FLOOD08791 - Laserfiche WebLink

Home Browse Search

construction technology, Certainly the costs of bringing all existing dams into conformity with lower risk levels would likely be unrealistically high. . the Limit truncation is .a little below, and the' Objective truncation is significantly below, the risk levels corresponding to the traditional dam design standards for some types of dams. For example, the Probable Maximum Flood standard, in the case of earth dams, represents a probability of failure of something a little less (after allowing for the conditional probability of breaching at low overtopping depths) than lE- 05 to 1 E-07 per annum if there is no freeboard above peak flood level (Lauren son and Kuczera, 1998). Other Aspects of Revised Criteria There will be a shift in focus from average Individual Risk to the Individual Risk for the person (or more typically, the group) most at risk. There is a view that average Individual Risk should be abandoned as an acceptable risk criterion, but the Working Group is yet to come to a fmal position on this question. ESTIMATION AND PRESENTATION OF RISKS TO LIFE There is now an improved understanding of how risks to life should be estimated and presented. The detailed treatment of this topic will be provided in the revised guidelines document. At this stage the aim is to give a broad indication of the new approaches. The main areas of change in assessing and presenting risks to life are: . an improved understanding of the concept of dam failure scenario . new procedures for estimating loss of life . better understanding of the difficulties in estimating loss of life . a better understanding of how Individual Risk is to be computed . a clearer understanding of how Individual Risk to the person most at risk is to be assessed . clarification of the rules for combining probabilities when obtaining the overall Individual Risk . clarification of the place of expected value of life loss (product of failure probability, f, and loss of life, N) in assisting appreciation of risks to life . an improved understanding of the quirks of FIN curves, and of how to interpret these curves. . clarification of the role of cost-to-save-a-life in assessing risks to life . a much improved understanding of how to present Societal Risks Detailed guidance on estimation and presentation of risks will be fully set out in the updated guidelines now in preparation. Only a brief mention of the main points follows: . failure scenario refers to a suite of states that defines the circumstances of each failure considered in a risk assessment. Typical states making up a scenario are loading (flood or earthquake magnitude), prior storage, failure mode (the relevant event tree), breach location (for example, main or saddle dam), concurrent downstream and tributary streamflow, time of day, season of the year. In theory, a detailed risk assessment may generate some hundreds of scenarios. Again in theory, each scenario yields two values, that for "f' (probability of failure per annum) and that for "N" (predicted loss of life). The fact that a study has only a relative few "f,N" pairs as output would generally be due to aggregation of scenarios in order to limit the cost of the study. . the DeKay and McClelland (1993) procedure for estimation of loss of life should be regarded as better than the US Bureau of Reclamation (1989) procedure set out in Appendix D of the 1994 Guidelines, notwithstanding that the former raises many difficulties, The procedure of Assaf et al (1997) may be an improvement on the DeKay and McClelland (D-M) approach, but there is no real experience of its application in Australia as yet (August 1998). Clearly Assaf et al regard their procedure as an improvement on the D-M and earlier methods, . Dr Mike DeKay (1997) has computed confidence limits for the regression equations of the DeKay and McClelland procedure. These are shown in Table I and are very wide. This is not surprising on