Programmation de L'algorithme de synthese de Gloushkov pour les tables de fluence des systemes sequentiels

A l'origine nous nous avions proposé programmer sur CAB-5DO la méthode de synthèse de Gloushkov pour la classe des automates asynchrones du type machine transfert, d'où le titre de la thèse. A sa place nous avons resous le même problême â l'aide d'une méthode algorihmique originale» Après une introduction on définit la nouvelle méthode, valable pour les machines asynchrones, ainsi que quelques propriétés intéréssantés des expressions itérées ( en particulier de l'événement universel )» Dans la suite on établit les organigrammes générales de synthèse et l'organisation du travail sur machine. Après la conclusion, où l'on résume les avantages de nôtre méthode, il y a txoisPnnexes, dans le premier desquels on fait d'une façon pratique le point sur ces avantages par rapport à la méthode de Gloushkov f dans.le deuxième on groupe des organigrammes très détaillés, les programmes correspondants et quelques résultats > et dans le troisième le programme traduit en fortran IV qui a été mis au point sur le calculateur IBM 360/44 du l . E. R. A. ( Centre d'Etudes et Récherches en Automatisme).

Preparing Exercise I-3: Optimization of cross-section tables using sensitivity coefficients in COBAYA3

Preparing Exercise I-3: Optimization of cross-section tables using sensitivity coefficients in COBAYA3

Effects of cross sections tables generation and optimization on rod ejection transient analyses

Best estimate analysis of rod ejection transients requires 3D kinetics core simulators. If they use cross sections libraries compiled in multidimensional tables,interpolation errors – originated when the core simulator computes the cross sections from the table values – are a source of uncertainty in k-effective calculations that should be accounted for. Those errors depend on the grid covering the domain of state variables and can be easily reduced, in contrast with other sources of uncertainties such as the ones due to nuclear data, by choosing an optimized grid distribution. The present paper assesses the impact of the grid structure on a PWR rod ejection transient analysis using the coupled neutron-kinetics/thermal-hydraulicsCOBAYA3/COBRA-TF system. Forthispurpose, the OECD/NEA PWR MOX/UO2 core transient benchmark has been chosen, as material compositions and geometries are available, allowing the use of lattice codes to generate libraries with different grid structures. Since a complete nodal cross-section library is also provided as part of the benchmark specifications, the effects of the library generation on transient behavior are also analyzed.Results showed large discrepancies when using the benchmark library and own-generated libraries when compared with benchmark participants’ solutions. The origin of the discrepancies was found to lie in the nodal cross sections provided in the benchmark.

Optimization of multidimensional cross-section tables for few-group core calculations

Multigroup diffusion codes for three dimensional LWR core analysis use as input data pre-generated homogenized few group cross sections and discontinuity factors for certain combinations of state variables, such as temperatures or densities. The simplest way of compiling those data are tabulated libraries, where a grid covering the domain of state variables is defined and the homogenized cross sections are computed at the grid points. Then, during the core calculation, an interpolation algorithm is used to compute the cross sections from the table values. Since interpolation errors depend on the distance between the grid points, a determined refinement of the mesh is required to reach a target accuracy, which could lead to large data storage volume and a large number of lattice transport calculations. In this paper, a simple and effective procedure to optimize the distribution of grid points for tabulated libraries is presented. Optimality is considered in the sense of building a non-uniform point distribution with the minimum number of grid points for each state variable satisfying a given target accuracy in k-effective. The procedure consists of determining the sensitivity coefficients of k-effective to cross sections using perturbation theory; and estimating the interpolation errors committed with different mesh steps for each state variable. These results allow evaluating the influence of interpolation errors of each cross section on k-effective for any combination of state variables, and estimating the optimal distance between grid points.