Refining PSO Applied to Electric Energy Cost Reduction in Water Pumping

Bruno Melo Brentan1 & Edevar Luvizotto Jr2

 

1 School of Civil Engineering, Architecture and Urban Studies, State University of Campinas (FEC – UNICAMP), Brazil; E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

2 School of Civil Engineering, Architecture and Urban Studies, State University of Campinas (FEC – UNICAMP), Brazil; E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

 

 

 

Abstract

The growing urban population growth necessitates that the water delivery sector uses safe, economical operations. In this context, an increasing number of operational routines has been tested that can adequately handle standard impositions and consumer's needs. The search for optimum routines for water pumping for startup and stopping and pump rotation variation has become increasingly common because of the need to reduce energy consumption, which therefore promotes the application of various optimization techniques. Among such techniques, special attention has been given to those inspired by nature, such as Particle Swarm Optimization (PSO), a technique based on the intelligence of groups, such as fish schools or insect swarms. This study presents a hybrid algorithm (simulator-optimizer) to determine optimized operational routines for pumping stations using PSO to define the number of pumps that are running (in nominal rotation) and rotations, which can therefore satisfy the operational restrictions at the time when the pumps are running. The performance evaluation is conducted by applying the model to an actual distribution network, where the energy cost was reduced by approximately 60%.

Keywords: Water supply system, energy efficiency, optimized operation, PSO

 

 

Introduction

 

One of the primary challenges that water supply companies currently face is the difficulty in dealing with urban growth, which is often disorganized, and maintaining safe, efficient water delivery. This growth incurs a constant need to expand the water delivery system, which increases energy consumption; this energy consumption is between two and three percent of the worldwide consumption (Vilanova and Balestieri, 2014)

Much of the delivery systems operational control is under the decision of operators who, by acting directly or by the command centre, load the operational rules based on their experience acquired from their career. However, the commands based on the operator's experience may lead to better performance when operations are executed without any knowledge of the system. Differently though, emergency situations, such as unpredicted stops of the pumping system or a duct rupture, may lead to emotionally affected decisions that are not always the most appropriate (Sandeep and Rakesh, 2011).

To reduce the reliance on empiricism in decision making, the literature proposes various tools that create responses from optimization mathematical models to aid operators in decision making. Among these models, hybrid models (optimizer-simulator) are capable of determining optimum manoeuvres for a given operational period.

Optimized operational routines related to pump start-up and stopping manoeuvres or rotation changes by a frequency inverter significantly reduces energy consumption, which has been reported in literature. The most common optimization techniques used to search for better operational routines are dynamic programming (Jowitt and Germanopoulos, 1992) and evolutionary techniques, such as genetic algorithms: (Cunha, 2009), (Andrade, et al., 2008), (Ribeiro, 2007), (Rodrigues, 2007), (Farmani et al., 2007), evolutionary multi-objective techniques (Wang, Chang and Cheng, 2009), (Barán, Lücken and Sotelo, 2005) or even particle swarm optimization (PSO) (Al-Ani, 2012).

The scan of the search area using meta-heuristics algorithms is arduous and imposes great computational efforts in problems with many variables. Hence, differently from the studies from literature, the present study develops an algorithm, hereby named Refining PSO, which consists of two steps. The objective of the algorithm is to reduce the computational effort for the simulation by decreasing the number of variables involved in the continuous stage (stage of rotation definition, phase two) once the operational status (shutdown or operating pump) is obtained in the first phase (discrete stage) by binary optimization. It is important to note that the strategy to define the pumps that are running versus the pumps that are not in operation avoids operational drawbacks because the search for only rotations may lead to low values, which in turn, can lead to undesirable physical phenomena, such as cavitation and vibration.