Analysis of the initial Zeus critical condition with MCNP and ENDF/B-VI

The Zeus experiment was designed to test the adequacy of 235U cross sections in the intermediate-energy range. For its initial configuration, the core contains thin, circular platters of highly enriched uranium (HEU) separated by similar platters of graphite. The cylindrical core is reflected by copper pieces shaped to produce a parallelepiped. Zeus achieved initial criticality on April 26, with ten HEU platters and an axial loading that is almost symmetrical. The critical configuration had 4 cm of graphite above the top HEU platter, 4 cm of graphite below the bottom HEU platter, and 8 cm of graphite between all but two adjacent HEU platters. The fourth and fifth platters, counting from the bottom, were separated by 6.5 cm of graphite and 60 mils (0.1524 cm) of aluminum. The aluminum shim was required because Zeus currently has an operating limit of 10cts of excess reactivity. A schematic of this configuration is shown. This configuration actually was very slightly supercritical, with a period of approximately1,100 s. This period corresponds to approximately1cents of excess reactivity and therefore to a value of keff between 1.0000 and 1.0001. For practical applications, therefore, the configuration can be considered to be exactly critical. A detailed MCNP model of the critical configuration was constructed, as were models for two slightly subcritical configurations. Prior to achieving criticality, measurements had been made for two very similar but slightly subcritical configurations. The first subcritical configuration had 8 cm of graphite between all adjacent HEU platters, while the second had only 7 cm between the fourth and fifth platters (but 8 cm between all others). Calculations were performed with MCNP4XS, a version of the MCNP Monte Carlo code that is intermediate between MCNP4B and the forthcoming MCNP4C, and a cross-section library derived from release 4 of ENDF/B-VI. These results show the incremental increase in reactivity from one configuration to the next, and the reactivity changes correspond quite well with the estimated subcritical margins. However, there is a bias of approximately0.01 Δk in the calculated values for keff