The study of subduction megathrust earthquakes, like the 2004 Sumatra-Andaman or the 2011 Tohoku earthquakes, generate heavy economic and human losses and represents, then, a major societal and scientific challenge. However, several limiting factors, related to the difficulty to analyze deformation offshore, to access deep source of earthquake and to integrate the characteristic time scales of seismic processes, must be overcome first. With this aim, we have developed an experimental approach to complement numerical modeling techniques that are classically used to analyze available geological and geophysical observations and measurements of subduction earthquakes. Our main objective is to validate a kinematically and mechanically first order scaled 2D and 3D analog model of a subduction zone capable of reproducing megathrust earthquakes and realistic seismic cycle deformation phases. By studying model kinematics and mechanical behavior, we expect to improve our understanding of seismic deformation processes and better constrain the role of physical parameters (fault friction, rheology, ...) as well as boundary conditions (loading rate, initial state of stress ...) on seismic cycle dynamics and megathrust earthquake dynamics. Another interesting issue is the potential use of 3D model kinematics as input data for tsunami numerical modeling works to improve our knowledge on how tsunami waves are generated. We also expect that results of this project will lead to significant improvement on how geophysical data, satellite observations and seismological records can be better interpreted.