Abstract Efficient and reliable power management is essential for modern drilling operations, where generator demands are highly dynamic and engine conditions frequently change; conventional generator management, reliant on static heuristics and rated capacities, often leads to suboptimal utilization, excess fuel consumption, and increased maintenance costs and to address these challenges, we present an integrated, real-time system that adaptively estimates generator capacity and optimizes power management in drilling environments1. At the core of the system is an Iteratively Weighted Least Squares Regression (IWLSR) model that continuously updates generator capacity using real-time operational data: it's a method that, coupled with dynamic data binning, robust outlier filtering, and continuous model adaptation, delivers significantly improved estimation accuracy, ensuring operational decisions reflect true generator capabilities. The system also incorporates an artificial intelligence (AI)-driven predictive module employing a Bi-directional Recurrent Neural Network (RNN), trained on extensive rig data to forecast short-term peak power demands several minutes in advance. The RNN's ability to capture nonlinear and non-stationary trends enables proactive generator dispatch, mitigating the risks associated with lagging, reactive responses. An adaptive generator management framework integrates these estimation and prediction engines, employing a cost-based optimization algorithm that considers fuel efficiency, run hours, cycle counts, and thermal constraints. This approach dynamically selects the optimal generator set, minimizing unnecessary cycling and emissions while maximizing reliability and asset life. Field deployments across multiple rigs have demonstrated substantial benefits: reductions in generator-related blackouts, accurate early detection of engine anomalies, and fuel savings of 3% to 7%. The adaptive, real-time system is resilient to varying operational conditions, representing a step-change improvement over the state of the art. This paper details the methodologies, field validation results, and operational advantages of this integrated approach, highlighting its transformative potential for sustainability, reliability, and cost-effectiveness in drilling power management.
Schueren et al. (Mon,) studied this question.