Strategies to develop warm solutions for real-time pump scheduling for water distribution systems

M. Fayzul K. Pasha, Kevin E Lansey

Research output: Contribution to journalArticle

15 Citations (Scopus)

Abstract

An optimal pump operation schedule that maintains satisfactory hydraulics conditions can generally reduce energy consumptions compared to the traditional trial and error based pump operation schedule. Linking an evolutionary based optimization algorithm with a hydraulic simulation model has gained attention for obtaining the optimal schedule. However, this technique requires significant computation time and thus has difficulty in real-time implementation. This paper presents several tactics to generate warm solutions that can be used in the initial population of the evolutionary algorithms to reduce the computation times. Strategies to generate warm solutions include the use of linear programming, surrogate model known as machine learning or meta-model, and historical pump schedule for similar demand pattern. Providing warm solutions from approximate methods or previous day’s results to stochastic search methods can improve solution convergence and offers significant computation time benefits. Results obtained from different strategies are compared.

Original languageEnglish (US)
Pages (from-to)3975-3987
Number of pages13
JournalWater Resources Management
Volume28
Issue number12
DOIs
StatePublished - Jul 12 2014

Fingerprint

Water distribution systems
pump
Scheduling
Pumps
hydraulics
Hydraulics
linear programing
Evolutionary algorithms
Linear programming
Learning systems
Energy utilization
simulation
water distribution system
method

Keywords

  • Energy saving
  • Evolutionary algorithm
  • Historical solution
  • LP model
  • Real-time pump operation
  • Support vector machine
  • Warmsolution

ASJC Scopus subject areas

  • Civil and Structural Engineering
  • Water Science and Technology

Cite this

Strategies to develop warm solutions for real-time pump scheduling for water distribution systems. / Pasha, M. Fayzul K.; Lansey, Kevin E.

In: Water Resources Management, Vol. 28, No. 12, 12.07.2014, p. 3975-3987.

Research output: Contribution to journalArticle

@article{a49328bca9994643a2b75c1c3b98e8b7,
title = "Strategies to develop warm solutions for real-time pump scheduling for water distribution systems",
abstract = "An optimal pump operation schedule that maintains satisfactory hydraulics conditions can generally reduce energy consumptions compared to the traditional trial and error based pump operation schedule. Linking an evolutionary based optimization algorithm with a hydraulic simulation model has gained attention for obtaining the optimal schedule. However, this technique requires significant computation time and thus has difficulty in real-time implementation. This paper presents several tactics to generate warm solutions that can be used in the initial population of the evolutionary algorithms to reduce the computation times. Strategies to generate warm solutions include the use of linear programming, surrogate model known as machine learning or meta-model, and historical pump schedule for similar demand pattern. Providing warm solutions from approximate methods or previous day’s results to stochastic search methods can improve solution convergence and offers significant computation time benefits. Results obtained from different strategies are compared.",
keywords = "Energy saving, Evolutionary algorithm, Historical solution, LP model, Real-time pump operation, Support vector machine, Warmsolution",
author = "Pasha, {M. Fayzul K.} and Lansey, {Kevin E}",
year = "2014",
month = "7",
day = "12",
doi = "10.1007/s11269-014-0721-0",
language = "English (US)",
volume = "28",
pages = "3975--3987",
journal = "Water Resources Management",
issn = "0920-4741",
publisher = "Springer Netherlands",
number = "12",

}

TY - JOUR

T1 - Strategies to develop warm solutions for real-time pump scheduling for water distribution systems

AU - Pasha, M. Fayzul K.

AU - Lansey, Kevin E

PY - 2014/7/12

Y1 - 2014/7/12

N2 - An optimal pump operation schedule that maintains satisfactory hydraulics conditions can generally reduce energy consumptions compared to the traditional trial and error based pump operation schedule. Linking an evolutionary based optimization algorithm with a hydraulic simulation model has gained attention for obtaining the optimal schedule. However, this technique requires significant computation time and thus has difficulty in real-time implementation. This paper presents several tactics to generate warm solutions that can be used in the initial population of the evolutionary algorithms to reduce the computation times. Strategies to generate warm solutions include the use of linear programming, surrogate model known as machine learning or meta-model, and historical pump schedule for similar demand pattern. Providing warm solutions from approximate methods or previous day’s results to stochastic search methods can improve solution convergence and offers significant computation time benefits. Results obtained from different strategies are compared.

AB - An optimal pump operation schedule that maintains satisfactory hydraulics conditions can generally reduce energy consumptions compared to the traditional trial and error based pump operation schedule. Linking an evolutionary based optimization algorithm with a hydraulic simulation model has gained attention for obtaining the optimal schedule. However, this technique requires significant computation time and thus has difficulty in real-time implementation. This paper presents several tactics to generate warm solutions that can be used in the initial population of the evolutionary algorithms to reduce the computation times. Strategies to generate warm solutions include the use of linear programming, surrogate model known as machine learning or meta-model, and historical pump schedule for similar demand pattern. Providing warm solutions from approximate methods or previous day’s results to stochastic search methods can improve solution convergence and offers significant computation time benefits. Results obtained from different strategies are compared.

KW - Energy saving

KW - Evolutionary algorithm

KW - Historical solution

KW - LP model

KW - Real-time pump operation

KW - Support vector machine

KW - Warmsolution

UR - http://www.scopus.com/inward/record.url?scp=85027922623&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85027922623&partnerID=8YFLogxK

U2 - 10.1007/s11269-014-0721-0

DO - 10.1007/s11269-014-0721-0

M3 - Article

AN - SCOPUS:85027922623

VL - 28

SP - 3975

EP - 3987

JO - Water Resources Management

JF - Water Resources Management

SN - 0920-4741

IS - 12

ER -