Partial discharge monitoring produces data, but data is not yet a decision. The value of a monitoring system is realised only when the PD readings change what the maintenance team does – and that change does not happen automatically. This article explains the decision chain that turns PD data into maintenance action: how thresholds are set, how trends are read, and how the results feed the asset lifecycle.
1. Setting the thresholds
An alarm is a comparison between a measurement and a threshold, and a poorly set threshold destroys the value of the system in one of two ways: it alarms so often that the team stops responding, or it never alarms and the first sign of trouble is the failure itself.
The practical approach is staged thresholds:
- Baseline phase – after commissioning, the system runs in measurement mode to establish the normal level at each point; the level varies with load, temperature and weather, so the baseline should cover a full load cycle
- Warning threshold – set above the normal range, triggering a review and a handheld confirmation visit
- Alarm threshold – set where the trend analysis shows rapid growth or the absolute level indicates a high-energy defect, triggering work planning
The thresholds are then tuned with experience – and they will differ between sensor types and between sites.
2. Reading the trend, not the number
The single most valuable habit in PD monitoring is to watch the trend. A stable reading at a moderate level is often more reassuring than a low reading that is rising. The trend questions are:
- Is the level growing, stable or falling?
- Is the growth linear (slow ageing) or exponential (a defect developing)?
- Does the activity correlate with load, temperature or weather?
- Does the pattern change – new phase-resolved characteristics, new frequencies?
A trend that accelerates is a defect developing; a trend that is flat at a moderate level can be scheduled for the next planned outage.
3. Confirming and diagnosing
An alarm from the online system is a reason to look, not a verdict. The confirmation step uses the other tools:
- A handheld detector confirms the reading and localises the source at the panel, the phase or the joint
- The phase-resolved pattern and the multi-channel comparison identify the defect type
- For cables, an OWTS test locates the discharge along the route
The output of this step is a diagnosis: what the defect is, where it is, and how urgent it is.
4. Planning the work
With a diagnosis, the work is planned rather than reacted to:
- Urgent defects – high energy, accelerating – are scheduled promptly with the right crew and spares
- Moderate defects are scheduled into the next planned outage, combining the repair with other work on the same bay
- Every diagnosis is logged, so the next reading at that point is compared with the history
5. Feeding the asset lifecycle
Over time, the PD history becomes an asset record. The data supports decisions beyond the single repair:
- Replacement versus repair – a panel or cable with repeated failures is a replacement candidate
- Budget allocation – the trends identify the asset classes that need investment
- Design feedback – recurring failure modes point to a design or installation practice that should change
Summary
PD monitoring pays off through the decision chain: staged thresholds that are tuned to the site, trend analysis that reads growth rather than level, confirmation and diagnosis with the portable and testing tools, planned work instead of emergency response, and asset records that guide the lifecycle. HUWOR supplies the full toolkit – online sensors, gateways, platforms and handheld detectors – as one family. Contact the HUWOR sales team to build your decision chain.