Deep Mixing Method, a deep in-situ admixture stabilization using cement slurry, has been developed in Japan to reinforce soft alluvial clays. In practice, treated soil columns are over-lapped each other to manufacture a huge improved ground in-situ. Wall type improved ground which consists long wall and short wall, is comparatively economical than block type. Due to the large difference of the engineering characteristics between treated soil and untreated soft soil, treated soil
mass is assumed to behave as a rigid structure buried in soft ground. But because interaction between trated soil and soft ground within the improved ground is not exclusively studied, the wall type improvement is not widely applied in-situ.
　In this study, the interaction between soft soil and wall type improved ground is investigated by centrifuge model tests. In the model tests, improved ground is subjected to back filling to cause sliding failure in 50g acceleration field. The model tests were carried out changing the stiffness of the ground overlaid by the improved ground and filling speed. The test results are compared with the current design procedure which is based on the block type improved ground. It is found in the study that the current design procedure can evaluate well to sliding failure of the wall type improved ground.
　In this paper, investigation of stresses acting on the surface of the improved ground is described in detail as well as the procedure of centrifuge model testing.