Design of Ram-Compacted Bearing Base Piling Foundations by Simple Numerical Modelling Approach and Artificial Intelligence Technique

Abstract

Over the past 20\xa0years or so, ram-compacted bearing base (RBB) piling foundations have been widely used in China as one of the typical enlarged base piling solutions that provide higher bearing capacity at lower costs compared to traditional piles and other enlarged base piles. However, the current design of RBB piles relies on semi-empirical equations that are typically conservative and inconsistent, often necessitates costly pilot pile loading tests before construction. This paper proposes a design method for RBB piles based on finite-element (FE) numerical modelling, validated with historic field cases, thus mitigating the limitations associated with the available empirical methods. The study also uses the developed FE modelling process to perform a comprehensive parametric study simulating a set of 1008 combinations of hypothetical data cases involving a wide range of various RBB piling configurations and soil types. The database obtained from the parametric study was then used in a sophisticated artificial intelligence/machine learning analysis via evolutionary polynomial regression (EPR) technique to return a standalone design tool in the form of relatively simple equations for predicting the RBB pile capacity. The developed EPR equations were validated with three historic cases, and the results confirmed reliable predictions.