1
2
Benchmark for reactive transport codes
3
in the context of complex cement/clay
4
interactions
5
Nicolas C.M. Marty1, Philippe Blanc1, Olivier Bildstein2, Francis Claret1,*, Benoit
6
Cochepin3, Su Danyang4, Eric C. Gaucher1, Diederik Jacques5, Jean-Eric Lartigue2, K.
7
Ulrich Mayer4, Johannes C.L Meeussen6, Isabelle Munier3, Ingmar Pointeau2, Liu
8
Sanheng5 and Carl Steefel7
9
10
11
12
13
14
15
16
17
18
19
20
21
1 BRGM,
45060 Orleans Cedex, France
Cadarache, F-13108 Saint-Paul-lez-Durance, France
3 ANDRA, 1/7 Rue Jean Monnet, F-92298 Châtenay-Malabry Cedex, France
4
Department of Earth, Ocean and Atmospheric Sciences. University of British
Columbia, 2207 Main Mall, Vancouver BC Canada.
5 Belgian Nuclear Research Centre SCK.CEN, Boeretang 200, Mol Belgium B-2400
6 Nuclear Research and Consultancy Group, PO Box 25, NL-1755 ZG Petten, The
Netherlands
7 Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA
2 CEA,
*Corresponding author f.claret@brgm.fr
1
22
Abstract
23
The use of subsurface and underground geological settings for engineering solutions
24
such as CO2 storage, geothermal energy and nuclear waste repositories will greatly
25
increase the occurrence of claystone/concrete interactions. Due to contrasting
26
geochemical conditions (including Eh, pH, solution composition) this interface is subject
27
to steep concentration gradients and is highly reactive. Predicting long term changes
28
(1,000 to 100,000 years) in these materials is thus crucial for assessing the behaviour of
29
such infrastructures. Experiments cannot provide sufficiently reliable information for
30
such a long time scale and although natural and archaeological analogues can be very
31
helpful, modelling is a unique tool to analyse and test different evolution scenarios. In
32
order to rely on such calculations, it is of paramount importance to demonstrate that the
33
results obtained are not dependent on the numerical reactive transport code used to
34
perform the modelling. In order to address this issue, a benchmark has been
35
established. Seven international teams have participated in this benchmark exercise. All
36
reactive transport codes used (TOUGHREACT, PHREEQC with two different ways of
37
handling transport, CRUNCH, HYTEC, ORCHESTRA, MIN3P-THCm) gave very similar
38
patterns in terms of predicted concentrations of solutes and minerals. The
39
benchmarking exercise reinforces the use of reactive transport modelling in a
40
performance assessment perspective.
3
41
1. Introduction
42
Radioactive waste repositories will use a significant quantity of cement: for the support
43
building of the access galleries and storage cells, for concrete plugs and for
44
containment materials for low to intermediate radioactive wastes. Several European
45
countries have chosen clay formations as possible host rocks (Landais, 2006). To
46
ensure sealing of the access galleries, a mixture of bentonite and sand will not only be
47
used in claystone, but also in granitic host rocks. Numerous cement/clay interfaces will
48
thus be present in a radioactive waste repository. The chemistry of the cementitious
49
material and of the clay media differ substantially, with high pH and low pCO2 in the
50
former and neutral pH and high pCO2 in the latter material. The mineralogy in both
51
materials is also complex and particularly reactive (Gaucher and Blanc, 2006; Savage,
52
2011; Savage et al., 2007). Numerous papers have described the results of reactive
53
transport modelling at this interface, with various degrees of complexity. Some authors
54
have adopted a purely thermodynamic approach (Adler et al., 1999; Gaucher et al.,
55
2004; Trotignon et al., 2006; Wang et al., 2010) while others have opted to couple
56
thermodynamic and kinetic approaches (De Windt et al., 2008; De Windt et al., 2004;
57
Fernandez et al., 2010; Marty et al., 2009; Savage et al., 2010; Savage et al., 2002;
58
Soler, 2003; Soler et al., 2011; Steefel and Lichtner, 1994, 1998; Trotignon et al., 2007;
59
Vieillard et al., 2004; Watson et al., 2009). The topic of cement/clay interactions thus
60
provides an excellent opportunity for testing reactive transport codes for their accuracy,
61
robustness, completeness and numerical stability.
62
This study presents a benchmark exercise open to the community of reactive transport
63
modellers. The problem has been therefore described in sufficient detail to allow its
4
64
reproduction whatever the reactive transport code used. The main challenge of the
65
exercise is linked to the complexity of the chemical reactions introduced in the test
66
cases: chemical speciation, a large number of mineral phases, ion exchange reactions
67
and dissolution/precipitation kinetics. In the frame of this benchmark, geochemical
68
evolution of a concrete/clay interface has been simulated using several codes:
69
TOUGHREACT (Xu et al., 2006; Xu et al., 2011) PHREEQC (Parkhurst and Appelo,
70
1999), CRUNCH (Steefel and Yabusaki, 1996), HYTEC (van der Lee et al., 2002, 2003;
71
van der Lee and Lagneau, 2004), ORCHESTRA (Meeussen, 2003) and MIN3P-THCm,
72
derived from MIN3P (Mayer et al., 2002). For PHREEQC, transport was addressed in
73
two ways: the use of mixed factors inside the PHREEQC code itself or using an external
74
module in which the diffusive transport equation in axisymmetric coordinates is solved
75
using a finite difference method.
76
Each code has its own specific features and capabilities (e.g. discretization schemes,
77
implementation of kinetic rate laws). Therefore this study was not designed to examine
78
small discrepancies between the codes but rather to make sure that all results obtained
79
by the various codes agreed in predicting the same mineralogical and chemical
80
changes considering (i) steep pH and Eh gradients, (ii) highly complex mineralogies
81
(both approximation of local equilibriums and reaction kinetics have been considered),
82
and (iii) the same mesh and transport parameters. Within this benchmark exercise a
83
focus on the calculation time is not performed. Indeed, some software products use
84
parallelised calculations and modelling is not necessarily processed on the same
85
computer, the results cannot be compared in terms of their efficiency to solve reactive
86
transport. Furthermore, this benchmark was proposed by an advanced reactive-
5
87
transport modeling team, not by code developers. Thus, the main purpose here is to
88
increase confidence in the safety analysis based on numerical modelling in the case of
89
nuclear waste management, but also for the safety of CO2 storage where cementitious
90
boreholes and clay cap rocks interact (Gherardi et al., 2012). The long time scale
91
considered in this benchmark is required for safety analysis, including testing different
92
storage scenarios (1,000 to 100,000 years).
93
2. Simulation description
94
2.1.
Geometry
95
One-dimensional radial geometry was chosen for modelling the sealing of a radioactive
96
waste repository site. Considering interacting volumes, the host rock pore water can be
97
assimilated as an infinite source, whereas the quantity of concrete is limited. A
98
heterogeneous mesh with a refined spatial resolution of 0.05 m focused on the
99
concrete/claystone interface was considered. Details of the spatial discretization are
100
given in the figure 1.
101
The mesh size was selected to ensure a satisfactory compromise between a spatial
102
resolution coherent with the expected geochemical processes, especially at the
103
interface, and computation time. Indeed, in the case of a sequential non iterative
104
approach (SNIA), the time step is proportional to the size of a grid cell, which strongly
105
increases the calculation time as the spatial resolution increases. For example, using
106
PHREEQC, to ensure that temporal and spatial discretisation produces a physically
107
correct solution, the value of the time step must at least comply with the Neumann
108
criteria in 1D geometry:
6
t 
x 2
3 Dp
1
109
where Dp (m2 s-1) is the pore diffusion coefficient with a maximum value of 1.4 10-
110
10
111
is worthy to notice that this not a requirement for the global implicit method, however
112
even with this approach, they are practical limits to the time steps due to stiffness and
113
other undetermined issues.
114
Taking into account this criterion, maximum time steps (Δtmax) of 3.1558 106 s must be
115
obeyed for a spatial discretisation of 0.05 m at the clay/concrete interface. However,
116
due to the complexity of the benchmark problem, smaller Δtmax are sometimes applied
117
in order to ensure the numerical convergence of performed simulations.
118
119
m2 s-1 in the claystone system, ∆x (m) and ∆t (s) refer to the space and time steps. It
2.2.
Mineralogical and chemical conditions
2.2.1. Concrete model
120
The mineralogical composition of the concrete considered is representative of an early-
121
age concrete (pH~13.2), i.e. with alkalis (Na+, K+). The concrete plug was modelled as
122
ordinary Portland cement (CEM I) consisting of portlandite, CSH with Ca/Si=1.6,
123
ettringite and minor quantities of hydrotalcite and monocarboaluminate (Blanc et al.,
124
2010a, b). The high calcite content of the material is due to the calcareous nature of the
125
aggregate used to prepare the concrete (table 1).
126
The period required to attain fully hydrated conditions of the concrete plug should be a
127
few thousand years after closure of the repository (Burnol et al., 2006). However, this
128
period was disregarded and simulations started considering fully saturated conditions
7
129
with an interstitial fluid equilibrated with the cement phases (table 2). The dissolved
130
components C, Si, Mg, Al, S and Fe are in equilibrium with mineralogical phases:
131
calcite, CSH, hydrotalcite, monocarboaluminate, ettringite and C3FH6 (Fe-hydrogarnet),
132
respectively.
133
Alkalis are not controlled by mineral phases since Na2O and K2O are undersaturated in
134
concrete water pore. Their concentrations were deduced from concrete water analysis
135
and are fixed when modelling starts. Lothenbach & Winnefeld (2006) measured values
136
varying with time between 320 and 650 mM for potassium and 26 and 65 mM for
137
sodium. Trotignon (2006) used sodium and potassium concentrations of 250 mM for
138
predictive simulations. The alkaline charge supplied by these authors is slightly higher
139
than that used in this study ([K] =140 mM and [Na] = 60 mM).
