Evolutionary Polynomial Regression-Based Models to Estimate Stability of Gravity Hunched Back Quay Walls

Abstract

This paper presents a new approach, based on Evolutionary Polynomial Regression (EPR), for predicting the safety factors of a quay wall against sliding, overturning and bearing capacity failure as functions of the soil’s shear strength parameters, geometry of the wall and loading conditions. To this end, a database of around 80000 data sets was created based on a conceptual model, employing a MATLAB-aided program. Based on input and output values of this database and employing EPR, three different models for the estimation of safety factors were developed and their results were compared with the values in the database. Investigation into the performance of the developed models indicates that these models are capable of estimating the stability of quay walls with a precision of around 95%. Parametric analyses were performed on the models to identify parameters with a key role in the stability of quay walls. The parametric studies were indicative of the models’ ability in capturing the effects of individual parameters on the wall safety factors, therefore proving the models helpful as tools in the preliminary design of gravity quay walls.