Résumé On propose une étude expérimentale du comportement d'un enrobé bitumineux. L'appareillage utilisé, qui est rapidement décrit, permet d'effectuer des essais triaxiaux de révolution régulés en température. Les résultats confirment que le comportement de ce matériau, qui comprend une composante visqueuse très marquée, est qualitativement le même pour les différentes valuers des variables de sollicitation: vitesse de déformation axiale, température et pression de confinement. Différents paramètres mécaniques ont été analysés. On précise l'évolution de ces paramètres, au nombre de 11, en fonction des variables de sollicitation A partir des résultats présentés et de ceux obtenus lors de campagnes d'essais antérieures, on propose un critère de contrainte maximale (appelée généralement contrainte de rupture). L'obtention d'un tel critère, pour les enrobés bitumineux, est nouveau.

Summary An experimental study of the behaviour of bituminous mixes is presented. The tests were performed using a triaxial apparatus, with temperature regulation, developed in our laboratory. First, two classes of cyclic stress path are analysed. The evolution of 11 mechanical parameters characterizing modulus, particular stresses, and volume variations which are obtained from the experimental data are linked with the chosen variables: temperature, confining pressure and rate of the axial strain. These variables are kept constant during each test. The reaction of the material is different for each value of the selected variable and the shape of the curves is the same as that obtained for non-cohesive granular materials. We observe, in particular, during compression, a contraction phase followed by an increase in the volume of the sample. The effects of the strain rate show the viscous behaviour of this kind of material. The values of the maximum stress are also studied. Following two realistic hypotheses, a maximum stress criterion (generally called the failure stress criterion) is given. The first hypothesis is linked with the conical shape of the governing formulation. It appears as a 3D extrapolation of our experimental data. The second hypothesis assumes that the material is isotropic. It was verified experimentally for compression and extension tests, on sample cubes in various directions. These tests show a slight anisotropy, which is neglected. Nevertheless, a method is given to take it into account, considering rotation of the cone. The criterion developed involves two cones, each with the trisectrix as axis and of equilateral triangular section extending into the principal stress space. The first cone arises when the stresses are positive (compression), and the second one when at least on stress is negative. The intersection of the two cones is on the coordinate planes. For a given temperature, this formulation contains five constants, and three of them define the strain rate effect. Such a criterion is, to the best of our knowledge, new.