Field Facilities for Groundwater Flow and Tracer Transport in Fractured Shale Bedrock

A field facility has been developed at the Oak Ridge National Laboratory, in Oak Ridge Tennessee (USA) to investigate groundwater flow and transport processes in fractured shale bedrock. The study site is located in the southeast portion of Waste Area Grouping 5 (WAG 5) on the Oak Ridge Reservation, and it is underlain by the Maryville Limestone which consists of interbedded shale-limestone sequences that dip 30-35° southeast (Figure 1 and Figure 2). The bedrock has a greater proportion of shale relative to limestone, and thus is highly fractured. Most fractures are a few centimeters to a meter in length, and they are highly interconnected.

Figure 3: A view of the experimental field facility, standing midway within the site, and looking out towards the tracer injection well (inside the blue tarped tent) and the field containing the buried waste trenches.	Figure 4: A view of the experimental field facility, standing near the tracer injection well and looking out across the groundwater monitoring wells that form a 35 m long strike parallel transect that ends at a cross-cutting tributary.

Investigations have focused on shallow groundwater flow within the saturated bedrock 2-9 m below the ground surface. The site is near the valley floor where converging groundwater flows west to east along geologic strike towards a cross-cutting tributary drainway (Figures 3 and 4). Fracture orientations and connectivity give rise to extensive preferential flow regimes within the bedrock. Groundwater passing through the site is heavily contaminated with ³H, ⁹⁰Sr, and DNAPLs which have originated from upslope contaminant waste trenches (Figure 2).

A 35 m long transect of multilevel groundwater monitoring wells has been established between the boundary of the buried waste trenches and a seep that drains into the cross-cutting tributary (Figure 5). Three wells 4014, 4165, and 2046 have been drilled (1.6, 7.6, and 18.4 m from the trench boundary, respectively) down to a depth of 9 m and point dilution measurements were performed as a function of depth to determine specific discharge variations within the formation (Figure 6). Groundwater discharge measurements revealed a fast-flowing fracture regime within the bedrock that was surrounded by a slower flowing matrix regime. Monitoring wells 4165 and 2046, located at 7.6 and 18.4 m from the trench boundary, were subsequently instrumented with nine Solinst multilevel sampling ports that were isolated with pressurized gel filled packers (Figure 7). The monitoring well located 3 m from the buried waste trenches is used as a tracer injection well to be discussed momentarily. An additional 24 drive-point wells were also installed along the transect (Figure 5) using a track-mounted pneumatic hammer (Figure 8). The pneumatic hammer was able to drive the wells to various depths within the shale bedrock so that fracture and matrix regimes could be monitored (Figure 9).

Figure 7: Phil Jardine inspects a Solinst multilevel sampling well equipped with the capabilities of extracting groundwater samples and monitoring dissolved gas tracers via passive diffusion.	Figure 8: Norm Farrow and Charlie Lamb (right to left) relax after installing 24 drive point wells into the shale bedrock using a track-mounted pneumatic hammer.

The injection well, 4014, located 1.6 m from the buried waste trenches, is used to dispense tracers into a fracture zone 5.2 to 6.1 m from the ground surface. The zone is isolated using a rock/bentonite plug on the lower end and a single straddle packer on the upper end. Tracers are added under natural gradient using a sophisticated computer driven delivery system that maintains a steady state or transient pulse of solute. A variety of tracers can be used including inorganic and organic solutes, radioactive tracers, and dissolved gases.

Figure 9: Cross-section of the experimental field facility showing the location and sampling depth of all groundwater monitoring wells.

Previous investigations at the site used multiple nonreactive tracers to quantify the extent of preferential flow and matrix diffusion within the shale bedrock. Three tracers (Br, He, and Ne), which have different molecular diffusion coefficients, were added to the injection interval as a steady-state pulse for 180 d under natural gradient. The spatial and temporal distribution of the tracers was monitored in the 33 wells for a period of 550 d. Groundwater levels were also measured in the drive point wells at least once a week, with the frequency increasing during storm events. Continuous pressure-head measurements were also performed at several locations throughout the site.

