Triaxial testing at elevated pressures has received an increasing amount of attention in recent years, not only for the measurement of design parameters for deep engineering structures, but also to gain a deeper understanding and a complete characterization of the behavior of many natural soils and soft rocks. The paper describes the design of three high pressure triaxial apparatus with capacities between 5 and 70 MPa, each of which is automated for full stress path capability and yet has an emphasis on economy of cost. The apparatus are also designed to measure stiffness from very small strains (0.0001%) to large strains (beyond 10%) over the whole working range of pressures. Examples of data for both soils and soft rocks are used to show the apparatus performance and to highlight the importance of testing over an extended range of pressures.

New methods of geotechnical design in sands are often based on a state parameter type of approach, requiring the in situ state of the soil to be quantified relative to its critical state line. One difficulty in the application of these methods is then the correct identification of the location of this line, particularly at lower stress levels where sands strain soften and strains tend to localize within the sample. This paper examines the shearing behavior of two sands of diverse mineralogy by means of triaxial testing. The approach adopted has been to use existing techniques and apparatus that might reasonably be implemented in general practice. Recommendations are made as to the type of apparatus and test that should be used, and the corrections necessary to the data in order to identify as accurately as possible the critical states.

Many advanced soil models rely on the current state relative to normal and critical state lines to describe soil behavior. The position of these lines, therefore, requires an accurate estimation of the specific volume or void ratio. A series of one-dimensional compression tests was performed both on a coarse and a fine grained soil to investigate the experimental accuracy of the initial specific volume. This was obtained comparing independent calculations of the initial specific volume that were based on redundant measurements of height and weight of the specimen, both at the beginning and at the end of the test. The redundancy in the measurements was a key factor to obtain independent calculations. It was found that the excess water, such as may be stored in the filter papers, was the main cause of inaccuracy, when gross errors did not occur. Two novel confining rings having a closed-base were designed to reduce this effect. Although this was possible for the coarse grained soils tested, the fine grained soils retained more water due to the higher suction at the end of the test and water adsorption could not be avoided. The assumption of saturation is shown to be far from accurate, meaning that both the bulk unit weight and the water content should be measured independently to obtain a reliable measurement of the specific volume. The specific volume of the intact soil was found to be less accurate than when reconstituted. The experimental scatter was compared with the theoretical accuracy obtained from the error propagation theory. Good agreement was found between the theoretical and experimental accuracy.