Application of Machine Learning to Predict Transient Sand Production in the Karazhanbas Oil Field, Ustyurt–Buzachi Basin (West Kazakhstan)

Shabdirova A. Kozhagulova A. Minh N.H. Zhao Y.
October 2023 Springer

Natural Resources Research
2023 #32 Issue 5 1975 - 1986 pp.

The aim of this research is to investigate the potential of machine learning (ML) in predicting sand production behaviors during oil extraction from an unconsolidated sandstone reservoir in Kazakhstan. The study is based on data from 43 wells, which include crucial parameters like reservoir depth, thickness of producing zone, fluid flow rate and water cut value. The study focused on three types of sand production behavior, namely transient sand production (TR), multiple-peak sand production and non-TR. The research utilized the random forest (RF) algorithm to forecast sanding behavior based on production and field parameters. The results of the study demonstrate that the RF algorithm is capable of predicting TR behaviors accurately, particularly when the training dataset includes information from TR wells. This information is useful in developing effective sand management strategies. The algorithms accuracy was highest when the complete set of input data was used, although the thickness and depth of the reservoir are less critical for its performance. This study is unique in its attempt to predict sand volume using ML algorithms, as earlier studies had concentrated solely on forecasting the sanding onset.

Karazhanbas oil field , Machine learning , Sand production , Sand volume

Text of the article Перейти на текст статьи

School of Engineering and Digital Sciences, Nazarbayev University, Astana, Kazakhstan
Oil and Gas Faculty, Atyrau Oil and Gas University, Atyrau, Kazakhstan
Fulbright University Vietnam, Ho Chi Minh City, Viet Nam

School of Engineering and Digital Sciences
Oil and Gas Faculty
Fulbright University Vietnam

10 лет помогаем публиковать статьи Международный издатель

Книга Публикация научной статьи Волощук 2026 Book Publication of a scientific article 2026