ArticlePDF Available

Abstract and Figures

Liquid elemental mercury occurrence in the subsurface as DNAPL is reported worldwide in proximity of several industrial facilities, such as chlor-alkali plants. Insight into Hg0 DNAPL infiltration behaviour is lacking and, to date, there are no experimental observations of its infiltration and distribution in water saturated porous media, except for capillary pressure-saturation column experiments. To better understand the processes governing elemental mercury DNAPL flow behaviour, a series of flow container experiments were performed using mercury DNAPL (in sands and glass beads) and PCE (in sands). While liquid Hg0 was not able to infiltrate in the sand filled container due to an overall lower permeability of the sample and a defect of the setup, in the glass beads experiment mercury DNAPL infiltration occurred. Dual gamma ray measurements showed that, in glass beads, liquid Hg0 preferentially migrated towards higher porosity zones. As for PCE, infiltration and distribution of Hg0 DNAPL are strongly affected by the heterogeneities within the porous formation. However, compared to other DNAPLs, liquid Hg0 shows a strong attenuation potential of gamma rays. Finally, numerical simulations of the glass beads experiment showed an overall good agreement with the experimental results, highlighting that, among the factors influencing the prediction of liquid Hg0 migration in water saturated porous media, the most critical are (i) the knowledge of the inflow rate, (ii) the reliable estimation of the porous formation permeability, and (iii) the accurate representation of the correlation between retention properties and intrinsic permeability.
This content is subject to copyright. Terms and conditions apply.
Infiltration and Distribution of Elemental Mercury DNAPL
in Water-Saturated Porous Media: Experimental
and Numerical Investigation
Andrea DAniello &Niels Hartog &Thomas Sweijen &
Domenico Pianese
Received: 24 July 2017 /Accepted: 26 December 2017
#Springer International Publishing AG, part of Springer Nature 2018
Abstract Liquid elemental mercury occurrence in the
subsurface as dense non-aqueous phase liquid
(DNAPL) is reported worldwide in proximity of several
industrial facilities, such as chlor-alkali plants. Insight
into Hg
0
DNAPL infiltration behavior is lacking and, to
date, there are no experimental observations of its infil-
tration and distribution in water-saturatedporous media,
except for capillary pressure-saturation column experi-
ments. To better understand the processes governing
elemental mercury DNAPL flow behavior, a series of
flow container experiments were performed using mer-
cury DNAPL (in sands and glass beads) and tetrachlo-
roethylene (PCE) (in sands). While liquid Hg
0
was not
able to infiltrate in the sand-filled container due to an
overall lower permeability of the sample and a defect of
the setup, in the glass beads experiment mercury
DNAPL infiltration occurred. Dual gamma ray mea-
surements showed that, in glass beads, liquid Hg
0
pref-
erentially migrated towards higher porosity zones. As
for PCE, infiltration and distribution of Hg
0
DNAPL are
strongly affected by the heterogeneities within the po-
rous formation. However, compared to other DNAPLs,
liquid Hg
0
shows a strong attenuation potential of gam-
ma rays. Finally, numerical simulations of the glass
beads experiment showed an overall good agreement
with the experimental results, highlighting that, among
the factors influencing the prediction of liquid Hg
0
migration in water-saturated porous media, the most
critical are (i) the knowledge of the inflow rate, (ii) the
reliable estimation of the porous formation permeability,
and (iii) the accurate representation of the correlation
between retention properties and intrinsic permeability.
Keywords Elemental mercury.DNAPL .
Tetrachloroethylene (PCE) .Multiphaseflow .Numerical
modeling .Gamma radiation
1 Introduction
Mercury is recognized worldwide as a hazardous con-
taminant, and the consequences of its release into the
environment can be detrimental for human beings and
ecosystems (US EPA 2007; World Chlorine Council
2011). Mercury can be found in the environment as a
result of natural processes, like volcanic activity, fires,
movement of rivers, lakes, streams, oceanic upwelling,
and biological processes (Risher 2003). However, since
the industrial revolution, anthropogenic activities like
mercury mining (UNEP 2002), mercury amalgamation
for gold and silver mining (Lechler et al. 2000; UNEP
Water Air Soil Pollut (2018) 229:25
https://doi.org/10.1007/s11270-017-3674-0
A. DAniello (*):D. Pianese
University of Naples Federico II, Department of Civil,
Architectural and Environmental Engineering, via Claudio 21,
80125 Naples, Italy
e-mail: andrea.daniello@unina.it
N. Hartog
KWR Watercycle Research Institute, Groningenhaven 7,
Nieuwegein, The Netherlands
N. Hartog :T. Sweijen
Utrecht University, Department of Earth Sciences, Environmental
Hydrogeology Group, Princetonplein 9, 3584 CC Utrecht,
The Netherlands
2002; Hylander and Meili 2003), wood preservation
(Bollen et al. 2008), chlor-alkali and thermometer-
manufacturing plants (Biester et al. 2002;Hylander
and Meili 2003; Bernaus et al. 2006; Arbestain et al.
2009; Brooks and Southworth 2011;Milleretal.2013)
considerably contributed to its distribution into the en-
vironment (Risher 2003). In particular, the chlor-alkali
industry is among the principal consumers of mercury as
well as the most important producer of wastes contain-
ing mercury (Pirrone et al. 2010). In 1996, almost 40%
of the mercury world production was due to the chlor-
alkali industry (Sznopek and Goonan 2000). According
to Euro Chlor (1999), most of the chlor-alkali plants
using liquid Hg
0
cells were located in Western Europe,
and the Hg
0
consumption was about 100 tons per year.
According to Lindley (1997), almost 12,000 tons of
liquid Hg
0
was in use as electrodes in European chlor-
alkali plants. Furthermore, the fate of the 90% of liquid
Hg
0
bought by Western European chlor-alkali plants is
not reported (Euro Chlor 1999) and, most probably,
considerable amounts of liquid Hg
0
are still stored in
wastes in proximity of the chlor-alkali facilities
(Hylander and Meili 2003). According to the global
mercury demand by use category, the global chlor-
alkali mercury demand decreased from 23% in 2000 to
15% in 2005 due to the progressive reduction in the use
of mercury cells technology (Pirrone et al. 2010). How-
ever, in 2010, the chlor-alkali production based on mer-
cury cells still represented 21% of the total world capac-
ity (Pirrone et al. 2010).
Liquid elemental mercury has been found worldwide
in the subsurface as dense non-aqueous phase liquid
(DNAPL) in several chlor-alkali sites (Scanlon et al.
2005; Di Molfetta and Fracassi 2008;Deebetal.2011;
ITRC 2012; Brooks and Southworth 2011;Golder
Associates 2011; Sweijen et al. 2014). Once in the
subsurface, Hg
0
DNAPL acts as a long-lasting source
of contamination (Davis et al. 1997), thanks to its low
aqueous solubility, 0.07 mg/l (Eichholz et al. 1988).
Depending on the site geochemistry, liquid mercury
may also undergo speciation (Leterme and Jacques
2013;González-Fernándezetal.2014; Leterme et al.
2014) and methylation (Compeau and Bartha 1985;
Gilmour et al. 1992), thus allowing for the formation
of organic and inorganic mercury compounds which
may be more toxic, mobile, and soluble than liquid
mercury itself. In addition, Hg
0
may also form in sub-
surface environments or sediments from other mercury
species (Bouffard and Amyot 2009; Richard et al.
2016). However, it may also release contaminants
through volatilization (Walvoord et al. 2008).
Despite its extensive use in petroleum engineering
for mercury porosimetry (Wardlaw and Taylor 1976;
Vavra et al. 1992; Pittman 1992; Smith et al. 2002;
Newsham et al. 2004;Ruthetal.2013), only few studies
(Eichholz et al. 1988; Devasena and Nambi 2010;
Sweijen et al. 2014;D'Aniello2017;DAniello et al.
2018, Under Review) are available on liquid Hg
0
flow
behavior in porous media as a DNAPL. Thanks to its
extraordinary density and surface tension, of about 8
and 15 times higher than tetrachloroethylene (PCE)
(Schwille 1988;CRC2014), liquid elemental mercury
infiltration behavior deviates with respect to more ex-
tensively studied DNAPLs, such as trichloroethylene
(TCE) and PCE, (Devasena and Nambi 2010;DAniello
2017;DAniello et al. 2018, Under Review). As a matter
of fact, mercury DNAPL behaves as a nonwetting phase
with respect to both air and water (DAniello 2017;
DAniello et al. 2018, Under Review); therefore, its
infiltration occurs only when it overcomes an entry
head, both in fully (Devasena and Nambi 2010)and
partially water-saturated porous media (DAniello
2017;DAniello et al. 208, Under Review). In particu-
lar, in fully water-saturated systems, Hg
0
has a high
capacity to infiltrate (DAniello 2017;DAniello et al.
2018, Under Review), and its infiltration behavior is
mainly governed by gravitational and capillary forces,
being practically independent of viscous forces
(Devasena and Nambi 2010). Moreover, mercury
DNAPL entrapment in fully water-saturated porous me-
dia results in much lower residual saturation than other
DNAPLs, like TCE and PCE (Devasena and Nambi
2010).
