Time-Series Forecasting of a CO2-EOR and CO2 Storage Project Using a Data-Driven Approach

This study aims to develop a predictive and reliable data-driven model for forecasting the fluid production (oil, gas, water) of existing wells future infill CO2-enhanced oil recovery (EOR) CO2 storage projects. Several models were investigated, such as auto-regressive (AR), multilayer perceptron (MLP), long short-term memory (LSTM) networks. The trained based on static dynamic parameters daily while considering inverse distance neighboring wells. developed evaluated using walk-forward validation compared quality metrics, span, variation in horizon. AR demonstrates convincing generalization performance across various time series datasets with but varied horizon eight LSTM has shorter strong generalizability robustness consistency. MLP shortest most other models. exhibits promising when is from an well similar features well. offers alternative physics-driven traditional modeling costly laborious.