06.12.2012 Views

Simulation of Flow in Porous Media and Applications in Waste ...

Simulation of Flow in Porous Media and Applications in Waste ...

Simulation of Flow in Porous Media and Applications in Waste ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Workshop on<br />

<strong>Simulation</strong> <strong>of</strong> <strong>Flow</strong> <strong>in</strong> <strong>Porous</strong> <strong>Media</strong> <strong>and</strong><br />

<strong>Applications</strong> <strong>in</strong> <strong>Waste</strong> Management<br />

<strong>and</strong> CO2 Sequestration<br />

October 3-7, 2011<br />

as part <strong>of</strong> the<br />

Radon Special Semester 2011 on<br />

Multiscale <strong>Simulation</strong> & Analysis <strong>in</strong> Energy <strong>and</strong> the Environment


This workshop will focus on mathematical <strong>and</strong> computational issues <strong>in</strong> subsurface flow. Subsurface flow<br />

problems are <strong>in</strong>herently multiscale <strong>in</strong> space due to the large variability <strong>of</strong> material properties <strong>and</strong> <strong>in</strong> time<br />

due to the coupl<strong>in</strong>g <strong>of</strong> many different physical processes, such as advection, diffusion, reaction <strong>and</strong> phase<br />

exchange. Subsurface flow models still need considerable development. For example, nonequilibrium<br />

effects, entrapped air, anomalous dispersion <strong>and</strong> hysteresis effects can still not be adequately described.<br />

Moreover, parameters <strong>of</strong> the models are difficult to access <strong>and</strong> <strong>of</strong>ten uncerta<strong>in</strong>, l<strong>in</strong>k<strong>in</strong>g this workshop<br />

to the Workshops 2 <strong>and</strong> 4. Computational issues <strong>in</strong> subsurface flows <strong>in</strong>clude the treatment <strong>of</strong> strong<br />

heterogeneities <strong>and</strong> anisotropies <strong>in</strong> the models, the efficient solution <strong>of</strong> transport-reaction problems with<br />

many species, treatment <strong>of</strong> multiphase-multicomponent flows <strong>and</strong> the coupl<strong>in</strong>g <strong>of</strong> subsurface flow models<br />

to surface flow models given by shallow water or Stokes equations. With respect to energy <strong>and</strong> the environment,<br />

<strong>in</strong> particular the modell<strong>in</strong>g <strong>and</strong> simulation <strong>of</strong> radioactive waste management <strong>and</strong> sequestration<br />

<strong>of</strong> CO2 underground have ga<strong>in</strong>ed high <strong>in</strong>terest <strong>in</strong> the community <strong>in</strong> recent years. Both applications provide<br />

unique challenges rang<strong>in</strong>g from modell<strong>in</strong>g <strong>of</strong> clay materials to treat<strong>in</strong>g very large scale models with<br />

high-performance comput<strong>in</strong>g.<br />

The workshop will br<strong>in</strong>g together key numerical mathematicians whose <strong>in</strong>terest is <strong>in</strong> the analysis <strong>and</strong><br />

computation <strong>of</strong> multiscale subsurface flow <strong>and</strong> practitioners from eng<strong>in</strong>eer<strong>in</strong>g <strong>and</strong> <strong>in</strong>dustry whose <strong>in</strong>terest<br />

is <strong>in</strong> the applications <strong>of</strong> these core problems. Particular problems to be considered will be (i) The design<br />

<strong>of</strong> accurate methods for problems with unresolved (or unresolvable) scales <strong>and</strong>/or uncerta<strong>in</strong>ty <strong>in</strong> data;<br />

(ii) The efficient <strong>and</strong> robust coupl<strong>in</strong>g <strong>of</strong> different models describ<strong>in</strong>g porous media <strong>and</strong> free flow (such as<br />

Darcy, Br<strong>in</strong>kman, Stokes, Richards <strong>and</strong> shallow water); (iii) Fast l<strong>in</strong>ear algebra solvers for subsurface<br />

flow problems <strong>in</strong> heterogeneous media; (iv) <strong>Applications</strong> <strong>of</strong> the above methods <strong>and</strong> ideas <strong>in</strong> energy <strong>and</strong><br />

the environment, such as deep geological disposal <strong>of</strong> radioactive waste or carbon capture <strong>and</strong> storage<br />

underground.<br />

The workshop will also <strong>in</strong>clude a special session on “Fuel Cell Model<strong>in</strong>g” which leads to similar<br />

mathematical problems as subsurface flow simulation.<br />

Workshop Organizers<br />

Peter Bastian, Heidelberg University, Germany<br />

Johannes Kraus, Johann Radon Institute, Austria<br />

Robert Scheichl, University <strong>of</strong> Bath, UK<br />

Mary F. Wheeler, University <strong>of</strong> Texas at Aust<strong>in</strong>, USA


Welcome<br />

to L<strong>in</strong>z <strong>and</strong> thank you very much for participat<strong>in</strong>g <strong>in</strong> the sixth RICAM Special Semester on Multiscale<br />

<strong>Simulation</strong> & Analysis <strong>in</strong> Energy <strong>and</strong> the Environment, hosted by the Johann Radon Insitute for<br />

Computational <strong>and</strong> Applied Mathematics (RICAM) from October 3 to December 16, 2011.<br />

Technological advances have greatly improved our quality <strong>of</strong> life. However, they br<strong>in</strong>g with them a<br />

huge surge <strong>in</strong> energy requirements which <strong>in</strong> turn puts at risk our entire bio-sphere. It is <strong>of</strong> paramount<br />

importance to predict these risks <strong>and</strong> to develop better solutions for the future. One <strong>of</strong> the central tasks<br />

is the accurate simulation <strong>of</strong> multiphase flow above <strong>and</strong> under ground. The risk analysis <strong>and</strong> uncerta<strong>in</strong>ty<br />

quantification, as well as the assimilation <strong>of</strong> data require statistical tools <strong>and</strong> efficient solvers for stochastic<br />

<strong>and</strong> determ<strong>in</strong>istic PDEs as well as for the associated <strong>in</strong>verse problems. The key features that make it<br />

extremely hard to predict these physical phenomena accurately are the multiple time <strong>and</strong> length scales<br />

that arise, as well as the lack <strong>of</strong> <strong>and</strong> uncerta<strong>in</strong>ty <strong>in</strong> data. Because <strong>of</strong> the highly vary<strong>in</strong>g scales <strong>in</strong>volved, the<br />

resolution <strong>of</strong> all scales is currently impossible even on the largest supercomputers. While there is a fairly<br />

long history <strong>of</strong> empirically successful robust computational techniques for certa<strong>in</strong> multiscale problems,<br />

the rigorous (numerical) analysis <strong>of</strong> such methods is <strong>of</strong> extremely high current <strong>in</strong>terest.<br />

The goal <strong>of</strong> the special semester is to provide a stimulat<strong>in</strong>g environment for civil eng<strong>in</strong>eers, hydrologists,<br />

meteorologists <strong>and</strong> other environmental scientists to address together with mathematicians work<strong>in</strong>g at<br />

the cutt<strong>in</strong>g edge <strong>of</strong> rigorous numerical analysis for multiscale (direct <strong>and</strong> <strong>in</strong>verse) problems the emerg<strong>in</strong>g<br />

challenges <strong>in</strong> the quantitative assessment <strong>of</strong> the risks <strong>and</strong> uncerta<strong>in</strong>ties <strong>of</strong> atmospheric <strong>and</strong> subsurface<br />

flow, focus<strong>in</strong>g <strong>in</strong> particular on<br />

• <strong>Simulation</strong> <strong>of</strong> <strong>Flow</strong> <strong>in</strong> <strong>Porous</strong> <strong>Media</strong> <strong>and</strong> <strong>Applications</strong> <strong>in</strong> <strong>Waste</strong> Management <strong>and</strong> CO2 Sequestration<br />

• Large-Scale Inverse Problems <strong>and</strong> <strong>Applications</strong> <strong>in</strong> the Earth Sciences<br />

• Data Assimilation <strong>and</strong> Multiscale <strong>Simulation</strong> <strong>in</strong> Atmospheric <strong>Flow</strong><br />

• Wave Propagation <strong>and</strong> Scatter<strong>in</strong>g, Direct <strong>and</strong> Inverse Problems <strong>and</strong> <strong>Applications</strong> <strong>in</strong> Energy <strong>and</strong><br />

the Environment<br />

• Multiscale Numerical Methods <strong>and</strong> their Analysis <strong>and</strong> <strong>Applications</strong> <strong>in</strong> Energy <strong>and</strong> the Environment<br />

• Stochastic Modell<strong>in</strong>g <strong>of</strong> Uncerta<strong>in</strong>ty <strong>and</strong> Numerical Methods for Stochastic PDEs<br />

Specific activities planned for the Special Semester are<br />

• 4 thematic workshops address<strong>in</strong>g some <strong>of</strong> the key topics <strong>of</strong> the Special Semester;<br />

• Special Lecture Series on ”Multilevel Methods for Multiscale Problems”;<br />

• Graduate Sem<strong>in</strong>ar on ”Multiscale Discretization Techniques”;<br />

• Wednesday Research Kitchen;<br />

• Public Lecture by Pr<strong>of</strong>. Jörn Behrens (KlimaCampus, Universität Hamburg) on<br />

“Tsunami Früh-Warnung: Mathematik und Wissenschaftliches Rechnen im Dienste der Sicherheit”.<br />

We s<strong>in</strong>cerely hope that you enjoy your stay <strong>in</strong> L<strong>in</strong>z!<br />

Local Organiz<strong>in</strong>g Committee Program Committee<br />

Robert Scheichl, Bath & RICAM (Chair) Peter Bastian, University <strong>of</strong> Heidelberg, Germany<br />

Jörg Willems, RICAM (Coord<strong>in</strong>ator) Mike Cullen, MET Office, Exeter, UK<br />

Johannes Kraus, RICAM (Co-Coord<strong>in</strong>ator) He<strong>in</strong>z Engl, RICAM & University <strong>of</strong> Vienna, Austria<br />

Erw<strong>in</strong> Karer, RICAM (Co-Coord<strong>in</strong>ator) Mel<strong>in</strong>a Freitag, University <strong>of</strong> Bath, UK<br />

Ivan G. Graham, University <strong>of</strong> Bath, UK<br />

Ulrich Langer, RICAM & University <strong>of</strong> L<strong>in</strong>z, Austria<br />

Markus Melenk, TU Vienna, Austria<br />

Robert Scheichl, University <strong>of</strong> Bath, UK (Chair)<br />

Mary F. Wheeler, University <strong>of</strong> Texas at Aust<strong>in</strong>, USA


Contents<br />

Information 2<br />

Workshop Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2<br />

Social Events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2<br />

Restaurants <strong>and</strong> Cafes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2<br />

General Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2<br />

Program 5<br />

Posters 7<br />

Abstracts 8<br />

Abstracts for Posters 18<br />

List <strong>of</strong> Participants 22<br />

1


Information<br />

Workshop Information<br />

Registration. The workshop registration will be on October 3rd, 2011 from 8:30 - 9:10 next to the<br />

sem<strong>in</strong>ar room SP2 416 on the 4th floor <strong>of</strong> the Science Park Build<strong>in</strong>g 2 (see floor plan). Participants<br />

that arrive later <strong>in</strong> the week can register at the special semester <strong>of</strong>fice SP2 456.<br />

Registration Fee. Non-<strong>in</strong>vited participants are k<strong>in</strong>dly asked to pay the registration fee <strong>in</strong> cash upon<br />

registration.<br />

Campus plan <strong>and</strong> overview map as well as a floor plan <strong>of</strong> the 4th floor <strong>of</strong> the workshop venue (Science<br />

Park Build<strong>in</strong>g 2) are located on the next pages.<br />

Sem<strong>in</strong>ar room. The workshop will take place <strong>in</strong> sem<strong>in</strong>ar room SP2 416 on the 4th floor <strong>of</strong> the Science<br />

Park Build<strong>in</strong>g 2 (see floor plan).<br />

Program. A time schedule for the workshop is located on the backside <strong>of</strong> this booklet.<br />

C<strong>of</strong>fee breaks. The c<strong>of</strong>fee breaks will be <strong>in</strong> the corridor <strong>of</strong> the 4th floor <strong>of</strong> the Science Park Build<strong>in</strong>g 2.<br />

Internet access. There will be an extra <strong>in</strong>formation sheet regard<strong>in</strong>g <strong>in</strong>ternet access available at registration.<br />

Social Events<br />

Welcome Reception & Poster Session. Monday, October 3th, 2011, 5:00 pm, at the 4th floor <strong>of</strong> the<br />

Science Park Build<strong>in</strong>g 2.<br />

Conference D<strong>in</strong>ner. Thursday, October 6th, 2011, 7:00 pm, at the restaurant “Kepler’s“, situated <strong>in</strong><br />

the Mensa build<strong>in</strong>g.<br />

Restaurants <strong>and</strong> Cafes<br />

• Mensa Markt (lunch time only) - Ma<strong>in</strong> canteen <strong>of</strong> the University (see campus plan)<br />

• KHG Mensa (lunch time only) - Smaller canteen - good traditional food (see overview map: “KHG<br />

L<strong>in</strong>z”)<br />

• Pizzeria “Bella Casa” - Italian <strong>and</strong> Greek restaurant (located next to the tram stop)<br />

• Ch<strong>in</strong>ese restaurant “Jadegarten” - (located close by the tram stop, adjacent to “Bella Casa”)<br />

• Asia restaurant “A2” - (located beh<strong>in</strong>d the Science Park on Altenbergerstrasse)<br />

• “Chat” cafe - c<strong>of</strong>fee, dr<strong>in</strong>ks <strong>and</strong> s<strong>and</strong>wiches (located <strong>in</strong> the “Hörsaaltrakt” - see overview map)<br />

• Cafe “Sassi” - c<strong>of</strong>fee, dr<strong>in</strong>ks <strong>and</strong> small snacks (located <strong>in</strong> the build<strong>in</strong>g “Johannes Kepler Universität”<br />

- see overview map)<br />

• Bakery “K<strong>and</strong>ur” - bakery <strong>and</strong> small cafe (located opposite the tram stop)<br />

General Information<br />

Accommodation. The arranged accomodation for <strong>in</strong>vited participants is the “Sommerhaus” hotel. You<br />

can f<strong>in</strong>d its location <strong>in</strong> the overview map on page 4.<br />

Special Semester Office: Room SP2 456. The special semester adm<strong>in</strong>istrator is Susanne Dujard<strong>in</strong>.<br />

Audiovisual & Computer Support. Room SP2 458, Wolfgang Forsthuber or Florian Tischler.<br />

2


Orientation/ Local Transport. From the railway station you have to take tram number 1 or 2 <strong>in</strong><br />

direction “Universität”. It takes about 25 m<strong>in</strong>utes to reach the desired end stop “Universität”.<br />

In order to get to the city center <strong>of</strong> L<strong>in</strong>z (“Hauptplatz”) <strong>and</strong> back you have to take aga<strong>in</strong> tram<br />

number 1 or 2 (about 20 m<strong>in</strong>utes). For more <strong>in</strong>formation see www.ricam.oeaw.ac.at/location/.<br />

