Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2022 Mar 31;12(1):5486.
doi: 10.1038/s41598-022-09495-0.

Machine learning to predict effective reaction rates in 3D porous media from pore structural features

Min Liu  1   2 , Beomjin Kwon  3 , Peter K Kang  4   5
Affiliations

Machine learning to predict effective reaction rates in 3D porous media from pore structural features

Min Liu et al. Sci Rep. .

Abstract

Large discrepancies between well-mixed reaction rates and effective reactions rates estimated under fluid flow conditions have been a major issue for predicting reactive transport in porous media systems. In this study, we introduce a framework that accurately predicts effective reaction rates directly from pore structural features by combining 3D pore-scale numerical simulations with machine learning (ML). We first perform pore-scale reactive transport simulations with fluid-solid reactions in hundreds of porous media and calculate effective reaction rates from pore-scale concentration fields. We then train a Random Forests model with 11 pore structural features and effective reaction rates to quantify the importance of structural features in determining effective reaction rates. Based on the importance information, we train artificial neural networks with varying number of features and demonstrate that effective reaction rates can be accurately predicted with only three pore structural features, which are specific surface, pore sphericity, and coordination number. Finally, global sensitivity analyses using the ML model elucidates how the three structural features affect effective reaction rates. The proposed framework enables accurate predictions of effective reaction rates directly from a few measurable pore structural features, and the framework is readily applicable to a wide range of applications involving porous media flows.

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interests.

Figures

Figure 1
Figure 1
The normalized local surface reaction rates shown in 2D slices (top) of two different 3D porous media (bottom).
Figure 2
Figure 2
The schematic of the ML-based framework that combines 3D pore-scale reactive transport, Random Forests (RF), and Artificial Neural Network (ANN) models.
Figure 3
Figure 3
(a) The R2 for importance estimations as a function of the number of instances. Inset: the coefficient of variations of 11 pore structural features; (b) the importance ranking of 11 pore structural features estimated by RF model trained with 300 instances. The error bars represent ± standard error which is obtained from fivefold cross-validations.
Figure 4
Figure 4
(a) ANN predictions (R2) as a function of the number of instances. Inset: the impact of the number of input features on ANN predictions using 300 instances. Features are removed one by one starting from the least important feature; (b) the testing and validation performance (inset) of ANN predictions using the three pores structural features from 300 instances. The values represent normalized reaction rates.
Figure 5
Figure 5
(a) Importance estimation of pore structural features at three different Pe (0.1, 1, and 10) with DaII = 0.03 (the corresponding Da values are 0.3, 0.03, 0.003) and (b) the corresponding prediction performance of ANN models using the three pore structural features from 300 instances; (c) importance estimation of pore structural features at DaII = 0.03 and 0.27 (the corresponding Da values are 0.3, 2.7) with Pe = 0.1 and (d) the corresponding prediction performance of ANN models using the three pore structural features.
Figure 6
Figure 6
The effect of specific surface, pore sphericity and coordination number on normalized effective reaction rates (Rnorm) under different Pe and Da conditions: (a–c) at Pe = 0.1, Da = 0.3, DaII = 0.03; (d–f) at Pe = 1, Da = 0.03, DaII = 0.03; (g–i) at Pe = 0.1, Da = 2.7, DaII = 0.27. The values of coordination number (4.6) in the left column, pore sphericity (0.76) in the middle column, and specific surface (0.13) in the right column are fixed to their average values in the dataset.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. Norouzi AM, Babaei M, Han WS, Kim K-Y, Niasar V. CO2-plume geothermal processes: A parametric study of salt precipitation influenced by capillary-driven back flow. Chem. Eng. J. 2021;425:130031.
    1. Erfani H, Joekar-Niasar V, Farajzadeh R. Impact of microheterogeneity on upscaling reactive transport in geothermal energy. ACS Earth Space Chem. 2019;3:2045.
    1. Maher K, Steefel CI, DePaolo DJ, Viani BE. The mineral dissolution rate conundrum: Insights from reactive transport modeling of U isotopes and pore fluid chemistry in marine sediments. Geochim. et Cosmochim. Acta. 2006;70:337.
    1. Lee W, et al. Spatiotemporal evolution of iron and sulfate concentrations during riverbank filtration: Field observations and reactive transport modeling. J. Contam. Hydrol. 2020;234:103697. - PubMed
    1. Zhi W, et al. From hydrometeorology to river water quality: Can a deep learning model predict dissolved oxygen at the continental scale? Environ. Sci. Technol. 2021;55:2357. - PubMed

Publication types