RBF_Cover (for eps) - National Water Research Institute
RBF_Cover (for eps) - National Water Research Institute
RBF_Cover (for eps) - National Water Research Institute
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
2<strong>RBF</strong><br />
The <strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong> Presents<br />
The Second International<br />
Riverbank Filtration Conference<br />
Riverbank Filtration:<br />
The Future Is NOW!<br />
PROGRAM & ABSTRACTS<br />
Edited by:<br />
GINA MELIN, <strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong><br />
September 16-19, 2003<br />
Hilton Cincinnati Netherlands Plaza ✦ Cincinnati, Ohio USA
2<strong>RBF</strong><br />
The <strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong> Presents<br />
The Second International<br />
Riverbank Filtration Conference<br />
Riverbank Filtration :<br />
The Future Is NOW!<br />
PROGRAM & ABSTRACTS<br />
Edited by:<br />
GINA MELIN, <strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong><br />
September 16-19, 2003<br />
Hilton Cincinnati Netherlands Plaza ✦ Cincinnati, Ohio USA
Published by the<br />
NATIONAL WATER RESEARCH INSTITUTE<br />
NWRI-2003-10<br />
10500 Ellis Avenue ✦ P.O. Box 20865<br />
Fountain Valley, Cali<strong>for</strong>nia 92728-0865<br />
(714) 378-3278 ✦ Fax: (714) 378-3375<br />
www.NWRI-USA.org
Conference Planning Committee<br />
Chair: CHITTARANJAN RAY, Ph.D, P.E., University of Hawaii at Mañoa<br />
Conference Coordinator: GINA MELIN, <strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong><br />
EDWARD J. BOUWER, Ph.D.,<br />
The Johns Hopkins University<br />
PAUL ECKERT, Ph.D.,<br />
Stadtwerke Düsseldorf<br />
WILLIAM D. GOLLNITZ,<br />
Greater Cincinnati <strong>Water</strong> Works<br />
THOMAS GRISCHEK, Ph.D.,<br />
University of Applied Sciences<br />
Dresden<br />
DAVID L. HAAS, P.E.,<br />
Jordan, Jones & Goulding<br />
THOMAS HEBERER, Ph.D.,<br />
Technical University of Berlin<br />
STEPHEN HUBBS, P.E.,<br />
Louisville <strong>Water</strong> Company<br />
Conference Sponsors<br />
HENRY C. HUNT, CPG,<br />
Collector Wells International, Inc.<br />
RUDOLF IRMSCHER, Ph.D.,<br />
Stadtwerke Düsseldorf AG<br />
RONALD B. LINSKY,<br />
<strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong><br />
JÜRGEN SCHUBERT,<br />
Stadtwerke Düsseldorf AG<br />
RODNEY A. SHEETS,<br />
United States Geological Survey<br />
THOMAS SPETH, Ph.D., P.E.,<br />
United States<br />
Environmental Protection Agency<br />
✦ Environmental & <strong>Water</strong> Resources <strong>Institute</strong> of the American Society of Civil Engineers<br />
✦ Greater Cincinnati <strong>Water</strong> Works<br />
✦ International Association of <strong>Water</strong>works in the Rhine Catchment Area<br />
✦ Jordan, Jones & Goulding<br />
✦ Louisville <strong>Water</strong> Company<br />
✦ Stadtwerke Düsseldorf AG<br />
✦ United States Environmental Protection Agency<br />
✦ United States Geological Survey<br />
in collaboration with<br />
✦ International Association of Hydrogeologists Commission<br />
on Management of Aquifer Recharge<br />
i
Foreword<br />
In 1999, the <strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong> (NWRI) invited American and European<br />
experts to the first International Riverbank Filtration Conference, held in Louisville,<br />
Kentucky, to discuss and promote riverbank filtration, a low-cost, relatively unknown water<br />
treatment technology in the United States. The result of this conference was Riverbank<br />
Filtration: Improving Source-<strong>Water</strong> Quality, a 365-page book written by over 30 experts and<br />
jointly published by NWRI and Kluwer Academic Publishers in 2002.<br />
The editors of Riverbank Filtration — namely, Chittaranjan Ray, Gina Melin, and Ronald<br />
Linsky — saw a need to expand on the topics raised in the book, specifically because the<br />
United States Environmental Protection Agency is developing the proposed Long Term 2<br />
Enhanced Surface <strong>Water</strong> Treatment Rule, which applies to public water-supply systems that<br />
use either surface water or groundwater under the direct influence of surface water as their<br />
raw-water source. As a result, NWRI organized the Second International Riverbank<br />
Filtration Conference, which featured over 40 presenters from around the world and was held<br />
in Cincinnati, Ohio, in September 2003.<br />
This conference came about through the dedication and assistance of numerous individuals<br />
and organizations worldwide. NWRI gratefully acknowledges the ef<strong>for</strong>ts of all those involved<br />
with planning, organizing, and sponsoring the conference, especially the 15-member<br />
Conference Planning Committee. NWRI also extends special thanks to the conference<br />
speakers, moderators, and panelists, whose expertise provided invaluable insight into the<br />
status and needs of riverbank-filtration technology. Finally, NWRI would specifically like to<br />
thank the Greater Cincinnati <strong>Water</strong> Works and Jordan, Jones & Goulding <strong>for</strong> providing<br />
countless hours and manpower towards organizing this conference.<br />
The extended abstracts provided at the conference were the contributions of conference<br />
presenters. Abstracts were edited only when obvious errors were detected or when printing<br />
requirements necessitated action. The opinions expressed within the abstracts are those of<br />
individual authors and do not necessarily reflect those of the sponsors.<br />
NWRI would like to extend sincere thanks to Gina Melin, Editor and Conference<br />
Coordinator, and Tim Hogan, Graphics Designer, <strong>for</strong> their ef<strong>for</strong>ts in bringing the abstracts to<br />
press and ensuring that the quality of each and every abstract reached its fullest potential.<br />
Ronald B. Linsky<br />
Executive Director<br />
<strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong><br />
iii
Conference Schedule & Abstracts Contents<br />
Wednesday, September 17, 2003<br />
All sessions will take place in the Continental Room<br />
7:00 am Registration Mezzanine Level<br />
7:00 am Speaker Breakfast Rosewood Room<br />
8:00 am Welcome Continental Room<br />
Ronald B. Linsky, <strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong>, Cali<strong>for</strong>nia<br />
8:15 am Keynote Address<br />
Riverbank Filtration: The American Experience . . . . . . . . . . . . . 1<br />
Edward J. Bouwer, Ph.D., The Johns Hopkins University, Maryland<br />
8:45 am Session 1: Costs<br />
Moderated by William D. Gollnitz, Greater Cincinnati <strong>Water</strong> Works, Ohio<br />
The Costs and Benefits of Riverbank-Filtration Systems . . . . . . . . . . . . . . 3<br />
Stephen A. Hubbs, P.E., Louisville <strong>Water</strong> Company, Kentucky<br />
9:15 am Session 2: Operations<br />
Moderated by Chittaranjan Ray, Ph.D., P.E.,<br />
University of Hawaii at Mañoa, Hawaii<br />
Bridging <strong>Research</strong> and Practical Design Applications . . . . . . . . . . . . . . . . . 7<br />
David L. Haas, P.E., Jordan, Jones & Goulding, Inc., Georgia<br />
Construction and Maintenance of Wells <strong>for</strong> Riverbank Filtration . . . . . . . 17<br />
Henry C. Hunt, CPG, Collector Wells International, Inc., Ohio<br />
Aquifer Storage and Recharge Pretreatment:<br />
Synergies of Bank Filtration, Ozonation, and Ultraviolet Disinfection . . . 23<br />
Robert S. Cushing, Ph.D., Carollo Engineers, Florida<br />
Evolution from a Conventional Well Field<br />
to a Riverbank-Filtration System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29<br />
John D. North, Cedar Rapids <strong>Water</strong> Department, Iowa<br />
11:15 am Session 3A: Hydraulic Aspects<br />
Moderated by Rudolf Irmscher, Ph.D., Stadtwerke Düsseldorf, Germany<br />
Groundwater Flow and <strong>Water</strong>-Quality – A Flowpath Study<br />
in the Seminole Well Field, Cedar Rapids, Iowa . . . . . . . . . . . . . . . . . . . . . 35<br />
Douglas J. Schnoebelen, Ph.D., United States Geological Survey, Iowa<br />
The Use of Aquifer Testing and Groundwater Modeling<br />
to Evaluate Changes in Aquifer/River Hydraulics at<br />
the Louisville <strong>Water</strong> Company . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39<br />
David C. Schafer, David Schafer & Associates, Minnesota<br />
12:15 pm Lunch Rosewood Room<br />
v
vi<br />
1:30 pm Session 3B: Hydraulic Aspects<br />
Moderated by Rudolf Irmscher, Ph.D., Stadtwerke Düsseldorf, Germany<br />
Plugging in Riverbank-Filtration Systems:<br />
Evaluating Yield-Limiting Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43<br />
Stephen A. Hubbs, P.E., Louisville <strong>Water</strong> Company, Kentucky<br />
Application of Different Tracers to Evaluate the Flow Regime<br />
at Riverbank-Filtration Sites in Berlin, Germany . . . . . . . . . . . . . . . . . . . . 49<br />
Dr. Gudrun Massman, Free University of Berlin, Germany<br />
2:30 pm Session 4: Siting<br />
Moderated by Henry C. Hunt, CPG,<br />
Collector Wells International, Inc., Ohio<br />
Siting and Testing Procedures <strong>for</strong> Riverbank-Filtration Systems . . . . . . . . 57<br />
Samuel M. Stowe, P.G., CPG, International <strong>Water</strong> Consultants, Inc., Ohio<br />
<strong>Water</strong> Quality Management <strong>for</strong> Existing Riverbank Filtration Sites<br />
along the Elbe River in Germany. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63<br />
Prof. Dr.-Ing. Thomas Grischek, University of Applied Sciences Dresden, Germany<br />
3:30 pm Session 5: Dynamics<br />
Moderated by Prof. Dr.-Ing. Thomas Grischek,<br />
University of Applied Sciences Dresden, Germany<br />
Using Models to Predict Filtrate Quality at Riverbank-Filtration Sites –<br />
What Is the Adequate Level of Modeling?. . . . . . . . . . . . . . . . . . . . . . . . . . 69<br />
Chittaranjan Ray, Ph.D., P.E., University of Hawaii at Mañoa, Hawaii<br />
The 100-Year Flood of the Elbe River in 2002<br />
and Its Effects on Riverbank-Filtration Sites . . . . . . . . . . . . . . . . . . . . . . . 81<br />
Dipl.-Ing. Matthias Krueger, Fernwasserversorgung Elbaue-Ostharz GmbH, Germany<br />
Temporal Changes of Natural Attenuation Processes<br />
During Bank Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87<br />
Paul Eckert, Ph.D., Stadtwerke Düsseldorf AG, Germany<br />
An Update of the City of Guelph’s Response to Regulation 459/00:<br />
Effective Natural In Situ Filtration of Several Groundwater<br />
Under the Direct Influence of Surface-<strong>Water</strong> Supplies . . . . . . . . . . . . . . . 91<br />
Dennis E. Mutti, P.E., Associated Engineering Limited, Canada<br />
On Bank Filtration and Reactive Transport Modeling . . . . . . . . . . . . . . . . 93<br />
Dr.-Ing. Ekkehard Holzbecher, Humboldt University, Germany<br />
6:00 pm Reception Pavilion Foyer<br />
6:30 pm Dinner Pavilion Ballroom<br />
Dinner Speaker:<br />
Hydraulic Sensitivities and Reduction Potential Correlated<br />
with the Distance Between the Riverbank and Production Well . . . . . . . . 99<br />
Bernhard Wett, Ph.D., University of Innsbruck, Austria
Thursday, September 18, 2003<br />
All sessions will take place in the Continental Room<br />
7:00 am Speaker Breakfast Rosewood Room<br />
8:00 am Keynote Address<br />
Riverbank Filtration: The European Experience . . . . . . . . . . . . . .<br />
Prof. Dr.-Ing. Martin Jekel, Technical University of Berlin, Germany<br />
105<br />
8:30 am Session 6: Microorganisms<br />
Moderated by Edward J. Bouwer, Ph.D., The Johns Hopkins University, Maryland<br />
Using Microscopic Particulate Analysis <strong>for</strong> Riverbank Filtration . . . . . . . 111<br />
Jennifer L. Clancy, Ph.D., Clancy Environmental Consultants, Inc., Vermont<br />
Transport and Removal of Cryptosporidium Oocysts<br />
in Subsurface Porous Media. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115<br />
Menachem Elimelech, Ph.D., Yale University, Connecticut<br />
Laboratory and Field Strategies <strong>for</strong> Assessing Pathogen Removal<br />
by Riverbank Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117<br />
Monica B. Emelko, Ph.D., University of <strong>Water</strong>loo, Canada<br />
Fate of Disinfection Byproduct Precursors and Microorganisms<br />
During Riverbank Filtration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123<br />
W. Joshua Weiss, The Johns Hopkins University, Maryland<br />
Assessment of the Microbial Removal Capabilities<br />
of Riverbank Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129<br />
Vasiliki Partinoudi, University of New Hampshire, New Hampshire<br />
11:00 am Session 7: Organics Removal<br />
Moderated by Richard J. Miltner, P.E.,<br />
United States Environmental Protection Agency, Ohio<br />
Riverbank Filtration: A Very Efficient Treatment Process<br />
<strong>for</strong> the Removal of Organic Contaminants?. . . . . . . . . . . . . . . . . . . . . . . . . 137<br />
Dr.-Ing. Heinz-Jürgen Brauch, DVGW-Technologiezentrum Wasser, Germany<br />
Organics Removal by Riverbank Filtration<br />
at the Greater Cincinnati <strong>Water</strong> Works Site . . . . . . . . . . . . . . . . . . . . . . . . 143<br />
Jeffrey Vogt, Greater Cincinnati <strong>Water</strong> Works, Ohio<br />
12:00 pm Lunch Pavilion Ballroom<br />
Lunch Speaker:<br />
Potential Uses of Riverbank Filtration<br />
<strong>for</strong> Regulatory Compliance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .<br />
Stig Regli, United States Environmental Protection Agency, Washington, D.C.<br />
149<br />
vii
viii<br />
1:15 pm Session 8: Emerging Contaminants Removal<br />
Moderated by Monica B. Emelko, Ph.D., University of <strong>Water</strong>loo, Canada<br />
