Using customer flow meter data to inform water companies

Traditional domestic water meter

Water meters are fitted in every home and are vital for customer billing and water conservation. Historically, meter readings are taken manually and infrequently, only when a new bill needs to be generated. However, due to technological advancements, water meters have now become much more advanced. Automated meter readings (AMR) and automated metering infrastructure (AMI) systems are increasingly being adopted by utility providers.  Such meters reduce the manual labour required for water metering and the costs to consumers, while also providing much larger amounts of data on water flow. With large datasets, there is the potential to generate and gather large amounts of information from customer meters. Such information can benefit more flexible payment plans for customers (further reducing costs to consumers) and knowledge on the performance of water networks. This is extremely useful to water providers as such knowledge facilitates improvements to the quality and consistency of water services.

With such large data-sets and variations on what is considered “normal-use”, converting this data into valuable and informative information is a challenge. One method for analysing water flow datasets, such as those produced by customer water meters, is Comparison of Flow Pattern Distribution (CFPD) analysis. This method works by firstly ensuring that all data is in a consistent format and comparable, then ordering each dataset from smallest to largest and finally plotting the datasets alongside one another against a reference dataset. This facilitates the comparison of datasets to identify both inconsistent and consistent changes over temporal and spatial scales. This method has previously been applied at whole-system levels, but not for the analysis of domestic customer data.

Transforming big data into valuable information to inform water companies and direct management practices

Research by STREAMer Lucy Irons has attempted to apply this method to domestic customer datasets. In the analysis, datasets from 2000 Anglian Water customers were evaluated using CFPD analysis to test if this the information generated could be used to identify events of inconsistent change, such as a period of inoccupancy. Two approaches were tested: block analysis and rolling average reference (RAR) analysis. The block analysis was able to identify different types of leakages at the customer site, as well as periods of non-occupancy.  RAR analysis was able to also identify leaks and provide information on their duration.

The study was able to successfully demonstrate the potential use of CFPD analysis when applied to customer meter datasets to generate valuable information, with further refinements needed before application by water companies. If such analysis were able to be refined and effectively utilised by water companies, this would progress and facilitate better management of water leaks, reducing costs to both water companies and customers, and reducing water stress and water loss.

References:

Irons, L. M. et al. (2015) Data driven analysis of customer flow meter data, Procedia Engineering. Elsevier, 119, pp. 834–843.

Download paper