To date, except for capillary pressure-saturation col-
umn experiments (Devasena and Nambi 2010;
DAniello 2017;DAniello et al. 2018, Under Review),
there are no experimental observations of Hg
0
DNAPL
infiltration and distribution in water-saturated porous
media, where it is expected to have a high capacity to
infiltrate (DAniello 2017;DAniello et al. 2018, Under
Review). Therefore, to better understand the processes
governing elemental mercury DNAPL flow behavior in
water-saturated porous media under more realistic sce-
narios, and to compare them with those of other
DNAPLs, a series of flow container experiments in fully
water-saturated stratified sands and glass beads is here
described. To allow a quantitative analysis of Hg
0
and
PCE infiltration and distribution behavior, non-intrusive
25 Page 2 of 17 Water Air Soil Pollut (2018) 229:25
measurements of samples porosity, PCE and Hg
0
satu-
rations were performed with a dual gamma ray system.
Finally, numerical simulations were performed to assess
to what extent the conventional multiphase flow model-
ing can allow for the prediction of Hg
0
DNAPL migra-
tion in water-saturated porous media.
2MaterialsandMethods
2.1 Porous Media and Fluids
Two types of glass beads (Glaswarenfabrik Karl Hecht
2013;Retsch2013), and two different sands, namely
Filtersand 0.20.5 mm and Silversand S60, (FILCOM-
Sibelco Group 2015), were used. On the basis of Went-
worth classification (1922), the two types of glass beads
are classified as coarse (d
50
= 1 mm; uniform distribu-
tion) and medium (d
50
= 0.375 mm; 0.2000.500 mm),
while the two sands as medium sands and will be
referred as to medium sand 1 (MS1) and medium sand
2(MS2). The particle size distribution of MS1 (d
50
=
0.39 mm; 0.2000.500 mm) is wider than MS2 (d
50
=
0.26 mm; 0.1250.355 mm). For each granular material,
information regarding intrinsic permeability (Table 1)
was obtained experimentally (DAniello 2017;
DAniello et al. 2018, Under Review) by means of the
falling head method (Bear 1972).
Demineralized water was used as the wetting fluid,
while laboratory grade (99%) PCE (Acros Organics)
and 99.9 + % redistilled liquid elemental mercury (Hg
0
)
(Alfa Aesar) were used as the nonwetting fluids
(Table 2).
The capillary pressure-saturation, P
c
(S), data
(Table 3) were obtained with column experiments per-
formed under stable flow behavior (DAniello 2017;
DAniello et al. 2018, Under Review).
2.2 Flow Container Experiments
Two poly-methyl methacrylate (PMMA) flow con-
tainers (Fig. 1) were used for the DNAPL infiltration
experiments. The one filled with sand was used to
observe PCE and Hg
0
DNAPL infiltration, while the
one filled with glass beads was used only for Hg
0
DNAPL. Therefore, due to the preferential wettability
of PCE to PMMA, the two front walls of the container
filled with sand were layered with a glass plate of
0.5 cm. Conversely, the two side walls, sufficiently far
from the DNAPL source, were not expected to come in
contact with PCE and were not covered with glass. For
the same reason of the front walls, the two PMMA inlet
walls, of 1 cm thickness each, were layered with a glass
plate of 1 mm. More details about the flow containers
can be found in DAniello (2017).
The first flow container (Fig. 1, left) was mainly
filled with MS1 and, to study the effects of macro
heterogeneities on PCE and Hg
0
migration, a rectangu-
Tabl e 1 Porous media properties
Porous medium Particle
density
(g/cm
3
)
Porosity Intrinsic
permeability
(10
7
cm
2
)
Coarse glass beads 2.65 0.410 45.15
Medium glass
beads
2.65 0.440 9.43
Medium sand 1 2.67 0.368 1.61
Medium sand 2 2.65 0.391 1.79
Tabl e 2 Fluids properties
Parameter Unit Hg
0
PCE Water
Density g/cm
3
13.5
a
1.63
b
0.998
Dynamic viscosity 10
3
Pa s 1.55
c
0.89
b
0.98
Surface tension Dynes/cm 485
d
32.6
e
72
Interfacial tension with water Dynes/cm 375
c
47.8
b
a
CRC (2014)
b
Pennel et al. (1996)
c
US DOE (2001) and references therein
d
Adamson and Gast (1997)
e
Mercer and Cohen (1990)
Tabl e 3 P
c
(S) data
Sample System Porosity α
(cm
1
)
nEntry head
(cm of
NAPL)
Coarse GB Hg
0
-water 0.410 0.0167 5.58 2.84
Medium GB Hg
0
-water 0.440 0.0071 6.83 8.36
MS1 PCE-water 0.353 0.0517 13.41 9.90
MS2 PCE-water 0.375 0.0365 28.99 15.77
MS1 Hg
0
-water 0.354 0.0087 7.97 6.19
MS2 Hg
0
-water 0.375 0.0049 12.83 12.51
αand nare the van Genuchten-Mualem model (Mualem 1976;
van Genuchten 1980)parameters
GB glass beads
Water Air Soil Pollut (2018) 229:25 Page 3 of 17 25
lar lens of 10 cm length by 5 cm height of MS2 was
located in the middle of the container, at 9 cm from the
bottom (Fig. 1, left). The water table was fixed at 1 cm
from the top layer of MS1, located at 25 cm from the
bottom. The two inlet walls were placed 2.5 cm deep
from the top of the MS1 layer, in the middle of the
container, with a spacing of 1.5 cm between each other.
The sand in the container was packed under water-
saturated conditions to avoid any entrapped air between
the sand particles. The flow container was initially filled
halfway with water and then, after having thoroughly
mixed the sandwith water in another container, the sand
was gradually poured into it. To maintain a fully water-
saturated system, a water head over the top layer of the
sand was kept at all times. At each filling step, each sand
layer was tamped in the same manner to compact the
porous medium. To prevent accumulation of sand in the
bottom valve, a nylon filter was put in the bottom hole.
The packing procedure described above was also
used for the PCE experiment. To create visual contrast
between PCE and water, thus allowing for visual obser-
vation of PCE infiltration, the water was dyed with
Bromophenol Blue (Acros Organics), a water soluble
pH indicator. As a result, the water in the container
showed a dark blue color. The water was dyed rather
than the PCE to prevent affecting PCE properties (Ap-
pendix 5.1).
The second sample (Fig. 1, right) consisted of a
water-saturated stratified glass beads system, made of
two layers of coarse glass beads with a medium glass
beads layer of 6 cm height in between, located at 8 cm
from the bottomof the container. The packing procedure
was the same of the sand sample used for the Hg
0
DNAPL experiment, except that glass beads were
poured in the container without any additional compac-
tion besides of the one induced by their own weight. The
two inlet walls, located 2 cm deep from the top of the
upper coarse glass beads layer, had a thickness of
0.5 cm, and the spacing between the two was of 1 cm.
The water table was fixed at 1.5 cm from the top of the
upper coarse glass beads layer, located at 25 cm from the
bottom.
For each experiment, a fixed amount of DNAPL was
carefully added between the inlet walls, thus allowing
the DNAPLs to build enough head to infiltrate and
distribute in the porous samples.
2.3 Dual Gamma System: Setup, Principles,
and Measurements
The Gamma Ray Attenuation (GRA) technique
(Ferrand et al. 1986; Hofstee et al. 1998a;Hofstee
et al. 1998b; Barataud et al. 1999; Johnson et al. 1999;
Baytaşand Akbal 2002; Pires et al. 2005; Pires and
Bereira 2014) is essentially based on the interaction of
radiation with matter. Depending on how the radiation
travels through the sample and, in particular, on how the
radiation gets absorbed, it is possible to determine, in a
Fig. 1 Schematic representation of the flow container samples (dimensions in cm): with sands (left) and glass beads (right)
25 Page 4 of 17 Water Air Soil Pollut (2018) 229:25
non-invasive way, properties related to the structure of
the porous medium and fluids saturations within the
pores.
The dual gamma ray system used consists of two
radioactive sources, namely Americium 241 (
241
Am)
and Cesium 137 (
137
Cs), which emit gamma rays, and
a sodium iodide (NaI) detector (Fig. 2).
The attenuation of
241
Am and
137
Cs gamma rays,
induced by the absorption of a medium, can be de-
scribed by Beer-Lambert law:
Iδ¼Iδ
0eμδx
 ð1Þ
where δ= Am, Cs, I
0
is the incident intensity of gamma
rays, Iis the intensity of gamma rays after attenuation
through a medium of length x(path length), and μis the
linear attenuation coefficient of the material. The atten-
uation coefficient, which is dependent on the physical
density of the material, measures the probability per unit
length of a photon to be absorbed or scattered while
interacting with a sample. In the case where the sample
is shielded from the gamma radiation by the presence of
the additional thickness of a container, the incident
intensity is called background intensity and must be
measured in advance. More details regarding the dual
gamma ray measurements can be found in the Appendix
(Section 5.2).
Non-intrusive measurements of the porosity field
followed the packing of the flow container samples
(Section 2.2). These measurements were performed
over the whole samples by means of the dual gamma
ray system, on the basis of a regular grid with 1-cm
spacing. DNAPLs saturation profiles were measured
24 h after the end of the infiltration process, established
when there was no further change of DNAPL level in
the inlet.
