Strategies for real time pump operation for water distribution systems

M. F K Pasha, Kevin E Lansey

Research output: Chapter in Book/Report/Conference proceedingConference contribution

10 Citations (Scopus)

Abstract

Energy which is the largest expense for water utility industries is greatly associated with the pump operation. An optimum pump operation schedule may reduce the cost significantly while maintaining the hydraulics in an acceptable range. To date evolutionary optimization algorithms are linked with the hydraulic simulation model to obtain the optimum pump operation schedule. However, this process of finding optimum solution requires extensive time consuming simulations. Therefore, computational time is the main constraint to develop a real time optimum controller for the pump operation. To reduce the computational time researchers are developing different strategies. This paper examines few strategies to develop the real time optimum controller. To reduce the computational time, optimization algorithm can start from a warm solution. Warm solution can be defined as the most recent solution which can be applicable for the time period for which the solution is being sought. Also, the relationships among pump operation energy, flow, demands and tank water levels can be linearized to get a linear program (LP) solved quickly. Metamodeling approach also can be used to approximate the nonlinear function. In this case, support vector machine (SVM), a type of artificial neural networks (ANNs) is trained with the expensive simulation data and then the model is linked with the evolutionary optimization algorithm to optimize the pump schedule. Results obtained by different approaches presented here are compared with results that obtained by a conventional hydraulic simulation model, EPANET linked with the evolutionary optimization algorithm, SFLA. Results are comparable to each other while the approaches proposed in this paper can find the near-optimal solutions with significantly minimum time.

Original languageEnglish (US)
Title of host publicationWater Distribution Systems Analysis 2010 - Proceedings of the 12th International Conference, WDSA 2010
Pages1456-1469
Number of pages14
DOIs
StatePublished - 2012
Event12th Annual International Conference on Water Distribution Systems Analysis 2010, WDSA 2010 - Tucson, AZ, United States
Duration: Sep 12 2010Sep 15 2010

Other

Other12th Annual International Conference on Water Distribution Systems Analysis 2010, WDSA 2010
CountryUnited States
CityTucson, AZ
Period9/12/109/15/10

Fingerprint

pump
hydraulics
simulation
energy flow
artificial neural network
water distribution system
water level
industry
cost
energy
water

Keywords

  • artificial neural networks
  • energy saving
  • LP model
  • Real time pump operation

ASJC Scopus subject areas

  • Water Science and Technology

Cite this

Pasha, M. F. K., & Lansey, K. E. (2012). Strategies for real time pump operation for water distribution systems. In Water Distribution Systems Analysis 2010 - Proceedings of the 12th International Conference, WDSA 2010 (pp. 1456-1469) https://doi.org/10.1061/41203(425)130

Strategies for real time pump operation for water distribution systems. / Pasha, M. F K; Lansey, Kevin E.

Water Distribution Systems Analysis 2010 - Proceedings of the 12th International Conference, WDSA 2010. 2012. p. 1456-1469.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Pasha, MFK & Lansey, KE 2012, Strategies for real time pump operation for water distribution systems. in Water Distribution Systems Analysis 2010 - Proceedings of the 12th International Conference, WDSA 2010. pp. 1456-1469, 12th Annual International Conference on Water Distribution Systems Analysis 2010, WDSA 2010, Tucson, AZ, United States, 9/12/10. https://doi.org/10.1061/41203(425)130
Pasha MFK, Lansey KE. Strategies for real time pump operation for water distribution systems. In Water Distribution Systems Analysis 2010 - Proceedings of the 12th International Conference, WDSA 2010. 2012. p. 1456-1469 https://doi.org/10.1061/41203(425)130
Pasha, M. F K ; Lansey, Kevin E. / Strategies for real time pump operation for water distribution systems. Water Distribution Systems Analysis 2010 - Proceedings of the 12th International Conference, WDSA 2010. 2012. pp. 1456-1469
@inproceedings{5f52bdc3372e45668171f1571707de0d,
title = "Strategies for real time pump operation for water distribution systems",
abstract = "Energy which is the largest expense for water utility industries is greatly associated with the pump operation. An optimum pump operation schedule may reduce the cost significantly while maintaining the hydraulics in an acceptable range. To date evolutionary optimization algorithms are linked with the hydraulic simulation model to obtain the optimum pump operation schedule. However, this process of finding optimum solution requires extensive time consuming simulations. Therefore, computational time is the main constraint to develop a real time optimum controller for the pump operation. To reduce the computational time researchers are developing different strategies. This paper examines few strategies to develop the real time optimum controller. To reduce the computational time, optimization algorithm can start from a warm solution. Warm solution can be defined as the most recent solution which can be applicable for the time period for which the solution is being sought. Also, the relationships among pump operation energy, flow, demands and tank water levels can be linearized to get a linear program (LP) solved quickly. Metamodeling approach also can be used to approximate the nonlinear function. In this case, support vector machine (SVM), a type of artificial neural networks (ANNs) is trained with the expensive simulation data and then the model is linked with the evolutionary optimization algorithm to optimize the pump schedule. Results obtained by different approaches presented here are compared with results that obtained by a conventional hydraulic simulation model, EPANET linked with the evolutionary optimization algorithm, SFLA. Results are comparable to each other while the approaches proposed in this paper can find the near-optimal solutions with significantly minimum time.",
keywords = "artificial neural networks, energy saving, LP model, Real time pump operation",
author = "Pasha, {M. F K} and Lansey, {Kevin E}",
year = "2012",
doi = "10.1061/41203(425)130",
language = "English (US)",
isbn = "9780784412039",
pages = "1456--1469",
booktitle = "Water Distribution Systems Analysis 2010 - Proceedings of the 12th International Conference, WDSA 2010",

