Integrated risk analysis from source to tap: Case study Göteborg

Other conference contribution, 2008

To achieve an efficient risk management of a drinking-water system the entire system has to be considered, from source to tap. An important part of risk management is to identify hazards and estimate risks, i.e. to conduct risk analyses. In order to provide a relevant basis for evaluating risks and efficiently prioritising risk reduction options, a risk analysis needs to properly consider interaction between different parts and components of the system. This is especially important in complex systems. Logic tree models have the capability of properly reflect system functionality as well as facilitating quantification of risk levels. A fault-tree model was therefore constructed for an integrated and probabilistic risk analysis of the drinking-water system in Göteborg, Sweden. The main (top) event studied in the analysis was supply failure, which included quantity and quality failures. Quantity failure occurs when no water is delivered to the consumer and quality failure when water is delivered, but unfit for human consumption according to existing water-quality standards. Hard data and expert judgements were used for estimating probabilities of events, consequences and uncertainties of estimates. Monte Carlo simulations were used for the calculations in order to facilitate uncertainty analysis of risk levels. The risk analysis provided information on the probability of failure, rate of failure and mean down time of the system. The number of people affected was also included in the fault tree and risk levels were expressed as Costumer Minutes Lost. The primary aims of this paper were to apply a fault-tree method, for integrated and probabilistic risk analysis of drinking-water systems, on the system in Göteborg and show how the results can be used. The results showed, for example, that the raw water part contributes most to the total risk level and that the distribution part includes frequent failures that most often have a short duration and affect a small number of people. The method was found to facilitate a quantitative and integrated risk analysis of the drinking-water system and the results provide information not only on risk levels, but also on the dynamic behaviour of the system. In addition, the method is capable of relevant handling interaction of system components. Furthermore, it provides transparency and facilitates for formal updating when new information becomes available. Hence, it is concluded that the method provides useful information for discussing and evaluating risks as well as possible risk reduction options.