Investigation of soil response characteristics of dry sandy deposits in wide range of base excitation

This research utilized seismic response measured in a series of centrifuge tests under a variety of excitation conditions that were carried out with various mechanically stabilized earth (MSE) walls. Altogether six different walls were tested in the large centrifuge at the National Geotechnical Centrifuge at University of California, Davis. The walls were subjected to as many as 17 base motions ranging from cosine waves to realistic earthquake like excitations which were scaled from motions recorded in Kobe (1995) and Santa Cruz (Loma Prieta, 1989). In the initial tests the maximum base acceleration was small, but subsequently increased as the test progressed; the largest event had excitation of as much as 0.9g. This centrifuge test series yielded a large database of soil response histories under a variety of excitation conditions. The thesis presented here uses this rich database of soil behavior to explore the characteristics of soil nonlinearity. Soil response characteristics in terms of amplification and permanent deformation (settlement) have been investigated with help of DESRA2M. A modified version of the time-domain code DESRA2 has been used to calculate nonlinear soil response of soft sediments near the surface (Lee and Finn 1978). DESRA2 is a direct nonlinear method in which the nonlinear hysteretic behavior of soil is modeled. Modifications to original program DESRA2 were such that some of the recently available nonlinear soil behavior data can be readily incorporated. The modifications have been chosen such that only a minimum number of soil parameter are used in the model, while retaining many of the convenient features of the original model. These response parameters are important design concerns for geotechnical engineers interested in assessing structural damage. Measured and Computed peak accelerations from base shakings from 1995 Kobe earthquake scaled to yield a peak base excitations from 0.13g to 0.90g have been compared. The coefficient of variation (R 2) in log10 units between computed and measured peak acceleration is 0.95. This high value of R2 = 0.95 indicates a good correlation. Given the fact that the data base from the centrifuge tests in as many as 51 and the base excitation level varied within a wide range of 0.13 to 0.9g, this level correlation is remarkable. Acceleration response spectra of computed and measured acceleration histories all cases were also compared. A damping ratio of 5% was used. The differences between the computed and measured spectral accelerations were virtually negligible above a structural period of 1.0sec. The international building code has provision for amplification (or deamplification) of acceleration response in soil. The case of dry sandy deposits fall into either Site Class C or D. The soil class that is appropriate for the soil deposit used in the centrifuge tests is Soil Class D. One of the outputs from DESRA2M is the surface settlement. The computed and measured incremental surface settlements for all the cases selected in the comparison of the acceleration response were studied. It may be noted that computed and measured settlements are similar especially in the intermediate level excitations.