Detecting leaks, infrastructure planning and accurate billing are the obvious benefits of smart meters and IoT sensors. But if you dive a little deeper into the data, there’s a lot more that can be learnt by reading between the lines.
Some of the real value of big data comes when you stop looking at individual sensors or meters and start comparing relationships between different types of data.
Even more interesting is when machine intelligence can learn what the “normal” behaviour of a network is, and send out an alert when something changes. For example, in many tropical regions, high rainfall events can lead to wet weather overflows from the sewer. If a 40mm rainfall event is typically needed to trigger a sewer overflow, but in subsequent events, the amount required to trigger the overflow reduces to 30mm, then 20mm, this points to a change in the capacity of the sewer network.
For one North Queensland utility, this apparent pattern in decreasing performance led the team to go out and flush the pipe to discover that it had been blocked by building rubble. This type of monitoring of data becomes impossible to do manually when there are hundreds, or even thousands, of sensors in the network, and it is only achievable when using specialised software that can sort and analyse such large volumes of data.
This analysis gets even more complex when analysing data across multiple types of sensors. For example, when tracking the source of odour complaints, data from weather stations about wind speed and direction, H2S concentrations in the air, and flow and level data from the sewer network can provide the utility with the information required to identify the source and cause of the odours, and take corrective actions such as changing pumping patterns or dosing levels.
The nature of data analysis is changing. Scada systems tend to generate alerts when a single sensor moves out of a defined range, but IoT systems can generate data from thousands of sensors. It is only by applying big data tools and techniques that the real insights will be identified.