On the benchmarking of reactive transport codes

Proceedings Karlsruhe, March 2014

TRePro III 2014 is the sixth meeting in the series of the Karlsruhe Geochemical Workshops. The first workshop was held in 1997 at the Forschungszentrum Karlsruhe. Its main topic was “Geochemical modelling – radiotoxic and chemical toxic substances in natural aquatic systems”. The second meeting was held in 1999 in Speyer focusing on “Mineral/water interactions close to equilibrium”. TRePro 2002 held in 2002 in Karlsruhe dealt with “Modelling of coupled transport reaction processes”. SoPro 2005 held again in Karlsruhe treated sorption processes on oxide and carbonate minerals. TRePro III 2014 takes up the idea of TRePro 2002 and TRePro II 2009. The intention of TRePro III 2014 is to serve as a forum for discussing the latest developments in modelling of coupled reaction and transport processes. The main idea of the workshop is to enhance the discussion between experimentalists and modellers from different domains. Without several persons at INE the workshop would not have been possible: Special thanks go to Bernhard Kienzler for initiating and to Horst Geckeis for constantly supporting the organisation of the workshop. The helpful input from various departments at KIT who were involved in the organisation is gratefully acknowledged. Finally we thank all contributors to this book of abstracts.

06.-07. March 2014 Olivier Bildstein, Horst-Jürgen Herbert, Florian Huber, Johannes Lützenkirchen, Vanessa Montoya, Wilfried Pfingsten,

List of content WORKSHOP AGENDA ........................................................................................................................................ I ABSTRACTS ...................................................................................................................................................... 0 MINERAL REACTIVITY AT CLAY/CONCRETE INTERFACE: INPUT AND NEED OF FUTURE RESEARCH ARISE FROM BOTH REACTIVE TRANSPORT MODELLING AND EXPERIMENTATIONS .............................................................. 1 FRANCIS CLARET ..................................................................................................................................................... 1 REACTIVE TRANSPORT MODELLING OF OXYGEN GAS DIFFUSION AND CONSUMPTION IN A DISPOSAL CELL OF RADIOACTIVE WASTE ....................................................................................................................................... 5 LAURENT DE WINDT, JEROME CORVISIER AND FRANÇOIS MARSAL ................................................................................... 5 HOW CAN THE RADIONUCLIDE DIFFUSION BE AFFECTED BY POROSITY CHANGES?.......................................... 9 I. FATNASSI, S. SAVOYE, P. ARNOUX, P. GOUZE, O. BILDSTEIN, V. DETILLEUX AND C. WITTEBROODT ..................................... 9 ATTACHMENT OF PARTICLES AT SURFACES. THEORY, EXPERIMENT AND APPLICATION ................................. 17 NIKOLA KALLAY..................................................................................................................................................... 17 TRANSPORT AND RETENTION OF MANIFACTURED NANOPARTICLES IN WATER-SATURATED POROUS MEDIA UNDER DIFFERENT CONDITIONS: MEASUREMENTS AND MODELING ............................................................. 19 IVAN TOLONI, FRANÇOIS LEHMANN AND PHILIPPE ACKERER .......................................................................................... 19 A NEW STOCHASTIC APPROACH FOR THE SIMULATION OF FOULING: FROM SINGLE PARTICLE DEPOSITION TO CLOGGING ...................................................................................................................................................... 21 CHRISTOPHE HENRY AND JEAN-PIERRE MINIER ........................................................................................................... 21 ION ASSOCIATION AND HYDRATION IN AQUEOUS ELECTROLYTE SOLUTIONS – THE VIEW OF DIELECTRIC SPECTROSCOPY .............................................................................................................................................. 27 RICHARD BUCHNER AND GLENN HEFTER .................................................................................................................... 27 UP-SCALING OF DIFFUSION COEFFICIENTS IN SATURATED AND UNSATURATED CLAYS .................................. 31 THOMAS GIMMI AND SERGEY V. CHURAKOV ............................................................................................................. 31 REACTIVE-TRANSPORT PARAMETERS ADJUSTMENT AND SENSITIVITY ANALYSIS: APPLICATION TO CONCRETE CARBONATION ............................................................................................................................................... 37 OLIVIER BILDSTEIN, PASCAL THOUVENOT, AMANDINE MARREL, ISABELLE MUNIER AND BENOIT COCHEPIN ............................ 37 REACTIVE TRANSPORT MODELLING OF GEOTHERMAL RESERVOIRS ............................................................... 41 REMIS CANNEPIN AND HELGE C. MOOG .................................................................................................................... 41 ON THE UPSCALING OF CHEMICAL TRANSPORT IN FRACTURED ROCK ........................................................... 43 VLADIMIR CVETKOVIC ............................................................................................................................................ 43 THE NEED FOR PERFORMANCE WHEN SIMULATING REACTIVE TRANSPORT IN HETEROGENEOUS MEDIA ..... 45 VINCENT LAGNEAU AND OLIVIER REGNAULT ............................................................................................................... 45 REACTIVE TRANSPOR MODELING OF THE INTERACTIONS OF CORROSION PRODUCTS AND BENTONITE: LONGTERM GEOCHEMICAL EVOLUTION AND H2 GAS GENERATION ....................................................................... 51 JAVIER SAMPER, ACACIA NAVES, LUIS MONTENEGRO, ALBA MON AND TIANFU XU ........................................................... 51 3+

A WRAP UP OF MODELING SORPTION PROCESSES OF EU UNDER VARYING GEOCHEMICAL CONDITIONS ... 57 S. BRITZ, U. NOSECK, V. BRENDLER, W. DURNER AND D. ZACHMANN ............................................................................ 57 COMPUTATIONAL CHALLENGES FOR REACTIVE TRANSPORT MODELLING: INTERFACING AND HIGH PERFORMANCE COMPUTING ......................................................................................................................... 65 JORGE MOLINERO ................................................................................................................................................. 65

ALGORITHM FOR FLUID VELOCITY FIELD QUANTIFICATION FROM IMAGE SEQUENCES IN COMPLEX GEOMATERIALS .............................................................................................................................................. 67 NICO KORN AND JOHANNA LIPPMANN-PIPKE ............................................................................................................. 67 COUPLING GROUNDWATER FLOW AND TRANSPORT IN FEFLOW WITH ENVIRONMENTAL GEOCHEMISTRY IN PHREEQC ........................................................................................................................................................ 71 LAURIN WISSMEIER, THOMAS KÄMPFER AND STEFAN WILHELM .................................................................................... 71 WATER-SOLID INTERACTIONS AT THE PORE SCALE ........................................................................................ 73 LIONEL MERCURY, KIRILL SHMULOVICH, ISABELLE BERGONZI AND JEAN-MICHEL MATRAY .................................................. 73 REACTIVE TRANSPORT MODELLING APPLICATIONS AND DEVELOPMENTS FOR PROJECTS IN THE OIL AND GAS INDUSTRY....................................................................................................................................................... 79 TIM J. TAMBACH, HENNING PETERS, CHRIS PENTLAND AND JEROEN R. SNIPPE ................................................................. 79 A SEQUENTIAL T-H-M-C ALGORITHM TO SIMULATE PHYSICAL LINKS BETWEEN STRESS AND GEOCHEMICAL ENVIRONMENT EVOLUTION ........................................................................................................................... 85 ALAIN DIMIER, JÉRÔME GAOMBALET AND AHMED BOUGACHA ...................................................................................... 85 EXPERIMENTAL BENCHMARKS FOR THE VERIFICATION AND VALIDATION OF REACTIVE TRANSPORT CODES 87 JENNA POONOOSAMY, GEORG KOSAKOWSKI AND LUC VAN LOON.................................................................................. 87 POSTERS ......................................................................................................................................................... 89 THE EFFECTS OF POROSITY CLOGGING ON THE TRANSPORT PROPERTIES OF POROUS MATERIALS UNDER GEOCHEMICAL PERTURBATIONS: EXPERIMENTAL APPROACH AND MODELING............................................. 91 A. CHAGNEAU, F. CLARET, B. MADÉ, M. WOLF, F. ENZMANN AND T. SCHÄFER ............................................................... 91 MODELING LONG-TERM LEACHING EXPERIMENTS OF FULL SCALE CEMENTED WASTES: EFFECT OF SOLUTION COMPOSITION ON DIFFUSION ....................................................................................................................... 95 C. BORKEL, V. MONTOYA AND B. KIENZLER ............................................................................................................... 95 DEPOSITION OF COLLOIDS ON GRANITE SURFACE: INFLUENCE OF MACROSCOPIC SURFACE HETEROGENEITIES AND EU(III) CONCENTRATION ........................................................................................................................ 99 GOPALA KRISHNA DARBHA, JOHANNES LÜTZENKIRCHEN, CORNELIUS FISCHER AND THORSTEN SCHÄFER ................................ 99 REACTIVE TRANSPORT CALCULATIONS ON POROSITY CHANGES AT MATERIAL INTERFACES ........................ 101 GEORG KOSAKOWSKI ........................................................................................................................................... 101 INTERACTION OF TRIVALENT METAL IONS WITH ALUMINIUM(HYDR)OXIDES .............................................. 103 T. KUPCIK, N. HUITTINEN, TH. RABUNG, J. LÜTZENKIRCHEN, H. GECKEIS AND TH. FANGHÄNEL ......................................... 103 BENCHMARK FOR NONLINEAR SORPTION PROCESSES OF CS MIGRATION THROUGH OPALINUS CLAY USING A SINGLE SPECIES (COMSOL) AND MULTI-SPECIES (MCOTAC) REACTIVE TRANSPORT MODEL ......................... 107 WILFRIED PFINGSTEN AND ANDREAS JAKOB ............................................................................................................. 107 UNCERTAINTY PROPAGATION OF LINEAR-FREE-ENERGY-RELATIONSHIP SOPRTION PARAMETERS TO THEIR APPLICATION IN REACTIVE TRANSPORT CALCULATIONS .............................................................................. 115 ALBERT RIERA AND WILFRIED PFINGSTEN................................................................................................................. 115 SPECIATION OF NEPTUNIUM ALONG DIFFUSION PATHWAYS IN OPALINUS CLAY USING MICRO-XAFS AND MICRO-XRF................................................................................................................................................... 119 JONATHAN ROSEMANN, SAMER AMAYRI, UGRAS KAPLAN, JAKOB DREBERT, DANIEL GROLIMUND, TOBIAS REICH ................. 119 RADIONUCLIDE TRANSPORT IN A NATURAL SHEARZONE AT THE GRIMSEL TEST SITE (GTS, SWITZERLAND) 125 T. SCHÄFER, I. BLECHSCHMIDT, M. BOUBY, S. BÜCHNER, J. BRENDLÉ, G. DARBHA, H. GECKEIS, T. KUPCIK, R. GÖTZ, W. HAUSER, S. HECK, F. HUBER, M. LAGOS AND A. MARTIN ....................................................................................................... 125

SORPTION STUDIES OF ACTINIDES/LANTHANIDES ONTO CLAY MINERALS UNDER SALINE CONDITIONS ...... 129 ANDREAS SCHNURR, RÉMI MARSAC, THOMAS RABUNG, JOHANNES LÜTZENKIRCHEN AND HORST GECKEIS .......................... 129 TC(VII) IMMOBILIZATION ON GRANITIC ROCKS FROM ÄSPÖ (SWEDEN) AND NIZHNEKANSKY MASSIF (RUSSIA) ..................................................................................................................................................................... 133 YURY TOTSKIY, FLORIAN HUBER, DIETER SCHILD, THORSTEN SCHÄFER, STEPAN KALMYKOV AND HORST GECKEIS .................. 133

Workshop Agenda 18.30

05.03.2014 Coming Together @ Kofflers Heuriger/ Rüppur 06.03.2014

8.00

Workshop Registration

8.30

Welcome Address

B. Kienzler

Session A Chair: O. Bildstein 8.40 – 9.25

MINERAL REACTIVITY AT CLAY/CONCRETE INTERFACE : INPUT AND NEED OF FUTURE RESEARCH ARISE FROM BOTH REACTIVE TRANSPORT MODELLING AND EXPERIMENTATIONS

F. Claret (inv.)

9.25 – 9.50

REACTIVE TRANSPORT MODELLING OF OXYGEN GAS DIFFUSION AND CONSUMPTION IN A DISPOSAL CELL OF RADIOACTIVE WASTE

L. De Windt

9.50 – 10.15

HOW CAN THE RADIONUCLIDE DIFFUSION BE AFFECTED BY POROSITY CHANGES?

I. Fatnassi

10.15 – 10.45

Coffee Break Session B Chair: T. Schäfer

10.45 – 11.30

ATTACHMENT OF PARTICLES AT SURFACES. THEORY, EXPERIMENT AND APPLICATION

N. Kallay (inv.)

11.30 – 11.55

TRANSPORT AND RETENTION OF MANIFACTURED NANOPARTICLES IN WATERSATURATED POROUS MEDIA UNDER DIFFERENT CONDITIONS: MEASUREMENTS AND MODELING

I. Toloni

11.55 – 12.20

A NEW STOCHASTIC APPROACH FOR THE SIMULATION OF FOULING: FROM SINGLE PARTICLE DEPOSITION TO CLOGGING

C. Henry

12.30 – 13.30

Lunch

I

Session C Chair: M. Altmaier 13.30 – 14.15

ION ASSOCIATION AND HYDRATION IN AQUEOUS ELECTROLYTE SOLUTIONS – THE VIEW OF DIELECTRIC SPECTROSCOPY

R. Buchner (inv.)

14.15 – 14.40

UP-SCALING OF DIFFUSION COEFFICIENTS IN SATURATED AND UNSATURATED CLAYS

T. Gimmi

14.40 – 15.05

REACTIVE-TRANSPORT PARAMETERS ADJUSTMENT AND SENSITIVITY ANALYSIS: APPLICATION TO CONCRETE CARBONATION

O. Bildstein

15.05 – 15.30

REACTIVE TRANSPORT MODELLING OF GEOTHERMAL RESERVOIRS

R. Cannepin

15.30 – 16.00

Coffee Break

Session D Chair: H. Moog 16.00– 16.45

ON THE UPSCALING OF CHEMICAL TRANSPORT IN FRACTURED ROCK

V. Cvetkovic (inv.)

16.45 – 17.10

THE NEED FOR PERFORMANCE WHEN SIMULATING REACTIVE TRANSPORT IN HETEROGENEOUS MEDIA

V. Lagneau

17.10 – 17.35

REACTIVE TRANSPOR MODELING OF THE INTERACTIONS OF CORROSION PRODUCTS AND BENTONITE: LONG-TERM GEOCHEMICAL EVOLUTION AND H2 GAS GENERATION

J. Samper

17.35 – 18.00

A WRAP UP OF MODELING SORPTION PROCESSES OF Eu3+ UNDER VARYING GEOCHEMICAL CONDITIONS

S. Britz

19.00

Poster Session with Dinner 07.03.2014 Session E Chair: G. Kosakowski

8.30 – 9.15

II

COMPUTATIONAL CHALLENGES FOR REACTIVE TRANSPORT MODELLING: INTERFACING AND HIGH PERFORMANCE COMPUTING

J. Molinero (inv.)

9.15 – 9.40

ALGORITHM FOR FLUID VELOCITY FIELD QUANTIFICATION FROM IMAGE SEQUENCES IN COMPLEX GEOMATERIALS

N. Korn

9.40 – 10.05

COUPLING GROUNDWATER FLOW AND TRANSPORT IN FEFLOW WITH ENVIRONMENTAL GEOCHEMISTRY IN PHREEQC

L. Wissmeier

10.05 – 10.30

REACTIVE TRANSPORT MODELLING APPLICATIONS AND DEVELOPMENTS FOR PROJECTS IN THE OIL AND GAS INDUSTRY

T. Tambach

10.30 – 11.00

Coffee Break

Session F Chair: W. Pfingsten 11.00 – 11.45

WATER-SOLID INTERACTIONS AT THE PORE SCALE

L. Mercury (inv.)

11.45 – 12.10

A SEQUENTIAL T-H-M-C ALGORITHM TO SIMULATE PHYSICAL LINKS BETWEEN STRESS AND GEOCHEMICAL ENVIRONMENT EVOLUTION

A. Dimier

12.10 – 12.35

EXPERIMENTAL BENCHMARKS FOR THE VERIFICATION AND VALIDATION OF REACTIVE TRANSPORT CODES

J. Poonoosamy

12.35

End of Workshop Lunch

III

Abstracts

TRePro III – Workshop on Modelling of Coupled Reactive and Transport Processes, Karlsruhe, 05-07.03.2014

MINERAL REACTIVITY AT CLAY/CONCRETE INTERFACE: INPUT AND NEED OF FUTURE RESEARCH ARISE FROM BOTH REACTIVE TRANSPORT MODELLING AND EXPERIMENTATIONS Francis Claret BRGM, 3 avenue Claude Guillemin BP 36009, 45060 Orléans Cedex 2-France *Corresponding author: [email protected] Introduction In many engineered concepts dealing with deep repository for radioactive wastes, concrete will be used to build access structures, galleries as well as vaults and waste packages for Intermediate Level Wastes (ILW). As the host rock can be clay rock (e.g. Callovo-Oxfordian layer in the east part of France, Opalinus Clay) concretes will be in contact will clayish materials. In addition, as swelling clays are planned to be used as sealing materials, another interface exist in the repository with clay materials such as bentonite. Due to the chemical disequilibrium between concrete and clays, chemical reaction can modify both chemical and physical properties of these materials (i.e. mineral phases, diffusion coefficient, …). In order to have the knowledge to assess the long term behaviour of concrete/clays interfaces and the evolution of the properties with time, (short term) experiments and (long term) predictive modelling have to be performed. Simulations based on pure thermodynamic model or based on thermodynamic/kinetic model led to the same major result: porosity clogging with by-products coming from cement /clay chemical interaction and re-precipitation at the interface of chemical compounds filling in the pore space at the interface. If porosity clogging and interface closure is predicted (see the review Gaucher et Blanc (2006)), the time occurrence is strongly dependant on calculation hypothesis (Marty et al., 2009). With regards to this problem, experiments are necessary to validate calculations (e.g. reactivity pathways, mineral dissolution and precipitation) and to make sure that these results can be used to assess the long term behaviour with less incertitude. Results & Discussion This paper focuses on experiments, their modelling, and long term modelling. On all the mentioned previous topics, recent works conducted at BRGM in close collaboration with Andra and international teams will be discussed (Gaboreau et al., 2012; Grangeon et al., 2013a; Grangeon et al., 2013b; Jenni et al.; Lerouge et al.; Marty et al., Submitted-a; Marty et al., Submitted-b; Marty et al., Submitted-c; Marty et al., 2009). The obtained inputs deal with, (i) clay concrete interface characterisation from a mineralogical and petrophysical point of views, (ii) a better understanding and description of the structure of C-S-H the main binding phase in concrete, (iii) collection of new kinetics data on C-S-H used to predict the 1


behavior of radioactive waste repositories, (iv) impact of the modelling assumptions on the modelled reaction pathways and (v) the reactive transport codes benchmarking in a the context of complex cement/clay interactions. Based on the described results the need of future research on the topic of clay/concrete interaction will be discussed. Acknowledgements Some of the results presented in this workshop were collected during some projects granted by ANDRA in the framework of the ANDRA/BRGM scientific partnership. This work has been conducted with a lot of colleagues at BRGM and in many others institutions. My first attempt was to put them all on the author list, but I have quickly realized that half of a page will be necessary. Thanks to all of them, I can assure that this past and future collaborations have been and will be very fruitful and exciting to make this research a success. References Gaboreau, S., Lerouge, C., Dewonck, S., Linard, Y., Bourbon, X., Fialips, C.I., Mazurier, A., Pret, D., Borschneck, D., Montouillout, V., Gaucher, E.C., Claret, F., 2012. In-situ interaction of cement paste and shotcrete with claystones in a deep disposal context. American Journal of Science 312, 314-356. Gaucher, E.C., Blanc, P., 2006. Cement/clay interactions - A review: Experiments, natural analogues, and modeling. Waste Manage 26, 776-788. Grangeon, S., Claret, F., Lerouge, C., Warmont, F., Sato, T., Anraku, S., Numako, C., Linard, Y., Lanson, B., 2013a. On the nature of structural disorder in calcium silicate hydrates with a calcium/silicon ratio similar to tobermorite. Cement and Concrete Research 52, 31-37. Grangeon, S., Claret, F., Linard, Y., Chiaberge, C., 2013b. X-ray diffraction: a powerful tool to probe and understand the structure of nanocrystalline calcium silicate hydrates. Acta Crystallogr B B69, 465-473. Jenni, A., Mäder, U., Lerouge, C., Gaboreau, S., Schwyn, B., 2014. In situ interaction between different concretes and Opalinus Clay. Physics and Chemistry of the Earth, Parts A/B/C. Lerouge, C., Claret, F., Tournassat, C., Grangeon, S., Gaboreau, S., Boyer, B., Borschnek, D., Linard, Y., 2014. Constraints from sulfur isotopes on the origin of gypsum at concrete/claystone interfaces. Physics and Chemistry of the Earth, Parts A/B/C. Marty, N.C.M., Blanc, P., Bildstein, O., Claret, F., Cochepin, B., Su, D., Gaucher, E.C., Jacques, D., Lartigues, J.E., Mayer, K.U., Meussen, J.C.L., Munier, I., Pointeau, I., Sanheng, L., Steefel, C., Submitted-a. Benchmark for reactive transport codes in the context of complex cement/clay interactions. Computational Geosciences. Marty, N.C.M., Grangeon, S., Warmont, F., Lerouge, C., Submitted-b. Alteration of nanocrystalline Calcium Silicate Hydrate (C-S-H) at pH 9.2 and room temperature. Cement and Concrete Research. 2


Marty, N.C.M., Munier, I., Gaucher, E.C., Tournassat, C., Gaboreau, S., Vong, C.Q., Giffaut, E., Cochepin, B., Claret, F., Submitted-c. Simulation of cement/clay interactions: feedback on the increasing complexity of modelling strategies. Transport in Porous Media. Marty, N.C.M., Tournassat, C., Burnol, A., Giffaut, E., Gaucher, E.C., 2009. Influence of reaction kinetics and mesh refinement on the numerical modelling of concrete/clay interactions. J Hydrol 364, 58-72.

