Enhancing Real-Time Stuck Pipe Prediction Using Trigger Coefficients: A New Approach for Realtime Well Engineering

Abstract

Abstract Conventional approaches to addressing stuck pipe scenarios in oil field operations have typically depended on established tools and procedures. While previous research has delved into the causes of drill pipe sticking and suggested adjustments in operational methodologies, current methods often fall short in predicting incidents comprehensively, particularly in light of diverse wellbore conditions. This paper presents a novel methodology that merges principles of physics, data science, and uncertainty modeling to offer more resilient and precise solutions for managing real-time pipe sticking occurrences. The methodology put forward embraces diverse modes of pipe sticking mechanisms, encompassing mechanical, geometrical, differential, keyseat, cuttings packoff, and geomechanical factors. Individual coefficients are assigned to monitor various parameters, with their weights dynamically tuned in real-time according to operational conditions and surface/downhole measurements. The independence of these parameters facilitates the prediction of mechanisms that contribute to pipe sticking occurrences. By integrating engineering and data science models through microservices, multiple calls during calculations are enabled, accommodating the expected nonlinear behavior of variables across various drilling conditions. In real-time analysis, the findings indicate a fast convergence towards optimal prediction, ensuring predictability and favorable outcomes, particularly in test wells with inherent uncertainties. Interestingly, the study sheds light on the multifaceted factors contributing to pipe sticking conditions, even when initially perceived as differential sticking. Notably, in wells with minor doglegs, secondary geometric factors such as borehole torsion emerge as significant contributors to sticking. This leads to increased resistance against pipe movement in both axial and radial directions, elucidating the role of mechanical factors outlined by the well profile energy coefficient. Moreover, trend analysis reveals a rising trend in pipe sticking conditions, attributed to uncertainties in input data. Furthermore, the integration of high-resolution lithology inputs enhances the accuracy of stuck pipe predictions. The prescriptive analytics approach comprises two modules: an inferencer and a recommender, providing operational parameters for intervention. The inference output offers practical recommendations to mitigate specific stuck pipe conditions. Additionally, in certain wells, the pack-off model and risk calculations incorporate insights from three primary models: cuttings bed buildup, caving volume concentration, and heightened friction on the drill string.