Speeding up stochastic analysis of bulk water supply systems using a compression heuristic

Abstract

It is possible to analyse the reliability of municipal storage tanks through stochastic analysis, in which the user demand, fire water demand and pipe failures are simulated using Monte Carlo analysis. While this technique could in principle be used to find the optimal size of a municipal storage tank, in practice the high computational cost of stochastic analyses made this impractical. The purpose of this study was to develop a compression heuristic technique to speed up the stochastic analysis simulations. The compression heuristic uses a pre-run to characterise the failure behaviour of a tank under demand-only conditions, and the stochastic simulations are then only run for periods in which fire demand or pipe failures affect the tank. The compression heuristic method was found to be accurate to within 5% of the full stochastic analysis method. The compression heuristic was also found to be faster than the full stochastic method when more than 27 systems were analysed, and thus allowed genetic algorithm optimisation to be practical by reducing the optimisation simulation time by 75%.

Keywords: stochastic analysis; Monte Carlo; optimisation; genetic algorithm; reliability