Enhancing Energy Efficiency and Operability in Offshore Drilling Operations: Validation and Benefits of a Machine Learning-Based Vessel Motion Prediction Application

Abstract This paper delves into the framework of a new vessel motion prediction application that leverages advanced machine learning (ML) techniques in conjunction with metocean forecasts to predict vessel motions as well as thruster loads. The paper illustrates the validation process, comparing predicted vessel motions and thruster utilization against actual records across several time intervals within a drilling campaign. The discussion encompasses the key role of diverse input factors in shaping prediction accuracy, and outlines strategies to address the inherent uncertainties associated with methods, weather forecasts, and the stochastic nature of the underlying problem. Our innovative approach for vessel motion prediction forecasting employs physics-informed neural networks. These networks were trained using second-order time domain simulations, encompassing environmental variables like waves, wind, and current, along with a vessel dynamic positioning (DP) control system model. Our methodology marks a significant advancement in the field, offering a robust framework for efficiently developing performant motion prediction models grounded in the underlying physics. The validation procedure yielded promising results, with the predicted responses trending well with the observed responses, and remaining near the expected range of responses attributed to machine learning model uncertainty and stochastic uncertainty innate to the marine environment. Discrepancies are noted and addressed with possible ways forward to improve the model. Central to the paper's narrative is the exploration of the benefits stemming from the motion prediction application's implementation. The application's capacity to enhance operational and energy efficiency is highlighted in this paper. This work contributes to making sustainable drilling practices available by providing a tool predicting the future vessels response both in terms of motions and thruster loads which can be used to optimize the power plants efficiency while maintaining safe operations. This technology will contribute to bringing the industry closer to its objectives of reduced environmental impact.