Water managers need annual replacement/rehabilitation plans for critical pipes. The development of reliable and robust failure models for assessing a pipe's propensity to fail, and how to assign criticality to an individual pipe segment, represents an integral part of a decision support systems' formulation. The Evolutionary Polynomial Regression (EPR) model identification technique has been recently used for developing aggregate failure models for both water distribution and sewer systems. This paper presents two recent improvements in the use of EPR for predicting pipe failures: (1) the Multi-Case EPR Strategy (MCS-EPR) and (2) a methodology of using EPR-derived models for predicting failure likelihood at the pipe level. The MCS-EPR permits the analysis of several systems/cases simultaneously. This facilitates the selection of the most general model structures containing only the key explanatory factors describing the underlying problem physics rather than noise in observed data (i.e., over-fitting). Deriving pipe-level failure prediction from aggregate EPR/MCS-EPR models is based on the interpretation of model structure and coefficients. Both proposed improvements are applied to 48 water quality zones of a UK water distribution network

Pipe level burst prediction using EPR and MCS-EPR

Giustolisi, Orazio;Berardi, Luigi
2007

Abstract

Water managers need annual replacement/rehabilitation plans for critical pipes. The development of reliable and robust failure models for assessing a pipe's propensity to fail, and how to assign criticality to an individual pipe segment, represents an integral part of a decision support systems' formulation. The Evolutionary Polynomial Regression (EPR) model identification technique has been recently used for developing aggregate failure models for both water distribution and sewer systems. This paper presents two recent improvements in the use of EPR for predicting pipe failures: (1) the Multi-Case EPR Strategy (MCS-EPR) and (2) a methodology of using EPR-derived models for predicting failure likelihood at the pipe level. The MCS-EPR permits the analysis of several systems/cases simultaneously. This facilitates the selection of the most general model structures containing only the key explanatory factors describing the underlying problem physics rather than noise in observed data (i.e., over-fitting). Deriving pipe-level failure prediction from aggregate EPR/MCS-EPR models is based on the interpretation of model structure and coefficients. Both proposed improvements are applied to 48 water quality zones of a UK water distribution network
Computer and Control in Water Industry, CCWI 2007
978-0-415-45415-5
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11589/14429
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • Scopus 15
  • ???jsp.display-item.citation.isi??? ND
social impact