Ensemble Machine Learning Assisted Reservoir Characterization Using Field Production Data–An Offshore Field Case Study

Estimation of fluid saturation is an important step in dynamic reservoir characterization. Machine learning techniques have been increasingly used recent years for prediction workflows. However, most these studies require input parameters derived from cores, petrophysical logs, or seismic data, which may not always be readily available. Additionally, very few incorporate the production reflection properties and also typically frequently reliably measured quantity throughout life a field. In this research, random forest ensemble machine algorithm implemented that uses field-wide injection data (both at surface) as only to predict time-lapse oil profiles well locations. The optimized using feature selection based on importance score Pearson correlation coefficient, combination with geophysical domain-knowledge. workflow demonstrated actual field structurally complex, heterogeneous, heavily faulted offshore reservoir. model captures trends three half historical production, injection, simulated future four deviated locations over 90% R-square, less than 6% Root Mean Square Error, 7% Absolute Percentage each case.