Numerical modeling of arsenic mobility - In-Situ Remediation of Arsenic-Contaminated Sites - page 45

this mixture of mineral components can then be predicted in an equilibrium calculation, without
requiring fitting of experimental data for the specific mixture (Payne et al. , 2013).
The GC approach acknowledges that the sorption properties of a natural sediment may be too
complex to be quantified in terms of the contributions by individual mineral phases to adsorption
and/or that the contribution of single mineral components to the macroscopic sorption behavior
of the sediment may not simply be additive. Following the GC approach, a surface complexation
model for natural sediment is established by fitting experimental sorption data for the mineral
assemblage as a whole (Bond et al. , 2008; Hyun et al. , 2009; Payne et al. , 2004). This may
not provide an accurate representation of surface complexation reactions at the molecular scale;
however, the GC approach allows for an accurate description of the macroscopic sorption behavior
of the natural system at hand.
Surface complexation models have been incorporated into subsurface transport models for
several field-scale reactive transport modeling studies. Appelo et al. (2002) provided an excellent
example of the usefulness of SCMs to simulate arsenic transport in Bangladesh groundwaters.
They used a modified version of the Dzombak and Morel (1990) surface complexation model to
illustrate mobilization of arsenic through desorption by bicarbonate. Also, Postma et al. (2007)
investigated the mobilization of arsenic in a floodplain aquifer in Vietnam. They constructed a
1D reactive transport model with PHREEQC, which simulated to mobilization of arsenic through
reductive dissolution of Fe-oxides, followed by subsequent re-adsorption on the surface of the
remaining Fe-oxides. The employed database was compiled from different sources. Aqueous
complexation reactions for arsenic were adopted from a compilation provided by Langmuir et al.
(2006), while arsenic adsorption and desorption reactions were simulated with the database of
Dzombak and Morel (1990), extended with additional sorption reactions for carbonate (Appelo
et al. , 2002) and silica species (Swedlund and Webster, 1999). Another simulation example is a 1D
PHREEQC reactive transport model that was developed by Charlet et al. (2007). It demonstrated
the role of phosphate, bicarbonate and aqueous iron in mobilizing an As plume within a 3000-m
aquifer section in West Bengal. A combination of surface complexation constants was adopted
from Dzombak and Morel (1990) and Appelo et al. (2002) and the surface complexation capacity
of the aquifer was coupled to the simulated amount of ferrihydrite. Similarly, Stollenwerk et al.
(2007) successfully used PHREEQC for a 1D reactive transport modeling study to simulate the
fate of arsenic. Again, adsorption reactions were modeled using mostly the surface complexation
model of Dzombak and Morel (1990), while equilibrium constants for the surface complexation
reactions for adsorption of arsenic, phosphate, silica and bicarbonate on aquifer sediments were
fitted to experimental data. The model was used to demonstrate the capacity of oxidized sediments
for remediating As-contaminated groundwater at a site near Dhaka, Bangladesh.
More recently a small number of multi-dimensional reactive transport modeling studies of
the fate of arsenic have been reported. A 2D modeling study of As in groundwater discharging
to Waquoit Bay, MA was undertaken by Jung et al. (2009). The performance of SCM versus
empirical formulations in simulating the observed groundwater As distribution was evaluated
using the simulation code PHT3D (Prommer et al. , 2003). Using the same code Wallis et al.
(2010) developed a process-based description of the coupled physical and geochemical processes
controlling the fate of arsenic during a deep well injection experiment in the Netherlands. The
numerical model was further expanded to subsequently simulate the fate of arsenic in an aquifer
storage and recovery (ASR) system in Florida (Wallis et al. , 2011). In the latter study the numerical
modeling illustrated that pyrite oxidation and the precipitation/dissolution of amorphous iron-
oxides (HFO) together with competitive displacement of As from sorption sites on HFO by
competing anions were the key chemical processes that controlled the mobility of arsenic at the
investigated aquifer storage and recovery (ASR) site. A detailed assessment of arsenic partitioning
among mineral phases, surface complexes and aqueous phases during injection, storage and
recovery was generated. Adsorption of arsenic was simulated using the diffuse double-layer model
and arsenic-HFO surface complexation data was obtained from Dzombak and Morel (1990) and
Appelo et al. (2002). In these models, the sorption capacity of the aquifer was linked to the
amount of ferrihydrite.