Driveability and performance of long driven piles founded in saprolites

Driven steel H-piles are commonly used in Hong Kong as the foundation to support structures. These piles are traditionally installed with termination criteria derived from Hiley's formula. However, the assumptions in Hiley's formula are not suitable for long piles. In this research, the application of wave equation analysis in deriving termination criteria for long steel H-piles founded in granitic saprolites was studied and implemented. A methodology to develop final set tables for long driven piles using wave equation analysis was outlined. To verify the performance of piles final set using the wave equation approach, two preliminary piles, PP-1A2-2 and PP-1B3-1, were installed based on the new termination criteria. Dynamic and static load tests were carried out on these two piles to confirm the adequacy of pile capacity. Both the pile settlement values at twice the design load and the residual settlement values were well below allowable limits. Hence, the proposed methodology can provide a cost-effective yet safe solution to long piles. Soil parameters for wave equation analyses of driven piles in 6 types of common soils in Hong Kong were also derived from a comprehensive correlation study. With these parameters, final set tables for piles of different lengths can be developed for pile construction. A Hong Kong Driven Pile Database was developed in this project using Microsoft Access. The database contains information of over 1500 piles, with complete records of 308 static load tests, 1276 dynamic tests and 257 CAPWAP analyses. It is one of the largest databases for steel H-piles in the world. To assist pile design in future projects, comprehensive searching functions are developed and demonstrated with design examples. Using the piling data in the database, the reliability of pile capacity prediction by various dynamic methods was evaluated. It is found that Hiley's formula generally underestimates the capacity of long piles. The predictions from Case's method and CAPWAP analysis are less scattered but these methods also underestimate the pile capacity. An evaluation study on 348 cases of steel H-piles founded on saprolites was also conducted to study the accuracy of driving stress predictions by the methods proposed by ICE (1954) and Bowles (1988), and by wave equation analysis. It is found that the wave equation analysis gives the best predictions among the three methods. The mean and standard deviation of the measurement-to-prediction ratio are 0.98 and 0.085, respectively. The load transfer in 14 instrumented long piles founded in granitic saprolites was also studied. The distributions of shaft resistance with depth were developed and the characteristics of the mobilization of both toe and shaft resistances were also studied. The Beta values and toe capacity factor Nq for granitic saprolites were back calculated. For Beta values, the mean is 0.63 while the standard deviation is 0.38. For toe capacity factor Nq, the mean and standard deviation are 98 and 50, respectively. Local movements of 3.3-8.3% of pile diameter are required for substantial mobilization of the toe resistance. For substantial mobilization of shaft resistance, local movements of 1.0-8.0% of pile diameter are required. These values are in a wider range than those for concrete piles. The pile capacity changes with time after construction, which is referred to as pile setup. The setup effect of steel H-piles founded in granitic saprolites, however, has not been sufficiently studied in the literature. In this project, the setup effect was studied through a series of dynamic tests and static load tests carried out on piles PP-1B3-1 and PP-1A2-2 up to 106 days after the end of driving. Setup in shaft resistance was observed while relaxation occurred at the pile toe. The setup factor Ai for completely decomposed granite was evaluated. The mean of Ai is 0.58, which is slightly higher than those reported by Bullock et al. (2005), based on relatively short concrete piles in sand and clay. The setup factor tends to decrease with depth