Transformer Commissioning

Ground Studies and Land Analysis

Before a transformer ever sees high voltage, the ground it sits on and the grid beneath it require the same rigor as the electrical commissioning of the unit itself. A power transformer is a major source of physical weight, sustained vibration, environmental oil hazard, and, during a fault, massive current.

1.1 Electrical Grounding Studies

A transformer needs a low-impedance path to earth to clear faults safely and protect personnel. An inadequate ground grid allows a system fault to energize the surrounding soil, creating lethal touch and step voltages.

Soil Resistivity Testing

Soil's ability to conduct electricity is measured with the Wenner four-electrode method: four equally spaced stakes are driven into the ground to inject current and measure voltage drop. Because moisture and temperature vary with depth, a multi-layer soil resistivity model is built rather than relying on a single surface reading.

The risk: if actual soil resistivity is higher than the original design assumed, due to dry sand, rock layers, or seasonal freezing, the grounding grid will fail to dissipate a fault quickly. This can destroy transformer neutrals and elevate local ground potential to dangerous levels.

Ground Grid Integrity and Continuity Testing

A high DC test current, typically 10 to 30 amps, is injected between the transformer tank ground pads and the main substation ground grid to measure resistance, which must be exceptionally low, generally under 0.1 to 0.5 ohms.

The risk: grounding risers missed or poorly welded during construction. Without a robust connection to the main grid, an internal fault will drive the entire metallic enclosure to system voltage, causing catastrophic flashovers and endangering anyone nearby.

1.2 Geotechnical and Structural Land Analysis

A large power transformer can weigh anywhere from 50 to over 500 tons. The foundation must be engineered to prevent movement over a service life of 40 years or more.

Geotechnical Soil Borings and Bearing Capacity

Deep core samples are drilled at the pad location to determine ultimate bearing capacity and settlement characteristics.

The risk: transformers vibrate continuously at twice the power line frequency, 120 Hz or 100 Hz, from magnetostriction in the core. Sustained vibration can compact or liquefy loose, unstable soils over time. Differential settlement, where one side of the pad sinks even half an inch relative to the other, imposes severe mechanical stress on internal winding structures, misaligns oil levels, and can crack rigid high-voltage bus ducts.

Seismic Hazard Analysis

Local peak ground acceleration is mapped to design an adequate anchoring system.

The risk: a transformer is top-heavy, with a heavy oil-filled conservator tank and massive porcelain bushings sitting at the top. Seismic loading can snap an unanchored unit off its foundation or rupture oil containment valves.

1.3 Civil and Environmental Land Analysis

Large transformers hold tens of thousands of gallons of combustible, environmentally hazardous mineral oil. Civil containment systems must be fully certified before commissioning proceeds.

Oil Containment and Drainage Analysis

A spill containment pit is designed and tested directly beneath the transformer, typically a deep concrete basin filled with washed fire-quenching stone connected to an oil-water separator.

The risk: a catastrophic internal arc can rupture the tank wall. If the containment pit is cracked or unsealed, thousands of gallons of burning oil will flood the rest of the substation, destroying adjacent equipment and reaching the local water table.

The mitigation: a water-tightness test on the containment basin, and structural verification that emergency firewalls are sound where the transformer sits close to other critical infrastructure.

1.4 Ground and Land Study Checklist

Table 1. Ground and land study sign-off metrics required before final energization.

‍

Standard Commissioning Protocols

These are the baseline tests. A transformer that has not cleared every item below should not be energized under any circumstances.

• Insulation Resistance and Polarization Index (PI). Measures winding insulation health. A PI below roughly 1.5 signals moisture ingress or contamination in the paper insulation.
• Transformer Turns Ratio (TTR) and Vector Group Verification. Confirms the internal windings were not damaged or shifted in transit and that phase relationships match the substation design.
• Winding Resistance. Measured across all tap positions to catch loose internal connections or broken strands in the winding transposition.
• Oil Quality and Dissolved Gas Analysis (DGA). A baseline fluid sample establishing dielectric breakdown strength, moisture content, and trace gas levels before the unit sees voltage.
• Protection and Control Functional Checks. Testing of the Buchholz relay, sudden pressure relays, winding and oil temperature indicators, and interlocking circuits.

Every one of these tests produces a single reading, or a small set of readings, taken once. None of them observe the transformer under sustained load, real ambient variation, or the mechanical stress of daily thermal cycling.

Esoteric Considerations: What Standard Tests Miss

These are the nuanced, easily overlooked physical and chemical phenomena that cause failures months or years after a clean commissioning report.

3.1 Core Saturation and Residual Magnetism

When a transformer is de-energized, the magnetic flux in the iron core does not return to zero. A substantial amount of residual magnetism stays trapped in the core.

