numerical modeling of contaminant dispersion from ... · (3d entities), the fractures (2d...

$: Numerical modeling of contaminant dispersion from ... · (3D entities), the fractures (2D entities), the fracture intersections (pipes - 1D entities) and pipe intersections (0D entities),$
Numerical modeling of contaminant dispersion from underground mine repositories

A. Peratta & V. Popov Wessex Institute of Technology, Southampton, UK

Abstract

One of the tasks of the FP5 EC project Low Risk Disposal Technology (LowRiskDT) was to determine the capacity of reference mine repositories to provide safe and permanent isolation of chemical waste. Calculations were made of the isolating capacity of reference repositories in terms of the concentration of hazardous ion species appearing in the far-field groundwater. Relevant flow and dispersion models were applied to the reference cases representing different rock structures and flow path patterns. The release and migration of dissolved hazardous chemical species with special respect to groundwater protection are determined. A numerical tool has been developed that can be used for analysis of transport of contaminants from underground repositories. The tool can take into account fracture zones, non-homogeneous domains, and can be used to model the processes in the far field, near field and on the interface of the far field and near field. The waste-isolating capacity has been estimated for underground mines in two types of geological media: crystalline rock and limestone. The repository consists of a large room filled with hazardous waste embedded in clay, where in the case of crystalline host rock, there is also a tunnel extending from the room which is also used as a repository, and there a number of fracture zones that intersect the domain and some of them intersect the excavation-disturbed zone (EDZ). Two types of hazardous waste were considered: Dichlorvos and batteries/zinc, but in this paper only the results for Diclorvos are reported. Keywords: hazardous waste disposal, underground mines, numerical modelling.

1 Introduction

This paper contains details on the safety assessment of the proposed LowRiskDT approach for disposal of hazardous waste in abandoned underground mines. The

Waste Management and the Environment II, V. Popov, H. Itoh, C.A. Brebbia & S. Kungolos (Editors)© 2004 WIT Press, www.witpress.com, ISBN 1-85312-738-8

safety criterion related to the waste isolating capacity of the mine repositories is evaluated taking into account the quality of the groundwater on a certain distance from the mine in the direction of the flow of the groundwater. Since no particular mine has been considered, a scenario is created where two different types of host rock media are considered: crystalline rock and limestone. All the considered parameters of the models were chosen in a conservative way or worst-case scenario, which contributes towards quicker dispersion. Both cases of mine repositories in crystalline rock and limestone were of similar geometry. The mine in crystalline rock consists of a room and a tunnel. Around the room and the tunnel excavation disturbed zone (EDZ) was considered in the model. Eighteen fracture zones intersect the domain, of which three intersect the EDZ and serve as fast tracks for transport of contaminants. The mine in limestone consists of a room and no EDZ and no fracture zones were considered. The analysis for both mine repositories was done for two types of chemicals: Dichlorvos and batteries/zinc, but in this paper only the results for Diclorvos are reported. Two types of analysis were performed, long-term, for periods of up to 600000 years and short-term, for periods of up to few thousand years. A numerical tool has been developed for modelling flow and transport in fractured porous media [1]. The numerical tool is based on the discrete-fracture model [2] for both flow and transport. The flow model is based on the Darcy law and the transport model is based on the advection-diffusion equation with reaction. The numerical approach used to solve the partial differential equations is the multi-domain boundary element dual reciprocity method [3], [4].

POROUSMATRIX

POROUSMATRIX

POROUSMATRIX

POROUSMATRIX

PIPE FRACTURE

3D BLOCK 3D BLOCK Porous matrixPorous matrix

2D SURFACE2D SURFACEFractureFracture

1D PIPE1D PIPEFracture IntersectionFracture Intersection

0D MPC0D MPCPipe IntersectionsPipe Intersections

Figure 1: Coupling of matrix blocks, fractures and fractures intersections.

2 Numerical solver used for flow and transport simulation The main features of the developed computer code are: The solver is based on the discrete fracture model; The model for flow is based on the Darcy flow and

192 Waste Management and the Environment II


for the transport on the advection-dispersion equation with reaction; The rock (3D entities), the fractures (2D entities), the fracture intersections (pipes - 1D entities) and pipe intersections (0D entities), have been implemented and coupled in a 3D code; The numerical approach used is the boundary element method (BEM) with domain decomposition for the flow and boundary element - dual reciprocity method multi domain approach (BE DRM-MD) for the transport; The computer code is implemented in such way that many sub-domains with different geometries and properties can exist in a singe model; Automatic time step selection is implemented; The code shows high accuracy and capability to integrate geometry with small details inside large-scale models. Figure 1 shows the way that matrix blocks, fractures and fracture intersections interact. Fractures can be modelled as 2D or 3D entities.

