| Description |
The purpose of this research was to evaluate the uplift capacity of single Rammed Aggregate Piers (RAPs) using numerical analysis and compare the results to the measured values from Full-Scale Tests (FSTs) conducted in the field. Each single RAP, uplift plate, and surrounding matrix soil was analyzed using a 3-dimensional Finite Difference Method (FDM). An important aspect of the RAP numerical modeling is construction sequence, which was simulated with three different techniques: Lift Compaction, Cavity Expansion, and the newly proposed Cavity Expansion with Variable Displacement technique. The constitutive model parameters were estimated from in situ tests and material properties reported in the literature. Two constitutive models were used in the numerical analyses - an elastic-perfectly plastic Mohr-Coulomb model and a Plastic-Hardening model that employs nonlinear stress-strain behavior in the elastic and plastic ranges. The RAP numerical analyses were performed with FLAC3D (Fast Lagrangian Analysis of Continua 3D) models, and the numerical models were validated against the result of five full-scale RAPs uplift tests that were performed in May and June of 1998 underneath the I-15 bridges near South Temple Street in Salt Lake City, Utah. The five tested RAPs had the same diameter of 3 ft (0.91 m) and heights of 3, 6, 9, 12, and 15 ft (0.91, 1.83, 2.74, 3.66, and 4.57 m). The FLAC3D models predicted the limiting uplift capacity of the RAPs measured during the FSTs with a 1% to 40% error depending on the RAP construction technique and the RAP length. The average accuracy of the three iv construction methods is nearly the same for each RAP length. The numerical models showed that the numerically expensive RAP construction process could be replaced with a horizontal stress initialization with insignificant loss of accuracy. The numerical model successfully captured the slight bulging outward of the lower portion of the RAP and formation of a short failure wedge near the surface at the maximum uplift force that were observed in the FSTs. The FLAC3D model with the Plastic-Hardening constitutive model could not predict the load-displacement of the FSTs with the same level of accuracy of the FLAC3D model with Mohr-Coulomb constitutive model. The lower accuracy of the more sophisticated Plastic-Hardening model is likely due to the estimation of the numerous parameters of this constitutive model with average material properties reported in the literature. |