2.4 Numerical Modeling: Hg
0
DNAPL Infiltration
and Distribution in Glass Beads
To assess to what extent numerical modeling can de-
scribe Hg
0
DNAPL infiltration and distribution in
water-saturated porous media, the glass beads flow con-
tainer experiment was simulated with the NAPL module
of the Groundwater Dynamics Analysis (GDAn) code
(DAniello 2017). GDAn is a two-dimensional finite
element model on unstructured triangular mesh meant
for the analysis of groundwater, multiphase flow, and
solute transport dynamics in porous media. In particular,
the NAPL module solves numerically the governing
equations of multiphase flow of immiscible fluids in
porous media (Abriola and Pinder 1985; Parker et al.
1987;Parker1989), by using the Picard iteration
scheme with an incremental solution procedure (Istok
1989). The model is based on the extended Darcyslaw
for multiphase flow (Bear 1972); therefore, it describes
NAPLs infiltration and distribution in porous media at
the representative elementary volume scale.
The input parameters of the numerical simulations
were mostly taken from Tables 1,2,and3. In addition
to these data, the irreducible water saturation was set
to 0.05, while the maximum entrapped NAPL satu-
ration to 0.10 (DAniello 2017). Specific storage
coefficients were determined assuming a compress-
ibility of 4.6e10 Pa
1
for water and of 3.7e11 Pa
1
for Hg
0
(Young et al. 2012), and of 1.4e11 Pa
1
for
glass beads (Luo et al. 2015), and an average porosity of
0.380 based on dual gamma ray measurements. The
scaling parameter was assumed equal to the ratio be-
tween water surface tension and Hg
0
interfacial tension
with water, on the basis of Lenhard and Parker (1987)
scaling theory, resulting in a value of 0.192. To describe
the P
c
-S and k
r
-S (relative permeability-saturation) rela-
tionships between elemental mercury and water, the van
Genuchten-Mualem model (Mualem 1976;van
Genuchten 1980) was used.
To describe the variability of intrinsic permeability
with porosity and of capillary retention parameters with
intrinsic permeability, the Kozeny-Carman formula
(Kozeny 1927;Carman1938,1956; Bear 1972; Fitts
2002) and the scaled form of the generalized Leverett
function (Leverett 1941; Kueper and Frind 1991;
Dekker and Abriola 2000) were applied. In this way,
each cell of the numerical domain was set to different
values for porosity, intrinsic permeability, and van
Genuchten αparameter. The van Genuchten nparame-
ter was assumed constant within the same glass beads
type. With the information derived from the gamma ray
Fig. 2 Schematic illustration of
the dual gamma ray setup
Water Air Soil Pollut (2018) 229:25 Page 5 of 17 25
measurements, the porosity field (Fig. 7) was deduced
for the whole domain and the Kozeny-Carman formula
(Eq. 2) was used to determine the intrinsic permeability
(k)ineachcell:
k¼d2
50
SF2
ϕ3
1ϕðÞ
2ð2Þ
with ϕthe porosity, and SF the particles shape factor,
(SF = 20.7), determined as the best fit with the data
(DAniello 2017). Then, the scaled form of the general-
ized Leverett function (Eq. 3) was used to determine the
van Genuchten αparameter in each cell. In particular,
for any pair ofpermeability values, the van Genuchtenα
parameter can be scaled as:
α2¼α1
k2
k1

γ
ð3Þ
where α
1
is the van Genuchten parameter and k
1
is the
intrinsic permeability of the reference samples (Tables 1
and 3), α
2
and k
2
are the values to be determined in each
cell, while γis a fitting parameter (equal to 0.5 for the
conventional Leverett function). To verify to what ex-
tent the spatial variability of intrinsic permeability plays
a significant role in the numerical prediction of Hg
0
DNAPL infiltration and distribution, the γparameter
was subjected to a sensitivity analysis, assuming the
value of 0.5 of the conventional Leverett function, and
the very large values of 0.75 and 0.95, thus considering
three different numerical scenarios.
The spatial domain was discretized in 1273 nodes
and 2413 unstructured triangular cells, suitably refined
in proximity of the DNAPL reservoir and of the top of
the medium glass beads layer. Across the whole domain,
the water hydraulic head was set to 26.5 cm, and the
boundary nodes on the top of the domain, outside the
inlet, were constrained to this value assuming a Dirichlet
boundary condition. Between the inlet walls, a volume
of 10 ml of liquid Hg
0
was injected into the domain with
an average infiltration rate of 0.099 ml/s, over a time of
101 s. On all other boundaries, a no-flux-boundary
condition was assumed for both water and elemental
mercury. The solution was advanced over time with a
variable time step ranging from 0.1 to 2 s depending on
convergence history, while convergence of the solution
was ensured by an absolute tolerance of 10
5
m.
3 Results and Discussion
3.1 Liquid Hg
0
Attenuation Potential of Gamma Rays
Liquid Hg
0
showed a strong attenuation potential of
gamma rays if compared to demineralized water and
PCE (Table 4). As a matter of fact, for all the path
lengths considered (0.1 to 4 cm), liquid Hg
0
shielded
241
Am radiation completely.
The Americium peak practically disappears (Fig. 3)
even with 1-mm thickness of elemental mercury be-
tween the radioactive source and the detector, and only
Cesium radiation is able to pass through liquid Hg
0
.
Even though Cesium has a stronger capacity to pen-
etrate matter than Americium, liquid Hg
0
shows a re-
markably high attenuation coefficient with respect to
this gamma radiation, about 17 times higher than water
(Table 4). According to Eq. 1, this implies that 0.5 cm of
liquid Hg
0
is sufficient to attenuate Cesium radiation of
about 47%, whereas 2 cm can induce an attenuation of
92%. However, according to the thickness (45 cm) and
average porosity (0.3600.380) of the porous samples
used for the Hg
0
DNAPL flow container experiments, at
100% of mercury saturation, liquid Hg
0
can reach a
maximum thickness of about 2 cm within the samples,
thus enabling the use of the Cesium source to determine
its complete range of saturation.
Furthermore, with the current dual gamma ray sys-
tem, elemental mercury saturation can be correctly de-
rived from Eq. 1only when the porous medium is dry or
fully water saturated. As a matter of fact, when a three-
phase system of air, NAPL, and water is encountered
within a porous medium, the determination of the liquid
phases saturation is usually performed by solving simul-
taneously Eq. 1for both radioactive sources, but, in the
case of liquid mercury, this is not possible because of the
complete attenuation of Americium.
Therefore, when analyzing Hg
0
DNAPL flow behav-
ior in porous media by means of non-intrusive techniques
as GRA, attention must be paid to the choice of the
Tabl e 4 Fluids attenuation coefficients
Fluid μ
Am
(cm
1
)μ
Cs
(cm
1
)
Demineralized water (DW) 0.186 0.075
DW + Bromophenol 0.185 0.072
PCE 0.633 0.113
Hg
0
1.280
25 Page 6 of 17 Water Air Soil Pollut (2018) 229:25
sources of radiation. In particular, the sources should be
strong enough to penetrate liquid Hg
0
to at least some
extent, and the sample thickness should not exceed a
certain path length to allow a correct characterization of
Hg
0
DNAPL saturation range within the porous medium.
3.2 PCE and Hg
0
Flow Container Experiments
with Sands
The main difference between the two PCE and Hg
0
infiltration experiments was found in the early stage of
the infiltration process. While PCE infiltrated and dis-
tributed within the water-saturated porous sample, liquid
Hg
0
did not. When elemental mercury was poured be-
tween the inlet walls, the sand particles of the top layer
were displaced and started floating over the top of the
Hg
0
surface, due to their lack of cohesion and lower
density. More mercury was added to build a sufficient
head to overcome the sand entry head but, rather than
infiltrating in the porous sample, Hg
0
found a preferen-
tial pathway between the inlet walls and the front walls
of the container, thus spreading over the top surface of
MS1. Then the experiment was stopped. Conversely,
12 ml of PCE infiltrated over a time of 436 s, under an
initial head of 2.16 cm. Unfortunately, Bromophenol
Blue did not allow the visualization of the PCE infiltra-
tion front within the stratified porous medium; hence, no
track of it over time is available.
Elemental mercury is more prone to infiltrate than
PCE in water-saturated porous media (DAniello 2017;
DAniello et al. 2018, Under Review) as it shows lower
DNAPL entry heads (Table 3). Therefore, the reason of
such a sharp difference in the infiltration behavior ex-
hibited by the two DNAPLs in the flow container ex-
periments is not immediately clear. Most probably, the
difference in the porosity field of the two flow container
samples played a major role.
Dual gamma ray measurements of the porosity field
(Figs. 4and 5) revealed significant differences among the
two samples. The sample used for the Hg
0
flow container
experiment showed a porosity remarkably lower than the
one used for the infiltration of PCE. Such a strong differ-
ence in porosity likely resulted in a sharp difference in
intrinsic permeability, thus inducing differences in entry
head. In particular, this difference in porosity was more
pronounced in proximity of the inlet, namely where the
infiltration should have occurred. Therefore, Hg
0
did not
infiltrate because it was unable to build a sufficient head
to overcome the higher entry head induced by a locally
lower intrinsic permeability. Hence, the increase in
DNAPL pressure, following the further addition of Hg
0
,
resulted in the overcome of the resistance exerted by
water between the inlet and the front walls rather than
that of water in the sand, thus inducing the Hg
0
escape
through a preferential pathway. To prevent the experi-
ment failure, the inlet walls should have been sealed with
the front walls of the container, thus forcing liquid Hg
0
to
build a head within the inlet space. In addition, whether
the Hg
0
head was higher than the inlet walls height, the
setup could have been easily modified to allow for inlet
walls of different dimensions and shape.