3


4


REACTIVE TRANSPORT MODELLING OF OXYGEN GAS DIFFUSION AND CONSUMPTION IN A DISPOSAL CELL OF RADIOACTIVE WASTE Laurent De Windt1*, Jérôme Corvisier1 and François Marsal2 1 2

Mines-ParisTech, Centre de Géosciences, 77305 Fontainebleau Cedex, France IRSN, PRP-DGE/SEDRAN/BERIS, B.P. 17, 92262 Fontenay-aux-Roses Cedex, France

* Corresponding author: [email protected] Introduction The oxic transient in geological radioactive waste disposals is a key issue for the performance of metallic components that may undergo high corrosion rates under such conditions. The environment in a deep geological disposal of high-level radioactive waste (HLW) will be initially aerated, due to the air introduced into the disposal cell during its excavation and waste emplacement. In addition, it is assumed that the ventilation of handling drifts will renew oxygen at the front of the disposal cell over a period from a few years to many decades. So far, the reducing conditions generally prevailing in deep geological environments are supposed to be quickly restored after the closure of such facilities because of the consumption of the oxygen by corrosion of the carbon steel (C-steel) of waste overpacks, oxidation of pyrite of the host rock and, to a probably lesser extent, microbial activities. However, a previous study carried out in-situ in the argillite formation of Tournemire (France) has suggested that oxic conditions could have lasted several years (Gaudin et al., 2009). Modelling water/gas/mineral interactions is a key issue in several applications of reactivetransport codes in radioactive waste disposals. A multiphase reactive transport model was performed to analyse the balance between the kinetics of pyrite oxidative dissolution, the kinetics of carbon steel corrosion and oxygen gas diffusion when carbon steel components are emplaced in the geological medium. Two cases were modelled: firstly, the observations made in Tournemire have been reproduced, and the model established was then applied to a disposal cell for high-level waste (HLW) in an argillaceous formation, taking into account carbon steel components and excavated damaged zones (EDZ). Coupling gas diffusion to reactive transport The reactive transport code HYTEC (van der Lee et al., 2003) is based on a finite element discretization in a representative elementary volume approach and a sequential iterative operator-splitting method for coupling between chemistry and transport. A simplified twophase dynamics of mass transfer can be simulated, namely a reactive transport modelling in the water phase linked to diffusion in the gas phase under water unsaturated conditions. The mass balance is made over the aqueous (dissolved) phase, the solid phases (including the mineral and fixed fractions) and eventually the gas phase. The gas diffusion coefficient is 5


a function of the water saturation state (variants of the Aachib and Millington-Quirk equations). Specific gas modules incorporating the perfect gas law and classical cubic equations of state (Redlich-Kwong, Peng-Robinson) and their appropriate analytic solvers have been developed and implemented. The Henry constants are either tabulated or their dependency upon pressure and temperature expressed as theoretical functions. HYTEC was applied to the 2D-configuration of boreholes and HLW disposal cell (Fig. 1) including gas diffusion, aqueous chemistry, cation exchange and dissolution/precipitation processes. Kinetic formulations were introduced for C-steel corrosion and pyrite dissolution under both oxic and anoxic conditions. Two modelling scenarios were considered according to the degree of water saturation and the voids between the cell components.  scenario 1 considers the residual voids between re-compacted argillite and host rock in the Tournemire borehole on the one hand, and between C-steel liner and argillite host rock in the HLW disposal cell on the other hand, allow for air diffusion; the initial water saturation state does not evolve; the Tournemire tunnel and the handling drift, to which many disposal cells are connected, are constantly ventilated with air (open system).  scenario 2 represents an intermediate stage between scenario 1 and a full resaturation state with perfect contacts between all the materials. In a variant (so-called scenario 1b) the Tournemire borehole and the disposal cell are not submitted to air ventilation (closed system).

Zoomed view

Figure 1: HLW disposal cell: 2D-cylindrical modelling grid (rotation around the length axis L); BC stands for boundary conditions, EDZ for the excavated damaged zone.

6


Results & Discussion In a closed system, modelling leads to a complete and fast consumption of oxygen in both configurations (Fig. 2). Modelling results are more consistent with the in situ test in Tournemire while considering residual voids between materials and/or a water unsaturated state allowing for oxygen gas diffusion (open conditions, scenario 1). In such a scenario 1 for HLW disposal cells, the cell environment globally remains oxidizing (Fig. 3), with an oxic corrosion of carbon steel and precipitation of goethite). Oxygen gas diffuses along the 20 m cell length. However, the partial oxygen pressure is below 0.01 bar at the back of the cell. In scenario 2, when the system is under an intermediate resaturation state, a redox contrast occurs between the reducing conditions at the back of the disposal cell (with anoxic corrosion of C-steel, magnetite formation and H2 production, Fig. 4) and oxidizing conditions at the front. Such a permanent differential of redox potential along the disposal cell might enhance corrosion in the zone where the redox front takes place or deeper in the cell. The extent of the oxidizing/reducing front in the disposal cell is strongly dependent on the gas diffusion coefficient in partially saturated zones. A: Tournemire borehole, closed system O2(aq) concentration

B: HLW disposal cell, closed system O2(aq) concentration

Figure 2: Dissolved oxygen consumption under water-saturation scenario 1b, i.e. a closed system without air ventilation, calculated at the back of the Tournemire borehole and the middle of the HLW disposal cell; sensitivity analysis on the rate constant of pyrite oxidation and steel oxic corrosion. The potential effect of microbial activities on oxygen consumption (respiration) and sulphate reduction should be undertaken in further calculations to strengthen this study, as well as the complex and coupled effects of the temperature gradient on water resaturation, multiphase transport and chemical processes.

7


Figure 3: HLW disposal cell: steady-state 2D-profiles of dissolved oxygen for scenario 1 (top) and scenario 2 (bottom).

Figure 4: Zoom of HLW disposal cell: 2D-profiles hydrogen partial pressures calculated after 100 y for scenario 2. References De Windt, L., Marsal, F., Corvisier, J. Pellegrini, D., in press. Modeling of oxygen gas diffusion and consumption during the oxic transient in a disposal cell of radioactive waste. Appl. Geochem., http://dx.doi.org/10.1016/j.apgeochem.2013.12.005 Gaudin, A., Gaboreau, S., Tinseau, E., Bartier, D., Petit, S., Grauby, O., Foct, F. and Beaufort, D., 2009. Mineralogical reactions in the Tournemire argillite after in-situ interaction with steels. Appl. Clay Sc. 43, 196-207. van der Lee, J., De Windt, L., Lagneau, V., Goblet, P., 2003. Module-oriented modeling of reactive transport with HYTEC, Comput. Geosc. 29, 265-275.

8


HOW CAN THE RADIONUCLIDE DIFFUSION BE AFFECTED BY POROSITY CHANGES? I. Fatnassi1,2,*, S. Savoye1, P. Arnoux1, P. Gouze2, O. Bildstein3, V. Detilleux4 and C. Wittebroodt5 1

CEA, DEN, DPC, Laboratory of Radionuclides Migration Measurements and Modeling, France 2 Géosciences Montpellier, UMR 5243, CNRS, Université de Montpellier 2, France 3 CEA, DEN, DTN, Laboratory for modeling of transfers in the Environment, France 4 Bel V, Belgium 5 IRSN, France. * Corresponding author: [email protected] Introduction Disposal in deep geological clay formations is one of the solutions chosen for managing the fate of high and intermediate level as well as long-lived nuclear wastes. The long-term evolution of these deep repositories should be, for a major part, governed by geochemical processes that can irreversibly modify the containment properties of the materials used in the multi-barrier system, and especially at their interfaces. For instance, mineral dissolution and precipitation can change their transport properties significantly by enlarging or clogging the pore space. However, addressing the feedback of porosity changes in the long-term simulations coupling chemistry and transport is still an issue, especially because of the lack of quantitative experiments used to calibrate the numerical models (Kosakowski et al., 2009). Up to now, even though diffusion is expected to be the main transport process taking place in the context of the nuclear waste geological disposal, only experiments carried out under advective conditions are described in literature for investigating the porosity clogging effect. For example, Tartakovsky et al. (2008) and Katz et al. (2011) used quasi-two dimensional flow cells packed with quartz sand and evidenced precipitation of calcium carbonate that clearly impacts the transport properties of their cells. The objective of the current study is to assess the ability of some numerical codes (two chemistry-transport codes, HYTEC and CRUNCH and one multiphysics software, Freefem++) to reproduce experimental results obtained from diffusion/precipitation experiments. Material & Methods The experimental setup consists of a modified through-diffusion cell, in which the cell body has a quasi two-dimensional shape and is made of plexiglass® and glass for visualizing the porosity clogging, if any. Different porous media (e.g. compacted sand, glass frit) were tested.

9


The initial transport properties of the different materials were determined by performing through-diffusion experiments with deuterium-enriched solutions before clogging. Afterwards, a solution containing Na2C2O4 was injected in the outlet reservoir, while another solution containing CaCl2 and deuterium was put into the inlet reservoir for studying the impact of the clogging by CaOx precipitation on the diffusive transport of the deuterium.

Figure 1: Modified through-diffusion cell used for investigating diffusion/clogging processes. Numerical methods Governing equations The system is represented by a 1D or 2D geometry. In both cases, the inlet (resp outlet) reservoir is taken into account with a fixed Ca (resp. Oxalate) concentration on the left (resp. right) side of the simulation box. All elements (Ca2+, C2O42-, Na+, Cl-) can diffuse with a specific constant diffusion value (except for HYTEC, with a unique diffusion coefficient). The precipitation reaction present in our study is:

The mass conservation of each element through a saturated porous medium is governed by:

The porosity evolution due to precipitation is given by :

Reaction rate is related to the specific surface area and elements concentrations by:

10


Where ci are element concentration,  is the porosity, R is the precipitation reaction rate, S is the specific surface area, De is the effective diffusion coefficient. 1D simulations with CRUNCH and HYTEC HYTEC and CRUNCH are built on two distinct approaches. The main differences concerns the way reaction and transport processes are coupled (sequential iterative approach vs. global implicit) and the kinetics and the reactive surface are described (Table 1). Table 1: Differences between Hytec and Crunch for kinetics and reactive surface area. Hytec Kinetic laws

Crunch

rs   AS * k rate * (1  rs: dissolution/precipitation reaction rate // Qs: ion activity product // As: specific surface area

Surface area

krate: dissolution/precipitation rate constant K: equilibrium constant

Abulk  Abulk,0 * (

Abulk  As * C  C : particle concentration 

As

: initial specific surface

(m2.g -1 porous medium) 

Abulk

: bulk surface (m2.m

QS ) K

-3



w : porosity



w0 : initial porosity Abulk,0 :The initial



w 2/3 ) w0

bulk area (m2.m

-3

porous

medium).

of porous medium). 

Abulk

: The bulk area (m2.m -3 porous medium).

Moreover, CRUNCH and HYTEC allows to numerically set a residual porosity value, i.e. a threshold value under which further clogging is not allowed anymore. Multiphysics simulations with FreeFem++ Modelling at pore scale was carried out by means of a multiphysics code (FreeFem++, Hecht et al., 2012). This approach enables the implementation of physical laws at a microscopic scale, such as Classical Nucleation Theory which is useful for describing the precipitation of secondary mineral. With this approach, only the pore space where clogging occurs is modelled in 2D and 3D. FreeFem++ enables adaptative meshing which allows for a very high mesh density around precipitation and a coarse density anywhere else in order to save computational time. In a first approach, we have used specific surface area and diffusivity given by the following equations:

11


Results & Discussion Sensitivity analyses with the CRUNCH Code (compacted sand configuration) The sensitivity of several key parameters (surface area, space and time discretization, residual porosity, cementation factor, rate constant, etc…) was studied. Figure 2 shows the strong impact of the cementation factor (on the left) and the residual porosity (on the right) on the behaviour of the deuterium flux calculated in the outlet reservoir.

Figure 2: Evolution of deuterium flux to changes of the cementation factor and residual porosity.

Comparison of the CRUNCH/HYTEC with the experimental results on compacted sand When using a unique value for the diffusion coefficient, results show that HYTEC leads to a faster decrease of the deuterium flux than CRUNCH, likely related to their different laws used for describing kinetics (Figure 3 and Table 1). Moreover, the use of distinct diffusion coefficient for each species in CRUNCH strongly delayed the flux drop (10 days / 5 days), in accordance with the lower diffusion coefficient for Ca2+ and Ox than that of HDO. Figure 3 shows that the injection of Ca and Ox clearly impacts the experimental HDO flux, which, however, does not tend towards zero. This could be explained by the growth process which can produce secondary minerals with a residual (microporous?) porosity. Such a behaviour was not successfully reproduced by CRUNCH and HYTEC for which a residual porosity should be fixed and a better description of the transient state has to be achieved.

12


Clogging modeled by CRUNCH with multi-diffusive component

Figure 3: Comparison of the experimental fluxes with and without clogging with simulations.

Figure 4: (From left to right): Oxalate concentration, Ca concentration, and porosity values within the pores at t ≈ 7 days. First results obtained with the multiphysics code, FreeFem++ Multiphysics codes allow for the integration of coupled set of equations representing multiple simultaneous physical phenomena. In this work, a first attempt was performed, aiming at estimating the key mechanisms that can occur at the pore scale. Figure 4 (left and middle) shows respectively the oxalate and calcium concentrations within the pores, for which the vertical diffusion front can be clearly observed. That is due to the oxalate and Ca consumption during the fast precipitation reaction. On the left (resp. right) of the front, oxalate concentration (resp. Ca) remains zero. On the right figure, the porosity strongly decreases in the narrow front where the CaOx precipitates. Due to the porosity drop, HDO diffusivity drastically decreases and a concentration discontinuity appears in the clogged area (Figure 5, left). Lastly, the HDO flux in the outlet increases until the clogging starts and therefore the flux tends towards zero, as predicted by CRUNCH and HYTEC (Figure 5, right).

13


Figure 5: Tracer (HDO) concentration at t ≈ 7 days (Left) and HDO flux in the outlet. Conclusions Two reactive transport codes, HYTEC and CRUNCH, were used for studying their ability to describe the porosity and diffusivity changes associated to chemical transformations in a one- dimensional domain. They led to distinct simulation results because of the different internal models and algorithms they use, such as those describing the secondary precipitation processes and especially, the surface area laws. Experimental results show that the tracer flux did not tend towards zero after 15 days when porosity clogging is expected to occur. That can be explained by the occurrence of a small (but not zero) residual porosity within the porous medium. These experimental data were not satisfactorily reproduced by CRUNCH and HYTEC. Indeed, better simulations require to set a non-zero residual porosity value. Nevertheless, this remains a semi-empirical treatment of the clogging. A complementary approach is under progress which consists in modelling the system at the pore scale with a multiphysics code, FreeFem++. It allows the integration of several crystal germination and growth processes and 3-D simulations with different laws for homogeneous and heterogeneous crystal growth should complete the current 2-D calculations. Acknowledgements This work received financial support from the CEA, IRSN and BEL V. References Katz, G.E, Berkowitz, B, Guadagninib, A., Saaltink, M. W. Experimental and modeling investigation of multicomponent reactive transport in porous media. J. Contam. Hydrol. 120-121, 27-44. Kosakowski, G., Blum, P., Kulik, D.,,Pfingsten, W., Shao, H. ,Singh, A., 2009. Evolution of generic clay/cement interface: first reactive transport calculations utilizing a Gibbs energy minimization based approach for geochemical calculation. J. Environ. Sci. Sustain. Soc. 3, 41-49.

14


Hecht, F., 2012. New development in FreeFem++. J. Numer. Math. 20, no. 3-4, 251–265. 65Y15. Tartakovsky, A. M., Redden, G., Lichtner, P. C., Scheibe, T. D., Meakin, P. 2008. Mixinginduced precipitation: Experimental study and multiscale numerical analysis. Wat. Resour. Res. 44, W06S04.

