In tribology, the numerical modeling has become an indispensable tool for studying a contact to overcome the experimental limitations. To have a better understanding of the phenomena involved, the models are no longer located at a single scale, but involve several ones, more than ever, making the concept of tribological triplet as a unavoidable concept. Working with this philosophy and approach based on the Non Smooth Contact Dynamics framework, which we remind some outlines, we propose to cross two steps~: model that can offer quantitative results and that implement the first ingredient to perform a homogenization at a contact level. In the first case, the study of coupling finite elements/discrete elements within the same simulation aims to propose models that are more "realistic". Even if the interface is already present in the contact and not going to evolves, it can highlight the use of measurement tool of spot particles via dynamic instabilities and allows to have not only qualitative results but also quantitative ones since the comparison with the experimental strain rates are in very good agreement. In the second case, the study of VER in tribological charges is performed to extend the homogenization techniques to contact problems in order to overcome the interface description on large scales by finding a way to homogenize the heterogeneous behavior of the interface and make a dialogue with the continue behavior of bodies in contact by send up, in a sense, average values of the microscopic scale to the macroscopic scale and in the other sense, use local data of the macroscopic scale as boundary conditions at the microscopic scale.
