rachid ababou - 12-13-14 june 2006, toulouse,...

IAHR International Association

of Hydraulic Research

PRE-PROCEEDINGS Book of Abstracts

AIRH Association Internationalede Recherche Hydraulique

IAHR-GW2006 International Ground Water Symposium

Conférence Internationale Eaux Souterraines

12-13-14 June 2006, Toulouse, France http://www.iahr-gw2006.cict.fr

GROUNDWATER HYDRAULICS IN COMPLEX ENVIRONMENTS: 1. Heterogeneity and upscaling; 2. Surface-subsurface coupling;

3. Chemically active transport; 4. Hydromechanics and density effects.

Organized and sponsored by IAHR - Groundwater Hydraulics Section Co-sponsored by IAHS, ASCE/EWRI, SHF, INPT, UPS, CNRS, Région Midi-Pyrénées

IAHR-GW2006 PRE-PROCEEDINGSBook of Abstracts

PRELIMINARY PROGRAM OF SESSIONS (VERSION 3) IAHR-GW2006 “GROUNDWATER IN COMPLEX ENVIRONMENTS”

12-13-14 JUNE 2006 TOULOUSE, FRANCE http://www.iahr-gw2006.cict.fr/

1

1

IAHR-GW2006

Conference Topics (Themes) Theme 1. Heterogeneity and Upscaling. Theme 2. Coupling of Surface/Subsurface Flow. Theme 3. Chemically Active Transport Phenomena Theme 4. Coupled hydro-mechanics and density effects.

Remarks about the program of sessions ORAL communications comprise regular talks (20 mn and 5 mn of questions) and keynote

lectures (40 mn and 5 mn discussion). POSTER communications are listed at the end of this program; they will be presented during

coffee breaks every day (12-13-14 June 2006). SESSIONS are organized around the four themes of the conference, with two sessions/themes

in parallel every time. For example: Theme 1 & Theme 2 in parallel sessions in the morning, then Theme 1 & Theme 3 in the afternoon. The Keynote Lectures are to be attended by all participants in the main auditorium.

The names of some authors or co-authors are underlined merely to indicate the main contact of the submitted paper or the expected speaker. This indication is subject to change.

MONDAY - 12 JUNE 2006 MONDAY - 12 JUNE 8h – 8h 30 Welcome & Registration 8h 30 – 9h 15 Keynote Lecture 1 (KL1) Probabilistic representation of heterogeneous geologic media. Représentation des milieux géologiques hétérogènes par des méthodes probabilistes Ghislain de MARSILY (Académie des Sciences & SISYPHE, Univ. Paris VI, Paris, France)

Monday morning sessions : Themes 1 & 2 9h 30 – 9h 55 (S1) Th1 (paper ID:2 & ID:95) Flow and transport modelling and efficiency of remediation in a fractured and karstic aquifer : a case study. Claudia CHERUBINI, Concetta I. GIASI Th2 (Paper ID:3 & 104) Coupling continental glaciations with groundwater flow models – Surface/subsurface interactions over the Canadian landscape during the Wisconsinian glaciation. LEMIEUX J.-M., SUDICKY E.A., PELTIER W.R. & L. TARASOV



2

2

9h 55 –10h 20 S 2 MONDAY (CONTINUED) Th1 (Paper ID:18)

Stochastic modeling of groundwater flow in the saprolite of a tropical gneissic watershed. MUDDU SEKHAR, A. CHAUDHURI, S. FLEURY and M. DESCLOITRES Th2 (Paper ID:29) Modeling of solute transport in a stream – aquifer system. Kathrin NANOU-GIANNAROU, K. SPANOUDAKI and A. I. STAMOU

10h 20 – 10h 50 Coffee break and POSTERS Monday morning (continued)

10h 50 – 11h 15 S 3 Th1 (Paper ID:68) A comparison of three finite volume methods for capturing irregular boundaries and heterogeneity in groundwater flow simulations. Dalila LOUDYI, Roger A. FALCONER, BINLIANG LIN Th2 (Paper ID:26)

A Depth-Continuous, Moisture Content-Discretized Interactive Infiltration Model Cary A. TALBOT, Fred L. OGDEN

11h 40 – 12h 05 S 4 Th1 (Paper ID:24)

Estimation of Flow Parameters in Heterogeneous Leaky Aquifers Nadim K COPTY, Angelos N. FINDIKAKIS, M. Savas SARIOGLU, Paolo TRINCHERO, Xavier SANCHEZ-VILA Th2 (Paper ID:28 & 105) Use of Environmental Tracer Data for Groundwater Modeling Giorgio Amsicora ONNIS, Rolf ALTHAUS, Roland PURTSCHERT, Stephan KLUMP, Rolf KIPFER,Harrie-Jan Hendricks FRANSSEN, Fritz STAUFFER and Wolfgang KINZELBACH

12h 05 – 12h 30 S 5 Th1 (Paper ID:66) Effect of heterogeneity and anisotropy on DNAPL migration in fractured plane SHIBANI JHA and M.S. MOHAN KUMAR Th2 (and Th4) (Paper ID:96)

The effect of near shore surface water bodies on submarine groundwater discharge Vassilios KALERIS



3

3

12h 30 – 12h 55 S 6 MONDAY (CONTINUED)

Th1 (Paper ID:34) Multi-Scale Characterization of an Heterogeneous Aquifer Through the Integration of Geological, Geophysical and Flow Data: A Case Study. B. BOURBIAUX, J.P. CALLOT, F. GAUMET, M. GUITON, R. LENORMAND and J.L. MARI Th2 (Paper ID:37)

Pathline-Calibrated Groundwater Flow Models of Nile Valley Aquifers, Esna, Upper Egypt Tom H. BRIKOWSKI , Abdallah FAID

12H 55– 14H LUNCH Monday afternoon: Themes 1 & 3

14h – 14h 45 Keynote Lecture 2 (KL 2) Simulating Complex Flow and Contaminant Transport Dynamics in an Integrated Surface-subsurface Modelling Framework. E.A. SUDICKY, J.M.LEMIEUX, J.P. JONES, A.E. BROOKFIELD, D. COLAUTTI, Y.-J. PARK (Department of Earth Sciences, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1) and R. THERRIEN, T. GRAF (Département de Géologie et de Génie Géologique, Université Laval, Québec, Canada G1K 7P4).

14h 45 – 15h 10 S 7 Th1 (Paper ID:42)

Quantifying groundwater model prediction uncertainties at a small scale highly heterogeneous remediation site Heinz J. THEIS Th3 (Paper ID:1) Hydraulically controlled combined vertical circulation of groundwater and alcohol Ulf MOHRLOK, Klaas HEINRICH

15h 10 – 15h 35 S 8 Th1 (Paper ID:45) Homogeneization of smaller scale heterogeneity at Äspö (Sweden) granitic site within a model for transfers of radionuclides at large temporal scales. Christophe GRENIER, Christian LAGUERRE, Gilles BERNARD-MICHEL.

Th1: Paper ID: 11 Title: Comparison of asymptotic semi-analytical solutions to solute transport in heterogeneous porous media Authors: Christophe C. FRIPPIAT and Alain E. HOLEYMAN



4

4

15h 35 – 16h S 9 MONDAY (CONTINUED)

Th1 (Paper ID:X1)

Stochastic Modelling of Unsaturated Flow and Transport: Numerical Experiments, Macroscopic Behavior and Upscaling. VEENA S. SORAGANVI, M. S. MOHAN KUMAR (Indian Institute of Sciences, Bangalore, India) and R. ABABOU (IMFT, Toulouse). Th3 (Paper ID:xxxx) Xxxx

16h – 16h 30 Coffee break and POSTERS Monday afternoon (continued)

16h 30 – 16h 55 S 10 Th1: (Paper ID:60 & 107) Effects of spatial heterogeneity and upscaling methods on hydrodynamic transport coupled with geochemical reactions. Marco DE LUCIA, Vincent LAGNEAU and Chantal DE FOUQUET Th3 (Paper ID:13)

Dissolution and precipitation processes in porous media: a pore scale model C.J. VAN DUIJN, V.M. DEVIGNE, T.L. VAN NOORDEN, I.S. POP

16h 55 – 17h 20 S 11 Th1 (Paper ID:65)

Asynchronous particle tracking for contaminant migration in heterogeneous 3D media with unstructured tetrahedral mesh Minh-Phuong LAM, Regina NEBAUER, Rachid ABABOU Th3 (Paper ID:14 & 106) Transverse effective dispersion coefficients in a chemically heterogeneous medium with flow fluctuations. Vanessa ZAVALA-SANCHEZ, Marco DENTZ and Xavier SANCHEZ-VILA

17h 20 – 17h 45 S 12 Th1 (Paper ID:77) Random and mixed random walk / finite volume methods for solute and heat transport in heterogeneous media. Gérald DEBENEST, Jean-F. THOVERT Th3 (Paper ID:16)

Crystal dissolution and precipitation in porous media flow: variable geometry. T.L. van NOORDEN, I.S. POP and C.J. van DUIJN

19 H: COCTAIL PARTY AT THE CITY HALL (to be confirmed)



5

5

TUESDAY 13 JUNE 2006 TUESDAY 13 JUNE

8h 30 – 9h 15 Keynote Lecture 3 (KL3) Multiscale Analysis of Biological Processes in Porous Media. Brian WOOD, Oregon State University, Corvallis, OR 97330 USA

Tuesday morning: Themes 1 & 2

9h 30 – 9h 55 S 13 Th1 (Paper ID: 30) Predicting the tracer plume development in fractured porous media by applying a double continuum approach Beyer MATTHIAS; Mohrlok ULF Th2 (Paper ID:38)

Flow over and within a porous bed computed using a macroscopic formulation of a low-Reynolds-number k-epsilon turbulence model. Mahmud AHSAN, Mark A. COTTON and Peter K. STANSBY

9h 55 –10h 20 S 14 Th1 (Paper ID:7)

Scale-dependence of dispersivity in upscaled transport models. Daniel FERNÀNDEZ-GARCIA, J. Jaime GÓMEZ-HERNÁNDEZ Th2 (Paper ID:46) Why, when and how we need to apply conjunctive water management of surface and groundwater: the case of the Charente basin, France. Fabien CHRISTIN, Gilles BELAUD, Christian LEDUC, Pierre-O. MALATERRE, Patrick LE GOULVEN

10h 20 – 10h 50 Coffee break and POSTERS

10h 50 – 11h 15 S 15 Th1 (Paper ID:12) Block-Upscaling of Transport: A comparison of ADE and mobile/immobile approach Matthias WILLMANN, Xavier SÁNCHEZ-VILA, and Jesus CARRERA Th2 (Paper ID: 93)

Groundwater flow modeling of perched Karstic aquifers and springs and its implication for determining precipitation-recharge relationship. Menachem WEISS and Haim GVIRTZMAN



6

6

11h 40 – 12h 05 S 16 TUESDAY (CONTINUED)

Th1 (Paper ID:21) Sub-gridded dual-porosity models: accurately modelling matrix-fracture transfers in fractured porous media C. FAMY, B. BOURBIAUX, P. LEMONNIER and M. QUINTARD Th2 (Paper ID:52)

Modelling Surfactant-Induced Flow and Contaminant Transport in the Vadose Zone. Scott E. MUNACHEN

12h 05 – 12h 30 S 17 Th1 (Paper ID:25)

Solute transport by a shear-thinning fluid in a channel flow: influence of the geometrical disorder Victor J. CHARRETTE, Elisa EVANGELISTA, Ricardo CHERTCOFF, Harold AURADOU, Jean-Pierre HULIN and Irene IPPOLITO Th2 (Paper ID:59) Methodology of the hydraulic and hydrodynamic modelling of aquifers-stream interactions. Noureddine GAALOUL

12h 30 – 12h 55 S 18 Th1 (Paper ID:44 & ID:69) Geostatistical modelling for the quantification of uncertainties on the unsaturated zone and the groundwater transfer. S. MAZUEL, C. DE FOUQUET, M. KRIMISSA, J.-P. CHILES Th1 (Paper ID: 101 & 103) An efficient local time stepping and Discontinuous Galerkin scheme for the advective transport equation in porous media Charbel-P. EL SOUEIDY, A. YOUNES & P. ACKERER

12H 55– 14H LUNCH

Tuesday afternoon: Themes 1 & 3

14h – 14h 45 Keynote Lecture 4 (KL 4) Paper ID:90 Analytical and numerical modeling of flow and transport in highly heterogeneous three-dimensional aquifers : ergodicity, gaussianity and anomalous behaviour. Igor JANKOVIC, Dept of Civil, Struct. & Envir. Engg., SUNY State Univ of New York at Buffalo, NY, USA.



7

7


Th1 (Paper ID:35 & ID:62)

A new method to invert hydraulic pumping tests for the identification of fractal characteristics and homogenization scale in fractured aquifers. Stephane BERNARD, Anne KACZMARYK, Gilles POREL, Frederick DELAY Th3 (Paper ID: 81) Role of sorption processes in reactive transport. Philippe BEHRA

15h 10 – 15h 35 S 20 Th1 (Paper ID:20) Understanding flow and mass transfer through an enhanced geothermal system (EGS) created into a deep fractured basement in the Rhine graben. Dominique BRUEL, Clement BAUJARD Th3 (Paper ID:31)

On Constrained Minimisation Techniques For Solving General Geochemical Reactive Transport Problems M. A. SBAI, MENJOZ A., AZAROUAL M.

15h 35 – 16h S 21 Th1 (Paper ID:49)

Permeability of porous media containing fracture networks with a power-law size distribution. V.V. MOURZENKO, I. BOGDANOV, J.-F. THOVERT, P.M. ADLER Th3 (Paper ID:47) Scaling kinetic reactions in heterogeneous media : from laboratory – to pilot – to field scale O. ATTEIA and C. GUILLOT

16h – 16h 30 Coffee break and POSTERS

16h 30 – 16h 55 S 22 Th1 (Paper ID:70) Impact of sedimentary structures with inclined couplets on macrodispersion in gravel aquifers. Fritz STAUFFER Th3 (Paper ID:53) Identification of groundwater endmembers: implications for the impact of liming on groundwater. Christina WEYER, Gunnar LISCHEID, Luc AQUILINA, Anne-Catherine PIERSON-WICKMANN



8

8


Th1 (Paper ID:74)

Modeling of flow and solute transport in a heterogeneous aquifer, Kerbala City, Iraq N. SAMANI and A. RAOOF Th3 (Paper ID:54) Micro-pore model for Cesium transport in sandy-clayed porous media. Sébastien CADALEN, Michel QUINTARD

17h 20 – 17h 45 S 24 Th1 (Paper ID:75) A new approach to study solute mixing in heterogeneous porous media. Paulo HERRERA and Roger BECKIE Th3 (Paper ID:56)

Development of Simplified Scaling Relationships for the Estimation of Subsurface Biological Parameters Using Moment Analysis. Mohamed M. MOHAMED and Kirk HATFIELD

20h 00 : Dinner banquet, downtown Toulouse (Hotel d’Assézat - fondation Bemberg)



9

9

WEDNESDAY 14 JUNE 2006

9h – 9h45 Keynote Lecture 5 (KL 5) Paper ID:89

Ancient and modern hydrology : the common ground and a few challenging tasks. Gedeon DAGAN, Faculty of Engineering, Tel Aviv University, Tel Aviv, Israël.

Wednesday morning: Themes 2 & 4

10h-10h25 Th4 (Paper ID:39) The flow systems and the groundwater interactions in the multi-aquifer system of the Carmel Coast region- Israel. Joseph GUTTMAN Th2 (Paper ID:61) Atmospheric controls and soil moisture inputs for a coupled model of surface-subsurface interactions. Claudio PANICONI, M. PUTTI

10h25-10h50 Th4 (Paper ID:67) Utilisation de l'effet barométrique pour la détermination du coefficient d'emmagasinement Frédéric LALBAT and Olivier BANTON Th2 (Paper ID:80) Coupled modeling of the functioning of continental surfaces from local to regional scales (SEVE : Sol-Eau-Vegetation-Energie). Modélisation couplée du fonctionnement des surfaces continentales aux échelles locales à régionales (SEVE : Sol Eau Végétation Energie). Valérie BORRELL-ESTUPINA(1), Isabelle BRAUD(9,2), Gérard DEDIEU(1), Aaron BOONE(1,3), Flora BRANGER(2), Yves BRUNET(7), Isabelle CALMET(4), Nadia CARLUER(2), André CHANZY(6), Philippe CHIBAUDEL(1), Jean-Dominique CREUTIN(9), Hendrik DAVI1, Alexandre ERN(5), Florence HABETS(3), Frédéric HECHT(11), Jérôme JAFFRE(12), Philippe LAGACHERIE(10), Jean-Claude MENAUT(1), Patrice MESTAYER(4), Roger MOUSSA(10), Joël NOILHAN(3), Jérôme OGEE(7), Albert OLIOSO(6), Laurent PREVOT(10), Fabrice RODRIGUEZ(8), Marc VOLTZ(10). 1CESBIO UMR5126 Toulouse, 2CEMAGREF Lyon, 3CNRM Toulouse, 4LMF Ecole Centrale de Nantes UMR 6598, 5ENPC-CERMICS Marne-la-Vallée, 6INRA-CSE Avignon, 7INRA-EPHYSE Bordeaux, 8LCPC Nantes, 9LTHE Grenoble, 10LISAH Montpellier, 11 INRIA Roquencourt, 12Laboratoire J.-L. Lions Paris (France).



10

10

10h50 – 11h20 Coffee Break and POSTERS WEDNESDAY (CONTINUED)

11h20-11h45: Th4 (Paper ID:73) Multiscale Multiphase Mass Transfer in Porous Media: Integrated Numerical and Experimental Studies. Souheil M. EZZEDINE, Russell L. DETWILER, Walt W. Jr. MCNAB Th2 (Paper ID:82)

Modeling coupled surface / subsurface flow interactions : implementation and comparison of three models based on Darcy, Boussinesq / Saint Venant and Boussinesq / diffusive wave, with application to the Garonne floodplain, Midi-Pyrénées, France. ABABOU R.(1), AL-BITAR A. (1), PEYRARD D.(2), QUINTARD M.(1), SANCHEZ-PEREZ J.M.(2), SAUVAGE S.(2), VERVIER P.(2), WENG P.(3). (1) Institut de Mécanique des Fluides de Toulouse ; (2) Laboratoire d’Ecologie des Hydrosystèmes (Toulouse) ; (3) Bureau des Recherche Géologiques et Minières (BRGM).

11h45-12h10: Th4 (Paper ID:83) Thermo-Hydro-Mechanical simulation of a 3D fractured porous rock (coupled matrix-fracture hydraulics). Rachid ABABOU, Israel CANAMON and Fco. Javier ELORZA

Th3: Paper ID: 57 Title: An Adaptive time stepping method based on Richardson Extrapolation : Application to batch chemistry, unsaturated flow and transport modelling. Authors: B. BELFORT, J. CARRAYROU, F. LEHMANN

12h10-12h35: Th1 (Paper ID:97) Interpretation of interference pumping tests in a fractured limestone (Poitiers - France) using fractal and dual-media approaches : Homogenization scale of hydraulic parameters. Anne KACZMARYK and Frederick DELAY Th3 (Paper ID:64)

NAPL dissolution in porous media : non-equilibrium effects due to saturation heterogeneities. C. TATHY, B. MABIALA and M. QUINTARD

12H 55– 14H: LUNCH 14h – 14h45: RESERVED FOR PROGRAMMATIC MEETINGS (IAHR, SHF, …)



11

11

Wednesday afternoon: Themes 1 & 4 WEDNESDAY (CONTINUED)

14h 45 – 15h 10: Th1 (Paper ID:76) Assessing model uncertainty in groundwater contamination models David DRAPER and Bruno MENDES Th4 (and Th2) (Paper ID :X3 ) Seawater intrusion modeling in heterogeneous costal aquifers with recharge and surface flow coupling via diffusive wave, Boussinesq and sharp interface equations. A.AL-BITAR and R.ABABOU (IMFT, Toulouse)

15h 10 – 15h 35 Th1 (Paper ID: 84) A numerical global upscaling technique for building reservoir models. Mickaele LE RAVALEC-DUPIN and Ludovic RICARD Th4 (Paper ID:32)

Framework for a process-based salinisation risk assessment: solute recycling versus primary groundwater salinisation. MILNES E., PERROCHET P., RENARD P.

15h 35 – 16h Th1 (Paper ID:79) Is there any hope for transverse dispersion ? M. DENTZ, E. ABARCA, J. CARRERA and X. SANCHEZ-VILA Th4 (Paper ID:17)

Modeling of the Groundwater Flow and Tranport of Reactive Solutes in the Salt Crust Aquifer, Salar de Atacama, Chile. CARLOS VASQUEZ G., JOSE MUÑOZ P.



12

12

16h – 16h 30 Coffee break and POSTERS WEDNESDAY (CONTINUED)

16h 30 – 16h 55 Th1 (Paper ID:X2)

Contaminant source identification with the “RAW” scheme : particle tracking with reverse anti-diffusion walk. R. ABABOU (1), A.C. BAGTZOGLOU (2) (1) Institut de Mécanique des Fluides de Toulouse ; (2) Univ. Connecticut, Storrs, CN, USA. Th4 (Paper ID:23) Numerical Modeling as a Tool to Investigate the Feasibility of Artificial Recharge to Prevent Possible Saltwater Intrusion into the Bangkok Coastal Aquifers System. Phatcharasak ARLAI , Manfred Koch, .Sucharit Kuntanakulwong, and Werapol Bejranonda 17h : CLOSING SESSION, and…FAREWELL.



13

13

ALL POSTER PAPERS

POSTER PAPERS ARE SCHEDULED DURING COFFE BREAKS

EVERY DAY (12-13-14 JUNE 2006) P Paper ID: 5 Advantages of geostatistical simulation in hydrogeology using krigeage aproach study case (sand aquifer of Biskra, Algeria) A.H. MESSAMEH , S. TEKKOUK , A. BOUGUERNE P Paper ID: 6 Integration of recharge data from a hydrological water balance model into a 3D FE model for a crystalline basement area in the tropical river catchment of the Upper Ouémé (Benin / West Africa) Tobias EL-FAHEM, Simone GIERTZ, Barbara REICHERT & Bernd DIEKKRÜGER P Paper ID: 27 (Th1) Mixing and spreading of a solute in a non Newtonian fluid flow inside a fracture A. BOSCHAN, H. AURADOU, J.P. HULIN, I. IPPOLITO, R. CHERTCOFF P Paper ID: 88 (& ID33) (Th2) Continual estimation of surface and underground flows in river basins Valeriy KLENOV P Paper ID: 36 Investigating the effect of forest stand volume on soil surface erosion by Geographic Information System(GIS). H.R.MASKANI , A.MERAJI. P Paper ID: 41 Prediction of future water levels of the pleistocene aquifer system of Wadi El-Assiuti area, eastern desert, Egypt Hossam H. ELEWA



14

14

POSTER PAPERS (CONTINUED)

P Paper ID: 51 Shallow Groundwater Flow in Unsaturated Hillslopes and its Implications for Landslide Mobilisation Scott E. MUNACHEN P Paper ID:78 Should dispersion describe mixing or spreading ? J. CARRERA, M. DENTZ and X. SANCHEZ-VILA P Paper ID: 71 Coupling of wetlands to sea by groundwater-borne nutrient transport Joseph Sebastian PAIMPILLIL P Paper ID:86 (Late abstract) A finite volume analysis on unstructured grids of aquifers in connection with rivers and lakes A. NJIFENJOU, A.J. KINFACK P (A or P -- Late abstract) Th2 Effet d’un fossé en travers sur l’écoulement hydrodynamique d’une nappe superficielle peu profonde. Application sur le site expérimental de la Jaillière (44, France). T.H. DEBIECHE , C.V. ADAMIADE, N. CARLUER

Paper ID:87 (Th2) Estimation of spatial-temporal variability of groundwater recharge by using fully coupled SWAT-MODFLOW model. IL-MOON CHUNG, NAM-WON KIM, JEONGWOO LEE, YOO-SEUNG-WON

Paper ID:9 (Th3) Coupled Geochemical and Transport Modeling of pH-Dependent Biodegradation of Organic Contaminants in a Pyrite-Rich Aquifer M. Savas SARIOGLU and Nadim K. COPTY

KEYNOTE LECTURE

Probabilistic representation of heterogeneous geologic media. La représentation des milieux géologiques hétérogènes

par des méthodes probabilistes.

Ghislain De MARSILY, Académie des Sciences, Université Paris 6, Paris, France.

ABSTRACT On peut aborder le problème de l’hétérogénéité en s’efforçant de définir une perméabilité équivalente homogène, par prise de moyenne, ou au contraire en dé-crivant la variation dans l’espace des propriétés des roches à partir des observations géologiques et des mesures locales. Les techniques disponibles pour une telle description sont soit continues, comme l’approche Géostatistique, soit discontinues, comme les modèles de faciès, Booléens, ou bien par Indicatrices ou Gaussiennes Seuillées, ou enfin Markoviens. Ces modèles de faciès sont mieux capables de prendre en compte la connectivité des strates géologiques, telles que les chenaux enfouis à forte perméabilité, ou au contraire les faciès fins de barrières de perméabilité, qui ont une influence importante sur les écoulement, et, plus encore, sur le transport. Les modèles génétiques récemment apparus ont la capacité de mieux incorporer dans les modèles de faciès les observations géologiques, chose courante dans l’industrie pétrolière, mais insuffisamment développée en hydrogéologie. On conclut que les travaux de recherche ultérieurs devraient s’attacher à développer les modèles de faciès, à les comparer entre eux, et à mettre au point de nouvelles méthodes d’essais in situ, comprenant les méthodes géophysiques, capables de reconnaître la géométrie et les propriétés des faciès. La constitution d’un catalogue mondial de la géométrie et des propriétés des faciès aquifères, ainsi que des méthodes de reconnaissance utilisées pour arriver à la détermination de ces systèmes, serait d’une grande importance pratique pour les applications.

FLOW AND TRANSPORT MODELLING AND EFFICIENCY OF REMEDIATION IN A FRACTURED AND KARSTIC AQUIFER: A CASE STUDY Claudia Cherubini1, Concetta I. Giasi2 1 PhD Student Polytechnical University of Bari [email protected] 2 Full Professor Polytechnical University of Bari [email protected] ABSTRACT The study of contaminants propagation in fractured and karstic aquifers shows uncertainties caused by the conditions of anisotropy of the medium and by the presence of cavities and residual products that could make fluid flow and solute transport unforeseeable. The present paper proposes the study carried out on a contaminated area located in the city centre of Bari (Italy) in which flow modeling has been implemented, starting from the use of a finite - difference computer code and considering the conditions of anisotropy of the medium by means of a reconstruction of the hydraulic conductivity map. A verification of the reconstruction of the so obtained map has been effectuated through the comparison with the map of the “draining degree” coming from the RQD (Rock Quality Designation) values obtained from boreholes. Those values, that could be related to hydraulic conductivity of the rocky mass interested by a network of karstic canals and fractures by means of empiric formulas, have been utilized in order to achieve an indirect estimation of the dynamics of subterranean drainage. The choice of the indicator pollutant has taken into account not only the activities carried out in more than a century in the study area, but also the properties of the contaminant itself. The studied model has been realized by means of the computer code Modflow (GWV4), supported by the description of hydraulic conductivity obtained through interpolation. The study has afterwards considered the opportunity of remediation through Pump&Treat for a period of two years and a subsequent check of the performed contaminant demolition. The analysis of the results of flow and the application of MT3DMS, allows to confirm an evident coherence of the hydrogeological model with the hydrodynamic one, showing capture zones of flow, and therefore of the contamination. The results of a study carried out in the area, based on a very accurate geomechanical analysis of the rock fracturing degree lead to different results. The contamination does not appear circumscribed inside the area at each depth. That should be taken into consideration in setting up remediation strategies: in order to be effective, they need to be applied to specified depths. Keywords: karstic, fractured, draining, pollution, remediation.

Impact of the Wisconsinian Glaciation on Canadian Continental

Groundwater Flow J.-M. Lemieux, E.A. Sudicky, University of Waterloo, Department of Earth Sciences, 200 University Avenue West, Waterloo, Ontario, Canada, N2L 3G1 W.R. Peltier, L. Tarasov University of Toronto, Department of Physics, 60 St. George Street, Toronto, Ontario, Canada, M5S 1A7 During the last glacial period (75 kyr – 10 kyr), the Canadian landscape was almost entirely covered with ice. The Laurentide ice-sheet, the largest of the three North-American ice sheets, reached a thickness of about 4 km and the force exerted by its weight on the earth’s crust was sufficient to cause a depression of the surface of about 1 km and an over-pressurization of porewater fluids. These dramatic conditions are suspected to have had a large impact on the groundwater flow system over the whole continent. Although an analysis of the evolution of groundwater flow systems during glacial periods is relevant to a number of problems, such as the long-term stability of high-level spent nuclear-fuel repositories located at depth, very few studies have been conducted to assess the impact of glaciation on deep-seated groundwater flow systems, particularly in a North-American context. A transient, highly heterogeneous three-dimensional groundwater flow model including the effect of the advective-dispersive redistribution of shield brines was constructed in order to capture the impact of the advance and retreat of the ice sheet over the Canadian landscape. The model is driven by a thermomechanical ice-sheet model of the last glacial cycle [Tarasov and Peltier, 2004] which provides the transient boundary conditions that includes the spatio-temporal distribution of the glacial ice, the elevation of the surface topography, meltwater rates, permafrost thicknesses, as well as temporal changes in sea level along the coastal margins. The evolving surface water drainage patterns and features such as proglacial lakes are also incorporated based on the hydrologic routing calculations performed by Tarasov and Peltier [2005]. The treatment of physical processes related to the influence of the ice sheet on the groundwater flow system such as hydromechanics, isostasy, subglacial melting and permafrost formation are also discussed. Simulation results show that hydraulic heads at depth below the ice sheet increase by several hundred meters and groundwater flow directions also change dramatically from what is observed today. Infiltration of subglacial meltwater also plays a key role in the increase of subsurface hydraulic heads as the meltwater is driven into the subsurface by the weight of the ice.

1

Stochastic modeling of groundwater flow in the saprolite of a tropical gneissic watershed

M. Sekhar1,2, A. Chaudhuri1, S. Fleury2 and M. Descloitres2

1Department of Civil Engineering, Indian Institute of Science, Bangalore, 560 012, India 2Indo-French Cell for Water Sciences, Indian Institute of Science, Bangalore, 560 012, India.

Abstract The groundwater flow in the granitic gneissic crystalline rocks, which are one of the abundant formations in the peninsular India is governed by a complex and irregular saprolite (i.e. weathered zone) along with the underlying deeper fractured formation. Integrating geophysical measurements in hydrogeological modeling provides a better understanding of fluid flow processes in such complex groundwater environments. The present paper discusses the probabilistic behavior of groundwater flow in the saprolite of a small experimental watershed using integration of stochastic modelling with information from various types of geophysical investigations. Investigations are made for assessing groundwater behavior at the small watershed scale (i.e. < 5 km²) at Moole Hole watershed of Nugu river basin in South India. The groundwater system is described by a highly foliated weathered zone of thickness ranging from 20 m to 35 m underlying a red and black soil system and overlying a fresh fractured gneissic rock. A 2D electrical imaging at several profiles (twelve numbers) in the watershed indicates a complex heterogeneous structure of the saprolite in the regolith-protolith substratum. The interface between the saprolite and the fresh rock is very irregular and interestingly the organization of these zones follow a near vertical structure. From such an organization, it is hypothesized that some portions of the saprolite react quickly to the recharge water and transmit the same through the lateral groundwater flow occurring within the saprolite. Figure 1 presents an electromagnetic map of the Moole Hole watershed. The electromagnetic conductivity (in mS/m) is shown using 4 classes. The first class (3-9 mS/m, shown in white color) represents the zones where the saprolite (i.e. the weathered zone) is close to the surface (< 3 m). Its thickness can be estimated using the 2D electrical imaging survey. The upper class (20-50 mS/m, shown in black color) represents the black clayey soil distribution whose surface is almost impermeable to infiltration. Below these black soils, the saprolite could be close or far to the surface. This is only known from 2D electrical imaging survey. Between those classes, i.e. between 9 and 20 mS/m, the 2D electrical imaging survey suggests that the saprolite is deeper than 10 m. Figure 2 presents an example of the 2D electrical imaging survey for the profiles 1 and 5. The resistivity classes (four in number) are shown from the knowledge of the weathering index versus resistivity relationship obtained in borewells. The first class (10-80 Ohm) corresponds to very clayey materials, almost impermeable. The second one (80-150

2

ohm.m) corresponds to loamy material, which can be categorized as poorly conductive from hydraulic point of view. The third one (150-500 ohm.m) corresponds to the saprolite, a loamy to sandy material, porous and hydraulically more conductive (in the range of 2x10-6 to 2x10-5 m/s according to magnetic resonance sounding and slug test in borewells). This saprolite could also be fractured in some parts. The last class (above 500 ohm.m) corresponds to the fractured and to fresh rock, mainly gneissic. The profiles 1 and 5 represent extreme cases: the profile 1 exhibits high variations of the saprolite depth and thickness, while the profile 5 exhibits a situation where the saprolite is deep and remains thin. From the topographic map, electromagnetic map information and using the results given by the 2D electrical imaging surveys, the 3D geometry of the saprolite is generated using a stochastic model. The hydraulic conductivity is assumed to be random field and described by the mean, standard deviation and correlation (or spectral) function. The distribution of hydraulic conductivity of the saprolite obtained using Magnetic resonance soundings and slug tests is used in the stochastic model for generating hydraulic conductivity field in the 3D weathered zone. The stochastic analysis of flow in the 3D weathered zone is performed using the Monte Carlo Simulation (MCS) method in a region of 1 km x 1 km near the outlet of the watershed using the hydraulic gradient obtained through the monitoring well network in this region. The model is used to study the behavior of flow in the saprolite and also to analyse the effect of correlation structure on the probabilistic behavior of groundwater fluxes.

Figure 1 Electromagnetic map of Moolehole watershed showing the depth to the saprolite and its spatial structure. Figure 2 The 2D electrical resistivity along profiles 1 and 5 shows the structure of the thickness of saprolite in the regolith-protolith substratum.

Conductivity EM 31 (mS/m)

3 9 15 30 50

123

45

50 100 150 200 250 300 350 400 450 500horizontal distance (m)

800

820

840

Elev

atio

n (m

)

Profile 1

50 100 150 200 250 300 350 400 450

820

840

860Profile 5

S N

S N

80 150 500

calculated resistivity (Ohm.m)

clayey materialfractured / fresh rock

saproliteloamy material

10 3000

MODELING OF SOLUTE TRANSPORT IN A STREAM – AQUIFER SYSTEM

KATHRIN NANOU-GIANNAROU, KATHRIN SPANOUDAKI AND ANASTASIOS I. STAMOU

Laboratory of Applied Hydraulics, Department of Water Resources, Hydraulic and Maritime Engineering, School of Civil Engineering,

National Technical University of Athens, Iroon Polytechniou 5, 15780 Athens, Greece

email: [email protected]

ABSTRACT

An Integrated Surface water – Groundwater Model (ISGM) is presented, describing flow and pollution interaction between a stream and the surrounding groundwater aquifer. The hydrodynamic part of the ISGM has also been presented in [1] and [2]. In this work the integrated model is further developed to allow the simulation of solute transport in the stream, solute transport in the aquifer and solute interchange between the stream and the aquifer. The ISGM consists of (a) a 3-D surface water sub-model, FLOW-3DL [3], successfully applied in numerous modelling studies of hydrodynamics and water quality in estuarine and coastal waters and (b) a 3-D saturated groundwater flow sub-model, both based on the finite difference method and using orthogonal grids. The momentum and mass conservation equations are the governing equations for both surface and groundwater flows and the advection-diffusion equation is used as the governing equation for solute transport in the linked stream-aquifer system. The newly developed groundwater sub-model of the ISGM is verified with benchmark cases and existing analytical solutions for solute transport in a steady uniform flow field assuming a homogeneous and isotropic aquifer and point or distributed sources of constant concentration or constant flux rate ([4], [5], [6], [7]). Finally, the ISGM is applied to a hypothetical stream-aquifer system showing the interaction between a confined aquifer and a partially penetrating stream for more complex flow fields where analytical solutions do not exist. Cases of both steady and non-steady flow are examined, representative of the possibilities offered by the use of integrated models. For the steady flow cases, a constant inflow to the stream is assumed and different situations of groundwater head elevations relative to stream stage combined with stream and/or groundwater contamination are addressed. For the non-steady flow cases, a tidal wave in the stream is assumed and various scenarios of solute release in the stream and the aquifer are studied.

Key words: integrated modelling, stream-aquifer interaction, solute transport, surface water, groundwater

REFERENCES

[1] Spanoudaki, K., Nanou-Giannarou, K. and Stamou, A.I., (2005). 3-D Modeling of River-Groundwater Interactions. EWRA, 6th International Conference, Menton, September 7-10. [2] Spanoudaki, K., Nanou-Giannarou, A., Stamou, A.I., Christodoulou, G., Sparks, T., Bockelmann, B. and Falconer, R.A., (2005). Integrated surface/subsurface water modeling. Global Network for Environmental Science and Technology, 9th Conference on Environmental Science and Technology, Rhodes, Greece, September 1-3. [3] Stamou, A. I., Noutsopoulos, C., Pipilis, K. G., Gavalaki, E., and Andreadakis, A., (1999). Hydrodynamic and Water Quality Modeling of Southern Evoikos Gulf- Greece. Global Nest the Int. J., vol. 1 (2), pp 5 -15. [4] Van Genuchten, M.Th. and Alves, W.J., (1982). Analytical solutions of the one-dimensional convective-dispersive solute transport equation. U.S. Department of Agriculture Technical Bulletin No. 1661, 151 pp.

[5] Wilson, J.L. and Miller, P.J., (1978). Two-dimensional plume in uniform ground-water flow. Journal of Hydraulics Division, ASCE, v.4, pp. 503-514. [6] Chan, S. and Javandel, I., (1996). Analytical solutions for solute transport in a vertical aquifer section. Journal of Contaminant Hydrology, v. 27, pp. 63-82. [7] Hunt, B., (1983). Mathematical Analysis of Groundwater Resources. Butterworths, Cambridge.

International Ground Water Symposium on“ Ground Water Hydraulics in Complex Environments”

(Heterogeneous Media, Coupled Processes, and Upscaling)

June 12-14, 2006Toulouse, France

A comparison of three finite volume methods for capturing irregularboundaries and heterogeneity in groundwater flow simulations

Dalila Loudyi, Roger A. Falconer, Binliang Lin

Cardiff School of Engineering, Cardiff University, Cardiff, U.K.

Abstract

Three finite volume methods to capture boundaries irregular geometry and heterogeneity

in groundwater flow simulations have been compared in this study. The two-dimensional

formulation has been considered. Three discretisations of the two-dimensional diffusion

equation, governing groundwater flow and for use with structured non-orthogonal

quadrilateral meshes, have been developed. The three methods rely on a cell-centred

finite volume approach, but show distinct differences in the choice of: gradient

approximation, head interpolations and control volume. A time implicit formulation has

been used in each model. The sparse system of linear equations that result from the

implicit formulation has been solved by using an iterative solver, based on the strongly

implicit procedure. Five test examples have been undertaken to compare the performance

of the newly developed methods against MODFLOW predictions and analytical results.

The accuracy of the results obtained was found to depend on the spatial and temporal

discretisations. One of the three developed methods proved its robustness, with regard to

mesh non-orthogonality and skewness, and was called the GWFV method. A discussion

about the performance of the new developed model has been included and the model has

been shown to perform well in comparison with MODFLOW.

