The repair changes the circuit in two ways. It introduces a new insulation element whose condition is unknown, and it disturbs the adjacent cable ends, which were prepared, cleaned and re-terminated as part of the joint. Both are inside the test object, and both are reasons why the whole circuit is normally tested rather than the joint alone.

What the repair changed and what must be re-proved

A joint restores the conductor continuity, the insulation and the screen or sheath, and it does so in a factory-controlled assembly brought to site and completed in field conditions. The dielectric withstand of the joint depends on the skill of the installation, on the cleanliness of the work, and on the moisture and contamination conditions during assembly, all of which are less controlled than a factory process.

What has to be re-proved is that the circuit can withstand the voltage it will see in service with an adequate margin, and that the repaired section does not contain a defect that will develop under thermal cycling or under the mechanical movement of the cable in service.

A withstand test establishes the first of those directly. It does not establish the second, because a defect that is not stressed to failure during the test may still develop later. That is the reason a diagnostic measurement performed alongside the test carries value beyond the pass-or-fail result.

Method choice after a repair

The method follows the insulation type first. Extruded insulation, which includes the cross-linked polyethylene and ethylene propylene rubber constructions that dominate modern distribution and transmission cable, is designed for alternating stress, and the methods applied to it are alternating methods. Direct voltage is not appropriate for this insulation type because of the field distribution and the trapped charge it leaves behind.

Within the alternating methods, the choice is between a very low frequency test and a resonant test. Very low frequency testing suits circuits of moderate capacitance and is portable, which matters when the repair is in a location with limited access. A resonant test suits circuits of high capacitance or where a test frequency closer to power frequency is required, at the cost of a larger setup.

Where paper-insulated cable with a metallic sheath is involved, the applicable standard may permit or require a direct voltage test, and the repair should be tested on the basis of the standard under which the circuit was designed rather than on the basis of the newer practice for extruded cable.

Voltage and duration for the repaired section

A test after a repair is normally treated as an acceptance test, because the repaired circuit is in the state a new circuit would be in once the joint is complete. The test voltage and duration therefore follow the acceptance values for the cable class rather than the moderated values that may be applied to an aged circuit during maintenance testing.

The distinction matters for aged circuits. Where a repair is made on a cable that has been in service for decades, there is a legitimate argument for applying a test at the acceptance level to re-prove the repaired section, and a competing argument that the older cable sections may not survive the acceptance stress. The resolution is normally to test at the acceptance level and accept the outcome, because a cable section that cannot survive an acceptance test is a section that is likely to fail in service. Where the operator chooses a moderated level, the decision and its justification belong in the test plan and in the record.

Documenting the repair before testing

The test record is only interpretable alongside the repair record. The joint type and manufacturer, the installation date and conditions, the personnel, and the ambient temperature and humidity during assembly all describe the conditions under which the new insulation was completed.

The conditions during assembly are not administrative detail. Moisture and contamination during joint assembly are among the most common causes of early joint failure, and a joint made in adverse conditions is a joint whose test result should be interpreted with that context. Where the assembly conditions were outside the recommended range, the test result does not fully retire the concern, and the monitoring interval after the repair should reflect that.

The repair record should also state which cable sections were disturbed, how much of the route was opened, and whether any cable was moved to gain access. Those facts determine which parts of the circuit are outside their previously verified condition and are the reason the test covers the whole circuit rather than the new joint.

Acceptance criteria agreed with the client

The acceptance criteria should be agreed before the test, and they should name the standard and the edition that provides the test voltage and duration for that cable class. Where the operator’s specification is more demanding than the standard, the specification governs, and where the two differ the difference should be resolved before the test rather than after a marginal result.

The criteria also cover the observation during the test. A test that holds voltage for the full duration without breakdown passes. A test in which the applied voltage cannot be maintained, or in which the current rises progressively, has not passed even if no breakdown occurred. Stating the expected current behaviour and the limits within which it should remain turns the pass criterion into something the field team can apply.

The field test requirements for the relevant cable classes are set out in IEC 60502-2 for medium voltage extruded cables, in IEC 60840 for the higher voltage classes and in IEC 62067 for the highest extruded classes, with the general field test framework in IEC 60060-3.

Repeat faults in the same section

A second failure in the same section of route is a different problem from a first failure, and the test programme should reflect that. Two failures at different joints on the same circuit may indicate a systemic installation problem, a cable that has been disturbed, or a route condition that is damaging the cable. Two failures at the same joint indicate a problem with that joint or with the way it was installed.

Where a section has failed more than once, the test after the second repair should be accompanied by a diagnostic measurement, because the withstand test alone will not distinguish a sound repair from one whose defect is dormant at the test stress. Partial discharge measurement during the test is the most direct way to obtain that evidence.

The history should be reviewed at the point of repair rather than after the next failure. Where the section contains several joints installed at different times, the record of which have failed and when is what indicates whether the problem is localised or distributed.

Correlating the withstand test with diagnostics

Running a partial discharge measurement during the withstand test converts a binary result into a diagnostic one. The withstand test establishes that the insulation survives the applied stress; the discharge measurement indicates whether there is an active defect at that stress level, which is the information that determines the monitoring interval after the repair.

Dielectric loss measurement, where the arrangement allows it, adds information about the loss behaviour of the repaired section. Both measurements are more valuable when a previous result exists for comparison, which is an argument for measuring them at commissioning as well as after repairs.

The measurement configuration matters for the discharge test in particular. The noise environment at a repair site is not controlled, and the detection threshold has to be established in that environment rather than assumed from a laboratory value. The configuration used, the threshold achieved and the noise conditions belong in the record, because a discharge result without them cannot be compared with the next one.

Sign-off and handover records

The handover record should contain the repair details, the test method with its waveform and frequency, the test voltage and duration, the measurement arrangement including the divider used, the observed current behaviour, the diagnostic results where measured, and the acceptance criteria applied.

The record should also state the state in which the circuit was left: the joints installed, any temporary connections removed, the sheath and bonding arrangement restored, and the earthing removed. Those items are where the difference between a tested circuit and a circuit ready for service is found, and they are the items that a subsequent inspection cannot reconstruct.

Where the test was performed with a method or at a level that departs from the plan, the departure and the reason belong in the record, because the next test on the same circuit will be compared with this one. Two cable references worth keeping with the record are the construction and installation documentation published by cable manufacturers such as Nexans and the field-test configuration material published by high-voltage test equipment suppliers such as Haefely, while the accumulated field experience behind current practice is coordinated through CIGRE study committees. The withstand testing range is grouped on the high-voltage insulation withstand test hub, and the fault location work that precedes a repair is covered under cable fault testing.