The discontinuity observed in the PCE saturation
profile (Fig. 6) also suggests a dependency of the infil-
tration behavior on the porous medium structure. The
Fig. 3 Typical (gray line) and 1-
mm thickness Hg
0
sample (black
line) dual gamma ray detected
signals
Water Air Soil Pollut (2018) 229:25 Page 7 of 17 25
dual gamma ray detected the presence of PCE every-
where in the lowest part of the sample, with a maximum
saturation of 0.094, while almost no PCE was detected
in the top part. Most likely, for PCE, the effect of micro
heterogeneities was even more pronounced than macro
heterogeneities. As a result, PCE migration probably
developed under an unstable flow regime, enhanced
by the micro heterogeneities present within the sample,
thus resulting in fingers formation (Poulsen and Kueper
1992; Illangasekare 1998; Mayer and Hassanizadeh
2005). This would explain why the dual gamma ray
system was not able to perform a continuous measure-
ment of the PCE front and why PCE was found in the
finer sand. Fingers are expected to be thin, of the order
of the pore sizes; hence, the dual gamma ray beam had
locally either missed the fingers or smeared the satura-
tion value through the whole thickness of the porous
medium, thus resulting in an overall negligible mea-
sured PCE saturation in the top part of the sample.
3.3 Hg
0
Flow Container Experiment with Glass Beads
The porosity field and the elemental mercury saturation
profile of the glass beads flow container experiment are
depicted in Figs. 7and 8.
Fig. 4 PCE flow container
sample: measured porosity field
Fig. 5 Hg
0
flow container
sample: measured porosity field
25 Page 8 of 17 Water Air Soil Pollut (2018) 229:25
The final distribution of elemental mercury (Fig.
8) in the highest porosity zone (Fig. 7) shows that its
infiltration and distribution were affected by the het-
erogeneities present within the sample. The infiltra-
tion front was fast and reached the top of the medium
glass beads layer in 31 s. After 101 s, the elemental
mercury source was depleted, and the distribution
occurred. No elemental mercury was able to pene-
trate the medium glass beads layer, and a flat pool, of
about 1-cm height, formed on top of this layer. Static
equilibrium was visually observed after 112 s from
the beginning of the infiltration process. The
maximum DNAPL saturation measured at the end
of the experiment was of 0.368, where the Hg
0
pool
was formed.
3.4 Numerical Modeling of the Hg
0
Flow Container
Experiment with Glass Beads
Numerical simulations, performed with GDAn
(Section 2.4), showed an overall good agreement with
the experimental results. The main features of Hg
0
DNAPL migration, such as the magnitude of the infil-
tration time, the stretched shape of the infiltration front
Fig. 6 Measured PCE saturation
profile
Fig. 7 Glass beads flow
container sample: measured
porosity field
Water Air Soil Pollut (2018) 229:25 Page 9 of 17 25
along the vertical axis as a result of the predominance of
gravity over capillary forces, and the spreading above
the medium glass beads layer without penetrating it,
were well captured.
The inspection of Figs. 9,10,and11,wherethe
progression of the Hg
0
infiltration front is reported
over time for each permeability scenario, shows that
themoretheγparameter (Eq. 3) increases, the sharper
becomes the variation of the van Genuchten αparam-
eter with intrinsic permeability, thus resulting in a
better prediction of the infiltration and distribution
phenomena. With increasing γ, the shape of the infil-
tration front becomes less symmetric, moving to-
wards areas with higher porosity and intrinsic perme-
ability, as observed in the experiment (Fig. 8), where
Hg
0
mainly migrated in the left part of the container
once the medium glass beads layer was reached.
However, numerical simulations were not able to
perfectly replicate the experimental observations. As
a matter of fact, the simulated infiltration front was
Fig. 8 Measured Hg
0
saturation
profile in glass beads
Fig. 9 γ=0.50Simulated Hg
0
infiltration front (Hg
0
saturation= 0.03) over time (10
100 s) every 10 s
25 Page 10 of 17 Water Air Soil Pollut (2018) 229:25
slower than the observed one, reaching the medium
glass beads layer in about 40 s rather than 31 s, and
achieving the static equilibrium configuration in
about 180 s rather than 112 s.
Furthermore, due to the more pronounced lateral
spreading of the simulated Hg
0
infiltration front above
the medium glass beads layer, the resulting maximum
saturations (Figs. 12,13,and14) were lower than the
measured ones, ranging between 0.114 and 0.128 with γ
ranging from 0.95 to 0.5.
However, to numerically reproduce the final distri-
bution of elemental mercury observed in the experiment
(Fig. 8), it was essential to take into account the spatial
variability of both intrinsic permeability and retention
properties (Eqs. 2and 3) across the whole simulation
domain. As a matter of fact, with increasing γ,elemental
Fig. 10 γ=0.75Simulated
Hg
0
infiltration front (Hg
0
saturation= 0.03) over time (10
100 s) every 10 s
Fig. 11 γ=0.95Simulated
Hg
0
infiltration front (Hg
0
saturation= 0.03) over time (10
100 s) every 10 s
Water Air Soil Pollut (2018) 229:25 Page 11 of 17 25
mercury was found at higher saturations in the left part
of the domain, while it reached lower saturations in the
right part, where the Hg
0
pool became thinner (Figs. 12,
13,and14).
The discrepancies found between the experimental
results andthe numerical predictions are probably due to
the lack of knowledge of the Hg
0
inflow rate over time
and to the low agreement of the correlation between the
measured porosity field with the estimated intrinsic
permeability and retention properties. As a matter of
fact, considering an average value of the inflow rate
makes it practically impossible to predict the position
of the infiltration front over time, and it may lead to
substantial overestimations of the extent of Hg
0
migra-
tion. In absence of any measurement, the prediction of
the inflow rate in a specific porous system requires a
detailed knowledge of the locations of varying perme-
ability lenses in the proximity of the DNAPL source
(Kueper and Gerhard 1995). Even though the non-
intrusive measurements performed with the dual gamma
ray system allowed a detailed description of the porosity
field, the relationship of porosity with intrinsic perme-
ability, as well as with retention properties, would have
required more experimental determinations to guarantee
a more reliable estimation. The information based on
Kozeny-Carman formula and on the modified Leverett
Fig. 12 γ=0.50Simulated
Hg
0
saturation at 180 s
Fig. 13 γ=0.75Simulated
Hg
0
saturation at 180 s
25 Page 12 of 17 Water Air Soil Pollut (2018) 229:25
function proved to be not sufficient outside the porosity
range defined with the hydraulic conductivity and P
c
(S)
experiments on glass beads. Nevertheless, the three
scenarios showed improvements with increasing γ,thus
confirming numerical predictions dependency on the
correlation between intrinsic permeability and retention
properties, as found for organic DNAPLs by Dekker and
Abriola (2000).
4 Conclusions
As for other DNAPLs, infiltration and distribution of
Hg
0
DNAPL in water-saturated porous media are affect-
ed by the heterogeneities present within the porous
formation. In the flow container experiment with sands,
liquid Hg
0
would have required to build a higher head
than PCE to infiltrate due to the presence of a lower
permeability zone beneath the inlet location. However,
the experiment was not successful due to a defect of the
setup that allowed liquid mercury to find a preferential
pathway through the inlet walls and the front walls of
the container. As a result, liquid Hg
0
spread over the top
of the sample, and no infiltration occurred. Conversely,
the Hg
0
flow container experiment with glass beads was
successful. In this experiment, visual observations and
dual gamma ray measurements showed that liquid Hg
0
preferentially migrated towards higher porosity zones.
As porosity is in relation with intrinsic permeability,
hence with retention properties, elemental mercury
DNAPL infiltration in water-saturated porous media is
likely to occur through higher permeability lenses,
where the entry heads are expected to be lower.
Numerical simulations of the glass beads experiment
showed an overall good agreement with the experimen-
tal results as the main features of Hg
0
DNAPL infiltra-
tion and distribution in water-saturated porous media
were well captured. However, numerical simulations
were not able to perfectly replicate the experimental
observations, despite the detailed description of the
porosity field performed with the dual gamma ray sys-
tem. The knowledge of the inflow rate, the reliable
estimation of the porous formation permeability, and
the accurate representation of the correlation between
P
c
(S) curves and intrinsic permeability proved to be
critical for the prediction of liquid Hg
0
infiltration and
distribution in water-saturated porous media as for or-
ganic DNAPLs, like PCE.
However, if compared to other DNAPLs, such as
PCE, liquid Hg
0
shows a strong attenuation potential
of gamma rays. Therefore, when analyzing Hg
0
DNAPL flow behavior in porous media with non-
intrusive techniques as GRA, the sources of radiation
should be strong enough to penetrate liquid Hg
0
to at
least some extent, and the sample thickness should not
exceed a certain path length to allow a correct charac-
terization of Hg
0
DNAPL saturation range within the
porous medium.
Acknowledgements The authors wish to thank the two anony-
mous reviewers for their constructive comments which helped to
improve this manuscript.
Fig. 14 γ=0.95Simulated
Hg
0
saturation at 180 s
Water Air Soil Pollut (2018) 229:25 Page 13 of 17 25
Funding information This research was financially supported
by the P.O.R. Campania FSE 2007/2013-2014/2020.
