Probabilistic deterioration modelling and time-dependent reliability analysis of coastal defences

The prediction of coastal flood defence performance deterioration in the future plays an essential role in the reliability analysis and management process of these structures. The climate change effects and sea level rise will increase the hydraulic loads and frequency of the future extreme events, which lead to a more challenging performance deterioration prediction. The main failure mechanisms in coastal flood defences, e.g. wave overtopping and piping will be affected due to the change in hydraulic parameters and deterioration processes, which may decrease the reliability of the structures. Also, the uncertainties arising from the mentioned problems lead to a more expensive and inefficient management strategy to protect the lands, people and properties against floods. Hence, continuous innovations in flood asset management and structural reliability analyses methods are necessary to improve the accuracy and efficiency of the future performance evaluation in a changeable environment, and then to decrease the maintenance and management costs. In practice, parameters of probabilistic deterioration models can be estimated using available data from routine inspections and observations. For coastal defences, the condition assessment results and observations are collected using different inspection strategies, and deterministic grade-based deterioration curves are available in order to estimate structures residual life with respect to the structural conditions. However, probabilistic approaches to model the stochastic deterioration process with consideration of the changes in hydraulic loading parameters are not studied yet in coastal flood defences. This thesis proposes probabilistic state-based deterioration models and time-dependent reliability analyses for coastal flood defences, which help to predict the future performance deterioration and condition grades of the assets. The proposed deterioration models have four components: 1) using homogeneous (i.e. stationary) and non-homogenous (i.e. time-dependent and non-stationary) Markovian models to forecast a coastal defence deterioration process, 2) employing stochastic theory and simulation techniques to establish the Markovian transition probability matrices for the assets, 3) utilising performance-based reliability models to predict the reliability of the structures regarding the predicted deteriorations and projected changes in hydraulic variables, and 4) utilising renewal replacement theory, and sequential decision-making model to select optimal multi-objective maintenance actions under partial information. The deterioration models were established through a process of conversion from deterministic data and conventional framework in coastal flood defences to a probabilistic system. The process of the system conversion is described, and the validation of the conversion is demonstrated by a number of numerical examples. The elements of the time-dependent transition probability matrices are calculated using Weibull distributed waiting times and non-linear optimisation techniques. The limit state equations for the reliability-based model are updated regarding the sea level rise and deterioration models, and the future performance is predicted using time-dependent reliability analyses. The validation and efficiency of the time-dependent reliability analysis are demonstrated by case studies. A Partially Observable Markov Decision Process is utilised to provide the optimal maintenance strategies for each time interval. Cost of imperfect information is adopted to control the optimisation process at each time-step based on the stochastic deterioration model. The obtained results from the proposed method are examined by experimental and field data available. The applicability of the method is demonstrated by numerical examples, and the results show that the proposed methodology is capable of assessing the structural performance in the future and also can provide multi-objective optimised repair and inspection schedule during the lifetime of coastal defence structures. The knowledge gained in this study contributes to the better understanding of the performance deterioration of flood defences. Furthermore, the methodology presented in this study could be helpful in assessing the actual state of deterioration and to decrease the cost of management for flood defences