Many centrifuge model tests on saturated model ground have been carried out for investigation on the liquefaction phenomenon. However, there were few investigations in which the degree of model ground saturation prepared by various techniques was measured and the performance of the techniques was discussed. In the present study, the model ground preparation techniques for manufacturing fully saturated model ground were discussed. A model sandy ground was prepared by three various techniques and the degree of model ground saturation was estimated by measuring the P-wave velocity propagating through the ground. The degree of model ground saturation was greatly influenced by the type of preparation technique, and a saturated model ground prepared by percolating CO2 gas followed by fluid in a vacuum condition was recommended for manufacturing fully saturated ground. Then, the modeling of models tests on three different scaled models were carried out on the saturated model ground manufactured by the recommended technique, which demonstrates the repeatability and reliability of this centrifuge model tests for the simulation on prototype behavior.
　Recently, the grid-type ground improvement by the deep mixing method (DMM) has been applied for the liquefaction prevention. The grid made by treated columns is expected to function to restrict the pore water pressure generation between the grids due to the reduction of shearing deformation. The existing guideline on the grid spacing is pointed out to have limitations for practical aplications. In the present study, a series of dynamic centrifuge model tests and numerical analyses were conducted to investigate the effect of gird spacing on liquefaction prevention. The seismic ground response and the pore water pressure generation in the ground were measured in detail. A new rational design guideline on the grid spacing was proposed. Further, a new arrangement of grid type improvement was also proposed for more economical design.