The effects of loading histories on the stress-strain behaviour of an artificially soft rock, compacted cement-mixed well-graded gravelly soil, in multiple-step loading (ML) triaxial compression (TC) tests were evaluated. Associated with an increase in the strain until the peak stress state with an increase in the confining pressure (σh) in single-step loading (SL) TC tests, all the loading steps in ML tests increasing σh, loading was terminated consistently during strain-hardening approaching the peak stress state with volume contraction, On the other hand, associated with a decrease in the strain until the peak stress state with a decrease in σh in SL TC tests, in ML tests decreasing σh, at the first and second loading steps, loading had to be terminated far before the peak stress state during strain-hardening. At the third and subsequent steps, in particular at the last step, the stress-strain relation quickly entered into the strain-softening regime exhibiting large dilation while significant effects of damage by preceding loading steps were observed. A comparison of the pre-peak stress-strain relation at respective steps of the ML tests with those during SL TC tests revealed that the stress-strain behaviour in the ML tests is affected by; 1) stiffening by cyclic loading due to elastoviscoplastic properties; 2) effects of σh on the stress-strain behaviour; and 3) damage effects that have taken place during preceding loading history, in particular those by previous shearing into the strain softening regime and unloading causing large negative shear strains. The shear strengths at different σhs could be determined based on results from relevant ML tests using a single specimen.
To evaluate the Mohr-Coulomb failure criterion (the M-C failure criterion) by using a single undisturbed sample of soil or rock, multiple-step loading triaxial compression test (ML TC test) consisting of a series of consolidation and shearing steps has been employed [1–2]. A series of ML TC tests on undisturbed samples from a given site can provide a spatial variability of soil or rock strength represented by the M-C failure criterion. Despite the above, details of loading histories comprising a sequence of consolidation and TC loading/unloading at different confining pressures are usually poorly stated by different researchers. For example, Kim and Ko (1979)  reported only the relevance of a ML test method on the stress-strain property of a certain rock type. Akai et al. (1981)  found that, in a number of ML TC tests on siltstone and tuff, it often became difficult to continue TC loading, because, soon after the specimen state became near failure, the specimen exhibited abrupt post-peak strain-softening before increasing the confining pressure. They controlled the lateral strain during TC loading to avoid abrupt failure. However, effects of the details of the test procedure were not studied. In the studies by Cain et al. (1986)  and Crawford and Wylie (1987) , volumetric strain was used to detect imminent failure at respective TC loading steps in ML test. However, the effects of precedent loading histories were not reported.