Papers by Daniele Sciannandrone
Le sujet de ce travail de thèse est l’application de la méthode de caractéristiques longues (MOC)... more Le sujet de ce travail de thèse est l’application de la méthode de caractéristiques longues (MOC) pour résoudre l’équation du transport des neutrons pour des géométries à trois dimensions extrudées. Les avantages du MOC sont sa précision et son adaptabilité, le point faible était la quantité de ressources de calcul requises. Ce problème est même plus important pour des géométries à trois dimensions ou le nombre d’inconnues du problème est de l’ordre de la centaine de millions pour des calculs d’assemblage.La première partie de la recherche a été dédiée au développement des techniques optimisées pour le traçage et la reconstruction à-la-volé des trajectoires. Ces méthodes profitent des régularités des géométries extrudées et ont permis une forte réduction de l’empreinte mémoire et une réduction des temps de calcul. La convergence du schéma itératif a été accélérée par un opérateur de transport dégradé (DPN) qui est utilisé pour initialiser les inconnues de l’algorithme itératif and p...
International audienceA three-dimensional (3D) axial capabilities have been recently implemented ... more International audienceA three-dimensional (3D) axial capabilities have been recently implemented in the TDT-MOC solver (Method of Characteristics solver), module of the French deterministic core physics analysis code APOLLO3 currently under development. This solver has already been validated on a two plans hexagonal sub-assembly test-case and showed quite good performances. In the framework of the ASTRID project, the CFV core presents strong heterogeneities, on both radial and axial directions, that are not treated correctly with the usual 2D solvers. The aim of this paper is to quantify the approximation made with non-fully-representative 2D cluster geometries by using the 3D TDT-MOC solver on a whole 3D heterogeneous CFV sub-assembly. Results obtained with 3DTDT-MOC are compared and validated using the continuous energy Monte-Carlo TRIPOLI-4 code and are discussed. Improvements, mainly focus on the homogenization and energy-collapsing methods, are also pointed out
Variance reduction is a key ingredient for solving radiation-protection problems with Monte-Carlo... more Variance reduction is a key ingredient for solving radiation-protection problems with Monte-Carlo particletransport codes. Many variance-reduction methods require the definition of an importance map and exhibit optimal performance if the importance map is given by the adjoint flux, the solution of the adjoint Boltzmann equation. This paper presents the implementation of the Consistent Adjoint-Driven Importance Sampling (CADIS) methodology via a coupling between the TRIPOLI-4 R Monte-Carlo particle-transport code and the IDT deterministic solver, Additionally, we describe the implementation of a new TRIPOLI-4 R score that makes it possible to estimate the adjoint flux during a direct Monte-Carlo calculation. These new features are expected to simplify the solution of difficult shielding problems.
Nuclear Science and Engineering, 2019
In radiation protection studies, the goal is to estimate the response of a detector exposed to a ... more In radiation protection studies, the goal is to estimate the response of a detector exposed to a strongly attenuated radiation field. Monte Carlo (MC) particle transport codes give the possibility to efficiently solve for such responses using several variance-reduction (VR) methods that help allocating more CPU time to the simulation of highly contributing histories. The TRIPOLI-4® MC particle transport code offers two main methods, the exponential transform and adaptive multilevel splitting (AMS), which rely on the definition of a suitable importance map. In this paper, we present an implementation of a generalized Consistent Adjoint Driven Importance Sampling (CADIS) methodology for TRIPOLI-4. The implementation relies on coupling with the IDT code, a deterministic solver for the Boltzmann adjoint transport equation, for the generation of importance maps. We study the performance of both VR methods present in TRIPOLI-4 in this setting. In particular, to our knowledge, this is the first time that a CADIS-like methodology has been applied to AMS.
The first collision source calculation has recently been implemented in the IDT discrete ordinate... more The first collision source calculation has recently been implemented in the IDT discrete ordinates nodal and short characteristics transport code. Several methods are developed and compared with the multigroup discrete ordinates transport code PARTISN and the continuous in energy Monte Carlo code TRIPOLI4 R ©. The paper describes five different options: 1) a non-conservative scheme based on the tracking through the mesh centers, 2) a non-conservative averaging of mesh corner values, 3) cell based balance scheme with the surface averaged interface currents, 4) semi-analytic approach based on a conservative local-cell balance with numerical surface integration and 5) method of characteristics (MOC) based conservative scheme. The comparisons are performed on simple one group 3D test cases, one with a simple homogeneous geometry and another with very strong material heterogeneities. The detector responses are obtained for a 42-group 3D gamma shielding problem and compared. Results show ...
