Shake Table Tests: Seismic Soil-Pile-Superstructure Interaction

The earthquake simulator at PEER UC Berkeley's Richmond Field Station is the site of an ongoing study investigating the performance of pile-supported structures in soft clays. The project is sponsored by the California Department of Transportation (Caltrans), and is part of a coordinated research effort by the UC Berkeley Geotechnical Group examining the seismic response of piles. A series of shake table scale-model tests is being performed to provide insight into a variety of aspects of seismic soil-pile-superstructure interaction (SSPSI) and also to generate a data set with which to calibrate an advanced SSPSI numerical modeling tool presently being developed at UC Berkeley.

The current advanced state-of-practice of SSPSI analysis consists of substructuring the problem into site response, soil-pile interaction, and structure-foundation response components, sequentially solving each subset of the problem. But SSPSI is a system with nonlinear and interdependent processes, and therefore this research program has focused on developing a fully coupled analytical method that analyzes the entire SSPSI system. The validation of such a tool requires detailed benchmark studies, but unfortunately there is a lack of well-documented case histories of instrumented pile-supported structures subjected to strong shaking. The shake table tests at PEER were undertaken to produce such a data set.

The prototype case was defined as 16" diameter concrete-filled steel pipe piles driven into a deposit of soft young San Francisco Bay mud, in single-pile and pile group configurations. Scaling relations were developed relating model to prototype performance, and were formulated to capture the nonlinear properties of SSPSI. Discrete scaling factors are listed in table 1 in terms of the geometric scaling factor, lambda (lambda = 8 for this project). Model piles consisting of 2" diameter x 0.028" thick wall aluminum tubes were selected to meet moment-curvature scaling criteria. A model soil composed of kaolinite, bentonite, and fly ash, at approximately 130% water content, was designed to satisfy scaling criteria of undrained shear strength, shear modulus, modulus degradation, and damping characteristics.

Table

A particular challenge of geotechnical scale-model testing is devising a containment vessel for the soil deposit being tested. The ideal container is one that allows the model soil deposit to respond in the same manner as the free-field prototype condition, without imposing significant boundary conditions. The model container specially designed for this project is shown in figure 1, mounted to the PEER shake table.

Fig. 1. Model test container mounted on shake table with soil mixer/pump in background

The container was designed with a cylindrical geometry to provide 3-D capability, thereby taking advantage of the shake table's recently upgraded capabilities and allowing deformation modes not attainable in centrifuge testing. The container consists of a rubber membrane bolted to a top ring and a base plate; the top ring is supported by four struts with universal joints that provide lateral flexibility in simple shear and allow multi-directional shear deformation. Circumferential Kevlar bands provide radial stiffness, and internal shear strips accommodate complementary shear stresses developed in the soil during shaking.

Five independent model setups were tested during November/December 1997, and five additional models were tested in July/August 1998. This series included free-field tests without piles, tests with single piles, and 2 x 2, 3 x 3, and 5 x 3 pile groups. Figure 2 depicts a test of two 3 x 3 pile groups with different columns and identical superstructure head masses.

Fig. 2. Test series 1.2. Two 3 x 3 pile groups with different column heights.


The test series was designed to

The models were subjected to uniaxial and biaxial input base motions consisting of low-level sinesweeps at 0.05 g, a Loma Prieta earthquake Yerba Buena Island (YBI) record scaled from 0.2 g to 0.3 g, and a Kobe earthquake Port Island record scaled from 0.1 g to 1.4 g. Thirty-five accelerometers were deployed in downhole arrays in the soil and mounted to the pile caps and superstructures; seven pairs of strain gages were fixed to the model piles to detect bending and axial strains; and wire potentiometers measured superstructure deflections. Up to 143 channels of data were acquired in each test. T-bar, vane shear, and shear-wave propagation tests were conducted to establish the in-situ soil shear strength and shear-wave velocity profiles. In addition, a suite of pile head loading tests was performed, including static lateral loading, pile head lateral impact, pile head forced vibration, and static and cyclic axial loading tests.

Preliminary test results were recently reported in the Proceedings of the 5th Caltrans Seismic Research Workshop, and will be described in only general terms here. First, the PEER shake table performed very well in reproducing the programmed command signals. Second, the model container appears to be doing a satisfactory job minimizing boundary effects, since the observed site response strongly correlates to the theoretical free-field response, as computed by SHAKE 91. Third, the single-pile tests have exhibited characteristics of both kinematic and inertial SSPSI, thereby meeting established benchmarks. Figure 3 presents bending moment envelopes of single piles with varying head masses subjected to YBI90 at 0.2 g; the concentration of moment near the head of pile S1 (160-lb head mass) illustrates inertial interaction, and higher bending moments at depth in pile S4 (10-lb head mass) reflect kinematic interaction.

Fig. 3. Test 1.15 Pile bending moment envelopes illustrating inertial and kinematic interaction

Fourth, the pile group tests have been successful in contrasting the structural response of different period, cap embedment, water impoundment, and group effect conditions. Figure 4 depicts the response of two 3 x 3 pile groups with identical head masses but different column heights (H/B = 1 and 3). The biaxial shaking tests and the correlation of head loading tests to seismic response were only recently completed and those data are currently being processed.

Fig. 4. Test 1.26. Accelerometer 5% damped response spectra of structures S1 and S2, and soil

In summary, an advanced testing capability for SSPSI problems has been developed at PEER UC Berkeley with the potential for a wide range of applications. Issues such as batter pile performance, large diameter drilled shaft response, pile spacing effects, pile raft foundation behavior, influence of column ductility on foundation demands, etc., are among future topics to be studied. The development of the numerical analysis tool in the GeoFEAP platform is progressing under Thomas Lok. Acknowledgements are extended to Professors Michael Riemer, Raymond Seed, and Juan Pestana; to Don Clyde for shake table operations; and to Kevin Mercer, Wes Neighbour, and Changri Yin for their testing assistance.

—Philip Meymand
Graduate Student Researcher
University of California, Berkeley

Reference
California Department of Transportation. [1998].
Proceedings of the 5th Caltrans Seismic Research Workshop. [Sacramento, Calif.].