Taxi Numbers.<br />

+43 732 6969 Oberösterreichische Taxigenossenschaft<br />

+43 732 2244 2244 L<strong>in</strong>zer Taxi<br />

+43 732 781463 Enzendorfer Taxi & Transport<br />

+43 732 2214 L<strong>in</strong>zer Taxi<br />

+43 732 660217 LINTAX TaxibetriebsgesmbH<br />

Further important phone numbers.<br />

+43 (0)732 2468 5222 RICAM & Special Semester Office (Susanne Dujard<strong>in</strong>)<br />

+43 (0)732 2468 5250/5255 RICAM IT Support (Florian Tischler/ Wolfgang Forsthuber)<br />

+43 (0)732 2457-0 Reception <strong>of</strong> Hotel Sommerhaus<br />

133 General emergency number for the police<br />

144 General emergency number for the ambulance<br />

More <strong>in</strong>formation about RICAM can be found at www.ricam.oeaw.ac.at. See also the Special<br />

Semester webpage www.ricam.oeaw.ac.at/specsem/specsem2011/ for additional <strong>in</strong>formation.<br />

Figure 1: 4th floor <strong>of</strong> Science Park Build<strong>in</strong>g 2.<br />

Figure 2: Campus plan<br />

3


Figure 3: Overview map<br />

4


Program<br />

Monday, October 3rd<br />

08:30 - 09:10 Registration<br />

09:10 - 09:20 Open<strong>in</strong>g<br />

09:20 - 10:10 Mary F. Wheeler (The University <strong>of</strong> Texas at Aust<strong>in</strong>)<br />

10:10 - 10:40 C<strong>of</strong>fee Break<br />

“Evaluat<strong>in</strong>g Long Term CO2 Storage <strong>in</strong> Sal<strong>in</strong>e Aquifers”<br />

10:40 - 11:30 Jan Mart<strong>in</strong> Nordbotten (University <strong>of</strong> Bergen)<br />

11:30 - 14:00 Lunch Break<br />

“Multiscale Model<strong>in</strong>g for CO2 Storage”<br />

14:00 - 14:50 Mart<strong>in</strong> Vohralík (Université Pierre et Marie Curie (Paris 6))<br />

14:50 - 15:20 C<strong>of</strong>fee Break<br />

“Robust a Posteriori Error Control <strong>and</strong> Adaptivity for Multiscale, Mult<strong>in</strong>umerics,<br />

<strong>and</strong> Mortar Coupl<strong>in</strong>g”<br />

15:20 - 16:10 Barbara Wohlmuth (Fakultät Mathematik, TU München)<br />

“A New Coupl<strong>in</strong>g Concept for Two-Phase Compositional <strong>Porous</strong> <strong>Media</strong> <strong>and</strong> S<strong>in</strong>gle-<br />

Phase Compositional Free <strong>Flow</strong>”<br />

17:00 Poster Session & Welcome Reception<br />

Tuesday, October 4th<br />

08:30 - 09:20 Helge K. Dahle (University <strong>of</strong> Bergen)<br />

“CO2-Migration: Effects <strong>and</strong> Upscal<strong>in</strong>g <strong>of</strong> Caprock Topography.”<br />

09:20 - 10:10 Laurence Halpern (LAGA. Université Paris 13)<br />

10:10 - 10:40 C<strong>of</strong>fee Break<br />

“Optimized Schwarz Waveform Relaxation for Reactive Transport Problems”<br />

10:40 - 11:30 Anthony Michel (IFP Energies nouvelles)<br />

11:30 - 14:00 Lunch Break<br />

“Time-Space Doma<strong>in</strong> Decomposition for Reactive Transport <strong>in</strong> <strong>Porous</strong> <strong>Media</strong> ”<br />

14:00 - 14:50 Andrew Cliffe (University <strong>of</strong> Nott<strong>in</strong>gham)<br />

14:50 - 15:20 C<strong>of</strong>fee Break<br />

“What should we do with Radioactive <strong>Waste</strong>?”<br />

15:20 - 16:10 Robert Marschall<strong>in</strong>ger ( ÖAW GIScience)<br />

“Multi-scale 3D <strong>and</strong> 4D Modell<strong>in</strong>g <strong>and</strong> <strong>Simulation</strong> <strong>in</strong> Geosciences”<br />

5


Wednesday, October 5th<br />

08:30 - 09:20 Jürgen Becker (Fraunh<strong>of</strong>er ITWM)<br />

“Pore-Scale Modell<strong>in</strong>g <strong>of</strong> <strong>Porous</strong> Layers Used <strong>in</strong> Fuel Cells”<br />

09:20 - 10:10 Andro Mikelić (Université Lyon 1)<br />

10:10 - 10:40 C<strong>of</strong>fee Break<br />

“Effective Pressure Interface Law for Transport Phenomena between an Unconf<strong>in</strong>ed<br />

Fluid <strong>and</strong> a <strong>Porous</strong> Medium us<strong>in</strong>g Homogenization”<br />

10:40 - 11:30 Jürgen Fuhrmann (Weierstrass Institute Berl<strong>in</strong>)<br />

11:30 - 14:00 Lunch Break<br />

“Electrochemical Processes <strong>and</strong> <strong>Porous</strong> <strong>Media</strong>: Mathematical <strong>and</strong> Numerical Model<strong>in</strong>g”<br />

14:00 - 14:50 Marco Discacciati (Universitat Politècnica de Catalunya, Barcelona)<br />

14:50 - 15:20 C<strong>of</strong>fee Break<br />

“Coupl<strong>in</strong>g Free <strong>and</strong> <strong>Porous</strong>-<strong>Media</strong> <strong>Flow</strong>s: Model<strong>in</strong>g, Analysis, <strong>and</strong> Numerical Approximation”<br />

15:20 - 16:10 Ivan Yotov (University <strong>of</strong> Pittsburgh)<br />

Thursday, October 6th<br />

“Mortar Multiscale Framework for Stokes-Darcy <strong>Flow</strong>s”<br />

08:30 - 09:20 Jocelyne Erhel (INRIA; Campus Universitaire de Beaulieu, Rennes, France)<br />

“Numerical <strong>and</strong> Stochastic Models <strong>of</strong> <strong>Flow</strong> <strong>in</strong> 3D Discrete Fracture Networks”<br />

09:20 - 10:10 Ala<strong>in</strong> Bourgeat (Université Lyon 1 - UCB, France)<br />

10:10 - 10:40 C<strong>of</strong>fee Break<br />

“Model<strong>in</strong>g Compressible Multiphase <strong>Flow</strong> <strong>and</strong> Transport <strong>in</strong> Saturated-Unsaturated<br />

<strong>Porous</strong> <strong>Media</strong>: Phase Appearance-Disappearance. Application to Gas Migration<br />

<strong>in</strong> Underground Nuclear <strong>Waste</strong> Repository”<br />

10:40 - 11:30 Alex<strong>and</strong>re Ern (Université Paris-Est, CERMICS)<br />

11:30 - 14:00 Lunch Break<br />

“Discont<strong>in</strong>uous Galerk<strong>in</strong> Method for Two-Component Miscible Liquid-Gas <strong>Porous</strong><br />

<strong>Media</strong> <strong>Flow</strong>s”<br />

14:00 - 14:50 Peter Bastian (Universität Heidelberg)<br />

14:50 C<strong>of</strong>fee Break<br />

“Numerical Solution <strong>of</strong> Compositional Two-Phase <strong>Flow</strong> <strong>in</strong> <strong>Porous</strong> <strong>Media</strong>”<br />

19:00 Conference D<strong>in</strong>ner<br />

6


Friday, October 7th<br />

08:30 - 09:20 Yalch<strong>in</strong> Efendiev (Texas A&M University)<br />

“Local-Global Multiscale Model Reduction Techniques for <strong>Flow</strong>s <strong>in</strong> Heterogeneous<br />

<strong>Porous</strong> <strong>Media</strong>”<br />

09:20 - 10:10 Clemens Pechste<strong>in</strong> (Johannes Kepler Universität L<strong>in</strong>z)<br />

10:10 - 10:40 C<strong>of</strong>fee Break<br />

“An Abstract Two-Level Additive Schwarz Method for Systems with High Contrast<br />

Coefficients”<br />

10:40 - 11:30 Ralf Kornhuber (Freie Universität Berl<strong>in</strong>)<br />

11:30 Clos<strong>in</strong>g<br />

Posters<br />

“Coupled Surface <strong>and</strong> Saturated/Unsaturated Ground Water <strong>Flow</strong> <strong>in</strong> Heterogeneous<br />

<strong>Media</strong>”<br />

The poster session will take place on the 4th floor <strong>of</strong> the 2nd Science Park Build<strong>in</strong>g. It will start<br />

at 5:00 pm on Monday, October 3rd.<br />

Sergey Alyaev (Universitetet i Bergen)<br />

“Multiscale <strong>Simulation</strong>s <strong>of</strong> Non-Darcy’s <strong>Flow</strong>s”<br />

Radim Blaheta (Institute <strong>of</strong> Geonics AS CR, Ostrava, CZ)<br />

“Micromechanics <strong>of</strong> Geomaterials <strong>and</strong> Geocomposites.”<br />

Thomas Carraro (University <strong>of</strong> Heidelberg)<br />

“Modell<strong>in</strong>g, <strong>Simulation</strong> <strong>and</strong> Optimization <strong>of</strong> the Microstructure <strong>of</strong> SOFC <strong>Porous</strong> Cathodes”<br />

Christian Engwer (Institut für Numerische und Angew<strong>and</strong>te Mathematik, Universität Münster)<br />

“Numerical Upscal<strong>in</strong>g <strong>in</strong> <strong>Porous</strong> <strong>Media</strong>”<br />

Ivan Georgiev (Radon Institute for Computational <strong>and</strong> Applied Mathematics (RICAM), L<strong>in</strong>z)<br />

“Precondition<strong>in</strong>g <strong>of</strong> Non-Conform<strong>in</strong>g FEM Systems”<br />

Christian Goll (University <strong>of</strong> Heidelberg)<br />

“Design <strong>of</strong> Numerical Methods to Simulate Models <strong>of</strong> a Solid Oxide Fuel Cell”<br />

Sridhara Nayak (Indian Institute <strong>of</strong> Technology, Kharagpur)<br />

“Impact on the Surface Temperature due to the Modifications <strong>of</strong> Underly<strong>in</strong>g L<strong>and</strong> Surface Conditions:<br />

A Study over Western India”<br />

Rebecca Neumann (University <strong>of</strong> Heidelberg)<br />

“Model<strong>in</strong>g Two-phase <strong>Flow</strong> with Disappear<strong>in</strong>g Gas Phase”<br />

Joerg Willems (Radon Institute for Computational <strong>and</strong> Applied Mathematics (RICAM), L<strong>in</strong>z)<br />

“Robust Preconditioners for General SPD Operators”<br />

7


Abstracts<br />

“Numerical Solution <strong>of</strong> Compositional Two-Phase <strong>Flow</strong> <strong>in</strong> <strong>Porous</strong> <strong>Media</strong>”<br />

Peter Bastian<br />

Universität Heidelberg<br />

Interdiszipl<strong>in</strong>äres Zentrum für Wissenschaftliches Rechnen<br />

Im Neuenheimer Feld 368, D-69120 Heidelberg<br />

In this talk I present two different methods for solv<strong>in</strong>g two-phase flow problems <strong>in</strong> porous media.<br />

The first method is a fully-coupled discont<strong>in</strong>uous Galerk<strong>in</strong> scheme for <strong>in</strong>compressible two-phase flow<br />

<strong>in</strong> heterogeneous media. It is based on a wett<strong>in</strong>g-phase pressure/capillary pressure formulation <strong>and</strong><br />

assumes the diffusion-dom<strong>in</strong>ated regime. These primary variables are cont<strong>in</strong>uous when both phases<br />

are present <strong>and</strong> simplify the formulation <strong>of</strong> the consistency <strong>and</strong> penalty terms at media discont<strong>in</strong>uities.<br />

The scheme can h<strong>and</strong>le unstructured, nonconform<strong>in</strong>g grids <strong>and</strong> tensor permeabilities <strong>in</strong> multiple space<br />

dimensions. For the time discretization diagonally implicit Runge-Kutta schemes are employed to<br />

obta<strong>in</strong> a formally higher-order method <strong>in</strong> space <strong>and</strong> time. Numerical comparison with a cell-centered<br />

f<strong>in</strong>ite volume method us<strong>in</strong>g full upw<strong>in</strong>d<strong>in</strong>g <strong>and</strong> implicit Euler shows that the new scheme achieves<br />

very accurate solutions on rather coarse grids.<br />

The second scheme treats compositional two-phase flow with phase disappearance. Based on a<br />

nonwett<strong>in</strong>g-phase pressure/capillary pressure formulation the scheme can h<strong>and</strong>le the disappearance<br />

<strong>of</strong> the nonwett<strong>in</strong>g phase without switch<strong>in</strong>g <strong>of</strong> variables or add<strong>in</strong>g complementarity constra<strong>in</strong>ts. <strong>Applications</strong><br />

<strong>of</strong> the method to the MoMas phase disappearance benchmark <strong>and</strong> a CO2 sequestration<br />

benchmark are presented.<br />

The last part <strong>of</strong> the talk will illustrate <strong>in</strong> some detail the s<strong>of</strong>tware implementation <strong>of</strong> these <strong>and</strong> other<br />

numerical schemes with<strong>in</strong> the DUNE s<strong>of</strong>tware framework (http://www.dune-project.org). This<br />

framework <strong>of</strong>ffers great flexibility for implement<strong>in</strong>g state-<strong>of</strong>-the-art discretization schemes <strong>and</strong> solvers<br />

while reta<strong>in</strong><strong>in</strong>g high performance <strong>and</strong> easy parallelization.<br />

This presentation is jo<strong>in</strong>t work with Olaf Ippisch <strong>and</strong> Rebecca Neumann.<br />

“Pore-Scale Modell<strong>in</strong>g <strong>of</strong> <strong>Porous</strong> Layers Used <strong>in</strong> Fuel Cells”<br />

Jürgen Becker<br />

Fraunh<strong>of</strong>er ITWM,<br />

Fraunh<strong>of</strong>er-Platz 1,<br />

67663 Kaiserslautern<br />

In a proton exchange membrane (PEM) fuel cell a proton-conduct<strong>in</strong>g polymer membrane separates<br />

anode <strong>and</strong> cathode. The membrane, together with catalyst <strong>and</strong> diffusion layers on both sides, forms<br />

the membrane electrode assembly (MEA). This MEA consists <strong>of</strong> several porous layers on both the<br />

cathode <strong>and</strong> the anode side:<br />

• the gas diffusion layer (GDL), usually a non-woven or woven carbon paper,<br />

• the micro-porous layer (MPL), usually made <strong>of</strong> carbon agglomerates, <strong>and</strong><br />

• the catalyst layer (CL), which consists <strong>of</strong> carbon agglomerates, plat<strong>in</strong>um particles <strong>and</strong> polymer<br />

electrolyte.<br />

The physical properties <strong>of</strong> such a porous layer are described by its macro-homogeneous properties, e.g.<br />

porosity, diffusivity, permeability <strong>and</strong> conductivity. Improv<strong>in</strong>g the performance <strong>of</strong> the cell is possible<br />

by improv<strong>in</strong>g the cell as a whole, but also by optimiz<strong>in</strong>g each <strong>of</strong> the layers to its requirements.<br />

