Recognizing abnormal fluid influx is the first and most important step for well control safety. However, due to the complexity of the wellbore fluid mechanism, uncertainty of the formation pressure and noises in the real-time measurement data, it is difficult to detect kicks by conventional approaches. The purpose of this paper is to propose a knowledge-data hybrid approach and build a reliable and easy-to-use tool to automatically detect kicks. Based on the knowledge of the kick main indicators, this work utilizes Long-Short moving window average algorithm combining advanced filtration technology to capture subtle change trends precisely and timely in noisy real-time measured surface data during drilling. Self-adaptive kick threshold value adopting approaches are proposed to improve the accuracy of kick detection for different data measuring technologies. The knowledge and inference logic of specific drilling operations and phenomena, such as wellbore ballooning effects and drilling fluid poring, are embedded in the data analysis technologies to reduce the false alarms of the kick detection system. A reliable and practical intelligent kick detection tool is built based on this knowledge-data-dual-driven method. The must-have input set of this tool only contains 8 channels, including pit volumes, outlet flowrate, stand pipe pressure, and etc, which makes the tool flexible and widely applicable. The tool consumes both real-time and archived drilling data, and provide audible and easy-to-understand visual alarms when kick is detected. Moreover, the tool supports different outlet flowrate measurement technologies, including Paddle meter, Radar meter and Coriolis meter. Extensive testing process using historical data manifests that the kick detection rate is 100% (correctly detected 16 kicks out of total 16 kicks), and the overall detection time is 6.1 mins earlier comparing to that of conventional methods. The tool was also deployed in a real-time operation center for more comprehensive verification. During the trial, 1 kick was experienced which was detected accurately and timely by the tool. The false alarm rate of the tool is lower than 3.5 times per 12 drilling hours for paddle meters. According to the demonstrated kick detection rate and false alarm rate, the tool is promising and beneficial for detecting kicks in early time automatically, which improves the well control safety and enables further automatic well control process.

You can access this article if you purchase or spend a download.