Procedures to model and solve probabilistic dynamic system problems

Probabilistic Safety Assessment (PSA), characterized by process-behaviours modelling and event likelihood calculation, has great importance for quantitative risk evaluation. PSA presents some difficulties for implementation, mainly when the analysis of a dynamic process is required. In this work, a set of procedures to formulate and solve Probabilistic Dynamic System Problems (PDSPs) is presented. Such procedures explain how events should be modelled and connected with each other to build a process model that makes it possible to answer two main questions: (i) What is the discrete probability of occurrence of a specific process event? And, given its occurrence (ii) What is the distribution of event time to occurrence? After answering these questions, the event-occurrence probability in a specific length of time, which is the main goal of PSA, is easily calculated. To explain this proposal, two PDSPs are solved: the pressure change in a vessel caused by failure of two valves and the change in holdup tank level caused by failure of two pumps and one valve.