A Depth-Continuous, Moisture Content-Discretized Interactive Infiltration Model

Cary A. Talbot Coastal & Hydraulics Laboratory US Army Engineer Research & Development Center Vicksburg, MS 39180 USA

Fred L. Ogden Department of Civil and Environmental Engineering, U-37 Univ. of Connecticut 309 F.L. Castleman Building Storrs, CT 06269 USA

As part of the on-going System-Wide Water Resources Program (SWWRP), research is being conducted by the US Army Engineer Research & Development Center (ERDC) in the development of coupled surface and subsurface flow interaction codes. In an effort to improve the computational efficiency and robustness of one-dimensional vadose zone flow calculations, alternatives to the Richards’ Equation (RE) are sought. This paper will describe a depth-continuous, moisture content-discretized interactive infiltration model under development at ERDC. The range of pore radii within a given soil is discretized into vertically continuous, interactive bins. The entry and vertical movement of wetting fronts in each bin are simulated by means of explicit infiltration approximations based on capillary and gravitational driving forces. Downward advancement of a wetting front within a bin can create pore-water deficits that are satisfied by capillary-driven inter-bin flow. Comparisons of model results with RE solutions and computational efficiency will be presented.

Estimation of Flow Parameters in Heterogeneous Leaky Aquifers NADIM K. COPTY, M. SAVAŞ SARIOĞLU Institute of Environmental Sciences, Bogazici University, Bebek 34342, Istanbul, Turkey e-mail: [email protected] ANGELOS N. FINDIKAKIS Bechtel National, Inc., 50 Beale Street, San Francisco, CA, 94105-1895, USA PAOLO TRİNCHERO & XAVIER SANCHEZ-VILA Department of Geotechnical Engineering and Geosciences, Technical University of Catalonia, Barcelona, Spain

Abstract

In natural geologic systems, confining layers overlying and/or underlying an aquifer are seldom completely impermeable; instead, most of them leak to some extent. This is particularly true in multi-layer and complex geologic systems. As a result the transient drawdown due to pumping is often dependent on the pumped aquifer properties as well as the hydrologic properties of the adjacent layers. The purpose of this paper is to examine the impact of heterogeneity of leaky aquifer systems on the analysis of pumping test data.

The aquifer system considered consists of two aquifers separated by an aquitard. A fully penetrating well is placed in one of the aquifers. The head in the second aquifer is assumed to be steady through the pumping test duration. The log-transmissivity of the pumped aquifer and the vertical capacitance of the aquitard are modeled as two independent multi-variate random spatial function with stationary first and second moments.

Monte Carlo simulations are used to simulate the time-dependent drawdown at the extraction well and nearby observation wells for different values of the statistical parameters defining the aquifer and aquitard flow properties. The simulated drawdown data at various distances from the well are then used to estimate the flow parameters (transmissivity, storativity and aquitard conductance) using (i) the inflection-point method developed by Hantush (1956) and (ii) a curve fitting approach based on the leaky aquifer type-curves developed by Walton (1962). These estimates are compared to each other and to the actual values used in the data generation. Results from this study indicate that the estimation of the flow parameters is dependent on the distance to the pumping well, and that the spatial variability of one parameter influences the estimation of the other parameters. Moreover, the comparison between the two analysis methods is influenced by the heterogeneity of the aquifer and aquitard in different manners. The implications of these results on the interpretation of pumping tests conducted in heterogeneous leaky and non-leaky aquifer systems are discussed.

Use of Environmental Tracer Data for Groundwater Modeling Giorgio Amsicora Onnis, Harrie-Jan Hendricks Franssen, Fritz Stauffer and Wolfgang Kinzelbach Institute of Hydromechanics and Water Resources Management Swiss Federal Institute of Technology, ETH Zürich e-mail: [email protected] Since the early ´50s, human activities such as nuclear power production, thermonuclear bomb testing and industrial processes released a range of environmental tracers into the atmosphere. The atmospheric concentrations of these tracers were measured at reference sites over the last decades. By measuring the concentrations in the groundwater and comparing the measured values with the known atmospheric growth curve (called atmospheric input function) it is possible to determine the groundwater age and gain other types of information about subsurface dynamics of water such as travel times, streamline information, ratio of fluxes, recharge rates, porosity. When considering thick unsaturated zones, the time spent by the tracer in the unsaturated zone results in a time lag – or delay - that must be considered in the calculation of the groundwater age. This effect is negligible for gaseous tracers which can diffuse to the water table in the unsaturated zone and distances to water tables of less than 10 m. But it gains in importance as deeper water tables are considered, resulting in delays which can be of 15 years or more (Cook and Solomon, 1995). The aim of the present work is to investigate the subsurface transport of the tracers 3H, 3He (as decay product of 3H), 85Kr and SF6 . First, we solve numerically the vertical advection-dispersion equation in the unsaturated zone (Fig. 1). Considering different tracers has the advantage of taking into account the different transport mechanisms in the subsurface. The water-bound 3H moves by advection following the seepage water, while the transport of the gas tracers like 3He, 85Kr and SF6 is advection dominated only in the few upper meters of the unsaturated zone, while afterwards they are diffusion dominated. A combination of several tracers may make their application to the flow modelling more reliable.

0 5 10 15 20 25 30 35 40 45 50 55

Fig.1 85Kr (years 1950-2004) and 3H (years 1978-2004) time series for different depths of the unsaturated zone.

Crossing the unsaturated zone also results in a modification of the temporal concentration distributions of the tracers. The shape of the input function at the groundwater table may thus significantly differ from that of the atmospheric input function. Consequently, the concentration history at the bottom of the unsaturated zone must be reconstructed to give a correct input function for any meaningful transport modelling in the saturated zone. We apply our theoretical studies to the Baltenswil aquifer (Switzerland). The catchment of this reference site was selected as it has been relatively well studied in the last decade. The undulating topography of the site – a sandy-gravel formation formed in the Riss ice-age covered by moraine material – gives rise to a complicated unsaturated zone geometry that makes Baltenswil a challenging test site for the validation of our unsaturated/saturated zone transport model.

0

10

20

30

40

50

60

70

80

90

85Kr input function c(z,t)

Year

Con

cent

ratio

n [d

pm/c

cKr]

soil surface10 mt.20 mt.30 mt.40 mt.50 mt.

0 2 4 6 8 10 12 14 16 18 200

20

40

60

80

100

Year

Con

cent

ratio

n [T

U]

3H input function c(z,t)

soil surface10 mt.20 mt.30 mt.40 mt.50 mt.

mailto:[email protected]

The flow model for the saturated zone is based on a set of 100 multiple equally likely realizations of the log-transmissivity field generated with GCOSIM3D (Gomez-Hernandez and Journel, 1993) conditional on transmissivity (T) measurements. In addition, these logT fields are also conditioned to steady-state hydraulic head measurements by the Sequential Self-Calibrated Method (Gomez-Hernandez et al., 1997), implemented in the code INVERTO (Hendricks-Franssen, 2001). For transport modeling in the saturated zone we need to provide the correct input function at the groundwater table. We thus reconstruct the concentration history at the bottom of the unsaturated zone according to the spatially variable thickness, which ranges from 0 to 50m in the domain of interest. This is then used in a second step as concentration input flux for transport modeling in the saturated zone (performed with MT3DMS (Zheng, 1990)). Accounting for tracer transport in the unsaturated zone allows, even with the simple assumption of a 1D homogeneous porous medium, the simulated concentrations at observation locations to show a fair simultaneous match with the measured concentrations for each of the considered environmental tracers (Fig. 2).

Figure 2: Modelling results vs. measured data at Baltenswil pumping station.

Sensitivity analysis shows that atmospheric input function and heterogeneity-related unsaturated zone parameters (thickness, soil water and air content, effective diffusion coefficient) are more significant to the final results than saturated zone ones (transmissivity and porosity). This prominent role is explained in terms of different residence times in the two zones, which in the aquifer under study are on average 30yrs/3yrs for the unsaturated/saturated zone respectively. The unsaturated zone regulates the subsurface dynamics as an interface between atmosphere and groundwater, with a slower dynamics compared to the movement of tracers in the saturated zone. Nonetheless transmissivity may come into play because it may change the water flux direction and thus the tracer trajectories in the saturated zone. Since different paths are related to overlying unsaturated zone soil columns of different thickness, different paths may result in a different ratio between the saturated and unsaturated zone travel times. When this ratio becomes large, such as 25% or more, saturated zone parameters like porosity may come again into play, influencing and determining the final concentration in the groundwater at the observation locations. Different transmissivity realizations would also influence the catchment areas of the different pumping stations in the domain, and give thus the possibility to include or exclude certain areas within the domain for the transport modelling.

Effect of Heterogeneity and Anisotropy on DNAPL Migration in Fractured Plane

Shibani Jha1 and M.S.Mohan Kumar2

Department of Civil Engineering, Indian Institute of Science, Bangalore 560 012, India Email : [email protected] ; [email protected] The risk for groundwater contamination by DNAPLs is high because of their high rates of production and high frequency of usages worldwide, which increases the likelihood of uncontrolled spills and releases. Fractures and joints giving rise to complex geological environments mostly dominate the heterogeneous geological media. Here lies the importance to predict, through efficient numerical simulations, the multiphase behaviour of DNAPL and water within the rough walled fracture in order to facilitate enhanced energy resource recovery (petroleum, natural gas, geothermal water and steam) and environmental protection (chemical and radiation contamination in ground water aquifers) in a complex heterogeneous subsurface environments. The accidental release of DNAPLs at the ground surface can migrate through the unconsolidated layers, and pool on the surface of a fractured clay aquitard. In such cases the DNAPL may further enter open fractures that exist in the aquitard, and it may be transferred to lower aquifer in the geological system. Such situations can create the ground water contamination problem. These geological formations are highly heterogeneous and anisotropic in nature. The complexity of the simulation increases if the multiphase fluid displacements occur in such environment. To simulate the dynamic front propagation behaviour of the various phases in such environment, efficient numerical models are required which can also simulate the effect of heterogeneity and anisotropy on DNAPL migration. For more local scale problems in which heterogeneous properties of the fractured formation have a significant influence on the multiphase flow behaviour, a proper simulation model is required. It is important to study the effect of heterogeneity at different scales on DNAPL movement in fractured environment. Heterogeneity can be visualized at various scales ranging from layered, random to correlated heterogeneity. The effect of anisotropy on DNAPL propagation will also be studied and the DNAPL plume will show the preferential arrangement of paths in different directions. The effect of gravity drainage on DNAPL movement will also be studied through variation in fracture inclination. Sensitivities of these aspects on DNAPL movement will be of significance in order to identify the prominent process. A proper simulation of the DNAPL plume, which is a prerequisite for remediation of contamination in such complex fracture formation, will be done to demonstrate the effect of correlation lengths on DNAPL plume movement. This will identify the proper grid size selection along with proper choice of correlation lengths for accurate estimation of the plume. These studies will be conducted on several small domains under various domain properties and hydraulic conditions. The model is based on numerical discretization scheme, which solves the general equations of mass conservation for each phase. To solve the equations, with proper pressure and saturation constraints and constitutive relations, in terms of pressure and saturation, the boundary conditions in terms of dependent variables or the first derivative of the dependent variables normal to the boundary has to be specified. The state of the system at time=0 has to be specified in terms of dependent variables. The non-linearity

of the problem involved is treated with Picard’s method and a sequential approach is adopted to reach the iterative solution. The fracture permeability is derived by assuming two phase fluid displacement in a fracture system represented through incompressible parallel plate flow within two dimensional small sub-regions of constant aperture in the fracture plane. The fluid phase distribution is implicitly represented at each location within the domain. Portions of the fracture are partially occupied by aqueous phase and partially by DNAPL, depending on the local capillary pressure and global accessibility condition. The aperture distribution for the fracture is generated using the geostatistical methods like the Random Field Generator, assuming that the apertures follow a Log-normally distributed, two-dimensional, spatially correlated random field. The void space of a fracture, visualized as a two-dimensional heterogeneous system, circumvent the need to specify a physical correlation length by carrying out the various simulations in a fracture plane whose correlation lengths are found to be the fractions of the domain size. Even though the aperture distribution generated yields points in certain regions of the fracture plane with extremely small apertures, actual closure at these points is not attained. The comparative study of nodal spacing and correlation length of the aperture distribution, in both x and y directions, have to be studied which will show the intensity of heterogeneity and anisotropy in the fracture and its effect on DNAPL movement. Model results will be studied using the saturation contour plots, which will show the preferential paths chosen by DNAPL. With the increase of time, it can be seen that the DNAPL has invaded those regions of the fracture field, which it was unable to invade earlier. Also it can be seen that certain regions of the fracture remain void of DNAPL at all the times. The ability for DNAPL to enter into the small opening regions of a fracture depends upon whether or not the capillary pressure at the advancing front exceeds the local entry pressure of the fracture. This ability increases as a function of the depth of penetration into the fracture since this depth determines the maximum capillary pressure that can be generated at the advancing front. These processes are to be demonstrated with respect to DNAPL introduced into a fracture initially filled with water. Under the simultaneous flow of both the phases and for the condition of capillary force dominating over viscous drag, DNAPL can flow in a fracture field only if there are continuous interconnected pathways that avoid small apertures, which would be blocked by water. This situation will arise if a long range spatial correlation among the large apertures exits. And also it can be demonstrated through numerical experiments that a longrange spatial correlation among small apertures facilitates the water phase movement in the heterogeneous field. The effect of gravity in anisotropic field will be studied by systematically varying the fracture orientation through angles of dip from horizontal through vertical. The DNAPL plume behaviour will be represented in the extreme cases of vertical and horizontal fractures. To study the effect of anisotropy on DNAPL migration in a multiphase system, various discretized representations of fracture apertures with different correlation lengths have been numerically generated in a fracture plane of various sizes and properties. The model is to be studied for these fracture fields and it will indicate that the relative permeabilities of both the phases are very much sensitive to the nature and range of spatial correlation between apertures.

The effect of near shore surface water bodies on submarine

groundwater discharge by

Vassilios KALERIS Department of Civil Engineering

University of Patras, Greece e-mail: [email protected]

ABSTRACT

Submarine groundwater discharge (SGD) is investigated numerically for a coastal

aquifer, in which a surface water body (SWB) is embedded. The conditions

considered appear in aquifers with lagoons, flat lakes or wetlands near the coast. In

such cases SGD can contain terrestrial groundwater, recirculated seawater, and water

originating from surface water body as well. The estimation of these components of

SGD is important for material budgets for the near shore seawater. Further, the

exchange between a surface water body and the near shore groundwater influences

not only the water budget of SWB but its quality too, as the groundwater discharging

into SWB can contain water of marine origin. Owing to the fact that lagoons, flat

lakes or wetlands in the coastal area represent a source of livelihood for local

communities, may have a high recreational value and/or serve as habitats for rare

species, the study of their water and material budget is important.

The numerical simulations are performed using the code FAST-C(2D), developed by

Holzbecher (1998). The model is based on the equation of flow in porous media of

fluids with variable density and the transport equation of salinity (brine). These

equations are coupled, as salinity influences density. The equations are presented in

the paper in dimensionless form and the dimensionless parameters influencing the

process are discussed. For selected combinations of these parameters the main

features of flow are investigated.

Figure 1 shows schematically the partial fluxes in the system coastal aquifer – surface

water body for the conditions investigated in this study. The calculated flow patterns

show that terrestrial groundwater discharges to the surface water body and not to the

sea. Additionally, a part of the saltwater flux discharges in the surface water body.

Submarine groundwater discharge consists of the recirculated seawater and of the

surface water, which invades the aquifer at the seaward edge of the surface water

Fig. 1: Schematic representation of the partial fluxes in the system coastal aquifer – surface water body. (SFGD=flux of terrestrial groundwater, SGD=total submarine groundwater discharge, RSGDc=flux of recirculated sea water, SWI= flux of seawater intruding the aquifer, SWIS=flux of groundwater of marine origin discharging into the surface water body, SWEX= total flux of groundwater discharging into the surface water body, SWIN=flux of surface water discharging into the sea, RSW=flux of recirculated surface water, msl=mean sea level)

body. The results show that the portion of recirculated water in SGD increases if the

intrusion of sea water in the aquifer becomes stronger. This occurs (a) with decreasing

ratio between the net flux in the seaward zone of the aquifer and the terrestrial

groundwater flux, (b) with increasing ratio between buoyancy forces and the forces

due to ambient groundwater flow, and (c) with decreasing value of the relative width

and the salinity of the surface water body. The portion of groundwater of marine

origin discharging into the surface water body increases under the following

conditions: (a) sea water intrusion is strong, (b) the relative width of the surface water

body is large and (c) the distance of the surface water body from the coast is small.

The flux SWIS, which arises in the salinity transition zone seaward of the surface

water body, is a mechanism for the transfer of pollutants and tracers from the aquifer

into the surface water body, in addition to terrestrial groundwater. The portion of

surface water in SGD, is larger (a) the weaker the sea water intrusion into the aquifer

is and (b) the larger the relative width of the surface water body is.

More systematic investigations concerning the effect of surface water bodies should

consider that in case of near shore lakes or lagoons their length (the scale parallel to

the shore) and their width do not differ so much. For such investigations, three

dimensional models are required.

References

Holzbecher, E., 1998. Modeling Density –driven Flow in Porous Media. Springer,

Berlin.

SGDSFGD

SFGD

SWIS

SWIN

SWI

SWEX msl

RSW RSGDc

International Ground Water Symposium IAHR-GW2006 Abstract proposal

Multi-Scale Characterization of an Heterogeneous Aquifer Through the Integration of

Geological, Geophysical and Flow Data: A Case Study.

B. Bourbiaux, J.P. Callot, F. Gaumet, M. Guiton, R. Lenormand, J.L. Mari, IFP*

*Institut Français du Pétrole, 1-4 Av. de Bois Préau, F-92852 Rueil Malmaison Cedex, [email protected] This paper gives the current status of the integrated modelling study of an experimental

hydrogeological site that has been developed for several years near Poitiers city. The concerned aquifer, 20 to 130 meters in depth, consists of tight karstic carbonates of Middle Jurassic age. It lies on the borderline, named the "Poitou threshold", between the Paris and the Aquitaine sedimentary basins.

Around 30 wells have been drilled on this site, according to a NW-SE-oriented five-spot regular pattern, with a well spacing of 70 meters. Most wells dispose of documented drilling records and logs of various nature, among which gamma-ray, temperature, acoustic. In addition, two wells were entirely cored. Wellbore and surface seismic was also acquired in the vicinity of selected wells.

Regarding flow data, pumping tests were performed on most wells of the site with interference recording in all other wells. Vertical production profiles were also measured along several wellbores. Matrix properties and porous structure were characterised from porosity-permeability measurements on rock fragments/cuttings and from NMR.

Consistent information about the aquifer heterogeneity could be drawn through the cross-checking and confrontation of these static and dynamic observations/measurements at various scales, from seismic to pore scale.

Eventually, a geostatistical model was built and calibrated in terms of flow properties. This model constitutes the first support for the history matching and interpretation of the large-scale pumping and interference tests.

The next step of model calibration to flow data will evaluate different techniques to gradually alter (optimize) the remaining poorly-defined input of the model while respecting its ascertained features.

Main observations/results drawn from various information sources are summarized hereafter. Seismic. The acquisition of usable seismic data is particularly difficult for low-depth reservoirs

underlying a thick weathered zone, such as the aquifer studied herein. Different surface seismic surveys were however attempted with different acquisition schemes. It turned out that reservoir internal markers could not be identified by reflection because of an insufficient frequency band. However, refraction signals enable to get an image of the top surface of the aquifer. This is useful in the present case as the aquifer top surface is strongly disturbed as a result of the presence of caves of karstic origin filled in by clayey surface erosion material. Offset/Vertical Seismic Profiles were also acquired in the vicinity of one well in order to identify major reservoir markers. Weathered zone obstacle was overcome thanks to buried sources. This way, aquifer bottom limit was identified as well as two reservoir markers, which were confirmed as two major drains of heterogeneous rock structure by production and acoustic logs. Sedimentary information from drilling, logs and cores. Reservoir facies, diagenetic fingerprints and fractures were characterized thanks to a thorough analysis of the cores from two wells. Several markers were identified on the basis of macro/microfaunes and granulometry or mineralogy changes. For all other non-cored wells, drilling reports and gamma-ray logs were used to detect and correlate facies changes. Particular attention was given the detection/location of fractured/dolomitic/vuggy limestone beds and to the 3D occurrence of cherty limestones, as those sedimentary and diagenetic features revealed themselves to be in relation with the distribution of major aquifer drains in space. Aquifer heterogeneity results from diagenetic fingerprints on bioclastic limestones of various textures. These fingerprints include limestone dissolution, silice dissolution/precipitation and clay filling and form a complex network comprising two apparently-

disconnected horizontal highly-conductive drains, and vertical less-conductive bodies. Diagenetic phenomena seem to have occurred along a preferential NW-SE direction corresponding to the main direction of fracturing in the site region.

Structural features – Fracturing. The structural setting consists in a low-depth platform located at the threshold of two sedimentary basins, and overlying a basement subjected to successive faulting episodes in relation with remote orogenesis. Fractures affecting the aquifer formation were characterized on analogue outcrops located near the hydrogeological site, and from the observation of available cores from two wells. Outcrops enabled to identify fractures along three directions, among which the main NW-SE one associated with the regional tectonic history. Fractures are found as sub-vertical swarms with a high vertical extension rather than bed-controlled diffuse fractures. Fairly few fractures were observed on the cores from the two vertical wells. However, an unexpected fracture density contrast was found between those two wells, though very close from one another. This contrast was also observed regarding the occurrence of cherts. Fractures do not seem to constitute the main direct origin of flow heterogeneity, but probably controlled the development of diagenetic features strongly impacting flows.

Flow data. Flow properties were characterized at various scales. Each well was pumped while measuring interference in all other wells. Such field-scale measurements constitute a very rich data base which interpretation and modelling is still under way. As a very preliminary analysis, we considered the aquifer as homogeneous and estimated an average aquifer permeability from each well pumping results. Large productivity/permeability differences, in a ratio exceeding 10, are observed between wells. Production logging was also carried in a few wells. The latter revealed the presence of two dominating producing layers located around -85 and -55m, and a minor contribution to flow from the remaining reservoir formation. These two producing levels are related to diagenetic facies, i.e. fractured, moldic and/or cherty limestones. Petrophysical measurements on cores fragments or cuttings of the reservoir matrix revealed substantial porosities, 10-25%, but very low permeabilities, less than 0.1md, due to a complex multi-porosity structure as was revealed by NMR measurements.

Data integration into a geostatistical model – Preliminary interpretation of aquifer behaviour. A preliminary geostatistical model of the hydrogeological site was built on the basis of geological markers correlated between wells, and of vertical and horizontal distribution parameters of "lithotypes". Lithotypes reflect the heterogeneous nature of the aquifer but also its flow properties, in particular the location of preferential flow paths identified from drilling records and production logs. The 3D distribution of these "lithotypes" was simulated. 3D views and cross-sections enable to capture the geometry and connectivity of the main aquifer drains, that is two separated horizontal drains associated with a complex vertical network made up of diagenetic features such as cherts and argillaceous karst-filling.

Perspectives. The above-described geostatistical model will be enriched from further measurements, especially logs and flow-metering. Advanced procedures to calibrate this model will be applied in order to history match flow data – especially the whole interference data base. Such calibration may concern either the flow properties directly, or the geological features/objects responsible for preferential flow. In addition to these productivity aspects, storage capabilities of the aquifer should be investigated through tracer tests, completed by further characterization of matrix properties, and simulation on the previous flow-calibrated model.

Main paper contributions:

- A methodology of multi-scale static/dynamic data analysis and confrontation to capture and understand the 3D distribution of flows within fluid-bearing karstic carbonate reservoirs.

- Application to a hydrogeological site: data integration into a 3D geostatistical model. - Guidelines for the interpretation and prediction of the flow behaviour of heterogenous

aquifers.

Pathline-Calibrated Groundwater Flow Models of Nile Valley Aquifers, Esna,

Upper Egypt Tom H. Brikowski1, Abdallah Faid2

Abstract:

Strongly concentrated agriculture along the River Nile in Egypt, combined with hydrologic changes related to the construction of the Aswan High Dam in the 1970's, has led to increasing salinization and waterlogging of agricultural areas. Successful control and remediation of these problems requires accurate understanding of the shallow Quaternary aquifers within the Nile Valley. While extensive conceptual models have been developed by the Egyptian RIGW, published numerical models have yet to incorporate all features of the conceptual model. In particular, marine affinity of some shallow groundwaters within the valley (Cl- as the predominant anion) indicates significant leakage from deeper Cretaceous aquifers into the shallow Quaternary aquifers, a feature that is not present in current models.

In this study, groundwater profile modeling incorporating the bedrock leakage demonstrates that its shallow appearance requires hydraulic separation of surficial from deep-recharged zones of the Quaternary aquifer. This separation occurs near the boundary between reclaimed and traditional agricultural lands, which is also the primary site of waterlogging. Apparently, excessive recharge presumed to occur beneath the reclaimed lands does not penetrate deeply, and therefore might be easily remediated with shallow drains. Profound similarities exist between the Nile Valley salinization cases and the occurrence of shallow “nuisance water” in desert southwestern U.S. cities (e.g. Las Vegas). The U.S. experience with this problem may provide useful guidance in addressing Nile Valley salinization and waterlogging issues in the future. In general, irrigation-related recharge from the reclaimed lands in the Nile Valley may have a much more localized impact on traditional lands than previously thought.

Keywords: Groundwater modeling; Nile Valley; salinization; waterlogging

1 Corresponding Author: Geosciences Department FO-21, University of Texas at Dallas, P.O. Box 830688, Richardson, TX 75083-0688, [email protected] 2 National Authority for Remote Sensing and Space Sciences, El-Nozha El-Gedida, Cairo, Egypt e-mail:[email protected]

IAHR-GW2006 Keynote Lecture (E.A.Sudicky, J.M.Lemieux et al.)

Simulating Complex Flow and Contaminant Transport Dynamics in an Integrated Surface-subsurface Modelling Framework

E.A. Sudicky, J.M.Lemieux,

J.P. Jones, A.E. Brookfield, D. Colautti and Y.-J. Park Department of Earth Sciences, University of Waterloo

Waterloo, Ontario, Canada N2L 3G1

R. Therrien, and T. Graf Département de Géologie et de Génie Géologique, Université Laval

Québec, Canada G1K 7P4

Over the past several years, increasing attention has been directed towards understanding flow and contaminant transport exchange processes occurring at the interface between surface water and groundwater, particularly in the vicinity of riparian zones in riverine valleys and within the hyporheic zone. In this paper, we will examine these processes in the context of the HydroGeoSphere model, a recently-developed surface/subsurface control-volume finite element model. HydroGeoSphere is a fully-coupled 3D model that can simulate water flow and advective-dispersive solute transport over the 2D land surface and in the 3D subsurface under variably-saturated conditions. Full coupling of the surface and subsurface flow regimes is accomplished implicitly by simultaneously solving one system of non-linear discrete equations describing flow and transport in both flow regimes, as well as the water and solute exchange fluxes between continua. A number of applications of the model to catchments of various scales, ranging from the scale of an intensively-monitored rainfall-runoff-tracer experiment (~ 2000 m2) up to a regional-scale watershed of about 8000 km2, has illustrated the complexity of watershed dynamics. The model capabilities and its main features will be demonstrated here with several high-resolution 3D numerical simulations to examine the impact of an upland surficial contaminant source that forms a subsurface plume that migrates vertically through the unsaturated zone and is then transported laterally below the water table where it eventually discharges along a reach of a small stream. Results show that short-duration, high-intensity water and solute exchange fluxes across the streambed interface, along with sharp concentration peaks in the stream water, can arise during individual precipitation events. The hydraulic head and concentration variations induced by short-duration rainfall variations show a muted response with increasing depth below the streambed due to the natural smoothing in the hydraulic response because of subsurface storage effects and because of dispersion and diffusion of the solute. It is also demonstrated that the solute exchange fluxes can vary markedly in the streambed, both spatially and temporally, depending on the intensity of the groundwater discharge/recharge patterns along and across the streambed. The variability and sensitivity of these near-stream processes to the temporal resolution of rainfall input and the heterogeneity of the sediments may be significant for the prediction of health risks to aquatic habitats. The simulations stress the advantage of using a process-based model such as HydroGeoSphere for the prediction of the impacts of alternative water and land use management scenarios.

Quantifying groundwater model prediction uncertainties at a small scale highly heterogeneous remediation site Heinz J. Theis German Federal Institute of Hydrology Am Mainzer Tor 1 56068 Koblenz Numerical groundwater modelling is a widely accepted tool for the evaluation of remediation measures in contaminated groundwater aquifers. Quantifying the inherent uncertainties of the model results can help to optimise the design of these measures and avoid facility oversizing due to safety supplements. In our investigations we tried to identify an appropriate method with respect to the scale of the remediation site. Furthermore the method should be suitable for practical applications, that means the incorporation of complex boundary conditions, the integration of expert knowledge and other soft data. We developed the SUFIX method which combines stochastic simulation methods with numerical models and allows to take into account the propagation of uncertainties. Starting with data preparation leading to a geostatistical analysis and the stochastic simulation, the method eventually comes to the evaluation of the model results. A Bayesian sequential updating approach allows the inverse optimisation of input data corresponding to the different stages of the modelling process. For stochastic simulation purposes different methods were used, indicator-based and Gaussian based approaches. The application of SUFIX on two practical remediation cases showed clearly the interconnection between the stochastic simulation method and the scale as well as the grade of the subsurface heterogeneity. Among the most promising methods found for simulating the subsurface heterogeneity on a small scale were the Truncated Gauss method, but an approach based on a manually conducted zonation surprisingly showed comparable results. To investigate the scale dependence of these results, a second investigation was accomplished on a larger scale. The related results showed that the ranking of methods can be substantially different to the first case (Table 1 and Table 2). Besides the choice of the suitable imaging method for generating the subsurface heterogeneity, it appeared, that another basic condition must be satisfied: the design of a sound and consistent conceptual hydrogeologic model stands above every other effort to optimise the quality of a numerical groundwater model. For the design of a conceptual model the modeller should have a good knowledge of structures with a large scale influence, which are only obtainable by interpretation of e.g. layering, buried paleochannels etc.. The testing on two field sites proved that SUFIX complies with requirements for a method that should be applicable in practice.

Table 1: Ranking of imaging methods for subsurface heterogeneity (Small scale – highly heterogeneous)

Table 2: Ranking of imaging methods for subsurface heterogeneity (Large scale - homogeneous)

Hydraulically controlled combined vertical circulation of groundwater and alcohol Ulf Mohrlok, Klaas Heinrich Institute for Hydromechanics, University Karlsruhe, Kaiserstr. 12, D-76128 Karlsruhe, Germany, Tel.: +49 721 608 6517, e-mail: [email protected] Groundwater remediation by co-solvent flushing becomes more and more important, since co-solvents increase the solubility and mobility of NAPLs tremendously. Contamination sources are removed very quickly and efficiently by application of co-solvents. On the other hand the application of co-solvents is a hydraulic challenge since most of them are only poorly miscible with the groundwater or at least the mixtures have significant different physical properties than groundwater. This means that the application of co-solvents changes the flow resistance conditions, i.e. the hydraulic conductivity, generates buoyancy effects and has to deal with multi-phase flow features in case of immiscible fluids. In close cooperation with the Institute of Hydraulic Engineering and Research, University Stuttgart, a remediation technique has been developed injecting an alcohol cocktail miscible with groundwater within a vertical circulation flow field established by a groundwater circulation well (GCW) (Mohrlok et al., 2005). This technology is based on the partitioning of the alcohol cocktail into the DNAPL contamination and increasing the solubility and mobility while reducing the density. From the hydraulic perspective and for reduction of applied alcohols a technique has been developed for alcohol injection only into a part of the circulation around a GCW (Fig. 1). This circulation system is able to control hydraulically the combined circulation of groundwater and alcohol cocktail.

123456789101112131415

Q b,i

Q b,c

Q b,o

ABCDEFGHJLM

KQ t,i

Q t,c

Q t,osampling point

PCE

3.17 m

1.26

m

alcohol

water

water

Figure 1: Two-dimensional experimental set-up: injection (bottom) and extraction (top) unit with three chambers each defining an inner, central and outer circulation. The basic principle of that hydraulic control is to adapt the boundary conditions with respect to the different viscosities and densities of groundwater and alcohol cocktail in order to get a well defined circulation flow field similar to a pure groundwater circulation. Basically, the established pressure conditions has to compensate the influences of those parameter differences. Experiments on groundwater remediation of a PCE contamination source in a

two-dimensional set-up demonstrated successfully the applicability of such a hydraulic control (Fig. 1) and a very fast and efficient remediation of the PCE source. References Mohrlok, U., Heinrich, K., Greiner, Ph., Braun, J., Schnieders, J., Koschitzky, H.-P. (2005): Alcohol flushing in laboratory experiments: In-situ remediation of DNAPL contaminated groundwater. Proceedings of ConSoil 2005, 3-7 October 2005, Bordeaux, France.

Homogeneization of smaller scale heterogeneity at Äspö (Sweden) granitic site witin a model for transfers of radionuclides at large temporal scales. Christophe Grenier, Christian Laguerre, Gilles Bernard-Michel. Laboratoire des Sciences du Climat et de l’Environnement, Unité mixte CEA/CNRS. Orme des Merisiers, 91191 Gif sur Yvette Cedex. France. Corresponding author Christophe Grenier: [email protected] We present results obtained within the Task 6 modeling project hosted by the Swedish Nuclear Fuel and Waste Management Company (SKB) Task Force on Numerical Modeling of Flow and Solute Transport in fractured granitic rock. Modeling the transport of radionuclides in natural fractured media is done to characterize and assess the performance of a potential deep geologic repository. The present task objective is to provide a bridge between models based on detailed site investigation data and calibrated against tracer experiments (month scale), and models corresponding to a hypothetical repository postclosure time scale (hundred thousands of years). Postclosure models capture the most significant features and processes of radionuclide transport, and allow for a sensitivity analysis of uncertain parameters. A 200 m semi-synthetic fractured block was built based on Äspö granitic site in situ measurements and serves as a test case for modellers. It encloses 5600 fractures (from 200 m scale to meter scale) as well as information about the complexity associated with fractures. These are basically faulted zones (several flow channels along mylonitic structures and providing infilling materials) or simpler non faulted zones (roughly a single channel with fracture coating and relatively homogeneous matrix properties). The objective is to build a predictive model for large time scales (hundred thousands of years) where the importance of diffusion into matrix zones is strong and leads to major retention effects. We work within our Cast3M code for a Mixed Hybrid Finite Element scheme. We present our approach and show that all units of the system can be included in a model where major fractures are explicitly represented and minor structures and smaller scale heterogeneity are homogenized (fracture complexity and back ground fracturation). Transport of a series of tracers is modelled within an Eulerian approach with only slight changes of the equations as compared with classical advection dispersion diffusion terms. All units of the system are included within a single formalism: complexity of fractures is taken into account in a semi analytical manner relying upon solution of orthogonal 1D diffusion in a heterogeneous medium; back ground fracturing is homogenized considering results obtained from homogenization of regular geometry (sugar box like) considering semi analytical expressions relying on 2D and 3D diffusion into regular matrix blocks. The impact of such smaller scale units on transfer properties of the global block is studied for a range of parameter values from in situ measurements. The flow conditions are uniform and low in intensity, typical of post closure time scales. We show the diverse impacts of matrix diffusion on breakthrough curves exiting the system. The more theoretical question of the impact of a second level of fracturation in the sugar box geometry is studied in the framework of triple porosity: major fracturation, secondary imbedded fracturation, matrix blocks. Results show that smaller scale fracturation leads to larger retardation of the plume but different shapes of breakthrough curves, in particular lower tailing due to smaller penetration depths. The regimes of the system and impact in the global model are studied.

!" #

$$ %& $ #

$ '() $ $ # *+$

% & , * - & &$

$ , $

-+ .& + &

'/)

% $ " $

$ 0 ,

$ $ $ '1)2

$

$ "3

+ '4) $

$ '5)% $

$

3 $ $ + -

$

6" $ $ " '7) %

$ & β"

α" $ $ - 8." # &

$ & β9:α9;" ! - 8. #

β :-

, $ + $ %

%<=+ %<="';)2

8.

α ; $ &

."- ,$6+ $ ':)

% $ $

"$

$ 1:4 > , :?

@ -

$ $ 6 $

& 1? !

$ #

$ 1% $ " $ % *+$')%

$ #$ :

,! , *+$

!"#

$$ $ !*+$" $

$ $

$ $ .*

&$ %<=

$

*+$

<A

':) 3.= 0= * **% $ & >$

= <<(( ;":1::1;';) 3+ #3B C* #<08 A$

+ $ $ = <<7: ;;"') 0%8A% $;:D$ .$:///'1) .! 0E3:/7/'4) . ! > 0 0 $ :

%= <<5 ;:"1(415;'() !##CC=F% !G

$= <;:: ://7"15H(:'5) <!+0**

= <<:: ://4";7;1'7) %*!<8 *>.I

A ! = < < / ;"A::;/@;:=<:::

'/) J K 3 . $ & A + $G8 $0;4

STOCHASTIC MODELING OF UNSATURATED FLOW AND TRANSPORT :

NUMERICAL EXPERIMENTS, MACROSCOPIC BEHAVIOR AND UPSCALING.

R.ABABOU (1), VEENA S.SORAGANVI (2), M.S. MOHAN S.KUMAR (2) (1) Institut de Mécanique des Fluides de Toulouse, France

(2) Indian Institute of Science, Bangalore, India

PRELIMINARY ABSTRACT

Effects of spatial heterogeneity and upscaling methods on hydrodynamic transport coupled with geochemical reactions

Marco De Lucia

DICMA, Università di Bologna, Italy Ecole des Mines de Paris, Centre de Géostatistique, France

Chantal de Fouquet

Ecole des Mines de Paris, Centre de Géostatistique, France

Vincent Lagneau Ecole des Mines de Paris, Centre d'Informatique Géologique, France

Geostatistics provides a wide set of tools to deal with uncertainty and spatial heterogeneity when forecasting the behaviour of reactive groundwater transport in absence of abundant or precise field data. In a Monte-Carlo approach, geostatistical simulations of uncertain geological parameters such as log-permeability and porosity are drawn constrained to available observations, and given to a numerical hydrodynamical model which solves the reactive flow equations over a discretized domain to obtain confidence intervals for calculated concentrations. But: the geostatistical simulations are defined over a fixed (small) support, while the flow calculation, much more cpu-demanding, is generally run over less dense grids of arbitrary shape, therefore requesting an upscaling of non additive quantities such as permability; furthermore, the upscaling method strictly depends on the numerical scheme used to solve the transport problem and on the flow regime itself. Another point is the equivalence criterion: which property must the equivalent upscaled porous mean mantain with respect to the original one? The need to answer these questions arised while considering geochemical modelling with the help of Hytec, a 2D reactive flow model with a coupled geochemical engine, developed at the CIG to study groundwater pollution, geochemistry and safety assessment of waste disposals. Hytec, or better its transport module R2D2, uses a finite volumes scheme, discretizing the domain in a Voronoi tessellation composed of irregular polygons. In a preliminary step, we compared different fast upscaling techniques (e.g. simplified renormalisation, Matheron-proposed power mean) over a regular square grid and in a permeameter configuration to put in evidence the influence of variability and upscaling techniques over the transport, and to identify the equivalence criterion to be adopted in view of the geochemical core problem. The porous medium is supposed log-normal. We found out that discrepancies in tracer distributions and arrival times calculated by Hytec are highly affected by the ratio between the mesh dimension and the correlation length of the simulation, and that the upscaling effect becomes rapidly significant if this ratio increases. Tests upon irregular mesh are also run to try a new approximated techniques based on simplified renormalisation, and generally to identify the most satisfactory upscaled permeability for a normal Hytec application, at least for stationary flow conditions. These results provide effective directives for modelling an Uranium production field located in an acquifer whose behavior has to be forecasted both during the production life (complex flow regime, mainly induced by the injection/extraction wells) and in the post-exploitation phase. Agreat number of simulations for the production were run to identify confidence intervals for concentrations in the extraction wells according to the expected geochemical environment and variability of the porous media.

