Tracer transport experiments involving colloids that showed little tendency to attach to rock surfaces and a nonsorbing solute (iodide) were conducted in three different well-characterized natural fractures in tuff. The objective was to investigate the potential for using nonattaching colloids in tracer experiments to (1) obtain a measure of the dispersion in a fractured system in the absence of diffusion into the matrix or into stagnant zones along the fracture walls, and (2) provide insights into predicting colloid transport in fractures under conditions of minimal deposition.
In the experiments, the colloids always arrived earlier in the effluent than the iodide, which we believe is evidence of (1) hydrodynamic chromatography and/or (2) the fact that the colloids experience a smaller effective volume in the fracture because they diffuse too slowly to enter low-velocity regions (dead zones) along the rough fracture walls. The iodide also approached the inlet concentration in the effluent more slowly than the colloids, with the concentration at a given elution volume being greater at higher flow rates. By contrast, the rate of approach of the colloid concentration to the inlet concentration did not vary with flow rate. We attribute this behavior to matrix diffusion of the iodide, with the colloids being too large/nondiffusive to experience this phenomenon. Dispersion of all tracers was greatest in the fracture of widest average aperture and least in the fracture of narrowest aperture, which is consistent with Taylor dispersion theory.
The tracer experiments were modeled/interpreted using a three-step approach that involved (1) estimating the aperture distribution in each fracture using surface profiling techniques, (2) predicting the flow field in the fractures using a localized parallel-plate approximation, and (3) predicting tracer transport in the fractures using particle-tracking techniques. Although considered preliminary at this time, the model results were in qualitative agreement with the experiments.