15


16


ATTACHMENT OF PARTICLES AT SURFACES. THEORY, EXPERIMENT AND APPLICATION Nikola Kallay Division of Physical Chemistry, Department of Chemistry, Faculty of Science, University of Zagreb, Horvatovac 102A, 10000 Zagreb, Croatia *Corresponding author: [email protected] Colloid particles in aqueous solutions of electrolytes may be attached on surfaces. In absence of surface active agents, the attachment process is usually an irreversible kinetic process governed by electrostatic forces. In such a case interaction energy is a sum of short range repulsion, van der Waals attraction and electrostatic attraction or repulsion. At close distance van der Waals attraction prevails so that deep energy minimum is developed. In the case of electrostatic repulsion the electrostatic barrier is developed which reduces rate of attachment and even prevents this process. Attachment (adhesion) may be related to heterocoagulation, since in almost all cases particles are of different composition with respect to the surface (substrate). In most of the cases surface charge depends on pH. Surface and particles exhibit different properties, i.e. different isoelectric points. Accordingly, one may observe three pH regions. Due to electrostatic repulsion, the adhesion will be prevented in (very) acidic region (positive charge of both surface and particles) and in (very) basic region (negative charge of both surface and particles). The repulsion could be reduced and attachment promoted by increasing electrolyte concentration. Between isoelectric points the charge of the surface will be of opposite sign with respect to particles and the electrostatic attraction will cause rapid attachment. There are different experimental methods for examination of the adhesion process. The simple and efficient one is the packed column technique. The column is filled with beads (substrate) and one measures the colloid particle concentration of inlet and outlet suspension. Small colloid and nano-particles undergo diffusional mechanism of attachment which is theoretically described so that one may interpret and predict the attachment rate. There are different aspects of adhesion experiments. One may study effect of magnetic forces, multilayer deposition, detachment of particles etc. Also, adhesion experiments may be applied for determination of isoelectric points of conductive metallic surfaces as well to follow corrosion process. In the lecture the theoretical background will be explained, the experimental techniques will be described. Also, some representative experiments will be shown.

17


18


TRANSPORT AND RETENTION OF MANUFACTURED NANOPARTICLES IN WATER-SATURATED POROUS MEDIA UNDER DIFFERENT CONDITIONS: MEASUREMENTS AND MODELING Ivan Toloni*, François Lehmann and Philippe Ackerer Laboratoire d’Hydrologie et de Géochimie de Strasbourg, UdS-CNRS UMR 7517 Université de Strasbourg, 1 rue Blessig, 67000 Strasbourg, FRANCE * Corresponding author: [email protected] Transport and kinetics retention of titanium dioxide (TiO2, rutile) nanoparticles were investigated in water-saturated porous media. Experiments were carried out under a range of ionic strength (IS) and water velocity (U) in laboratory columns, packed with quartz sand. Columns were packed as uniformly as possible in order to get the same hydrodynamic parameters for each experiment (porosity, dispersivity) and all column experiments were conducted at least in duplicate. Conductivity, pH and UV-absorption (280 nm) were measured automatically all along the experiments for both inlet and outlet flows by means of on-line sensors. The break through curves (BTC) had typical blocking controlled shapes with concentration increasing in time (Fig. 1). Mass retention decreased with an augmentation of U and increased with an augmentation of the IS of the solution. The three BTC corresponding to different U were modeled coupling the Convective-Dispersive Equation with a kinetic deposition term. A Langmuirian dynamics was proposed for kinetic deposition, coherently with the blocking mechanism that controls the BTC shape. The deposition term depends on two parameters: the deposition coefficient (kd [s-1]) and the maximum solid phase concentration (Smax, [mg/g]). The parameters were optimized for each BTC through the resolution of an inverse problem using the software HYDRUS 1D. pH 10 50 mg/L

1.0

tracer 63; 3 mM; 3mL/min 74; 3 mM; 0,5 mL/min 67; 5 mM; 18 mL/min 65; 5 mM; 3 mL/min 71; 5 mM; 0,5 mL/min 67 model fit 65 model fit 71 model fit

C/C0 (-)

0.8

0.6

0.4

0.2

0.0 0

1

2

3

4

5

Vp (-)

Figure 1: Ionic strength and water velocity effect: break through curves and modeling.

19


20


A NEW STOCHASTIC APPROACH FOR THE SIMULATION OF FOULING: FROM SINGLE PARTICLE DEPOSITION TO CLOGGING Christophe Henry1* and Jean-Pierre Minier2 1

Institute of Fluid-Flow Machinery, Polish Academy of Sciences, ul. Fiszera 14, 80-231 Gdansk, Poland 2 EDF R&D, Mécanique des Fluides, Energie et Environnement, 6 quai Watier, 78400 Chatou, France *Corresponding author: [email protected] Introduction Particulate fouling generally arises from the continuous deposition of colloidal particles on initially clean surfaces, a process which can even lead to a complete blockage of the fluid cross-section. This accumulation generally affects the normal operating conditions, and therefore the performances, of various components involved in many industrial situations, such as in heat exchangers (Schwarz (2001)). The main purpose of the present study is to emphasise how the key physical interactions (i.e. particle-fluid, particle-particle and particle-surface) are intertwined throughout the whole fouling process. Then, a new stochastic modelling approach for colloidal fouling, regarded here as a continuous process ranging from single-particle deposition up to clogging effects, is presented. Methods Phenomena and interactions at play To understand, model and simulate the complete fouling phenomenon, fouling is often addressed as resulting from the coupling and the interplay of four elementary phenomena which can be considered separately: deposition, resuspension, agglomeration and clogging (i.e. multilayer deposition). Fouling thus results from the intricate coupling between particlefluid, particle-particle and particle-surface interactions. A recent review of experimental works on particle clogging has revealed that particle accumulation manifests itself by different fouling patterns (Henry et al., 2012b). Thus, as depicted in Figure 1, it is worth distinguishing three typical fouling patterns: I.

blocking effects (when particle-particle interactions are strongly repulsive) leading only to monolayer formation;

21


II.

multilayer deposition (with attractive particle-particle and particle-substrate interactions) with no necessary specific structure of the resulting accumulation of deposited particles;

III.

induction effects (when particle-surface prevent deposition except in some local areas while particle-particle interactions are weakly repulsive) leading to the formation of isolated surface clusters that eventually merge to form arches.

Figure 1: Sketch of the typical situations for the deposit structure.

Modelling approach For the sake of simplicity and as a first step in the whole modelling approach, the model has been developed considering the case of dilute suspensions only (no agglomeration), oneway coupling (no fluid motions around particles) and without detachment (since we wanted to concentrate on the modelling steps related to the formation and growth of the deposit itself). Following previous studies on the early stages of fouling (Minier and Peirano (2001), Guingo and Minier (2008a), Guingo and Minier (2008b), Henry et al. (2011), Henry et al. (2012a)), the new stochastic model for particulate clogging is based on a two-stage scenario: I.

particle transport by the fluid is modelled in the framework of one-particle pdf approaches;

II.

the attachment step is described using the DLVO theory coupled with a statistical description of the surface facing the particle (both for the presence of surface roughness or for the presence of already deposited particles as sketched in Figure 2).

The coupling between the hydrodynamic transport and the attachment step follows an energy-balance approach, where the kinetic energy of a particle impacting the surface is compared to the energy barrier encountered upon interacting with the fouled or clean surface.

22


Figure 2: Drawing of the attachment step, leading either to single particle deposition (left) or to the formation of a larger cluster (right). Reprinted from Henry et al. (2012b). Copyright 2012 with permission from Elsevier. Results & Discussion This new stochastic modelling approach has been shown to retrieve the three typical fouling patterns that have been observed experimentally (i.e. monolayer, induction and multilayer formation). The present approach has also been validated qualitatively by comparing numerical results to recent experimental measurements of fouling. Finally, the model has been shown to be able to reproduce the blockage of a complete flow section: for that purpose, the deposition of Brownian particles in a rectangular mesh was recorded. As seen in Figure 3, the particles initially deposit on a clean surface, eventually forming clusters whose size grows continuously until complete blockage as more and more particles deposit. These first results are encouraging and indicate that it is worth pursuing the development of such modelling approaches. Yet, there is still several issues that need to be addressed before coming up with a complete model for particle clogging: two-way coupling models are needed to capture the influence of clusters on the fluid motions, a model for particle/cluster re-entrainment has to be developed with a proper description for the deposit morphology and cohesion forces. Conclusions The qualitative good agreements between experimental data and numerical results show that a modelling approach, which treats accurately and separately the effects of the three physical interactions, has the potential to address properly the challenging issue of particulate fouling. Therefore, a single modelling framework for the description of the whole fouling process is possible and the present model opens interesting possibilities for future simulations of fouling in complex industrial situations. 23


Figure 3: Clogging of a microchannel through sequential deposition of small particles (a,b,c) until the flow passage becomes blocked by a larger cluster (d). Reprinted from Henry et al. (2012b). Copyright 2012 with permission from Elsevier. Acknowledgements The authors are grateful to Jacek Pozorski, Pr. at the Institute of Fluid Flow Machinery, Polish Academy of Science, for fruitful discussions. The authors also thank Grégory Lefèvre, CdR at LECIME Chimie ParisTech. References Guingo M. and Minier J.P., 2008. A stochastic model of coherent structures for particle deposition in turbulent flows. Phys. Fluids. ,20(5): 053303 Guingo M. and Minier J.P., 2008. A new model for the simulation of particle resuspension by turbulent flows based on a stochastic description of wall roughness and adhesion forces. Aerosol Science, 39: 957-973 Henry C., Minier J.P., Lefèvre G. and Hurisse O. 2011. Numerical study on the deposition rate of hematite particles on polypropylene walls: role of surface roughness. Langmuir, 27:4603-12

24


Henry C., Minier J.P. and Lefèvre G., 2012. Numerical study on the adhesion and reentrainment of nondeformable particles on surfaces: the role of surface roughness and electrostatic forces. Langmuir, 28: 438-52 Henry C., Minier J.-P., Lefèvre G., 2012b. Towards a description of particulate fouling: from single particle deposition to clogging. Advances in Colloid and Interface Science, 185186: 34-76 Minier J.P. and Peirano E., 2001. The pdf approach to turbulent and polydispersed two-phase flows. Phys. Rep., 352(1–3): 1–214. Schwarz T., 2001. Heat transfer and fouling behaviour of siemens PWR steam generators— long-term operating experience. Exp. Therm. Fluid. Sci., 25: 319-27

25


26


ION ASSOCIATION AND HYDRATION IN AQUEOUS ELECTROLYTE SOLUTIONS – THE VIEW OF DIELECTRIC SPECTROSCOPY Richard Buchner1* and Glenn Hefter 2 1 2

Institut für Physikalische und Theoretische Chemie, Universität Regensburg, Germany Chemistry Department, Murdoch University, Australia

*Corresponding author: [email protected] Introduction Ion solvation and ion association are key features of electrolyte solutions in any solvent: their balance largely determines the structure, thermodynamics and dynamics of such solutions. Despite being intensively studied over many years using a large number of techniques, the interplay between solvation and association effects are still intriguing. Regarding ion association processes, the results obtained with different techniques show considerable discrepancies, which were recently highlighted by Hefter (2006) and Marcus and Hefter (2006). The aim of this contribution is to show what kind of information can be provided on ion association and hydration by dielectric relaxation spectroscopy (DRS). Materials & Methods DRS probes the response of the sample to a time-dependent electric field. For electrolyte solutions this response mainly occurs in the microwave region of the electromagnetic spectrum and essentially arises from ion conduction and dipole reorientation (Buchner and Hefter (2009)). The technique and the associated data analysis will be briefly introduced. Results & Discussion This presentation will focus on the dipolar contribution to the dielectric spectra, which enables DRS to detect all species (solvent molecules, ions, ion aggregates) possessing a permanent dipole moment, provided their lifetime is at least comparable to their rotation time (Figure 1). In the DR spectra a (dipolar) solvent contribution is always present. From its amplitude, effective hydration numbers, Z, can be determined. As will be discussed, Z is not necessarily identical to the coordination numbers determined, e.g., with scattering experiments or computer simulations, as it reflects the nature and strength of ion-solvent interactions.

27


Figure 1: Dielectric loss spectra, (), of aqueous Al2(SO4)3 solutions at 25 °C indicating the approximate contributions from the solvent and from solute ion pairs.

Figure 2: (a) Solute contribution to dielectric loss spectra, (), of aqueous CuSO4 solutions at 5 °C indicating the location of the 2SIP, SIP and CIP relaxations; (b) amplitudes of the 2SIP, SIP and CIP modes; (c) ion-pair concentrations, including the total concentration of ion pairs relevant for thermodynamics (after Akilan et al. (2006)). Most simple inorganic ions do not possess a dipole moment. Therefore, the appearance of a distinct solute mode is generally an indication for the presence of sufficiently long-lived (>1 ns) ion pairs although larger (dipolar) aggregates may also be present (Schrödle et al. (2008)). A major advantage of DRS is its unique ability to distinguish ion-pair species of differing hydration states: contact (CIP, “inner-sphere”), solvent-shared (SIP, “outer-sphere”) and solvent-separated (2SIP, “outer-outer-sphere”) ion pairs, see Figure 2. The concentrations of the species are quantitatively accessible from the amplitudes of the resolved modes while the corresponding peak frequencies aids in their identification. This distinguishes DRS from other spectroscopic methods, which are generally only sensitive to CIPs, and from thermodynamic methods and conductivity measurements, which detect the 28


overall degree of ion association but give no molecular-level information on the species present. References Akilan, C., Hefter, G., Rohman, N., Buchner, R., 2006. Ion Association and Hydration in Aqueous Solutions of Copper(II) Sulfate from 5 to 65 °C by Dielectric Spectroscopy. J. Phys. Chem. B, 110(30): 14961-14970. Buchner, R. and Hefter, G., 2009. Interactions and dynamics in electrolyte solutions by dielectric spectroscopy. Phys. Chem. Chem. Phys., 11(40): 8984-8999. Hefter, G., 2006. When spectroscopy fails: The measurement of ion pairing. Pure Appl. Chem. 78(8): 1571-1586. Marcus, Y. and Hefter, G., 2006. Ion Pairing. Chem. Rev., 106(11): 4585-4621. Schrödle, S., Wachter, W., Buchner, R., Hefter, G., 2008. Scandium Sulfate Complexation in Aqueous Solution by Dielectric Relaxation Spectroscopy. Inorg. Chem., 47(19): 86198628.

29


30


UP-SCALING OF DIFFUSION COEFFICIENTS IN SATURATED AND UNSATURATED CLAYS Thomas Gimmi 1,2* and Sergey V. Churakov 1 1

Laboratory for Waste Management, NES, Paul Scherrer Institute, CH-5232 Villigen, Switzerland 2 Rock-Water Interaction, Institute of Geological Sciences, University of Bern, CH-3012 Bern, Switzerland * Corresponding author: [email protected] Introduction Claystones are considered in several countries as potential host rocks for deep geological disposal of radioactive waste, and compacted clays as backfill material for engineered barriers. Molecular diffusion is the dominant transport mechanism for water and ions in claystones and compacted clays under typical natural hydraulic conditions. The sealing efficiency of the engineered or geologic barriers depends largely on the diffusion coefficients of these materials. Accordingly, diffusion coefficients have to be determined experimentally for these materials for all relevant situations. This is mostly done by tracer experiments in the laboratory. In order to build up confidence for long-term predictions of the performance of the barriers, a thorough understanding of the basic molecular mechanisms of diffusion in the porous matrix as well as of the interplay between structural and chemical properties of clays and solutes has to be proven. The experimental studies need thus to be backed up and complemented by theoretical and modelling studies. Experiments have clearly shown that pore-scale features like local pore morphology (e.g., interparticle pore, interlayer pore) as well as local chemical and mineralogical properties (e.g., mineral surface charge) affect the large-scale transport of ions and tracers (e.g., Glaus et al., 2007; Porion et al., 2007; González Sánchez et al., 2008; Gimmi and Kosakowski, 2011). Including such pore-scale features directly in transport models at the continuum scale is challenging both because of conceptual issues (all parameters should actually be defined at the scale of the grid cells in the model) as well as because of numerical problems arising from the large disparity of scales. An efficient way to proceed is to use a step-wise upscaling, where ‘effective’ transport parameters (e.g., diffusion coefficients) needed as input for larger-scale simulations are obtained by simulations for smaller scales (e.g., Rotenberg et al., 2007; Bourg and Sposito, 2010; Churakov and Gimmi, 2011; Tyagi et al., 2013). The smaller-scale simulations can be based on a different modelling concept than the largerscale simulations and can account for all desired small-scale features. In this way, it is possible to link, for instance, molecular dynamics simulations of specific local environments to pore-scale simulations of heterogeneous domains, or pore-scale simulations to continuum-scale simulations. Such up-scaling approaches have so far mainly focused on cases where the whole pore space is filled with pore water. Under natural conditions, claystones considered as potential host 31


rocks as well as clay backfills are indeed water saturated. The construction of an underground repository or generally the contact with a dry atmosphere leads typically to some degree of desaturation of the rocks (Matray et al., 2007). A water-saturated state of the host rock and backfill materials is then achieved again some time after closure of the repository. At the earlier stages, however, or when considerable amounts of gas are produced during transient phases, for instance through corrosion processes, unsaturated conditions may prevail and affect the evolution of geochemical reactions. Diffusive transport of solutes in the pore water under unsaturated conditions is therefore also of interest. Typically, diffusion coefficients of solutes in partly saturated porous materials are clearly lower than under saturated conditions. Recently, Savoye et al. (2010, 2012) investigated the diffusion of a water tracer and of anions and a cation through partially saturated samples from the Callovo-Oxfordian claystone, a formation considered as potential host rock in France. They found that the steady-state flux of HTO and anions decreased by a factor of up to 7 and 50, respectively, and that the diffusion coefficient of Cs decreased by a factor of up to ~60 for water saturations down to ~0.8. The reasons for the different behaviour of the neutral, anionic and cationic tracers are not fully clear at present, but it is hypothesized that they are related to pore-scale features. Churakov (2013) investigated the local diffusion of water and ions in water films of a desaturated pore by molecular simulations. The results of the modelling are qualitatively consistent with the experimentally observed trends but a proper account for pore-scale features is needed for a quantitative description at the scale of the experiments. Material & Methods Here we derived up-scaled diffusion coefficients for saturated and partially water saturated, comparably large model clay samples. The model clay samples define the arrangement of different solids and pores at the pore scale as a map on a grid (Churakov and Gimmi, 2011). The microstructure maps were generated according to an algorithm (Tyagi et al., 2013) that considers particle shape, particle size distribution and pore size distribution. Diffusion coefficients at the sample scale were then obtained by random walk simulations. First, we investigated the effect of the coefficient of variation of the pore size distribution on saturated diffusion coefficients. Second, the water retention function, that is, the relation between the water potential and the water content, of the model clay samples was determined with an approach that considers the effects of capillary and adsorptive forces, and diffusion coefficients of unsaturated samples were subsequently obtained. Similarly as in Churakov and Gimmi (2011), local diffusion coefficients to be used for the random walk simulations were derived from molecular dynamics simulations for saturated and unsaturated pore configurations (Churakov 2013). We focus at present on diffusion of an inert tracer (e.g., HTO) through relatively simple 2D, high-density smectite samples. These samples are composed of clay particles with a roughly bimodal particle size distribution. All particles contain interlayers with a width of about two water layers (0.5 nm).

32


Results & Discussion In order to get approximately realistic drainage patterns, a large variability of pore sizes and thus comparably large pore maps are required, which is computationally demanding. Figure 1a shows a part of a pore map, which has a total size of 2 µm times 2 µm, Figure 1b greyvalue coded pore sizes of the same part. The pore map was generated based on a lognormal distribution of the interparticle pore sizes with a mean width of 3 nm and a coefficient of variation of 3. The effective pore size distribution derived from the water retention function for this sample, which includes also the interlayer pores of 0.5 µm width, is shown in Figure 1c. The resulting sample-scale diffusion coefficients derived for saturated samples (Figure 2) decrease with increasing coefficient of variation of the underlying lognormal interparticle pore size distribution. This can be interpreted as an increase of tortuosity (and possibly a decrease of connectivity) caused by the increasing variability of the interparticle pore sizes.

Figure 1: (a) A part of a 2-µm x 2-µm pore map used to derive up-scaled diffusion coefficients. (b) Grey-value coded pore size distribution of the part of the pore map shown in a. (c) Effective pore size distribution derived from the water retention function of the pore map.

