Dry-Type Transformer FRA: Adapting Techniques for Air-Cored and Cast-Resin Windings
Dry-type transformers (open-wound, vacuum pressure impregnated, or cast resin) are widely used in commercial buildings, industrial facilities, wind turbines, and underground substations where oil-filled units present fire risks. However, applying a Transformer Frequency Response Analyzer to dry-type units requires modified techniques and different interpretation criteria compared to oil-filled transformers. This article addresses the specific challenges and best practices for dry-type FRA testing.
Fundamental Differences Affecting FRA
Dry-type transformers differ from oil-filled units in several ways that impact FRA:
No dielectric oil: The winding insulation is air or solid cast resin, which has lower permittivity (εr ≈ 3–4 for resin vs. εr ≈ 2.2 for oil). This affects inter-winding capacitance and shifts resonant frequencies upward by 10–30%.
Higher self-resonant frequencies: Lower dielectric constant and smaller physical size result in first resonant peak typically at 50–200 kHz vs. 5–20 kHz for oil-filled transformers.
No core immersion: The core is exposed to air, making low-frequency measurements more susceptible to external magnetic fields and humidity effects.
Terminal accessibility: Dry-type units often have exposed busbars or cable lugs rather than bushings, requiring different lead connection methods.
Smaller physical size: Windings are more rigidly supported in cast-resin designs, making them less susceptible to minor displacement but more susceptible to crack propagation.
Expected FRA Signature Characteristics
Healthy dry-type transformers exhibit distinct spectral features:
Low-frequency band (10 Hz – 5 kHz): Lower amplitude than oil-filled due to higher winding resistance and air-core coupling. Typical amplitude is -30 to -50 dB vs. -10 to -30 dB for oil-filled.
Mid-frequency band (5 kHz – 200 kHz): More resonant peaks (typically 8–15 peaks) compared to oil-filled (3–6 peaks) due to the distributed nature of dry-type windings.
High-frequency band (200 kHz – 10 MHz): Faster roll-off (30–40 dB per decade) because of higher series resistance in cast-resin windings.
These differences are normal. Do not compare dry-type signatures to oil-filled references; use only same-model, same-construction baselines.
Lead Configuration for Exposed Busbars
Dry-type transformers rarely have standard bushings. Adapt lead connections:
Use alligator clips or C-clamps with insulated handles to connect to busbars or cable lugs.
Maintain consistent connection points—mark the exact location on the busbar (e.g., 5 cm from the end) for repeatability.
Avoid connecting leads directly to bolt threads or uneven surfaces that create variable contact resistance.
For cast-resin units with embedded terminals, use the provided test points or manufacturer-recommended connection method.
Document connection methods with photographs for every test.
Interpretation Differences: What Constitutes a Deviation
Dry-type transformers have different baseline stability and deviation thresholds:
Temperature effect: Dry-type FRA signatures are less temperature-sensitive than oil-filled (0.1 dB per 10°C vs. 0.5 dB for oil-filled). A 10°C difference between tests is acceptable without compensation.
Humidity effect: Cast-resin surfaces absorb moisture, increasing surface capacitance and shifting high-frequency resonances. Limit humidity-related variation by testing at similar relative humidity (±20%) or correcting using empirical models.
Correlation coefficient thresholds: Use tighter thresholds for dry-type: CC < 0.92 indicates possible anomaly (vs. 0.85 for oil-filled) because dry-type signatures are naturally more stable when undamaged.
Amplitude deviation: A 2 dB change in dry-type is significant (vs. 3 dB for oil-filled) due to lower baseline amplitudes.
Case Example: Detecting Cast Resin Crack with FRA
A 2.5 MVA cast-resin transformer serving a data center experienced a lightning surge on the primary side. No external damage was visible. DGA (air sampling) was normal. FRA testing revealed:
A sharp 5 dB notch at 180 kHz that was not present in the baseline
Phase A correlation coefficient = 0.78 (vs. 0.96 baseline)
Mid-band amplitude reduction of 3 dB on Phase A only
Acoustic emission testing localized a crack in the cast-resin encapsulation on Phase A. The crack allowed partial discharge, which would have progressed to a phase-to-ground fault. The transformer was replaced before failure, avoiding a data center outage costing $500,000 per hour. FRA detected the crack that visual inspection and DGA missed.
Testing Wind Turbine Dry-Type Transformers
Wind turbines use dry-type transformers in the nacelle. Special considerations:
Motion during test: Nacelle yaw and blade pitch movements introduce mechanical noise. Lock the yaw system and park blades before FRA testing.
Vibration isolation: Use additional averaging (50–100 sweeps) to overcome low-level vibration from wind-induced nacelle motion.
Limited access: Nacelle transformers are often compact. Use short, flexible test leads and a handheld FRA instrument.
Baseline at commissioning: Perform FRA during turbine commissioning while the transformer is new and the nacelle is stationary. Store baseline in the turbine control system database.
Moisture Detection in Cast-Resin Windings
Cast resin can absorb moisture over time, reducing surface resistivity and increasing dielectric loss. FRA detects moisture as:
Progressive elevation of the high-frequency noise floor (>500 kHz) by 3–10 dB
Broadening of resonant peaks in the 100–500 kHz range
Reduction in the amplitude of the highest-frequency resonant peak
If these patterns appear, perform insulation resistance and polarization index testing to confirm moisture. Drying the transformer (low-temperature oven or hot air) may restore the FRA signature.
Recommended Test Parameters for Dry-Type
Optimize FRA instrument settings for dry-type transformers:
Frequency range: 20 Hz to 10 MHz (upper range can be reduced from 25 MHz because dry-type resonances rarely exceed 10 MHz)
Sweep points: 500–1,000 (sufficient for the narrower frequency range)
Averaging: 10–20 sweeps (higher than oil-filled due to environmental noise susceptibility)
Output voltage: 10–20 V peak-to-peak (dry-type windings may require higher signal to overcome air-core attenuation)
The Transformer Frequency Response Analyzer is fully applicable to dry-type transformers when test techniques and interpretation criteria are appropriately adapted. For asset managers responsible for dry-type fleets, FRA provides critical early warning of resin cracking, winding displacement, and moisture ingress—failure modes often missed by conventional testing.
