Data-driven Bayesian network model for early kick detection in industrial drilling process

Kick, or hydrocarbon influx, is one of the significant challenges during the drilling operation. A kick happens when the formation pressure exceeds the hydrostatic pressure of mud weight. Detection of a kick at an early stage spares more time to take necessary actions to prevent its growth and mitigate the potential well blowout. There are varieties of methods applied for early kick detection. The conventional method entails monitoring surface parameters which leads to delay in the detection. Some recent works show the ability to employ monitoring of downhole parameters to realize early kick detection. Data-driven Bayesian Network (BN) has shown to solve problems in complex systems where the knowledge about the system is not adequate to apply a model-based method. Data-driven BN creates a model based on historical data, which is usually available, unlike expensive, and often insufficient, expert knowledge. Using the data obtained in a laboratory-scale experiment, this paper presents the application of data-driven BN model in using downhole parameters to early kick detection. (C) 2020 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.