Fluid flow and solute transport simulations in tight geologic formations: Discrete fracture network and continuous time random walk analyses
Oluwaseun J. Akomolafe,Behzad Ghanbarian,Jeffrey D. Hyman
DOI: https://doi.org/10.1016/j.jhydrol.2024.131109
IF: 6.4
2024-04-07
Journal of Hydrology
Abstract:Understanding the effect of fracture network heterogeneity on flow and transport in tight formations with low permeability values is of great importance in a wide variety of applications, such as groundwater pollution and storage as well as water production. Although numerous studies have investigated solute transport in fractured media, investigation about scale effects on flow and transport in low-permeability aquifers is limited. More specifically, we are not aware of any study that has addressed how the continuous time random walk (CTRW) model parameters vary with scale and fracture network properties. To this end, we performed a set of numerical simulations in sparse fracture networks of sizes L = 20, 35, and 50 m under two different volumetric fracture counts (p30 = 0.05 and 0.1) using the discrete fracture network approach. The fractures in the networks were elliptical in shape, and their radii followed a truncated power-law distribution with exponents α = 1.5, 2, and 2.5. We simulated fluid flow using the Reynolds equation, and solute transport was simulated using particle tracking. The solute transport behavior was quantified by fitting the solution of the CTRW transport equation to the arrival time distributions. Results showed that as the truncated power law exponent α increased, the permeability of the networks decreased. In all the fracture networks studied, we found non-Fickian solute transport behavior deduced from small values of β ( 0.69≤β≤1.25 ), the exponent in the truncated power-law arrival time distribution in the CTRW model. In our fracture networks, the aperture size spanned less than one order of magnitude meaning that the fracture permeability distribution was relatively narrow. Therefore, the observed non-Fickian behavior should be attributed to sparse fracture networks and low connectivity. We found that the effective permeability, keff , at p30 = 0.05 was more scale-dependent compared to that at p30 = 0.1 because the fracture networks with p30 = 0.05 were less connected and, thus, more heterogeneous. Our simulation also disclosed the scale dependence of the CTRW model parameters. We conducted multiple linear regression analysis to link the CTRW model parameters to the network size as well as the geometrical and topological properties of the network. Results of regression analysis showed that all the CTRW model parameters except the average solute velocity were significantly dependent on the system size L (p-value < 0.05). Further investigations using a broader range of fracture network properties, such as broader aperture size distributions, smaller and larger network sizes and volumetric fracture counts are still required.
geosciences, multidisciplinary,water resources,engineering, civil