Dissolution and precipitation processes in porous media: a pore scale model

C.J. van Duijn1, V.M. Devigne2, T.L. van Noorden1, I.S. Pop1,1 Dept. of Math. and Comp. Sci., TU Eindhoven, P.O. Box 513, 5600 MB Eindhoven, The Netherlands

2 Centre SITE, ENS des Mines, 158 Cours Fauriel 42100 St-Etienne, FRANCEe-mail: C.J.v.Duijn, T.l.v.Noorden, [email protected], [email protected]

ABSTRACT

We discuss a pore scale model for precipitation and dissolution processes in a porous medium. Thevoid region is occupied by a fluid in which cations and anions are dissolved. Under certain conditions,these ions can precipitate and form a crystalline solid, which is attached to the surface of the grains (theporous skeleton) and thus is immobile. The reverse reaction of dissolution is also possible.This model is considered in [1] and represents the pore–scale analogue of the one proposed in [2]. Itconsists of several components: the Stokes flow and the transport of the dissolved ions by convectionand diffusion, which are processes in the void space of the medium, and the dissolution/precipitationreactions occuring on the surface of the grains (the porous skeleton).The particularity of the model is in the description of the dissolution and precipitation processes in-volving a multi–valued dissolution rate. In mathematical terms, the model is a system of partial andordinary differential equations. The part related to the chemistry is a coupled system of convection–diffusion equations for the concentration of the ions, and of an ordinary differential inclusion definedonly on the grain boundary and describing the concentration of the precipitate.In this talk we assume that the flow geometry as well as the fluid properties are not affected by thechemical processes. Moreover, the Peclet and Damkohler numbers are assumed to be of moderate order,so that all the time scales involved are of the same order of magnitude. First we discuss some qualitativeproperties of the model for general domains, and then consider simpler geometries. In particular, if thevoid space is a strip, with dissolution and precipitation occurring at the lateral boundaries, we investigatethe formation of a dissolution front. This front is moving in the flow direction and separates the regionwhere the precipitate is present from the one where no crystals are encountered.As a first step for a rigorous justification of the macroscopic model we let the ratio between the thicknessand the length of the strip vanish. In the limit we end up with the upscaled transport–reaction modelproposed in [2].We conclude our talk with numerical experiments sustaining the theoretical results. Some details con-cerning the numerical algorithm will also be presented.

REFERENCES

[1] C.J. van Duijn, I.S. Pop, “Crystal dissolution and precipitation in porous media: pore scaleanalysis”, J. Reine Angew. Math., Vol. 577, pp. 171–211, (2004).

[2] P. Knabner, C. J. van Duijn, S. Hengst, “An analysis of crystal dissolution fronts in flowsthrough porous media. Part 1: Compatible boundary conditions”, Adv. Water Res., Vol. 18,pp. 171–185, (1995).

Abstract for IAHR-GW2006 conference (M-P. Lam et al.)

1/2

Asynchronous particle tracking for contaminant migration in heterogeneous 3D media with unstructured tetrahedral mesh

Minh-Phuong LAM(1,2), Regina NEBAUER(1), Rachid ABABOU(2) (1) EDF/LNHE - Laboratoire National d’Hydraulique & Environnement, 78401 Chatou, France

(2) IMFT - Institut de Mécanique des Fluides de Toulouse, 31400 Toulouse, France Contaminant transport in porous media is usually modeled macroscopically by solving an advection-dispersion equation in an eulerian framework, e.g. finite volumes or finite elements on fixed grids:

( ) ( ) 0)( =∇•∇−•∇+∂

∂ CCtC DU θθθ

3IR⊂Ω∈x (Eq.1)

where C(x,t) is concentration (kg of solute/m3 of solvent), U(x,t) is water velocity vector (m/s), θ(x) is porosity or water content (m3/m3), and D(x,t) is the local hydrodynamic dispersion tensor (m2/s) which may be reduced to isotropic diffusion for testing purposes. The velocity field is assumed given from a flow calculation based on Darcy’s law (θU = -K(x) ∇ H) and mass conservation (e.g., div(θU) = 0). It can be highly variable in space (and time). Solutions of the eulerian transport PDE are difficult to obtain (for several reasons). In contrast, lagrangian particle tracking (LPT) methods are not affected by numerical dispersion and do not require algebraic solvers. The solute plume is discretized into concentration packets (particles). The model tracks each particle based on the following time-explicit displacement algorithm (advection step plus random walk step for diffusion) :

( )( )nnn tttttt ZDDDUXX •∆+∆∇•+•∇++=∆+ 2)ln()()( θ (Eq.2), where X(t) is the particle position vector (x(t),y(t),z(t)) at time t, Zn is a purely random vector sequence, and √D stands for the square-root of the dispersion matrix (Dij), which can be defined in various ways. Note that (U,D,θ) are all space-dependent. In spite of the simplicity of Eq.2, there still remain several delicate choices to be made in the implementation of the method. The difficulties and the techniques to overcome them are discussed in this paper, keeping in mind the main objective, i.e., matrix-fracture systems. Different particle tracking and interpolation/projection techniques are being implemented in the ESTEL3D code (EDF/LNHE), based on darcian velocity fields obtained from a finite element method with unstructured tetrahedral mesh. The method chosen here to implement both steps of particle pushing is asynchronous in time : each particle is pushed through the discrete tetrahedral mesh during a variable time step that is not known in advance, and the different particles have different time stepping “schedules”. In contrast, synchronous time stepping was used in otherlagrangian particle schemes (e.g. Fadili et al., Math. Geol., 1999). Some features of our approach are outlined below.

1. Interfaces and discontinuities. At interfaces between blocks, e.g. due to stratification, or fracturing and faulting, both D and θ are discontinuous. Thus, strictly speaking, div(D) and grad(lnθ) will diverge in eq.(2). There are several methods to overcome this difficulty, such as particle reflection techniques, but they are not easily applicable for complex 3D geometries. Smoothing and interpolation techniques were tested with some success for high contrast, irregular matrix-joint systems in 2D (e.g. Spiller et al, IAMG Proc., 2002). The need for non-synchronous time-stepping was highlighted. Here, the smoothing method is adapted and tested further in 3D, using asynchronous time stepping, on unstructured mesh imposed from the hydraulic problem.

2. Particle tracking across interfaces using smoothing techniques Here, we smooth interfacial discontinuities by replacing D and θ in a cell by their volume-weighted mean values, calculated over the cell itself and its four neighbours. This spatial smoothing, performed over the entire domain, does not affect homogeneous regions. Each smoothing operation is equivalent to applying a spatial filter. The number of filters to be applied becomes the control parameter for smoothing the interfaces. The transport parameters (U,θ,D) are P0-discretized, i.e., piecewise constant per element. The LPT algorithm estimates a particle displacement with (U,θ,D) evaluated at the starting point of the particle. The procedure to detect “overshoot” and stop the particle when it reaches a cell face, is explained in figure 1 (1 particle & 3 tetrahedra).

3. Local versus global mesh element coordinate systems. A local element coordinate system is used to improve particle displacement accuracy and CPU time.

4. Asynchronous time stepping.

Abstract for IAHR-GW2006 conference (M-P. Lam et al.)

2/2

Figure 1 : Schematic showing particle trajectory with overshoot treatment (shown in plane view).

We use an “asynchronous” time step for particle displacements. Each particle “P” moves independently of the others along its own pathway, while sampling the mesh-based velocity field and other material properties along the way (cf. fig. 1). A global output time step ∆tSYNC is needed for synchronizing the outputs of particle positions, moments, concentrations, etc. The asynchronous particle time-steps ∆tP are constrained by the relation:

( )∑+

=∆=∆1

1

PN

nPSYNC ntt (Eq.3)

where NP is the number of cells entirely crossed by particle “P”.

5. Post-processing : interpolating particles to concentrations. Once the particle trajectories are computed, the ESTEL3D code returns all particle coordinates at each time tSYNC. One of the methods for obtaining the concentration field C(x,t) from particle positions XP(t) is as follows. The domain is partitioned into sub-domains, such as hexahedral boxes, or spherical shells (for isotropic diffusion), independent of the mesh used for transport parameters. At selected times tn=tSYNC(n), the number of particles residing in each sub-domain is counted, normalized by the total number of particles in the domain, and weighed by particle masses. This yields the estimated field C*(x,tn) distributed over discrete sub-domains.

6. Example test : pure isotropic diffusion in 3D space There is no velocity in the diffusion problem presented here, however, we still use the unstructured tetrahedral mesh in order to test the asynchronous scheme. In figure 2, we compare the concentration field C(r,tn) obtained from our model with the analytical solution for pure diffusion in an infinite domain in IR3. The simulations are performed with 10 000 particles injected in the middle of the homogeneous medium at t=0. The concentration field was obtained by counting particles on spherical shells of thickness ∆rC. The optimal value of ∆rC depends on the average size of the tetrahedral elements. The resulting concentration is plotted as a function of radial distance from the origin (r2=x2+y2+z2) after 10, 100, 1000 and 10 000 years. Clearly, the particle diffusion model gives satisfactory results, quite close to the analytical gaussian plume. The corresponding particle clouds are also analyzed via spatial moments, and visualized in 3D (not shown here for lack of space).

Figure 2: Comparison of computed & analytical concentration fields for 3D isotropic diffusion : ESTEL3D particle model (wavy color curves) vs. analytical (smooth black curves) at t = 10, 100, 1000, 10000 years.

222 , , θDU111 , , θDU

333 ,D , θU

1

Effective dispersion in temporally fluctuating flow through a chemically heterogeneous medium.

Vanessa Zavala-Sánchez, Marco Dentz and Xavier Sánchez-Vila

Department of Geotechnical Engineering and Geosciences, Technical University of Catalonia (UPC), Barcelona, Spain

In this work we investigate the effective transport of a reactive solute through a chemically heterogeneous medium. We focus on spatially fluctuating equilibrium sorption properties, which here are characterized by a random retardation factor. The medium is physically homogeneous. The resulting spatially uniform flow is fluctuating in time. Groundwater flow in aquifers, in fact, fluctuates on a range of scales including hyper annual climatic fluctuations, seasonal and irrigation cycles, daily barometric fluctuations, for example. This kind of simplified model might be appropriate to describe the transport of an organic solute in an aquifer, which is comparatively homogeneous with respect to the hydraulic conductivity, but exhibits a strongly varying organic carbon content which determines the retardation factor. A similar problem was formulated by Attinger et al. (1999), who investigated the temporal behaviour of a solute cloud in a chemically heterogeneous porous medium under steady state flow conditions. The full perturbation theory result shows that the heterogeneities of the medium change, in a quantitatively relevant way, the behaviour of the longitudinal components of the dispersion tensor corresponding to the direction of advective transport, whereas the corresponding transverse parts are only weakly influenced by the stochastic fluctuations of the retardation factor. In the long time limit, the transverse dispersion coefficient reduces to the initial pore-scale dispersion coefficient; all deviations from this value for shorter times are also of the order of this pore-scale dispersion coefficient. Here we show that temporal fluctuations together with spatial heterogeneity and local dispersion cause macroscopic (i.e., independent of local dispersion) transverse spreading. Following the general perturbation approach developed by Attinger, we study the symbiotic impact of chemical medium heterogeneities and the fluctuations of the flow field on the effective large scale transport. In a stochastic modelling framework the fluctuating retardation factor is modelled as a spatial random field. Temporal fluctuations of the flow velocity are represented as a temporal random process. To quantify the influence of space fluctuations of the retardation factor and time fluctuations of the flow velocity on the effective transport behaviour, we study the effective centre of mass velocity and effective dispersion coefficients. In a stochastic model these observables are defined as averages over all typical realizations of the stochastic processes under consideration. We define the effective dispersion coefficient in terms of the time and space ensemble average of the time derivative of the second centred moments of the (properly normalised) non-adsorbed solute concentration in one typical realization of the two random processes:

(2) (1) (1)1( ) ( ) ( ) ( )2

effij ij i j

dD t m t m t m tdt

= − (1)

This effective quantity characterises physical spreading in a typical disorder realization as opposed to the frequently considered ensemble dispersion coefficient, which is derived from the ensemble averaged concentration distribution and quantifies also the artificial spreading due to centre of mass fluctuations. Using a second order perturbation approach in the fluctuations of the random fields, ( )eff

ijD t is given by the sum of the contributions due to local dispersion, the interaction of local dispersion and spatial heterogeneity (Attinger et al., 1999), and a contribution due to the interaction of local dispersion, spatial fluctuations of the retardation factor and temporal velocity fluctuations,

( ) ( )( ) ( ) ( )eff s eff t effijD t D D t D tδ δ= + + (2)

2

The resulting perturbation theory expression for the contributions of spatial fluctuations of the retardation factor and temporal velocity fluctuations, can be evaluated explicitly in the limit of large Peclet numbers, which yields,

2

( ) 2

22

2

22

1 1

2 4 411 1

t RR

uu u

D D

effDt tR

σ πδ σ τ κ

κκ

τ τ

= −+

+ + +

⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦

(3)

Where we defined the advection time scale 1 /u l uτ = , which measures the time for advective transport of the solute over one correlation length of the medium. Furthermore, we defined the dispersion time scale 2 /D l Dτ = , which characterises the time for dispersive solute transport over

one correlation length l . The non-dimensional Kubo number / uκ τ τ= compares the correlation time

τ to the advection time scale uτ . In order to derive explicit results, we specified a spatial and a

temporal correlation functions, both as a Gauss-shaped correlation functions, where 2RRσ denotes the

spatial variance and R the mean value of the retardation factor, and 2uuσ the variance of time

fluctuations. Figure 1 illustrates the time evolution of the contributions to the transverse dispersion coefficient

( ) ( )t effD tδ in d=2, for an inverse Peclet number 310−∈= . Furthermore, it is shown the temporal

behaviour of the contributions to the longitudinal ( ) ( )s effLD tδ and transverse ( ) ( )s eff

TD tδ effective dispersion coefficients under steady state conditions. The behaviour of the transverse effective dispersion coefficient in a time-dependent flow field is qualitatively and quantitatively different from the one observed in steady flow. The stochastic perturbative analysis of transport in a steady flow field yields transverse asymptotic dispersion coefficient of the order of the local dispersion (i.e. microscopic). In a temporally fluctuating flow field, however, the effective transverse quantity evolves to a macroscopic asymptotic value which is of the same order of magnitude as the contribution to the longitudinal dispersion coefficient developed under steady flow conditions. Note that the asymptotic

long-time value for ( ) ( )t effD tδ in d=2 is approached according to ( ) 1/ Dt τ −, whereas ( ) ( )s eff

LD tδ

approaches its final asymptotic value as ( ) 1/ 2/ Dt τ −.

FIGURE 1. Effective dispersion coefficients under steady state flow (Attinger et al,1999), and transverse effective dispersion

coefficient under transient flow conditions, in d=2 as a function of time for a point –like initial condition; (with 3

/ 10u D

τ τ−

∈= = ,

22 2/ 0.1RR uu u

R σ τσ = , 1κ = ).

Rel

ativ

e di

sper

sion

coe

ffici

ent

D(t)

/D

( )s effLD Dδ+

( )t effTD Dδ+

/ ut τ

( )s effT t

D Dδ→∞

+

RANDOM AND MIXED RANDOM WALK/FINITE VOLUME METHODS FOR SOLUTE AND HEAT TRANSPORT IN

HETEROGENEOUS MEDIA

Debenest Gerald *, Thovert Jean-François 1

* Institut de Mécanique des Fluides, 31400 Toulouse France 1 Laboratoire de Combustion et de Détonique, 86361 Chasseneuil France.

[email protected] Abstract: This paper deals with the problem of transport phenomena in medium where high contrasts in terms of capacities, diffusivities or porosity. Several approaches were made in the last decades to propose a way to tackle this problem. We can quote the work made by Uffink (1) who was the first to develop a theorical way to allow Random Walk to be efficient in the case of discontinuous diffusivity fields. More recently, Labolle et.al (1), deals with the transport phenomena when both diffusivity and porosity can vary in space. In terms of transport, he solved the problem given in the next system of equations:

where Ji is the diffusive flux, Di the effective diffusion value in phase i, c the concentration and εi the porosity. He has achieved this for high contrast in diffusivity and little one in terms of porosities. That’s the aim of this paper, to show how we can also use a RW algorithm to deal with this problem. Random walk is a common approach in transport phenomena. We have recently developed a 3D microscale numerical tool (3) for smouldering process in packed beds of solid fuel. It incorporates several mechanisms such as:

1. Transport of chemical reactive and inert species by convection and diffusion in the fluid phase

2. Reaction on the heterogeneous surface of the sphere depending on the local conditions

3. Heat generation and its transport by convection diffusion in the fluid phase and by conduction in the solid phase.

The extreme contrast of capacity existing between solid and fluid phase in thermal transport (up to 1000) makes the Labolle method not efficient to treat that kind of problem. That’s why we devised a new method to do this. At first, we will show the standard RW algorithm used to treat this kind of transport in 1D medium with both diffusivities and capacities discontinuous. One example is given in figure 1. In a second part we will show how we can mix two numerical methods in order to circumvent overwhelming computational requirements by applying RW only in a part on the

0. =∇+∂∂

∇−=

Jitc

ciDJi

iε

domain and coupling it with a finite volume scheme in the other part. One example is given in figure 2 with the same kind of contrast than this used for 1D case. This second case is a stationary situation when a unit drop of temperature is imposed upon a square unit sample which contains a gaseous inclusion. On the left, the result (fig 2.a) is given for the coupling method (RW/ Finite Volume) and on the right (fig 2.b) the standard deviation from the solution of a full VF method.

Figure 2a: steady heat flow in a two dimensional medium with a gaseous inclusion in the middle

Figure 2b: standard deviation between coupling method and full VF calculation.

References: (1) Uffink, G. J. M, 1985, A random walk for the simulation of macrodispersion in s stratified aquifer, in: Relation of groundwater Quality and Quantity, IAHS Publ. 146, Int. Assoc of Hydro. Sci., Gentbrugge, Belgium, pp. 103-114; (2) Labolle EM., J Quastel and G.E. Fogg, 1998, Water Resours. Res., 34, 1685-1693. (3) Debenest G, Mourzenko V.V.M, Thovert J-F, Smouldering in fixed bed of oil shale grains. A three dimensional microscale numerical model, 2005, Combustion Theory and Modelling, 113-135.

Figure 1: temperature profiles when a burst of heat is released at a distance δ from the junction of solid (left) and gas (right) domains. Dots are numerical results and line are the analytical solution. We have imposed a contrast between gas and solid phase equal to Dg/Ds~103 and Cs/Cg~104. The temperature profiles are given at two times Dst/δ =0.1 and 0.2.

Crystal dissolution and precipitation in porous media flow: variablegeometry

T.L. van Noorden and I.S. Pop and C.J. van DuijnDepartment of Mathematics and Computer Science, Technische Universteit Eindhoven, PO Box 513,

5600 MB, Eindhoven, The Netherlandse-mail: [email protected], [email protected]

ABSTRACT

In this work we propose a pore scale model for crystal dissolution and precipitation in a porous medium.Our investigations are a first step towards an upscaled model for crystal formation in reactive porousmedia flows and are motivated by the work in [1,2].

We consider a porous medium that is fully saturated by a fluid in which cations M1 (e.g. sodium ions)and anions M2 (e.g. chlorine ions) are dissolved (see Fig. 1). In a precipitation reaction, n particles ofM1 and m particles of M2 can precipitate in the form of one particle of crystalline solid (e.g. sodiumchloride) attached to the surface of the grains (the porous matrix). The reverse reaction of dissolutionis also possible. As a result of the precipitation and dissolution of crystals the geometry of the flowdomain may change.

The model we propose consists of a system of coupled partial differential equations on a variabledomain. The changes in the flow domain, which occur due to the dissolution and precipitation of crystalsare modeled by a Stefan-like boundary condition. This describes the movement of the interface betweenthe fluid and the crystal layer.

For simple geometries we show that solutions exist and we study their qualitative behavior using bothanalytic and numerical techniques.

Crystals

− +

+

−

+

−

−+

−

+−

Ions

Grain

Figure 1: Saturated porous medium with ions dissolved in the fluid and crystals attached to the grainsurface. The ions and the crystals may undergo a precipitation/dissolution reaction.

REFERENCES

[1] P. Knabner, C.J. van Duijn, S. Hengst, “An analysis of crystal dissolution fronts in flowsthrough porous media. Part 1: Compatible boundary conditions”, Adv. Water Res., Vol. 18,(1995).

[2] C.J. van Duijn, I.S. Pop, “Crystal dissolution and precipitation in porous media: pore scaleanalysis”, J. reine angew. Math., Vol. 577, (2004).

IAHR-GW2006 KeyNote Lecture

Multiscale Analysis of Biological Processes in Porous Media.

Brian WOOD Oregon State University, Corvallis, OR 97330 USA

[email protected]

ABSTRACT

Although we often think of 'bioremediation' when discussing biological processes in porous media, there are a large number of other important natural and engineered biological processes that occur in such systems. Specific examples include wastewater treatment (e.g., trickling filters or anaerobic packed bed reactors), water filtration, immobilized cell bioreactors, biofilters for removing air contaminants, and even scaffolds for growing tissues. Tremendous advances have been made in the recent past in biology in general, and this has made a significant impact in terms of research on biological processes in porous media. In this talk, the multiscale structure of biological systems in porous media will be discussed, and several specific examples of upscaling in such systems will be presented. The discussion will include a survey of relevant theory (from the volume averaging perspective), and an overview of some new methods that are being adopted for making measurements of these complex systems. Comparisons of theory with experimental data will be presented where possible.

Predicting the tracer plume development in fractured porous media by applying a double continuum approach

Beyer, Matthias; Mohrlok, Ulf Institute for Hydromechanics, University Karlsruhe (TH), Kaiserstr. 12, 76128 Karlsruhe, Germany,

phone: 0049-721-6084106, [email protected]

fracture networks, transport simulation, double continuum model

Flow and transport in fractured porous media is very complicated due to their extremely

heterogeneous structure. Large fractures provide preferential pathways for regional fluid

movement leading to a fast contaminant transport, whereas the small fractures as well as the

rock matrix are relevant in terms of storage and retardation, as their permeability values are

some orders of magnitude smaller. The interactions between these systems are characterized

by specific exchange processes.

To predict meso and macro scale plume developments in fracture matrix systems, a double

continuum approach is chosen, as the detailed knowledge of the discontinuity geometries are

not required, but the heterogeneous transport behavior can be adequately represented by two

overlapping and interacting continua. The first continuum represents the large fracture

system, whereas the second continuum represents the small fracture system or the matrix

respectively. These two continua possess different flow, transport and storage parameters and

are connected by respective exchange terms.

An existing double continuum approach has been further developed for steady state flow

conditions, with the focus on mass exchange terms between the fractures and the matrix

system. This approach is implemented in the “Double-Continuum MT3D”- program and has

been applied to transport problems in fractured porous systems with different fracture network

geometries.

The parameter determination for the fracture continuum is the most important step to calibrate

the double continuum model and to predict the plume development. Of particular interest

were the determination of hydraulic parameters by fracture network characteristics. The

transversal dispersion coefficient and hydraulic conductivity are directly dependent on the

angle between the fracture direction and hydraulic gradient. The higher the angle, the higher

is the transversal dispersion coefficient and the lower will be the resulting effective hydraulic

conductivity. These relationships have been investigated and quantified in terms of different

fracture networks.

1

Flow over and within a porous bed computed using a macroscopic

formulation of a low-Reynolds-number k-ε turbulence model

Mahmud Ahsan, Mark A. Cotton+ and Peter K. Stansby

School of Mechanical, Aerospace and Civil Engineering (MACE) The University of Manchester

Manchester M60 1QD U.K.

(+ Author for correspondence. e-mail: [email protected] Tel: +44 161 306 5752)

Summary: Computations of the flow over and within a porous bed are undertaken using a ‘macroscopic’ representation of a low-Reynolds-number two-equation (k-ε) turbulence model. The formulation is extended in the porous region to include the Darcy and Forchheimer terms within the mean flow and turbulence transport equations. The effect of specifying alternative free surface boundary conditions on the ε-equation is investigated and it is found that turbulence profiles in the free-flow region are sensitive to this prescription. There are significant differences between the present results and those of Cotton, Reedha and Stansby (2005) for channel flow over an impermeable bed. The flow distribution between the free-flow and porous region is influenced strongly by the value assigned to the ‘Forchheimer coefficient’, cF (results not shown in the present Abstract). Flow description and solution methodology The measurements of Prinos et al. (2003) supply the target data for the present computational study of water flow over and within a porous bed. The experimental arrangement consists of two-dimensional (x-z) flow through a lower porous region of depth hp and an upper unrestricted region of depth hf. The porous region consists of a regular matrix of rods, i.e. cylinders in cross-flow, in which the rows and columns of the matrix are configured to lie in horizontal and vertical planes. Prinos et al. reported hot-film measurements of mean velocity and turbulence profiles taken along a vertical plane bisecting two adjacent columns of rods. ‘Microscopic’ turbulence model calculations in which the flow around the individual rods was resolved were also presented by Prinos et al.

Within a macroscopic representation the flow is spatially fully-developed and, following Getachew et al. (2000), the x-momentum equation may be written

+−−

∂∂+

∂∂+

∂∂−=

∂∂ k

UU

32UU

KcφU

Kνφ

zU)νν(

zxp

ρ1

tU

2/1F2

t (1)

where U is the fluid velocity, φ and K the porosity and permeability of the porous medium, and cF the Forchheimer coefficient. In the free flow region, phz ≥ , the final two terms of equation (1), the Darcy and Forchheimer extensions to the momentum equation, are set to zero.

The basic vehicle adopted here to supply the eddy viscosity, νt in equation (1) is the ‘low-Reynolds-number’ two-equation turbulence model of Launder and Sharma (1974) in which transport equations are carried for the turbulent kinetic energy, k and its rate of viscous dissipation, ε. The Reynolds number to which reference is made in describing the Launder-Sharma model is a local Reynolds of turbulence of the form Ret = εν/k 2 . Details of the modelled k- and ε-transport equations are given in the full paper.

2

Results To give an indication of the full set of results, Fig. 1 shows turbulent kinetic energy profiles for Case 250-50 of Prinos et al. (2003) in which the Reynolds number of the free-flow region, Ref = ν/hU ff,bulk , is equal to 1.04 × 104. Computed profiles from the present programme

of research are shown together with the hot-film data of Prinos et al. and their microscopic turbulence model results. cF is set to 0.3 throughout the mean flow and turbulence model equations. The four profiles generated using the macroscopic formulation of the model differ only in the free surface boundary condition applied to the ε-transport equation (these being the widely-adopted zero gradient condition, a new form developed by Cotton et al., 2005 based on a near-surface limiting form of the k-equation, and two variants of a non-local specification proposed by Celik and Rodi, 1984 and 1988). The present macroscopic results tend to under-predict the measured values very close to the porous medium/free-flow interface. Over most of the free-flow depth the macroscopic values of k are higher than the data (although the present results are in somewhat better agreement with the microscopic computations of Prinos et al.). The profile generated using the zero gradient boundary condition is clearly distinct from the other macroscopic distributions of k. Differences are apparent between the present results and the earlier calculations of Cotton et al. for flow over an impermeable bed.

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

k/Ubulk,f2

(z-h

p)/h

f

zero gradientCelik & Rodi (1984), α = 0.18Celik & Rodi (1988), α = 0.43limiting formExpt. data, Prinos et al. (2003)Micro. calc., Prinos et al. (2003)

Fig. 1 Turbulent kinetic energy profiles for Case 250-50 (cF = 0.3). References Celik, I. & Rodi, W., 1984. Simulation of free-surface effects in turbulent channel flows. PhysicoChemical

Hydrodynamics, 5, 217-227. Celik, I. & Rodi, W., 1988. Modeling suspended sediment transport in nonequilibrium situations. ASCE J.

Hydraulic Eng., 114, 1157-1191. Cotton, M. A., Reedha, D. & Stansby, P. K., 2005. Low-Reynolds-number two-equation turbulence modelling

for open channel flow: development and evaluation of free surface boundary conditions on the dissipation rate equation. J. Hydraulic Res., 43, 631-642.

Getachew, D., Minkowycz, W. J. & Lage, J. L., 2000. A modified form of the κ-ε model for turbulent flows of an incompressible fluid in porous media. Int. J. Heat Mass Transfer, 43, 2909-2915.

Launder, B. E. & Sharma, B. I., 1974. Application of the energy-dissipation model of turbulence to the calculation of flow near a spinning disc. Lett. Heat Mass Transfer, 1, 131-138.

Prinos, P., Sofialidis, D. & Keramaris, E., 2003. Turbulent flow over and within a porous bed. ASCE J. Hydraulic Eng., 129, 720-733.

This abstract was submitted to International Ground Water Conference IAHR-GW2006

12-14 June 2006

SCALE-DEPENDENCE OF DISPERSIVITY IN UPSCALED TRANSPORT MODELS

Daniel Fernàndez-Garcia1

J. Jaime Gómez-Hernández1

1Polytechnic University of Valencia, Ingeniería Hidráulica y Medio Ambiente, Camino de Vera s/n., 46022 Valencia, Spain, Email: [email protected] ABTRACT

Prediction of the fate and transport of dissolved contaminates in groundwater is required in conducting risk analysis and in decision-making in problems of hazardous waste management and remediation of contaminated sites. In order to efficiently make such predictions in hydrogeologically complicated field settings, analytical or numerical models that solve the groundwater flow equation in conjunction with the advection-dispersion equation are often employed based on a coarse computational discretization of the domain (with grid-blocks larger than the scale of heterogeneity). Upscaling is the process by which small scale information (e.g. core samples) is transferred to a larger support scale given, for instance, by the numerical model grid-blocks. Since solute transport is largely dominated by small-scale fluctuations of the velocity field, the impact of suppression of the velocity variability within grid-blocks in the upscaling process is important. Although the scale-dependence of dispersivity associated with transport of solute through small-scale lnK random fields has been largely investigated in the literature, the corresponding scale-dependence of dispersivity associated with upscaled transport models (involving a coarse discretization) is still an open question. We present Monte Carlo transport simulations that elucidate how the process of upscaling impacts on the scale-dependence of dispersivity. To achieve this, we compare the scale-dependence of dispersivity associated with a transport model before and after upscaling the underlying heterogeneity. The upscaling process not only consists in estimating grid-block hydraulic conductivity tensors, but also incorporates a block dispersivity tensor that accounts for the suppression of heterogeneity within grid-blocks. Block hydraulic conductivity tensors are calculated to preserve the mean fluxes at the block scale, while block dispersivity tensors are estimated from simulated uniform flow tracer tests that only samples the portion of heterogeneity defined by the grid blocks.

WHY, WHEN AND HOW WE NEED TO APPLY CONJUNCTIVE WATER MANAGEMENT OF SURFACE AND GROUNDWATER:

THE CASE OF THE CHARENTE BASIN, FRANCE

Fabien Christin, Gilles Belaud, Pierre-Olivier Malaterre, Christian Leduc, Patrick Le Goulven

UMR G-EAU - "Water management, Actors and Uses" Cemagref - 361 rue Jean-François Breton

BP 5095, 34196 MONTPELLIER cedex 5, France email : [email protected]

In a context of water scarcity, managers are dealing with quantitative and qualitative management of surface and groundwater resources. In the Charente river basin, the water needs can exceed the natural and artificial water supply. In this case, managers are faced with an imbalance during the low flow period, between the available water resource and the quantity required for different uses. The water consumption represents 136 Mm3 in the Charente basin of which 11 Mm3 is for drinking water, 0.4 Mm3 is for industry and more than 124 Mm3 is used for irrigation. In France, the development of irrigation since the 1970s is an important cause of the imbalance between available resources and demands. Added to high sensitivity of the resource to climatic conditions, the water demand causes very low flows to occur in different parts of the Charente basin which lead to negative ecological impacts on river ecosystems, as well as economic impacts based on the agricultural yield due to water restrictions that are placed on the irrigation. Due to these water resource problems, Charente basin managers decided to implement several solutions: (i) water rules were imposed to decrease irrigation water withdrawals as restricted level, (ii) new water resources were created with the construction of two dams which represent 22.4 Mm3 and (iii) deficit management tools were implemented. Management tools would provide essential information for their decisions on allocation of water resources. Indeed, managers are faced with the task of identifying optimal or near-optimal water management solutions in highly complex hydrosystems. In the Charente basin, the model used, Tableau de Bord de la Resource en eau (TBR), is a water surface model with hydrologic, hydraulic and water demand (irrigation) modules. So, the TBR only considered the surface component and not the groundwater component. Over a long period of time, scientists and water managers have looked at groundwater and surface water as two separate entities. Through the Charente basin example, the presentation explains why we have a water scarcity situation and why, in this case, a conjunctive water management is particularly needed and not just a surface water model as the TBR. Studies have highlighted the importance of the groundwater component in the Charente Basin and the important exchange between the aquifer and the river, as well as the influence of irrigation pumping on the rivers flows. In more details, the presentation is broken down into three parts:

Water resources and the management situation in the Charente Basin, Quantification of the interactions of the surface water and groundwater components, How better knowledge of the surface and groundwater interaction could lead to improved water

management in the Charente basin (flow forecasting, effectiveness of dam water releases). Key-words: surface and groundwater relationship, water management, aquifer-river interaction,

Charente

Block-Upscaling of Transport: A comparison of ADE and

mobile/immobile approach

Matthias Willmann, Xavier Sánchez-Vila, and Jesus Carrera

Department of Geotechnical Engineering and Geosciences, Technical University of

Catalonia, Jordi Girona1-3, 08034 Barcelona, Spain

Real aquifers are heterogeneous over an evolving range of scales, and problems in

hydrogeology are usually too complex to be solved at the pore scale. Thus upscaled

transport equations are needed which allow to model large scale heterogeneities on the

numerical grid and small scale heterogeneities within the transport equation.

Upscaled equations exist for large travel times derived in a stochastic framework. Dagan

(1994) and Rubin et al. (2003) introduced two upscaled equations with the same shape as

the ADE but modeling the heterogeneity as a function of block size within the (macro-)

dispersion tensor. Their approach is based on the assumption of ergodicity and this makes

their solution only valid for large travel times, which is of limited importance in many

field applications.

At shorter time scales, so-called anomalous or non-Fickian behavior is reported. This

behavior cannot be modeled by the ADE. The most prominent anomalies are the scale

effect of dispersivity, the tailing of breakthrough curves and the directional dependence

of porosity.

This behavior can be represented by various models conceptualizing the aquifer as a

superposition of a mobile and one or more immobile zones. The main limitation of such

models resides in the fact that their parameters are determined by some best-fit

procedures against field data and are not easily related to measurable quantities of clear

physical meaning.

Our work aims at comparing the capabilities of the upscaled ADE and the

mobile/immobile approach to reproduce the actual concentration field within a

heterogeneous field at small time scales.

The approach is based on the following steps. First, we model a plume on single

realizations of heterogeneous transmissivity fields with a very fine grid resolution and

solving the ADE with constant local-scale dispersivity. We assume this model to

reproduce a perfectly known reality where the ADE is valid (i.e., at the pore scale). The

transmissivity field created using a Gaussian Random Simulator is based on a nested

variogram in order to reproduce heterogeneities at different scales.

Then, some measurements of concentration are taken at various times and locations

within this reference field and transmissivity is upscaled, as a function of a progressively

increasing size of the grid blocks.

Both the concentration observations and the upscaled transmissivities are finally

embedded within an inverse modeling procedure to estimate the transport parameters as a

function of the numerical grid size adopted. Two different inverse models are employed:

(1) An ADE where the (macro-) dispersion coefficient is estimated, and (2) an ADE with

an exchange term between mobile and immobile zone where the exchange parameters are

estimated. The procedure is then repeated for various block sizes of transmissivity and for

different heterogeneity structures. The estimation results are then compared with the

original, fine grid, solution.

We show for short time scales that the mobile/immobile approach is reproducing much

better the original field than the upscaled ADE. We then propose an empirical

relationship to describe the dependence upscaled mass transfer parameters: (1)

measurable geostatistical properties of transmissivity as mean, variance, and correlation

length; and (2) the block size of the numerical model.

This would allow the modeler to take a better choice of upscaled input parameters into a

transport model in heterogeneous aquifers.

IAHR-GW2006-Theme 2 June, 2006; Toulouse, France

Groundwater Flow Modeling of Perched Karstic Aquifers and Springs and its Implication for Determining Precipitation-Recharge Relationship

Menachem Weiss and Haim Gvirtzman

Inst. of Earth Sci., Hebrew University, Jerusalem 91904, Israel

The carbonate rocks at the Judea and Samaria Mountains, Israel, are composed of thick bedded karstic limestone and dolomite, and thin marl formations. Groundwater accumulates preferentially on top of the relatively impermeable marl layers and creates perched aquifers. The karstic character of the carbonate rocks, in conjunction with the marly aquitards, causes diversion of groundwater flow to discrete springs. These perched-karstic springs are common across the Israeli mountain landscape and some have discharges on the order of 100,000 m3 per year. Using detailed topographic and geologic maps, we estimated the recharge areas of various springs (at a scale of 1-5 square kilometers). For many of the springs, there is an extensive database, tens of years long, of spring discharge (hydrographs). Detailed records of daily precipitation are also available from meteorological stations adjacent to the springs.

In this study, we developed numerical groundwater flow models, using the finite-difference MODFLOW code. The models accurately simulate the groundwater flow field within the perched-karstic subaquifers, as well as the spring hydrographs. The models account for the two separate flow domains, matrix and karstic conduits; the latter being simulated by using the drain package within MODFLOW. The perched nature of the aquifer was simulated in a way that the upper sections of the modeled domain oscillate between saturated and unsaturated conditions. The leakage from the perched aquifer through the lower aquitard into the deeper aquifer was simulated by assuming atmospheric pressure lower boundary conditions. During the calibration process, the following parameters were estimated: (1) the horizontal hydraulic conductivity within the upper, perched aquifer layer; (2) the vertical leakance between the upper, perched aquifer layer, and the lower regional aquifer layer; (3) the storage coefficient in the upper, perched aquifer layer; (4) the karst conduit conductivity

and the appropriate spatial geometry within the upper, perched aquifer; and (5) the recharge.

Although precipitation can be quantitatively well-defined, the conversion from precipitation to subsurface recharge values is very vague and many hydrologists are constantly dealing with this problem. Using the numerical models and the long-tern precipitation and discharge records, we were able to calculate the transient behavior of recharge, leakage and spring discharge. Preliminary results from such numerical modeling calibration efforts show that peak spring discharges cannot be replicated by using a simple percentage of the observed transient precipitation data. However, by considering the assumed changes in evapotranspiration rates in unusually wet or dry years, the model can be reasonably calibrated. Thereby, a new hydrometeorological model (a Precipitation-Recharge function) that accounts not only for normal years but also for exceptionally wet and dry years was quantified. The new model circumvents the problem of overestimating recharge in relatively dry years and underestimating recharge in relatively wet years.

To date, four springs have been analyzed:

1. Ein Al Matwi near Ariel at the Samaria Mountains. The recharge area is 2.0 km2 and the average yearly discharge is 100,000 m3.