3 Safety assessment of mine repositories The evaluation of waste isolating capacity of mine repositories regarding the flow and transport aspects is conducted for two different geological media, crystalline rock and limestone. In the case of crystalline rock the model consists of a large room, a tunnel, a EDZ of variable thickness around the room and the tunnel, intersecting fracture zones, and chemicals which are embedded in compacted clay inside the room and the tunnel. In the case of limestone the model consists of a large room in which the chemicals are embedded in compacted clay.

3.1 Case of mine and tunnel in crystalline rock

Figure 2 shows a room with a tunnel filled with hazardous chemical embedded in clay. The dimension of the room is 100m×50m×50m. The size of the modeled domain is 700m×800m×600m. The length of the tunnel is 150m and the cross section is 5×5m. Around the room and the tunnel there is EDZ with thickness of 3m around the room and 1m around the tunnel. There are 18 fracture zones in the model, with an aperture of 1m. For simplicity of mesh generation the fracture zones are perpendicular to each other. The computer code can cope with any geometry of fractured porous media. The only restrictions are related to computational resources. For each porous block the average hydraulic properties are considered. This is considered to be a conservative case since three fractures intersect the EDZ of the room and two intersect the EDZ of the tunnel, which speeds up the transport of the chemicals, which would leak out of the repository. The boundary conditions for flow are: on the top surface atmospheric pressure; on the bottom surface impermeable boundary conditions; one of the vertical surfaces has got 5% overpressure in respect to the hydrostatic pressure, while the other three vertical surfaces are with hydrostatic pressure. Saturated flow is assumed. The results for the flow in the case of crystalline rock are shown in Figure 3. The hydraulic head is represented as a density plot over the fracture network, whereas the velocity field is represented as a vector plot. Typical values of the velocity in the rock matrix are two orders of magnitude lower than in the fracture

Waste Management and the Environment II 193


network, which is due to the difference in hydraulic conductivities in rock and fractures. The influence of the EDZ can be seen in Figure 3. Because of the higher hydraulic conductivity the flow is directed towards the room on the inflow part of the room/EDZ, in the figure bottom-left side of the room, and away from the room on the outflow part of the room/EDZ, in the figure top-right side of the room.

O1 O2

g

O3x

y

Ground level

Lateral cut L-L

Figure 2: Position of the room, tunnel, fracture zones and observation well and

observation points in the model.

Figure 3: Density plot of hydraulic head and vector plot of Darcy velocity in the fracture network.

Next, results for the transport are shown. The transport is calculated for 2,2-dichloroacetic acid (DCA), product of decomposition of Dichlorvos. The



(a) (b)

(a) (b)

distribution coefficient Kd is taken to be = 0, therefore, the retardation factor R is 1. The initial concentration inside the repository is 10000ppm.

Figure 4: (a) Normalized concentration distribution of DCA in the considered domain after 127000 years; and (b) after 317000 years.

Concentration profiles in the well

-0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

-600 -500 -400 -300 -200

Depth [m]

Con

cent

ratio

n [p

pm]

2006008001000120014001600180020002200240026002800

Time [yr]

0

2000

4000

6000

8000

10000

12000

0 500000 1000000

Time (yr)

Con

cent

ratio

n (p

pm)

0

0.5

1

1.5

2

2.5

3

3.5

Room

Well

Figure 5: (a) DCA concentration variation in time in the observation well (b) DCA concentration variation in time in the middle of the room, point O1, shown in the left axis, and in the well, point O2, shown in the right axis.

Figures 4(a) and 4(b) show the character of the transport. In this case the transport of the contaminant is different than the transport of plume of contaminant in groundwater. In the case of transport of plume in groundwater, due to dispersion, the size of the plume increases while the concentration decreases, but preserves its shape. In this case, since the hazardous waste is embedded in clay of low hydraulic conductivity, the chemical is slowly released in time, the process having little resemblance with transport of contaminant plume in groundwater. It can be seen that by t=100000 years the maximum concentration in the rock is still close to the repository since the concentration gradient inside the repository is still high enough to induce significant flux of



