Long-term Oil Production Forecasting A Comparative Analysis of Reservoir Simulation Models and Machine Learning Approaches

Abstract

Our research contributes valuable insights into the application of machine learning, particularly LSTM, for long-term oil production prediction. The comparative analysis of different models sheds light on their respective strengths and weaknesses. As we move forward, the integration of multi-variate prediction models holds the potential to further refine and advance the accuracy of predictions in the dynamic field of oil production. Besides, the algorithm can be checked over gas wells to see effect of the model in that way. As mentioned previously, the gas wells contain high degree of cyclicity due to seasonality phenomena. As an additional information for long-short term memory machine learning algorithm, it is believed that it might have a positive impact on the predictions.