ETPI Progress Report - January 1998

Experimental Progress (PI - Jardine)

Field-scale transport experiment involving multiple chelated radionuclides

The natural gradient injection of Br-, 57Co(II)EDTA, 51Cr(III)EDTA, and 109CdEDTA was initiated at the WAG 5 field facility on November 21, 1997 within the fractured interval of a shale bedrock. The injection will continue for 180 d, and the tracer plumes will be monitored for 550 d using an array of multilevel wells that were instrumented within a faster flowing fracture regime and a slower flowing matrix regime (see http://research.esd.ornl.gov/hydrology/WAG5/). The study has been continuous for 60 d and the tracer plumes are migrating preferentially along strike-parallel fractures (Figs. 1-3). Horizontal and transverse dispersion of the tracers is evidenced by increased spreading of the tracer plumes within the fracture and matrix intervals. Transverse dispersion within 6 m from the source is minimal as noted by the large mass recoveries of Br- within the fracture regime (e.g. wells 10 and 4165-D).

The transport of the chelated radionuclides, 57Co(II)EDTA, 51Cr(III)EDTA, and 109CdEDTA were significantly retarded relative to the nonreactive Br- tracer (Figs 1-4). This is an important finding since it is generally assumed that chelated radionuclides move conservatively along fracture pathways. The mechanism of retardation could be the result of geochemical or microbial processes. Physical processes such as diffusion into the bedrock matrix cannot account for the mass loss since the chelated radionuclides are expected to move more slowly into the matrix relative the smaller Br- ion. We are in the process of unraveling the mechanism of retardation. Also, the oxidation of Co(II)EDTA to Co(III)EDTA does not appear to be an important process and this is advantageous from a remedial perspective since Co(III)EDTA is extremely stable and difficult to immobilize within the subsurface.

The slow migration of the tracers into the bedrock matrix can be seen in figure 4. At 6 m from the source, bromide and the chelated radionuclides can be first detected within the matrix at 20 d. The slow migration of the tracers into the matrix is driven by concentration gradients between the fracture regime and the matrix regime. It will be interesting to see how microbial and geochemical processes alter the fate and transport of the chelated radionuclides as they continue to migrate farther into the matrix.

Please see http://research.esd.ornl.gov/hydrology/WAG5/ for results and discussions of previous nonreactive multiple tracer studies at this experimental site.

Figure 1: Observed breakthrough of Br-, 57Co(II)EDTA, 51Cr(III)EDTA, and 109CdEDTA within the fracture regime, 6 m from the source (well 4165-D).
Figure 2: Observed breakthrough of Br-, 57Co(II)EDTA, 51Cr(III)EDTA, and 109CdEDTA within the fracture regime, 6 m from the source (well 10).
Figure 3: Observed breakthrough of Br- and 57Co(II)EDTA within the fracture regime, 9 m from the source (well 16).
Figure 4: Observed breakthrough of Br- and 57Co(II)EDTA, 6 m from the source and 0.1 m into the matrix (well 7).

Laboratory experiments involving multiple chelated radionuclides

As described in our July 1997 progress report, the fate and transport of 57Co(II)EDTA, 58Co(II)EDTA, 109CdEDTA, and 51Cr(VI) are being investigated in undisturbed soil columns of fractured weathered shale from the Oak Ridge Reservation. Our final multispecies transport experiment is in progress (Fig. 5) involving the migration of Br-, 57Co(III)EDTA, 109CdEDTA, and 51Cr(VI). The chelated radionuclides were retarded relative to nonreactive Br- with Cr > CdEDTA > Co(III)EDTA. The Co(III)EDTA is retarded via sorption reactions to solid phase, whereas the CdEDTA is retarded via solid phase induced dissociation of the metal-chelate complex (see July 1997 progress report). Chromium is significantly more retarded relative to the other tracers and this is due primarily to solid phase sorption reactions and redox transformations of Cr(VI) to the more highly reactive Cr(III) species. The reduction reaction is advantageous from a remedial perspective since Cr(III) is less toxic and less mobile than Cr(VI). The redox reaction is most likely catalyzed by soil organic carbon residing on the surface of clay minerals and oxides. To test this hypothesis, we loaded an undisturbed soil column with natural dissolved organic carbon to produce a monolayer coverage of organic carbon on the soil surface (Fig. 6). When compared to control columns (Fig. 5 and 7), it is apparent that the Cr(VI) retardation increases nearly 4 fold with the addition of solid phase organic carbon, most likely due to the enhanced formation of Cr(III). Since the Cr(III) is strongly sorbed, we cannot detect it in the effluent; however, we plan to take the soil columns to the Stanford Synchrotron Radiation Laboratory in Stanford, CA and perform X-ray Absorption Spectroscopy on the solid phase. These results will definitively reveal what proportion of the solid phase Cr is as Cr(III) and Cr(VI).

