A vast amount of data of various types is collected by water utility companies, and in practice, they are utilized in numerous ways for regulatory reporting, informing operational decisions, and strategic planning activities. Between the process of collecting data in its raw form, through to applications, several intermediary steps are involved in transforming and extracting insights from the data to help inform decision-making. However, significant gaps still exist in evidencing how these datasets can be more effectively utilized for solving problems related to water distribution networks. This research project concentrates on leveraging a variety of datasets to improve decision-making in the optimization of water distribution networks, with Scottish Water as the industrial partner and a key collaborator. 

Between the process of collecting data in its raw form and its use in specific business applications, several intermediary steps are involved in transforming and extracting insights from the data to help inform decision-making. In this project, a thorough understanding of currently employed (and occasionally overlooked) data types, practical business applications, and the intermediary analysis layers is gained. This reveals opportunities for greater utilization of certain data types that serve useful purposes but are often underutilized, along with the great potential for the adoption of data-driven modelling approaches. Expanding on this opportunity space, one key proposed objective focuses on identifying complex (and often unapparent) relationships between datasets obtained from water distribution networks through pattern recognition techniques. Subsequently, a novel method is proposed for performing network diagnostics to detect and locate faulty elements, such as pumps, preventing severe operational issues. This proactive approach is a shift from the commonly used reactive approach to failures by identifying and addressing potential root causes before they become significant problems.