Figure 2: Diffusion coefficients obtained for large, saturated pore maps generated based on increasing coefficients of variation of the interparticle pore size distribution.

33


Figure 3: Diffusion coefficients as a function of water saturation for a comparably small clay pore map with uniform interparticle pores. Figure 3 presents diffusion coefficients as a function of the water saturation of the sample, derived for a smaller sample map with smaller average pore size. Clearly, the diffusion coefficients decrease with increasing desaturation, broadly consistent with experimental observations. Further work will include simulation of unsaturated diffusion coefficients of HTO for the large sample maps, as well as investigations of the effect of partial drainage on diffusion of anions or cations. Conclusions The results obtained so far are promising. The saturated and unsaturated diffusion coefficients show similar tendencies as the experimental data. Increasing the broadness of the pore size distribution at a given mean pore size tends to decrease diffusion coefficients, and decreasing the saturation of a sample clearly reduces the diffusion coefficients. At present, simulations for larger unsaturated samples with a broader range of pore sizes, better representative for real samples, are still outstanding. We think that our up-scaling method can contribute significantly to the understanding and interpretation of diffusion in saturated and unsaturated clay samples. Acknowledgements Partial financial support of our work by Nagra, the Swiss National Cooperative for the Disposal of Radioactive Waste, is kindly acknowledged. References Bourg I.C. and Sposito G., 2010. Connecting the molecular scale to the continuum scale for diffusion processes in smectite-rich porous media. Environmental Science & Technology 44, 2085-2091. Churakov S.V. (2013) Mobility of Na and Cs on montmorillonite surface under partially saturated conditions. Environental Science and Technology, 47, 9816-982. 34


Churakov S.V. Gimmi Th. (2011) Up-Scaling of Molecular Diffusion Coefficients in Clays: A Two-Step Approach. Journal Physical Chemistry C, 115 (14), 6703-6714. Gimmi T. and Kosakowski G., 2011. How Mobile Are Sorbed Cations in Clays and Clay Rocks? Environmental Science & Technology 45, 1443-1449. Glaus M.A., Baeyens B., Bradbury M.H., Jakob A., Van Loon L.R. and Yaroshchuk A., 2007. Diffusion of Na-22 and Sr-85 in montmorillonite: Evidence of interlayer diffusion being the dominant pathway at high compaction. Environmental Science & Technology 41, 478-485. González Sánchez F., Jurányi F., Gimmi T., Van Loon L., Unruh T. and Diamond L.W., 2008. Translational diffusion of water and its dependence on temperature in charged and uncharged clays: A neutron scattering study. Journal of Chemical Physics 129. González-Sánchez F., Gimmi T., Jurányi F., Van Loon L. and Diamond L.W., 2009. Linking the Diffusion of Water in Compacted Clays at Two Different Time Scales: Tracer ThroughDiffusion and Quasielastic Neutron Scattering. Environmental Science & Technology 43, 3487-3493. Matray,J.-M.;Savoye,S.;Cabrera,J.Desaturationandstructure relationships around drifts excavated in the well-compacted Tournemire’s argillite (Aveyron, France). Eng. Geol. 2007, 90, 1–16. DOI 10.1016/j.enggeo.2006.09.021.Porion P., Michot L.J., Faugere A.M. and Delville A., 2007. Influence of confinement on the long-range mobility of water molecules within clay aggregates: A H-2 NMR analysis using spin-locking relaxation rates. Journal of Physical Chemistry C 111, 13117-13128. Rotenberg B., Marry V., Dufreche J.-F., Giffaut E. and Turq P., 2007. A multiscale approach to ion diffusion in clays: Building a two-state diffusion-reaction scheme from microscopic dynamics. Journal of Colloid and Interface Science 309, 289-295. Savoye S., Page J., Puente C., Imbert C. and Coelho D., 2010. New Experimental Approach for Studying Diffusion through an Intact and Unsaturated Medium: A Case Study with Callovo-Oxfordian Argillite. Environmental Science & Technology 44, 3698-3704. Savoye S., Beaucaire C., Fayette A., Herbette M. and Coelho D., 2012. Mobility of Cesium through the Callovo-Oxfordian Claystones under Partially Saturated Conditions. Environmental Science & Technology 46, 2633-2641. Tyagi M., Gimmi Th., Churakov S.V. (2013) Multi-scale micro-structure generation strategy for up-scaling transport in clays. Advances in Water Resources, 59, 181-195.

35


36


REACTIVE-TRANSPORT PARAMETERS ADJUSTMENT AND SENSITIVITY ANALYSIS: APPLICATION TO CONCRETE CARBONATION Olivier Bildstein1*, Pascal Thouvenot1, Amandine Marrel2, Isabelle Munier3 and Benoit Cochepin3 1

CEA, DEN, DTN/SMTA/LMTE, Cadarache, F-13108 Saint Paul-lez-Durance, France CEA, DEN, DER/SESI/LSMR, Cadarache, F-13108 Saint Paul-lez-Durance, France 3 Andra, DRD, F-92298 Châtenay-Malabry, France 2

*Corresponding author: [email protected] The modelling of atmospheric carbonation of concrete material is an important part of the understanding of the phenomenological evolution of the geological disposal of low- and intermediate-level long-lived radioactive waste during the operating/reversibility period. Concrete carbonation is a complex process which involves the coupling of water flow (drying), transport in the aqueous and gaseous (mainly CO2) phases, dissolution of primary minerals (mainly portlandite, CSH and ettringite), and precipitation of secondary minerals such as calcium carbonate, other CSH phases (Thouvenot et al., 2013). A stochastic methodology was developed which aims at adjusting the reactive transport parameters of such systems thanks to the results of accelerated carbonation experiments at 20°C for different water content (corresponding to experiments at different relative humidities, HR) (Drouet, 2010). Indeed, a large number of simulation parameters used in calculations, either taken from the literature or based on “expert estimated” values, are fraught with significant uncertainty. It is, in particular, the case for parameters expressing the dependence of the diffusion coefficients with respect (a) to porosity  and (b) to water a 1 b saturation Sl (e.g. the Millington-Quirk law : Deff  D0 S l ). It is also the case for the

dissolution and precipitation kinetics constants of many mineral phases, which can have a strong impact on simulation results. Moreover, there are also uncertainties concerning the model used to represent the decrease of the chemical reactivity when the free water content is low in the cementitious materials (the dissolution/precipitation rate of minerals is simply multiplied by a coefficient Rs depending on the water saturation, with 0 ≤ Rs(RH) ≤ 1). In order to take into account the uncertainty of such input parameters and to optimize the model calibration with experiments on accelerated concrete carbonation, a stochastic approach is used to perform sensitivity analyses and calibration. The objective is to identify the most influential parameters affecting simulation results, whether in a direct way or in interaction with one another. This type of approach potentially requires a large number of simulations. In most cases, the complexity of the calculations and the high computing time for the simulations with reactive-transport codes (Toughreact code in this case, LBNL) make it difficult to directly carry out this set of simulations. To solve this problem, a statistic approach is proposed with the building of a metamodel based on kriging or Gaussian process (GP) from a set of “numerical experiments” in order to approximate the results of the reactive-transport code in a predetermined domain of variation of the input parameters 37


(Table 1) (e.g. Sacks et al., 1989, Kleijnen and Sargent, 2000, Iooss et al., 2006). The parameters of the GP metamodel are estimated using the maximum likelihood method (Marrel, 2008). Table 1: Variation interval for the kinetics, diffusion, and reactivity input parameters (reference values from literature). Input parameters log k portlandite log k CSH 1.6 log k calcite Millington-Quirk a Millington-Quirk b Reactivity coefficient Rs

Minimal values -10 -11 -8 1 2,1 0

Maximal values -6 -7 -4 3 6,3 1

Reference values -7,8 -8,8 -5,8 2 4,2 1

The GP metamodel validity is evaluated using 3 indicators: the quantity of mineral towards the surface of the concrete slab and inside the slab, and the depth reached by the carbonation front (Fig. 1). These indicators are evaluated for the portlandite and calcite profiles measured in experiments conducted at 54%, 63%, 70%, and 80% HR (8 criteria).

Figure 1: Carbonation front in concrete at the end of the experiment: a typical mineralogical profile for primary minerals (e.g. portlandite, CSH1.6). Indicators used to evaluate the metamodel validity (in red). Using this metamodel, a first calibration can be carried out, determining the values of the input parameters which minimize the difference between the mineralogical profiles predicted by the metamodel and the experimental observations (Fig. 2). The profiles predicted by the metamodel are satisfactorily close to the observed profiles, confirming the quality of the calibration. However, the profiles corresponding to the optimized parameters simulated with Toughreact are slightly different from those predicted by the metamodel, pointing at some areas for improvement (e.g. more simulations in the learning step). 38


Figure 2: Portlandite (left) and calcite (right) volume fraction profile for different RH: experimental data (black) are compared with values calculated using the PG metamodel (blue) and using the Toughreact code with the optimized parameters determined with the metamodel (red). A global sensitivity analysis was also performed in order to quantify how the uncertainties on input parameters can impact on the uncertainty of the results obtained with the Toughreact code. This analysis, using the Sobol indices (Sobol, 1993), shows the very strong dependence 39


of the results on portlandite dissolution kinetics and the absence of any effect of calcite precipitation kinetics. It also shows that only the ‘b’ parameter of the Millington-Quirk law influences the position of the carbonation front, and that, only in interaction with the dissolution kinetics of portlandite and CSH1.6. In addition, the study shows in a counterintuitive way that the reactivity coefficient is not very sensitive: this can be due to the uniform law assigned for this parameter, a choice made in an effort to build a metamodel without strong constraints. The mineralogical profiles obtained using Toughreact with the whole set of input parameters in the learning base also show the existence of a large number of simulations where the portlandite and calcite profiles are flat, a situation that is never observed in the experiments. These situations can correspond to cases where the reactivity of the system is overall low, for example, when the kinetics of the portlandite or the reactivity coefficient is very small. This point opens a critical reflexion concerning the effects of the interval of variations chosen for each input parameters, and in this study, in particular, regarding the value of the portlandite dissolution kinetics which should probably be shifted to higher values. References Drouet, E., 2010. Impact de la température sur la carbonatation des matériaux cimentaires. Prise en compte des transferts hydriques. PhD thesis ENS Cachan, France. Iooss, B., Van Dorpe, F., Devictor, N., 2006. Response surfaces and sensitivity analyses for an environmental model of dose calculations. Reliability Engineering & System Safety 91 (10–11): 1241–1251. Kleijnen J.P.C., Sargent, R.G., 2000. A methodology for fitting and validating metamodels in simulation. Europ. J. Operational Res. 120(1): 14-29. Marrel, A., Iooss, B., Van Dorpe, F., Volkova, E., 2008. An efficient methodology for modeling complex computer codes with Gaussian processes. Computational Stat. & Data Analysis 52(10): 4731-4744. Sacks, J., Welch, W.J., Mitchell, T.J., and Wynn, H.P., 1989. Design and analysis of computer experiments. Stat. Sci., 4: 409-435. Sobol, I.M., 1993. Sensitivity estimates for non linear mathematical models. Math. Modelling and Computational Exp. 1 :407–414. Thouvenot, P., Bildstein, O., Munier, I., Cochepin, B., Poyet, S., Bourbon, X. and Treille E., 2013. Modeling of concrete carbonation in deep geological disposal of intermediate level waste, EPJ Web of Conferences 56, 05004, DOI: 10.1051/epjconf/20135605004.

40


REACTIVE TRANSPORT MODELLING OF GEOTHERMAL RESERVOIRS Remis Cannepin and Helge C. Moog Gesellschaft für Anlagen- und Reaktorsicherheit (GRS) mbH, Bereich Endlagersicherheitsforschung, Abteilung Prozessanalysen, Theodor-Heuss-Str. 4, 38122 Braunschweig *Corresponding author: [email protected]

For geothermal applications, given the long period of time during which the power plants must be exploited, simulation tools capable of accurately predicting the behavior of the thermal reservoirs over the years would be of great interest. One of the key capabilities of such modeling programs must be the ability to calculate chemical reactions and their influence on the reservoir properties. Some reservoir solutions, however, do exhibit a high ionic strength, which leads to difficulties in estimating the influence of pressure on the system equilibria. Recently, a new version of PHREEQC has been released, which implements a method to correct partial molal volumes of aqueous species with respect to ionic strength (Appelo 2013). For the present work a small database (GeoDat) has been compiled. It is based on THEREDA R-01, but has been extended with respect to some relevant phases, aqueous species and pertinent Pitzer parameters. Possibilities and limitations of the database are demonstrated with selected examples along with a comparison using other databases. Afterwards, application of the database to reactive transport models of existing reservoirs will be demonstrated. References Appelo 2013: C. A. J. Appelo, D. L. Parkhurst, and V. E. A. Post: Equations for calculating hydrogeochemical reactions of minerals and gases such as CO2 at high pressures and temperatures. Geochimica et Cosmochimica Acta 125, 49–67.

41


42


ON THE UPSCALING OF CHEMICAL TRANSPORT IN FRACTURED ROCK Vladimir Cvetkovic Water Resources Engineering, KTH (Royal Institute of Technology), 10044 Stockholm, Sweden. *Corresponding author: [email protected] The impact of flow heterogeneity on chemical transport from single to multiple fractures is investigated. The emphasis is on the dynamic nature of the specific surface area (SSA) due to heterogeneity of the flow, relative to a purely geometrical definition. It is shown how to account for SSA as a random variable in modelling multi-component reactions. The flow dependent SSA is interpreted probabilistically, following inert tracer particles along individual fractures. Upscaling to a fracture network is proposed as a time-domain random walk based on the statistics of SSA for single fractures. Statistics of SSA are investigated for three correlation structures of transmissivity, one classical multi-gaussian, and two nonGaussian. The coefficient of variation of single fracture SSA decreases monotonously with the distance over the fracture length; the CV of the upscaled SSA reduces further such that after ca 20 fractures it is under 0.1 for a disconnected field, and around 0.2 for connected and multi-gaussian fields. This implies that after 10-20 fractures, uncertainty in SSA is significantly reduced, justifying the use of an effective value. A conservative, lower bound for the dimensionless upscaled effective SSA was found to be 1, suitable for all heterogeneity structures, assuming the cubic hydraulic law applicable. The effect of flow heterogeneity is also illustrated for a dissolution-precipitation reaction example.

43


44


THE NEED FOR PERFORMANCE WHEN SIMULATING REACTIVE TRANSPORT IN HETEROGENEOUS MEDIA Vincent Lagneau1* and Olivier Regnault2 1 2

MINES ParisTech, Centre de Géosciences AREVA Mines

*Corresponding author: [email protected] Introduction In situ Recovery (ISR), or solution mining, is an alternative technique for recovering (usually) deep, poor grade ores. The process involves a series of injection and producer wells, and the circulation of a leaching solution (acidic, basic and/or oxidizing). The leaching solution interacts with the ore body, and solubilizes the desired element. E.g. since 2009, ISR has become the leading technique for uranium exploitation. Reactive transport simulation is an efficient tool to investigate the behavior of ISR exploitation: indeed, the whole process is based on reacting fluid circulation in porous media. The simulation can help understand the processes at stake, optimize the ore dissolution rate, minimize the residual ore after production, and minimize the use of reagents. It can also help understand and predict the long-term impact of the exploitation. A first stage of the simulation is to understand and calibrate the phenomenon at stake: pH buffering by several mineral phases, dissolution of in situ oxidized iron bearing phases and oxidative dissolution of uranium, kinetic controls. The simulation of an exploitation then faces several challenges: strong reactivity of the fluids and fast fluid flow are numerically challenging. Moreover, the large number of wells imposes to cover large domains. Finally, a correct description of the heterogeneity of the medium is key for the simulation to be predictive: 1D or 2D, or 3D with homogenized sections of space fail to reproduce production data. As a result, large simulations, 3D with over 105 cells, have to be undertaken. Material & Methods Code Hytec Hytec is a reactive transport code developed at MINES ParisTech (van der Lee et al 2003, Lagneau and van der Lee 2010). The code solves the chemical equations for most geochemical reactions, including aqueous complexation, redox, dissolution/precipitation, sorption (using several formalisms), each at equilibrium or under kinetic control. The hydrodynamic module copes with flow (saturated, unsaturated or two-phase, at steady- or transient-state), transport (water and gas phases), and heat transfer. The resolution is performed using a finite volume discretization over a Voronoi mesh (nearest neighbour) and a semi-implicit adaptive time-discretization.

45


The coupling between chemistry is transport is performed using a sequential iterative approach. The coupling with Flow and heat transfer are solved is explicit. Geometrical model The ore body lies within a heterogeneous aquifer formation. The underlying heterogeneity is two-fold and derives from the genesis of the ore (Dahlkamp 1991, Hobday and Galloway 1999). In a first time, the sedimentation under continental conditions builds a reduced, channelized aquifer. Second, an oxidizing front moves through the aquifer; mobile U(VI) in the oxidized part of the aquifer is then precipitated into U(IV) low solubility minerals such as uraninite. These genetic processes lead to two superimposed heterogeneities: lithofacies coarser or finer sands down to clays, chemical facies (oxidized, mineralized and reduced), and finally uranium grade within the mineralized facies. The model is based on block-model of the ore deposit, using geological observations and geostatistical reconstruction. A pattern of injection and production wells is set over the geometry. Geochemical model A geochemical model is devised to represent all three geochemical facies (oxidized, mineralized and reduced): Table. 1. Secondary minerals deriving from the interaction with Fe(III)-bearing sulfuric acid are included in the model. Finally, kinetics are set following literature data and tuned according to laboratory (e.g. Ben Simon 2011) and field observations (production data). Table 1: Mineralogical model for the facies of the simulation; kinetically controlled minerals denoted with a *. Minerals

Oxidized

Mineralized

Reduced

Uraninite Calcite Goethite Cristobalite* Montmorillonite-Mg* Kaolinite*

0 0 3.5 wg‰ 660 wg‰ 23 wg‰ 95 wg‰

variable 0.5 wg‰ 0 830 wg‰ 30 wg‰ 20 wg‰

0 0 0 830 wg‰ 30 wg‰ 20 wg‰

Results & Discussion The geochemical model was applied to simplified geometry in 1D and 2D with limited heterogeneity (i.e. a limited number of geometrically defined geochemical zones). The results or correct in terms of processes: progression of a pH front then oxidative dissolution of the uranium, and increased Fe(III) due to acidic dissolution of iron hydroxide in the oxidized zones. However, the production curves (pH, uranium in the production wells) are not representative of field observations: the front is to sharp and the tail distribution disappears too quickly. 46


Therefore, full 3D simulations were undertaken. Here again, a limited description of the heterogeneity failed to reproduce the observations data. Finally, a full heterogeneity model was undertaken: distribution of facies (oxidized, mineralized and reduced), distribution of uranium grade within the mineralized facies and permeability distribution according to the lithofacies: Figure. 1. The simulation follows the exploitation conditions: wells position, injection and production rate history (including closing wells, or newly developed injectors), and concentration of the injection fluids (acidity, majors and Fe(III) content).

Figure 1: Example of uranium distribution in the system – accumulation (i.e. integral concentration over the height of the aquifer) on the xy axis (in meters). The wells are figured: green=injection, red=production. The simulation was run over a 2-year period. The processes from the homogeneous simulations are reproduced, but the underlying heterogeneity produces interesting results. For instance, a streamline can flow through several geochemical facies; thus the fluid can sometimes be recharged in Fe(III) in situ, increasing its U-oxidative potential. Also, the fluid produced at a production well is the result of the mixing between streamline through U-rich and U-poorer areas, short and longer pathways (leading to different breakthrough rimes), fast and slower paths (leading to different kinetic controls). The mixing then yields an integrative concentration, much more dispersed than the homogeneous simulations. Hence, lower and wider U peak and much longer U concentration tail. Comparison with field data fits correctly, both in terms of uranium produced and acid consumption (balance of acid injected and reproduced): Figure. 2. This result is all the more satisfying that the number of fitting parameters is reduced to three parameters only.

