ABSTRACT: In order to reexamine the strain softening behavior of rocks, sandstone is tested using a stiff testing machine. The results show that the strain softening behavior is not an intrinsic property of rock materials but rather depends on the structure of the specimen used in the tests. A model based on the test results reveals that brittleness in the strain softening region is related to a difference between the failure and residual strengths, which does not always increase with confining pressure.
1 INTRODUCTION
The subject of stability of underground structures, especially rock bursts, is important in both mining and civil engineering. Stiff testing machines have thrown light on the strain softening behavior of rocks with regard to the stability. One of the characteristic results obtained from experiments using stiff testing machines is that the negative slope of the stress-strain curves in the softening region decreases with an increase in confining pressure(Jaeger & Cook 1979). However, stress-strain curves whose negative slope gradually increases with confining pressure are also found in some published papers (Kinoshita et al. 1975, Gowd & Rummel 1980, Farmer 1983). In this paper, firstly, conventional triaxial tests performed to investigate the strain softening behavior of rocks are described. Next, a model based on their results is constructed and the effect of confining pressure on the softening behavior is discussed.
2 EXPERIMENT AND ITS RESULTS
Ainoura sandstone from Nagasaki prefecture, Japan was tested under a conventional triaxial condition, using the stiff testing machine with a servo-control system. Cylindrical specimens(diameter 50mm by length 100mm) were prepared. A displacement transducer was set up outside of the triaxial cell for measuring the axial displacement, which was controlled by using as a feed-back signal and from which the apparent axial strain e1a was estimated. Simultaneously two cross strain gauges were attached to the surface of the specimen to measure local strains: the axial strain e1 and lateral strain e3. Since compressive stress is considered positive in this paper, the axial stress corresponds to the maximum principal stress s1. Figure 1 shows the differential stress-apparent axial strain curves which were obtained from monotonic loadings at confining pressure s3 from 0 to 100MPa. Although the effect of confining pressure on the failure and residual strengths is as previously mentioned, it is noticed that when confining pressure exceeds 30MPa compared to a uniaxial condition or lower confining pressures the strain softening behavior cannot be stably controlled because of its brittleness. This interesting behavior which is not concerned with a failure mode of splitting but with that of shearing will be discussed later, in detail. The differential stress, apparent axial strain, local axial strain and local lateral strain obtained from cyclic loading at confining pressure of 50MPa are represented with loading time as the axis of abscissas in Figure 2. Locations of the cross strain gauges attached to the specimen and a fracture plane formed by shearing are also shown in this figure. After peak stress is attained, the local strains do not necessarily act according to the apparent strain but rather to the differential stress.
