Use of Monte Carlo Simulation in Long-Term Capital Planning of Rehabilitation of Water Distribution Networks

Long-term capital planning of intervention programs for water distribution networks involves the selection of pipe for intervention by applying the appropriate strategies such as replacement, lining, or cathodic protection as examples. A proposed program consists of both the selection of pipe and the time at which the interventions are to be performed. The assessment of the costs and benefits of long-term programs necessarily require a prediction model to forecast water main break rates into the future. However, in proposing a rehabilitation scenario the use of predicted failure rates can be problematic in the immediate short-term. There is always some variation expected between predicted failure rates and observed failure rates. Failure history rather than predicted failure rates is the preferred method for selecting pipes for intervention in the short-term. Replacing a pipe that has not failed cannot be justified on the grounds that it is predicted to fail. Therefore, long-term planning models should transition from pipe selection based on past failure history for the immediate present to selection based on predicted failure rates in the distant future. This paper describes how Monte Carlo simulation can be used to shift from short-term, history-based modeling to long-term, prediction-based modeling in a single planning scenario. The method is demonstrated using examples from the City of Hamilton, Ontario.