Real-time prediction of formation lithology using drilling parameters: an example from Ca Tam oilfield

Abstract

Construction of stratigraphic column is an important stage in minerals exploration and researching the historical development of geological processes. Besides, determining and identifying the boundaries of lithological layers also helps a lot in minimizing the risk of drilling complications and incidents as well as increasing efficiency in drilling. In this study, the authors focus mainly on applying machine learning algorithms to classify lithology and identify stratigraphy directly from the real-time drilling data of 02 wells in the Ca Tam oil field. The proposed model has high accuracy, this result demonstrates the great superiority and effectiveness of applying this method. The model using the Fuzzy c-means algorithm has predicted and identified relatively accurately the three main lithological groups in the study area: sandstone, claystone, and clay. The study's encouraging findings demonstrate the need for further focus and funding on this new strategy in the future to raise the effectiveness of oil and gas well drilling in Vietnam.