Influence of stress relief due to deep excavation on capacity of pile foundations

Pile foundations are commonly used to support tall buildings with a deep basement in major cities such as Hong Kong, Shanghai and London. The design of pile foundations is often based on load tests carried out at the ground surface, prior to basement excavation. A sleeve may be used in a load test to eliminate shaft resistance within the entire depth of the planned excavation. In such a conventional load test, however, the effects of stress relief due to basement excavation on pile performance cannot be captured. Systematic research investigating capacity and stiffness of piles with and without considering stress relief effects is not available in the literature. There is no simple theoretical solution to calculate the capacity of a pile foundation subjected to stress relief resulting from deep excavation.;The objectives of this research are to investigate the capacity and fundamental mechanisms of pile foundations with and without subjecting to stress relief. Two major research methodologies, namely centrifuge modelling and distinct element modelling (DEM), were adopted. The first series of centrifuge model tests included in-flight load tests on ten instrumented single piles conducted both at ground surface prior to excavation and at the formation level after completion of a 20 m-deep (prototype) in-flight excavation in dry sand. Each pile was 1.6 m in diameter, 30 m in length and instrumented at eight levels in the shaft to measure axial loads. Two different pile-soil interfaces, namely non-dilatant and dilatant interfaces, were used to model the response of piles in loose sands/normally consolidated clays and dense sands/overconsolidated clays, respectively. Governing mechanisms of pile-soil interface subjected to unloading were explained and quantified by DEM study of shearbox tests under constant normal stiffness, constant normal load and constant volume conditions. The second and third series of centrifuge tests were to investigate 3x3 elevated pile groups (PGs) and piled rafts (PRs) subjected to stress relief, respectively. Both non-dilatant and dilatant pile-soil interfaces were considered. During each test, in-flight load test was conducted at the formation level after a 20 m-deep excavation. The piles were connected by a rigid cap, which was elevated by 1 m and in contact with bottom of excavation for each PG and PR, respectively.;It is found that the capacity of a single pile with non-dilatant interface is reduced by 20% after subjecting to a 20 m-deep excavation. The reduction in shaft resistance is proportional to the magnitude of stress relief. Conventional load test on a sleeved pile prior to excavation gives non-conservative pile capacity. For a pile with dilatant interface, dilation of the pile-soil interface induces an increase in normal stress acting on the pile shaft. The magnitude of dilation is 30% higher after 20 m of stress relief. A semi-empirical equation is established to obtain the normal stress increase for a given dilatant pile-soil interface. Calculated pile shaft resistance based on the equation agrees well with measured data from centrifuge model tests. Furthermore, observed group efficiency of a PG subjected to stress relief due to excavation is only 0.7, which should be considered when interpreting load test results from a single pile. Capacity of a pile foundation is improved if a raft is in contact with soil. The raft carries up to 23% of total load at a settlement of 10% pile diameter. Moreover, effective stress in the soil is increased by raft load and therefore capacity of each pile in the PR is higher than a corresponding single pile or a pile in an elevated PG. The raft-soil-pile interaction is verified and quantified three-dimensionally based on Boussinesq solution. Both measured and calculated results show that shaft resistance increase is up to 100 kPa for the centre pile in a 3x3 group within a depth of half of raft width. It should be noted that the findings are from small-scale model piles with diameter of 16 mm. Extrapolation of the results to prototype must consider the influence of pile diameter on shaft resistance.