Surrogate Modeling Framework for Cantilever Sheet Pile Wall Stability Assessment: Case Study Campus Ullevål
摘要
The design of cantilever sheet pile walls (C-SPWs) is predominantly carried out in a deterministic manner, with limited explicit consideration of uncertainty and few systematic means for optimizing design solutions. This study addresses these limitations by developing and applying a surrogate modeling framework for the stability assessment of C-SPWs. The objectives are threefold: (i) to identify the governing geometrical and constitutive parameters controlling C-SPW stability; (ii) to generate response surfaces as practical engineering tools for balancing resources and design requirements in the selection of structural design variables; and (iii) to enable fully probabilistic, reliability-based stability assessment with explicit consideration of geotechnical parameter uncertainty. The framework combines strength-reduction finite element analyses with polynomial chaos expansion surrogate models to efficiently approximate the factor of safety over a defined design variable space. It is demonstrated using a real excavation project at Campus Ullevål (Oslo, Norway), supported by C-SPWs and lime–cement stabilization in sensitive marine clay. The results show that, for the investigated case, the undrained shear strength of the natural clay formations governs C-SPW stability, while the influence of lime–cement stabilization parameters is secondary within the considered design space. The surrogate models further enable rigorous uncertainty propagation and reliability analyses, clarifying how reductions in epistemic uncertainty contribute to meeting target reliability levels. The study provides practical guidance on the validity range of the adopted two-dimensional modeling assumptions and outlines future extensions to address imperfect interface behavior and spatial variability not captured in the present study.