2.2.2. Callovo-Oxfordian claystone model
140
141
The mineralogy and the pore water composition of the clayey host rock are
142
representative of the Callovo-Oxfordian claystone formation (COx) located in North-East
143
France where the French Nuclear Agency Andra is operating an underground research
144
laboratory (URL) (Delay et al., 2007). The complexity of the mineralogy of the COx
145
formation is included in the model following the recommendations of Gaucher et al.
146
(2009).
147
The constituent phases of claystone selected here (table 3) were established according
148
to the “Chlorite(CCa2)/Illite(IMt2)” model proposed by Gaucher et al. (2009). An
149
apparent dry density of 2.3 g cm-3 has been calculated from the COx mineralogical
150
composition. The mass ratios of different phases were extracted from the BRGM
151
mineralogical database (Lerouge et al., 2006) considering the C2b2 level. Illite and
8
152
illite/smectite minerals (I/S) were identified as the main clay phases of the COx
153
claystone. The simulations therefore incorporate a small amount of montmorillonite in
154
order to reflect the presence of the interstratified mineral. This modelling strategy
155
slightly affects the COx pore water composition established by Gaucher et al. (2009).
156
Microcline was also introduced in the simulations, but the precipitation reaction was
157
inhibited as the mineral cannot form at low temperatures (Gaucher et al., 2009). Quartz
158
precipitation was under kinetic control and amorphous silica was included as a potential
159
secondary phase. The modelling strategy is coherent with the amorphous form of CSH
160
considered in the concrete plug.
161
The model considers ion exchange on illite and smectite reflected in the selectivity
162
exchange coefficients on the two exchangers (Tournassat et al., 2009; Tournassat et
163
al., 2007).
164
The pore water composition resulting from the calculation (table 4) corresponds closely
165
to geochemical analyses performed in the URL at 25 °C (Vinsot et al., 2008).
166
2.2.
Physical properties
167
Considering the very low permeability of the two media, the mass transport calculation
168
only follows Fick’s law:
f i   De gradCi
2
169
where De is the effective diffusion coefficient (m2 s-1) that depends on the properties of
170
the diffusing chemical species, the pore fluid, and the porous medium and C i refers to
171
the concentration of the species i (mol m-3). In this study, the same molecular diffusion
172
coefficient has been considered for all the species. It is worthy noticing that no
9
173
excavated damaged zone (EDZ) was considered here. Lower effective diffusion was
174
assumed for the cement plug relative to the surrounding clay stone (table 5). At the
175
interfaces, depending on the code capability or the choice made by the modeller, either
176
the arithmetic or the harmonic mean of the effective diffusion coefficients was used. It
177
should be noted that this choice has only limited and very local impact on mass
178
transport and the simulation results, because porosity changes and related effects on
179
transport properties were not considered.
180
181
3. Chemical reactions
3.1.
Thermodynamic database
182
The THERMODDEM thermodynamic database (Blanc et al., 2012) was used for this
183
benchmark (http://thermoddem.brgm.fr/). This database is available in different formats:
184
PHREEQC, TOUGHREACT, CRUNCH, Geochemical workbench, and CHESS
185
(geochemical module used by HYTEC). For this study, the UBC team converted the
186
database to the MIN3P-THCm format. ORCHESTRA is able to read the PHREEQC
187
format directly. The data of the 31 minerals considered in the simulations (including
188
primary and secondary phases) are presented in table 6.
189
The THERMODDEM database includes the hydrated radius a0 parameters for the
190
extended Debye-Hückel activity-composition model. The latter was used with codes
191
PHREEQC2, iPHREEQC3 and CRUNCH. For the ORCHESTRA and HYTEC code, the
192
modelling teams have choose to use the Davies activity-composition model instead,
193
which do not imply parameters specific to the aqueous complexes. TOUGHREACT
194
uses a different version of the extended Debye-Hückel model, after Helgeson et al.
10
195
(1981). The effective hydrated radiuses, consistent with such model were then provided
196
instead. MIN3P-THCm is using a specific version of the extended Debye-Hückel model.
197
Eventually, such differences have only a little impact on the results since the ionic
198
strength obtained in the simulation do not oversome a value of 0.3.
199
3.2.
Reaction rates
200
It is widely known that mineral dissolution rates depend on several kinetic parameters.
201
Generally, the effects that physical and chemical parameters exert on mineral
202
weathering rates (temperature, pH, catalysis/inhibition by aqueous species and solution
203
saturation state) are incorporated in a general form of mineral dissolution. TST
204
(Transition State Theory) kinetic laws are included in most geochemical codes. Its
205
formulation is implemented in several codes (TOUGHREACT, CRUNCH, HYTEC and
206
MIN3P-THCm). Dissolution/precipitation rates of a mineral (n) at different pH values and
207
constant temperature are given by Lasaga et al. (1994):
rn   k n An 1  n

3
208
where positive rn values indicate dissolution and negative values denote precipitation
209
(mol s-1 kg-1w), kn is the rate constant (mol m-2 s-1), which is temperature dependent, An
210
is the reactive surface area (m2 kg-1w) and Ωn is the mineral saturation ratio.
211
Parameters θ and η must be determined experimentally in order to describe the rate
212
dependency on the state of saturation. However, these parameters are only rarely found
213
for mineral dissolution, because reactions are usually studied far from equilibrium.
214
The dissolution constant (k in the above equation) is expressed as:
11
  Eanu
nu
k  k 25
exp 
 R
  Eai
1 
1
i
 
   k 25 exp 
 T 298.15  i
 R
1 
1
n
 
 aijij
 T 298.15  j
215
where Ea (J mol-1) is the activation energy and k25 is the rate constant at 25 °C. The
216
superscripts nu and i refer to reactions under neutral pH conditions and other
217
conditions, respectively; j refers to the species index involved in one mechanism, a is
218
the activity of the species and n is a power term.
219
Note that rate equations can be provided directly in the input files of ORCHESTRA and
220
PHREEQC. This strategy makes these codes extremely flexible. TST laws can
221
therefore be easily reproduced.
222
Apart from several reactions at local equilibrium, dissolution/precipitation reactions of
223
primary phases were controlled as much as possible by kinetic laws. The kinetic
224
parameters shown in table 7 and table 8 were extracted from a literature review.
225
3.3.
Ion exchange
226
Claystone exchange properties are dominated by those of I/S (Gaucher et al., 2009). A
227
CEC of 17.4 meq 100 g-1 was considered. Taking into account a rock grain density of
228
2.65 g.cm-3, this corresponds to 2.1 mol.L-1 of exchangers. The exchange assemblage
229
was limited to Na, K, Mg and Ca cations. Claystone were initially equilibrated with the
230
COx pore water shown in table 4. The selectivity coefficients available in table 9 were
231
calculated using exchange models given by Tournassat et al. (2007; 2009). The
232
coefficients use the Gaines and Thomas convention (Gaines and Thomas, 1953;
233
Sposito, 1981).
12
4
234
Although some codes can handle variation in the total amount of exchanger (e.g.
235
calculated from the density, the porosity or the volume fraction of a mineral), within this
236
benchmark the total exchanger concentration remains constant.
237
238
239
4. Benchmark cases
4.1.
Validation
of
transport
properties
and
cationic
exchange models (case 1)
240
The radial geometry describe in figure 1 has been used for simulations performed in
241
absence of mineralogy. Only cement and COx fluids as well as the exchange capacity
242
of the COx formation have been considered. The examination of modelled profiles (Cl
243
concentration and exchanger composition) at 1,000 years has been used for the
244
validation of transport properties and cationic exchange models.
245
4.2.
Full mineralogy with slow reaction rates (case 2)
246
The extrapolation of available laboratory rates to “natural system” is not yet supported
247
by field studies (laboratory experimental feldspar dissolution rates are 2-5 orders of
248
magnitude times faster than “natural rates”, Velbel, 1993; White and Brantley, 2003;
249
Zhu, 2005). Kinetic parameters reported in this study have been established form flow-
250
through experiments and proposed reaction rates are probably overestimated.
251
Simulations were then performed using surface areas 3 orders of magnitude lower than
252
the values given in table 8. It should be noted that even if several parameters can
253
explain the discrepancy between natural and experimental rates (Maher et al., 2006),
254
the decrease in reactive surface areas (A) used in equation 3 is strictly equivalent to a
13
255
decrease in rate constants, whatever the exact cause of this decrease, because
256
reactive surface areas are supposed constant in calculations.
4.3.
257
Full mineralogy with fast reaction rates (case 3)
258
The last benchmark exercise uses reactive surface areas given in table 8 without any
259
modification. Simulations test the code capabilities to manage with efficiency the time
260
step necessary to reach a correct numerical convergence.
261
5. Numerical codes
5.1.
262
TOUGHREACT
263
Part of the benchmark was performed using the TOUGHREACT two-phase non-
264
isothermal reactive transport code (Xu et al., 2004). This code was developed by
265
introducing reactive geochemistry into the framework of TOUGH2 V2 (Pruess et al.,
266
1999).
267
TOUGHREACT calculations usually adopt the sequential non-iterative approach (SNIA).
268
The selected resolution computation procedure for one time step is a sequence
269
comprising one non-reactive transport step followed by a batch chemistry step. The
270
reactive transport was therefore solved using the SNIA. This approach could lead to
271
numerical errors but is frequently used, mainly to save computation time. Full details on
272
the numerical methods and simulator capabilities of TOUGHREACT are given in Xu et
273
al. (2004) and Xu et al. (2004; 2011).
Concrete/clay
interactions
were
14
simulated
using
the
EOS3
module.
