The transport of Cd in alluvial gravels was investigated using batch tests
and miscible displacement experiments with large and small columns. Batch t
ests indicates that Cd adsorption fitted the Freundlich isotherm reasonably
well and was highly nonlinear. The sorption of Cd under batch conditions w
as 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 in
jected 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, 1
0 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 pr
esence of a non-equilibrium component for the Cd transport. This was furthe
r evidenced by decreases in Cd concentrations when flow was interrupted. Un
der 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, wer
e 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 sol
utes. The retardation factor (R) values were estimated based on results fro
m batch tests and the time moment method. For comparison purposes, the R va
lues of the Cd data were also simulated from the two-site models. The value
s of the partitioning coefficient (beta) and the mass transfer coefficient
(omega) were estimated from two-region/site models. Modelling results sugge
st 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 th
at the degree of chemical non-equilibrium increased with increasing pore-wa
ter velocities, however the change was not major for the flow rates investi
gated (18-61 m/day); and (4) equilibrium models could provide apparently go
od fits to skewed BTCs by overestimating dispersivity, which incorporates a
ll the effects from the dispersion and diffusion within both mobile and imm
obile liquid phases and/or type-1 and type-2 sorption sites. Comparatively,
non-equilibrium models described the Cd data better and provided a more co
rrect explanation for the spreading and asymmetry of the Cd BTC. (C) 1999 E
lsevier Science B.V. All rights reserved.