DEFORMATION RESTRAINT AND THE MECHANICS OF SOIL BEHAVIOR

Statistical analysis of data from the literature indicates that variability of natural soil deposits often is a predominant factor contributing to inaccuracy of geotechnical predictions. A review of 46 case histories showed that settlement predictions for shallow foundations by consolidation theory resulted in errors ranging from 100 percent underpredictions to 30 percent overpredictions. Prediction error was also found to be a random process described by the Gaussian probability distribution function. With variability reconciled, consolidation theory was found to underpredict settlements by 22 percent. Although geotechnical design methodologies which include soil variability have been proposed, obstacles to practical application appear to be the absence of well defined probabilistic design goals and the inability of obtaining sufficient data to define variability with existing tests.;The degree of radial restraint was also found to have significant influence on axial stress-strain response of the soil. A fivefold difference was observed for restraints imposed by previous K-Test apparatus. A methodology in which the restraint function is defined from elastic theory is proposed, and experimentation illustrates that such systematic definition results in deformation response which is in close agreement with that realized from Lambe's stress path method.;In this research, probabilistic design targets for ultimate bearing capacity of shallow foundations were developed from criteria forming the basis of the current American Concrete Institute building code. Deformation restraint testing which involves subjecting cylindrical specimens to variable confining stress dictated by the amount of radial deformation was investigated as a potential method for rapid, single-specimen evaluation of limit strength parameters. The Iowa K-Test is a convenient form of deformation restraint testing where the degree of radial restraint or the restraint function is arbitrarily established by the character of the apparatus. An apparatus which allows imposition of an infinite variety of restraint functions was developed, and tests on loess suggests the existence of a unique, restraint function which allows development of a stress path corresponding to a multispecimen, constant-confining-stress strength envelope. This unique function was also found to be consistent with a theoretical stress-strain relation developed from Drucker's hypothesis.