Discussion of "Bearing capacity prediction of spatially random c – ϕ soils"

Abstract

Popescu The writer would like to commend the authors for a particularly interesting analytical and numerical study of the effect of spatial variability of soil properties on the bearing capacity. It is generally acknowledged that the problem of accounting for the stochastic spatial variation of soil properties is a particularly complex one, both from the datacollection point of view and from the methodological point of view. Generally accepted techniques have not yet been established to accomplish these tasks. To initiate a constructive discussion on this subject, some comments are provided here on the selected range of coefficients of variation (COVs) of soil properties and on the methodology suggested by the authors to simulate the stochastic soil properties. Range of coefficients of variation Confidence in design soil properties, and therefore in predicted structural response, is affected by a series of uncertainties arising from several sources, such as intrinsic spatial variability (i.e., actual variation of soil properties from one location to another), measurement errors, and insufficient field information. Those sources of uncertainty are often considered together as one source of variability in the response. It was observed by the authors, and by other researchers, that the intrinsic spatial variability of soil properties affects not only the variability of the predicted structural response, but also the mechanical behaviour itself. As explained, for example, by Focht and Focht (2001), because of the presence of weaker zones in a soil deposit, “the actual failure surface can deviate from its theoretical position to pass through weaker material so that the average mobilized strength is less than the apparent average strength.” It appears that different types of uncertainties (insufficient information, measurement errors, etc., on the one hand, and actual spatial variability, on the other hand) affect the structural response in different manners, and they may have to be analyzed separately. It is the understanding of the writer that this was in fact the intention of the authors, namely to analyze the effects of intrinsic spatial variability alone. A number of researchers have provided information on ranges of COVs for spatial variability of different soil properties, as obtained from in situ soil investigations. In most cases, those ranges also include variability due to measurement errors. There is, however, a very comprehensive study in this respect (Phoon and Kulhawy 1999), mentioned also by the authors, in which approximate guidelines are provided for the COV of intrinsic spatial variability, separated from other sources of uncertainty. It is the writer’s opinion that, until more results become available, the guidelines provided by Phoon and Kulhawy (1999) are a good indication of the order of magnitude for the COV of soil variability. COVs as high as 5 (i.e., one order of magnitude larger than those usually reported in the literature) are assumed in the paper for the soil parameters. In justifying this range, the authors present a very interesting set of considerations related to soil variability at various scales, starting from the soil particle scale (of the order of micrometres) and going to the regional scale (of the order of kilometres). It is demonstrated that, within such limits, the range of COVs assumed in the paper is reasonable. The writer believes, however, that with respect to intrinsic soil spatial variability, the scales applicable here are bounded by the volume of soil involved in the bearing capacity failure and by the finite element mesh size. 367