Tensile failure appears widely around underground tunnels and openings. Unfortunately investigations on tensile deformation of rocks are much less popular than those on compression. Test results from quite a few authors have shown that for some rocks, Young's modulus in tension, ET, is much smaller than that in compression, EC. As a consequence, large errors may arise in the stress values obtained from traditional methods for such rocks.
Test results were presented on deformation behaviour of several rock types in both direct tension and compression. It was found that for the rocks of which ET « EC, the deformation in tension is nonlinear and plastic though their behaviour in compression is approximately linear and elastic.
Two examples are presented to show how errors would appear by using a linear bi-elastic model to a structure where the deformation in tension is plastic. These examples show that plastic behaviour in tension has great Influence on stress values and failure. Therefore, further investigations on tensile deformation are necessary to be conducted, for more rocks.
It is assumed in most analytical formulas and numerical techniques, such as finite element, boundary element and distinct element methods that the values of Young's modulus and Poisson's ratio in tension are equal to those in compression, although some authors have reported that this assumption is not correct for many rocks (Haimson et al. 1974).
Hawkes et al. (1973) conducted some tests of direct tension and compression with three rock types. Their results show that, the ratios of the Young's modulus at 50% of failure stress in compression and the modulus in tension were respectively 9, 2 and 1.
Stimpson and Chen (1993) made some uniaxial tension/compression load tests with different rocks. Their results show that the ratios of average Young's modulus under direct tension, ET, and the modulus under compression, EC, equal 0.5, 1.0,0.7 and 0.3–0.4 In four rock types.
Jaeger et al. (1976) concluded that for the majority of rocks, the value of Young's modulus in tension is generally less than that in compression.
Results obtained by Sundaram et al. (1980) with finite element method shown that errors may cause by calculating the stress values of the rock sample of Brazilian tests by using traditional formula. When EC/ET ratio is 2, the error is over 10%; When EC/ET ratio is 10, the error is as high as 40%. Similar results also appear in the condition of beam (Timoshenko, 1958), hydraulic extension test (Haimson et al. 1974) and ring test (Chen et al. 1993).
Unfortunately, published data on the deformation of rocks under tension are much less than those under compression. Today, the knowledge on tensile deformation behaviour of rocks still needs to be developed.
A new test frame was first developed with which direct tension tests could be performed followed a compressive phase at the same rock specimen. With this way, the error caused by different rock samples of the tests in compression and tension can be avoided.