The risk: re-energizing at the wrong point on the voltage wave allows this residual flux to combine with the incoming system flux, driving the core into deep saturation. The resulting inrush current can reach up to ten times rated current, violently stressing the windings and triggering false protection trips or mechanical deformation.

The mitigation: perform a core demagnetization procedure with a variable DC power supply before final SFRA testing and energization.

3.2 Core and Winding Mechanical Shifts (Transit Harmonics)

A transformer can look pristine from the outside while its internal, multi-ton winding assemblies have microscopically shifted or warped from low-frequency vibration during rail or road transport.

The risk: geometric shifts alter the internal capacitance between winding layers. The unit may survive initial energization, but its ability to withstand a major external short circuit later in life is compromised.

The mitigation: Sweep Frequency Response Analysis (SFRA), covered in detail in Section 4.

3.3 Static Electrification During Oil Filling

Pumping thousands of gallons of insulating oil into a large tank creates friction between the moving oil and the solid cellulose insulation, generating high levels of static electricity.

The risk: if oil is pumped too quickly, static charge accumulates on paper insulation surfaces faster than it can dissipate. It can discharge as an internal flashover through the oil, destroying insulation before the unit is ever connected to the grid.

The mitigation: meticulous tank grounding, oil flow velocities held below roughly one meter per second, proper vacuum-filling procedure, and a strict oil rest period of 24 to 48 hours before voltage is applied. Section 5 details the full vacuum-filling and moisture-removal process.

3.4 CT Saturation and Polarity Misalignment

Current transformers mounted inside the bushings feed current signals to the protective relays.

The risk: a single CT winding wired backward, or a CT that saturates prematurely during a commissioning test, causes the differential protection relay to interpret normal load current as an internal fault.

The mitigation: strict primary injection testing. Drive actual high current through the primary bushings and confirm the current reaching the protection relays matches, in both magnitude and phase angle, what the math dictates.

Interpreting SFRA Graph Deviations

Sweep Frequency Response Analysis injects a low-voltage swept sine signal, typically 20 Hz to 2 MHz, across the winding and measures the response. Because the winding-core assembly behaves as a distributed network of inductors, capacitors, and resistances, its frequency response is effectively a mechanical fingerprint. Any physical change in winding geometry shows up as a shift in that fingerprint, long before it would show up in DC resistance, TTR, or even DGA.

4.1 Reading the Frequency Bands

Low frequency (20 Hz–10 kHz): dominated by the core. Deviations here point to core deformation, shorted turns, or residual magnetism issues rather than winding movement. A large low-frequency shift with everything else stable often means demagnetization was incomplete.

Mid frequency (10 kHz–100 kHz to 1 MHz): dominated by the winding-to-winding and winding-to-ground capacitance and the main winding inductance. This is the band most sensitive to axial and radial winding displacement, hoop buckling, and winding-to-winding movement, and it is the band that matters most after a rough transport event or a through-fault.

High frequency (above roughly 1 MHz): dominated by the winding's internal geometry, lead and tap connections, and stray capacitance. Deviations here usually point to loose connections, lead movement, or tap changer contact issues rather than bulk winding shift.

4.2 Quantifying the Deviation

Comparing the commissioning trace to the factory baseline is done by correlation coefficient across each frequency band, not by eye alone. A common industry practice bins the comparison into three tiers:

Correlation coefficient above 0.99 in a band: no significant mechanical change. Normal manufacturing and measurement variation.

Correlation coefficient between roughly 0.98 and 0.99: a caution zone. Investigate further with a repeat measurement, a check of test lead connections and grounding, and comparison against sister-unit traces if available before concluding there is a defect.

Correlation coefficient below roughly 0.98, particularly in the mid-frequency band: indicates a probable mechanical change. Cross-check against winding resistance, TTR, and a visual or internal inspection before committing to energization.

4.3 Common Deviation Signatures

A uniform vertical shift across all frequencies: usually a test setup or grounding artifact, not a transformer defect. Re-test before drawing conclusions.

A new resonance peak appearing in the mid-frequency band: classic signature of axial winding displacement or a partial short between adjacent turns.

Peak-frequency shift without amplitude change: consistent with a change in winding-to-winding capacitance, often from clamping pressure loss or minor radial movement.

Divergence that appears only on one phase: points to a localized, single-phase mechanical event rather than a design or manufacturing-wide characteristic, and should be treated with more urgency than a deviation present on all three phases.

The value of SFRA is entirely comparative. A trace with no factory baseline, or no sister-unit trace, cannot be interpreted with confidence. The commissioning record should always specify what the current trace is being compared against and the correlation coefficients per band, not simply pass or fail.