Optimization <strong>in</strong>cludes improv<strong>in</strong>g the micro-structure <strong>of</strong> each layer as material connectivity <strong>and</strong> pore<br />

morphology have a major impact on the properties.<br />

Therefore, <strong>in</strong> this talk I will present<br />

1. how to create 3D virtual structure models <strong>of</strong> each layer,<br />

2. how to determ<strong>in</strong>e the effective properties <strong>of</strong> the layer numerically, <strong>and</strong><br />

3. how to validate this approach.<br />

The 3D structure models are created us<strong>in</strong>g stochastic processes; For example, GDL models are created<br />

by plac<strong>in</strong>g fibres r<strong>and</strong>omly us<strong>in</strong>g an anisotropic direction distribution. These fibres may be straight<br />

or bent, where the bend<strong>in</strong>g is determ<strong>in</strong>ed by its own r<strong>and</strong>om process. For MPL <strong>and</strong> CL, agglomerate<br />

structures <strong>of</strong> carbon particles are created r<strong>and</strong>omly.<br />

8


Us<strong>in</strong>g these 3D models, the effective properties <strong>of</strong> the layers can be predicted by solv<strong>in</strong>g the appropriate<br />

partial differential equations on the pore-scale model. To determ<strong>in</strong>e s<strong>in</strong>gle-phase permeability, Stokes’<br />

equation is solved. Two-phase permeability values, which are <strong>of</strong> <strong>in</strong>terest due to the production <strong>of</strong><br />

water at the cathode, are determ<strong>in</strong>ed by comb<strong>in</strong><strong>in</strong>g s<strong>in</strong>gle-phase simulations with the pore morphology<br />

method. The effective diffusivity is determ<strong>in</strong>ed solv<strong>in</strong>g Laplace’s equation. For MPL <strong>and</strong> CL, also<br />

Knudsen diffusion is considered, as pore sizes become comparable to the mean free path <strong>of</strong> the gas.<br />

To validate this approach I compare experimentally measured GDL properties with numerically determ<strong>in</strong>ed<br />

properties. For this purpose, a tomography image <strong>of</strong> the GDL is used as model to ensure<br />

micro-structural geometric similarity between experiment <strong>and</strong> simulation.<br />

“Model<strong>in</strong>g Compressible Multiphase <strong>Flow</strong> <strong>and</strong> Transport <strong>in</strong> Saturated-Unsaturated <strong>Porous</strong> <strong>Media</strong>: Phase<br />

Appearance-Disappearance. Application to Gas Migration <strong>in</strong> Underground Nuclear <strong>Waste</strong> Repository”<br />

Ala<strong>in</strong> Bourgeat<br />

Université Lyon 1 - UCB, France<br />

Motivated by modell<strong>in</strong>g the gas migration <strong>in</strong> an underground nuclear waste repository, we derive a new<br />

compositional model <strong>of</strong> compressible multiphase flow <strong>and</strong> transport <strong>in</strong> porous media, with <strong>in</strong>terphase<br />

mass transfer, The liquid phase is composed <strong>of</strong> water <strong>and</strong> dissolved hydrogen <strong>and</strong> the gas phase is<br />

composed <strong>of</strong> vaporized water <strong>and</strong> hydrogen. One <strong>of</strong> the ma<strong>in</strong> difficulty appear<strong>in</strong>g <strong>in</strong> the usual models<br />

is their <strong>in</strong>adequacy to take <strong>in</strong> account both fully <strong>and</strong> partially water saturated situations ; lead<strong>in</strong>g<br />

to numerical problems or unphysical numerical constra<strong>in</strong>ts. The new unified model<strong>in</strong>g <strong>of</strong> fully <strong>and</strong><br />

partially water saturated porous materials, presented <strong>in</strong> our talk, is based on fundamental pr<strong>in</strong>ciples<br />

<strong>of</strong> fluid mechanic <strong>and</strong> thermodynamic; <strong>and</strong> its ma<strong>in</strong> <strong>in</strong>terest is to make possible a unified numerical<br />

treatment <strong>of</strong> fully <strong>and</strong> partially water saturated situations. Under adequate assumptions, the existence<br />

<strong>of</strong> solutions for equations correspond<strong>in</strong>g to this ”unified fully partially saturated” formulation could<br />

be proved. We present numerical simulations show<strong>in</strong>g the efficiency <strong>of</strong> this modell<strong>in</strong>g.<br />

“What should we do with Radioactive <strong>Waste</strong>?”<br />

Andrew Cliffe<br />

School <strong>of</strong> Mathematical Sciences,<br />

University <strong>of</strong> Nott<strong>in</strong>gham<br />

Most radioactive waste arises from civil nuclear power programmes. Op<strong>in</strong>ions are sharply divided when<br />

it comes to nuclear power: some countries <strong>in</strong>tend to phase it out altogether, others are heavily reliant<br />

on it for electricity generation <strong>and</strong> will be for the foreseeable future. However, everyone is agreed<br />

that radioactive wastes must be disposed <strong>of</strong> <strong>in</strong> a safe manner. This talk will review the options that<br />

have been considered for safe long-term disposal <strong>of</strong> radioactive waste <strong>and</strong> expla<strong>in</strong> why deep geological<br />

disposal is now the (almost) universally preferred solution. A deep geological disposal facility is a<br />

comb<strong>in</strong>ation <strong>of</strong> an eng<strong>in</strong>eered <strong>and</strong> a natural system <strong>and</strong> any analysis <strong>of</strong> the safety performance <strong>of</strong> such<br />

a facility must take <strong>in</strong>to account the uncerta<strong>in</strong>ties <strong>in</strong>herent <strong>in</strong> natural systems. The techniques used<br />

to deal with uncerta<strong>in</strong>ty will be outl<strong>in</strong>ed. The important case <strong>of</strong> the uncerta<strong>in</strong>ties aris<strong>in</strong>g from limited<br />

<strong>in</strong>formation about rock properties <strong>and</strong> their impact on transport <strong>of</strong> radionuclides from a deep disposal<br />

facility back to the human environment will be considered <strong>in</strong> some detail, illustrat<strong>in</strong>g the importance<br />

<strong>of</strong> modern developments <strong>in</strong> uncerta<strong>in</strong>ty quantification <strong>and</strong> the solution <strong>of</strong> stochastic partial differential<br />

equations.<br />

“CO2-migration: Effects <strong>and</strong> Upscal<strong>in</strong>g <strong>of</strong> Caprock Topography.”<br />

Helge K. Dahle<br />

Dept. <strong>of</strong> Mathematics,<br />

University <strong>of</strong> Bergen, Norway<br />

In prospective sites for CO2 storage, such as a sal<strong>in</strong>e aquifer or ab<strong>and</strong>oned petroleum reservoir, the<br />

subsurface conditions are such that CO2 is a supercritical fluid <strong>and</strong> slightly soluble <strong>in</strong> water <strong>and</strong> thus<br />

forms a separate fluid phase that is less dense <strong>and</strong> much less viscous than the resident br<strong>in</strong>e. Because<br />

<strong>of</strong> the viscosity <strong>and</strong> density differences the CO2 will be captured under the caprock <strong>and</strong> then migrate<br />

as a th<strong>in</strong> plume under the <strong>in</strong>fluence <strong>of</strong> gravity. It follows that the dynamics <strong>and</strong> distribution <strong>of</strong> the<br />

9


CO2 plume will depend on the topography <strong>of</strong> the caprock.<br />

In this presentation we demonstrate how caprock topography may impact plume migration <strong>and</strong> distribution.<br />

In particular we note how caprock roughness reduces migration speeds. Open issues to be<br />

discussed are how subscale caprock roughness may be upscaled, <strong>and</strong> how roughness can be parameterized.<br />

The presentation is based on jo<strong>in</strong>t work with Halvor M. Nilsen, SINTEF ICT, Norway <strong>and</strong> Sarah<br />

Gasda, Uni Research, Norway<br />

“Coupl<strong>in</strong>g Free <strong>and</strong> <strong>Porous</strong>-<strong>Media</strong> <strong>Flow</strong>s: Model<strong>in</strong>g, Analysis, <strong>and</strong> Numerical Approximation”<br />

Marco Discacciati<br />

Laboratori de Càlcul Numèric (LaCàN).<br />

Universitat Politècnica de Catalunya (UPC BarcelonaTech),<br />

Barcelona, Spa<strong>in</strong>.<br />

marco.discacciati@upc.edu<br />

In this talk we consider the model<strong>in</strong>g <strong>and</strong> numerical simulation <strong>of</strong> coupled free/porous-media flows.<br />

We present different model<strong>in</strong>g approaches <strong>in</strong>volv<strong>in</strong>g the Navier-Stokes equations to represent the<br />

<strong>in</strong>compressible fluid <strong>and</strong> Darcy <strong>of</strong> Fochheimer equations <strong>in</strong> the porous region. We discuss <strong>and</strong> compare<br />

different coupl<strong>in</strong>g strategies based on the so-called penalization method or on the application <strong>of</strong> the<br />

transmission conditions <strong>of</strong> Beavers-Joseph-Saffman.<br />

After giv<strong>in</strong>g some results on the well-posedness <strong>of</strong> the result<strong>in</strong>g coupled models, we illustrate a doma<strong>in</strong><br />

decomposition framework to set up possible iterative methods to compute their f<strong>in</strong>ite element solution.<br />

We discuss the effectiveness <strong>and</strong> robustness <strong>of</strong> such algorithms show<strong>in</strong>g several numerical results.<br />

F<strong>in</strong>ally, we present some possible applications <strong>of</strong> the models that we have studied.<br />

“Local-Global Multiscale Model Reduction Techniques for <strong>Flow</strong>s <strong>in</strong> Heterogeneous <strong>Porous</strong> <strong>Media</strong>”<br />

Yalch<strong>in</strong> Efendiev<br />

Texas A&M University<br />

The development <strong>of</strong> numerical algorithms for simulations <strong>of</strong> flow processes <strong>in</strong> large-scale highly heterogeneous<br />

porous formations is challeng<strong>in</strong>g because properties <strong>of</strong> natural geologic porous formations<br />

(e.g., permeability) display high variability <strong>and</strong> complex spatial correlation structures which can span<br />

a hierarchy <strong>of</strong> length scales. It is usually necessary to resolve a wide range <strong>of</strong> length <strong>and</strong> time scales,<br />

which can be prohibitively expensive, <strong>in</strong> order to obta<strong>in</strong> accurate predictions <strong>of</strong> the flow, mechanical<br />

deformation, <strong>and</strong> transport processes under <strong>in</strong>vestigation. In practice, some types <strong>of</strong> coarsen<strong>in</strong>g (or<br />

upscal<strong>in</strong>g) <strong>of</strong> the detailed model are usually performed before the model can be used to simulate<br />

complex processes. Many approaches have been developed <strong>and</strong> applied successfully when a scale separation<br />

adequately describes the spatial variability <strong>of</strong> the subsurface properties (e.g., permeability) that<br />

have bounded variations. The quality <strong>of</strong> these approaches deteriorates for comp! lex heterogeneities<br />

without scale separation <strong>and</strong> high contrast. In this talk, I will describe multiscale model reduction<br />

techniques that can be used to systematically reduce the degrees <strong>of</strong> freedoms <strong>of</strong> f<strong>in</strong>e-scale simulations<br />

<strong>and</strong> discuss applications to preconditioners <strong>and</strong> coupl<strong>in</strong>g to global model reduction tools. Numerical<br />

results will be presented that show that one can improve the accuracy <strong>of</strong> multiscale methods<br />

by systematically add<strong>in</strong>g new coarse basis functions, obta<strong>in</strong> contrast-<strong>in</strong>dependent preconditioners for<br />

complex heterogeneities, <strong>and</strong> get reduced order models at low cost.<br />

10


“Numerical <strong>and</strong> Stochastic Models <strong>of</strong> <strong>Flow</strong> <strong>in</strong> 3D Discrete Fracture Networks”<br />

Jocelyne Erhel<br />

INRIA; Campus Universitaire de Beaulieu,<br />

Rennes, France;<br />

Jocelyne.Erhel@<strong>in</strong>ria.fr<br />

This is jo<strong>in</strong>t work with Jean-Raynald de Dreuzy (CNRS, UMR Geosciences, Rennes), Gérald<strong>in</strong>e Pichot<br />

(INRIA, Rennes) <strong>and</strong> Baptiste Poirriez (INSA, UMR IRISA, Rennes).<br />

Underground water is naturally channelled <strong>in</strong> fractured media, where <strong>in</strong>terconnections can be very<br />

<strong>in</strong>tricate. Discrete Fracture Networks are based on a geometrical model, where fractures are 2D<br />

doma<strong>in</strong>s, for example ellipses, which form a 3D network. Stochastic models rely on probability laws<br />

to def<strong>in</strong>e the geometry, <strong>in</strong> particular the length distribution <strong>of</strong> fractures. In our model, we def<strong>in</strong>e a<br />

power law, which <strong>in</strong>duces a large range <strong>of</strong> scales. This multiscale model is quite challeng<strong>in</strong>g to h<strong>and</strong>le,<br />

s<strong>in</strong>ce a few large fractures can <strong>in</strong>teract with many small fractures. Physical laws are quite simple, with<br />

Poseuille’s flow equations <strong>in</strong>side each fracture <strong>and</strong> cont<strong>in</strong>uity <strong>of</strong> pressure <strong>and</strong> flux at the <strong>in</strong>tersections<br />

between the fractures. Boundary conditions are applied at the borders <strong>of</strong> the fractures <strong>and</strong> <strong>of</strong> the<br />

network, which is limited by a f<strong>in</strong>ite cube. Our numerical model relies on a mixed hybrid f<strong>in</strong>ite element<br />

method, which is well suited for such problems. With a conform<strong>in</strong>g mesh, cont<strong>in</strong>uity conditions are<br />

easy to apply, but with a non conform<strong>in</strong>g mesh, mortar relations must be devised. A non conform<strong>in</strong>g<br />

mesh allows mesh<strong>in</strong>g the fractures <strong>in</strong>dependently, <strong>in</strong> order to adapt the mesh step to the fracture<br />

scale. Because the network is generated r<strong>and</strong>omly, mesh generation can fail when small angles appear<br />

at the <strong>in</strong>terface <strong>of</strong> a fracture (the <strong>in</strong>terface is the set <strong>of</strong> <strong>in</strong>tersections with other fractures). We have<br />

developed an orig<strong>in</strong>al discretization procedure, with a two-step algorithm, act<strong>in</strong>g first at the network<br />

level then at the fracture level. This method removes small angles so that mesh generation is feasible<br />

