Skip to main content Skip to secondary navigation

A deep learning modeling suite to provide fast prediction for CO2 storage problems

Main content start was developed by Gege Wen at Stanford University, advised by Prof. Sally M. Benson. CCSNet provides Synthetic Heterogeneous, Homogeneous, Purely layered, and User upload isotropic permeability maps. The isotropic cases are predicted with pre-trained convolutional neural network models [1]. Anisotropic permeability maps are supported with the Synthetic Heterogeneous option, predicted with pre-trained enhanced Fourier neural operators [2].

[1] Wen, Gege, Catherine Hay, and Sally M. Benson. "CCSNet: A deep learning modeling suite for CO2 storage." Advances in Water Resources 155 (2021): 104009. DOI:

[2] Wen, Gege, Zongyi Li, Kamyar Azizzadenesheli, Anima Anandkumar, and Sally M. Benson. "U-FNO–An enhanced Fourier neural operator-based deep-learning model for multiphase flow." Advances in Water Resources (2022): 104180. DOI: (arxiv, pressblogpost)