Research · 07
Probabilistic Safety Assessment for Advanced Reactors
차세대 원자로 확률론적 안전성 평가
For reactor types that have no established framework yet, such as the molten salt reactor, we work out which accident paths exist and how often they can occur. Rather than stopping at counting branches, we run the analysis codes themselves so that timing comes with the answer.
Card figures are concept diagrams.
Probabilistic safety assessment (PSA) uses event trees and fault trees to work out which combinations of failures lead to an accident, and answers in numbers such as core damage frequency. Light water reactors have decades of established procedure and failure-rate data behind them; the molten salt reactor does not, because the fuel is also the coolant and accidents therefore unfold differently.
We build PSA models for the molten salt reactor from the ground up: defining the initiating events and accident sequences particular to this reactor type, and treating internal events such as component failures and human error alongside external events such as earthquakes, flooding, and fire.
Then we put time back in. An event tree only splits on whether a safety function succeeded or failed; it carries no record of what happened when, or in which order. In a reactor whose fuel is its own coolant and moves on its own, that order changes the outcome. So we are extending the work into dynamic probabilistic safety assessment (DPSA): running the system analysis codes themselves thousands of times over varied conditions, letting each accident actually play out, and quantifying risk from the resulting population of runs. RAVEN drives those repeated runs and turns them back into probabilities.
How the plant responds as a whole is drawn as a state-based model. Each component, system, and operator is represented by the state it is currently in, and the conditions that move it to another state are linked together (EMRALD). Drawn this way, you can see where an accident sequence branches and how much time is left to act before it does. Timing of the sequence is what this part of the work is for.
We are also widening the target. A floating molten salt reactor, moored at sea, does not share the conditions of a land-based plant: the hull rolls and heels, and being away from shore changes how long outside help takes to arrive. We are building a separate PSA model that accounts for this.
All three of these are work in progress; none has published results yet.
What we work on
- 01Defining the initiating events specific to a molten salt reactor and building the event trees
- 02Internal event analysis covering component failures and human error
- 03Assessing how external events such as earthquakes and flooding affect safety functions
- 04Simulation-based risk quantification through repeated code runs — dynamic PSA (RAVEN)
- 05State-based modeling of the integrated plant response and the timing of accident sequences (EMRALD)
- 06A PSA model for the floating molten salt reactor