<strong>and</strong> leads to a good aspect ratio. On the other h<strong>and</strong>, this approach creates geometrical configurations<br />

where more than two fractures can share an <strong>in</strong>tersection edge. This br<strong>in</strong>gs up a new difficulty for the<br />

mortar method, where new relations must be def<strong>in</strong>ed. This spatial discretization leads to a symmetric<br />

positive def<strong>in</strong>ite l<strong>in</strong>ear system, with a large sparse matrix. We have <strong>in</strong>vestigated the efficiency <strong>of</strong> several<br />

l<strong>in</strong>ear solvers on parallel <strong>and</strong> distributed computers. S<strong>in</strong>ce the network can be easily partitioned <strong>in</strong>to<br />

subdoma<strong>in</strong>s (the fractures), we have developed a hybrid solver based on doma<strong>in</strong> decomposition. This<br />

approach decouples <strong>in</strong> some sense the flow at the fracture scale from the <strong>in</strong>teractions at the network<br />

scale. The Schur complement, which gathers the unknowns at the <strong>in</strong>tersections <strong>of</strong> the network, is solved<br />

with a preconditioned conjugate gradient. We comb<strong>in</strong>e Neumann Neumann, def<strong>in</strong>ed at the fracture<br />

scale, with a global preconditioner, def<strong>in</strong>ed at the network scale, based on a deflation formulation.<br />

The numerical model <strong>and</strong> the solver are implemented <strong>in</strong> the s<strong>of</strong>tware MPFRAC, which is embedded<br />

<strong>in</strong>to the scientific platform H2OLab. Our numerical experiments highlight the efficiency <strong>of</strong> this Schur<br />

solver.<br />

“Discont<strong>in</strong>uous Galerk<strong>in</strong> Method for Two-Component Miscible Liquid-Gas <strong>Porous</strong> <strong>Media</strong> <strong>Flow</strong>s”<br />

Alex<strong>and</strong>re Ern<br />

Université Paris-Est, CERMICS,<br />

Ecole des Ponts, 6& 8 Av. B. Pascal,<br />

77455 Marne-la-Vallée cedex 2, France<br />

We consider two-component (typically, water <strong>and</strong> hydrogen) miscible liquid-gas porous media flows<br />

<strong>in</strong>clud<strong>in</strong>g gas-phase (dis)appearance, as motivated by hydrogen production <strong>in</strong> underground repositories<br />

for radioactive waste. Follow<strong>in</strong>g Smaï’s Ph.D. thesis (Université Claude Bernard Lyon, 2009),<br />

we formulate the govern<strong>in</strong>g equations <strong>in</strong> terms <strong>of</strong> liquid pressure <strong>and</strong> dissolved hydrogen density as<br />

ma<strong>in</strong> unknowns, lead<strong>in</strong>g mathematically to a nonl<strong>in</strong>ear elliptic-parabolic system <strong>of</strong> PDEs, <strong>in</strong> which the<br />

equations degenerate when the gas phase disappears. We develop a discont<strong>in</strong>uous Galerk<strong>in</strong> method<br />

for space discretization, comb<strong>in</strong>ed with a backward Euler scheme for time discretization <strong>and</strong> an <strong>in</strong>complete<br />

Newton method for l<strong>in</strong>earization. Numerical examples deal with gas-phase (dis)appearance,<br />

ill-prepared <strong>in</strong>itial conditions, <strong>and</strong> heterogeneous problem with different rock types. This is jo<strong>in</strong>t work<br />

with I. Mozolevski (Universidade Federal de Santa Catar<strong>in</strong>a, Brazil).<br />

11


“Electrochemical Processes <strong>and</strong> <strong>Porous</strong> <strong>Media</strong>: Mathematical <strong>and</strong> Numerical Model<strong>in</strong>g”<br />

Jürgen Fuhrmann<br />

Weierstrass Institute for Applied Analysis <strong>and</strong> Stochastics<br />

Mohrenstr. 39<br />

10117 Berl<strong>in</strong>, Germany<br />

Electrochemical effects <strong>in</strong> porous media are ubiquitous <strong>in</strong> nature, <strong>in</strong>dustry <strong>and</strong> everyday life. Important<br />

examples are electrochemical energy conversion <strong>and</strong> storage devices like fuel cells <strong>and</strong> batteries,<br />

where porous electrodes allow to provide a large <strong>in</strong>terface area between electrolytes <strong>and</strong> solid state<br />

conductors.<br />

In electrochemical systems, flow, transport, reactions <strong>and</strong> electric field are tightly coupled. To <strong>in</strong>troduce<br />

the general sett<strong>in</strong>g <strong>of</strong> model<strong>in</strong>g <strong>in</strong> electrochemical devices, we start with the Nernst-Planck-<br />

Poisson system describ<strong>in</strong>g the movement <strong>of</strong> ions due to advection, diffusion <strong>and</strong> the self-consistent<br />

electric field established by the distribution <strong>of</strong> charged particles. This system is coupled to electrochemical<br />

reactions on electrode surfaces. As outside <strong>of</strong> boundary layers, electro-neutrality prevails,<br />

this system can be converted to Ohm’s law with special boundary reaction terms. In a porous medium,<br />

a further upscal<strong>in</strong>g step results <strong>in</strong> the porous electrode equations which are a base <strong>in</strong>gredient <strong>of</strong> models<br />

<strong>of</strong> electrochemical systems conta<strong>in</strong><strong>in</strong>g porous structures.<br />

As a result we present a multiphase-multicomponent ansatz <strong>in</strong> porous media augmented by equations<br />

describ<strong>in</strong>g the electrostatic potential, which <strong>in</strong>fluences transport <strong>and</strong> reactions <strong>in</strong> various ways.<br />

The implicit Euler, Voronoi box based f<strong>in</strong>ite volume method allows to derive a framework for the<br />

numerical implementation <strong>of</strong> mathematical models based on reaction-diffusion-convection systems.<br />

Particular advantages <strong>of</strong> the method are unconditional stability, positivity, discrete maximum pr<strong>in</strong>ciple,<br />

local <strong>and</strong> global mass conservation, <strong>and</strong> efficient ways to solve stationary <strong>and</strong> time dependent<br />

cases. It relies on the ability to create Delaunay meshes conform<strong>in</strong>g to <strong>in</strong>terior <strong>and</strong> exterior boundaries.<br />

We mention challenges connected with the resolution <strong>of</strong> boundary layers <strong>and</strong> h<strong>and</strong>l<strong>in</strong>g <strong>of</strong> anisotropies.<br />

The talk concludes with results from applications which have been implemented us<strong>in</strong>g the approach<br />

presented.<br />

“Optimized Schwarz Waveform Relaxation for Reactive Transport Problems”<br />

Laurence Halpern<br />

LAGA. University Paris 13.<br />

93430 Villetaneuse, France<br />

Schwarz waveform relaxation algorithms have been <strong>in</strong>troduced <strong>in</strong> the last decade to solve l<strong>in</strong>ear problems<br />

<strong>of</strong> various types <strong>in</strong> parallel by doma<strong>in</strong> decomposition <strong>in</strong> space. Each subdoma<strong>in</strong> evolves with<br />

its own time grid <strong>in</strong> time w<strong>in</strong>dows, <strong>and</strong> the iteration between the subdoma<strong>in</strong>s takes place at the end<br />

<strong>of</strong> the time w<strong>in</strong>dow. Much attention has been paid earlier to the design <strong>of</strong> <strong>of</strong> fast algorithms for<br />

l<strong>in</strong>ear problems, by choos<strong>in</strong>g the type <strong>and</strong> the coefficients <strong>of</strong> the transmission conditions between the<br />

subdoma<strong>in</strong>s [1].<br />

In the reactive transport process, a small time scale is needed <strong>in</strong> a reactive zone localized <strong>in</strong> the<br />

neighbourhood <strong>of</strong> mobile <strong>in</strong>terfaces between water <strong>and</strong> gas, whereas larger time scales are <strong>in</strong>volved <strong>in</strong><br />

the mildly reactive zone. This is the motivation <strong>of</strong> the SHP-CO2 project with the french ANR <strong>and</strong><br />

the IFPEN. With<strong>in</strong> this project, we have <strong>in</strong>troduced various types <strong>of</strong> Schwarz waveform relaxation<br />

algorithms for non-l<strong>in</strong>ear systems [2,3]. They are coupled with Krylov acceleration <strong>and</strong> Newton<br />

algorithms for the non-l<strong>in</strong>ear part. The aim <strong>of</strong> this talk is to present our latest progress from a<br />

theoretical po<strong>in</strong>t <strong>of</strong> view. Numerical aspects will be evaluated on toy problems. These results are a<br />

jo<strong>in</strong>t activity with the PhD student Florian Haeberle<strong>in</strong> <strong>and</strong> Anthony Michel at IFP, Filipa Caetano<br />

at University Paris 11, J’er’emie Szeftel at ENS, Mart<strong>in</strong> G<strong>and</strong>er at Geneva University. <strong>Applications</strong><br />

<strong>of</strong> the strategies to real problems will be presented by Anthony Michel.<br />

References<br />

[1] D. Bennequ<strong>in</strong>, M. G<strong>and</strong>er <strong>and</strong> L. Halpern. A Homographic Best Approximation Problem with<br />

Application to Optimized Schwarz Waveform Relaxation. Math. Comp. 78 (2009), no. 265, 185223.<br />

[2] F. Caetano, M. G<strong>and</strong>er, L. Halpern <strong>and</strong> J. Szeftel. Schwarz waveform relaxation algorithms for<br />

semil<strong>in</strong>ear reaction-diffusion. Networks <strong>and</strong> heterogeneous media Volume 5, Number 3, pp 487–505,<br />

September 2010.<br />

[3]F. Haeberle<strong>in</strong>, L. Halpern & A. Michel, Krylov subspace accelerators for non-overlapp<strong>in</strong>g Schwarz<br />

waveform relaxation methods applied to coupled nonl<strong>in</strong>ear reactive transport systems <strong>in</strong> the context<br />

<strong>of</strong> CO2 geological storage simulation. Proceed<strong>in</strong>g DD20, 2011, submitted.<br />

12


“Coupled Surface <strong>and</strong> Saturated/Unsaturated Ground Water <strong>Flow</strong> <strong>in</strong> Heterogeneous <strong>Media</strong>”<br />

Ralf Kornhuber<br />

Freie Universität Berl<strong>in</strong>,<br />

Institut für Mathematik,<br />

Arnimallee 6, 14195 Berl<strong>in</strong><br />

Richards equations for saturated/unsaturated groundwater flow is based on state equations relat<strong>in</strong>g<br />

saturation to capillary pressure. The numerical solution <strong>of</strong> the result<strong>in</strong>g degenerate parabolic problems<br />

typically suffers from strong nonl<strong>in</strong>earities <strong>and</strong> ill-condition<strong>in</strong>g <strong>in</strong> the presence <strong>of</strong> strongly vary<strong>in</strong>g<br />

saturation. As a remedy, we suggest a solver-friendly discretization based on Kirchh<strong>of</strong>f transformation<br />

which can be re<strong>in</strong>terpreted <strong>in</strong> physical variables <strong>in</strong> terms <strong>of</strong> suitable quadrature rules. In this way<br />

ill-condition<strong>in</strong>g is separated from the numerical solution process. We show convergence <strong>and</strong> provide<br />

error estimattes This approach is extended to heterogeneous state equations by doma<strong>in</strong> decomposition<br />

methods based on nonl<strong>in</strong>ear transmission conditions. We suggest suitable coupl<strong>in</strong>g conditions <strong>of</strong><br />

ground water flow <strong>and</strong> shallow water equations for surface water flow. The coupled system is solved<br />

by a Dirichlet-Neumann-type iteration account<strong>in</strong>g for the multiple time scales. The performance <strong>of</strong><br />

the algorithms is illustrated by numerical computations.<br />

“Multi-scale 3D <strong>and</strong> 4D Modell<strong>in</strong>g <strong>and</strong> <strong>Simulation</strong> <strong>in</strong> Geosciences”<br />

Robert Marschall<strong>in</strong>ger<br />

ÖAW GIScience,<br />

Schillerstr. 30,<br />

5020 Salzburg<br />

Geoscience comput<strong>in</strong>g typically deals with three-dimensional (“3D”) spatially referenced data; <strong>in</strong><br />

process modell<strong>in</strong>g, time enters as an extra, fourth dimension. Mostly, volumes are <strong>in</strong>volved; thus,<br />

geoscience modell<strong>in</strong>g <strong>and</strong> simulation necessitates volume data structures that can h<strong>and</strong>le large ranges<br />

<strong>of</strong> object scale, geometric precision <strong>and</strong> shape complexity. Solid modell<strong>in</strong>g is state-<strong>of</strong>-the-art for portray<strong>in</strong>g<br />

volume objects because it backs up the unambiguous representation <strong>of</strong> arbitrarily shaped 3D<br />

objects <strong>and</strong> provides automated analysis <strong>and</strong> Boolean relationship <strong>of</strong> the <strong>in</strong>volved solid objects. Much<br />

as <strong>in</strong> the 2D realm, where areal data can be represented either <strong>in</strong> vector or raster form, <strong>in</strong> the 3D<br />

doma<strong>in</strong>, boundary representation (“B-Rep”) <strong>and</strong> voxel modell<strong>in</strong>g (“VM”) are used to portray volume<br />

data. B-Rep aggregates the total solid model from self-conta<strong>in</strong>ed, complexly shaped solid objects<br />

<strong>and</strong> has its strengths <strong>in</strong> the quasi-natural presentation <strong>and</strong> <strong>in</strong> a straightforward, manual edit<strong>in</strong>g <strong>of</strong><br />

objects. By contrast VM, with the roots <strong>in</strong> 2D raster data process<strong>in</strong>g, is based on tesselat<strong>in</strong>g the<br />

whole model universe. VM, as the st<strong>and</strong>ard output format <strong>of</strong> geoscience <strong>in</strong>terpolation <strong>and</strong> simulation<br />

algorithms, excels <strong>in</strong> represent<strong>in</strong>g cont<strong>in</strong>uous spatial variation <strong>and</strong> <strong>in</strong> analytical flexibility. Both<br />

B-Rep <strong>and</strong> VM are generically scale-<strong>in</strong>variant; i.e., the full scale range covered by geosciences – from<br />

microscopic to megascopic objects – can be described with B-Rep <strong>and</strong> VM. Abstract<strong>in</strong>g the geoscience<br />

analysis, modell<strong>in</strong>g <strong>and</strong> simulation frameworks <strong>in</strong> a 3D modell<strong>in</strong>g universe or <strong>in</strong> a space-time<br />

(hyper)prism, exhaustiveness/sparsity <strong>and</strong> precision/error <strong>of</strong> orig<strong>in</strong>al (measured or acquired) data can<br />

be explored <strong>in</strong> space <strong>and</strong> space/time. In the macro or mega scales, because <strong>of</strong> the high costs <strong>of</strong> 3D<br />

subsurface data acquisition (e.g., drill<strong>in</strong>g or geophysical campaigns), the associated modell<strong>in</strong>g universes<br />

<strong>and</strong> space-time (hyper)prisms conta<strong>in</strong> only sparse <strong>and</strong> clustered orig<strong>in</strong>al data. This is why,<br />

for the macroscopic/megascopic scales, reliable <strong>and</strong> efficient 3D/4D modell<strong>in</strong>g <strong>and</strong> <strong>in</strong>terpolation <strong>and</strong><br />

simulation algorithms are m<strong>and</strong>atory to fill the spatial/spatiotemporal “gaps” <strong>in</strong> the respective modell<strong>in</strong>g<br />

universes or space-time hyperprisms. Especially <strong>in</strong> the macro- <strong>and</strong> mega scales, geostatistics<br />

plays a central role <strong>in</strong> provid<strong>in</strong>g stochastic-based estimation <strong>and</strong> simulation <strong>of</strong> spatial <strong>and</strong> space-time<br />

data <strong>and</strong> associated <strong>in</strong>certa<strong>in</strong>ty. Toward the other end <strong>of</strong> the geoscience scale range, data acquisition<br />