}

TY - GEN

T1 - Strategies for real time pump operation for water distribution systems

AU - Pasha, M. F K

AU - Lansey, Kevin E

PY - 2012

Y1 - 2012

N2 - Energy which is the largest expense for water utility industries is greatly associated with the pump operation. An optimum pump operation schedule may reduce the cost significantly while maintaining the hydraulics in an acceptable range. To date evolutionary optimization algorithms are linked with the hydraulic simulation model to obtain the optimum pump operation schedule. However, this process of finding optimum solution requires extensive time consuming simulations. Therefore, computational time is the main constraint to develop a real time optimum controller for the pump operation. To reduce the computational time researchers are developing different strategies. This paper examines few strategies to develop the real time optimum controller. To reduce the computational time, optimization algorithm can start from a warm solution. Warm solution can be defined as the most recent solution which can be applicable for the time period for which the solution is being sought. Also, the relationships among pump operation energy, flow, demands and tank water levels can be linearized to get a linear program (LP) solved quickly. Metamodeling approach also can be used to approximate the nonlinear function. In this case, support vector machine (SVM), a type of artificial neural networks (ANNs) is trained with the expensive simulation data and then the model is linked with the evolutionary optimization algorithm to optimize the pump schedule. Results obtained by different approaches presented here are compared with results that obtained by a conventional hydraulic simulation model, EPANET linked with the evolutionary optimization algorithm, SFLA. Results are comparable to each other while the approaches proposed in this paper can find the near-optimal solutions with significantly minimum time.

AB - Energy which is the largest expense for water utility industries is greatly associated with the pump operation. An optimum pump operation schedule may reduce the cost significantly while maintaining the hydraulics in an acceptable range. To date evolutionary optimization algorithms are linked with the hydraulic simulation model to obtain the optimum pump operation schedule. However, this process of finding optimum solution requires extensive time consuming simulations. Therefore, computational time is the main constraint to develop a real time optimum controller for the pump operation. To reduce the computational time researchers are developing different strategies. This paper examines few strategies to develop the real time optimum controller. To reduce the computational time, optimization algorithm can start from a warm solution. Warm solution can be defined as the most recent solution which can be applicable for the time period for which the solution is being sought. Also, the relationships among pump operation energy, flow, demands and tank water levels can be linearized to get a linear program (LP) solved quickly. Metamodeling approach also can be used to approximate the nonlinear function. In this case, support vector machine (SVM), a type of artificial neural networks (ANNs) is trained with the expensive simulation data and then the model is linked with the evolutionary optimization algorithm to optimize the pump schedule. Results obtained by different approaches presented here are compared with results that obtained by a conventional hydraulic simulation model, EPANET linked with the evolutionary optimization algorithm, SFLA. Results are comparable to each other while the approaches proposed in this paper can find the near-optimal solutions with significantly minimum time.

KW - artificial neural networks

KW - energy saving

KW - LP model

KW - Real time pump operation

UR - http://www.scopus.com/inward/record.url?scp=84856143281&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84856143281&partnerID=8YFLogxK

U2 - 10.1061/41203(425)130

DO - 10.1061/41203(425)130

M3 - Conference contribution

SN - 9780784412039

SP - 1456

EP - 1469

BT - Water Distribution Systems Analysis 2010 - Proceedings of the 12th International Conference, WDSA 2010

ER -