The role of load variations in assessing credible dike failure probabilities

Abstract

Assessing dike safety is of key interest to societies in low-lying areas, but results can be implausible when, for example, they contradict the observed performance of the dike. To improve credibility, load monitoring data can be incorporated using reliability updating techniques. This paper investigated the role of load variations in reliability updating and assessing credible failure probabilities. It was found that the impact of reliability updating increases when load variations are small, as a large contribution to failure probabilities comes from relatively frequent load levels, of which the conditional failure probabilities are reduced most through reliability updating. Moreover, a credibility check was introduced for dikes that have been stable for decades, where load levels with return periods of up to 10 years are not expected to contribute more than 50% to the failure probability, indicating an imbalance between load variation and strength uncertainty. This imbalance occurs when the inverse gradient of the fragility curve exceeds 1.5 times the decimate height of the load. Many Dutch dikes, including canal dikes and dikes along the large lakes and delta regions, have small decimate heights. For these dikes, strength uncertainties must be sufficiently small to obtain credible failure probability estimates.