Transport and Attenuation of Pharmaceutical Residues<br />
During Bank Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151<br />
Andy Mechlinski, Technical University of Berlin, Germany<br />
Attenuation of Pharmaceuticals During Riverbank Filtration . . . . . . . . . . 155<br />
Traugott Scheytt, Ph.D., Technical University of Berlin, Germany<br />
The Fate of Bulk Organics and Emerging Contaminants<br />
During Soil-Aquifer Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159<br />
Dr. Jörg E. Drewes, Colorado School of Mines, Colorado<br />
Ethylenediaminetetraacetic Acid Occurrence and Removal<br />
Through Bank Filtration in the Platte River, Nebraska . . . . . . . . . . . . . . . 163<br />
Jason R. Vogel, Ph.D., United States Geological Survey, Nebraska<br />
3:15 pm Session 9: Public Policy and Regulatory<br />
Moderated by Richard J. Miltner, P.E.,<br />
United States Environmental Protection Agency, Ohio<br />
Riverbank Filtration as a Regional Supply Option<br />
<strong>for</strong> the United States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167<br />
Leo Gentile, P.G., CPG, Jordan, Jones & Goulding, Inc., Georgia<br />
Application of the Long Term 2 Enhanced Surface <strong>Water</strong><br />
Treatment Rule Microbial Toolbox at Existing <strong>Water</strong> Plants . . . . . . . . . . . 173<br />
Richard A. Brown, Environmental Engineering and Technology, Inc., Virginia<br />
Draft Protocol <strong>for</strong> the Demonstration of Effective Riverbank Filtration . . . . 175<br />
William D. Gollnitz, Greater Cincinnati <strong>Water</strong> Works, Ohio<br />
Source <strong>Water</strong> Protection and Riverbank Filtration<br />
in the Dyje River Basin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181<br />
Prof.-Dr. Petr Hlavinek, Brno University of Technology, Czech Republic<br />
5:15 pm Session 10: <strong>Research</strong> Needs Panel Discussion<br />
Moderated by Ronald B. Linsky, <strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong>, Cali<strong>for</strong>nia<br />
Panelists:<br />
Kellogg J. Schwab, Ph.D.,<br />
Johns Hopkins Bloomberg School of Public Health, Maryland<br />
Peter Fox, Ph.D., Arizona State University, Arizona<br />
Monica B. Emelko, Ph.D., University of <strong>Water</strong>loo, Canada<br />
Prof. Dr.-Ing. Thomas Grischek,<br />
University of Applied Sciences Dresden, Germany<br />
Prof. Vladimir Rojanschi, Ecological University Bucharest, Romania<br />
6:30 pm Reception Rosewood Room
Friday, September 19, 2003<br />
7:00 am Speaker Breakfast Rosewood Room<br />
8:00 am Session 11: Case Studies “Lessons Learned”<br />
Moderated by Stephen Hubbs, P.E., Louisville <strong>Water</strong> Company, Kentucky<br />
Greater Cincinnati <strong>Water</strong> Works<br />
Flowpath Study Field Design: Methodology and Evaluation. . . . . . . . . . . . 187<br />
Bruce Whitteberry, P.G., Greater Cincinnati <strong>Water</strong> Works, Ohio<br />
The Hungarian Experience with Riverbank Filtration . . . . . . . . . . . . . . . . 193<br />
Ferenc Laszlo, Ph.D.,<br />
<strong>Institute</strong> <strong>for</strong> <strong>Water</strong> Pollution Control, <strong>Water</strong> Resources <strong>Research</strong> Centre, Hungary<br />
Nitrate Pollution of a <strong>Water</strong> Resource –<br />
15 18<br />
N and 0 Study of Infiltrated Surface <strong>Water</strong> . . . . . . . . . . . . . . . . . . . . . 197<br />
Frantisek Buzek, Ph.D., Czech Geological Survey, Czech Republic<br />
Microbial Growth in Artificially Recharged Groundwater:<br />
Experiences from a 4-Year Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203<br />
Ilkka T. Miettinen, Ph.D., <strong>National</strong> Public Health <strong>Institute</strong>, Finland<br />
Evaluation of the Existing Per<strong>for</strong>mance of Infiltration Galleries<br />
in the Alluvial Deposits of the Parapeti River . . . . . . . . . . . . . . . . . . . . . . 207<br />
Dip.-Eng. Alvaro Camacho Garnica,<br />
Bolivian Association of Sanitary Engineers, Bolivia<br />
Sensitivity and Implication of Microscopic Particulate Analysis –<br />
A Collector Well Owner’s Perspective. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215<br />
Barry C. Beyeler, City of Boardman, Oregon<br />
Combined Use of Surface <strong>Water</strong> and Groundwater<br />
<strong>for</strong> Drinking-<strong>Water</strong> Production in the Barcelona Metropolitan Area. . . . . 217<br />
Jordi Martín-Alonso, Barcelona’s <strong>Water</strong> Company, Spain<br />
11:30 am Conference Wrap-Up<br />
Edward J. Bouwer, Ph.D., Johns Hopkins University, Maryland<br />
Martin Jekel, Ph.D., Technical University of Berlin, Germany<br />
ix
Acronyms<br />
ADA ß-alaninediacetic acid<br />
AOC Assimilable organic carbon<br />
AOX Adsorbable organic halogen<br />
ASR Aquifer Storage and Recovery<br />
DBP Disinfection byproduct<br />
DOC Dissolved organic carbon<br />
DTPA Diethylenetrinitrilopenataacetic acid<br />
EDTA Ethylenediaminetetraacetic acid<br />
GAC Granular activated carbon<br />
GC Gas chromatography<br />
GWUDI Groundwater under the direct influence of surface water<br />
HAA Haloacetic acid<br />
HPI Hydrophilic carbon<br />
HPO-A Hydrophobic acids<br />
LT2ESWTR Long Term 2 Enhanced Surface <strong>Water</strong> Treatment Rule<br />
MAP Microbially available phosphorous<br />
MS Mass spectrometry<br />
NASRI Natural and Artificial Systems <strong>for</strong> Recharge and Infiltration<br />
NTA Nitrilotriacetic acid<br />
NOM Natural organic matter<br />
NPEC Nonylphenolpolyethoxycarboxylate<br />
O&M Operation and maintenance<br />
PDTA 1.3-propylenedinitrilotetraacetic acid<br />
PhAC Pharmaceutically active compound<br />
PHREEQC pH redox equilibrium equation<br />
<strong>RBF</strong> Riverbank filtration<br />
THM Trihalomethane<br />
TOC Total organic carbon<br />
USEPA United States Environmental Protection Agency<br />
USGS United States Geological Survey<br />
UV Ultraviolet<br />
xi
xii<br />
Units of Measure<br />
cfu Colony-<strong>for</strong>ming unit.<br />
ft Foot<br />
m Meter<br />
m 3 /d Cubic meters per day<br />
m 3 /s Cubic meters per second<br />
MGD Million gallons per day<br />
mg/L Milligrams per liter<br />
ntu Nephelometric turbidity unit<br />
pCi/L Picocurie per liter<br />
pfu Plaque-<strong>for</strong>ming unit.<br />
µg/L Micrograms per liter
Conference<br />
Abstracts
Keynote Presentation<br />
Riverbank Filtration: The American Experience<br />
Edward J. Bouwer, Ph.D.<br />
The Johns Hopkins University<br />
Baltimore, Maryland<br />
Riverbank filtration (<strong>RBF</strong>) is a process during which surface water is subjected to subsurface flow<br />
prior to extraction from vertical or horizontal wells. Most <strong>RBF</strong> systems are located along<br />
riverbanks. Alternative systems can involve lakes and infiltration ponds. The objective of this<br />
keynote address is to provide an overview of the water-quality improvements possible with <strong>RBF</strong>,<br />
the motivation <strong>for</strong> <strong>RBF</strong> and its promise in the United States, and some of the remaining<br />
challenges <strong>for</strong> the reliable implementation of this technology.<br />
During infiltration and soil passage, surface water is subjected to a combination of physical,<br />
chemical, and biological processes, such as filtration, dilution, sorption, and biodegradation, that<br />
can significantly improve raw-water quality. Transport through alluvial aquifers is associated with<br />
a number of water-quality benefits, including the removal of microbes, pesticides, total organic<br />
carbon (TOC), dissolved organic carbon (DOC), nitrate, and other contaminants. In comparison<br />
to most groundwater sources, alluvial aquifers that are hydraulically connected to rivers are<br />
typically easier to exploit (shallow) and more highly productive <strong>for</strong> drinking-water supplies. As<br />
reflected by several recently published reviews (Journal of Hydrology, 2002; Ray et al., 2002a;<br />
Tufenkji et al., 2002; and Ray et al., 2002b), <strong>RBF</strong> is receiving increased attention, especially in the<br />
United States. One motivation <strong>for</strong> the increased applications of <strong>RBF</strong> is the need <strong>for</strong> drinking-water<br />
utilities to meet increasingly stringent drinking-water regulations, especially with regard to:<br />
• The provision of multiple barriers <strong>for</strong> protection against microbial pathogens.<br />
• Tighter regulations <strong>for</strong> disinfection byproducts (DBPs), such as trihalomethanes (THMs)<br />
and haloacetic acids (HAAs).<br />
A second motivation is the ability of <strong>RBF</strong> to provide continuous treatment and to buffer against<br />
accidental spills and terrorist events.<br />
Since <strong>RBF</strong> is a natural treatment system and has a long history of use in Europe, it is sometimes<br />
viewed as a simple process. This simplicity is appealing; however, the geochemical, biological, and<br />
hydrologic factors that control the removal of dissolved and particulate contaminants during <strong>RBF</strong><br />
are complex. We have learned from experience with groundwater remediation that the effectiveness<br />
of subsurface processes is inherently site-specific. Many of the papers presented at this conference<br />
will address the factors influencing water-quality improvements possible with <strong>RBF</strong> systems.<br />
One question of interest to users can be stated as, “Is <strong>RBF</strong> with post-disinfection an acceptable, or<br />
even preferable, alternative to conventional drinking-water treatment?” In my laboratory, the<br />
reduction of DBP precursors upon <strong>RBF</strong> was compared with that obtained using a bench-scale<br />
Correspondence should be addressed to:<br />
Edward J. Bouwer, Ph.D.<br />
Professor, Department of Geography and Environmental Engineering<br />
The Johns Hopkins University<br />
3400 N. Charles Street • Baltimore, Maryland 21218 USA<br />
Phone: (410) 516-7437 • Fax: (410) 516-8996 • Email: Bouwer@jhu.edu<br />
1
2<br />
conventional treatment train on corresponding river waters. The river waters were subjected to a<br />
treatment train consisting of coagulation, flocculation, sedimentation, filtration, and ozonation.<br />
This research showed that <strong>RBF</strong> per<strong>for</strong>ms as well as or better than a bench-scale conventional<br />
treatment train (based on coagulation chemistry <strong>for</strong> optimum turbidity removal) with respect to<br />
the removal of natural organic matter (NOM), particularly precursor material <strong>for</strong> THM4 and<br />
HAA6 concentrations. Consequently, a potential major benefit of <strong>RBF</strong> is as a pretreatment step<br />
<strong>for</strong> controlling DBPs. A shift from chlorinated to brominated DBPs occurred during <strong>RBF</strong>. Since<br />
brominated DBPs tend to have greater toxicity than chlorinated DBPs, the shift from chlorinated<br />
to brominated DBP species caused reductions in the calculated risk <strong>for</strong> bank-filtered waters in this<br />
study to be lower than corresponding reductions in THM concentrations. Nonetheless, the data<br />
demonstrate the ability of <strong>RBF</strong> to reduce theoretical risk due to THMs <strong>for</strong>med upon chlorination<br />
in all cases and with substantially better per<strong>for</strong>mance than the bench-scale conventional<br />
treatment train.<br />
A remaining challenge <strong>for</strong> the reliable implementation of <strong>RBF</strong> as a technology is the dynamic<br />
nature of these complex hydrologic systems. Transient flows in rivers influence flow and removal<br />
processes during ground passage, which can affect the quality of bank-filtered waters. A more<br />
detailed understanding of the fundamental processes that govern contaminant transport in <strong>RBF</strong><br />
systems will lead to the reliable design, implementation, and operation of <strong>RBF</strong> systems.<br />
REFERENCES<br />
Journal of Hydrology (2002). Special Issue on Bank Filtration, 266(3-4): 139-284.<br />
Ray, C., T. Grischek, J. Schubert, J.Z. Wang, and T.F. Speth (2002a). “A perspective of riverbank filtration.”<br />
Jour. AWWA, 94(4): 149-160.<br />
Ray, C., G. Melin, and R.B. Linsky, editors (2002b). Riverbank Filtration: Improving Source <strong>Water</strong> Quality,<br />
Kluwer Academic Publishers, Dordrecht.<br />
Tufenkji, N., J.N. Ryan, and M. Elimelech (2002). “The promise of bank filtration.” Environmental Science<br />
and Technology, 36(21): 423A-428A.<br />
ED BOUWER has taught environmental engineering courses at The Johns Hopkins<br />
University since 1985. His research interests include factors that influence the biotrans<strong>for</strong>mation<br />
of organic contaminants, bioremediation <strong>for</strong> the control of organic contaminants<br />
at waste sites, biofilm kinetics, the interaction between biotic and abiotic processes,<br />
groundwater contamination, biological processes design in wastewater, industrial and<br />
drinking-water treatment, and the transport and fate of microorganisms in porous media.<br />
At present, he is an editor <strong>for</strong> or is on the editorial board of several journals, including the<br />
Journal of Contaminant Hydrology, Biodegradation, and Environmental Engineering Sciences. Bouwer received a<br />
B.S. in Civil Engineering from Arizona State University, and both an M.S. and Ph.D. in Environmental<br />
Engineering and Science from Stan<strong>for</strong>d University.
