Field testing and monitoring of full-scale drilled shafts subjected to lateral loads

In highway structures, drilled shafts have been commonly used as foundations to resist both vertical and lateral loads. Analysis of laterally loaded drilled shafts requires that the soil-shaft interaction mechanisms be properly considered. Traditionally, the p-y curve has been used to represent the lateral soil-shaft interactions in the computer analysis. The method for determining the p-y curves has been developed using a limited number of load test results in the early 70s. There is a need to critically re-evaluate the existing p-y curve criteria and to extend the range of applications to the drilled shafts of large diameters.; The objectives of the research are to develop a new method for determining the p-y curves based on the Standard Penetration Testing (SPT) N values and to utilize the neural network techniques to develop a predictive model of the drilled shafts under lateral loads. As a separate objective, an instrumentation program was developed and implemented in the Hamilton-126-6.61 bridge to monitor the foundation (drilled shafts and driven piles) behavior during installation and in-service conditions.; To accomplish the objectives, a total of nine (9) lateral load tests have been conducted at three Ohio Department of Transportation (ODOT) construction sites. The drilled shafts have been fully instrumented with vibrating wire strain gages and inclinometer casing has been embedded as well. Based on these lateral load test results, correlation has been developed to enable the establishment of p-y curves using the SPT N values.; As part of a comprehensive study of the drilled shaft behavior under bridge loads, an instrumentation program has been developed and implemented on a deck-on-steel stringer three span bridge in Hamilton County, Ohio. Both vibrating wire strain gages and inclinometer casing have been embedded in the foundation structures during construction. Information collected from this instrumentation-monitoring project furnishes important insights on the actual stresses and deflections of the foundation structures, which can then be used in the development of the LRFD (Load and Resistance Factor Design) approach for bridge foundations.; In analyzing the data collected from the load tests, the neural network techniques have been employed to develop an optimized computational algorithm for predicting the behavior of the laterally loaded drilled shafts using the SPT N values. The variations of the neural network topology and the training algorithms have been extensively studied in this dissertation. The final result is a recommendation of the most efficient neural network computational algorithm specifically tailored to the prediction of the drilled shaft behavior.