Geothermics 70 (2017) 324–338

Contents lists available at ScienceDirect

Geothermics
journal homepage: www.elsevier.com/locate/geothermics

Causes of abundant calcite scaling in geothermal wells in the Bavarian MARK

Molasse Basin, Southern Germany
⁎
Christoph Wannera, , Florian Eichingerb, Thomas Jahrfeldc, Larryn W. Diamonda
a
Rock-Water Interaction Group, Institute of Geological Sciences, University of Bern, Baltzerstrasse 3, CH-3012 Bern, Switzerland
b
Hydroisotop GmbH, Woelkestrasse 9, D-85301 Schweitenkirchen, Germany
c
Renerco Plan Consult GmbH, Ganghoferstraße 66, D-80339 München, Germany

A R T I C L E I N F O A B S T R A C T

Keywords: The carbonate-dominated Malm aquifer in the Bavarian Molasse Basin in Southern Germany is being widely
Scales exploited and explored for geothermal energy. Despite favorable reservoir conditions, the use of geothermal
Boiling wells for heat and power production is highly challenging. The main difficulty, especially in boreholes > 3000 m
Cavitation deep with temperatures > 120 °C, is that substantial amounts of calcite scales are hindering the proper operation
Centrifugal pump
of the pumps within the wells and of the heat exchangers at the surface. To elucidate the causes of scaling we
Geochemical modeling
present an extensive geochemical dataset from the geothermal plant in Kirchstockach. Based on chemical
Fluid inclusions
Chiller analyses of wellhead water samples, chemical and mineralogical analyses of scales collected along the uppermost
800 m of the production well, and chemical analyses of gas inclusions trapped in calcite-scale crystals, four
processes are evaluated that could promote calcite scaling. These are (i) decompression of the produced fluid
between the reservoir and the wellhead, (ii) corrosion of the casing that drives pH increase and subsequent
calcite solubility decrease, (iii) gas influx from the geothermal reservoir and subsequent stripping of CO2 from
the aqueous fluid, and (iv) boiling within the geothermal well. The effectiveness of the four scenarios was
assessed by performing geochemical speciation calculations using the codes TOUGHREACT and CHILLER, which
explicitly simulate boiling of aqueous fluids (CHILLER) and take into account the pressure dependence of calcite
solubility (TOUGHREACT). The results show that process i causes notable calcite supersaturation but cannot act
as the sole driver for scaling, whereas ii and iii are negligible in the present case. In contrast, process iv is
consistent with all the available observations. That is, scaling is controlled by the exsolution of CO2 upon boiling
at the markedly sub-hydrostatic pressure of 4–6 bar within the production well. This process is confirmed by the
visible presence of gas inclusions in the calcite scales above the downhole pump, where the production fluid
should nominally have been in the homogeneous liquid state. Whereas minor calcite scaling may have been
triggered by fluid decompression within the production well, we conclude that the abundant scaling along the
pump casing is due to cavitation induced by operating the pump at high production rates.

1. Introduction Scaling phases precipitate when they become supersaturated along the
geothermal loop due to gas exsolution or due to a solubility decrease
The precipitation of mineral phases from geothermal fluids (i.e., induced by changes in temperature and/or pressure as well as mixing
scales) may pose a serious economic risk to the successful operation of between aqueous solutions with different chemical compositions
geothermal plants (Thomas and Gudmundsson, 1989). Typical scaling (Bozau et al., 2015; García et al., 2006; Thomas and Gudmundsson,
phases include carbonates (Ámannsson, 1989; Arnórsson, 1989; Benoit, 1989). Typical problems associated with scaling include clogging of
1989; Lindal and Kristmannsdóttir, 1989), amorphous silica and sili- pipes and wells, reduced efficiency of pumps and heat exchangers, re-
cates (Gunnarsson and Arnórsson, 2005 Kristmannsdóttir, 1989; duced reinjection capacity as well as accumulation of hazardous ma-
Zarrouk et al., 2014), sulphates (Regenspurg et al., 2015) as well as terials (e.g., Pb- and Ra-bearing scales) that require costly disposal
mixed metal oxides and sulfides (Gallup, 1989; Regenspurg et al., 2015; (Bozau et al., 2015; García et al., 2006; Scheiber et al., 2013).
Wilson et al., 2007), which are often associated with corrosion of parts Over the past 15 years the Bavarian Molasse Basin in southern
of the geothermal plant (Honegger et al., 1989; Mundhenk et al., 2013). Germany has become a veritable hotspot for geothermal power

⁎
Corresponding author.
E-mail address: [email protected] (C. Wanner).

http://dx.doi.org/10.1016/j.geothermics.2017.05.001
Received 21 November 2016; Received in revised form 27 March 2017; Accepted 2 May 2017
0375-6505/ © 2017 Elsevier Ltd. All rights reserved.
C. Wanner et al. Geothermics 70 (2017) 324–338

Fig. 1. Geographical and geological overview of the studied site. (a)

Relief map showing the location of the geothermal plant at
Kirchstockach. (b) Schematic cross section through the Bavarian
Molasse Basin illustrating that the Malm aquifer dips southward and
that geothermal wells are aimed to intersect permeable fracture zones
(modified from Mayrhofer et al., 2014).

production. Currently 22 geothermal power plants are being operated calcite scaling is hindering the proper operation of the pumps within
(Dussel et al., 2016; Lentsch et al., 2015; Mayrhofer et al., 2014). Most the wells and of the heat exchangers at the surface. Worldwide, the
of them are located within the greater Munich area. The thermal re- formation of calcite scales is mostly attributed to boiling of the geo-
servoir for all these plants is the Upper Malm aquifer of Upper Jurassic thermal fluid or to CO2-stripping via degassing of sparingly soluble
age, which is about 500–600 m thick. This aquifer is dominated by gases that have a lower solubility than CO2 (Arnórsson, 1989; Benoit,
carbonates and has elevated fracture-porosity owing to karstification 1989; Simmons and Christenson, 1994). Thus, the formation of calcite
during the late Cretaceous and early Cenozoic. During the subsequent scales in the plants SE and S of Munich is an unexpected phenomenon,
Alpine orogeny the Malm unit was situated within the evolving because all the plants are being operated at wellhead pressures of
northern foreland depression and hence it was progressively buried by 10–20 bar to exceed the experimentally determined degassing pres-
the erosional products of the uplifting Alps, creating the Molasse Basin. sures, which are in the range of 2–10 bar.
As a consequence, the upper Malm aquifer now dips southward and This study aims at identifying the main processes controlling calcite
reservoir temperatures chiefly depend on the depth of the aquifer at a scaling in geothermal plants in the area SE of Munich in order to aid in
given location (Fig. 1). A maximum reservoir temperature of about optimizing the operation of these plants. To do so we focus on a par-
150 °C is observed to the south where the reservoir depth reaches ticular production period from December 2014 to March 2015 at the
4–5 km. The various geothermal plants produce fluids at rates between Kirchstockach plant. This period was chosen because we have good data
40 and 140 L/s. Most of the plants are producing thermal energy. on the amount of scales formed over this period, on the corresponding
However, SE and S of Munich, where the production temperature production rates and on the evolution of the chemical fluid composition
is > 120 °C, electrical energy is produced as well. Despite the favorable at the wellhead. In addition, we have found liquid and gas inclusions in
reservoir conditions, the successful operation of the plants located SE calcite crystals precipitated from the geothermal fluid during the in-
and S of Munich is highly challenging. In about two thirds of the plants vestigated production period.
producing thermal water with temperatures > 120 °C, substantial

325
C. Wanner et al. Geothermics 70 (2017) 324–338

Fig. 2. Schematic of the production well of the geothermal plant at Kirchstockach, illustrating (a) the vertical well profile including a simplified overview of the stratigraphy; (b) the
detailed well and pump design (projected into the vertical axis); (c) the nominal pressure profile corresponding to the situation where the intake of the pump is placed 800 m below the
surface.

2. Site description the radial bearings. At the beginning of our study period (December 7,
2014) a new riser pipe (pump casing in Fig. 2b) was installed and a
The binary cycle geothermal plant at Kirchstockach is located about pump with a production rate of 80–90 L/s was placed at a depth of
15 km SE of downtown Munich (Fig. 1). It became commercially op- 800 m. After this period the pump and the overlying pump casing
erational in March 2013 and it consists of a geothermal doublet with (Fig. 2b) were removed and replaced.
one production and one injection well, plus the power plant at the Substantial amounts of scales were observed not only in the pump
surface, yielding an installed capacity of 5.5 MWel. Both wells reach but also along the retrieved pump casing of the production well and in
depths of about 3900 m (Fig. 2) and were drilled subvertically through installations of the power plant at the surface, such as within the heat
the top 2.5 km of sediments. At greater depths the wells deviate hor- exchanger and at filter cartridges (Fig. 3). Whereas the casing is coated
izontally to intersect multiple steeply dipping fracture zones (Cacace with scales up to ∼1 mm thick, veritable “scale rocks” with weights on
et al., 2013; Moeck and Mraz, 2015) (Figs. 1 and 2). As a consequence the order of hundreds of kg are formed over time in tubes of the geo-
of the deviations the horizontal distance between the production and thermal plant at the surface (Fig. 3c). These “rocks” are formed by the
injection wells within the reservoir is roughly 2 km, whereas at the accumulation of platy scales that had initially formed on the surface of
surface the two wells are only 20 m from each other. The inner dia- the tubes before being ripped off by turbulence. Such “scale rocks” have
meter of the production well (inside the casing) is 12.25 inches (31 cm) to be removed and filter cartridges have to be replaced regularly to
along the top 2063 m and 8.5 inches (21.6 cm) along the reminder of ensure the proper operation of the geothermal plant.
the well. Along the section within the Malm reservoir no casing is in- The Kirchstockach geothermal plant is being operated at 16–18 bar
stalled (i.e., open hole completion). The plant is being operated using fluid pressure, which is notably higher than the experimentally de-
an electrical submersible pumping system with an 8 m long, multi-stage termined degassing pressure of the wellhead fluid (2–10 bar).
centrifugal pump that is placed in the production well at variable Nevertheless, tiny gas bubbles are observed through a small window at
depths between 450 and 900 m. Typical production rates are between the wellhead. Unfortunately, no device is installed to capture this gas
80 and 130 L/s at a constant temperature of 135 °C. The reservoir phase.
temperature has been measured at 140–141 °C, thus ascent of the fluid The pressure profile along the production well at Kirchstockach is
through the production well results in only 5–6 °C of cooling. Since the shown in Fig. 2c. Beginning in the reservoir at about 300 bar, the
start of the commercial power production in March 2013 until the end pressure falls linearly with decreasing depth according to a hydrostatic
of our study period on March 15, 2015, the pump had to be replaced 7 gradient (Fig. 2c). Without pumping, the hydrostatic head on the
times due to technical problems. In most cases, scaling likely caused the aquifer stabilizes a water table at ∼200 m below the surface. Therefore,
failure of various parts of the pumping system, such as the hydraulics of during our study period, the pump was placed at 800 m depth to boost

