In this experimental and numerical thesis we have studied the behavior of a granular assembly in two different situations. In the first part, a sand track is subjected to repeated passages of vehicles, under certain conditions a regular pattern of ripples appears spontaneously: this is known as the washboard road instability. This phenomenon, very common on dirt roads, is of course annoying for drivers but is also very dangerous because of the lack of adhesion it causes. We reproduced this instability with a laboratory scale set-up and also thanks to soft spheres molecular dynamics simulations. With these experimental and numerical tools we have highlighted some properties of instability. In particular, we have measured the dispersion relation of the pattern and shown that this phenomenon follows pitchfork instability. By measuring the stress acting on the vehicle we were then able to build a linear stability analysis that predicts quantitatively the threshold of the instability and the wavelength of the pattern. Finally, we have studied more complex cases where several vehicles were simultaneously on the track. We have also considered the case where the track was wet and therefore cohesive. In a second part, we have studied the behavior of a silo filled with grains subjected to temperature variations. A slow creep motion of the grains in the column is observed. Two flow regimes are observed according to the amplitude of temperature cycles. We obtained an expression of the critical temperature between the two regimes as a function of the microscopic parameters of the grains (elasticity and surface roughness). We have also studied the motion of each individual grain in order to interpret the macroscopic dynamics of the pile.