47


Figure 2: Comparison of simulation results and field data: uranium concentration produced by the field mixing of all the producer wells (top), acid consumption (bottom). Conclusions The simulation of ISR exploitation gives very encouraging results, in that simulations can be performed using a limited number of fitting parameters. These simulations open the way to predictive simulations or a tool to test new production ideas. However, the simulation can only perform accurately if the geometrical description is sufficiently accurate. This leads to massive simulations (105+ cells), including numerous computationally challenging reacting fronts. These simulations were made possible by a combination of improved management of the code parallelization and performing computers. 48


References Ben Simon, R., 2011. Tests de lessivage acide de minerais d’uranium et modélisations géochimiques des réactions. Application à la récupération minière in situ (ISR). PhD thesis Mines ParisTech. Dahlkamp, F.J., 1991. Uranium ore deposits. Springer, Berlin Heidelberg New York. Hobday, D..K, Galloway, W.E., 1999. Groundwater processes and sedimentary uranium deposits. Hydrology journal, 7:127-138. Lagneau V. and J. van der Lee, J., 2010. HYTEC results of the MoMas reactive transport benchmark. Computational Geosciences, 14:435–449 van der Lee, J., De Windt, L , Lagneau, V. and Goblet, P., 2003. Module-oriented modeling of reactive transport with HYTEC. Computers and Geosciences, 29:265-275.

49


50


REACTIVE TRANSPOR MODELING OF THE INTERACTIONS OF CORROSION PRODUCTS AND BENTONITE: LONG-TERM GEOCHEMICAL EVOLUTION AND H2 GAS GENERATION Javier Samper1*, Acacia Naves1, Luis Montenegro1, Alba Mon1 and Tianfu Xu2 1 2

Civil Engineering School. University of A Coruña, 15071 A Coruña, Spain College of Environment and Resources, Jilin University, Changchun, Jilin 130021, China

* Corresponding author: [email protected] Introduction The assessment of the long-term performance of the engineered barrier system of a highlevel radioactive waste repository requires the use of reactive transport models for the interactions of the canister-corrosion products and the clay materials of the barrier (Lu et al., 2011; Xu et al. 2008). Here we present a reactive transport model for the long-term hydrochemical evolution of the bentonite porewater in the bentonite barrier of a spent-fuel, carbon-steel canister repository in granite. Numerical Model A 1D radial model of a disposal cell has been used to simulate canister corrosion and the long term evolution of the bentonite. It considers the canister and the bentonite (Figure 1). Both of them are assumed to be homogeneous. The canister external radius is 0.45 m and its thickness is 0.10 m. The external radius of the bentonite barrier is 1.2 m. Water flow and solute transport through the granite has been simulated with a prescribed water flux of granite water parallel to the axis of the gallery. The model accounts for canister corrosion, the chemical interactions of corrosion products and bentonite, mineral dissolution/precipitation, surface complexation of Fe2+ and H+ on three types of sorption sites and cation exchange reactions of Ca2+, Mg2+, Na+, K+ and Fe2+. It considers also the generation of H2(aq) which is allowed to diffuse through bentonite. Solute transport through the granite is simulated with a prescribed water flux parallel to the axis of the gallery which flushes the bentonite/granite interface. Long-term simulations (1Ma) were performed for the reference scenario at a constant temperature of 25ºC and for a set of variant scenarios. Although large corrosion rates are expected to take place just immediately after waste emplacement due to the presence of oxygen, the anoxic corrosion will be about 1 to 2 μm/year (King, 2008). A corrosion rate of 2 μm/year is assumed in the reference scenario. Surface complexation reactions in the bentonite are modelled with a triple sorption site model. The total concentration of sorption sites is 0.322 mol/L. There are 3 types of sorption sites. The first type of sites corresponds to the strong sites which have a large binding affinity but a small concentration (0.0079 mol/L). The other two types are the weak sites (the socalled weak 1 and weak 2) have binding constants weaker than those of the strong sites although their concentrations (0.16 mol/L) are larger than those of the strong sites. Table 1 51


lists the protolysis constants at 25ºC. Proton surface complexation plays a major role in controlling the pH of the bentonite porewater.

Figure 1: Sketch and finite element grid of the 1D axi-symmetric model. Table 1: Protolysis constants for surface complexation reactions for a triple-site model (Bradbury and Baeyens, 2005) at 25ºC. Reaction

Reaction

Log Kint

S   S   S OH 2    S OH  H

-4.5

S

S    S S O H    S OH

7.9

S

Log Kint

  W2  OH 2    S OH  H

W2

-6.0

S    S 2  S OFe  H    S OH  Fe 0.6

S

S   S 2  S OFeOH  2 H    S OH  Fe 10.0  H 2O

S

  W2 O H    S OH 10.5

W2 

  W1  OH 2    S OH  H

W1

-4.5

  W1 O H    S OH 7.9

W1 

S    S 2 2    SW 1OH3.3  S OFe(OH )2  3H   Fe  2 H O  SW 1OFe   H    Fe   S OH20.0 2

Model Results Model results for the reference scenario indicate that canister corrosion leads to a marked increase in pH and the concentration of dissolved Fe2+. The canister is fully corroded after 5·104 years for a constant corrosion rate of 2 μm/y. The largest pH in the bentonite is almost 9.5 at 2·105 years. Most of the released Fe2+ diffuses from the canister into the bentonite where it precipitates mainly as magnetite (Figure 2). Siderite precipitation is much smaller than magnetite precipitation due to the limited availability of dissolved bicarbonate. Precipitation of corrosion products progresses as Fe2+ diffuses into the bentonite. It takes place in a 7-8 cm thick volume of bentonite around the canister/bentonite interface and 52


leads to a very significant decrease of bentonite porosity. Released Fe2+ also sorbs by surface complexation and undergoes cation exchange. The competition of Fe2+ and H+ for the sorption sites near the canister/bentonite interface causes several sorption fronts which induces fronts on pH, Eh, Fe2+ concentration and mineral dissolution/precipitation (Figure 3). Therefore, Fe2+ sorption plays a very relevant role in the geochemical evolution of bentonite. Only a small part of the Fe2+released by canister corrosion is sorbed at exchanged sites. The evolution of the concentration of exchanged Fe2+ is related to that of sorbed and dissolved Fe2+. 0.45 0.4

102y

0.35

Porosity

0.3 0.25 104 y

0.2

106 y

0.15 0.1 0.05 0 4.5

4.75

5

5.25 r (dm)

5.5

5.75

6

Figure 2: Spatial distribution of magnetite precipitation in bentonite at t = 103, 104, 105 and 106 years (left) and spatial distribution of the change in bentonite porosity due to mineral dissolution and precipitation at selected times. r is the radial distance to the axis of the disposal cell.

Figure 3: Time evolution of pH and the dissolved concentration of Fe2+ at r = 1 and 50 cm from the canister/bentonite interface. Fronts caused by competition for the sorption sites can be observed at 1 cm from the interface after 80 years. Conservative species such as Cl- present a pattern of decreasing concentration with time in the bentonite because they diffuse from the bentonite into the granite. Dissolved cations, Ca2+, Mg2+, Na+ and K+, show trends similar to those of conservative species but they are also subjected to mineral dissolution/precipitation and cation exchange processes. Calcite dissolves in most of the bentonite except near the canister where it precipitates due to the increase in pH induced by canister corrosion. Dissolution/precipitation of quartz and gypsum are not significant. The computed concentrations of exchanged cations in the bentonite vary with time due to changes in cation porewater concentration. The concentration of exchanged Ca2+ increases after 1 Ma while those of Na+ and Mg+ decrease. The partial 53


pressures of H2(g) have been calculated from the computed activities. They increase while the canister is being corroded until significantly high values and decrease once the canister has been fully corroded. The effects of mineral dissolution and precipitation in porosity have been evaluated. The reduction of bentonite porosity due to mineral precipitation near the canister/bentonite interface could be large and result in clogging of the bentonite pores. Model predictions indicate that the changes in porosity will be negligible in the rest of the buffer (Figure 3) Figure 4 shows the time evolution of computed concentration of sorbed species in the bentonite at r = 4.6 dm. The competition between the dissolved Fe 2+ and H+ for the weak 1 sorption sites produces two sorption fronts. The first one occurs at t = 80 years when the concentration of SW1OH decreases while that of SW1OFe+ increases. The concentration of the surface complex SW1O- of weak 1 sorption sites increases after 3000 years producing a decrease in the concentration of SW1OFe+. There is no competition between Fe2+ and H+ for the weak 2 sorption sites because the model does not consider Fe 2+ sorption on such sites. This is the main reason why the concentration of SW2OH is the largest during most of the simulated time. The sorption front at t = 80 years is shown also as an increase in the concentration of SW2OH and a decrease of SW2OH2+. The concentration of the strong sites is much smaller than those of the complexes attached to the weak sites. Sorbed complexes at r = 4.6 dm 0.18

S(s)-HO

S(w2)-OH

0.16

S(s)-OH2+ S(s)-O-

0.14

mol/L

0.12

S(w1)-OH

S(s)-O Fe+

S(w1)-OFe+

S(s)-O FeOH S(s)-O Fe(OH)2-

0.1

S(w1)-OH

0.08

S(w1)-OH2+

0.06

S(w1)-O-

S(w1)-O-

0.04

S(w1)-O Fe+

0.02 0 1.E+01

S(w2)-OH

S(w2)-O-

S(w2)-OH2+ S(w2)-O1.E+02

1.E+03 1.E+04 Time (years)

1.E+05

1.E+06

Figure 4: Time evolution of the computed concentrations of the sorbed in the bentonite at r = 4.6 dm. Sensitivity analyses Prediction uncertainties were evaluated for the following parameters: 1) The corrosion rate; 2) The effective diffusion coefficient De of the dissolved species in the bentonite; 3) The water flow at the bentonite/granite interface; 4) The cation exchange selectivities and 5) The chemical compositions of the bentonite and granite porewater compositions. Sensitivity runs were performed for corrosion rates ranging from 0.1 to 5 μm/year. The larger the corrosion rate, the sooner and the closer to the canister/bentonite interface the sorption, pH and Eh fronts take place (Figure 5). The larger the corrosion rate, the larger the magnetite and siderite concentration close to the canister/bentonite interface but the smaller their 54


penetration into the bentonite. Consequently, the larger the corrosion rate, the faster the porosity reduction close to the canister but the smaller the thickness of affected bentonite. In addition, the smaller the corrosion rate the smaller the peak of hydrogen pressure in the bentonite. 20000 0.1 μm/y

Pressure of H2(g) (atm)

18000

0.5 μm/y

16000

1 μm/y

14000

2 μm/y

12000

3 μm/y

10000

5 μm/y

8000 6000 4000

2000 0 1.00E+03

1.00E+04

1.00E+05

1.00E+06

Time (years)

Figure 5: Sensitivity of magnetite precipitation (left) and computed time evolution of the H2(g) partial pressure in the bentonite (right) to changes in the corrosion rate. Conclusions The interactions of corrosion products and bentonite may lead to a drastic decrease of bentonite porosity and large H2(g) pressures. The main conclusions of the sensitivity analyses include: 1) The larger the corrosion rate, the larger the pH, the larger the concentration of precipitated magnetite near the canister/bentonite, the larger the zone where corrosion products precipitate and the larger the H2(g) partial pressure; 2) The De of bentonite affects the concentration of the dissolved Fe2+ and the precipitation of the corrosion products; 3) The computed concentrations of dissolved species, the sorption fronts and the concentrations of precipitated magnetite and siderite are very sensitive to the water flow in the granite; 4) The thickness of bentonite affected by pore clogging is sensitive to all the investigated parameters; 5) The cation selectivities affect mostly the concentration of exchanged cations. However, the computed pH, Eh and the concentrations of dissolved and precipitated species lack sensitivity to the selectivities; and 6) Model results are sensitive to the chemical compositions of the bentonite and granite porewaters. The general patterns of pH, Eh, H2(g) pressure and magnetite precipitation, however, are similar to those of the reference run. Acknowledgments The research leading to these results has received funding from the European Atomic Energy Community’s Seventh Framework Programme (FP7/2007-2011) under grant agreement 249681. This work was partly funded by ENRESA (Spain) and a project from the Ministry of Economy and Competitiveness (Project CGL2012-36560). We acknowledge the contributions of Juan Carlos Mayor from ENRESA.

55


References Lu, C., J. Samper, B. Fritz, A. Clement and L. Montenegro 2011. Interactions of corrosion products and bentonite: An extended multicomponent reactive transport model. Phys Chem Earth 36, 1661–1668. King, F. 2008. Corrosion of carbon steel under anaerobic conditions in a repository for SF and HLW in Opalinus Clay. Nagra Technical Report 08-12. Xu, T., R Senger and S Finsterle 2008. Corrosion-induced gas generation in a nuclear waste repository: Reactive geochemistry and multiphase flow effects. Applied Geochemistry, 23(12):3423–3433.

56


A WRAP UP OF MODELING SORPTION PROCESSES OF Eu3+ UNDER VARYING GEOCHEMICAL CONDITIONS S. Britz1*, U. Noseck1, V. Brendler2, W. Durner3 and D. Zachmann4 1

GRS Braunschweig, Theodor-Heuss-Str. 4, 38122 Braunschweig, Germany Helmholtz-Zentrum Dresden-Rossendorf, P.O.Box 510119, D-01314 Dresden, Germany 3 TU Braunschweig, Institut für Geoökologie, Langer Kamp 19c, 38106 Braunschweig, Germany 4 TU Braunschweig, Institut für ökologische und nachhaltige Chemie, Hagenring 30, 38106 Braunschweig, Germany 2

*corresponding author: [email protected]

Introduction In safety assessments for radioactive waste repositories detailed knowledge about surface processes regarding long-term safety relevant elements (e.g. Am3+) is of fundamental importance. Surface reactions such as sorption as well as other processes lead to retardation and influence radionuclide migration. These processes occur in nature and therefore, they must be considered for realistic transport calculations. Amongst others they function as a natural barrier and might reduce contaminant dissemination of potentially hazardous pollutants in natural environments. So far, the Kd-concept has been applied using distribution coefficients constant in time and space to describe radionuclide transport in the far field of a repository. This approach does not take temporally and spatially changing geochemical conditions into account. In order to include varying geochemical conditions the smart Kd-concept based on thermodynamic sorption models was developed (Brendler et al., 2002). Combining the state-of-the-art codes r3t (radionuclide, reaction, retardation, and transport) with d 3f (distributed, density-driven flow) it is possible to model transport processes as a function of important environmental parameters e.g. pH, pCO2, ionic strength, Ca concentration, DIC, and radionuclide concentration. By including the smart Kd-concept into r3t, multidimensional Kd-matrices are calculated for each radionuclide and sediment a-priori, which are subsequently applied for reactive transport calculations (Stockmann et al., 2014). To model sorption processes that are controlled by geochemical conditions robust data sets of so called surface complexation parameters (SCP) are required. These parameters, such as protolyses constants (pK-values), specific surface area (SSA) and surface site density (SSD) as well as stability constants of surface complexes (logK-values) are derived from measurements. Here SCP are obtained by fitting experimental data sets applying the geochemical speciation code PhreeqC (Parkhurst and Appelo, 1999) in combination with the parameter estimation code UCODE (Poeter et al., 2005). The SCP are iteratively optimized by UCODE to obtain the best fitted data set. Thereby derived SCP are subsequently applied as fixed parameters in reactive transport models. 57


Literature studies revealed that for important mineral phases such as muscovite and orthoclase nearly no SCP data sets are available at present (Noseck et al., 2012). Hence, we performed extensive laboratory studies to derive experimental data. This study describes the approach to assess surface complexation parameters of Eu3+ (as a homologue for trivalent actinides). Experimental data of titration, batch and column experiments are discussed and evaluated. Exemplarily, we demonstrate the approach for muscovite. However, this method may be applied to any mineral or sediment of interest. Material & Methods Material Muscovite was mined in China and distributed by Normag GmbH (Homburg, Germany). Information regarding preconditioning and mining procedures of the vendor were considered confidential and thus not accessible. However, due to its origin naturally occurring impurities were represented in all samples offering realistic experimental conditions. Therefore, extensive geochemical analyses but also physical characterizations were performed (XRD, REM, EDX). The mineral was sieved and a grain size fraction of 63 – 200 µm was used in all experiments. Experiments To derive all data for reactive transport simulations three experimental set ups were necessary. Titration experiments and batch experiments were conducted to obtain SCP. Column experiments provided transport data that was applied to finally model reactive transport processes under varying geochemical conditions. The experimental set up for potentiometric titration experiments as well as data processing follows Lützenkirchen et al. (2012). Potentiometric titration experiments For minerals, which are not purified, via discontinuous potentiometric mass titrations an equilibrium pH (pHequ) is determined referring to the solid-liquid ratio (M/V [g L-1]) at which the pH of the suspension is independent of additionally added solid and independent of the initial pH value of the suspension. For purified minerals this pH equals the point of zero net proton charge (pHpznpc). To determine pHequ of the applied muscovite rising M/V ratios were prepared in degassed 10-5 mol L-1 NaClO4 background solution under argon (up to a ratio of 900 g L-1). pH was measured while the samples were flushed with argon to minimize atmospheric CO2 bias. Equilibrium pH was collected after 24 h, 7 d, 14 d, and 6 months (Figure 1a)). Following mass titration experiments, electrolyte titrations (discontinuous) were conducted. This titration technique determines the influence of ionic strength on the pHequ of unpurified minerals. A M/V ratio of 500 g L-1 was applied representing the solid content of pHequ (Figure 1a)). Increasing amounts of NaClO4 salt were added to the suspensions; final NaClO4 concentrations ranging from 10-6 mol L-1 to 3 mol L-1. pHequ of these samples change according to the applied electrolyte concentrations. The new equilibrium pH is set to the point of zero net proton and hydroxide consumption for muscovite for the 58