Session 1: Costs<br />
The Costs and Benefits of Riverbank-Filtration Systems<br />
Stephen A. Hubbs, P.E.<br />
Louisville <strong>Water</strong> Company<br />
Louisville, Kentucky<br />
Henry C. Hunt, CPG<br />
Collector Wells International, Inc.<br />
Columbus, Ohio<br />
Jürgen Schubert<br />
Stadtwerke Düsseldorf<br />
Düsseldorf, Germany<br />
The Benefits of Riverbank Filtration<br />
The history of <strong>RBF</strong> in modern times is connected to the experience of disease outbreaks in Europe<br />
in the 1890s, with specific reference to the cholera epidemic in Hamburg, Germany, in 1892. The<br />
preference <strong>for</strong> groundwater, <strong>RBF</strong>, and artificial recharge to surface water stems from this<br />
experience, noting much more wholesome water from these sources than from surface water.<br />
A much earlier reference to <strong>RBF</strong> (albeit far less scientific) is found in the Bible:<br />
The fish that were in the Nile died, and the Nile became foul, so that the Egyptians<br />
could not drink water from the Nile … So all the Egyptians dug around the Nile <strong>for</strong><br />
water to drink, <strong>for</strong> they could not drink of the water of the Nile. (Exodus 7, 21-24)<br />
Thus, it appears that the benefits of <strong>RBF</strong> are anything but new!<br />
The benefits of <strong>RBF</strong> have been cataloged in recent publications to include:<br />
• Particle removal.<br />
• Pathogen removal.<br />
• Organic and inorganic chemical removal.<br />
• Peak smoothing in spills.<br />
• Reduction in DBP <strong>for</strong>mation.<br />
• Production of a more biologically stable water.<br />
When faced with the decision to choose an advanced treatment technology <strong>for</strong> its two treatment<br />
plants, the Louisville <strong>Water</strong> Company in Louisville, Kentucky, sought to compare these benefits<br />
against the benefits of in-plant treatment techniques from both a treatment efficiency perspective<br />
and cost-based perspective; however, a comprehensive, quantitative measure was difficult to develop.<br />
Correspondence should be addressed to:<br />
Stephen A. Hubbs, P.E.<br />
Vice President, New Technology<br />
Louisville <strong>Water</strong> Company<br />
550 South Third Street • Louisville, Kentucky 40202 USA<br />
Phone: (502) 569-3675 • Fax: (502) 569-0813 • Email: SHubbs@lwcky.com<br />
3
4<br />
The Louisville <strong>Water</strong> Company also wanted to involve the public in the overall decision process,<br />
so a simple communication tool was desired that was capable of ranking various treatment<br />
technologies with regards to the risks they were designed to reduce. Treatment effectiveness was<br />
simply identified by a series of “+” signs, with one “+” being effective and two “++” being highly<br />
effective. A negative assignment (“–”) indicated that the treatment process had an overall<br />
negative impact on the risk of a selected component, and a “0” indicated no impact.<br />
The risks evaluated in the analysis included: pathogens, such as Giardia and Cryptosporidium;<br />
DBPs; tastes and odor (2-methylisoborneol and Geosmin); synthetic organic chemicals like<br />
atrazine and Aloclor; and river-borne spills of industrial chemicals. Operational benefits included:<br />
reduced regrowth in the distribution system, reduced main breaks from avoidance of extremely<br />
cold water, avoidance of zebra mussels, and reliability associated with simple operation. These risks<br />
and benefits were used to compare the <strong>RBF</strong> process to other treatment technologies, including<br />
conventional treatment, ozone, ultraviolet light (UV), granular activated carbon (GAC), biological<br />
active carbon, combinations of all of these, and membranes.<br />
This matrix was recently modified to include risks associated with the presence of radon and<br />
arsenic in groundwater. Radon has been detected in <strong>RBF</strong> water in Louisville at levels near the<br />
maximum contaminant level of 300 picocuries per liter (pCi/L). The radon level leaving the<br />
treatment plant is below the detection limit of 50 pCi/L, with the reduction the result of<br />
out-gassing in open treatment basins and blending with surface water. Arsenic has not been<br />
detected in <strong>RBF</strong> water at Louisville. Radon and arsenic risk factors do not exist in Ohio River<br />
source water. The treatments <strong>for</strong> radon (aeration) and arsenic (flocculation with ferric) effectively<br />
reduce these risk factors at an increase in treatment costs.<br />
A matrix comparing these risks and operational benefits is presented in Table 1. This analysis<br />
indicates which treatment combinations provide specific water-quality benefits. From this analysis,<br />
it was obvious that <strong>RBF</strong> was capable of matching the benefits of in-plant treatment options.<br />
Table 1. Comparison of Risks and Operational Benefits<br />
River + Ozone River + GAC River + GAC<br />
Benefits <strong>RBF</strong> + UV and UV + UV + Membranes<br />
Particle Removal + + + +<br />
Microbial Removal + + + +<br />
DBP Reduction + + + +<br />
Taste and Odor + + + +<br />
Spill Dampening ++ 0 ++ +<br />
Iron/Manganese – + + 0<br />
SOC Removal + + + +<br />
AOC/BDOC Control + 0 + +<br />
Nitrification Control + + + +<br />
Temperature ++ 0 0 0<br />
Operability + 0 0 +<br />
Residuals 0 0 0 +<br />
Multiple Barriers ++ ++ ++ ++<br />
Reduced Chemicals + 0 0 +<br />
Radon – 0 0 0<br />
Arsenic 0 0 0 0<br />
TOTAL 13 9 12 13<br />
AOC = Assimilable organic carbon. BDOC = Biodegradable organic carbon. SOC = Synthetic organic chemical.
Cost-Evaluation of Riverbank Filtration<br />
The capital cost of a <strong>RBF</strong> system depends on many factors, including aquifer characteristics, type<br />
of well-screen installation (vertical or horizontal), aesthetic considerations in facility design, and<br />
distance to the population served. The operational costs vary as a function of water quality and<br />
required treatment, lift required in pumping, and pump and well-screen maintenance costs. The<br />
ability of a stream to support a given well-field capacity can be calculated as the yield per unit<br />
length of riverbank. This capacity is influenced by the composition of the riverbed, riverbed<br />
scouring characteristics, stream-water quality, and width of the river.<br />
Large river systems situated in glacial sands and gravels (such as the Ohio, Mississippi, and Rhine<br />
rivers) can sustain yields up to 8 million gallons per day (MGD) per 1,000 feet (ft) of riverbank.<br />
Typical installations in these aquifers include 1.5-MGD vertical wells spaced on 200-ft centers, or<br />
15-MGD horizontal collector wells spaced at 2,000-ft centers.<br />
The 20-MGD horizontal collector well system constructed in Louisville in 1999 cost $5 million.<br />
The system included:<br />
• Seven laterals that are 200 to 240 ft in length.<br />
• A 21-ft diameter, 100-ft deep caisson.<br />
• A pump house and controls with one constant speed and one variable speed pump<br />
(both 10 MGD).<br />
• Two thousand feet of 42-inch discharge piping to the plant.<br />
The pump house was designed to include architectural features complimentary to surrounding<br />
residential neighborhoods. This system can peak at over 20 MGD in warm weather and is operated<br />
year-round at 17 MGD.<br />
When this system was being considered, alternative treatment techniques <strong>for</strong> surface-water treatment<br />
were also evaluated. Critical treatment functions included efficiency in removing Giardia and<br />
Cryptosporidium, ability to remove 2-methylisoborneol and Geosmin, and DBP reduction. Suitable<br />
alternatives were determined from the matrix and evaluated <strong>for</strong> cost. This process was repeated<br />
twice, by two separate consultants: once in the initial planning stages of the 20-MGD project<br />
(1995), and again just be<strong>for</strong>e a contract was let to design a 45-MGD expansion of the system<br />
(2002).<br />
The surface-water treatment process selected in the final cost analysis as providing comparable<br />
benefits to the overall benefits of <strong>RBF</strong> included: conventional treatment, ozone and biological<br />
treatment in GAC-capped filters, and UV/chlorine/chloramine disinfection. Treatment costs,<br />
however, were also estimated <strong>for</strong> separate elements of this treatment scheme (as was the cost of<br />
membranes) <strong>for</strong> the 180-MGD treatment plant at Crescent Hill.<br />
The results of this cost comparison are included in Table 2. The capital costs were amortized over<br />
20 years in this analysis, and the operation and maintenance (O&M) costs include the treatment<br />
costs to reduce hardness from 220 to 160 milligrams per liter (mg/L). Based on this analysis, the<br />
decision was made to proceed with the development of <strong>RBF</strong> <strong>for</strong> the entire capacity of the 60-MGD<br />
B.E. Payne <strong>Water</strong> Treatment Plant, and to continue considering <strong>RBF</strong> <strong>for</strong> the larger 180-MGD<br />
capacity Crescent Hill Treatment Plant.<br />
5
6<br />
Treatment Capital Cost Annual O&M Cost Present Worth Cost<br />
Alternative ($ million) ($ million) ($ million)<br />
<strong>RBF</strong> Conventional 103 1.82 112<br />
<strong>RBF</strong> + UV 116 2.34 130<br />
River + Ozone UV 70 6.36 152<br />
River + GAC + UV 114 5.46 160<br />
River + Membranes<br />
+ GAC + UV<br />
Table 2. Results of the Cost Comparison<br />
204 4.96 251<br />
Capital Costs of Alternative Well-Field Design: Hard-Rock Tunnel Collector<br />
The difficulty of obtaining riverfront property in developed areas prompted the Louisville <strong>Water</strong><br />
Company to consider alternative designs to the traditional well-field configuration. The design<br />
selected <strong>for</strong> constructing the 45-MGD addition to the existing 20-MGD <strong>RBF</strong> capacity includes<br />
the construction of a large-diameter (10- to 12-ft) horizontal tunnel in bedrock, below the waterbearing<br />
sand and gravel aquifer. Vertical wells will be constructed at 200-ft centers, and will<br />
penetrate the bedrock and discharge by gravity into the tunnel. A single pump station located at<br />
the treatment plant will extract up to 45 MGD from the tunnel, connecting the 30 vertical wells<br />
in the system. This construction technique minimizes the impact on landowners, with visible<br />
construction activities limited to drilling and developing the individual wells. This design also<br />
allows total flow to be distributed to each well, evenly distributing riverbed plugging stresses across<br />
6,000 ft of riverbank.<br />
The cost of this system was compared to the cost of developing a conventional well field with<br />
submersible pumps and a collecting header. The cost-estimate <strong>for</strong> the bedrock tunnel system was highly<br />
impacted by the assigned cost of the hard-rock tunnel. The final design included a cement-lined tunnel<br />
to protect against intermittent layers of shale encountered in the massive limestone <strong>for</strong>mation.<br />
The current estimate <strong>for</strong> this hard-rock tunnel-vertical well extraction system is $33 million <strong>for</strong><br />
45-MGD capacity. The system involves approximately 6,000 ft of riverbank and hard-rock tunnel.<br />
The comparable conventional vertical well field has been estimated at $23 million. Current plans<br />
are to design and construct the hard-rock vertical well system, noting advantages in expandability,<br />
future connection to an additional 180-MGD system <strong>for</strong> the larger treatment plant, and general<br />
constructability in the developed riverfront area.<br />
STEVE HUBBS is a Professional Engineer with 28 years of experience at the Louisville<br />
<strong>Water</strong> Company in Louisville, Kentucky. In the early 1980s, he began researching riverbank<br />
filtration as an alternate source of water <strong>for</strong> the Louisville <strong>Water</strong> Company, specifically<br />
looking at the reduction in disinfection byproduct precursors, river-borne organics, and<br />
mutagenicity in the riverbank-filtration process. His work continued in the 1990s with a<br />
focus on pathogen reduction, and his research is now being conducted on the hydraulic<br />
connection between the riverbed and aquifer, with a focus on riverbed plugging dynamics<br />
and their influence on sustainable yields from high-capacity riverbank-filtration systems. Hubbs received an<br />
M.S. in Environmental Engineering from the University of Louisville, and is currently enrolled in the Ph.D.<br />
program in Civil and Environmental Engineering at the University of Louisville, focusing on the hydraulics<br />
of riverbank-filtration systems.
Session 2: Operations<br />
Bridging <strong>Research</strong> and Practical Design Applications<br />
David L. Haas, P.E.<br />
Jordan, Jones & Goulding, Inc.<br />
Norcross, Georgia<br />
Michael J. Robison, P.E.<br />
Jordan, Jones & Goulding, Inc.<br />
Norcross, Georgia<br />
David R. Wilkes, P.E.<br />
Jordan, Jones & Goulding, Inc.<br />
Norcross, Georgia<br />
Background<br />
The Louisville <strong>Water</strong> Company operates two water filtration plants: the B.E. Payne <strong>Water</strong> Treatment<br />
Plant (Payne Plant), which has a treatment capacity of 60 MGD (227,000 cubic meters per day [m 3 /d])<br />
and is located in eastern Jefferson County, Kentucky; and the Crescent Hill <strong>Water</strong> Treatment<br />
Plant, which is located closer to downtown Louisville and has a treatment capacity of 180 MGD<br />
(682,000 m 3 /d). Both are conventional surface-water treatment plants drawing their supply from<br />
the Ohio River. The Payne Plant has two 60-inch (1.5-meter [m]) raw-water intake pipes located<br />
on plant property. The intake <strong>for</strong> the Crescent Hill <strong>Water</strong> Treatment Plant is located near the<br />
intersection of Zorn Avenue and River Road, approximately 7 miles (4.4 kilometers) downstream<br />
from the Payne Plant intake.<br />
In 1999, the Louisville <strong>Water</strong> Company started operating a horizontal collector well at its<br />
Payne Plant. This horizontal collector well consists of a 16-ft (4.9-m) diameter caisson.<br />
The caisson was constructed down to the top of bedrock and is approximately 100-ft (30.5-m)<br />
deep. There are seven 12-inch (30.5-centimeter) collector laterals radiating out horizontally from<br />
the caisson. A pump station was constructed on top of the caisson above the flood plain to pump<br />
water to the Payne Plant <strong>for</strong> further treatment. The design capacity of this well is 15 MGD<br />
(57,000 m 3 /d).<br />
Jordan, Jones & Goulding, Inc. was retained by the the Louisville <strong>Water</strong> Company to implement<br />
<strong>RBF</strong> <strong>for</strong> the remainder of the Payne Plant. This second phase of <strong>RBF</strong> will provide an additional<br />
45 MGD (171,000 m 3 /d) of capacity. Ultimately, the Louisville <strong>Water</strong> Company desires to extend<br />
<strong>RBF</strong> technology to the Crescent Hill <strong>Water</strong> Treatment Plant.<br />
This paper provides a summary of issues that were faced in taking the concept of providing<br />
additional <strong>RBF</strong> capacity <strong>for</strong> the Payne Plant through design. There were three primary areas that<br />
Correspondence should be addressed to:<br />
David L. Haas, P.E.<br />
Senior Project Manager<br />
Jordan, Jones & Goulding, Inc.<br />
6801 Governors Lake Parkway • Norcross, Georgia 30071 USA<br />
