Multistage (i.e. multiple confining pressure) triaxial testing on one rock specimen aims to produce a peak strength criterion for the material whilst avoiding the need to perform several single-stage tests on separate samples. Notwithstanding the economy this brings, there is the question whether the damage that is progressively induced as the specimen is loaded to a critical stress at each stage could result in the multistage test underestimating the strength of the material. To explore this, we use the local degradation approach (Fang & Harrison, 2002) to firstly replicate single-stage and customary multistage triaxial testing, and then perform a hypothetical test in which the confining pressure is increased before damage occurs. This modelling confirms that the established multistage test does underestimate the strength of rock, and suggests how significant improvements may be made by developing and adopting a new testing procedure.


Knowledge of the axial stress – axial strain behaviour of rock as a function of confining pressure is of central importance in rock engineering, particularly as it allows establishment of a peak strength criterion for the material. Most such criteria are given as functions of the peak major stress sustainable for a given value of minor principal stress.

To both characterise and quantify a peak strength criterion, testing of cylindrical specimens in triaxial compression is commonly undertaken. The simplest approach is to prepare a number of specimens, and test each one to identify the peak axial stress (i.e. major principal stress) at a particular value of confining pressure (i.e. minor principal stress). However, this is costly both in terms of material requirements and time required for specimen preparation and testing. It also introduces possible errors associated with inter-specimen variability due to heterogeneity. To overcome these difficulties, multistage testing has been developed (Dolinar et al., 1982; ISRM, 1983). This uses a single specimen that is compressed axially under a sequence of increasing values of confining pressure, with the next higher confining pressure being applied once peak strength has been attained — the so-called ISRM Type II test.

Although there are obvious advantages to the methodology, particularly in terms of reduced cost and time required, it is long been known that rock sustains significant microstructural damage prior to reaching peak strength (Hallbauer et al., 1973; Lockner et al., 1991). The effect of this on the ability of multistage testing to produce peak strength parameters seems not to have been subject to rigorous analysis. Here, we apply the local degradation approach (Fang & Harrison, 2002) to investigate the discrepancy in the strength properties obtained by multistage and single-stage testing, and show how the multistage test could be modified to reduce the errors.


Even within intact heterogeneous rock, cracks are found throughout the material in the form of flaws and defects in between grains and particles, and mechanical breakdown of rock occurs through the exploitation and propagation of these (Yuan & Harrison, 2006).

This content is only available via PDF.
You can access this article if you purchase or spend a download.