Practical application of penalty-free evolutionary multi-objective optimisation of water distribution systems

Alemtsehay G. Seyoum, Tiku T. Tanyimboh, Calvin Siew

Research output: Contribution to conferenceProceeding

Abstract

Evolutionary algorithms are a commonly applied optimisation approach in water distribution systems. However, the algorithms are time consuming when applied to large optimisation problems. The aim of this paper is to evaluate the application of a penalty-free multi-objective evolutionary optimisation algorithm to solve a real-life network design problem. The optimization model uses pressure-dependent analysis that accounts for the pressure dependency of the nodal flows and thus avoids the need for penalties to address violations of the nodal pressure constraints. The algorithm has been tested previously using benchmark optimisation problems in the literature. In all cases, the algorithm found improved solutions and/or the best solution reported previously in the literature with considerably fewer function evaluations. In this paper, a real-life network with over 250 pipes was considered. The network comprises multiple sources, multiple demand categories, many fire flows and involves extended period simulation. Due to the size and complexity of the optimization problem, a high performance computer that comprises multiple cores was used for the computational solution. Multiple optimisation runs were performed concurrently. Overall, the algorithm performs well; it consistently provides least cost solutions that satisfy all the system requirements quickly. The least-cost design obtained was over 40% cheaper than the existing network in terms of the pipe costs.

Conference

Conference9th World Congress of the European Water Resources Association, Water Resources Management in a Changing World: Challenges and Opportunities
Abbreviated titleEWRA2015
CountryTurkey
CityIstanbul
Period10/06/1513/06/15

Fingerprint

Water distribution systems
Multiobjective optimization
Pipe
Costs
Function evaluation
Evolutionary algorithms
Fires

Keywords

  • penalty-free multi-objective evolutionary optimisation
  • water distribution systems
  • pressure-dependent analysis
  • genetic algorithm
  • high performance computing
  • redundant binary codes

Cite this

Seyoum, A. G., Tanyimboh, T. T., & Siew, C. (2015). Practical application of penalty-free evolutionary multi-objective optimisation of water distribution systems. 9th World Congress of the European Water Resources Association, Water Resources Management in a Changing World: Challenges and Opportunities, Istanbul, Turkey.
Seyoum, Alemtsehay G. ; Tanyimboh, Tiku T. ; Siew, Calvin. / Practical application of penalty-free evolutionary multi-objective optimisation of water distribution systems. 9th World Congress of the European Water Resources Association, Water Resources Management in a Changing World: Challenges and Opportunities, Istanbul, Turkey.7 p.
@conference{e553ab9f626740028fdf1a826fc12152,
title = "Practical application of penalty-free evolutionary multi-objective optimisation of water distribution systems",
abstract = "Evolutionary algorithms are a commonly applied optimisation approach in water distribution systems. However, the algorithms are time consuming when applied to large optimisation problems. The aim of this paper is to evaluate the application of a penalty-free multi-objective evolutionary optimisation algorithm to solve a real-life network design problem. The optimization model uses pressure-dependent analysis that accounts for the pressure dependency of the nodal flows and thus avoids the need for penalties to address violations of the nodal pressure constraints. The algorithm has been tested previously using benchmark optimisation problems in the literature. In all cases, the algorithm found improved solutions and/or the best solution reported previously in the literature with considerably fewer function evaluations. In this paper, a real-life network with over 250 pipes was considered. The network comprises multiple sources, multiple demand categories, many fire flows and involves extended period simulation. Due to the size and complexity of the optimization problem, a high performance computer that comprises multiple cores was used for the computational solution. Multiple optimisation runs were performed concurrently. Overall, the algorithm performs well; it consistently provides least cost solutions that satisfy all the system requirements quickly. The least-cost design obtained was over 40{\%} cheaper than the existing network in terms of the pipe costs.",
keywords = "penalty-free multi-objective evolutionary optimisation, water distribution systems, pressure-dependent analysis, genetic algorithm, high performance computing, redundant binary codes",
author = "Seyoum, {Alemtsehay G.} and Tanyimboh, {Tiku T.} and Calvin Siew",
year = "2015",
month = "6",
day = "13",
language = "English",
note = "9th World Congress of the European Water Resources Association, Water Resources Management in a Changing World: Challenges and Opportunities, EWRA2015 ; Conference date: 10-06-2015 Through 13-06-2015",

