Scaling relations for soil studies indicate that certain model behaviors will represent correctly those of the prototype if the model is subjected to an acceleration proportional to the model scale. The most convenient way of accomplishing this is to perform model tests in a large centrifuge. Lateral cyclic load tests have been carried out on a 1/100scale instrumented model pile imbedded in a saturated sandy silt at 100 g in a centrifuge. The history of displacements and moments is presented and discussed. An analysis is presented to explain the observed softening behavior of the pile.
In design and construction of a variety of pile-supported structures, including offshore drilling and production platforms, greater attention is being paid to axial and lateral structural response to cyclic loading due to wave or earthquake excitation. Analytical and numerical methods have been suggested to handle single (and to some extent) repetitive load cycles, 6 but experimental verification of the techniques is scarce.3 Thi9 develops because of the difficulty of conducting full-scale field tests, and the inadequacy of conventional laboratory model tests correctly to model the prototype load, stress, and displacement conditions. In addition, the nonuniformity of natural soils renders the results of field tests less general than is desired for a confirmation of theoretical behaviors. For the very large steel piles, diameters 4 to 6 ft and wall thicknesses up to 2 in., used in offshore prototype construction, tests under single or cyclic axial and lateral load applications are virtually impossible. The information that such tests would supply, however, is urgently needed.
A way out of these difficulties is to perform correctly scaled model tests, in which the model soil responds correctly to prototype level stresses and strains. This can be accomplished by carrying out such tests in a centrifuge, as described below. A number of cyclic lateral load tests on appropriately scaled model piles has been concluded; the soils included fine sand, dry and saturated, and a saturated sandy silt. The test results are presented and discussed in the paper.
If the scaling relations required to characterize tests of geotechnical models are established, it is found that, because of the general dependence of the mechanical properties of soil on the ambient stresses, and the importance of gravity-induced stresses, scaling can be satisfied only under special conditions. In the special case of soil, it is inconvenient or impossible to construct a model material, and a real soil usually is employed in model tests. In that case, the scaling conditions require that the soil model be subjected to a higher gravitational acceleration than the prototype. The ratio of the accelerations in model and prototype structures is inversely proportional to the ratio of their linear dimensions; that is to say, a 1/100-scale soil model must be subjected to one hundred times the earth's gravitational acceleration, and so on.