<strong>in</strong> the micro-scale has become straightforward <strong>and</strong> affordable. Recent advances <strong>in</strong> destruction-free<br />

materials analysis like micro computed tomography (“µCT”) or Neutron Tomography (“NT”) make<br />

the (micro)structures <strong>in</strong> geological materials accessible as VMs: paleontologic fossil associations, hard<br />

rock textures <strong>and</strong> sediment structures can be portrayed <strong>in</strong> detail. Today, the micro-scale geoscience<br />

modell<strong>in</strong>g universes are exhaustively filled with orig<strong>in</strong>al data! Object based image analysis <strong>in</strong> its multidimensional<br />

flavour (“xd-OBIA”), is essential for mean<strong>in</strong>gful classification <strong>of</strong> geoscience µCT data:<br />

petrographical, sedimentological or paleontological expert knowledge can be successfully <strong>in</strong>corporated<br />

<strong>in</strong> the classification process to build precise 3D microstructure VMs from orig<strong>in</strong>ally noisy, s<strong>in</strong>gle channel<br />

µCT image stacks. Comb<strong>in</strong><strong>in</strong>g the above-mentioned aspects <strong>and</strong> methods, a novel testbed with<br />

stunn<strong>in</strong>g research opportunities is opened, cross-fertiliz<strong>in</strong>g different research fields: xd-OBIA <strong>in</strong> comb<strong>in</strong>ation<br />

with µCT <strong>and</strong> NT can be applied to create precise VMs <strong>of</strong> geoscientific micro-objects. These<br />

13


exhaustive data sets can be utilized as “Gold” St<strong>and</strong>ards, the spatial variability <strong>of</strong> which can be<br />

quantified by geostatistical variography <strong>and</strong> multipe-po<strong>in</strong>t geostatistical tra<strong>in</strong><strong>in</strong>g images. Sequentially<br />

th<strong>in</strong>n<strong>in</strong>g out the exhaustive data, geostatistical <strong>in</strong>terpolation <strong>and</strong> simulation can be benchmarked, <strong>in</strong><br />

a controlled environment <strong>of</strong> natural structures, to re-build the “Gold” St<strong>and</strong>ard. In the end, with the<br />

well-known paradigm <strong>of</strong> scale-<strong>in</strong>variance <strong>of</strong> geological structures <strong>in</strong> m<strong>in</strong>d, relevant geostatistical modell<strong>in</strong>g<br />

<strong>and</strong> simulation <strong>in</strong>sights derived from geoscience microstructures will hopefully be propagated<br />

back to the macro- <strong>and</strong> mega scales to create more precise 3D <strong>and</strong> 4D geoscience models from sparse<br />

<strong>in</strong>put data.<br />

“Time-Space Doma<strong>in</strong> Decomposition for Reactive Transport <strong>in</strong> <strong>Porous</strong> <strong>Media</strong> ”<br />

Anthony Michel<br />

IFP Energies nouvelles,<br />

1 et 4 avenue de Bois-Preau,<br />

92852 Rueil-Malmaison Cedex - France<br />

In this talk I will present applications <strong>of</strong> time-space doma<strong>in</strong> decomposition techniques to multi-species<br />

reactive transport problems. This is a jo<strong>in</strong>t work with F. Haeberle<strong>in</strong> <strong>and</strong> L. Halpern which contributes<br />

to the ANR-SHPCO2 project on High Performance <strong>Simulation</strong> for CO2 Geological Storage. The<br />

theoretical background will be presented <strong>in</strong> another talk by L. Halpern.<br />

“Effective Pressure Interface Law for Transport Phenomena between an Unconf<strong>in</strong>ed Fluid <strong>and</strong> a <strong>Porous</strong><br />

Medium us<strong>in</strong>g Homogenization”<br />

Andro Mikelić<br />

Université Lyon 1,<br />

Département de mathématiques <strong>and</strong> Institut Camille Jordan,<br />

Villeurbanne, France<br />

We present model<strong>in</strong>g <strong>of</strong> the <strong>in</strong>compressible viscous flows <strong>in</strong> the doma<strong>in</strong> conta<strong>in</strong><strong>in</strong>g unconf<strong>in</strong>ed fluid<br />

<strong>and</strong> a porous medium. For such sett<strong>in</strong>g a rigorous derivation <strong>of</strong> the Beavers-Joseph-Saffman <strong>in</strong>terface<br />

condition was undertaken by Jäger <strong>and</strong> Mikelić [SIAM J. Appl. Math. 60 (2000), p. 1111-1127] us<strong>in</strong>g<br />

the homogenization method. So far the <strong>in</strong>terface law for the pressure was conceived <strong>and</strong> confirmed<br />

only numerically. In this article we derive the Beavers <strong>and</strong> Joseph law for a general body force by<br />

estimat<strong>in</strong>g the pressure field approximation. Different than <strong>in</strong> the Poiseuille flow case, the velocity<br />

approximation is not divergence-free <strong>and</strong> the precise pressure estimation is essential. F<strong>in</strong>ally, this new<br />

estimate allows us to justify rigorously the pressure jump condition us<strong>in</strong>g the Navier boundary layer,<br />

already used to calculate the constant <strong>in</strong> the law by Beavers <strong>and</strong> Joseph.<br />

This is a jo<strong>in</strong>t work with Anna Marc<strong>in</strong>iak-Czochra (IWR <strong>and</strong> BIOQUANT, Universität Heidelberg,<br />

Germany)<br />

“Multiscale Model<strong>in</strong>g for CO2 Storage”<br />

Jan Mart<strong>in</strong> Nordbotten<br />

Department <strong>of</strong> Mathematics<br />

University <strong>of</strong> Bergen<br />

The large spatial <strong>and</strong> temporal scales over which the evolution <strong>of</strong> stored CO2 must be assessed, forces<br />

us to consider coarse spatial <strong>and</strong> temporal discretizations. Such discretizations cannot reasonably be<br />

expected to be convergent to the st<strong>and</strong>ard equations <strong>of</strong> porous media flow, which are valid on much<br />

f<strong>in</strong>er scales, <strong>and</strong> have solution structures that cannot be resolved on coarse grids.<br />

Therefore, there is a need to develop coarse-scale models that are suitable to be discretized by the<br />

computational grids we can afford.<br />

In this talk, we discuss the framework for develop<strong>in</strong>g such coarse models, emphasiz<strong>in</strong>g the development<br />

<strong>of</strong> both coarse 3D <strong>and</strong> 2D models for CO2 storage <strong>in</strong> geological formations. We will pay particular<br />

attention to cases where the parameter functions change not only quantitatively, but also qualitatively,<br />

<strong>and</strong> highlight the novel computational considerations that arise as a consequence there<strong>of</strong>.<br />

14


“An Abstract Two-Level Additive Schwarz Method for Systems with High Contrast Coefficients”<br />

Clemens Pechste<strong>in</strong><br />

Institute <strong>of</strong> Computational Mathematics<br />

Johannes Kepler University<br />

Altenberger Str. 69, 4040 L<strong>in</strong>z (A)<br />

clemens.pechste<strong>in</strong>@numa.uni-l<strong>in</strong>z.ac.at<br />

Robust doma<strong>in</strong> decomposition solvers for the f<strong>in</strong>ite element discretization <strong>of</strong> scalar elliptic problems<br />

with heterogeneous (high contrast, or multiscale) coefficients have been studied for some time. This<br />

<strong>in</strong>cludes both iterative substructur<strong>in</strong>g methods <strong>and</strong> overlapp<strong>in</strong>g Schwarz methods. Us<strong>in</strong>g weighted<br />

Po<strong>in</strong>caré <strong>in</strong>equalities <strong>and</strong>/or spectral theory, robustness can theoretically be guaranteed <strong>in</strong> a large<br />

variety <strong>of</strong> cases. However, there is only little theory available for systems <strong>of</strong> PDEs, such as l<strong>in</strong>earized<br />

elasticity or problems <strong>in</strong> H(curl).<br />

In this talk, we consider an abstract framework for overlapp<strong>in</strong>g Schwarz methods for variationally<br />

posed systems <strong>of</strong> PDEs. Essentially, we only require positive semi-def<strong>in</strong>ite element stiffness matrices<br />

<strong>and</strong> their connectivity. As usual, the local subspaces are def<strong>in</strong>ed on overlapp<strong>in</strong>g subdoma<strong>in</strong>s. The key<br />

<strong>in</strong>gredient is an abstract coarse space constructed from a particular eigenproblem <strong>in</strong> the overlap <strong>of</strong> each<br />

subdoma<strong>in</strong> as well as an algebraic partition <strong>of</strong> unity. We provide a rigorous <strong>and</strong> robust convergence<br />

theory <strong>and</strong> show some numerical results for the cases <strong>of</strong> Darcy <strong>and</strong> l<strong>in</strong>earized elasticity.<br />

The talk is on jo<strong>in</strong>t work with V. Dolean (Université de Nice), P. Hauret (Michel<strong>in</strong>), F. Nataf (Universié<br />

Pierre et Marie Curie), R. Scheichl (University <strong>of</strong> Bath), <strong>and</strong> N. Spillane (Université Pierre et Marie<br />

Curie).<br />

“Robust a Posteriori Error Control <strong>and</strong> Adaptivity for Multiscale, Mult<strong>in</strong>umerics, <strong>and</strong> Mortar Coupl<strong>in</strong>g”<br />

Mart<strong>in</strong> Vohralík<br />

Laboratoire Jacques-Louis Lions,<br />

Université Pierre et Marie Curie (Paris 6),<br />

B.C. 187, 4 place Jussieu,<br />

75252 Paris, France<br />

We consider discretizations <strong>of</strong> a model elliptic problem by means <strong>of</strong> different numerical methods applied<br />

separately <strong>in</strong> different subdoma<strong>in</strong>s, termed mult<strong>in</strong>umerics, coupled us<strong>in</strong>g the mortar technique.<br />

The grids need not match along the <strong>in</strong>terfaces. We are also <strong>in</strong>terested <strong>in</strong> the multiscale sett<strong>in</strong>g, where<br />

the subdoma<strong>in</strong>s are partitioned by a mesh <strong>of</strong> size h, whereas the <strong>in</strong>terfaces are partitioned by a mesh <strong>of</strong><br />

much coarser size H, <strong>and</strong> where lower-order polynomials are used <strong>in</strong> the subdoma<strong>in</strong>s <strong>and</strong> higher-order<br />

polynomials are used on the mortar <strong>in</strong>terface mesh. We derive several fully computable a posteriori<br />

error estimates which deliver a guaranteed upper bound on the error measured <strong>in</strong> the energy norm.<br />

Our estimates are also locally efficient <strong>and</strong> one <strong>of</strong> them is robust with respect to the ratio H/h under<br />

an assumption <strong>of</strong> sufficient regularity <strong>of</strong> the weak solution. The present approach allows bound<strong>in</strong>g separately<br />

<strong>and</strong> compar<strong>in</strong>g mutually the subdoma<strong>in</strong> <strong>and</strong> <strong>in</strong>terface errors. A subdoma<strong>in</strong>/<strong>in</strong>terface adaptive<br />

ref<strong>in</strong>ement strategy is proposed <strong>and</strong> numerically tested. This is a jo<strong>in</strong>t work with Gerg<strong>in</strong>a V. Pencheva<br />

<strong>and</strong> Mary F. Wheeler (UT Aust<strong>in</strong>) <strong>and</strong> Tim Wildey (S<strong>and</strong>ia National Labs).<br />

“Evaluat<strong>in</strong>g Long Term CO2 Storage <strong>in</strong> Sal<strong>in</strong>e Aquifers”<br />

Mary Fanett Wheeler<br />

Director, Center for Subsurface Model<strong>in</strong>g,<br />

Institute for Computational Eng<strong>in</strong>eer<strong>in</strong>g <strong>and</strong> Sciences,<br />

The University <strong>of</strong> Texas at Aust<strong>in</strong>,<br />

201 East 24th Street,<br />

ACE 5.324 — Campus Mail C0200,<br />

Aust<strong>in</strong>, TX 78712<br />

Currently the world obta<strong>in</strong>s more than 80% <strong>of</strong> its energy (coal, oil, gas) for the global economy from<br />

the subsurface. The byproducts <strong>of</strong> consum<strong>in</strong>g these fuels such as greenhouse gas accumulation <strong>in</strong> the<br />

atmosphere are serious <strong>and</strong> potentially devastat<strong>in</strong>g. Renewables such as solar energy <strong>and</strong> w<strong>in</strong>d farms<br />

may take many decades to develop before becom<strong>in</strong>g economically feasible alternatives capable <strong>of</strong> replac<strong>in</strong>g<br />

or reduc<strong>in</strong>g fossil energy usage. A major hope for the near future is geologic sequestration,<br />

15


a proven means <strong>of</strong> permanent CO2 greenhouse-gas storage. This method <strong>in</strong>volves <strong>in</strong>ject<strong>in</strong>g CO2 ,<br />

generally <strong>in</strong> supercritical form, directly <strong>in</strong>to underground geological formations. Oil <strong>and</strong> gas fields,<br />

sal<strong>in</strong>e formations, unm<strong>in</strong>eable coal seams, <strong>and</strong> sal<strong>in</strong>e-filled basalt formations are considered as storage<br />

sites. Various physical processes such as highly impermeable caprock <strong>and</strong> geochemical trapp<strong>in</strong>g mechanisms<br />

would prevent the CO2 from escap<strong>in</strong>g to the surface. Unfortunately, it is difficult to design<br />

<strong>and</strong> manage such efforts. Predictive computational simulation may be the only means to account for<br />

the lack <strong>of</strong> complete characterization <strong>of</strong> the subsurface environment, the multiple scales <strong>of</strong> the various<br />

<strong>in</strong>teract<strong>in</strong>g processes, the large areal extent <strong>of</strong> storage sites such as sal<strong>in</strong>e aquifers, <strong>and</strong> the need for<br />

long time predictions.<br />

In this presentation we discuss multiscale, multiphysics algorithms for accurately predict<strong>in</strong>g the fate <strong>of</strong><br />

<strong>in</strong>jected CO2 <strong>in</strong> conditions governed by multiphase flow, rock mechanics, multicomponent transport,<br />

thermodynamic phase behavior, chemical reactions with<strong>in</strong> both the fluid <strong>and</strong> the rock, <strong>and</strong> the coupl<strong>in</strong>g<br />

<strong>of</strong> all these phenomena over multiple time <strong>and</strong> spatial scales. Both theoretical <strong>and</strong> computational<br />

results will be presented.<br />

“A New Coupl<strong>in</strong>g Concept for Two-Phase Compositional <strong>Porous</strong> <strong>Media</strong> <strong>and</strong> S<strong>in</strong>gle-Phase Compositional<br />

