[en] The theoretical work follows two main lines: A. Code development; B. Theoretical work on fragmentation. Two computer codes have been developed. The first code contains a heat transfer model (during the vaporization phase) based on the inverse Leidenfrost phenomenon (which has been observed experimentally in water). The exact solution of the heat diffusion equation in a sphere is included in the code. The code accounts for the time history of each fuel particle by means of specially averaged temperature values. The presence of fission gases can also be taken into account. A size distribution of fuel particles has also been incorporated in the code as well as the effect of the friction due to the channel walls and that of the pressure losses at channel outlet. An extensive parametric study has been carried out with this code. The main conclusions are the following: 1. Total mechanical work strongly decreases with the fragmentation and/or mixing time constants. 2. Vapour blanketing during the vaporization phase is effective only if accompanied by a relatively slow process of fragmentation and mixing. In this case total mechanical work strongly decreased with degree of vapour blanketing. 3. Total mechanical work rises with initial length of sodium piston. 4. Time to empty the 120 cm long channel is 15-20 msecs. for values of the fragmentation and/or mixing time constants of the order of 5-10 msecs. 5. Effects due to particle size distribution and gas content are important only fora rapid fragmentation and mixing process. It must be painted out that (as far as the gas is concerned) this conclusion is valid only within the limits of the effects (due to the gas) which have been considered in the model. Propagation effects can be analysed by using the second code. The interaction region can be subdivided into an arbitrary number of sections, each containing fuel and coolant. The thermal conductivity of the liquid sodium has also been taken into account, as well as the presence of non condensable gases. For the heat transfer process an effective thermal conductivity of the liquid/vapour mixture accounts for partial vapour blanketing of the fuel particles. Numerical results are in general consistent with those obtainable from the first code. A survey of the various fragmentation mechanisms of UO₂ in sodium was done. It was concluded that 'vapour bubble and collapse' seems to be the most probable fragmentation mechanism in the case of UO₂/sodium systems. For this reason work was focussed on modelling this mechanism. A preliminary model was given. Work is progressing. It is planned to couple the FCI theoretical model No. 1 developed at Karlsruhe to the american REXCO code which is available at Ispra. Work is being completed. The FCI subroutine is already available