• Stratigraphy-aware random fields quantify spatial variability in adjacent pits. • Homogeneous models underestimate max wall displacements by 28–50%. • Kriging and random field analyses reveal localized high-risk deformation zones. • Simultaneous excavation increases deformation interaction by 30–40%. • Provides a probabilistic workflow for risk assessment in variable soils. Urban deep excavations in soft soils face significant deformation risks due to spatial variability in ground conditions, yet conventional deterministic models often oversimplify this heterogeneity. This study systematically investigates the influence of spatial variability on the deformation behavior of adjacent foundation pits, a prevalent yet underexplored scenario in urban geotechnical projects. Building on established geostatistical approaches, this study introduces stratigraphy-aware random fields that maintain geological consistency while capturing parameter uncertainty. Using field data from Nanjing as a case study, this study compares three modeling approaches: homogeneous, Kriging-interpolated, and random field-based simulations. The findings indicate that conventional homogeneous models underestimate maximum wall displacements by 28-50%, with random field analyses revealing localized high-risk zones undetectable through traditional methods. Excavation sequence analysis further shows that simultaneous excavation increases deformation interactions by 30-40% compared to staggered approaches. These findings provide engineers with a validated workflow for probabilistic risk assessment, particularly for projects requiring deformation control in variable soils. The study advances geotechnical practice by demonstrating how spatial variability quantitatively impacts excavation performance, while the methodology offers adaptable tools for additional geotechnical applications involving soil heterogeneity.
Shad et al. (Sun,) studied this question.