[en] Experiments done in several LMR-type critical assemblies simulated core axial expansion, core radial expansion and bowing, coolant expansion, and control driveline expansion. For the most part new experimental techniques were developed to do these experiments. Calculations of the experiments basically used design-level methods, except when it was necessary to investigate complexities peculiar to the experiments. It was found that these feedback reactivities generally are overpredicted, but the predictions are within 30% of the experimental values. 14 refs., 2 figs., 4 tabs

[en] The probability that reactivity feedbacks fail to prevent damage is computed by propagating data and modeling uncertainties through transient calculations, with these uncertainties being constrained by experimental evidence. Screening processes are used to identify the most important parameters and accident initiators. The response surface method is used to facilitate the error propagation and a Monte Carlo rejection technique is used to force the parameter variations to be consistent with the observed distribution of experimental quantities. The reliability of the failure probability estimates is evaluated. This process is applied to ATWS events in the PRA for the EBR-II reactor. The loss-of-normal-power (LONP), loss-of-flow and transient overpower accidents without scram were found to warrant detailed analysis and a complete analysis has been made for the first of these. Six parameters are primarily responsible for the LONP outcome variations. The conditional probability of minor core damage from LONP without scram is 1.2 x 10^-2. The uncertainty in this estimate is a factor of 2. This damage estimate would be an order of magnitude higher if experimental information about feedbacks in EBR-II was not used. the conditional probability of major core damage from LONP without scram is <10^-6. 20 refs., 1 fig., 3 tabs

[en] One of the parameters usually measured in a fast reactor critical assembly is the reactivity associated with inserting a small sample of a material into the core (sample worth). Local heterogeneities introduced by the worth measurement techniques can have a significant effect on the sample worth. Unfortunately, the capability is lacking to model some of the heterogeneity effects associated with the experimental technique traditionally used at ANL (the radial tube technique). It has been suggested that these effects could account for a large portion of what remains of the longstanding central worth discrepancy. The purpose of this paper is to describe a large body of experimental data - most of which has never been reported - that shows the effect of radial tube-related local heterogeneities

[en] The Experimental Breeder Reactor II (EBR-II) shut down on September 30, 1994 and defueling is under way. Longstanding safety rules for changing subassemblies in the EBR-II reactor grid include requirements on neutron count rates. A different safety strategy is needed for the defueling operations because the count rates will become too low as the Sb-Be neutron source decays and fuel removal reduces neutron multiplication. A key component of the new strategy is to establish a deeply subcritical configuration and then confirm the deep subcriticality. The calculational and experimental approaches used to verify that the reactor has been made deeply subcritical are described in this paper. All the approaches make use of a reference loading. Just before the final shutdown, the core was reloaded to have a small (∼25 cents) excess reactivity and essentially the most compact geometry possible. Criticality and other measurements were made to characterize the loading in the last startup, Run 170B

[en] The idea of treating an accident initiator in a probabilistic manner is developed. Instead of using a bounding value, the rod reactivity insertion in an unprotected transient overpower accident is described realistically as a distribution of insertion magnitudes. The initiator analysis uses EBR-II Operating Instructions and data files of rod worths and position histories. The average insertion magnitude is found to be 16 cents, which is only about hall of the feedback reactivity from zero to full power. The probability of inserting 130 cents or more, the Technical Specification limit, is less than 10^-6. The initiator characteristics are then propagated through a probabilistic analysis of the reactivity feedback response to the initiator. This analysis shows that reactivity feedbacks reduce by four orders of magnitude the probability of a rod run-in event resulting in substantial core damage, in addition to the more than five order of magnitude margin afforded by the scram system

[en] Material worth measurements are routinely performed in fast reactor critical programs. Until recently, such measurements were usually made by inserting cylindrical samples via stainless steel tubes. Such measurements pose difficult problems for the analyst since the geometry is very complicated. An alternative measurement utilizes a drawer oscillator to remove or insert a fuel plate. In such a measurement, samples are large enough to cause significant distortions of the local flux. Consequently, it is important to account for local flux distortions when calculations of such experiments are performed. Traditional analysis techniques have not addressed many of the complicated phenomemon associated with the measurements, and large (10 to 20%) discrepancies exist between calculated and experimental sample worths. The intent of this paper is to describe new analysis techniques which have been developed to compute accurately sample worths for complicated geometries and large perturbing samples

[en] Improvements to startup criticality predictions for the EBR-II reactor have been made. More exact calculational models, methods and data are now used, and better procedures for obtaining experimental data that enter into the prediction are in place. Accuracy improved by more than a factor of two and the largest ECP error observed since the changes is only 18 cents. An experimental method using subcritical counts is also being implemented

[en] One of the long-standing problems from LMFBR critical experiments is the central worth discrepancy, the consistent overprediction of the reactivity associated with introducing a small material sample near the center of an assembly. Reactivity (sample worth) experiments in ZPR-9, assembly 34, the Uranium/Iron Benchmark (U/Fe), were aimed at investigating this discrepancy. U/Fe had a large, single-region core whose neutronics was governed almost entirely by ²³⁵U and iron. The essentially one-dimensional plate unit cell had one 1.6 mm-wide column of 93% enriched uranium (U(93)) near the center, imbedded in about 50 mm of iron and stainless steel. The neutron spectrum was roughly comparable to that of an LMFBR, but the adjoint spectrum was much flatter than an LMFBR's. The worths of four different fissile materials were measured and the worth of U(93) was measured using several different experimental techniques

[en] Scores of critical assemblies were constructed, over a period of about three decades, at the Argonne National Laboratory ZPR-3, ZPR-6, ZPR-9, and ZPPR fast critical assembly facilities. Most of the assemblies were mockups of various liquid-metal fast breeder reactor designs. These tended to be complex, containing, for example, mockups of control rods and control rod positions. Some assemblies, however, were 'physics benchmarks'. These relatively 'clean' assemblies had uniform compositions and simple geometry and were designed to test fast reactor physics data and methods. Assemblies in this last category are well suited to form the basis for new criticality safety benchmarks. The purpose of this paper is to present an overview of some of these benchmark candidates and to describe the strategy being used to create the benchmarks

[en] The technical feasibility of using the TREAT reactor facility for boron neutron capture therapy (BNCT) research was assessed. Using one-dimensional neutronics calculations, it was shown that the TREAT core neutron spectrum can be filtered to reduce the undesired radiation (contamination) dose per desired neutron more effectively than can the core spectra from two prominent candidate reactors. Using two-dimensional calculations, it was demonstrated that a non-optimized filter replacing the TREAT thermal column can yield a fluence of desired-energy neutrons more than twice as large as the fluence believed to be required and, at the same time, have a contamination dose per desired neutron almost as low as that from any other candidate facility. The time, effort and cost required to adapt TREAT for a mission supporting BNCT research would be modest