A Machine Learning Approach for Gas Kick Identification

C. E. Obi*, Y. Falola, K. Manikonda, A. R. Hasan, I. G. Hassan, M. A. Rahman

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

8 Citations (Scopus)

Abstract

Warning signs of a possible kick during drilling operations can either be primary (flow rate increase and pit gain) or secondary (drilling break and pump pressure decrease). Drillers rely on pressure data at the surface to determine in-situ downhole conditions while drilling. The surface pressure reading is always available and accessible. However, understanding or interpretation of this data is often ambiguous. This study analyzes significant kick symptoms in the wellbore annulus both under static (shut in) and dynamic (drilling/circulating) conditions. We used both supervised and unsupervised learning techniques for flow regime identification and kick prognosis. These include an artificial neural network (ANN), support vector machine (SVM), K-nearest neighbor (KNN), decision trees, K-means clustering, and agglomerative clustering. We trained these machine learning models to detect kick symptoms from the gas evolution data collected between the point of kick initiation and the wellhead. All the machine learning techniques used in this work made excellent predictions with accuracy greater than or equal to 90%. For the supervised learning, the decision tree gave the overall best results, with an accuracy of 96% for air influx cases and 98% for carbon dioxide influx cases in both static and dynamic scenarios. For unsupervised learning, K-means clustering was the best, with Silhouette scores ranging from about 0.4 to 0.8. The mass rate per hydraulic diameter and the mixture viscosity yielded the best types of clusters. This is because they account for the fluid properties, flow rate, and flow geometry. Although computationally demanding, the machine learning models can use the surface/downhole pressure data to relay annular flow patterns while drilling. There have been several recent advances in drilling automation. However, this is still limited to gas kick identification and handling. This work provides an alternative and easily accessible primary kick detection tool for drillers based on data at the surface. It also relates this surface data to certain annular flow regime patterns to better tell the downhole story while drilling.

Original languageEnglish
Pages (from-to)663-681
Number of pages19
JournalSPE Drilling and Completion
Volume38
Issue number4
DOIs
Publication statusPublished - Dec 2023
Externally publishedYes

Fingerprint

Dive into the research topics of 'A Machine Learning Approach for Gas Kick Identification'. Together they form a unique fingerprint.

Cite this