Abstract

Elastic properties of carbonate reservoir and nonreservoir (tight carbonate, shale etc.) may overlap which make it difficult to characterize the carbonate reservoir using conventional interpretation techniques. Artificial intelligence (AI)/machine learning (ML) based approach of G&G data integration & interpretation are making pace in E&P business and is helping a lot to utilize voluminous available G&G data to decipher valuable information. Present study is an example of use of Machine Learning (ML) techniques to delineate the carbonate reservoir in Bassein Formation of Vasai East field of Western Offshore Basin, Mumbai. Bassein Formation of Mid to Late Eocene age is dominated by limestone and proved to be hydrocarbon bearing in Vasai East field. Negative surprises also occurred in the field due to facies variation. Porous limestone of Bassein Formation and intervening/overlying shale with significant overlap in P-impedance could not be discriminated using inversion studies. Although the use of ML approach helped to capture the hidden pattern in data and to delineate the reservoir. A 3-D facies volume was generated using ML technique with three types of facies i.e., porous limestone, tight limestone and shale. Based on the study, polarity reversal due to lateral lithological variation was also identified and mapped. Structural interpretation were refined by incorporating the polarity reversal. Attribute analysis and paleo-structural analysis based on refined interpretation is used to know the reservoir extent. Facies volume along with paleo-structure analysis is used to understand the depositional setup and hydrocarbon distribution in the study area. Results of the study can be used to mitigate the risk in exploration and exploitation of hydrocarbon.