MODELING OF THE LABORATORY TESTS OF INTERACTION OF THE NaNO3-NaOH FLUIDS WITH SANDSTONE ROCKS FROM DEEP RADIONUCLIDE REPOSITORY SITE, USING TOUGHREACT A.V. Kiryukhin 1, E.P. Kaymin2, E.V. Zakharova2, А.А. Zubkov3 1Institute of Volcanology and Seismology FEB RAS, Piip-9, Petropavlovsk-Kamchatsky, Russia, 683006 2- Institute of Physical Chemistry and Electrochemistry RAS, Leninsky-31, Moscow, Russia, 119991 3- Siberia Chemical Plant, Kurchatova-1, Seversk, Russia ABSTRACT TOUGHREACT modeling was used to reproduce laboratory tests with sandstone samples collected from deep radionuclide repository site in Siberia Chemical Plant. Laboratory test includes injection of alkaline fluids into sandstone samples at 70оС. Modeling results were compared with observed test data (mineral phase change, transient concentration data at the outlet of sample column). Some minerals were restrain in the model to precipitate or dissolve according to laboratory test results. Model and test convergence in mineral phases (Na-smectite and kaolinite precipitation in the model, quartz, microcline, chlorite and biotite dissolution in the model) were obtained. Nevertheless it was not found possible to generate sodium carbonates in the model (while sodium clearly observed in the test). Transient chemical concentration data at the outlet of sample column match Na only. It’s concluded that the model should be improved in the future work to better match observed data. Chemical Input Data Initial mineral fractions are shown in Table 1, parameters of kinetic water-rock chemical interaction assigned the same as in (Kiryukhin et al, 2004) paper (E aenergy of activation, kJ/kmole), while other parameters were corrected during model calibration (Table 2). Chemical compositions of the initial solution (natural pore fluids) and injected fluid are shown in Tables 3 and 4, correspondingly. MODEL CALIBRATION Calibration Data Two tests with duration of 79 days and 42 days correspondingly were performed with injection mass flux at average level 2.50 10-5 kg/s m2. During test (at times 9, 16, 23, 30, 32, 58, 79 days) sampling of fluid took place at column outlet for chemical analysis (Na, Al, Si, Ca, Mg, K, Sr). Microprobe analysis of mineral composition of samples after testing was done. Microprobe analysis performed based on Link INCA ENERGY200 to electronic scan facility CamScan MV-2300. A.A. Grafchikov (Institute of Experimental Mineralogy RAS) took participation in this analysis. c d a INTRODUCTION TOUGHREACT is a computer code capable to simulate thermal-hydrodynamicchemical (THC) processes including multiphase nonisothermal transport and kinetics of the rock - fluid chemical interaction. The THC processes and secondary minerals associations observed in some drilled geothermal fields of the recent volcanic activity areas have been successfully reproduced by TOUGHREACT simulations (Xu, T. and Pruess, K., 2001, A.V. Kiryukhin et al, 2004). Similar processes took place during waste radionuclide fluids injection in sandstones aquifers. When liquid radionuclide waste injected in layer type reservoirs (Siberia Chemical Plant (SCP), (Seversk), Mining-Chemical Plant (MCP) (Zheleznogorsk) – chemical interaction with natural pore fluids and clay minerals of the deep repository site took place. New secondary minerals created, while primary minerals dissolve, temperature increase due to radiogenic heat release (A.I. Rybalchenko et al, 1994). Monitoring of the hydrogeological parameters in the wells, as well as laboratory experiments at P-T conditions corresponding to physical and chemical processes in repository sites conducted to get reliable information on processes there (А.А. Zubkov et al, 2002, E.P. Kaymin et al, 2004). Reliable numerical model needed to forecast process of migration of radionuclides and to guaranty safety condition in repository sites too. In this study, TOUGHREACT was used to reproduce laboratory experiment of the process of technogenic alteration observed in sandstones (obtained from deep repository site) as a result of chemical interaction during NaNO3-NaOH fluid injection in rock samples at temperature 70оС. Modeling results calibrated against observed secondary minerals, generated during laboratory experiment and identified based on microprobe analysis, and against transient chemistry data of fluids, discharged from the core outlet during laboratory experiment. MODEL SETUP In this study, TOUGHREACT was used to reproduce laboratory experiment of the process of technogenic alteration observed in sandstones. In the model , adjective and diffusive transport of aqueous chemical species is considered. Mineral dissolution/precipitation can proceed at equilibrium and /or under kinetic conditions, according to the following rate law: r = kS (1-Q/K) exp(Ea/(R*298.15)-Ea/(RT)), where k – kinetic constant of the chemical dissolution/precipitation at 250C, mole/s·m2; S – specific reactive surface area, m 2/m3; Q – is activity product; K – is equilibrium constant for mineral-water interaction; Ea – is the activation energy, kJ/kmole; R – is the gas constant, kJ/kmole K, and T – is temperature, K. Temperature effects are also considered for geochemical reaction calculations in which equilibrium and kinetic data are functions of temperature. CONCLUSIONS b e f g h Scan electron images of samples (E.P. Kaymin data): a- Chlorite (Chl) replacement by montmorillonite (Mont), b- Muscovite (Ms) replacement by kaolinite (Kaol), c- Biotite (Bt) replacement by montmorillonite (Mont), d- K-feldspar (Kfs) replacement by montmorillonite (Mont) , e- grains of magnetite hosted in clay minerals,f- magnetite (white) inside of montmorillonite (grey) replaced biotite grain, g- sodium or trona (Na) release in form of crust and regions