5.2.
274
PHREEQC
5.2.1. PHREEQC2
275
276
Reactive transport performed with PHREEQC2 (Parkhurst & Appelo, 1999) is also
277
based on a sequential non-iterative approach (SNIA). Mixing factors between two
278
adjacent cells are used to calculate the mass transport by diffusion. Mixing factors are
279
defined explicitly for each cell in PHREEQC (Appelo et al., 2010; Parkhurst and Appelo,
280
1999):
mixfij 
Di tAij f be
5
hijV j
281
where Di is the solute diffusion coefficient, Aij is the surface area between adjacent cells,
282
fbc is a boundary condition correction factor, hij is the distance between the midpoints of
283
adjacent cells and Vj is volume of the central cell.
284
5.2.2. iPHREEQC
285
iPHREEQC is a set of modules developed specifically to allow easy integration of
286
PHREEQC into other software (Charlton and Parkhurst, 2011). To simulate the
287
interaction between the concrete and clay, the diffusive transport equation in
288
axisymmetric coordinates is solved using a finite difference method and the reaction is
289
calculated with PHREEQC via the iPHREEQC modules. Coupling is performed with a
290
sequential non-iterative approach (SNIA), i.e. the same method as used in PHREEQC.
291
The code is further parallelised with a message passing interface (MPI) using the
292
concept of domain decomposition. When the model starts, each processor is assigned
293
with a certain number of cells and each processor creates an iPHREEQC module to
15
294
perform the geochemical calculation task for all the cells residing on that processor. At
295
the domain boundaries, messages which are primarily chemical compositions are
296
transmitted from one processor to another (Charlton and Parkhurst, 2011).
297
5.3.
CRUNCH
298
CRUNCH (or CrunchFlow) is a software package for simulating multi-component multi-
299
dimensional reactive transport in porous media. It incorporates most of the features
300
previously found in the GIMRT/OS3D package into a single code (Steefel, 2008; Steefel
301
and Yabusaki, 1996) and uses an automatic read of a thermodynamic and kinetic
302
database. The global implicit approach (GIMRT) is used to simulate diffusive reactive
303
transport since it can take larger time steps than sequential methods once the system
304
relaxes into a quasi-stationary state. After a unique reactive transport time step, mineral
305
volumes and surface areas are updated.
306
5.4.
HYTEC
307
The HYTEC code (van der Lee et al., 2002, 2003) is used for reactive transport
308
modelling in porous media under saturated and unsaturated conditions. HYTEC is
309
based on a finite volume scheme with representative elementary volumes (REV) for
310
mass transport and a sequential iterative operator-splitting method for coupling between
311
chemistry and transport. The HYTEC 4.0.4 is the version used in this study
312
5.5.
ORCHESTRA
313
In ORCHESTRA, reactive transport processes are implemented by a mixing-cell
314
concept. This implies that transport systems can be composed of (well-mixed) cells and
16
315
connections between these cells. The cells contain the information on the local physical
316
and chemical composition, while the connections between the cells contain the mass
317
transport equations (diffusion, convection etc.). Both the connections between the cells,
318
but also the literal equations involved in mass transport are defined in text input files.
319
Effectively this results in a finite difference scheme. In ORCHESTRA it is possible to
320
choose a SIA approach (predictor corrector method), but for this benchmark a non-
321
iterative approach was used (SNIA), solved with a fourth order Runge Kutta method.
322
The kinetic reactions were solved using the same time step as the transport processes.
323
ORCHESTRA is coded in Java, which results in an executable programme that runs on
324
different operating systems (e.g. MS Windows, Linux, Unix, Solaris, OSX). Java also
325
contains standard language constructs for parallel programming, which facilitates
326
parallelisation. As a result, ORCHESTRA makes efficient use of multi-processor
327
hardware. In contrast with the iPHREEQC method for parallelisation that uses domain
328
decomposition with nodes/cells predefined for each processor, the ORCHESTRA
329
approach uses a single node pool, from which nodes are taken by each available
330
processor. The advantage of the iPHREEQC method is that it is more suitable for
331
distribution work over different physical machines. The advantage of the ORCHESTRA
332
method is better load balancing for systems with strongly varying calculation times per
333
node.
334
In ORCHESTRA it is possible to obtain real-time graphic output for any
335
chemical/physical parameter (e.g. concentration or pH profiles) during a calculation,
336
which makes it possible to detect oscillations or unexpected behaviour before a run has
337
finished.
17
338
5.6.
MIN3P-THCm
339
MIN3P-THCm (Mayer et al., 2002; Mayer and MacQuarrie, 2010) is a general purpose
340
multi-component reactive transport code for variably saturated porous media. It has a
341
high degree of flexibility to a wide range of hydrogeological and geochemical problems.
342
The chemical processes included are homogeneous reactions in the aqueous phase as
343
well as heterogeneous reactions. Reactions within the aqueous phase and dissolution-
344
precipitation reactions can be considered as equilibrium or kinetically controlled
345
processes. All kinetically controlled reactions can be described as reversible or
346
irreversible reaction processes. Different reaction mechanisms for dissolution-
347
precipitation reactions are considered, which can be subdivided into surface controlled
348
and transport controlled reactions.
349
The code is based on the direct substitution approach (DSA) and uses the global
350
implicit method (GIM) for solving the model equations, which considers reaction and
351
transport simultaneously. Spatial discretisation is performed on the basis of the finite
352
volume method and allows simulations to be conducted in one, two, and three spatial
353
dimensions. To maximise versatility, the model formulation includes a generalised
354
framework for kinetically controlled reactions, which can be specified through a
355
database together with equilibrium processes. Full details on the numerical model and
356
capabilities are given in Mayer et al. (2002) and Mayer and MacQuarrie (2010). The
357
code was recently extended for non-isothermal systems, highly saline conditions,
358
deformation due to surface loading and to include atmospheric boundary conditions
359
(Bea et al., 2010; Bea et al., 2012). The revised code is entitled MIN3P-THCm and the
360
simulations for this benchmark was conducted with version 1.0.48.0.
18
361
6. Results and discussion
6.1.
362
Case 1
363
The first proposed case has been performed for the validation of transport properties.
364
Benchmark exercise
365
concentration profiles).
has been evaluated from geochemical evolutions (e.g.
6.1.1. Cl concentration profiles
366
367
As in our model the concrete pore water is free from Cl, this element can be used as a
368
tracer for the validation of physical properties of the considered porous media. The
369
perfect agreement between Cl concentration profiles (cf. figure 2) prevents the
370
possibility of modeller’s mistakes or software’s incoherencies for the mass-transport
371
processing.
6.1.2. Exchanger composition
372
373
Exchanger composition at 1,000 years has been reported on figure 3. Gaines-Thomas
374
exchange convention has been used for the calculations. Weak discrepancies observed
375
on exchanger composition are then mainly attributed to various activity models
376
implemented in each code (see section 3.1). Such effect is well illustrated on the Na-
377
exchanged profiles where 3 groups can be identified:
378
-
ORCHESTRA and HYTEC using the Davies equation,
379
-
MIN3P-THCm making a compromise between Davies and Debye-Huckle
380
equation,
19
381
382
-
PHREEQC2 and iPHREEQC3, TOUGHREACT and CRUNCH using the
extended Debye-Huckle equation.
6.2.
383
Case 2
384
The second exercise aims to test the efficiency of each code for the processing of
385
complex geochemistry such as one encountered in claystone/concrete interactions.
386
Numerical results are compared in terms of mineralogical and fluid chemistry
387
transformations. Exchanger composition evolutions can be extracted from code input
388
files available in electronic supplementary data.
6.2.1. Mineralogical changes and pH evolutions
389
390
Mineralogical transformations are shown in volume distribution diagrams with the
391
various mineralogical phases considered in the system according to the distance.
392
Changes in pH are also shown on the secondary axis of ordinates (red curve). Results
393
obtained after 10,000 years of concrete/claystone interactions are shown in figure 4 for
394
TOUGHREACT, figure 5 for PHREEQC2, figure 6 for iPHREEQC3, figure 7 for
395
CRUNCH, figure 8 for HYTEC, figure 9 for ORCHESTRA and figure 10 for MIN3P-
396
THCm.
397
Geochemical transformations resulting from contact between clay and concrete media
398
have been well described and have also been the subject of numerous publications
399
(Gaucher and Blanc, 2006; Savage et al., 2007; Savage, 2011). Therefore, the aim of
400
this study is not to provide a detailed description of the expected geochemical
401
processes. In brief, the main reactions observed are:
20
402

C3FH6, monocarboaluminate, CSH1.6 and portlandite in the concrete;
403
404
Dissolution of smectite (weak), quartz and dolomite inside the claystone and

Precipitation of calcite, saponite and straetlingite in the claystone and ettringite,
saponite, ferrihydrite , magnetite and CSH1.2 and 0.8 in the concrete.
405
406
Although the formulations and specific capabilities of each code are not strictly identical,
407
the numerical results indicate strong similarities in nature, amount and extent of the
408
materials modification. The models also show a tendency to clog the porosity in
409
response to mineralogical transformations.
410
6.2.2. Pore water evolution
411
Only very small differences are observed for the K, Na and Al concentration profiles
412
(figure
413
processes/parameters responsible for these discrepancies cannot be identified with
414
certainty. Independently of the code used, all pH profiles are very similar (figure 4 to
415
figure 10).
416
11).
Due
6.3.
to
the
complexity
of
the
calculations
performed,
the
Case 3
417
After a brief presentation of code-specific simplifications and modifications made on the
418
benchmarking exercise, numerical results are compared in terms of mineralogical
419
transformations.
420
extracted from code input files available in electronic supplementary data.