Phone: (678) 333-0242 • Fax: (678) 333-0828 • Email: DHAAS@JJG.com<br />
7
8<br />
design engineers needed to address:<br />
• Type of <strong>RBF</strong> collection system.<br />
• Hydrogeologic yield.<br />
• <strong>Water</strong> quality and treatment.<br />
Type of <strong>RBF</strong> Collection System<br />
First, it was necessary to select the most appropriate type of <strong>RBF</strong> technology to satisfy the technical<br />
needs of the project economically, as well as other goals such as community acceptance of the<br />
project. The construction of additional horizontal collector wells, similar to the existing well, was<br />
considered unacceptable because 15 to 20 of these collector wells would be required to satisfy the<br />
total treatment capacity <strong>for</strong> both plants. The construction of this many wells, each with an aboveground<br />
pump station, would be costly, as well as aesthetically unpleasing.<br />
Three options were considered that would use tunnels in conjunction with <strong>RBF</strong> to satisfy the<br />
needs of the project:<br />
• Option 1 consisted of driving a tunnel through the alluvial deposits that overlay bedrock<br />
and installing collector laterals from within the tunnel (soft-ground tunnel option). This<br />
design concept is depicted in Figure 1.<br />
• Option 2 included installing a series of horizontal collector wells (Figure 2) that would<br />
be capped at grade and connected together using a deep tunnel through the bedrock.<br />
• Option 3 consisted of installing vertical collectors connected together using a tunnel<br />
through the bedrock (Figure 3).<br />
With any of these options, the number of above-ground pump stations would be minimized,<br />
because water from horizontal or vertical collectors would be conveyed through the tunnel to a<br />
single pump station located on plant property.<br />
In addition to cost, factors that were considered in selecting the type of <strong>RBF</strong> collection system<br />
included construction and maintenance.<br />
Construction Considerations<br />
The two main construction issues that were evaluated <strong>for</strong> Option 1 were the feasibility of<br />
constructing the tunnel in the alluvial deposits along the riverbank and the feasibility of installing<br />
laterals safely from within the tunnel. With regard to Options 2 and 3, the main construction issue<br />
evaluated was rock quality.<br />
To evaluate these factors, a detailed subsurface investigation was conducted that included the<br />
following elements:<br />
• Large-diameter bucket auger borings to bedrock.<br />
• Grain-size analysis of composite samples obtained from the bucket auger borings.<br />
• Core borings of bedrock materials.<br />
• Examination of rock outcrops near the site.<br />
Large-diameter bucket auger borings were used to collect representative samples of alluvium.<br />
Because alluvium contains significant amounts of coarse gravel, cobbles, and possibly boulders, the
Ohio River<br />
Southern Caisson<br />
(capped at grade)<br />
19-ft Diameter<br />
Soft Soft-Ground<br />
Ground Tunnel unnel<br />
Tunnel<br />
Caisson<br />
Capped<br />
at Grade<br />
Ground Surface<br />
Pump<br />
Station<br />
Lagoon<br />
Existing Collector Well<br />
Existing 48-Inch<br />
Raw <strong>Water</strong> Transmission Line<br />
New 54-Inch<br />
Raw <strong>Water</strong> Transmission Line<br />
Lagoon<br />
Lagoon<br />
Clay<br />
Lagoon<br />
Sand and Gravel<br />
14-ft Diameter Tunnel with 25 Laterals<br />
Bedrock<br />
Plan View 1A<br />
Existing (Two)<br />
60-Inch Diameter<br />
<strong>Water</strong> Intake Pipes<br />
Northern Caisson<br />
(Pump Station)<br />
32-ft Diameter Shaft<br />
200-ft Laterals<br />
Existing B.E. Payne<br />
<strong>Water</strong> Treatment Plant<br />
15+0 10+0 5+00 0+00<br />
Figure 1. Soft ground tunnel concept (Option 1).<br />
Section View 1B<br />
large-diameter bucket auger was considered the most appropriate sample collection method.<br />
Grain-size distribution curves were developed based on sieve analyses of the bucket auger samples.<br />
The results of the rock-core boring program indicated the presence of limestone and shale. A<br />
particularly good layer of limestone occurred from 130- to 157-ft (40- to 48-m) below ground<br />
surface. Limestone beds in this unit tend to be on the order of 2- to 3-ft (0.6- to 0.9-m) thick,<br />
interbedded with 3- to 4-inch (7- to 10-centimeter) thick layers of shale. All of the limestone in<br />
this layer has a rock quality designation in the range of 90 to 100 percent, which according to<br />
Deere and Deere (1989) is classified as “excellent.”<br />
9
10<br />
Ohio River<br />
Pump Station<br />
Collector Well<br />
19-ft Diameter<br />
Drop Shaft<br />
4-ft Diameter<br />
Based on the results of the subsurface investigation, the construction of either type of tunnel<br />
system (soft ground or bedrock) was determined to be feasible.<br />
Maintenance Considerations<br />
Collector<br />
Well<br />
Collector<br />
Hard Rock<br />
Hard Roc Rock Tunnel unnel<br />
Lagoon<br />
200-ft<br />
Diameter Laterals<br />
11 per Collector Well<br />
7-ft Diameter<br />
Tunnel<br />
Existing<br />
Collector Well<br />
Lagoon<br />
Existing 48-Inch<br />
Raw <strong>Water</strong><br />
Transmission Line<br />
Lagoon<br />
Lagoon<br />
Pump Station<br />
32-ft Diameter Shaft<br />
Existing (2) 60-Inch Diameter<br />
<strong>Water</strong> Intake Pipes<br />
New 54-Inch Raw <strong>Water</strong><br />
Transmission Line<br />
200-ft<br />
Diameter Laterals<br />
11 per Collector Well Sand and<br />
Gravel<br />
24-ft Diameter<br />
Shaft<br />
in Bedrock<br />
Collector Well<br />
19-ft Diameter<br />
8+00 4+00 0+00<br />
Figure 2. Hard-rock tunnel concept using horizontal collector wells (Option 2).<br />
Existing B.E. Payne<br />
<strong>Water</strong> Treatment Plant<br />
Ground<br />
Surface<br />
Bedrock<br />
Plan View 2A<br />
One of the unique challenges of Option 1 would be the long-term maintenance of collector<br />
laterals. Over time, it is anticipated that the laterals would need to be cleaned to restore their<br />
yield, as would be the case with any well screen. Several options were considered that would allow<br />
individual laterals to be taken out of service <strong>for</strong> cleaning while the rest of the tunnel system<br />
remained in operation. The selected option consisted of a “wet/dry” tunnel system, with water<br />
being conveyed through carrier pipes in the “wet” side and O&M being per<strong>for</strong>med from within<br />
Clay<br />
Drop Shaft<br />
4-ft Diameter<br />
Section View 2B
TBM Exit<br />
Construction<br />
Vertical Collectors<br />
Installedon 200-ft Centers<br />
<strong>RBF</strong> Pump Station<br />
Ohio River<br />
Existing B.E. Payne<br />
<strong>Water</strong> Treatment Plant<br />
Capped Well Capped Well<br />
Clay<br />
Sand &<br />
Gravel<br />
Aquifer<br />
Bedrock<br />
Well Casing<br />
<strong>Water</strong> Table Level<br />
Wellscreen<br />
Tunnel<br />
Figure 3. Hard-rock tunnel concept using vertical collectors (Option 3).<br />
TBM Construction<br />
Existing Collector<br />
Well/Pump Station<br />
Approx. 2,000 ft<br />
Tunnel Extension<br />
(no wells)<br />
Existing Low Lift<br />
Pump Station<br />
Plan View 3A<br />
the “dry” side. This construction option would allow laterals to be maintained individually, while<br />
all other laterals remain in service. Figure 4 illustrates the “wet/dry” tunnel concept.<br />
Maintenance requirements <strong>for</strong> Option 2 would involve taking an entire caisson out of service <strong>for</strong><br />
lateral rehabilitation. This concept would reduce the effective capacity of the overall system by<br />
the capacity of an entire caisson, or approximately one-third of a three-caisson system.<br />
Option 3 would consist of individual wells that could be individually taken out of service and<br />
rehabilitated. A packer could be used to isolate the vertical collector from the tunnel. Based on<br />
maintenance considerations, Option 3 was preferred.<br />
3B<br />
To Pump<br />
Station<br />
Section View<br />
11
12<br />
Figure 4. Wet/dry tunnel concept.<br />
Hydrogeologic Yield<br />
Computer modeling was used to predict the safe yield of the aquifer and to optimize the placement<br />
of collectors <strong>for</strong> all of the options. For Option 1, collector laterals would be positioned 60 ft<br />
(18.3 m) on center on the riverside of the tunnel (Schafer, 2000). Using a 1,500-ft (457-m) long<br />
tunnel with laterals located on 60-ft (18.3-m) centers resulted in an estimated yield ranging from<br />
44 to 61 MGD (167,000 to 231,000 m 3 /d), depending on water temperature. For Option 2, the<br />
optimum number of laterals per collector well was determined to be 10 to 11 (Schafer, 2000) and<br />
the estimated yield ranged from 37 to 51 MGD (140,000 to 193,000 m 3 /d), depending on water<br />
temperature, <strong>for</strong> a system consisting of three new collector wells. For Option 3, approximately<br />
30 vertical collectors would be installed approximately 200-ft (67-m) apart, resulting in an<br />
estimated yield ranging from 38 to 44 MGD (144,000 to 167,000 m 3 /d) (Shafer, 2002).<br />
Data from the existing collector well and pumping tests were used as the basis <strong>for</strong> selecting<br />
leakance and transmissivity values that were used in calculating the predicted yields. Figure 5<br />
shows measured leakance data from the existing collector. As shown, leakance declined over the<br />
initial 2-year period of operation from over 2.0 to 0.148 inverse days due to clogging in the<br />
riverbed. Within the past year, measured leakance values appear to have stabilized and increased<br />
slightly. To be conservative in predicting yields <strong>for</strong> the new <strong>RBF</strong> addition, the lowest observed<br />
leakance value was used.<br />
<strong>Water</strong> Quality and Treatment<br />
Tunnel Cross-Section<br />
Gate Valves with<br />
Removable Elbow<br />
Precast<br />
Concrete<br />
Tunnel Segments<br />
12-Inch Lateral<br />
Concrete Fill<br />
Two 48-Inch Carrier Pipes<br />
<strong>Water</strong> quality obtained from the collector wells in many locations has been documented as having<br />
many advantages over that obtained directly from surface-water sources (Sontheimer, 1980;<br />
Hubbs, 1981; Wang et al., 1995; <strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong>, 1999; Kuehn and Mueller,<br />
2000). In the Louisville <strong>Water</strong> Company’s case, the existing <strong>RBF</strong> collector well has shown many
Leakance (inverse days)<br />
10<br />
1<br />
0.1<br />
10/1/1999 3/31/2000 9/30/2000 3/31/2001 9/30/2001 3/31/2002 9/30/2002<br />
favorable improvements in water quality compared to the Ohio River source (Wang et al., 2002),<br />
including:<br />
• Lower turbidity.<br />
• Lower TOC.<br />
• Increased minimum temperature.<br />
• Reduced microbial contaminants.<br />
Date<br />
Figure 5. Measured leakance values corrected to 68-degrees Fahrenheit (October 1999).<br />
Table 1 summarizes typical water quality from the collector well compared to local groundwater<br />
and Ohio River sources.<br />
Table 1. Characteristics of River <strong>Water</strong> and Groundwater in the Project Area<br />
Infiltrated Bedrock<br />
Parameters River <strong>Water</strong> Groundwater Groundwater<br />
pH 1 7.7 to 7.9 7.4 to 7.5 7.2 to 7.3<br />
Total Hardness (mg/L) 1 90 to 205 280 to 290 530 to 582<br />
Total Alkalinity (mg/L) 1 50 to 110 235 to 250 260 to 280<br />
TOC (mg/L) 1 2.1 to 4.9 0.3 to 0.6 0.4 to 0.7<br />
Turbidity (ntu) 1 2 to 1,500 5.0
14<br />
The turbidity of well water is typically less than 0.08 nephelometric turbidity units (ntu), which<br />
could allow the Louisville <strong>Water</strong> Company to bypass the coagulation treatment process and use<br />
direct filtration <strong>for</strong> treating water. Reduced turbidity will result in lower coagulant dosages and less<br />
sludge production. The reduction of organic material in the water will provide lower levels of<br />
DBPs in the finished water.<br />
With minimum water temperatures of the Ohio River source near freezing, water main breaks are an<br />
issue during the winter months. By having a more moderate water temperature and using more <strong>RBF</strong><br />
water, this will ultimately result in fewer main breaks. The Louisville <strong>Water</strong> Company also noted the<br />
lack of Geosmin and 2-methylisoborneol in well water during a taste and odor episode that occurred<br />
in 1999, resulting in the ability to eliminate the need to add powdered activated carbon.<br />
Well water, however, has other water-quality issues that need to be addressed in the design of the<br />
system. Data from the Louisville <strong>Water</strong> Company shown in Table 1 indicates the following<br />
treatment issues:<br />
• Increased hardness.<br />
• Presence of radon.<br />
• Low dissolved oxygen.<br />
Additionally, water quality of the aquifer downstream from the Payne Plant has been documented<br />
as having elevated levels of iron (Schafer, 2000). This will be a consideration as the Louisville<br />
<strong>Water</strong> Company proceeds with implementing <strong>RBF</strong> <strong>for</strong> the Crescent Hill <strong>Water</strong> Treatment Plant,<br />
but is not a factor <strong>for</strong> the current design project at the Payne Plant. Because the Payne Plant is<br />
already a softening plant, the increased hardness of <strong>RBF</strong> water will be treated using existing basins.<br />
Radon levels of 180 pCi/L have been noted in the existing collector well. While this is below the<br />
anticipated maximum contaminant level of the future Radon Rule (300 pCi/L), the Louisville<br />
<strong>Water</strong> Company desires to set a goal of 0 pCi/L, because the surface-water source currently<br />
supplying water to its customers does not contain radon. Treatment <strong>for</strong> radon will involve the<br />
installation of an in-line aerator.<br />
Dissolved oxygen from well water is less than 0.1 mg/L. Currently, as water flows through<br />
the basins of the existing treatment plant, oxygen levels are naturally increased up to about<br />
5 mg/L. The same in-line aerator used <strong>for</strong> radon removal can be used to impart higher levels of<br />
oxygen into the water, if needed.<br />
Summary<br />
The results of the subsurface evaluation show that both tunnel options are feasible <strong>for</strong> <strong>RBF</strong>. With<br />
either Option 1 or 3, the collector laterals would be evenly spaced along the entire length of the<br />
collection system, providing <strong>for</strong> more even hydraulic head distribution across the riverbank and,<br />
thus, better overall yield from the installation. With Option 2, the hydraulic head would be<br />
focused at the collector wells, resulting in less than maximum yield from the aquifer system.<br />
Maintenance flexibility would be greater <strong>for</strong> either Option 1 or 3, which would allow each<br />
lateral/vertical collector to be maintained independently of the operation of the remaining<br />
laterals/vertical collectors. Cost was a determining factor in selecting Option 3. In optimizing the<br />
design of the selected <strong>RBF</strong> system, the design engineers considered both water-quantity and waterquality<br />
issues. <strong>Water</strong>-quantity predictions were made using conservative assumptions based on<br />
data from the existing collector well. <strong>Water</strong>-quality issues were identified and appropriate<br />
treatment technologies are being incorporated into the design.
