Power line inspections tend to require more equipment and perform more functions as the need to improve quality of service with limited budget calls for ever more efficient inspection solutions. Simultaneously, current outsourcing and asset management trends lead grid managers to consider their lines as a single, comprehensive entity that should be examined and surgically maintained abandoning the model of a sequence of uncoordinated equipment and parts that are maintained by different people.
These were the motives behind a new approach to asset management where lines are envisaged as individuals and grids are a population of power lines and substations. Together, both types form the interdependent structure of a graph with edges (lines) and vertices (substations). This way, the reliability of each “individual” depends on the unified status evaluation for reliability.
In order to compare points of interest produced with different functions carried in a common inspection, a small set of classification rules is applied to all inspection functions so that all issues can be expressed in a common format. Some classification computation methods are illustrated for thermography, vegetation and visual inspections. Benefits and current limitations are also mentioned.
A risk index is computed from this semi-quantitative classification for points of interest located at each tower and each span. Then, the method is generalised to the whole power line (sequence of towers ad spans) and subsequently to groups of power lines, arranged as power grids. Equipped with such data, maintenance managers can distribute maintenance resources in a more efficient way across the grid and across maintenance functions addressing two problems: 1) given a certain maintenance budget, which is the optimal distribution of resources that minimises the probability of failure and increases reliability and 2) what’s the minimum amount of resources required to attain a given reliability threshold and how are these resources distributed.
This paper describes the implementation of these methods in Portugal. Early results suggest that operators with high quality inspection and maintenance could expect significant efficiency gains from selective maintenance actions in limited sections of their grids based on risk assessment as opposed to blind, periodical maintenance of whole lines based on elapsed time.
As operators move from systematic, regular activities on their grids to more statistical-based inspection and maintenance schedules, it is necessary to store as much information as possible and process it with time and geographic references to reliably estimate probabilities of failure, to optimise maintenance and inspection dates.