[en] This thesis is devoted to a numerical study of the quantum dynamics of the Fermi accelerator which is classically chaotic: it is particle in a one dimensional box with a oscillating wall. First, we study the classical dynamics: we show that the time of impact of the particle with the moving wall and its energy in the wall frame are conjugated variables and that Poincare surface of sections in these variables are more understandable than the usual stroboscopic sections. Then, the quantum dynamics of this systems is studied by the means of two numerical methods. The first one is a generalization of the KKR method in the space-time; it is enough to solve an integral equation on the boundary of a space-time billiard. The second method is faster and is based on successive free propagations and kicks of potential. This allows us to obtain Floquet states which we can on one hand, compare to the classical dynamics with the help of Husimi distributions and on the other hand, study as a function of parameters of the system. This study leads us to nice illustrations of phenomenons such as spatial localizations of a wave packet in a vibrating well or tunnel effects. In the adiabatic situation, we give a formula for quasi-energies which exhibits a phase term independent of states. In this regime, there exist some particular situations where the quasi-energy spectrum presents a total quasi-degeneracy. Then, the wave packet energy can increase significantly. This phenomenon is quite surprising for smooth motion of the wall. The third part deals with the evolution of a classical wave in the Fermi accelerator. Using generalized KKR method, we show a surprising phenomenon: in most of situations (so long as the wall motion is periodic), a wave is localized exponentially in the well and its energy increases in a geometric way. (author). 107 refs., 66 figs., 5 tabs. 2 appends

[en] A new family of stationary coherent states for the two-dimensional harmonic oscillator is presented. These states are coherent in the sense that they minimize an uncertainty relation for observables related to the orientation and the eccentricity of an ellipse. The wavefunction of these states is particularly simple and well localized on the corresponding classical elliptical trajectory. As the number of quanta increases, the localization on the classical invariant structure is more pronounced. These coherent states give a useful tool to compare classical and quantum mechanics and form a convenient basis to study weak perturbations. (author)

[en] This paper is an introduction to the physics of Accelerator Driven Subcritical Reactors (ADSR) and some technologies associated with them. The basic neutronics is presented with a specific discussion of modifications with respect to that of critical reactors. The fuel evolution in ADSR's is discussed, including the influence of reactivity surges and drops on the limitation of the design reactivity. The application of ADSRs to nuclear waste management is examined and the different options are discussed. Finally, some practical proposals are briefly discussed

[en] The following topics were dealt with: undercriticality, general principles, accelerator assistance, future perspectives

[en] Simulations of new reactor designs, such as generation IV concepts, require three dimensional modeling to ensure a sufficiently realistic description for safety analysis. If precise solutions of local physical phenomena (DNBR, cross flow, form factors,...) are to be found then the use of accurate 3D coupled neutronics/thermal-hydraulics codes becomes essential. Moreover, to describe this coupled field with a high level of accuracy requires successive iterations between neutronics and thermal-hydraulics at equilibrium until convergence (power deposits and temperatures must be finely discretized, ex: pin by pin and axial discretization). In this paper we present the development and simulation results of such coupling capabilities using our code MURE (MCNP Utility for Reactor Evolution), a precision code written in C++ which automates the preparation and computation of successive MCNP calculations either for precision burnup and/or thermal-hydraulics/thermic purposes. For the thermal-hydraulics part, the code COBRA is used. It is a sub-channel code that allows steady-state and transient analysis of reactor cores. The goal is a generic, non system-specific code, for both burn-up calculations and safety analysis at any point in the fuel cycle: the eventual trajectory of an accident scenario will be sensitive to the initial distribution of fissile material and neutron poisons in the reactor (axial and radial heterogeneity). The MURE code is open-source, portable and manages all the neutronics and the thermal-hydraulics/thermic calculations in background: control is provided by the MURE interface or the user can interact directly with the codes if desired. MURE automatically builds input files and other necessary data, launches the codes and manages the communication between them. Consequently accurate 3D simulations of power plants on both global and pin level of detail with thermal feedback can be easily performed (radial and axial meshing grids are managed by MURE). A comparison to an NEA benchmark of a heterogeneous PWR MOX/UO₂ core is presented. Results for hot full-power conditions show an agreement of our simulations with the benchmark (the accuracy of the results are within the errors of the benchmark). The temperature dependent cross sections for MCNP have been provided for each isotope using NJOY99. The convergence of coupled field in heterogeneous configuration is obtained after around five iterations; the Shannon entropy effect which affects neutron source convergence is attenuated using a large number of source particles and inactive cycles. (authors)

[en] This document deals with the quantification of the minimum thermal power level for a demonstrator and the definition of the physical criteria which define the representative character of a demonstrator towards a power reactor. Solutions allowing to keep an acceptable flow in an industrial core, have also been studied. The document is divided in three parts: the representativeness elements, the considered solutions and the characterization of the neutrons flows at the interfaces and the dose rates at the outer surface of the vessel. (A.L.B.)

