The ageing of water networks results in an increase in water pipe breaks in addition to a decrease in hydraulic capacity. Considering the complexity of these proce sses combined with the huge investments municipalit ies will have to make to maintain an adequate service level, it is imperative to develop tools that will assist water supply utilities managers in selecting, among the availabl e options, those that will minimize the total cost s on the long term. This article presents a new strategy for the optimal replacement of water pipes. It integrates t he wo key elements involved in the deterioration of water sup ply services, namely the structural integrity and t he hydraulic capacity. The objective function used to define opt imal solutions comprises two terms: one related to repair costs and another to replacement costs. The optimal solution must minimize this function under the con straints that all node demands and pressure are satisfied. T he model used to estimate the probability of pipe b reak occurrences considers time intervals between succes sive pipe breaks as a random variable described by probability density functions. A Bayesian approach is used to estimate the model parameters values. Ne twork hydraulics are modeled using Epanet2.0, and a genet ic algorithm (GA) is used to seek the optimal solut i n. The validation and the performance evaluation of the pr oposed strategy have been realized by generating st ochastic pipe breaks on a water pipe network. The network “l ifetime” has been subdivided into five-year time in tervals. The planning schedule for the next five years is de fined at the beginning of these time intervals (i.e . which pipes are replaced, and when they are replaced according to the optimization results). At the beginning of e ach of these periods, parameters values of the pipe break model are re-evaluated according to break records a vail ble at that time. Once the water pipe network is upgrad ed, pipe break records are extended to the next fiv e years. This process of identifying water pipes to be repla ced for the next five years is repeated until the e nd of the network “lifetime”. Results are reported for two hy pothetical water networks of 100 and 250 pipes, res pectively.
[1]
Uri Shamir,et al.
An Analytic Approach to Scheduling Pipe Replacement
,
1979
.
[2]
Marc Parizeau,et al.
Open BEAGLE: A New C++ Evolutionary Computation Framework
,
2002,
GECCO.
[3]
J. Villeneuve,et al.
Optimal replacement of water pipes
,
2003
.
[4]
Barry J. Adams,et al.
LONG-TERM PLANNING METHODOLOGY FOR WATER DISTRIBUTION SYSTEM REHABILITATION
,
1998
.
[5]
D. Cox,et al.
Analysis of Survival Data.
,
1986
.
[6]
Graeme C. Dandy,et al.
Optimal Scheduling of Water Pipe Replacement Using Genetic Algorithms
,
2001
.
[7]
L. Dridi.
Développement et validation d'une approche de remplacement des conduites pour les réseaux d'eau potable.
,
2005
.
[8]
Patrick Eisenbeis.
Modelisation statistique de la prevision des defaillances des conduites d'eau potable
,
1994
.
[9]
Robert M. Clark,et al.
Cost models for Water supply distribution systems
,
2002
.
[10]
Thomas M. Walski,et al.
Predicting Internal Roughness in Water Mains
,
1988
.
[11]
Larry W. Mays,et al.
Optimal Rehabilitation Model for Water‐Distribution Systems
,
1993
.
[12]
Barry J. Adams,et al.
Selection and scheduling of rehabilitation alternatives for water distribution systems
,
1998
.
[13]
Dragan Savic,et al.
WATER NETWORK REHABILITATION WITH STRUCTURED MESSY GENETIC ALGORITHM
,
1997
.