Prototype probabilistic models for WTI2017

The Dutch primary flood defences are periodically tested against statutory safety standards. Currently, policymakers are contemplating a move towards safety standards defined in terms of maximum allowable probalilities of flooding. To facilitate such a move, a new probabilistic framework, Hydra-Ring, for assessing the safety of flood defences is being developed in the project WTI 2017. In order to concentrate the development of probabilistic computation techniques and knowledge on uncertainties within the WTI-project, a new sub-project (“cluster’) was initiated in 2013. In order for the client to monitor the progress in probabilistic modelling, it was agreed upon that cluster C would provide the current report which summarizes the progress of the first 9 month of 2013. The activities with respect to probabilistic modelling can be divided into two categories: Improvement on generic probabilistic computation techniques and implementation of prototype probabilistic models for various failure mechanisms. The work described in this report is inevitably “work-in-progress”. Furthermore, the prototype probabilistic models so far could not make use of the failure mechanisms models that will be used in WTI 2017, as these models are still being developed by other WTI0clusters. Therefore, predecessors or simplified failure models were used instead. In spite of these two disclaimers, significant progress was made. First of all on “system reliability techniques”, i.e. techniques that compute failure probabilities of a whole system of flood defences. For instance, a technique was developed that allows the use of smaller dike sections in failure computation, which is very beneficial for areas without large statistically homogeneous dike stretches. Furthermore, an improved version of the “Hohenbichler” method was proposed, which plays a role in the system reliability analysis of Hydra-Ring. Both improved methods on system reliability were implemented and successfully tested. Significant progress was also made on the implementation of prototype probabilistic models for failure mechanisms “instability of asphalt revetment” and “instability of block revetment” , and “piping”. These models ran successfully and provided reliable results, even in combination with the fast but potentially unstable probabilistic computation method “FORM”.