Safety Approach of BORAX Type Accidents in French Research Reactors

Most of pool type French research reactors are designed to withstand an explosive BORAX accident, defined as a pressure load on the pool walls. The purpose of this paper is to present the approach implemented at IRSN to analyse this accident by linking safety assessment and supporting studies. Examples of recent work on Jules Horowitz Reactor (JHR) and ORPHEE will be presented. Although all aspects of the accident are addressed, we will focus on the first two frames of the transient: the reactivity insertion and the consequences on the core. The first step of the BORAX analysis is to identify the most penalizing plausible reactivity insertion. This means characterising the sequences of events that can induce a reactivity surge and evaluate the worth of such variation. Neutronic computations are then required to quantify the reactivity increase. To comply with the geometrical specificities of research reactors, IRSN chose to use the homemade Monte Carlo code MORET5. The control rod worth calculations on the JHR were in good agreement with the operator results, whereas in ORPHEE, IRSN demonstrated that the beam channels reactivity worth was largely. In both cases the obtained results allowed an interesting dialogue with the operator and were used in the conclusions of the safety assessment. Following the accidental sequence of events, the second stage analysed by IRSN is the power transient occurring in the core and the consequences on the fuel. IRSN applied on JHR a homemade simplified model based on point kinetics and standard thermal balance equations to compute power evolution taking into account the temperatures of the fuel for feedback reactivity. As heat exchange coefficients between cladding and water for such fast transients are unknown, IRSN took the conservative hypothesis of adiabatic heating of the plates. The comparison the JHR power pulse calculation results against SPERT experimental measurements enabled IRSN to be optimistic about the possibility that a slow reactivity insertion would not lead to severe consequences on the core. It also highlighted a lack of knowledge about fast transient physical processes and the need of validated tools if a refined simulation is to be carried out. (author)