Three-Parameter Lognormal Distribution to Estimate Ultimate Bearing Capacity of Pile Foundations with Extrapolation of Load-Settlement Curves

The bearing capacity of piles has a large variability when estimated from prior geotechnical investigation alone. On-site tests are effective in geotechnical risk management, particularly for ground whose performance is not well known and/or for new piling methods. The authors have been attempting to develop simple static load tests with press-in piles. However, there are issues in estimating the ultimate bearing capacity; (1) it is difficult to apply a large load owing to the limitation of piling machines and reaction piles; (2) excessive load may affect the performance of piles. Thus, there is an urgent need for a method to extrapolate load-settlement curves and estimate ultimate capacity. Accordingly, we investigated the relationship between the measured ultimate capacity, the estimated ultimate capacity by Chin's extrapolation, and the maximum load used for the estimation, based on a database of pile load tests. It was found that the ratio of the measured load to the estimated one follows a three-parameter (TP) lognormal distribution well with the maximum load as the lower bound. Moreover, the effect of the TP lognormal distribution on the reliability of the structure is modeled and discussed.

Keywords: pile foundation, bearing capacity, probability distribution, load-settlement curve, reliability-based design