2. Ein Delbah near Dura at the Judea Mountains. The recharge area is 0.6 km2 and the average yearly discharge is 36,000 m3.

3. Ein Chaniya at the southwestern part of Jerusalem. The recharge area is 2.6 km2 and the average yearly discharge is 110,000 m3.

4. Ein Harrashah near Ramallah at the Samaria Mountains. The recharge area is 1.4 km2 and the average yearly discharge is 80,000 m3.

The first 3 springs are located at the contact between the Aminadav Formation (karstic limestone perched aquifer) on top and the Moza Formation (relatively impermeable marl aquiclude) below. The fourth spring is located at the contact between the Kfira Formation aquifer and the Qatana Formation aquitard. Because of the different geological settings, the modeling efforts have also allowed us to better understand the spatial variations of the hydrological parameters.

Sub-gridded dual-porosity models: accurately modelling matrix-fracture transfers in fractured porous media

C. Famy1, B. Bourbiaux1, P. Lemonnier1 and M. Quintard2

Description of the paper

Flow simulations in fractured media are often based on the “dual-porosity” concepti,ii, in which the representation of both fracture and matrix media is very simplified. At the reservoir model gridblock-scale, most simulators use an approximate pseudo-steady-state mass exchange formulation resulting from an upscaling of the matrix block-scale flow. This formulation is reasonably predictive for single-phase flows, but generally inaccurate for multiphase flows. This is mainly due to the impact of non-linearities and the coupling between several physical mechanisms, especially capillarity and gravity, that do not yield the same homogenised flow behaviour at the matrix block scale. Furthermore, multiphase exchanges require more accurate upscaled formulations. A numerical approach to overcome this limitation consists in sub-gridding the matrix blocksiii, but this method is still unused in practical situations because of its high computational cost.

This paper describes the optimisation of this sub-gridding technique in the capillary imbibition

case, by taking into account the physical specificities of this mechanism and with a criterion of minimal computational cost. Implemented in a conventional flow simulator, this technique allows the calculation of very reliable exchange terms between matrix blocks and fractures. Results, observations and conclusions

The study of the capillary imbibition mechanism on a single matrix block (possibly anisotropic) allows one to create an optimised irregular multi-dimensional sub-grid of the matrix block, which is turned into a one-dimensional one to reduce the number of cells, thus the computational cost (Fig. 1).

Fig 1 – Sub-gridding methodology: 2D example

This sub-gridding methodology has been validated by comparison with reference fine-grid simulations, for various rock-fluid properties and anisotropic flow conditions. The reference solutions are reproduced very accurately, including the detailed evolution of matrix-fracture transfer rates with

1 Institut Français du Pétrole (I.F.P.), 1 & 4 avenue de Bois-Préau 92852 Rueil-Malmaison Cedex 2 Institut de Mécanique des Fluides de Toulouse (I.M.F.T.), allée du Professeur Camille Soula, 31400 Toulouse

time. Figure 2 presents a comparison, in the 2D case, between the reference fine-gridded model (solid line), the 2D model with an optimal sub-grid (dashed and dotted line), the model involving an optimised one-dimensional sub-grid (dashed line) and the conventional dual-porosity model (dotted line).

Fig 2 - Prediction of the kinetics of capillary imbibition of a matrix block

using the sub-gridded models and a dual-porosity model: 2D example Since the sub-gridding methodology provides accurate exchange terms, it has been implemented

in a conventional flow simulator dedicated to fractured porous media. Additional efficiency was obtained through the choice of a proper algorithm. Matrix unknowns are pre-eliminated, except those involved in the exchange flux at matrix-fracture boundary. This pre-elimination allows one to solve first a non-linear system of same size as the one solved in the case of a conventional double-porosity model. The matrix pressure and saturation are then computed separately, by solving a small linear system per sub-gridded matrix block. This coupling method has been especially chosen to minimise the CPU time required by the sub-gridding methodology.

The accuracy of the simulated exchange terms ensures reliable production profiles to be predicted from fractured porous media.

Applications

The sub-gridding methodology improves the calculation of matrix-fracture exchanges driven by capillary forces. Therefore, short- and long-term flows of water-drive fractured media can be reliably predicted. This methodology can be applied to other multiphase flow problems solved with an industrial fractured reservoir simulator. Technical contributions

The main technical contributions of this work are the development of an optimised method for sub-gridding matrix blocks, and the implementation of this sub-gridding methodology into a conventional flow simulator dedicated to fractured media.

i G.I. Barenblatt, I.P. Zheltov and I.N. Kochina, Basic concepts in the theory of seepage of homogeneous liquids in fissured rocks, PMM vol. 24, n°5 (1960). ii J.E. Warren and P.J. Root, The behavior of naturally fractured reservoirs, SPE Journal (1963). iii K. Pruess and T.N. Narasimhan, A practical method for modeling fluid and heat flow in fractured porous media, SPE Journal (1985).

MODELLING SURFACTANT-INDUCED FLOW AND CONTAMINANT TRANSPORT IN THE VADOSE ZONE

Scott E. Munachen, Geohazard Research Centre

Abstract A fundamental understanding of the impact of surfactants on unsaturated flow and contaminant transport is critical to the design of systems for the efficient remediation of the vadose zone. Although flow through porous media is commonly assumed to be independent of solute concentration, dissolved organic surfactants can have a direct and significant effect on the flux through unsaturated due to the dependence of capillary pressure on interfacial surface tension. This paper presents the results from a series of physical model tests designed to investigate the influence of solute concentration-dependent surface tension on unsaturated flow phenomena. Two-dimensional unsaturated flow perturbations caused by surfactant-induced capillary pressure gradients were observed in large-scale miscible displacement experiments using an instrumented flow cell packed with sand. A 1-butanol surfactant solution containing a dye tracer was applied at a constant rate from a point source located on the soil surface above an unconfined synthetic aquifer with ambient groundwater flow and a capillary fringe of approximately 0.5 m. The migration of the contaminant plume through the vadose zone was tracked via a transparent observation panel using close range photogrammetry and digital image correlation. Over thirty instrumentation stations comprising time domain reflectometry probes and tensiometers measured in-situ moisture content and pressure head, respectively. The data indicate two distinct flow phenomena associated with concentration-dependent surface tension in the vadose zone. Firstly, there is a transient flow perturbation associated with the advance of the solute front beneath the point source. Large surfactant concentration gradients associated with the advance of the solute front generate capillary pressure gradients that produce locally high fluxes. Secondly, because the air-entry pressure decreases in proportion to the reduction in surface tension induced by the surfactant, there is a localised depression of the capillary fringe as the contaminant displaces pore-water in the vicinity of the phreatic surface. The height of the depressed capillary fringe is proportional to the product of the height of the initial capillary fringe and the relative surface tension of the surfactant solution. The horizontal transport of surfactant in the depressed capillary fringe, driven primarily by the ambient groundwater flow, caused the propagation of a wedge-shaped drying front in the down-gradient direction. Comparison of dye transport mechanisms during experiments with and without surface-active compounds indicated that because surfactant-induced drainage decreased the storage capacity of the vadose zone, the dye breakthrough time to the water table was more than twice as fast when the contaminant solution contained surfactant.

The research clearly indicates that accurate simulation of unsaturated flow of compounds that significantly depress surface tension can only be achieved by including the effects of surface tension on pressure head and the associated hydraulic parameters. It is therefore imperative that concentration-dependent surface tension be integrated within current conceptual and numerical models to fully assess situations involving the transport of organic solutes.

Solute transport by a shear thinning fluid in a channel flow: influence of the geometrical disorder Victor J. Charrette1, Elisa Evangelista1, Ricardo Chertcoff1, Harold Auradou2, Jean-Pierre Hulin2 and Irene Ippolito1

1 Grupo de Medios Porosos, Facultad de Ingeniería, Universidad de Buenos Aires, Paseo Colón 850, 1063 Buenos Aires, Argentina 2 Laboratoire Fluides Atomatique et Systèmes Thérmiques, Bat 502, Campus Universitaire, 91405 Orsay Cedex, France Subsurface fluids recovery, whether oil and gas, or contaminants, requires understanding the way in which fluid(s) flow within porous and fractured rocks and soil. Tracer test is a common way to investigate the connectivity and heterogeneity of the pore space and to determine transport properties such as the dispersivity. The dilution of the dissolved tracers results from their spreading under the combined effects of the velocity fluctuations in the bulk of the medium and of the molecular diffusion that allows tracer particles to move from one streamline to another. The classical approach assumes that dispersion can be treated as a fickian process and characterized by a dispersion coefficient independant of the path length. One can identify different regimes of dispersion according to the Peclet number which is the ratio between the typical times for molecular diffusion and convection. In the present work, we report experimental measurements of dispersion in a channel geometry. Dispersion in three different geometries leading to preferential flow paths has been investigated : these include an empty glass tube and the same tube filled of glass beads with two different packing geometries (the tube diameter is 3 mm and its length 1.5 m). In the second configuration the tube is filled with 2.5 mm diameter beads lined up in contact with each other at the bottom of the tube. In the third configuration, the tube is tilted while being filled, leading to a disordered and more compact packing. Experimentally, a steady flow is first established in the tube, then the fluid is displaced by the same fluid but dyed. The dilution of the dye is observed by using a high resolution CDD camera of high resolution for measuring the time variation of the light absorption by the fluid at the outlet of the tube. The dispersion coefficient is obtained from the experiments by fitting the breakthrough curves. In order to modify the velocity field and, therefore the dispersion characteristics, the rheology of the fluid was modified by adding polymer (xanthan) to the solution. To a first approximation, the viscosity of this type of shear-thinning fluid varies following a power law of the shear rate. When the tube does not contain beads, mixing arises from the parabolic velocity profile; velocity is then highest at the center of the tube and zero at the boundary. Because of molecular diffusion, a tracer particle may «jump» from one streamline to an other : its velocity increases as it moves towards the center of the tube and decreases towards the walls. This random motion leads to the well known Taylor dispersion process in which the dispersion coefficient varies as the square of the Peclet number. This behavior is clearly visible in figure 1 where the dispersivity ld (ratio of the dispersion coefficient by the Peclet number Pe) varies linearly to Pe. When beads are arranged in line inside the tube, the situation is somewhat different. Fluid flows freely in the space between the top of the bead layer and the tube walls and the flow field is periodic. Very low velocity zones are encountered between the beads in the bottom part of the tube section. These stagnant regions represent a significant part of the total volume and tracer particles can only move in and out of these zones through molecular diffusion : this effect is marked by a tailing (retardation) effect in the breakthrough curves. In this case the breakthrough curves must be separated into two components corresponding respectively to dispersion in the flowing zones and diffusive exchange with the stagnant regions. Using this approach, one defines an asymptotic dispersion coefficient Das (it represents the dispersion coefficient for a system of same structure but

much longer length for which the convection-diffusion equation is again valid.) Figure 1 displays the variations of the dispersivity ld as a function of the Péclet number : as for the empty tube, the dispersivity ld increases linearly with Pe but is significantly larger. This seems to be associated to the larger size of the stagnant zones when beads are present which increases the characteristic diffusive exchange time, and therefore the dispersion coefficient. In this case, the dispersivity verifies ld ∝ Pe0.5 corresponding to a behaviour intermediate between geometrical (ld ∝ const) and Taylor dispersion (ld ∝ Pe). Although this behaviour is not fully understood, the disorder of the flow field in this type of array results in much easier mixing betweenflow lines as in the periodic array. Preliminary experiments with polymer solutions aimed at better understanding these effects have been performed : they show that dispersivity variations remain the same in the periodic array while ld increases with the concentration in the disordered array. Figure 1 : Variations of the the characteristic length ld = D/U as a function of the Peclet number Pe = Ud/Dm (d is the diameter of the tube, U is the mean velocity flow). These experiments were realized for a solution of 10 % glycerol in water (Newtonian fluid). For an empty tube (filled circles) : ld ∝ Pe indicating that Taylor dispersion dominates. For the periodic bead array, one has also ld ∝ Pe for Pe < 2000. For higher Pe values convective mixing becomes dominant (Re > 5). For a disordered bead array,ld ∝ Pe0.5 (white triangles corresponds to experiments performed with a solution of 70 % glycerol in water, 20 cp of viscosity).

0.1

1

10

100

1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06

Pe

Ld (c

m)

Methodology of the hydraulics and hydrodynamics Modelling aquifers-stream interactions

Noureddine GAALOUL

National Research Institute for Rural Engineering Water and Forestry (I.N.R.G.R.E.F), Rue Hédi Karray B.P.10-2080 Ariana - TUNISIE

: +216.98.43.57.72 E-mail: [email protected] Abstract The variations of level of the rivers modify the piezometric dimensions of the aquifer. A thorough knowledge of the aquifer - rivers interactions makes it possible to improve the stock management out of water on a scale a basin. In regional scale, the determination of the transfer’s aquifer - rivers is intended to predict the heights of aquifer (humidity condition, inundation) following the river modifications (management, weather conditions). The quantification of these exchanges to the regional ladder is complex; the analytical solutions are often impossible. From that time, mathematical models relating to ground water, to surface water and to there interaction are implemented. The resolution techniques and the precautions linked to the implement (acquisition of data, setting, ..) hydraulic models to free surface and hydrodynamics are approached. To take into account of the aquifer-stream interactions in the models, several procedures are conceivable. The hydraulic behavior of irrigation canals shows that these systems are complex, with a dynamic characterized by important time lags, strong non-linearity and numerous Interactions between different consecutive sub-systems. A good knowledge of system dynamics is needed to design an automatic controller for an irrigation canal. It is possible to split a canal into sub-systems composed of reach with a cross regulator at is downstream end. Cross structure dynamic, for example a gate, is simple to model. On the other hand reach dynamics is modeled by a set of partial derivative hyperbolic equations: Saint-Vansant’s equations. Linearisation of Saint-Vansant’s equations near a steady flow, allows obtaining a reach transfer matrix that has the advantage to keep the distributed parameter system characteristics and therefore the infinite state space dimension. The accuracy of this transfer matrix is evaluated for two kinds of reach behaviour : a short reach with waves propagation, a long reach with delay and damped wave motion. This evaluation is made by using a model based on a finite difference approximation of Saint-Venant’s equations by Preissmann’s scheme. Fluctuations in discharge and water depth in a canal pool are due to two physical phenomena occurring in the flow: wave propagation (perturbation), mass transport (long waves). If gate movements are slow, water depth and discharge change slowly and the flow is unsteady gradually varied. Assume that the flow is one-dimensional, streamline curvature is small and velocity is uniform over the cross section, the flow can be modeled very accurately by Saint Venant equations.

The fluctuations in discharges and depths in a reach depend on many parameters such as reach length, bed slope, cross section, roughness, and initial water profile. In order to characterize reach behavior without studying the influence of each variable, the flow is characterized using dimensionless numbers. To transform Saint-Vansant’s equations in dimensionless form each variable is divided by a constant reference with the same dimension. These problems can be resolved by the principal techniques, which are hydraulics and hydrodynamics modelling. Mathematical models are developed to solve flow computations in surface and groundwater by the equations of Saint Venant and Laplace. Those models differ by the chosen method of computation and discretization. The modelling of aquifer-stream interactions is realised through two approaches: the integration method and the unilateral method. By coupled models, groundwater levels and river stages are computed simultaneously. The interactions are taken into account by the law of Darcy. An integration model (a surface model coupled to a groundwater model) is developed by the implicit method of finite differences. Tests, undertaken on the surface models and on the integration model, show the importance of the time steps on the results validity. On the other hand, a unilateral model, easier to use, is developed to solve regional problems. The river is simulated either as a set potential, or as a flow, generated by the water head difference between the river and the aquifer. The latter one is know, and is supposed not to change by leakage out of the aquifer. This model has allowed to resolve a practical problem: it was possible to define the zones exposed to groundwater rises due to decennial or century floods in the Cap Bon region in the North of Tunisia. Key words: Hydraulic model, Hydrodynamic model, aquifer-rivers interactions, Tunisia.

Geostatistical modelling for the quantification of uncertainties on the unsaturated zone and the groundwater transfer

S. MAZUEL (1 & 2), C. de FOUQUET (1), M. KRIMISSA (2), J.-P. CHILÈS (1)

(1) Centre de géostatistique – Ecole des Mines de Paris – 35, rue Saint-Honoré 77305 Fontainebleau Cedex - FRANCE (2) EDF-R&D Laboratoire National d’Hydraulique et Environnement – Groupe P 78 – 6, quai Watier 78401 Chatou Cedex - FRANCE

Uncertainties regarding the development of pollutant plumes are related, among other factors, to uncertainties associated with the geological medium, the flow and the source term. A case study was conducted on a former industrial site, whereby the pollutant transfer in the unsaturated zone was examined within a small domain. Two sampling surveys were completed to evaluate both the source term and physical parameters. Due to technical constraints the two sampling surveys were completed on different parts of the site, whose locations were approximately 10 meters apart along a horizontal plane, and 5 meters apart along a vertical plane. Within the first part of the site, collected data was used to conduct an exploratory data analysis to characterize the spatial variability of the physical (bulk density, porosity, moisture, water content) and chemical parameters (pH, organic carbon) measured in both field and laboratory settings (CEA/SAT). Results showed an anisotropic distribution (principal directions x and depth) for the water content and an isotropic distribution for bulk density. It was also noted that the lateral variability in each level is greater than the variability between the means of each level. Therefore this structure cannot be represented by a model with constant values on horizontal layers. The results of this data analysis were then used to create geostatistical models and to estimate (by kriging) or simulate the spatial distribution of the parameters in this zone.

Omnidirectional variogram of the porosity

Within the second part of the site, the source term was sampled and a new exploratory data analysis was completed to characterise the spatial concentration distribution. Concentration samples were comprised of two different volumes (50 and 500 cm3). Therefore, important questions were: can we use the same analysis (ex. probability distribution) for samples of differing volume, and can we estimate or simulate the pollutant plume with all data? Because of the presence of high values a log transformation of the translated data was used ( data was transformed to Ln(1+C) where C is the concentration). Experimental variograms were computed and variogram models were fitted. Indicator variogram for the first, second (median), and third quartiles were tested to better understand the influence of low, medium and high values on the pollutant plume dispersion. A separate statistical analysis for the two sample sizes showed that the sample volume does not affect the coefficient of variation (standard deviation / mean). This suggests that the data from both the 50 cm3 and 500 cm3 samples can be analysed together. After the data exploratory phase, a fast Fourier transform algorithm was used to construct geostatistical simulations of flow parameters (saturated water content, residual water content, saturated hydraulic conductivity, water retention curve and the relative hydraulic conductivity curve following the Van Genuchten law). The simulations obtained were then used as input data for an unsaturated zone flow study with the finite element code ESTEL (EDF R&D). Spatial distribution models were tested with different type of variograms, ranges and standard deviations. The influence of the spatial variation of flow parameters in the unsaturated zone was examined through a sensitivity study, using various flow and boundary conditions. Hydraulic flow velocities were then compared between a heterogeneous models and a homogeneous model for the unsaturated zone. For some cases, velocities between the two models were strongly different because of the presence of preferential flow paths in the heterogeneous models. Finally, the pollution source term was also simulated, using the geostatistical model previously fitted to the experimental data. Historical pluviometry was used to impose boundary conditions. Transport computations, and thus adsorption/desorption coefficients, were approximated by a generalized Kd approach. Two different pollutants, one with a high Kd, and the other with a low Kd, were used to help model pollutant transport. The resulting transport was examined and the respective ratios of various uncertainties were then evaluated.

An efficient local time stepping-Discontinuous Galerkin scheme for the advective transport equation in porous media

C. EL SOUEIDY, A. YOUNES & P. ACKERER Institut de Mécanique des Fluides et des Solides, Université Louis Pasteur-CNRS-UMR 7507, 2 rue Boussingault, 67000 Strasbourg, France

Hyperbolic conservations laws model a great number of physically interesting phenomena such as gas dynamics, shallow water flow and advective transport of contaminant in porous media. Such equations often have sharp, moving fronts and other local, fine scale features. Locally conservative methods such as Discontinuous Galerkin methods have the advantages that they are explicit in time and, thus, easy to implement, approximate shocks or sharp fronts accurately and with no oscillations, and are globally mass conservative.

Currently, one drawback of the explicit time integration methods is the global CFL

condition. Small elements or locally large velocities greatly reduce the CFL time step, even if they occur in a small region of the domain. Another possibility is to use one of a variety of unconditionally stable implicit methods. However these methods for advection are generally more numerically diffuse than explicit schemes, and they lead to nonsymmetric systems of equations which can be difficult to solve.

In this work, we present an efficient local time stepping method that can provide a

substantial speedup with explicit time stepping. The main advantages of the method are its simplicity of implementation and its computational efficiency. The new approach reduces the cost of using an explicit procedure while maintaining accuracy by allowing for variable time-steps. The time step is allowed to vary spatially, then the CFL condition can be satisfied locally and larger time steps can be taken in regions of relative inactivity. This approach was motivated by the one-dimensional procedure for conservation laws described in Osher and Sanders (1983).

Numerical experiments show that the local time stepping scheme exhibit similar accuracy and stability to the global time stepping, and gives almost identical solutions as the traditional scheme for some complex flow fields with nonuniform meshes and spatially varying velocities.

Analytical and Numerical Modeling of Flow and Transport

in Highly Heterogeneous Three-dimensional Aquifers:

Ergodicity, Gaussianity and Anomalous Behavior

Igor Jankovic

Department of Civil, Structural and Environmental EngineeringState University of New York at Buffalo, Buffalo, USA

Flow and transport through highly heterogeneous porous formations areexamined in a numerical laboratory and with aid of a simple analytic model.

Multi-indicator conductivity structure is used to represent a 3D isotropicmedium of finite integral scale, I. The hydraulic conductivity follows a log-

normal distribution. The conductivity model contains spherical inclusionsthat are embedded in a homogeneous background. The inclusions representspatially correlated fluctuations in conductivity, while the background rep-

resents conductivity fluctuations over much smaller scales that show up asnugget in conductivity variogram. The effective conductivity is hence used

as the background conductivity. The statistical structure of the model, rep-resented by the mean, variance and two-point correlation, can be adjusted to

match any isotropic statistical structure of aquifers by changing conductivity,size and placement of inclusions. The present study focuses on inclusions of

constant size that are placed on a periodic lattice.Numerical solutions of flow and transport are obtained using the Analytic

Element Method (Strack, 1989, Jankovic and Barnes, 1999). Disturbance

potential due to each inclusion is represented with analytic functions thatcontain 289 degrees of freedom. The total potential equals the potential

due to uniform flow of intensity U plus sum of all disturbance potentials.100, 000 inclusions are used to create a heterogeneous body 220I long and

110I wide (Figure 1). Three variances of log-conductivity, σ2Y , are used for

inclusions: 2, 4 and 8. Flow simulations are carried out on a massively parallelsupercomputer cluster with single-processor equivalent CPU time of over 6

years per simulation. Contaminant transport is then simulated using 40, 000

particles in a stationary core 121I long and 44I wide (both in y and in z

direction) placed inside the flow domain. Concentration in the injection plane(x = 0) was proportional to local flux. Breakthrough curves are computed

1

for a number of control planes. Details of the numerical simulations can befound in Jankovic et al, 2006.

220 I

y

x

z

121 I

110I

44 I

U

Figure 1: Numerical setup. Solid lines are intersections of constant head surfaces and thez = 0 plane.

Analytical solutions of flow and transport are based on classical Maxwell’ssolution for a single spherical inclusion in uniform flow as presented in Fiori

et al, 2006. The conductivity structure behind the analytical solutions isidentical to that of the numerical solutions. This simplified analytic model

accounts for interactions between inclusions in an approximate manner; theseinteractions are accounted for explicitly in the numerical solution. The sim-plified model captures the entire breakthrough curves high precisions for all

investigated values of σ2Y and for all control planes.

Equivalent macrodispersivity, αLeq, is computed by fitting a Gaussian

model to breakthrough curves using least squares method and from the tem-poral moments (mean and variance) of breakthrough curves. In the Gaussian

model (e.g. Dagan, 1989), relative mass flux µ(t, x) at time instance t and

2

control plane at x is expressed as:

µ(t, x) = ∂M∂t

= 12(x

t+ U) Z( x−Ut

4αLeqUt) ;

Z( x−Ut4αLeqUt

) = 1(4παLeqUt)1/2 exp[−

(x−Ut)2

4αLeqUt] (1)

where αLeq is equivalent macrodispersivity that corresponds to the controlplane at x. M(t, x) is a mass arrival function, defined as the relative mass of

solute that has moved beyond x at time t (e.g. Rinaldo et al, 1989, Cvetkovicand Shapiro, 1990, Cvetkovic and Dagan, 1994). αLeq is computed by least

squares matching of µ(t, x) that comes from numerical simulations or theanalytic model with µ(t, x) predicted by Equation 1.

In laboratory or in the field it is customary to identify an equivalentmacrodispersivity with the aid of the temporal moments of µ(t, x):

T(x) =

∞∫0

tµ(t, x)dt ;

σ2t(x) =

∞∫0

(t − T)2µ(t, x)dt (2)

αLeq can then be defined as:

αLeq =σ2

tU2

2x(3)

The integration limits in Equations 2 can be set a finite value to simulateincomplete sampling (recovery) of µ(t, x). This corresponds to values of M in

Equation 1 less than unity (e.g. 99.5% recovery corresponds to M = 0.995).Results for σ2

Y = 8 are presented on Figure 2.

The findings indicate that breakthrough curves display a thin, but per-sistent tail caused by the presence of low-conductive zones. The tail haslarge impacts on second and higher moments of travel time. This explains

the high sensitivity of αLeq on the recovery fraction observed on Figure 2.Furthermore, the tail was not ergodic even for the large domain used in the

experiment. The rest of the plume is ergodic and displays a Gaussian-like(symmetric) shape that is well captured by the Gaussian model (Equation 1).

Modeling and prediction of the tail is of limited practical use due to imprecisecharacterization of low-conductive zones, measurement errors and influence

3

of molecular diffusion. Nonetheless, use of moment-based macrodispersioncoefficient in highly heterogeneous aquifers is questionable.

For the largest simulated variance value (σ2Y = 8), the equivalent macrodis-

persivity grows with distance from the source; it does not level-off, as shown

on Figure 2. The transport is hence non-Fickian, but it does not appear tobe anomalous since tendency to Fickianity is observed. However, whethertransport is Fickian or anomalous may be a theoretical question of limited

practical significance for many highly heterogeneous aquifers because of thelarge ”setting” times.

0.0

3.0

6.0

9.0

12.0

0 20 40 60 80 100 120x/I

aLeq/I99.9% recovery

99.5% recovery

least squares fit

Figure 2: Equivalent macrodispersivity vs. the distance from the injection plane for σ2Y = 8.

Acknowledgments This paper is based upon work partially supportedby the National Science Foundation under Grant EAR-0218914. The authors

also wish to thank the Center for Computational Research, University atBuffalo, for resources and assistance in running numerical simulations.

4

References

[1] Cvetkovic, V.D., and A.M. Shapiro, Mass arrival of sorptive solute inheterogenous porous media, Water Resour. Res., 26, 2057-2067, 1991.

[2] Cvetkovic, V., and G. Dagan, Transport of kinetically sorbing solute

by steady random velocity in heterogeneous porous formations, Journ.Fluid Mech., 265, 189-215, 1994

[3] Dagan, G., Flow and Transport in Porous Formations, Springer-Verlag,1989.

[4] Fiori, A., I. Jankovic and G. Dagan, Modeling Flow and Transport inHighly Heterogeneous Three-Dimensional Aquifers: Ergodicity, Gaus-

sianity and Anomalous Behavior. Part 2: Approximate Semi-AnalyticalSolution, accepted in Water Resources Research, 2006

[5] Jankovic, I., and R. Barnes, Three-dimensional flow through large num-bers of spherical inhomogeneities, Journ. Hydrology, 226, 224-233, 1999.

[6] Jankovic, I., A. Fiori and G. Dagan, Modeling Flow and Transport in

Highly Heterogeneous Three-Dimensional Aquifers: Ergodicity, Gaus-sianity and Anomalous Behavior. Part 1: Conceptual Issues and Numer-

ical Simulations, accepted in Water Resources Research, 2006

[7] Rinaldo, A, Marani, A., and A. Bellin, On mass response functions,

Water Resour. Res., 25, 1603-1617, 1989.

[8] Strack, O.D.L., Groundwater Mechanics, Prentice-Hall,Englewood Cliffs

NJ, 1989.

5

A new method to invert hydraulic pumping tests for the identification of fractal characteristics and homogenization scale in fractured aquifers

Stephane Bernard1, Anne Kaczmaryk1, Gilles Porel1 and Frederick Delay1*

1-UMR 6532, CNRS, University of Poitiers, Earth Sciences Building, 40 Avenue du Recteur Pineau, F-86022 Poitiers – France * Corresponding author

Hydraulic interference tests have been widely used in hydrogeology and petroleum engineering to assess the hydraulic diffusivity of underground reservoirs. The basic principle is to pump a constant flow rate in a well and to monitor the pressure transients in another well. In fractured rocks, the classical Jacob solution may appear unsuited because the pressure draw-down often increases more rapidly than linearly with ln t (Bogdanov et al., 2003). We propose an approximated analytical solution to radial flow in fractal fractured media combining a 2D Jacob solution with scaling laws for the permeability kf and the fracture porosity φf of the medium. It is written in dimensioned variables with a limited number of parameters that are fitted easily on experimental data by inversion using a Gauss-Newton procedure (Delay et al., 2004).

The analytical method and its inversion have been applied to a series of interference pumping

tests carried out over the hydrogeological experimental site in Poitiers (France). It encloses about 40 wells spatially set up as nested five-spot systems (a well at the center and four wells at the corners of a square) with lag distances between wells ranging from 50 to 300 m. The wells are bored in a vertically fractured Jurassic limestone of about 100 m in thickness that typically behaves as a regional confined aquifer.

An example of simulation plus inversion on experimental data is given in Figure 1. As stated

above, the behavior of pressure transients is fully compatible with hydrodynamic parameters kf and φf varying in time (or equivalently in space) during the propagation of the pressure depletion. These parameters are modeled as follows : kf = k0t-α; φf = φ0t-γ.

Several calculations have shown the model to be very sensitive to α which imposes its pre-

evaluation before identifying the other parameters. This is done by a log-log plot of a slightly modified derivative of the pressure draw-down with respect to ln t. After average oscillations at early times (fig. 1), the curve stabilizes and its slope gives the value of α (e.g., 0.288). Then, γ, k0, φ0 are directly identified on ∆p versus ln t by the Gauss-Newton algorithm. Note that the fractal dimension calculated from fitted values of α and γ is in the range [1.90, 1.99] which is in agreement with values encountered in 2D regular percolation networks.

0.01

0.10

1.00

100 1000 10000 100000 1000000

time (s)

dP /

d ln

t

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

10 100 1000 10000 100000 1000000

time (s)

P (1

05 Pa)

Figure 1. Inversion of an experimental interference pumping test. Left – Pre-evaluation of α with the log-log plot of the modified derivative p ln t∂∂∆ as a function of t. The slope gives the value of α. Right – Best fit for the pressure draw-down versus ln t: dots = experimental curve, solid line = solution from the analytical approximation.

More than 120 experimental draw-down curves have been inverted (Bernard et al., 2005). Each curve is fitted with its own parameters k0, φ0, α, γ and these parameters can be reinterpreted in terms of averaged long-term permeability and porosity using kf 0 f 0(t) k t ; (t) t−α= φ =φ − γ with, e.g., t = 48 hours. The evolution of kf(48 h) and φf(48 h) with the lag distance between observed and pumped wells is reported in Figure 2.

0 50 100 150 200 250 3000.0

1.0x10-10

2.0x10-10

3.0x10-10

4.0x10-10

5.0x10-10

6.0x10-10

7.0x10-10

8.0x10-10

9.0x10-10

1.0x10-9

0 50 100 150 200 250 300

1.0x10-12

2.0x10-12

3.0x10-12

4.0x10-12

5.0x10-12

6.0x10-12

7.0x10-12

φ f(t =

48 h

)

distance (m)

k f(t

= 48

h)

distance (m) Figure 2. Long-term permeability kf (t = 48 hours) and porosity φf (t = 48 hours) deduced from the fitting of

and f 0k k t −α= f 0 t− γφ =φ over the whole series of hydraulic tests carried out in this work. kf is almost constant whatever the lag distance between pumped and observed wells. A pseudo-homogenization scale is reached for distances of 100-200 m. φf shows a clear decrease with the lag distance between pumped and observed wells. A homogenization scale might be reached for distances beyond 300 – 500 m only

The long-term permeability remains almost constant with the lag distance between wells whereas the fracture porosity decreases. The permeability would rapidly reach a homogenization scale, even if the medium is fractal, provided the pressure depletion has propagated enough through the fracture field. On the other hand, there is a clear decreasing power law for the fracture porosity φf and its homogenization scale is probably much larger. Note that the hydraulic diffusivity, proportional to the ratio kf/φf, increases with the lag distance between wells whereas it is assumed to decrease in fractal media. This discrepancy motivated the search for an alternative to the fractal interpretation. A dual-media approach (2k, 2φ) in radial coordinates has been tested. The first runs show that several pressure transients monitored on the site can be fitted with this model while keeping its parameters constant over space. Moreover, the interpretation with a fractal approach of draw-down simulations from a (2k,2φ) model shows the macroscopic hydraulic diffusivity to increase with the lag distance between wells. Thus, it can be questioned on the hydraulic behavior of the site : a mix between a fractal fracture network and a dual media ? Further works are needed to get the answer, all the more it is not proven that radial flow is able to grasp the scaling laws predicted by theory and based on laminar parallel flow. Bernard, S., Delay, F. and Porel, G., August 2005. A new method of data inversion for the

identification of fractal characteristics and homogenization scale from hydraulic pumping tests in fractured aquifers. Submitted to J. Hydrol.

Bogdanov, I.I., Mourzenko, V.V., Thovert, J.-F. and Adler, P.M., 2003. Pressure draw-down well tests in fractured porous media. Water Resour. Res., 39(1): 1021-1040.

Delay, F., Porel, G. and Bernard, S., 2004. Analytical 2D model to invert hydraulic pumping tests in fractured rocks with fractal behavior. Geophys. Res. Letters, 31(16).

ROLE OF SORPTION PROCESSES IN REACTIVE TRANSPORT

Philippe BEHRA

Institut National Polytechnique Toulouse – Ecole Nationale Supérieure des Ingénieurs en Arts Chimiques et Technologiques – Laboratoire de Chimie Agro-Industrielle – UMR 1010 INRA/INP-ENSIACET – 118, route de Narbonne – 31077 TOULOUSE Cedex 04 (France) Email: [email protected]

In ground water, transport of contaminants are controlled by exchange processes,

which occur at very small interfaces. The role of interfaces is due to their reactivity and their

very high surface areas per volume unit, which regulate many natural reactions at the water-

mineral or living organism-water interfaces. Solid surface interactions play an important role

in the geochemical fate of trace elements, such as the regulation of heavy metal or metalloid

ions in waters and soils, sorption processes, dissolution processes and leaching of acidic

soils, organic matter oxidation catalysed by Fe and Mn oxides….

The fate of contaminants in an open system can be modelled by a mass balance equation, i.e.

a transport equation, which is the sum of two terms which depend on space (x,y,z) and time

(t), the so-called transient advective-dispersive mass transport equation. From this equation,

in homogeneous saturated porous media, we can define the retardation factor, R, which is

often used to describe the relative motion of reactive solutes with respect to inert or

conservative tracer:

R = 1 + ρs ( ∂ Cs/ ∂ [C]) (1)

([C]: aqueous concentration [mol L–1]; Cs: solid phase concentration [mol g–1]; ρs: apparent

solid density depending on pore volume)

In definition of retardation factor, equation (1), appears ( ∂ Cs/ ∂ [C]). It is thus

necessary to know the relationship between Cs and [C] to be able to solve reactive transport

equation. Different sorption mechanisms are proposed in literature depending on the type of

solid phase, reactive solutes and scales… Moreover, the behaviour of reactive solutes or non-

conservative chemicals is controlled by the speciation in the overall given system, i.e. by the

chemical composition (redox, pH, ligand concentrations…) and by the physico-chemical

properties (ionic strength, surface charge, hydrophilic-hydrophobic properties of chemicals

and minerals…) of the different phases.

In this talk, after some definitions and development on reactive transport, we will

compare the behaviour of different heavy metals such as cadmium, copper, lead, mercury,

silver… and organometallic compounds such as the organotin products in the presence of a

natural aquifer quartz sand at various scales, in order to show the role of the aqueous and

surface complexation of the different compounds which are present both in the aqueous

phase and at the solid-liquid interfaces. The aim is to show why it is necessary for modelling

transport of trace elements to have a better knowledge of the properties and the composition

of potentially reactive surfaces and of their evolution in space and time due to the presence

of contaminants.

ACKNOWLEDGEMENTSParts of the work were performed either at the Institut de Mécanique des Fluides et des Solides (UMR 7507 Université Louis Pasteur-CNRS, Strasbourg – France) and/or at the University La Sapienza in Rome. These works were supported by Deutsche Forschungsgemeinschaft, CNRS (« Programme Environnement, Vie & Sociétés »), CONACYT, Université Louis Pasteur, Strasbourg, Région Alsace, Galileo programme.

UNDERSTANDING FLOW AND MASS TRANSFER THROUGH AN ENHANCED GEOTHERMAL SYSTEM (EGS) CREATED INTO A DEEP FRACTURED BASEMENT IN THE RHINE GRABEN. Dominique BRUEL, Ecole des Mines de Paris, Fontainebleau, [email protected] Clément BAUJARD, Ecole des Mines de Paris, Fontainebleau, [email protected] Geothermal energy stored in crystalline rock in deep underground is widely available for massive electricity production. To reach its full potential, the extraction of heat over a prolonged period may need to create an engineered geothermal reservoir, and past experiences have shown how geology and natural conditions had to be taken into consideration. The initial idea to develop closed systems, using parallel artificially created fractures by hydraulic fracturing has not been successful. An advanced concept consisting in massive hydraulic stimulations of the pre-existing fracture pattern to increase the overall permeability of the rock mass resulted in a more volumetric flow through the rock. Long term circulation tests using a multiple well system highlighted the difficulties in controlling the geometry of the created reservoir and in insuring a satisfactory mass balance in between inlet and outlet. Tapping into faulted areas, indeed a far more restricted target, allows however to combine some favourable conditions: - the natural fracture and fault system has a significant permeability, - fractures are likely to be easily stimulated by hydraulic experiments due to the tectonic

history, - the open nature of the systems allows a potential fluid resource and enable the use of

down hole pumps in the production wells, so that the global fluid mass balance can be achieved.

