Multi-objective optimization of energy efficiency and pressure management in large water distribution networks

Abstract

Many utilities around the world are required to constantly improve their operations. This is a job which is never fulfilled. Every day demand changes, new elements appear in a system, old elements are replaced and even new sensor technologies are implemented in the field. To better manage these dynamic changes utilities have been shifting towards a digitalization of their daily operations. Within this landscape many different issues which are faced by a utility can be addressed with the help of decision support systems. Such systems may be of particular relevance for issues pertinent to diverse stakeholders such as: operators, customers, decision makers, government and environmental agencies. Among the issues which are more relevant to operators are the continuous supply of water demands, and at an acceptable water quality level. As long as this is achieved, most utilities satisfy their customers. However, this does not come without costs. Reduction of operational costs is always in the loop of discussion of every utility, associated with operational actions such as reducing leakages, reducing energy consumption, reducing pipe breaks and overall improvement of asset management.
This research deals with two different aspects of operations of water distribution networks related to energy minimization and pressure management. Both problems are dealt through model-based optimization via pump scheduling and water network sectorization. The research findings are a stepping stone to improved operation of large water distribution systems where assets are known and there is a need for improvement of energy use. Both optimization problems have been performed using simulations made with EPANET 2.0, linked to heuristic algorithms. Due to the large number of optimization runs, High Performance Computing (HPC) was also employed, using the national computing grid of the Netherlands for research and educational institutions, named SURFSara HPC Cloud, and, to some extent using Microsoft Azure®.....