Rock behaviour under multiaxial compression

Abstract

An experimental study has been carried out to investigate the behaviour of rock under multiaxial compression and assess the influence of both the stress conditions and test configuration on the apparent characteristics of this behaviour. Over three hundred specimens of Springwell sandstone, of various forms, have been tested using different loading techniques and most encountered stress fields. Cubes and thick-walled hollow cylinders have been subjected to uniaxial, biaxial, triaxial and polyaxial compression, and solid cylinders have been subjected to standard uniaxial and triaxial compression. Extensive work has initially been put into designing and developing the testing facilities required. A new multiaxial hollow cylinder test apparatus has been devised using a Hoek triaxial cell and specially designed system for the application of internal pressure, major modifications have been made to an existing multiaxial cubical test apparatus, and appropriate testing arrangements and procedures have been developed. Prior to initiating the experimental programme, characterisation tests have been conducted to determine the fundamental properties of the rock, and non-destructive ultrasonic wave velocity tests have been utilised together with statistical methods to examine any inherent variations in the specimens used. A remarkable agreement has emerged between the rock static and dynamic Young's moduli and the results have confirmed that the Springwell sandstone can practically be regarded as linear elastic, homogeneous and isotropic. The concept of the multiple failure state triaxial test has been utilised and extended to conduct multiple failure state polyaxial tests. While the concept remains useful, strain results obtained beyond the first failure state are likely to be inconsistent. Results of cube tests have been found to be highly influenced by the boundary conditions. Although the use of PTFE sheets can reduce the effect of friction between the specimen and the loading platens, it may equally have a weakening effect on the test specimen. The theory of elasticity has been found adequate to calculate the stresses in the hollow cylinders but remains inexact when deviation from linear behaviour occurs prior to failure. When the outer and inner surfaces of the cylinder are not perfectly concentric, the effect on the test results has proved negligible. Hollow cylinders have been found to provide an alternative means for measuring the rock indirect tensile strength. Results of biaxial and polyaxial tests on both cubes and hollow cylinders have confirmed the marked influence of the intermediate principal stress on the apparent strength of rock. Comparison of results from multiaxial tests on cubes, hollow and solid cylinders have shown that the apparent strength, deformability and failure characteristics of the rock are remarkably influenced by the stress conditions imposed as well as the test configuration used. Available failure criteria have advantages and disadvantages, but none of them can explain the diversity of the results obtained. On the whole, the results appear to cast doubt on common conceptions of rock failure and ultimately pose the question of how realistic current testing techniques are in their prediction of the actual behaviour of rock.