As part of our Asset management and maintenance activities, we carry out patrolling activities to identify and mitigate any third-party threats to our network. The transition to hydrogen will require increased maintenance and patrolling, the aim of this project is to identify the high-risk areas of the network and prioritise patrolling activities to allow a more efficient maintenance regime. This will be achieved using a predictive model.
Advanced analytics allows us to optimize the patrolling frequency based on the probability of unannounced third-party works on the assets. It can be further utilized with novel data sources to optimise vegetation, marker posts and all other patrolling activities. Where linked to satellite and other advanced imaging systems it could negate the need for costly patrolling activities.
Benefits
Better Decision making around patrolling frequencies
Risk based approach allows prioritisation of high-risk areas on the network
Negates the need for costly patrolling activities
Allows identification of hotspot regions where further novel technologies such as vibration detection can be deployed
Reduced environmental impact (helicopter flights frequencies)
Learnings
Outcomes
In order to highlight the tangible benefits of a risk-based dynamic patrolling approach, a cost-benefit analysis was conducted. The possible benefits can be assessed on two different scales: network risk, and patrolling costs:
· Network Risk is the risk level noted against the Network based on the volume of events flagged. The network risk is measured by comparing the (expected) number of days that the network is exposed to events with the (expected) number of days that the network is not exposed. Figure 2 shows this in a visual way. The value of the network risk can vary between 0 and 1. A value of 0 indicates that there is no network risk which is only achievable when all events are instantaneously spotted, something that can never be achieved in practice. A value of 1 indicates that no event was detected, and thus that all events lasted for their full duration. For values between those extremes, the picture is more nuanced. For example, a network risk value of 0.5 may indicate that 50% of all events were detected instantaneously, and that the remaining 50% of the events were not detected at all. However, it could also be that all events were detected exactly in the middle of their duration (and thus the network was still exposed for 50% of the time). The real computation is slightly more involved as the risk factor[1] and the risk profile[2] were also taken into account, but it provides already an intuitive interpretation of the network risk.
· Patrolling costs are the overall value of monitoring the Transmission Network via helicopters. The maintenance/patrolling costs are measured in terms of the distance that needs to be patrolled by helicopters over a full year (under a specific patrolling scheme).
[1] The risk factor identifies how risky a certain event is for the network. Indeed, different events exist with varying levels of risk to the network.
[2] The risk profile shows how the risk of the event evolves over the duration of the event. For some events, the majority of the network risk is born in the first days of the events. For other events, the majority of the network risk would be born at the end of the event duration. For simplicity, we have assumed that all events have a constant risk profile, i.e. all days during the event are as risky to the network.
Lessons Learnt
During the project, some lessons were learnt that could be useful for future projects.
