Water authorities are taking steps to ensure that smart-meter data is stripped of publicly identifiable information before being fed into analytics engines that are expected to significantly improve leak detection and customer information in an award-winning smart-meter program in Townsville, Queensland.
The Townsville Smart Water Pilot program – currently being run by the Townsville City Council and IBM as a byproduct of Townsville’s 2011 receipt of an IBM Smarter Cities Challenge grant – is testing the collection, analysis and use of water-usage information from meters that have so far been installed in 210 homes across the Townsville suburbs of Aitkenvale and Bushland Beach. IBM has been analysing usage information collected from Taggle Systems-designed wireless meters, which have been designed with ten-year batteries to minimise maintenance costs, to provide real-time usage data and trend information more accurate than what has been available in the past.
The system’s ability to pick out anomalous water distribution – a sign of a crack in the water utility’s underground network of pipes – is expected to lead to better water conservation and network maintenance programs; a similar program previously run by IBM and the US city of Dubuque, Iowa reduced water consumption by 6.6 percent and improved leak detection eight times over previous methods.
“In the past people were drawing consumption data up into head-ends that were sitting on the communications network, and bringing it into an Access database or Excel spreadsheets, and trying to do some analysis,” Glen Garner, senior managing consultant at IBM Global Business Services’ Centre of Competency for Energy & Utilities, told CSO Australia.
“We’ve transformed it into a very sophisticated model that enables you to do some very neat mathematics to understand where the asset is at.”
While it has been getting nods from industry – most recently, with the receipt of a Smart Infrastructure Project award at Infrastructure Partnerships Australia’s 2013 National Infrastructure Awards – the system also reflects a growing trend towards collection and aggregation of large volumes of data about citizens’ behaviour – and that has necessitated a proactive approach to ensuring the data cannot be easily matched to citizens’ houses.
Real-time readings of a household’s water usage could, for example, be used by thieves to determine when a family was home and not home, or away on holidays. Similar concerns have plagued rollouts of electricity smart meters, such as those being deployed across Victoria. It’s a real-time reminder of the security implications of big-data aggregation and analysis, says Garner, who notes that the power of big data has been tempered by the need to preserve citizens’ privacy.
“We make sure the data we get from Taggle is blinded,” he said, noting that IBM only has a customer number that is associated with each meter. “We don’t want to make a connection between the physical address and person that is the customer in our system; we just do the analytics.”
Macro trends in usage information can prove helpful for analytic purposes, however: for example, the project team is looking into ways of using pre-provided demographic information to compare households based on their number of residents.
Households with over six residents have distinctive water-usage patterns, Garner said: “we are able to group like residents together, but again we don’t have residents’ details; we just know that they’re in a demographic group that’s similar.”
If necessary, council systems can cross-match pilot participants’ customer numbers with ratepayer information to provide interactivity further down the chain – but this sort of activity would happen within the council and subject to its normal privacy controls.
“This lets us have weekly usage contests, for example, and the person with the lowest usage of a demographic area might get a prize. Those are the sorts of things you can do while keeping the privacy concept out there.”