Calibration of partial factors for the pile efficiency serviceability limit state design in geosynthetic-reinforced and pile-supported embankment

Geosynthetic-reinforced and pile-supported (GRPS) embankments are widely adopted on soft ground, where non-uniform load transfer governs pile performance and serviceability. Reliable estimation of pile efficacy is therefore essential, yet the BS 8006 and CUR 226 design methods have not been systematically evaluated or calibrated against field observations. This study develops a reliability-based serviceability design framework for pile efficacy in GRPS embankments through calibration of correction and partial factors. A database comprising 34 independent data points from 23 instrumented full-scale embankments across 18 projects is assembled to assess the predictive performance of existing models. Correction factors are introduced for both the arching-induced pile efficacy Ea and the total pile efficacy E, resulting in improved accuracy and removal of systematic bias. Model biases are statistically characterized under different levels of design information, and corresponding action and resistance factors are calibrated. A design example demonstrates the practical implementation of the proposed framework. The results show that the calibrated models provide unbiased and reliability-consistent estimates of pile efficacy, offering a rational basis for serviceability design of GRPS embankments.