contaminant. At t=300000 years the leakage from the room has decreased significantly, so the highest concentration in the rock is away from the room, due to transport of contaminant, which leaked at some previous time. Both effects in the transport of the chemicals are evident, the advection and the dispersion. The timescale of the whole process of release of the chemical from the repository is of the order of magnitude of 500000 years. In Figure 5(a) the short-term analysis of the concentration of DCA in the observation well is shown. It can be seen that after 1000 years the maximum concentration in the well is below 1ppm. Figure 5(b) shows the change of concentration in time in the middle of the room, point O1 in Figure 2, shown in the left axis, and in the well, point O2 in Figure 2, shown in the right axis. It is important to notice that the situation in O2 would be different than what is shown in Figure 5 due to several factors: (i) It was considered that the clay is saturated at t = 0 y. In the case of the Friedland Ton clay with density at water saturation of 1900 kg/m3, practically important wetting of the volume of clay-embedded waste will not commence until 4000 years after application, providing that the waste mass in the big room in this case is surrounded by a 100 cm “liner” of Friedland Ton with the assumed density [5, 6]. Therefore, at time t = 3000 years the chemical would have not left the repository at all, or very little would have leaked. (ii) Not all of the Dichlorvos would transform/decay into DCA. (iii) There would be decay of DCA. (iv) Fully saturated flow was considered, which would not be the case in general (v) The flow velocity vector has got a significant upward (towards the surface) component, which in general may not be the case. Even under such conservative/unrealistic assumptions, the concentration of DCA in O2, according to the results of the simulations, would not exceed 1 ppm in three thousand years.

3.2 Case of mine in limestone

The second case is a mine repository in limestone. The geometry of the mine and domain is the same one that is shown in Figure 2, with the difference that in this case the following was excluded from the model: fractures, tunnel and EDZ. The observation well and observation points O1 to O3 remain in the same place, see Figure 2. The results for flow are given in Figure 6. Since there are no fractures in this model, the hydraulic head and the velocity field are shown in an arbitrary plane. The velocities in the limestone are of the order of 1m/year, while inside the repository the velocities are of order of 0.1mm/year. This shows that the difference in the velocities is of four orders of magnitude, which is in agreement with the difference in hydraulic conductivities in the clay and limestone. The transport in the clay is mainly by diffusion, while in the limestone it is combined, advection and dispersion. The EDZ was not included in the model since it adds complexity and numerical burden to the computations, while it was judged that the EDZ would not change the results in any significant way since there were no fractures included in the model that would be connected to the EDZ.



(a) (b)

Figure 6: Hydraulic head and velocity field for the case of mine repository filled with clay in limestone.

The modeling conditions for Dichlorvos were the same ones that were used in the case of mine repository in crystalline rock.

Figure 7: Normalized concentration distribution of DCA in limestone around mine repository after: (a) 200 years and (b) 2000 years.

Figures 7(a) and 7(b) show the results for leakage of DCA from the repository and its transport through limestone. All the results are for short-term analysis, up to 2000 years. The process of leakage and transport is similar to the one in crystalline rock where the chemical is slowly released mainly by diffusion, because of the low hydraulic conductivity of the Friedland Ton clay, and after



that it is relatively rapidly transported through the limestone due to both, advection and dispersion. In this sense the process looks like a quasi steady-state as the distribution of the DCA in the space looks similar for normalized concentration, what changes are the concentrations which decrease due to the decrease of DCA inside the repository and with this the concentration gradients decrease, which in turn reduces the out-flux of DCA. There is difference in the processes of transport of DCA once it leaves the repository, depending on whether the mine is in crystalline rock or limestone. In crystalline rock the main transport is conducted through fractures and fracture zones, since there the hydraulic conductivity is much higher than in the rock. The crystalline rock slows down the transport by absorbing the chemical, which penetrates the rock mainly by diffusion. In the case of repository in crystalline rock the transport will be mainly defined by the characteristics and distribution of the fractures and fracture zones. In the case of the limestone the transport through the rock is rapid, compared to crystalline rock, due to much higher hydraulic conductivity.

0

1

2

3

4

5

6

0 1000 2000 3000 4000

Time [yr]

Con

cent

ratio

n [p

pm] y = -400m

y = 0

Figure 8: Concentration in function of time in the observation well for DCA

in limestone, point O2, and at the surface, point O3.

