In the framework of surveillance program dosimetry, the main parameter in the determination of the fracture toughness and the integrity of the reactor pressure vessel (RPV) is the fast neutron fluence on pressure vessel. Its calculated value is extrapolated using neutron transport codes from measured reaction rate value on dosimeters located on the core barrel.
EDF R&D has developed a new 3D tool called EFLUVE3D based on the adjoint flux theory. This tool is able to reproduce on a given configuration the neutron flux, fast neutron fluence and reaction rate or dpa results of an exact Monte Carlo calculation with nearly the same accuracy. These EFLUVE3D calculations does the Source*Importance product which allows the calculation of the flux, the neutronic fluence (flux over 1MeV integrated on time) received at any point of the interface between the skin and the pressure vessel but also at the capsules of the pressurized water reactor vessels surveillance program and the dpa and reaction rates at different axial positions and different azimuthal positions of the vessel as well as at the surveillance capsules. Moreover, these calculations can be carried out monthly for each of the 58 reactors of the French current fleet in challenging time (less than 10mn for the total fluence and reaction rates calculations considering 14 different neutron sources of a classical power plant unit compared to more than 2 days for a classic Monte Carlo flux calculation at a given neutron source).
The code needs as input:
- for each reaction rate, the geometric importance matrix produced for a 3D pin by pin mesh on the basis of Green’s functions calculated by the Monte Carlo code TRIPOLI;
- the neutron sources calculated on assemblies data (enrichment, position, fission fraction as a function of evolution), pin by pin power and irradiation. These last terms are based on local in-core activities measurements extrapolated to the whole core by use of the EDF core calculation scheme and a pin by pin power reconstruction methodology.
This paper presents the fundamental principles of the code and its validation comparing its results to the direct Monte Carlo TRIPOLI results. Theses comparisons show a discrepancy of less than 0,5% between the two codes equivalent to the order of magnitude of the stochastic convergence of Monte Carlo results.
FRED: a fast Monte Carlo code on GPU for quality control in Particle Therapy
