Dataset for: Data-Driven Characterization of PISA-Based Soil Reaction Curves for Large Diameter Monopiles in Sand

Large diameter monopiles are the prevailing foundation concept for offshore wind turbines. While the PISA (Pile Soil Analysis) methodology has established a new standard for accurately predicting the lateral response of these rigid piles, its application typically requires computationally time consuming 1D Finite Element Analysis (FEA) for site-specific calibration. This study presents a data-driven methodology to derive simplified, PISA-consistent design equations for cohesionless soils, eliminating the need for complex numerical simulations in routine design. A comprehensive dataset consisting of 1,000 numerical case studies was generated, covering a wide range of soil stiffness, internal friction angles, and pile geometries. Using automated data mining algorithms, the study analyzed over 115,000 individual data points representing the lateral soil reaction. A hyperbolic model was employed to characterize the non-linear soil response, and the key parameters—initial stiffness (Kini) and ultimate resistance (pult)—were extracted using non-linear least squares optimization. Multiple log-linear regression analysis was performed to establish robust correlations between the PISA-based curve parameters and the soil-geometry properties. The derived model was optimized by enforcing physical constraints to ensure geotechnical consistency. The proposed formulations achieved exceptional statistical accuracy, with coefficients of determination (R2) of 0.99 for ultimate resistance and 0.90 for initial stiffness. These results demonstrate that the developed semi-empirical equations provide a practical and highly accurate tool for engineers to estimate PISA-quality soil reactions for preliminary design.