Centrifuge Study of Offshore Platform Response to Earthquake Excitations

Abstract

Abstract The paper describes a series of specialized centrifuge-shake table tests performed at the University of California at Davis (UC Davis) to investigate the behavior of offshore steel-jacket template structures subjected to earthquake excitations. The program tested two single piles and a truss-braced four-leg template with ungrouted piles. The tests were performed in a slightly overconsolidated clay deposit. The first (axial) pile was loaded axially at variable displacement rates to determine rate effects on pile capacity. The second (bending) pile, designed to isolate the lateral response, included a mass at the top to develop a similar dynamic response as the template structure. Various types of dynamic tests were performed. Small amplitude "step wave" base acceleration pulses were applied to estimate the natural period and damping of the bending pile and template structures. A controlled harmonic ground motion excitation was then applied to investigate the dynamic response of the structures through a range of input frequencies (frequency sweep test). Finally, both small (elastic level) and large (ductility level) earthquake motions were applied. Accelerations were measured both in the soils (free field) and on the structures. In addition, the bending pile and a pile on the template structure were strain gauged to determine bending moments. Displacements and pore pressures were also measured throughout the test program. The results obtained during the testing were analyzed by Energo Engineering with a series of numerical models using the Capacity Analysis Program (CAP), software that has roots in the initial SPSS/INTRA joint industry development for seismic analysis of offshore platforms Arnold et al. (1977). The axial test results showed an increase in pile capacity of about 14% per log cycle. This could potentially suggest an increase in axial pile capacity of 60 to 80% vs. static capacity, neglecting cyclic degradation effects, for earthquake loading conditions. Model predictions for the natural period of the structures were considerably improved using lateral p-y springs stiffer than specified by API. These stiffer springs are based on model tests performed for conductors. For the small elastic-level earthquake excitation, predictions with respect to accelerations and bending moments were considerably improved when the measured free-field depth-varying accelerations were used as input and radiation damping was included in the predictions. For the elastic level excitations, about a 33% amplification was observed between the input acceleration from the shake table and the measured accelerations near the top of the clay deposit. However, for the ductility-level earthquake excitation, which had peak input acceleration from the shake table of 0.46g, acceleration near the top of the deposit (3.9 m below mudline, BML) was reduced to 0.07 g. These results were consistent with ground response predictions which fully accounted for large amounts of hysteretic damping. These combined results suggest that for very large earthquakes with typical offshore clay deposits, a limiting acceleration level may be reached near the top of deposit due to significant energy dissipation in the soil. Finally the template structure foundation was observed to fail in the frequency sweep test when resonance caused sufficient dynamic amplification. The failure was predicted with CAP prior to the centrifuge test. The settlement of the template structure was also accurately predicted with CAP.