Fluid chemistry and exchanger composition evolutions can be
21
421
6.3.1. Modification of the benchmark conditions
422
As expected, fast reaction rates proposed in the benchmark problem resulted in
423
numerical difficulties for the majority of the codes used. In order to perform the
424
calculations, the benchmark specifications described above had to be modified to
425
varying degrees. The modelling teams had to deal with the specific capabilities of each
426
code. The main modifications concern the number of mineralogical phases under kinetic
427
control and the maximum time step size (table 10). When not considered as kinetic
428
reactions, the mineralogical phases were processed assuming local equilibrium.
429
6.3.2. Mineralogical changes and pH evolutions
430
Results obtained after 10,000 years of concrete/claystone interactions are shown in
431
figure 12 for TOUGHREACT, figure 13 for PHREEQC2, figure 14 for iPHRREQC3,
432
figure 15 for CRUNCH, figure 16 for HYTEC, figure 17 for ORCHESTRA and figure 18
433
for MIN3P-THCm.
434
Observed small differences are likely dominated by code-specific modifications of the
435
initial benchmarking proposal in order to perform the calculation. For example as shown
436
in table 10, the number of phases initially expected to be considered under kinetic
437
control has been decreased for most models. These modifications mainly concern the
438
processing at local equilibrium for the fastest reaction rates (calcite, dolomite and
439
siderite) to reduce stiffness in the system of equations. The relevance of a kinetic
440
formulation for carbonates is questionable for long-term predictive modelling in a purely
441
diffusive system. Besides, considering a mesh refinement of 5 cm at the
442
concrete/claystone interface, these minerals could be processed with the approximation
443
of local equilibrium (Marty et al., 2009). At first glance, one could argue that whatever
22
444
the relevance of the proposed kinetics, numerical codes should be able to process the
445
proposed exercise. However, numerically, it is difficult to obtain a correct estimation of
446
reaction rates for fastest rate constants when the mineral saturation ratio is close to one
447
(Ω in equation 3). It could reduce accuracy and cut the time step, so an equilibrium or
448
quasi equilibrium approach can be well justified. The difficulties encountered are
449
probably due to the capacity of codes to manage the time step efficiently enough to
450
solve
451
convergence criterion) may differ slightly from one modelled case to another and then
452
may also affect numerical results (possible source of discrepancies). This issue is
453
particularly true for codes using a SNIA approach for solving reactive-transport
454
equations.
455
Compared with the previous results (see mineralogical changes obtained from the case
456
2 in section 6.2.1), the highest availability of silica (realized from clay dissolution)
457
promotes the formation of zeolite (Clinoptilolite(Ca): Ca0.55(Si4.9Al1.1)O12:3.9H2O) instead
458
of straetlingite (Ca2Al2SiO2(OH)10:2.5H2O). Other mineralogical transformations (e.g.
459
saponite precipitations) have been identified in the previous case. Whatever the
460
considered reaction rates (case 2 vs. case 3), alkaline plume diffusion (i.e. pH > 9) is
461
limited at the first 10 centimeters inside the claystone at 10,000 years of simulated time.
462
reactive-transport
equations. The
selected
calculation
parameters
(e.g.
7. Conclusion
463
Although the code formulations differ substantially (numerical scheme, activity
464
correction model, polynomial equation for equilibrium-constant calculations...), very
465
limited discrepancies in the numerical results can be observed. Indeed, independent of
23
466
the code used, the mineralogical profiles in the concrete zone indicate strong similarities
467
in terms of volume fractions and composition of the constituent phases and in terms of
468
solutes concentrations. Critical analysis has demonstrated the robustness of the
469
obtained results regarding the geochemical evolution at the cement/clay interface.
470
Taking into account the feedback of the porosity evolution on the diffusion coefficient
471
would certainly be a good subject for another benchmark. Even though, this question
472
has been tackled in an advective system by the benchmark proposed by Xie et al (this
473
issue), evaluating the effect for a purely diffusive system and with more complex
474
geochemical transformation will be of great interest. Another topic that could be
475
investigated in the future concerns the modelling of such an interface taking into
476
account the heterogeneity of the porous media (e.g. excavated disturbed zone) and its
477
associated transport parameters.
478
479
Acknowledgments: Subsurface environmental simulation benchmarking studies have
480
been initiated by Carl Steefel and Steve Yabusaki. We thank the LBNL, for the
481
organisation of SS Bench I workshop at Berkeley, the Graduate Institute of Applied
482
Geology of National Central University and the National Science Foundation of Taiwan,
483
for the organisation of SS Bench II workshop at Taiwan and the Helmholtz Centre for
484
Environmental Research - UFZ for the organisation of SS Bench III workshop at Leipzig.
485
24
486
References:
487
Adler, M., Mader, U.K., Waber, H.N., 1999. High-pH alteration of argillaceous rocks: an
488
experimental study. Schweiz Miner Petrog 79, 445-454.
489
Appelo, C.A.J., Van Loon, L.R., Wersin, P., 2010. Multicomponent diffusion of a suite of
490
tracers (HTO, Cl, Br, I, Na, Sr, Cs) in asingle sample of Opalinus Clay. Geochim
491
Cosmochim Ac 74, 1201-1219.
492
Bea, S.A., Mayer, K.U., Macquarrie, K.T.B., 2010. Hydromechanical and geochemical
493
coupling
within
an
intercratonic
sedimentary
basin
affected
494
glaciation/deglaciation events. Geochim Cosmochim Ac 74, A63-A63.
by
495
Bea, S.A., Wilson, S.A., Mayer, K.U., Dipple, G.M., Power, I.M., Gamazo, P., 2012.
496
Reactive Transport Modeling of Natural Carbon Sequestration in Ultramafic Mine
497
Tailings. Vadose Zone J 11.
498
Blanc, P., Bourbon, X., Lassin, A., Gaucher, E.C., 2010a. Chemical model for cement-
499
based materials: Temperature dependence of thermodynamic functions for
500
nanocrystalline and crystalline C-S-H phases. Cement and Concrete Research 40,
501
851-866.
502
Blanc, P., Bourbon, X., Lassin, A., Gaucher, E.C., 2010b. Chemical model for cement-
503
based materials: Thermodynamic data assessment for phases other than C-S-H.
504
Cement and Concrete Research 40, 1360-1374.
505
Blanc, P., Lassin, A., Piantone, P., Azaroual, M., Jacquemet, N., Fabbri, A., Gaucher,
506
E.C., 2012. Thermoddem: A geochemical database focused on low temperature
507
water/rock interactions and waste materials. Appl Geochem 27, 2107-2116.
25
508
Burnol, A., Blanc, P., Xu, T., Spycher, N., Gaucher, E.C., 2006. Uncertainty in the
509
reactive transport model response to an alkaline perturbation in a clay formation.,
510
in: Laboratory, L.B.N. (Ed.), TOUGH Symposium 2006, Berkeley, California.
511
Charlton, S.R., Parkhurst, D.L., 2011. Modules based on the geochemical model
512
PHREEQC for use in scripting and programming languages. Comput Geosci 37,
513
1653-1663.
514
De Windt, L., Marsal, F., Tinseau, E., Pellegrini, D., 2008. Reactive transport modeling
515
of geochemical interactions at a concrete/argillite interface, Tournemire site
516
(France). Phys Chem Earth 33, S295-S305.
517
De Windt, L., Pellegrini, D., van der Lee, J., 2004. Coupled modeling of
518
cement/claystone interactions and radionuclide migration. J Contam Hydrol 68,
519
165-182.
520
Delay, J., Rebours, H., Vinsot, A., Robin, P., 2007. Scientific investigation in deep wells
521
for nuclear waste disposal studies at the Meuse/Haute Marne underground
522
research laboratory, Northeastern France. Phys Chem Earth 32, 42-57.
523
Fernandez, R., Cuevas, J., Mader, U.K., 2010. Modeling experimental results of
524
diffusion of alkaline solutions through a compacted bentonite barrier. Cement and
525
Concrete Research 40, 1255-1264.
526
Gaines, J.G.L., Thomas, H.C., 1953. Adsorption Studies on Clay Minerals. II. A
527
Formulation of the Thermodynamics of Exchange Adsorption. The Journal of
528
Chemical Physics 21, 714-718.
529
530
Gaucher, E.C., Blanc, P., 2006. Cement/clay interactions - A review: Experiments,
natural analogues, and modeling. Waste Manage 26, 776-788.
26
531
532
Gaucher, E.C., Blanc, P., Matray, J.M., Michau, N., 2004. Modeling diffusion of an
alkaline plume in a clay barrier. Appl Geochem 19, 1505-1515.
533
Gaucher, E.C., Tournassat, C., Pearson, F.J., Blanc, P., Crouzet, C., Lerouge, C.,
534
Altmann, S., 2009. A robust model for pore-water chemistry of clayrock. Geochim
535
Cosmochim Ac 73, 6470-6487.
536
Gherardi, F., Audigane, P., Gaucher, E.C., 2012. Predicting long-term geochemical
537
alteration of wellbore cement in a generic geological CO2 confinement site:
538
Tackling a difficult reactive transport modeling challenge. J Hydrol 420, 340-359.
539
Landais, P., 2006. Advances in geochemical research for the underground disposal of
540
high-level, long-lived radioactive waste in a clay formation. J Geochem Explor 88,
541
32-36.
542
Lasaga, A.C., Soler, J.M., Ganor, J., Burch, T.E., Nagy, K.L., 1994. CHEMICAL-
543
WEATHERING RATE LAWS AND GLOBAL GEOCHEMICAL CYCLES. Geochim
544
Cosmochim Ac 58, 2361-2386.
545
Lerouge, C., Michel, P., Gaucher, E.C., Tournassat, C., 2006. A geological,
546
mineralogical and geochemical GIS for the Andra URL: A tool for the water-rock
547
interactions modelling at a regional scale., in: SKB (Ed.), FUNMIG - 2nd annual
548
workshop, Stockholm, Sweden.