REFERENCES<br />
Deere, D.U., and D.W. Deere (1989). “Rock Quality Designation (RQD) After Twenty Years.” Contract<br />
Report No. 6L-89-1, U.S. Army Corps of Engineers.<br />
Hubbs, S.A. (1981). “Organic reduction — Riverbank Infiltration at Louisville.” Proceedings, American<br />
<strong>Water</strong> Works Association Annual Conference, St. Louis, Missouri.<br />
Kuehn, W., and U. Mueller (2000). “Riverbank Filtration: An Overview.” Journal AWWA, 92(12): 60.<br />
<strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong> (1999). Abstracts, International Riverbank Filtration Conference, November<br />
4-5, 1999, Louisville, Kentucky.<br />
Schafer, D. (2001). Evaluation of Collector Well Production Capacity B.E. Payne <strong>Water</strong> Treatment Plant, David<br />
Schafer & Associates, Inc., Project Report.<br />
Schafer, D. (2000). Hydraulics Analysis of Groundwater Extraction at the B.E. Payne <strong>Water</strong> Treatment Plant,<br />
David Schafer & Associates, Inc., Project Report.<br />
Sontheimer, H. (1980). “Experience with Riverbank Filtration Along the Rhine River.” Journal AWWA,<br />
72(7): 386.<br />
Wang, J., J. Smith, and L. Dooley (1995). “Evaluation of Riverbank Infiltration as a Process <strong>for</strong> Removing<br />
Particles and DBP Precursors.” Proceedings, American <strong>Water</strong> Works Association <strong>Water</strong> Quality Technology<br />
Conference, New Orleans, Louisiana.<br />
Wang, J.Z., S.A. Hubbs, and R. Song (2002). Evaluation of Riverbank Filtration as a Drinking <strong>Water</strong> Treatment<br />
Process, American <strong>Water</strong> Works Association <strong>Research</strong> Foundation Report Number 90922.<br />
Wang, J.Z. (2003). Personal communication.<br />
DAVID HAAS is a Senior Project Manager with Jordan, Jones & Goulding, Inc., an<br />
Atlanta-based consulting firm that offers engineering, management, and planning<br />
services. Haas has over 18 years of experience in municipal water supply, treatment, and<br />
distribution system projects. Currently, he is the Project Manager <strong>for</strong> the Louisville <strong>Water</strong><br />
Company’s 45-million gallons per day riverbank-filtration project at the B.E. Payne <strong>Water</strong><br />
Treatment Plant in Louisville, Kentucky. He is also a contributing author of Riverbank<br />
Filtration: Improving Source-<strong>Water</strong> Quality, jointly published by Kluwer Academic<br />
Publishers and the <strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong> in 2002. Haas received both a B.S. and M.S. in<br />
Environmental Engineering from the University of Louisville. He is a Professional Engineer in the States of<br />
Georgia, Kentucky, and Tennessee.<br />
15
Session 2: Operations<br />
Construction and Maintenance of Wells<br />
<strong>for</strong> Riverbank Filtration<br />
Henry C. Hunt, CPG<br />
Collector Wells International, Inc.<br />
Columbus, Ohio<br />
By description, <strong>RBF</strong> implies that we are designing something that will allow water to be infiltrated<br />
from a surface-water source through riverbank (and river-bottom) deposits. This allows physical<br />
(e.g., suspended) particles to be filtered out of source water in an attempt to “pretreat” raw water<br />
be<strong>for</strong>e it reaches a treatment plant or enters a distribution system with some sort of primary<br />
treatment, such as chlorination.<br />
While the term “<strong>RBF</strong>” is a relatively new term in the United States, well and gallery systems have<br />
been used in international settings to develop water supplies using induced infiltration dating back<br />
to the 1800s. Many well and infiltration systems have relied on <strong>RBF</strong> to provide recharge into<br />
alluvial aquifers to replace groundwater pumped at sites all across the United States, but only<br />
recently have these sites been coined as “<strong>RBF</strong>” sites. <strong>Water</strong>-supply facilities at such sites typically<br />
include vertical wells, infiltration galleries, and radial collector wells.<br />
What makes <strong>RBF</strong> work is that the water level in the aquifer is drawn down by pumping from a<br />
well or gallery system located adjacent to a surface-water source, and water from the surface-water<br />
body is then induced to infiltrate into the aquifer as hydraulic gradients are developed from<br />
pumping. The schematic in Figure 1 shows this general relationship <strong>for</strong> horizontal collector wells<br />
and conventional vertical wells.<br />
Plan<br />
Elevation<br />
Screened<br />
Pipe<br />
River River<br />
Horizontal Vertical<br />
Figure 1. Well systems develop capacity through induced infiltration (Ray et al., 2002a).<br />
Flow<br />
Correspondence should be addressed to:<br />
Henry C. Hunt, CPG<br />
Senior Project Manager/Hydrogeologist<br />
Collector Wells International, Inc.<br />
6360 Huntley Road • Columbus, Ohio 43229 USA<br />
Phone: (614) 888-6263 • Fax: (614) 888-9208 • Email: hchunt@collectorwellsint.com<br />
Flow<br />
17
18<br />
Riverbank Filtration Suitability<br />
A <strong>RBF</strong> system is designed to infiltrate water from an adjacent surface-water source, using streambed<br />
and riverbank deposits to naturally filter out suspended materials from source water. The first (and<br />
obvious) requirement is that the facilities be placed in close proximity to a source of recharge, such<br />
as a river. During the feasibility and siting stages of a project, a number of criteria must be<br />
considered, including:<br />
• Availability of water from a surface-water source that can recharge the aquifer.<br />
• An efficient hydraulic interconnection between the river and aquifer.<br />
• Suitable water quality in the surface-water source.<br />
• Sustainable flow in the river to match anticipated withdrawal rates.<br />
A detailed hydrogeologic investigation is required to verify that aquifer conditions are favorable<br />
<strong>for</strong> considering a <strong>RBF</strong> facility to meet project water demands. The investigation must evaluate the<br />
geology of the aquifer and the interconnection between the river and groundwater in the aquifer<br />
to determine the feasibility of inducing infiltration from the river, evaluate possible well designs,<br />
and determine likely well yields. This type of investigation typically includes exploratory test<br />
drilling, aquifer (pumping) testing, and data analysis to develop the needed in<strong>for</strong>mation <strong>for</strong> each<br />
prospective project site. Based on the results of this investigation, well designs are developed and<br />
alternatives are compared <strong>for</strong> feasibility, yield, and cost. These alternative designs are generally<br />
discussed below.<br />
Vertical Wells<br />
Vertical wells represent the most conventional method <strong>for</strong> developing a groundwater supply in the<br />
country. These wells consist of a vertical borehole that is usually completed with a screen and riser<br />
casing to allow water to enter from the <strong>for</strong>mation and be pumped from the borehole via a pumping<br />
system installed within the riser casing. A diagram of a typical vertical well constructed in an<br />
unconsolidated (e.g., sand and gravel) aquifer is shown in Figure 2. Vertical wells can be used<br />
effectively when small to moderate yields are needed, or when a system is growing slowly over a<br />
longer period of time, such that adding a well to the system every year or so meets growing water<br />
demands. To meet larger demands, a series of vertical wells must be installed, spreading along<br />
riverfront areas or grouped into well-field clusters.<br />
Vertical wells can be constructed using a variety of drilling methods, including mud rotary, reverse<br />
circulation, cable tool, bucket-auger, dual-rotary, and air rotary. The method selected <strong>for</strong> each site<br />
will take into account a combination of well criteria, including well depths, diameters, water-table<br />
elevations, screen requirements, and potential problems caused by geologic conditions<br />
encountered. These wells are constructed by drilling a vertical borehole, and then installing the<br />
desired well riser casing and screen in the borehole. In some cases, an artificial gravel-pack filter<br />
is also added to help prevent the intrusion of sand and silt from the <strong>for</strong>mation during pumping.<br />
Collector Wells<br />
A collector well, also called a horizontal collector or radial well, differs from a vertical well in that<br />
the well screen is installed horizontally into the aquifer <strong>for</strong>mation from a central rein<strong>for</strong>ced<br />
concrete caisson that serves as the wet well pumping station. These wells are constructed by<br />
sinking sections of rein<strong>for</strong>ced concrete (called lifts) into the aquifer adjacent to the river until the<br />
lower portion of the caisson reaches the design elevation <strong>for</strong> installing the well screen. Individual
Land<br />
Surface<br />
lengths of well screens are then projected out into the aquifer in a variety of patterns near the base<br />
of the alluvial aquifer or at another pre-selected horizon within the aquifer. Where <strong>RBF</strong> is desired,<br />
the pattern of lateral well screens is predominantly beneath the river. This allows the “pumping<br />
center” <strong>for</strong> the collector well to be shifted closer to the river, usually in an attempt to increase the<br />
percentage of river water that is infiltrated into the aquifer. This design capability usually permits<br />
the highest percentage of river water to be achieved in a riverbank setting. The ability to install<br />
the well screens horizontally in the aquifer beneath the river permits longer lengths of well screen<br />
to be installed, per site, typically maximizing the yield possible from each well site. It is common<br />
<strong>for</strong> a collector well to produce a yield equivalent to multiple vertical wells from the same well site.<br />
A general schematic of a typical collector well is shown in Figure 3.<br />
Well Selection<br />
<strong>Water</strong><br />
Pump to<br />
<strong>Water</strong><br />
Treatment<br />
Plant<br />
Well<br />
Casing<br />
Pump House<br />
Electric Motor<br />
Approximately<br />
1-m High<br />
Rotating<br />
Shaft<br />
Cone of Depression<br />
Pump Bowl<br />
<strong>for</strong> a Turbine Pump<br />
Well Screen<br />
Gravel Pack<br />
<strong>Water</strong><br />
Figure 2. Typical components of a vertical well (Ray et al., 2002b).<br />
Original <strong>Water</strong> Surface<br />
The hydrogeologic investigation determines the hydraulic characteristics of the aquifer <strong>for</strong>mation<br />
necessary to determine the potential yield possible from either a collector well or a series of<br />
vertical wells. This data allows a comparison to be made of well designs to meet project water<br />
demands, usually considering a single collector well versus a series of vertical wells, with<br />
connecting pipeline, electrical service, access roads, etc., to produce the equivalent yield (it is<br />
common <strong>for</strong> a collector well to produce a yield equal to 5 to 10 vertical wells). As the capital costs<br />
to install the “complete system” are compared, it is often very competitive to consider a collector<br />
well to meet moderate to very large capacities, and more competitive to consider a vertical well<br />
when lower capacities are required (<strong>for</strong> example, when incremental increases in capacity are<br />
projected over a number of years).<br />
River<br />
19
20<br />
Laterals<br />
Alternate Systems<br />
In addition to vertical wells and radial collector wells, infiltration galleries can be used to develop<br />
filtered surface-water supplies through <strong>RBF</strong>. These can include trenching parallel or beneath the<br />
river and installing screened gallery pipes that can deliver filtered surface water to sumps and wet<br />
wells. It is also possible to use horizontal directionally drilled wells to induce infiltration from an<br />
adjacent river. These systems are often installed under low head conditions, such that lower per<br />
foot yields are obtained, requiring longer gallery lengths to meet higher capacity needs. It is often<br />
difficult to per<strong>for</strong>m effective well-screen maintenance on these systems.<br />
Well Maintenance<br />
Pump Shaft<br />
Pump<br />
Pump House<br />
Central<br />
Collection<br />
Caisson<br />
Well Screens<br />
(in Laterals)<br />
Figure 3. Typical components of a radial collector well (Ray et al., 2002b).<br />
Land Surface<br />
It is normal <strong>for</strong> well screens to become plugged with chemical (mineral) precipitates and biological<br />
growths (e.g., iron bacteria) over time in alluvial aquifers. The rate of plugging can be exacerbated<br />
if the screen design creates excessive entrance velocities through the slot openings in the well<br />
screens, so it is important that proper screen design considers the water quality and hydraulic<br />
characteristics of the aquifer to maintain entrance and approach velocities within acceptable<br />
ranges to minimize this rate of plugging and extend the intervals between well cleanings. Since<br />
the lineal footage of well screen in a collector well is longer than <strong>for</strong> a vertical well, entrance<br />
velocities are minimized so that the intervals between required maintenance can be extended.<br />
This typically results in lower O&M costs over the life of the collector well.<br />
Well maintenance can be accomplished using a variety of methods including mechanical,<br />
chemical, or a combination of methods. The most effective method to restore well-screen<br />
openings and well efficiency will vary from well field to well field, and is selected based upon<br />
details of the well construction (e.g., screen design), groundwater quality, past results, the nature<br />
of plugging (mineralogic versus biological), and other factors. The use of an ongoing monitoring<br />
and record-keeping program should allow you to track well per<strong>for</strong>mance, identify operating trends,<br />
and predict optimal times <strong>for</strong> per<strong>for</strong>ming maintenance.
Summary<br />
Well systems can be constructed in alluvial aquifer systems at many locations to induce infiltration<br />
from adjacent surface-water sources to provide a pre-filtered raw-water supply. The design of these<br />
systems should be selected after evaluating site-specific aquifer conditions, water-quality<br />
objectives, and project water demands to ensure that the most effective system is selected. There<br />
are many well systems operating in the United States and overseas, demonstrating that <strong>RBF</strong> can<br />
be used effectively to meet water-quality and capacity objectives.<br />
REFERENCES<br />
Ray, C., T. Grischek, J. Schubert, J. Wang, and T. Speth (2002a). “A perspective of riverbank filtration.”<br />
Journal AWWA, 94(4): 149-160.<br />
Ray, C., G. Melin, and R. Linsky, editors (2002b). Riverbank Filtration: Improving Source-<strong>Water</strong> Quality,<br />
Kluwer Academic Publishers, Dordrecht.<br />
HENRY HUNT is a Senior Project Manager and Hydrogeologist with Collector Wells<br />
International, Inc., which specializes in the design, construction, and rehabilitation of<br />
water-supply systems <strong>for</strong> public drinking water, industrial process, or power plant cooling<br />
water. Hunt has over 25 years experience involving riverbank-filtration systems, ranging<br />
from the inspection and rehabilitation of existing well and infiltration systems to the<br />
design and construction of new water wells and riverbank-filtration facilities. He has<br />
special expertise concerning the inspection, evaluation, siting, and testing of radial<br />
collector wells along many alluvial valleys across the United States and overseas. In addition, he has authored<br />
a number of papers regarding water-supply development, infiltration galleries, radial collector wells, and<br />
seawater collector wells, including several chapters in the 2002 publication Riverbank Filtration: Improving<br />
Source-<strong>Water</strong> Quality and the newly revised American <strong>Water</strong> Works Association Manual 21 on Groundwater.<br />
Hunt received a B.A. in Geology from Lafayette College in Pennsylvania, with a minor in Civil Engineering.<br />
21
Session 2: Operations<br />
Aquifer Storage and Recovery Pretreatment:<br />
Synergies of Bank Filtration, Ozonation,<br />
and Ultraviolet Disinfection<br />
Robert Stanley Cushing, Ph.D., P.E.<br />
Carollo Engineers<br />
Sarasota, Florida<br />
R. David G. Pyne, P.E.<br />
ASR Systems<br />
Gainesville, Florida<br />
David R. Wilkes, P.E.<br />
Jordan, Jones & Goulding<br />
Norcross, Georgia<br />
Introduction<br />
Aquifer Storage and Recovery (ASR) is a technique in which water is stored in subsurface aquifers<br />
during high-supply, low-demand periods <strong>for</strong> use during low-supply, high-demand periods. As<br />
indicated in Figure 1, water is pumped into a confined aquifer, displacing the native water, and<br />
<strong>for</strong>ms a zone in the aquifer comprised primarily of injected water. <strong>Water</strong> is subsequently extracted<br />
from the aquifer when needed to supplement finished water-production capacity.<br />
Native<br />
Ground<br />
<strong>Water</strong><br />
Buffer<br />
Zone<br />
Stored<br />
<strong>Water</strong><br />
ASR Well<br />
Confining Layer Confining Layer<br />
Figure 1. Schematic representation of ASR.<br />
Stored<br />
<strong>Water</strong><br />
Target Storage Volume<br />
Confining Layer<br />
Correspondence should be addressed to:<br />
Robert S. Cushing, Ph.D., P.E.<br />
Partner<br />
Carollo Engineers, P.C.<br />
401 North Cattlemen Road • Suite 306 • Sarasota, Florida 34232 USA<br />
Phone: (941) 371-9832 • Fax: (941) 371-9873 • Email: RCushing@carollo.com<br />
Buffer<br />
Zone<br />
Native<br />
Ground<br />
<strong>Water</strong><br />
23
24<br />
Traditionally, ASR has been used to store potable water to meet seasonal limitations in source-water<br />
