Four direct shear (DS) apparatuses having different sizes with the specimen lengths ranging from 40 to 800 mm were constructed in the study. The vertical and shear stresses acting on the shear zone were measured as accurately as possible confirming its importance. Noticeable effects of specimen shape were observed. The effects of specimen size were evaluated by performing constant pressure DS tests on a fine poorly graded sand (Toyoura sand) in the small, semimedium, medium and large DS apparatuses and a well-graded sandy gravel in the medium DS apparatus. The residual shear strength of Toyoura sand was independent of the specimen size and initial density. Due likely to specimen size effects on both progressive failure and boundary mechanical restraint, the peak strength decreased with an increase in the specimen size. As the specimen size increased with dense Toyoura sand and as the particle size increased in the medium DS tests, the shear displacement at the peak stress and the ultimate volume increase at the residual state consistently increased while the postpeak strain softening became slower. These specimen size effects can be attributed to the thickness of shear zone and the number of shear bands included in the shear zone.

Laboratory triaxial compression tests were carried out to investigate the mechanical behavior of dense sand and geogrid-reinforced granular soils. The tested sand having its mean particle size (D50) equal to 0.6 mm was adopted. Three geogrids with different longitudinal and transverse nominal strengths were used. The dimensions of the cylindrical soil specimen were 70 mm (diameter) × 160 mm (height). The relative density was equal to 70% for all tests. The reinforced sand specimens with one or two geogrid layers were sheared under effective confining pressures (σ′3) equal to 50 kPa. The test results of unreinforced sand indicate the general stress-strain behavior of dense sand when sheared, whereas the deviatoric stress reaches its peak value, after which it gradually decreases to ultimate value (σ1 - σ3)ult. The difference of effective confining pressure indicates that the peak of deviatoric stress Δσd = (σ1 - σ3) increases with the increase in effective confining pressure (σ′3), while the peak principal stress ratio (σ′1/σ′3) decreases with the increase (σ′3). The friction angle (ϕ′)and cohesion (c′), defined by analytical and graphical methods for unreinforced sand. Geogrid as reinforcement increasing peak shear strength. The increasing peak shear strength is more pronounced with a higher number of geogrid and the geogrid with higher stiffness. Increased in confining stress inside reinforced soil mass (Δσ3R) can be interpreted by cohesive reinforced soil (CR).

Reinforced sand shows larger dilatancy associated with a larger volume of shear zone in direct shear tests, compared to unreinforced sand. It is likely that multiple shear bands develop forming a thicken shear zone, resulting in larger dilatancy. Air-dried sand specimens reinforced with phosphor bronze strips installed perpendicularly to the shear direction were tested by using a medium scale direct shear test apparatus. The test results were analyzed with a new mechanical model using estimated thicknesses of shear zone as one on the major input parameters.

To clarify the influence of boundary insulation on the development of frozen soils, seven artificial ground freezing (AGF) models of Ottawa sand were performed with different thermal impedance Z-values of the boundary insulation system. Results of testing show that the greater the Z-values, the larger the frozen soil areas for a given time of freezing. When the Z-value is more than 2.27 m2 K/W, however, the extent of frozen soils appears to be stabilized for the AGF model with dimensions of 100 cm (L) by 100 cm (W) by 15 cm (H). Two-dimensional finite element analyses were also conducted to verify the performance of boundary insulation systems of the physical model. Numerical simulations show the development of frozen soils is significantly affected by the ambient temperature if no insulation is covered on the boundary of the AGF model, and the growth of frozen soils would cease at a freezing time of less than 6 h. However, when the Z-value of the boundary insulation system is more than 2.27 m2 K/W, the pattern of the temperature field and the size of frozen soils in the model resemble those for a system with a perfectly insulated boundary with an elapsed time of freezing up to 24 h. Hence, it can be concluded that the thermal impedance Z-value should be at least 2.27 m2 K/W for the boundary insulation system of the current AGF model to be free from the influence of an ambient environment and have a similar temperature response as the perfectly insulated model for a test time up to 24 h.