This work is include in the national project SOLCYP. This project, support by ANR, focus on the observation and modeling of behavior of pile under cyclic loading with large number of cycles (about one million). This type of solicitations leads to an important change in the internal state of the interface soil/pile and classic models aren't adapted to deal with this issue. Here, internal state of material is taken into account with a number of cycle limited (easier to study).In geometrical, interface issue is pretty difficult, mainly due to the fact that mechanical interface can be shifted from the geometrical contact to the mass of soil. In extreme ways, concrete pile will have a behavior mainly linked to the behavior of soil when steel pile behavior can be assessed with coulomb friction.Two internal variables have been identifies: the first one can be regarded as isotropic and can be easily assessed when the second one is considered as anisotropic harder to quantify. Generally, just the first variable is taken into the account to forecast the evolution of material properties and behavior. But this is no more than an approximation. In fact, anisotropic variable has an action as well as the isotropic one. But, depending to the type of solicitations, these actions can be strongly linked or really uncorrelated. Evolution of parameters set is given depending these variables.This work has made possible the design of two models: the first one is a mass soil model. It needs 20 parameters, identified from four triaxial and isotropic tests, and is able to reproduce the soil behavior under numerous loading paths (monotonic, cyclic, drained, undrained) and different initial states (density, mean pressure...). The second one is an adaptation of the soil model for rough interface issues. It needs 14 parameters and is also able to forecast interface behavior under different initial state and limit conditions. Both, confirmed for some dozens of cycles, will be used in the large number of cycles issues formulation.
