G. Noguere

Retroactive Generation of Covariance Matrix of Nuclear Model Parameters Using Marginalization Techniques

Abstract An uncertainty propagation methodology relying on marginalization techniques was recently developed to produce covariance matrices between existing model parameters involved in describing neutron-induced reactions. This work has been implemented in the nuclear data assimilation tool CONRAD. The performance of the code was demonstrated through simplified test cases based on a Reich-Moore description of the 155Gd(n,γ) reaction. Results are compared with those produced via Monte Carlo techniques.

Interpretation of Fission Product Oscillations in the MINERVE Reactor, from Thermal to Epithermal Spectra

Abstract To take into account the reactivity loss in spent fuels, an experimental program was set up in 1993 at CEA-Cadarache, France, oscillating separated fission products (FPs) in the MINERVE reactor. Reactivity worth measurements of small samples allow the extraction of information about nuclear data of the studied isotopes. A fully validated calculation scheme has been implemented to interpret pile-oscillation measurements. Therefore, calculation over experiment ratios can be accurately transposed to trends in the integral capture cross section of the FPs. With the European JEFF3.1.1 library, results show a discrepancy below 3% for several nuclides: 155Gd, 149, 152Sm, 143Nd, and 95Mo, but improvements may be needed for some others: 133Cs, 103Rh, 99Ru, and 153Eu. Based on the Integral Data Assimilation technique, we propose new thermal cross-section values, (348 ± 14) b and (42 478 ± 1793) b, for two of the most absorbing nuclides, 143Nd and 149Sm, respectively.

Zero Variance Penalty Model for the Generation of Covariance Matrices in Integral Data Assimilation Problems

Abstract This paper presents simple models developed to generate covariances between observable and latent variables. The methodology consists of using “variance penalty” terms as a measure of the contribution of the latent-variable uncertainties to the variance of a given calculated quantity z. This approach provides a useful understanding of how the observable and latent variables are related to each other and ensures the positive-definiteness of the covariance matrix. This work has been implemented in the nuclear data assimilation tool CONRAD. Performances of analytic and Monte Carlo models are illustrated with covariances calculated for neutron-induced capture reactions on stable xenon isotopes (124Xe, 126Xe, 128Xe, 129Xe, 130Xe, 132Xe, and 134Xe).

The Use of Nuclear Data as Nuisance Parameters in the Integral Data Assimilation of the PROFIL Experiments

Abstract The PROFIL and PROFIL-2 experiments were carried out in the fast reactor PHENIX. They were designed to provide integral information on neutron cross sections [(n,γ), (n,2n), and (n,f)] of several fission products and actinides. Previous interpretations report integral results with unrealistic small uncertainties that only take into account the statistical contribution. This work presents an uncertainty propagation technique able to include systematic uncertainties due to neutron fluence scaling. Such a technique consists of marginalizing analytically the uncertainties of the nuclear data (nuisance parameters) involved in the fluence scaling procedure. For the capture cross sections of 235U, 238U, and 239Pu, the interpretation of the PROFIL and PROFIL-2 experiments with the international library JEFF-3.1.1 provides excellent results equal to 1.000, 1.019, and 0.982, respectively, with a relative uncertainty close to 1.5% (1σ).

Improved mixed oxide fuel calculations with the evaluated nuclear data library JEFF-3.2

Abstract An overestimation of the keff values for mixed oxide (MOX) fuels was identified with Monte Carlo (TRIPOLI-4) and deterministic (APOLLO2) calculations based on the Joint Evaluated Fission and Fusion (JEFF) evaluated nuclear data library. The overestimation becomes sizeable with Pu aging, reaching a reactivity change of Δρ≈+700 pcm for integral measurements carried out with MOX fuel containing a large amount of americium. This bias was observed for various critical configurations performed in the zero-power reactor EOLE of the Commissariat à l’énergie atomique et aux énergies alternatives (CEA), Cadarache, France. The present work focuses on the improvements achieved with the new 239Pu and 241Am evaluated nuclear data files available in the latest version of the JEFF library (JEFF-3.2). The resolved resonance range of the plutonium evaluation was reevaluated at Oak Ridge National Laboratory (ORNL), Oak Ridge, Tennessee, with the SAMMY code in collaboration with CEA Cadarache. The resonance parameters of the americium evaluation were obtained with the REFIT code in collaboration with the research institutes Institute for Reference Materials and Measurements (IRMM), Geel, Belgium, and Institut de recherche sur les lois fondamentales de l’Univers (Irfu), Saclay, France.

On the Use of Zero Variance Penalty Conditions for the Generation of Covariance Matrix between Model Parameters

Abstract This paper presents models developed to generate covariances between observable, latent and nuisance variables. The methodology consists in using “variance penalty” terms as a measure of the contribution of the uncertainty of the latent and nuisance variables to the variance of a given calculated quantity z . The Resonance Parameter Covariance Matrix for 241 Am obtained with the Zero Variance Penalty model (ZVP) is compared with the corresponding one provided by Monte-Carlo.