Analysis of piping at hydraulic structures through an event tree approach

Abstract

The Netherlands has always been threatened by water. To protect flood-prone areas flood defenses have been constructed. To guarantee that they keep meeting the requirements systematic assessments take place. Since 2017, these assessments are performed according to the ‘Wettelijke beoordelingsinstrumentarium’ resulting in a safety judgment category (I_V, which is the highest safety category, to VII_V, the lowest safety category). The category depends on the difference between the norm and the actual failure probability of the assessed dike trajectory. To determine this failure probability each applicable failure mechanism is assessed. However, the assessment of piping at hydraulic structures does not result in a failure probability. The objective of this thesis is to find a method that can be used to obtain a failure probability for this failure mode. The current assessment strategy for piping at hydraulic structures is based on a comparison between the critical difference in hydraulic head and the occurring difference in hydraulic head over the structure. This strategy has only two possible (non-probabilistic) assessment outcomes, II_V and V_V (instead of the entire range of outcomes for probabilistic analyses). Currently, flood defense assessment is mainly based on the application of fault trees. Another possible method is to use event trees to quantify the probability of failure due to piping. Both methods clarify and relate all sub-events before failure happens, both qualitative and quantitative. The event tree method is chosen because an event tree can describe multiple consequences and is not limited to Boolean operators. To test the probabilistic event tree method a case study of the Marksluis chamber lock was conducted. In the event tree approach the piping failure mode is divided in three main phases: A) geotechnical part, B) emergency response and C) remaining strength of the structure. Three different event tree variants have been designed. Variant 1 (shown in the picture) is rather extensive and serves solely qualitative purposes while variant 2 and 3 are more concise and have quantification of the failure probability as goal. The geotechnical part is quantified using Monte Carlo and FORM analyses. However, the other two phases cannot be quantified with a model. The remaining strength of the structure is quantified by expert knowledge and it can be concluded that the upper limit of the conditional probabilities of structural failure and breach formation is 0.1. For the Marksluis the difference between the norm and the actual probability of failure due to piping is more than a factor 30. This leads to assessment category I_V. By using an event tree approach the probability of failure of different types of hydraulic structures due to piping can be quantified. This quantification is based on both models and expert knowledge. Adjustment of the WBI 2017 assessment procedure is useful for failure modes that cannot be assessed in a probabilistic manner yet. By implementing an event tree approach it becomes easier to think of what sub-events must be assessed and how this can be done in a probabilistic manner.