Transformer through-fault assessment starts with the exposure: which windings carried fault current, its waveform and duration, and any repeated events. A satisfactory ratio test or quiet gas report does not exclude every mechanical consequence. Select evidence for the unresolved questions; no single result automatically authorises return to service.
Preserve the Event Before It Becomes a Summary Number
Obtain relay oscillography, event timestamps, breaker operation records, protection indications and the network configuration. Confirm instrument-transformer ratios and the location at which each current was measured. A feeder current is not automatically the current in every transformer winding.
Distinguish peak current, RMS current and duration. Preserve waveform quality and evidence of measurement saturation. The OEM may need these records alongside design data and previous fault history.
IEC 60076-5:2006 addresses thermal and dynamic withstand under external short circuits. Only its catalogue scope was reviewed here. It is not a post-event field clearance procedure; no numerical criterion is reproduced.
Keep earlier winding concerns, faults and transport anomalies in the investigation file. Missing records mean incomplete history, not absence of prior exposure.
Match Each Test to an Unresolved Question
Charles Sweetser's paper on SFRA and leakage reactance explains their complementary use in investigating winding movement. Its cases support combining evidence, not issuing a verdict from one curve or transferring a case-specific threshold to every design.
The following original table helps define the test request.
| Unresolved question | Evidence to review | Limitation to retain |
|---|---|---|
| What stress did the transformer experience? | Oscillography, clearing times and topology | Incomplete waveforms can limit reconstruction |
| Has winding geometry changed? | Comparable SFRA and leakage-reactance measurements | Setup and reference quality affect interpretation |
| Has a current path changed? | Winding resistance with temperature and tap identity | A terminal reading does not locate every defect |
| Has the ratio or connection behaviour changed? | Ratio and phase-displacement results | Agreement does not exclude all mechanical damage |
| Is there chemical evidence of an electrical or thermal event? | Correctly timed DGA and sample history | A single sample may not represent the full sequence |
| Are external components or auxiliaries affected? | Documented inspection and relevant functional checks | Scope must match the actual installation |
The responsible engineer selects the tests and records their purpose. This is not a mandatory universal package.
Worked Example: The Ratio Passes, the Investigation Continues
Consider two hypothetical external faults several minutes apart. The summary lists only the first event's larger current. The second has lower reported current but a different clearing duration. Retain both waveforms rather than discarding part of the exposure.
Turns-ratio results match their references, but one phase's SFRA differs from its pre-event baseline. The ratio result does not prove unchanged winding geometry.
The SFRA records used different tap positions. A matched comparison removes much of the difference. Evaluate the remaining repeatable change with leakage reactance and the OEM's knowledge of the winding arrangement.
The incident record now separates an explained setup effect from the residual observation. Neither "ratio passed" nor "SFRA changed" is sufficient operating clearance.
If the pre-event file is unavailable, a sister transformer offers context, not a replacement historical baseline.
Define the Evidence Needed to Close the Incident
Request an engineering disposition containing the reconstructed events, tests, valid comparisons, unresolved anomalies and ownership of follow-up. Include original files, not just screenshots or green indicators.
Link chemical findings to the DGA assessment and retain events in condition monitoring. Preserve reporting bounds and missing samples as such.
Close each event-related question with evidence or a named follow-up. Neither a health score nor an elapsed waiting period independently clears the transformer.
Talk to an engineer about reviewing the incident evidence.
Sources and scope: IEC 60076-5:2006, catalogue scope only; Sweetser, Applying Sweep Frequency Response Analysis and Leakage Reactance to Determine Mechanical Faults in a Power Transformer, full public paper reviewed. Sources rechecked 29 September 2026. Table and incident are original, non-normative examples.
Cover: AI-generated editorial illustration of generic equipment, not a photograph of an actual incident or a test setup.
