This thesis introduces new software to interactively construct multi-layer models and bedding sequences, populate layer-by-layer properties, and enable the fast simulation of nuclear logs. The method consists of modifying simultaneously layer thicknesses and properties to rapidly simulate the outcome (nuclear logs) for comparison to field logs. I include applications which appraise the numerical simulation of gamma-ray, density, compensated neutron, and photoelectric factor logs. An analogous application for sonic modeling is considered as well which uses a modified version of Wyllie’s slowness averaging equation. The procedure is tested for the case of vertical wells and horizontal layers. Examples of application include 6 synthetic and 5 field cases. Additionally, the software is implemented in combination with other formation-evaluation procedures to interpret resistivity and nuclear logs. Simulations of nuclear logs for synthetic models can be used to improve the assessment and interpretation of field data. Interactive modeling and simulation of nuclear logs provides a very good agreement with field logs with an average error of 3.9%. The order of logs to be matched as well as the data available are significant factors in the accuracy of the match. Numerical simulation and matching of field logs using fast modeling procedures is a reliable method to improve the inference of static and dynamic petrophysical properties of rock formations. / text

Les accélérateurs d’ions lourds sont un outil incontournable pour la recherche en physique nucléaire. Ils sont également utilisés pour diverses applications. Il est nécessaire de caractériser la production des neutrons secondaires dans les accélérateurs afin de garantir un fonctionnement sûr en toutes circonstances. Cependant, les données expérimentales sont très rares voire inexistantes. Pour certaines données, on note des divergences entre différentes publications. Des désaccords sont aussi observés entre les mesures et les calculs. Toutes ces raisons justifient le programme Thick Target Neutron Yields (TTNY) dont l’objectif est de mesurer des spectres doublement différentiels (énergie, angle) des neutrons générés par l’interaction des ions lourds (12≤Afaisceau≤208 et 4,6 MeV/nucléon≤Efaisceau≤95 MeV/nucléon) sur cibles épaisses (natC, natCu et natNb). Deux techniques de mesure ont été utilisées : Activation et Temps de vol. Cela permet d’avoir une meilleure confiance dans les mesures, d’étudier les limites expérimentales et de consolider les conclusions que l’on peut en tirer. Les mesures sont comparées à des simulations effectuées dans ce travail avec les codes Monte-Carlo les plus utilisés en calcul nucléaires : PHITS (japonais), FLUKA (européen (CERN/INFN)) et MCNP (américain). Ces comparaisons ont permis d’évaluer la qualité des codes dans les énergies étudiées et pour les masses des noyaux explorées. Elles ont permis aussi de conclure sur les incertitudes systématiques et les éventuelles évolutions à apporter aux modèles physiques de ces codes. / Heavy-ion accelerators are an essential tool for nuclear physics research. They are also adopted in several applications. It is necessary to characterize the secondary neutrons production in order to guarantee a safe operation in every circumstance in accelerators. However, experimental data are very rare or even non-existent. For some data, we notice disagreements between different publications. Disagreements are also observed between measurements data and simulations. For all these reasons, we established the program Thick Target Neutron Yields (TTNY). This program aims to measure the double differential neutron spectra (energy, angle) generated by the interactions of heavy-ions (12≤Abeam≤208 and 4.6 MeV/nucleon≤Ebeam≤95 MeV/nucleon) on thick targets (natC, natCu and natNb). Two measurements methods were adopted: Activation and Time of Flight. This choice allows having a better confidence on the measurements, studying experimental limits and consolidating the conclusions that could be drawn from the experimental results. The measurements are compared to the simulations performed with some Monte-Carlo widely used in nuclear simulation: PHITS (Japanese), FLUKA (European (CERN/INFN)) and MCNP (American). These comparisons allowed evaluating the modeling quality of heavy-ion reactions for the energies and masses explored in this work. We also conclude on the systematic uncertainties and on the potential improvements to be introduced to physics models of these codes.