[en] The future of nuclear energy may require breeding and optimized waste management. Innovative technologies have to be explored, in order to reduce considerably the ore consumption and the associated waste production. Sodium-cooled fast reactors seem to be the most achievable technology in the coming decades, and can play an important role to launch the generation 4 technologies. However, the standard sodium-cooled reactors face to the problem of the positive void coefficient and a major Minor Actinides (MA) production if transmutation is not considered. In this context, we perform neutronics studies on innovative (or evolutive) sodium-cooled reactors. These studies are based on MURE (MCNP Utilities for Reactors Evolution), a C++ object-oriented evolution code that couple the Monte-Carlo transport code MCNP with a fuel depletion code under given conditions (constant power, refueling, reactivity adjustment,..). By construction, MURE is very versatile and offers the possibility to interact with the system during the evolution. Different ways of evolution such as predictor-corrector methods, hybrid multi-group binning approach are used to speed up MCNP run time (at least a factor 30), - In a first part, a 'reference' case, a SFR without recycling MA, is presented; a propagation of statistical error is shown during the whole evolution and several methods are compared (predictor-corrector, hybrid multi-group binning,..). Different configurations of a fast sodium cooled reactor (SFR, ∼1 GWe) are investigated, as for example a U/Pu core MA-loaded Uranium blankets have been studied and compared to a 'reference' standard U/Pu systems. The evolution is performed until equilibrium. Safety parameters during the whole evolution are studied, particularly void coefficient degradation because of presence of minor actinides. The radiotoxicity of the waste leaving the reactor has been calculated and compared to the 'reference' case. We also consider the use of thorium fuels, using U or Pu as fissile material. A self-breeding Th/U configuration has been found, using thorium blankets and is characterized by a negative void coefficient. Comparison of produced wastes in different strategies (transmutation of MA or not) will be presented. (authors)

[en] Research on ADS, related new fuels and their ability for nuclear waste incineration leads to a revival of interest in nuclear cross-sections of many nuclides in a large energy range. Discrepancies observed between nuclear databases require new measurements in several cases. A complete measurement of such cross-sections including resonance resolution consists in an extensive beam time experiment associated to a long analysis. With a slowing down spectrometer associated to a pulsed neutron source, it is possible to determine a good cross-section profile in an energy range from 0.1 eV to 40 keV by making use of a slowing-down time lead spectrometer associated to a pulsed neutron source. These measurements performed at ISN (Grenoble) with the neutron source GENEPI requires only small quantities of matter (as small as 0.1 g) and about one day of beam by target. We present cross-section profile measurements and an experimental study of the self-shielding effect. A CeF₃ scintillator coupled with a photomultiplier detects gamma rays from neutronic capture in the studied target. The neutron flux is also measured with a ²³³U fission detector and a ³He detector at symmetrical position to the PM in relation to the neutron source. Absolute flux values are given by activation of Au and W foils. The cross-section profiles can then be deduced from the target capture rate and are compared with very detailed MCNP simulations, which reproduce the experimental set-up and provide also capture rates and flux. The method is then applied to ²³²Th, of main interest for new fuel cycle studies, and is complementary to higher energy measurements made by D. Karamanis et al. (CENBG). Results obtained for three target thicknesses will be compared with simulations based on different data bases. Special attention will be paid to the region of unresolved resonances (>100 eV). (author)

[en] The possible interest of accelerator driven subcritical reactors for minor actinides incineration is examined. The physics of neutron multiplying systems is recalled. The differences between critical and subcritical reactors' control are described, with emphasis on the importance of the delayed neutrons fraction. The minor actinides fuel evolution is studied with the conclusion that fast neutron spectra are clearly more efficient then thermal neutron spectra. It is, also, shown that characteristic times for incineration should be in the order of 10 years. The number of minor actinides incinerators necessary for 60 PWRs is estimated to be about 6 with total thermal power of 9 GW. These reactors will, also, be able to transmute essentially all ⁹⁹Tc and ¹²⁹I produced by the 60 PWR. The excess electricity cost for MA incineration is estimated to be about 5%. (authors)

[en] Projects dealing with future reactors based on new fuels and able to incinerate nuclear waste require good knowledge of numerous cross sections. In order to resolve nuclear database discrepancies, capture cross-section profiles between 0.1 eV and 30 keV have been measured for different materials using a lead-slowing-down-time spectrometer in association with a pulsed neutron generator. The measurement of the neutron flux with a ²³³U fission detector and a ³He counter, and careful analysis of the E-t correlation compared to very precise Monte Carlo simulations, brought new information on the lead scattering cross section. Capture profiles for reference materials (gold, tantalum, indium, and silver), core materials (thorium and technetium), and structure materials (manganese and nickel) were measured with a CeF₃ scintillator and photomultiplier for different thicknesses. Areas of agreement and disagreement between experimental results and simulations using different databases have been determined with a precision of 5%. Correction tables are given for some elements. This method opens an efficient way for revisiting (n, γ) databases, and it allows rapid error evaluation and sensitivity studies