supply or to optimize the use of water-treatment capacity. For example, the Peace River Regional <strong>Water</strong><br />
Supply Authority in Florida draws and treats water from the Peace River at a rate limited by minimum<br />
flows and levels. By storing treated water through ASR during high river flow periods, the Authority<br />
has been able to more than double the usable yield from this source, while avoiding adverse impacts to<br />
sensitive downstream ecosystems. Mt. Pleasant <strong>Water</strong> Works in South Carolina supplements their<br />
water supply using reverse-osmosis treatment of brackish groundwater. ASR allows the reverse-osmosis<br />
plant to operate at a relatively constant rate, minimizing the production cost from this facility.<br />
Emerging ASR applications include storing water types other than finished potable water, including<br />
partially treated surface water and reclaimed wastewater, as well as different uses <strong>for</strong> ASR production<br />
water (such as water-treatment plant source-water, irrigation, and environmental applications).<br />
Biological and physiochemical processes during ASR storage can remove certain contaminants,<br />
including THMs and HAAs, leading to ASR applications that provide treatment, as well as storage.<br />
<strong>Water</strong>-quality goals <strong>for</strong> water to be injected into an ASR well vary with application. Generally, the<br />
least restrictive goal is to produce a water with chemical and particle characteristics that prevent<br />
excessive loss of permeability in the injection well aquifer. Some applications require injection water<br />
to meet primary (or public health-related drinking water) standards to protect aquifer quality or to<br />
meet ASR production requirements. Treatment to secondary or aesthetic water-quality standards may<br />
also be required. The most restrictive water-quality goals require treatment <strong>for</strong> trace, unregulated<br />
contaminants (e.g., low levels of endocrine disrupting compounds). Treatment system complexity,<br />
O&M requirements, and cost are all strongly related to source-water quality and finished water-quality<br />
goals, with ASR pretreatment costs ranging well over an order of magnitude, depending on the<br />
particular application.<br />
One general treatment approach <strong>for</strong> meeting ASR pretreatment goals consists of bank filtration<br />
followed by ozonation and/or UV disinfection (Figure 2). This process combination offers the<br />
improved reliability of a multiple-barrier approach while maintaining cost and operational<br />
efficiencies due to the synergistic nature of the individual unit processes. The following discussion<br />
provides a summary of treatability study results, engineering analysis, and cost-estimates <strong>for</strong> this<br />
process configuration.<br />
Lake Okeechobee<br />
Figure 2. Integrated bank filtration/ozonation/UV disinfection process train.<br />
Discussion<br />
Bank Filtration<br />
Ozone Contactor/<br />
Reaction Basin<br />
UV Disinfection<br />
Traditional treatment approaches, such as membrane filtration or conventional treatment,<br />
produce process residuals that generally have substantial cost and operational implications. In<br />
addition, these traditional approaches require substantial chemicals <strong>for</strong> process O&M, an aspect
that is important when considering both reliability and security. A process train comprised of bank<br />
filtration followed by ozonation and/or UV disinfection produces no residuals and has minimal<br />
chemical requirements. This process combination offers a number of benefits, including cost,<br />
flexibility, and multiple barriers to pathogens and other contaminants.<br />
For this process combination, the barriers to pathogens operate by distinctly different mechanisms,<br />
enhancing the robustness of the multiple-barrier approach and providing the first example of<br />
process synergy. Bank filtration provides physical removal and physical/chemical/biological<br />
inactivation of pathogens, ozonation provides chemical disinfection, and UV provides<br />
disinfection through a fundamentally different mechanism.<br />
While bank filtration removes color and taste- and odor-causing compounds, ozonation provides<br />
a means of further improving these water-quality characteristics. Ozonation also increases<br />
UV transmittance, decreasing the capital and operational cost of the UV disinfection process.<br />
Case Study Description and Objectives<br />
As part of the Comprehensive Everglades Restoration Program, the U.S. Army Corps of Engineers<br />
— in conjunction with the South Florida <strong>Water</strong> Management District — have initiated pilot<br />
studies to evaluate potential treatment options <strong>for</strong> an ASR project that includes hundreds of ASR<br />
wells. Treatment systems will be designed to treat surface water prior to injection into groundwater<br />
storage zones, with ultimate capacity approaching 1.5-billion gallons of water per day.<br />
The purpose of the pilot-testing project was to demonstrate surface-water treatment technologies<br />
to meet multiple water-quality goals prior to injection into ASR wells. Pilot-test data and<br />
engineering analysis were used to develop optimized design criteria, as well as capital and O&M<br />
costs <strong>for</strong> full-scale treatment systems. The specific primary objectives of the study were to:<br />
• Determine raw-water quality during the pilot study.<br />
• Collect and analyze unit process and treatment train per<strong>for</strong>mance data.<br />
• Establish process feasibility and design criteria <strong>for</strong> full-scale water treatment systems.<br />
• Determine preliminary capital and operating cost estimates <strong>for</strong> each of the unit processes<br />
considered.<br />
Methodology<br />
Two parallel treatment trains were operated during most of the study:<br />
• Treatment Train 1 — Simulated bank filtration/ozonation.<br />
• Treatment Train 2 — Simulated bank filtration/UV disinfection.<br />
During later stages of the study, two additional trains were tested. Simulated bank filtration was<br />
bypassed and a direct surface-water intake (with wire mesh filtration followed by ozonation and<br />
UV disinfection) was evaluated. Figure 3 illustrates the treatment trains, with dashed lines labeled<br />
“Alternative Treatment Train” showing the supplemental testing.<br />
<strong>Water</strong> was taken from the St. Lucie Canal (eastern shore of Lake Okeechobee) through an intake<br />
header that consisted of slotted polyvinyl chloride pipe. The water was pumped from the intake<br />
pipe through a centrifugal pump and into the simulated bank-filtration system, consisting of a<br />
wedge-wire filter (Parker TruClean filtration system) followed by a gravity sand filter<br />
(10-ft × 10-ft area, with 36-inches of 0.45 to 0.55-millimeter diameter sand). Note that original<br />
25
26<br />
plans <strong>for</strong> a pilot-scale bank-filtration intake were changed by the U.S. Army Corps of Engineers<br />
and South Florida <strong>Water</strong> Management District. Simulated bank-filtration results were analyzed<br />
directly and extrapolated to consider the impacts of full-scale bank-filtration intake.<br />
Results<br />
Decant<br />
<strong>Water</strong> Source Submersible<br />
Pump<br />
Figure 3. Pilot process flow schematic.<br />
Residual<br />
Wedge Wire<br />
Filter<br />
Settling Tank<br />
Residual<br />
Sand Filter<br />
Solids Land<br />
Applied On-Site<br />
Alternate Treatment Train<br />
Holding<br />
Tank<br />
Alternate Treatment Train<br />
Raw-water quality during the study (August 2002 to October 2002) was generally consistent with<br />
historical trends. TOC levels were high, averaging 27 mg/L, with a maximum concentration of<br />
49 mg/L. Correspondingly, true color was also elevated, averaging 96 color units, with a maximum<br />
of 236 color units. Turbidity was variable, ranging from 5 to 57 ntu, with an average of 15 ntu.<br />
The simulated bank-filtration unit process provided particle removal averaging 91 percent and<br />
turbidity removal averaging 78 percent. The removal of dissolved constituents (e.g., TOC and<br />
color) was minimal. As expected, due to the short residence time in the simulated bank-filtration<br />
system (1 to 3 hours), particle removal through the simulated system was less than expected<br />
through a full-scale bank-filtration system, and microbial activity and the associated removal of<br />
organic material was not significant.<br />
Ozone was capable of removing color to meet and exceed the secondary maximum contaminant<br />
level of 15 color units and to reduce UV absorbance. Due to the high ozone doses required to<br />
maintain a residual and rapid decay rates, achieving disinfection credit through ozonation was not<br />
feasible. In addition, with bromide concentrations approaching 300 micrograms per liter (µg/L),<br />
achieving significant disinfection and maintaining bromate concentrations below the maximum<br />
contaminant level of 10 micrograms per milliliter would be infeasible; however, ozone doses used<br />
to reduce color and improve UV transmittance (from a minimum of 13-percent UV transmittance<br />
in raw water to greater than 65-percent UV transmittance in ozonated water).<br />
Due to the synergistic relationship between ozone and UV disinfection, full-scale design criteria<br />
<strong>for</strong> the two processes were analyzed concurrently to minimize water production costs. As ozone<br />
dose goes up, the cost to purchase and operate the ozone system rises; however, due to the increase<br />
in UV transmittance with higher doses, UV-disinfection system costs decrease. An average ozone<br />
dose of approximately 6 mg/L and maximum dose of 11 mg/L produced a UV transmittance of<br />
65 to 72 percent and a minimum production cost (amortized capital, plus O&M). The UV system<br />
UV<br />
O 3<br />
Effluent<br />
to Canal
was designed to deliver a dose of 140 millijoules per square centimeter to meet all disinfection<br />
goals <strong>for</strong> Giardia, Cryptosporidium, and viruses at a design UV transmittance of 60 percent.<br />
The cost of a 5-MGD bank filtration/ozonation/UV-disinfection treatment facility is approximately<br />
$4,759,000 <strong>for</strong> design and construction and $147,000 per year <strong>for</strong> O&M (or $0.20 per 1,000 gallons<br />
of water produced). This translates into a normalized capital cost of $0.95 per gallons per day of<br />
capacity and a production cost of $0.70.<br />
BOB CUSHING has 14 years of experience in applied environmental science and<br />
engineering. Throughout his career, he has coupled fundamental concepts with sound<br />
engineering practices to provide creative, innovative, and enduring solutions to challenges<br />
faced by water and wastewater utilities. Cushing has been responsible <strong>for</strong> numerous<br />
successful treatment facility planning and design projects, as well as studies and programs<br />
<strong>for</strong> improving distribution system water quality. Cushing has practiced nationally,<br />
providing service to a broad cross-section of the industry, from some of the largest and most<br />
visible utilities (e.g., New York City and Washington, D.C.) to very small applications with important and<br />
unique issues (e.g., Oray <strong>National</strong> Fish Hatchery, Utah). He has also been responsible <strong>for</strong> introducing and<br />
applying advanced technologies, most notably ultraviolet disinfection <strong>for</strong> potable water treatment. An<br />
internationally recognized expert in ultraviolet disinfection, he is responsible <strong>for</strong> seminal applied research in<br />
this area, and is project manager and a primary author <strong>for</strong> the United States Environmental Protection<br />
Agency’s Ultraviolet Disinfection Guidance Manual. Cushing received a B.S. in Petroleum Engineering and an<br />
M.S. and Ph.D. in Civil Engineering from the University of Texas at Austin.<br />
27
Session 2: Operations<br />
Evolution from a Conventional Well Field<br />
to a Riverbank-Filtration System<br />
John D. North<br />
Cedar Rapids <strong>Water</strong> Department<br />
Cedar Rapids, Iowa<br />
Douglas J. Schnoebelen, Ph.D.<br />
United States Geological Survey<br />
Iowa City, Iowa<br />
Shelli Grapp<br />
Cedar Rapids <strong>Water</strong> Department<br />
Cedar Rapids, Iowa<br />
Roy E. Hesemann<br />
Cedar Rapids <strong>Water</strong> Department<br />
Cedar Rapids, Iowa<br />
The City of Cedar Rapids is located in east-central Iowa and has a population of 121,000. The<br />
Cedar River alluvial aquifer is the sole source of drinking water <strong>for</strong> the City. Within the last few<br />
years, the Cedar River — which drains a major agricultural watershed — has experienced extended<br />
periods of elevated levels of nitrate and herbicides (notably, atrazine and its degradation<br />
byproducts) that were above the maximum contaminant level <strong>for</strong> drinking water. Fortunately,<br />
during these extended periods of compromised water quality in the river, <strong>RBF</strong> enabled the Cedar<br />
Rapids <strong>Water</strong> Department to obtain and treat adequate quantities of water that met all drinkingwater<br />
standards. The Cedar Rapids <strong>Water</strong> Department and the United States Geological Survey<br />
(USGS) have been collaborating in an ongoing study of the river and its hydraulic interconnection<br />
with the City’s wells. A primary focus of this study is to identify operational and development<br />
strategies that will optimize <strong>RBF</strong> and the water-quality benefits it af<strong>for</strong>ds.<br />
The Cedar River originates in southern Minnesota and generally flows in a southeasterly direction<br />
through east-central Iowa until it discharges to the Iowa River in Louisa County. The Cedar River<br />
drains a major agricultural watershed, which also includes several urban areas. The entire<br />
watershed encompasses 7,800 square miles, in which approximately 6,500 square miles are<br />
upstream from Cedar Rapids’ well fields. Land use in the watershed is predominantly agricultural<br />
(approximately 90 percent), with major crops being corn and soybeans (Kalkhoff et al., 2000;<br />
Schnoebelen and Schulmeyer, 1998; Schulmeyer, 1995). Major urban areas in the watershed<br />
include:<br />
• Albert Lea and Austin in Minnesota.<br />
• Mason City, Clear Lake, Charles City, Forest City, Waverly, Cedar Falls/<strong>Water</strong>loo, and<br />
Cedar Rapids/Marion in Iowa.<br />
Correspondence should be addressed to:<br />
John D. North<br />
<strong>Water</strong> Utility Director, Cedar Rapids <strong>Water</strong> Department<br />
1111 Shaver Road NE<br />
Cedar Rapids, Iowa 52402 USA<br />
Phone: (319) 286-5912 • Fax: (319) 286-5911 • Email: J.North@Cedar-Rapids.org<br />
29
30<br />
The Cedar River and its water quality have been the subject of extensive studies and monitoring<br />
by the Cedar Rapids <strong>Water</strong> Department, USGS, Iowa Geological Survey Bureau, and other local<br />
agencies. These studies have documented several major water-quality issues that affect the river’s<br />
suitability <strong>for</strong> the three designation uses established by the Iowa Department of Natural Resources:<br />
• Primary contact recreation (e.g., swimming).<br />
• Wildlife, fish, and aquatic life.<br />
• Potable water source.<br />
<strong>Water</strong>-quality challenges come from both point and non-point sources and include excessive soil<br />
erosion/sedimentation, elevated levels of nitrate and phosphorous, and fecal coli<strong>for</strong>m bacteria<br />
counts above the acceptable levels <strong>for</strong> recreational swimming. The 57-mile segment of the Cedar<br />
River upstream from the City’s well field to LaPorte City has been placed on the State’s list of<br />
impaired streams due to elevated levels of nitrate and fecal coli<strong>for</strong>m bacteria.<br />
The most daunting challenge <strong>for</strong> the Cedar Rapids <strong>Water</strong> Department is the increasing trend of<br />
elevated levels of nitrate, especially in the spring and early summer. During these periods, monthly<br />
nitrate levels are routinely in the range of 10.0 to 14.7 mg/L as nitrogen (N). The drinking-water<br />
standard <strong>for</strong> nitrate is 10.0 mg/L (as N), and a single exceedance constitutes a violation.<br />
Fortunately, as illustrated in Figure 1, a 2- to 3-mg/L reduction in nitrate levels is generally<br />
accomplished as the water moves from the river to the wells. This natural reduction, combined<br />
with the monitoring and management of individual wells, has enabled the Cedar Rapids <strong>Water</strong><br />
Department to avoid any violations of the drinking-water standard <strong>for</strong> nitrate.<br />
16.0<br />
14.0<br />
12.0<br />
10.0<br />
8.0<br />
6.0<br />
4.0<br />
2.0<br />
0.0<br />
Jan-02<br />
Feb-02<br />
Mar-02<br />
Apr-02<br />
May-02<br />
Jun-02<br />
Monthly Nitrate Results (mg/L as N)<br />
Highest Reported Values<br />
Jul-02<br />
Aug-02<br />
Sep-02<br />
Oct-02<br />
Nov-02<br />
Dec-02<br />
Jan-03<br />
Feb-03<br />
Mar-03<br />
Apr-03<br />
May-03<br />
Jun-03<br />
Jul-03<br />
Cedar River<br />
NW <strong>Water</strong> Plant<br />
MCL Standard<br />
Figure 1. Highest recorded nitrate levels: Cedar River as compared to the well water supplies to Cedar<br />
Rapids’ NW <strong>Water</strong> Plant (Source: Cedar Rapids <strong>Water</strong> Department).<br />
Prior to 1963, the City of Cedar Rapids obtained its water supplies directly from the Cedar River.<br />
River water underwent treatment processes similar to those used today — lime softening,<br />
disinfection by chloramination, filtration, and the addition of fluoride and phosphate <strong>for</strong> corrosion<br />
control. This provided <strong>for</strong> safe drinking water, but failed to avoid the intermittent aesthetic<br />
problems (taste and odor) primarily associated with elevated algae levels in the river. This<br />
prompted the City to construct and convert its water supplies to a series of shallow wells along the<br />
Cedar River.
