The stated objective of this paper “to develop a methodology for evaluating undrained strength of sand, which governs the initiation of flow failures of a natural slope or earth structure, based on both laboratory and field testing” is ambitious, particularly when put within the context of previous unsuccessful attempts to attain the same objective. The culmination of the authors’ paper is their Fig. 19, which essentially relates (i) undrained strength of “clean” sands (fines content less than 5%) in a unique way to cone tip resistance and vertical stress level, and (ii) undrained strength of sands with a fines content greater than 5% in a unique way to tip resistance, sleeve friction, and vertical stress level. Given the multitude of factors that affect both cone penetration test (CPT) tip resistance and undrained strength of loose sands and the extreme sensitivity of undrained strength to some of these factors, the writer must question whether any such unique relationship is plausible. The process that led to the development of Fig. 19 includes numerous assumptions and simplifications. Many of these are noted by the authors and discussed individually, but others are only implicit. Although some of these simplifications and assumptions may be reasonable on their own, others are certainly not universally accepted. No attempt has been made to evaluate the cumulative potential errors resulting from these simplifications and assumptions. The writer is concerned that these may be so large as to render the proposed general correlation of limited practical use. Such is the stature of the authors within the geotechnical community, however, that it is likely that Fig. 19 could be widely used despite the authors’ own caution that estimated strengths are “only approximate and tentative.” The writer is aware of situations where other published CPT correlations with various parameters have even been programmed into the processing software for CPT output and presented, in otherwise factual reports, as profiles of the parameters in question, divorced from any qualifications and reservations that may have accompanied the correlation in the original publication. In the present case, the potentially important assumptions and simplifications can be listed as follows: (1) The results of laboratory undrained triaxial extension and “simple shear” tests from boundary measurements can be taken as a true reflection of behaviour in extension and simple shear in the field with no allowance for potential errors due to strain nonuniformity. The practical implausibility of the results of the triaxial extension tests is illustrated by the authors’ Fig. 8, in which it is suggested that sand at a relative density as high as 80% may have a ratio of undrained strength to initial mean strength as low as 0.15! Some may take the view that most of the apparent discrepancy between triaxial extension and compression tests could be due to errors in the extension tests. The fact that the simple shear tests apparently show intermediate response between triaxial extension and triaxial compression may simply be due to an intermediate degree of error. (2) Tests on Toyoura Sand can be used for interpretation of all case histories, despite the extensive evidence that different sands at the same relative density will not necessarily behave in the same manner. (3) The shear stress at the laboratory-measured quasi-steady state is the relevant strength with regard to flow slides. Even though, at some densities, strength will increase at higher strains in the laboratory, it is assumed that this would not occur in the field. The only justification given is that “it is unknown if hardening is possible in such conditions.” Some workers have suggested that the phenomenon of quasi-steady state is exaggerated by (or even due to) laboratory test details. Whether or not one agrees with this, some uncertainty must be present. (4) The effect of fines content can be dealt with by a consideration of the effect of fines content on CPT tip resistance – density relationships without consideration of the potential effect on other sand behaviours such as during undrained flow. This essentially implies that behaviour of all sands is uniquely related to relative density. (5) CPT tip resistance can be “normalized” for overburden pressure and “corrected” for grain characteristics using the method of Robertson and Wride (1998). (6) The “correction” factor (Kc) for cyclic conditions may be used for static conditions.
Read full abstract
Feb 1, 2002
S Teachavorasinskun + 2
Open Access