Effects of simplifying fracture network representation on inert chemical migration in fracture‐controlled aquifers

Abstract

[1] While it is widely recognized that highly permeable 'large-scale' fractures dominate chemical migration in many fractured aquifers, recent studies suggest that the pervasive 'small-scale' fracturing once considered of less significance can be equally important for characterizing the spatial extent and residence time associated with transport processes. A detailed examination of chemical migration through fracture-controlled aquifers is used to advance this conceptual understanding. The influence of fracture structure is evaluated by quantifying the effects to transport caused by a systematic removal of fractures from three-dimensional discrete fracture models whose attributes are derived from geologic and hydrologic conditions at multiple field sites. Results indicate that the effects to transport caused by network simplification are sensitive to the fracture network characteristics, degree of network simplification, and plume travel distance, but primarily in an indirect sense since correlation to individual attributes is limited. Transport processes can be 'enhanced' or 'restricted' from network simplification meaning that the elimination of fractures may increase or decrease mass migration, mean travel time, dispersion, and tailing of the concentration plume. The results demonstrate why, for instance, chemical migration may not follow the classic advection-dispersion equation where dispersion approximates the effect of the ignored geologic structure as a strictly additive process to the mean flow. The analyses further reveal that the prediction error caused by fracture network simplification is reduced by at least 50% using the median estimate from an ensemble of simplified fracture network models, and that the error from network simplification is at least 70% less than the stochastic variability from multiple realizations.

Reactive fluid flow in CO2 storage reservoirs: A 2‐D pore network model study

“…Note that further analyses and results are presented in dimensionless form. Pore diameters are log‐normally distributed with a coefficient of variation COV≈0.4 for most sediments; we also test COV = 1.0 and 1.5 to better reflect fractured rock conditions …”

“…The coefficient of variation in pore size (COV) is larger in fractured rock masses than in sediments . An additional set of simulations with COV = 1.0 and 1.5 is conducted to examine the role of pore size variability on the evolution of mean tube diameter and flow rate.…”

Influence of fracture scale heterogeneity on the flow properties of three‐dimensional discrete fracture networks (DFN)

“…Because of the finite size of the systems studied, the latter analysis will likely hold for DIST networks but is incomplete for SHORT networks at threshold. In fact, for DIST networks, fractures forming the connected structures range over a limited size interval [ de Dreuzy et al , 2001c; Wellman et al , 2009]. Increasing the system size or equivalently enhancing the system resolution will not issue dramatic changes in the system connectivity, but rather add smaller fractures to the existing main connected paths.…”

Linking Structural and Transport Properties in Three‐Dimensional Fracture Networks

“…Fracture apertures are determined by correlating their width to the fracture radii using a positively correlated power law relationship: where γ = 5.0 × 10 −4 and β = 0.5 are dimensionless parameters. This correlation between fracture size and aperture is a common assumption in DFN models (Bogdanov et al, ; de Dreuzy et al, ; Hyman et al, ; Frampton & Cvetkovic, ; Joyce et al, ; Wellman et al, ).…”