These ideas are the main rationales for the European research project running at Soultz sous Forêts (France), in the Rhine graben geological context. Progress and lessons learned, accumulated since the project started in 1987, led to the present day reservoir, developed at a depth of 5 km. It consists into one central injection well, GPK3, and two deviated production wells, about 650 m apart to the north (GPK2) and to the south (GPK4). Temperature at casing shoe is about 180°C and is nearing 200 °C at the bottom of the wells. Our work focuses on the quantitative understanding of a series of hydraulic experiments related to the sequence of development of this deep reservoir, using a numerical tool integrating a hydro-mechanical coupling in fractured systems. Our first purpose is to explain the recorded data obtained successively during the stimulation of the wells GPK2 (2000), GPK3 (2003) and GPK4 (2004) by simulating the fluid circulations induced by injections at high flow rates. The success of these hydraulic stimulations requires the shearing a great number of fractures. Within a fracture, sliding motion is accompanied by some dilation, which improves the hydraulic permeability, and by some acoustic emission, which helps to define the spatial location of the rupturing process. All of these measurements, pressure records, flow logs, migration rate of the induced seismicity, extension of the seismic cloud, breakthrough of the pressure wave at the other wells, breakthrough of the injected mass at other wells, are used as constraints to calibrate the numerical models. A normal faulting stress regime is assumed at the reservoir depth. To account for the density contrast in between injected cold fresh waters and the formation fluid which is a heavy brine, a two immiscible fluid approach is implemented in the numerical code Fracas, available at the Centre of Geological Informatics of the Paris School of Mines. This code takes advantage of a Discrete Fracture Network formulation and has already been used to replicate the thermo-hydro-mechanical behaviour of an earlier reservoir, developed at

same location at the depth of 3 km and circulated at a rate of 25 kg/s during 4 months in 19971. It was consequently decided to enhance this code in order to allow the simulation of multiphase flows of fluids of different physical and thermodynamical properties and to use it to underline eventual density driven flows. The formulation and the numerical resolution of multiphase flows in discrete fracture networks is presented together with an evaluation of the accuracy of the IMPES algorithm, using two documented test cases2 (the Buckeley-Leverett test case and the problem of the coastal aquifer). The set-up of the geometrical model of the fracture network of Soultz-sous-Forêts, based on field observations, will be precisely described in order to quantify the influence of the injected fluid density on hydraulic stimulations. The main outcome of this modelling effort of the three stimulation phases is the description of a possible geothermal reservoir with enhanced properties, which in turn can be used in a second set of calculations to predict the hydraulic behaviour of the global 3 well system during subsequent circulation tests. A long term low rate, 15 kg/s, fully automated and balanced experiment is underway at Soultz site. Figure 1 shows the calculated saturation distribution in fluid 1 (the injected fluid at the central bore hole) at the centre of the fractures of a potential fracture network, after a period of 1 week of re-injection. This first result compares favourably with the most recent in situ observations, demonstrating that injected fluid is recovered after 3.5 days into GPK2 and that about 70% of the fluid is produced by GPK2, while ~30% comes out from GPK4. The results obtained are of first importance to feedback the discussions on the internal structure of the reservoir and prepare further plan to optimize any additional stimulation experiment.

Figure 1: Saturation in fluid 1 (injection fluid), calculated at the centre of the fractures forming a possible network. after 7 days of re-injection in the central well (GPK3). Note that the saturation is maximum (red=1) close to the injection well, and minimum (white=0.01) close to the GPK2 well (green trace), where the breakthrough occurs after 3.5 days. No breakthrough had happened in GPK4 (red trace), to the south after 1 week of circulation.

1 Bruel (2002), Oil and Gas Science and Technology, vol57 n°5, pp.459-470 2 C.Baujard and D.Bruel (2004) 18 SWIM Ground Water and Saline Intrusion, Cartagena, Spain, (Ed. Araguas, Custodio and Manzano) IGME, pp. 63-76.

1

On Constrained Minimisation Techniques For Solving General Geochemical Reactive Transport Problems

M. A. Sbai*

1French Geological Survey (BRGM), 3, Av. Claude Guillemin, BP 6009, 45060, Orléans cedex 2

Abstract In the framework of reactive transport modelling several approaches have been proposed. Many research efforts are devoted to families of sequential operator-splitting and fully (or global) implicit methods coupling solute transport and geochemical equations solvers respectively. However, there is little works advocated to numerical solvers genuinely efficient for this task. We have developed a new geochemical engine based on robust non-linear optimisation technology for mixed algebraic-differential equations stemming from thermodynamic equilibrium and kinetic processes. It is shown how our constrained minimisation algorithm for solving thermodynamic equations eliminates the need for the search process of dissolved or non existing mineral phases, and how general geochemical constraints are fetched directly into it. In the paper we also discuss predictor-corrector techniques for solving the coupled set of geochemical equations. These techniques contribute significantly to the general efficiency of reactive transport simulations because the number of iterations is significantly reduced by comparison to the Newton-Raphson method. Another numerical advantage of this approach is guaranteeing total concentrations positivity of all species, even those present in small amounts. To extend this feature globally, the geochemical engine is coupled with a higher-order Godunov method for finite volume computations in non-rectangular 3D grids. The C/C++ computer package featuring these new developments is designed in a flexible manner allowing it to be plugged with other available solute transport packages. It has gone through intensive benchmarks and comparisons with other reactive transport models. Finally, we close the paper by an application of the computer model to CO2 injection in a deep French geothermal reservoir surrounding Paris. The main aquifer strata known as the ‘Dogger’ has an active record of geothermal development since the seventies. Currently, it is one possible target for large scale injection of CO2 in its supercritical form. We guided two sets of simulations, near the wells for typical injection periods of 50 years, and in the far field for a few thousands years, to show the contrast in the flow and geochemical transport processes throughout. The contrasted behaviour suggests that it is possible to spatially decouple modelling near and far from the injection wells in the Dogger context.

* Corresponding Author: [email protected], +33 (0)2 38 64 35 27

Permeability of porous media containing fracture networks with a power-law size distribution

V.V. Mourzenko (1), I. Bogdanov (2), J.-F. Thovert (1), P.M. Adler (3)

(1) Laboratoire de Combustion et de Détonique, SP2MI, BP 30179, 86962 Futuroscope-Chasseneuil, France

(2) Institut de Physique du Globe de Paris, tour 24, 4 place Jussieu, 75252 Paris Cedex 02

(3) Laboratoire Sisyphe, tour 24, 4 place Jussieu, 75252 Paris Cedex 05, France

Fractures and fracture networks strongly influence the flow properties of many geological formations, such as aquifers and oil reservoirs. The determination of their effective permeability is addressed here by solving numerically the flow equations in a three-dimensional discrete description of the fracture network and of the embedding matrix. The methodology follows that of Bogdanov et al. (2003). The flow is governed by Darcy's equation in the rock matrix, with permeability Km ; it is also described by a two-dimensional Darcy law in the fractures, which are plane polygons randomly located and oriented in space with transmissivity σ. In this earlier work, the fractures were assumed to have all the same size R. The numerical solution of the flow equations is conducted on a tetrahedral mesh with conforms with and explicitly contains the fractures. The reduced macroscopic permeability mKKK =' of the fractured medium was shown to depend on two parameters only, the dimensionless fracture conductivity mRKσσ =' and a dimensionless fracture density ρ' based on the concept of excluded volume. However, real fracture networks are generally polydisperse, with a fracture size distribution which very often obeys a power-law of the form ( ) aRRn −= α . The properties of such networks in an impermeable matrix have been studied by Mourzenko et al. (2004, 2005). A generalized dimensionless fracture density ρ'3 which involves a shape factor was introduced and proved to be an adequate percolation parameter. In these terms, the critical density is nearly invariant, over a wide range of shape and size distribution parameters. A general expression for the network permeability was also proposed, which is the product of the volumetric surface area, weighted by the individual fracture conductivities, and of a fairly universal function of the dimensionless network density, which accounts for the influences of the fracture shape and size distributions. The present work is the extension and synthesis of these earlier studies and it finally addresses the full complexity of flow in permeable fractured media, by accounting for the matrix flow and for the size polydispersity of the fractures. The data in Figure 1, for very conducting fractures and a=2, show the importance of the percolation status of the fracture network. Two branches are observed for low and high densities which correspond to situations with percolating and non percolating networks, and which can be modeled by using our earlier results. A transition takes place in the intermediate range, where both situations coexist. Figure 2 contains results for two exponents a=1.5 and a=2.5, and for various fracture shapes. It illustrates the success of the dimensionless parameters in unifying the data over a wide range of fracture shape and size distributions.

Figure 1:

The fractured medium effective permeability 'K as a function of the dimensionless density ρ'3 for networks of hexagonal fractures with σ' =104 and a=2. Dots are the results for individual realizations. Lines are averages over the cases of percolating (blue) or non percolating (red) networks. In the middle range, the green line is the overall average.

Figure 2:

The fractured medium effective permeability 'K as a function of the dimensionless density ρ'3, when σ' =104. Lines are averages for networks of hexagonal fractures with a=1.5 (______) or 2.5 (- - -). Symbols are for triangles ( ), squares ( ), 20-gones ( ) and rectangles with aspect ratio 4 ( ) with a=1.5, and for squares with a=2.5 ( ).

References: Bogdanov I., V. Mourzenko, J.-F. Thovert, P.M. Adler – "Effective permeability of fractured porous media in steady-state flow", Water Resourc. Res., 39, 1023, doi:10.1029/ 2001WR000756 (2003).

Mourzenko V.V., Thovert J.−F., Adler P.M. - "Macroscopic permeability of three−dimensional fracture networks with power−law size distribution", Phys. Rev. E 69, 066307 (2004)

Mourzenko V.V., Thovert J.−F., Adler P.M. - "Percolation of three-dimensional fracture networks with power-law size distribution", Phys. Rev. E, 72, 036103 (2005).

Scaling kinetic reactions in heterogeneous media : from laboratory – to pilot – to field scale

ATTEIA OLIVIER & GUILLOT DE SUDUIRAUT CHARLOTTE Institut EGID, 1 Allée Daguin 33607 Pessac Cedex [email protected] Abstract . Biodegradation reactions are more and more investigated due to their importance in the field of soil and ground water remediation. However most of the field transport studies show that the fickian dispersion occurs only at local scale and that the global dispersion in the field is mainly linked to heterogeneity in velocity distribution. After a brief description of the major reactions that occur at contaminated sites, and how they can be treated in reactive transport problems, several laboratory or pilot scale experiments illustrate the concept of dispersivity and help to give the most common values. Then we present a simplified approach to generate 3D random concentration field from a parallelepiped source, that allows for calculation of reactions. It is shown that the equivalent dispersivity of a reactive plume is significantly smaller than the one for the tracer. Moreover a pseudo first order rate of degradation arises from an instantaneous reaction scheme that is due to the heterogeneous distribution of permeability. Concerning the chlorinated solvents behaviour, it is shown from experimental data that the separate sources of organic substances and chlorinated solvents that may prevail in the field will have dramatic influence on the field scale degradation kinetics.

Abstract Submission International Ground Water Symposium IAHR, 12 -16 June 2006, Toulouse, France

“Ground Water Hydraulics in Complex Environments”

Impact of sedimentary structures with inclined couplets on macrodispersion in gravel aquifers

Fritz Stauffer Institute of Hydromechanics and Water Resources Management

ETH Zurich, CH-8093 Zurich / Switzerland [email protected]

Investigations in outcrops of heterogeneous gravel deposits in North-Eastern Switzerland (Jussel et al., 1994) revealed that distinct sedimentary structures appear as more or less horizontal lenses and layers within extended patches of bimodal gravel deposits. However, two out of the observed facies element types, the open framework/ bimodal structures and part of the combined structures of grey gravel (bimodal gravel without silt) and brown gravel (bimodal gravel with silt), consist of inclined couplets. They are the results of flu-vial processes with the filling of troughs and channels. Together, these two facies elements account for about 10% of the sediment volume. Results of three-dimensional stochastic facies type transport modelling in a uniform mean flow field over a distance of 100m of a saturated aquifer were compared with the results of analytical unimodal and bimodal sto-chastic macrodispersion models (Stauffer and Rauber, 1998). In the latter the inclination of couplets is neglected. The bimodal approach clearly showed better correspondence with the numerical facies model results. This was primarily due to the pronounced bimodal probability density of the hydraulic conductivity values, which was observed in the field. However, differences were still present and striking. Can they be attributed to the inclina-tion of the sediment couplets? This inclination manifests itself as a strong anisotropy of the hydraulic conductivity tensor with inclined principal axes of the facies element. In this paper the role of sedimentary structures with inclined couplets is investigated by a series of three-dimensional numerical flow and transport experiments. Single lenses of sedimen-tary structures with inclined principal axes of the hydraulic conductivity tensor are inves-tigated as well as groups of lenses, which are imbedded in both homogeneous and hetero-geneous background matrix material. Particle tracking on models with presence of such structures shows pronounced lateral (horizontal and vertical) deflexions thus highly affect-ing the development of the macrodispersion tensor. These observations can better explain the results of the stochastic facies modelling of Stauffer and Rauber (1998) in a qualitative and also in a semi-quantitative manner. The results enable a better understanding of the transport processes, which are responsible for the macrodispersion effects in heterogene-ous gravel aquifers at the local scale.

Identification of groundwater endmembers: implications for the impact of liming on groundwater

Christina Weyer1,2, Gunnar Lischeid1,3, Luc Aquilina4,5, Anne-Catherine Pierson-Wickmann4,6

1 University of Bayreuth, Chair of Ecological Modeling, Dr.-Hans-Frisch-Straße 1-3, 95440

Bayreuth, Germany 2 [email protected], 3 [email protected]

4 University of Rennes 1, Géosciences Rennes, Campus de Beaulieu, CS 74205, 35042 Rennes cedex, France,

5 [email protected], 6 [email protected] In catchments with base poor bedrock, like granite, water-rock interactions are often insufficient to counteract the acidification of the catchments caused by acidifying deposition. To attenuate the harmful effects of acidification the catchments are often limed. While the effect of catchment liming on forest soils and stream water has been studied by many authors (e.g. Gunn et al., 2001; Frank and Stuanes, 2003; Lorz, Hruska and Kram, 2003), there are very few studies on the impact of liming on groundwater quality and thus on its buffering capacity. To study that phenomenon, we analysed 87Sr/86Sr ratios, major and trace element ratios in different water components, in major minerals and in leaching samples of dolomite and crushed drilling core samples (soluble trace minerals). Our study took place in a forested granitic catchment that was limed several times in the past.

Figure: Keeling diagram (87Sr/86Sr ratios vs reciprocal Sr concentration) of different water components, major and trace minerals. 1 from Irber et al. (1997) Stream water samples reflected the contribution of three different groundwater endmembers (see figure). The endmembers were identified using a Keeling diagram (87Sr/86Sr ratios vs reciprocal Sr concentration) and by a principle component analysis. The deep groundwater endmember with high 87Sr/86Sr ratios and high Sr concentrations could be ascribed to feldspars and biotite, the wetland groundwater endmember with low 87Sr/86Sr ratios and low Sr concentrations to precipitation. However, the endmember of the shallow groundwater outside the wetlands, characterized by low 87Sr/86Sr isotope ratios and higher Sr and Ca

concentrations than those of the wetland groundwater endmember, was not so clearly to identify. Higher Ca/Na ratios in all groundwater samples than that of plagioclase - often considered as the major geochemical Ca source in granitic catchment - indicated the presence of an additional Ca source. Trace minerals rich in Ca like apatite are discussed in literature to play a major role in the chemical weathering of granitoid rocks (White et al., 1999; Oliva et al., 2004) and consequently for the chemical composition of groundwater in silicate catchments, despite their minor fraction in the bedrock (Clow et al., 1997). Irber et al. (1997) found apatite to be easily soluble, especially at acid conditions. However, 87Sr/86Sr ratio analysis of our leaching experiment with crushed drilling core samples did not correspond with apatite dissolution. It is more likely, that liming is the endmember of the shallow groundwater endmember outside the wetlands. Because only the areas outside the wetlands were limed, this hypothesis corresponded with higher Sr and Ca concentrations and higher Ca/Na ratios in the groundwater outside the wetlands than in the wetland groundwater. The impact of liming on deep groundwater was less pronounced than on shallow groundwater because of the application of lime on the surface. In addition, water-rock interactions and thus mineral dissolution are enhanced in deep groundwater due to its longer residence time. As there are no wetlands in the upper part of the catchment, runoff samples of a spring of that part of the catchment reflected only the contribution of the deep groundwater endmember and the shallow groundwater outside the wetlands. References: Clow D.W., Mast M.A., Bullen T.D., Turk J.T. (1997): Strontium 87/strontium 86 as a tracer of mineral weathering reactions and calcium sources in an alpine/subalpine watershed, Loch Vale, Colorado. Water Resources Research 33(6), 1335-1351 Frank J., Stuanes A.O. (2003): Short-term effects of liming and vitality fertilization on forest soil and nutrient leaching in a Scots pine ecosystem in Norway. Forest Ecology and Management 176 (1-3), 371-386 Gunn J, Sein R, Keller B, et al. (2001): Liming of acid and metal contaminated catchments for the improvement of drainage water quality. Water, Air and Soil Pollution 130 (1-4), 1439-1444 Irber W., Förster H.-J., Hecht L., Möller P., Morteani G. (1997): Experimental, geochemical, mineralogical and O-isotope constraints on the late-magmatic history of the Fichtelgebirge granites (Germany). Geologische Rundschau 86, Suppl.: S110-S124 Lorz C., Hruska J., Kram P. (2003): Modeling and monitoring of long-term acidification in an upland catchment of the Western Ore Mountains, SE Germany. Science of the Total Environment 310 (1-3), 153-161 Oliva P., Dupré B., Martin F., Viers J. (2004): The role of trace minerals in chemical weathering in a high-elevation granitic watershed (Estibère, France): Chemical and mineralogical evidence. Geochimica et Cosmochimica Acta 68 (10), 2223-2244 White A.F., Bullen T.D., Vivit D.V., Schulz M.S., Clow D.W. (1999): The role of disseminated calcite in the chemical weathering of granitoid rocks. Geochimica et Cosmochimica Acta 63(13/14), 1939-1953

Modeling of flow and solute transport in a heterogeneous

aquifer, Kerbala City, Iraq

N. Samani and A. Raoof Department of Earth Sciences, Shiraz University, Shiraz 71454, Iran,

[email protected]

An investigation was carried out to evaluate the hydrogeological conditions of the Center of Kerbala City (CKC) aquifer which encompasses the holy shrines of Al-Hussein and Al-Abbas. A three-dimensional finite-difference groundwater flow model was implemented to conceptualize flow conditions and establish the hydrogeological budget of the aquifer. The shallow and contaminated part of this aquifer comprises serious environmental problems that accelerated during the golf war. The city, as well as the shrines, has suffered from these problems for thirty years. The groundwater system consists of an upper phreatic, middle semi-pervious, and a lower semi-confined (leaky) aquifer, respectively of about 5, 2, and 30 meters average thickness. Chemical characteristics and fluctuations of groundwater levels, as well as the numerical simulation of the aquifer, indicate that the upper layer is recharged by leakage of drinking water pipe network, sewers, and septic tanks. Also, more than 80 % of the recharge of the upper layer is transmitted to the lower aquifer via the middle layer, while the rest discharges into the drains surrounding the study area. Three principal mathematical models were constructed and several add-on packages were implemented. MODFLOW program was applied to simulate the flow and hydraulic condition of the system. Then, The efficiency of various dewatering schemes such as pumping wells, Horizontal drains, and cut-off walls (diaphragms) were evaluated by flow model around the holy shrines. Based on the simulation results, a general 50% reduction of the Annual leakages will result to a desired water table drawdown of 1m. The drains, if constructed, would lower the water table just a few meters away from the drain axis, and therefore, does not meet the needs. Simulation by MODFLOW indicates that lowering the water table below the shrines can be carried out by pumping wells in the lower layer rather than the drains in the superficial layer and cut-off walls. Therefore, pumping the lower aquifer around each shrine is suggested to reach the desired drawdown and maintain the lowered water table. The GWM (Ground-Water Management) Package and the Modular Groundwater Optimizer (MGO) Code were used to optimize location and number of pumping wells and, time and rate of pumping. The first package is a gradient base optimization method while the second package

uses the Genetic Algorithm. The results of the MGO code were more reliable than that of GWM Package. The Subsidence and Aquifer-System Compaction (SUB) Package was used for estimating the subsidence due to water table decline. The calibrated flow model and the MT3DMS code were jointly used for simulating contaminant transport based on the advection and dispersion in groundwater system. Mass transport simulation by MT3DMS indicates that, if chloride input to the aquifer ceased, after 20 years a reduction of 82% of the chloride concentration in the aquifer will occur. Contaminant migration from some alternative point sources was also simulated.

!" #$%& '()*% +,,-

./ 012 34/0250 46 /73829: :;<= 34/0:48 >2 <07?@ 012 A219B;4: 46 9 34/09C ;/9/0 ;/ 9 <9/?@D389@2? <4;8 E F2930;B2 <7:6932 34/09C ;/90;4/ ;< 9 34CG8252 G12/4C 2/4/ H C9/@ C780;34CG4/2/0312C ;398 :2930;4/< 39/ 4 337: 4/ 9 C780;DG19<2 I4> E J1;< G9G2: ?;<37<<2< 012 ;/02:930;4/ A20>22/0:9/<G4:0 9/? 312C ;398 G:4G2:0;2< E K4: 01;< H >2 G70 47: ;/02:2<0 4/ 012 0:9/<G4:0 46 L2<;7C 90B2:@ 84> 34/32/0:90;4/ ;/ 9/ ;?298;M2? G4:47< C2?;9 34CG4<2? 46 K4/09;/2A8297 <9/? > ;01 9 62>G2:32/0< 46 .88;02 ?7 N7@ <907:902? 46 >902: EJ12 C 93:4DG4:2 <3982 G:4A82C ;< C4?2882? A@ 9 O:9;/ 46 ;88;02 <7::47/?2? > ;01 >902: PG19<2αQ ;/ >1;31 9?B230;4/ 9/? ?;R<7;4/ 4 337:

∂cα

∂t= −~∇ ·

[

~uαcα − Dα~∇cα

] PSQJ12 389@ O:9;/ PG19<2

βQ ;< 98<4 9 G4:47< C 2?;9 E T4:G0;4/ 46 L2<;7C 4/ .88;02 ?7 N7@ ;<9 >288 =/4>/ :2930;4/ H 012:C4?@/9C ;3 9/? =;/20;3 <07?;2< 19B2 A22/ ?:;B2/ 04 319:9302:;M201;< 1202:4O2/247< :2930;4/ E U0 012 C ;3:4DG4:2 <3982 H 012 <4:A;/O C 239/;<C 2 ;< C 4?282? A@ 9?;R7<;B2 G:4 32<< 347G82? > ;01 9 <4:A ;/O :902 4/ :2930;B2 <7:6932<∂s

∂t= k+c − k−s

PVQ>12:2

c;< 012 A78= 34/32/0:90;4/ 9/?

s012 <4:A 2? 34/32/0:90;4/ EU 89:O2 :9/O2 46 312C ;398 9/? O24C 20:;398 G:4G 2:0;2< 19B2 A22/ 25G84:2? 04 ?202:C ;/2 012?4C 9;/ 46 B98;0@ 46 012 2W7;8;A:;7C 1@G4012<;< P14C4O2/247< <48702 34/32/0:90;4/ 90 012 C ;3:4DG4:2 <3982 Q E X@ 7<;/O 012 O2/2:98 B487C 2 9B2:9O;/O 7G<398;/O 0231/;W72 >2 C4?28 012 0:9/<G4:09/? <4:G0;4/ ;/ 012 389@ A@ 9 :209:?2?D?;R7<;B2 2W790;4/ 90 012 C93:4DG4:2 <3982 E

(1 + Rβ)∂cβ

∂t= ~∇ ·

[

Dβ~∇cβ

] PYQJ1;< 0>4D2W790;4/ C4?28 O;B2 :;<2 04 9 34CG20;0;4/ A20>22/ 9?B230;4/ 9/? ?;R7<;4/ ;/ 012C 93:4DG4:2 E X@ <48B;/O 012 G4:2D<3982 G:4A82C >2 39837 8902 2R230;B2 G:4G 2:0;2< 9< ?;<G2:<;4/ 4:25319/O2 :902 A20>22/ I7;?2 9/? <48;? G19<2 E X@ ;/3:29<;/O 012 N23820 /7CA2: 9/? :209:?90;4/69304: H ?;R2:2/0< G9002:/< 39/ A2 4A<2:B2? 9/? 9 34::2890;4/ A20>22/ 0:9/<G4:0 9/? 312C ;398G:4G2:0;2< 647/? E

S

A new approach to study solute mixing in heterogeneous porous media Paulo Herrera and Roger Beckie

Department of Earth and Ocean Sciences, The University of British Columbia 6339 Stores Road, Vancouver, British Columbia

CANADA , V6T 1Z4

A good understanding of solute mixing in natural geological formations is important for many practical applications. For example, reaction rates of natural biodegradation processes and dilution of contaminants in aquifers are controlled by the rate of mixing. In porous media solute mixing is produced by mass transfer processes that occur at the pore scale. Under certain conditions, continuum or large scale heterogeneity of the medium can enhance those pore scale mechanisms. Thus, solute mixing in heterogeneous porous media is produced by the combined action of molecular diffusion, local scale and continuum scale flow velocity heterogeneity. So far, theoretical and numerical studies about mixing have had problems to model mechanisms that take place at such different scales. In most cases solute mixing is modeled by using a unique coefficient of dispersion that takes account for all sub-grid scale heterogeneity. This approach has two important drawbacks. First, it overestimates physical mixing and reaction rates. Second, it makes impossible to distinguish the contribution of local and macro scale heterogeneity to the overall solute mixing. We developed a new numerical approach based on particle methods that overcomes those issues. In this model heterogeneity at the local and macro scales is explicitly included by using two different dispersion coefficients. That makes possible to study separately the effect of local and macro scale dispersion on solute mixing. We have used this new numerical method to study mixing mechanisms under different scenarios, to quantify the effect of local and macro scale heterogeneity in solute mixing, and to verify previous theoretical investigations.

Development of Simplified Scaling Relationships for the Estimation of Subsurface Biological Parameters Using Moment Analysis

Mohamed M. Mohamed1 and Kirk Hatfield2

Abstract

Modeling contaminant transport and biodegradation in the subsurface requires estimation

of flow, transport and biological parameters. These parameters are often determined in

the laboratory; and then transformed to reflect field-scale processes. Many field tests

have been developed to evaluate field-scale (macro-scale) flow and transport parameters.

Estimation of biological parameters, on the other hand, is still limited to laboratory-scale

experiments. Using laboratory-based parameters in numerical models to simulate field

scale processes could result in false predictions of contaminant distribution in the

subsurface; and, therefore, inaccurate remediation action. Scaling relationships between

spatial moments of both contaminant plume and bacterial populations are developed in

this paper. These relationships are then used to estimate the biological parameters such as

the microbial maximum growth rate (µmax), the substrate half saturation coefficient (KS),

and the yield coefficient (YS). The finite element numerical model METABIOTRANS is

used to test these relations through several testing problems. Results from these problems

indicate the validity of such relations especially as the time interval between field

sampling events decreases.

1Civil and Environmental Eng. Dept., United Arab Emirates University, PO Box 17555, Al-Ain, UAE. 2 Civil Engineering Dept., University of Florida, 345 Weil Hall, P.O. Box, 116580, Gainesville, FL 32611, USA.

Keynote Lecture IAHR-GW2006, Toulouse 12-14 June 2006

1

ANCIENT AND MODERN HYDROLOGY: THE COMMON GROUND AND A FEW CHALLENGING TASKS

Gedeon DAGAN

Faculty of Engineering Tel Aviv University

Israel

The presentation is inspired by a rich mosaic unveiled recently at the Tzipori archeological site in Israel. It depicts an allegoric festival, celebrating the rising of the Nile. The detail of interest is a symbolic representation of a Nilometer, a pillar with engraved units that served to measure and record the level of the Nile. The device stimulated my interest in exploring the similarity between ancient and modern hydrology. The Nile River has always been the backbone of Egypt and without its annual inundation, ancient Egypt would never have come into being. Priests made painstaking records of the rising and falling of the sacred river, to predict the flooding of the Nile with great accuracy. Nilometers could be traced back to 3000 BC (first dynasty) and records were used for prediction that served the Pharaoh’s administration for policy making (taxation and storage). The priests that analyzed and interpreted the records are the ancient hydrologists, making hydrology one of the oldest professions in the world. An indirect testimony of the role played by the river regime and its forecast is provided by the story of Joseph in the Bible. Joseph interpreted Pharaoh's dream and predicted a seven years period of abundance, followed by an equal one of draught. A recent analysis of modern records indicates indeed the presence of such a cycle (Kondrashov et al, Geophys. Res. Lett., 32, 2005). There are four constituents of ancient hydrology that are shared with modern hydrology: (i) Hydrology is a quantitative discipline that deals with data and with mathematical analysis; (ii) Hydrology is an applied science whose concerns and aims were related to the needs of society; (iii) Hydrology dealt with prediction under uncertainty and ancient hydrologists had to use some sort of sophisticated time series analysis in order to predict the occurrence of extreme events; and (iv) Hydrology is intertwined with economy, political and social issues as its predictions had a serious impact on the sustainability and wellbeing of society.

Keynote Lecture IAHR-GW2006, Toulouse 12-14 June 2006

2

These components are also shared by stochastic modeling of subsurface flow and transport, a discipline that has evolved in the last three decades. In this approach, models of water flow and contaminant transport account for the spatial variability of the properties of natural formations. This variability is subjected to uncertainty and it is modeled as stochastic. The common ground with ancient hydrology and a few challenging issues faced by stochastic modeling are discussed subsequently along the preceding four themes. (i) Stochastic modeling is highly mathematical and it uses the tools of statistical continuum theory, statistical physics, geostatistics, advanced numerical methods and risk analysis. A challenging task is communication with the more descriptive disciplines (geochemistry, geohydrology) and taking advantage of geochemical and geological data. A related task is to convey the results in a simple form to practitioners. (ii) Stochastic modeling is motivated by applications, primarily by the need to understand, predict and correct soil and groundwater pollution. The interest in exploring advanced mathematical and physical concepts notwithstanding, application to real life is of paramount importance (see Dagan, EOS, Transactions AGU, 83, 53, 2002).. (iii) By definition stochastic modeling deals with randomness and uncertainty. It is concerned with complex random spatial variability rather than time series. A challenge of a theoretical nature is to make an impact on related disciplines of basic science (heterogeneous media, transport by random fields). A related task is to strengthen the connection with risk analysis and econometrics. (iv) Contaminant transport and groundwater pollution are topics of great social impact. Like in Ancient Egypt, hydrological prediction has a significant impact on the well-being of society and these issues are of a social, political and economic nature. The challenge is to account for these issues in the developments of the theory and its applications.

The flow systems and the groundwater interactions in the multi-aquifer system of the Carmel Coast region- Israel Joseph Guttman Mekorot- Israel National Water Co. 9 Lincoln St. Tel-Aviv 61201, Israel, [email protected] Abstract The study area includes two aquifers that are in geological and hydrological connection. In the eastern part of the area the aquifer is the Cenomanian limestone-dolomite aquifer that builds the Carmel Mountain and westward to it, is the Pleistocene sandstone aquifer that is nestled between Mt Carmel and the sea-shore.

The exploitation from these two aquifers began to be extensively developed in the 1950’s for drinking and agricultural purposes. During the years, the salinity in these aquifers rises, while the most serious signs of acute water degradation were found in the Pleistocene sandstone aquifer. The water quality in the Pleistocene aquifer has declined so much in the interval that the water is no longer suitable for drinking purposes. Today the water from this aquifer is exploited primarily for refilling fishponds and for irrigation. Further salinity rises will limit it application to irrigation, for its salinity is approaching the maximum limits of several of the salt sensitive crops, such as the banana plantations. The Cenomanian limestone-dolomite aquifer has also exhibited a rise in salinity over time, but it has generally been much more subdued.

The Pleistocene sandstone aquifer is lying closest to the coastline has experienced the greatest deterioration in water quality. The common assumption was that the source of salinity is from sea intrusion. But, the result of a geohydrological investigation, employing a wide suite of isotopic techniques shows that the salinization and the continued degradation of the groundwater resources is not related to sea water intrusion, but, rather contaminants come from fish farming, fertilization and other agricultural practices carried out on, or immediately above, the aquifer sandstone along with sea spray contributions. The limestone dolomite aquifer is indirectly influenced by the same salinization processes when the groundwater flow direction is reversed during heavy summer exploitation from this aquifer

The groundwater in the Pleistocene sandstone aquifer is strongly enriched in δ18O and δ2H, compared to the groundwater of the limestone-dolomite aquifer Fig. 1). The main part of the isotopic deviation of the groundwater within the Pleistocene sandstone aquifer from the EMWL (the Eastern Mediterranean Water Line) is the result of the high amount of leakage from the fishponds into the Pleistocene sands. The fishponds undergo strong evaporation, which enriches the residual water in 2H and 18O. Wells, which have been drilled in close proximity to the fishponds, are chemically and isotopically identical to the fish pond water. The wells exploiting the Pleistocene aquifer are used primarily for making up the loss in the fishponds caused by evaporation and downward

1

mailto:[email protected]

leakage. This process repeatedly recycles large percentages of the same water. Recycling of the leakage leads to continue isotopic enrichment (Fig. 1) and salinization. The Pleistocene aquifer in particular suffers also from high concentrations of nitrate. After many years of cultivation and high fertilizer use, significant amounts of nitrogen may become stored in the unsaturated zone. This mass of nitrogen dissolves into the aquifer by winter rainfall and irrigation return flow.

The results of the investigation show that there are two aquifer water types, having considerable interchange between them (Fig. 1). Infiltration from the fishponds and the recycled irrigation water are the principal sources of recharge of the Pleistocene sandstone aquifer and they are more important than the direct rainfall over the aquifer or the lateral discharge from the limestone-dolomite aquifer. During the winter, the limestone-dolomite aquifer receives directly fresh rainfall water. The water level in the limestone-dolomite aquifer has a higher head than in the Pleistocene aquifer and groundwater flows from this aquifer into the Pleistocene aquifer in places where contact between the two aquifers exists. During the summer months, owing to massive pumping and the small aquifer storage, the dynamic water levels in the limestone-dolomite aquifer drop below the Pleistocene sandstone aquifer water levels. At this time the groundwater flow is reversed introducing higher salinity into the limestone-dolomite aquifer. This process recurs yearly. The end result leads to a slow and persistent rise in salinity.

The study illustrates the various dynamic processes taking place in this multi-aquifer system and the influence of human activities on the water deterioration. It is seen that if these processes continue, the salinization of the aquifers will accelerate and will be no longer suitable for drinking purposes and for irrigation of the salt sensitive crops.

-35.00

-30.00

-25.00

-20.00

-15.00

-10.00

-5.00

0.00

5.00

10.00

15.00

-7.00 -6.00 -5.00 -4.00 -3.00 -2.00 -1.00 0.00 1.00 2.00

δ18O (0/00)

δ2 H ( 0

/00)

Limestone Aquifer Pleistocene Aquifer

Fishpond Water

Mixing Line

EMWL (Gat and Dansgaard, 1972)

"Aquifer Water"

Evaporation

Fig. 1: The Oxygen-18 and Deuterium in the study area

2

C. Paniconi [email protected] Institut National de la Recherche Scientifique - Centre Eau, Terre et Environnement (INRS-ETE), Université du Québec, 490 de la Couronne, Québec, G1K 9A9, Canada M. Putti [email protected] Dipartimento di Metodi e Modelli Matematici per le Scienze Applicate, University of Padova, Via Belzoni 7, 35131 Padova, Italy Atmospheric controls and soil moisture inputs for a coupled model of surface-subsurface interactions We will describe recent work on a model that integrates land surface (1D overland and channel flow) and subsurface (3D soil and groundwater flow) processes via a coupling term that is computed as the balance between atmospheric forcing (rainfall and potential evaporation) and the amount of water that can actually infiltrate or exfiltrate the soil. This is one of several different approaches in recent hydrological modeling research that seeks to better represent the interactions in the atmosphere-land surface-subsurface continuum. The same diffusion wave equation is used for both hillslope runoff (rill flow) and channel flow, with the drainage network extracted from digital terrain data and with the distinction between these two surface flow regimes based on a threshold value of upstream drainage area. Renewed interest in physically-based, distributed hydrological models owes much to the promise of readily available observation data at relevant spatio-temporal resolutions (e.g., rainfall radar, active or passive microwave for soil moisture). To this end we will also describe the data assimilation module currently implemented in the coupled model and some of the issues relating to its use and further development, and some ongoing work to assess the potential of synthetic aperture radar (SAR) imagery from the Envisat and Radarsat satellites for soil moisture estimation.

Use of the barometric effect for the determination of the storage coefficient. Utilisation de l’effet barométrique pour la détermination du coefficient

d’emmagasinement

Frédéric LALBAT et Olivier BANTON (Laboratoire d’Hydrogéologie, Université d’Avignon, France)

Résumé

L’aquifère molassique du bassin de Carpentras (Vaucluse, France) renferme une nappe captive sollicitée par de nombreux forages à usage domestique ou agricole. Ces ouvrages ne permettent généralement pas la pratique d’essais de pompage pour accéder à l’emmagasinement. Pour déterminer ce paramètre hydrodynamique à des fins de modélisation, nous avons étudié l’effet de la barométrie sur la piézométrie. En effet l’emmagasinement est directement lié au coefficient d’efficience barométrique (rapport entre la variation de la pression atmosphérique et celle du niveau d’eau mesuré dans le forage). En six points de mesure répartis sur le bassin, le coefficient d’efficience barométrique a été évalué en suivant quatre méthodes : méthode de Clark, régression linéaire entre les mesures de barométrie et de piézométrie, régression linéaire entre les incréments et déconvolution. De ces méthodes on obtient différentes valeurs du coefficient d’efficience barométrique caractéristiques du comportement à court terme ou à long terme. L’enregistrement de la piézométrie sur les forages implantés dans une zone très captive de l’aquifère, présente de fortes perturbations diurne et semi-diurne dues aux marées terrestres. Ces composantes périodiques sont bien caractéristiques d’une nappe captive. Or le rapport des coefficients d’efficience barométrique à court et à long terme indiquerait un comportement de nappe libre. Cette apparente contradiction s’explique par le bruit induit par les marées terrestres et est à rapprocher des observations d’une étude récente d’un aquifère captif de Floride. Il apparaît donc que le calcul du coefficient d’efficience barométrique pour un système captif sensible aux effets de marée ne peut pas être effectué de façon automatique sans leur prise en compte. Notre approche s’intéresse aux développement de filtres et de techniques d’analyse spectrale.

International conference IAHR-GW2006 on "GroundWater in Complex Environments", 12-14/06/06, Toulouse (France).

MODELISATION DU FONCTIONNEMENT DES SURFACES CONTINENTALES AUX ECHELLES LOCALES A REGIONALES

SEVE : SOL EAU VEGETATION ENERGIE Valérie Borrell1, Isabelle Braud9,2, Gérard Dedieu1, Aaron Boone1,3, Flora Branger2, Yves Brunet7, Isabelle Calmet4, Nadia

Carluer2, André Chanzy6, Philippe Chibaudel1, Jean-Dominique Creutin9, Hendrik Davi1, Alexandre Ern5, Florence Habets3, Frédéric Hecht11, Jérôme Jaffré12, Philippe Lagacherie10 , Jean-Claude Menaut1, Patrice Mestayer4, Roger

Moussa10, Joël Noilhan3, Jérôme Ogée7, Albert Olioso6, Laurent Prévot10, Fabrice Rodriguez8, Marc Voltz10 1CESBIO UMR5126 Toulouse, 2CEMAGREF Lyon, 3CNRM Toulouse, 4LMF Ecole Centrale de Nantes UMR 6598, 5ENPC-CERMICS Marne-la-Vallée, 6INRA-CSE Avignon, 7INRA-EPHYSE Bordeaux, 8LCPC Nantes, 9LTHE Grenoble, 10LISAH

Montpellier, 11 INRIA Roquencourt, 12Laboratoire J.-L. Lions Paris Mots clés : Modélisation couplée cycles eau-carbone-azote. Hétérogénéités spatiales et temporelles. Bassin versant. Assimilation de données. Ecohydrologie.

