Data Envelopment Analysis Energy Efficiency Water Services Artificial Neural Networks Water-energy Nexus Operating Characteristics

External authors:

• Maria Molinos-Senante ( Universidad de Valladolid )
• Alexandros Maziotis ( Pontificia Universidad Catolica de Chile , New York Coll )

Abstract:

Understanding how energy efficient the water services are and what drives inefficiency can greatly assist water utilities in delivering sustainable services. This study employs a neural network (NN) approach to measure the energy efficiency of water services in relation to the volume of drinking water supplied and the number of connected properties. Unlike other non-parametric approaches, NN allows capturing the complex relationships and dependencies between various factors influencing energy efficiency of water companies. An empirical application for English and Welsh water utilities embracing water only companies (WoCs) and water and sewerage companies (WaSCs) over 2008-2020 was conducted. The average energy efficiency score was found to be 0.411, indicating that water utilities could potentially save 0.54 kWh per cubic meter of drinking water supplied. Notably, WaSCs exhibited better energy performance compared to WoCs, with energy efficiency scores of 0.559 and 0.239, respectively. Nevertheless, based on the volume of water delivered, WaSCs could save 0.65 kWh/m3 whereas WoCs potential energy savings are 0.24 kWh/m3. Energy efficiency remained relatively stable across the years, with average values of 0.440, 0.388 and 0.454 for the periods 2008-2010, 2011-2015, and 2016-2020, respectively. The analysis conducted using decision tree methods highlighted the relevance of water treatment quality and the source of raw water as key variables influencing the energy efficiency of water utilities. These findings can be valuable for policymakers, enabling them to gain a deeper understanding of the driving factors behind energy efficiency in water service provision.