Accident studies for the Gas-cooled Fast Reactor (GFR) in support of fuel safety studies

The Gas-cooled Fast Reactor (GFR) is one of the six systems selected by the Generation IV International Forum and is also studied by the French Commissariat a l'Energie Atomique et aux Energies Alternatives (CEA), in parallel of the Sodium-cooled Fast Reactor (SFR), as its long term alternative. This internship deals with the study of severe accidents in the frame of the preliminary design. The objective is, for some unlikely severe accidental situations, to prevent a core degradation extension and to limit its consequences. The internship has consisted in adopting a multidisciplinary approach, for some accidents, by coupling the results of 'monodisciplinary' calculation codes with simplified analytical models, as a supplement of previous studies. Among these severe accidents is the Unprotected Loss Of Coolant Accident (ULOCA) which appears to be difficult to control for long times, partly because of a penalizing neutronics effect of nitrogen injection, modeled during transient calculations by a reactivity insertion and a reduction of the coolant void effect coefficient. A strategy of accident control could be foreseen but should not be enough to maintain the clad temperature below 1850 deg. C (liquid formation temperature) longer than 1h20 after the accident beginning. Air ingress transients (breaks on the primary circuit and on the close containment), with scram actuated and nitrogen injection, have been studied regarding SiC oxidation. A simplified analytical model, using CATHARE2 calculations results, has been developed to estimate the silica layer growth rate and the possible transition towards an active oxidation mode synonymous with clad degradation. Studied situations (10'' primary break, rupture of the cross-duct...) do not seem to be able to severely damage the cladding. The application of Eremeev diffusion model at the fuel/liner interface is a method to follow the kinetics of the tungsten liner degradation and then the risk of losing the confinement of gaseous fission products function. While non-perfect, preliminary results show, in the example of an ULOCA (3'' break), an estimated W₂C layer growth rate of 1 to 2 μm/h, on condition that no liquid phase appears. With the given hypothesis and data, the diffusion process at the fuel/liner interface doesn't seem to really damage the liner, but avoiding a liquid phase formation at 1850 deg. C is critical. Severe accident scenarios of core melting and corium boiling have been studied with a simple analytical model coupling neutronics, heat transfers and thermodynamics. The study has allowed to calculate first orders of magnitude (kinetics of melting, boiling and pressure rising) and should ideally be improved with a more detailed modeling of physical phenomenon. (author)