AI-based simulation of settlement and failure modes in additively manufactured models of soft soil foundations

Scaled 3D-printed models of soft soil foundations, present new opportunities for researching failure factors and settlement behaviour under various load scenarios. Simulation Framework based on Artificial Intelligence (SF-AI) is presented to investigate settlement and forecast failure modes in soft soil foundations. With the goal to improve forecast accuracy and reduce need on laborious physical testing, data-driven learning algorithms are combined with numerical and experimental observations. To detect non-linear correlations between material qualities, applied loads, and foundation reactions, deep neural networks and ensemble methods are trained by high-resolution settlement data and deformation patterns obtained from 3D-printed test models. By evaluating temporal and geographical trends in displacement fields, AI-driven system might predict shear failure, bearing capacity collapse, and excessive differential settlement. The proposed method speeds soft soil foundation provides a digital twin for geotechnical verification model and improvement. To enhance geotechnical engineering safety, dependability, and sustainability, additive manufacturing and AI-driven simulation are suggested to anticipate settlement and failure in complex soil–structure systems. SF-AI, foundation modelling is correct which grows from 4.8 to 12.0 mm with load, decreases with width and stiffness, differential settlement is 0–1.9 mm, shear strain 0.45–1.8%, prediction errors 2–3% and safety factors drop 1.35–0.9.