Free <strong>Flow</strong>”<br />

Barbara Wohlmuth<br />

Fakultät Mathematik, TU München<br />

Jo<strong>in</strong>t work with Ra<strong>in</strong>er Helmig, Andreas Lauser, Klaus Mosthaf (IWS Uni Stuttgart).<br />

<strong>Flow</strong> <strong>and</strong> transport processes <strong>in</strong> doma<strong>in</strong>s composed <strong>of</strong> a porous medium <strong>and</strong> an adjacent free-flow<br />

region appear <strong>in</strong> a wide range <strong>of</strong> <strong>in</strong>dustrial, environmental <strong>and</strong> medical applications. Although <strong>of</strong> high<br />

relevance, the mathematical model<strong>in</strong>g <strong>and</strong> the numerical simulation rema<strong>in</strong>s challeng<strong>in</strong>g. This talk<br />

addresses two issues: firstly how to transfer <strong>in</strong>formation between the two sub-system <strong>and</strong> secondly<br />

how how to solve numerically phase changes with<strong>in</strong> a sub-system.<br />

Two basic strategies for the description <strong>of</strong> mass <strong>and</strong> momentum transfer <strong>in</strong> coupled free <strong>and</strong> porousmedium<br />

flow on the Darcy scale can be identified. In the s<strong>in</strong>gle-doma<strong>in</strong> approach, one set <strong>of</strong> equations<br />

is assumed to be valid <strong>in</strong> the whole doma<strong>in</strong> <strong>and</strong> the coupl<strong>in</strong>g is realized via a transition zone, where<br />

material parameters are varied. In the two-doma<strong>in</strong> approach two sets <strong>of</strong> equations are used with<strong>in</strong> the<br />

different subdoma<strong>in</strong>s. To obta<strong>in</strong> a well-posed sett<strong>in</strong>g, additional transfer conditions at the <strong>in</strong>terface<br />

have to be imposed. So far these concepts have been developed for s<strong>in</strong>gle-phase s<strong>in</strong>gle-component systems<br />

describ<strong>in</strong>g the coupl<strong>in</strong>g for mass <strong>and</strong> momentum. However, <strong>in</strong> many applications compositional<br />

multi-phase flow occurs. Here, we generalize the two-doma<strong>in</strong> approach for a non-isothermal twophase<br />

two-component model <strong>and</strong> <strong>in</strong>troduce suitable coupl<strong>in</strong>g conditions. The <strong>in</strong>terface is assumed to<br />

be simple <strong>in</strong> the sense that it has no thickness <strong>and</strong> cannot store mass, momentum or energy. Coupl<strong>in</strong>g<br />

conditions for mass, momentum <strong>and</strong> energy are def<strong>in</strong>ed based on flux cont<strong>in</strong>uity <strong>and</strong> thermodynamic<br />

equilibrium. Moreover, the coupl<strong>in</strong>g concept employs the Beavers-Joseph concept <strong>in</strong> the knowledge<br />

<strong>of</strong> its limitations to parallel, s<strong>in</strong>gle phase flow. The presented model may also be used <strong>in</strong> applications<br />

similar to evaporation, e.g. the design <strong>of</strong> <strong>in</strong>dustrial dry<strong>in</strong>g processes.<br />

Many practically relevant multi-phase problems for flow <strong>and</strong> transport <strong>in</strong> porous media require knowledge<br />

<strong>of</strong> the composition <strong>of</strong> the <strong>in</strong>volved fluid phases <strong>in</strong> order to adequately approximate the underly<strong>in</strong>g<br />

physical processes. One major issue <strong>of</strong> many contemporary multi-phase multi-component models are<br />

phase transitions. A sound <strong>in</strong>corporation <strong>of</strong> these transitions <strong>in</strong>to the flow model is essential s<strong>in</strong>ce<br />

the appearance or disappearance <strong>of</strong> a fluid phase changes the underly<strong>in</strong>g physics <strong>of</strong> the local problem.<br />

If not h<strong>and</strong>led properly, this tends to cause numerical oscillations when solv<strong>in</strong>g the strongly nonl<strong>in</strong>ear<br />

system <strong>of</strong> partial differential equations. Currently, there are two common classes <strong>of</strong> multi-phase,<br />

multi-component models when it comes to determ<strong>in</strong><strong>in</strong>g the local thermodynamic state. The first class<br />

uses a flash calculation which determ<strong>in</strong>es the local thermodynamic state from the overall mass <strong>of</strong> the<br />

<strong>in</strong>dividual components. Flash calculations are stable <strong>in</strong> regard to phase transitions, but exhibit two<br />

major drawbacks: Firstly, to get well-def<strong>in</strong>ed fluid pressures, they require to assume that all fluid<br />

phases are compressible; secondly, they tend to be computationally expensive because, <strong>in</strong> general,<br />

they imply that a non-l<strong>in</strong>ear system <strong>of</strong> equations <strong>of</strong> the size <strong>of</strong> all thermodynamic quantities relevant<br />

for the partial differential equations needs to be solved. The second class <strong>of</strong> models locally adapts the<br />

set <strong>of</strong> primary variables to the local phase state. This externally imposed phase state is then used for<br />

the calculation until the results become physically <strong>in</strong>consistent – which is <strong>in</strong>dicated, for example, by<br />

negative saturations. When such a condition is detected at a given spatial location, the set <strong>of</strong> present<br />

phases is altered <strong>and</strong> the primary variables are switched to physically mean<strong>in</strong>gful quantities. For this<br />

reason, this approach is usually referred to as primary variable switch<strong>in</strong>g (PVS). Compared to flash<br />

16


calculations, this approach is locally more efficient, s<strong>in</strong>ce the thermodynamic state can be calculated<br />

<strong>in</strong>dependently for each fluid phase. However, quite <strong>of</strong>ten numerical oscillations can be observed which<br />

then result <strong>in</strong> extremely small time-steps or even a break down <strong>of</strong> the algorithm. Hence, the second<br />

goal <strong>of</strong> the present contribution is to meet the dem<strong>and</strong> for a robust <strong>and</strong> efficient implicit numerical<br />

scheme that is capable <strong>of</strong> h<strong>and</strong>l<strong>in</strong>g phase transitions.<br />

To do so, we formulate the conditions for the local presence <strong>of</strong> fluid phases as a set <strong>of</strong> so-called complementarity<br />

or Karush–Kuhn–Tucker (KKT) conditions. Mathematically, this leads to the structure <strong>of</strong><br />

variational <strong>in</strong>equality. As similar conditions also occur <strong>in</strong> other important applications like contact or<br />

obstacle problems, there exist many approaches <strong>of</strong> treat<strong>in</strong>g these complimentarity constra<strong>in</strong>ts. Here we<br />

make use <strong>of</strong> the fact that the KKT conditions can equivalently be reformulated as a non-differentiable<br />

but semismooth equation, termed nonl<strong>in</strong>ear complementarity function (NCP). Comb<strong>in</strong>ed with the<br />

balance equations, a system <strong>of</strong> nonl<strong>in</strong>ear equations is obta<strong>in</strong>ed that can be solved by means <strong>of</strong> a<br />

semismooth Newton method with locally superl<strong>in</strong>ear convergence. As the KKT conditions <strong>and</strong> the<br />

non-l<strong>in</strong>earities <strong>of</strong> the material are h<strong>and</strong>led with<strong>in</strong> the same Newton loop, a nested iteration for the<br />

determ<strong>in</strong>ation <strong>of</strong> the local phase state is avoided.<br />

“Mortar Multiscale Framework for Stokes-Darcy <strong>Flow</strong>s”<br />

Ivan Yotov<br />

Department <strong>of</strong> Mathematics, University <strong>of</strong> Pittsburgh<br />

We discuss numerical model<strong>in</strong>g <strong>of</strong> Stokes-Darcy flows based on the Beavers-Joseph-Saffman <strong>in</strong>terface<br />

conditions. The doma<strong>in</strong> is decomposed <strong>in</strong>to a series <strong>of</strong> small subdoma<strong>in</strong>s (coarse grid) <strong>of</strong> either Stokes<br />

or Darcy type. The subdoma<strong>in</strong>s are discretized by appropriate Stokes or Darcy f<strong>in</strong>ite elements. The<br />

solution is resolved locally (<strong>in</strong> each coarse element) on a f<strong>in</strong>e grid, allow<strong>in</strong>g for non-match<strong>in</strong>g grids<br />

across subdoma<strong>in</strong> <strong>in</strong>terfaces. Coarse scale mortar f<strong>in</strong>ite elements are <strong>in</strong>troduced on the <strong>in</strong>terfaces to<br />

approximate the normal stress <strong>and</strong> impose weakly cont<strong>in</strong>uity <strong>of</strong> the velocity. Stability <strong>and</strong> a priori<br />

error analysis is presented for fairly general grid configurations. By elim<strong>in</strong>at<strong>in</strong>g the subdoma<strong>in</strong> unknowns<br />

the global f<strong>in</strong>e scale problem is reduced to a coarse scale <strong>in</strong>terface problem, which is solved<br />

us<strong>in</strong>g an iterative method. We precompute a multiscale flux basis, solv<strong>in</strong>g a fixed number <strong>of</strong> f<strong>in</strong>e<br />

scale subdoma<strong>in</strong> problems for each coarse scale mortar degree <strong>of</strong> freedom, on each subdoma<strong>in</strong> <strong>in</strong>dependently.<br />

Tak<strong>in</strong>g l<strong>in</strong>ear comb<strong>in</strong>ations <strong>of</strong> the multiscale flux basis functions replaces the need to solve<br />

any subdoma<strong>in</strong> problems dur<strong>in</strong>g the <strong>in</strong>terface iteration. Numerical results for coupl<strong>in</strong>g Taylor-Hood<br />

Stokes elements with Raviart-Thomas Darcy elements are presented.<br />

This is jo<strong>in</strong>t work with Ben Ganis, University <strong>of</strong> Texas at Aust<strong>in</strong>, Vivette Girault, Paris VI, <strong>and</strong><br />

Danail Vassilev, Rensselaer Polytechnic Institute<br />

17


Abstracts for Posters<br />

“Multiscale <strong>Simulation</strong>s <strong>of</strong> Non-Darcy’s <strong>Flow</strong>s”<br />

Sergey Alyaev<br />

Universitetet i Bergen, Matematisk <strong>in</strong>stitutt,<br />

Postboks 7803, 5020 BERGEN<br />

The st<strong>and</strong>ard approximation for flow-pressure relationship <strong>in</strong> porous media is Darcy’s law. It is<br />

justified <strong>and</strong> gives good results for slow flows, however <strong>in</strong> a number <strong>of</strong> applications with faster flow<br />

regimes it breaks down. These applications <strong>in</strong>clude flows that may occur near wells <strong>and</strong> <strong>in</strong> fractured<br />

regions <strong>in</strong> subsurface. Moreover, those flows are common for <strong>in</strong>dustrial <strong>and</strong> near surface porous media.<br />

Darcy’s law can be derived by means <strong>of</strong> homogenization from the Stokes’ equation on the pore scale.<br />

Stokes’ equation does not take <strong>in</strong>to account non-l<strong>in</strong>ear effects <strong>of</strong> the flows that are <strong>of</strong> importance for<br />

flows with larger Reynolds numbers. The st<strong>and</strong>ard cont<strong>in</strong>uum mechanics approach for those cases<br />

is us<strong>in</strong>g non-l<strong>in</strong>ear Navier-Stokes equations for the f<strong>in</strong>e scale. To perform larger scale simulations <strong>of</strong><br />

non-l<strong>in</strong>ear flow solv<strong>in</strong>g the Navier-Stokes equation is not feasible due to time <strong>and</strong> memory limitations.<br />

As an alternative we apply multiscale techniques.<br />

We present a heterogeneous multiscale method that on the coarse scale only assumes conservation<br />

<strong>of</strong> mass on control volumes, that is, no phenomenological Darcy-type relationship for velocity is<br />

presumed. The fluid fluxes are <strong>in</strong>stead provided by a f<strong>in</strong>e scale Navier-Stokes mixed f<strong>in</strong>ite element<br />

solver. This methodology, that we will refer to as multiscale simulation is different from a common<br />

multiscale modell<strong>in</strong>g procedure <strong>of</strong> upscal<strong>in</strong>g. The difference is, that <strong>in</strong> the case <strong>of</strong> MS modell<strong>in</strong>g<br />

one chooses the coarse scale model <strong>in</strong> advance based on physical reason<strong>in</strong>g <strong>and</strong> then gets effective<br />

parameters from solv<strong>in</strong>g some sort <strong>of</strong> f<strong>in</strong>e scale equations. The advantage <strong>of</strong> MS simulation is that<br />

the results <strong>of</strong> it will co<strong>in</strong>cide with Darcy’s law (with upscaled parameters) without <strong>in</strong>creas<strong>in</strong>g the<br />

computational time asymptotically for the slow flows. Yet it will deliver better results for faster flow<br />

regimes without the need <strong>of</strong> choos<strong>in</strong>g another coarse scale model by a user.<br />

We consider application <strong>of</strong> the methodology <strong>in</strong> the context <strong>of</strong><br />

• Weakly compressible near-well flows with oscillat<strong>in</strong>g boundary conditions;<br />

• <strong>Porous</strong> media with large pores lack<strong>in</strong>g scale separation;<br />

• Particle transport <strong>in</strong> porous media.<br />

The work is be<strong>in</strong>g carried out <strong>in</strong> collaboration with Jan M. Nordbotten <strong>and</strong> Eirik Keilegavlen.<br />

“Micromechanics <strong>of</strong> Geomaterials <strong>and</strong> Geocomposites.”<br />

Radim Blaheta<br />

Institute <strong>of</strong> Geonics AS CR,<br />

Studentska 1768,<br />

708 00 Ostrava-Poruba, Czech Republic<br />

The poster addresses three topics. The first one is the micro FEM analysis <strong>of</strong> processes <strong>in</strong> geomaterials<br />

<strong>and</strong> geocomposites (which arose from grout<strong>in</strong>g for improv<strong>in</strong>g the geomaterials). We consider processes<br />

<strong>of</strong> deformation <strong>and</strong> flow <strong>in</strong> porous media <strong>and</strong> the FEM modell<strong>in</strong>g should give answer to the questions<br />

what are the effective properties, how the changes <strong>in</strong> microstructure <strong>in</strong>fluence the macroscale behaviour<br />

<strong>and</strong> also how to optimize the microstructure. The second topic is the iterative solution (jo<strong>in</strong>t work with<br />

O. Axelsson <strong>and</strong> V. Sokol). The micro FEM simulations lead to the solution <strong>of</strong> large-scale systems,<br />

which are frequently very ill-conditioned due to oscillations <strong>and</strong> high jumps (contrast) <strong>in</strong> material<br />

coefficients. Therefore, we are motivated <strong>in</strong> seek<strong>in</strong>g parallelizable <strong>and</strong> robust iterative solution methods<br />

<strong>and</strong> we shall present here some results concern<strong>in</strong>g the variants <strong>of</strong> multi-level Schwarz methods with<br />

coarse spaces constructed by aggregations. We also touch the construction <strong>of</strong> block preconditioners for<br />

mixed FEM solution <strong>of</strong> Darcy flow problems. The third topic is <strong>in</strong>verse analysis (jo<strong>in</strong>t work with R.<br />