326
C. Wanner et al. Geothermics 70 (2017) 324–338

Fig. 3. Photographs of scales formed at (a) the pump, (b) along the pump casing and (c) in installations of the geothermal plant at the surface.

the fluid pressure from ∼50 bar at its intake (where the exact pressure milling the samples in an agate mortar. To identify accessory minerals,
depends on production rate as well as depth) to ∼100 bar at its outlet the carbonate fraction of the powders was first removed by dissolution
(kink in Fig. 2c). Above the pump the pressure falls again according to a in a few mL of a 2N HCl solution. The remaining sample was then
linear hydrostatic gradient, reaching the wellhead at the target pressure suspended in ethanol and loaded onto silica plates. Following eva-
of 16–18 bar. During operation the water table between the riser pipe poration of the ethanol the XRD analyses were repeated. Minerals were
and the well casing (Fig. 2b) lies at 200–250 m depth, depending on the identified from the diffractograms using the Panalytical software “High
production rate. Score Plus”.
Elemental analyses were performed at Hydroisotop GmbH on the 10
3. Methods scale samples for which the mineralogical composition had been de-
termined. The samples were digested in aqua regia (HNO3 + 3HCl) at
3.1. Sampling 60 °C prior to measuring Na+, K+, Mg2+, Ca2+ and SO42+ by ion
chromatography using a coupled DIONEX ICS-1500 system. Total dis-
3.1.1. Scales solved iron was determined using a WTW MultiLab P5 photometer,
The thickness of the scales was measured on each of the 68 casing while the Cu concentration was measured by ICP-MS using a Perkin
tubes when the pump casing (Fig. 2b) was taken out of the well on Elmer Elan 6000 system. The CO32− concentration was calculated as-
March 15, 2015. Because this casing had been free of scales when it was suming that Ca2+ and Mg2+ are exclusively derived from dissolved
first installed in the well in December 2014, these measurements per- carbonate minerals (CaCO3 and MgCO3 components). The analytical
tain to scales precipitated exclusively during the investigated produc- uncertainty is ± 5%.
tion period (December 7, 2014 to March 15, 2015). Prior to its in-
stallation in the well, about 45% of the length of casing had been coated 3.2.2. Analyses of wellhead fluid samples
with a thin plastic film to protect it from corrosion and to test the effects The chemical composition of wellhead waters was determined at
of corrosion on scaling. After measuring their thicknesses, roughly 50 g Hydroisotop GmbH. Concentrations of major cations and anions were
of scales per tube was sampled for analysis. In addition, samples of measured by ion chromatography (IC) using a coupled DIONEX ICS-
scales were retrieved from the intake of the pump and from the first 1500 system. Based on the observation that HCO3− is by far the
stage of the pumping system. dominant base in our wellhead samples, its concentration was de-
termined by alkalinity titration performed on-site during sample col-
3.1.2. Wellhead samples lection, whereas the concentration of CO2(aq) and other dissolved gases
The produced thermal water was regularly sampled at the wellhead. was obtained by quantitative degassing (see below) followed by gas
To do so, the thermal water was cooled down to < 40 °C under op- chromatography using a Shimadzu GC-17A WLD/FID system. The
erational pressure (16–18 bar) in a specially designed device employing analytical uncertainty is ± 5%. Total sulfide concentrations were de-
air cooling at a low water flow rate of ∼10 mL/s. Subsequently the termined by zinc-acetate precipitation followed by ICP-MS analysis.
thermal water was drained under atmospheric pressure to take samples
for analysis of chemical and in-situ physicochemical parameters. The 3.2.3. Analyses of fluid inclusions in calcite
parameters pH, electrical conductivity and redox potential were de- Platy calcite scales ∼1 mm thick were embedded in epoxy resin and
termined on-site using WTW probes. Alkalinity was determined by ti- then cut parallel to the growth direction of the calcite crystals (i.e.
tration also on-site. Additionally, thermal water was drained into perpendicular to the well casing). The resulting ∼1 mm wide strips of
stainless steel pressure vessels under in-situ pressures and temperatures. parallel calcite crystals were then polished on both sides to produce
These samples were then quantitatively degassed in a closed atmo- transparent microscope sections ∼100 μm thick. Petrographic ex-
sphere in the laboratory. The degassing pressure was experimentally amination of their fluid inclusions was performed using an Olympus
determined by slowly releasing pressure under in-situ temperature and BX51 polarizing microscope. Observations were made in normal
observing the formation of bubbles via a specially designed viewing transmitted light and in UV epi-illumination, which allows identifica-
system. tion of fluorescent hydrocarbons.

3.2. Analyses of scales and wellhead fluid samples 3.2.3.1. Microthermometry. Heating experiments were performed on
inclusions consisting of liquid + gas bubbles to determine their
3.2.1. Mineralogical and chemical analyses of scales homogenization temperature (Th), i.e. the equilibrium temperature at
The mineralogical composition of the scales was determined at the which they transform to a homogeneous (single-phase) liquid state. The
University of Bern on 10 of the 68 samples originating from the pump measurements were carried out at the University of Bern using a
casing overlying the pump (Fig. 2) and on the two samples collected Linkam MSD-600 heating-cooling stage mounted on an Olympus BX51
from the pumping system. The main minerals were identified by X-ray microscope. The inclusions were viewed during measurements through
powder diffraction (XRD) using a Panalytical X’Pert Pro diffractometer an Olympus 100x/0.80 LM PlanFI objective lens. The stage was
with Cu-Kα radiation (λ = 1.54598 Å). Powders were prepared by calibrated against the temperature of phase transitions in synthetic

327
C. Wanner et al. Geothermics 70 (2017) 324–338

Table 1
Mineralogical and chemical composition of scales from various locations within the
production well.

Location of Main Accessory mineral Chemical compositiona

scaling mineral phases
formation phase

Pump Calcite Bornite (Cu5FeS4) Ca (wt%): 36.9 ± 0.2

b
Digenite (Cu9S5) CO32− (wt%): 59.6 ± 0.6
Chalcopyrite (CuFeS2) Mg (wt%): 1.66 ± 0.12
Fe (wt%): 0.97 ± 0.48
c
S (wt%): 0.76 ± 0.45
K (wt%): 0.02 ± 0.01
Cu (wt%): 0.10 ± 0.02
Na (wt%): 0.03 ± 0.01
Mg/Ca: (mol ratio) 0.07

Uncoated casing Calcite Magnetite (Fe3O4) Ca (wt%): 37.2 ± 0.3

b
Pyrrhotite (FeS) CO32− (wt%): 59.2 ± 0.4
Pyrite (FeS2) Mg (wt%): 1.32 ± 0.21
Fe (wt%): 1.87 ± 0.62
c
S (wt%): 0.36 ± 0.19
K (wt%): 0.05 ± 0.04
Cu (wt%): < 0.01
Na (wt%): < 0.01
Mg/Ca: (mol ratio) 0.06

Coated casing Calcite – Ca (wt%): 37.8 ± 0.1

Fig. 4. Log(K) of calcite-fluid equilibrium (CaCO3 + H+ = HCO3− + Ca2+) as a func- b
CO32− (wt%): 60.3 ± 0.2
tion of pressure and temperature, illustrating that calcite solubility is strongly dependent Mg (wt%): 1.46 ± 0.03
on temperature and only weakly dependent on pressure. Symbols denote log(K) values Fe (wt%): 0.26 ± 0.19
c
tabulated in the Soltherm.H06 database (Reed and Palandri, 2006), whereas solid lines S (wt%): 0.07 ± 0.02
refer to the interpolation calculated by TOUGHREACT according to Eqs. (1) and (2) as K (wt%): 0.06 ± 0.05
well as the parameters listed in Table S2 (supplementary data). Cu (wt%): < 0.01
Na (wt%): < 0.01
Mg/Ca: (mol ratio) 0.06
fluid inclusions, such that Th values reported herein are accurate to
a
within ± 0.5 °C. The inclusions were photographed before and after Average composition; full analyses are provided in the electronic appendix.
heating to check their phase proportions for signs of possible stretching. Uncertainties refer to the standard deviation (1σ) derived from measurements of multiple
samples.
Replicate measurements were also performed on the same inclusions as b
Calculated from Mg and Ca, since electron microprobe analyses of individual crystals
a further check on stretching. show that all Mg and Ca is derived from Mg-bearing calcite.
c
Calculated from SO42−.