Today, Cedar Rapids obtains all of its water supplies from a series of shallow wells constructed in<br />
the sand and gravel alluvium along the Cedar River. The City now has four well fields consisting<br />
of 45 vertical wells and four horizontal collector wells. Total current production capacity of the<br />
wells is generally assumed to be approximately 65 MGD, but output will vary depending on river<br />
level and recent operating conditions. Withdrawal rates now average about 36 MGD, with a<br />
maximum demand of 50 MGD. Approximately 80 percent of the water is used <strong>for</strong> grain processing<br />
and other wet industries.<br />
Vertical wells are drilled to the top of the bedrock with depths ranging from 42 to 72 ft and are<br />
located about 30 to 900 ft from the river. The two original horizontal collector wells placed in<br />
service in 1995 consist of a 13-ft diameter center column extending to a depth of approximately<br />
60 ft (about 2 ft above bedrock) and six lateral screens extending about 200 ft from the central<br />
column. The two collector wells placed in service this year are of similar construction, except <strong>for</strong><br />
the use of a 16-ft diameter center column. All of the well laterals were constructed so as not to<br />
extend under the river channel except under high water conditions.<br />
The USGS and other agencies have conducted extensive studies of the Cedar River and the<br />
alluvium that serves as the drinking-water supply <strong>for</strong> Cedar Rapids. These studies have determined<br />
that the Cedar River is generally a “gaining stream” — that is, water will typically move through<br />
alluvium to the river; however, pumping of the wells will reverse the normal flow pattern and<br />
induce the infiltration of water from the Cedar River into the alluvium. The USGS found that<br />
induced filtration from the Cedar River supplies approximately 74 percent of the recharge to the<br />
City’s wells. The balance of the recharge is from the underlying aquifer (21 percent) and the<br />
infiltration of precipitation (5 percent). The USGS also determined that the travel times <strong>for</strong> the<br />
movement of water from the Cedar River to the nearest vertical well ranged from 7 to 17 days<br />
(Schulmeyer, 1995; Schulmeyer and Schnoebelen, 1998).<br />
The movement of water through sand and gravel <strong>for</strong>mations due to pumping is commonly called<br />
induced <strong>RBF</strong>. <strong>RBF</strong> has been widely used in Europe as a pretreatment process <strong>for</strong> almost 100 years,<br />
and there has been recent interest and research concerning its use in North America. The natural<br />
filtration and biological activity that occur during the movement of water through the alluvium<br />
af<strong>for</strong>ds several water-quality benefits. These include a reduction in turbidity and microbiological<br />
improvements with the removal/reduction of viruses, bacteria, and protozoan organisms<br />
(e.g., Giardia and Cryptosporidium), as well as a reduction in the levels of nitrates, herbicides, and<br />
other potential contaminants. USGS studies of Cedar Rapids’ wells have confirmed that, “The<br />
filtering efficiency of the aquifer is equivalent to a 3-log reduction rate or 99.99-percent reduction<br />
in particulates” (Schulmeyer, 1995).<br />
When the Cedar Rapids <strong>Water</strong> Department and USGS initiated their cooperative study in 1992, it<br />
was restricted to a 231-square mile area along the Cedar River that encompasses the City’s four well<br />
fields. Initially, the primary focus was to develop an understanding of the hydraulic interconnection<br />
of the river and wells and the factors that affect the quantity and quality of the water drawn from the<br />
wells. For example, studies regarding new well development would simply attempt to identify sites<br />
that would yield significant water with relatively low levels of iron and manganese. There was very<br />
little consideration given to “<strong>RBF</strong>” and enhancing the water-quality benefits it af<strong>for</strong>ds.<br />
Beginning about 1995, the scope of the study was expanded to include additional physical and<br />
chemical parameters, as well as the addition of biological monitoring (e.g., Microscopic Particulate<br />
Analysis). This was prompted by two primary considerations. The ongoing cooperative study<br />
confirmed that well water was consistently of higher quality than the river; however, the study also<br />
confirmed that the movement of recharge water through alluvium could also readily transport<br />
31
32<br />
contaminants from the river to the wells. Additionally, Iowa regulatory officials issued a preliminary<br />
determination that the newly constructed collector wells were groundwater under the direct<br />
influence of surface water (GWUDI) sources.<br />
To date, major study activities and work products of the USGS/Cedar Rapids <strong>Water</strong> Department<br />
cooperative study include:<br />
• Identification and mapping of contaminant sources in the immediate vicinity of the<br />
City’s wells.<br />
• Development of a regional groundwater model covering 231 square miles, which<br />
simulates groundwater flow under steady-state conditions.<br />
• Construction of a detailed groundwater flow model, which simulates groundwater flow<br />
and well recharge under transient conditions (a computer model is now complete and<br />
being tested).<br />
• Completion of extensive water-quality monitoring (physical and chemical parameters) of<br />
the river and wells.<br />
This research, along with the Cedar Rapids <strong>Water</strong> Department’s Source <strong>Water</strong> Assessment study,<br />
has documented the importance of the river and the potential contaminant threats it poses to our<br />
water supplies. They have also confirmed the protection and other benefits af<strong>for</strong>ded by <strong>RBF</strong>.<br />
Consequently, the cooperative study was recently expanded to include the entire Cedar River<br />
<strong>Water</strong>shed. The Cedar Rapids <strong>Water</strong> Department, USGS, and Iowa Geological Survey Bureau<br />
have partnered to conduct three separate comprehensive synoptic studies of water quality of<br />
samples at 64 locations throughout the watershed. A dye tracer/time-of-travel study has been<br />
completed on the lower main stem of the Cedar River, with plans <strong>for</strong> more dye tracing on other<br />
sections at low flow conditions, as well as possible Lagrangian sampling. In Lagrangian sampling,<br />
the same mass of water is sampled as it moves downstream. The objective of this study will be to<br />
validate time-of-travel models and to determine the fate of nitrogen compounds as they are<br />
transported down the river (D.J. Schnoebelen, 2002).<br />
In summary, the Cedar Rapids <strong>Water</strong> Department has attempted to develop a better understanding<br />
of the river and its wells, which would allow the implementation of management and<br />
protection programs <strong>for</strong> its drinking-water supplies. Although we did not fully understand its role<br />
until just recently, <strong>RBF</strong> is a valuable pretreatment process and is an integral component of our<br />
multi-barrier strategy <strong>for</strong> protecting water quality. It has enabled the Cedar Rapids <strong>Water</strong><br />
Department to avoid any violations of the drinking-water standard <strong>for</strong> nitrate that would<br />
necessitate the construction and operation of costly nitrate removal facilities.<br />
Without <strong>RBF</strong>, the Cedar Rapids <strong>Water</strong> Department would not have been able to supply quality<br />
water that meets all drinking-water standards while maintaining the competitive rates required by<br />
its major customers, the local industries. Although very beneficial, it must be acknowledged that<br />
<strong>RBF</strong> does not completely remove contaminants, as evidenced by the presence of nitrates and<br />
herbicides in Cedar Rapids’ well supplies. <strong>RBF</strong>, along with watershed protection programs, must<br />
be a key element of Cedar Rapids’ multi-barrier approach to the protection of its water supplies<br />
and public health.
REFERENCES AND ACKNOWLEDGEMENTS<br />
The following references and publications were used in the research and preparation of this report<br />
(though not all are cited in the text):<br />
<strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong><br />
• Ray, C., G. Melin, and R.B. Linsky, editors (2002). Riverbank Filtration: Improving Source-<strong>Water</strong><br />
Quality, Kluwer Academic Publishers, Dordrecht.<br />
United States Geological Survey<br />
• Becher, K.D., S.J. Kalkhoff, D.J. Schnoebelen, K.K. Barnes, and V.E. Miller (2001). <strong>Water</strong>-Quality<br />
Assessment of the Eastern Iowa Basins – Nitrogen, Phosphorous, Suspended Sediment and Organic<br />
Carbon in Surface <strong>Water</strong>, 1996-98, U.S. Geological Survey <strong>Water</strong>-Resources Investigations Report<br />
01-4175.<br />
• Kalkhoff, S.J., K.K. Barnes, K.D. Becher, M.E. Savoca, D.J. Schnoebelen, E.M. Sadorf, S.D. Porter,<br />
and D.J. Sullivan (2000). <strong>Water</strong> Quality in the Eastern Iowa Basins, Iowa and Minnesota, 1996-98,<br />
U.S. Geological Survey Circular 1210.<br />
• Schnoebelen, D.J. (2002). Written correspondence.<br />
• Schnoebelen, D.J., D.E. Christiansen, and K.D. Becher (2002). “City of Cedar Rapids: Source<br />
<strong>Water</strong> Assessment of Public Drinking-<strong>Water</strong> Supplied from Ground-<strong>Water</strong>.”<br />
• Schulmeyer, P.M. (1995). Effect of the Cedar River on the Quality of the Ground-<strong>Water</strong> Supply <strong>for</strong> City<br />
Rapids, Iowa, U.S. Geological Survey <strong>Water</strong>-Resources Investigations Report 94-4211.<br />
• Schulmeyer, P.M., and D.J. Schnoebelen (1998). Hydrogeology and <strong>Water</strong> Quality in the Cedar Rapids<br />
Area, Iowa, 1992-1996, U.S. Geological Survey <strong>Water</strong>-Resources Investigations Report 97-4261.<br />
Iowa Department of Natural Resources<br />
• “Progress Report One — Cedar River Assessment Survey,” June 2001.<br />
• “Source <strong>Water</strong> Assessment and Protection Program and Implementation Strategy <strong>for</strong> the State of<br />
Iowa,” March 2000.<br />
Iowa Department of Natural Resources – Geological Survey Bureau<br />
• Seigely, L.S., and M.P. Skopec. Written reports of three comprehensive synoptic monitoring programs<br />
conducted at 62 sites on the Cedar River on August 26, 2000; June 2, 2001; and June 7, 2003.<br />
Cedar Rapids <strong>Water</strong> Department<br />
• Grapp, S.J., J.D. North, and R.E. Hesemann (2002). “City of Cedar Rapids Source <strong>Water</strong> Assessment.”<br />
JOHN NORTH has 30 years of experience in water-quality monitoring and in the<br />
operation and maintenance of water systems. For the past 11 years, he has worked <strong>for</strong> the<br />
City of Cedar Rapids in Iowa, which draws its water supplies from a series of shallow<br />
alluvial wells located along the Cedar River and drains a major agricultural watershed. At<br />
present, he is the <strong>Water</strong> Utility Director <strong>for</strong> the City of Cedar Rapids <strong>Water</strong> Department.<br />
Under his direction, the Cedar Rapids <strong>Water</strong> Department has conducted several<br />
comprehensive water-quality research projects and other studies designed to ensure the<br />
continued availability of an ample supply of safe drinking water <strong>for</strong> Cedar Rapids. Since 1993, the utility has<br />
partnered with the United States Geological Survey in a comprehensive ongoing study of the City’s well<br />
fields and their hydraulic interconnection with the Cedar River. A major focus of the study has been to<br />
optimize the benefits of <strong>RBF</strong>. North received a B.S. in Chemistry from Loras College. He has dualcertification<br />
as a Grade IV water and wastewater plant operator.<br />
33
Session 3: Hydraulic Aspects<br />
Groundwater Flow and <strong>Water</strong> Quality – A Flowpath<br />
Study in the Seminole Well Field, Cedar Rapids, Iowa<br />
Douglas J. Schnoebelen, Ph.D.<br />
United States Geological Survey<br />
Iowa City, Iowa<br />
Michael J. Turco<br />
United States Geological Survey<br />
Lincoln, Nebraska<br />
John D. North<br />
Cedar Rapids <strong>Water</strong> Department<br />
Cedar Rapids, Iowa<br />
In Iowa, alluvial aquifers near major rivers are a source of water <strong>for</strong> many communities. The City<br />
of Cedar Rapids withdraws water from wells completed in the Cedar River alluvium, a shallow<br />
alluvial aquifer adjacent to the Cedar River. The City of Cedar Rapids is located within Linn County<br />
in east-central Iowa, and water <strong>for</strong> the City is supplied by four well fields (East, Northwest,<br />
Seminole, and West well fields) along the Cedar River. The City has a population of about 121,000,<br />
and several large industries are major water users. Currently, per capita water usage in the City is<br />
nearly three times the national average. The City is committed to providing both a high quality<br />
and quantity of water to its customers. The USGS and Cedar Rapids <strong>Water</strong> Department have<br />
been working together in an ongoing research program to better understand water quality and flow<br />
in the Cedar River and alluvial well fields. Work has been done on both a basin and well-field<br />
approach and has involved dye tracing/time-of-travel studies on the Cedar River, water-quality<br />
sampling, geochemical modeling, and groundwater-flow modeling.<br />
The effect of land use in the Cedar River Basin on both surface-water and groundwater quality is<br />
an important issue. The Cedar River Basin upstream from Cedar Rapids is approximately<br />
6,500 square miles. Upstream land use in the Cedar River Basin is over 90-percent agriculture.<br />
Corn and soybeans are the major crops. Livestock raised in the area include beef and dairy cattle,<br />
as well as hogs. Runoff from agriculture is of concern, particularly during the spring and early<br />
summer when many chemicals are applied to cropland. Triazine and acetanilide herbicides are<br />
commonly applied in the Cedar River Basin, and these herbicides are water soluble and can be<br />
transported to streams and infiltrate to groundwater. In addition, several studies in eastern Iowa<br />
have identified nutrients as a major contaminant that has impaired water quality (Goolsby and<br />
Battaglin, 1993; Hallberg et al., 1996; Schnoebelen et al., 1999; Kalkhoff et al., 2000). In general,<br />
the majority of nitrogen inputs in the Cedar River Basin are from chemical fertilizers and animal<br />
manure (Becher et al., 2000). High nitrate levels (greater than 10.0 mg/L) in the Cedar River are<br />
of particular concern to municipal water operators. The lower Cedar River is listed on the Iowa<br />
total maximum daily load list <strong>for</strong> nitrate upstream of Cedar Rapids, Iowa. The Cedar River is the<br />
Correspondence should be addressed to:<br />
Douglas J. Schnoebelen, Ph.D.<br />
<strong>Research</strong> Hydrologist/<strong>Water</strong>-Quality Specialist<br />
United States Geological Survey<br />
Federal Building, Room 269 • 400 South Clinton • Iowa City, Iowa 52240 USA<br />
Phone: (319) 358-3617 • Fax: (319) 358-3606 • Email: djschnoe@usgs.gov<br />
35
36<br />
source of most nitrate detected in the Cedar River alluvial aquifer because of induced infiltration<br />
from the river due to pumping (Schulmeyer and Schnoebelen, 1998; Boyd, 2000).<br />
An unconsolidated surficial layer of glacial till, loess, and Cedar River alluvium (alluvial aquifer)<br />
overlies carbonate bedrock of Devonian and Silurian age (bedrock aquifer) in the study area. The<br />
alluvial aquifer typically consists of a sequence of coarse sand and gravel at the base, grading<br />
upwards to finer sand, silt, and clay near the surface. The sand and gravel contain carbonate, shale,<br />
and ferro-magnesium rich rock fragments. The thickness of the alluvial aquifer ranges from about<br />
2 to 30 m. The alluvial aquifer is recharged by the infiltration of water from the Cedar River,<br />
precipitation, and seepage from the underlying bedrock and adjacent hydrogeologic units. In areas<br />
under the influence of municipal pumping, groundwater flow is from the Cedar River toward the<br />
well fields; in areas outside the influence of municipal pumping, groundwater flow is toward the<br />
Cedar River. Results from a regional groundwater flow model indicated that approximately<br />
74 percent of the water pumped from the alluvial well fields is recharged from the Cedar River,<br />
approximately 21 percent is recharged from adjacent underlying hydrogeologic units, and approximately<br />
5 percent of the water is from infiltrating precipitation (Schulmeyer and Schnoebelen,<br />
1998). Currently, a more detailed groundwater model in the study area indicates that, in some<br />
places, up to 90 percent of the water pumped from the alluvial well fields is recharged from the<br />
Cedar River.<br />
The water quality in the alluvial aquifer within the well field has been characterized with samples<br />
collected from both monitoring and municipal wells at various times since 1992 (Boyd, 2000;<br />
Schulmeyer and Schnoebelen, 1998; Schnoebelen and Schulmeyer, 1996). Calcium, magnesium,<br />
and bicarbonate are the dominant ions. In addition, nitrate, sulfate, silica, iron, and manganese are<br />
present in significant concentrations in certain wells or at certain times of the year. Previous work in<br />
the Seminole well field indicated some detections of herbicides and their degradates (breakdown<br />
products) in shallow monitoring wells (3.8- to 6-m deep) completed in the alluvium as water moved<br />
from the river into the alluvial aquifer (Boyd, 2000). Atrazine was the most commonly detected<br />
herbicide in this study. Acetochlor, cyanazine, and metolachlor were also detected, but at smaller<br />
concentrations than atrazine. Acetanilide degradates were detected at greater frequencies and at<br />
greater concentrations than their corresponding parent compounds. Fewer numbers of detections of<br />
herbicide compounds were found in wells completed deeper in the alluvium.<br />
The infiltration of water with large nitrate concentrations into the alluvial aquifer from the Cedar<br />
River affects groundwater quality. Recent research was conducted along a flowpath to study <strong>RBF</strong><br />
through a natural wetland area. Groundwater modeling helped locate the flowpath study. The<br />
study examined the role of a natural wetland in reducing nitrate concentrations as water moves<br />
from the Cedar River. A real challenge <strong>for</strong> the Cedar Rapids <strong>Water</strong> Department is the increasing<br />
trend of nitrate concentrations in the Cedar River. Nitrate concentrations in the Cedar River<br />
during the spring are often more than 10 mg/L and can reach 20 mg/L. A 2- to 3-mg/L reduction<br />
in nitrate often occurs as water moves from the river to the well, but in some wells, this may not<br />
reduce nitrate concentrations below the 10.0-mg/L maximum contaminant level. Sampling in<br />
wells along a flowpath occurred quarterly over a period of about 4 years. A comparison of water<br />
chemistry was made from water analyses from:<br />
• The river.<br />
• A monitoring well upgradient of the wetland area and river.<br />
• Wells in the wetland area.<br />
• Wells between the wetland area and river.
