WWT award finalist

It’s for the Supply Interruption Detection Service that we provide which is a collaboration between i2O Water and Artesia.

i2O has been developing a network monitoring and advanced analytics software service; it takes pressure and flow data which water companies have been collecting for many years without doing much with it, derives insight from it, and visualises that insight in a variety of actionable ways.

Artesia carries out data focused consultancy projects for the water industry and had developed a machine learning anomaly detection algorithm and service, which uses the same water company data that i2O collects, that emailed event notifications to its clients. The algorithm is called eVader (event, alarm, detection and reporting).

These two were brought together into a single solution to make the algorithm more readily accessible to a large number of clients, reduce the lead time on alarms and improve the visibility of detected network events to clients.

The key benefits of the technology are reduced “customer minutes lost”, a key supply interruption metric, reduced leakage, and reduced water lost from the network.  Being able to respond quickly to events also allows water companies to reduce carbon emissions from vehicles and pumping of lost water, and abstraction from the environment.

The water industry relies on customers in the main to advise them of network issues i.e., after they have happened and after damage has occurred and water has been lost from the system.

Artificial Intelligence can spot patterns in data much better than humans can, much quicker, and tirelessly. What Artesia’s algorithm does is to build a model of ‘normal’ from a time-series dataset. It requires no manual intervention and doesn’t need to be manually ‘trained’. When new data is received, it is checked against this model to see whether it is normal, and if not, an amber (warning) or a red (alert) flag is raised. New data is also used to dynamically update the model, which is the part of the process that is often called “machine-learning”; the algorithm is continuously improving its ability to identify anomalies, for each location.

The algorithm is able to categorise the anomalies that it detects. This gives the potential for these to be handled differently: for different people to be made aware of them and to take the appropriate action.

The software service is delivered by i2O’s network monitoring and advanced analytics solution iNet.  It is hosted in AWS using the latest technologies, ensuring a reliable, secure, and cost-effective service.

Detected event alerts are emailed within minutes of data being received.  Yorkshire Water saved more than 300 l/s over a single 12-week period using the system which is now being run on its entire network.

WWT has published a full list of finalists for all awards.