3.2.3.2. Liberation and analysis of gas inclusions. About 7–13 g of scales

were crushed in an evacuated piston-cylinder device to liberate gases
trapped in fluid inclusions (Eichinger et al., 2010; Hämmerli, 2009; temperature effects on mineral solubilities, TOUGHREACT considers
Dublyansky, 2012). Before crushing, the sample chamber was change in equilibrium constants of geochemical reactions as a function
evacuated, flushed twice with He to avoid any air contamination and of pressure via the relation:
filled with He to around 400 mbar. During crushing the equipment was
heated to 150 °C to avoid gas sorption on the freshly crushed mineral P − P0
Log (K )T , P = Log (K )T , P0 − ΔV ⋅
surfaces. After crushing, the device was directly fitted to a gas 2.303⋅R⋅T (1)
chromatograph and the pressure and concentrations of the main where Log(K)T,P and Log(K)T,P0 refer to the decadic logarithms of the
inorganic gas species (CO2, N2, Ar, O2) and hydrocarbons were equilibrium constant K for a given reaction at temperature T (K) and at
measured by GC-WLD and GC-FID (Shimadzu GC-17A), respectively. the pressure P (bar) of reaction and at the reference pressure P0 (vapor-
The volumes of the extracted gas species per gram of sample were saturation pressure of pure water at T); R denotes the gas constant
calculated with respect to the weight of the sample, the gas pressure, (83.14 bar cm3 mol−1 K−1); ΔV (cm3/mol) is the temperature-depen-
the total volume of the piston-cylinder device, and the temperature. Air dent volume change associated with the given reaction (average ΔV
contamination was quantified by the oxygen concentration and over the pressure interval P0 to P), which is computed from regression
measured gas concentrations were corrected accordingly. coefficients aP to eP:

ΔV = aP + bP·T + cP·T2 + dP/T + eP/T2 (2)

3.3. Geochemical modeling
Values of Log(K) and ΔV coefficients for reactions relevant to the che-
3.3.1. Aqueous speciation calculations mical system of interest are listed in the supplementary data. Fig. 4
Aqueous speciation calculations were performed using the software illustrates the effects of pressure and temperature on calcite solubility.
TOUGHREACT V3 (Xu et al., 2014) to compute the saturation state of Owing to the temperature- and pH-dependent speciation of dis-
calcite and dolomite in wellhead water samples. TOUGHREACT is a solved carbon (HCO3− + H+ = CO2(aq) + H2O), the pH value, HCO3−
modeling code that allows simulation of fully coupled multi-phase and and CO2(aq) concentrations at the wellhead (135 °C) vary from those
multi-dimensional reactive transport as well as zero-dimensional spe- measured on-site (20 °C). For the chemical composition of our wellhead
ciation calculations. Its capabilities have been recently tested in an samples both carbon species are relevant, which means that any
extended benchmark study (Steefel et al., 2015 and references therein). changes in pH on the order of the measurement uncertainty ( ± 0.1)
For this study, aqueous speciation was calculated using the tempera- significantly affect the relative concentrations of CO2(aq) and HCO3−.
ture-dependent thermodynamic data in the Soltherm.H06 database Consequently, on-site pH measurements (Table 2) were not used to
(Reed and Palandri, 2006). In addition to taking into account reconstruct the speciation of the wellhead fluid. Instead, to compute

328
 Table 2
C. Wanner et al.

Chemical analyses of wellhead samples collected during October 2014 and September 2015. Also shown are geochemical speciation calculations performed to reconstruct the composition under reservoir conditions as well as to simulate the linear
decompression scenarios.

KST-1 KST-2 KST-3 KST-4 KST-5 KST-6 KST-7 KST-8 KST-9 KST-10 KST-11 KST-12 Reservoir recon- Linear P-drop Linear P-drop
structiona scenarioa + calcite prec.a

Sampling date 16.10. 13.01. 03.02. 03.03. 31.03. 28.04. 26.06. 10.07. 10.09. 15.09. 17.09. 24.09. n.a n.a n.a
2014 2015 2015 2015 2015 2015 2015 2015 2015 2015 2015 2015

In-situ Production rate L/s 91 89.9 87.8 82.2 84.4 76.4 106 128.5 120 120 130 130 n.a. n.a. n.a.
measurements
Temperature °C 135 134 134.0 134.4 133.8 135.6 134.2 135 135.1 135 135 135 140 135 135
Pressure bar 16.4 17.3 17.2 17.1 18 16 18 18 17 19 19 19 300 18 18
Degassing pressure bar 8 10 7 4 5 2 3 4 3.5 2.5 3.5 3.5 n.a. n.a. n.a.
pH at 20 °C – 6.26 6.35 6.29 6.37 6.35 6.42 6.42 6.36 6.38 6.27 6.40 6.30 n.a. n.a. n.a.
pH (wellhead P,T)a – 6.58 6.65 6.61 6.54 6.61 6.68 6.59 6.60 6.66 6.55 6.66 6.67 6.44 6.55 6.52
Eh mV −109 −199 −114 −166 −132 −52 −105 −122 −81 −57 −89 −95 n.a. n.a. n.a.
O2 mg/l < 0,1 < 0,1 < 0,1 < 0,1 < 0,1 < 0,1 < 0,1 < 0,1 < 0,1 < 0,1 < 0,1 < 0,1 n.a. n.a. n.a.

Water analysis Na+ mg/L 113 116 117 117 116 118 115 116 114 118 118 117 117 117 117
K+ mg/L 16.2 16.8 16.9 16.8 17.1 16.7 16.5 16.5 16.5 16.8 16.7 16.9 16.9 16.9 16.9
Ca2+ mg/L 6.1 15.6 16.3 13.6 15.6 10.7 13.6 16.2 15.5 3.9 16.2 18.9 18.9 18.9 13.7
Mg2+ mg/L 1.9 2.1 2.1 2 2 2 2 1.9 1.9 2.1 2 2 2 2 2
a
HCO3− mg/kg 195 226 233 227 228 222 221 228 225 200 228 241 243 242 229
a
CO2(aq) mg/kg 133 126 142 164 141 120 145 146 126 144 128 132 173 174 179
DIC mmol/L 6.25 6.64 7.12 7.51 7.01 6.41 6.98 7.11 6.61 6.58 6.73 7.03 8.0 8.0 7.87
Cl− mg/L 75.3 76.2 77.3 76.8 75.6 73.5 74.8 77.6 74.2 78.3 78.2 74.7 74.7 74.7 74.7

329
F− mg/L 3.2 3.6 3.5 3.1 3.7 3.4 3.8 3.4 n.m. 3.8 4.0 3.8 3.8 3.8 3.8
SO42− mg/L 6.1 6.7 6.1 5.4 7.1 7.9 6.2 5.8 5.9 6.3 6 5.7 5.7 5.7 5.7
TDS mg/L 431.4 470.7 483.7 469.1 477.1 458.6 457.3 474.9 471.0 437.2 484.1 493 487.7 487.5 468.3
HS− + S2− mg/L 6.14 8.2 5.8 7.4 7.3 6.7 n.m n.m 7.48 8.1 8.1 7.8 7.8 7.8 7.8
d
DOC mg/L 2 0.78 1.13 1.4 n.m 1.7 n.m n.m n.m n.m n.m 1.2 n.s. n.s. n.s.
e
SIcalcite – −0.33 0.17 0.16 0.02 0.12 0.05 0.04 0.13 0.18 −0.52 0.20 0.30 0.02 0.18 0.00
SIdolomite – −0.40 0.26 0.23 −0.02 0.15 0.17 0.05 0.13 0.26 −0.61 0.30 0.42 −0.07 0.24 −0.01

Gas analysisb H2 mmol/kg < 0.01 0.01 0.01 0.01 0.01 < 0.01 < 0.01 0.01 < 0.01 n.m. n.m. n.m. n.s. n.s. n.s.
Ar mmol/kg 0.01 0.02 0.02 0.02 0.03 0.01 0.01 0.01 0.01 0.01 0.02 0.02 n.s. n.s. n.s.
N2 (124)c mmol/kg 0.55 0.83 0.88 0.78 0.94 0.30 0.64 0.81 0.41 0.39 0.50 0.36 n.s. n.s. n.s.
CO2 (1370)c mmol/kg 2.70 2.74 2.97 3.31 2.91 2.51 2.83 3.06 2.51 2.93 2.58 2.87 n.s. n.s. n.s.
Methane (CH4) mmol/kg 1.02 1.71 1.76 1.65 1.84 0.83 1.12 1.54 1.01 0.99 1.24 1.04 n.s. n.s. n.s.
(191)c
H2S mmol/kg 0.19 0.26 0.18 0.23 n.m. n.m. n.m n.m n.m n.m n.m n.m n.s. n.s. n.s.
Ethane (C2H6) mmol/kg 0.01 0.03 0.02 0.02 0.03 0.01 0.02 0.02 0.01 0.01 0.02 0.01 n.s. n.s. n.s.
Propane (C3H8) mmol/kg 0.01 0.01 0.01 0.01 0.01 < 0.01 0.01 0.01 n.m n.m n.m < 0.01 n.s. n.s. n.s.
Butane (C4H10) mmol/kg < 0.01 0.01 0.01 0.00 0.01 < 0.01 < 0.01 0.01 < 0.01 n.m n.m < 0.01 n.s. n.s. n.s.
Total gas content mmol/kg 4.31 5.35 5.69 5.80 5.76 3.67 4.64 5.45 3.97 4.33 4.38 4.31 n.s. n.s. n.s.

n.m.: not measured; n.s.: not simulated, n.a.: not applicable.

a
Obtained from geochemical modeling.
b
Concentrations refer to gases dissolved in wellhead samples.
c
Equilibrium concentration at reservoir P,T conditions (300 bar, 140 °C) in the presence of a corresponding gas phase.
d
Dissolved organic carbon (DOC).
e
Corresponds to the saturation index of the proposed solid solution (Ca0.94Mg0.06CO3).
Geothermics 70 (2017) 324–338
C. Wanner et al. Geothermics 70 (2017) 324–338

