Search

Water mains are the essential component of a water supply system (WSS) buried underground, and they account for 80% of total system expenditures. Deterioration of this asset, as a result of an aggressive soil environment, is inevitable over a pipe's service life. Although an increasing number of studies have estimated the remaining service life (RSL) of metallic pipes, most of them either are data intensive or consider limited soil and pipe parameters. In this paper, a Bayesian belief network (BBN) model is proposed to handle the problem of varying data availability and the dependency between parameters. First the proposed approach uses a combination of empirical data, experimental data, expert opinion, and a mathematical model to predict soil corrosivity and pit depth. Then a simple programming logic is used to predict RSL. Finally, the performance of the model is evaluated using a BBN sensitivity analysis. Monte Carlo simulations (MCSs) are performed using randomly generated input from measured statistical parameters (i.e., mean and standard deviation) to indicate the effect of input parameters on metallic pipe RSL and safety index (SI). [ABSTRACT FROM AUTHOR]