Please use this identifier to cite or link to this item:
https://hdl.handle.net/2440/80941
Type: | Conference item |
Title: | Optimal control of pumping at a water filtration plant using genetic algorithms |
Author: | Simpson, A. Sutton, D. Keane, D. Sherriff, S. |
Citation: | International Conference on Computing and Control for the Water Industry : Water industry systems modelling and optimization applications CCWI '99 13-15 September 1999, Exeter, United Kingdom/ Dragan Savic and Godfrey Walters (eds.): 10 p |
Publisher: | University of Exeter, Centre for Water Systems, 2005 |
Issue Date: | 1999 |
ISBN: | 0863802494 |
Conference Name: | International Conference on Computing and Control for the Water Industry (1999 : Exeter, UK) |
Statement of Responsibility: | A.R. Simpson, D.C. Sutton, D.S. Keane and S. J. Sheriff |
Abstract: | At many water filtration plants water is delivered by pumps from a clear water storage to nearby tanks in other parts of the water distribution system. The control of the pumps in terms of startup and shut-down is usually achieved by trigger levels based on tank low and high water levels. This paper investigates the use of genetic algorithm optimisation to optimise the operation of pumps with the objective of minimising the pumping cost whilst staying within the operating constraints of minimum and maximum water levels for the tanks. Results for a case study are presented. Two different formulations of the genetic algorithm are considered. In both cases, it is assumed that the demand for the day has been forecast and that the water treatment output has been set to this value. Penalty costs are applied to degrade the fitness of solutions in a genetic algorithm population when the tank level drops below or rises above contracted minimum and maximum levels. In addition, a penalty cost is applied when a pump starting limitation is violated. In the first formulation, the trigger levels for the lower tank are optimised. The optimised trigger levels were found to be dependent on the starting level in the upper tank. In the second formulation, the trigger levels in the upper tank are optimised. The outcome of the genetic algorithm optimisation for the second case shows that the lower tank is too small to enable any feasible selection of trigger levels for the upper tank. The results show that there is potential for savings in pumping costs by using genetic algorithm optimisation for real-time operation of water filtration plant pump operations. |
Rights: | Copyright status unknown |
Appears in Collections: | Aurora harvest 4 Civil and Environmental Engineering publications |
Files in This Item:
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.