Most of these systems are for detecting leaks/bursts at district metered area (DMA)
level. DMAs are designed to be hydraulically isolated areas that are generally
permanent in the system. Automated online analysis systems have the potential
to be a useful tool for real-time identification of small- to medium-sized bursts.
Their use promotes a more proactive approach to leakage management, with
awareness of leakage incidences soon after they occur and before the customer
is seriously impacted. Such systems make it feasible to identify and hence find
and fix leaks that would previously have become background leakage, providing
the potential to reduce the so-called Economic Level of Leakage. Mounce et al. [28]
review approaches for event detection in WDS measured time series data, with
a focus on data-driven methodologies for leak detection. Wu and Liu [29] also
provide a more extensive summary of data-driven approaches and their performance.
Mounce and Boxall [30] describe an online system pilot implemented with a UK
water company using an ANN and fuzzy logic system for detection of leaks/bursts
at DMA level. This AI system was not reliant on any special hardware or network
configuration and produces intelligent ‘smart alarms’. The system was subsequently
commercialised as FlowSure (Servelec Technologies Ltd) demonstrating how
academic research can have real-world impact.
Smart water network technologies have the potential to deliver an improved
service to customers and cost-effective performance improvements for the
water industry. Sensor technology and the ‘big data’ they generate combined
with advanced ML techniques are providing new opportunities for deeper
understanding of WDS. SmartWater4Europe (SW4EU) was a European FP7
demonstration project. Four demo sites comprise solutions for leakage control,
water quality management and energy optimisation incorporating sensors, data
processing, modelling and analytics technologies. The UK demo site (TWIST) in
reading investigated how emerging technologies can be used to create a SWN with
near real-time notification of performance to enable proactive management and
intervention (particularly for leakage). Technologies utilised during the pilot
have included installation of flow instruments through full-bore hydrants,
instruments capable of high-resolution monitoring (thus enabling the identification
of pressure transients), AMR customer smart meters as well as traditional sensors.
Three network leakage algorithms to detect leakage and other abnormalities
(as soon as they occur) in the water network were tested and assessed: AURA
BED alerts as described in [28, 31], dynamic bandwidth monitoring (DBM) and
Netbase envelopes (developed by Crowder Consulting). Having a standard approach
to test different algorithms allows an objective comparison of their effectiveness.
Increasing amounts of SWN data are only of real business value if this
valuable resource is ultimately used to inform and support decision-making,
i.e. data to information to insight to action. Projects such as SW4E allow the
exploration, at demo and full WDS pilot scale, of deploying multiple smart
network technologies, both hardware and software, and the multiplicative synergy
between them. The deployment of a smart water network has its own challenges
such as large network data stores, false positives, limited analytical capability, pipe
location and condition, failure prediction, meter coverage, response to failures, etc.
18
S. R. Mounce
level. DMAs are designed to be hydraulically isolated areas that are generally
permanent in the system. Automated online analysis systems have the potential
to be a useful tool for real-time identification of small- to medium-sized bursts.
Their use promotes a more proactive approach to leakage management, with
awareness of leakage incidences soon after they occur and before the customer
is seriously impacted. Such systems make it feasible to identify and hence find
and fix leaks that would previously have become background leakage, providing
the potential to reduce the so-called Economic Level of Leakage. Mounce et al. [28]
review approaches for event detection in WDS measured time series data, with
a focus on data-driven methodologies for leak detection. Wu and Liu [29] also
provide a more extensive summary of data-driven approaches and their performance.
Mounce and Boxall [30] describe an online system pilot implemented with a UK
water company using an ANN and fuzzy logic system for detection of leaks/bursts
at DMA level. This AI system was not reliant on any special hardware or network
configuration and produces intelligent ‘smart alarms’. The system was subsequently
commercialised as FlowSure (Servelec Technologies Ltd) demonstrating how
academic research can have real-world impact.
Smart water network technologies have the potential to deliver an improved
service to customers and cost-effective performance improvements for the
water industry. Sensor technology and the ‘big data’ they generate combined
with advanced ML techniques are providing new opportunities for deeper
understanding of WDS. SmartWater4Europe (SW4EU) was a European FP7
demonstration project. Four demo sites comprise solutions for leakage control,
water quality management and energy optimisation incorporating sensors, data
processing, modelling and analytics technologies. The UK demo site (TWIST) in
reading investigated how emerging technologies can be used to create a SWN with
near real-time notification of performance to enable proactive management and
intervention (particularly for leakage). Technologies utilised during the pilot
have included installation of flow instruments through full-bore hydrants,
instruments capable of high-resolution monitoring (thus enabling the identification
of pressure transients), AMR customer smart meters as well as traditional sensors.
Three network leakage algorithms to detect leakage and other abnormalities
(as soon as they occur) in the water network were tested and assessed: AURA
BED alerts as described in [28, 31], dynamic bandwidth monitoring (DBM) and
Netbase envelopes (developed by Crowder Consulting). Having a standard approach
to test different algorithms allows an objective comparison of their effectiveness.
Increasing amounts of SWN data are only of real business value if this
valuable resource is ultimately used to inform and support decision-making,
i.e. data to information to insight to action. Projects such as SW4E allow the
exploration, at demo and full WDS pilot scale, of deploying multiple smart
network technologies, both hardware and software, and the multiplicative synergy
between them. The deployment of a smart water network has its own challenges
such as large network data stores, false positives, limited analytical capability, pipe
location and condition, failure prediction, meter coverage, response to failures, etc.
18
S. R. Mounce