In order to understand the transport mechanism of solute through naturally fractured rocks, it is important to quantitatively characterize the influence of varying surface roughness on fluid flow and solute transport. Rough-walled fracture geometry models with different joint roughness coefficients (JRC) were generated based on MATLAB pixel analyses of ten standard Barton profiles. Fluid flow and solute transport in the 2D rough-walled fractures were simulated by solving the Navier-Stokes-equation (NSE) and the transport equation for different pressure drops (i.e., 5, 10, and 20 Pa). The simulation results show that the evolution of the solute-concentration field within rough-walled fractures is significantly dependent on surface roughness. Analysis of the breakthrough curves (BTCs) and residence time distributions (RTDs) indicated that rough fracture surfaces with large JRCs played a significant role in weakening the non-Fickian transport characteristics (i.e., early arrival and long tail) under the same pressure drop. It was found that the solute-concentration-distribution index (CDI), i.e., a metric for quantifying the longevity of the tail, increased with the JRCs and decreased with an increase in pressure drops. This result demonstrates that decreasing the surface roughness increases the Péclet number (Pe) and enhances advection process in solute transport, resulting in an increase in the non-uniform concentration distribution and shortened the long tail. Inverse modeling of the BTCs shows that rough fracture surfaces with large JRCs decrease the effective dispersion coefficient and Pe, suggesting that rough fracture surfaces decrease the advection and dispersion processes and delay the early arrival. These results provide more comprehensive understanding of the role of surface roughness in solute transport through fractures. Based on the relationships between JRC value and effective dispersion coefficient, a prediction method was established to predict the non-Fickian transport and the JRC value, and the practical cases further proved the feasibility of the prediction method.
Keywords: Early arrival; Fracture surface roughness; Joint roughness coefficient; Long tail; Solute transport.