Journal of Computational Physics, 2022
Variance reduction is a key ingredient for solving radiation-protection problems with Monte-Carlo... more Variance reduction is a key ingredient for solving radiation-protection problems with Monte-Carlo particle-transport codes. Many variance-reduction methods require the definition of an importance map and exhibit optimal performance if the importance map is given by the adjoint flux, the solution of the adjoint Boltzmann equation. This paper presents the implementation of the Consistent Adjoint-Driven Importance Sampling (CADIS) methodology via a coupling between the TRIPOLI-4 R Monte-Carlo particle-transport code and the IDT deterministic solver, Additionally, we describe the implementation of a new TRIPOLI-4 R score that makes it possible to estimate the adjoint flux during a direct Monte-Carlo calculation. These new features are expected to simplify the solution of difficult shielding problems.
20th Topical Meeting of the Radiation Protection and Shielding Division (RPSD 2018), Aug 26, 2018
Annals of Nuclear Energy
Abstract In recent years a solver based on the Method of Characteristics (MOC) allowing the treat... more Abstract In recent years a solver based on the Method of Characteristics (MOC) allowing the treatment of 3D extruded geometries has been developed inside the TDT code of APOLLO3. The standard Step Characteristics (SC) approximation is used and results show an excellent agreement with Monte-Carlo simulations. However a fine mesh refinement is needed to converge axially the strong flux gradients customarily appearing in 3D reactor physics applications. In this paper an improvement of this method is proposed: the results of the previous work show that much of the flux variations are likely to be represented by a polynomial basis along the vertical direction. Since most of the geometrical and physical heterogeneities are radially located, the SC approach is preserved to represent the solution over the radial plane. As a matter of fact the strong irregularities in the geometrical meshes prevent from an efficient use of a polynomial expansion. On the contrary along the axial direction the computational meshes assume a Cartesian shape, well suited for a polynomial representation of sources and fluxes. A convenient polynomial development in this direction allows us to approximate the strong flux slopes without the help of a large number of axial meshes. Finally we will show how synthetic accelerated results for the fourth generation fast breeder reactor ASTRID perform with respect to computational time, memory efficiency and precision against Monte-Carlo calculations. We will consider assembly geometries that span the full-height of the reactor and that are characterized by the strong heterogeneity conceived to enhance the axial leakage which improves the safety features of this new concept, especially under void conditions.
EPJ Web of Conferences
Recently a solver based on the Method Of Characteristics (MOC) for 3D extruded geometries has bee... more Recently a solver based on the Method Of Characteristics (MOC) for 3D extruded geometries has been developed in the APOLLO3 R project [1]. In this method the domain is divided in regions and the Step Characteristics (SC) approximation is used in each of them to represent the solution of the transport problem. Since the biggest degree of heterogeneities is found along the radial direction, the idea proposed in this paper is to keep the SC approximation to compute the solution over the radial plane and to implement a polynomial expansion of the flux along the vertical direction. In fact the results of the previous works [1] show that the flux gradient along the z axis are likely to be represented by a classical polynomial basis. In this paper we weigh up the benefits of using a less refined mesh.
SNA + MC 2013 - Joint International Conference on Supercomputing in Nuclear Applications + Monte Carlo, 2014
The solution of the neutron transport equation in 3D domains is a high resource demanding task. O... more The solution of the neutron transport equation in 3D domains is a high resource demanding task. One of the most successful method of solution of this equation is the Method of Characteristics (MOC). However, the huge number of trajectories constitutes a serious issue in terms of memory needs and in terms of number of floating point operations. In this paper, we propose some techniques to compress tracking data and to speed up the MOC solver. These methods take advantage of regularities of 3D prismatic geometries. The new algorithms have been implemented in the 3D MOC solver TDT and applied to a typical UOx/MOx colorset.