549
550
Lothenbach, B., Winnefeld, F., 2006. Thermodynamic modelling of the hydration of
Portland cement. Cement and Concrete Research 36, 209-226.
551
Marty, N.C.M., Munier, I., Gaucher, E.C., Tournassat, C., Gaboreau, S., Vong, C.Q.,
552
Giffaut, E., Cochepin, B., Claret, F., Submitted. Simulation of cement/clay
27
553
interactions: feedback on the increasing complexity of modelling strategies.
554
Transport in Porous Media.
555
Marty, N.C.M., Tournassat, C., Burnol, A., Giffaut, E., Gaucher, E.C., 2009. Influence of
556
reaction kinetics and mesh refinement on the numerical modelling of concrete/clay
557
interactions. J Hydrol 364, 58-72.
558
Mayer, K.U., Frind, E.O., Blowes, D.W., 2002. Multicomponent reactive transport
559
modeling in variably saturated porous media using a generalized formulation for
560
kinetically controlled reactions. Water Resour Res 38.
561
Mayer, K.U., MacQuarrie, K.T.B., 2010. Solution of the MoMaS reactive transport
562
benchmark with MIN3P-model formulation and simulation results. Computational
563
Geosciences 14, 405-419.
564
565
Meeussen, J.C.L., 2003. ORCHESTRA: An object-oriented framework for implementing
chemical equilibrium models. Environ Sci Technol 37, 1175-1182.
566
Parkhurst, D.L., Appelo, C.A.J., 1999. User's guide to PHREEQC (version 2) - a
567
computer program for speciation, reaction-path, 1D-transport, and inverse
568
geochemical calculations. US Geol. Surv. Water Resour. Inv. Rep. 99-4259, 312p.
569
Pruess, K., Oldenburg, C., Morodis, G., 1999. TOUGH2 user’s guide, Version 2.0, in:
570
571
572
LBNL-43134, L.B.N.L.R. (Ed.), Berkeley, p. 197.
Savage, D., 2011. A review of analogues of alkaline alteration with regard to long-term
barrier performance. Mineral Mag 75, 2401-2418.
573
Savage, D., Benbow, S., Watson, C., Takase, H., Ono, K., Oda, C., Honda, A., 2010.
574
Natural systems evidence for the alteration of clay under alkaline conditions: An
575
example from Searles Lake, California. Appl Clay Sci 47, 72-81.
28
576
577
Savage, D., Noy, D., Mihara, M., 2002. Modelling the interaction of bentonite with
hyperalkaline fluids. Appl Geochem 17, 207-223.
578
Savage, D., Walker, C., Arthur, R., Rochelle, C., Oda, C., Takase, H., 2007. Alteration
579
of bentonite by hyperalkaline fluids: A review of the role of secondary minerals.
580
Phys Chem Earth 32, 287-297.
581
Soler, J.M., 2003. Reactive transport modeling of the interaction between a high-pH
582
plume and a fractured marl: the case of Wellenberg. Appl Geochem 18, 1555-
583
1571.
584
Soler, J.M., Vuorio, M., Hautojarvi, A., 2011. Reactive transport modeling of the
585
interaction between water and a cementitious grout in a fractured rock. Application
586
to ONKALO (Finland). Appl Geochem 26, 1115-1129.
587
588
589
590
591
592
Sposito, G., 1981. The thermodynamics of soil solution. Oxford University Press, New
York.
Steefel, C.I., 2008. CrunchFlow, Software for modeling multicomponent reactive flow
and tranport, User 's manual.
Steefel, C.I., Lichtner, P.C., 1994. Diffusion and reaction in rock matrix bordering a
hyperalkaline fluid-filled fracture. Geochim Cosmochim Ac 58, 3595-3612.
593
Steefel, C.I., Lichtner, P.C., 1998. Multicomponent reactive transport in discrete
594
fractures - II: Infiltration of hyperalkaline groundwater at Maqarin, Jordan, a natural
595
analogue site. J Hydrol 209, 200-224.
596
Steefel,
C.I.,
Yabusaki,
S.B.,
1996.
597
Muticomponent-Multidimensional
598
Programmer's guide.
OS3D/GIMRT.
Reactive
29
Software
Transport.
User
for
Modeling
Manual
&
599
Tournassat, C., Gailhanou, H., Crouzet, C., Braibant, G., Gautier, A., Gaucher, E.C.,
600
2009. Cation Exchange Selectivity Coefficient Values on Smectite and Mixed-
601
Layer Illite/Smectite Minerals. Soil Sci Soc Am J 73, 928-942.
602
Tournassat, C., Gailhanou, H., Crouzet, C., Braibant, G., Gautier, A., Lassin, A., Blanc,
603
P., Gaucher, E.C., 2007. Two cation exchange models for direct and inverse
604
modelling of solution major cation composition in equilibrium with illite surfaces.
605
Geochim Cosmochim Ac 71, 1098-1114.
606
Trotignon, L., Devallois, V., Peycelon, H., Tiffreau, C., Bourbon, X., 2007. Predicting the
607
long term durability of concrete engineered barriers in a geological repository for
608
radioactive waste. Phys Chem Earth 32, 259-274.
609
610
Trotignon, L., Peycelon, H., Bourbon, X., 2006. Comparison of performance of concrete
barriers in a clayey geological medium. Phys Chem Earth 31, 610-617.
611
van der Lee, J., De Windt, L., Lagneau, V., Goblet, P., 2002. Presentation and
612
application of the reactive transport code HYTEC. Computational Methods in
613
Water Resources, Vols 1 and 2, Proceedings 47, 599-606.
614
615
van der Lee, J., De Windt, L., Lagneau, V., Goblet, P., 2003. Module-oriented modeling
of reactive transport with HYTEC. Comput Geosci 29, 265-275.
616
van der Lee, J., Lagneau, V., 2004. Rigorous methods for reactive transport in
617
unsaturated porous medium coupled with chemistry and variable porosity. Dev
618
Water Sci 55, 861-868.
619
Vieillard, P., Ramirez, S., Bouchet, A., Cassagnabere, A., Meunier, A., Jacquot, E.,
620
2004. Alteration of the Callow-Oxfordian clay from Meuse-Haute Marne
30
621
Underground Laboratory (France) by alkaline solution: II. Modelling of mineral
622
reactions. Appl Geochem 19, 1699-1709.
623
624
Vinsot, A., Mettler, S., Wechner, S., 2008. In situ characterization of the CallovoOxfordian pore water composition. Phys Chem Earth 33, S75-S86.
625
Wang, L., Jacques, D., De Cannière, P., 2010. Effects of an Alkaline Plume on the
626
Boom Clay as a Potential Host for Geological Disposal of Radioactive Waste.
627
SCK•CEN
628
http://publications.sckcen.be/dspace/bitstream/10038/1397/1/er_28final.pdf
ER-28,
Mol,
Belgium.
629
Watson, C., Hane, K., Savage, D., Benbow, S., Cuevas, J., Fernandez, R., 2009.
630
Reaction and diffusion of cementitious water in bentonite: Results of 'blind'
631
modelling. Appl Clay Sci 45, 54-69.
632
Xu, T., Sonnenthal, E., Spycher, N., Pruess, K., 2004. TOUGHREACT user's guide: a
633
simulation program for non-isothermal multiphase reactive geochemical transport
634
in variable saturated geologic media, in: LBNL-55460, L.B.N.L.R. (Ed.), Berkeley,
635
p. 192.
636
Xu, T., Sonnenthal, E., Spycher, N., Pruess, K., 2006. TOUGHREACT—A simulation
637
program for non-isothermal multiphase reactive geochemical transport in variably
638
saturated geologic media: Applications to geothermal injectivity and CO2
639
geological sequestration. Comput Geosci 32, 145-165.
640
Xu, T., Spycher, N., Sonnenthal, E., Zhang, G., Zheng, L., Pruess, K., 2011.
641
TOUGHREACT Version 2.0: A simulator for subsurface reactive transport under
642
non-isothermal multiphase flow conditions. Comput Geosci 37, 763-774.
643
31
644
TABLES
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
Table 1: Mineralogical composition of the concrete plug.
Table 2: Composition of the concrete pore water (PHREEQC calculation).
Table 3: Mineralogical composition of the Callovo-Oxfodian argillites
Table 4: Composition of the Callovo-Oxfordian porewater (PHREEQC calculation).
Table 5: Transport parameters
Table 6: Thermodynamic constants (25°C) and molar volume of minerals considered in
simulations.
Thermodynamic
data
were
extracted
from THERMODDEM
(http://thermoddem.brgm.fr).
Table 7: Kinetic parameters considered for dissolution reactions. Parameters describe
the pH dependence and deviation from equilibrium of the dissolution rates (see
appendix B in Xu et al., 2004).
Table 8: Kinetic parameters considered for precipitation reactions.
Table 9: Selectivity coefficients for exchange reactions. Extracted from Gaucher et al.
(2009)
Table 10: Simplifications adopted in order to perform the proposed benchmark exercise.
32
661
Table 1: Mineralogical composition of the concrete plug.
Minerals name in the
Structural formula
Volume
mol L-1 of
DDB
fraction (%)
solution
C3FH6
Ca3Fe2(OH)12
2.17
0.94
Calcite
CaCO3
71.43
129.45
CSH(1.6)
Ca1.60SiO3.6:2.58H2O
14.69
11.61
Ettringite
Ca6Al2(SO4)3(OH)12:26H2O
4.22
0.40
Hydrotalcite
Mg4Al2O7:10H2O
0.81
0.24
Monocarboaluminate
Ca4Al2CO9:10.68H2O
0.10
0.02
Portlandite
Porosity
Ca(OH)2
0.13
662
663
33
6.38
12.92
664
Table 2: Composition of the concrete pore water (PHREEQC calculation).
