutes with the resistance at one minute:
Polarization index
PI = R10 min / R1 min
R10 min: insulation resistance after 10 minutes of applied test voltage
R1 min: insulation resistance after 1 minute
PI describes how strongly the insulation polarizes, which is a function of insulation type, condition and temperature. A PI near 1.0 on oil-paper insulation usually warrants investigation, but a low PI is not proof of a fault: dry-type and resin-cast windings polarize weakly by nature, and a measurement taken at high temperature or with a short stabilization time can produce the same result on healthy insulation.
This is where insulation type matters more than any single threshold. Oil-paper, resin-cast and liquid-filled windings behave differently, and a limit copied from one type applied to another produces false alarms. Compare with the previous test on the same unit, at a similar temperature, with the same test voltage and duration.
An insulation result that contradicts the oil results and the visual inspection deserves a repeat measurement before it becomes a maintenance decision.
For oil-immersed transformers, the oil is part of the insulation system. A BDV test raises the voltage across a defined electrode gap until breakdown; the result is sensitive to moisture and to solid particles suspended in the oil.
What a low BDV means: the oil's dielectric strength has fallen, usually because of moisture ingress or particulate contamination. It does not tell you where the moisture came from, and it does not identify a winding fault. What it does not mean: a low BDV is not evidence of winding deformation, and a good BDV does not prove the insulation is healthy - a transformer with a hot spot and active arcing can still have acceptable bulk oil.
Sample handling decides the value of the result. A sample drawn from a dead-end valve, taken in wet weather without a closed sampling arrangement, or left standing before test can show contamination introduced by the sampling rather than present in the transformer. Record the sampling point, the oil temperature and the time between sampling and test.
BDV is a screening measurement, not a full oil assessment. When it falls, the useful next step is water content, acidity and interfacial tension, because these indicate whether the problem is moisture ingress, oxidation or contamination. Water content is the most directly actionable: it tells you whether the oil can be dried in place or whether the insulation itself has absorbed moisture and needs a different decision.
A rising water content with a stable BDV is an early warning. A falling BDV with normal water content points towards particulate contamination or sampling error. Reading the two together prevents the common mistake of treating a single low BDV as proof of a wet transformer.
Typical new-oil acceptance values sit around 30 kV and above for a 2.5 mm gap, but the applicable limit depends on voltage class, oil type and the utility's specification. Treat any single number as a starting point, not a universal pass mark.
Frequency response analysis sweeps the transformer's impedance over a frequency range and compares the resulting curve with a reference. It detects winding deformation, core movement and winding displacement that ratio and resistance tests cannot see.
The method is described in IEC 60076-18 and is inherently comparative. Without a baseline - factory data, an earlier test or a sister unit - a single curve has limited diagnostic value, because normal transformers differ in their response. Interpretation also requires experience: connection quality, earthing, tap position and even the test lead layout change the curve.
SFRA is not a routine test for every distribution transformer. It is used after a through-fault, a transport event, a repair, or when other results suggest mechanical movement. When it is used, repeat the measurement with identical connections before drawing a conclusion.
Some results should not be acted on until they are repeated. Repeat the measurement, on the same tap and at a recorded temperature, when:
A repeat test costs an hour. Replacing a transformer because of a contaminated bushing, a loose lead or an uncorrected temperature difference costs considerably more. Where the repeat confirms the result, the second measurement also gives you the trend point you will need next year.
The value of routine maintenance comes from combining tests, not from any one of them. The table maps common combinations to the most likely explanations.
| Combination of results | More likely explanation | Confirm with |
|---|---|---|
| Ratio deviation on one phase, resistance high on the same phase | Tap contact or winding connection fault | Tap sweep, inspection |
| Ratio normal, resistance imbalance, insulation normal | Joint or tap contact resistance | Repeat resistance at stable temperature |
| Insulation resistance low, PI low, oil BDV low | Moisture in the insulation system | Oil moisture content, drying decision |
| Insulation resistance low, oil BDV normal | Surface contamination, bushings or leads | Clean and repeat, check bushings |
| All electrical tests normal, DGA abnormal | Localized thermal or discharge fault not yet affecting bulk measurements | DGA trend, repeat sampling |
| Ratio and resistance normal, SFRA deviates | Winding movement without turn loss | Repeat SFRA, compare phases |
The table is a reasoning aid, not a diagnosis. Every row assumes the measurement was valid, the temperature was recorded, and the comparison basis was correct.
A maintenance report that cannot be compared with the next one is of limited use. Record the tap position, oil or winding temperature at the time of test, the test voltage and duration for insulation tests, the instrument and its calibration status, the reference temperature used for correction, and the factory or previous values the result is compared against. Add a short interpretation: what changed, by how much, and what the next action is.
There is no single figure that applies to every unit and every tap. Under IEC 60076-1, the limit at the principal tapping is the lower of ±0.5% of the specified ratio or one-tenth of the actual percentage impedance, with separate requirements for other tap positions and winding pairs. The purchase specification and the standard edition in force decide the acceptance criterion. A deviation inside tolerance can still matter if it has changed since the last test.
Copper resistance rises about 0.4% per degree Celsius. A transformer measured at 5 °C in winter and at 35 °C in summer will show roughly a 12% difference with no fault present. Correct both readings to the same reference temperature before comparing.
No. PI depends on insulation type, temperature and test duration. Dry-type and resin-cast windings polarize weakly, and a hot measurement can lower PI on healthy insulation. Compare with the unit's own history at a similar temperature before concluding that the insulation has deteriorated.
No. BDV measures the dielectric strength of the bulk oil against moisture and particles. It does not detect localized thermal or discharge faults, which dissolved gas analysis is intended to reveal. A transformer can pass BDV and still have an active internal fault.
When there is a reason: after a through-fault, during transport or after a repair, or when ratio, resistance or DGA results suggest mechanical movement. It requires a baseline and identical test connections, and it is interpreted by comparison rather than against a fixed limit.
Rarely. A single result should be repeated and supported by another test before an outage decision. The exception is a result that indicates an immediate safety risk, such as a very low insulation resistance with visible damage or an active internal fault confirmed by gas analysis.
Send the test values, the tap position, the oil temperature and the previous report. We will help you work out whether the change is a measurement artifact, a normal trend or something that needs an outage.
Scope note: this article describes general interpretation practice. Acceptance limits, test voltages, correction factors and maintenance intervals come from the applicable standards (including IEC 60076 series, IEC 60156, IEC 60422 and IEEE C57.12.90), the transformer manufacturer's instructions, the purchase specification and the utility's own policy. Values quoted here are typical examples, not universal requirements.
Envie sua consulta diretamente para nós