Learnings

Outcomes

1. The uncertainty in thermal image cameras is higher than previously thought. Even with proper calibration, the uncertainty in the cameras can be significant enough to cause spurious identification of defects in the image.
2. The environment within which the measurement is made can also significantly impact the uncertainty in the measurement. This can also lead to spurious interpretation and thus cause false defects to be identified which are not actually present. When studying the measurements made, the conditions of the surrounding equipment and the weather at time would need to be taken in to account to avoid false interpretations.
3. High voltage assets are generally of very complex geometries and are made from different materials. This can introduce further errors into the readings produced by the cameras. Each asset would have unique thermal behaviour which can influence the readings made by thermal cameras, thus lead to false positives.
4. By studying the combinations of the uncertainties, minimum thresholds for alarms have been identified. Having alarm thresholds below these recommended thresholds is likely to cause a very high false positive rate, which would in term lead to an increased amount of unnecessary work. These thresholds depend more on the constructions of the assets than their function.
5. Using differential temperature measurements could introduce more uncertainties in the measurements. This is because while differential measurement could decrease some uncertainties, it would increase others. For the applications assessed in this project, it resulted in a significant increase in uncertainty.
6. Attempting to make thermal images that capture multiple assets in the same frames could cause the quality to suffer heavily from the size of source effect. In principle, the asset of interested should occupy a substantial part of the frame for the measurement to be accurate. For example, when taking an image for a three-phase power system asset, if that image captures all the three phases’ devices in one single frame, that would mean that each phase could only take a third of the size of the frame at best. This would cause the quality to suffer from the size of source effect significantly.
7. According to how thermal cameras are currently used by NGET, only when large temperature differences are recorded, would they be considered as reliable indicators of defects. This would require the defects to produce significant heat that could cause a large temperature rise. This does not match all the defects or failure modes that are known.
8. Electrical insulators, e.g., bushings, CSEs, etc., are also tend to be good thermal insulators. This means that, in order for the thermal cameras to reliably detect measurable surface temperatures, the core temperature of these insulators would need to significantly higher. It is important to consider what kind of core temperatures are needed to produce such surface temperatures and whether such high core temperatures would cause rapid deterioration in these assets.
9. For the thermal cameras to provide reliable detections for defects, the symptoms related to the failure modes would need to be consistent while not being intermittent.

Recommendations for further work

1. Further investigations into the uncertainty introduced by different types assets would provide clarity on what kind of variability the asset managers should expect when studying thermal images from a range of assets.
2. Further reviews of different applications could also be performed, including the use of thermal cameras for imaging smaller assets which are further away from the cameras, such as hot joints on overhead lines.

Lessons Learnt

1. Data collection is imperative to obtaining strong results. Collecting and collating the required data early in the project helps to ensure that the project can be delivered on time.
2. Contingency planning for working at home should be considered at the start of a project; to avoid high impact low probability events from disrupting timely completion.

Dissemination

This project was disseminated at the ENIC 2020 through a virtual presentation.