Concrete pore water chemical
Element (total
compositionat 25°C (mol kgconcentration) 1
w)
3.8 10-5
4.6 10-7
2.2 10-5
1.4 10-1
1.5 10-9
1.9 10-3
6.0 10-2
9.8 10-4
5.3 10-5
13.2
-2.8
-13.1
Al
Fe
Si
K
Mg
Ca
Na
S(VI)
TIC
pH
pe
log PCO2 (atm)
665
666
34
667
Table 3: Mineralogical composition of the Callovo-Oxfodian argillites
Minerals name in the
Structural formula
DDB
Calcite
Celestite
Chlorite(Cca-2)
Dolomite
Illite(IMt2)
Microcline
Montmorillonite(HcCa)
Pyrite
Quartz(alpha)
Siderite
Porosity
CaCO3
SrSO4
(Mg2.964Fe1.927Al1.116Ca0.011)(Si2.633Al1.367)O10(OH)8
CaMg(CO3)2
(Na0.044K0.762)(Si3.387Al0.613)(Al1.427Fe0.376Mg0.241)O10(OH)2
K(AlSi3)O8
Ca0.3Mg0.6Al1.4Si4O10(OH)2
FeS2
SiO2
FeCO3
0.18
668
669
35
Volume
fraction
(%)
0.23
0.01
0.02
0.04
0.33
0.03
0.08
0.01
0.25
0.01
mol L-1
of
solution
27.99
0.69
0.41
2.76
10.77
1.37
2.75
1.06
50.86
1.10
670
671
Table 4: Composition of the Callovo-Oxfordian porewater (PHREEQC calculation).
Calculated COx
Gaucher et al.
In situ
Element
chemical composition
(2009)
concentrations*
at 25°C (mol kg-1w)
(mol kg-1w)
(mol kg-1w)
-8
-8
Al
8.4 10
3.0 10
-Fe
6.8 10-5
4.7 10-5
(1.5 ± 1.1) 10-5
Si
1.8 10-4
1.8 10-4
1.4 10-4
-4
-4
Sr
2.3 10
2.1 10
(2.5 ± 0.2) 10-4
-4
-3
K
5.1 10
1.0 10
(9.0 ± 3.0) 10-4
Mg
5.1 10-3
5.4 10-3
(5.9 ± 1.1) 10-3
-3
-3
Ca
7.6 10
8.5 10
(7.6 ± 1.4) 10-3
-2
-2
Na
4.0 10
4.3 10
(5.6 ± 0.4) 10-2
-2
-2
Cl
4.1 10
4.1 10
(4.1 ± 1.1) 10-2
-2
-2
S(VI)
1.1 10
1.5 10
(1.9 ± 0.4) 10-2
TIC
3.8 10-3
2.4 10-3
(4.2 ± 0.6) 10-3
pH
7,0
7.2
7.2 ± 0.2
pe
-2.8
-3.0
-3.4 ± 0.5
log PCO2 (atm)
-1.8
-2.2
-1,9
* Values extracted from Vinsot et al. (2008).
672
36
673
Table 5: Transport parameters
Porosity
Effective diffusion
Material
(%)
coefficient (m2 s-1)
COx claystone
18
2.6 10-11
Concrete plug
13
9 10-12
674
675
37
676
677
678
Table 6: Thermodynamic constants (25°C) and molar volume of minerals considered in
simulations. Thermodynamic data were extracted from THERMODDEM
(http://thermoddem.brgm.fr).
Molar
Minerals
Structural formula
log K25
volume
(cm3 mol-1)
Amorphous silica
SiO2
-2.70
29.00
Brucite
Mg(OH)2
17.11
24.63
Clinoptilolite(Ca)
Ca0.55(Si4.9Al1.1)O12:3.9H2O
-2.11
209.66
CSH(1.6)
Ca1.60SiO3.6:2.58H2O
28.00
84.68
CSH(1.2)
Ca1.2SiO3.2:2.06H2O
19.30
71.95
CSH(0.8)
Ca0.8SiO2.8:1.54H2O
11.05
59.29
C3FH6
Ca3Fe2(OH)12
72.38
154.50
Ettringite
Ca6Al2(SO4)3(OH)12:26H2O
57.01
710.32
Ferrihydrite(2L)
Fe(OH)3
3.40
34.36
Gypsum
CaSO4:2H2O
-4.60
74.69
Hydrotalcite
Mg4Al2O7:10H2O
73.76
227.36
Fe(OH)2
Fe(OH)2
12.85
24.48
Magnetite(am)
Fe3O4
14.59
44.52
Monocarboaluminate
Ca4Al2CO9:10.68H2O
80.57
261.96
MordeniteB(Ca)
Ca0.515Al1.03Si4.97O12:3.1H2O
-2.92
209.80
Portlandite
Ca(OH)2
22.81
33.06
Pyrite
FeS2
-23.59
23.94
Pyrrhotite
FeS
-3.68
18.20
Saponite(Ca)
Ca0.17Mg3Al0.34Si3.66O10(OH)2
28.07
138.84
Saponite(FeCa)
Ca0.17Mg2FeAl0.34Si3.66O10(OH)2
26.54
139.96
Straetlingite
Ca2Al2SiO2(OH)10:2.5H2O
49.67
215.63
Calcite
CaCO3
1.85
36.93
Celestite
SrSO4
-6.62
46.25
(Mg2.964Fe1.927Al1.116Ca0.011)(Si2.633Al1.367)
Chlorite(CCa-2)
61.33
211.92
O10(OH)8
Dolomite
CaMg(CO3)2
3.53
64.37
Gibbsite(am)
Al(OH)3
10.58
31.96
(Na0.044K0.762)(Si3.387Al0.613)(Al1.427Fe0.376
Illite(IMt2)
11.52
139.18
Mg0.241)O10(OH)2
Microcline
K(AlSi3)O8
0.04
108.74
Montmorillonite(HcCa)
Ca0.3Mg0.6Al1.4Si4O10(OH)2
7.28
132.48
Quartz(alpha)
SiO2
-3.74
22.69
Siderite
FeCO3
-0.27
29.38
679
38
680
681
Table 7: Kinetic parameters considered for dissolution reactions. Parameters describe the pH dependence and deviation from equilibrium
dissolution rates (see appendix B in Xu et al., 2004).
nu
OH 
Mineral
A
k 25
E anu
k 25H 
E aH 
k 25
E aOH 
n OH 
nH
2 -1
(m g )
(mol m-2 s-1) (kJ mol-1) (mol m-2 s-1)
(kJ mol-1) (mol m-2 s-1)
(kJ mol-1)
Illite(IMt2) (1)
30
3.3 10-17
35
9.8 10-12
0.52
36
3.1 10-12
0.38
48
(2)
-15
-11
-12
Montmorillonite(HcCa)
8.5
9.3 10
63
5.3 10
0.69
54
2.9 10
0.34
61
Chlorite(CCa-2) (3)
0.003
6.4 10-17
16
8.2 10-9
0.28
17
6.9 10-9
0.34
16
Quartz(alpha) (4)
0.05
6.4 10-14
77
---1.9 10-10
0.34
80
Celestite(5)
40
2.2 10-8
34
1.4 10-6
0.10
33
---Calcite(6)
0.7
1.6 10-6
24
5.0 10-1
1.00
14
---Dolomite(7)
0.1
1.1 10-8
31
2.8 10-4
0.61
46
---Siderite(8)
2.7
2.1 10-9
56
5.9 10-6
0.60
56
---Gibbsite(9)
1
-----3.1 10-6
1
48
Microcline(10)
0.1
1.0 10-14
31
1.7 10-11
0.27
31
1.4 10-10
0.35
31
682
Kinetic data extracted from Marty et al. (accepted)
39
of the
θ
η
1
0.17
1
1
0.49
1
0.16
1
1
0.09
1
10.34
1
1
2.06
1
2.10
1
1
2.35
683
684
Table 8: Kinetic parameters considered for precipitation reactions.
Additional mechanism
k 25pre.
E apre.
A
i
i
Mineral
E ai
k 25
ni
-1
(m2 g-1) (mol m-2 s-1)
(kJ mol )
(kJ mol-1)
(mol m-2 s-1)
Illite(IMt2)(11)
30
6.2 10-14
66
----Montmorillonite(HcCa)
8.5
0
0
----Chlorite(CCa-2)
0.003
0
0
----Quartz(alpha) (12)
0.05
3.2 10-12
50
----Celestite(13)
40
5.1 10-8
34
----Calcite(14)
0.7
1.8 10-7
66
HCO31.9 10-3
1.6
67
Dolomite(15)
0.1
9.5 10-15
103
----Siderite(16)
2.7
1.6 10-11
108
----(17)
-6
Gibbsite
1
0
0
OH
3.1 10
1
48
Microcline
0.1
0
0
----Kinetic data extracted from Marty et al. (accepted)
41
θ
η
0.06
1
1
4.58
0.5
0.5
1
1
1
1
1.68
1
1
0.54
2
2
1
1
1
1
685
Table 9: Selectivity coefficients for exchange reactions. Extracted from Gaucher et al. (2009)
Illite/smectite model
(Gaucher et al., 2009)
log k exNa / Ca
log k exNa / Mg
log k exNa / K
0.7
0.7
1.2
686
687
42
688
Table 10: Simplifications adopted in order to perform the proposed benchmark exercise.