(i.e., pump, coated and uncoated casing). However, there are differ-
ences with respect to accessory minerals (Table 1). In the pump scales,
the Cu-sulfides bornite (Cu5FeS4), digenite (Cu9S5) and chalcopyrite
(CuFeS2) were identified, which presumably are products of partial
corrosion of the bronze (i.e., Cu-bearing) radial bearings of the pump.
Particularly interesting is the identification of bornite because it can
occur as two stable polymorphs: (i) as a low-temperature (< 200 °C)
orthorhombic phase and (ii) as a high-temperature (> 265 °C) cubic
phase (Grguric et al., 1998). In our case, the structure of the identified
bornite is clearly orthorhombic (see diffractograms provided in the
supplementary data), although the presence of some of the cubic
polymorph cannot be excluded because the diffraction planes of the
cubic phase are also present in the structure of the orthorhombic
polymorph. In the scales precipitated along the uncoated tubes, the Fe
sulfides pyrite (FeS2) and pyrrhotite (FeS) as well as the iron oxide
magnetite (Fe3O4) were identified. Judging from the black undersides
of the platy scales it is clear that these phases occur only on the surface
in contact with the casing. It is thus likely that their formation is due to
corrosion of the Fe-bearing casing, which occurs only during the initial
stage of scaling. The occurrence of magnetite and pyrrhotite explains
Fig. 5. Depth-dependent thickness of calcite scales along the pump casing. The scales
the magnetic property of the scales collected from the uncoated casing.
precipitated during the investigated operational period from December 2014 to March
2015. Bars denote the thickness of scaling on each of the 68 casing tubes.
In contrast, calcite was the only mineral identified in scales precipitated
on the coated casing, demonstrating that the coating successfully pro-
tects the casing from corrosion.
saturation states of calcite and dolomite at the wellhead accurately, the The mineralogy of the scales is also reflected by their elemental
pH value and the HCO3− and CO2(aq) concentrations were simulta- compositions (Table 1). Calcium is the most abundant element in all the
neously reconstructed by speciation calculations performed at wellhead scales, in accord with the observed predominance of calcite. The con-
conditions (135 °C, 18 bar). To do so, dissolved inorganic carbon (DIC) stant Mg/Ca ratio (0.06–0.07 on a molar basis) of the calcite was
concentrations were fixed to the sum of CO2(aq) and HCO3− con- confirmed by electron microprobe analyses of individual crystals. This
centrations obtained from the concentration measurements of dissolved demonstrates that the scales are in fact a CaCO3–MgCO3 solid solution
gases and from the on-site alkalinity titration, respectively, while ap- Calcite
(i.e., Ca0.93-0.94Mg0.06-0.07CO3) with an ideal MgCO3 activity, aMgCO3 , of
plying a charge-balance constraint on pH. As HCO3− is the dominant 0.06–0.07, rather than a mixture between calcite and dolomite. No
anion, charge balance was effectively obtained by varying the HCO3− distinct variation in the chemical composition with depth is observed
concentrations as a function of pH. along the 800 m long casing. However, scales from the uncoated casing
show significant Fe and S contents reflecting the identified Fe-sulfides
3.3.2. Boiling calculations and Fe-oxides, whereas Fe and S concentrations are essentially zero in
Geochemical effects of degassing possibly occurring within the scales from the coated casing. Scales from the pump show detectable Cu
geothermal loop were assessed by performing boiling calculations using concentrations and the highest of all S concentrations, consistent with
the computer code CHILLER (Reed and Spycher, 2006). In addition to the identified Cu-sulphides (bornite, digenite, chalcopyrite).
the cited CHILLER manual, details of the numerical approach can be The fact that the sampled scales formed exclusively during the in-
found in Reed (1997, 1998). Unlike TOUGHREACT, CHILLER allows vestigated production period (December 2014–March 2015) provides
computation of the equilibrium distribution of chemical species be- an opportunity to quantify the amount of Ca and CO2 that precipitated
tween aqueous, gas and mineral phases rather than just between aqu- along the riser pipe and to determine the corresponding average calcite
eous and solid phases. Similarly to speciation calculations using deposition rate. This requires determining the total mass of scalings by
TOUGHREACT, it calculates aqueous activity coefficients using the integrating their thickness over the entire pump casing (Fig. 5) and by
extended Debye–Hückel approach (Helgeson et al., 1981), it uses gas taking into account production rates over the investigated operational
fugacities as discussed by Spycher and Reed (1988) and it employs the period (supplementary data). Treating the precipitates as pure calcite
same thermodynamic database (i.e., Soltherm.H06). results in total, time-integrated losses from the production fluid of
0.24 mg/L Ca and 0.27 mg/L CO2, which yields an average calcite de-
4. Results position rate of 7.12 × 10−13 mol/L/s.

4.1. Scale samples 4.2. Wellhead water samples

The thickness of scales precipitated along the pump casing during Concentrations of aqueous species and dissolved gases measured on
the 99-day operational period varies from 0.3 to 1.1 mm (Fig. 5). In wellhead samples collected before, during and after the investigated
general, the thickness is slightly lower along the sections of casing production period are listed in Table 2. Also listed are reconstructed
coated by plastic. The outer surfaces of the platy scales (i.e. the surfaces wellhead pH values and computed calcite and dolomite saturation in-
originally in contact with the thermal water) are rough, being made up dices (SI) at wellhead conditions (T = 135 °C, P = 18 bar). Although its
of the euhedral terminations of tiny calcite crystals (∼150 μm diam.). effect is only minor (∼0.01), the presence of MgCO3 in the calcite solid
The inner surfaces of platy scales originally in direct contact with the solution has been taken into account in the listed SIcalcite values, cal-
plastic coating of the casing are very smooth, white and non-magnetic, culated according to Reed (1992):
whereas those originally in contact with the uncoated casing tubes are SICa0.94 Mg0.06 CO3 = log (0.094⋅10 SIcalcite + 0.06⋅10 SIMagnesite) (3)
rough, dark grey and magnetic. The thickness of scales precipitated at
the intake of the pump and on the impeller of the first stage of the pump The thermal water can be described as a weakly mineralized
is also on the order of 1 mm. Na–HCO3 type, which is typical for thermal water occurring in the
The mineralogy is dominated by calcite in all three scaling types Upper Malm aquifer within the Bavarian Molasse Basin. Interestingly,

330
C. Wanner et al. Geothermics 70 (2017) 324–338

the saturation state of wellhead samples with respect to calcite and UV epi-illumination, (iii) rarer light yellow-brown petroleum inclusions
dolomite is variable. In most cases, wellhead samples are super- containing ∼80 vol% hydrocarbon liquid + ∼20 vol% immiscible gas
saturated with respect to calcite and dolomite (SI > 0). However, and that exhibit yellow-orange fluorescence under UV epi-illumination;
three samples were undersaturated with respect to both carbonate (iv) inclusions that are mixtures of types i and ii, i.e. with 10–30 vol%
phases (SI < 0). The saturation indices of calcite and dolomite reflect blue-fluorescing liquid + 70–90 vol% immiscible gas bubbles. Types i
the variable concentrations of the pertinent dissolved species, as illu- and ii are the most abundant. All four types of inclusions are distributed
strated in Fig. 6a–c by the linear correlations between the calcite sa- throughout the crystals, indicating that they were trapped simulta-
turation index and the amount of total dissolved solids (TDS), between neously during crystal growth, i.e. they are all primary inclusions. No
the Ca2+ concentration and TDS, as well as between the HCO3− con- healed fractures were found (which could possibly host secondary fluid
centration and TDS. In contrast, very little variation with changing TDS inclusions) and no aqueous fluid inclusions were found, presumably
is observed for other major species, such as Na+ and Cl− (Fig. 6d). The owing to the slow growth rate of calcite. In contrast to scales from the
total dissolved gas content varies between 3.67 and 5.80 mmol/kg casing, calcite crystals precipitated at the first stage of the pumping
(Table 2) and it mainly consists of CO2, CH4, and N2. A significant H2S system host only petroleum inclusions showing blue or brownish
concentration was also observed whereas only trace amounts of H2, fluorescence under UV epi-illumination, whereas single-phase gas in-
ethane, propane and butane were measured. clusions are absent. Moreover, the calcite crystals precipitated at the
pump contain smaller and far fewer inclusions than crystals pre-
4.2.1. Reconstruction of reservoir conditions cipitated along the casing.
The fact that calcite and dolomite dissolution rates are rather fast The occurrence of gas inclusions has important implications because
(Palandri and Kharaka, 2004 and references therein) implies that che- it demonstrates that a free gas phase is present within the production
mical equilibrium is likely to prevail in the Malm aquifer. Therefore, the well. Further, the occurrence of petroleum inclusions implies that im-
observation that in most cases wellhead fluid samples are not in equi- miscible oil droplets are present as well. The type (iv) inclusions con-
librium with respect to calcite and dolomite (Table 2, Fig. 6a) suggests sisting of mixtures of variable amounts of gas and oil are expected when
that the wellhead samples do not reflect reservoir conditions. Moreover, both end-member phases coexist during inclusion entrapment
the occurrence of calcite scalings along the production well demon- (Diamond, 2003a). It is likely that these inclusions derived from a
strates that some of the Ca2+ and HCO3− dissolved under reservoir minor influx of oil from petroleum pockets in the Jurassic carbonates
conditions precipitated from the upflowing thermal water. The forma- (e.g., Malm) and/or overlying Cenozoic units of the Bavarian Molasse
tion of calcite scalings is also reflected by the correlations observed Basin (Mayrhofer et al., 2014; Veron, 2005).
between the calcite saturation index and the dissolved calcite species In summary, the simultaneous presence of gas inclusions and pet-
(Fig. 6a–c), suggesting that low concentrations of Ca2+ and HCO3− at roleum inclusions shows that the produced fluid at Kirkstockach is ac-
the wellhead are due to greater precipitation of scales at depth. tually a 3-phase mixture of mutually saturated water + gas + oil. The
Accordingly, speciation calculations using TOUGHREACT were total fraction of free gas is likely to be low, as gas bubbles are seen only
performed to reconstruct the chemical composition at reservoir condi- occasionally in the viewing port at the wellhead. Similarly, the oil
tions (140 °C, 300 bar). To do so the Ca2+ concentration was fixed at fraction must also be low, as minor petroleum films are only seldom
the maximum value observed at the wellhead (KST-12: 18.9 mg/L, observed during fluid sampling at the wellhead.
Table 2) because this is likely the sample that experienced the least
amount of calcite precipitation within the production well. Moreover, 4.3.2. Microthermometry
the concentration of total dissolved inorganic carbon (DIC = H- No phase transitions were observable in the type-i gas inclusions.
CO3− + CO2(aq)) was numerically increased until calcite and dolomite Only type-ii and -iii petroleum inclusions (with ∼20 vol.% gas bubbles)
reached nearly perfect saturation (SI ≈ 0). The corresponding reservoir were measured. All undergo bubble-point transitions, i.e. the gas
pH was reconstructed by the total ionisable hydrogen ion balance bubble shrank progressively upon heating and dissappeared at the
(Truesdell and Singers, 1974), which means that the total H+ con- temperature recorded as the homogenizaton temperature (Th), where-
centration was fixed to the sum of the value calculated for KST-12 and upon the inclusions consisted entirely of hydrocarbon liquid. None of
the amount of DIC added to the system. Applying this procedure yielded these inclusions showed a change in absolute size of their gas bubbles
a reservoir pH of 6.44, which is slightly lower than the values re- after their first cycle of heating to Th and cooling back to room tem-
constructed for the wellhead samples (6.54–6.68), and a reservoir DIC perature. All the homogenizaton temperatures (Th) were reproducible
of 8 mmol/kgH2O (Table 2), which is 0.5–1.8 mmol/kgH2O higher than on individual inclusions. These two findings rule out stretching of the
DIC concentrations measured in wellhead samples. The calculated in- inclusions despite the ∼30–90 bar rise in internal pressure upon
crease in DIC in KST-12 is due largely to an increase in CO2(aq) heating from room temperature to Th. It follows that the Th results can
(≈41 mg/kg CO2(aq)), whereas the HCO3− concentration remains es- be considered representative of the properties of the production fluid.
sentially unchanged, demonstrating that reservoir conditions (i.e., cal- The Th of inclusions within calcite crystals precipitated along the
cite and dolomite equilibrium) can be obtained by solely adding CO2 first stage of the pumping system (808 m depth), along casing tube #62
that was lost due to degassing during upflow. (717 m depth) and along tube #30 (341 m depth) are 128.2 ± 6.4 °C
(n = 3), 130.7 ± 7.8 °C (n = 8) and 137.6 ± 1.5 °C (n = 5), respec-
4.3. Fluid inclusions tively. No systematic difference in Th was observed between blue- and
brown-fluorescing petroleum inclusions.
4.3.1. Petrography The facts that the four inclusion types are coeval, they were trapped
Fluid inclusions were found in calcite crystals in scales collected in a heterogeneous (i.e. multi-phase) state and their host crystals be-
from coated and uncoated casing tubes as well as from the pump. A have rigidly (isochorically) during microthermometry, lead to
detailed petrographic investigation was performed on scales collected straightforward interpretations of the Th values (see Diamond, 2003a,b
from casing tubes #30 (coated, 341 m depth) and #62 (uncoated, for the principles). A first case is the scale sample at tube #30 (∼460 m
717 m depth) as well as from the first stage of the pumping system above the pump), in which petroleum was clearly saturated with re-
(808 m depth). Four types of inclusions were observed at room tem- spect to a free gas phase during entrapment. It follows that the mea-
perature in scales from the two casing types (Fig. 7): (i) colorless, sured Th values (137.6 ± 1.5 °C) must correspond directly to the
single-phase gas inclusions that do not fluoresce under UV epi-illumi- trapping temperature of the inclusions. Indeed, as expected from
nation, (ii) colorless petroleum inclusions that consist of ∼80 vol% li- theory, the results precisely match the 137 °C temperature of the pro-
quid + ∼20 vol% immiscible gas and that fluoresce light blue under duction fluid inferred for tube #30 when taking into account the minor

331
C. Wanner et al. Geothermics 70 (2017) 324–338

Fig. 6. Variation in compositions of wellhead fluids collected between October 2014 and September 2015. SI: saturation index. TDS: total dissolved solids. (a–c) Linear correlations
between the calcite saturation index and TDS, Ca2+ and TDS, and HCO3− and TDS. (d) Demonstration that other major species (Na+, Cl−) do not contribute to the observed TDS
variation.

temperature decrease occurring in the production well during upflow. 4.3.3. Crushing and gas analyses
The second case is the scale sample from the pump, in which no gas The composition of gas inclusions liberated from fluid inclusions
inclusions were found. Theory dictates that the measured Th values during crushing of calcite scales is fairly constant (Table 3). Gases lib-
( 130.7 ± 7.8°C) must be lower than the trapping temperature of the erated from all four samples mainly consist of N2 and minor CO2. Me-
inclusions. While this expected behavior is confirmed by the results thane and other hydrocarbon gases are present only in trace amounts.
(Tfluid at the pump is estimated at 137.6 °C), the small difference be- The N2/CO2 ratio varies from 2.4 to 6.9, but no clear trend with sam-
tween Th and Ttrapping (3–16 °C) suggests the petroleum was close to pling depth is observed. It should be noted that H2O gas was not ana-
saturation with respect to a free gas phase. The third case is the sample lyzed and that the values listed in Table 3 would have to be adjusted to
from tube #62, some 80 m above the pump. Here gas inclusions are lower values if it were actually present in the inclusions. The N2/CO2
present and the highest Th values match the temperature of the pro- ratio, however, would remain the same. Since N2 gas can be liberated
duction fluid precisely, but other Th values are up to 15 °C lower, by the degradation of organic matter (Littke et al., 1995), the observed
suggesting slight undersaturation with respect to gas, as observed at the high N2 gas concentrations likely originate from the same hydrocarbon
pump. These results suggest increasing gas saturation with height above source as the petroleum inclusions (Fig. 7).
the pump and they rule out the possibility that the production fluid was
overheated near the pump.

332
C. Wanner et al. Geothermics 70 (2017) 324–338

Fig. 7. Examples of primary gas and petroleum in-

clusions observed in calcite crystals that precipitated
along tube #30. Microphotographs (a) and (c) taken
under normal transmitted light. Microphotographs
(b), (d) and (e) taken under UV reflected light. (b)
and (d) show the same fields of view as (a) and (c),
demonstrating that the petroleum inclusions are
fluorescent, whereas the gas inclusions are not. (e)
Petroleum inclusions showing both orange and blue
fluorescence.

5. Discussion temperature is decreasing from about 140 to 135 °C (Table 2, “Linear P-

drop scenario”). This calculation demonstrates that, despite the rather
5.1. Scenarios for scale formation low sensitivity of calcite solubility to pressure variations (Fig. 4) and
the minor temperature drop occurring in the production well, decom-
The presence of Fe corrosion phases as well as CO2-bearing gas in- pression of the reservoir fluid may yield notable calcite supersaturation
clusions demonstrate that corrosion of the uncoated casing and CO2 at the wellhead. However, the increase in SIcalcite associated with de-
degassing of the thermal water are occurring in the production well. compression is significantly lower than the maximum value computed
Eqs. (4) and (5) illustrate that both processes are associated with a pH for the wellhead samples (SIcalcite = 0.30, Table 2). Moreover, calcite
increase due to the consumption of protons (i.e., H+): precipitation induced by decompression would reset the saturation
index to zero, but this conflicts with the observation that certain
Fe0(s) + 2H+ → Fe2+ + H2 (4)
wellhead samples are undersaturated with respect to calcite (Table 2).
HCO3− + H → CO2(g) + H2O
+
(5) Finally, the linear decompression scenario does not consider the pre-
sence of a free gas phase in the production well, which is clearly de-
Since calcite solubility decreases with increasing pH, both corrosion monstrated by our fluid inclusion evidence (Fig. 7). This means that
and CO2 degassing promote the precipitation of calcite in the produc- calcite precipitation constitutes the only DIC sink in this scenario. Nu-
tion well. In addition, calcite precipitation is favored by decompression merically fixing the calcite saturation index to zero by removing equal
during upflow in the well (Figs. 2 and 4). Based on these processes amounts of Ca2+ and HCO3− during decompression decreases the Ca2+
identified to promote calcite precipitation (i.e., pressure drop, corro- and DIC concentrations in the production well by 0.13 mmol/kgH2O due
sion, CO2 degassing) we propose 4 scenarios for the formation of the to calcite precipitation (Table 2, “linear P-drop + calcite prec.”). The
calcite-dominated scales (Fig. 8): decrease in DIC accounts for only 13 mol% of the amount that had to be
numerically added to reconstruct the water composition at reservoir
• Linear decompression scenario: the precipitation of calcite is con- conditions. It follows that another process controlled the DIC content
trolled by the solubility decrease associated with the pressure drop and that decompression between the reservoir and the wellhead cannot
of ca. 280 bar occurring when the thermal water is pumped from the have been the sole or even main driver for scaling within the production
Malm aquifer to the surface. well.
• Corrosion scenario: the precipitation of calcite is controlled by the
corrosion of the casing and the associated increase in pH and de-
crease in calcite solubility. 5.2.2. Corrosion scenario
• Gas influx scenario: the precipitation of calcite is controlled by an The proportion of typical corrosion products such as Fe oxides and
Fe sulfides in scales sampled along the production well is rather low
influx of a free gas phase into the production well, which based on
gas analyses of crushed fluid inclusions is dominated by N2. As a (Table 1). Most importantly, these corrosion phases occur only at the
consequence, CO2 dissolved in the thermal water may be stripped interface to the uncoated casing, implying that corrosion accompanies
into the gas phase, inducing a pH increase (Eq. (5)) and hence a only the initial stage of calcite scaling. Once the casing is covered (i.e.,
decrease in calcite solubility.
• Gas exsolution scenario: the precipitation of calcite is controlled by Table 3
Chemical composition of gases liberated from liquid and gas inclusions during crushing of
boiling of the produced thermal water within the pump at 800 m
calcite scales collected from various casing tubes along the production well.
depth within the production well, due to a major, non-hydro-
statically controlled pressure drop or to a major temperature in- a
Gas Casing #3 Casing #7 Casing #32 Casing #60
crease. (z = −35 m) (z = −82 m) (z = −365 m) (z = −694 m)

