Characterization and predictive modeling of directional piling lateral offsets for offshore conductors

Abstract

Directional piling refers to a construction method in which the conductor shoe is pre-chamfered to facilitate directional deviation during the piling process. It effectively avoids inter-well collisions during operations. An experimental setup was designed to clarify the offset mechanism of offshore conductors. Directional piling for conductors was tested on a small scale. The study investigated the effects of shoe angle, penetration depth, conductor size, and soil properties on the offset characteristics during directional piling. The results showed that all these parameters have a strong correlation with the conductor's offsets. Numerical simulations were additionally conducted based on the experiments. A robust dataset was constructed by combining experimental results and numerical simulation data. Following feature extraction and parameter preprocessing, a directional piling offset prediction model for offshore conductors was developed using the multilayer perceptron (MLP) regression model. A 97 % fitting rate was attained by the prediction model. The model was applied to a well group in the South China Sea. The prediction model's accuracy was validated when the lateral offset calculation error was kept within ±0.2m of the measured data. The research findings will provide higher precision and efficiency for offshore well construction operations, offering a solid theoretical foundation for conductor anti-collision strategies.