The transport of Cd in alluvial gravels was investigated using batch tests and miscible displacement experiments with large and small columns. Batch tests indicates that Cd adsorption fitted the Freundlich isotherm reasonably well and was highly nonlinear. The sorption of Cd under batch conditions was essentially complete within a few minutes, followed by a time-dependent slow sorption in weeks. A large column (200 cm long, 19 cm diameter) was injected with pulse solutions containing Cd, Br, Cl, and rhodamine WT at pore-water velocities of 20 and 61 m/day; while the small column (18 cm long, 10 cm diameter) was injected with Br and Cd pulses at pore-water velocities of 18 and 58 m/day. Breakthrough curves (BTCs) of Cd were highly asymmetric while those of other solutes were essentially symmetric, suggesting the presence of a non-equilibrium component for the Cd transport. This was further evidenced by decreases in Cd concentrations when flow was interrupted. Under all conditions investigated, asymmetry in Cd BTCs essentially resulted from chemical non-equilibrium as physical non-equilibrium was not present. Both equilibrium and non-equilibrium models, as incorporated in CXTFIT, were used to describe observed BTCs of Cd. Dispersivity was obtained from the nonreactive solute Br data, and it was then held constant for the other solutes. The retardation factor (R) values were estimated based on results from batch tests and the time moment method. For comparison purposes, the R values of the Cd data were also simulated from the two-site models. The values of the partitioning coefficient (beta) and the mass transfer coefficient (omega) were estimated from two-region/site models. Modelling results suggest that (1) Cd transport was mainly controlled by rate-limited sorption; (2) the scale effect on Cd transport was minor for the reasonably well sorted material investigated, with a slight increase in the degree of equilibrium in the large column. Small columns may be used to obtain fast experimental results for homogeneous medium. However, if heterogeneity and preferential flow is important, larger columns are recommended. (3) it was evidenced that the degree of chemical non-equilibrium increased with increasing pore-water velocities, however the change was not major for the flow rates investigated (18-61 m/day); and (4) equilibrium models could provide apparently good fits to skewed BTCs by overestimating dispersivity, which incorporates all the effects from the dispersion and diffusion within both mobile and immobile liquid phases and/or type-1 and type-2 sorption sites. Comparatively, non-equilibrium models described the Cd data better and provided a more correct explanation for the spreading and asymmetry of the Cd BTC. (C) 1999 Elsevier Science B.V. All rights reserved.
