5 Conclusions
Ten different metrics that could be potentially used to characterise the resilience of a water resources management system to uncertain future changes (i.e. a given scenario of supply and demand) were investigated. An in-depth analysis of the metrics was carried out on a real-world case study of Bristol Water’s supply system, including an examination of metric sensitivity and correlation, and a detailed examination of the behaviour of water deficit periods. The results obtained lead to the following key recommendations for the selection of an appropriate resiliencebased performance metric:
1. Simultaneous use of multiple metrics covering different aspects of resilience and related water deficit events is recommended. This does not seem to be the case currently in the literature (where typically a single metric is used to assess water system’s resilience) nor in engineering practice (where quantitative resilience metrics are not used much, if at all).
2. Metric M2 (“the duration of longest water deficit period” metric) was observed to be one of the more informative and comprehensive performance metrics followed closely by metric M7 (“the water deficit of greatest magnitude recorded” metric). However, the analysis demonstrated a relatively high correlation between the two metrics and therefore considering just a single metric may prove sufficient to evaluate the resilience of a water resource system. A duration based metric would be a more logical assessment metric to use of the two types, as it is the duration of temporary water restrictions that most impact on customers and supply, whereas the magnitude of water deficit events is of less direct concern to customers and water companies so long as the magnitude is maintained within acceptable threshold levels. All this seems consistent with the resilience metric suggested by Hashimoto et al. (1982) now supported with actual evidence.