Annals of Nuclear Energy, 2015
ABSTRACT Specialized tracking techniques have been developed for a step-characteristics transport... more ABSTRACT Specialized tracking techniques have been developed for a step-characteristics transport solver. These techniques strongly exploit the regularities of extruded 3D axial geometries to reduce the number of operations and the memory requirements of the transport solver. The method of compound trajectories has been applied to exactly treat geometrical boundary conditions with a minimum number of trajectories. The chord classification method consists in identifying recognized types of chords that share the same length. It allows a consistent reduction of the number of chords that have to be stored and also an important speed up of the transport iterations thanks to the pre-computed values of escape coefficients. An alternative storage method has also been implemented: in this method one trajectory is represented by a partial record which is used to reconstruct the trajectories on-the-fly. Numerical tests show good improvements for both memory and computational time. Optimized methods have also been developed for horizontal and vertical directions, which are particularly advantageous in terms of computational costs. In the same contest, a modified version of the Minimized-Bickley quadrature (Sanchez et al., 2002) containing these directions is proposed. The new methods have been tested by comparison with standard quadrature sets. Results show that for some cases the special directions improve the efficiency of the solver, while for some others the opposite effect is obtained.
Epj Web of Conferences, 2021
A new Discrete Ordinates transport solver for unstructured tetrahedral meshes is presented. The s... more A new Discrete Ordinates transport solver for unstructured tetrahedral meshes is presented. The solver uses the Discontinuous Galërkin Finite Element Method with linear or quadratic expansion of the flux within each cell. The solution of the one-group problem is obtained with non-preconditioned fixed-point or GMRES iterations. Precision and performances of the solver are evaluated on the 3D Radiation Transport Benchmark Problems proposed by Kobayashi, showing very good agreement with the reference and good computing times in serial execution.
The topic of our research is the application of the Method of Long Characteristics (MOC) to solve... more The topic of our research is the application of the Method of Long Characteristics (MOC) to solve the Neutron Transport Equation in three-dimensional axial geometries. The strength of the MOC is in its precision and versatility. As a drawback, it requires a large amount of computational resources. This problem is even more severe in three-dimensional geometries, for which unknowns reach the order of tens of billions for assembly-level calculations.The first part of the research has dealt with the development of optimized tracking and reconstruction techniques which take advantage of the regularities of three-dimensional axial geometries. These methods have allowed a strong reduction of the memory requirements and a reduction of the execution time of the MOC calculation.The convergence of the iterative scheme has been accelerated with a lower-order transport operator (DPN) which is used for the initialization of the solution and for solving the synthetic problem during MOC iterations...
A polynomial expansion in the axial direction has been implemented in the Method Of Characteristi... more A polynomial expansion in the axial direction has been implemented in the Method Of Characteristics framework, for the solution of the mono-group, steady state neutron transport equation for 3D extruded geometries. This work has been realized in the context of the APOLLO3 R project and, in particular, in the Two & three Dimensional Transport (TDT) module. In this paper we weigh up benefits and disadvantages of this approach, when comparing to the classical Step Characteristics (SC) approximation, already available in APOLLO3 R .
Nuclear Science and Engineering
Abstract In this paper we describe some recent developments in the Method of Characteristics (MOC... more Abstract In this paper we describe some recent developments in the Method of Characteristics (MOC) for three-dimensional (3D) extruded geometries in the nuclear reactor analysis code APOLLO3®. We discuss the parallel strategies implemented for the transport sweep of the MOC solver in the OpenMP framework, and introduce the 3D version of the DPN operator that is customarily used in APOLLO2 to accelerate MOC convergence. In order to provide good physical results, we have also coupled the MOC with the self-shielding environment of APOLLO3®. We describe, in particular, the coupling techniques necessary to implement a full subgroup cross-section self-shielding method and a specialized version of the Tone self-shielding technique. In this framework, we use part of the tracking method used for the 3D calculation to provide the two-dimensional Collision Probability Method (CPM) coefficients necessary to produce the self-shielding calculations. We will show some important computational speedups also in the CPM of APOLLO3® with respect to the APOLLO2 CPM equivalent implementation, including the parallelization issue. Finally, we will compare our approach toward a Monte Carlo calculation of a fast breeder reactor hexagonal assembly representing a fertile-fissile interface.
Uploads
Papers by Daniele Sciannandrone