Risk analyses for a water supply system: Occoquan Reservoir, Fairfax and Prince William Counties, Virginia, USA / Des analyses aléatoires d'un système d'approvisionnement en eau: Occoquan Reservoir, Fairfax et Prince William Counties, Virginia, USA

Abstract

This report demonstrates two techniques for evaluating risks in the operation of a water supply system. Both rely on reconstructed historical streamflow data to develop estimates of the probabilities of certain specific events occurring in the future. These techniques are applied to the Occoquan Reservoir which was experiencing an unprecedented low level of storage in the autumn of 1977. The two techniques are used respectively to evaluate the overall adequacy of the existing reservoir and to evaluate the risks in the 1977 crisis. In the first technique, the general risk analysis model (GRAM), simulations of the reservoir's contents are carried out under a set of assumptions about withdrawal rates and emergency procedures. The results of the GRAM simulation for the Occoquan Reservoir are in the form of estimates of the probabilities that in any year contain emergency procedures will have to be invoked. These estimates are given for a range of rates of withdrawals and for four different stages of emergency actions. Also given are the estimated probabilities of entering emergency conditions in one year given that an emergency has occurred in the previous year. Due to the year-to-year persistence of low flows, these latter (conditional) probabilities are higher than the former (marginal) probabilities. The second technique used is position analysis. In this procedure probability distributions of future storages are estimated under existing storage conditions and an assumed rate of withdrawal from the reservoir. The position analysis which was initialized at the 1 October 1977 conditions indicates that the probability of entering a Stage III emergency (the prohibition of all uses of water non-essential to life, health, and safety) in the autumn of 1977 or winter of 1978 was 10 per cent at that time. With a reduction in water use by 8 million gallons per day, however, this probability would have fallen to 4 per cent. If a long reconstructed historical streamflow record is available to a water supply agency, then the agency will have the capability to undertake its own risk analyses. It can carry out comparisons of alternative operating policies by using techniques such as GRAM. It can also evaluate the short-term risks of different plans of operation during crisis situations by using position analysis.