Resolution
Phases under kinetic
Code
Remarks
scheme
control
PHREEQC2
SNIA
Carbonates,
Microcline,
Chlorite(Cca-2),
gibbsite and
iPHREEQC3 &
Illite(IMt2),
Quartz(alpha),
celestite
external
SNIA
Montmorillonite(HcCa)
processed
at the
transport
local
equilibrium
module
Celestite, Chlorite(Cca-2),
Carbonates
SNIA (but SIA
Gibbsite(am), Illite(IMt2) ,
TOUGHREACT
processed at the
possible)
Montmorillonite(HcCa),
local equilibrium
Quartz(alpha), Microcline
Acidic
All minerals listed in tables
mechanism
6, 7 and 8. Equilibrium
CRUNCH
Global implicit
(unnecessary)
phases treated as quasion reaction rate
equilibrium reactions.
not considered
SIA
Celestite, Chlorite(Cca-2),
Illite(IMt2), Microcline
Montmorillonite(HcCa),
Quartz (alpha), Siderite
Calcite
processed at
local equilibrium
MIN3P
DSA
All minerals listed in tables
6, 7 and 8. Equilibrium
phases treated as quasiequilibrium reactions.
Reactive
surface area of
calcite
decreased by 2
orders of
magnitude
ORCHESTRA
SNIA (but SIA
possible)
Microcline, Chlorite(Cca-2),
Illite(IMt2), Quartz(alpha),
Montmorillonite(HcCa)
(Identical to PHREEQC)
Carbonates
processed at
local equilibrium
HYTEC
689
690
691
SNIA: Sequential non-iterative approach
SIA: Sequential iterative approach
DSA: direct substitution approach
43
692
FIGURES CAPTIONS
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
Figure 1: System modelled with a heterogeneous mesh with a spatial resolution of 0.05
m at the concrete/claystone interface
Figure 2: Cl concentration profiles at 1,000 years obtained with TOUGHREACT,
PHREEQC2, iPHREEQC3, CRUNCH, HYTEC, ORCHESTRA and MIN3P-THCm (case
1).
Figure 3: Exchanger compositions at 1,000 years obtained with TOUGHREACT,
PHREEQC2, iPHREEQC3, CRUNCH, HYTEC, ORCHESTRA and MIN3P-THCm (case
1).
Figure 4: Mineralogical and pH changes obtained with TOUGHREACT after 10,000
years of concrete/claystone interactions (case 2).
Figure 5: Mineralogical and pH changes obtained with PHREEQC2 after 10,000 years
of concrete/claystone interactions (case 2).
Figure 6: Mineralogical and pH changes obtained with iPHRREQC3 after 10,000 years
of concrete/claystone interactions (case 2).
Figure 7: Mineralogical and pH changes obtained with CRUNCH after 10,000 years of
concrete/claystone interactions (case 2).
Figure 8: Mineralogical and pH changes obtained with HYTEC after 10,000 years of
concrete/claystone interactions (case 2).
Figure 9: Mineralogical and pH changes obtained with ORCHESTRA after 10,000 years
of concrete/claystone interactions (case 2).
Figure 10: Mineralogical and pH changes obtained with MIN3P-THCm after 10,000
years of concrete/claystone interactions (case 2).
Figure 11: Si, Al, Ca, Na, Mg, K, Cl and S(6) concentrations obtained with
TOUGHREACT, PHREEQC2, iPHREEQC3, CRUNCH, HYTEC, ORCHESTRA and
MIN3P-THCm after 10,000 years of concrete/claystone interactions.
Figure 12: Mineralogical and pH changes obtained with TOUGHREACT after 10,000
years of concrete/claystone interactions (case 3).
Figure 13: Mineralogical and pH changes obtained with PHREEQC2 after 10,000 years
of concrete/claystone interactions (case 3).
Figure 14: Mineralogical and pH changes obtained with iPHREEQC3 after 10,000 years
of concrete/claystone interactions (case 3).
Figure 15: Mineralogical and pH changes obtained with CRUNCH after 10,000 years of
concrete/claystone interactions (case 3).
Figure 16: Mineralogical and pH changes obtained with HYTEC after 10,000 years of
concrete/claystone interactions (case 3).
Figure 17: Mineralogical and pH changes obtained with ORCHESTRA after 10,000
years of concrete/claystone interactions (case 3).
Figure 18: Mineralogical and pH changes obtained with MIN3P-THCm after 10,000
years of concrete/claystone interactions (case 3).
44
733
734
735
Figure 1: System modelled with a heterogeneous mesh with a spatial resolution
736
of 0.05 m at the concrete/claystone interface
737
45
1,000 years
PHREEQC2
iPHREEQC3
TOUGHREACT
CRUNCH
ORCHESTRA
MIN3P-THCm
HYTEC
Cl concentration (mol L-1)
0.05
0.04
0.03
0.02
0.01
0
0
2
4
6
8 34
36
38
40
42
Distance (m)
738
739
Figure 2: Cl concentration profiles at 1,000 years obtained with TOUGHREACT,
740
PHREEQC2, iPHREEQC3, CRUNCH, HYTEC, ORCHESTRA and MIN3P-THCm (case
741
1).
742
46
743
744
Figure 3: Exchanger compositions at 1,000 years obtained with TOUGHREACT,
745
PHREEQC2, iPHREEQC3, CRUNCH, HYTEC, ORCHESTRA and MIN3P-THCm (case
746
1).
747
47
748
10 000 years - TOUGHREACT
1.2
14
1
0.8
0.6
10
0.4
8
0.2
0
6
1
2
4
3
Distance (m)
749
5
pH
Volume fraction
12
Amorphous silica
Brucite
Calcite
Celestite
Chlorite(Cca-2)
Clinoptilolite(Ca)
CSH(1.6)
CSH(1.2)
CSH(0.8)
C3FH6
Dolomite
Ettringite
Fe(OH)2
Ferrihydrite(2L)
Gibbsite(am)
Gypsum
Hydrotalcite
Illite(IMt2)
Magnetite(am)
Microcline
Monocarboaluminate
Montmorillonite(HcCa)
MordeniteB(Ca)
Portlandite
Pyrite
Pyrrhotite
Quartz(alpha)
Saponite(Ca)
Saponite(FeCa)
Siderite
Straetlingite
750
Figure 4: Mineralogical and pH changes obtained with TOUGHREACT after 10,000
751
years of concrete/claystone interactions (case 2).
752
48
10 000 years - PHREEQC2
1.2
14
1
0.8
0.6
10
0.4
8
0.2
0
6
1
2
3
4
Distance (m)
753
5
pH
Volume fraction
12
Amorphous silica
Brucite
Calcite
Celestite
Chlorite(Cca-2)
Clinoptilolite(Ca)
CSH(1.6)
CSH(1.2)
CSH(0.8)
C3FH6
Dolomite
Ettringite
Fe(OH)2
Ferrihydrite(2L)
Gibbsite(am)
Gypsum
Hydrotalcite
Illite(IMt2)
Magnetite(am)
Microcline
Monocarboaluminate
Montmorillonite(HcCa)
MordeniteB(Ca)
Portlandite
Pyrite
Pyrrhotite
Quartz(alpha)
Saponite(Ca)
Saponite(FeCa)
Siderite
Straetlingite
754
Figure 5: Mineralogical and pH changes obtained with PHREEQC2 after 10,000 years
755
of concrete/claystone interactions (case 2).
756
49
10 000 years - iPHREEQC3
1.2
14
1
0.8
0.6
10
0.4
8
0.2
0
6
1
2
3
4
Distance (m)
757
5
pH
Volume fraction
12
Amorphous silica
Brucite
Calcite
Celestite
Chlorite(Cca-2)
Clinoptilolite(Ca)
CSH(1.6)
CSH(1.2)
CSH(0.8)
C3FH6
Dolomite
Ettringite
Fe(OH)2
Ferrihydrite(2L)
Gibbsite(am)
Gypsum
Hydrotalcite
Illite(IMt2)
Magnetite(am)
Microcline
Monocarboaluminate
Montmorillonite(HcCa)
MordeniteB(Ca)
Portlandite
Pyrite
Pyrrhotite
Quartz(alpha)
Saponite(Ca)
Saponite(FeCa)
Siderite
Straetlingite
758
Figure 6: Mineralogical and pH changes obtained with iPHRREQC3 after 10,000 years
759
of concrete/claystone interactions (case 2).
760
50
10 000 years - CRUNCH
1.2
14
1
0.8
0.6
10
0.4
8
0.2
0
6
1
2
3
4
Distance (m)
761
5
pH
Volume fraction
12
Amorphous silica
Brucite
Calcite
Celestite
Chlorite(Cca-2)
Clinoptilolite(Ca)
CSH(1.6)
CSH(1.2)
CSH(0.8)
C3FH6
Dolomite
Ettringite
Fe(OH)2
Ferrihydrite(2L)
Gibbsite(am)
Gypsum
Hydrotalcite
Illite(IMt2)
Magnetite(am)
Microcline
Monocarboaluminate
Montmorillonite(HcCa)
MordeniteB(Ca)
Portlandite
Pyrite
Pyrrhotite
Quartz(alpha)
Saponite(Ca)
Saponite(FeCa)
Siderite
Straetlingite
762
Figure 7: Mineralogical and pH changes obtained with CRUNCH after 10,000 years of
763
concrete/claystone interactions (case 2).
764
51
10 000 years - HYTEC
14
1.2
1
0.8
10
0.6
0.4
8
0.2
6
0
1
2
3
4
Distance (m)
765
5
pH
Volume fraction
12
Amorphous silica
Brucite
Calcite
Celestite
Chlorite(Cca-2)
Clinoptilolite(Ca)
CSH(1.6)
CSH(1.2)
CSH(0.8)
C3FH6
Dolomite
Ettringite
Fe(OH)2
Ferrihydrite(2L)
Gibbsite(am)
Gypsum
Hydrotalcite
Illite(IMt2)
Magnetite(am)
Microcline
Monocarboaluminate
Montmorillonite(HcCa)
MordeniteB(Ca)
Portlandite
Pyrite
Pyrrhotite
Quartz(alpha)
Saponite(Ca)
Saponite(FeCa)
Siderite
Straetlingite
766
Figure 8: Mineralogical and pH changes obtained with HYTEC after 10,000 years of
767
concrete/claystone interactions (case 2).
