skip to main content


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

 
more » « less
Award ID(s):
2053370 2046015
NSF-PAR ID:
10364682
Author(s) / Creator(s):
; ;
Publisher / Repository:
Nature Publishing Group
Date Published:
Journal Name:
Scientific Reports
Volume:
12
Issue:
1
ISSN:
2045-2322
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Abstract

    Considering heterogeneity in porous media pore size and connectivity is essential to predicting reactive solute transport across interfaces. However, exchange with less‐mobile porosity is rarely considered in surface water/groundwater recharge studies. Previous research indicates that a combination of pore‐fluid sampling and geoelectrical measurements can be used to quantify less‐mobile porosity exchange dynamics using the time‐varying relation between fluid and bulk electrical conductivity. For this study, we use macro‐scale (10 s of cm) advection–dispersion solute transport models linked with electrical conduction in COMSOL Multiphysics to explore less‐mobile porosity dynamics in two different types of observed sediment water interface porous media. Modeled sediment textures contrast from strongly layered streambed deposits to poorly sorted lakebed sands and cobbles. During simulated ionic tracer perturbations, a lag between fluid and bulk electrical conductivity, and the resultant hysteresis, is observed for all simulations indicating differential loading of pore spaces with tracer. Less‐mobile exchange parameters are determined graphically from these tracer time series data without the need for inverse numerical model simulation. In both sediment types, effective less‐mobile porosity exchange parameters are variable in response to changes in flow direction and fluid flux. These observed flow‐dependent effects directly impact local less‐mobile residence times and associated contact time for biogeochemical reaction. The simulations indicate that for the sediment textures explored here, less‐mobile porosity exchange is dominated by variable rates of advection through the domain, rather than diffusion of solute, for typical low‐to‐moderate rate (approximately 3–40 cm/day) hyporheic fluid fluxes. Overall, our model‐based results show that less‐mobile porosity may be expected in a range of natural hyporheic sediments and that changes in flowpath orientation and magnitude will impact less‐mobile exchange parameters. These temporal dynamics can be assessed with the geoelectrical experimental tracer method applied at laboratory and field scales.

     
    more » « less
  2. The study of thermal convection in porous media is of both fundamental and practical interest. Typically, numerical studies have relied on the volume-averaged Darcy–Oberbeck–Boussinesq (DOB) equations, where convection dynamics are assumed to be controlled solely by the Rayleigh number ( Ra ). Nusselt numbers ( Nu ) from these models predict Nu – Ra scaling exponents of 0.9–0.95. However, experiments and direct numerical simulations (DNS) have suggested scaling exponents as low as 0.319. Recent findings for solutal convection between DNS and DOB models have demonstrated that the ‘pore-scale parameters’ not captured by the DOB equations greatly influence convection. Thermal convection also has the additional complication of different thermal transport properties (e.g. solid-to-fluid thermal conductivity ratio k s / k f and heat capacity ratio σ ) in different phases. Thus, in this work we compare results for thermal convection from the DNS and DOB equations. On the effects of pore size, DNS results show that Nu increases as pore size decreases. Mega-plumes are also found to be more frequent and smaller for reduced pore sizes. On the effects of conjugate heat transfer, two groups of cases (Group 1 with varying k s / k f at σ  = 1 and Group 2 with varying σ at k s / k f  = 1) are examined to compare the Nu – Ra relations at different porosity ( ϕ ) and k s / k f and σ values. Furthermore, we report that the boundary layer thickness is determined by the pore size in DNS results, while by both the Rayleigh number and the effective heat capacity ratio, $\bar{\phi } = \phi + (1 - \phi )\sigma$ , in the DOB model. 
    more » « less
  3. null (Ed.)
    The modelling of natural convection in porous media is receiving increased interest due to its significance in environmental and engineering problems. State-of-the-art simulations are based on the classic macroscopic Darcy–Oberbeck–Boussinesq (DOB) equations, which are widely accepted to capture the underlying physics of convection in porous media provided the Darcy number, $Da$ , is small. In this paper we analyse and extend the recent pore-resolved direct numerical simulations (DNS) of Gasow et al. ( J. Fluid Mech , vol. 891, 2020, p. A25) and show that the macroscopic diffusion, which is neglected in DOB, is of the same order (with respect to $Da$ ) as the buoyancy force and the Darcy drag. Consequently, the macroscopic diffusion must be modelled even if the value of $Da$ is small. We propose a ‘two-length-scale diffusion’ model, in which the effect of the pore scale on the momentum transport is approximated with a macroscopic diffusion term. This term is determined by both the macroscopic length scale and the pore scale. It includes a transport coefficient that solely depends on the pore-scale geometry. Simulations of our model render a more accurate Sherwood number, root mean square (r.m.s.) of the mass concentration and r.m.s. of the velocity than simulations that employ the DOB equations. In particular, we find that the Sherwood number $Sh$ increases with decreasing porosity and with increasing Schmidt number $(Sc)$ . In addition, for high values of $Ra$ and high porosities, $Sh$ scales nonlinearly. These trends agree with the DNS, but are not captured in the DOB simulations. 
    more » « less
  4. Abstract