particular electrolyte concentration of 10-2 mol L-1 NaClO4 (pHpznphc, Lützenkirchen et al.(2012)). Equilibrium was reached after 6 months. Obtained information was employed to conduct continuous potentiometric acid-base titration. 500 g L-1 muscovite were equilibrated in 10-2 mol L-1 NaClO4 over 44 d to reach a constant pH of 3.8 (addition of HNO3). Applying this equilibrated suspension a titration was conducted via the addition of 0.1 mol L-1 NaOH using a combined glass electrode (Metrohm, Syntrode, 6.0248.020) in combination with an automatic titration device (Titrando 808, software Tiamo 2.3, equilibrium criterion 0.1 mV/min). Resulting data was corrected for the determined pHpznphc of muscovite. The net surface proton excess is determined. According calculations are described elsewhere (Lützenkirchen et al., 2012). Batch experiments Muscovite batch experiments were conducted applying two M/V ratios (1/0.02, 1/0.08 g L -1), four ion concentrations (10-5 mol L-1, 10-6 mol L-1, 10-7 mol L-1, 10-8 mol L-1 Eu3+), and four initial pH values ranging between 4 – 9. As an inert background electrolyte 10-2 mol L-1 NaClO4 was used. Initial pH values were adjusted over approximately three months by adding NaOH and HCl. After the pH was constant over at least 24 h Eu3+ was spiked to set up the target concentrations of the suspensions. Sorption equilibrium of Eu 3+ was reached after additional 24 h. Samples were then centrifuged and filtered with 0.02 µm syringe filters. Eu 3+ concentrations in the supernatant were determined via ICP-MS (Thermo Electron Corporation, X Series, SV 1001C). For more information see Noseck et al. (2012). Column experiment Column experiments were conducted with specifically developed perfluoralkoxy polymer columns (length 12.2 cm, ID 4.1 mm). Muscovite was packed into the column to a constant bulk density of b = 1.39 g cm-3 throughout the entire column length. The column was saturated and preconditioned over four months. The background electrolyte solution (10-2 mol L-1 NaClO4) was injected from the bottom of the column via a peristaltic pump (const. flow rate of 0.21 ± 0.01 ml min-1). Samples were collected in fractions every 13 min. The percolating solution was then changed to a peak solution that additionally contained 9.06*10-6 mol L-1 NaBr (tracer) and 10-5 mol L-1 Eu3+. After 1.5 pore volumes, the solution was changed back to the background electrolyte. The experiment lasted for 14 days. Simultaneous sample analyses via ICP MS yielded the tracer break-through curve. Due to high retardation of the Eu3+ by muscovite no Eu3+ breakthrough was observed, hence the spatial distribution of Eu in the column needed to be determined. Therefore, the column was split into 0.5 cm slides and the samples were leached in 5% HNO3 for at least 24 h. The suspensions were filtered with 0.2 µm syringe filters and the leachates analysed via ICP MS. Modelling approach The geochemical speciation code PhreeqC in combination with the parameter estimation code UCODE were applied to fit the experimental data and to estimate the SCP. First, obtained corrected potentiometric titration curves yielded pK-values and SSD. Determined SCP (pK-values, SSD, SSA) were applied in subsequent calculations of batch experiment data. Finally, within reactive transport models SCP derived from titration and batch experiments 59


were applied as fixed data sets. Differences between modeled reactive transport data compared to experimental ones can be adapted by fitting transport boundary conditions. Verification calculations are conducted with STANMOD (Simunek et al., 1999). Results & Discussion So far, batch and titration experiments have been successfully completed. Column experiments with muscovite are still ongoing. Evaluation via PhreeqC and STANMOD are planned in early 2014. Results will be presented during the workshop. Combination of mass-, electrolyte-, and potentiometric titration accounts for mineral dissolution, cation-exchange, and changes in edge surface charge. Figure 1 shows results from potentiometric mass (a)) and electrolyte (b)) titrations of muscovite. Data sets with different equilibration times are presented (24 h, 7 d, 14 d, 6 months). Since equilibration was not definitely reached after 24 h only data from equilibration times of 6 months (electrolyte titration) and 14 d (mass titration) are discussed here. The plateau of the potentiometric mass titration (Figure 1 a)) equaled pHequ = 7.8 referring to a M/V ratio of 500 g L-1. The reached pH equilibrium is independent of the initial pH value of the suspension and independent of an increase of the M/V ratio.

Figure 1: a) Muscovite mass titration - determination of pHequ = 7.8. Equilibrium was reached after 14 d. b) Muscovite electrolyte titration (M/V = 500 g L-1), determination of pHpznphc = 7.8. Blank titration (black x) shows pH development of the samples without mineral influences. Equilibrium was reached after 6 months. After 24 h or 7 d suspensions of the electrolyte titration had not reached the pHequ determined via mass titration. Hence, pH measurements were repeated after 6 months. Meanwhile, samples were stored under Ar, dark, and cool. After 6 months suspensions showed no decrease of the pHequ referring to an electrolyte concentration of 10-2 mol L-1 NaClO4, which represented the chosen electrolyte concentration of batch and column experiments. Hence, the obtained pHpznphc for the applied muscovite was 7.8. It should be noted that conducted mass titration experiments under equal geochemical conditions with orthoclase also showed no differences in pHpznphc compared to pHequ. Therefore, it can be assumed that electrolyte influences are significantly apparent at NaClO4 concentrations > 0.1 mol L-1. 60


ƩH+,OH- [µmol/L]

Figure 2 displays corrected (x) and original (o) continuous acid-base potentiometric data of muscovite. The point of zero net proton and hydroxide consumption assessed via continuous acid-base muscovite titration of pH 4.7 (Figure 2) was shifted to pHpznphc = 7.8. Original data was recalculated (Lützenkirchen et al., 2012). Consideration of the pHequ and pHpznphc are significant to assess robust SCP. 16 12

pH pznphc = 7.8

8 4

pH

0

-4 -8 -12

3

4

5

6

7

8

9

10

11

pot. titration data corrected data

Figure 2: Continuous potentiometric acid-base titration of muscovite in 0.01 mol L-1 NaClO4: corrected (x, net proton surface excess) and uncorrected (o, relative proton surface excess) data (M/V 500 g L-1 ≙ 10.8 m2 500 g-1 mineral).The proton and hydroxide consumption vs. pH is displayed. Figure 3 illustrates batch experiment data of muscovite. Results follow expectations: explicit correlations between M/V ratio and sorbate concentrations as well as correlations between initial ion concentration and pH were observed. The sorption edge was strongly pH dependent and restricted to the pH range between 4 and 6. Minimum sorbate concentrations at low pH values did not fall below 20% due to cation exchange.

Figure 3: Batch experiment data set of muscovite in 10-2 mol L-1 NaClO4 background electrolyte. M/V ratio [g ml-1]. The application of the geochemical speciation code PhreeqC is scheduled for upcoming months. Input files have already been developed. However, due to pending data of column experiments modelling muscovite SCP and transport processes have not been started with yet. Results will be presented during the workshop.

61


Conclusions Major objective of the overall project is to couple changing, geochemically driven reactions with transport processes in sediment and rock formations through the application of the smart-Kd-concept (Stockmann et al., 2014). Therefore, robust sorption data sets are necessary. Since data gaps were identified this study contributes lacking sorption and transport data. As a homologue for long-term safety relevant actinides (Pu3+, Cm3+ and Am3+, etc.) sorption processes and characteristics of Eu3+ are analysed and discussed. Here, titration and batch experiment data of muscovite were introduced giving a first overview of Eu sorption processes. We found that titration experiment data need correction: the point of zero net proton charge determined by the continuous potentiometric titration is corrected according to the pHequ. The influence of ionic strengths is less important here due to its low concentration of 0.1 mol L-1 (pHequ = pHpznphc). Only titration experiments with electrolyte concentrations > 0.1 mol L-1 would need significant correction (pHequ ≠ pHpznphc). Batch experiments showed expected sorption tendencies, namely an increase of sorption with rising pH values, increasing M/V ratio and decreasing element concentration. The effect of cation exchange capacity of muscovite was also identified. Future evaluation of assessed titration and batch experiment data with the geochemical speciation code PhreeqC in combination with the parameter estimation code UCODE will yield necessary SCP in order to model coupled reactive transport processes in porous media. Collected and modelled data will be introduced and discussed. Acknowledgements Experiments are part of a cooperation project between HZDR and GRS Braunschweig and is funded by the German Federal Ministry of Economics and Technology (BMWi) under contract numbers 02 E 10518 and 02 E 10528 as well as 02E 11072A and 02E 11072B. We would like to thank all project members for constructive discussions. References Brendler, V., Arnold, T., Nordlinder, S., Zänker, H., Bernhard, G., 2002. Speciation and Sorption for Risk Assessment: Modelling and Database Application. In: H.D. Schulz and G. Teutsch (eds). Geochemical Processes. Conceptual Models for Reactive Transport in Soil and Groundwater. Wiley VCH, Weinheim: p.79-94. Fein, E., 2004. Model for transport and retention in porous media. GRS-192, BMWA-FKZ 02E9148/2, Gesellschaft für Anlagen- und Reaktorsicherheit (GRS) mbH, Braunschweig: 319 p. Lützenkirchen, J., Preočanin, T., Bauer, A., Metz, V., Sjöberg, S., 2012. Net surface proton excess of smectites obtained from a combination of potentiometric acid-base, mass and electrolyte titrations. Colloids and Surfaces A: Physiochemical and Engineering Aspects, 412: 11-19. Noseck, U., Brendler, V., Flügge, J., Stockmann, M., Britz, S., Lampe, M., Schikora, J., Schneider, A., 2012. Realistic integration of sorption processes in transport codes for 62


long-term savety assessments. GRS-297, BMWi-FKZ 02E10518, Gesellschaft für Anlagen- und Reaktorsicherheit (GRS) mbH, Braunschweig: 293 p. Parkhurst, D. L., Appelo, C. A. J., 1999. User’s guide to PHREEQC (Version 2) – A computer program for speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations. U.S. Geological Survey Water-Resources Investigation Report, 99-4259: 312 p. Poeter, E. P., Hill, M. C., Banta, E. R., Mehl, S., Christensen, S., 2005. UCODE_2005 and six other computer codes for universal sensitivity analysis, calibration, and uncertainty evaluation. U.S. Geological Survey Techniques and Methods, 6-A11: 299 p. Simunek, J., van Genuchten, M. Th., Sejna, M., Toride, N., Leij, F. J., 1999. The STANMOD computer software for evaluating solute transport in porous media using analytical solutions of convection-dispersion equation. U. S. Salinity laboratory, Agricultural Research Service, U. S. Department of Agriculture, Riverside, California: 20 p. Stockmann, M., Flügge, J., Schikora, J., Noseck, U., Brendler, V. Smart Kd-concept in reactive transport modeling for radioactive waste repositories. To be published 2014.

63


64


COMPUTATIONAL CHALLENGES FOR REACTIVE TRANSPORT MODELLING: INTERFACING AND HIGH PERFORMANCE COMPUTING Jorge Molinero Amphos 21 Consulting, Passeig de Garcia i Faria, 49-51. 08019 Barcelona *Corresponding author: [email protected]

Introduction iMaGe stands for interfacing Multiphysics and Geochemistry and aims to launching a platform to foster the development of software solutions for communicating geosciences modelling codes. The idea behind iMaGe is combining and optimizing the capabilities of different specific codes that are readily available. Normally these codes have been developed over years and are largely tested and verified. The primary iMaGe goal is to achieve the best performance for modelling complex coupled phenomena in the subsurface environment, while avoiding code writing from scratch. We always look for a compromise between innovation and avoiding reinventing the wheel. Interface Comsol-Phreeqc (iCP). Application to CO2 sequestration iCP stands for interface Comsol-Phreeqc. It is an interface to couple Comsol and Phreeqc. The result of this interface is a tool for solving a wide range of multiphysics and chemical problems. It is written in Java and uses the IPhreeqc C++ dynamic library and the Comsol Java API. Furthermore, it takes advantage of the multicore computer architecture by balancing the computational load over different threads. In this work we present a modelling exercise in the framework of underground injection of CO2 in the Duero Basin (Spain). The Duero basin in NW Spain is one the most promising basin for CO2 storage in the Iberian Peninsula due to the existence of favourable deep aquifers close to large CO2 emission point sources. A number of projects are presently active either for scientific research (e.g., the Hontomín site, OXI-CFB300 EPRR project) or commercial purposes (e.g., Sahagún and Los Páramos projects). The project called Los Páramos intends to assess the injection of CO2 in a group of dome-shaped structures with an estimated total capacity of 200 Mt (ranked 2nd in the Iberian Peninsula, IGME 2010). These domes were studied in the past for hydrocarbon exploration and a large body of information is available from seismic profiles (over 170 km) and 3 deep wells. The Los Páramos site is emplaced in the San Pedro Folded Band (SPFB) that consists mainly of thick-skinned thrusts of Mesozoic rocks (Triassic and Upper Cretaceous) sealed by a thick (1200-1500 m), undeformed cover of 65


Tertiary claystones. Dome-like structures are related to thrusts leading to favourable reservoirs. The target horizon for CO2 storage is the Utrillas Fm sandstone with high porosity (13-20%) and thickness (225-250 m). In three of the domes, the Utrillas Fm is below -800m, allowing thus the storage of CO2(sc). This sandstone hosts an aquifer containing saline water, up to 50 g·L-1, according to the data from drill wells. The presence of saline groundwater is explained by water interaction with Triassic evaporite layers just underlying the Utrillas Fm sandstones. The CO2 storage at Los Paramos site is planned via injection of supercritical CO 2 (CO2(sc)) in the Utrillas Fm. In general, the next four trapping mechanisms are expected, which are of increasing importance through time (1) structural, (2) residual saturation, (3) dissolution, and (4) mineral. The prediction of the mass of CO2 stored through time in any storage systems is an essential parameter in the pre-injection assessment of a geological storage. For safety reasons, it is relevant to know the mass of CO2 trapped under the different trapping mechanisms. In this work, storage quantification in the Dome B of Los Páramos site has been performed by using multiphase transport simulations with COMSOL Multiphysics. Model results predict well the amount of CO2 trapped as residual phase and the onset of the formation of CO 2-rich brine fingers and their extent and evolution. Interface DarcyTools-PFLOTRAN (iDP). Application to grout degradation in the deep geological repository in Sweden. iDP stands for interface DarcyTools-PFLOTRAN. It is an interface to couple DarcyTools and FLOTRAN. The main target is to develop an integrated reactive transport platform consisting of Darcy Tools (complex flow simulations) and PFLOTRAN (High Performance Computing for geochemical simulations) in order to incorporate DarcyTools (ability) with the reactive transport capabilities of PFLOTRAN such as the high performance computing (HPC) with the help of its internal libraries such as PETSc. The interface is written in a modular, crossplatform and extensible way using Fortran and Python. iDP has been applied for the first time in the Mare Nostrum III, a new facility of the Barcelona Supercomputing Centre. A total of 25,000 processor cores during 5 days were used to solve a large-scale (100 Mcells), long-term (10,000 years) simulation of the hydrogeochemical behaviour of an hyperalkaline plume produced by the dissolution of grout used during the construction of a deep geological repository for spent nuclear fuel. The simulation integrates the complex 3D groundwater flow accounting for the Discrete Fracture Network (DFN) of the site, and the complexity of the geochemical system involved in cement grout dissolution and secondary minerals precipitation within the flowing fractures.

66


ALGORITHM FOR FLUID VELOCITY FIELD QUANTIFICATION FROM IMAGE SEQUENCES IN COMPLEX GEOMATERIALS Nico Korn1 and Johanna Lippmann-Pipke1* 1

Helmholtz-Zentrum Dresden-Rossendorf, Institute of Resource Ecology, Permoser Str. 15, 04318 Leipzig, Germany *Corresponding author: [email protected] Introduction For reactive transport modelling in geosciences a velocity field v(xi,t) with i = 1, 2, or 3 is required. This velocity field can be a) obtained by definition, b) calculated on the basis of a given geometry, a set of partial differential equations and initial and boundary conditions or, c) obtained from observations. By now, various tomographic methods have been applied to observe fluid flow also in dense geological material under realistic conditions. These are magnetic resonance imaging (MRI, e.g. (Greiner et al., 1997), neutron transmission tomography (e.g. (Pleinert and Degueldre, 1995), X-ray computed tomography (CT) (e.g. (Goldstein et al., 2007; Klise et al., 2008), electrical resistivity tomography (ERT), e.g. (Bowling et al., 2006; Gheith and Schwartz, 1998) and last but not least positron emission tomography (PET), e.g. (Khalili et al., 1998; Kulenkampff et al., 2008; Richter et al., 2005). Still, the extraction of quantitative velocity fields from observed concentrations fronts that pass through complex geological media is not a trivial task. One option to solve this kind of problem is to inject a tracer pulse into a sample, record image sequences of the tracer's flow through the sample and generate local break through curves (BTC). Numerical simulations – e.g. on the basis of finite difference methods (e.g. (Yoon et al., 2008) – may then vary hydraulic conductivity, porosity and dispersivity values within appropriate ranges and evaluate the model fits of the data over different scales. The authors conclude that predicting water flow at fine scales (relative to permeability variations) is very challenging and that this may have large implications for modelling reactive transport, where reactant residence time and mixing can be greatly impacted by water flowpaths. Material & Methods To overcome such problems that accompany the fitting of parameter values such as hydraulic conductivity, porosity or dispersivity in 3D numerical flow models, we designed, implemented and tested a new algorithm. It is conceptualized for application to real-world 3D PET image series of transport process observation in geological media that may be affected to some degree by noise, image artifacts and detection limits. Our algorithm does 67


not need prior information about the internal geometry of the sample, but only the global flow rate and the geometric boundaries of the sample. Still, the foundation of the algorithm is the continuity equation. Its validity serves as an optimization criterion to fit segments of flow paths to the images. In this way, the network of flow paths is recovered and the velocity can be computed using a robust and universal approach. We model the flow path network inside a rock sample as a network of flow path segments. Each segment is a straight and typically a short part of a single flow path. For these segments, we assume that the fluid is incompressible and that there are no sinks and sources within a segment. These assumptions are typically true for water in a closed flow path. As a first step, the algorithm identifies regions that show a significant increase in mass at some point in time (maxima in a BTC). At such regions nodes are placed, that are to be connected later with segments of the flow path network. For each straight flow path segment, it is sufficient to use a 1D model, which greatly reduces computation time without sacrificing much accuracy, and makes the algorithm more robust against noise. For the algorithm, a segment is represented by a cylindrical tube that completely covers the flow path. Because there are no sinks and sources, the flow rate is constant when the tube covers exactly one flow path, but varies when it does not. Therefore, we can use the variation of the flow rate as an optimization criterion to decide where to place a tube, i.e. which of the aforementioned nodes to connect. Finally, we can compute the velocity field from the flow rate and cross-sectional area of the tube. Results & Discussion For validating the algorithm we simulated a non-reactive tracer experiment in COMSOL Multiphysics® on a synthetic fracture network as a benchmark model for the algorithm. (Later, the image sequences obtained from the transport simulation are to be replaced by the PET image sequences). A velocity field (derived using the cubic law) was used to simulate transport of a conservative tracer. The resulting image sequence was provided to our new algorithm, which then computed the underlying velocity field.

Figure 1: Left: Synthetic fracture network (hight= 8.7cm, width=13.6cm). Right: Simulated transport of a rectangular tracer pulse applied from a inflow region at the bottom and allowed to exit through a central outflow boundary at the top. Total duration of experiment is 936s or 15min. 68


Figure 1, left, shows the network used for the flow simulation; figure 1, right, the resulting image sequence returned by the non-reactive transport simulation. Figure 2 shows a comparison of the velocity field obtained by our algorithm versus the velocity field underlying the non-reactive transport simulation. Our algorithm is able to recover key features and reproduces a velocity field that agrees well in most regions with that of the simulation. From the perspective of contaminant transport, it is worth noting that the velocity field is never underestimated, even when parts of the network cannot yet be fully reconstructed by the algorithm (white areas in figure 2, left).

0.0015m/s

0 m/s

Figure 2: Left: Velocity field from our algorithm. Right: Velocity field of simulation that generated the transport process images as shown in figures 1, right. Conclusion Here we introduced our new algorithm (work in progress) that estimates velocity distributions from image sequences. It is robust against noise and static image artefacts, and requires no prior knowledge about the, possibly complex geometry of the sample. The current run-time for the example shown is well under ten seconds (using an Intel Core 2 6600 consumer pc ). These properties make the new algorithm universally suitable for tracer experiments for a wide range of applications. The obtained velocity distributions (fig. 2, left) can directly be used for further reactive transport modelling. References Bowling, J.C., Zheng, C., Rodriguez, A.B. and Harry, D.L., 2006. Geophysical constraints on contaminant transport modeling in a heterogeneous fluvial aquifer. Journal of Contaminant Hydrology, 85(1–2): 72-88. Gheith, H.M. and Schwartz, F.W., 1998. Electrical and visual monitoring of small scale threedimensional experiments. Journal of Contaminant Hydrology, 34(3): 191-205.

