Calibration of resistance factors for axially loaded concrete piles driven into soft soils

The evaluation of axial load resistance of piles driven into soft Louisiana soils based on reliability theory is presented. Forty-two square precast, prestressed, concrete piles that were tested to failure were included in the investigation. The predictions of pile resistances were based on static analysis (a-method for clay and Nordlund method for sand) and three cone penetration test (CPT) direct methods: the Schmertmann, De Ruiter-Beringen, and Bustamante-Gianeselli methods. In addition, dynamic measurements with signal matching analysis of pile resistances using CAPWAP, which is based on the measured force and velocity signals obtained near the pile top during driving, were evaluated. The Davisson and modified Davisson interpretation methods were used to determine the measured ultimate load-carrying resistances from pile load tests. The predicted ultimate pile resistances obtained by using the different prediction methods were compared with the measured resistances determined from pile load tests. Statistical analyses were carried out to evaluate the capability of the prediction design methods to estimate the measured ultimate pile resistance of driven piles. The results showed that the static method overpredicted the pile resistance, whereas the dynamic measurement with signal matching analysis (CAPWAP end-of-driving and 14-day beginning-of-restrike) underpredicted the pile resistance. Of the three direct CPT methods, the De Ruiter-Beringen method was the most consistent prediction method with the lowest coefficient of variation. Reliability-based analyses, by using the first-order, second-moment method, also were conducted to calibrate the resistance factors (phi) for the investigated pile design methods. The resistance factors for different design methods were determined and compared with AASHTO recommendation values. The calibration showed that De Ruiter-Beringen method has a higher resistance factor (phi(De Ruiter) = 0.64) than the other two CPT methods.