Vacuum Filling and Moisture Calculation

Moisture is the single largest driver of dielectric failure in a new transformer, and it is introduced almost entirely during transport, storage, and the oil-filling step itself. The vacuum-filling process exists to remove both moisture and air from the cellulose insulation and the tank before the unit ever sees voltage.

5.1 The Vacuum-Filling Sequence

Step 1, dry-out check and evacuation: the tank is evacuated to a target vacuum level, typically below 1 to 3 torr (roughly 1 to 4 mbar) for large power transformers, and held to confirm the vessel does not leak back above a specified rate over a set hold period. This step pulls entrained air and residual moisture vapor out of the paper insulation, which is far more absorbent of moisture than the oil itself.

Step 2, hold under vacuum: the tank is held at deep vacuum for a duration driven by winding mass and insulation thickness, often 12 to 24 hours or more for large units, allowing moisture bound in the paper to migrate out as vapor.

Step 3, oil introduction under vacuum: dry, degassed oil is introduced from the bottom of the tank while vacuum is maintained on the vessel, at a controlled fill rate. Filling under vacuum, rather than at atmospheric pressure, prevents air bubbles from becoming trapped in the winding insulation and paper wraps, which would otherwise become partial-discharge sites once the unit is energized.

Step 4, final vacuum hold and degassing: once full, the unit is held under vacuum again to allow any remaining entrained gas bubbles to rise out of the oil column. This is the step most often shortened under schedule pressure, and it is the step the industry's veteran commissioning engineers flag most consistently as non-negotiable.

Step 5, oil rest period: 24 to 48 hours minimum before energization, allowing static charge generated during filling to bleed to ground and allowing any microscopic bubbles to fully clear.

5.2 Calculating Moisture Content and Target Levels

Moisture is tracked two ways during and after filling: directly in the oil, and indirectly in the paper insulation, since paper holds roughly 200 to 300 times more moisture than the oil in equilibrium with it at a given temperature.

Oil moisture (Karl Fischer titration): reported in parts per million by weight. New, properly processed oil in a large power transformer should test below roughly 10 ppm at fill, with an acceptance threshold typically set at or below 15 to 20 ppm depending on voltage class and specification.

Percent saturation: because water solubility in oil rises sharply with temperature, raw ppm alone is not meaningful without the oil temperature at the time of sampling. Percent saturation, the measured ppm divided by the oil's saturation limit at that temperature, is the number that should actually be tracked and trended over time.

Estimating paper moisture from oil moisture: equilibrium curves (commonly presented as Oommen or similar moisture-equilibrium charts) relate oil ppm and temperature to the equivalent percent moisture by dry weight in the cellulose. A target for new, properly dried insulation is below 0.5 percent moisture by dry weight in the paper; anything above roughly 2 percent significantly accelerates cellulose aging and lowers the dielectric strength margin under load.

Vacuum level as a moisture proxy: during the dry-out hold, the vacuum vessel's ability to hold pressure over time, rather than continuing to rise as more vapor is pulled off, is itself evidence that moisture evolution from the paper has substantially completed. A vacuum that keeps degrading over an extended hold means the insulation is still releasing moisture and the hold should be extended, not cut short to meet a schedule.

The golden rule applies here more than anywhere else in commissioning: the physical time required for bound moisture to migrate out of multi-inch-thick pressboard and for microscopic bubbles to rise out of a multi-thousand-gallon oil column cannot be compressed by a tighter project schedule. Every shortened vacuum hold is a decision to accept elevated infant mortality risk in exchange for schedule.

What Could Go Wrong and How to Avoid It

Table 2. Primary commissioning failure modes, root causes, and mitigations.

The golden rule of commissioning: never rush the vacuum hold and oil rest phases. The physical time required for microscopic air bubbles to rise out of the oil and for static charges to dissipate cannot be accelerated by a tight project schedule.

The Infant Mortality Gap

Transformer failure rates over time follow the classic bathtub curve: an elevated failure rate in the first one to three years of service, a long low-failure-rate plateau, and a rising rate again as the unit approaches end of life from wear-out mechanisms. Commissioning is built almost entirely to manage the front edge of that curve, and it does so well for the defects it is designed to catch. It does not manage the defects that only manifest once the unit is carrying real load.