I. CONTEXTE Les évolutions des activités humaines et les modifications possibles du climat auront très probablement des répercussions sur la ressource en eau, la production des cultures, les pollutions de l'environnement, les risques hydrologiques, la préservation de la biodiversité… Proposer une gestion durable du territoire qui anticipe ces impacts passe par une meilleure compréhension du fonctionnement des surfaces continentales. Cette amélioration peut, entre autre chose, venir d’une modélisation complète de ce fonctionnement qui tienne compte de la complexité de la surface et qui permette de simuler les processus couplés impliqués dans le cycle de l'eau, du carbone, de l'azote et des substances polluantes aux échelles de la parcelle, du paysage et de la région. L’objet de SEVE est de proposer une démarche devant aboutir à une telle modélisation modulaire et évolutive, en abordant en priorité l’échelle du paysage. Ce projet, à fort caractère pluridisciplinaire, est soutenu par le programme ACI/ECCO (INSU-Ministère de la Recherche). A ce jour, le Groupe SEVE a réalisé les cahiers des charges de la modélisation SEVE et de sa version 0, qui contiennent en particulier un dossier d’architecture logicielle destiné à définir les concepts qui présideront à la réalisation informatique du modèle, ainsi que la proposition de solutions scientifiques et techniques pour la réalisation de la version 0 et des travaux préliminaires à la construction de cette version.

II. SPECIFICATIONS ET ARCHITECTURE GENERALES DE SEVE L’outil de modélisation SEVE repose sur le principe que les surfaces continentales sont le siège, d’une part, de processus d’évolution des éléments constitutifs de ces surfaces et, d’autre part, de processus de transfert d’eau, d’énergie et d’autres substances qui permettent des échanges entre ces différents éléments. L’architecture générale de SEVE rend compte de ces deux types de processus en distinguant explicitement : • Les « Objets » qui représentent les différents éléments constitutifs des surfaces continentales et qui sont

caractérisés par des propriétés physiques et par des modèles d’évolution. • Les « Modules de transfert » qui conceptualisent les processus régissant les flux échangés entre les Objets. La modélisation des flux échangés par les modules de transfert s’appuie sur une discrétisation spatiale (ou maillage) des surfaces continentales qui doit être cohérente avec la représentation en Objets. Les unités ainsi obtenues sont appelées Unités Fonctionnelles (UF), qui sont éventuellement re-découpées en Unités de Calcul (UC) en fonction des contraintes numériques ou de l’hétérogénéité de certains paramètres ou variables. Cette architecture, issue de l’expression des besoins, s’articule autour de 4 grandes composantes : • La structuration des surfaces continentales, destinée notamment à identifier les Objets et le maillage sur

lesquels s’appuiera la modélisation (appelée « segmenteur ») • Les modules de transfert (dans la basse atmosphère, dans le sol/sous-sol (infiltration, écoulements sub-

surfaciques/souterrains…), verticaux de surface (interception, évaporation, transpiration…), latéraux concentrés/non concentrés en milieux urbains et naturels)

• Les modules de description de l’évolution du paysage rattachés aux Objets (fonctions de croissance, de développement et de structure de la végétation, fonction de production des eaux usées, fonction de production de chaleur anthropique, évolution du manteau neigeux, évolution d’une litière…)

• L’architecture permettant le couplage dynamique entre ces différents modules et son pilotage par un «Superviseur » (faire communiquer les différents modules entre eux (synchronisation des modules, échanges d’information à leur interface), adapter les maillages et interpoler temporellement les grandeurs échangées, assurer la convergence des processus couplés itérativement). De plus, le superviseur doit permettre à la modélisation d’être modulaire et évolutive. Le couplage réalisé doit être dynamique (la séquence des modules à exécuter est définie par les réactions des modules aux différents événements qui se produisent) et le superviseur

International conference IAHR-GW2006 on "GroundWater in Complex Environments", 12-14/06/06, Toulouse (France). doit gérer la parallélisation des calculs. Il devra reconnaître les langages de programmation Fortran, C et C++. Dans la mesure du possible, la modélisation SEVE devra être constituée de logiciels libres.

La version 0 est une première étape de construction de l’outil avec des objectifs plus modestes et des hypothèses simplificatrices. Elle simule les cycles de l’eau, de l’énergie et du carbone (flux lents et rapides) depuis l’échelle de la parcelle vers celle du paysage avec des processus dont les représentations peuvent être simplifiées.

Fig 1. Schéma de principe de l’architecture de la modélisation SEVE

Pour la deuxième étape de SEVE, qui traite de la construction de la version 0 de SEVE à l’échelle du paysage, différents travaux préliminaires sont en cours de réalisation dans différents laboratoires du groupe. Ils visent à évaluer la faisabilité de différents choix scientifiques et à analyser des solutions techniques possibles. Ces travaux présentent un état d’avancement différent suivant les disciplines. Un certain nombre d’entre eux seront présentés dans l’article associé (dont la modélisation couplée de l'évolution de la végétation et des transferts verticaux (Davi, CESBIO 2005) et la prise en compte de l’hétérogénéité spatio-temporelle de l’espace cultivé pour la simulation de crue (Moussa, Chahinian (MHYDAS), LISAH 2005).

III. CONCLUSIONS

Le premier bénéfice du projet SEVE résulte des échanges approfondis entre communautés voisines. Chacun ayant acquis une meilleure connaissance des travaux, des objectifs (et du langage) des autres, le groupe de travail est maintenant structuré et fonctionne efficacement. Dans le paysage international, le projet SEVE est original et novateur. Des avancées conceptuelles importantes ont été réalisées et des travaux en cours dans les laboratoires préparent la construction de la version 0. Cette version 0 devrait fournir la base pour développer les versions ultérieures. Au-delà de la V0, des développements informatiques spécifiques et lourds sont à prévoir dans les années à venir. Tous les problèmes scientifiques ne sont pas résolus, et le développement des versions successives de SEVE devra s’accompagner de recherches destinées à lever les verrous. Ceci concerne par exemple le passage raisonné de l’échelle du paysage à l’échelle de la région, les méthodologies de spatialisation et de désagrégation des variables d’entrée et des paramètres, la typologie des paysages et des bassins versants, la conceptualisation de processus sous-maille. Référénces citées dans le résumé R. Moussa, M. Voltz, P. Andrieux, 2002 : Effects of the spatial organization of Agricultural management on the hydrological behaviour of a farmed catchment during flood events, Hydrological Processes, 16(2), 393-412. N.Chahinian 2004. Paramétrisation multi-critère et multi-échelle d’un modèle hydrologique spatialisé de crue en milieu agricole. Thèse de doctorat de l’Université Montpellier II soutenue le 23 janvier 2004, 238 p. Référénce du projet V. Borrell, I. Braud, G. Dedieu, A. Boone, F. Branger, Y. Brunet, I. Calmet, M. Carluer, A. Chanzy, P. Chibaudel, J. D. Creutin, H. Davi, A. Ern, F. Habets, F. Hecht, J. Jaffré, P. Lagacherie, J. C. Menaut, P. Mestayer, R. Moussa, J. Noilhan, J. Ogée, A. Olioso, L. Prevot, F. Rodriguez, M. Voltz. (2005). Modélisation du fonctionnement des surfaces continentales aux échelles locales à régionales –SEVE : Sol, Eau, Végétation, Energie. Programme National coordonné ANR «ECCO» : premier colloque de restitution scientifique. 6/12/05. Action Thématique PNBC- Session II, p157-163.

International Groundwater Symposium on Groundwater Hydraulics in Complex Environments Toulouse, June 12-14, 2006

Multiscale Multiphase Mass Transfer in Porous Media: Integrated Numerical and Experimental Studies

Souheil M. Ezzedine1, Russell L. Detwiler2, Walt W. McNab Jr.3

1ERD, Lawrence Livermore National Laboratory, P.O. Box 808, L-530, Livermore, California, 94551; Telephone (925) 422-0565; Fax (925) 424-3155; Email [email protected]

2E&E, Lawrence Livermore National Laboratory, P.O. Box 808, L-201, Livermore, California, 94551; Telephone (925) 422-6229; Fax (925) 423-1057; Email [email protected]

3ERD, Lawrence Livermore National Laboratory, P.O. Box 808, L-530, Livermore, California, 94551; Telephone (925) 423-1423; Fax (925) 424-3155; Email [email protected]

Many modeling efforts described in the literature assume a first-order, linear-driving-force

model to represent the chemical dissolution process at the non-aqueous/aqueous phase liquid (NAPL/APL) interface. This assumption raises the questions: where, in relation to a region of pure NAPL, does the "bulk aqueous solution" region begin, how does it behave and how does it impact the dissolution and mass transfer processes? The answers are assumed to be associated with an arbitrary, predetermined boundary layer, which separates the NAPL from the surrounding solution. The mass transfer rate is primarily limited by diffusion of the component through the boundary layer. Representing mass flux as a rate-limiting process is equivalent to assuming diffusion through a poorly defined stationary boundary layer with an instantaneous local equilibrium and linear concentration profile. Some environmental researchers have enjoyed success explaining their data using mass transfer correlations between the dissolved mass flux and the concentration gradient. These correlations are usually expressed in terms of the modified Sherwood number as a function, for example, of NAPL saturation and the Reynolds and Peclet numbers. However, this approach ignores the details of local hydrodynamics, and thus, the resulting correlations depend strongly on the experimental conditions and the scale at which they are conducted. Thus, a more general theory for NAPL dissolution in natural systems is needed.

In this study, we seek to fundamentally enhance the understanding and modeling of NAPL

dissolution. Dissolution of NAPL takes place at the pore scale level while remediation technologies are applied at the field scale. Building models that effectively represent the pore scale processes requires developing techniques for quantifying these processes and their impact on large-scale mass transfer rates. We present an integrated experimental and computational approach aimed at quantitatively investigating the role of pore structure, entrapped NAPL distribution and hydrodynamic conditions on pore scale mass transfer rates. This provides a systematic first step towards our ultimate goal of bridging, in a consistent way, the scale disparity between the pore scale processes and the field scale applications.

At the pore scale, a series of dissolution experiments have been conducted in smooth or

rough glass wall Hele-Shaw-like cells. Pore space morphology and the evolving geometry of entrapped ganglia were measured using light transmission techniques. Experiments were conducted using CO2 gas and TCE, representing unit-like cells for the upscaling closure problem. Tracking the NAPL-water interface under different flow rates allowed us to collect a large set of data on chemical dissolution rates, enabling us to calculate a lumped dissolution mass transfer coefficient of CO2 and TCE in water.

Lumped mass transfer rates are not ideal for modeling studies because the transfer surface

area is inherently embedded, hence unknown, and yet will change dynamically as dissolution

proceeds. As such, we are attempting to define and estimate an intrinsic mass transfer coefficient. Intrinsic mass transfer is defined as the outcome of the interaction of five major energies and forces that control the interfacial dissolution kinetics: solvation energy, interfacial tension energy, electrostatic energy, thermal fluctuation energy, and viscous forces that accelerate the chemical transfer across the APL/NAPL interface. By digital calculation of surface areas we successfully estimated APL/NAPL interface surface area, and hence calculated the intrinsic mass transfer. Results show a consistent intrinsic mass transfer rates regardless of the hydrodynamics.

In conjunction with the experiments, a numerical model was developed that introduces a

new approach in developing non-linearly coupled flow and transport equations in order that includes a specific description of interface dissolution processes. The governing NAPL/APL evolution equations were formulated along principles similar to the Stefan problem of moving interfaces. In addition to accounting for the continuous erosion of NAPL/APL interface due to the dissolution, ganglia of NAPL are allowed to move freely, under the influence of prevailing physical forces. Our numerical simulations were obtained using adaptive mesh refinement (AMR), Galerkin Finite Elements (GFEM) technique to solve the coupled nonlinear flow and transport equations simultaneously. The numerical model solves Navier-Stokes flow and transport with a state-of-the art level-set interface-tracking scheme. Adaptive mesh refinement allowed us to accurately track the interface between the ganglion and the APL.

Using published values of intrinsic physicochemical properties of TCE and CO2, this model

was used to simulate the pore scale experiments. Simulations indicated good agreement with the experiments. Once the evaluation of the numerical model was achieved, we numerically evaluated the mass transfer rates of TCE and CO2 for the staged experiments. Using a weighted average of the mass transfer rate at the interface with the numerically computed fluid velocity at the interface allowed us to further eliminate the velocity effect and obtain a velocity-averaged intrinsic mass-transfer rate for TCE. Results reaffirmed our initial assessment of the consistency of the intrinsic mass transfer rates.

To help interpret meso (Darcy) scale dissolution experiments of CO2 and TCE in a 2D meso

scale slab porous media, we have developed a 3D numerical model for packing synthetic deterministic and stochastic building-block periodic porous media. These synthetic porous media allow us to numerically upscale the pore scale processes to the meso scale as an intermediary step to the field-scale upscaling. Generating periodic porous media enable us to build meso scale synthetic porous media and to apply our numerical model to them. Given an initial mass of NAPL distributed in the synthetic porous media one can simulate the dissolution of the NAPL and quantify the instantaneous mass-transfer rates. Using different porous media packing, we successfully determined an upscaled effective mass transfer coefficient by relating the proportion of the remaining NAPL mass to the ratio of the average mean dissolved concentration in the system and the solubility limit of the NAPL. Furthermore, by changing only the diameter of the soil grains in deterministic synthetic porous packing, we were able to study the role of morphology of the pore space and entrapped NAPL on the hydrodynamics and the dissolution of NAPL. Currently, we are validating our meso-scale numerical upscaling results to the meso scale experiments while building a theoretical and numerical volume-averaging formulation of the field scale effective mass transfer which will be applied to a large set of data collected at Lawrence Livermore National Laboratory superfund site cleanup efforts. Work performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory (LLNL) under Contract W-7405-Eng-48 and funded by LDRD 04-ERD-01 and computations were conducted using the Livermore Computation Resources (LCR). UCRL-ABS-215074

IAHR-GW 2006 : Abstract “Modeling coupled surface/subsurface…” (Ababou et al.)

1

Modeling coupled surface/subsurface flow interactions : implementation and comparison of three models based on

Darcy, Boussinesq/Saint Venant, and Boussinesq/diffusive wave, with application to the Garonne floodplain, Midi-Pyrénées, France.

Ababou R.(1), Al-Bitar A., Peyrard D.(2), Quintard M.(1), Sanchez-Perez J.M.(2), Sauvage S.(2), Vervier P.(2), Weng P.(3)

PRELIMINARY ABSTRACT

Progress is needed towards better understanding environmental surface/subsurface flow interactions in alluvial river systems, and their effects on mass transport. For example, in the case of an alluvial plain like the Garonne river around the city of Toulouse, one needs to explore the physical parameters of the stream/soil/aquifer system (wet zone) that can influence hyporheic exchanges and their effects on biochemical nutrient uptake processes. Thus, there is a need for a modeling tool that could simulate two way interactions betweeen surface water and groundwater, through the stream’s hyporheic zone, in order to quantify exchanges between these three compartments of the hydrosystem.

The objective of the present study is to contribute to the development and application of mathematical and numerical models for simulating flow, and also solute transfer, in a floodplain where surface/subsurface interactions are important. In this paper, three distinct plane flow models (“2D models”) are tested, taking into account to various degrees some of the interactions that occur between various media: surface water (or more generally free surface open water bodies), saturated porous media, and unsaturated soils. Model (1). To test the approach based solely on Darcy’s law (without full coupling with surface water flow), we used the MARTHE model developed by BRGM (Thiery, 1990, 1993a,b; Thiery and Amraoui, 2001) is used. It processes three-dimensional (3D) flow by solving a discretized form of Darcy’s equation (Darcy, 1856) in a saturated domain, or Richards’ equation (Richards, 1931) in an unsaturated domain. Model (2). The second model, called HYPORE, uses Saint Venant equation for surface water, coupled with Boussinesq/Dupuit equation for free surface flow in a porous medium (aquifer). This model is solved numerically in a finite element framework using a commercial package, the COMSOL Multiphysics Software ( http://www.comsol.com ), previously known as “FEMLAB”. Model (3). The third model is based on the coupling capabilities of the variably saturated finite volume 2D/3D Bigflow code. It is here used in 2D mode, i.e., with vertically averaged flow equations (plane flow). The Boussinesq equation is coupled with the kinematic-diffusive wave equation using a single model equation with distributed parameters (see references).

To test validity of these three models, each of them was applied at a reach scale (8 km), including an instrumented site (called Monbequi) and the floodplain of the Garonne river (southwest of France). At this site 20 piezometers are installed and measured water level during 2 years in continue. In different hydrologic conditions, concentrations in conservative element (as chloride) were measured in each piezometer. Each model permits to quantify interactions between surface and subsurface flow during stationary and no stationary period (as flood period) and to simulate conservative transport in order to compare the validity of these three approaches.


2

Some of the results obtained thus far with the coupled approaches (Models 2 and 3) are illustrated in Figures 2.A and Figure 2.B. References.

ABABOU R., G. TRÉGAROT (2002) : Coupled Modeling of Partially Saturated Flows : Macro-Porous Media, Interfaces, and Variability. Published in Proceedings CMWR 2002, Computational Methods in Water Resources, 23-28 June 2002, Delft, The Netherlands, 8pp.

COMSOL MULTIPHYSICS… WENG P., SÁNCHEZ-PÉREZ J.M., SAUVAGE S., VERVIER P ET GIRAUD F. (2003) : Hydrological

modelling to characterise the riparian wetland of a large alluvial river (Garonne river, France). Hydrological processes, 17, 2375-2392.

Keywords : 2D flow modeling, Saint-Venant, Dupuit, Boussinesq, diffusive wave, groundwater-stream interactions, hyporheic zone, surface-subsurface coupling, solute transport.

Figure 1: Schematic representation of flow-transport interactions in a meandering stretch of the Garonne river (flow and solute transport through the hyporheic zone and pebble banks, showing also interactions with groundwater through river bed and banks,…).

D. Peyrard

Flux from surface water to hyporheic zone

Flux from hyporheic zone to surface waterLateral flux

Flux from ground water to surface water

A

B

C

D

A B C D

Porous media

Impermeable substratum


3

Figure 2: Numerical simulation of coupled stream/aquifer interactions during a synthetic flood event in a river stretch with two coupled flow models: Fig.2.A (top & middle): 2D plane surface/subsurface module of BigFlow (BF-Python).

Fig.2.B (below right): 2D plane surface/subsurface module of HYPORE code (solved with COMSOL Multiphysics) simulating flow and transport of chloride concentrations.

Cl- in mg.l-1

Alluvial plain

Garonne river

Thermo-Hydro-Mechanical modeling of 3D fractured rocks with coupled matrix-fracture hydraulics.

Rachid ABABOU(1), Israel CAÑAMÓN(1,2), Fco. Javier ELORZA(2)

(1) INSTITUT DE MÉCANIQUE DES FLUIDES DE TOULOUSE, FRANCE (2) UNIV. POLITEC. DE MADRID, SCHOLL OF MINES, MADRID, SPAIN

EXTENDED ABSTRACT

We present a preliminary study of fractured rock, including fracture network reconstruction and numerical flow simulations, as a first step towards a fully coupled Thermo-Hydro-Mechanical (T-H-M) analysis of a fractured granite formation located at the Grimsel Test Site (GTS, Switzerland), where the FEBEX experiment is located. FEBEX is an experiment to test the T-H-M behavior of a crystalline high-level waste repository.

The aim of the preliminary simulations presented below is to reproduce the hydraulic behavior of the fractured medium using either single or dual continuum approaches to the fractured porous rock. The macroscale continuum equations and coefficients are obtained by upscaling from the local Darcy equation (matrix) and Poiseuille-type equation (fractures) up to block scale, where each homogenized block contains ideally many fractures. But, to obtain the upscaled equations requires knowledge of the morphology of the 3D fracture network. The overall procedure can be summarized as follows. First, the 3D network is obtained by a statistical reconstruction method (or

inversion method) based on various fracture statistics and on detailed observations of fracture traces on tunnel drifts and boreholes.

Secondly, the domain of interest is partitioned into sub-domains, where fractured rock is represented as a set of single-fractured ‘blocks’. The tensorial upscaled coefficients are computed at the scale of the sub-domains based on superposition approximations.

Thirdly, the system of continuum PDE’s are solved numerically for initial-boundary conditions, with a numerical mesh finer than block scale, using (here) 3D finite element PDE solvers in FEMLAB® [3].

In this study, we focus mostly on hydraulic upscaling (although some preliminary results on Thermo-Hydro-Mechanics may also be presented). A set of 3D numerical flow experiments with either single or dual continuum equations is presented, along with the analytical method used for upscaling hydraulic coefficients. The flow simulations are performed using the FEMLAB® (now COMSOL Multiphysics) commercial software. More generally, the fully coupled THM involves a tensorial non-orthotropic (rank 4) PDE system to be solved with COMSOL® multiphysics.

Figure 3: Reconstructed fractured network.

Acknowledgements

This work has been partially supported by the FEBEX project under contract number FIKW-CT-2000-0016 with the European Commission and ENRESA.

References 1. Ababou R., A. Millard, E. Treille, M. Durin, and F. Plas : Continuum Modeling of Coupled Thermo-Hydro-Mechanical

Processes in Fractured Rock. Comput. Methods in Water Resources, Kluwer Academic Publishers, A. Peters et al. eds. Vol.1, Ch.6, pp.651-658 (1994).

2. Barenblatt G. , Zelthov I., Kochina I. Basic concepts in the theory of seepage of homogeneous liquids in fissured rocks. J. Appl. Math. 24 (1960), 213-240.

3. COMSOL, 2004: FEMLAB User’s Guide 3.1. COMSOL AB. October 2004. 581 pp. 4. Dershowtiz W. et al. (1992): FRACMAN v.2.3 : Interactive discrete feature data analysis, geometric modelling, and

exploration simulation – User Doc. Redmond WA: Golder Assoc. 1992. 5. Dykhuisen, R.C. (1990). A new coupling term for dual-porosity models. Water Resour. Res., 1990, 26(2):351-356. 6. Keusen, H.R., Ganguin, J., Schuler, P., Buletti, M. (1989): Grimsel Test Site: Geology. NAGRA Tech. Report. NTR 87-

14E. (Ref [6]) 7. Kfoury M., R. Ababou, B. Noetinger and M. Quintard, 2004. Matrix-fracture exchange in fractured porous media:

stochastic upscaling. Comptes-Rendus Académie Sciences (Paris). C.R. Mecanique, 2004, Vol.332, pp.679-686. 8. Pardillo J., Campos R., Guimerá J. (1997): Caracterización geológica de la zona de ensayo FEBEX(Grimsel-Suiza).

CIEMAT-IMA-M-2-01. 9. Stietel A., Millard A., Treille E., Vuillod E., Thoraval A., Ababou R.: Continuum Representation of Coupled Hydro-

Mechanical Processes of Fractured Media: Homogenisation and Parameter Identification. In : ‘Devts. Geotech. Engg.’: Coupled Thermo-Hydro-Mecha. Processes (DECOVALEX). O.Stephansson, L.Jing, C-F.Tsang eds. 79:135-164, Elsevier (1996).

10. Warren J.E. and P.J. Root (1963). The behavior of naturally fractured reservoirs. Soc. Petrol. Eng. J., 3(5), 245-255, 1963.

X Y Z

IAHR-GW2006 Toulouse, France – June 2006

* Corresponding author : [email protected]

An Adaptive time stepping method based on Richardson Extrapolation

Application to batch chemistry, unsaturated flow and transport modelling

B. Belfort, J. Carrayrou* and F. Lehmann

Institut de Mécanique des Fluides et des Solides, UMR ULP-CNRS 7507

2 rue Boussingault – 67000 Strasbourg, France. Abstract: Modelling of groundwater flow and transport is of interest in many sciences and engineering

applications. Ordinary- or Partial Differential Equations (ODE and PDE) are commonly used

for describing unsteady phenomena. The resolution of these equations through numerical

approximation leads to temporal, and often spatial discretizations. Analysis of simulation

results can require, on the one hand, an evaluation of discretization errors for its own

reliability. On the other hand, a grid adaptation process can take into account or be based on

this error monitoring.

In our work, the local temporal truncation error is estimated by following the general

Richardson extrapolation (Shampine, 1985). It consists in solving the same equation first in

one large time step and secondly in two half time steps. The difference between the solutions

obtained in one and two time steps allows to evaluate the temporal error. The second aspect of

the method is to control this error by using an efficient and automatic adjustment of the time

step size.

Significantly different computational codes can be improved with the new algorithm. Hence,

various numerical tests have been carried out in the domain of:

kinetic chemistry in batch system, which is described with coupled ODE:

( )i1 i n

dc f c ,..., c ,..., cdt

=

reactive transport, written under the instantaneous equilibrium assumption (Rubin 1983,

Steefel et al., 1996):

( ) ( ) ( )sTd TfD Td U Td

t∂ ω + ρ

= ∇ ⋅ ⋅ ∇ − ⋅ ∇ ∂

unsaturated water flow, which can be described using the mixed form of

Richards’equation (Celia et al., 1990):

( ) ( ) ( ) v

h. K h . h z f

t∂θ

− ∇ ∇ − = ∂

If the algorithm seems, at a first sight, to require more calculation time, some optimisations

have been proposed to improved its efficiency. Besides, different strategies are developed to

conserve good mass balance error or to add physical meaning for the extrapolated solution.

Finally, comparisons with fixed time step method or adaptive method without extrapolation

prove the advantages of the method as regards to the accuracy and reliability.

References : Celia, M. A., Bouloutras, E. T. and Zarba, R. L.: 1990, A general mass-conservative numerical solution for the

unsaturated flow equation, Water Resour. Res. 26 (7), 1483-1496.

Rubin J.: 1983, Transport of reacting solutes in porous media : relation between mathematical nature of problem

formulation and chemical nature of reaction, Water Resour. Res. 19 (5), 1231-1252.

Shampine, L. F.: 1985, Local error estimation by doubling, Computing 34 (2), 179-190.

Steefel, C. I. and McQuarrie, K. T. B.: 1996, Approaches to modelling of reactive transport in porous media. In

Reactive Transport in Porous Media. P. C. Lichtner, C. I. Steefel, E. H. Oelkers, Eds., Reviews in

Mineralogy, Mineralogical Society of America, Washington. 34, 82-129.

Valocchi, A. J., Street, R. L. and Roberts, P. V.: 1981, Transport of ion-exchanging solutes in groundwater:

Chromatographic theory and field simulation, Water Resour. Res. 17 (5), 1517-1527.

Batch kinetic test case: evolution of the temporal relative error versus the

required CPU time.

Reactive transport test case (Valocchi et al., 1981):

evolution of the temporal relative error versus the required CPU time

Definition of the temporal and spatial relative error

measure (TREM and SREM) Unsaturated flow test case (Celia et al., 1990):

evolution of the global error versus the required CPU time

1 10 100 1000

1E-6

1E-5

1E-4

1E-3

0.01

0.1

1

SR

EM

ref (

%)

CPU Time (s)

Adaptive time step with extrapolation Adaptive time step without extrapolation Fixed time step

Nb time steps

t ref ,tref

ref ,tt 1

y yTREM

y=

−= ∑

where yref,t refers to the reference solution at time t.

refref

refdomain

y ySREM dz

y−

= ∫

where yref is the reference solution and i indexes the cells of the spatial mesh

1 10 100 10001E-5

1E-4

1E-3

0.01

0.1

1

TREM

ref (-

)

CPU Time (-)

Adaptive time step with extrapolation Adaptative time step without extrapolation Fixed time step

0.1 1 101E-6

1E-5

1E-4

1E-3

0.01

0.1

1

TREM

ref (-

)

CPU Time (-)

Explicit Richardson Runge Kutta 2 Richardson Runge Kutta 4 Richardson Explicit fixed Runge-Kutta 2 fixed Runge-Kutta 4 fixed

Interpretation of interference pumping tests in a fractured limestone (Poitiers – France) using fractal and dual-media

approaches : Homogenization scale of hydraulic parameters

ANNE KACZMARYK & FREDERICK DELAY UMR 6532 CNRS, University of Poitiers, Earth Sciences Building,

40 Avenue du Recteur Pineau, F-86022 Poitiers - France e-mail: [email protected]

Abstract Interference pumping tests remain an important tool to evaluate hydrodynamic parameters of a reservoir but require specific interpretation methods in porous fractured rocks. Automatic inversion which lowers subjectivity in fitting models to data has been tested on three approaches well suited to fractured rocks: the fractal single medium, the homogeneous dual medium and the fractal dual medium. Sensitivities to parameters need to be re-scaled to get reliable and stable inverse solutions. The three models have been applied to an important set of data from a fractured limestone in Poitiers (France). Results are discussed in terms of homogenization scale which has some importance if parameters drawn from interference testing are re-used, e.g., in predicting production capacity or in conditioning numerical models of the reservoir. Key words: fractured reservoir; pumping test; fractal medium; dual medium; inverse problem.

NAPL dissolution in porous media: Non-equilibrium effects due to saturation heterogeneities

C. Tathy1, B. Mabiala1, and M. Quintard2

Extended Abstract

Risk analysis associated to the pollution of an aquifer by trapped hydrocarbons (referred to as NAPL for Non-Aqueous Phase Liquid) and the modelling of natural attenuation is largely dependent on the type of model used for the NAPL phase dissolution in the source zone. NAPL concentration in the water at the exit of the source zone may depend largely on the dissolution model. The total time required for complete dissolution of the source zone is also dependent upon this model. Non-equilibrium effects may arise when considering the dissolution of the trapped NAPL in the porous medium at the pore-scale. This is most often approximated at the Darcy-scale by a non-equilibrium model involving a mass exchange coefficient. The characteristic time associated to this exchange term is on the order of l2/D, where l is characteristic of the trapped phase cluster size, and D is the molecular diffusion coefficient. It can be as large as some hours, especially at very low saturation at the end of the dissolution process. It is shown in this paper that non-equilibrium effects may arise at the large-scale when considering Darcy-scale saturation heterogeneities. These heterogeneities may be due to medium heterogeneities, or to the initial repartition of the NAPL phase, because of fingering for instance.

Direct numerical simulation

The problem is studied first by performing Darcy-scale numerical experiments on various 1D and 2D systems, including stratified media. One example of the results typically obtained is presented in Figure 1 to Figure 4.

ωη

Sor,ω

Sor,η

Figure 1. Geometry and initial conditions. Legend

t = 11.57 days

t = 33.56 days

t = 92.59 days

Figure 2. Saturation fields (strata length : 10 m)

Legend

1 Ecole Nationale Supérieure Polytechnique, BP 69, Brazzaville, Congo 2 Institut de Mécanique des Fluides de Toulouse (I.M.F.T.), allée du Professeur Camille Soula, 31400 Toulouse

t = 11.57 days

t = 33.56 days

t = 39.35 days

t = 52.08 days

t = 92.59 days

Figure 3. Concentration fields

0 20 40 60 80 100 1200.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

t (days)

Figure 4. Averaged concentration versus equilibrium concentration at x=10m

The results show first that the water concentration of the NAPL component at the exit of the source zone may be significantly lower than the equilibrium value. It is also shown that the characteristic time-scale for the total dissolution may be much larger than the small-scale counterpart. Large-scale averaging: 1D cases

When direct numerical simulations cannot be performed, a large-scale model would be very useful. Given the above considerations, it seems difficult to derive such a model in a very general framework. In this paper, the volume averaging technique is applied to 1D systems with Darcy-scale heterogeneities. A large-scale model is derived from the Darcy-scale equations in the two following limit cases: small and large Darcy-scale Damkhöler numbers. The resulting models in both cases have the mathematical structure of a non-equilibrium model. It is shown how to calculate the resulting mass exchange term from the Darcy-scale heterogeneities. One of the important finding is that the obtained values have a very different behaviour compared to the Darcy-scale usual correlations. The large-scale correlations are also very different between the two limit cases. The resulting large-scale models are compared favourably to Darcy-scale direct simulations.

Acknowledgement: This work received financial support from program RITEAU (project:

CIDISIR).

C/Ceq

Assessing model uncertainty in groundwater contamination models

David Draper and Bruno Mendes Department of Applied Mathematics and Statistics

University of California, Santa Cruz Santa Cruz, CA 95064 – USA email ([email protected])

The issue of uncertainty of models in groundwater contamination has always been present in the minds of modelers. Everyone is well aware of the difficulties in characterizing geophysical systems, and how different models can fit observable data equally well and yet offer quite different extrapolations (predictions) away from the data. We begin with an uncertainty assessment framework we developed earlier [www.ams.ucsc.edu/~draper/writings.html items 43 and 55], in which all forms of uncertainty in groundwater contamination modeling can be broken down hierarchically into four types:

• scenario (there may be uncertainty about relevant inputs to the physical process under study),

• structure (conditional on scenario, the precise mathematical form of the best

model to capture all relevant physical processes (advection, diffusion, ...) may not be known,

• parametric (conditional on scenario and structure, the model will typically have

one or more unknown physical constants that need to be estimated from data), and

• predictive (conditional on scenario, structure, and parameters, the model predictions may still not agree perfectly with the observed data).

Two of the most fundamental goals of scientific work are accurate prediction of future data and accurate (well-calibrated) uncertainty assessments for the predictions. If scenario and structural uncertainty are present, it will not be enough in creating well-calibrated uncertainty assessments merely to qualitatively compare the results of different scenario and structural choices; it is necessary to combine uncertainty across (or between) these different choices in addition to quantifying uncertainty within (conditional on) them. A Bayesian approach to solving this problem appears most natural to us. To show the applicability of our approach we chose a data set available from a reputable source [Gonzalez et al., Groundwater Hydraulics, 1984], and well established analytical/ numerical modeling methods [Javandel et al., "Groundwater Transport: Handbook of Mathematical Models, 1984]. In

the work on which I will report in this talk, we increase the physical realism of the modeling by making a direct comparison of one-, two-, and three-dimensional advection-diffusion models on the same data set. In each case we use Markov Chain Monte Carlo methods to account fully for parameter uncertainty given the specific model structure; we compare the models with a log-scoring criterion [www.ams.ucsc.edu/~draper/ writings.html, item 71] which rewards models that make accurate and well-calibrated predictions of the observable data; we create composite predictive distributions that capture not only within-model parametric and predictive uncertainty but also between-model uncertainty; and we demonstrate that these composite predictive distributions are better calibrated than those obtained by ignoring model uncertainty or treating it in a qualitative way. Our approach can be applied quite generally, e.g., not only in modeling research but also in risk analysis for accidental contaminations.

Seawater intrusion modeling in heterogeneous costal aquifers with recharge and surface flow coupling via

diffusive wave, Boussinesq and sharp interface equations. A. AL-BITAR1 and R. ABABOU

Institut de Mécanique des Fluides de Toulouse (IMFT), Allée du Professeur Camille Soula, 31400 Toulouse, France.

Introduction In some coastal aquifers one may have streams or rivers going into the sea. In this case

modelling saltwater intrusion requires taking into consideration the surface and subsurface flow interactions. Here we develop 2D plane model aimed to represent the process under sharp interface conditions between the fresh /saltwater zones. The model is implemented in a finite volume numerical code BIGFLOW (Ababou and Bagtzoglou, 1993) and tested in simple configurations domains.

Surface flow model The surface water body flow is modelled using the diffusive kinematic wave. This model

is obtained from the N-S system by two steps: First by vertical integration to get the 2D St-Venant system and second by neglecting the inertial terms in the St-Venant system. In the resulting model there is a direct relation between the velocity and the hydraulic head thus the St-Venant system is reduced to a single PDE. The main parameter is the Manning-Strickler friction coefficient or the Chezy coefficient.

Subsurface flow model As for the subsurface flow it is represented by the 2D vertically integrated Boussinesq

flow equation considering an underlying saltwater zone. In natural environments a certain transition zone exists between the freshwater zone and the saltwater zone. The process is generally modelled using a density dependent solute transport model (Ackerer et al. 1999, Diersch and Kolditz 2002). The other approach is to consider the freshwater and the saltwater as two immiscible fluids separated by a so called 'sharp interface'. The resulting nonlinear PDE system can be solved by interface tracking or by a continuous approach. In the last approach one gets an equivalent two-phase flow system with pseudo-capillary pressure (Huyakorn et al. 1996, Holm and Langtangen 1999). Now if the saltwater zone is considered as immobile, the Badon Ghijben-Herzberg condition is obtained. This condition gives a direct relation between the freshwater hydraulic head, the freshwater/saltwater density contrast and the position of the saltwater zone considering hydrostatic conditions in the saltwater. Thus the system of PDE is reduced to one PDE with the freshwater hydraulic head as unknown.

The major inconveniences of the solute transport models are the computational loads (Johannsen et al. 2002) and the difficult evaluation of the dispersion coefficients, more specifically of the macrodispersion. Another factor is that in most cases the width of the transition zone, at regional scale, is small compared to the uncertainty of the interface position due to the heterogeneity of the conductivity field. Dagan and Zeitoun (1998) investigated the effect of stratification under Boussinesq approximations. Albitar and Ababou (2005) investigated the effect of plane heterogeneity under Boussinesq approximation. The sharp interface models major weakness is the less physical representation of the processes on small scales and thus the inability to incorporate chemical processes.

1 Corresponding author: [email protected]

Single generic equation coupling To enable a single equation representation of the coupled system, the kinematic wave

PDE is written as a diffusive process in the form of the Darcy flow equation with a non-linear coefficient. Then we consider one continuous fresh water body in permanent equilibrium with a saltwater body, the system is then reduced to a single diffusive equation with an equivalent conductivity coefficient for the river-aquifer system. The hydraulic head which is the only unknown coincides with the water level in the river or in the aquifer indifferently. This model has been implemented using the generic single equation code BIGFLOW. Simulations were done on simple test cases, such as the simplified meander configuration shown in Fig.1.

Fig 1 Numerical results in a simplified meandering channel. Light green color : topographic map; darker blue color : water level (hydraulic head), and dark red color : saltwater/freshwater interface.

References Ababou R, Bagtzoglou AC (1993). BIGFLOW: a Numerical Code for Simulating Flow in Variably Saturated, Heterogeneous Geologic Media Theory and User’s Manual Ver.1.1. NUREG/CR-6028, US NRC Report, Washington DC. Albitar A, Ababou R (2005). Random Field Approach to Seawater Intrusion in Heterogeneous Coastal Aquifers, in GeoENV: Geostatistics for Environmental Applications, Renard P., Demougeot-Renard H., Froidevaux R. (Eds.), ISBN: 3-540-26533-3, Springer. Ackerer P, Younes A, Mose R (1999). Modeling variable density flow and solute transport in porous medium: 1. Numerical model and verification. Transp Porous Media 35(3):345–73. Dagan G, Zeitoun DG (1998). Seawater-freshwater interface in a stratified aquifer of random permeability distribution. J Contam Hydrol. 29: 185–203 Diersch H J G, Kolditz O(2002). Variable-density flow and transport in porous media: approaches and challenges. Adv. Water Resour. 25: 899-944 Holm E J , Langtangen H P (1999). A method for simulating sharp fluid interfaces in groundwater flow. Adv. Water Resour. 23: 83-95 Huyakorn, P S , Wu Y S, and Park N S (1996). Multiphase approach to the numerical solution of a sharp interface saltwater intrusion problem, Water Resour. Res. 32(1): 93-102. Johannsen, K., W. Kinzelbach, S. Oswald, and G. Wittum (2002). The saltpool benchmark problem - numerical simulation of saltwater upconing in a porous medium. Adv. Water Resour 25:335-348.

Keywords: Surface/subsurface coupling, saltwater intrusion, sharp interface, aquifers.

