Review and assessment of sand sample preparation techniques from both theoretical and experimental viewpoints are presented. Sample densities obtained by air pluviation are shown to be sensitive to rate of pouring and drop height. Terminal velocity is reached at a very small drop height, and homogeneous samples of the same initial density tend to be formed by pluviation of uniform sand in water. Uniformly dense samples obtained by vibration of loose pluviated samples show no detectable difference in behavior when compared to samples densified by control of drop height only. Effective confinement during densification by vibration appears to prevent formation of a loose top layer. A loose top layer in an otherwise dense sample leads to a marked decrease in liquefaction resistance. Preparation of triaxial sand samples by pluviation in water is recommended because it results in initially saturated specimens, and homogeneous samples of desired densities can be replicated without difficulty.

Tests performed in a permeameter have been used to evaluate the gradient ratio test method. The evaluation is based on observed variations of water head and permeability in the composite soil and geotextile specimen with respect to time and with imposed hydraulic gradient. A water pluviation technique was used to reconstitute homogeneous, saturated sand specimens in the permeameter at any targeted density. An energy dissipator was used to prevent disturbance of the top of the specimen by inlet water at large flow rates. Some physical clogging of the soil may result from a blinding action due to fines in the recirculating water. Biological clogging may also occur, but is eliminated by treatment of the water with an algicide.

An experimental study aimed at a direct comparison of the undrained behavior of sand using specimens reconstituted by different techniques is presented. It is shown that at identical initial void ratio and effective stress state, the moist-tamped sand is potentially liquefiable, but in the water-deposited state may even be dilative. Water-deposited specimens are shown to be very uniform in contrast to the large nonuniformities that usually occur on moist tamping, rendering their results questionable from the standpoint of laboratory element tests. A direct comparison of the behavior of truly undisturbed sand specimens retrieved by in-situ ground freezing and their corresponding reconstituted counterparts after consolidating to identical initial states is also presented in support of the contention that the fabric that ensues on water pluviation closely simulates that of the natural alluvial and hydraulic fill sands, enabling the use of reconstituted specimens as substitutes for the expensive undisturbed frozen specimens for material characterization.

Fundamental assumptions necessary for assessment of volume change because of membrane penetration in triaxial tests on granular soils, by methods that use dummy rod inclusions or assume isotropic behavior of sand in loading, have been critically examined and shown to be invalid. Reliable estimates of membrane effects are essential in the development of constitutive models of soil behavior. Alternative methods, which do not violate any critical assumptions, are proposed for estimating membrane penetration and supported by experimental evidence.

The gradient ratio test is used to assess soil-geotextile compatibility in filtration applications. Experience gained in performing the test is summarized for ten soils and four non-woven geotextiles, and the potential for application of the test method is demonstrated. Details of a new, modified test device are then presented which address a perceived limitation of the existing configuration of the test device. Results of a preliminary series of tests, performed to commission the modified gradient ratio test device, suggest the filtration behaviour of the geotextile is essentially independent of normal stress.

A combination hydraulic-pneumatic loading system for triaxial testing of soils under a variety of loading modes is described. The system permits both monotonic and cyclic loading under stress or strain control. Anisotropic consolidation and stress path control during monotonic or cyclic shear are important features of the system. A combination stress- and strain-controlled loading on the same specimen can also be carried out. Typical test results are presented in support of the capabilities of the loading system.

Stress nonuniformities across the wall of hollow cylindrical torsional shear (HCT) specimens under generalized stress conditions are assessed taking into account the nonlinearity in soil behavior. The domain of stress space that results in acceptable levels of nonuniformities is thus delineated for a given specimen geometry. It is shown that previous analyses assuming linear elastic soil grossly overestimate stress nonuniformities in HCT specimens. The HCT device is thus suitable for investigating soil behavior under generalized stresses over a much larger domain of stress space than that thought earlier.

Equations based on elastic thin cylindrical shell compression theory are developed for correcting stresses for membrane strength in a triaxial test. The corrections are expressed in terms of the unstretched dimensions, the diameter and thickness of the membrane, the axial and volumetric strain of the cylindrical membrane cavity referenced to its unstretched dimensions, together with E and v of the membrane rubber. The development specifically considers changes in membrane thickness as they occur during the straining process, and in particular its influence on the specification of E of the membrane. Direct experimental verification of the derived stress correction is provided in support of their validity.

Strain path control during loading in addition to the often-desired stress path control is shown to constitute a versatile method of characterizing stress-strain response of soils in the laboratory. It allows probing stress or strain increments to be applied in any direction at an ambient effective stress state to study the stress history and stress-path-dependent response. The capture of even the potential strain softening and the post peak behavior is made possible with no difficulty. Some unique capabilities of a triaxial and a multiaxial hollow cylinder torsion apparatus of these types are demonstrated by results of tests on a saturated sand.