Acceptance Testing a 15 kV XLPE Feeder With a New Splice

Refuse the DC test on this cable and document why.

This is solid-dielectric XLPE, not paper-insulated lead cable. DC testing is inappropriate here for three reasons:

1. Space charge. A DC test drives charge into the insulation and leaves a trapped field that can combine with the AC service stress at energization and precipitate a failure in a cable that just 'passed'. The risk is worst at joints and terminations - exactly the new components being commissioned.
2. Wrong stress distribution. DC and AC do not distribute stress the same way through a dielectric. A DC test at 35 kV (about 4 U0 for this 8.7 kV line-to-ground cable) can overstress parts of the insulation while understressing the defects that matter in service.
3. Standards practice. Current cable testing practice uses AC withstand at power frequency or 0.1 Hz VLF for XLPE acceptance, with PD and dielectric-loss diagnostics. DC hi-potential for XLPE is a legacy requirement, not a current acceptance criterion.

What to do: submit a technical clarification to the specification, test instead at 0.1 Hz VLF at 3 U0 for the acceptance duration required by the governing standard and project spec, and get the deviation approved in writing before the Friday energization rather than after.

Reject the splice. A withstand test is a go/no-go breakdown test, not a defect detector.

The PD data:

• Inception at 6.1 kV, which is 0.70 × U0 (8.7 kV)
• About 250 pC at the 8.7 kV operating stress

The cable will operate continuously at 8.7 kV line-to-ground. Inception below the operating voltage means the discharge is active in normal service, and 250 pC is well above the level expected of a sound joint. The defect - likely a void or a lifted interface in the heat-shrink - will erode the insulation until it fails in service, probably soon after energization when the cable is hot and loaded.

A VLF withstand at 3 U0 only proves the insulation did not break down during that test. It cannot prove absence of a discharge site that is already active at one-third of the test stress. This is exactly why PD is run alongside the withstand on new joints.

Disposition: do not energize. Locate the discharge (time-domain reflectometry plus acoustic or ultrasonic detection to confirm it is at the splice rather than at a termination), cut out and rebuild the splice, and repeat both the VLF withstand and the PD scan before acceptance.

Safety first, because the hazards here are the ones people get hurt by.

Safety controls:

• The far end is 900 ft away in an unlocked room. Post a guard, barricade it, and use a documented two-person or communicated-watch arrangement. Every cable test failure mode includes someone at the remote end touching a conductor that is being raised to 26 kV.
• Lock out and tag the feeder at both ends, verify absence of voltage, and apply personal grounds before connecting test leads.
• After any DC or VLF test, discharge and ground the conductor for a defined period before handling. A DC-tested cable holds a lethal charge long after the set is switched off - another reason not to run the DC test at all here.
• Ground the test set and the cable shield per the set manufacturer's instructions, and keep the test area roped with the set operator inside it.

Documentation:

• Test voltage, frequency, waveform, and duration actually applied, with the standard and spec clause used
• Ambient temperature and humidity at the time of test
• PD detector calibration, coupler location, PDIV, extinction voltage, and PD magnitude versus voltage
• A written record of the rejected 1985 DC requirement and the approved deviation, plus the disposition of the splice and the retest results