macla nº 20.

julio ‘15
revista de la sociedad española de mineralogía
59

Mixing-Induced Precipitation in Brine

Mining: Reactive Transport Modeling
/ ALEJANDRO GARCÍA-GIL (1), MAR GARCÍA-ALCARAZ (2), ENRIC VAZQUEZ (2), CARLOS AYORA (2*)

(1) Departamento de Geología. Universidad Zaragoza, Pedro Cerbuna, 12. 50009 Zaragoza
(2) Grupo Hidrología Subterránea, UPC-CSIC, Instituto de Diagnóstico Ambiental y Estudios del Agua, Jordi Girona 18, 08034 Barcelona

INTRODUCTION planes (probably stratification planes). It

is 12 to 15 m thick and has high
Continental brines have become the hydraulic conductivities. (II) A less
leading raw material for lithium permeable body of 2 to 8 m thick acting
production worldwide. Lithium-bearing as an aquitard which consist of halite
brines are found in salt aquifers (mainly that may have gypsum levels. (III) Unit B
halite) in the nucleus of salars (salt is a semi-confined aquifermade up of
flats). Because of extremely high halite, 8 to 11 m thick and highly
evaporation rates, the primary discharge conductive. (IV) A lowpermeability body
occurs by evaporation keeping the water is at the bottom of the mined system.
table below the surface.
Two different brines are identified in
Brines are exploited through pumping aquifers “A” and “B” with a transition
wells or ditches excavated in halite brine in the aquitard separating those
aquifers and evaporation of the brine in aquifers. Brine extraction is carried out
fig 2. Gypsum precipitates around in a well.
solar ponds to further concentration. In via brine pumping fields, i.e. horizontal
this type of mining, such as that in drains (1 km length) connected with
REACTIVE TRANSPORT MODELING
Atacama (Chile), a fast and drastic loss vertical extraction well (Fig. 1). Hundreds
in the hydraulic efficiency of the of pumping wells are distributed
pumping wells threatens the feasibility according to a regular grid. Due to A reactive transport modeling of a
of the exploitation. We postulate here consistency of halite rock, no casing is transversal section of the horizontal
that the hydraulic efficiency loss is due used. drain has been used to investigate the
to salt precipitation induced by brine processes leading to salt precipitation in
mixing. The effects of mixing solutions the surroundings of a drain. To solve the
has only been considered to explain multisolute reactive transport RETRASO
carbonate dissolution at the seawater code (Saaltink et al., 1998) has been
intrusion front in coastal aquifers used.
A
(Sanford & Konikow, 1989), and two
explain the deep cavity formation B For each time increment, the porosity is
required to allocate Mississippi Valley- updated according to the molar volume
fig 1. Sketch of the brine exploitation, with two of the precipitated phases. Porosity
type deposits (Corbella et al., 2004). aquifers, A and B, separated by an aquitard. The
area within the dotted line represents the modeled variation modifies permeability in the
The aim of this work is to study the domain. simulations according to
feasibility of the hypothesis of brine Kozeny‟sequation:
mixing and mineral precipitation in the
brine exploitation of the Atacama Salar, During the brine extraction, a drop in the
by means of reactive transport brine level in the well is observed. The
modeling. level drop is caused by an exponential
decrease inhydraulicconductivity and
causes a rapid decrease in the brine where is the initial porosity and
STUDY SITE pumping efficiency. This level drop is the intrinsic permeability for matrix .
satisfactorily modeled assuming an
The Atacama Salar is a closed saline „skin effect‟, i.e., the existence of a low The model domain is 40 x 500 m
basin which forms part of the current permeability skin around the pumping section normal to the drain. Because
inner fore arc of northern Chile. With a well. Indeed, massive precipitation of the symmetry of the problem under
1400 km2 outcrop area, the salt nucleus salts is observed in and around the wells consideration only half of the
constitutes a multilayer halite aquifer. It and drains (Fig. 2). transversal section to the drain is
can be distinguish (Fig. 1): (I) Unit A is an modeled (see modeled section in Fig.
unconfined aquifer mainly composed of 1). The model simulates 90 days of
pure halite with several discontinuity pumping with a pumping rate of 0.01

palabras clave: Salmuera, Mezcla, Precipitación, Yeso, Halita, key words: Brine, Mixing, Precipitation, Gypsum, Halite, Atacama,
Atacama, Permeabilidad, Colmatación, Transporte Reactivo Permeability, Clogging, Reactive Transport
resumen SEM 2015 * corresponding author: [email protected]
 macla nº 20. julio ‘15
revista de la sociedad española de mineralogía
60

L/s in the center of the drain. No flow is

considered through the top and bottom
boundaries, and constant head at the
A
boundary opposite to pumping. Pitzer
ion-ion interaction model has been used
in the geochemical calculations.

