Dynamic multivariate loss and risk assessment of process facilities

Dynamic risk assessments (DRA) are the next generation of risk estimation approaches that help to enable safer operations of complex process systems in changing environments. By incorporating new evidences from systems in the risk assessment process, the DRA techniques ensure estimation of current risk. This thesis investigates the existing knowledge and technological challenges associated with dynamic risk assessment and proposes new methods to improve effective implementation of DRA techniques. Risk is defined as the combination of three attributes: what can go wrong, how bad could it be, and how often might it happen. This research evaluates the limitations of the methodologies that have been developed to answer the latter two questions. Loss functions are used in this work to estimate and model operational loss in process facilities. The application of loss functions provides the following advantages: (i) the stochastic nature of losses is taken into account; and (ii) the estimation of the operational loss in process facilities due to the deviation of key process characteristics (KPC) is conducted. Models to estimate reputational loss and significant elements of business interruption loss, which are usually ignored in the literature, are also provided. This research also presents a methodology to develop multivariate loss functions to measure the operational loss of multivariate process systems. For this purpose, copula functions are used to link the univariate loss functions and develop the multivariate loss functions. Copula functions are also used to address the existing challenge of loss aggregation for multiple-loss scenarios. Regarding the dynamic estimation of the probability of abnormal events, the Bayesian Network (BN) has usually been used in the literature. However, integrated safety analysis of hazardous process facilities calls for an understanding of both stochastic and topological dependencies, going beyond traditional BN analysis to study cause-effect relationships among major risk factors. This work presents a novel model based on the Copula Bayesian Network (CBN) for multivariate safety analysis of process systems, which addresses the main shortcomings of traditional BNs. The proposed CBN model offers great flexibility in probabilistic analysis of individual risk factors while considering their uncertainty and complex stochastic dependence. The research outcomes provide advanced methods for critical operations, such as the offshore operations in harsh environments, to be used in continuous improvement of processes and real-time risk estimation. Application of the proposed dynamic risk assessment framework, along with a proper safety culture, enhances the day-to-day risk-informed decision making process by constantly monitoring, evaluating and improving the process safety performance