Algorithmic network monitoring for a modern water utility: a case study in Jerusalem

摘要

We report on the design, deployment, and use of TaKaDu, a real-time algorithmic WaternInfrastructure Monitoring solution, with a strong focus on water loss reduction and control.nTaKaDu is provided as a commercial service to several customers worldwide. It has been in use atnHaGihon, the Jerusalem utility, since mid 2009. Water utilities collect considerable real-time datanfrom their networks, e.g. by means of a SCADA system and sensors measuring flow, pressure,nand other data. We discuss how an algorithmic statistical solution analyses this wealth of rawndata, flexibly using many types of input and picking out and reporting significant events andnfailures in the network. Of particular interest to most water utilities is the early detectionncapability for invisible leaks, also a means for preventing large visible bursts. The system alsondetects sensor and SCADA failures, various water quality issues, DMA boundary breaches,nunrecorded or unintended network changes (like a valve or pump state change), and other events,nincluding types unforeseen during system design. We discuss results from use at HaGihon,nshowing clear operational value.