A NUMERICAL GLOBAL UPSCALING TECHNIQUE FOR BUILDING RESERVOIR MODELS

Mickaële Le Ravalec-Dupin and Ludovic Ricard

Institut Français du Pétrole 1 & 4, avenue de Bois Préau 92852 Rueil-Malmaison France

The use of numerical models for investigating flow in underground reservoirs has become common practice in hydrology and petroleum engineering over the last 30 years. Very detailed geological models are built on grids, which contain up to 108 gridblocks. Although computers are growing ever more powerful with more precision, fluid flow simulations on such grids are tremendously CPU-time consuming. To make flow simulations tractable, the number of gridblocks has to be reduced. Some type of averaging or upscaling is therefore required to estimate the equivalent properties to be attributed to the coarse gridblocks (Figure 1). The focus of this talk is mainly concerned with the upscaling of absolute permeabilities. The development of numerical upscaling techniques was motivated by the idea of being consistent with flow physics. Basically, the equivalent permeabilities of coarse gridblocks are derived from flow simulations. They are strongly influenced by the size of the domain over which flow simulation is carried out. Upscaling techniques are said to be local when flow simulation is reduced to the target coarse gridblock. One speaks about techniques with an extended neighborhood when flow simulation is run over the target coarse gridblock plus some neighboring coarse gridblocks. Last, upscaling techniques are said global when flow simulation is run over the entire fine grid. The computed equivalent permeabilities also depend on the boundary conditions applied at the limits of the simulation domain. These boundary conditions are usually arbitrary. Here, we propose a global upscaling technique inspired by the boundary condition perturbation approach introduced by the works of Pickup et al. (1992). The underlying idea is to reduce the dependency on boundary conditions and to better capture the connectivity of the neighboring geological bodies The proposed global approach calls for a flow simulation over the entire fine grid. In practice, this step is feasible only if flow simulation is fast. Because of this requirement, we first developed a pseudo-spectral solver using Fast Fourier transforms. As an example, it requires about 30 seconds to approximate pressure gradients and velocities over a grid containing 1000×1000 gridblocks. Solving flow equations yields a gradient pressure, which is then used to estimate the boundary conditions applied to the coarse gridblocks. Next, this pressure gradient is perturbed leading to the estimation of the equivalent permeability (Figure 2). The resulting equivalent permeability is a full tensor. Numerical experiments are then carried out on synthetic geological models to investigate the interest of the method proposed. Finally, these results are compared to those of other numerical upscaling techniques. Reference Pickup, G., Jensen, J., Ringrose, P., and Sorbie, K., 1992, A method for calculating permeability tensor using perturbed boundary conditions, 6th ECMOR.

Ku pη

= − ∇ eqKU P

η= − ∇

. 0u∇ = . 0U∇ = Figure 1. Going from the fine scale to the coarse scale.

Figure 2. Computed equivalent permeability tensor.

FRAMEWORK FOR A PROCESS-BASED SALINISATION RISK ASSESSMENT: SOLUTE RECYCLING

VERSUS PRIMARY GROUNDWATER SALINISATION

Ellen Milnes Centre d’hydrogeologie Université de Neuchâtel Rue Emile Argand 11 CH 2007-Neuchâtel Tél: ++41 32 718 26 77 Pierre Perrochet Centre d’hydrogeologie Université de Neuchâtel Rue Emile Argand 11 CH 2007-Neuchâtel Tél: ++41 32 718 25 77 Philippe Renard Centre d’hydrogeologie Université de Neuchâtel Rue Emile Argand 11 CH 2007-Neuchâtel Tél: ++41 32 718 25 37 Keywords : coupled salinisation processes, irrigation salinisation, transport simulation, process-based risk assessment Groundwater salinisation in irrigated areas can either be caused by primary salinisation processes

(dissolution of geogenic salts, agricultural inputs, seawater intrusion etc.) or by solute recycling from

irrigation and by evaporative processes. In contrast to primary salinisation processes, solute recycling

from irrigation does not add any solutes to the system, but may lead to salinisation by re-distribution

of extracted solutes. Since the respective remedial or conservation measures may be very different, a

crucial issue is to correctly identify the spatial distribution and the superposition of the different

processes. We therefore propose a simulation procedure which allows decomposition of the

measureable bulk salinities into primary salinisation and solute recycling components. The spatial

impact of different salinisation mechanisms can thereby be evaluated separately, leading to a

framework for a process-based salinisation risk assessment methodology.

In a first step, a simulation procedure is presented that allows direct calculation of the solute recycling

process. Since the amount of the re-distributed salt load is a function of the extracted salt load, a

procedure has to be adopted, in which the the strength of the solute source in the advection-dispersion

reflects the extracted salt load at irrigation wells. Applying the transfer function theory to describe the

solute recycling process yields an independent relationship for the strength of the solute source. The

impulse response is first expressed in Laplace space as the sum of the n-fold convolutions of the

average travel time probability density function between the irrigated surface (volume) and the

extraction wells. The strength of the solute source is then obtained by a convolution product of the

impulse response with the primary solute mass flux captured by the well. At late times, the solute

source strength is a function of the capture probability and the primary solute mass flux only.

The proposed simulation procedure allows direct simulation of salinisation induced by solute recycling

and can thus be combined with simulations of primary salinisation processes. Thereby, the overall

‘present state’ salinity distribution can be simulated, which has to be calibrated to fit the measured

‘present state’ bulk salinity distribution. Re-running the (calibrated) simulation without the coupled

process of solute recycling yields the ‘present state’ primary salinisation component, which can be

deducted from the overall salinity distribution to obtain the ‘present state’ solute recycling component.

In a second step, the salinity distribution obtained from the direct simulation procedure for late times

yields the overall salinisation potential, which can be decomposed into a solute recycling potential

(RP) and primary salinisation potential (PP).

The decomposed ‘present state’ salinity distributions and salinisation potentials, respectively, are used

to calculate a salinisation risk index. The risk index is defined as the potential of further salinisation

induced by the respective salinisation process and is obtained by deducting the components of the

‘present state’ salinisation from the respective salinisation potentials and normalising them with the

salinisation potential. The thereby obtained risk index maps reveal areas which are prone to further

salinity increase due to solute recycling and primary salinisation, respectively, and are strictly related

to the hydraulic setting for which the salinisation potentials were established. Modification or

optimisation of an exploitation scheme leads to modified risk index distributions. Risk index maps

obtained for a modified exploitation scheme highlight areas which will suffer further salinisation from

areas for which the modified exploitation scheme will have a remediating effect.

Comparing the solute recycling and primary salinisation risk index distributions with the ‘present

state’ salinity distribution allows delimitation of areas which require remediation (i.e. areas with low

risk indices and high salinities) and areas which require conservation (i.e. areas with salinities close to

the exploitation limit with high risk indices). The risk mapping procedure is illustrated on an example

from Cyprus (Akrotiri aquifer).

The described process-based framework for a salinisation risk assessment methodology allows

identification of the spatial distribution of the superimposed salinisation processes and can therefore

be envisaged as tool to design appropriate remedial and conservation measures.

Is there any hope for transverse dispersion?

Dentz, M., E. Abarca, J. Carrera and X. Sánchez-Vila

Department of Geotechnics and Applied Geosciences, School of Civil engineering, Technical University of Catalonia, Barcelona, Spain

ABSTRACT Transverse dispersion is a key process in controlling not only contaminants spreading but also mixing, and therefore reaction rates, as well as seawater intrusion (i.e., penetration, width of the mixing zone and inland seawater flux). It is as important, if not more, than longitudinal dispersion. It is usually quantified by the transverse dispersion coefficient, which is hard to evaluate. In fact, stochastic results suggest that its macroscopic value is negligible, while, in practice, it is often taken to be a sizeable proportion of longitudinal dispersion, about one fifth to one tenth. Yet, field evidence supporting either assumption is surprisingly scarce. Here, we argue that, sea water intrusion data can be used to estimate field scale transverse dispersivity values. Moreover, we summarize results showing that time fluctuations in velocity lead to macroscopic dispersivity. The resulting computed transverse dispersivity is consistent with the values used in practice.

MODELING OF THE GROUNDWATER FLOW AND TRANPORT OF REACTIVE SOLUTES IN THE SALT CRUST AQUIFER. SALAR DE ATACAMA, CHILE.

Carlos Vásquez García1, José Francisco Muñoz Pardo2 Keywords: model, transport reactive, brine aquifer, SHEMAT. This paper presents the investigation on the modeling of the groundwater flow in the salt crust aquifer of the nucleus at the Salar de Atacama, located to 2.300msnm in the Atacama Desert, Chile (Figure 1).

Figure 1: Geographic location of study zone.

The nucleus of Salar de Atacama is a homogenous body (90% halite), 1.100km2 of area and 650m of mean thickness. The aquifer contents brine with high levels of carbonate, sulfate, lithium, potassium, manganese and boron. Mining companies extract brine with pumping and injection wells, soon to concentrate it by evaporation (2.130mm/year of potential evapotranspiration). Groundwater flow and transport of reactive solutes model must consider the variations density, viscosity, porosity and permeability. The flow, transport and chemical reactions equations are hard to solve because they are nonlinear and connected. This work recognizes and describes the important characteristics (hydrology, climate, area, topography) and processes in the zone of study, for example the chemical interaction between solutes and fluid, and between these and the ground due to precipitation and dissolution chemical’s reactions. The software SHEMAT (version 7.1, 2003) was used to represent and simulate this process.

1 MSc. student, Dept. de Ingeniería Hidráulica y Ambiental, Pontificia Univ. Católica de Chile, Ave. Vicuña Mackenna 4860, 6904411 Santiago, Chile (author) E-mail: [email protected] 2 Civil engineering, Dr. Ing., Dept. de Ingeniería Hidráulica y Ambiental, Pontificia Univ. Católica de Chile, Ave. Vicuña Mackenna 4860, 6904411 Santiago, Chile (supervisor). E-mail: [email protected]

SHEMAT is a three-dimensional hidrogeochemistry modeling software developed in Aachen University of Technology (RWTH), Germany. This software solves the differential equations of flow, transport of solutes and heat with a scheme of finite differences with nodes trims in the cells. SHEMAT connects the chemical component given by precipitation and dissolution chemical reactions (in equilibrium or kinetic) through a source term in the equation of transport. The aspects of SHEMAT that have greater interest in this investigation are:

• It considers the coefficients of activity of the chemical equations using virials coefficients of Pitzer, methodology specially adapted for fluid with high ionic force.

• It allows the variation of properties of the fluid (viscosity and density) and of the aquifer (porosity and permeability) based on the concentration, temperature and pressure. In the case of the permeability the user can choose a direct relation with the porosity between some options.

• The data base available in the program includes chemical parameters of the ions Na+, K+, Mg2+, H+, Cl-, SO4

2-, OH, HCO3-, CO3

2-, CO2, H2O, Sr and Ba (valid between 0 and 90ºC), and it can be modified by any user.

Finally, model SHEMAT is applied to the brine’s extraction on the Salar de Atacama, including the chemical component (precipitation and dissolution reactions) like agent of modification of properties of the fluid and aquifer. The model (Figure 2) consists of a rectangular dominion of dimensions 15km x 25km with a unique layer of 60m with impermeable bottom and batteries of pumping and injection wells until 30m. The validated model would allow simulating ion concentrations in wells and the levels of water table under different scenes, of way to evaluate the behavior in the medium and long operated term of the aquifer.

Figure 2: Nucleus representation of halite aquifer.

60m

25km

15km

60m

15km

Q1 Q2

30m

®2005 Google – Imagery ©2005 DigitalGlobe10km

R.ABABOU & A.C.BAGTZOGLOU : EXTENDED ABSTRACT FOR IAHR-GW2006 (2 PP.)

1

SOURCE IDENTIFICATION WITH THE “RAW” SCHEME : REVERSE ANTI-DIFFUSION WALK

R. ABABOU (a), A.C. BAGTZOGLOU (b) (a)Institut de Mécanique des Fluides de Toulouse, France

(b)University of Connecticut, Storrs, CN, U.S.A.

INTRODUCTION In view of increasing demands for clean drinking water, it is highly desirable to identify pollution sources accurately, as well as to backtrack the pollution source, to recover the spatial extent the plume at different times, and ultimately, to reconstruct the contamination plume release time history. This can be formulated as an advection-diffusion inverse problem. Its difficulty is compounded by the fact that geologic media are highly heterogeneous. In this context, the objectives of this paper are to contribute, first, to the problem of source position identification. The reader is referred to Atmadja & Bagtzoglou for a review on mathematical methods for pollution source identification. They developed a PDE-based method called Marching-Jury Backward Beam Equation (MJBBE) to recover plume spatial distributions and release history between known initial and final plumes. Another relevant method is the Quasi-Reversibility (QR) approach, based on a PDE that is numerically stable even with negative time steps. However, numerical difficulties arise with most PDE-based source identification and plume reconstruction methods. For this reason, we developed a novel scheme using a “reverse”, anti-diffusive particle random walk, in a lagrangian framework. This scheme should achieve source identification without having to deal with ill-posed or poorly conditioned inverse PDE problems. It is based on a “censored” random walk whose properties are tailored to produce just the right amount of “fickian anti-diffusion”, in order to force the present (“final”) diffusion plume to refocus backwards in time, at the correct rate, towards its “initial state” or source. We name it “Reverse Anti-diffusive Walk” (RAW). The objectives are to generalize its use for heterogeneous field pollution with advective transport and tensorial dispersion. The particle-based RAW scheme is now briefly described in the case of pure isotropic (anti)diffusion. Glimpses of the analytical theory are given in 1D; numerical results are displayed both in 1D and 2D. FORMULATION OF PARTICLE-BASED ANTI-DIFFUSIVE SCHEME (“RAW”) Continuous time formulation. The basis of the anti-diffusion scheme is a continuous time, censored, non-local random walk (here described in the 1D case for simplicity) :

( ) ( ) ( )otherwise

SSifdttDtdUDtdX GLOBALLOCAL

022 −==

=ξ (3.1)

( )( )( ) ( )( )tXtXSignS

tdUSignS

GLOBAL

LOCAL

−=

= ( )( )3.3

2.3

Several remarks are in order. First, anti-diffusion is forced by the minus sign in “SLOCAL= - SGLOBAL”. Secondly, non-locality is due to the fact that local displacement depends on mean displacement, or center of mass of the plume, via equation (3.3). Thirdly, although this scheme is formulated here in continuous time, it is physically more meaningful in discrete time (as will be seen). Discrete time formulation & adaptive time-stepping. Using an explicit time discretization of eqs.(3) along with a specific adaptive time-step, ∆t = ∆tn , we are able to show by probabilistic arguments that a constant anti-diffusion rate (-D1) can be achieved, where D1 may be different from the pseudo-diffusion rate D0 that adjusts the intensity of jumps in the censored random walk. We obtain the following theoretical results based on random process theory. Briefly, constant rate anti-diffusion is achieved, with decreasing dispersion variance, as follows: ( ) ( ) 11

21

2 20 ++ −= nXnX tDt σσ , where (tn) is a discrete time related to an adaptive time-stepping scheme (∆tn). In the limit n→∞, we find ∆tn →0 and we obtain the asymptotic time: ( ) .20 1

2 Dtt XSUPn σ=→ We can also deduce, from the maximum anti-diffusion time tSUP, the maximum amount of variance reduction that has occurred during that time. The result is of the form:

( ) ( ) ( )ββσσ −−∆−= 112 1120

2 nn tDt , and satisfies ( ) 0lim 22 ==

∞→MINnn

t σσ .

This anti-diffusion procedure becomes 100% efficient in the limit as n→∞ and ∆tn →0, provided that 0<β<1, which is always satisfied if D0≤D1. Choosing D0=D1 gives β≈0.5947; the variance reduction ratio is about 0.5% after 10 time steps, and on the order of 10-6 after only 25 steps.

R.ABABOU & A.C.BAGTZOGLOU : EXTENDED ABSTRACT FOR IAHR-GW2006 (2 PP.)

2

NUMERICAL RESULTS : PLUME NARROWING & SOURCE FOCUSING IN 1D & 2D The figures below illustrate the numerical implementation (in MATLAB™) of our anti-diffusion particle scheme in 1D and 2D. The program was also used to generate the “final” Gaussian plume. Note. In all results below, “t” is “backward time”, i.e., the plume evolves into the past as “t” increases. 1D results. In the 1D test presented here: D0=1; D1=0.1; thus ω = D1/D0 << 1. The adaptive time step was calculated numerically, i.e., based on the theoretical equations but using numerical moments of the particle cloud (not theoretical moments). The total number of adaptive anti-diffusion time-steps was nMAX=50. The initial (or rather final) plume at time “t”=0 was Gaussian, with N=10 000 particles.

Particle distributions in 1D space. Top: “final plume” starting at t = 0 (“t” is backward time). Bottom: “quasi-initial plume” after 50 adaptive anti-diffusion steps.

Space-time plot of 1D particle cloud during anti-diffusion. Bottom line: t=0 (final plume). Top line: narrow quasi-initial plume (after 50 steps).

The 1D results (only some are shown here) indicate that: (i) : for ω= D1/D0 ≥ 1 (not shown), concentration profiles are not gaussian although they do converge to a Dirac; (ii): for ω= D1/D0 << 1, concentration profiles are near gaussian at all times and they converge to a Dirac at the source.

2D isotropic results. Starting with a 2D isotropic Gaussian plume C(x,y), we analyze the plume as a particle histogram vs. radial distance to the center of mass (r). Indeed, if C(x,y) is bivariate Gaussian, C(r) must follow a Rayleigh distribution. It turns out that the Rayleigh distribution fits quite well the numerical plumes, i.e., they remain close to isotropic Gaussian as they become narrower during anti-diffusion (a rather satisfactory result). Finally, the analytical (theoretical) and computed (numerical) dispersion variance during the 2D isotropic anti-diffusion process (not shown here for lack of space): dispersion variance decreases almost linearly with time, in an anti-fickian fashion, as required.

Radial particle plot, and fitted Rayleigh distribution, C(r), during 2D anti-diffusion starting from Gaussian plume on top.

Space-time particle plot for the 2D isotropic anti-diffusion problem (time axis is vertical).

CONCLUSION. To sum up, the proposed particle-based “RAW” scheme, with adaptive time-stepping, appears to be efficient in refocusing a present-day (“final”) particle cloud back on the original point source at a near constant (fickian) anti-diffusion rate. Moreover, if the “observed” cloud is Gaussian, the intermediate particle clouds during the refocusing process are also nearly Gaussian.

Numerical Modelling as a Tool to Investigate the Feasibility of

Artificial Recharge to Prevent Possible Saltwater Intrusion into the Bangkok Coastal Aquifer System

P. ARLAI1 , M. KOCH1 and S. KOONTANAKULWONG2 1 Department of Geohydraulics and Engineering Hydrology, University of Kassel, Kurt-Wolters Str.3, Kassel, Germany 2 Department of Water Resources Engineering, Chulalongkorn University, Bangkok, Thailand e-mail: [email protected]; [email protected] Abstract. Bangkok Metropolis and the vicinity provinces have experienced a tremendous increase of groundwater consumption in recent years. This groundwater is being extracted from the main eight water-bearing units underneath the lower Chao-Praya basin, with the consequence that piezometric heads in the aquifer system have been significantly reduced in some of the layers and head gradients have built up that appear to induce widespread migration of saltwater from its source regions into the producing areas of the aquifer system, leading to saltwater contamination there. Because of this imminent groundwater quality problem, Thai water authorities are presently considering the implementation of a comprehensive aquifer system restoration (ASR) program that will most likely comprise the option of artificial recharge. One of the major goals of the present paper is, therefore, to numerically investigate in a preliminary fashion the feasibility of artificial aquifer recharge and, if so, to propose optimal design strategies to that regard. The analysis of the available groundwater quality data shows that there are basically two kinds of polluting saltwater sources in the aquifer, namely, (1) classical seawater intrusion along the coast of the Gulf of Thailand, (2) vertical saltwater leakage from the overlaying, ubiquitous marine clay formations that has, in some parts of basin, protruded to the lower aquifer layers. The numerical water balance study reveals that the second source is with 34% of the inflow indeed the dominant groundwater polluter, compared with only 6% for the seawater intrusion. From the inspection of measured chloride profiles, in conjunction with flow and solute transport modeling, we are able construct an innovative map that allows to discriminate between four types of contaminated zones, namely, one of seawater intrusion, one of mixing between seawater and saltwater, one of shallow vertical saltwater intrusion and one of deep saltwater intrusion. This map clearly shows that seawater dominantly intrudes along the coastline west of the Chao-Praya river, while vertical saltwater intrusion mainly pollutes groundwater in the vicinity of Bangkok, Nonthaburi, Pathum Thani and some parts of Samut Prakan. Because of the dual origins of the saline waters above, we find that classical groundwater recharge alone is not able to improve the groundwater quality in the Bangkok aquifer system in the future, but that more complex water management strategies are needed, such as policy- or “non-constructive” measures, as well as a combination of policy- and constructive (use of recharge- and clean-up wells) measures. For that purpose a set of 31 different schemes is analyzed by means of a flow and transport model that, although it neglects the density-dependence of the saline water, provides a first guess of the overall feasibility and efficiency of a particular aquifer restoration scheme. It turns out that the best scenarios are the ones where all pumps in the lower layers are shut off, or where the pump rates are reduced to 60% of their 2002-values. An even more efficient, but also costlier scenario is one which uses such a water policy change in conjunction with several recharge and clean-up wells. A very interesting option appears to be the elimination of all pumps in one of the lower aquifer layers, as this affects the existing groundwater users the least, while still significantly reducing the area of contamination in the target year 2032. Keywords Bangkok Aquifer System, Groundwater Management; Recharge and Remediation Schemes; Sea- and Saltwater Intrusion; Modeling Flow and Solute Transport; MODFLOW, MT3DMS

ADVANTGES OF GEOSTATISTICAL SIMULATION IN HYDROGEOLOGY USING KRIGEAGE APROACH

STUDY CASE (SAND AQUIFER OF BISKRA, ALGERIA) A.H MESSAMEH1 , S. TEKKOUK 2 , A. BOUGUERNE 3

1 Hydrogeology, Hydrology research laboratory ( LARHYSS) Science, Engineering Faculty,University .P.o Box 145, R.P 07000, Biskra. Algeria Tél/Fax: +213 33 733204 [email protected] or [email protected] 2 Physics Department, University of Girona, Campus de Montilivi, 17071-Girona Catalonia-Spain , Tél :++34 972418371, Fax :++ 34 972418098 [email protected] or [email protected] 3 Hydraulic Deartment, Engineering Faculty, Batna university, 05000, Batna, Algeria Tèl/Fax:++213 33 869919, [email protected] ABSTRACT: Knowledge of quantifying and simulating aquifer’s pollutant transport by improving the physical-

chemical process background involved in the pollutant’s dynamic. In order to get idea about the spatial

and temporal of aquifer’s salinity evolution, Implementation of such simulation tools is needed for allowing

better comprehension to estimate the Impact on the hydrogeologic aquifer’s management behavior.

Obviously, measured networks are of limited density and the tremendous gaps in the statistical

series have a negative effect to the prediction of different hydraulic parameters using such mathematical

models. Therefore, suggestion for applying geostatistical analysis method is primordial.

Using variables regionalized theory, (Krigeage) is based on this theory permitting to use some recorded

punctual information to reconstitute on each spatial point the structured phenomenon intensity.

The major measured variables values are nonlinear transformation. Thus the simulation is

necessary in geostatistic. Krigeage approach might be useful to give an estimate for physical parameters,

boundary condition and define hydraulic charge, concentration fields.

The aim of this study is to investigate the chemical water quality and the different relationships

between the water chemistry and structural context for helping decision maker to drilling wells for water

supply purposes, taking into account water chemical analysis of the Mio -Pliocene aquifer at average

depth. The Data are coming from water service authority of Biskra province.

The validation of the method has been established from the capacity to reproduce simple

analytical solutions.

Two studies case has been investigated, one is micro-scale aquifer (1000 x 1000 m2), the other

case is macro-scale (sand aquifer of Biskra).

The simulation results show that the Gaussian Model expresses better the typical piezometric continuity.

Key world: Simulation, Geostatistic, Kriging, Regionalized variables, Gaussian

INTERET DE SIMULATION GEOSTATISTIQUE PAR KRIGEAGE EN HYDROGEOLOGIE – CAS DE LA NAPPE DES SABLES DE BISKRA (ALGERIE)

A.H MESSAMEH1 , S. TEKKOUK 2 , A. BOUGUERNE 3

1 Laboratoire de Recherche en Hydraulique Souterraine et de Surface LARHYSS Faculté des Sciences et Sciences de l'Ingénieur, Université de Biskra B.P 145 , R.P 07000, Biskra. Algérie Tél/Fax: +213 33 733204 [email protected] ou [email protected] 2 Physics Department, University of Girona, Campus de Montilivi, 17071-Girona Catalonia-Spain Espagne Tél :++34 972418371, Fax :++ 34 972418098 [email protected] ou [email protected] 3 Hydraulic department, Engineering Faculty, Batna university, 05000, Batna, Algeria Tèl/Fax:++213 33 869919, [email protected] RESUME:

Savoir quantifier et simuler le transport de polluant dans les aquifères, en améliorant la

connaissance des processus physico-chimique intervenant dans la dynamique d'une pollution. Afin

d'étudier l'évolution spatio-temporelle de la salinité des aquifères; nécessite l'élaboration d'outils de

simulation permettant de mieux comprendre et d'évaluer l'impact des aménagements et de sollicitations

sur le comportement hydrogéologique des aquifères.

La densité limitée des réseaux de mesure, les lacunes fréquentes dans les séries statistiques font

que l'estimation des différents paramètres hydrauliques par les modèles mathématiques est entachée

d'erreurs. Pour cela, nous proposons d'appliquer une méthode d'analyse géostatistique

(krigeage) basée sur la théorie des variables régionalisées, permettant de reconstituer, en tout point,

l'intensité d'un phénomène structuré dans l'espace, à partir de quelques informations enregistrées

ponctuellement.

En géostatistique, les simulations sont nécessaires pour tout problème impliquant des

transformations non-linéaires des variables mesurées. L'approche de krigeage peut servir à l'estimation

des paramètres physiques et les conditions frontières et la définition des champs de charge hydraulique et

de concentration.

L’objectif de notre étude est d’étudier les caractéristique de la qualité chimique des eaux, les

différentes relations entre le chimisme de l’eau et leur contexte structural pour aider a l’implantation des

forages destinée à l’A E P. Grâce a les donnés brutes des analyses chimiques des eaux de la nappe Mio-

pliocène a moyenne profondeur disponible au niveau de la D. H. W de Biskra

La validité de la méthode a été établie à partir de la capacité de reproduire des solutions

analytiques simples.

Deux études de cas, l'une sur un aquifère à petite échelle (1000 x 1000 m 2 ) et l'autre sur un

cas à grande échelle ( nappe des sables de Biskra ) sont présentées.

Les résultats de simulation indiquent que le modèle Gaussien exprime mieux la continuité typique

de la piézométrie.

MOTS CLES: Simulation, géostatistique, krigeage, variable régionalisée, modèle gaussien

Integration of recharge data from a hydrological water balance model into a 3D FE model for a crystalline basement area in the tropical river catchment of the Upper Ouémé (Benin / West Africa) Tobias El-Fahem1, Simone Giertz2, Barbara Reichert1

& Bernd Diekkrüger2 1Geological Institute, University of Bonn, 53115 Bonn, Germany e-mail: [email protected] 2 Geographical Institute, University of Bonn, 53115 Bonn, Germany Key words groundwater flow model; finite elements; hydrological water balance model, data acquisition; regolith; fractured aquifer; Benin; Ouémé catchment Abstract

The Republic of Bénin is one of the poorest countries in the world. It is situated at the northern coast of the Gulf of Guinea. Its major river, the Ouémé, has its source in the centre of the country and flows towards the Atlantic in the South. Like in most developing countries the population is still fast growing with increasing concerns about the availability of potable water.The pressure on the available water resources in the region is high and especially in the dry season where no surface water is available the water demand in the rural areas strongly depends on exploitable groundwater.

This study is part of the IMPETUS West Africa project. IMPETUS is a multi-disciplinary research approach in order to describe the impact of climatic change on the water resources of Benin. A special regard is given to the river catchment of the Upper Ouémé which covers an area of around 14.500 km². In this context a numerical groundwater flow model is used to simulate future scenarios of water availability in the Ouémé catchment to facilitate groundwater management for the model area in the future. The software used for modelling is FEFLOW 5.1 (WASY, 2004). As only poor field data are available in the region intense field work and data gathering has been carried out by the IMPETUS-project to obtain sufficient information for model calibration and validation. In addition to FEFLOW the hydrological model UHP-HRU is applied in order to simulate the water balance and the hydrological processes (Giertz et al. 2006). The model simulates surface and soil processes as evapotranspiration, surface runoff, percolation, interflow and groundwater recharge. The spatial discretization of the model is carried out according to the HRU concept (hydrologic response units). The subdivision in HRUs depends on topography, soil and land use. To perform a data coupling of groundwater and surface water models the recharge from UHP-HRU is integrated into the FEFLOW model. Comparison with recharge estimation from groundwater hydrographs and soil chloride data has been carried out. The simulation of climate change scenarios by UHP-HRU revealed that the increasing temperature and the change in rainfall patterns - predicted by the climate REMO for 2025 - cause a decreasing groundwater recharge and river discharge for some areas of the Upper Ouémé. With the application of numerical models the availability of future water resources and its vulnerability to recharge and exploitation can be evaluated.Furthermore the groundwater modelling contributes to a better understanding of the hydrogeological system. The study area is situated on the crystalline basement of the West African craton. A conceptual hydrogeological model

has been developed by Fass (2004) during the first IMPETUS phase from 2000–2003 for a 30 km² sized sub-catchment of the Ouémé river. Hydrochemical sampling and stable isotope analysis proves the applicability of this conceptual model for most of the Upper Ouémé catchment. Two aquifers of different characteristics are found in the study area: a regolith aquifer and a fractured basement aquifer. The regolith thickness depends on its morphological position. It is less at hill tops and valley bottoms and thicker at hill slopes. The average thickness ranges from 15 to 20 m. Although the regolith shows a porosity of around 2-5 %, its permeability of 1-9 x 10-7 ms-1 is only weak. The groundwater flow in the crystalline basement occurs only in fractures and faults. Fracture openings vary from mm to dm scale; the length of fractures ranges from m to km scale. In general the depth of fractures is around 40 to 60 m. Porosity is low with 0.1–0.2 %. Its transmissivity strongly depends on the fracture aperture. In general it is around 10-4 - 10-6 m2 s-1 (Jacquin & Seygona, 2004). A continuous monitoring system with 12 automatic measuring water level devices had been installed in April 2005 in observation wells and boreholes of pumps all over the study area. Measurements are carried out in a 3 hour interval. Additionally snap shot measurements of the water table are regularly made at the end of dry and rainy season in wells and boreholes in the catchment. During the rainy season the groundwater table rises in the regolith to a height of 3 to 4 m below ground and drops back to a base level of 10 to 15 m below ground during the dry season. Isotope investigation and hydrochemical analyses showed a stagnant water reservoir in the fractures with ages of around 50 years. The regolithic groundwater is composed of more recent waters from precipitations. Only in areas where intensive groundwater pumping is carried out the fractures contain younger water. Water from fractures does not contribute to the river discharge A first stationary model was calibrated to represent the groundwater conditions at the end of the dry season. Furthermore the model was calibrated for the end of the rainy season. Based on daily measurements from the divers validation will be done to connect both stationary models via a transient model. Boundary conditions for the connecting time steps between dry and rainy season are interpolated using linear algorithms for the nodal points. So far the integration of the intermittent flow of the rivers (only rainy season) is not satisfactory solved. At the present state pumping has only small effects on the behaviour of the water table in the model and is outweighed by the impact of climatic conditions. In order to implement the groundwater demand demographic data from the national census and a water consumption study (Schopp, 2004) are used to calculate the total consumption of groundwater for each census settlement. The successful application of groundwater flow models, especially in developing countries, is strongly limited by the availability of data and knowledge about the hydrogeology of the model area. Thus the presented model in its current state can only be a rough approach to show the general behaviour of groundwater flow in the study area and its reaction towards changes in climate and exploitation schemes. New data are obtained by ongoing field work and allow the increasing refinement of the model. Acknowledgements This work was supported by the Federal German Ministry of Education and Research (BMBF) under grant No. 07 GWK 02 and by the Ministry of Science and Research (MWF) of the federal state of Northrhine-Westfalia under grant No. 223-21200200.

REFERENCES Fass, T. (2004) Hydrogeologie im Aguima Einzugsgebiet in Benin / Westafrika. Electronic dissertation at the Faculty of Mathematics and Science of the University of Bonn, Germany. Giertz, S., Diekkrüger, B., Jaeger, A., El-Fahem, T. & M. Schopp (2006): An interdisciplinary scenario analysis to assess the water availability and water demand in the Upper Ouémé catchment in Benin, Advances in Geosciences, Submitted. Jacquin, F. & Seygona, Z. Y. (2004) Contribution à l’étude du fonctionnement hydrodynamique des aquifères du bassin versant de la Donga. Internal report to ORE AMMA/CATCH, IRD, Cotonou, Bénin. Schopp, M.: Wasserversorgung in Benin unter Berücksichtigung sozioökonomischer und soziodemographischer Strukturen - Analyse der Wassernachfrage an ausgewählten Standortendes Haute Ouémé. PhD-Thesis, University of Bonn. http://hss.ulb.uni-bonn.de/diss_online/landw_fak/2005/schopp_marion/index.htm, 2004. WASY (2004) FEFLOW version 5.1, Finite Element Subsurface Flow and Transport Simulation System. Institute of Water Resources Planning and System Research Ltd, Berlin, Germany.

Mixing and spreading of a solute in a non Newtonian fluid flow inside a fracture

A. Boschan, H. Auradou, J.P. Hulin, I. Ippolito and R. Chertcoff

Subsurface fluid flow in many low-permeability geological formations occurs primarily through fracture networks. In order to model such systems, one needs to understand flow in single fractures which are the building blocks of the network. Among the many parameters which influence fluid motion at this scale, we focus here on the geometry of the fractures, specifically on the effect of correlations between the roughness of the two opposite walls on the transport of a solute within the fracture gap. The analysis of the surfaces of both faults and fresh fractures has established that their roughness cannot be described by a finite set of typical wavelength values, but is instead self-affine. The self-affine character of the fracture surfaces has been observed over a significant range of length scales, in a broad variety of materials and for both natural and man-made fractures. We will show that the self-affine nature of the fracture walls controls the aperture field, and has thus an important influence on fluid flow due to the resulting hydrodynamic boundary conditions. Here, we consider two complementary transparent self-affine fracture surfaces with parallel mean planes, which are separated in order to open the fracture and shifted laterally to introduce a shear displacement. Previous studies have shown that the shear displacement strongly influences the void geometry and, therefore, the transport property of the fracture. For instance, the permeability along the direction of the shear is reduced while permeability normal to the shear is enhanced. This effect is related to the appearance of large ridges at right angle to the shear. In the present work, flow is in the direction perpendicular to the shear and the transport of a dye solute is studied by a light absorption technique. The size of the fracture walls is 171 mm by 85 mm and the difference in height between the lowest and the highest points is 3 mm. Futhermore, the normal distance between the two mean planes of the fracture walls is set to 0.75 mm and the lateral (shear) displacement between the two surfaces is 0.33 mm. Experimentally, the fracture is first filled with a transparent polymer solution (scheroglucan) which is pumped out of the fracture at a constant flow rate and replaced by the same liquid but dyed. Images of the invaded region are taken at constant time intervals using a high resolution cooled 12 bits digital camera. Prior calibration measurements are used to relate the light absorbance to the dye concentration. Figure 1 shows a color representation of the tracer distribution during one experiment. Two mechanisms of tracer transport were identified : mixing and spreading. We will show that each process can be studied separately by varying the width (perpendicular to the flow) of the region of interest (ROI). If a width of one pixel (200 microns) is considered, we will show that solute transport can be accurately represented with the classical advection dispersion equation using a constant value for the dispersion coefficient. The latter is found to be related to the mixing of the tracer in the gap of the fracture. Within the fracture aperture, the velocity is maximal in the center of the fracture and drops at the fracture wall. This velocity variation takes place over a relatively small distance (0.75mm here) and a tracer particles can sample the velocity field through molecular diffusion. This results in a mixing mechanism for the tracer known as Taylor dispersion. This process was carefully studied and its dependence on both the flow rate and the polymer concentration was analysed.

If the ROI width is increased up to the fracture width (85 mm) a complete different behavior is observed : the coefficient of dispersion is found to increased linearly with the travel distance. This scale dependence of the dispersion is explained by the occurrence of channels oriented along the flow direction and created by the shear displacement. In order to study this mechanism, the geometry of the iso concentration lines (c/co=0.5) represented in figure 1b was carefully analysed. The structure of the front will be presented in detail. More precisely the development of the digitations as a function of time and their spatial correlation will be discussed. The effect of the polymer concentration on the typical size of the front will also be considered. The dispersion characteristic obtained when fluids flow in a fracture will be compared to the mixing behavior in another similar system. The latter consists in a flat plate set above a rough surface with randomly distributed spikes; this contrasts with the previous case for which the surface roughness displayed long range correlations. As for fractures, the surface do not touch and the spikes pertubate the flow. The influence of roughness on dispersion in this configuration and in the previous one will be compared.

Figure 1: Tracer concentration map obtained for one experiment. The red arrow show the flow direction and the size of the picture is 85 mm by 40 mm. The red domain indicates zone where the fracture gap is fully saturated by the dyed. The blue domain corresponds to non dyed region while the green strip locates mixing domains. The tracer concentration variation in this strip along the flow direction where carefully studied. The black line in the right figure indicates the iso concentration line c/co = 0.5.

Continual estimation of surface and underground flows in river basins

Valeriy Klenov

98-5-80 Profsouznaya, 117485 Moscow, Russia Tel.: 007-095-335-5448, E-mail: [email protected]

The nearly all tasks of water resources management require for assessment of surface and

underground water resources in its dynamics, for the purpose of prediction of water related

accidents, and for estimation of unexpected response of the Environment on human activity.

For these and other goals, it was worked out the 2D Digital River Basin (DRB). The DRB

contains a current state of this Nature System being written in a multi-layer Digital Matrix

(15-20 layers) of variables and parameters. Activation of the DRB is a Virtual System (VS),

which follows up and counts governing processes over an area in analogue with the processes

in Nature System as follows: surface and underground water flows, sediments, and pollution

transport, and others. The discussed are (1) properties of the DRB and the VS, (2) case study

for the small tributary of Moscow River, and (3) conditions of the DRB install.

Ground and watertight surfaces layers determine structure of flows. The DRB converts

precipitation and other exterior influences to following processes: surface flows, water

penetration through unsaturated zone, flows in aquifer, and others water related processes.

The computed Surface and Underground water penetrates a Virtual System as a real flow

penetrates a real Nature System. For example, a gradual filling, overfilling of a reservoir

outcomes breakdown of dam, and catastrophic flood downstream. The DRB calculates flows

for each time step and through each cell by method of the Genetic Matrixes, by estimation of

water/mass exchanges between all neighbor cells (Klenov, 1999, 2004), in accordance to

elevation gradients, water thickness, and water delay. Underground flows are simultaneously

evaluated. The continual estimation of all kinds of flows over the area is the enliven Virtual

System. The Virtual System provides monitoring of processes, of water resources, and

computes response of the area on external natural and human pressure and impacts.