N2 Vol% 73.6 70.1 87.4 82.4

CO2 Vol% 25.5 29.6 12.6 16.8
5.2. Assessing scenarios of scale formation CH4 Vol% 0.60 0.23 0.31 0.2
N2/CO2 – 2.89 2.37 6.91 4.89
b
5.2.1. Linear decompression scenario Sum 99.7 99.8 99.9 99.5
Aqueous speciation calculations using the reconstructed reservoir a
Vapor (i.e, H2O gas) was not determined and does not appear in the analyses.
fluid yield a change in the calcite saturation index of 0.16 when the b
Difference to 100% corresponds to minor amounts of methane, ethane, propane,
pressure drops from 300 to 18 bar in the production well while the butane, and pentane.

333
C. Wanner et al. Geothermics 70 (2017) 324–338

Fig. 8. Scenarios of the formation of calcite scales: (a) linear decompression scenario in which calcite precipitation occurs only due to the solubility decrease associated with the linear
pressure decrease of ca. 280 bar as the thermal water is pumped from the Malm aquifer to the surface; (b) Corrosion scenario in which calcite precipitation is induced by corrosion of the
casing, which increases pH and hence lowers calcite solubility; (c) Gas influx scenario in which calcite precipitation is controlled by influx of an N2-rich gas phase into the production well,
which subsequently strips CO2 from the thermal water; (d) Gas exsolution scenario in which calcite precipitation is caused by gas exsolution from the produced thermal water due to a
localized pressure drop, most likely located within the centrifugal pump.

passivated) by calcite, corrosion is slowed down because the casing is equilibrium concentration under reservoir P–T conditions (300 bar,
no longer in contact with the upflowing thermal water. Moreover, 140 °C). Such an unusual, localized plumbing configuration could per-
scaling occurs along the coated tubes as well, which are unaffected or haps be imagined for the Kirchstockach site, but the fact that substantial
only slightly affected by corrosion (Table 1). Similarly to the linear scaling is observed in two thirds of the geothermal plants SE of Munich
decompression scenario, the corrosion scenario does not consider the demonstrates that the phenomenon is not just local. Therefore, we
presence of a free gas phase in the production well, as inferred from consider influx of a disequilibrium gas an unlikely scenario for scaling.
fluid inclusion analyses (Fig. 7). In summary, corrosion is unlikely to
have been the main driver for scaling within the production well. 5.2.4. Gas exsolution scenario
Similar to the gas influx scenario, the gas exsolution scenario is
5.2.3. Gas influx scenario consistent with the presence of gas inclusions within calcite crystals
In contrast to the two previous scenarios, this one considers the (Fig. 7). The difference is that this scenario considers the formation of
presence of a free gas phase in the upflowing thermal water and it is the free gas phase within the production well itself (Fig. 8d) and not
thus consistent with the presence of gas inclusions within calcite crys- within the geothermal reservoir. Boiling could in principle be induced
tals (Fig. 7), as well as with the need to numerically add CO2 in order to by overheating of the pump at 800 m depth prior to scale formation,
reconstruct the composition of the reservoir fluid. If N2, which con- although the water would have to reach about 250 °C at that depth to
stitutes the major component of the gas inclusions (Table 3), entered intersect the gas-saturation curve, which seems unlikely at flow rates of
the production well as a free gas phase (Fig. 8c), it would strip dissolved ∼2 m/s. In fact, significant heating of the water seems to be ruled out
CO2 from the thermal water as it approaches gas–water chemical because the gas-saturated fluid inclusions at the pump depth homo-
equilibrium. Since CO2 stripping is associated with a pH increase (Eq. genize below 138 °C, and because the orthorhombic polymorph of
(5)), it would immediately lower the solubility of calcite in the up- bornite, which is stable only below 200 °C, occurs as a corrosion pro-
flowing thermal water. Accordingly, the influx of a free gas phase into duct on the pump itself.
the base of the production well could cause significant calcite super- This leaves only a major pressure drop, e.g. due to fast operation of
saturation and possibly drive scaling. The problem with this scenario, the pump impeller, as a conceivable cause of gas exsolution. To assess
however, is that the concentration of major gases dissolved in wellhead whether such a pressure drop is a feasible scenario, boiling calculations
samples (CO2, CH4, N2) are several orders of magnitude lower than the were performed using CHILLER. Simulations were run using the re-
equilibrium concentration at the reservoir pressure of 300 bar constructed reservoir fluid (Table 2) as initial composition and by
(Table 2). This implies that the gas influx scenario is only possible if an continuously lowering the pressure at a constant temperature of 137 °C,
N2–CO2-rich gas phase somehow ingressed from a separate reservoir corresponding to the inferred trapping temperature of the gas inclu-
that is hydrologically decoupled from the Malm aquifer, with partial sions, until the saturation pressure of pure water was reached (3.3 bar).
mixing of the gas and water first occurring near to or within the pro- In addition, based on the finding that the gas inclusions have high N2 to
duction well. If there were a hydrological connection between the CO2 ratios (Table 3), simulations were run for variable initial N2 con-
postulated gas reservoir and the Malm aquifer, the dissolved con- centrations in order to reconstruct the dissolved N2 concentration prior
centration under reservoir conditions would correspond to the to gas exsolution. The behavior of methane and other gases was not

