Soil Dynamics and Liquefaction 2000

The paper presents a non associated elasto/viscoplastic constitutive equation for saturated sand. This constitutive equation assumes: instantaneous elastic response, yield surfaces in the sense of viscoplasticity (i.e., after each reloading following with delay in time the loading path), it describes both compressibility and dilatancy and the constitutive equation is non associated (i.e., the viscoplastic potential does not coincide with the yield function). It is shown that in this case a strip in the constitutive domain exists where short-term repeated loading/unloading pulse can produce loosing of stability, i.e., liquefaction.

Despite numerous past attempts and research efforts, comparisons of physical test results on the dynamic interaction of foundations with sandy soils with analytical theories have continued to be a subject of considerable debate and inconsistency. To determine the source of the difficulties that besiege the physical problem, a systematic investigation was performed using the centrifuge scaled modeling method to examine at a fundamental level the dynamic behavior of surface foundations under both vertical and horizontal excitations. Finally exposed by the experimental approach, a basic reason for the continuing difficulty in characterizing the dynamic behavior of foundations on sandy soils is discovered to be a key incompatibility of present analytical frameworks with the observed physical soil-foundation behavior. While a deeper physical understanding is required prior to a proper theoretical resolution, a conceptually new but practically simple analytical framework which can explicitly recognize and rationally accommodate the granular soil dynamics problem is found to be possible using the instrument of Impedance Modification Factors.

The results presented in this paper were developed as part of a larger project analyzing the behavior of full-scale laterally loaded piles in liquefied soil, the first full-scale testing of its kind. This paper presents the results of a series of full-scale tests performed on deep foundations in liquefiable sand, both before and after ground improvement, where controlled blasting was used to liquefy the soil surrounding the foundations. Data were collected showing the behavior of laterally loaded piles before and after liquefaction. After the installation of stone columns, the tests were repeated. Based on the results of these tests, it can be concluded that the installation of stone columns can significantly increase the density of the improved ground as indicated by the cone penetration test. The stone columns were found to significantly increase the stiffness of the foundation system, by more than 2.5 to 3.5 times that in the liquefied soil. However, in non-liquefied ground, the improvement from stone columns could be more than compensated for by increasing the piles. In liquefied soil, however, more than doubling the number of piles or increasing shafts diameters by 50 percent did not nearly match the improved performance of the treated ground. This study provides some of the first full-scale quantitative results on the improvement of foundation performance due to stone columns in a liquefiable deposit.

This paper describes the findings of an experimental study supported by the U.S. Army Centrifuge Research Center and Engineer Earthquake Engineering Research Program (EQEN) into the behavior of saturated sands under high initial effective confining stresses subjected to strong ground shaking. The research was conducted using the Army Centrifuge at the U.S. Army Engineering Research and Development Center (ERDC), located in Vicksburg MS, formerly known as the Waterways Experiment Station (WES). The centrifuge studies have shown that the generation of excess pore pressure is capped at a level below 100 percent for vertical effective confining stresses exceeding around 3 atmospheres (atm, or 300 KPa). This limit reduces at higher confining stresses. If verified, the potential benefits from this finding for the design of remediation works for large earth dams could be substantial. The paper describes the equipment used for the experiments, the research program, and presents the initial results, contrasting the development of excess pore pressure at low confining stress with that at high confining stress. Possible consequences for a hypothetical dam are discussed.

Liquefaction of sand, clean or with some fines content, has been extensively studied over the last three decades and is currently a phenomenon reasonably predictable. The study of cohesive soils behavior during and immediately after cyclic loading is especially difficult because of their structure variability and the major influence of structural characteristics on dynamic properties, which makes testing of reconstituted samples of little interest in evaluation of natural deposits. There are several categories of cohesive soils potentially liquefiable, which are considered in this paper: clayey silts or silty clays of low plasticity meeting the "Chinese Criteria"; highly sensitive clays; collapsible loess. Data from literature for better understanding of their susceptibility to liquefaction (triggering, post-cyclic strength) are summarized.

Whether the presence of silt adversely or beneficially affects liquefaction and the collapse potential of silty soils and how to evaluate cyclic strength behavior of a sand containing different silt contents are contentious issues. The purpose of this work is to investigate this question. Stress controlled undrained cyclic triaxial tests were conducted on specimen prepared by mixing a sand with silt in different proportions. The cyclic stress ratio (CSR=0.2) and confining pressure (100 kPa) were maintained constant. Relationship between no. of cycles required to cause liquefaction (at 5% axial strain) versus void ratio, and newly introduced equivalent void ratio indices based on intergrain contact density considerations are presented. Cyclic strength correlates well with the latter indices.

A new empirical method for evaluating liquefaction-induced vertical displacements is presented, which is based on results obtained from a Multi-Linear Regression (MLR) analysis of field performance records. Previously published techniques for predicting earthquake-induced settlements are based primarily on results derived through laboratory cyclic triaxial testing. Consequently, settlements obtained by such methods account only for the effect of the volumetric strains. In the field, observed vertical displacements are a function not only of the volumetric strains but also the vertical component induced by lateral spreading and the effects of sand boils. In the present study, a MLR model based on field data is developed for predicting liquefaction-induced vertical displacements. A database consisting of 50 measurements of vertical displacement forms the basis for the MLR model and analysis.

