The number of spent nuclear fuel assemblies taken from nuclear power plants and to be stored in existing storage pools is increasing. Therefore, there is a need to optimize the storage configurations. The computer codes and cross sections used to analyze proposed storage configurations must be validated through comparison with experimental data. Restrictive values of ksafe, caused by limited data, can prevent optimal storage utilization. As a collaborative effort between Westinghouse Safety Management Solutions, Oak Ridge National Laboratory (ORNL), Georgia Institute of Technology, and the University of Missouri Research Reactor (MURR), more than 120 experiments were performed using four highly enriched MURR fuel assemblies. The 252Cf-source-driven noise analysis technique developed at ORNL was used as the measurement method for these experiments. This method is based on calculating a specific ratio of measured auto-power and cross-power spectral densities. Twenty-two unique configurations from the MURR experimental program were analyzed for benchmarking purposes.

These subcritical experiments were described and analyzed in this paper to provide new measurements to increase the amount of data available for benchmarking criticality codes and cross sections for systems that are far from critical (keff < 0.9).

All aspects of the experimental apparatus designed for the experiment program are thoroughly described to enable calculational modeling. Measured and calculated results for the 22 configurations of interest are given. Thorough perturbation studies on measurement uncertainties (e.g., fuel spacing and composition) were performed to determine the uncertainty on the ratio and keff values. Inferred keff values ranged from 0.648 ± 0.005 to 0.860 ± 0.006. A simplified benchmark model is described that consists of the four fuel assemblies, four 3He detectors, detector drywells, and the water reflector. For these measurements, the calculated ratio and keff values agreed with the measurement results within the measurement uncertainty. All of the analyzed configurations were considered acceptable for validation of computer codes and cross sections.