334
C. Wanner et al. Geothermics 70 (2017) 324–338

simulated because they are present only in traces in the gas inclusions. been described for similar liquid-dominated geothermal systems but
at shallower pump installation depths than that at the Kirchstockach
5.2.4.1. Simulated boiling of the reservoir fluid in the production geothermal site (Aksoy, 2007; Aksoy et al., 2008).
well. Fig. 9 illustrates that the composition of the exsolved gas as 4. Perturbation of the saturation states of calcite and dolomite. Upon
well as the remaining liquid phase depends on the dissolved N2 exsolution of N2 and steam at the pump, aqueous CO2 is partially
concentration at reservoir conditions. If the reservoir N2 stripped into the gas phase (Fig. 9g–i), immediately causing an in-
concentration is equal to the maximum N2 concentrations measured crease in calcite and dolomite supersaturation (SI > > 0, Fig. 9d).
at the wellhead (1 mmol/kgH2O, Table 2), degassing starts at about This triggers calcite precipitation or alternatively accelerates on-
5.5 bar and the gas phase consists of mainly steam (H2O) and minor going calcite precipitation, whereas dolomite fails to precipitate
amounts of N2 and CO2. In contrast, if the N2 concentration in the owing to kinetic limitations (see rate compilation by Palandri and
reservoir is 3 or 5 times larger, degassing starts at 9 or 14 bar, Kharaka, 2004).
respectively, and the gas phase is initially dominated by N2. With 5. Persistence of calcite supersaturation and subsequent scaling at the
progressive degassing (i.e., decreasing pressure), the steam fraction pump, along the casing and within the geothermal plant at the
increases while the N2 fraction decreases. In contrast, the CO2 fraction surface. Scales form along the entire pump casing (Fig. 5) because
of the gas phase shows only minor variations at pressures > 4 bar, the observed calcite precipitation rate along the uppermost 800 m of
which means that CO2 is continuously stripped into the gas phase. the casing (7.12 × 10−13 mol/L/s) is too slow to set the produced
Degassing of N2 and CO2 yields a pH increase as well as decreases in DIC thermal water back to calcite saturation. Similarly, redissolution of
and N2 concentrations in the liquid phase. As a consequence, calcite gaseous CO2 potentially occurring when the pressure increases to
becomes significantly more supersaturated than in the linear about 100 bar above the pump is too slow to reattain equilibrium.
decompression scenario (Table 2) and its saturation index increases Therefore, CO2 remains at least partly in the gas phase, as demon-
with continued degassing. The dependence of the aqueous composition strated by the presence of CO2-bearing gas inclusions at shallow
on the initially dissolved N2 concentration is such that at a specific depths in the well (Fig. 7, Table 3). Consequently, wellhead samples
pressure, pH and calcite saturation index correlate positively with the are still supersaturated with respect to calcite (Table 2), which ex-
reservoir N2 concentration, while the DIC concentration correlates plains why scaling is currently ongoing in the geothermal plant at
negatively. These correlations occur because, owing to the low N2 the surface (Fig. 3c).
solubility in aqueous fluids, high reservoir N2 concentrations yield an
increase of the gas saturation pressure and thus a higher gas fraction at 5.2.4.3. Open questions. While our calculations nicely predict many of
a specific degassing pressure. our observations by invoking boiling of the reconstructed reservoir fluid
In general, our boiling simulations (Fig. 9) simultaneously re- at 4–6 bar (Fig. 9, Table 4), they predict gas volume fractions of up to
produce the ranges of parameters measured on wellhead samples (e.g., 10%, which is certainly higher than the gas fraction observed at the
pH, DIC, SIcalcite, [N2(aq)]), but only at low degassing pressures of wellhead. Consequently, some of the gas phase must redissolve into the
4–6 bar. The same applies for the N2/CO2 ratios of the gas inclusions. aqueous phase before reaching the wellhead to yield wellhead gas
The composition of the wellhead sample that appears to be least af- fractions lower than those at the pump. In contrast, some of the CO2
fected by scale formation (i.e., max. [Ca] and max. SIcalcite, Table 2: must remain in the gas phase to maintain calcite supersaturation and to
KST-12) is nicely approximated at a degassing pressure of 4.6 bar by form gas inclusions as discussed above. Dissolution of CO2 gas is in fact
setting the reservoir N2 concentration to 3 mmol/kg (Table 4). A unique notoriously slow at T < 150 °C and this phenomenon has long plagued
reconstruction of the reservoir N2 concentration is not possible, how- solubility experiments in pure water and in salt solutions (e.g. as
ever, because other combinations of reservoir N2 concentrations and reviewed by Diamond and Akinfiev, 2003; Akinfiev and Diamond,
degassing pressures can yield similar fits as long as the degassing 2010). Experimentalists typically allow anywhere between 30 min and
pressure is < 6 bar. several hours to equilibrate percolating CO2 gas with aqueous solutions,
even in small-volume containers equipped with mechanical stirrers
5.2.4.2. Implications for the Kirchstockach geothermal power plant. Based (e.g. Malinin and Saveleva, 1972). In the production well at
on the observations that wellhead concentrations and N2/CO2 ratios of Kirchstockach, gas bubbles take only ∼7 min to rise from the pump
gas inclusions are closely approximated by our boiling calculations to the surface plant. Partial rather than complete CO2 equilibration is
(Table 4), we postulate the following sequence of processes to explain therefore to be expected.
calcite scaling at Kirchstockach: At Kirchstockach there is no possibility to closely monitor the pump
in order to verify the occurrence of boiling and gas redissolution related
1. Carbonate minerals and fluid are at thermodynamic equilibrium to cavitation and the subsequent pressure increase above the pump.
within the Malm aquifer owing to high reaction rates and suffi- However, based on the rather wide variety of internally consistent
ciently long residence times. As a consequence, thermal water that observations at hand (gas inclusions, absence of temperature increase in
enters the production well at 3300–3900 m depth (Fig. 2a and b) is the production well, geochemical modeling vs. wellhead fluid compo-
saturated with respect to calcite and dolomite (i.e., sitions, experimental degassing pressures) the pressure has to be as low
SIcalcite = SIdolomite = 0). as 4–6 bar somewhere within the production well and we are confident
2. Decompression of the fluid during upflow within the production that cavitation at the pump is the most likely reason. Furthermore, our
well induces notable calcite supersaturation (Table 2) that may in- observations demonstrate that boiling and gas redissolution are highly
itiate calcite precipitation far below the intake of the pump. dynamic. For instance, three of our 13 wellhead samples are under-
3. A major pressure drop to 4–6 bar occurs somewhere in the pro- saturated with respect to calcite (Table 2, Fig. 6a), which can only
duction well. This pressure is far below the operational pressure of happen if calcite precipitation upon boiling occurs fast enough to set
the plant at the surface (16–18 bar) and far below the pressure at the the calcite saturation index back to zero, and if some redissolution of
intake of the pump (ca. 50 bar, Fig. 2b and c). The interior of the CO2 takes place during further upflow in the production well. Finally,
pump is the only location within the geothermal loop where such a variations in Ca2+ and HCO3− concentrations observed at the wellhead
low pressure is imaginable (Fig. 2c). Therefore, we hypothesize that (Fig. 6) demonstrate that calcite precipitation rates along the produc-
fast rotation of the impeller in the pump induces a highly localized tion well are highly variable. The actual reason for the varying pre-
pressure drop and subsequent exsolution of gas consisting of mainly cipitation rate is not yet resolved, although it might be related to flow
steam and N2 (Fig. 9). Such gas exsolution within centrifugal pumps turbulences within the well and/or to a varying CO2 redissolution rate
is a known phenomenon called cavitation (Rayner, 1995), and it has above the pump.

335
C. Wanner et al. Geothermics 70 (2017) 324–338

Fig. 9. Simulated evolution of properties of the liquid and the coexisting gas phase exsolving from the reconstructed reservoir fluid (Table 2) when the pressure decreases to the steam-
saturation pressure (Psat = 3.3 bar) at T = 137 °C and at variable N2 concentrations. (a) Gas fraction. (b) pH. (c) Dissolved organic carbon (DIC) concentration. (d) Calcite saturation
index. (e) Dissolved N2 concentration. (f) Volumetric N2/CO2 ratio. (g)–(i) Gas composition. Gray bands show the range of the y-axis parameter measured in wellhead fluid samples (b–e)
or in gas inclusions (f).

Unfortunately, we do not know if and to what extent scaling has analyses of fluid inclusions trapped in the scales, to unravel the causes
occurred at greater depths below the pump. However, once this is of substantial amounts of calcite scaling. Geochemical modeling shows
known, it should be possible using our approach to assess whether that the calcite solubility decrease associated with decompression of the
decompression plays a significant role next to cavitation in controlling produced thermal water along the hydrostat during flow from the re-
scale formation within the production well and in the geothermal plant servoir to the wellhead cannot fully explain calcite supersaturation
at the surface. observed at the wellhead. This suggests that another process is driving
scaling as well. Only minor amounts of pyrrhotite and magnetite scales
6. Summary and conclusions are present, implying that corrosion of the casing is limited and hence
its influence on calcite solubility is negligible. The occurrence of N2-rich
We have presented an extensive geochemical dataset from the gas inclusions demonstrates that a free gas phase is present within the
Kirchstockach geothermal well, including chemical analyses of well- well, from the surface down to at least the depth of the downhole-pump
head samples, chemical and mineralogical analyses of scales, as well as (800 m). Stripping of CO2 from the thermal water by this gas phase is

336
C. Wanner et al. Geothermics 70 (2017) 324–338

Table 4
Comparison of wellhead fluid composition predicted from boiling calculations with that determined on the sample that is least affected by scaling formation (KST-12). Also shown is the
simulated N2/CO2 ratio of the coexisting gas phase in comparison to the ratio measured in gas inclusions within calcite (Table 3).

Liquid phase Gas phase

Pdegas DIC (mmol/ HCO3− (mg/ CO2 (mg/ N2(aq) (mmol/ pH SIcalcite Gas fraction [H2O] (vol. [N2] (vol. [CO2] (vol. (N2/CO2)gas
(bar) kg) kg) kg) kg) (Vol%) %) %) %)

a
Simulation 4.6 7.03 241 132 0.44 6.69 0.35 8.4 73.20 19.35 7.45 2.60
b
KST−12 – 7.03 241 132 0.36 6.67 0.30 – – – – 2.37–6.9

– Unknown.
a
[N2]reservoir = 3 mmol/kg.
b
Sample with max. SIcalcite and [Ca] implying minimum amounts of scaling formation.

