Desarrollo local en un hábitat afectado

This research consisted of assessing the accuracy of the implementation of Vinsome and Westerveld (1980) semi-analytical method to the process of matrix diffusion for finite embedded heterogeneity scenarios in groundwater chemical transport modeling.

The execution of the semi-analytical method in a Visual Basic program in Excel® was tested against experimental results from published laboratory scale flow chamber studies, and with simple geometry numerical simulations developed in MT3DMS. The results obtained showed good-to-excellent visual and quantitative agreement, indicating good accuracy of the matrix diffusion semi-analytical/numerical method for most practical purposes.

A FORTRAN version of the semi-analytical method, REMChlor-MD tested against fine grid numerical simulations provided great results, for systems with highly complex heterogeneities present. This becomes particularly important taking into account that this type of setup is closer to real life scenarios, where the exact heterogeneity of field sites is usually unknown.

Three geometric parameters were used in the semi-analytical method for finite

embedded heterogeneities: the high permeability material volume fraction (Vf), the

high/low permeability material interface area (Amd), and the characteristic average diffusion

length (L). A geometrical relationship was defined to reduce the number of matrix diffusion

parameters to define in the upcoming REMChlor-MD version to only two: Vf and L. The

results obtained using this “2-parameter” approach provided a decent match with the fine- grid simulation data without calibration, and a good match with small adjustments to L.

The semi-analytical method implementation proved to be extremely efficient, providing great match to experimental results and numerical simulations with run times ranging from fractions of seconds up to less than three minutes, depending on the size of the model and the simulated periods of time. This is outstanding considering the fine grid MT3DMS numerical simulations used as base of comparison using a little under three million gridblocks took up to 70h to run the mass transport simulation. The efficiency of the semi-analytical method is due to the fitting function approximation for the low permeability areas, allowing the effects of complex heterogeneity to be approximated in a coarse grid.

Recommendations

Some recommendations for future research based on this study include:

• Compare REMChlor-MD and MT3DMS for T-PROGS scenario incorporating sorption and degradation.

• Consider more realistic scenarios for the implementation of the method, such as field sites to study and develop the parameterization of the semi-analytical method with field data.

• Assess the feasibility of a fitting function containing a multiple time-dependent penetration depth to improve the response of the semi-analytical method around the concentration reversal period.

• Implement the matrix diffusion semi-analytical/numerical method in commercial chemical transport models like MT3DMS (Zheng and Wang, 1999).

Appendix A: Cubic spline interpolation

The simulated data in section 5.2 was equally distributed in 0.5 days spacing whereas the experimental dataset had an independent variable that was not equally distributed. The program SRS1 Cubic Spline for Excel from © SRS1 Software (SRS1 Software LLC, 2015) was used to interpolate the data from the semi-analytical model results to match the time series of the experimental data.

The graphs shown in Figure A.1 through Figure A.3 present the spline interpolation results alongside the original simulated data for the different cases studied in sections 5.2 and 5.3.

a)

b)

Figure A.1 Comparison of cubic spline with one gridblock simulated data in suspended clay lenses case for a) Bromide and b) Fluorescein.

b)

Figure A.2 Comparison of cubic spline with 50-gridblock simulated data in suspended clay lenses case for a) Bromide and b) Fluorescein.

a)

b)

Figure A.3 Comparison of cubic splines with simulated data in diffusion length study for a) Bromide and b) Fluorescein.

Appendix B: Borehole data for T-PROGS simulation

The data for the boreholes used in Chapter 7 shown below was taken from the “Borehole Editor” window in GMS: