Energy rating of a water pumping station using multivariate analysis

Among water management policies, the preservation and the saving of energy demand in water supply and treatment systems play key roles. When focusing on energy, the customary metric to determine the performance of water supply systems is linked to the definition of component-based energy indicators. This approach is unfit to account for interactions occurring among system elements or between the system and its environment. On the other hand, the development of information technology has led to the availability of increasing large amount of data, typically gathered from distributed sensor networks in so-called smart grids. In this context, data intensive methodologies address the possibility of using complex network modeling approaches, and advocate the issues related to the interpretation and analysis of large amount of data produced by smart sensor networks. In this perspective, the present work aims to use data intensive techniques in the energy analysis of a water management network. The purpose is to provide new metrics for the energy rating of the system and to be able to provide insights into the dynamics of its operations. The study applies neural network as a tool to predict energy demand, when using flowrate and vibration data as predictor variables.