69


Goldstein, L., Prasher, S.O. and Ghoshal, S., 2007. Three-dimensional visualization and quantification of non-aqueous phase liquid volumes in natural porous media using a medical X-ray Computed Tomography scanner. Journal of Contaminant Hydrology, 93(1–4): 96-110. Greiner, A., Schreiber, W., Brix, G. and Kinzelbach, W., 1997. Magnetic resonance imaging of paramagnetic tracers in porous media: Quantification of flow and transport parameters. Water Resources Research, 33(6): 1461-1473. Khalili, A., Basu, A.J. and Pietrzyk, U., 1998. Flow visualization in porous media via positron emission tomography. Physics of Fluids, 10: 1031-1033. Klise, K.A., Tidwell, V.C. and McKenna, S.A., 2008. Comparison of laboratory-scale solute transport visualization experiments with numerical simulation using cross-bedded sandstone. Advances in Water Resources, 31(12): 1731-1741. Kulenkampff, J., Richter, M., Gründig, M. and Seese, A., 2008. Observation of transport processes in soils and rocks with Positron Emission Tomography. Geophysical Research Abstracts, 9: 02754. Pleinert, H. and Degueldre, C., 1995. Neutron radiographic measurement of porosity of crystalline rock samples: a feasibility study. Journal of Contaminant Hydrology, 19(1): 29-46. Richter, M., Gründig, M., Zieger, K., Seese, A. and Sabri, O., 2005. Positron Emission Tomography for modelling of geochmical transport processes in clay. Radiochimica Acta, 93: 643-651. Yoon, H., Zhang, C., Werth, C.J., Valocchi, A.J. and Webb, A.G., 2008. Numerical simulation of water flow in three dimensional heterogeneous porous media observed in a magnetic resonance imaging experiment. Water Resources Research, 44: W06405.

70


COUPLING GROUNDWATER FLOW AND TRANSPORT IN FEFLOW WITH ENVIRONMENTAL GEOCHEMISTRY IN PHREEQC Laurin Wissmeier 1*, Thomas Kämpfer 1 and Stefan Wilhelm 1 1

AF Consult Switzerland

* Corresponding author: [email protected] Introduction We present PHREEQC4FEFLOW, a software tool for simulations of subsurface flow and solute transport in two and three-dimensional domains in combination with comprehensive intra-phase and inter-phase geochemistry. PHREEQC4FEFLOW is a plugin for the professional subsurface flow and transport simulator FEFLOW, which employs the interface version of the geochemical modelling framework PHREEQC as a reaction engine. FEFLOW is routinely used by state authorities and industry to assess groundwater related issues such as extraction capacity, vulnerability and pollution control. The coupling with PHREEQC gives major advantages over FEFLOW’s built-in reaction capabilities, i.e., the soil solution is speciated from its element composition according to thermodynamic mass action equations with ion activity corrections. State-of-the-art adsorption models such as surface complexation with diffuse double layer calculations as well as equilibrium and kinetic mineral reactions are accessible. In addition, PHREEQC provides a framework to integrate user-defined kinetic reactions with possible dependencies on solution speciation (i.e., pH, redox state, saturation indices, and ion activities), allowing for modelling of microbially mediated reactions. Extensive compilations of geochemical reactions and their parameterization are accessible through associated databases. Material & Methods We will discuss the coupling procedures with the implemented split operator, the treatment of negative component concentrations and the convergence criteria for transport. In addition we will comment on the program capabilities but also on important limitations. Results & Discussion In a first application, we present the validation of the coupling using an example for transient infiltration with pesticide degradation, where the kinetic degradation can be simulated entirely by FEFLOW. The comparison of results from FEFLOW alone and the PHREEQC4FEFLOW coupling shows small differences that can be attributed to the employed non-iterative sequential split operator. In the following, we demonstrate the versatility of 71


PHREEQC4FEFLOW by modelling problems from the fields of agricultural management as well as pollution control and waste disposal.

72


WATER-SOLID INTERACTIONS AT THE PORE SCALE Lionel Mercury1*, Kirill Shmulovich2, Isabelle Bergonzi1 and Jean-Michel Matray3 1

Institut des sciences de la Terre d’Orléans, UMR 7327 Université d’Orléans/ CNRS/ BRGM, 1A rue de la Férollerie, 45071 Orléans, France 2 Institute of Experimental Mineralogy, Russian Academy of Science. 142432 Chernogolovka, Russia. 3 IRSN, DEI/SARG/LETS, BP 17, 92262 Fontenay-aux-Roses cedex, France. *Corresponding author: [email protected]

Introduction Water and aqueous solution confined in restricted volumes (pore, channels, intra-solid cavities, etc) are largely encountered everywhere in nature, and their thermochemical properties define the possible driving forces of their interactions with any phase(s) of interest locally present. At given (T,Ptotal) pairs, liquids are considered to be bulk materials, except when the confinement reaches the nanometric scale. Despite this very usual choice, a growing number of evidences points to the existence of an intermediate-sized domain (1 μm - 50 nm) in which the interaction behaviors seems to change, most probably due to thermodynamic driving forces. This contribution reports on dedicated recent experiments, the measurements of which can be interpreted assuming modified reactive properties in the system. Capillarity Capillary water arises in the porous networks that are not liquid-saturated and displays liquid-air interface concave toward air. Practically speaking, it can be encountered in soils (featuring most of the immobile water) and trees (sap ascent), in aquifers depleted of their original fluids, in CO2 storage reservoirs, in the aerated zone around tunnels, etc. The thermodynamic properties of liquid can be easily corrected through a V.dP term, since capillary water is characterized by a decreasing internal pressure, and many theoretical predictions can be done accordingly (e.g. Mercury and Tardy, 2001; Mercury et al., 2003). The pore size domain of influence of such effect, as predicted by Laplace law, is around 8-10 nm, that restricts its applicability in natural settings to the lowest existing pores, those of clayey materials for instance. However, we performed a series of experiment which shows that the capillary effects may act at a larger scale (Bouzid et al., 2011a and b).

73


Precipitation in capillary solution The first experiment (Bouzid et al., 2011a) was done by drying a filtration membrane, after its imbibition by a quasi-saturated NaCl solution. After sufficient evaporation, first we observed precipitation of a quite perfect cubic halite, completely filling the round pore (Fig. 1). Later on, while evaporation promoted capillary water contacting the cubic halite, we observed the precipitation of a secondary curved halite. Our thermodynamic interpretation (phase diagram, Fig. 1b), is that the capillary liquid properties modifies the equilibrium constant and then induces a driving force toward the precipitation of a curved “capillary” solid, because at pressure equilibrium (“isobaric” sketch) with capillary liquid. It means that we were able to promote capillary precipitation, and the “concave” curvature of this secondary halite supports this proposition.

Figure 1: Left, normal and capillary halite precipitated in a pore initially saturated with aqueous solution and submitted to a drying sequence. Right, corresponding phase diagram showing the change in the equilibrium constant of the reaction in anisobaric (normal solid + capillary liquid) and isobaric (capillary liquid and capillary solid) situations. It is interesting to outline the dimensions of the original pore (10 μm) and the size of the capillary bridges (600-800 nm). Indeed, it indicates that the mechanism, though capillary, does not require to be nanoscopic to become significant. Whenever a drying system has got corners, wedges, etc, to locate capillary water, the precipitation of solids with new shape appears possible. Another point of interest, though more or less outside the geochemical approach here developed, is the geomechanical consequence of such precipitate on the container itself (here a polycarbonate polymer). It deforms according to the capillary traction exerted at the vertices of the cubic halite where the capillary bridges (initially liquid, then solid) take place. Due to the precipitation of the capillary solid, the stress fossilizes. Dissolution in capillary solution The second experiment (Bouzid et al., 2011b) was carried out in a capillary tube, initially filled with a NaCl quasi-saturated brine. The first event was a precipitation of massive halite 74


at the liquid-air interface, isolating a liquid column in-between from outside (and so called “NaCl cork”) (Fig. 2 left).

Figure 2: Left, capillary tube ( = 80 μm) filled with NaCl brine and closed by two NaCl massive precipitates, with (upper, t0) and without (lower, t0 + 56 days) in-between dendritic solid. Right, decreasing size of the massive solids, showing the dissolution effect with time. Initially, this liquid column was assumed to be bulk liquid at equilibrium with the NaCl corks. But, with time, we observed a continuous decreasing size of the corks, firstly rapid, then more progressive (Figure. 2 right), highlighting that the solid was initially under-saturated with respect to halite. Additionally, after three months, occluded liquid cavitated demonstrating that it was not a bulk liquid but a superheated one, most probably due to the persistence of a capillary liquid ring all around the corks (see discussion in Bouzid et al., 2011b). Using previous phase diagram (Fig. 1 right), and remembering that a capillary and a superheated liquids have exactly the same properties, the interpretation is straightforward: dissolution is driven by the increasing solubility of bulk halite in superheated/capillary liquid due to an increasing equilibrium constant of this anisobaric system And that change is achieved in a 80 μm wide channel. Solid curvature Pore networks can be developed by solid grains stacking, each individual grain having its own shape and size. As a consequence, a pore is a cavity whose solid walls have a certain curvature at the pore scale. We built an experiment to measure the solubility of such solid as a function of its curvature (Mercury et al., submitted), owing to a pressure membrane extractor enabling to sequentially extract solutions from thinner and thinner pores. Measurements were performed with a silica paste whose pore network was first imaged by cryo-SEM (Fig. 3 left). The solid matrix was concavely curved toward the pore-filling solution, and the measured silica content was decreasing with the decreasing pore size (Fig. 3 right). Thermodynamic calculations based on the same isobaric-anisobaric concepts fitted satisfactorily the results (Figure. 3 right).

75


Figure 3: Silica paste pore network (left) in which (right) silica concentration (normalized to the initial concentration, symbols) is measured as a function of the extraction pressure, here converted into the corresponding pores sizes, and thermodynamic recalculation (lines). In the present discussion, an interesting point is the 150 nm (1/R 6 μm-1) diameter of the pores wherein the pore-size-controlled effect becomes visible. Despite small, such pores are outside the nanoscopic world, and so the pore-size-dependent solubility may have a real impact in natural settings. Interfacial water The water-rock interactions take place at the very contact between the two phases, and again, it is usually assumed that the interfacial domain is a matter of molecular layers, in the 1-5 nm range. Obviously, there are numerous reasons, both theoretical and experiment, to restrict the interfaces to such limited distance. Yet, we performed recently an infra-red experiment which opens some questions about this consensus. Micro-IR maps using synchrotron radiation source (at Soleil synchrotron, SMIS beamline) were performed on water occluded in a quartz fragment (fluid inclusion, Fig. 4) (Bergonzi et al., 2013). Measuring the dangling OH band is usual at the 1-5 nm scale with adapted IR spectroscopy. Here above, classic IR micro-spectroscopy captured this band, this time at the 0.5 μm (at least) scale. To illustrate the consequence of such findings, we converted the IR bands into thermodynamic values (Fig. 5), using a partition function recently developed (Bergonzi et al., submitted). Interfacial water appears to be 1 kJ/mol less reactive than the bulk, over a thickness (at least 500 nm) that makes this shift highly relevant for water-solid interactions.

76


Figure 4: Two IR maps (right) of the sampled inclusion (left). The colour relates to the absorbance scale, taken (middle) at the main peak of the OH-stretching band: 3400 cm-1, featuring the bulk liquid (e.g. Maréchal, 2007), and (right) at the dangling-OH band: 3680 cm-1, characteristic of hydrophobic interfaces (e.g. Miranda and Shen, 1999).

Figure 5: Gibbs free energy as a function of the location in the map. The dotted line displays the bulk value (Dorsey, 1968). Conclusions Physico-chemical measurements on geological-type samples tend to indicate that liquid and solids at the pore scale may behave in different ways than expected from bulk behaviors. The lengths characteristic of such (mesoscopic ?) behaviors are (at least) in the 30-500 nm range, and then cannot be linked to nanoscopic driving forces. We propose that geochemical transformations are sensitive (in relevant conditions) to geometrical confinement at a much higher scale than usually assumed. As a consequence, the reactive term of water-rock interactions in (not so) thin pores, channels, or cavities, probably requires to be re-examined considering this proposition.

77


Acknowledgements This work has received financial support from the French Agency for Research (Agence Nationale de la Recherche), grant CONGE-BLAN10-61001, and from Region centre, grant 201100070577 SIRE. It also benefited from the beamtime allocation n°20130094 from SOLEIL synchrotron. The support of labex Voltaire, grant ANR-10-LABX-100-01 is also acknowledged. References Bergonzi, I., Mercury L., and Jamme, F., 2013. Thermodynamic properties of interfacial water from its infrared signature. Proc. of the “16th Int. Conference on the Properties of Water and Steam”, 1-5 Septembre 2013, Londres (UK), 000-000. Bouzid, M., Mercury, L., Lassin A., Matray, J.-M., and Azaroual, M., 2011a. In-pore tensile stress by drying-induced capillary bridges inside porous materials. J. Colloid Interf. Sci., 355: 498-506. Bouzid, M., Mercury, L., Lassin, A., et Matray, J.-M., 2011b. Salt precipitation and trapped liquid cavitation in micrometric capillary tubes. J. Colloid Interf. Sci., 360: 768-776. Maréchal, Y., 2007. The hydrogen bond and the water molecule, Elsevier, 332 p. Dorsey, N.E., 1968. Properties of ordinary water-substance, in all its phases: water-vapor, water and all the ices. American Chemical Society Monograph Series. New York; Hafner, 673 p. Mercury, L., and Tardy, Y., 2001. Negative pressure of stretched liquid water. Geochim.Cosmochim. Acta, 65: 3391-3408. Mercury, L., Azaroual, M., Zeyen, H., and Tardy, Y., 2003. Thermodynamic properties of solutions in metastable systems under negative or positive pressures. Geochim. Cosmochim. Acta, 67: 1769–1785. Mercury, L., Bouzid, M., Matray, J.-M., Défarge, C., submitted. Pore-size-controlled solubility in aquifers: role of the curvature of the host pores. Earth Planet. Sci. Lett. Miranda, P.B., and Shen, Y.R., 1999. Liquid interfaces: a Study by Sum-Frequency vibrational spectroscopy. J. Phys. Chem. B, 103: 3292-3307.

78


REACTIVE TRANSPORT MODELLING APPLICATIONS AND DEVELOPMENTS FOR PROJECTS IN THE OIL AND GAS INDUSTRY Tim J. Tambach*, Henning Peters, Chris Pentland and Jeroen R. Snippe Shell Global Solutions International BV, Kessler Park 1, 2288 GS Rijswijk, The Netherlands *corresponding author: [email protected]

Introduction Reactive transport modelling (RTM) is becoming more recognized as a tool in the oil and gas industry. This is mainly due to increasing importance of subsurface projects involving production and injection of more reactive fluids and gases, for example storage of CO 2 and H2S, and enhanced oil recovery (EOR). Also water flooding assessments sometimes require RTM. Several assumptions that were appropriate for modelling of conventional oil and gas reservoirs are not always valid for such more complex projects. Our in-house Shell reservoir simulator MoReS was coupled with PHREEQC (Parkhurst and Appelo, 2013), which was recently extended (PHREEQC3) to include pressure dependency of geochemical reactions. This coupling improves forecast modelling, development and deployment of RTM in new areas of interest at elevated pressure (P) and temperature (T) conditions encountered in subsurface reservoirs (Wei, 2012). The software was successfully benchmarked with TOUGHREACT-Pitzer (Zhang et al., 2006), which is important for confidence in simulation results (Pentland et al., 2013). Although modelling tools are in place, there is a need for improved understanding of fluid-rock interactions at high T, P, and salinity. Experimental data for mineral interaction with CO 2, H2S and impurities are scarce at such elevated conditions. However, equilibrium constants (K) for mineral and speciation reactions are key parameters to model mineral dissolution and precipitation. It is also known that several current geochemical databases lead to quantitatively different results even when using the same simulator (Dethlefsen et al., 2011). Geochemical data is not only used by reservoir engineers in subsurface applications, but also links to ongoing work done by engineers from wells and surface (i.e. production chemists, production technologists, pipeline engineers, and process engineers). Part of the ongoing RTM development is to provide one consistent geochemical database to end-users from several disciplines working in our integrated projects.

79


Material & Methods Coupling with PHREEQC Partitioning of components between the water and hydrocarbon phases and chemical interactions between fluids and reservoir rock were included by coupling PHREEQC to our inhouse reservoir simulator MoReS. Traditionally, RTM can be carried out in two ways: (1) sequential calls to the flow and geochemical solver (i.e. explicit coupling); (2) incorporation of the geochemical equations into the flow solver (i.e. fully implicit coupling). However these approaches can have drawbacks in terms of accuracy/consistency and computational speed, respectively. In our simulator we developed a hybrid approach between (1) and (2) to overcome these issues. Software benchmark We used our in-house RTM simulator to model a case study of CO2 injection for 25 years into a deep saline aquifer (DSA) and compared the results with those from TOUGHREACT-Pitzer using ECO2N (Pruess, 2005). TOUGHREACT-Pitzer is a dedicated version of TOUGHREACT (Xu et al., 2006) for problems involved in aqueous solutions of high-ionic strength. Geochemical reactions were carried out at 1 bar, as TOUGHREACT does not take into account pressure dependency of mineral reactions. The aquifer is described using a 2D radial model with grid refinement close to the well, a radius of 8 km, and 43 layers with unique porosity and horizontal permeability. The Pitzer ion-interaction ionic activity model (Zhang and Spycher, 2006) is used for accurately taking into account hyper saline conditions. Geochemical database For construction of the database we used the Helgeson-Kirkham-Flowers (HKF) equation of state, which is the underlying model for SUPCRT92 (Johnson et al., 1992) and subsequent other geochemical databases (Aradóttir et al., 2012). Equations were built into MATLAB and equilibrium constants (K) of relevant reactions were computed as a function of T. The data were then fitted according to the following polynomial model: ,

(1)

where parameters A1 to A6 are fit constants. This model is used by PHREEQC and describes the temperature dependence of the log K. Pressure dependence of the log K is taken into account by PHREEQC3 using an equation for the specific volume (Vm,inf) of aqueous species at infinite dilution (Johnson et al., 1992): (2)

In this equation, a1 to a4 are species dependent thermodynamic pressure constants, Pb is the pressure in bar, TK is the temperature in Kelvin, W is the species dependent Born coefficient, and Q is the Born function describing the pressure dependence of the dielectric 80


constant of water. PHREEQC simulations using the database were carried out at infinite dilution to compute speciation. We compared the log K, as well as speciation results, with results from the commercial software OLI Stream Analyzer, which is frequently used for our well/surface applications. Results & Discussion Coupling with PHREEQC Our new approach ensures an implicit scheme of gas solubility between flow and geochemical solvers, while other (mostly ionic) aqueous species are transported explicitly because they remain in the brine (no dissolution in the gas or oil phase). In this way the number of components that need to be solved fully implicitly during one time step is limited and computationally efficient.