in montmorillonite (Mont),h- sodium or trona (Na) release in form of regions in montmorillonite (Mont). Note: black space is polymeric matrix. Flow and Solution Input Data According to the Laboratory test data 700С isothermal conditions with mass flux 2.50 10-5 kg/s m2 and pressure 3.0MPa were assigned in the model. Sample porosity assigned 0.2. The length of the model correspond to the length of test sample 15 cm. 1-D numerical grid generated includes 32 elements: B 1 – source of injected fluid (element volume 5.00E+20 m3, R 1- R 30 elements represents sandstone column of 15 cm length, each element with width of 0.005 m, and D 1 – inactive element with specified pressure 3.0 MPa, which correspond to discharge from the column outlet. Figure above: Modeling results of mineral fraction change along injection steamline in the sandstone column. Figure left: Modeling (lines) and laboratory test (solid circles) match (transient chemical concentrations of fluids). Modeling Results TOUGHREACT modeling of the laboratory test (run #7) yield the following results: (1) Mineral phase fractions change. By the end of 79 day alkaline solution injection: quartz - dissolve (from 9.2 10-5 to 9.8 10-5), microcline dissolve (from 3.1 10-6 to 3.2 10-6), albite-low - dissolve (to 1.4 10-6), Na-smectite (montmorillonite) precipitate in the middle and outlet part of column (up to 1.8 10-3), kaolinite – precipitate in the middle and outlet part of column (up to 1.1 10-5), chlorite – dissolve (3.0 10-9), muscovite – dissolve everywhere (from 6.3 10-5 to 7.3 10-5). Secondary mineral phases (Na-smectite and kaolinite) were formed during first 9 days only. No sodium carbonate precipitations was obtained in the model (while abundant sodium carbonates observed during laboratory test). (2) Match of observed and modeling transient chemical concentrations of fluid sampled from sandstone column outlet (run #7)shows the following. pH match show the same trend of model and experiment, while absolute modeling values 2.6 units greater. Na concentrations from model match those from the experiment. К match show model 3 times less values than experiment. Ca match model and experiment show the same trend, while absolute values in the model 2-3 orders less than experiment. Mg match show the same trend in model and experiment, while, model absolute values are 1-2 orders less. Al match show model yield 4 order greater compare experiment. Si match show convergence in the first times, while later model concentrations 3 times greater experiment. It was also found that change of rate constants of mineral precipitations (kS) for Na-smectite and kaolinite (Table 3) has no effect on рН and outlet discharge transient chemical concentrations. (1) TOUGHREACT modeling was used to reproduce laboratory tests with sandstones samples collected from deep radionuclide repository site in Siberia Chemical Plant. Laboratory test include injection of alkaline fluids into sandstones samples at 70оС. (2) Model and test convergence in mineral phases (Na-smectite and kaolinite precipitation in the model, quartz, microcline, chlorite and biotite dissolution in the model) were obtained due to restrain for some minerals to precipitate /dissolve. Nevertheless it was not found possible to generate sodium carbonate in the model (while sodium clearly observed in the test). Transient chemical concentrations data at the outlet of sample column match Na only. pH match show the same trend of model and experiment, while absolute modeling values 2.6 units greater. Ca and Mg match model and experiment show the same trend, while absolute values in the model 2-3 orders less than experiment. (3) The main reason model and laboratory test mis-match seems to be TOUGHREACT not taking into account mineral/mineral chemical reactions. In laboratory test was found K release to solutions, and Al consumed by secondary minerals due to biotite, K-feldspars, muscovite replacing by clay minerals, глинистыми минералами. If such reactions will be implemented in TOUGHREACT, then convergence of modeling and laboratory test data may improve. ACKNOWLEDGEMENTS We express our gratitude to T. Xu, N.Spycher (Lawrence Berkeley National Laboratory) for valuable comments and suggestions, B.N. Ryzhenko and О.А. Limantseva (GeoChi RAS), I.B. Slovtsov (IVS FEB RAS) for additional thermodynamic calculations. This work was supported by Siberia Chemical Plant, FEB RAS project 06-I-ОНЗ-109 and RFBR project 06-05-64688-а. REFERENCES: А.А. Zubkov, О.V. Makarova, V.V. Danilov, Е.V. Zakharova, Е.P. Kaymin, К.А. Menyailo, А.I. Rybalchenko, Technogenic geochemical processes during injection of the liquid radionuclide waste into sandstones layer type reservoirs, Geoecology, Engineering Geology, Hydrogeology, Geocriology, 2002, №2, p.133-144. A.V. Kiryukhin, М.Y. Puzankov, I.B. Slovtsov et al, Thermal-Hydrodynamic-Chemical modeling processes of secondary mineral precipitation in production zones of geothermal fields (in print “Vulcanologia and Seismologia”, 32 p.). Rybalchenko A.I., Pimenov M.K., Kostin P.P. et al. Е.P. Kaymin, Е.V. Zakharova, L.I. Konstantinova, А.А. Deep injection of liquid radionuclide waste. Graphchikov, L.Y. Аranovich, V.М. Shmonov, Study of the Мoscow, IzdAT publ., 1994, 256 p. (in Russian). interaction of alkaline radionuclide waste with sanstone rocks, Geoecology, Engineering Geology, Hydrogeology, Xu, T. and Pruess, K., 2001a, On Fluid Flow and Geocriology, 2004, №5, p.427-432. Mineral Alteration in Fractured Caprock of Magmatic Hydrothermal Systems, Journal of A.V. Kiryukhin, T. Xu, K. Pruess, J. Apps, I. Slovtsov, Geophysical Research, 106 (B2), 2121-2138. Thermal-Hydrodynamic-Chemical (THC) Modeling Based on Geothermal Field Data, Geothermics, v.33, No.3, 2004, p. 349-381.