Hrtus <strong>and</strong> R. Kohut). We are <strong>in</strong>terested <strong>in</strong> micro FEM analysis, which uses <strong>in</strong>put based on assign<strong>in</strong>g<br />

the local material properties accord<strong>in</strong>g to tomography scans (provid<strong>in</strong>g the material distribution) <strong>and</strong><br />

properties <strong>of</strong> <strong>in</strong>dividual material components. These properties are determ<strong>in</strong>ed with the aid <strong>of</strong> tests<br />

<strong>of</strong> macroscale behaviour on the specimens with known material distribution <strong>and</strong> solution <strong>of</strong> <strong>in</strong>verse<br />

problems for identification <strong>of</strong> properties <strong>of</strong> selected material components. We describe this approach<br />

<strong>and</strong> provide some results for the case <strong>of</strong> coal-PE res<strong>in</strong> geocomposites.<br />

18


“Modell<strong>in</strong>g, <strong>Simulation</strong> <strong>and</strong> Optimization <strong>of</strong> the Microstructure <strong>of</strong> SOFC <strong>Porous</strong> Cathodes”<br />

Thomas Carraro<br />

Institute for Applied Mathematics, University <strong>of</strong> Heidelberg<br />

The performance <strong>of</strong> a solid oxide fuel cell (SOFC) is strongly affected by electrode polarization losses,<br />

which are related to the composition <strong>and</strong> the microstructure <strong>of</strong> the porous materials. A model that can<br />

decouple the effects associated to the geometrical arrangement, shape <strong>and</strong> size <strong>of</strong> the particles on one<br />

side <strong>and</strong> the material properties on the other can give a relevant improvement <strong>in</strong> the underst<strong>and</strong><strong>in</strong>g<br />

<strong>of</strong> the underly<strong>in</strong>g processes.<br />

A porous mixed ionic <strong>and</strong> electronic conduct<strong>in</strong>g (MIEC) cathode was reconstructed by a tomography<br />

technique based on focused ion beam coupled with scann<strong>in</strong>g electronic microscope (FIB/SEM). The<br />

detailed geometry <strong>of</strong> the microstructure is used for 3D calculations <strong>of</strong> the electrochemical processes <strong>in</strong><br />

the electrode. The area specific resistance (ASR) <strong>of</strong> the reconstructed porous cathode is calculated as<br />

a performance <strong>in</strong>dex. A model based on the f<strong>in</strong>ite element method (FEM), which complexity requires<br />

the use <strong>of</strong> high performance comput<strong>in</strong>g techniques (HPC), has been developed.<br />

In this work we show the comparison <strong>of</strong> the 3D microstructure model with a 1D homogenized one.<br />

A trusted reduced model by homogenization can be used to optimize the design <strong>of</strong> experiments <strong>and</strong><br />

<strong>in</strong>crease the precision <strong>of</strong> a fitt<strong>in</strong>g procedure. First steps towards the optimal experimental design for<br />

impedance measurements for SOFC are shown.<br />

“Numerical Upscal<strong>in</strong>g <strong>in</strong> <strong>Porous</strong> <strong>Media</strong>”<br />

Christian Engwer<br />

Institut für Numerische und Angew<strong>and</strong>te Mathematik<br />

Universität Münster<br />

For the simulation <strong>of</strong> transport processes <strong>in</strong> porous media effective parameters for the physical processes<br />

on the target scale are required. Numerical upscal<strong>in</strong>g can help where experiments are not<br />

possible, or hard to conduct.<br />

Bastian <strong>and</strong> Engwer proposed an Unfitted Discont<strong>in</strong>uous Galerk<strong>in</strong> (UDG) method for PDEs on doma<strong>in</strong>s<br />

with a complicated geometric shape. This method is well suited for simulations on the pore<br />

scale. The method uses f<strong>in</strong>ite element meshes which are significantly coarser then those required by<br />

st<strong>and</strong>ard conform<strong>in</strong>g f<strong>in</strong>ite element approaches <strong>and</strong> is flexible enough to be used for elliptic, hyperbolic<br />

<strong>and</strong> parabolic problems. Essential boundary conditions are <strong>in</strong>corporated us<strong>in</strong>g Discont<strong>in</strong>uous<br />

Galerk<strong>in</strong> discretization on a cut-cell mesh.<br />

We apply this method to numerical upscal<strong>in</strong>g us<strong>in</strong>g direct simulation on the pore scale <strong>and</strong> present<br />

different applications. The method is robust with respect to computations on coarse meshes.<br />

“Precondition<strong>in</strong>g <strong>of</strong> Non-Conform<strong>in</strong>g FEM Systems”<br />

Ivan Georgiev<br />

Computational Methods for Direct Field Problems,<br />

Radon Institute for Computational <strong>and</strong> Applied Mathematics (RICAM),<br />

Altenberger Strasse 69,<br />

A-4040 L<strong>in</strong>z, AUSTRIA,<br />

ivan.georgiev@oeaw.ac.at<br />

Jo<strong>in</strong>t work with J. Kraus, RICAM.<br />

We present different approaches for precondition<strong>in</strong>g <strong>of</strong> l<strong>in</strong>ear algebraic systems obta<strong>in</strong>ed from nonconform<strong>in</strong>g<br />

discretizations <strong>of</strong> scalar elliptic <strong>and</strong> l<strong>in</strong>ear elasticity problems. We consider <strong>in</strong>terior penalty<br />

discont<strong>in</strong>uous Galerk<strong>in</strong> discterizations <strong>of</strong> l<strong>in</strong>ear elasticity equations <strong>and</strong> we propose a space splitt<strong>in</strong>g<br />

which gives rise to uniform preconditioners. For the problem with Dirichlet boundary conditions<br />

imposed on the entire boundary the l<strong>in</strong>ear non-conform<strong>in</strong>g f<strong>in</strong>ite elements provides a lock<strong>in</strong>g-free<br />

discretization. We devise an optimal order multilevel method for the considered 3D pure displacement<br />

elasticity problem. F<strong>in</strong>aly we present a two-level method for the scalar elliptic problems with highly<br />

vary<strong>in</strong>g coefficients. The proposed generalized hierarchical basis yields a robust splitt<strong>in</strong>g with respect<br />

to mesh size <strong>and</strong> jump discont<strong>in</strong>uities <strong>of</strong> the coefficients.<br />

19


“Design <strong>of</strong> Numerical Methods to Simulate Models <strong>of</strong> a Solid Oxide Fuel Cell”<br />

Christian Goll<br />

Institute <strong>of</strong> Applied Mathematics, University <strong>of</strong> Heidelberg, Germany<br />

The motivation <strong>of</strong> this work is to model the behaviour <strong>of</strong> a reactive gas mixture <strong>in</strong> the anode <strong>and</strong> the<br />

overly<strong>in</strong>g gas channel <strong>of</strong> a solid oxide fuel cell (SOFC). Via impedance measurements, polarization<br />

losses aris<strong>in</strong>g dur<strong>in</strong>g the operation <strong>of</strong> the fuel cell can be identified. To achive this, a coupled system<br />

<strong>of</strong> PDEs has to be solved numerically. This system conta<strong>in</strong>s (Navier-)Stokes <strong>and</strong> Darcy equations<br />

to characterize the flow as well as the Stefan-Maxwell/Dusty-Gas model to describe diffusion <strong>and</strong><br />

reaction <strong>of</strong> the species.<br />

Over the last years, a lot <strong>of</strong> work has been put <strong>in</strong>to the coupl<strong>in</strong>g conditions for the Stokes/Darcy-system<br />

along the <strong>in</strong>terface between fluid <strong>and</strong> porous region. In 2001, Jaeger, Mikelić <strong>and</strong> Neuss proposed a<br />

set <strong>of</strong> conditions for the lam<strong>in</strong>ar viscous flow over a porous bed, which actually allows to decouple the<br />

problem <strong>and</strong> make it easier to compute. We compare these ’decoupl<strong>in</strong>g conditions’ numerically with<br />

a fully coupled approach.<br />

“Impact on the Surface Temperature due to the Modifications <strong>of</strong> Underly<strong>in</strong>g L<strong>and</strong> Surface Conditions: A<br />

Study over Western India”<br />

Sridhara Nayak<br />

Center for Oceans, Rivers, Atmosphere <strong>and</strong> L<strong>and</strong> Sciences;<br />

Indian Institute <strong>of</strong> Technology, Kharagpur;<br />

Kharagpur 721 302, INDIA<br />

Jo<strong>in</strong>t work with M. M<strong>and</strong>al.<br />

The important anthropogenic <strong>in</strong>fluences on global warm<strong>in</strong>g are greenhouse gases <strong>and</strong> changes <strong>in</strong> l<strong>and</strong><br />

use & l<strong>and</strong> cover. It is well accepted that the greenhouse gases are ma<strong>in</strong> causes <strong>of</strong> global warm<strong>in</strong>g<br />

<strong>and</strong> recent climate change, but changes <strong>in</strong> l<strong>and</strong> use & l<strong>and</strong> cover are also contribut<strong>in</strong>g to a certa<strong>in</strong><br />

extent. The changes <strong>in</strong> l<strong>and</strong> use & l<strong>and</strong> cover modify the underly<strong>in</strong>g l<strong>and</strong> surface conditions. The<br />

modifications <strong>in</strong> l<strong>and</strong> surface conditions change the <strong>in</strong>teraction between l<strong>and</strong> surface <strong>and</strong> atmosphere,<br />

i.e., the exchange <strong>of</strong> energy <strong>and</strong> moisture between l<strong>and</strong> surface <strong>and</strong> atmosphere. Presence <strong>of</strong> vegetation<br />

modifies surface albedo, absorption <strong>of</strong> solar radiation, sensible <strong>and</strong> latent heat fluxes. Hence,<br />

br<strong>in</strong>gs <strong>in</strong> changes <strong>in</strong> the earth’s radiation balance. The changes <strong>in</strong> l<strong>and</strong> surface conditions modify the<br />

evapotranspiration <strong>and</strong> hence moisture exchange. The exchange <strong>of</strong> energy <strong>and</strong> moisture contributes<br />

significantly <strong>in</strong> determ<strong>in</strong><strong>in</strong>g local, regional <strong>and</strong> even global climate. The modifications <strong>in</strong> l<strong>and</strong> surface<br />

conditions result <strong>in</strong> emission or removal <strong>of</strong> CO2 <strong>in</strong> the atmosphere <strong>and</strong> thus contribut<strong>in</strong>g regional<br />

warm<strong>in</strong>g or cool<strong>in</strong>g. In this study we have two objectives. First attempt has been made to <strong>in</strong>vestigate<br />

surface temperatures dur<strong>in</strong>g 1973-2009 over Western India <strong>and</strong> the impact on the surface temperature<br />

due to the modifications <strong>in</strong> l<strong>and</strong> surface conditions the over the region. This is based on temperature<br />

datasets from observation as obta<strong>in</strong>ed from India Meteorological Department <strong>and</strong> NCEP/NCAR reanalysis<br />

(NNRP1). The impact on the surface temperature due to the modifications <strong>in</strong> l<strong>and</strong> surface<br />

conditions is estimated based on deviation <strong>in</strong> temperature <strong>in</strong> the observation <strong>and</strong> reanalysis datasets.<br />

The observed temperature datasets <strong>in</strong>dicates warm<strong>in</strong>g over Western India is 0.13C per decade. The<br />

result shows the modifications <strong>in</strong> l<strong>and</strong> surface conditions dur<strong>in</strong>g 1973-2005 contribut<strong>in</strong>g to this warm<strong>in</strong>g<br />

is 0.05C per decade. Second attempt has been made to <strong>in</strong>vestigate the changes <strong>in</strong> l<strong>and</strong> use & l<strong>and</strong><br />

cover over the region dur<strong>in</strong>g the period 1975-2005. This is <strong>in</strong>vestigated based on four l<strong>and</strong> use & l<strong>and</strong><br />

cover datasets as obta<strong>in</strong>ed from Global L<strong>and</strong> Cover Facility (GLCF). The change <strong>in</strong> l<strong>and</strong> surfaces is<br />

<strong>in</strong>vestigated by quantify<strong>in</strong>g the l<strong>and</strong> covered with various l<strong>and</strong> use & l<strong>and</strong> cover types. The comparison<br />

<strong>of</strong> the change <strong>in</strong> surface temperature with the change <strong>in</strong> l<strong>and</strong> surface <strong>in</strong>dicates the warm<strong>in</strong>g due<br />

to the modifications <strong>in</strong> underly<strong>in</strong>g l<strong>and</strong> surface conditions is because <strong>of</strong> the reduction <strong>of</strong> area under<br />

open forest <strong>and</strong> subsequent <strong>in</strong>crease <strong>of</strong> the area under agricultural l<strong>and</strong>.<br />

20


“Model<strong>in</strong>g Two-phase <strong>Flow</strong> with Disappear<strong>in</strong>g Gas Phase”<br />

Rebecca Neumann<br />

University <strong>of</strong> Heidelberg<br />

Interdiscipl<strong>in</strong>ary Center for Scientific Comput<strong>in</strong>g<br />

Im Neuenheimer Feld 368<br />

69120 Heidelberg<br />

Carbon Capture <strong>and</strong> Storage (CCS) is a recently discussed new technology, aimed at allow<strong>in</strong>g an<br />

ongo<strong>in</strong>g use <strong>of</strong> fossil fuels while prevent<strong>in</strong>g the produced CO2 to be released to the atmosphere. CSS<br />

<strong>in</strong>volves two components (water <strong>and</strong> CO2) <strong>in</strong> two phases (liquid <strong>and</strong> gas). To model the process,<br />

a multiphase flow equation is used. One <strong>of</strong> the big problems aris<strong>in</strong>g <strong>in</strong> two-phase flow simulations<br />

is the disappearance <strong>of</strong> the gas phase, which leads to a degeneration <strong>of</strong> the equations satisfied by<br />

the saturation. A st<strong>and</strong>ard choice <strong>of</strong> primary variables, which is the pressure <strong>of</strong> one phase <strong>and</strong> the<br />

saturation <strong>of</strong> the other phase, cannot be applied here.<br />

We developed a new approach us<strong>in</strong>g the pressure <strong>of</strong> the nonwett<strong>in</strong>g phase <strong>and</strong> the capillary pressure<br />

as primary variables. We implemented this new choice <strong>of</strong> primary variables <strong>in</strong> the DUNE simulation<br />

framework <strong>and</strong> present the numerical results for some test cases.<br />

“Robust Preconditioners for General SPD Operators”<br />

Joerg Willems<br />

Computational Methods for Direct Field Problems,<br />

Radon Institute for Computational <strong>and</strong> Applied Mathematics (RICAM),<br />

Altenberger Strasse 69,<br />

A-4040 L<strong>in</strong>z, AUSTRIA,<br />

joerg.willems@ricam.oeaw.ac.at<br />

An abstract sett<strong>in</strong>g for robustly precondition<strong>in</strong>g symmetric positive def<strong>in</strong>ite (SPD) operators is presented.<br />