RESULTS AND DISCUSSION

According to field observations, the

reactive transport calculations predict a
thin mixing zone and gypsum B -50
-40
precipitation in the aquitard between -30
units A and B (Fig. 3). However, most of -20
-10
gypsum precipitation takes place 0.0
around the drain (Fig. 3). Precipitation is
not symmetrical. There is a zone with a
low rate of gypsum precipitation which
can be called “no reactive channel” (Fig.
C D
3). Halite precipitation is two orders of
magnitude lower than gypsum (not
represented). Porosity decreases 0.05
towards the drain from 5%, as the initial 0.04
porosity, to 0.18 % in the drain wall. 0.03
0.02
Within the “no reactive channel” the 0.01
porosity remains above 2%. 0.005
0.002
Pumping generates hydraulic gradients
forcing the upper brine to flow through fig.3 .A)Zones
Zoneswith
withbrine
brineAA(green),
(green),BB(white)
(white)and
andmixture
mixtureofofboth
both(orange)
(orange)marked
markedby bythe
theconcentration
concentrationofofa
fig 3. A)
the aquitard unit towards the lower a conservative
conservative element.
element. Isopotential
Isopotential linesare
lines areplotted
plottedininthe
thebackground.
background. B)B) Mass
Mass ofof gypsum
gypsum precipitated
precipitated
aquifer (Unit B). This flow fieldproduces (mol/m ). Flow
(mol/m3).
3 Flow lines
lines plotted
plotted in
in the
the background.
background. C) C) Detail
Detail of
of B
B around
around the
the drain.
drain. D)
D) Porosity
Porosity.All
.Allthe
thegraphics
graphics
correspondtotoresults
resultsafter
after90
90days
dayscalculation.
calculation.
a mixing of the brines A and B. The correspond
mixing of two brines in equilibrium with amount of halite precipitates. The mixture has exhausted most of
a salt produces supersaturation and
its precipitation potential and forms a
12 12 12 12

10 10 10 10
x-na x-nh4
x2-ca x2-mg
8 x2-ca 8 x2-mg
68 68

46 46

24 24 2

0 02 0 0

precipitation of the salt. Indeed, Fig. 4

0.0 0.2 0.4 0.6 0.8 1.0 0.0 2.0 4.0 6.0 8.0 0.0 2.0 4.0 6.0 8.0 0.0 2.0 4.0 6.0 8.0

„non-reactive‟ channel.
shows the representation of the gypsum 0.5
equilibrium equation (x*y=K). Although
0.4 As hydraulic conductivity changes with
[SO4] mol/kgw

activities should be represented instead

porosity, after approximately 15 days
molalities, the total salinity is almost 0.3 top and bottom environments of the
constant, and therefore the activity
drain has undergone mineral clogging,
coefficients and the activity of water can 0.2
and very small flow rates cross clogged
be assumed constant and included in K. 0.1 areas. In this setting, ascending and
Any brine equilibrated with gypsum,
descending flow paths are forced to flow
such as A and B, will plot within the 0.0
0.00 0.10 0.15 0.20 towards the “no reactive channel”. The
equilibrium line. Any brine with higher 0.05
[Ca] mol/kgw thickness of this channel progressively
Ca and/or SO4 will be supersaturated. In
fig 4. Ideal mixing of two brines (A and B) that are in wanes, and the efficiency of the well
contrast, any mixture of two brines equilibrium with gypsum. If mixing is non-reactive, exponentially drops.
equilibrated with gypsumwill plot within theresulting mixture will be supersaturated.
the straight segment AB, and therefore Returning to equilibrium requires precipitation.
REFERENCES
in the supersaturation field. Because of
gypsum stoichiometry, precipitation will Close to the drain, descending brine. A
(path 1 of Fig. 3) invades the aquifer Corbella, M., Ayora, C.,Cardellach, E. (2004):
move the mixture to the equilibrium Hydrothermal mixing, carbonate
around a 1:1 slope line. The distance with brine B, high mixing occurs and
massive gypsum (and minor halite) dissolution and sulfide precipitation in
from the conservative line to the Mississippi Valley-type deposits:
equilibrium line determines the mass of precipitates in the area abovethe drain Mineralium Deposita, 39, 344-357.
gypsum precipitated. It is easy to see during the initial 15 days (Fig. 3). Saaltink M.W., Ayora C., Carrera J. (1998): A
that: 1) the more different Ca and SO4 Meanwhile, brine B flows from far mathematical formulation for reactive
concentrations are in the mixing brines, regions through more permeable aquifer ransport that eliminates mineral
the moreprecipitation occurs; and 2) B towards the drain. It mixes with concentrations. Water Resources
descending brine A, which has invaded Research, 3, 1649–1656.
50% mixing results in maximum Sanford, W.E., & Konikow, L.F. (1989):
amount of mass precipitated. The same the drain area, and precipitated gypsum
below the drain (path2 of Fig. 3). In Simulation of calcite dissolution and
discussion can be made for halite. porosity changes in saltwater mixing zones
However, Cl and Na concentration in A between these two paths, brines A and B in coastal aquifers: Water Resources
and B brines are relatively similar, their have been mixed up during their flow to
Research, 25, 655-667.
mixing line is almost coincident with the the drainprecipitating small amounts of
equilibrium curve, and only a small salts along the mixing (path 3 in Fig. 3).