Experimental investigations of driven pile group behaviour in Belfast soft clay

Abstract

In 1996, a site adjacent to Kinnegar Sewage Treatment Plant, near Belfast, was made available (by the Northern Ireland Department of the Environment) to Trinity College Dublin to perform load tests on driven piles. Since 1997, investigations at Kinnegar have included lateral load tests on single piles (Phillips 2002) and vertical load tests (both static and cyclic) on single piles and small pile groups (reported in this Thesis). Extensive site investigation data has also been assimilated over this period. Vertical load tests were carried out single reference piles and groups of five piles (centre to corner spacing of s/B=3). Each precast concrete driven pile was 250mm square and with an embedded length of 6m, much of its load was carried in shaft friction. Some of these piles contained internal instruments allowing load distributions and horizontal total stresses to be estimated. Slow static load tests (in tension and compression, for single piles and groups) were of the maintained load type with a final displacement rate of 0.24mm/hour for each load increment. One-way cyclic tension tests (on single piles and groups) were carried out at a rate of 1 cycle/minute for between 180 and 550 cycles. There are two distinct facets to group action in driven piles: (i) the effect of neighbouring installations and (ii) the effect of loading adjacent piles. Both effects are likely to determine how the load-displacement relationship of a pile within a group differs from that of an isolated single or reference pile. It is commonly assumed that the soil surrounding any group pile is stiffened due to adjacent piling (i.e. its shear modulus increases) leading to improved individual performance under load. However, although the maximum excess pore pressures and total stress associated with group installation are higher than for a single pile, these are transient and return to original values very quickly. In fact, total stresses immediately after the installation of a five-pile group fall below those expected for a single pile. Upon full equalization of horizontal effective stresses, the value for the group appears to fall slightly below that expected for a single pile. This suggests that adjacent pile installation may have a small negative effect on subsequent pile performance. Independent data also suggests either a neutral or small negative installation effect. Group action in Belfast soft clay appears to be governed predominantly by the load transferred between group piles when under load. Tests on tension and compression fivepile groups (s/B=3) are consistent in indicating that on average, the stiffness of the group is typically ≈45% that of a single pile at a typical working load per pile1 (alternatively, the group has displaced by over twice the amount of a single pile). The relative stiffnesses of the centre and corner piles (corresponding to a working load per pile) are more difficult to determine due to inconsistent pile cap flexibility; nevertheless the centre pile is typically ≈65%-85% as stiff as a corner pile at any stage2 of group loading. While relative pile capacities depend upon the flexibility of the pile-cap, the ultimate load in the centre pile falls 25-30% below that of a single pile for a rigidly capped group. A load test on one group pile (with settlement measurements on adjacent non-loaded piles) proved instructive. It provided data to show that non-linear superposition (of displacements) is unsuitable for predicting the response of closely spaced piles but may be acceptable when widely spaced. The importance of choosing non-linear soil models is emphasized, as linear elastic ‘interaction factors’ grossly overestimate the amount of load transferred between piles and may lead to inefficient piling solutions. This data also contributed to the development of a trendline relating the stiffness efficiency to the pile spacing/width ratio (s/B), a useful design guide for small groups of the Belfast configuration. Its use in tandem with a suitable indication of the cost of constructing a capped group indicates that the economy of the solution is relatively insensitive to spacing in the range 4B<s<8B; spacings closer than 4B are not recommended. Cyclic loading to ‘failure’ degrades the shaft capacity of driven single piles by 25%; further ‘damage’ is entailed for the five pile groups (35%). However, cycling to loads below 50% of the pile’s dynamic capacity (as would be the case for piles with a factor of safety >2) causes little cyclic degradation. Failure in one-way cyclic tension is governed by the accumulation of mean pile head displacements and is not amplitude-driven.