Quay walls are earth-retaining structures that provide berthing for ships, enabling safe loading, unloading, and mooring. As vital port infrastructure, they must withstand increasing depths, heavier loads, and complex conditions. Finite Element Modelling (FEM) is widely used to study their behavior, but accuracy is limited by uncertainties in soil properties, constitutive models, and structural assumptions.

This study develops and validates a FEM of a smart quay wall in the Amaliahaven Project, Port of Rotterdam, using field monitoring data from staged dredging. The quay wall is instrumented with inclinometers that recorded lateral displacements during dredging. Geotechnical parameters were derived from Cone Penetration Tests (CPTs) and empirical correlations, with the Hardening Soil (HS) model adopted as the primary constitutive model. The HSsmall model was also tested to evaluate the influence of small-strain stiffness.


Initial FEM results underestimated displacements. Sensitivity analysis identified friction angles and stiffness moduli of key layers as the most influential parameters. Guided calibration using inverse analysis improved agreement, achieving accuracy within ±30% across dredging phases.

A key discrepancy involved front wall rotation. The assumed hinge connection between front and combi wall produced unrealistic behavior, while modelling it as fixed reduced errors to 13%, suggesting actual site conditions lie between hinge and rigid assumptions. Comparison of HS and HSsmall showed small-strain stiffness had minimal influence due to the high stiffness of sand layers.

A parametric study examined dredging depth, surcharge loading, and soil variability. Dredging beyond −21 m NAP and surcharge loads above 40 kN/m² significantly increased wall displacements, while anchor forces always remained below design capacity, confirming structural safety.

This research demonstrates that combining sensitivity analysis with calibration enhances FEM accuracy and highlights the importance of realistic connection modelling.