Appendix
PCE Experiment with Sands: Bromophenol Blue
Addition to Water
Bromophenol Blue was used in a concentration of 1 g/l,
equal to 1/4 of bromophenol solubility in water (ONeil
2006). To avoid any presence of bromophenol granular
residuals, after several rounds of mechanical mixing, the
water was filtered using a 10-μm mesh nylon filter.
The choice of dying water rather than PCE was
dictated by the fact that oil soluble dyes like Sudan III,
Sudan IV, and Oil Red, generally used to dye PCE
(Schwille 1988; Hofstee et al. 1998a; Hofstee et al.
1998b), proved to influence PCE interfacial properties
(Tuck et al. 2003), reducing its entry pressure and af-
fecting its spreading behavior, thus enhancing its mobil-
ity, and, in the case ofSudan IV, to leave residuals on the
sediments as a result of PCE dilution (Hartog et al.
2010), thus affecting the porous medium characteristics
and, potentially, PCE flow behavior. Therefore, water
was dyed with Bromophenol Blue, generally used for
conservative tracer tests (Cirpka et al. 2006;Rockhold
et al. 2007; Chiogna 2010). Furthermore, as shown in
Table 4(Section 3.1), adding bromophenol to
demineralized water does not result in any appreciable
change in the water phase gamma ray attenuation.
Dual Gamma Ray Measurements: Additional Details
The NaI detector measures an average value, over the
beam cross section (6-mm diameter) and the path length,
of the number of arriving photons (counts) over their
energy, rather than intensities. Intensities are determined
by first summing up the count rates of the peak areas
(Fig. 3), from 40.4 to 82 keV for
241
Am, and from 623.5
to 740.9 keV for
137
Cs, and then by dividing them by the
measurement time. The measurement time was chosen
according to liquid Hg
0
high attenuation potential
(Section 3.1) and was set to 120 s (DAniello 2017).
The attenuation coefficients were measured by plac-
ing different amounts of cuvettes in a row, thus leading
to different thicknesses (path lengths, Eq. 1). To estab-
lish the background intensity, the cuvettes were mea-
sured empty first, then, with the aim of determining the
intensity measurements, each cuvette was filled with the
substance/material to be analyzed. Then, on the basis of
Beer-Lambert equation (Eq. 1), the attenuation coeffi-
cients were determined by means of a least square fitting
technique over the measurement points.
Polystyrene (PS) squared cuvettes, of 1-cm thickness
and 4-ml volume, were used for most of the measure-
ments, while PMMA was used for tetrachloroethylene,
due to the almost null resistance of PS to PCE. Due to
the high attenuation potential showed by liquid Hg
0
(Section 3.1), to completely investigate its behavior with
respect to gamma radiation, further measurements were
carried out for path lengths of 0.5 and 0.1 cm (1 mm).
For the half-centimeter path length, a squared cuvette, of
half-centimeter thickness, was used. Instead, for the 1-
mm path length, the desired Hg
0
thickness was obtained
by joining together three square glass plates, of 1-mm
thickness each, and creating a circular hole of 2-cm
diameter into the middle plate. The hole was connected
to the top by a channel of 2-mm width to allow the
filling with Hg
0
, after which the top port was sealed with
super glue. More details regarding the dual gamma ray
setup, its calibration, and operating principles can be
found in DAniello (2017).
References
Abriola, L. M., & Pinder, G. F. (1985). A multiphase approach to
the modeling of porous media contamination by organic
compounds: 1. Equation development. Water Resources
Research, 21(1), 1118.
Adamson, A. W., Gast, A. P. (1997). Physical Chemistry of
Surfaces. 6Ed, Wiley.
Arbestain, M. C., Rodriguez-Lado, L., Bao, M., Macias,F. (2009).
Assessment of mercury-polluted soils adjacent to an old
mercury-fulminate production plant. Applied and
Environmental Soil Science.
Barataud, F., Moyne, C., & Stemmelen, D. (1999). Measurement
of soil water diffusivity of an undisturbed forest soil using
dual-energy gamma radiation technique. Soil Science,
164(7), 493502.
Baytaş, A. F., & Akbal, A. (2002). Determination of soil param-
eters by gamma-ray transmission. Radiation Measurements,
35(1), 1721.
Bear, J. (1972). Dynamics of fluids in porous media. New York:
Dover Publications, Inc..
Bernaus, A., Gaona, X., van Ree, D., & Valiente, M. (2006).
Determination of mercury in polluted soils surrounding a
chlor-alkali plant: direct speciation by X-ray absorption spec-
troscopy techniques and preliminary geochemical character-
isation of the area. Analytica Chimica Acta, 565(1), 7380.
25 Page 14 of 17 Water Air Soil Pollut (2018) 229:25
Biester, H., Müller, G., & Schöler, H. F. (2002). Estimating distri-
bution and retention of mercury in three different soils con-
taminated by emissions from chlor-alkali plants: part I.
Science of the Total Environment, 284(1), 177189.
Bollen, A., Wenke, A., & Biester, H. (2008). Mercury speciation
analyses in HgCl
2
-contaminated soils and groundwater
implications for risk assessment and remediation strategies.
Water Research, 42(1), 91100.
Bouffard, A., & Amyot, M. (2009). Importance of elemental
mercuryinlakesediments.Chemosphere, 74(8), 10981103.
Brooks, S. C., & Southworth, G. R. (2011). History of mercury use
and environmental contamination at the Oak Ridge Y-12
Plant. Environmental Pollution, 159(1), 219228.
Carman, P. C. (1938). The determination ofthe specific surface of
powders. Journal of the Society of Chemical Industry, 57,
225.
Carman, P. C. (1956). Flow of gases through porous media.
London: Butterworths Scientific Publications.
Chiogna, G. (2010). Transverse mixing of conservative and reac-
tive tracers in porous media. Doctoral dissertation:
Universität Tübingen.
Cirpka, O. A., Olsson, Å.,Ju, Q., Rahman, M. D., & Grathwohl, P.
(2006). Determination of transverse dispersion coefficients
from reactive plume lengths. Ground Water, 44(2), 212221.
Compeau, G. C., & Bartha, R. (1985). Sulfate-reducing bacteria:
principal methylators of mercury in anoxic estuarine sedi-
ment. Applied and Environmental Microbiology, 50(2), 498
502.
CRC (2014). Handbook of Chemistry and Physics. 94th ed.
(Internet Version 2014).
DAniello, A. (2017). The Flow Behaviour of Elemental Mercury
DNAPL in Porous Media. PhD Thesis, Università degli Studi
di Napoli Federico II.
Davis, A., Bloom, N. S., & Que Hee, S. S. (1997). The environ-
mental geochemistry and bioaccessibility of mercury in soils
and sediments: a review. Risk Analysis, 17(5), 557569.
Deeb, R., Hawley, E., Kell, L., O'Laskey, R. (2011). Assessing
alternative endpoints for groundwater remediation at con-
taminated sites. ESTCP-Project ER-200832.
Dekker, T. J., & Abriola, L. M. (2000). The influence of field-scale
heterogeneity on the infiltration and entrapment of dense
nonaqueous phase liquids in saturated formations. Journal
of Contaminant Hydrology, 42(2), 187218.
Devasena, M., & Nambi, I. M. (2010). Migration and entrapment
of mercury in porous media. Journal of Contaminant
Hydrology, 117,6070.
Di Molfetta, A., Fracassi, F. (2008). Consulenza tecnica sulla
contaminazione in atto nell'area del polo industriale di
Bussi. Procura della Repubblica presso il Tribunale di
Pescara, procedimento n. 12/2006 R.G.N.R. mod. U.
Eichholz,G. G., Petelka, M. F., & Kury, R. L. (1988). Migration of
elemental mercury through soil from simulated burial sites.
Water Research, 22(1), 1520.
US EPA (2007). Treatment Technologies for Mercury in Soil,
Waste, and Water. EPA Report 542-R-07-003. US
Environmental Protection Agency, Office of Superfund
Remediation and Technology Innovation.
Euro Chlor (1999). The mercury cell technology in Western
Europe. Presentation at the OSPAR Workshop on the
Chlor-Alkali Industry, 2729 September 1999, Brussels.
Ferrand, L. A., Milly, P. C. D., & Pinder, G. F. (1986). Dual-
gamma attenuation for the determination of porous medium
saturation with respect to three fluids. Water Resources
Research, 22(12), 16571663.
FILCOM (2015). Technical Datasheet Filtersand 0.20.5 mm and
Silversand S60.
Fitts, C. R. (2002). Groundwater science. Academic Press.
Gilmour, C. C., Henry, E. A., & Mitchell, R. (1992). Sulfate
stimulation of mercury methylation in freshwater sediments.
Environmental Science & Technology, 26(11), 22812287.
Golder Associates (2011). 2010 conceptual site modelbotany.
Report No. 107623162_001_R_Rev0.
González-Fernández, B., Menéndez-Casares, E., Meléndez-
Asensio, M., Fernández-Menéndez, S., Ramos-Muñiz, F.,
Cruz-Hernández, P., & González-Quirós, A. (2014).
Sources of mercury in groundwater and soils of west Gijón
(Asturias, NW Spain). Science of the Total Environment, 481,
217231.
Hartog, N., Cho, J., Parker, B. L., & Annable, M. D. (2010).
Characterization of a heterogeneous DNAPL source zone in
the Borden aquifer using partitioning and interfacial tracers:
residual morphologies and background sorption. Journal of
Contaminant Hydrology, 115(1), 7989.
