Track 3: growth of nuclear technology and research numerical and computational aspects of the coupled three-dimensional core/plant simulations: organization for economic cooperation and development/U.S. nuclear regulatory commission pressurized water reactor main-steam-line-break benchmark-I. 6. CEA-IPSN Participation in the MSLB Benchmark

The OECD/NEA Main-Steam-Line-Break (MSLB) Benchmark lets us compare state-of-the-art and best-estimate models used to compute reactivity accidents.A comprehensive study has been carried out by CEA and IPSN with the CATHARE, CRONOS2, and FLICA4 codes to assess the three-dimensional (3-D) effects in the MSLB accident and to explain the return-to-power (RTP) occurrence. The three exercises of the MSLB benchmark are defined with the aim of analyzing the space and time effects in the core and their modeling with computational tools. Point kinetics (exercise 1) simulation results in an RTP after scram, whereas 3-D kinetics (exercises 2 and 3) does not display any RTP. Our objective is to understand the reasons for the conservative solution of point kinetics and to assess the benefits of best-estimate models. First, the core vessel mixing model is analyzed; second, sensitivity studies on point kinetics are compared to 3-D kinetics; third, the core thermal-hydraulics model and coupling with neutronics is presented; finally, RTP and a suitable model for MSLB are discussed. Modeling of the vessel mixing is identified as a major concern for an accurate computation of MSLB. On one hand, the RTP in exercise 1 is driven by the mixing between primary loops, and on the other hand, the hot assembly power in exercise 3 depends on the inlet temperature map at assembly level. Vessel mixing between primary loops is defined by the ratio of the hot-leg temperature difference over the cold-leg temperature difference. Specifications indicate a ratio of 50%. Sensitivity studies on this ratio were conducted with CATHARE and point kinetics. Full mixing of the primary loops leads to a sooner and higher RTP, while no mixing results in a later and weaker RTP. Indeed, the intact steam generator (SG) is used to cool down the broken SG when both loops are mixed in the vessel, and the primary temperature decreases faster. In the extreme case of no mixing, only one-half of the primary circuit is cooled down after isolation of the intact SG, which reduces the consequence of the accident. For exercise 3, no mixing between loops would maximize the local power (cooler inlet temperature for the hot assembly) but would minimize the global reactivity insertion. Unfortunately, realistic core boundary conditions were not derived because state-of-the-art thermal-hydraulic calculations do not provide a suitable and reliable turbulence model for the core vessel. Point kinetics is based on a fixed spatial power distribution and on neutronic parameters used to perform the reactivity, as a function of Doppler, moderator, and control rods. Since in an MSLB accident the reactivity insertion is caused by an overcooling of the reactor, the moderator reactivity coefficient has to be accurately defined. The moderator reactivity can be correlated either to moderator density or temperature. It turns out that density and temperature are not equivalent for MSLB because both primary temperature and pressure vary within wide ranges during the accident. A temperature coefficient for moderator reactivity results in a power raise at the beginning of the transient, whereas a density coefficient first results in a power drop. This qualitative difference modifies the whole transient because scram is triggered either by high power level or by low primary pressure. When performing the reactivity coefficient with CRONOS2 during exercise 3, we found a range of ∼18 to 31 pcm/kg.m-3 with an initial value of 27 pcm/kg.m-3. Some of this large variation is related to the 3-D effects taking place in the core. The power distribution is strongly affected in the axial and radial directions by the scram (assumption of one rod stuck-out) and by the asymmetry of the core inlet temperature. Sensitivity studies on point kinetics show that a reduction of the reactivity coefficient by 20% is enough to cancel the RTP. Modification of the power shape displayed by 3-D kinetics tends to lower the moderator reactivity, which is over-predicted by point kinetics based on core initial conditions. Exercise 2 is devoted to core thermal hydraulics and neutronics coupling. To assess the main state-of-the-art models based on 3-D kinetics, we coupled CRONOS2 with 3-D or one-dimensional (1-D) thermal hydraulics per assembly (FLICA4) and with 3-D coarse-mesh thermal hydraulics (CATHARE). Comparison of the three calculations shows a weak effect of transverse flows in the core because primary pumps are operated at nominal conditions during all the transient. On the other hand, collapsing assemblies in a coarse mesh lead to a loss of accuracy, particularly for local information such as maximum fuel temperature or power peak. Both thermal hydraulics and neutronics have to be computed at the same scale (assembly) to ensure consistency between power and feedback. The RTP in exercise 1 mainly depends on the accuracy of point kinetics parameters and also on the vessel mixing between primary loops. On the other hand, best-estimate calculations (exercise 3) display significant 3-D effects in the core, which affect the moderator reactivity during the transient. Such effects cannot be modeled by point kinetics and seem to explain the occurrence of RTP in exercise 1 because of conservative point kinetics parameters. A suitable best-estimate model for MSLB can be achieved by coupling the 1-D thermal hydraulics of primary and secondary circuits, taking special care of the vessel mixing, with the 3-D coupled core kinetics and thermal hydraulics per assembly.