Pile-supported ports allow seawater to run below the deck, and thus may suffer structural damages during extreme coastal events such as hurricanes. These structural damages, in turn, may result to port closures that can cause significant economic losses. Risk analysis can predict the post-hazard functionality of ports though the structural damage assessment of these structures prior to coastal events. However, assumptions on the selected demand estimates may affect the estimated probability of structural damage. This research aims to shed light on the sensitivity of the wave model selection for the risk assessment of pile-supported ports when subjected to storm surge and waves. The examined structural damage is the uplift of the deck, and the risk assessment is conducted through the development of fragility curves for a typical deck-pile connection, for which fragility curves are developed for different wave models. Uncertainties are also considered in parameters affecting the demand and capacity of the examined deck-pile connection and are propagated through the Monte Carlo simulation using the Latin Hypercube Sampling. The results indicate changes to the uplift probability as a result of the selected wave model. Thus, wave model selection can alter the uplift failure probability. In addition, the study proposes parameterized fragility models to enable the uplift risk assessment across a region. The presented results aim to throw light on the proper model selection to produce more realistic risk assessment estimates towards the resilience of coastal infrastructure. / Thesis / Master of Applied Science (MASc)

Subject of the dissertation is a probability analysis on creation of embankment subsoil filtration deformations. Out of the broad topic of filtration deformations, the dissertation is focused at initiation of the process. At this phase, the overlying layers downstream of levee get uplifted, and eventually get breached. The process of downstream subsoil failure probability estimation is formulated using mathematical and statistical modelling. Within the scope of parametrical study the suggested method is applied onto a test environment. Results of the probability estimation are compared with results obtained using Limit state design method. The results of stochastic modelling can be used for improvement of Limit state design method precision.