Water Network Boundaries And Data Mining
In the book ‘Thinking in Systems’, Donella Meadows describes the importance of defining a system’s boundary in order to answer key questions. In centralised water systems, defining boundaries can be achieved using the location of flow meters. In broad terms data locations are referred to as:
1. Source –the dam, river or treatment plant
2. Network –pump stations, pipes and reservoirs that distribute water throughout rural and urban landscapes
3. Household (end-use) – Data collected by smart meters at properties
Historically, engineers used the source data because it alone captured the story of consumption. With this, decision-makers could generalise about a water system’s seasonal behaviours and annual trends.
As time has passed, the source data has provided the longest data record which can be useful to estimate system-wide probabilities of peak demand events, amongst other things. Furthermore, the source data plays a critical role in defining triggers to mitigate the negative impacts of drought. This includes triggers for water restrictions and construction of new infrastructure to improve system robustness.
As the source data was the only reliable record in the early days of data collection, it was associated with decision making with issues across the network, for example proposed development. However, the source data could not account for spatial variability within the network. Spatial variability of water consumption is now evident in the historical network data and is associated with behaviour of consumers, property type and size, the efficiency of water usage (technology) and key climate parameters temperature and rainfall (Beal & Stewart 2013; Chang, Praskievicz & Parandvash 2014; Gurung et al. 2014).
Data collection throughout the network started in the late 90’s with the invaluable introduction of real time SCADA monitoring systems (Williams, G & Kuczera, G 2015). SCADA systems revolutionised the day-to-day operation of the network as operator’s could now monitor and control the system in real time. Prior to this, if a pipe failure occurred for example, operators relied on the community to notify them of the pipe burst. Though, community notification to the utility has stood the test of time and to this day plays a valuable role in identifying issues quickly (Williams, GS 2016). Real time monitoring of the network has been invaluable at times of emergency storm situations and when the network is stressed due to peak demand events.
With the emergence of relatively cheap IoT sensors, system management can be improved by extending monitoring capabilities as well as ‘filling-gaps’ for existing monitoring programs. At the scale of subnetworks, a deeper understanding of spatial variability of consumption provides significant value.
In 2019 there is an inevitability that households will eventually have individual smart meters. Consumers with a deeper understanding of how they in many cases ‘unconsciously’ waste water should be motivated to adopt more water wise behaviours. This technology may also lead to a restructure in water bills where the peak and off-peak periods have different pricing structures, similar to the electricity sector.
Data mining of historical water network data can significantly improve the evidence-base for decision making for new infrastructure and operational management. FSA Data conducts data mining of historical network data to connect the dots with catchment behaviours and peak demand events to inform decision making.
Beal, CD & Stewart, RA 2013, 'Identifying residential water end uses underpinning peak day and peak hour demand', Journal of Water Resources Planning and Management, vol. 140, no. 7, p. 04014008.
Chang, H, Praskievicz, S & Parandvash, H 2014, 'Sensitivity of urban water consumption to weather and climate variability at multiple temporal scales: The case of Portland, Oregon'.
Gurung, TR, Stewart, RA, Sharma, AK & Beal, CD 2014, 'Smart meters for enhanced water supply network modelling and infrastructure planning', Resources, Conservation and Recycling, vol. 90, pp. 34-50.
Williams, GS & Kuczera, G 2015, 'Framework for forensic investigation of associations between operational states and pipe failures in water distribution systems', Journal of Water Resources Planning and Management, Vol. 142, Issue 3.
Williams, GS 2016, 'Forensic systems analysis linking pipe failures and operating states in water distribution systems', Ph.D. thesis, University of Newcastle, Australia, <http://hdl.handle.net/1959.13/1322527>.