PREDICTING WATER-MAIN BREAKAGE RATES

A methodology for quantifying the variation in pipe failure rates associated with temporal and spatial clustering of water-main failures is developed. In the first stage a cross-referencing scheme is used to determine the mean number of failures that occur subsequent to initial pipe breaks and the distribution of numbers of subsequent failures for specified intervals in time and space around these initial breaks. The observed distribution of numbers of subsequent failures is compared with the distribution provided by the Poisson equation based on the mean number of subsequent failures. The second stage uses nonlinear regression to derive the values for coefficients of an equation that captures the changes in the mean number of subsequent failures with variation in the time and space intervals. The equation thus obtained through regression is used to further derive equations for the variation in the failure rate as a function of time or space. The methodology is demonstrated by application to the water-main pipe failure data base of the city of Winnipeg, Canada.