calculated to lower calcite solubility significantly, and it is therefore the Cacace, M., Blöcher, G., Watanabe, N., Moeck, I., Börsing, N., Scheck-Wenderoth, M.,
Kolditz, O., Huenges, E., 2013. Modelling of fractured carbonate reservoirs: outline of
most likely cause of scaling in addition to decompression. An origin of a novel technique via a case study from the Molasse Basin southern Bavaria,
the gas in the reservoir seems unlikely based on the hydrogeological Germany. Environ. Earth Sci. 70, 3585–3602.
setting, because at the reservoir level the gas would have to be out of Diamond, L.W., Akinfiev, N.N., 2003. Solubility of CO2 in water from −1.5 to 100 °C and
from 0.1 to 100 MPa: evaluation of literature data and thermodynamic modelling.
equilibrium with the thermal water to explain the available chemical Fluid Phase Equilib. 208, 265–290.
analyses. Boiling of the produced water within the geothermal well is Diamond, L.W., 2003a. Systematics of H2O inclusions. In: In: Samson, I., Anderson, A.,
the most plausible mechanism to generate the gas phase. Marshall, D. (Eds.), Fluid Inclusions: Analysis and Interpretation. Mineralogical
Association of Canada. Short Course, vol. 32. pp. 55–79.
Boiling calculations using CHILLER successfully predicted the che- Diamond, L.W., 2003b. Introduction to gas-bearing, aqueous fluid inlcusions. In: In:
mical composition of wellhead samples as well as the N2/CO2 ratios of Samson, I., Anderson, A., Marshall, D. (Eds.), Fluid Inclusions: Analysis and
gas inclusions when the boiling pressure was set to a low pressure of Interpretation. Mineralogical Association of Canada. Short Course, vol. 32. pp.
101–158.
4–6 bar. As the downhole pump is the only location within the geo-
Dublyansky, Y.V., 2012. Design of two crushing devices for release of the fluid inclusion
thermal loop where such a low pressure is imaginable, we hypothesize volatiles. Cent. Eur. J. Geosci. 4, 219–224.
that fast rotation of the centrifugal impeller induces cavitation, gen- Dussel, M., Lüschen, E., Thomas, R., Agemar, T., Fritzer, T., Sieblitz, S., Huber, B., Birner,
erating the gas phase which leads to CO2-stripping and hence accel- J., Schulz, R., 2016. Forecast for thermal water use from Upper Jurassic carbonates in
the Munich region (South German Molasse Basin). Geothermics 60, 13–30.
erating scaling. It follows that scaling and its related technical problems Eichinger, F., Meier, D.B., Hämmerli, J., Diamond, L.W., 2010. Stable Isotope Signatures
could be limited if the pump is operated at lower production rates and/ of Gases Liberated from Fluid Inclusions in Bedrock at Olkiluoto. Posiva Working
or if the well design at the pump is optimized to avoid cavitation. Report 2010–88. Olkiluoto, Finnland.
Gallup, D.L., 1989. Iron silicate scale formation and inhibition at the salton sea geo-
thermal field. Geothermics 18, 97–103.
Acknowledgments García, A.V., Thomsen, K., Stenby, E.H., 2006. Prediction of mineral scale formation in
geothermal and oilfield operations using the Extended UNIQUAC model: part II.
Carbonate-scaling minerals. Geothermics 35, 239–284.
The Swiss Competence Center of Energy Research–Supply for Grguric, B.A., Putnis, A., Harrison, R.J., 1998. An investigation of the phase transitions in
Electricity (SCCER-SoE) supports research in geothermal energy at the bornite (Cu5FeS4) using neutron diffraction and differential scanning calorimetry.
University of Bern. Mark Reed and Nic Spycher kindly provided us a Am. Mineral. 83, 1231–1239.
Gunnarsson, I., Arnórsson, S., 2005. Impact of silica scaling on the efficiency of heat
copy of CHILLER. Special thanks go to Lukas Aschwanden for assisting
extraction from high-temperature geothermal fluids. Geothermics 34, 320–329.
CW in the fluid inclusion lab and to Arthur Adams for electron mi- Hämmerli, J., 2009. Gas-Rock Interaction at the Olkiluoto Investigation Site, Finland, Msc
croprobe analyses. We also appreciate the analytical support of Andrej Thesis. University of Bern, Switzerland.
Helgeson, H.C., Kirkham, D.H., Flowers, G.C., 1981. Theoretical prediction of the ther-
Voropaev and Stephan Wechner (Hydroisotop) and the on-site help
modynamic behavior of aqueous electrolytes by high pressures and temperatures; IV
from Holm Bremer and Igor Hajdu. Finally, we thank the editor, Eva Calculation of activity coefficients, osmotic coefficients, and apparent molal and
Schill, and two anonymous reviewers for their constructive comments standard and relative partial molal properties to 600 °C and 5 kb. Am. J. Sci. 281,
that greatly improved the manuscript. 1249–1516.
Honegger, J.L., Czernichowski-Lauriol, I., Criaud, A., Menjoz, A., Sainson, S., Guezennec,
J., 1989. Detailed study of sulfide scaling at la courneuue nord a geothermal ex-
Appendix A. Supplementary data ploitation of the Paris Basin, France. France Geothermics 18, 137–144.
Kristmannsdóttir, H., 1989. Types of scaling occurring by geothermal utilization in
Iceland. Geothermics 18, 183–190.
Supplementary data associated with this article can be found in the Lentsch, D., Dorsch, K., Sonnleitner, N., Schubert, A., 2015. Prevention of casing failures
online version at http://dx.doi.org/10.1016/j.geothermics.2017.05. in ultra-deep geothermal wells (Germany). In: Proceedings, World Geothermal
001. Congress. Melbourne, Australia, 19–25 April 2015.
Lindal, B., Kristmannsdóttir, H., 1989. The scaling properties of the effluent water from
Kizildere power station, Turkey, and recommendation for a pilot plant in view of
References district heating applications. Geothermics 18, 217–223.
Littke, R., Krooss, B., Idiz, E., Frielingsdorf, J., 1995. Molecular nitrogen in natural-gas
accumulations—generation from sedimentary organic-matter at high-temperatures.
Akinfiev, N.N., Diamond, L.W., 2010. Thermodynamic model of aqueous CO2–H2O–NaCl
AAPG Bull. 79, 410–430.
solutions from −22 to 100 °C and from 0.1 to 100 MPa. Fluid Phase Equilib. 295,
Malinin, S.D., Saveleva, N.I., 1972. Exerimental investigatioin of CO2 solubility in NaCl
104–124.
and CaCl2 solutions at temperatures of 25, 50 and 75 degrees and elevated CO2
Aksoy, N., Serpen, U., Filiz, S., 2008. Management of the Balcova–Narlidere geothermal
pressure. Geokhimiya 643.
reservoir, Turkey. Geothermics 37, 444–466.
Mayrhofer, C., Niessner, R., Baumann, T., 2014. Hydrochemistry and hydrogen sulfide
Aksoy, N., 2007. Optimization of downhole pump setting depths in liquid-dominated
generating processes in the Malm aquifer, Bavarian Molasse Basin, Germany. Hydrol.
geothermal systems: a case study on the Balcova-Narlidere field, Turkey. Geothermics
J. 22, 151–162.
36, 436–458.
Moeck, I., Mraz, E., 2015. Multiphase fossil normal faults in geothermal systems: insights
Ámannsson, H., 1989. Predicting calcite deposition in krafla boreholes. Geothermics 18,
from the Bavarian Molasse Basin. Geotecton. Res. 70–72.
25–32.
Mundhenk, N., Huttenloch, P., Sanjuan, B., Kohl, T., Steger, H., Zorn, R., 2013. Corrosion
Arnórsson, S., 1989. Deposition of calcium carbonate minerals from geothermal water-
and scaling as interrelated phenomena in an operating geothermal power plant.
s—theoretical considerations. Geothermics 18, 33–39.
Corros. Sci. 70, 17–28.
Benoit, W.R., 1989. Carbonate scaling characteristics in Dixie Valley, Nevada geothermal
Palandri, J.L., Kharaka, Y.K., 2004. A compilation of rate parameters of water-mineral
wellbores. Geothermics 18, 41–48.
interaction kinetics for application to geochemical modeling, US Geological Survey.
Bozau, E., Häußler, S., van Berk, W., 2015. Hydrogeochemical modelling of corrosion
Report 2004-1068.
effects and barite scaling in deep geothermal wells of the North German Basin using
Rayner, R., 1995. Pump Users Handbook. Elsevier Advanced Technology, Oxford, UK
PHREEQC and PHAST. Geothermics 53, 540–547.

337
C. Wanner et al. Geothermics 70 (2017) 324–338

427 pp. Am. J. Sci. 294, 361–400.

Reed, M., Palandri, J.L., 2006. SOLTHERM.H06, a Database of Equilibrium Constants for Spycher, N.F., Reed, M.H., 1988. Fugacity coefficients of H2, CO2, CH4, H2O and of H2O-
Minerals and Aqueous Species. University of Oregon, Eugene, USA. CO2-CH4 mixtures: a virial equation treatment for moderate pressures and tempera-
Reed, M., Spycher, N., 2006. Users Guide for CHILLER: A Program for Computing Water- tures applicable to calculations of hydrothermal boiling. Geochim. Cosmochim. Acta
Rock Reactions, Boiling, Mixing and Other Reaction Processes in Aqueous Mineral- 52, 739–749.
Gas Systems. University of Oregon, Eugene, USA. Steefel, C.I., Appelo, C.A.J., Arora, B., Jacques, D., Kalbacher, T., Kolditz, O., Lagneau, V.,
Reed, M., 1992. Calculation of multicomponent chemical equilibria and reaction progress Lichtner, P.C., Mayer, K.U., Meussen, H., Molins, S., Moulton, D., Parkhurst, D.L.,
in systems involving minerals, gases and aqueous phase. Geochim. Cosmochim. Acta Shao, H., Simunek, J., Spycher, N., Yabusaki, S., Yeh, G.T., 2015. Reactive transport
46, 513–528. codes for subsurface environmental simulation. Comput. Geosci. 19, 445–478.
Reed, M., 1997. Hydrothermal alteration and its relationship to ore field composition. In: Thomas, D.M., Gudmundsson, J.S., 1989. Advances in the study of solids deposition in
Barnes, H.L. (Ed.), Geochemistry of Hydrothermal Ore Deposits. John Wiley & Sons, geothermal systems. Geothermics 18, 5–15.
New York, pp. 303–366. Truesdell, A.H., Singers, W., 1974. The calculation of aquifer chemistry in hot-water
Reed, M., 1998. Calculation of simultaneous chemical equilibria in aqueous-mineral-gas geothermal systems. J. Res. US Geol. Surv. 2, 271–278.
systems and its application to modeling hydrothermal processes. In: In: Richards, J., Veron, J., 2005. The Alpine Molasse Basin—review of petroleum geology and remaining
Larson, P. (Eds.), Techniques in Hydrothermal Ore Deposits Geology. Reviews in potential. Bull. Angew. Geol. 10, 75–86.
Economic Geology, vol. 10. pp. 109–124. Wilson, N., Webster-Brown, J., Brown, K., 2007. Controls on stibnite precipitation at two
Regenspurg, S., Feldbusch, E., Byrne, J., Deon, F., Driba, D.L., Henninges, J., Kappler, A., New Zealand geothermal power stations. Geothermics 36, 330–347.
Naumann, R., Reinsch, T., Schubert, C., 2015. Mineral precipitation during produc- Xu, T., Sonnenthal, E.L., Spycher, N., Zheng, L., 2014. TOUGHREACT V3.0-OMP
tion of geothermal fluid from a Permian Rotliegend reservoir. Geothermics 54, Reference Manual: A Parallel Simulation Program for Non-Isothermal Multiphase
122–135. Geochemical Reactive Transport. LBNL Manual. http://esd.lbl.gov/FILES/research/
Scheiber, J., Seibt, A., Birner, J., Genter, A., Moeckes, W., 2013. Application of a scaling projects/tough/documentation/TOUGHREACT_V3-OMP_RefManual.pdf.
inhibitor system at the geothermal power plant in Soultz-sous-Forêts: laboratory and Zarrouk, S.J., Woodhurst, B.C., Morris, C., 2014. Silica scaling in geothermal heat ex-
on-site studies. In: Proceedings, European Geothermal Congress. Pisa, Italy, June 3–7. changers and its impact on pressure drop and performance: Wairakei binary plant,
Simmons, S.F., Christenson, B.W., 1994. Origins of calcite in a boiling geothermal system. New Zealand. Geothermics 51, 445–459.

338