Correlation of site variability from SASW and CPT measurements

Abstract

Geotechnical site characterization programs are designed to reduce uncertainty and maximize the efficiency of geotechnical design. Ironically, however, many site characterization programs are inefficient themselves as they call for extensive intrusive testing conducted at regular intervals. This thesis attempts to improve the efficiency of such programs by drawing a correlation between variability in CPT and SASW measurements across a geotechnical site. With such a correlation, the practicing engineer could potentially modify the extent (and cost) of intrusive testing based on observed variability in surface wave dispersion. To examine this correlation, CPT and SASW data from the Labadie Utility Waste Landfill (UWL) near Labadie, Missouri were analyzed. Variability was quantified through coefficients of variation (COV) calculated amongst CPT measurements (q[subscript t] and f[subscript s]) for a particular depth (z) and SASW phase velocity (V[subscript ph]) for a particular wavelength (λ) at five locations within the UWL. Positive correlations were achieved utilizing a method developed in this thesis wherein q[subscript t] was converted to an equivalent corrected tip resistance (q[subscript teq]) by weighting q[subscript t] in the same manner that Rayleigh wave energy is weighted with depth. Utilizing this method, a linear regression between the mean COV of V[subscript ph] and q[subscript teq] yields a slope of 0.18, an intercept of 0.02, a coefficient of determination (R[superscript 2] ) of 0.751, and a p-value of 0.057. Removing outliers and COVs calculated in the highly variable upper 5 ft of soil yields a slope of 0.19, an intercept of 0.02, a R[superscript 2] of 0.961, and a p-value of 0.003. Finally, a relationship between q[subscript teq] and V[subscript ph] was developed that facilitates the estimation of dispersion curves from q[subscript t] alone. Utilizing this relationship, dispersion curves estimated from q[subscript t] were, on average, within 10% of those measured.