[en] Future sodium cooled fast reactors (SFRs) have to fulfill the GEN-IV requirements of enhanced safety, minimal waste production, increased proliferation resistance and high economical potential. This PhD project is dedicated to the evaluation of the impact of innovative fuels (especially minor actinides bearing oxide sphere-pac fuels) on the safety performance of advanced SFRs with transmutation option. The SIMMER-III code, originally tailored to mechanistically analyze later phases of core disruptive accidents, is employed for accident simulations. During the PhD project, the code has been extended for a better simulation of the early accident phase introducing the treatment of thermal expansion reactivity effects and for taking into account the specifics of sphere-pac fuels (thermal conductivity and gap conditions). The entire transients (from the initiating event to later accident phases) have been modeled with this extended SIMMER version. Within this PhD work, first the thermo-physical properties of sphere-pac fuel have been modeled and cast into SIMMER-III. Then, a new computational method to account for thermal expansion feedbacks has been developed to improve the initiation phase modeling of the code. The technique has the potential to evaluate these reactivity feedbacks for a fixed Eulerian mesh and in a spatial kinetics framework. At each time step, cell-wise expanded dimensions and densities are calculated based on temperature variations. Density factors are applied to the expanded densities to get an equivalent configuration (in reactivity) with original dimensions and modified densities. New cross sections are calculated with these densities and the reactivity of the equivalent configuration is computed. The developed methods show promising results for uniform and non-uniform expansions. For non-uniform expansions, model improvement needs have been identified and neutronics simulations have been carried out to support future SIMMER extensions. Preliminary results are encouraging. In the third part of the PhD, two core designs with conventional and sphere pac fuels are compared with respect to their transient behavior. These designs were established in the former CP-ESFR project: the working horse core and the optimized CONF2 core (with a large sodium plenum above the core for coolant void worth reduction). The two fuel design options are compared for steady state and transient conditions (Unprotected Loss of Flow accident, ULOF) either at beginning of life (BOL) or under irradiated conditions. Analyses for sphere-pac fuel reveal two main phases to consider at BOL. At start-up, the non-restructured sphere-pac fuel shows a low thermal conductivity compared to pellet fuel of same density. However, the fuel restructures quickly (in a few hours) due to the high thermal gradients and its thermal conductivity recovers. The fuel then shows a behavior close to the pellet one. The study also shows that the CONF2 core leads to a very mild transient for a ULOF accident at BOL. The large upper sodium plenum seems to effectively prevent large positive reactivity insertions. However, stronger reactivity and power peaks are observed under irradiated conditions or when americium is loaded in the core and lower axial blanket. This PhD work demonstrates, under current simulation conditions, that sphere-pac fuels do not seem to cause specific safety issues compared to standard pellet fuels, when loaded in SFRs. The accurate simulation of core thermal expansion reactivity feedbacks by means of the extended SIMMER version plays an important role in the accident timing (simulations confirm the expected delay in the first power peak) and on the energetic potential compared to the case where these feedbacks are omitted. The analyses also confirm the mitigating impact of a large sodium plenum on transients with voiding potential. The behavior of sphere-pac fuel in these conditions opens a perspective to its practical application in SFRs. (author)