Dry and moist tamping methods are widely used for the investigation of sandy soil liquefaction. However, the effect of these two tamping-based reconstitution methods on composite soil liquefaction behavior have been rarely assessed. In this study, the dry and moist tamping methods were assessed using loose silica sand matrix (HN31) containing various amounts of non-plastic silt particles (C500), and by performing a series of undrained triaxial compression tests. Microstructural observations using a high-resolution optical microscope were also performed on loose silty sand specimens that had the similar initial states to those for the undrained triaxial tests. Triaxial results show that for silty sand specimens prepared by dry tamping method, a global dilatant behavior was observed and the undrained shear strength decreased as the fraction of silt particles increased. By contrast, a total reversed phenomenon was observed for those prepared by moist tamping method: all specimens showed more or less limited liquefaction behavior and silt particles mitigated the contractive tendency, which provided beneficial contribution to the undrained shear strength. Microstructural observations show that in the dry tamping method, overall structure formed by the host sand and silt particles was regularly deposited, promoting a more stable strain-hardening response. Moreover, some silt particles were found to be located near sand-sand contact points, significantly loosening the sand matrix. On the contrary, by using moist tamping method, the sand was uniformly covered by bulked silt particles, and the aggregated sand formed the metastable structure, facilitating the occurrence of static liquefaction. Moreover, with the increase of fines fraction, this metastable structure was weakened, characterized by the filling of large inter-sand voids, leading to the increase of undrained shear strength. Thus, a critical value of total fines fraction F_c^Tot equalling to 21% was determined, which theoretically verified the insignificant effect of specimen preparation methods on the undrained shear strength. In that case, the metastable structure in specimen prepared by moist tamping method was expected to disappear and the overall response of specimen changed from the dominance of host sand to that of fines.

干式和湿式夯实方法被广泛用于沙质土壤液化的研究。但是，这两种基于夯实的重建方法对复合土壤液化行为的影响很少得到评估。在这项研究中，使用包含各种量的非塑料淤泥颗粒（C500）的疏松硅砂基质（HN31）并通过执行一系列不排水的三轴压缩试验，评估了干式和湿式夯实方法。还使用高分辨率光学显微镜对松散的粉砂样品进行了显微观察，这些样品的初始状态与不排水的三轴试验相似。三轴结果表明，对于用干夯法制备的粉砂样品，观察到整体膨胀行为，并且随着粉砂颗粒比例的增加，不排水的剪切强度降低。相比之下，通过湿夯法制备的样品则观察到完全相反的现象：所有样品都或多或少地显示出有限的液化行为，粉砂颗粒减轻了收缩趋势，这为不排水的剪切强度提供了有益的贡献。显微组织观察表明，在干夯法中，由主体砂和淤泥颗粒形成的整体结构规则地沉积，从而促进了更稳定的应变硬化响应。此外，发现一些淤泥颗粒位于砂-砂接触点附近，从而使砂基明显松散。相反，通过湿夯法，沙子被堆积的淤泥颗粒均匀地覆盖，聚集砂形成亚稳结构，有利于静态液化的发生。此外，随着细粒含量的增加，这种亚稳结构被削弱，其特征是填满了大的砂间空隙，导致不排水剪切强度的增加。因此，总罚款分数的临界值确定的F _c ^Tot等于21％，这从理论上证明了样品制备方法对不排水剪切强度的微不足道的影响。在这种情况下，通过湿夯法制备的样品中的亚稳结构有望消失，并且样品的整体响应从主体砂的优势变为细粉的优势。