The distinction matters because a test programme that repeats everything wastes the outage, and one that repeats too little leaves the equipment with evidence that no longer describes it.

What the repair changed

A repair changes the physical state of the equipment and therefore the validity of the evidence that described the previous state. The scope of the repeat testing follows from which elements were disturbed and which measurements depended on them.

A winding repair changes the resistance and ratio results at the taps it affected, and it may change the winding geometry, which invalidates the frequency response reference. A bushing replacement changes the dielectric results for the winding it belongs to and the capacitance of the bushing itself.

A cable joint replacement changes the dielectric condition of the whole circuit, because the test is applied from the ends and the new joint is inside the test object. A relay replacement changes the protection verification for every function the relay implements.

Tests that must be repeated in full

The tests that must be repeated are the ones whose evidence the repair invalidated. For a winding repair, that is the resistance and ratio measurements across the tap range, because the repair changed the conducting path and the turns relationship at the taps it affected.

For a cable repair, it is the withstand test on the completed circuit, because the test is applied from the ends and the repaired joint is part of the test object. There is no way to test the joint alone without a specialist arrangement.

For a protection repair, it is the settings verification and the functional test for every function the replaced device implements, followed by the scheme test where the device participates in one.

In each case the full test is required because the repair changed the object the test describes. A partial test of the repaired element alone leaves the interaction between the repaired element and the rest of the equipment unproved.

The standards that the repeat tests are judged against are the same as for the original tests: the transformer requirements in IEC 60076-1, the dielectric framework in IEC 60076-3, the switchgear requirements in IEC 62271-1 and the cable field test requirements in IEC 60502-2.

Tests that can be reduced with justification

A test can be reduced where its object was not disturbed and where its measured value could not have been affected by the repair. Insulation resistance measurements on a winding whose repair was mechanical, for example, are unaffected by the repair and remain valid evidence.

The reduction should be recorded with its reasoning. A line stating that the insulation resistance measurement from the previous test remains valid because the repair did not involve the insulation system is defensible; the same reduction left unstated looks like an omission.

The judgement should also account for the possibility that the repair disturbed something indirectly. A bushing replacement can disturb the connection arrangement, which affects resistance measurements at the terminals. The question is whether the measured quantity depends on anything the repair touched, not whether the repair nominally involved it.

Re-testing after component replacement

A component replacement is a repair in which the item replaced is defined rather than the damage being unknown. That makes the scope easier to establish: everything whose behaviour depended on the original component has to be re-verified.

Replacing a measurement transformer changes the ratio, polarity and burden evidence for the circuit it serves, and it changes the protection verification for every function that uses that circuit. Replacing a relay changes the settings verification and the functional tests for that device and the scheme tests it participates in.

The commissioning records for the replaced component remain valid as a description of the component that was removed. They do not describe the installation as it now stands, which is why the equipment history has to record the replacement as an event rather than simply replacing the old values.

Recording the repair in the asset history

The record should state the date, the scope of the repair, the cause where it was established, the components replaced, the tests performed afterwards and their results.

It should also state which of the earlier records remain valid. That statement is what allows a later engineer to know whether a measurement from three years ago describes the current equipment or an earlier configuration.

Where the repair changed the equipment in a way that affects the design assumptions, the change should be reflected in the design records as well as the maintenance records. A repair that changes a rating, a connection arrangement or a protection function is a modification rather than a maintenance activity.

Shortened intervals after a repeat failure

A second failure in the same equipment or at the same location indicates that the first repair addressed the symptom rather than the cause. The testing after the second repair should be accompanied by a diagnostic that addresses the mechanism, and the monitoring interval should be shortened until the trend is established.

The diagnostic depends on the equipment. A transformer that has failed twice may justify a frequency response comparison against the original fingerprint and a dissolved gas series. A cable that has failed twice in the same section justifies a discharge measurement alongside the withstand test.

The shortened interval should have a stated end point. A rule that returns the equipment to the normal schedule once a defined number of results are stable prevents the escalated programme from becoming permanent.

Coordinating re-test with energisation

The re-test sits between the repair and the return to service, and the sequence has to allow for it. A repair that is complete but untested cannot be energised, and the outage plan should show the test as a distinct stage with its own duration rather than as an activity absorbed into the repair.

Where the test requires conditions that the repair did not, such as a settled temperature or a defined settling period after oil processing, the plan should allow for them. A test performed before the equipment has reached the condition the acceptance criteria assume produces a result that has to be interpreted rather than accepted.

Sign-off after repair

The sign-off should state what was repaired, what was re-tested, what was accepted from earlier evidence with the reasoning, and the condition in which the equipment is returned to service.

Where a result was marginal and accepted, the acceptance should be recorded with its justification. Where a test was reduced, the reduction should be recorded with its basis. Those two statements are what allow the sign-off to be reviewed later, and they are what distinguish a complete record from a file of results. Independent acceptance practice is published by NETA, and repair and condition assessment practice for utility assets is coordinated through CIGRE study committees. The equipment range used for the repeat tests is listed under all electrical testing equipment, with the transformer range on the power transformer testing hub and the cable range on the cable fault testing hub.