A comparative analysis and rapid performance prediction of polymer flooding process by coupling reservoir simulation with neural networks

Accelerated technological progresses offer massive amounts of data, prompting decision making for any asset to be more complicated and challenging than before. Data-driven modelling has gained popularity among petroleum engineering professionals by turning big data into valuable insights that introduces fast and reliable decision making. In this study, a viscosifying polymer flooding performance-forecasting tool is developed using an artificial neural network-based data-driven model. A wide variety of reservoir and operational scenarios are generated to inclusively cover possible conditions of the process. Each scenario goes through no injection, water-only flooding, polymer followed by waterflooding and polymer-only flooding schemes. Neural network models were trained with three representative performance indicators derived from simulator outputs; efficiency, water-cut and recovery factor. Practicality of the tool in assessing probabilistic and deterministic predictions is demonstrated with a real polymer-flooding case of Daqing Oil Field.