Figure 8 shows the results for short-term analysis of DCA concentrations in two points, O3 and O2, see Figure 2 for the case of crystalline rock. It can be seen that unlike the case of mine in crystalline rock, here the maximum concentration is reached relatively quickly, after only few hundreds of years, and it decreases from then onward. The maximum concentration is just above 5 ppm reached in approximately 300 years, and drops below 1ppm in both points after four thousand years. This shows that the importance of the engineered barrier system (EBS) is much higher for mine repositories in limestone compared to crystalline rock. However, these results are obtained for a very conservative case and in reality it is unlikely that the concentrations would reach such level, because of the



factors mentioned before: (i) Not all of the Dichlorvos would transform/decay into DCA. (ii) There would be decay of DCA. (iii) Fully saturated flow was considered. (iv) It was considered that the clay is saturated at t = 0 y (v) the flow velocity vector has got a significant upward (towards the surface) component, which in general may not be the case.

4 Conclusions Crystalline rock The analysis of mine repository in crystalline rock shows that low concentrations of chemicals would appear in the groundwater not far from the mine repository and on the surface. In the case of DCA the maximum concentrations on the ground surface do not exceed 10 ppm, before t≈200000 years. However, these results are obtained for a very conservative case and in reality it is very unlikely that the concentrations would reach such level, because of the factors mentioned in the previous section. It must be mentioned that any analysis longer than few thousand years is unreliable since by that time tectonic as well as glacial processes may completely change the situation. The short-term analysis of DCA leakage and transport shows that the concentration of DCA in the observation well would not exceed 1ppm in three thousand years. This result is valid for very conservative case, since in reality there would be several factors, which were mentioned in the previous sections, that would reduce the concentration in the observation well.

Limestone There are similarities and differences in the processes of transport of DCA once it leaves the repository, depending on whether the mine is in crystalline rock or limestone. Both processes are similar in the process of release of the chemicals from the repository, the main mechanism for transport being diffusion in clay. The differences are in respect to the transport in the surrounding geological media. In crystalline rock the main transport is conducted through fractures and fracture zones, since there the hydraulic conductivity is much higher than in the rock. The crystalline rock slows down the transport by absorbing the chemical, which penetrates the rock mainly by diffusion. In the case of repository in crystalline rock the transport will be mainly defined by the characteristics and distribution of the fractures and fracture zones. In the case of the limestone the transport through the rock is rapid, compared to crystalline rock, due to much higher hydraulic conductivity and is due to both, advection and dispersion. The results for short-term analysis of DCA concentrations for the case of mine repository in limestone show that the maximum concentration in the observation well is reached relatively quickly, after only few hundreds of years, and it decreases from then onward. The maximum concentration is just above 5 ppm reached in approximately 300 years, and drops below 1ppm after four thousand years. These results, just like the ones in the case of mine repository in crystalline rock, are obtained for a very conservative case and in a reality it is unlikely that



the concentrations would reach such level, because of the factors mentioned above. The above risk analysis shows that the disposal in mine repositories in limestone could represent a safe option providing that the engineered barrier is designed such that sufficiently high insulation is provided. The LowRiskDT project provides sufficient information [6] for such task to be successfully completed. The analysis shows that the engineered barrier in the case of mine repository in limestone is more important than in the case of crystalline rock because of the ways of transport of chemicals through these two different types of geological media, which has been described in this paper.

Acknowledgements

This research was supported by the LowRiskDT project (Contract number EVG1-CT-2000-00020) – part of the FP5, Energy, Environment and Sustainable Development European Commission Programme.

References

[1] Peratta, A., Popov, V., ‘A new scheme for numerical modelling of flow and transport processes in 3D fractured porous media’, submitted to Advances in Water Resources.

[2] Steefel, C.I. and Lichtner, P.C. (1997), “Multicomponent reactive transport in discrete fractures: I. Controls on reaction front geometry”, Journal of Hydrology, 209, 186-199.

[3] Popov, V., Power, H. (1999), “The DRM-MD Integral equation method: An efficient approach for the numerical solution of domain dominant problems”, International Journal for Numerical Methods in Engineering, 44, 327-353.

[4] Popov, V., Power, H. (1999), “DRM-MD approach for the numerical solution of gas flow in porous media, with application to landfill”, Engineering Analysis with Boundary Elements, 23/2, pp. 175-188.

[5] Compilation of physical and physico/chemical data of clay materials and steel containers that are suitable for waste isolation, LowRiskDT Project, D2.2 Report, March 2002.

[6] Definition of methods for preparation and manufacturing of clay-based isolation materials, LowRiskDT Project, D2.3 Report, December 2002.



numerical modeling of contaminant dispersion from ... · (3d entities), the fractures (2d...

Documents