Keywords: pump scheduling, evolutionary computation, genetic algorithms, Pareto dominance, multiobjective optimisation, water supply systems.

Typically, a pumping station of a water supply system is composed of a set of different pumps. These pumps are used to supply reservoirs, located through out the community. In time, these reservoirs supply consumers. Pumps work in combination with each other in order to satisfy water demand from the community. This system must satisfy several hydraulic and technical restrictions. Thus, at a particular point in time, some pumps may be working while others may not. Hence, a pump schedule is the set of all pump combinations chosen for every time interval in a scheduling scope. Therefore, an optimal pump schedule is the one that optimises established objectives (i.e., cost of energy and maintenance) while fulfilling all restrictions of the system.

Some approaches to this problem have been presented showing that important savings can be made; especially when evolutionary algorithms are used [1,2,3]. The purpose of this work is to utilise several recognised optimisation algorithms to solve an optimal pump-scheduling problem and to compare their performance. Without lost of generality, a simplified hydraulic model was chosen. It consists of a water source, a pumping station with five pumps, a water reservoir and a main pipe to drive water from the station to the reservoir. Restrictions considered in this work include: maximum and minimum levels in the reservoir, water demand, technical characteristics of pump combinations and others [4].

An important contribution of this work is the analysis of four simultaneous minimisation objectives. The first one is the cost of electrical energy consumed by the pumps. The second one is the pump's maintenance cost. The third one is the level variation in the reservoir between the beginning and the end of the optimisation period, and finally, the maximum power peak, considering the cost of reserved power.

This work solves the optimal pump-scheduling problem using Multiobjective Evolutionary Algorithms (MOEAs). For the first time, six recognised MOEAs are applied on this problem: the Strength Pareto Evolutionary Algorithm (SPEA) [5], the Non Dominated Sorting Genetic Algorithm (NSGA) [6], its second version (NSGA2) [7], the Controlled Elitist Non Dominated Sorting Genetic Algorithm (CNSGA) [8], the Niched Pareto Genetic Algorithm (NPGA) [9] and the Multiple Objective Genetic Algorithm (MOGA) [10]. In order to satisfy hydraulic and technical restrictions, a heuristic algorithm was developed and combined with the above algorithms.

Multiobjective optimisation metrics [11] were used to compare the performance of MOEAs. Experimental results show that SPEA is the best method for this problem, although other algorithms may also be useful. Furthermore, SPEA's set of solutions provides pumping station engineers with many optimal pump schedules to choose from. Engineer's criteria can then be used to make a final selection, knowing other compromise alternatives.

1

G. Mackle, D. Savic and G. Walters, "Application of Genetic Algorithms to Pump Scheduling for Water Supply", GALESIA'95, London, 1995.

2

D. Savic, G. Walters and M. Schawb, "Multiobjective Genetic Algorithms for Pump Scheduling in Water Supply", ERASMUS, School of Engineering of Exeter-UK and Stuttgart University-Germany, 1997.

3

A. Sotelo and B. Barán, "Pumping Cost Optimization in Water Supply Systems Using an Evolutionary Multiobjective Combined Algorithm". XV Chilean Conference on Hydraulic Engineering, 337-347, Chile, 2001

4

F. Dolqachev and N. Pashkov, "Hydraulics and Hydraulic Machines", Mir, Moscow, 1985.

5

E. Zitzler and L. Thiele, "Multiobjective Evolutionary Algorithms: A Comparative Case Study and the Strength Pareto Approach", IEEE Transactions on Evolutionary Computation, 3(4), 257-271, 1999. doi:10.1109/4235.797969

6

N. Srinivas and K. Deb, "Multiobjective Optimisation Using Nondominated Sorting Genetic Algorithm", Evolutionary Computation, 2(3), 221-248, 1994. doi:10.1162/evco.1994.2.3.221

7

K. Deb, S. Agrawal, A. Pratab and T. Meyarivan, "A Fast Elitist Nondominated Sorting Genetic Algorithm for Multiobjecive Optimisation: NSGA-II", Parallel Problem Solving from Nature VI Conference, 849-858, Springer, 2000.

8

K. Deb and T. Goel, "Controlled Elitist Non-dominated Sorting Genetic Algorithms for Better Convergence", First International Conference on Evolutionary Multi- Criterion Optimisation, 67-81, Zurich, 2001.

9

J. Horn, N. Nafpliotis and D. Goldberg, "A Niched Pareto Genetic Algorithm for Multiobjective Optimisation", First IEEE Conference on Evolutionary Computation, IEEE World Congress on Computational Intelligence, 1, 82-87, New Jersey, 1994. doi:10.1109/ICEC.1994.350037

10

C. Fonseca and P. Fleming, "Genetic Algorithms for Multiobjective Optimization: Formulation, Discussion and Generalisation", Fifth International Conference on Genetic Algorithms, 416-423, California, 1993.

11

D. Van Veldhuizen, "Multi-objective Evolutionary Algorithms: Classification, Analysis and New Innovation", Ph.D. Thesis, Air Force Institute of Technology, Ohio, 1999.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £85 +P&P)