Probabilistic analysis of suction caisson foundations for offshore wind turbines under lateral loading considering spatial variability of the critical state stress ratio via SANISAND-MS model

Suction caissons are widely used as offshore foundations due to their ease of installation and cost-effectiveness. Conventional design methods often overlook soil spatial variability, which can lead to inaccurate estimates of ultimate limit state (ULS) reliability. This study employs a random finite element method combined with Monte Carlo simulation to investigate how the spatial variability of the critical state stress ratio $M_c$, as defined in the SANISAND-MS constitutive model, affects the response and ULS reliability of suction caissons in sandy soils under horizontal monotonic loading. The results reveal that $M_c$ plays a complex role in governing soil behavior, influencing both the critical friction angle and the dilatancy surface. A higher $M_c$ delays the onset of dilation, promoting contractive behavior, and leads to larger rotation of the suction caisson under horizontal loading. The spatial variability of $M_c$, characterized by the coefficient of variation (COV) and autocorrelation distance $\delta$, significantly affects both the variability and mean response of the foundation. As COV increases, the heterogeneity of $M_c$ intensifies, leading to greater scatter in caisson rotation but a reduced mean rotation compared to the homogeneous case. In contrast, increasing $\delta$ amplifies response variability while leaving the mean rotation largely unchanged. Moreover, the probability of failure $P_{\text{f}}$ is found to decrease with COV but increase with $\delta$. Neglecting the spatial variability of $M_c$ can therefore result in substantial overestimation of $P_{\text{f}}$, resulting in overly conservative and unnecessarily costly foundation designs. These findings highlight the importance of incorporating spatial variability in offshore foundation design to achieve more accurate and economical outcomes.