Gas kicks are often encountered during drilling the oil & gas formations. This paper proposes a data-driven method by employing machine learning for real-time detection of gas kicks. Firstly, logging data is recorded and compiled for gas cases. Secondly, artificial neural network is employed to detect gas kicks. Meanwhile, principal components analysis is used to reduce the dimensionality of parameters. At last, three gas kick accidents are studied to verify the power of the technology used. The method is shown to be highly reliable for identifying the gas kicks accidents at a shorter time than the reported detection using conventional techniques.
At present, the trend of oil exploration and development around the world is toward the ocean to deep water, and deepwater drilling faces many challenges such as high wellbore temperature pressure control requirements, poor borehole stability, narrow drilling fluid safety density window, and many other challenges. The deepwater environment makes the well control more complicated, and the drilling risk increases sharply. Once the blowout is out of control, the damage is huge, which will not only cause huge economic losses and casualties, but also cause great damage to the offshore ecological environment (BP, 2010; Zhang, Zhang, Yue, et al, 2018).
Therefore, the prevention and control of deepwater oil and gas well overflow and blowout is a major problem that needs to be solved urgently in offshore oil and gas development. Early detection of overflow (especially gas kick) plays an extremely important role in well control. Gas kick is the induction of a kick or directly leads to a blowout.
One of the main factors, and avoiding and preventing blowout accidents is an important guarantee for safe drilling. If the gas can be detected immediately after entering the wellbore, and effective measures are taken in time, not only can the blowout be avoided, but also the downhole complex accidents occurring in the well control process can be obviously reduced, thereby protecting the oil and gas layer, increasing the drilling rate and ensuring the drilling efficiency.