[en] For the original design of SUPER-PHENIX, only pool fires were analysed for secondary sodium because these were thought to be the most likely. However, after the sodium spray fire at the solar plant of ALMERIA, an analysis of the consequences of secondary spray fire was undertaken. According to the French Safety Authority, the most penalizing cases of sodium leak and fire must be taken into account for each type of consequences, up to the complete rupture of a main secondary pipe. The experimental data available were mainly based on sodium flowrates in the range of ten kilograms per second, which are far below the leak flowrates obtained in case of a complete rupture of a main secondary pipe, i.e. several tons of sodium per second during a short time interval; moreover, it was obviously not possible to perform sodium tests with such high flowrate conditions. Consequently a complete methodology for the prediction of the behaviour of large sodium spray fires has been developed: the two-dimensional code PULSAR, which solves the two phase flow Navier-Stokes equations with source terms of mass and energy, is first used to evaluate the physical behaviour of a spray of sodium droplets in a cell in diverse conditions and thus to determine the burning rate. This last value is then used as data in the FEUMIX code in which other phenomena such as the dynamic response of pressure relief systems are described, in order to determine the pressure transient in the cell. This approach has been successfully tested using the experimental data available from past and recent tests, particularly the high flowrates tests IGNA 3602 and IGNA 3604. This numerical approach has been applied to the analysis of the consequences of postulated large sodium leaks in SUPER-PHENIX and allowed us to justify the hypotheses used to design the protective measures implemented on the plant, and thus the demonstration of safety with regard to large sodium leaks. (author)