Extrapolation of pile capacity from non-failed load tests

The described state of practice calls for the ability to reliably estimate the ultimate pile capacity from non-failed load tests. A practical analytical method is proposed, capable of extrapolating the measured load-settlement relations beyond the maximum tested load. The proposed procedure, along with two other possible methods, are evaluated. The procedures are examined through a database of 63 piles load tested to failure. Loading is assumed to be known for only 25%, 33%, 50%, 75% and 100% of the actual ultimate capacity and, separately, for 25%, 33%, 50%, 75% and 100% of the load-settlement data points. The "known" data is then extrapolated using the actual measurements. For consistency, only one failure criterion (Davisson's) is applied. The obtained results are analyzed statistically to determine the accuracy all three methods.; It is shown that the accuracy of the proposed method is 0.99*0.26 (SD), 0.89*0.41, 0,74*0.46 and 0.64*0.44 when assuming 100%, 75%, 50%, 33%, and 25% of the known ultimate capacity. The obtained results for the 63 data cases suggest that even when the predicted ultimate capacity is four times the maximum actual load test, the associated risk is zero for exceeding the design load, when using the extrapolated value with a factor of safety of 2.0.; Case history analyses of six load-tested piles at two sites are presented. The analyzed cases indicate possible substantial savings when the ultimate capacity well exceeds the maximum applied load. Moreover, the method already demonstrates its enormous importance from both aspects, engineering and financial. (Abstract shortened by UMI.)...