The method belongs to the class <strong>of</strong> additive Schwarz preconditioners, <strong>and</strong> it requires only<br />

rather mild assumptions naturally satisfied by operators result<strong>in</strong>g from the discretization <strong>of</strong> several<br />

important partial differential equations. The term ”robust” refers to the property <strong>of</strong> the condition<br />

numbers <strong>of</strong> the preconditioned systems be<strong>in</strong>g <strong>in</strong>dependent <strong>of</strong> mesh parameters <strong>and</strong> problem parameters.<br />

Important <strong>in</strong>stances <strong>of</strong> such problem parameters are <strong>in</strong> particular (highly vary<strong>in</strong>g) coefficients.<br />

The core <strong>of</strong> this method is the construction <strong>of</strong> the coarse space based on the solution <strong>of</strong> local generalized<br />

eigenvalue problems. The abstract framework is applied to the scalar elliptic equation <strong>and</strong><br />

Br<strong>in</strong>kman’s equations with coefficients vary<strong>in</strong>g over several orders <strong>of</strong> magnitude. S<strong>in</strong>ce Br<strong>in</strong>kman’s<br />

equations model<strong>in</strong>g flows <strong>in</strong> highly porous media constitute a saddle po<strong>in</strong>t problem, the abstract<br />

framework is not immediately applicable. This difficulty is overcome by resort<strong>in</strong>g to the stream function<br />

formulation. The latter is, however, only used for the coarse space construction, while the actual<br />

solution process is carried out <strong>in</strong> the primal variables, i.e., velocity <strong>and</strong> pressure. Several numerical<br />

examples are presented to illustrate the properties <strong>of</strong> the method.<br />

21


List <strong>of</strong> Participants<br />

Alyaev Sergey Universitetet i Bergen cobxo3bot@gmail.com<br />

Andreev Roman SAM, ETH Zürich <strong>and</strong>reevr@math.ethz.ch<br />

Arbenz Peter ETH Zürich, Chair <strong>of</strong><br />

Computational Science<br />

arbenz@<strong>in</strong>f.ethz.ch<br />

Bastian Peter University <strong>of</strong> Heidelberg peter.bastian@iwr.uni-heidelberg.de<br />

Becker Jürgen Fraunh<strong>of</strong>er ITWM,<br />

Kaiserslautern<br />

juergen.becker@itwm.fhg.de<br />

Blaheta Radim Institute <strong>of</strong> Geonics AS CR,<br />

Ostrava, CZ<br />

blaheta@ugn.cas.cz<br />

Bourgeat Ala<strong>in</strong> UCB Lyon 1 ala<strong>in</strong>.bourgeat@univ-lyon1.fr<br />

Buck Marco Fraunh<strong>of</strong>er Institute for<br />

Industrial Mathematics<br />

marco.buck@itwm.fraunh<strong>of</strong>er.de<br />

Caiazzo Alfonso WIAS Berl<strong>in</strong> caiazzo@wias-berl<strong>in</strong>.de<br />

Carraro Thomas University <strong>of</strong> Heidelberg thomas.carraro@iwr.uni-heidelberg.de<br />

Challa Durga Prasad RICAM, L<strong>in</strong>z durga.challa@oeaw.ac.at<br />

Cliffe Andrew University <strong>of</strong> Nott<strong>in</strong>gham Andrew.Cliffe@nott<strong>in</strong>gham.ac.uk<br />

Collis Joe University <strong>of</strong> Nott<strong>in</strong>gham pmxjc2@nott<strong>in</strong>gham.ac.uk<br />

Dahle Helge K. University <strong>of</strong> Bergen, Norway helge.dahle@math.uib.no<br />

Dashti Masoumeh University <strong>of</strong> Warwick,<br />

Mathematics Institute<br />

M.Dashti@warwick.ac.uk<br />

Discacciati Marco Universitat Politècnica de<br />

Catalunya, Barcelona<br />

marco.discacciati@gmail.com<br />

Efendiev Yalch<strong>in</strong> Texas A&M University efendiev@math.tamu.edu<br />

Engl He<strong>in</strong>z RICAM & University <strong>of</strong> Vienna he<strong>in</strong>z.engl@oeaw.ac.at<br />

Engwer Christian Universität Münster christian.engwer@wwu.de<br />

Erhel Jocelyne Inria jocelyne.erhel@irisa.fr<br />

Ern Alex<strong>and</strong>re University Paris-Est, CERMICS ern@cermics.enpc.fr<br />

Frank Elisabeth Johannes Kepler University<br />

L<strong>in</strong>z<br />

elisabeth.frank@jku.at<br />

Freitag Mel<strong>in</strong>a University <strong>of</strong> Bath m.freitag@maths.bath.ac.uk<br />

Fuhrmann Jürgen Weierstrass Institute Berl<strong>in</strong> juergen.fuhrmann@wias-berl<strong>in</strong>.de<br />

Gahalaut Krishan P.S. RICAM, L<strong>in</strong>z krishan.gahalaut@oeaw.ac.at<br />

Georgiev Ivan RICAM, L<strong>in</strong>z ivan.georgiev@oeaw.ac.at<br />

Goll Christian University <strong>of</strong> Heidelberg christian.goll@iwr.uni-heidelberg.de<br />

Halpern Laurence Université Paris 13 halpern@math.univ-paris13.fr<br />

Hrtus Rostislav Institute <strong>of</strong> Geonics <strong>of</strong> the AS<br />

CR, v.v.i., Ostrava, CZ<br />

hrtus@ugn.cas.cz<br />

Kar Manas RICAM, L<strong>in</strong>z manas.kar@oeaw.ac.at<br />

Karer Erw<strong>in</strong> RICAM, L<strong>in</strong>z erw<strong>in</strong>.karer@oeaw.ac.at<br />

Kollmann Markus Doctoral Program<br />

Computational Mathematics,<br />

L<strong>in</strong>z<br />

markus.kollmann@dk-compmath.jku.at<br />

Kolmbauer Michael Johannes Kepler University<br />

L<strong>in</strong>z<br />

kolmbauer@numa.uni-l<strong>in</strong>z.ac.at<br />

Kornhuber Ralf FU Berl<strong>in</strong> kornhuber@math.fu-berl<strong>in</strong>.de<br />

Kraus Johannes RICAM, L<strong>in</strong>z johannes.kraus@oeaw.ac.at<br />

Langer Ulrich Johannes Kepler University<br />

L<strong>in</strong>z<br />

ulanger@numa.uni-l<strong>in</strong>z.ac.at<br />

Livshits Irene Ball State University ilivshits@bsu.edu<br />

Mahmoud Alaa Al-Azhar University, Faculty <strong>of</strong><br />

Science<br />

allaaly2@yahoo.com<br />

Marschall<strong>in</strong>ger Robert ÖAW GIScience robert.marschall<strong>in</strong>ger@oeaw.ac.at<br />

Mataln Marianne Montanuniversität Leoben marianne.mataln@unileoben.ac.at<br />

Melenk Jens Markus TU Wien melenk@tuwien.ac.at<br />

Michel Anthony IFP Energies nouvelles anthony.michel@ifpen.fr<br />

Mikelić Andro Université Lyon 1, Dépt<br />

mathématiques <strong>and</strong> ICJ<br />

mikelic@univ-lyon1.fr<br />

22


Nayak Sridhara Indian Institute <strong>of</strong> Technology,<br />

Kharagpur<br />

sridharanayakiitkgp@gmail.com<br />

Neumann Rebecca University <strong>of</strong> Heidelberg rebecca.neumann@iwr.uni-heidelberg.de<br />

Nguyen Trung Thanh RICAM, L<strong>in</strong>z trung-thanh.nguyen@oeaw.ac.at<br />

Nordbotten Jan Mart<strong>in</strong> University <strong>of</strong> Bergen Jan.Nordbotten@mi.uib.no<br />

Oberaigner Eduard Montanuniversität Leoben ero@unileoben.ac.at<br />

Park M<strong>in</strong>ho University <strong>of</strong> Nott<strong>in</strong>gham m<strong>in</strong>.park@nott<strong>in</strong>gham.ac.uk<br />

Pechste<strong>in</strong> Clemens Johannes Kepler University<br />

L<strong>in</strong>z<br />

clemens.pechste<strong>in</strong>@numa.uni-l<strong>in</strong>z.ac.at<br />

Peterseim Daniel Humboldt-Universität zu<br />

Berl<strong>in</strong>, Institut für Mathematik<br />

peterseim@math.hu-berl<strong>in</strong>.de<br />

Scheichl Robert University <strong>of</strong> Bath R.Scheichl@bath.ac.uk<br />

S<strong>in</strong>i Mourad RICAM, L<strong>in</strong>z mourad.s<strong>in</strong>i@oeaw.ac.at<br />

Teckentrup Aretha University <strong>of</strong> Bath, Department<br />

<strong>of</strong> Mathematical Sciences<br />

A.L.Teckentrup@bath.ac.uk<br />

Tomar Satyendra RICAM, L<strong>in</strong>z satyendra.tomar@oeaw.ac.at<br />

Ullmann Elisabeth University <strong>of</strong> Bath E.Ullmann@bath.ac.uk<br />

Vohralík Mart<strong>in</strong> LJLL, University Paris 6 vohralik@ann.jussieu.fr<br />

Wachsmuth Daniel RICAM, L<strong>in</strong>z daniel.wachsmuth@oeaw.ac.at<br />

Wheeler Mary F. The University <strong>of</strong> Texas at<br />

Aust<strong>in</strong><br />

mfw@ices.utexas.edu<br />

Willems Jörg RICAM, L<strong>in</strong>z joerg.willems@oeaw.ac.at<br />

Wohlmuth Barbara Technische Universität<br />

München<br />

barbara.wohlmuth@ma.tum.de<br />

Yang Huidong RICAM, L<strong>in</strong>z huidong.yang@oeaw.ac.at<br />

Yotov Ivan University <strong>of</strong> Pittsburgh yotov@math.pitt.edu<br />

Zikatanov Ludmil The Pennsylvania State<br />

University<br />

ltz@math.psu.edu<br />

23


Monday Tuesday Wednesday Thursday Friday<br />

Open<strong>in</strong>g (9:10) Helge K. Dahle Jürgen Becker Jocelyne Erhel Yalch<strong>in</strong> Efendiev<br />

8:30-9:20<br />

Local-Global Multiscale Model<br />

Reduction Techniques for <strong>Flow</strong>s<br />

Numerical <strong>and</strong> Stochastic<br />

Models <strong>of</strong> <strong>Flow</strong> <strong>in</strong> 3D Discrete<br />

Pore-Scale Modell<strong>in</strong>g <strong>of</strong> <strong>Porous</strong><br />

Layers Used <strong>in</strong> Fuel Cells<br />

CO2-Migration: Effects <strong>and</strong><br />

Upscal<strong>in</strong>g <strong>of</strong> Caprock<br />

<strong>in</strong> Heterogeneous <strong>Porous</strong> <strong>Media</strong><br />

Fracture Networks<br />

Topography<br />

Mary F. Wheeler Laurence Halpern Andro Mikelić Ala<strong>in</strong> Bourgeat Clemens Pechste<strong>in</strong><br />

9:20-10:10<br />

Model<strong>in</strong>g Compressible<br />

Effective Pressure Interface Law<br />

for Transport Phenomena<br />

An Abstract Two-Level Additive<br />

Schwarz Method for Systems<br />

Multiphase <strong>Flow</strong> <strong>and</strong> Transport<br />

<strong>in</strong> Saturated-Unsaturated <strong>Porous</strong><br />

Optimized Schwarz Waveform<br />

Relaxation for Reactive<br />

<strong>Media</strong>: Phase<br />

between an Unconf<strong>in</strong>ed Fluid<br />

<strong>and</strong> a <strong>Porous</strong> Medium us<strong>in</strong>g<br />

Evaluat<strong>in</strong>g Long Term CO2<br />

Storage <strong>in</strong> Sal<strong>in</strong>e Aquifers<br />

with High Contrast Coefficients<br />

Transport Problems<br />

Appearance-Disappearance.<br />

Homogenization<br />

Application to Gas Migration ...<br />

C<strong>of</strong>fee C<strong>of</strong>fee C<strong>of</strong>fee C<strong>of</strong>fee C<strong>of</strong>fee<br />

Jan Mart<strong>in</strong> Nordbotten Anthony Michel Jürgen Fuhrmann Alex<strong>and</strong>re Ern Ralf Kornhuber<br />

10:40-11:30<br />

Coupled Surface <strong>and</strong><br />

Time-Space Doma<strong>in</strong><br />

Saturated/Unsaturated Ground<br />

Water <strong>Flow</strong> <strong>in</strong> Heterogeneous<br />

Discont<strong>in</strong>uous Galerk<strong>in</strong> Method<br />

for Two-Component Miscible<br />

Electrochemical Processes <strong>and</strong><br />

<strong>Porous</strong> <strong>Media</strong>: Mathematical<br />

Decomposition for Reactive<br />

Transport <strong>in</strong> <strong>Porous</strong> <strong>Media</strong><br />

Multiscale Model<strong>in</strong>g for CO2<br />

Storage<br />

Liquid-Gas <strong>Porous</strong> <strong>Media</strong> <strong>Flow</strong>s<br />

<strong>and</strong> Numerical Model<strong>in</strong>g<br />

<strong>Media</strong><br />

Lunch Lunch Lunch Lunch Lunch<br />

Mart<strong>in</strong> Vohralík Andrew Cliffe Marco Discacciati Peter Bastian<br />

14:00-14:50<br />

Numerical Solution <strong>of</strong><br />

Coupl<strong>in</strong>g Free <strong>and</strong> <strong>Porous</strong>-<strong>Media</strong><br />

<strong>Flow</strong>s: Model<strong>in</strong>g, Analysis, <strong>and</strong><br />

Robust a Posteriori Error<br />

Control <strong>and</strong> Adaptivity for<br />

Compositional Two-Phase <strong>Flow</strong><br />

<strong>in</strong> <strong>Porous</strong> <strong>Media</strong><br />

What should we do with<br />

Radioactive <strong>Waste</strong>?<br />

Numerical Approximation<br />

Multiscale, Mult<strong>in</strong>umerics, <strong>and</strong><br />

Mortar Coupl<strong>in</strong>g<br />

C<strong>of</strong>fee C<strong>of</strong>fee C<strong>of</strong>fee C<strong>of</strong>fee<br />

Barbara Wohlmuth Robert Marschall<strong>in</strong>ger Ivan Yotov<br />

15:20-16:10<br />

Multi-scale 3D <strong>and</strong> 4D<br />

A New Coupl<strong>in</strong>g Concept for<br />

Two-Phase Compositional<br />

Mortar Multiscale Framework<br />

for Stokes-Darcy <strong>Flow</strong>s<br />

Modell<strong>in</strong>g <strong>and</strong> <strong>Simulation</strong> <strong>in</strong><br />

Geosciences<br />

<strong>Porous</strong> <strong>Media</strong> <strong>and</strong> S<strong>in</strong>gle-Phase<br />

Compositional Free <strong>Flow</strong><br />

17:00 Poster<br />

19:00 D<strong>in</strong>ner<br />

Each talk is 40 m<strong>in</strong>utes long. After the end <strong>of</strong> each talk, there are 10 m<strong>in</strong>utes for discussion.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!