The state of the art for evaluating post liquefaction flow deformations of earth structures is presented and the main parameters that affect the deformations are reviewed. The primary focus is on the estimation of residual strength and large displacement methods for analyzing the consequences of liquefaction.

Casitas Dam is a 335-foot high, zoned earthfill embankment located in a highly seismic area, near Ventura, California. The U.S. Bureau of Reclamation (Reclamation) concluded that the foundations of the dam were susceptible to liquefaction in the event of an earthquake, and that liquefaction of the foundations could lead to failure of the embankment and uncontrolled release of the reservoir. Reclamation developed designs for dam safety modifications to increase the seismic stability of the downstream section of the dam, but the planned modifications do not include treatment of the upstream foundations. Two-dimensional dynamic deformation analyses and three-dimensional post-earthquake stability analyses were performed to evaluate the expected performance of the untreated upstream slope during and after significant earthquake shaking. The deformation analyses indicated that significant deformations would be expected in the upstream section of the embankment dam, but that freeboard would be maintained at the downstream edge of the crest of the modified dam. The three dimensional stability analyses indicated that three-dimensional factors of safety for the upstream slope that were between 10% and 35% higher than the corresponding two dimensional factors of safety. Based on these results, the two-dimensional stability analyses may be over-estimating the deformations in the upstream section of the dam. This paper includes a discussion of the methods used in the analyses, the results obtained, and limitations and other issues that had to be addressed to complete the analyses.

In this paper, probabilistic methods developed through logistic regression analyses of field (SPT and CPT) data and those through a newly developed mapping function approach are evaluated and compared. Well-documented cases by U.S. Geological Survey are used in the evaluation and comparison. The results provide a basis for risk-based decisions regarding liquefaction potential.

Seismic zonation of geologic deposits susceptible to ground motion amplification is implemented by combining in situ measurements, geologic maps, and remote sensing imagery in a geographic information system (GIS). The National Earthquake Hazards Reduction Program (NEHRP) recommends classifying geologic deposits into six profile types based on the average shear wave velocity of the upper 30 meters. Programs such as GIS-HAZUS, created by FEMA, are used to assess seismic risk and estimate potential loss based on the NEHRP classification. Seismic zonation obtained from this project may be incorporated into GIS-HAZUS to improve loss estimates within the New Madrid Seismic Zone (NMSZ). Conventional seismic zonation entails obtaining field measurements of dynamic properties such as shear wave velocity. This approach requires extensive field testing which may not always be cost efficient, particularly in rural areas. Therefore, an approach for identifying soils susceptible to amplification is presented that merges in situ measurements of dynamic properties with geologic information and available remote sensing imagery. Geologic and soil data were used to identify deposits, susceptible to amplification. Remote sensing images were used to delineate alluvial deposits characterized by thick deposits of soft sediments. Geographic information systems (GIS) offer a platform for merging these various data types and assessing the spatial extent of soils susceptible to ground motion amplification. Seismic zonation of the New Madrid Seismic Zone (NMSZ) is currently underway in a cooperative effort with the Association of Central United States Earthquake Consortium (CUSEC) State Geologists and the U.S. Geological Survey. Available in situ measurements of shear wave velocity are continuously being collected and archived for this region. The resulting seismic zonation may be used to identify areas requiring additional field testing.

Analyses are performed using a non-linear dynamic analysis code NIKE3D with Ramberg-Osgood non-linear material model and interface algorithm that allows separation and frictional sliding. The earthquake-induced separation along the earth-concrete interface is one of the potential problems in composite dams. A hypothetical 30.48 m (100-ft) high composite dam is analyzed for its behavior under Koyna Earthquake of India, 1967. The concrete and soil slopes on both upstream and downstream sides are varied to observe their effects the dynamic upstream interface behavior. Maximum separation, separation depth, interface nodal accelerations and interface pressure are obtained. The analysis results indicate: 1) Soil-concrete interface can separate under strong shaking, 2) The separation can be significant and the separation along the upstream soil-concrete interface can reach a great depth, 3) Peak crest acceleration can be very high, 4) Maximum dynamic interface earth pressures are significantly higher than static interface earth pressures.

The current approach for using cone penetration test data for liquefaction potential assessment involves the comparison of the seismic stresses imparted into a soil mass during an earthquake to the resistance of the soil to liquefaction. The approach involves an indirect correlation of density and other soil parameters to the liquefaction resistance of the soil mass. A direct approach for evaluating the liquefaction potential of a soil deposit has been proposed through a test using a vibrating piezocone penetrometer. Calibration chamber tests in saturated clean sands indicate that the penetration resistance of a soil mass can be significantly reduced through vibration under certain density and stress conditions. This reduction in penetration resistance is associated with elevated pore water pressures, and is directly related to the liquefaction potential of the soil.