A Random Field Model for the Prediction of Changes in the Undrained Shear Strength of Petroleum Contaminated Clay Soils

摘要

Quite often grounds in urban areas are subjected to contamination due to leaks from underground storage tanks of gas stations and of heating oil, as well as from spills of trucks and tankers carrying crude oil, heating oil or gasoline. Moreover the ever growing urban population pushes city boundaries to areas where industries had operated and the ground is heavily contaminated by petroleum and/or its derivatives. These contaminants reduce the load carrying capacity of the soil, thus compromising the stability of structures. udA laboratory investigation has been carried out to determine the undrained shear strength of completely saturated contaminated clay. The undrained shear strength, Su, is the parameter required in the total stress analysis (TSA) of foundations emphasis is placed on the experimental determination of the undrained total stress analyses (short term analysis) parameters. Specifically, the effect of contaminants on the undrained shear strength Su, must be determined in order to verify the following two requirements:ud1. A foundation must not collapse or become unstable under any conceivable loading.ud2. Settlement of the structure must be within tolerable serviceability limits. udIndeed experimental tests, as well as a statistical model developed in this study confirm that contaminants deteriorate the undrained shear strength of the soil and have significant effects on the elastic moduli. Consequently the immediate bearing capacity of the soil is affected and the integrity or serviceability of the foundation may be jeopardized. udA statistical random field model, based on the undrained shear strengths obtained in the laboratory, was used to model the contaminated soil. A method was developed to predict the statistical properties of the excursion set of the Gaussian random field above high thresholds. A new heavy tailed random field called the Student Random Field was also introduced, for which the distribution of the size of one cluster of its excursion set was derived. The tail distribution of its supremum was also approximated. Finally, as previously mentioned, this random field theory is applied to real data obtained from a series of triaxial tests with 2, 4 and 6% crude oil, heating oil and gasoline contaminated soil. ud