Deep Convolutional Autoencoders for Robust Flow Model Calibration under Uncertainty in Geologic Continuity

We demonstrate the robustness of VAE handles diverse features and the effectness of combining VAE with gradient-based inversion.

Prerequisites

Python 3.6

MATLAB

Tensorflow 1.13

The MATLAB Reservoir Simulation Toolbox (MRST)

Eclipse

Data

Due to the large size of data files, the data files (realizations and PCA basis) are not uploaded. Please email me ([email protected]) for the access to them.

Citation

Please cite our paper if you find the codes useful

@inproceedings{jiang2020history,

title={History Matching under Uncertain Geologic Scenarios with Variational Autoencoders},

author={Jiang, A and Jafarpour, B},

booktitle={ECMOR XVII},

volume={2020},

number={1},

pages={1--14},

year={2020},

organization={European Association of Geoscientists & Engineers}

}

Acknowledgments

The authors also thank Syamil Mohd Razak for helping build the three-dimensional case study for this work.