Variability Analysis of Bearing Capacity of Piles with Slowly Varying Q–s Curves

Ultimate bearing capacity of foundation pile is of great importance for pile design and optimization, but it may not be feasible due to economic and safety considerations, especially the destructive test. A multi-parametric inversion method is proposed based on a hybrid algorithm that combines SA (simulated annealing algorithm) with Powell’s method, which utilizes the non-failed static load–settlement (Q–s) curves of 36 pipe piles from a specific field to invert the 16 surrounding soil parameters from a global optimization perspective. The results demonstrate that the maximum displacement value of the pile head and actual load-holding duration have negative correlation with predicted ultimate bearing capacity of piles with the same type at the same site. Their values range from 1.3 to 2.1 times of the ultimate bearing capacity of the pile body material based on the soil parameters inversion, which contributes to optimizing pile foundation design and furnishing data-driven support to AI algorithms for addressing intricate nonlinear problems.