| Soils encountered in construction are often unsaturated. If the magnitude of matric suction is quantified, it can be incorporated as a stress variable of the shear strength in the unsaturated state, which will influence the calculated safety of geo-structures. The objective of this study is to provide a detailed characterization of the suction profile in a North Carolina residual soil and propose an empirical model to predict soil suction based on basic soil indices. A field test site was chosen in Greensboro, North Carolina. In addition, three slopes of varying steepness were cut (0.25:1, 0.5:1, and 1:1) and one cantilever sheet pile wall was installed. The soil between the intended slopes and the sheet pile wall was excavated via stage excavation to a depth of 22 feet. Two major residual soil groups, a high plasticity silt (MH) and a low plasticity silt (ML), were encountered at the site. The suction of these soils, retrieved using thin-walled sampling tubes, was tested using tensiometers, the filter paper method, and pressure plate tests. Sixteen Fredlund Thermal Conductivity (FTC) suction sensors and sixteen moisture sensors were installed in the field for measurements over time. The monitoring process was from June 27, 2013 to March 10, 2014. Man-made infiltration (ponding of water) on the horizontal ground surface at the top of the slopes was maintained from February 6, 2014 to March 10th, 2014.;A comprehensive suction-related soil property database was established by testing soil from 64 Shelby tube samples obtained from soil borings. Twelve SWCCs were obtained using pressure plate tests. A field curve concept is demonstrated that considers the actual suction conditions of the soil, as measured from a tensiometer inserted into the Shelby tube, and the hysteresis of soil water characteristic curves (SWCCs). By using the actual one-point reading as a reflection point, the drying curve of SWCC can be adjusted to account for the actual field condition of the soil. The reliability of the paired T5 and T5x tensiometers used in this study was validated by testing uniformly laboratory-compacted soil samples.;An empirical suction prediction model was developed using multivariate analysis. The proposed statistical model for predicting the suction of the test-site residual soils better represents measured soil suction than models found in the literature. The proposed model was verified using limited data from Rahardjo (2012) who also studied soil water characteristic curves of residual soils.;For the field instrumentation, the field sensors were able to predict the changes in suction and water content quantitatively. FTC suction sensors captured the changes in suction, but the absolute values at the initial and final stages did not reflect the actual suction conditions measured on retrieved samples, at the beginning and end of the monitoring period. The water content sensor measurements are shown to be more reasonable. |
