Class A prediction versus performance of O-cell pile load tests in Sydney Sandstone

Patrick K. Wong and David Oliveira


As part of the Barangaroo South project, pile load tests were carried out on two prototype rock socketed piles (SC-01 and SC-02) in Sydney Sandstone to validate the design. The pile load tests were carried out using the bi-directional OCell method on 750 mm diameter piles having rock socket lengths of 7.85 m (SC-01) and 6.38 m (SC-02). The target O-Cell test load was 17 MN, giving a total potential maximum test load of 34MN (i.e. shaft plus base) on each pile. Once the testing reached the required target test load, the O-Cell was loaded to its maximum capacity of about 26 MN. Assuming the O-Cell load recorded beyond the calibrated range is valid, the maximum test load reached is equivalent to a total capacity of 52MN on each pile. Prior to the O-Cell testing of the first test pile (SC-02), a Class A prediction was carried out. In this paper, actual test pile load-deformation results are compared to the Class A prediction, together with a back-analysis of the O-Cell testing using numerical methods for calibration of rock stiffness and assessment of topload pile response.

The Class A prediction of SC-02 using the method given in Fleming (1992) and adopted foundation design parameters gave good match with the actual test results. SC-02 was back-analysed using numerical methods (program FLAC3D) to assess the equivalent secant modulus of the rock. Analyses of the top load pile response were then carried out using FLAC3D and program PIES (the software used for the foundation design of the project). By comparing the top-load analysis results with those inferred from the O-Cell test, the ultimate resistance parameters were confirmed but it was assessed that the design stiffness parameters adopted may be optimistic. Back-analysis of the test results also indicate that the top-load stiffness inferred from the O-Cell test may be over-estimated due to opposing displacement interaction effects. The conclusion is that actual pile stiffness at the serviceability load may be about 55% of the design estimate. However, pile top settlement is relatively small (i.e. ~ 2% of pile dia.) and the design estimate is considered reasonable in terms of accuracy for geotechnical design purposes.