Estimating Setup of Driven Piles into Louisiana Clayey Soils

Two types of mathematical models for pile setup prediction, the Skov-Denver model and the newly developed rate-based model, have been established from all the dynamic and static testing data, including restrikes of the production piles, restrikes, static, and statnamic tests of the test piles at the LA-1 relocation project. Pile testing data from other sites, such as Mo-Pac Railroad Overpass, Bayou Liberty, and Calcasieu River etc., have been used for model verification. Twenty-one out of the 115 restrike records of the production piles and three load testing records from the nine tested piles were obtained at or longer than two weeks after pile installation. The conventional Skov-Denver model is achieved with the setup parameter A equal to 0.57, and the normalized ultimate shaft capacity from the rate-based model is 1.846 on the basis of the entire restrike and load testing data. Based on the rate-based model with limited amount of long-term production pile restrike data, it is predicted that the ultimate shaft capacities of the piles were about twice the measured shaft capacities at the 24-hour restrike. In general, the piles at the LA-1 relocation project reached about 90~95 percent of their ultimate shaft capacities within two weeks after installation. Preliminary verification and prediction work has indicated that capacities of those piles at two-week or longer-time restrike or load testing were mostly under predicted if the entire database was used for the model prediction. The setup parameter and the normalized ultimate shaft capacity have turned to 0.65 and 1.985, respectively. Selected piles with restrike or load testing at or more than 200 hours after the end of driving have given the setup parameter of 0.65 and the normalized ultimate shaft capacity of 1.985. Predictions with the new rate-based model are improved. It demonstrates that long-term restrike or long-waiting load testing data have a profound and critically important role in improving reliability and accuracy of the prediction models. An empirical relationship, between the measured pile capacity at 24-hour restrike and the calculated pile capacity based on the Cone Penetration Test (CPT) log, has been established. It will make pile setup prediction operable without the 24-hour restrike data. As the last portion of the research project, a simple Load and Resistance Factor Design (LRFD) calibration of pile setup has been performed. Resistance factors have been achieved corresponding to different target reliability indices and dead load to live load ratios.