FLOOD08791 - Laserfiche WebLink

Home Browse Search

persons have the same minimum or zero Warning Time. Or there may be a nursing home where all residents have a high level of vulnerability. The suggested approach for estimating the risk to the person most at risk, is to identify the group most at risk and to then apply the D-M procedure as though that grou!, was the only Population at Risk. The ratio of estimated Loss of Life It> Population at Risk for that 1 group then provides the conditional probability of death to be used in computing the Individual Risk for ~ the person most at risk. In other words, the average risk over the most vulnerable group is taken to be the risk to the person most at risk. Of course, where there is special vulnerability combined with low Warning Time (for example, a nursing home with zero warning time) there may be a case to take conditional probability, given dam failure, as 1.0. o where there are several failure modes that are not mutually exclusive, the overall conditional probability of failure lies somewhere between the bounds of: the maximum conditional probability as lower bound the probability of the union of events, the several failure modes, as upper bound If any of the conditional probabilities are high (say greater than 0.01), the sum of the conditional probabilities is not sufficiently accurate as the union of events, and the correct formula mUSt be used. The union of events is to be computed on the conditional probabilities of failure (that is, before multiplying by the probability of the loading event). These factors need consideration in obtaining the overall Individual Risk. It is often not correct to simply sum the contributions to Individual Risk of each failure scenario. Example calculations will be given in the new guidelines. o expected value of life loss, "f. N", is a useful measure, especially for identifying the relative contributions of the various failure scenarios to the overall risk to life. However, ANCOLD does not favour the use of the overall expected loss of life as the single measure of Societal Risk The reason is that the total expected value of life loss hides the various life loss scenarios that can occur. The total expected value is a single number and there are an indefinite number of combinations of life loss scenarios that can produce that figure. In other words, the single number of total expected value of life loss tells a decision maker nothing of the range of life loss that could occur. In contrast, an "FIN" plot does give information about the many life loss scenarios that can occur. It is for this reason that ANCOLD prefers FIN plots as the main measure of Societal Risk, notwithstanding their acknowledged problems. It should be noted that the overall expected life loss cannot be overlaid onto FIN graphs and there is no means of comparative plotting of the two measures (expected value and FIN plots). Some incorrect overlays have been published in the literature. o It is the case that FIN curves are difficult to interpret for those who are not experienced in their use. Also the results can depend on the formulation of the risk problem in a given situation (Evans and Verlander, 1997), which means that FIN plots are not a totally satisfactory measure of risk. Finally, where several failure scenarios have the same "N" value, the FIN plot varies according to the order in which the scenarios are listed when computing the cumulative distribution function "F". Notwithstanding these problems, AJ'-lCOLD is satisfied that FIN plots are the most valuable and practical measure of Societal Risk that is presently available. The last of the problems mentioned can be overcome by combining the probabilities of all failure scenarios with the same "N" value, before computing "F". Societal Risk is to be taken as not meeting the criteria if any part of the FIN plot passes above the criterion line. If only a small portion passes above the line then only one or a few failure scenarios require risk reduction in order to meet the criterion. o the concept of cost-to-save-a-life is useful in evaluating remedial measures, It is the annualised cost of risk reduction measures divided by the reduction in expected life loss per annum that is achieved by those measures. This measure should be seen as a means of rating cost effectiveness of remedial action: It is not to be seen as placing a monetary value on human life. Thus where there are several options that reduce risk to a required level, that with the lowest cost-to-save-a-life would be preferred on grounds of cost effectiveness in reducing risk to life. This measure can also be used as a quantitative aid in applying the ALARP principle. Cost-to-save-a-life values typically cover a very wide range and are often very high. Figures published by the US Office of Management and Budget (1992 - Table C-2) show that US$30 to 50 million per life saved are moderate values.