Reply to the discussion by J. Chu, W.K. Leong, and W.L. Loke on "Defining an appropriate steady state line for Merriespruit gold tailings"

Abstract

Chu et al. provide a discussion of an issue that has long been a source of concern for anyone trying to characterize the potential for liquefaction of sandy material using laboratory testing, i.e., the method of specimen preparation. At the outset, we need to correct an apparent misunderstanding. In their discussion of our paper, Chu et al. query why the gold tailings is labeled as material that “would always exhibit dilative characteristics upon loading”, when our study showed that contractive behaviour was possible. The quoted phrase merely represented the conventional wisdom in South Africa prior to the Merriespruit failure and was based primarily on the results of laboratory testing carried out in commercial laboratories using slurry deposition techniques where little attention was paid to the effect of preparing specimens at a range of initial void ratios. Chu et al. question the method of moist tamping that was used in our experimental programme and quote various researchers to support their view. However, contrary to their claim, Dyvik and Hoeg (1999) found that specimens prepared by both moist tamping and by a slurry deposition technique produced contractive behaviour, whilst undisturbed specimens of the same silty fine sand showed dilative behaviour, even at a slightly higher initial void ratio. These authors attributed the difference to the existence of layering in the undisturbed specimen, which was apparently clearly evident. Vaid and Sivathayalan (2000) have shown results obtained from moist tamping preparation techniques that indicate a less uniform distribution of void ratio than was obtained from the water pluviation method and point to the large collapse settlement of moist tamped specimens upon initial flooding as evidence of the unsuitability of this technique. We would suggest that this component of settlement might be minimized by an appropriate choice of the moisture content at which the moist tamped specimen is prepared. We carry out this screening process as a matter of routine and have found large differences in the volume change upon saturation. It is thus misleading to talk about the moist tamping technique without being clear about the moisture content used. Chu et al. clearly favour the use of the water pluviation technique for specimen preparation, claiming that “water pluviation is considered to produce a soil fabric similar to that of natural sands,” despite the difficulties (which they acknowledge) of segregation that occur when fines are deposited together with sand. They also show results for “identical” specimens that were prepared using water sedimentation and moist tamping techniques and produced dilative and contractive behaviours, respectively. It is questionable whether void ratios can be determined to an accuracy of three decimal places, as implied by their results. We have certainly been unable to produce this level of accuracy. Their ‘identical” specimens may thus not be identical at all. The difference in behaviour could also be because of the almost unavoidable variation in local void ratio that occurs during specimen preparation. As shown by Frost et al. (1999), even the technique of specimen preparation by highly controlled air pluviation could not prevent some degree of nonhomogeneity of void ratio distribution. These authors suggest that global void ratio measures lead to erroneous determinations of the steady state, a problem that is exacerbated with dilative specimens. A key point that does not seem to have been addressed by Chu et al. relates to how we might explain the occurrence of extensive flow failures unless the material that flows is predominantly contractive in nature. The fact that moist tamping produces a structure that is susceptible to strain softening during undrained loading shows that such behaviour is of course possible. The primary question thus seems to be whether the structure developed by moist tamping is representative of what may occur in the field. An interesting factor that appears to have been overlooked by the proponents of water pluviation as a technique of specimen preparation is the nature of the water into which the solid particles