Every standard and esoteric test described above shares the same structural limitation: each one is a single measurement, taken once, under controlled or de-energized conditions, before the transformer experiences its actual operating environment. A transformer that passes IR, PI, TTR, winding resistance, DGA, SFRA, and every protection check can still fail in month four, because:

• A winding clamping structure that reads perfectly correlated on a de-energized SFRA trace can still loosen incrementally under the first several hundred thermal expansion and contraction cycles once the unit is carrying real load.
• A marginal connection that passes a static winding resistance measurement at ambient temperature can develop a hot spot only once it is carrying rated current and cycling with daily and seasonal ambient swings.
• A core grounding strap that is mechanically intact at commissioning can fatigue and loosen under sustained 100 to 120 Hz vibration, producing circulating currents and gassing that a one-time DGA baseline cannot foresee.
• Moisture that reads acceptable in the oil at fill, per Section 5, continues to migrate between paper and oil for months after energization as the unit heats and cools, and a single post-energization DGA sample only captures one point in that migration.

None of this reflects a failure of the commissioning process. It reflects what commissioning was designed to do: verify a snapshot, not observe a trajectory. Closing the infant mortality gap requires a fundamentally different kind of instrumentation, one that watches the transformer continuously through the exact window where these latent defects turn into failures.

How VIE Closes the Gap

VIE Technologies provides continuous, non-invasive, physics-based transformer health monitoring built on four-signal sensor fusion: triaxial vibration, thermal signature, magnetic field, and local weather. The architecture is designed specifically to observe the trajectory that commissioning cannot, starting the moment a unit is energized and continuing for the life of the asset.

8.1 Extending SFRA's Question Into Live Operation

SFRA answers whether the winding geometry has shifted, but only at the moment of the test, with the unit de-energized. VIE's triaxial vibration sensing observes the mechanical signature of the energized, loaded winding and core assembly continuously. A winding clamping structure loosening gradually under live thermal cycling shows up as a drifting vibration signature well before it would produce a measurable SFRA deviation on the next scheduled maintenance outage, and long before it produces an electrical fault. VIE does not replace SFRA. It picks up exactly where SFRA's de-energized snapshot leaves off.

8.2 Catching What a Single DGA Sample Cannot

A 24-hour post-energization DGA sample, per Section 6, is a single data point. VIE's continuous thermal and vibration monitoring is built to flag the physical precursors to gassing, developing hot spots from loosening connections, abnormal mechanical signatures from a loosening core ground strap, before they progress to the point of producing measurable dissolved gas. Because the architecture is physics-based rather than dependent on a learned baseline, it does not need months of historical data on a given unit to establish what normal looks like. It can begin flagging deviation from expected physical behavior from the first days of energization, which is precisely the window where infant mortality risk is highest.

8.3 Distinguishing Real Faults From Normal Variation

The local weather signal in VIE's four-signal fusion exists to solve a specific problem: a transformer's vibration, thermal, and magnetic signatures all vary with ambient temperature, wind loading on radiators, and seasonal load. Without that context, a monitoring system risks either false alarms on hot summer days or missed detections that get attributed to normal weather variation. Correlating all three physical signals against local weather lets VIE separate a genuine developing defect, of exactly the kind described in Sections 3 and 7, from routine environmental and loading variation, reducing false positives during the noisy first year of a unit's life when commissioning teams and operations staff can least afford chasing phantom alarms.

8.4 Fleet-Level Visibility During the Highest-Risk Window

Utilities and IPPs commissioning multiple units on a similar schedule, whether a substation rebuild, a data center campus buildout, or a multi-site rollout, face the same infant mortality exposure across every unit simultaneously. VIE's fleet-level monitoring, deployed across more than 1,000 transformers and 3 gigawatts under management, gives operations and engineering teams a single view across every newly energized unit during its highest-risk window, rather than relying on the next scheduled maintenance outage, often 6 to 12 months out, to catch a developing infant mortality failure.

8.5 A Non-Invasive Complement, Not a Replacement

VIE's sensors are non-invasive and physics-based by design, they observe the transformer from the outside using vibration, thermal, and magnetic signatures rather than requiring an internal connection or an outage to install. This makes continuous post-commissioning monitoring practical to deploy at the exact moment it matters most: immediately after the commissioning team signs off and the unit is turned over to operations, when the standard test protocols have finished asking their one-time questions and the transformer's real operating life, and its real infant mortality exposure, begins.

Conclusion

A clean commissioning report confirms that a transformer is safe to energize. It does not confirm that the unit will survive the year that follows. The standard tests, IR and PI, TTR, winding resistance, DGA, and protection checks, along with the esoteric considerations of residual magnetism, transit-induced mechanical shift, static electrification, and CT polarity, all address defects that are detectable in a single measurement taken before or at the moment of energization. Infant mortality failures live in the gap between that measurement and the unit's actual operating trajectory.

Rigorous commissioning, including careful SFRA interpretation against a factory baseline and a properly sequenced vacuum-fill and moisture-removal process, remains the non-negotiable foundation. Closing the infant mortality gap on top of that foundation requires continuous, physics-based observation of the transformer once it is live, exactly where VIE's four-signal sensor fusion is built to operate.