    Accurate prediction of physical alterations in carbonate reservoirs under dissolution is critical for development of subsurface energy technologies. The impact of mineral dissolution on flow characteristics depends on the connectivity and tortuosity of the pore network. Persistent homology is a tool from algebraic topology that describes the size and connectivity of topological features. When applied to 3D X‐ray computed tomography (XCT) imagery of rock cores, it provides a novel metric of pore network heterogeneity. Prior works have demonstrated the efficacy of persistent homology in predicting flow properties in numerical simulations of flow through porous media. Its ability to combine size, spatial distribution, and connectivity information make it a promising tool for understanding reactive transport in complex pore networks, yet limited work has been done to apply persistence analysis to experimental studies on natural rocks. In this study, three limestone cores were imaged by XCT before and after acid‐driven dissolution flow‐through experiments. Each XCT scan was analyzed using persistent homology. In all three rocks, permeability increase was driven by the growth of large, connected pore bodies. The two most homogenous samples saw an increased effect nearer to the flow inlet, suggesting emerging preferential flow paths as the reaction front progresses. The most heterogeneous sample showed an increase in along‐core homogeneity during reaction. Variability of persistence showed moderate positive correlation with pore body size increase. Persistence heterogeneity analysis could be used to anticipate where greatest pore size evolution may occur in a reservoir targeted for subsurface development, improving confidence in project viability.

     
    more » « less
  5. null (Ed.)
    In this paper, we consider an important problem for modeling complex coupled phenomena in porous media at multiple scales. In particular, we consider flow and transport in the void space between the pores when the pore space is altered by new solid obstructions formed by microbial growth or reactive transport, and we are mostly interested in pore-coating and pore-filling type obstructions, observed in applications to biofilm in porous media and hydrate crystal formation, respectively. We consider the impact of these obstructions on the macroscopic properties of the porous medium, such as porosity, permeability and tortuosity, for which we build an experimental probability distribution with reduced models, which involves three steps: (1) generation of independent realizations of obstructions, followed by, (2) flow and transport simulations at pore-scale, and (3) upscaling. For the first step, we consider three approaches: (1A) direct numerical simulations (DNS) of the PDE model of the actual physical process called BN which forms the obstructions, and two non-DNS methods, which we call (1B) CLPS and (1C) LP. LP is a lattice Ising-type model, and CLPS is a constrained version of an Allen–Cahn model for phase separation with a localization term. Both LP and CLPS are model approximations of BN, and they seek local minima of some nonconvex energy functional, which provide plausible realizations of the obstructed geometry and are tuned heuristically to deliver either pore-coating or pore-filling obstructions. Our methods work with rock-void geometries obtained by imaging, but bypass the need for imaging in real-time, are fairly inexpensive, and can be tailored to other applications. The reduced models LP and CLPS are less computationally expensive than DNS, and can be tuned to the desired fidelity of the probability distributions of upscaled quantities. 
    more » « less