Search

Regression models are presented for the dynamic properties of highly organic soils. The models are based on a database of triaxial and resonant-column/torsional-shear cyclic loading tests on thin walled tube samples mainly retrieved from the Sacramento-San Joaquin Delta. The soils in this database range from highly fibrous peat to amorphous organic clays with organic contents (OC) ranging from 14-81%, water contents ranging from 88-495%, total densities (p) ranging from 1.056-1.450 Mg/[m.sup.3], and effective consolidation stresses ([[sigma]'.sub.vc] ranging from 11-135 kPa. The secant shear modulus (G) and equivalent damping ratio ([xi]) were modeled as variables dependent on the shear strain amplitude ([[gamma].sub.c]), consolidation stress ([[sigma].sub.vc]), and OC. The residuals of the regression models were analyzed against other predictor variables including undisturbed density ([rho]), loading frequency (f), and number of loading cycles (N). A regression model for [rho] was developed, and conditional probabilities were used to improve the estimation of G and [xi] when p measurements were available. The database of in situ measurements of shear wave velocity ([V.sub.s]) was used to adjust the regression model for in situ conditions. Variances and correlations in the regression models are presented. CE Database subject headings: Soil properties: Dynamic properties; Organic matter; Peat; Shear modulus; Damping; Regression models.

The seismic demand on potentially liquefiable soils can be approximated by a series of uniform shear stress cycles. Procedures are reviewed for converting an arbitrary acceleration time history to a series of uniform cycles with amplitude = 0.65 of the peak. The number of cycles (N) at this amplitude is evaluated so as to represent a seismic demand for liquefaction triggering equivalent to that of the accelerogram. An assumed relationship between N and magnitude (m) underlies so-called magnitude scaling factors used to adjust the liquefaction resistance of soil for the effects of duration/magnitude. Scaling factors can alternatively be related directly to N, which enables the effects of factors other than m (for example, site-source distance r) to be quantified. We develop empirical models for N that are applicable to active tectonic regions and find a strong dependence on m and r and a weaker dependence on site condition and near-fault rupture directivity effects. The model for N is used to develop new scaling factors for soil liquefaction resistance that are distance-dependent and have a defined level of uncertainty.

Back-analyses of the recorded seismic response of the Operating Industries, Inc. landfill were performed to estimate strain-dependent shear modulus reduction and material damping relationships for solid waste. Two-dimensional dynamic finite element analyses were performed using the accelerograms from five earthquake events. The sensitivity of the computed results to variations in shear wave velocity, shear modulus reduction, damping, and Poisson's ratio was examined. The recorded and calculated motions at the top of the landfill were compared through an objective statistical analysis of the computed and recorded acceleration response spectra. Statistically significant findings from the analysis indicate that strain-dependent shear modulus reduction and material damping relationships for the solid waste at this site are best characterized by those of a medium plasticity clay.

It is often useful in earthquake engineering practice to characterize the frequency content of an earthquake ground motion with a single parameter. Three simplified frequency content parameters are examined: mean period ([T.sub.m]), predominant period ([T.sub.p]), and the smoothed spectral predominant period ([T.sub.o]). These frequency content parameters are calculated for 306 strong motion recordings from 20 earthquakes in active plate-margin regions. These data are used to develop a model that describes the magnitude, distance, and site dependence of these frequency content parameters. Nonlinear regression analyses are performed to evaluate model coefficients and standard error terms. The results indicate that the traditional [T.sub.p] parameter has the largest uncertainty in its prediction, and that previous relationships proposed to predict [T.sub.p] are inconsistent with the current data set. Moreover, [T.sub.m] is judged to be the best simplified frequency content characterization parameter, and it can be reliably estimated.