Enabling site-specific well leakage risk estimation during geologic carbon sequestration using a modular deep-learning-based wellbore leakage model

Abstract

Amid growing climate concerns, geologic carbon sequestration (GCS) is a promising technology for mitigating net carbon emissions by storing CO2 in reservoirs. Oil and gas brownfields are an attractive option for CO2 storage, but these sites have many historical wellbores from petroleum production that can provide potential leakage pathways for CO2 or formation brine. Therefore, risk management of GCS operations requires an assessment of potential well leakage. Due to the high uncertainty of subsurface systems, stochastic approaches are ideal for quantifying the range of risk behaviors, but they must be computationally efficient in the face of complex physics. Here, we develop a new deep learning wellbore model to predict the leakage of CO2 and brine through wellbores. Full physics numerical simulations were used to generate data sets. A complex regression problem was divided into sub-problems of classification and regression to improve model performance. Feature analysis quantifies the impact of each feature on model prediction and enables us to select only effective features for model training. The model shows high predictive performance across a wide range of geologic and injection conditions and well attributes. A case study illustrates how the model is applied to assess well leakage in GCS operations.