CCSNet: a deep learning modeling suite for CO2 storage

Gege Wen, Catherine Hay, Sally M. Benson, CCSNet: a deep learning modeling suite for CO2 storage, Advances in Water Resources(2021), doi:https://doi.org/10.1016/j.advwatres.2021.104009

Data sets

The data set is available at https://drive.google.com/drive/folders/1SVZFkaxkAIjcGKew3rzGTmKW5tSBUGf7?usp=sharing

Note: variable name is same as file name.

Train:

• Input: train_x.h5
• Gas saturation: trian_y_SG.h5
• Reservoir pressure: train_y_BPR.h5
• Initial pressure: train_y_P_init.h5
• xCO2 molar fraction: train_y_BXMF.h5
• yCO2 molar fraction: train_y_BYMF.h5
• Liquid phase density: train_y_BDENW.h5
• Gas phase density: train_y_BDENG.h5

Test:

• Input: test_x.h5
• Gas saturation: test_y_SG.h5
• Reservoir pressure: test_y_BPR.h5
• Initial pressure: test_y_P_init.h5
• xCO2 molar fraction: test_y_BXMF.h5
• yCO2 molar fraction: test_y_BYMF.h5
• Liquid phase density: test_y_BDENW.h5
• Gas phase density: test_y_BDENG.h5

Pre-trained models

The data set is available at https://drive.google.com/drive/folders/1SVZFkaxkAIjcGKew3rzGTmKW5tSBUGf7?usp=sharing

• Gas saturation: trained_models/SG_v1.h5
• Pressure buildup: trained_models/dP_v1.h5
• xCO2 molar fraction: trained_models/bxmf_v1.h5
• yCO2 molar fraction: trained_models/bymf_v1.h5
• Liquid phase density: trained_models/bdenw_v1.h5
• Gas phase density: trained_models/bdeng_v1.h5

Requirements

• tensorflow-gpu==1.15
• numpy==1.19.0

Web application

The pre-trained models are hosted at ccsnet.ai.

Please refer to https://youtu.be/5bIlfjyo6Jkfor a video demonstration.