Neutronics Experiments and Analysis Related to Strong Moderation Heterogeneity in LWRs

The Supercritical-Water-Cooled Reactor (SCWR) is the Generation IV reactor concept most closely related to current light water reactors (LWRs). The SCWR builds on the vast experience with today's LWRs and supercritical coal-fired power plants. Water at supercritical state is used as moderator and coolant, and reaches a temperature of about 500 °C at the core outlet, which enables a much higher thermal efficiency (∼ 44%) than possible with current-day LWRs. In the operating range of an SCWR, the supercrical water density can become as low as one seventh of the density of water at room temperature. In order to ensure a thermal neutron spectrum, large moderator regions are introduced into the fuel assembly designs, resulting in lattices with strong moderation heterogeneity. As such, the neutronics of these assemblies differs significantly from that of standard LWRs, and lies outside the validated domain of reactor physics codes. Previous comparisons between the deterministic code CASMO-4 and reference calculations carried out with the Monte Carlo code MCNP4C showed large differences in calculated pin-wise reaction rate distributions in perturbed SCWR lattices. The experimental validation of standard reactor physics codes for SCWR-representative neutronics conditions is thus clearly of key importance for the further development of this technology. The goal of this thesis is to provide an experimental database for such validation and to assess the level of performance of current-day codes for SCWR analysis. An SCWR-like fuel lattice, based on a Japanese assembly design proposal from 2001, has been investigated in the PROTEUS zero-power research reactor at the Paul Scherrer Institute. Measurements have been carried out on the unperturbed test lattice, as also on six other PROTEUS configurations corresponding to different types of perturbations of the reference lattice. The investigated perturbations include control rod related effects, moderator density changes, and the replacement of a fuel pin with gadolinium-poisoned fuel. For each experimental configuration, pin-wise distributions of the total fission rate (Ftot) and the 238U capture rate (C8), as well as of their ratio (C8/Ftot), have been obtained across the assembly and compared to computed values. The neutronics codes used for the calculations are the LWR assembly code CASMO-4E and the Monte Carlo code MCNPX. Additionally, the reactivity effects of removing individual pins from the unperturbed SCWR-like lattice have been measured and compared to predictions obtained using the two codes. The pin-wise reaction rate distributions predicted with MCNPX have been found to agree within two standard deviations with the measured values for the unperturbed, as well as all the perturbed, configurations. The 1σ uncertainty was in the order of 0.4%, 0.8%, and 2.2% for Ftot, C8, and C8/Ftot, respectively. MCNPX could thus be validated for the various SCWR-like conditions investigated and has, in turn, been used to validate CASMO-4E on simplified geometries. For the latter purpose, the unperturbed and perturbed SCWR-like lattices were modeled as reflected assemblies, and reaction rate distributions predicted with the two codes were compared. For the unperturbed lattice and for the configurations with local perturbations of the neutron absorption, the agreement between the codes was within ∼ 1% for all reaction rates. However, for lattices with locally perturbed moderation conditions, CASMO-4E was initially found to overestimate the reaction rates in the vicinity of the perturbations by up to 5%. The discrepancies were identified as resulting from the default leakage treatment in CASMO-4E, which applies a DB2 correction in a homogenized sense across the lattice. In this way, cases with a global leakage gradient were not being treated properly. Usage of the optional input card BZ2, which allows a region-wise leakage treatment, resolved this problem, and the codes then agreed mostly within 1% for all the configurations. The pin removal worth measurements in the unperturbed SCWR-like lattice have provided integral data complementary to the reaction rate distributions. MCNPX results were found to agree with experiment within the statistical uncertainty of typically ∼ 10% (1σ). The comparison of reflected-assembly results from MCNPX and CASMO-4E yielded significant differences in the calculated pin removal worths (up to 4.3σ). A decomposition analysis of these differences has indicated that the discrepancies, especially for the better moderated pin positions, are linked to the calculation of the additional moderator effect caused by the pin removal. Finally, the transferability of the code validation, carried out under the PROTEUS experimental conditions, to the most recently proposed SCWR assembly designs has been assessed. Since, under power-reactor conditions, the moderator and coolant water densities are considerably lower in the proposed SCWR assemblies than that in the PROTEUS test lattices, their neutron spectra are much harder. Assembly-averaged values of the integral parameters C8/Ftot and F8/Ftot have been found to be as much as 65% and 80% higher, respectively, than for the PROTEUS reference lattice. However, the agreement between CASMO-4E and MCNPX predictions for the proposed SCWR assembly designs is still very good, indicating that the accuracy of the deterministic calculations does not deteriorate markedly when considering these new designs under power-reactor conditions.