Project Overview:

First Test

• Purpose and General Configuration
• Overall Concept
• Model Preparation
• Soil Properties
• Properties of Structure Models
• Scale Factors
• Ground Motions
• Instrucmentaion and Measurements
• Data Processing

Second Test

• Purpose and General Configuration
• Overall Concept
• Model Preparation
• Soil Properties
• Properties of Structure Models
• Scale Factors
• Ground Motions
• Instrucmentaion and Measurements
• Data Processing

Third Test

Fourth Test

Purpose and General Configuration of Test SHD02

Centrifuge Test Series II (SHD02) investigates the seismic performance of three scale models of SFSI systems at liquefied ground sites. Nevada Sand, a well-characterized clean fine sand, is used as liquefiable model soil in this experiment. A series of four shaking events was applied to the model at a centrifuge acceleration of 55 g. This shaking was applied transversely with respect to the direction of the structural models.

Overall Concept

The 1.7 m long by 0.7 m wide by 0.7 m deep flexible shear beam container with uniaxial (horizontal) shaking mode earthquake simulator was employed. The container has internal dimensions of approximately 1.65 m x 0.79 m x 0.58 m (L x W x H). Using 55 g of spin acceleration, the internal dimensions of the shear beam container became equivalent to a prototype of 90.75 m x 43.4 m x 31.9 m (L x W x H). The soil model was 28 m deep. Figures below show the plan and cross section view of the model, respectively. All dimensions in the figures below are given in prototype units, unless specified otherwise.

Plan view of SHD02 model with prototype dimensions

Cross section view of SHD02 model; Structure A: base-line case, prototype 2-story building with rigid mat foundation; Structure B: a 2-story structure with increased width and length; and Structure C: a 4-story tall slender building with rigid mat foundation.

Model Preparation

Soil layers were placed using a barrel hopper, operated by a manual overhead crane. The placed density was controlled by maintaining a nearly constant drop height, rate, and mesh size. The non-liquefiable dense base sand layer and the liquefiable sand were prepared using dry-pluviation with subsequent saturation. Monterey sand was placed by dry-pluviation for the surficial fill soil. Soil was pluviated into the model container in successive layers, with each layer followed by leveling of the surface using a vacuum. Each lift corresponded to the elevation of a horizontal instrument array.
The model was saturated with a de-aired methyl cellulose solution of a viscosity higher than water to obtain realistic pore-pressure dissipation in the centrifuge model (Stewart et al., 1998). For saturation, the model container was placed on the centrifuge arm and metal saturation troughs were placed at both ends. Plastic tubes were extended to the bottom of the model container to ensure that the saturation proceeds from the bottom and ends toward the middle and upward. Before saturation, the model was evacuated under vacuum and the voids were flooded with CO2. Figures 3 and 4 are photographs taken during model preparation, which show images of model instrumentation, soil pluviation, and the final model configuration.

Soil Properties

Nevada sand is a fine, uniform, angular sand with mean grain size between 0.14 and 0.17 mm, and a coefficient of uniformity of 1.67, and less than 5% fines. We used un-improved Nevada sand in our tests, with Dr of approximately 40% and 86% for the liquefiable and non-liquefiable sand layers, respectively. The loose Nevada sand (Dr = 40%) was placed using a barrel hopper and the dense Nevada sand (Dr = 86%) using a box hopper, both manually operated by an overhead maintaining a nearly constant drop height. The minimum and maximum dry density (according to ASTM D4253-83 and D4254-83) and void ratio (according to Modified Japanese method) are summarized in the following table.

Properties of Nevada Sand – courtesy of EA Hausler (2002) and AM Kammerer (2000)

Monterey Sand

Monterey 0/30 sand, uniform coarse sand with mean grain size of approximately 0.4 mm, was placed on top of the liquefiable Nevada sand layer as the unsaturated fill material. The purpose of using this layer is to prevent capillary rise and liquefaction directly below the structures and to provide adequate bearing pressure at the foundation level of the buildings. This soil was placed using a hand held hopper at an established relative density of 86%.

Properties of Monterey 0/30 Sand – courtesy of EA Hausler (2002)

Propreties of Structural Models

All structures used for test series SHD01 are single-degree-of-freedom structures made of steel and aluminum. The foundations are made of solid blocks of aluminum and are assumed to be rigid. The blocks representing the mass of the structures and the two side walls or columns holding the mass are made of steel, due to steel’s relative high density and yield strength. There are three structures in the first model, each representing a mat foundation with a bearing pressure of approximately 76, 76, and 120 kPa for structures A, B, and C respectively. The following tables summarizes the properties and dimensions of these structures as well as their initial natural periods in prototype scale.

Scale Factors

The following table provides a lost of the scaling factors that was used to convert the data to prototype scale

Ground Motions

A sequence of shaking events was applied to the model at a centrifugal acceleration of 55 g. The shaking direction was along the width of the structures, transverse to their length. The first event applied to the model was a step wave used primarily to evaluate the small strain elastic response of the model and to verify that the instruments and the data acquisition systems function properly. The next events were a sequence of scaled versions of the fault-normal component (north-south) of ground motion recordings at a depth of 83 m obtained at the Kobe Port Island downhole array during the 1995 Kobe Earthquake.

Summary of Ground Motion Sequence in SHD01

Acceleration, velocity, displacement-time histories and the acceleration response spectrum of the Large Port Island shaking event (PGA = 0.55 g).

Instrumentation and Measurements

A thorough description of the centrifuge geotechnical facilities and instrumentation system at UC Davis is provided at their website: http://nees.ucdavis.edu. A vertical array of accelerometers was placed within the soil to measure the soil response during dynamic loading. Vertical LVDTs (linear variable differential transformer) are commonly used to measure ground surface settlements adjacent and at some distance from the buildings to observe the induced soil densification and soil property changes. Before and after shaking events, cone penetration and shear wave velocity testing were performed on the soil on the spinning centrifuge. Figure 6 is a three-dimensional representation of the model which shows the instrumentation layout. A logical combination of wired pore pressure transducers, accelerometers, and LVDTs with wireless accelerometers and an array of high-speed video cameras available at UC Davis were used to record data during the first test series, which are available as streaming data archived as part of the NEES data repository on NEES Central. In addition to these instruments, colored sand grids were used to facilitate measuring vertical strains when excavating each model at the end of the shaking events.

The plan view of Model SHD02; UC Davis Centrifuge; model scale showing instrumentation layout as well as the shaking direction

The cross section view of Model SHD02; UC Davis Centrifuge; model scale showing instrumentation layout

The 3-D view of model SHD02 showing instrumentation layout as well as relative size and shape of structures (blue: displacement, red: acceleration, purple: pore pressure transducers)

Data Processing

For each shaking event, data was collected at 4096 Hz frequency for approximately 15 seconds. Raw data files were saved in binary format and were subsequently converted to ASCII format using the LabView program. Data was saved starting a few seconds before shaking and ending a few seconds after each shaking event.