Application of machine learning algorithm to forecast production for fracture basement formation, Central arch, Bach Ho field

Abstract

Oil production forecast is a big challenge in the oil and gas industry. Simulation model and prediction results play an important role in field operation and management. Currently, dynamic simulation model, decline curve analysis are popular tools applied to forecast production. The dynamic simulation model shows a remarkable effect for sedimentary objects. However, production forecasting by this method for fracture basement formation sometimes gives unreliable results because the fracture basement formation is a complex of geological objects, which causes difficulties in predicting the geological characteristics. The decline curve analysis (DCA) method uses simple extrapolated mathematical functions to forecast oil production, therefore the results do not reflect the production operations such as opening/closing production interval.To avoid the disadvantages of these traditional methods, Vietnam Petroleum Institute (VPI) has studied the applicability of machine learning to forecast oil production for fracture basement formation of Bach Ho field. The study results show that the random forest model has improved the production forecast with low relative error (4%).