Glaswarenfabrik Karl Hecht (2013). Technical Datasheet 1 mm
Glass Beads.
Hofstee, C., Dane, J. H., & Walker, R. C. (1998a). Infiltration and
redistribution of perchloroethylene in stratified water-
saturated porous media. Soil Science Society of America
Journal, 62(1), 1322.
Hofstee, C., Oostrom, M., Dane, J. H., & Walker, R. C. (1998b).
Infiltration and redistribution of perchloroethylene in partial-
ly saturated, stratified porous media. Journal of Contaminant
Hydrology, 34(4), 293313.
Hylander, L. D., & Meili, M. (2003). 500 years of mercury
production: global annual inventory by region until 2000
and associated emissions. Science of the Total Environment,
304(1), 1327.
Illangasekare, T. H. (1998). Flow and entrapment of nonaqueous
phase liquids in heterogeneous soil formations, in physical
non-equilibrium in soil. Ann Arbor Press, Chelsea,
Michigan, 417423.
Istok, J. D. (1989). Groundwater modelling by the finite element
method. Water Resources Monograph, 13, American
Geophysical Union, 2000 Florida Avenue, NW,
Washington, DC 2000.
ITRC (2012). Using remediation risk management to address
groundwater cleanup challenges at complex sites. p 31.
Johnson, J. C., Sun, S., & Jaffé, P. R. (1999). Surfactant enhanced
perchloroethylene dissolution in porous media: the effect on
mass transfer rate coefficients. Environmental Science &
Technology, 33(8), 12861292.
Kozeny, J. (1927). Ueber kapillare Leitung des Wassers im Boden.
Akademie der Wissenschaften in Wien, 136(2a), 271.
Kueper, B. H., & Frind, E. O. (1991). Two-phase flow in hetero-
geneous porous media: 2. Model application. Wat er
Resources Research, 27(6), 10591070.
Kueper, B. H., & Gerhard, J. I. (1995). Variability of point source
infiltration rates for two-phase flow in heterogeneous porous
media. Water Resources Research, 31(12), 29712980.
Lechler, P. J., Miller, J. R., Lacerda, L. D., Vinson, D., Bonzongo,
J. C., Lyons, W. B., & Warwick, J. J. (2000). Elevated
Water Air Soil Pollut (2018) 229:25 Page 15 of 17 25
mercury concentrations in soils, sediments, water, and fish of
the Madeira River basin, Brazilian Amazon: a function of
natural enrichments? Science of the Total Environment,
260(1), 8796.
Lenhard, R. J., & Parker, J. C. (1987). Measurement and predic-
tion of saturation-pressure relationships in three-phase porous
media systems. Journal of Contaminant Hydrology, 1,407
424.
Leterme, B., Jacques, D. (2013). Literature Review on Mercury
Speciation Soil Systems Under Oxidizing Conditions.
Snowman Network, IMaHG Report, Project No. SN-03/08.
Leterme, B., Blanc, P., & Jacques, D. (2014). A reactive transport
model for mercury fate in soilapplication to different an-
thropogenic pollution sources. Environmental Science and
Pollution Research, 21(21), 1227912293.
Leverett, M. (1941). Capillary behavior in porous solids.
Transactions of the AIME, 142(01), 152169.
Lindley, A. A. (1997). An economic and environmentalanalysis of
the chlor-alkali production process: mercury cells and alter-
native technologies. European Commission, Directorate-
General III Industry, C-4, Chemicals, plastics, rubber,
Brussels, Belgium.
Luo, H., Du, Y., Hu, Z., Cooper, W. L., & Lu, H. (2015). High-
strain rate compressive behavior of glass beads under con-
finement. Experimental Mechanics, 55(5), 935950.
Mayer, A., & Hassanizadeh, S. M. (2005). Soil and groundwater
contamination: nonaqueous phase liquids, principles and
observations. American Geophysical Union.
Mercer, J. W., & Cohen, R. M. (1990). A review of immiscible
fluids in the subsurface: properties, models, characterization
and remediation. Journal of Contaminant Hydrology, 6(2),
107163.
Miller, C. L., Watson, D. B., Lester, B. P., Lowe, K. A., Pierce, E.
M., & Liang, L. (2013). Characterization of soils from an
industrial complex contaminated with elemental mercury.
Environmental Research, 125,2029.
Mualem, Y. (1976). A new model for predicting the hydraulic
conductivity of unsaturated porous media. Wate r R e s ourc e s
Research, 12(3), 513522.
Newsham, K. E., Rushing, J. A., Lasswell, P. M., Cox, J. C.,
Blasingame, T. A. (2004). A comparative study of laboratory
techniques for measuring capillary pressures in tight gas
sands. In SPE Annual Technical Conference and Exhibition.
Society of Petroleum Engineers.
O'Neil, M. J. (2006). The Merck Index: an encyclopedia of
chemicals, drugs, and biologicals. Whitehouse Station, NJ:
Merck and Co. Inc..
Parker, J. C. (1989). Multiphase flow and transport in porous
media. Reviews of Geophysics, 27(3), 311328.
Parker, J. C., Lenhard, R. J., & Kuppusamy, T. (1987). A para-
metric model for constitutive properties governing multi-
phase flow in porous media. Water Resources Research,
23(4), 618624.
Pennell, K. D., Pope, G. A., & Abriola, L. M. (1996). Influence of
viscous and buoyancy forces on the mobilization of residual
tetrachloroethylene during surfactant flushing.
Environmental Science & Technology, 30(4), 13281335.
Pires, L. F., & Pereira, A. B. (2014). Gamma-ray attenuation to
evaluate soil porosity: an analysis of methods. The Scientific
World Journal, 2014, ID 723041.
Pires, L. F., Bacchi, O. O. S., & Reichardt, K. (2005). Soil water
retention curve determined by gamma-ray beam attenuation.
Soil & Tillage Research, 82(2005), 8997.
Pirrone, N., Cinnirella, S., Feng, X., Finkelman, R. B., Friedli, H.
R., Leaner, J., Mason, R., Mukherjee, A. B., Stracher, G. B.,
Streets, D. G., & Telmer, K. (2010). Global mercury emis-
sions to the atmosphere from anthropogenic and natural
sources. Atmospheric Chemistry and Physics, 10(13),
59515964.
Pittman, E. D. (1992). Relationship of porosityand permeability to
various parameters derived from mercury injection-capillary
pressure curves for sandstone. The American Association of
Petroleum Geologist Bulletin, 76(2), 191198.
Poulsen, M. M., & Kueper, B. H. (1992). A field experiment to
study the behavior of tetrachloroethylene in unsaturated po-
rous media. Environmental Science & Technology, 26(5),
889895.
Retsch (2013). Technical Datasheet 0.100.25 mm and 0.20
0.50 mm Glass Beads.
Richard, J. H., Bischoff, C., Ahrens, C. G., & Biester, H. (2016).
Mercury (II) reduction and co-precipitation of metallic mer-
cury on hydrous ferric oxide in contaminated groundwater.
Science of the Total Environment, 539,3644.
Risher, J. F. (2003). Elemental mercury and inorganic mercury
compounds: Human health aspects. Geneva: Concise
InternationalChemicalAssessmentDocument,World
Health Organization.
Rockhold,M. L., Yarwood, R. R., Niemet,M. R., Bottomley, P. J.,
Brockman, F. J., & Selker, J. S. (2007). Visualization and
modeling of the colonization dynamics of a bioluminescent
bacterium in variably saturated, translucent quartz sand.
Advances in Water Resources, 30(6), 15931607.
Ruth, D., Lindsay, C., & Allen, M. (2013). Combining electrical
measurements and mercury porosimetry to predict perme-
ability. Petrophysics, 54(06), 531537.
Scanlon, B.R., Tachovsky, J. A., Reedy, R., Nicot, J.P., Keese, K.,
Slade, R. M., Merwad, V., Howard, M. T., Wells, G. L.,
Mullins, G. J., Ortiz, D. M. (2005). Groundwatersurface
water interactions in Texas. Implications for Water
Resources and Contaminant Transport (TCEQ), Bureau of
Economic Geology, the University of Texas at Austin, 129
131.
Schwille, F. (1988). Dense chlorinated solvents in porous and
fractured mediamodel experiments. Chelsea, MI: Lewis
Publishers.
Smith, J. D., Chatzis, I., Ioannidis, M. A. (2002). A new technique
for measuring the breakthrough capillary pressure. Society of
Core Analysis, 40, 2002.
Sweijen, T., Hartog, N., Marsman, A., & Keijzer, T. J. (2014). The
transport behaviour of elemental mercury DNAPL in saturat-
ed porous media: analysis of field observations and two-
phase flow modelling. Journal of Contaminant Hydrology,
161,2434.
Sznopek, J. L., Goonan, T. G. (2000). The materials flow of
mercury in the economies of the United States and the world.
US Geological Survey, Circular 1197.
Tuck, D. M., Iversen, G. M., & Pirkle, W. A. (2003). Organic dye
effects on dense nonaqueous phase liquids (DNAPL) entry
pressure in water saturated porous media. Wa t er Res o u rces
Research, 39(8).
UNEP. (2002). Global mercury assessment. Geneva: Switzerland.
25 Page 16 of 17 Water Air Soil Pollut (2018) 229:25
US DOE (2001). Mercury Contaminated Material
Decontamination Methods: Investigation and Assessment.
