Multi-Physics Simulations and Sensitivity Analysis of a Main Steam Line Break Transient in a Pressurized Water Reactor

This paper illustrates the first steps of the Best Estimate Plus Uncertainty methodology implemented for the study of a main steam line break (MSLB) accident in a four-loop pressurized water reactor. An MSLB is characterized by rapid depressurization of a steam generator, leading to uncontrolled cooling of the primary circuit and a core power increase because of moderator reactivity feedback. The phenomenology of this accident is multiphysics as it involves strong coupling between reactor physics and thermal hydraulics.

First, a Phenomena Identification and Ranking Table (PIRT) was established based on expert judgment to list all the main physical phenomena involved in the accident and rank them according to their influence on the physical quantity of interest: the minimum departure from nucleate boiling ratio (MDNBR). The PIRT analysis resulted in a list of 10 dominant phenomena and 10 uncertain parameters associated with such phenomena. Then, a one-at-a-time sensitivity analysis was performed for all the dominant phenomena identified in the PIRT. Based on such analysis, three phenomena emerged as the most influential: water mixing in the lower plenum, initial reactivity margin, and moderator reactivity feedback. The sensitivity analysis provided a better understanding of the accident sequence and allowed slightly revising and consolidating the PIRT. It also showed that the axial power profile plays an important role in determining the MDNBR value. Last, an uncertainty propagation study was realized on the MSLB transient based on the Monte Carlo method. Probability density functions (PDFs) of the uncertain parameters were estimated according to expert judgment. The study led to the evaluation of a lower bound of the quantity of interest, i.e. the MDNBR, by applying the Wilks method. Possibilities for improvement were also identified, and an inverse uncertainty quantification process based on measurement data is proposed to improve the estimation of PDFs associated with uncertain parameters.