Partial Discharge on the Electricity Network: A Utility's Guide to PD Surveys

Partial Discharge on the Electricity Network: A Utility's Guide to PD Surveys

1. Why PD belongs in network asset management

PD is a localised dielectric defect that discharges under operating stress. It does not cause an outage today; it indicates the mechanism that will.

For a network operator, that matters because of how cable failures arrive:

  • They are unplanned — they happen at the worst time, often under load or weather stress
  • They are location-expensive — finding the fault is a large share of the total cost
  • They are network-disruptive — a single feeder can affect many customers
  • They are increasing with age — large populations of 1990s–2000s extruded cable are reaching the end of their design life

The value of PD surveying is shifting work from the first category to the last: from emergency repair to planned intervention.

2. Prioritising which circuits to survey

Surveys cost money, so the programme must be risk-weighted. Five parameters are commonly used:

Parameter Why it matters Data source
Age Extruded cable degrades with time and thermal cycles Asset register
Fault history Repeat offenders indicate systematic weakness Outage records
Loading Thermal stress accelerates insulation aging SCADA / load records
Criticality Customer count, critical customers, network position Network model
Environment Moisture ingress, soil conditions, water tree exposure Route records, GIS

A practical approach: rank all circuits, start with the top decile, and re-rank each year as results come in. This captures the majority of avoidable risk with a fraction of the survey effort.

3. Choosing the survey method

Two families of measurement, often used together:

Offline PD testing (during an outage)

  • VLF excitation with calibrated PD measurement — see PD testing for commissioning for the full measurement sequence
  • Highest sensitivity and the only method that gives a traceable pC figure
  • Requires switching the circuit out of service and terminating the cable ends

Online PD survey (in service)

  • HFCT sensors clamped on cable earth/neutral paths, ultrasonic and TEV sensors on accessories
  • No outage required, so the whole network can be screened progressively
  • Lower sensitivity and higher noise, so best used for screening: find candidates, then confirm offline

A common programme uses online screening to shortlist accessory-hotspots and offline testing to quantify and locate defects on the shortlisted circuits.

Practical constraint: whichever method is used, results are only comparable over time if the method, voltage sequence and instrumentation stay consistent. Mixing methodologies mid-programme breaks the trend.

4. Ranking findings by severity and criticality

A PD result becomes a work order only after two independent judgements — the severity side follows the interpretation criteria for PD results:

Severity (the defect) Criticality (the circuit)
Magnitude and trend Customer impact
Location: termination / joint / cable body Redundancy of supply
PDEV margin against operating voltage Cost of unplanned outage
Reproducibility across tests Repair feasibility and access

High severity on a low-criticality spur may be monitored; moderate severity on a critical single-feed circuit may justify immediate planned intervention. A simple 2D matrix (severity × criticality) is enough to sequence work and to justify the budget.

5. Turning results into a maintenance plan

The programme output should feed three decisions:

  1. Replace / repair — defects confirmed at joints or terminations, or PD in the cable body, become planned work
  2. Monitor — stable findings enter a shorter retest cycle (e.g. annual), with the trend as the trigger
  3. Re-baseline — after any repair, record a new PD result and a fresh TDR trace with the correct velocity factor, because the asset has changed and old baselines no longer apply

Keep PD data in the asset register alongside test certificates, joint positions, and TDR traces — so that a finding can be correlated with as-built records and with later tests on the same circuit.

6. The cost case

A network PD programme is justified by avoided events, not by the value of the tests. The comparison is straightforward:

Cost of the programme Cost of one avoided failure
Survey labour and switching Emergency crew mobilisation
Test equipment (capex or service contract) Fault location (hours to days)
Data management and analysis Excavation and repair, often in winter
Planned repair works Unserved energy, regulatory penalties, customer compensation

For most operators the break-even is a small number of avoided unplanned failures per year — which is why PD surveying pays back fastest on exactly the circuits that have already failed once.

FAQ

How often should a network be PD surveyed? Common practice is an initial screening pass across the highest-risk circuits, then annual or biennial surveys on those, with shorter intervals for circuits showing activity. Circuits whose PD is stable and whose PDEV margin is comfortable can move to a longer cycle.

Can online PD testing replace offline testing? Not for quantification. Online methods screen a network efficiently with no outage, but the traceable pC measurements and defect location come from offline VLF testing with calibrated PD measurement. Use online to find candidates, offline to decide.

What is the biggest practical mistake in a PD programme? Changing method or instrumentation between tests, which makes the trend meaningless; and recording results without location, which makes them unactionable. Both are avoidable — the field-level versions of these errors are covered in common mistakes in on-site PD testing.

Where does PD usually originate? Overwhelmingly at accessories — joints and terminations — rather than in the middle of an intact extruded cable. This is good news operationally: the defect is usually at an accessible, repairable point, and it is often found by mapping PD against joint positions.

Related reading

Kembali ke blog

Tulis komentar