Abstract:
Contaminant dispersion in columnar basalt fractures was examined through laboratory experiments and mathematical modelling. A hexagonal discrete fracture pattern with variable fracture aperture and length was assumed to represent the columnar basalt. Artificial fractures made of Plexiglas were connected in various configurations to form intersections, fractures and a hexagonal fracture network of aperture sizes 0.2, 0.5, 1 and 2mm. Dye solutions and clean water were sent through each system and outlet concentrations and flow rates in each fracture were measured for various flow conditions. Systems were tested under laminar flows (Reynolds numbers 5-55) and high Peclet numbers (2,100-25,800) to achieve similar conditions prevailing in the basalt aquifers in the Auckland region. The degree of mixing in a four-way fracture intersection was found to be a function of inlet and outlet flow ratios, aperture and velocity. Discrete fracture network modelling carried out in previous studies has only considered two mixing mechanisms in fracture intersections, i.e. either proportionate routing or completely mixed. Laboratory experiments show that mixing may vary from zero to 35% for apertures between 0.5 - 2mm, while for fine fractures (0.2mm) higher mixing is possible due to advective mass transfer across the flow interface. The mixing mechanism was characterised using an idealised flow pattern considering advection, diffusion and dispersion as the transport mechanisms. The model showed that the mixing is mainly caused by transverse dispersivity which is linearly related to aperture and linearly related to velocity in logarithmic scale. Mixing was also characterised using mass transfer theory where the mass transfer coefficient was found varying linearly with both aperture and velocity in logarithmic scale. Single fractures with two intersections at the ends to represent typical hexagonal columnar basalt fracture system were tested. The mixing mechanism in the system was characterised using the two-dimensional advection-dispersion equation assuming an idealised flow pattern. In contrast to four-way intersections, mixing in three-way intersections does not appear to vary with the fracture aperture or flow characteristics. The major transport mechanism in the fracture system which causes mixing was identified as diffusion. Transverse dispersion is· the main cause of mixing in the intersection, but does not contribute significantly to the overall mixing. In all cases, intersection mixing was found to be less than 10% while the overall mixing reached as much as 100% for longer fractures of small apertures. For fractures larger than 0.5mm aperture, forced mixing does not homogenise within the fracture length and therefore two-dimensional transport must be considered. Accordingly, if similar conditions prevail, such as in columnar basalt, it is appropriate to analyse solute transport through discrete fracture networks by considering the two dimensional advection-diffusion in fractures. Using these findings, a mathematical model was developed to assess the contaminant dispersion through a hexagonal fracture network. The model was validated through the laboratory network model and was then applied in the field scale to estimate the longitudinal and transverse dispersivities of hexagonal columnar basalt fracture systems. Mean longitudinal and transverse dispersivity values resulted from the model were 0.7m and 0.4m which are comparable with the previous field study carried out in the area. The model could be applied to estimate dispersivities in hexagonal fracture systems with known characteristics such as columnar basalt, avoiding large costs which would involve field studies. Finally, a contaminant transport model for a basalt aquifer in the Auckland region (Onehunga) was developed based on the available hydro-geological information combined with the research findings. Model simulations were performed for various fracture configurations and for different potential contaminants to develop a family of contaminant-concentration profiles at the pumping well and to identify areas of high, medium and low risk in the region.