Page 213 - gas transport in porous media

Tidwell
208
tracer studies conducted by Chao and others (2000). Experiments involved a two-
dimensional horizontal laboratory tank measuring 244 by 122 by 6.35 cm filled
heterogeneously with different silica sands. Both uniform flow and convergent flow
tracer tests were performed using a conservative tracer. Tests involving uniform flow
yielded results consistent with classical stochastic theory that is the dispersivities
approached a constant value at large displacements. The convergent flow tests gave
very different results; specifically the dispersivities exhibited sustained scale depen-
dence across a wide range of displacements. For the convergent tests, the dispersivity
was also noted to be dependent on both the spatial extent of the source term and its
point of injection (even when the radial distance from the pumping well was constant).
To investigate transport scaling processes over a larger range of scales, experiments
have been conducted in the field under controlled conditions. These tests generally
included detailed descriptions of the subsurface geology followed by a natural and/or
forced gradient tracer experiment. The advection, dispersion and reaction of the
contaminant plume are then monitored over time through the sampling of a dense
network of boreholes. Efforts are then made to interpret the measured tracer behavior
with that predicted by stochastic transport models.
One of the best-known field tests was conducted at the Borden Site in Ontario,
Canada. Natural gradient experiments were conducted using a variety of conserva-
tive and nonconservative tracers in this relatively homogeneous (Var[lnk] = 0.29)
sand aquifer (Freyberg, 1986; Sudicky, 1986). In much the same manner, field tests
were conducted in Cape Cod, Massachusetts (Garabedian, 1987; Hess et al., 1991).
Natural-gradient tracer experiments conducted in this gravel aquifer extended over
a travel distance of about 275m. Additionally, Moltyaner and Killey (1988a, 1988b)
describe a 40 m long tracer test conducted in a fluvial sand aquifer at Chalk River,
Ontario, termed the Twin Lakes site.
In each of these tests, macrodispersivities calculated from the field data were
compared with that predicted by the stochastic theory of Gelhar and Axness (1983)
subject to the spatial distributions of conductivity characterizing each aquifer. For
each of the field tests, favorable comparisons were found. As predicted by theory,
longitudinal spreading dominated the evolution of the tracer plume with the transverse
dispersion playing a relatively weak role, particularly in the vertical direction.
More recently several natural-gradient tracer tests were conducted at the Colum-
busAir Force Base in Mississippi, commonly known as the MADE (Macrodispersion
Experiment) site. The MADE site is associated with a shallow alluvial aquifer that is
at least one order of magnitude more heterogeneous than the aquifers noted above.
Owing to the considerable heterogeneity in the measured hydraulic conductivity dis-
tribution the classical Fickian advection-dispersion model had difficulty reproducing
the measured tracer behavior (Zheng and Jiao, 1998; Eggleston and Rojstaczer, 1998).
Part of the problem was believed to be due to transport along preferential flow path-
ways. Similar studies of tracer dispersion and diffusion in fractured dolomite required
higher-order models to explain mass-transfer between the fractures and host matrix
(Haggerty et al., 2001).