High-resolution matrixes of water infiltration, water permeability, soil resistance, and other

parameters provide reforming of precipitation to surface flows, provide estimation of water

penetration through unsaturated zone, and underground flow over watertight. The

uncertainties appear because of data deficiency for watertight surface, non-conformity of

surface and underground basins, and due non-homogeneous sediment layers. The fast raster-

vector reforming of user’s selected matrixes to contour computer maps facilitates observation

of a real area through its virtual double. The DRB was primary applied for the protection of a

river basin against unexpected consequences of human impacts, in view to find decisions to

minimize hazardous erosion in a basin, and to estimate long time consequences of the human

impact. The object is a small tributary (a big ravine) of the Moscow River, with a zone of

building near watershed. The tasks of the case study were calculation of the Basin response on

human impacts under pressure of as follows: storm precipitation, dam building, well

withdraw, soil destruction, and flows of pollution. The DRB computed many scenarios of the

Basin protection. The case study also includes the estimation of surface – underground water

interaction. Outlets of computed groundwater correlate with the same in the nature. Incessant

computer mapping of selected layers of variables and distributed parameters shows dynamics

of surface and underground water/mass/pollution flows and their interactions, shows response

of aquifer on artificial reservoirs and on wells withdraw. The computing and mapping of the

aquifer surface correlate with the independently observed cross-sections of the area.

The most important property of Open Natural Systems is spatially distributed retard of flows

after exterior influences. The delay for water infiltration and underground flows extends the

common response delay of a system to exterior influence. The Virtual System estimates the

nearest Future by the Outstripping Monitoring, on a base of the System’s Present state. It state

includes underground water resources and its level. It is vital for incessant estimation of

impending disasters (floods, landslides, and outcomes of human activities) for any area and

with a high spatial resolution.

The conditions of the DRB and the Virtual System implementation are as follows:

- In the start, the initial Virtual System must be drowning near the Nature System’s current

state. The river net, aquifer, and reservoirs should be filled by water by preliminary running of

the DRB by continual precipitation until the Virtual Basin to be adequate the current Basin’s

state.

- For the continual estimation of the near Future, The Virtual Systems must be provided by

distributed regular data of exterior influences (precipitation, air temperature, and soil/rock

properties). The join of the Virtual System with the Digital Earth technology will be the

Moving Digital Earth (MDE).

- Far in the Future, the Past and Present influence decreases and the Future of exterior

pressure are not amply predicable. The ways look like in use scenarios of outer reasons until

the Outer Virtual Systems (OVS) will be worked out.

- References

Klenov, V.I. (1999). Simulation of Surface Water – Groundwater Interaction in River

Basins. Proceedings on ModelCARE 1999, Zurich, Switzerland.

Klenov, V.I. (2004) Moving Digital Earth for outstripping monitoring of disasters,

Proceedings of UDMS’04, Chioggia, Venice, Italy, 3.61-3.69.

Investigating the effect of forest stand volume on soil surface erosion by Geographic Information System(GIS). H.R.Maskani.1 , A.Meraji.2

1: Member of scientific board of Iranian Academic Center for Education ,Culture and Research, Guilan branch(A.C.E.C.R.)[email protected] 2: Member of scientific board of Iranian Academic Education , Center for Culture and Research(ACECR) , Guilan [email protected] One of the important problems in forest science is productive bed or forest soil protection, so investigate and research on different aspects of this topic is very important. This article tries to determine the relationship between surface erosion and tree cover, so after a primary study on watersheld domains, two nearby series with an area about 4580 ha were selected and by land inventory it was appeared that these two areas except trees stand volume and climate have approximately similar geological, pedological, steep, dominant direction and altitudinal extension, age and stand composition characteristics. So this area is really suitable to achieve the mentioned goals. After selecting statistics community, the maps the maps that are needed to use erosion model (PSIAC) were prepared and separated into 5.8*5.8 meters cells by use of ArcGIS\Arcmap software and 9 layers that were needed for each cell were prepared and its related measurements have been done and the erosion foe each cell was measured and classified into similar records and about 60 erosional sites for each seri achieved. The mentioned information exported to “excel” and final measurements were done. Results showed that trees covering and surface erosion has a mathematical and inverse ratio. Keywords:Erosion,GIS,Stand Volume.

PREDICTION OF FUTURE DRAWDOWN OF WATER LEVELS OF THE PLEISTOCENE AQUIFER SYSTEM OF WADI EL-ASSIUTI AREA, EASTERN

DESERT, EGYPT

Hossam Hamdy Elewa

National Authority for Remote Sensing and Space Sciences (NARSS), Cairo, Egypt com.@yahoo2hossh :Mail-E

ABSTRACT

Wadi El-Assiuti designates a remarkable dry drainage basin in Egypt with main channel reaching about 186 km in length. It includes potential soils for agricultural development if irrigation water is available. The Pleistocene groundwater aquifer of Wadi El Assiuti area represents the main water source used for irrigation and domestic purposes. This aquifer system has suffered from over consumption during the past few decades. At the proximity area of the Nile Valley (the downstream of this wadi; the old cultivated land), the Pleistocene sediments are mostly influenced by the Nile River recharge, and to a less extent by the surface runoff water from the Eastern Desert wadies. Traveling farthest inside the eastern tributaries of Wadi El-Assiuti, the influence of the River Nile diminishes, and the main recharging sources are those coming from the upward leakage from older formations or from the Red Sea mountainous range, outcropping to the far northeast of the study area. The superfluity of groundwater wells drilled in the recent years, with its consequent over consumption of the groundwater reservoir seems to be clear-cut cause for such levels drawdown and quantity deterioration. The constructed flow net map for the Pleistocene aquifer system in the area of study sketched a pattern of pressures and flow directions within the Pleistocene aquifer. The regional water flow is from east, northeast towards west, southwest directions, with some adverse minor directions at the north-central part of the study area, which are attributed to the structural-bearing and consumption rates of this part of Wadi El-Assiuti. The most important information provided by the equipotentiometric map is the determining of areas suffering from over pumping or piezometric lows. These areas are encountered at the central part of the study area, represented by closed contours. The equipotentiometric map, constructed to illustrate the present piezometric pressures of the Pleistocene aquifer in Wadi El Assiuti, expresses to a great extent the aquifer mismanagement criteria in the study area. However, using empirical formulae and assumptions, the expected drop of piezometric levels of groundwater in the Pleistocene aquifer, as future scenarios, was predicted for five, ten and twenty years of pumping operation with discharge rate of 70 m3 /hr. and operations of groundwater wells for 10 hr/day. The expected future drawdown is expected to reach its maximum value of about 12.4 m. b g. l. by the year of 2025. These future scenarios reflected the urgent need for a policy or management scheme to decrease the severe drawdown of water level in this important developmental area. Key words: Egypt, Eastern Desert, Wadi (dry valley) El-Assiuti, Pleistocene groundwater aquifer, Equipotentiometric map, future scenarios of water levels drawdown.

SHALLOW GROUNDWATER FLOW IN UNSATURATED HILLSLOPES AND ITS IMPLICATIONS FOR LANDSLIDE MOBILISATION

Scott E. Munachen, Geohazard Research Centre

Abstract High infiltration capacities associated with the residual soil mantle of steep, humid hillslopes and the presence of less permeable bedrock at depth create conditions that favour the mobilisation of shallow landslides. Typically, rainwater and snowmelt permeate the subsoil under gravity and capillary forces to an underlying low conductivity layer. The permeability contrast leads to the development of a perched water table, and downslope saturated flow ensues. In regions where shallow subsurface storm flow is the dominant means by which water reaches the channel all incident precipitation must pass through a largely unsaturated soil profile before contributing to runoff. Hence unsaturated zone processes may directly control the timing and magnitude of positive pore pressure development and slope instability. This paper presents the results of a series of field experiments designed to investigate the mechanisms by which rainfall signals propagate through an unsaturated soil profile. The behaviour of a small unchannelled headwater basin, driven to quasi-steady state by sprinkler-irrigation, was monitored by an instrumentation system comprising a network of tensiometers, piezometers, time domain reflectometry probes, discharge meters, and rain gauges. Analysis of the hydrologic response reveals that unsaturated zone dynamics play a primary role in dictating the spatio-temporal evolution of pore pressures and discharge from the hillslope. During initial infiltration some of the deeper tensiometers responded before the arrival of the advancing wetting front. With continued irrigation most tensiometers registered near-zero matric suctions before the majority of piezometers responded fully, and a stable groundwater flow field occurred only after steady state developed in the vadose zone. Interestingly, the data also indicate that the response time of the tensiometers was faster than a simple plug flow approximation. This time lag in the development of the pressure head field in unsaturated slopes is shown to be critical to understanding the mechanisms controlling their stability. The pore-water retention characteristics yielded near-zero pressure heads throughout the soil profile for slight but persistent rain. With the onset of steady discharge the unsaturated zone, saturated zone, and groudwater flux became delicately linked, such that a rapid increase in rainfall intensity led to a saturated zone response and peak discharge which occurred much faster than could have happened through advection alone. The precipitation spike produced a transient pressure wave that travelled relatively rapidly through the unsaturated zone, inducing a large change in hydraulic conductivity and the rapid effusion of stored pore-water. Hence, minor rainstorms which fall on nearly-saturated hillslopes can produce small variations in pressure head that are accompanied by correspondingly large changes in water content, giving rise to the transmission of pressure waves in response to

increased rainfall intensity and a relatively rapid response in the vadose zone. Such a coupled dynamic process is believed to be the underlying mechanism that enables short bursts of rainfall to induce slope instability.

Should Dispersion Describe Mixing or Spreading?

J. Carrera, M. Dentz, and X. Sánchez-Vila

Department of Geotechnics and Applied Geosciences, School of Civil engineering, Technical University of Catalonia, Barcelona, Spain

ABSTRACT Dispersion is the key process controlling pollutants spreading in spatially fluctuating velocity fields. We argue that in applications involving chemical reactions, one is more interested in evaluating the rate at which inflowing water, possibly containing contaminants, mixes with resident water. Both concepts, spreading and mixing, are equivalent in classical advection-dispersion transport formulations with uniform velocity, because dispersion is the only mechanism controlling the two processes. However, in heterogeneous velocity fields, velocity fluctuations promote spreading which in turn leads to increased mixing by interaction with local dispersion and molecular diffusion. That is, the two concepts are linked but different. Unfortunately, existing stochastic concepts on dispersion concentrate on spreading, rather than on mixing. In fact, traditional ensemble macroscopic dispersion measures not only spreading but also uncertainty on the plume centroid. This uncertainty is removed in the effective dispersion concept. Still, when the initial concentration distribution is extended, the rate of growth of effective dispersion is largely controlled by spreading, rather than actual mixing. Here we argue that, when one is interested in representing mixing, one should employ the effective dispersion obtained for an initial pulse input, regardless of the size of the initial plume.

Coupling of wetlands to sea by groundwater-borne nutrient transport (Joseph Sebastian Paimpillil1, K.K. Balachandran2 , T.Joseph2) 1Center for Earth Resources and Environment Management, K.K. Road, Near Parkland Apts, Cochin 17, India, E mail [email protected] 2 National Institute of Oceanography, Regional Center Cochin 14

Abstract In constructing water budget and mass flux estimations for coastal margins, submarine ground water discharge is often overlooked. The ground water discharge influences oceanic chemistry through discharges of nutrients. The coastal waters of Arabian Sea had indications of ground water seepage through the narrow strip of submerged porous lime shell beds running almost parallel to the coast. This supplies considerable quantities of nutrients and precondition the coastal waters for rich primary production. The long-term chlorophyll trend showed a “greening” of the near-shore waters. The poor sanitary facilities of the coastal belt is the main source of nutrients to ground water. The ground water fluxes depend on factors such as: Climatic (Monsoon) variability-which controls the fresh water discharge into backwaters providing the necessary force to overcome the frictional resistance of the porous lime shell deposits; human factors (land use mosaic, socio economic and sanitary conditions) and the tidal factor - controls the hydraulic difference between sea and brackish water. The significant quantity of ground water flow occurs during the monsoon months when water level in the backwater is high and the sea level remains at its annual low. With the heavy rains and flash floods linked with climate variability, such situations can occur in other seasons and at similar oceanic locations. Though the coastal nutrient enrichments, primary productivity boosting and a slow change in biodiversity were identified at few coastal pockets along west coast of India; the details of the exchange of coastal water and groundwater across the sediment-water interface deserve more attention. Key words: Nutrient enrichments, Ground water fluxes, primary productivity

A finite volume analysis on unstructured grids of

aquifers in connection with rivers and lakes

A. NJIFENJOU ∗ A. J. KINFACKEcole Nationale Superieure Polytechnique

University of Yaounde IBP 8390 Yaounde-Cameroon

ABSTRACT:

Modeling the interaction between surface and subsurface flows is a widely stud-ied subject in connection with environmental issues. Due to their conservativecharacter, the finite volume methods are suitable for addressing fluid flow prob-lems. However in their classical setting (one field formulation i.e. potentialformulation ) the finite volume methods display some limitations for accuratelymodeling hydraulic interactions between surface and subsurface flows. For ex-ample it is not easy to model the water penetration into an aquifer from a riverand/or a lake lying on grid blocks boundaries ( see Figure 1 below).

Based upon a two fields formulation (potential and velocity) a mixed hy-brid finite volume ( MHVF) is applied for accuretaly modeling the hydraulicconnection between surface and subsurface flows. This method is constructedon the principles of conservation, using continuity in flux and potential overgrid blocks boundaries as the basic properties. The flexibility of the MHFV liesessentially in these basic properties. In fact these properties play a key role inmodeling interactions between the aquifers and rivers or lakes at the level of gridblocks boundaries. This is the main innovation within this work concerning thefinite volume formulation of surface and subsurface flow. Numerical simulationbased upon these ideas are performed for a real aquifer. The results from thisnumerical experiment are validated by experimental observations •

Key words: finite volume method, groundwater problems, surface/subsurfaceinteraction •

∗Corresponding author: [email protected]

1

Figure 1: Left: an aquifer involving borings (cross points), and in connectionwith rivers (dotted lines) and lakes (bold points).Right: spatial discretization of the aquifer

SURFACE/SUBSURFACE INTERACTION MODELING:

Balance Equation :−div [k H grad H] = f in D

′(Ω)

where we have set: f = ΦPluie +∑

l∈L

ΦLacl δl +

∑r∈R

ΦRivierer δr +

∑f∈F

ΦForagef δf

Boundary conditions :

H = 0 on ΓDirichlet

∂H

∂n= 0 on ΓNeumann

References:A. Njifenjou and E. Kamgnia (2004), “Mixed Hybrid Finite Volume Theory

and Application to Aquifers Simulation”, Proceedings of the 7th African Con-ference on Research in Computer Science, November 22-25, 2004, Hammamet(Tunisia)

A. Njifenjou, E. Kamgnia and D. Bandji, Mixed hybrid finite volume theoryand application to aquifers. To appear in ARIMA Review (Revue Africaine dela Recherche en Informatique et Mathematiques Appliquees)

A. Njifenjou and I. M. Nguena, “A finite volume approximation for secondorder elliptic equations on quadrilateral grids with full matrix: derivation ofthe scheme and a theoretical analysis”, To appear in International Journal onFinite Volumes (IJFV)

2

1

Effet d’un fossé en travers sur l’écoulement hydrodynamique d’une nappe superficielle peu profonde

Application sur le site expérimental de la Jaillière (44, France)

T.H. Debieche 1 , C.V. Adamiade 2 et N. Carluer 3 Cemagref, groupement de Lyon, Unité de Recherche Qualité des Eaux et Prévention des Pollutions, Equipe Pollutions Diffuses. 3 bis, quai Chauveau, 69336 Lyon Cedex 09 1. [email protected] 2. [email protected] 3. [email protected] 1. Introduction Dans le grand ouest de la France, les sols peu profonds sur socle peu perméable induisent la formation de nappes superficielles, vulnérables aux pollutions de surface. Dans les zones agricoles, l’utilisation intensive des pesticides peut provoquer l’infiltration de ces polluants vers les eaux de la nappe et produire une pollution locale, pouvant ensuite rejoindre une nappe plus profonde par drainance, ou le réseau hydrographique par écoulement subsurfacique. Les fossés (éléments du bocage ou fossés de collecte du drainage par tuyaux enterrés) sont susceptibles d’influencer significativement les écoulements sur un versant et de modifier ainsi le transfert des pesticides et leur impact potentiel sur l’environnement (MARGOUM 2003; ADAMIADE 2004). Cette étude aborde l’influence d’un fossé en travers de la pente sur les écoulements au sein d’un versant afin de permettre de caractériser le rôle de collecte, transfert et dissipation que peut avoir un tel fossé sur les pesticides. Nous traiterons plus particulièrement l’évolution du système nappe-fossé au cours d’un événement pluvieux. 2. Présentation du site expérimental Le site expérimental ARVALIS-Institut du végétal de la Jaillière se situe en Loire–Atlantique (44)(fig. 1). Le versant instrumenté sur lequel se déroule l’étude s’étend sur une superficie de 4200 m2, constituant une bande de 60 m de largeur au dessus d’un ruisseau. Sa pente est de 3,2 %. Le dispositif expérimental est composé d’un fossé en travers, de 10m de longueur, situé au milieu de la zone d’étude ; un collecteur de ruissellement pour éviter l’intrusion des eaux du ruissellement vers le fossé ; six piézomètres situés au voisinage du fossé pour étudier la relation hydrodynamique entre la nappe et le fossé ; et quatre piézomètres répartis sur l’ensemble du site pour suivi l’évolution naturelle de la nappe.

2

Angers

Nantes

N

P6PE

S27

S28

S24

20

10

0

30

60

40

50

70

0 50 6020 30 4010

Ruisseau

Fossé

N

PE

S16

S23

S28S18

S14

S26

S21S19

Ruisseau

C

PS26

PS27

PS28

PS24PS23

Déversoir fossé

Déversoir ruissellement

Collecteur du ruissellement

Piézomètre de surface

données

Tuyau d’évacuation

Échelle

CC

PS18

Déversoir fossé


C

GouttièreBac-déversoir

Centrale d’acquisition des données


1 m

Toit du fosséBords du fossé

Piège ruissellementla JaillièreAngers

Nantes

N

P6PE

S27

S28

S24

20

10

0

30

60

40

50

70

0 50 6020 30 4010

Ruisseau

Fossé

N

PE

S16

S23

S28S18

S14

S26

S21S19

Ruisseau

CC

PS26

PS27

PS28

PS24PS23

Déversoir fossé


Collecteur du ruissellement

Piézomètre de surface

données


Échelle

CC

PS18

Déversoir fossé


CC

GouttièreBac-déversoir

Centrale d’acquisition des données


1 m

Toit du fosséBords du fossé

Piège ruissellementla Jaillière

Fig. 1 : Site expérimental de la Jaillière 3. Résultats Le suivi de l’évolution hydrodynamique de la nappe et du fossé, pendant la période du 2000 à 2005, a permet de montrer que le fossé draine la nappe, avec un débit qui varie entre 0 et 1L/s. Le débit est nul lorsque le niveau de la nappe est inférieur à celui du fossé et fort pendant les périodes, où le niveau de la nappe devient supérieur à celui du fossé, surtout pendant les périodes pluvieuses. Le suivi de l’évolution de la relation nappe-fossé pendant les épisodes pluviométriques montre que :

- la remonté de la nappe dans la partie amont du fossé est supérieure à celle de la partie avale, traduisant l’interception des écoulements venant de l’amont et le rôle de barrage du fossé. Le débit drainé par rabattement de la partie avale de la nappe est très faible ;

- La comparaison de l’évolution au cours d’un événement pluvieux de la piézométrie des piézomètres situées à l’amont immédiat du fossé et sur un transect non influencé par le fossé, indique à partir duquel niveau de la nappe l’influence du fossé devient significative.

4. Conclusion Cette étude montre l’effet des fossés sur le fonctionnement hydrodynamique de la nappe ainsi que leur capacité de drainer la nappe et de changer le trajet des eaux souterraines vers le réseaux hydrographique. Les eaux récupérées par le fossé, peuvent présenter un risque majeur sur la qualité des eaux de surface, si ces dernières présentent des fortes teneurs en pesticide. D’où l’intérêt de faire une attention particulière à la qualité des eaux des fossé, surtout pendant les périodes pluvieuses ou le lessivage du sol et le drainage sont importants. Bibliographie ADAMIADE, C.-V. (2004). Influence d'un fossé sur les écoulements rapides au sein d'un

versant. Application au transfert des produits phytosanitaires. Géosciences et ressources naturelles. Mention : Hydrologie, Université Pierre et Marie Curie. 239 pp.

MARGOUM, C. (2003). Contribution à l'étude du devenir des produits phytosanitaires lors d'écoulements dans les fossés : caractérisation physico-chimique et hydrodynamique. Environnement et santé. Grenoble, Université Joseph Fourier. Grenoble I. 243 pp.

Estimation of Spatial-temporal Variability of Groundwater Recharge by Using Fully Coupled SWAT-MODFLOW Model

IL-MOON CHUNG 1, NAM-WON KIM 2

JEONGWOO LEE 3, YOO-SEUNG WON 4

1 Senior Researcher, Korea Institute of Construction Technology, 412-712, Gyunggi-Do,Korea

(Tel: 82-31-910-0334, Fax: 82-31-910-0251, e-mail: [email protected]) 2 Research Fellow, Korea Institute of Construction Technology, 412-712, Gyunggi-Do,Korea

(Tel: 82-31-910-0256, Fax: 82-31-910-0251, e-mail: [email protected]) 3 Post Doctoral, Korea Institute of Construction Technology, 412-712, Gyunggi-Do, Korea

(Tel: 82-31-910-0343, Fax: 82-31-910-0251, e-mail: [email protected]) 4Researcher, River Information Center of Han River Flood Control Office, MOCT, Seoul, 137-049, Korea

(e-mail: [email protected] )

Abstract

Modelling the hydrologic response of a basin with a physically based approach requires the selection of a model that allows the simulation of an interaction between groundwater and surface water. This paper suggests a novel approach of integrating the quasi-distributed watershed model SWAT with the fully-distributed ground-water model MODFLOW. Since SWAT model has semi distributed features, its groundwater component hardly considers distributed parameters such as hydraulic conductivity, storage coefficient. Equally difficult is the detailed representation of groundwater recharge and head distribution. To solve these problems, the method of exchanging characteristics of the hydrologic response units(HRUs) in SWAT and cells in MODFLOW with a fully coupled manner is newly proposed. The linkage is completed by considering the interaction between stream network and aquifer to reflect boundary flow. This approach is applied to Musimcheon basin in Korea. This application demonstrates the combined model enables an estimation of sparial-temporal variability of groundwater recharge and an interaction between saturated zone and channel reaches, which plays an essential role in the runoff generation in the Musimcheon basin. The comprehensive results show the wide applicability of model which represents the temporal-spatial groundwater recharge and head distribution.

Key words SWAT, MODLFOW, spatial-temporal variation of recharge

Coupled Geochemical and Transport Modeling of pH-Dependent Biodegradation of Organic Contaminants in a Pyrite-Rich Aquifer

by

Murat Savas Sarioglu and Nadim K Copty

Bogazici Universitesi Cevre Bilimleri Enstitusu

Bebek, 34342 Istanbul, Turkiye

Abstract

Groundwater pH is one of the key factors influencing subsurface bacterial activity. Numerous studies have shown that the biodegradation of organic contaminants is generally maximum at or near neutral pH. However, in many groundwater systems, such as systems exposed to acid mine drainage, landfill leachate, alkaline lakes or effluents from cement factories, non-neutral pH conditions may occur. The purpose of this study is to develop a coupled reactive transport and geochemical model that explicitly incorporates the effect of spatial and temporal variations of the pH on the biodegradation of organic contaminants.

The developed model consists of two modules: a transport module and a geochemical module. The transport module solves the groundwater flow and transport equations for key components, such as hydrocarbon, dissolved oxygen, microbial mass and other reactive groundwater species influencing the hydrocarbon biodegradation and pH distribution. The geochemical module allows for the simulation of both kinetically defined as well as equilibrium-driven reactions involving the key species. The governing non-linear system of equations is solved using a sequential multi-step operator-splitting algorithm. A Crank-Nicholson finite-difference formulation is used to simulate the advective-dispersive transport of all species. An iterative procedure is used to solve the non-linear ordinary differential equations describing the kinetics of contaminant biodegradation and pyrite oxidation. The algebraic system of equations used to represent the geochemical equilibrium is solved using a Newton-Raphson iterative algorithm. Both modules account for heterogeneity in the definition of the hydrogeological and biochemical parameters.

For demonstration, the model is applied to a hypothetical pyrite-rich aquifer contaminated with petroleum hydrocarbons. Pyrite is a constituent of many soils and rocks and although non-soluble in water, it readily reacts with oxygen to give ferrous and ferric iron and sulfuric acid. Pyrite oxidation reactions have been observed to proceed both abiotically and biotically. A commonly used practice for the remediation of aquifers contaminated with petroleum hydrocarbons is the injection of oxygen for the enhanced aerobic biodegradation of the organic contaminant. However, the presence of pyrite may interfere with the intended purpose of the injected oxygen, leading to undesirable side effects. Specifically, because of competition for the injected oxygen, less oxygen would be available for the contaminant bioremediation. Moreover, the acidification of the subsurface environment may inhibit microbial activity, leading to reduced biodegradation of the organic contaminants.

The developed coupled geochemical and reactive transport model is used to quantify the above processes and assess the relative dominance of the hydrocarbon contaminant degradation and pyrite oxidation. Both abiotic as well as microbially-mediated pyrite-oxidation kinetics are incorporated in the model. The impact of heterogeneity as well as key parameters on the fate and transport of the organic contaminant is also evaluated. The example is used to demonstrate how additional measures such as the injection of an alkaline

solution with the oxygen or use of oxygen releasing barriers may optimize the remediation process.


of Hydraulic Research IAHR-GW2006


IAHR-GW2006 International Ground Water Symposium

Conférence Internationale Eaux Souterraines

12-13-14 June 2006, Toulouse, France http://www.iahr-gw2006.cict.fr

GROUNDWATER HYDRAULICS IN COMPLEX ENVIRONMENTS: 1. Heterogeneity and upscaling; 2. Surface-subsurface coupling;

3. Chemically active transport; 4. Hydromechanics and density effects.

Organized and sponsored by IAHR - Groundwater Hydraulics Section Co-sponsored by IAHS, ASCE/EWRI, SHF, INPT, UPS, CNRS, Région Midi-Pyrénées

Scientific objectives The purpose of the International Conference on “Ground Water Hydraulics in Complex Environments” is to bring together hydro-geologists and hydraulic engineers dealing with complex and coupled flow and transport problems in the subsurface, and to bring them together with scientists and engineers focused on groundwater, porous media and flow modeling. The Conference will provide a forum for dialog and exchange of ideas, which will enhance the ability of different groups of scientists to better describe and understand coupled flow and transport processes, or multi-physics problems, in heterogeneous geologic media and in complex environments. The Conference also aims at promoting fruitful dialog between researchers and practicing engineers and scientists worldwide. The Conference is co-sponsored by two major scientific universities, Université Paul Sabatier and Institut National Polytechnique de Toulouse.

Conference Topics (Thematic Sessions) Theme 1. Heterogeneity and Upscaling.

This theme is devoted to groundwater flow and transport in heterogeneous, discontinuous, fractured media: detailed modeling and upscaling approaches.

Theme 2. Coupling of Surface/Subsurface Flow. This theme focuses on the hydraulic coupling of surface and subsurface flows: stream-aquifer-soil interactions; perched water; flooding, runoff and re-infiltration.

Theme 3. Chemically Active Transport Phenomena This theme is devoted to reactive transport, chemical interactions, multiphase fluid displacements and other coupled transport problems in GW systems.

Theme 4. Coupled hydro-mechanics and density effects. This theme focuses on coupled hydro-mechanics and thermo-hydro-mechanics processes (THM), and on density-driven flows in groundwater systems.

Conference venue The International Ground Water Conference IAHR-GW2006 will be held in a modern conference center, “Météopole”, located in the French National Meteorological Center (Météo France), in the city of Toulouse, France.

The city of Toulouse The city of Toulouse is an architectural marvel of rose pink brick, located on the banks of the Garonne river, in the southwestern region of France known as Midi-Pyrénées. The city hosts the second largest student population in France, and its university is one of the oldest in Europe (the mathematician Fermat lived and worked here). The city is a major international technological and scientific pole on aeronautics, space, earth observation, geophysics and hydrology (Airbus, CNES, SPOT Image, Météo France, National Flood Warning Center…).

Conference steering committee • Rachid ABABOU [email protected]

Chair, IAHR-GW2006 Conference Institut National Polytechnique de Toulouse (INPT) Institut de Mécanique des Fluides de Toulouse (INP-ENSEEIHT / UPS / CNRS), Toulouse, France

• Angelos FINDIKAKIS [email protected] Chair, IAHR Groundwater Hydraulics Section, Bechtel National, Inc., San Francisco, USA

• Members of IAHR Ground Water Hydraulics Section • Members of SHF Ground Water Committee

(Comité Eaux Souterraines de la Société Hydrotechnique de France)

Organizing committee • Rachid ABABOU (IMFT, Toulouse) / Chair • Michel QUINTARD (IMFT, Toulouse) / Co-chair • Marie-Christine TRISTANI (IMFT, Toulouse) / Administration • Ahmad AL-BITAR (IMFT, Toulouse) / Media • José SANCHEZ-PEREZ (LEH, Toulouse) • Philippe RENARD (CHYN Neuchatel, Switzerland) • Philippe BEHRA (ENSIACET, Toulouse) • David LABAT (OMP-LMTG, Toulouse) • Valérie ESTUPINA-BORRELL (CESBIO, Toulouse) • Thierry POINTET (SHF Eaux Souterraines / BRGM Orléans) • David BAILLY (IMFT, Toulouse) • Hocine HENINE (IMFT & CESBIO, Toulouse) • Gérald DEBENEST (IMFT)

International scientific committee • Philippe ACKERER (IMFS Strasbourg, France) • Ross A.C.BAGTZOGLOU (U. Connecticut, Storrs) • Giovanni BARROCU (Univ. Cagliari, Italy) • Philip BINNING (Tech. Univ. of Denmark) • Jesus CARRERA (U. Politec. Barcelona, Spain) • Vladimir CVETKOVIC (Stockholm, Sweden) • Gedeon DAGAN (Tel Aviv, Israel) • Menachem ELIMELECH (U. of Yale, CT, USA) • Javier ELORZA (U. Politec. Madrid, Spain) • Jaime GOMEZ-HERNANDEZ (Valencia, Spain) • Rainer HELMIG (U. of Stuttgart, Germany) • Pierre HUBERT (CIG & Paris School of Mines) • Igor JANKOVIC (SUNY Buffalo, New York)

• Wolfgang KINZELBACH (ETH Zürich, CH) • Peter KITANIDIS (Stanford University, CA, USA) • Ghislain de MARSILY (Univ. of Paris 6, France) • Sekhar MUDDU (IIS, Bangalore, India) • Driss OUAZAR (EMI, Rabat, Morocco) • Claudio PANICONI (INRS Eau, Québec, Canada) • Karsten PRUESS (LBNL, Berkeley, USA) • Yoram RUBIN (Univ. California, Berkeley, USA) • Xavier SANCHEZ-VILLA (Barcelona, Spain) • Fritz STAUFFER (ETH Zürich, Switzerland) • Ed SUDICKY (Univ. of Waterloo, Canada) • Chin-Fu TSANG (LBNL, Berkeley, USA) • Albert VALOCCHI (U. of Illinois, Urbana-Ch.)

Conference calendar and deadlines December 2004 -First Announcement

July 2005 -Second Announcement, Call for Papers, Early Registration

1rst November 2005 -Deadline for abstract submittal (abstracts: 2 pages)

31 January 2006 -Notification of acceptance of abstracts

1rst March 2006 -Opening date for full paper submittal (papers: 12 pp.)

1rst March 2006 -Opening of registration http://www.iahr-gw2006.cict.fr

15 May 2006 -Closing of early registration / Opening of late registration

12 June 2006 -Deadline for full paper submittal (papers: 12 pages)

12-13-14 June 2006 -Conference IAHR-GW2006, Toulouse, France.

Fall 2006 -Review of selected papers for post-conference publication.

Registration fees Early registrations (by the 15th of May 2006): €490 for regular participants (♣) €390 for IAHR, IAHS and ASCE/EWRI members (♣ ) (♦ ) €250 for students (limited offer based on availability) (♣ ) (♥ ) Late registrations (after the 15th of May 2006): €550 for non-student participants (♣ ) €350 for students (depending on availability) (♣) (♥ ) (♦ ) Special membership fee: for the special fee to be applicable, society membership must be

demonstrated by providing sufficient data for verification (early registration only).

(♥ ) Students: a number of “half-price” registration fees will be available for students (including the banquet at no extra-cost) : students should apply by email for this fee, and student status must be demonstrated by providing sufficient data for verification.

(♣ ) Banquet: all registration fees include the banquet at no extra cost to the participants. .

On line information and registration Conference web site: http://www.iahr-gw2006.cict.fr Conference email address: [email protected]

Contact and administration of conference

Marie-Christine TRISTANI (IAHR-GW2006) Institut de Mécanique des Fluides de Toulouse, Allée du Professeur Camille Soula, 31400 Toulouse, France.

Email : [email protected]

Tel: +33 (0)5 61 28 58 31

Fax: +33 (0)5 61 28 59 92

Other contacts Rachid ABABOU (Chairman, IAHR-GW2006 conference):

Rachid ABABOU (Prof., INP de Toulouse) Institut de Mécanique des Fluides de Toulouse Allée du Professeur Camille Soula 31400 Toulouse, France.


Tel: +33 (0)5 61 28 58 45

Fax: +33 (0)5 61 28 58 99

Angelos FINDIKAKIS (Chairman, IAHR Section on Groundwater Hydraulics):

Angelos FINDIKAKIS Bechtel Systems & Infrastructure Inc. Mail stop 333/12/C34, P.O. Box 3965 San Francisco, CA 94119-3965, U.S.A.

Email: [email protected]

Tel. 001-415-768-8550

Fax 001-415-768-4898

Thierry POINTET (Chairman, SHF Ground Water Committee, France):

Thierry POINTET BRGM, Direction du Service public, Service Eau, 3, avenue C. Guillemin (BP 6009) 45060 Orléans Cedex 2, France.


Tel : +33 (0)2 38 64 36 09

Fax : +33 (0)2 38 64 34 46

Related conferences and events CMWR XVI International Conference (Copenhagen, Denmark) Date: 19-22 June 2006 (Monday-Thursday) Contact: Philip BINNING [email protected] Web site: http://www.cmwr-xvi.org/

SWICA III - SWIM 19 : Internat. Conferences on SaltWater Intrusion (Cagliari, Italy) Date: Summer 2006 (date to be announced) Contact: Giovanni BARROCU [email protected] Web site: http:// www.olemiss.edu/sciencenet/saltnet/conf-swica.html


of Hydraulic Research IAHR-GW2006


ADDRESS & MAP OF METEO POLE / BASSO CAMBO AREA Conference venue

The International Ground Water Conference IAHR-GW2006 will be held in a modern conference center, “MÉTÉOPOLE”, located in the French National Meteorological Center (Météo France), in the city of Toulouse (Basso Cambo district), France.

Getting to METEO POLE from downtown Toulouse by public transportation (recommended)

- take metro A direction "BASSO CAMBO" untill end of line.

- take bus N°8 direction "LYCEE POLYVALENT" get down at "METEO" stop.

For complete information about bus-metro schedules and maps, check out the following site: http://www.tisseo-connex.com/horaires/ (or else, see below).

Bus-metro map, and street address, of METEO POLE Toulouse

s METEOPOLE 42, av. G. Coriolis 31057 Toulouse, France Tél. : +33 (0)5 61 07 80 80

Instructions to arrive at METEO POLE (in french)

Vous arrivez en train Prenez le métro à la gare Matabiau jusqu'au terminus Basso Cambo, puis, avec le même ticket, le bus n° 8 direction Lycée Polyvalent. Le bus vous arrêtera devant Météo-France. Durée du trajet : environ 30 minutes.

Vous arrivez en voiture • Depuis l'autoroute Paris-Bordeaux, suivez la direction Foix-Tarbes, vous êtes sur le périphérique. Prenez la sortie 27 : La Cépière. Suivez la direction Cugnaux, puis les Pradettes, puis Météo-Cerfacs.

• Depuis l'autoroute Montpellier-Carcassonne, suivez la direction aéroport Blagnac, en restant sur le périphérique. Prenez la sortie 27 : La Cépière- Cugnaux-Les Pradettes.

• Depuis l'aéroport, prenez la direction Auch, puis la rocade Arc-En-Ciel direction Lardenne- Cugnaux-Tournefeuille, et sortez au Parc d'Activités Basso Cambo. La Météopole est fléchée depuis cette sortie.

Vous arrivez en avion • Le moyen le plus rapide est le taxi : durée du trajet environ 10 minutes.

• Des navettes aéroport/centre-ville partent toutes les 20 minutes, tous les jours. Vous descendez à la gare Matabiau ou à Jean-Jaurès, et vous prenez le métro direction Basso Cambo, et ensuite le bus n° 8 (voir ci-dessus).

Bus-metro map of METEO POLE / BASSO CAMBO

See map next page

Bordelongue

Le Chapitre

La Faourette

La Cépière

Lardenne

Purpan

La CrabePonts Jumeaux

Blagnac

25

26

27

28

29

230

31B

38

N 20 A 62Bordeaux

N 88Albi

N 128Castres

N 124Auch

N 113Carcassonne

A 61Montpellier

A 64Tarbes

Aéroportde ToulouseBlagnac

D15

D112

D50

D2

D23

D50

D632

D2

N

N 20

0 1 2 3km

ToulousePlan d’ensemble A 68

Albi

Météopole

Accès à la MétéopoleuVous arrivez en voiture

• Depuis l'autoroute Paris-Bordeaux, suivez ladirection Foix-Tarbes, vous êtes sur le périphérique.Prenez la sortie 27 : La Cépière. Suivez la directionCugnaux, puis les Pradettes, puis Météo-Cerfacs.

• Depuis l'autoroute Montpellier-Carcassonne,suivez la direction aéroport Blagnac, en restantsur le périphérique. Prenez la sortie 27 : La Cépière-Cugnaux-Les Pradettes.

• Depuis l'aéroport, prenez la direction Auch, puisla rocade Arc-En-Ciel direction Lardenne-Cugnaux-Tournefeuille, et sortez au Parcd'Activités Basso Cambo. La Météopole est flé-chée depuis cette sortie.

vVous arrivez en train

Prenez le métro à la gare Matabiau jusqu'au ter-minus Basso Cambo, puis, avec le même ticket, lebus n° 8 direction Lycée Polyvalent. Le bus vousarrêtera devant Météo-France.Durée du trajet : environ 30 minutes.

fVous arrivez en avion

• Le moyen le plus rapide est le taxi : durée du tra-jet environ 10 minutes.

• Des navettes aéroport/centre-ville partent toutesles 20 minutes, tous les jours. Vous descendez à lagare Matabiau ou à Jean-Jaurès, et vous prenez le métro direction Basso Cambo, et ensuite lebus n° 8 (voir ci-dessus).

M M

M

M

La Cépière27

La Crabe2

Purpan29

Lardenne28

La Faourette26

N 128Auch

Roca

deA

rc-e

n-Ci

el

AvenueD632

Chemin

Av.

Av.

Gasp. Coriolis

Route

Av. duGénéral

Eisenhower

du

du DrGrynfogel

de

seys

ses

Ramelet

deLardenne

Allée

des

Vita

relle

s

Moundi

R. C

-M P

erro

ud

Basso Cambo

Les Pradettes

Bellefontaine

Reynerie

Mirail

D50

D23

Météopole

Le Chapitre38

A64Tarbes

D15

Route

de

St-Simon

Gra

ph

ism

e:

TY

PI

42, av. G. Coriolis31057 Toulouse

Tél. : 05 61 07 80 80

rachid ababou - 12-13-14 june 2006, toulouse,...

Documents