Figure 5: Multicomponent displacement of Br-, 57Co(III)EDTA, 109CdEDTA, and 51Cr(VI) through an undisturbed column of fractured weathered shale at a flux of 2 cm/d, I=0.03, and pH=3.8 (this experiment is ongoing).
Figure 6: Displacement of 51Cr(VI) through an undisturbed column of fractured weathered shale that was loaded with a monolayer coverage of natural dissolved organic carbon. The flux was 2 cm/d, I=0.03, and pH=3.8 (this experiment is ongoing).
Figure 7: Displacement of 51Cr(VI) through an undisturbed column of fractured weathered shale at a flux of 2 cm/d, I=0.03, and pH=3.8.

Numerical Modeling Progress (PI - Gwo)

Integration of model development, performance enhancement, field-scale application, and comparison of data and modeling results are underway. Our emphasis on the integration of these modeling elements can be summarized as:

  1. Model development: A two-dimensional biohydrogeochemistry code has been developed by the Pennsylvania State University. The computer code has been numerically verified and its documentation has recently been completed. Development of a three-dimensional sibling of the code is planned for the second and third quarters of FY98 and the parallel version of the three-dimensional code will be made available by the end of the calendar year 1998.

  2. Model performance: We continue to enhance the multiregion groundwater flow code 3DMURF (see July 1997 report). Recent improvements include dynamic memory allocation, implementation of new data structure and search algorithm, and new preconditioner for linear matrix solvers. The new preconditioner (block diagonal preconditioning) alone accounts for 50% reduction in computational cost attributable to the linear matrix solvers. Similar enhancement is also seen in the 3DMURT code that calculates contaminant concentrations using the flow fields calculated by 3DMURF. The results of these studies were published in the 1997 High Performance Computing Conference and can be accessed on the web (see http://www.csm.ornl.gov/~g4p/hpc97_2/hpc97_2.html and http://www.csm.ornl.gov/~g4p/HPC_97.html).

  3. Model application: A three-dimensional, high-performance FORTRAN (HPF) version of the two dimensional HYDROGEOCHEM code developed at the Pennsylvania State University was applied to the Melton Branch experimental station at the Oak Ridge National Laboratory (ORNL). This model application demonstrated the capability of the high performance computer code to calculate coupled, complex hydrogeological and multispecies solute transport problems at field scale. Data on computational efficiency of the model were collected and suggestions to future improvements in the generic areas of field-scale hydrogeochemistry modeling and high-performance computing using HPF were provided. The findings of this research were presented in the 1997 American Geophysical Union Chapman Conference/Soil Science Society of America Outreach Conference and can be accessed on the web (see http://www.csm.ornl.gov/~g4p/chapman_1/chapman_1.html).

  4. Comparison of modeling results with field data: Comparison of data and modeling results is an indispensable modeling element that may help reveal or clarify the mechanisms of subsurface hydrogeochemical processes. Modeling results on the bromide transport at the Melton Branch site (Fig. 8) compared favorably with field data previously obtained from a soil-block bromide tracer injection experiment at the Walker Branch at ORNL (Fig. 9). This in turn suggests that the multiregion mechanisms implemented in 3DMURF and 3DMURT may adequately describe solute movement in field, undisturbed soils. The research result was also presented in the 1997 American Geophysical Union Chapman Conference/Soil Science Society of America Outreach Conference and can be found on the web (see http://www.csm.ornl.gov/~g4p/chapman_1/chapman_1.html).

    Figure 8
    Figure 9
ETPI Progress Reports / Comments? / webmaster@research.esd.ornl.gov