Machine Learning Forecasts Oil Rate in Mature Onshore Field Jointly Drivenby Water and Steam Injection

摘要

In this paper,we tackle an old problem - production forecast-using techniques that are relatively newto the reservoir engineer toolbox.The problem at hand consists of forecasting oil production in a matureonshore field simultaneously driven by water and steam injection.However,instead of turning to traditionalmethods,we deploy machine-learning algorithms which will feed on a plethora of historical data to extracthidden patterns and underlying relationships with a view to forecasting oil rate.No geological model and/ornumerical reservoir simulators will be needed,only 3 sets of time-series:injection history,production historyand number of producers.Two Machine-Learning algorithms are used:linear-regression and recurrentneural networks.