}

Seyoum, AG, Tanyimboh, TT & Siew, C 2015, 'Practical application of penalty-free evolutionary multi-objective optimisation of water distribution systems' 9th World Congress of the European Water Resources Association, Water Resources Management in a Changing World: Challenges and Opportunities, Istanbul, Turkey, 10/06/15 - 13/06/15, .

Practical application of penalty-free evolutionary multi-objective optimisation of water distribution systems. / Seyoum, Alemtsehay G.; Tanyimboh, Tiku T.; Siew, Calvin.

2015. 9th World Congress of the European Water Resources Association, Water Resources Management in a Changing World: Challenges and Opportunities, Istanbul, Turkey.

Research output: Contribution to conferenceProceeding

TY - CONF

T1 - Practical application of penalty-free evolutionary multi-objective optimisation of water distribution systems

AU - Seyoum, Alemtsehay G.

AU - Tanyimboh, Tiku T.

AU - Siew, Calvin

PY - 2015/6/13

Y1 - 2015/6/13

N2 - Evolutionary algorithms are a commonly applied optimisation approach in water distribution systems. However, the algorithms are time consuming when applied to large optimisation problems. The aim of this paper is to evaluate the application of a penalty-free multi-objective evolutionary optimisation algorithm to solve a real-life network design problem. The optimization model uses pressure-dependent analysis that accounts for the pressure dependency of the nodal flows and thus avoids the need for penalties to address violations of the nodal pressure constraints. The algorithm has been tested previously using benchmark optimisation problems in the literature. In all cases, the algorithm found improved solutions and/or the best solution reported previously in the literature with considerably fewer function evaluations. In this paper, a real-life network with over 250 pipes was considered. The network comprises multiple sources, multiple demand categories, many fire flows and involves extended period simulation. Due to the size and complexity of the optimization problem, a high performance computer that comprises multiple cores was used for the computational solution. Multiple optimisation runs were performed concurrently. Overall, the algorithm performs well; it consistently provides least cost solutions that satisfy all the system requirements quickly. The least-cost design obtained was over 40% cheaper than the existing network in terms of the pipe costs.

AB - Evolutionary algorithms are a commonly applied optimisation approach in water distribution systems. However, the algorithms are time consuming when applied to large optimisation problems. The aim of this paper is to evaluate the application of a penalty-free multi-objective evolutionary optimisation algorithm to solve a real-life network design problem. The optimization model uses pressure-dependent analysis that accounts for the pressure dependency of the nodal flows and thus avoids the need for penalties to address violations of the nodal pressure constraints. The algorithm has been tested previously using benchmark optimisation problems in the literature. In all cases, the algorithm found improved solutions and/or the best solution reported previously in the literature with considerably fewer function evaluations. In this paper, a real-life network with over 250 pipes was considered. The network comprises multiple sources, multiple demand categories, many fire flows and involves extended period simulation. Due to the size and complexity of the optimization problem, a high performance computer that comprises multiple cores was used for the computational solution. Multiple optimisation runs were performed concurrently. Overall, the algorithm performs well; it consistently provides least cost solutions that satisfy all the system requirements quickly. The least-cost design obtained was over 40% cheaper than the existing network in terms of the pipe costs.

KW - penalty-free multi-objective evolutionary optimisation

KW - water distribution systems

KW - pressure-dependent analysis

KW - genetic algorithm

KW - high performance computing

KW - redundant binary codes

UR - http://www.ewra2015istanbul.org/

M3 - Proceeding

ER -

Seyoum AG, Tanyimboh TT, Siew C. Practical application of penalty-free evolutionary multi-objective optimisation of water distribution systems. 2015. 9th World Congress of the European Water Resources Association, Water Resources Management in a Changing World: Challenges and Opportunities, Istanbul, Turkey.