Approaching Dynamic PSA within CANDU 6 NPP

The outline of this dissertation is going to present the applications that are the subject of the work and also the lay down of work content. Chapter 1 reviews the conventional PSA main concepts, summarizes a short introduction history of Dynamic PSA (DPSA) and presents a non-exhaustive DPSA state-of-the-art with the recent and future developments. Chapter 2 presents the first application of the thesis, which is actually an introduction in the context of the Integrated Dynamic Decision Analysis (IDDA) code, that represents the main tool used in the attempt of approaching the Dynamic PSA. Starting from a description that reflects the level of knowledge about the system, IDDA code is able to develop all the scenarios of events compatible with the description received, from both points of view: either logical construction, or probabilistic coherence. By describing the system configuration and operation in a logically consistent manner, all the information is worked out by the code and is made available to the analyst as results in terms of system unavailability, minimal cut sets, uncertainty associated. The code allows also the association of different consequences that could be of interest for the analyst. The consequences could be of any type, such as economical, equipment outage time, etc.; for instance it can be considered an outage time for certain components of the system and then is calculated the “expected risk”. The association of consequences provides the inputs for a good decision making process. Chapter 3 represents the core applications of the present work. The applications purpose is the coupling between the logic probabilistics of the system or plant and associated phenomenology of primary heat transport system of a generic CANDU 6 NPP. First application is the coupling between the logic-probabilistic model of EWS system and associated phenomenology of primary heat transport system of CANDU 6 NPP. The considered plant transient is the total Loss of Main Feed-water with or without the coincident failure of the Emergency Water Supply System. The second application is considering the CANDU 6 Station Blackout as plant transient-consequential condition, moreover the loss of all AC power sources existing on the site. The transient scenarios development consider the possibility to recover the offsite grid and the use of mobile diesel generators in order to mitigate the accident consequences. The purpose is to challenge the plant design and response and to check if the plant conditions of a severe accident are reached. The plant response is challenged for short and long periods of time. The IDDA code allows interfacing the logic-probabilistic model of the system with the plant response in time, therefore with the evolution in time of the plant process variables. This allows raising sequences of possible events related in cause-consequence reasoning, each one giving place to a scenario with its development and its consequences. Therefore this allows acquiring the knowledge not only of which sequences of events are taking place, but also of the real environment in which they are taking place. Associating the system sequences that lead to system unavailability on demand with the resulting phenomenology proves to be a useful tool for the decision making process, both in the design phase and for the entire power plant life time. Chapter 4 presents future possible applications that could be developed with the present Dynamic PSA approach. A particular application could be the optimization or development of robust plant emergency operating procedures. In fact it consists in the coupling between the logic-probabilistics of the plant configurations corresponding to the Emergency Operating Procedure (EOP) and the associated phenomenology of the primary heat transport systems with the consideration for the plant safety systems. The application could highlight those situations where the plant fails either because of hardware failures or system dynamics and furthermore to reveal those situations where changing of the hardware states brings the process variables of the system state out of the system domain. A timeline course should be created for the process variables characterizing the plant state and that should reveal the time windows that operators have at disposition for intervention, in order to avoid potentially catastrophic conditions. Some week points in the EOP could be identified and then resolutions to be provided for their improvement, on the basis of sensitivity analyses. Chapter 5 presents the conclusions and the insights of the work and outlines possible improvements in terms of the present methodology proposed.