Water & Wastewater Treatment

www.wwtonline.co.uk | WWT | JANUARY 2016 | 27 In the know Digging deeper: data Big data dilemma Linking the disparate sources of data that exist across the business can be the key to water companies' data challenges Mike Jones PRiNciPAL coNsULTANT, UTiLiTies UNiT iNfosYs I ncreasingly utility companies are being steered towards Big Data, and the benefits that can be derived from mining and analysing the vast quantity of customer and operational data they hold. Water companies have large physical asset bases, distributed in varying concentrations, over regions of mixed topography, geology and land use. As a result GIS have a potentially significant role to play in planning and executing capital and operational programmes. Large silos of structured and unstructured data are already a common feature for utilities such as volumes of data requires the highest specification computer hardware, with fast access to network drives. In this sense 'big data' is nothing new in the utilities sector. However, many 'Big Data' calls miss a fundamental issue, in that asset 'data' is expensive to obtain and consistently maintain. Where AMR (Automated Meter Reading) water metering is installed, customer usage data is readily available. However, most utilities' physical assets are geographically widely spread, sometimes in locations difficult to access, and the cost of gathering and maintaining data can thus be restrictively high. For example, a manhole survey can cost an average of £70 or more; with over 550,000km of sewers in the UK, and assuming 40m between manholes, a 1% validation survey would cost circa £10 million. Surveys can also have complex health and safety risks that need to be managed, such as working in confined spaces, at height, or in the middle of busy roadways with cars and lorries passing at speed just feet away. This is just one element of data, and surveys in complex situations can be extremely expensive, with elements such as underground, undersea and otherwise covered pipeline surveys sometimes costing in the region of £I million. For these reasons, asset data is o–en limited and of dubious quality. Sensors and instrumentation are improving data collection and data flow. Sensors are cheaper to install, run and maintain, and are more robust. Nonetheless, they are still relatively expensive items in terms of the up-front cost when making the water network more data-generative. With asset data o–en being limited, suspect, and costly to improve, and sensors and instrumentation expensive to deploy, smarter utilities such as water companies are looking to make better use of the information they already hold in order to understand the network, supply and demand, customer expectations and future preventative maintenance. By using a combination of engineering knowledge coupled with effective analytics, trends can be mapped and normal asset behaviour determined. Where data is readily available such analysis is relatively simple, however where asset data is limited, engineering knowledge and understanding can be used to define Now that utilities are using diverse data sources, its storage requires substantial iT infrastructure water companies. Beyond the more familiar IT ground of customer data, modelling of assets, pipeline flow data, operational data from treatment plants and other waypoints has stretched the limits of systems, both in terms of hardware and storage, since the 1980s (and earlier in a few cases). Now that utilities are using diverse data sources, including weather (e.g. radar rainfall), 3D mapping, instrumentation and third party sources, even the storage of data can require substantial IT infrastructure. Terabytes of stored files are now common within the water industry. Modelling large

• Cover