Shot peening is widely used in automotive and aeronautic industries to improve fatigue life of metallic components. Its beneficial effects are mainly due to the residual stress field caused by the plastic deformation of the near-surface region resulting from multiple shot impacts. It is therefore important to know the values of the induced residual stresses in order to predict the mechanical strength of the peened component, and to know how these stresses vary by changing the shot peening parameters. The problem is that experimental measurement of residual stress is costly and time-consuming, and generally involves semi-destructive techniques. These difficulties make assessment of compressive residual stresses in real (industrial) peened components very challenging. On the contrary, numerical simulation can provide an alternative way to deal with this task. Consequently, several shot peening models have been developed in the literature. Although these models were successfully applied to investigate important physical phenomena encountered in shot peening, their application to assess residual stresses resulting from a real shot peening test is still not within reach. Indeed, due to computation costs and the complexity of the process, they cannot be directly applied to simulate a complete shot peening experiment. Development of a robust methodology allowing these models to properly simulate such an experiment at minimal cost (i.e. using simplifying assumptions) is thus needed. The present paper aims to meet this need. First, a new discrete-continuum coupling model combining the strengths of the existing shot peening models was developed. To avoid expensive computation times, only major shot peening features are included in this model. Then, a comprehensive methodology explaining how this model can be applied to simulate a real shot peening experiment was proposed. To validate the developed model as well as the associated methodology, they were applied to simulate a real shot peening experiment from the literature. Relatively good results were obtained compared to experimental ones, with relatively little computation effort.Read less <