768
52
10 000 years - ORCHESTRA
1.2
14
1
0.8
0.6
10
0.4
8
0.2
0
6
1
2
3
4
Distance (m)
769
5
pH
Volume fraction
12
Amorphous silica
Brucite
Calcite
Celestite
Chlorite(Cca-2)
Clinoptilolite(Ca)
CSH(1.6)
CSH(1.2)
CSH(0.8)
C3FH6
Dolomite
Ettringite
Fe(OH)2
Ferrihydrite(2L)
Gibbsite(am)
Gypsum
Hydrotalcite
Illite(IMt2)
Magnetite(am)
Microcline
Monocarboaluminate
Montmorillonite(HcCa)
MordeniteB(Ca)
Portlandite
Pyrite
Pyrrhotite
Quartz(alpha)
Saponite(Ca)
Saponite(FeCa)
Siderite
Straetlingite
770
Figure 9: Mineralogical and pH changes obtained with ORCHESTRA after 10,000 years
771
of concrete/claystone interactions (case 2).
53
10 000 years - MIN3P-THCm
1.2
14
1
0.8
0.6
10
0.4
8
0.2
0
6
1
2
3
4
Distance (m)
772
5
pH
Volume fraction
12
Amorphous silica
Brucite
Calcite
Celestite
Chlorite(Cca-2)
Clinoptilolite(Ca)
CSH(1.6)
CSH(1.2)
CSH(0.8)
C3FH6
Dolomite
Ettringite
Fe(OH)2
Ferrihydrite(2L)
Gibbsite(am)
Gypsum
Hydrotalcite
Illite(IMt2)
Magnetite(am)
Microcline
Monocarboaluminate
Montmorillonite(HcCa)
MordeniteB(Ca)
Portlandite
Pyrite
Pyrrhotite
Quartz(alpha)
Saponite(Ca)
Saponite(FeCa)
Siderite
Straetlingite
773
Figure 10: Mineralogical and pH changes obtained with MIN3P-THCm after 10,000
774
years of concrete/claystone interactions (case 2).
775
54
776
777
Figure 11: Si, Al, Ca, Na, Mg, K, Cl and S(6) concentrations obtained with TOUGHREACT, PHREEQC2, iPHREEQC3,
778
CRUNCH, HYTEC, ORCHESTRA and MIN3P-THCm after 10,000 years of concrete/claystone interactions.
55
779
780
10 000 years - TOUGHREACT
1.2
14
1
0.8
0.6
10
0.4
8
0.2
0
6
1
2
3
4
Distance (m)
781
5
pH
Volume fraction
12
Amorphous silica
Brucite
Calcite
Celestite
Chlorite(Cca-2)
Clinoptilolite(Ca)
CSH(1.6)
CSH(1.2)
CSH(0.8)
C3FH6
Dolomite
Ettringite
Fe(OH)2
Ferrihydrite(2L)
Gibbsite(am)
Gypsum
Hydrotalcite
Illite(IMt2)
Magnetite(am)
Microcline
Monocarboaluminate
Montmorillonite(HcCa)
MordeniteB(Ca)
Portlandite
Pyrite
Pyrrhotite
Quartz(alpha)
Saponite(Ca)
Saponite(FeCa)
Siderite
Straetlingite
782
Figure 12: Mineralogical and pH changes obtained with TOUGHREACT after 10,000
783
years of concrete/claystone interactions (case 3).
784
56
10 000 years - PHREEQC2
1.2
14
1
0.8
0.6
10
0.4
8
0.2
0
6
1
2
3
4
Distance (m)
785
5
pH
Volume fraction
12
Amorphous silica
Brucite
Calcite
Celestite
Chlorite(Cca-2)
Clinoptilolite(Ca)
CSH(1.6)
CSH(1.2)
CSH(0.8)
C3FH6
Dolomite
Ettringite
Fe(OH)2
Ferrihydrite(2L)
Gibbsite(am)
Gypsum
Hydrotalcite
Illite(IMt2)
Magnetite(am)
Microcline
Monocarboaluminate
Montmorillonite(HcCa)
MordeniteB(Ca)
Portlandite
Pyrite
Pyrrhotite
Quartz(alpha)
Saponite(Ca)
Saponite(FeCa)
Siderite
Straetlingite
786
Figure 13: Mineralogical and pH changes obtained with PHREEQC2 after 10,000 years
787
of concrete/claystone interactions (case 3).
788
57
10 000 years - iPHREEQC3
1.2
14
1
0.8
0.6
10
0.4
8
0.2
0
6
1
2
3
4
Distance (m)
789
5
pH
Volume fraction
12
Amorphous silica
Brucite
Calcite
Celestite
Chlorite(Cca-2)
Clinoptilolite(Ca)
CSH(1.6)
CSH(1.2)
CSH(0.8)
C3FH6
Dolomite
Ettringite
Fe(OH)2
Ferrihydrite(2L)
Gibbsite(am)
Gypsum
Hydrotalcite
Illite(IMt2)
Magnetite(am)
Microcline
Monocarboaluminate
Montmorillonite(HcCa)
MordeniteB(Ca)
Portlandite
Pyrite
Pyrrhotite
Quartz(alpha)
Saponite(Ca)
Saponite(FeCa)
Siderite
Straetlingite
790
Figure 14: Mineralogical and pH changes obtained with iPHREEQC3 after 10,000 years
791
of concrete/claystone interactions (case 3).
792
58
10 000 years - CRUNCH
1.2
14
1
0.8
0.6
10
0.4
8
0.2
0
6
1
2
3
4
Distance (m)
793
5
pH
Volume fraction
12
Amorphous silica
Brucite
Calcite
Celestite
Chlorite(Cca-2)
Clinoptilolite(Ca)
CSH(1.6)
CSH(1.2)
CSH(0.8)
C3FH6
Dolomite
Ettringite
Fe(OH)2
Ferrihydrite(2L)
Gibbsite(am)
Gypsum
Hydrotalcite
Illite(IMt2)
Magnetite(am)
Microcline
Monocarboaluminate
Montmorillonite(HcCa)
MordeniteB(Ca)
Portlandite
Pyrite
Pyrrhotite
Quartz(alpha)
Saponite(Ca)
Saponite(FeCa)
Siderite
Straetlingite
794
Figure 15: Mineralogical and pH changes obtained with CRUNCH after 10,000 years of
795
concrete/claystone interactions (case 3).
796
59
10 000 years - HYTEC
1.4
14
1.2
12
0.8
10
0.6
0.4
8
0.2
0
6
1
2
3
4
Distance (m)
797
5
pH
Volume fraction
1
Amorphous silica
Brucite
Calcite
Celestite
Chlorite(Cca-2)
Clinoptilolite(Ca)
CSH(1.6)
CSH(1.2)
CSH(0.8)
C3FH6
Dolomite
Ettringite
Fe(OH)2
Ferrihydrite(2L)
Gibbsite(am)
Gypsum
Hydrotalcite
Illite(IMt2)
Magnetite(am)
Microcline
Monocarboaluminate
Montmorillonite(HcCa)
MordeniteB(Ca)
Portlandite
Pyrite
Pyrrhotite
Quartz(alpha)
Saponite(Ca)
Saponite(FeCa)
Siderite
Straetlingite
798
Figure 16: Mineralogical and pH changes obtained with HYTEC after 10,000 years of
799
concrete/claystone interactions (case 3).
800
60
10 000 years - ORCHESTRA
1.2
14
1
0.8
0.6
10
0.4
8
0.2
0
6
1
2
3
4
Distance (m)
801
5
pH
Volume fraction
12
Amorphous silica
Brucite
Calcite
Celestite
Chlorite(Cca-2)
Clinoptilolite(Ca)
CSH(1.6)
CSH(1.2)
CSH(0.8)
C3FH6
Dolomite
Ettringite
Fe(OH)2
Ferrihydrite(2L)
Gibbsite(am)
Gypsum
Hydrotalcite
Illite(IMt2)
Magnetite(am)
Microcline
Monocarboaluminate
Montmorillonite(HcCa)
MordeniteB(Ca)
Portlandite
Pyrite
Pyrrhotite
Quartz(alpha)
Saponite(Ca)
Saponite(FeCa)
Siderite
Straetlingite
802
Figure 17: Mineralogical and pH changes obtained with ORCHESTRA after 10,000
803
years of concrete/claystone interactions (case 3).
804
61
10 000 years - MIN3P-THCm
1.2
14
1
0.8
0.6
10
0.4
8
0.2
0
6
1
2
3
4
Distance (m)
805
5
pH
Volume fraction
12
Amorphous silica
Brucite
Calcite
Celestite
Chlorite(Cca-2)
Clinoptilolite(Ca)
CSH(1.6)
CSH(1.2)
CSH(0.8)
C3FH6
Dolomite
Ettringite
Fe(OH)2
Ferrihydrite(2L)
Gibbsite(am)
Gypsum
Hydrotalcite
Illite(IMt2)
Magnetite(am)
Microcline
Monocarboaluminate
Montmorillonite(HcCa)
MordeniteB(Ca)
Portlandite
Pyrite
Pyrrhotite
Quartz(alpha)
Saponite(Ca)
Saponite(FeCa)
Siderite
Straetlingite
806
Figure 18: Mineralogical and pH changes obtained with MIN3P-THCm after 10,000
807
years of concrete/claystone interactions (case 3).
62