The measurements of soil parameters relevant to tunnelling in clays: Discussion

Abstract

The authors have described a valuable series of tests in which they have attempted to simulate the stress paths in the vicinity of a tunnel-the stress paths having first been determined by means of an elastoplastic analysis. At first sight the approach seems eminently reasonable; however, it carries with it the assumption that the failure envelope can be obtained from standard tests. Burland and Fourie (1985) have described a simple method for failure in the walls of a shaft or tunnel, or in the zone in front of an excavation retaining wall, involves a rather special stress path. As illustrated in Fig. 1 for an overconsolidated soil, once the state of a soil element reaches failure it will travel along the failure envelope towards the origin. The authors show exactly the same behaviour in their Fig. 1 for three typical soil elements. Burland and Fourie have designated such stress paths passive or active stress relief paths. The question arises: Are we entitled to assume that such or active stress relief paths coincide with the failure envelope from routine tests on which the paths rise up to a failure line but do not travel along it? The problem of what strength to adopt for a condition of stress relief occurred during the design of the embedded cantilever retaining walls for a cut and cover tunnel at Bell Common in London clay (Hubbard et al. 1984). Since there appeared to be no information on this problem a conservative aphroach was adopted in which the wall was designed to have a factor of safety in excess of unity assuming the zone was operating at its fully softened strength (c' = 0; +I,' = 20). The experimental investigation of active and stress relief paths presents difficulties since one cannot impose an appropriate stress path on a sample by controlling the stresses alone. It is in fact a mixed boundary value problem in which the stresses and strains interact to keep the sample in a state of contained failure as the confining stresses are reduced. Burland and Fourie (1985) have described a simple method for investigating such stress paths using mercury columns connected either to the cell or the ram of a standard stress path cell (Bishop and Wesley 1975). Figure 2 shows the arrangement for a stress relief test and a full description of the operation has been given by Burland and Fourie (1985). In summary, the sample is brought to an initial stress 0 by standard procedures. Valve 1 is closed and 2 is opened and the sample is subjected to steadily reducing axial stresses. A stress path 0, 1 , 2 , 3 , 4 will be followed in which the latter part lies on a failure envelope. By varying the number of standpipes and their cross-sectional area So11 e lement B