Monitoring night flow measurements to detect potential leakages
Advanced Analytics | Custom Analytics + Software Solutions | Data Visualisation + Reporting Solutions
The LeakSense tool facilitates the exploration of night flows and potential leakage by water industry experts, aided by specific automatic flagging of anomalous night flows.
Harmonic was selected as the Data Science squad within the Waters Services Reform National Transition Unit (NTU). During this time, the squad developed a suite of data science solutions centred around the identification and reduction of water leakage. These solutions are now available for the benefit of all councils.
The Challenge
Leakage is a huge issue for councils facing ageing water distribution infrastructure. Identifying the source of leakage and prioritising areas for maintenance can be very challenging due to the hidden nature of water network infrastructure. Persistent low-level leaks often go undetected for long periods of time, leading to substantial losses and damage. Leakage detection can be enhanced by analysing flows at night time when we typically expect very little water usage.
The data provided to Harmonic indicated water network flows for Auckland on 15-minute readings at a consumer metering level.
The Solution
Harmonic developed LeakSense, a tool to facilitate early detection of areas with significant water leakage by analysis of night flows.
The LeakSense interactive dashboard facilitates a top-down approach to leakage detection, allowing experts to drill down from problematic metro areas all the way to individual meters. It allows visualisation of total water flows at the metro, local, or suburb levels, and displays historical data ranges to help identify flows outside the range of typical historical behaviour.
LeakSense presents three different time series flow data visualisations:
- Raw metering data to view potential spikes or changes in behaviour at the most granular level.
- Daily minimum night flows to spot historical trends in night flows and signs of leakage.
- Region flow balances from smart meter data, balancing inflows, consumption, and outflows to identify anomalous losses.
Harmonic also implemented an anomaly detection algorithm on the minimum night flows to flag unusual flow behaviour. These anomalies are summarised on the dashboard homepage to provide a quick summary of potential leakages and alert utilities on potential problems.
The Result
- Reducing the time spent by industry teams on spreadsheet construction and analysis of historical flows.
- Reducing the time between network leak onset and detection by making signals of leakage visible.
- Facilitating industry experts and decision-maker to more more effectively target leakage reduction and network efficiency improvement efforts.