26.02.2013 Views

Program - Brookhaven National Laboratory

Program - Brookhaven National Laboratory

Program - Brookhaven National Laboratory

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

FC 4 11:40 AM<br />

Propagation of Nuclear Data Uncertainties in PWR Burnup Pin-Cell Deterministic<br />

Calculations using the Total Monte Carlo Approach<br />

Francesco Ferragut, Pouya Sabouri, Adrien Bidaud<br />

Laboratoire de Physique Subatomique et Cosmologie, CNRS/Université Joseph Fourrier, Institut<br />

Polytechnique de Grenoble<br />

We present here the results of the PWR Burnup Pin-Cell Benchmark UAM I-1b based on the use of the<br />

TENDL library [1] with the deterministic transport code DRAGON [2]. The evaluation of the uncertainties<br />

uses the Total Monte Carlo method. Hundreds of TENDL evaluated files of of the most important<br />

nuclides for this benchmark ( 235 U, 238 U, 239 Pu, 241 Pu) were processed with NJOY and other codes to produce<br />

hundreds of WIMS libraries that DRAGON can read. Then, hundreds of burnup calculation where<br />

done, whose spread is an image of the uncertainties in the evaluated files that is itself an image of the<br />

uncertainties in the model parameters of TALYS. Our methodology gives very similar results than the full<br />

stochastic Total Monte Carlo, where a Monte Carlo code such as SERPENT [3] is used for solving the<br />

Boltzmann equation. In our case, thanks to the deterministic approach, we do not have to subtract the<br />

statistical noise to calculate the uncertainty in the outputs such as keff or nuclide densities. Of course,<br />

our methodology would be limited to applications that are within the recommended domain of use of the<br />

transport code. When compared to methods based on the sampling of the deterministic XS [4],[5] the<br />

advantage of our method is the consistency of the handling of the nuclear data all along the process. We<br />

do not have to generate discontinuities in the cross sections to obtained our energy integrated libraries.<br />

Our sampled libraries are build exactly as any classical groupwise libraries. Our group constants are build<br />

from sampled but continuous cross sections themselves evaluated from sampled pre-evaluation parameters.<br />

Furthermore, our approach is more consistent with the evaluation as we do not sample the lumped<br />

cross sections (scattering, absorption, fission and total), but we lump sampled - probably more physical<br />

- reactions such as elastic, inelastic scatterings and n,xn reaction for instance. Another advantage of the<br />

use of TENDL libraries in a deterministic code is that we can use perturbation theory at least at the<br />

first step of burn up so as to calculate not only the uncertainties but also the sensitivities as a function<br />

of energy of Keff or nuclide inventory changes. We can then better understand which reactions are more<br />

contributing to the uncertainties during fuel burnup. Furthermore, we have developed tools that can build<br />

the covariance matrices of the sampled groupwise nuclear data. The folding of the sensitivities obtained by<br />

Perturbation Theory with these calculated, self-shielded covariance matrices gives results very comparable<br />

to the observed uncertainties calculated with GPT. Thanks to this method, we can not only calculate the<br />

uncertainties but also track down the error down to the actual reactions described in the evaluation (and<br />

not on the lumped ND used in the deterministic code) and give the weight in the error of them and of the<br />

correlations between those reactions. Given the consistency of the method, our method not only complement<br />

the results of available other methods, but also allows for a more straightforward and more coherent<br />

link between the evaluation of the uncertainties given by the evaluator and their impacts calculated by the<br />

reactor physicist.<br />

Corresponding author: Pouya Sabouri<br />

[1] D. Rochman and A.J. Koning. TENDL-2011: TALYS-based Evaluated Nuclear Data Library, PHYSOR-<br />

2012 conference, Knoxville, Tenessee, USA, April 15-20, 2012. [2] G. Marleau, R. Roy and A. Hebert,<br />

DRAGON: A Collision Probability Transport Code for Cell and Supercell Calculations, Report IGE-157,<br />

Institut de genie nucleaire, ecole Polytechnique de Montreal, Montreal, Quebec (1994) [3] D. Rochman<br />

and A.J. Koning, Propagation of 235,236,238 U and 239 Pu nuclear data uncertainties for a typical PWR fuel<br />

element, accepted in Nucl. Technology, October 2011 [4] Artem Yankov, Makus Kleis, Marhew A. Jesse,<br />

85

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!