Gas kick and lost circulation risk identification method with multi-parameters based on support vector machine for drilling in deep or ultradeep waters

Abstract

In deepwater and ultra-deepwater drilling, formation pressures are complex and the drilling fluid density window is narrow, which makes it prone to well control incidents such as kicks and lost circulation. To ensure safe, efficient, and cost-effective drilling operations, real-time monitoring of kick and lost circulation incidents is essential. This paper, based on the classical symbolic aggregate approximation (SAX) method, comprehensively considers the average difference and slope characteristics of time series data for kick and lost circulation monitoring parameters, and proposes an improved SAX method for kick and lost circulation risk identification. Additionally, the paper establishes “kick standards” and “lost circulation standards” models, calculating the similarity measurement distance between the monitoring parameters and the standard models. These similarity measures are used as feature vectors in a support vector machine to develop a multi-parameter collaborative risk identification model for kicks and lost circulation. Verification with data from two wells in the South China Sea demonstrates that the model can accurately identify kick and lost circulation incidents, providing a new approach for real-time risk identification in deepwater drilling.