Liquefaction resistance of sand deposit improved with sand compaction piles

This paper reports results of in-situ tests and undrained cyclic shear tests on high-quality undisturbed samples obtained by the in-situ freezing method at three sites where foundation soils had been improved with sand compaction piles. Triaxial specimens obtained from the frozen samples were fully saturated in the triaxial cell and then subjected to cyclic loadings. In the current design practice of ground improvement with the sand compaction piles, the SPT N-value at a mid point of a rectangular area surrounded by four adjacent sand piles, at which the N-value, and thus the liquefaction resistance is considered to be the smallest, is set as a target for the degree of compaction. It was found, however, that distributions of N-values and N-d-values in a horizontal plane at a certain depth appeared to be rather random. This suggests that the N-value at the center of a rectangular area does not always provide a conservative evaluation. The relationship between liquefaction resistances and the N-value, which was developed based mainly on field evidences of earthquake-induced liquefaction of natural soil deposits and reclaimed lands, compared quite well with that obtained from tests on fully saturated high quality specimens and mean values of N-d at several locations in the improved sands. Degree of saturation of the frozen sample was revealed to be considerably low, in the range between about 70 and 90%. This fact indicates that the liquefaction resistances of improved sands are significantly higher than those obtained from the relationship which is available only for fully saturated soils.