Evaluating the safety of large dams by means of an average value of failures taken over a lang time-span of dam history leads to inadequate results, since it eliminates most of the declining trend. It appears that an analysis that evaluates the trend as such is required to obtain a proper assessment of the safety performance as a basis for failure prognosis and thereby for decision making an investments. Trends appear in dependencies an three different variables: yearof failure, year of construction,and age. Two different methods are presented to analyze these trends: one method for the trend analysis of the fatlure occurrence in time for a dam population, and a secend method for the combined analysis of trends in the construction year and age dependencies. The first method is applied to large dams in the U.S.A. and in Western Europe built after 1850, combined as well as for individual dam types, whereas the secend method is only applied to U.S. embankment dams because of its more extensive data requirements.
展开▼