DE-FG21-95EW55094.
van Genuchten, M. T. (1980). A closed form equation for
predicting the hydraulic conductivity of unsaturated soils.
Soil Science Society of American Journal, 44,892898.
Vavra, C. L., Kaldi, J. G., & Sneider, R. M. (1992). Geological
applications of capillary pressure: a review. The American
Association of Petroleum Geologist Bulletin, 76(6), 840850.
Walvoord, M. A., Andraski, B. J., Krabbenhoft, D.P., & Striegl, R.
G. (2008). Transport of elemental mercury in the unsaturated
zone from a waste disposal site in an arid region. Applied
Geochemistry, 23(3), 572583.
Wardlaw, N. C., & Taylor, R. P. (1976). Mercury capillary pressure
curves and the interpretation of pore structure and capillary
behaviour in reservoir rocks. Bulletin of Canadian Petroleum
Geology, 24(2), 225262.
World Chlorine Council (2011). Reduction of mercury emissions
and use from the chlor-alkali sector partnership.
Young, H. D., Freedman, R. A., Ford, A. L. (2012). Sears and
Zemanskys University Physics with Modern Physics.
Addison-Wesley, ISBN 9780321696861.
Water Air Soil Pollut (2018) 229:25 Page 17 of 17 25
... Simulations were performed with GDAn (full details in D'Aniello 2017), a 2D finite element model on unstructured triangular mesh meant for the analysis of groundwater (D'Aniello et al. 2019a(D'Aniello et al. , 2019b, multiphase flow (D'Aniello et al. 2018), and solute transport in porous media at the representative elementary volume (REV) scale. In particular, the WS module was used to compute the groundwater flow field, whereas the NAPL and the ST modules were used for MF and ST simulation sets, respectively. ...
Article
Full-text available
The potential environmental impacts on subsurface water resources induced by unconventional gas production are still under debate. Solving the controversy regarding the potential adverse effects of gas leakages on groundwater resources is therefore crucial. In this work, an interesting real‐world case is presented in order to give further insight into methane multiphase and transport behavior in the shallow subsurface, often disregarded compared to the behavior in the deep subsurface. Multiphase flow and solute transport simulations were performed to assess the vulnerability of an existing shallow unconfined aquifer with respect to a hypothetical methane leakage resulting from a well integrity failure of a former deep geothermal well. The analysis showed that migration of gaseous methane through the aquifer under examination can be extremely fast (of the order of a few minutes), occurring predominantly vertically upwards, close to the well. By contrast, dissolved methane migration is largely affected by the groundwater flow field and occurs over larger time scales (of the order of months/years), covering a greater distance from the well. Overall, the real concern for this site in case of gas leakages is the risk of explosion in the close vicinity of the well. Predicted maximum gaseous fluxes (0.89‐22.60 m3/d) are comparable to those reported for leaking wells, and maximum dissolved methane concentrations may overcome risk mitigation thresholds (7‐10 mg/L) in a few years. Therefore, surface and subsurface monitoring before decommissioning is strongly advised to ensure the safety of the site.
... Despite the worldwide importance and presence of liquid elemental mercury in the subsurface, only recently its DNAPL behaviour in porous media has been addressed. As for more extensively studied DNAPLs, such as creosote, carbon tetrachloride, trichloroethylene (TCE), and tetrachloroethylene (PCE), liquid elemental mercury requires to overcome an entry head to infiltrate in fully water saturated porous media (Devasena and Nambi, 2010), and it has a relatively high capacity to infiltrate as illustrated by numerical simulations (Sweijen et al., 2014;D'Aniello et al., 2018b) and by capillary pressure-saturation, P c (S), experiments under stable flow behaviour (D'Aniello et al., 2018a). However, in contrast with other DNAPLs, liquid elemental mercury was found to behave as a nonwetting phase with respect to both air and water, and therefore it requires to overcome an entry head to infiltrate in partially water saturated porous media (D'Aniello et al., 2018a). ...
Article
Industrial use has led to the presence of liquid elemental mercury (Hg0) worldwide in the subsurface as Dense NonAqueous Phase Liquid (DNAPL), resulting in long lasting sources of contamination, which cause harmful effects on human health and detrimental consequences on ecosystems. However, to date, insight into the infiltration behaviour of elemental mercury DNAPL is lacking. In this study, a two-stage flow container experiment of elemental mercury DNAPL infiltration into a variably water saturated stratified sand is described. During the first stage of the experiment, 16.3 ml of liquid Hg0 infiltrated and distributed into an initially partially water saturated system. Afterwards, during the second stage of the experiment, consisting of the simulation of a “rain event” to assess whether the elemental mercury already infiltrated could be mobilized due to local increases in water saturation, a significant additional infiltration of 4.7 ml of liquid mercury occurred from the remaining DNAPL source. The experiment showed that, under conditions similar to those found in the field, Hg0 DNAPL infiltration is likely to occur via fingers and is strongly controlled by porous medium structure and water saturation. Heterogeneities within the porous medium likely determined preferential pathways for liquid Hg0 infiltration and distribution, as also suggested by dual gamma ray measurements. Overall, this study highlights that the infiltration behaviour of mercury DNAPL is strongly impacted by water saturation. In the field, this may result in a preferential infiltration of Hg0 DNAPL in wetter areas or in its mobilization due to wetting during a rain event, as indicated by this study, or a groundwater table rise. This should be considered when assessing the likely distribution pathways of historic mercury DNAPL contamination as well as the remediation efforts.
Article
Full-text available
The flow of Dense Non-Aqueous Phase Liquid (DNAPL) in highly permeable porous media is characterized by a complex interplay between surface tension, viscous, gravity, and inertia forces. Gravitational effects in these systems have been particularly studied in the context of displacement instability, but little work has focused on the impact of gravitational and inertia forces on the stable displacement of DNAPL in highly permeable but non-fractured porous media. Here, we study the impact of the gravity and Forchheimer numbers on the stable displacement of DNAPL fronts in porous media. We first performed DNAPL injection experiments in bead packings of different sizes for different inlet flow rates and initial saturation. These experiments were accurately modeled using a Darcy-Forchheimer model combined with an Arbitrary Lagrangian-Eulerian tracking of the DNAPL front. Once validated against stable injections in glass beads, the model was then used to perform a broader parametric study than available with our experimental setup. We explored different injection and pumping scenarios over a range of dimensionless numbers. We found that gravity can significantly alter the fluid front in flows commonly found in contaminant hydrology. We estimate, however, that inertia will have a non-negligible impact upon the displacement of DNAPL only during active remediation techniques (e.g., free product pumping method) or pipe rupture events involving low-viscous DNAPL in highly-permeable porous media.
Article
Full-text available
Particular attention is paid to the risk of carbon dioxide (CO2) leakage in geologic carbon sequestration (GCS) operations, as it might lead to the failure of sequestration efforts and to the contamination of underground sources of drinking water. As carbon dioxide would eventually reach shallower formations under its gaseous state, understanding its multiphase flow behavior is essential. To this aim, a hypothetical gaseous leak of carbon dioxide resulting from a well integrity failure of the GCS system in operation at Hellisheiði (CarbFix2) is here modeled. Simulations show that migration of gaseous carbon dioxide is largely affected by formation stratigraphy, intrinsic permeability, and retention properties, whereas the initial groundwater hydraulic gradient (0.0284) has practically no effect. In two different scenarios, about 18.3 and 30.6% of the CO2 that would have been injected by the GCS system for 3 days could be potentially released again into the atmosphere due to a sustained leakage of the same duration. As the gaseous leak occurs, the aquifer experiences high pressure buildups, and the presence of a less conductive layer further magnifies these. Strikingly, the dimensional analysis showed that buoyant and viscous forces can be comparable over time within the predicted gaseous plumes, even far from the leakage source. Local pressure gradients, buoyant, viscous, and capillary forces all play an important role during leakage. Therefore, neglecting one or more of these contributions might lead to a partial prediction of gaseous CO2 flow behavior in the subsurface, giving space to incorrect interpretations and wrong operational choices.
Conference Paper
Full-text available
The Italian Decree no. 152/2006 requires local administrations to identify the subjects that caused contamination of soil and groundwater. This is a compulsory step to enable the characterization and remediation of a contaminated site, and yet, in old industrial areas impacted by multiple sources, it can prove very challenging. The aim of the present study is to propose a stochastic methodology to locate potential sources while taking into account the uncertainties in time (using a transient numerical flow model) and in space (with a forward particle tracking from a site suspected to host buried wastes) in the N-W sector of the Milano Functional Urban Area (FUA). The methodologic probabilistic framework (developed within the AMIIGA project package funded by Interreg Central EUROPE) could be used for a historical reconstruction of releases from contaminated sites (under uncertainties) in order to apply the guiding principle in the European environmental legislation (European Parliament, 2016), the Polluter Pays Principle.
Technical Report
Full-text available
In the present report, the main features of the recently updated version of the GDAn code are presented. In particular, a new module (GDAn-GAS) was implemented to model compressible and temperature dependent two-phase flows of gas and brine in porous media. In order to show the capabilities of the new module, the application of GDAn-GAS to a real-world case of a hypothetical gaseous methane leakage into a shallow aquifer is discussed.