Figure 1: Gas saturation (top) and porosity (bottom) with distance from the wellbore (located on the left side) after 1,000 years using MoReS-PHREEQC (left) and TOUGHREACT (right). Software benchmark Figure 1 shows some results of CO2 injection into a DSA computed with both our in-house RTM simulator and TOUGHREACT-Pitzer. Several reservoir engineering and geochemical results were obtained over time, but here we focus on the CO2 plume and the porosity after 1,000 years of simulation time. The computed extent and shape of the CO2 plume is comparable for both simulators, although minor differences occur. The computed change in 81


porosity is sensitive to mineral reactions and depends on mineral densities. Several dissolution and precipitation reactions were modelled, with an overall change in porosity that is quite similar for both simulators. The small remaining differences can be understood from remaining small differences in model input. Geochemical database Figure 2 shows the log K and molality of HCO3-, as computed with the HKF model and applied to PHREEQC. We also computed other species, but in view of space limitations these will not be discussed here. The results show that the computed log K of HCO3- is in agreement with the log K reported by OLI Stream Analyzer, which also holds for computation of its speciation at infinite dilution. Similar results were obtained for other species, although small deviations may occur as a result of exact thermodynamic data used. Our workflow is set up in such a way that the remaining difference can be resolved altogether, by choosing a single thermodynamic data repository. The results demonstrate the potential of our workflow for generating one consistent database.

Figure 2: Computed and fitted log K (left) as a function of T at P=1 bar (left) for HCO3-. Computed molality using PHREEQC3 and OLI as a function of T at P=300 bar (right). Conclusions Our in-house reservoir simulator was successfully coupled with PHREEQC and benchmarked against TOUGHREACT-Pitzer for a case study of CO2 injection into a DSA. A geochemical database was constructed, based on the HKF model, and tested using PHREEQC. Comparison of log K and speciation results with results from OLI Stream Analyzer show a good match and demonstrate the potential of using such a database in integrated projects. RTM applications and developments will be further deployed in future projects such as CO2/H2S storage and enhanced oil recovery. Acknowledgements We thank Lingli Wei, Guoxiang Zhang, and Hongmei Huang for valuable contributions. 82


References Aradóttir, E.S.P. , Sonnenthal, E.L., Jónsson, H., 2012. Development and evaluation of a thermodynamic dataset for phases of interest in CO2 mineral sequestration in basaltic rocks. Chem. Geol. 304-305: 26-38 Dethlefsen, F., Haase, C., Ebert, M., Dahmke, A., 2011. Uncertainties of geochemical modeling during CO2 sequestration applying batch equilibrium calculations. Environmental Earth Sciences: 1-13. Johnson, J.W., Oelkers, E.H., and Helgeson, H.C., 1992. SUPCRT92—A software package for calculating the standard molal thermodynamic properties of minerals, gases, aqueous species, and reactions from 1 to 5000 bar and 0 to 1000 °C: Comp. Geosci., v. 18(7): 899–947. Parkhurst, D.L., and Appelo, C.A.J., 2013. Description of input and examples for PHREEQC version 3—A computer program for speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations: U.S. Geological Survey Techniques and Methods, book 6, chap. A43: 497 p., available only at http://pubs.usgs.gov/tm/06/a43. Pentland, C.H., Zhang, G., Huang, H., Tambach, T.J., and J.R. Snippe, 2013. A Comparative Study of Reactive Transport Modelling - Using TOUGHREACT and MoReS for Modelling CO2 Sequestration. Sustainable Earth Science (SES) conference, Pau, France. Pruess K. 2005. ECO2N: A TOUGH2 Fluid Property Module for Mixtures of Water, NaCl, and CO2. Wei, L., 2012. Sequential coupling of geochemical reactions with reservoir simulations for waterflood and EOR Studies. SPE Journal 17 (2): 469-484. Xu, T., E. Sonnenthal, Spycher, N., and Pruess, K., 2006. TOUGHREACT—a simulation program for non-isothermal multiphase reactive geochemical transport in variably saturated geologic media: applications for geothermal injectivity and CO2 geologic sequestration. Comp. Geosci. 32(2): 145-165. Zhang, G. and Spycher, N., 2006. Reactive Geochemical Transport Modeling of Concentrated Aqueous Solutions: Supplement to TOUGHREACT User’s Guide for the Pitzer IonInteraction Model. Earth Sciences Division, Lawrence Berkeley National Laboratory University of California, Berkeley, CA 94720.

83


84


A SEQUENTIAL T-H-M-C ALGORITHM TO SIMULATE PHYSICAL LINKS BETWEEN STRESS AND GEOCHEMICAL ENVIRONMENT EVOLUTION Alain Dimier1*, Jérôme Gaombalet and Ahmed Bougacha 1

Eifer: European Institute for Energy Research, Karlsruhe Institute of Technology, Karlsruhe D-76131, Germany * Corresponding author: [email protected]

Introduction Considering actual geological application trends, like deep geothermal energy or shell gases, the development of a transient coupling algorithm between geochemistry and geomechanics appears to be relevant. The present study can be considered as the validation of the coupling algorithm, the next step will be the introduction of plastic deformations to consider the evolution over time of caprock materials. Abstract We will shortly present here the methodology to gather within one environment two open source software, PhreeqC (Parkust and Appelo, 1999) and Elmer, which have disjunctive application fields; the goal being to broaden the physics to be considered and be able, more specifically, to model the impact of the chemistry variations to the mechanic of soils. Those two software's are gathered in one environment to enable multi-component, reactive solute transport studies in three-dimensional saturated/ unsaturated ground-water flow systems; becoming so called Python modules (van Rosum, 2008). Based on Open Source, the resulting tool enables the study of natural and contaminated ground-water flow systems at a variety of scales ranging from laboratory experiments to local and regional field scales, that for the aforementioned physics. More specifically, here mechanics is coupled to Chemistry through porosity variations, the coupling being based on a transient sequential algorithm. We will discuss the way the stress field can be efficiently evaluated once the variations of the porosity field are retrieved from the geochemical solver. A direct comparison with analytical solutions issued from the literature (Jianhong et al. 2009) will be commented as an illustration of the phenomenology that can be presently modeled and as algorithm performances assessment. An algorithm implementing retroaction of mechanical stresses on chemistry will be discussed.

85


Thereafter examples related to the topic of CO2 storage safety will be presented giving some initial hints on the temporal evolution of the caprock, the goal of such studies being to give elements on caprock confinement properties in the context of CO2 storage. Acknowledgements That work has been partially financed by E.D.F. Contributions of J. Gaombalet and A. Bougacha have been made on an open source basis. References Parkhurst, D. L., Appelo, C. A. J., 1999. User’s guide to PHREEQC (Version 2) – A computer program for speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations. U.S. Geological Survey Water-Resources Investigation Report, 99-4259: 312 p. http://www.csc.fi/english Guido van Rossum 2008. Extending and Embedding the Python Interpreter, Release 2.5.2. Jianhong, Y. Wu F. Q., Sun J. Z. 2009. Estimation of the tensile elastic modulus using Brazilian disc by applying diametrically opposed concentrated loads, Int. J. Rock Mech. Min Sci., 46(3):568–76

86


EXPERIMENTAL BENCHMARKS FOR THE VERIFICATION AND VALIDATION OF REACTIVE TRANSPORT CODES Jenna Poonoosamy*, Georg Kosakowski and Luc Van Loon Laboratory for Waste Management, Department of Nuclear Energy and Safety, Paul Scherrer Institut, Villigen, Switzerland *Corresponding author:[email protected]

The evolution of porosity due to mineral alteration processes and the associated change of transport parameters are of major concern for natural geological environments or engineered underground disposal systems. One example is the long term storage of nuclear waste in a deep geological repository where concrete materials will be used for support structures, for cavern backfill or for waste conditioning. The diffusion of the cement pore water into the surrounding clay host rock might lead to the precipitation and dissolution of mineral phases near the clay-cement interface and might clog the porosity (Dauzeres et al., (2010), Gaborau et al., (2012)). Porosity clogging will stop diffusive and advective transport across the interface. It is thus important to understand how clogging is triggered and how the clogging process is influenced by transport, and vice versa. Several modeling studies have been carried out to predict the effect of the alkaline plume on clay barriers (Cochepin et al., (2008), De Windt et al., (2007), De Windt et al., (2004), Trotignon et al., (2005)). However in numerical models, which calculate porosity changes due to dissolution/precipitation of minerals, the temporal evolution of porosity is highly affected by assumed reaction kinetics laws and applied numerical mesh refinement (Marty et al., (2009)). In addition, such numerical models use empirical relations for linking porosity with transport properties (e.g. diffusivity and permeability). The numerical models need to be tested against experiments to validate the underlying concepts how the geochemical migration is affected by porosity changes. The aim of this work is to set 1D and 2D reactive transport experiments in simple media, where clogging occurs and its evolution with time can be easily assessed. These experiments are modeled with the reactive transport code OpenGeoSys-GEM (OGS-GEM) (Shao et al., (2009); Kosakowski and Watanabe, (2013)). As a first step, a simple experimental 2D benchmark of dimensions 0.1 m by 0.1 m involving a granular porous medium was developed. The flow field was characterized by conservative dye tracer tests. The experimental results were compared to a numerical model and good match between experiment and model was achieved. In a next step, experimental setup was modified in order to induce precipitation. A 0.01m thick layer of reactive celestite (SrSO4) was inserted between two layers of the granular porous medium. Barium chloride was injected into the tank. Barium chloride accelerates the dissolution of celestite, which is replaced by barite (BaSO4). Due to a higher molar volume of barite the porosity decreases near the reactive layer. During the course of the experiment, 87


changes in the flow field, suggesting localized permeability (and porosity) changes, were observed. Currently we are investigating the structure of the precipitations more closely. This is necessary in order to parametrize transport laws and kinetic rates, which is an important step to describe the reactive experimental system with a numerical model. References Cochepin, B., Bildstein, O., Steefel, C., Lagneau, V., & Van der Lee, J., 2008. Approaches to modelling coupled flow and reaction in a 2D cementation experiment. Advances in Water Resources, 31, 1540-1551. Dauzeres, A., Le Bescop, P., Sardini, P., & Cau Dit Coumes, C., 2010. Physico-chemical investigation of clayey/cement-based materials interaction in the context of geological waste disposal: Experimental approach and results. Cement & Concrete Research, 40, 1327-1340. De Windt, L., Badredinne, R., & Lagneau, V., 2007. Long-term reactive transport modelling of stabilized/solidified waste: from dynamic leaching tests to disposal scenarios. Journal of Hazardous Materials, 207, 529-536. De Windt, L., Pellegrini, D., & Van der Lee, J., 2004. Coupled modeling of cement/claystone interactions and radionuclide migration. Journal of Contaminant Hydrology, 68, 165-182. Gaborau, S., Lerouge, C., Dewonck, S., Linard, Y., Bourbon, X., Fialips, C., et al., 2012. In-Situ Interaction of cement paste and shocrete with claystones in deep disposal context. American Journal of Science, 312, 314-356. Kosakowski, G., & Wanatabe, N., 2013. OpenGeoSys-Gem: A numerical tool for calculating geochemical and porosity changes in saturated and partially saturated media. Physics and Chemistry of the Earth. Marty, N. C., Tournassat, C., Burmol, A., Giffaut, E., & Gaucher, E., 2009. Influence of reaction kinetics and mesh refinement on the numerical modelling of concrete/clay interactions. Journal of Hydrolology, 364, 58-72. Shao, H., Dmytrieva, S., Kolditz, O., Kulik, D., Pfingsten, D., & Kosakowski, G., 2009. Modeling reactive transport in non-ideal aqueous–solid solution system. Applied Geochemistry, 24(7), 1287–1300. Trotignon, L., Didot, A., Bildstein, O., Lagneau, V., & Margerit, Y., 2005. Design of a 2D cementation experiment in porous medium using numerical simulation. Oil & Gas Science and Technology - Rev. IFP, 60, 307-318.

88


Posters

89


90


THE EFFECTS OF POROSITY CLOGGING ON THE TRANSPORT PROPERTIES OF POROUS MATERIALS UNDER GEOCHEMICAL PERTURBATIONS: EXPERIMENTAL APPROACH AND MODELING A. Chagneau1,2,3, F. Claret3, B. Madé4, M. Wolf5, F. Enzmann5 and T. Schäfer1,2 1

Institute for Nuclear Waste Disposal (INE), Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany 2 Freie Universität Berlin, Berlin, Germany. 3 BRGM, 3 avenue Claude Guillemin, BP 36009, Orléans Cedex 2, France. 4 ANDRA, 7 rue Jean Monnet, 92298 Chatenay-Malabry Cedex. 5 Johannes Gutenberg-Universität Mainz, Mainz, Germany The focus of the present study is to characterize the effects of porosity clogging on the effective diffusivity of porous materials under geochemical perturbation. A systematic experimental approach was used, and coupled to reactive transport modelling. The experiments were performed in transparent Plexiglas diffusion cells, composed of a 50 mm long column filled with purified sea sand and two reservoirs of 20 mL capacitance. Prior to the mineral precipitation experiments, the transport properties were characterized by means of conservative radiotracer (HTO) tests. The reservoirs and pore space were filled with a NaCl background electrolyte, and one reservoir was spiked with the tracer. From these experiments, the effective diffusivity (De) of the porous material was found equal to (4.34 ± 0.15)×10-10 m² s-1, and the porosity (ε) equal to 0.417 ± 0.017. A mercury intrusion porosimetry measurement performed on 2 samples confirmed these values, giving porosities of 0.404 and 0.427. The precipitating phase chosen for the porosity clogging was celestite (SrSO 4), for its wellknown simple and fast precipitation mechanisms. For these experiments, one reservoir was filled with a solution bearing the cation (SrCl2), and the other by a solution bearing the anion (Na2SO4). The cation reservoir was spiked with the radiotracer (HTO). The precipitation of the salt occurred very fast and close to the middle of the cell. The precipitate took the shape of a very thin disk (less than 1 mm in thickness), perpendicular to the direction of diffusion and visible to the naked eye after about 10 days. After this delay, the precipitation front did not appear to grow any further. The diffusion of HTO continued at a very slow and constant rate after precipitation for several weeks, indicating that the pore space was not fully clogged. Similar experiments were performed with a compacted clay (Na-exchanged Illite du Puy), following the experimental protocol of VanLoon et al. (2003), with a cocktail of radiotracers (HTO, 36Cl and 85Sr). The tracers diffusion observed was about 2 (for HTO) to 8 (for 36Cl) times lower than in absence of precipitation, clearly indicating a porosity reduction. The simple through diffusion experiments (no precipitation) in compacted sea sand were well reproduced by the geochemical modeling code CRUNCHFLOW. However, the model could not reproduce the HTO diffusion curves obtained for precipitation experiments. The 91


challenge of the present work now lies in the development of the reactive transport model, and more precisely in the law describing the relationship between the D e and the porosity. In the CRUNCHFLOW code, the Archie’s law ( , where D0 is the diffusivity of the species in pure water, and m is a fitting parameter known as the cementation factor) is used, which is known to be unsuited for low porosities (e.g. Navarre-Sitchler et al., 2009). References Van Loon, L.R., Soler, J.M., Bradbury, M.H., 2003 Diffusion of HTO, 36Cl− and 125I− in Opalinus Clay samples from Mont Terri: Effect of confining pressure, Journal of Contaminant Hydrology, Volume 61, Issues 1–4, Pages 73-83. Navarre-Sitchler, A., Steefel, C. I., Yang, L., Tomutsa, L., Brantley, S. L., 2009. Evolution of porosity and diffusivity associated with chemical weathering of a basalt clast. Journal of Geophysical Research: Earth Surface 114, 2156-2202.

92


93


94


MODELING LONG-TERM LEACHING EXPERIMENTS OF FULL SCALE CEMENTED WASTES: EFFECT OF SOLUTION COMPOSITION ON DIFFUSION C. Borkel, V. Montoya and B. Kienzler Institute for Nuclear Waste Disposal (INE), Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany *Corresponding author: [email protected]

Cementitious materials are commonly used to solidify/stabilize low to intermediate level liquid and solid nuclear waste prior to the disposal in a geological repository. The degradation of these cement based materials in contact with solutions depends on many factors, such as the cement composition, liquid/solid ratio, time, temperature, pH, solution composition and renewal rate. The present work focuses on the modeling of leaching experiments performed with cemented waste products enclosed with tap water for more than 30 years in a specific drift of the Asse II salt mine. Simulated cemented waste has been produced using Ordinary Portland Cement (CEM I 42.5R) with a water/cement ratio of 0.235 L kg-1 and a mixture of chemicals mimicking the products from reprocessing of spent fuel by the Purex process (mainly NaNO 3, but also Na-citrate, Na-tartrate, Na2HPO4·12H2O and Na-oxalate). The cemented waste simulate was additionally doped with Cs and has a total mass of 192 kg. After complete hydration, the waste was immersed in 335 L canisters filled with 239 L of tap water. The initial composition of the leachant was (in mol L-1) K: 0.013; Na: 0.013; Ca: 0.002; Mg: 0.001; Cl: 0.025 and SO4: 0.001 and pH 7 (Kienzler et al. 2002). Solution composition and pH were monitored for more than 30 years.

Figure. 1: Conceptual model used for the calculations. A homogeneous monolith of solidified cemented waste simulate is introduced in a container with the leaching solution (pH = 7). The modeling code Phreeqc (Parkhurst and Appelo, 2013) is used to simulate the leaching tests taking into account diffusive transport of solutes in 2 D and the focus is diffusive release of the cation Cs. The thermodynamic database used in calculations is ThermoChimie 95


(Duro et al. 2012), and the ionic strength corrections are made using the SIT approach to adequately take into account ionic strength around 0.5 M. In Figure. 1 the conceptual model used in the calculations is represented schematically. The concentration gradients between the pore solution and the leaching solution cause diffusion of chemical species basically from the pore solution to the surrounding, dilute water. In the modeling the effect of the charge balance constraint is assessed and results are compared to the analytical solution. XRD analysis have been done to determine the solid phases present in cement and are used to help outlining strengths and weaknesses of the different models. References Duro L., Grivé M. and Giffaut E., 2012. ThermoChimie, the ANDRA Thermodynamic Database, MRS Proceedings, 1475 Kienzler B., Schlieker M., Schüßler W., Metz V., Hentschel D., Nies C., Kerner N., Seither A., Meyer H., Bracke G. 2002 Langzeit Auslaug- und Korrosionsexperimente an zementierten 1:1 Gebinden in der Schachtanlage Asse Probennahme und Auswertung Wissenschaftliche Berichte, FZKA-6716 Parkhurst D.L. and Appelo C.A.J., 2013. Description of Input and Examples for PHREEQC Version 3—A Computer Program for Speciation, Batch-Reaction, One-Dimensional Transport, and Inverse Geochemical Calculations. U.S. Geological Survey Techniques and Methods, book 6, chap. A43

96


97


98


DEPOSITION OF COLLOIDS ON GRANITE SURFACE: INFLUENCE OF MACROSCOPIC SURFACE HETEROGENEITIES AND Eu(III) CONCENTRATION Gopala Krishna Darbha1*, Johannes Lützenkirchen1, Cornelius Fischer2 and Thorsten Schäfer1 1

Institute for Nuclear Waste Disposal (INE), Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany 2 MARUM / FB Geowissenschaften, Universität Bremen, D-28359 Bremen, Germany *Corresponding author: [email protected] Interaction of colloids with surfaces is of primary importance to answer many key questions to industry and environmental processes. Deposition experiments were conducted with artificially patterned(Darbha et al. 2012) and natural surfaces(Darbha et al. 2010; Darbha, Fischer et al. 2012) to predict the impact of surface topography (roughness) and its significance in estimation of colloid interaction forces. Overall, an increased colloid retention with an increase in the roughness of the substrate was observed. The negative asperities (pits) are preferred sites for colloid deposition compared to flat surfaces. For granodiorite, roughness played an important role when [Eu]