Alkali-Aggregate Reaction (AAR) is a pathology of concrete, of chemical origin, leading to damage, swelling and cracking of the material. In a structure, such as dams or nuclear structures, this pathology can lead to major disorders and thus be the cause of a loss of the properties or a drop in its bearing capacity. It is therefore essential to be able to predict the behaviour of these damaged structures and to be able to understand and model the mechanisms of this pathology. The origin of the alkali-aggregate reaction is, in the first place, linked to the chemical attack of the aggregates by the basic solution of the concrete, and in particular by the hydroxyl and alkaline ions of the interstitial solution. As a result of this attack, the silicate ions thus released into the solution can react with the alkali and calcium ions to form new phases within the aggregates. The combination of the attack on the aggregates and the formation of these precipitates in the pores of the aggregates or in the porosity at the aggregate/paste interface leads to cracking of these constituents of the concrete and thus to damage to the material. In order to explain the differences observed for a concrete of identical formulation but containing aggregates of different mineralogical composition and to be able to predict them, we propose to account for the difference in the nature of the aggregates through two complementary characterizations: 1. Experimental characterisation: The objective will be to characterise as precisely as possible the properties of the aggregates: mineralogy, reactivity, porosity, transport properties and cracking during RAG. 2. Numerical characterisation: The objective will be to model the degradation mechanisms of the aggregate and at the 'paste/aggregate' interface, giving a detailed description of the sequence of phenomena: transport in the aggregates and the surrounding paste, dissolution, gel precipitation, internal cracking, crack propagation and expansion of the concrete.​