Contribution to the modelling of solid state sintering

This thesis deals with the simulation of the sintering of nuclear fuel on a pellet scale. The goal is to develop numerical tools which can contribute to a better understanding of the physical phenomena involved in the sintering process. Hence, a multi scale approach is proposed. First of all, a Discrete Element model is introduced. It aims at modeling the motion of particles on a Representative Elementary Volume scale using an original Discrete Element Method. The latter is a Non Smooth Method called Contact Dynamics. Recently, there have been numerous papers about the simulation of sintering using Discrete Element Method. As far as we know, all these papers use smooth methods. Different studies show that the results match well experimental data. However, some limits come from the fact that smooth methods use an explicit scheme which needs very small time steps. In order to obtain an acceptable time step, the mass of particles have to be dramatically increased. The Non Smooth Contact Dynamics uses an implicit scheme, thus time steps can be much larger without scaling up the mass of particles. The comparison between smooth and non smooth approaches shows that our method leads to a more realistic representation of rearrangement. An experimental validation using synchrotron X-Ray microtomography is then presented, followed by a parametric study on the sintering of bimodal powders that aims at showing the capacity of this model. The second part presents a mechanical model on the sub-Granular scale, using a Finite Element method. This targets a better understanding of the behavior of two grains in contact. The model is currently being developed but the first results already show that some parameters like the shape of the surface of the neck are very sensitive. In the future, the Non smooth Contact Dynamics model of sintering may be improved using the results obtained by the sub-Granular scale mechanical model. (author)