Controversy exists over the prediction of driven pile capacity in low plasticity, low over-consolidation ratio (OCR) clays. This paper examines this question and includes a detailed re-evaluation of load tests on driven piles performed at two low OCR sites, Lierstranda and OnsØy. Pile capacity predictions performed with the Imperial College Pile (ICP) design method are reported that included site-specific interface ring-shear tests. Additional test data are cited from other relevant case histories. While it is known that the ICP method can be applied safely in low OCR settings, including those with low plasticity, a class of normally consolidated, low plasticity clays exists in which the medium term axial capacities are significantly overestimated using either the conventional American Petroleum Institute (API) or ICP approaches. Arching is proposed as a possible mechanism to explain the low capacities measured in two cases. Counteracting positive time effects are also discussed. Cone penetration tests (CPTs) appear to provide a good diagnostic tool for identifying cases where capacity over-predictions might be anticipated.
A recent review by Clausen and Aas1 of the Norwegian Geotechnical Institute (NGI) raised concerns over the reliability of ICP and API predictions for three low OCR clays; Pentre (Shropshire, UK), OnsØy and Lierstranda (both close to Oslo, Norway). Pentre and Lierstranda are low plasticity, low over-consolidation ratio (OCR) silty clays, whereas OnsØy is a medium to high plasticity, low OCR clay. Reviews by Chow2 and Aldridge3 of the main text API procedures4 for assessing pile axial capacity in clay did not reveal any significant skewing with respect to soil plasticity index. Chow's database review of the application of the Imperial College Pile (ICP) procedures, proposed by Jardine and Chow5 and updated by Jardine et al. in 20056 led to a similar lack of apparent bias. This paper addresses the controversy by re-evaluating the ICP method capacity predictions for these sites and considering additional tests in other low plasticity, low OCR deposits.
The ICP method was developed from research performed by Jardine7, Bond8, Lehane9 and Chow10. Chow10 assessed the method's reliability in clays with a database of 55 high quality pile load tests with plasticity indices from 12 to 84% (including low plasticity deposits at Pentre, Croke Park and Cowden). Jardine et al.6, extended the data base to 68 pile load tests, and to ensure a good spread of data across a variety of different clays they limited the number of tests within a single geotechnical deposit to 10. Jardine et al.'s6 study reported a mean Qd/Qm (calculated pile capacity divided by measured pile capacity) of 1.03 and a coefficient of variation (CoV) of 0.20, which was a significant improvement on the API predictions for the same database. No bias was found for the ICP method with respect to plasticity index.
Clausen and Aas1 of NGI performed an assessment of pile capacity prediction of a database of piles driven in clay using the API4 and ICP5 methods.