Article
Full-text available
The progressive increase of impervious surfaces induced by urbanization altered significantly the natural hydrological cycle of urban catchments. To face the need of more sustainable and effective solutions for stormwater management and planning, Low Impact Development (LID) practices have been frequently proposed to support existing urban drainage systems. Among LID infrastructures, design and management of stormwater infiltration facilities are still characterised by a high degree of uncertainty. Since the stochastic heterogeneity of the soil may affect significantly their hydraulic performance, it is crucial to understand whether it should be accounted for in the design process. To this aim, numerical experiments under transient variably water saturated conditions were performed. Four infiltration facilities of different bottom length subjected to the same spatial variability of the intrinsic permeability field were considered. Simulations showed that the effects of stochastic heterogeneity on the hydraulic performance are dependent on the dimensions of the facility and on the correlation lengths of the intrinsic permeability field. These effects may potentially undermine the capacity to capture stormwater. To be reduced, the bottom dimensions of the facility should be higher than the horizontal correlation lengths of the intrinsic permeability field. Most strikingly, whether the stochastic heterogeneity is considered or not, the volume infiltrated through the bottom follows an increasing power law with increasing bottom length, while the average infiltration rate at the bottom follows a decaying power law.
Article
Full-text available
Urbanization has led to a substantial change in the hydrological cycle of urban catchments. Increased runoff and urban flooding, decreased direct subsurface infiltration and groundwater recharge, deterioration of water quality are among the major effects of this alteration. To alleviate these effects, Low Impact Development (LID) practices have been frequently adopted for stormwater management. Among LID infrastructures, infiltration facilities are particularly challenging to design and model due to the considerable amount of uncertainties related to the hydrogeological configuration of installation sites. To date, analysis on how soil heterogeneity, groundwater table depth, and thickness of the unsaturated zone affect the hydraulic performance of infiltration facilities are lacking. To address this knowledge gap, a series of numerical experiments under transient variably water saturated conditions were performed for a hypothetical infiltration facility. Numerical simulations showed that i) infiltration rates increase considerably as the initial depth of the groundwater table increases, ii) the contribution of the bottom of the facility to the infiltration of water is generally higher than the sides, iii) the presence of a less conducting soil layer at a short depth from the bottom of the facility reduces infiltration rates dramatically, iv) the complete clogging of the bottom of the facility has a dramatic impact on the hydraulic performance, v) the stochastic heterogeneity of the soil controls the overall stormwater infiltration process through the facility, and the hydraulic performance may largely deviate from the case when heterogeneity is absent.
Article
Mercury is a contaminant of global concern due to its harmful effects on human health and for the detrimental consequences of its release in the environment. Sources of liquid elemental mercury are usually anthropogenic, such as chlor-alkali plants. To date insight into the infiltration behaviour of liquid elemental mercury in the subsurface is lacking, although this is critical for assessing both characterization and remediation approaches for mercury DNAPL contaminated sites. Therefore, in this study the infiltration behaviour of elemental mercury in fully and partially water saturated systems was investigated using column experiments. The properties affecting the constitutive relations governing the infiltration behaviour of liquid Hg⁠0, and PCE for comparison, were determined using P⁠c(S) experiments with different granular porous media (glass beads and sands) for different two- and three-phase configurations. Results showed that, in water saturated porous media, elemental mercury, as PCE, acted as a non-wetting fluid. The required entry head for elemental mercury was higher (from about 5 to 7 times). However, due to the almost tenfold higher density of mercury, the required NAPL entry heads of 6.19 cm and 12.51 cm for mercury to infiltrate were 37.5% to 20.7% lower than for PCE for the same porous media. Although Leverett scaling was able to reproduce the natural tendency of Hg⁠0 to be more prone than PCE to infiltrate in water saturated porous media, it considerably underestimated Hg⁠0 infiltration capacity in comparison with the experimental results. In the partially water saturated system, in contrast with PCE, elemental mercury also acted as a nonwetting fluid, therefore having to overcome an entry head to infiltrate. The required Hg⁠0 entry heads (10.45 and 15.74 cm) were considerably higher (68.9% and 25.8%) than for the water saturated porous systems. Furthermore, in the partially water saturated systems, experiments showed that elemental mercury displaced both air and water, depending on the initial water distribution within the pores. This indicates that the conventional wettability hierarchy, in which the NAPL has an intermediate wetting state between the air and the water phases, is not valid for liquid elemental mercury. Therefore, for future modelling of elemental mercury DNAPL infiltration behaviour in variably water saturated porous media, a different formulation of the governing constitutive relations will be required.
Thesis
Full-text available
This thesis focuses on the characterization of the flow behaviour of elemental mercury (Hg0) DNAPL (Dense NonAqueous Phase Liquid) in porous media. In the subsurface, Hg0 DNAPL can act as a long lasting source of contamination, causing detrimental consequences for the human health and the environment. Therefore, insight into the flow behaviour of elemental mercury in porous media is needed and is critical in assessing the control of contaminant spreading as well as remediation approaches. However, the scientific literature on Hg0 DNAPL is still very limited and, to date, it remains unclear to what extent the validity of the classical constitutive relations, used to describe DNAPLs flow behaviour, as well as the scaling theory, holds for elemental mercury. These issues become crucial in partially water saturated porous media, where liquid Hg0 is likely to behave as a nonwetting phase with respect to both air and water. To address these knowledge gaps, experimental and numerical analysis were performed. In particular, the properties affecting the constitutive relations governing liquid Hg0 infiltration behaviour were explored first, using capillary pressure-saturation, Pc(S), experiments in different granular porous media, and in two- and three-phase fluid systems. Then, the infiltration and (re)distribution behaviour of Hg0 DNAPL was studied in variably water saturated stratified porous media with flow container experiments and dual gamma ray measurements of porosity and fluid saturations. Experimental results indicated that elemental mercury infiltration is strongly controlled by the porous medium water content and can be triggered by its changes due to, for example, rain events. Finally, a new theoretical formulation of elemental mercury retention properties in variably water saturated porous media was proposed and, to assess to what extent numerical modelling can predict elemental mercury migration in porous media, the flow container experiments were simulated using GDAn, the code developed by the Author.
Article
Full-text available
This paper provides an up-to-date assessment of global mercury emissions from anthropogenic and natural sources. On an annual basis, natural sources account for 5207 Mg of mercury released to the global atmosphere, including the contribution from re-emission processes, which are emissions of previously deposited mercury originating from anthropogenic and natural sources, and primary emissions from natural reservoirs. Anthropogenic sources, which include a large number of industrial point sources, are estimated to account for 2320 Mg of mercury emitted annually. The major contributions are from fossil-fuel fired power plants (810 Mg yr<sup>−1</sup>), artisanal small scale gold mining (400 Mg yr<sup>−1</sup>), non-ferrous metals manufacturing (310 Mg yr<sup>−1</sup>), cement production (236 Mg yr<sup>−1</sup>), waste disposal (187 Mg yr<sup>−1</sup>) and caustic soda production (163 Mg yr<sup>−1</sup>). Therefore, our current estimate of global mercury emissions suggests that the overall contribution from natural sources (primary emissions + re-emissions) and anthropogenic sources is nearly 7527 Mg per year, the uncertainty associated with these estimates are related to the typology of emission sources and source regions.
Book
Groundwater Science, 2E, covers groundwater's role in the hydrologic cycle and in water supply, contamination, and construction issues. It is a valuable resource for students and instructors in the geosciences (with focuses in hydrology, hydrogeology, and environmental science), and as a reference work for professional researchers. This interdisciplinary text weaves important methods and applications from the disciplines of physics, chemistry, mathematics, geology, biology, and environmental science, introducing you to the mathematical modeling and contaminant flow of groundwater. New to the Second Edition: New chapter on subsurface heat flow and geothermal systems Expanded content on well construction and design, surface water hydrology, groundwater/ surface water interaction, slug tests, pumping tests, and mounding analysis. Updated discussions of groundwater modeling, calibration, parameter estimation, and uncertainty Free software tools for slug test analysis, pumping test analysis, and aquifer modeling Lists of key terms and chapter contents at the start of each chapter Expanded end-of-chapter problems, including more conceptual questions Two-color figures Homework problems at the end of each chapter and worked examples throughout Companion website with videos of field exploration and contaminant migration experiments, PDF files of USGS reports, and data files for homework problems PowerPoint slides and solution manual for adopting faculty.
Article
From the 1940s through the 1980s, the work of such researchers as Purcell, Leverett, Swanson, and Thomeer provided methods for predicting flow properties, particularly permeability, based on capillary pressure curves measured on rock-mercury-vacuum systems. Although these methods met with considerable success, they have been largely overshadowed, particularly in recent years, by the rapid development of network and direct models. This paper reports on a study of 24 samples from an Offshore Ghana formation of Turonian age. The paper demonstrates that the work of Purcell, in particular, can be used to provide primary predictions of permeability (that is, predictions without any flexible fitting parameters) provided that mercury porosimetry and formation factor measurements are available, either on identical or companion samples. The predictions are good to within a range of factors between 0.5 and 2, with a mean error of less than 35%. This accuracy is remarkable considering that the range of permeability is seven decades and that the data were not collected for the specific purpose of the study. An experimental protocol is suggested that should improve further the already excellent results. The applicability of the method to drill cuttings is discussed. Also, a suggestion for a method of applying the proposed technique to digital rock results is presented. Finally, the paper explores reasons behind the success of the predictions and suggests what types of reservoirs are expected to provide similar successes.