The results of this study indicated that a portion of the tracer plume moved preferentially along fast flowing, strike parallel fractures, while another portion of the plume slowly migrated into the surrounding bedrock matrix. This is observed in Figure 10 which shows the breakthrough of Br^-, 6 m from the source within the fracture and matrix regimes. Tracer breakthrough within the fracture regime shows a rapid concentration increase and approach to steady-state mass transport. The concentration gradient between the fracture and matrix regimes results in the slow migration of tracer into the bedrock matrix (Figure 10). Molecular diffusion is most likely the rate limiting process controlling tracer migration into the matrix, since Br^- concentrations decrease exponentially with greater distances into the matrix. Calculated effective diffusion coefficients, however, are 20× larger than the true molecular diffusion coefficient for Br^-. This suggest advective contributions towards solute transport within the bedrock matrix.

The true test of diffusion limited mass transfer can be derived from the multiple tracers that have different molecular diffusion coefficients. All three tracers migrated preferentially along strike, and their concentrations in the fracture regime quickly reached a consistent steady-state value (Figure 11). The fact that all three tracers had a similar reduced concentration in the fracture regime implied that any observed differences in tracer breakthrough into the matrix would solely be a function of their molecular diffusion coefficients. The breakthrough of the tracers 6 m from the source and 0.8 m into the matrix relative to the fracture is shown in Figure 12 a. The movement of He and Ne into and from the matrix was more rapid than Br^-, and this is consistent with the larger molecular diffusion coefficient of the dissolved gases relative to Br^-. These results confirmed the contribution of matrix diffusion to the overall physical nonequilibrium process that controls contaminant transport in the shale bedrock.

At greater distances from the source, the contribution of matrix interactions are still prominent, and tracer breakthrough profiles remain suggestive of a diffusion mechanism, although at first glance this may not be apparent (Figure 12 b, c). At 13 m from the source and 0.6 m into the matrix, the three tracers eventually breakthrough nearly simultaneously, with the concentration of gas tracers eventually surpassing Br^- (Figure 12 b). This is followed by tracer washout after the input pulse was terminated at 180 d. At 23 m from the source and 0.1 m into the matrix, the movement of Br^- into the matrix is actually more rapid than that of the nobel gas tracers, which is exactly opposite of what was observed 6 m from the source. This apparent paradox is caused by preferential loss of gas tracers to the rock matrix closer to the source. Thus, Br^- remains within the advective flow field (fracture regime) for a longer time period, allowing it to be transported greater distances. Having been transported farther down gradient, Br^- experiences the first opportunity to begin diffusing into the matrix at distances further from the source. Eventually, He and Ne arrive at the same locations and also begin to diffuse into the matrix, lagging behind that of Br^- (Figure 12 b, c). Because of the faster diffusion rate of the noble gases, the movement of He and Ne into the matrix is more rapid and, if given enough time, the He and Ne breakthrough curves will eventually cross over and surpass the Br^- breakthrough curves (see Figure 12 b as an example).

Numerical simulation of the experimental data in Figures 12 a, b, c, using a one-dimensional fracture flow model, revealed similar trends as those observed (Figure 13 a, b, c). Model input parameters are based on independent field and laboratory observations using subsurface media similar to that at the WAG 5 field facility (Dreier et al., 1987; Wilson et al., 1992; McKay et al., 1996; Dorsch et al., 1996). Since advective flow partially contributes to tracer migration in the matrix, parameter z (distance into the matrix) was smaller than that observed in order to account for the shorter travel distance within the matrix. Model results show that for conditions close to the source, the migration of Br^- into the matrix was slower relative to the gas tracer, which is consistent with the smaller molecular diffusion coefficient for Br^- (Figure 13 a). Model results also showed that for conditions far from the source, Br^- migrated into the matrix more rapidly than the gases, since the latter were preferentially lost to the matrix closer to the source due to their larger molecular diffusion coefficients (Figure 13 c).

The experimental and numerical results of this study are consistent with contaminant discharge concentrations within the matrix and fracture regimes at the site. Figure 14 shows seasonal groundwater ³H concentrations profiles within the fractured and matrix regimes 7.6 m from the buried waste trenches. The ³H concentration profiles could be described as sinusoidal functions with local maximum and minimum concentrations within the fracture regime occurring during early December and early April, respectively. The sinusoidal functions for the matrix regime progressively shifted to later times in the year with increased distances into the matrix (Figure 14). The time lags in the concentration profiles of the matrix relative to the fracture were similar to measured first-arrival times of the nonreactive tracers that "diffused" from the fracture into the surrounding matrix. These results show that the multiple tracer test served as an adequate means of quantifying the rate of contaminant storage and depletion within secondary bedrock sources. The mass transfer rates have application in risk assessment modeling and in the design of improved remedial strategies targeted at contaminant removal in hard to reach sources within the bedrock matrix.