In addition, a comparison of water-chemistry data from a municipal well located near the wetland<br />
area and one located nearest the river were compared in terms of water chemistry from previous<br />
sampling work (Schulmeyer and Schnoebelen, 1998). Results show that nitrate concentrations<br />
were 4 to 6 times lower in samples from monitoring wells completed in the wetland area than in<br />
the Cedar River or groundwater in the upland area; however, iron and manganese concentrations<br />
in samples from the monitoring wells in the wetland areas were an order of magnitude higher<br />
when compared to the river or upland well. <strong>Water</strong> samples from the wells and the Cedar River<br />
generally displayed similar trends (high in the spring and low in the fall), while iron and manganese<br />
concentrations were more variable.<br />
As water moves from the river towards the monitoring wells, microorganisms obtain energy <strong>for</strong><br />
metabolic processes by catalyzing the oxidation of organic matter with a progressive series of<br />
reducing reactions (Stumm and Morgan, 1981). Nitrate can be reduced to elemental nitrogen (N 2)<br />
by denitrification (Equation 1) or to ammonium (NH 4+) by reduction (Equation 2). Since<br />
ammonium was only detected in small quantities (less than 0.80 mg/L), denitrification most likely<br />
is the predominant process.<br />
4NO 3 – + 5CH2O + 4H + = 2N 2(g) + 5CO 2 + 7H 2O (Equation 1)<br />
NO 3 – + 2CH2O + 2H + = NH 4 + + H2O + 2CO 2<br />
(Equation 2)<br />
Reduction then proceeds from nitrate (NO 3 – ) to Mn +4 , Fe +3 , SO4 –2 , CO2, and N 2. The reduced<br />
<strong>for</strong>ms of iron (Fe II) and manganese (Mn II) are more soluble in water and are more mobile than<br />
oxidized <strong>for</strong>ms (Hem, 1985) and, under anoxic conditions, are stable. As nitrate in groundwater<br />
is depleted, iron and manganese reduction begins. The reduction of Fe +3 to Fe +2 and Mn +4 to<br />
Mn +2 from aquifer grain coatings can cause large concentrations of these ions in groundwater.<br />
Ferrihydrite and manganite (MnOOH) occurring as oxyhydroxide coatings on clay and silt<br />
particles are the most likely oxidized <strong>for</strong>ms of iron (Fe +3 ) and manganese (Mn +3 and Mn +4 ) in the<br />
alluvial aquifer. Oxidized <strong>for</strong>ms of iron and manganese might occur in the aquifer as crystalline<br />
minerals, such as hematite (Fe 2O 3) and hausmannite (Mn 3O 4). Iron and manganese may<br />
co-precipitate with carbonate minerals to cause well fouling.<br />
<strong>Research</strong> in the Seminole well field indicates that the location of a well in or near natural wetland<br />
areas may benefit from the natural reduction of nitrate concentrations, with the disadvantage of<br />
increased iron and manganese concentrations. Future expansions of the well fields may take<br />
advantage of natural wetland areas to help reduce nitrate concentrations. In Iowa, most wetlands<br />
have been drained, but alluvial wetlands associated with bottomland <strong>for</strong>ested and oxbow lake<br />
areas may persist as they are subject to periodic flooding and are often not suitable <strong>for</strong> sustained<br />
agriculture.<br />
REFERENCES<br />
Becher, K.D., D.J. Schnoebelen, and K.K.B. Akers (2000). “Nutrients discharged to the Mississippi River<br />
from Eastern Iowa <strong>Water</strong>sheds, 1996-97.” Journal AWWA, 36(1): 161-173.<br />
Boyd, R.A. (2000). “Herbicides and herbicide degradates in shallow groundwater and the Cedar River near<br />
a municipal well field, Cedar Rapids, Iowa.” The Science of the Total Environment, 241-253.<br />
Goolsby, D.A., and W.A. Battaglin (1993). “Occurrence, distribution, and transport of agricultural chemicals<br />
in surface water of the Midwestern United States.” Selected papers on agricultural chemicals in water resources of<br />
the midecontinental United States, D.A. Goolsby, L.L. Boyer, and G.E. Mallard, compilers, U.S. Geological<br />
Survey Open-File Report 93-418, p. 1-25.<br />
37
38<br />
Hallberg, G.R., D.G. Riley, J.R. Kantamneni, P.J. Weyer, and R.D. Kelley (1996). Assessment of Iowa safe<br />
drinking water act monitoring data, 1988-1995, Iowa City, University of Iowa Hygienic Laboratory <strong>Research</strong><br />
Report 97-1, 132 p.<br />
Hem, J.D. (1985). Study and interpretation of the chemical characteristics of natural water, third edition, U.S.<br />
Geological Survey <strong>Water</strong>-Supply Paper 2254, p. 264.<br />
Kalkhoff, S.J., K.K. Barnes, K.D. Becher, M.E. Savoca, D.J. Schnoebelen, E.M. Sadorf, S.D. Porter, and D.J.<br />
Sullivan (2000). <strong>Water</strong> Quality in the Eastern Iowa Basins, Iowa and Minnesota, 1996-98, U.S. Geological<br />
Survey Circular 1210, 37 p.<br />
Schnoebelen, D.J., K.D. Becher, M.W. Bobier, and T. Wilton (1999). Selected nutrients and pesticides in streams<br />
of the Eastern Iowa Basins, 1970-95, U.S. Geological Survey <strong>Water</strong>-Resources Investigations Report 99-4028,<br />
105 p.<br />
Schnoebelen, D.J., and P.M. Schulmeyer (1996). Selected hydrogeologic data in the Cedar Rapids area, Benton<br />
and Linn Counties, Iowa, October 1992 through March 1996, U.S. Geological Survey Open-File Report<br />
96-471, 172 p.<br />
Schulmeyer, P.M., and D.J. Schnoebelen (1998). Hydrogeology and water-quality in the Cedar Rapids Area,<br />
Iowa, 1992-96, U.S. Geological Survey <strong>Water</strong> Resources Investigations Report 97-4261, 77 p.<br />
Stumm W., and J.J. Morgan (1981). Aquatic Chemistry — An introduction emphasizing chemical equilbria in<br />
natural waters, second edition, Wiley-Interscience Publisers, New York, 780 p.<br />
DOUG SCHNOEBELEN is a <strong>Research</strong> Hydrologist with the United States Geological<br />
Survey and has worked on a variety of groundwater and surface-water projects over the last<br />
13 years. He has served as the groundwater specialist on the White River <strong>National</strong> <strong>Water</strong><br />
Quality Assessment project in Indiana and as the surface-water specialist <strong>for</strong> the Eastern<br />
Iowa Basins <strong>National</strong> <strong>Water</strong> Quality Assessment project in Iowa. In addition, he has been<br />
the Iowa District <strong>Water</strong>-Quality Specialist since 1994, and is an Adjunct Professor in the<br />
Geoscience Department at the University of Iowa. His research has focused on the use of<br />
isotopes and bore hole geophysics in groundwater and, more recently, on the fate and transport of agricultural<br />
chemicals in surface water and groundwater across eastern Iowa. In particular, he has focused on the fate and<br />
transport of several pesticide degradate compounds. He has been involved with ongoing research in riverbank<br />
filtration and geochemical modeling in the Cedar Rapids well field since 1999. Schnoebelen received a B.S.<br />
in Geology from the University of Iowa, an M.S. in Geology from the University of Tennessee, and a Ph.D.<br />
in Geology, with a minor in Environmental Science, from Indiana University.
Session 3: Hydraulic Aspects<br />
The Use of Aquifer Testing and Groundwater Modeling<br />
to Evaluate Changes in Aquifer/River Hydraulics at the<br />
Louisville <strong>Water</strong> Company<br />
David C. Schafer<br />
David Schafer & Associates<br />
Stillwater, Minnesota<br />
In 1999, the Louisville <strong>Water</strong> Company in Louisville, Kentucky, constructed a 20-MGD radial<br />
collector well as part of a pilot study to evaluate using <strong>RBF</strong> to reduce treatment costs and to ensure<br />
high-quality water production. The water company currently withdraws up to a total of 240 MGD<br />
of surface water from the Ohio River at the Zorn Pumping Station and B.E. Payne <strong>Water</strong> Treatment<br />
Plant (both in Louisville, Kentucky), and is exploring ways to further improve water-supply<br />
quality.<br />
A primary objective of the pilot project was to identify processes that would meet or exceed<br />
expected water-quality regulations. Considerations included:<br />
• Removing synthetic organics.<br />
• Reducing turbidity.<br />
• Eliminating taste and odor during summer months.<br />
• Attenuating water temperature extremes in the distribution system.<br />
Although several options were evaluated (such as GAC and membrane filtration), <strong>RBF</strong> was the<br />
only one that addressed all of the Louisville <strong>Water</strong> Company’s concerns.<br />
In addition to evaluating water quality, a secondary objective was to assess aquifer/river hydraulics<br />
and monitor the hydraulic per<strong>for</strong>mance over time to detect whether or not changes in capacity<br />
occurred. Although it was not expected that the well structure would lose hydraulic efficiency,<br />
there was some concern that riverbed materials adjacent to the well could clog or compact in<br />
response to groundwater withdrawal.<br />
The collector well was constructed about 120 ft from the shore of the Ohio River at the B.E. Payne<br />
<strong>Water</strong> Treatment Plant site. It incorporated a 16-ft inside diameter concrete caisson extending to<br />
a depth of about 105-ft below land surface. The caisson penetrated the glacial sand and gravel<br />
aquifer at the site, bottoming out on the underlying shale and limestone bedrock. Seven 12-inch<br />
diameter stainless steel horizontal well-screen laterals were installed near the base of the sand and<br />
gravel aquifer — four laterals 240 ft in length and three 200 ft in length. The four longer laterals<br />
were oriented toward the Ohio River, while the other three shorter laterals extended 1) upriver,<br />
2) downriver, and 3) away from the river.<br />
Correspondence should be addressed to:<br />
David C. Schafer<br />
President<br />
David Schafer & Associates<br />
9955 North 101st Street • Stillwater, Minnesota 55082 USA<br />
Phone: (651) 762-8281 • Fax: (651) 762-8335 • Email: dschafer@prodigy.net<br />
39
40<br />
Two 10-MGD capacity pumps were installed to produce water from the collector well. A variablefrequency<br />
driver was used to power one of the two pump motors to allow the pumping plant to<br />
readily adapt to changing flow and head requirements.<br />
In addition to the collector well, about a dozen on-land piezometers were installed around the<br />
collector well at distances up to 1,200-ft away. Also, a nested piezometer was installed about<br />
80-ft offshore, adjacent to the well site. The piezometers were used to monitor aquifer water levels<br />
during the testing and operation of the collector well to provide the data needed to quantify<br />
aquifer characteristics.<br />
Following well construction, a 70-day constant-rate pumping test was conducted at a pumping rate<br />
of 19.4 MGD. <strong>Water</strong> levels were monitored in both the collector well and observation wells. The<br />
maximum drawdown observed in the collector well during the test was about 27 ft. The pumping<br />
test data were used to quantify hydraulic properties — transmissivity and river leakance — and to<br />
identify baseline per<strong>for</strong>mance characteristics.<br />
The data were analyzed by constructing a groundwater flow model of the site and adjusting model<br />
input parameters until the model faithfully replicated water levels observed during the pumping<br />
test. The USGS groundwater flow code, MODFLOW, was selected <strong>for</strong> the modeling study. The<br />
aquifer and Ohio River were represented in a model grid covering an area 4,000 ft × 14,000 ft. The<br />
model domain was subdivided into 119 rows, 53 columns, and 8 model layers. The use of a large<br />
number of model layers permitted the realistic simulation of vertical resistance to groundwater<br />
flow — both water exiting the river and entering collector well laterals. Results of the pumping<br />
test and modeling ef<strong>for</strong>t revealed an aquifer transmissivity of 204,000 gallons per day per foot and<br />
a river leakance of 2.35 inverse days at the prevailing groundwater and surface-water temperatures.<br />
The collector well was used as the pumping well in five subsequent pumping tests conducted over<br />
a 3-year period from 1999 to 2002. Data from each pumping test were analyzed using groundwater<br />
modeling to update the computed values of transmissivity and leakance. The prevailing groundwater<br />
and surface-water temperatures were different <strong>for</strong> each test that was conducted. Because water<br />
temperature affects flow characteristics, it was necessary to correct the results of each test <strong>for</strong><br />
temperature effects so that per<strong>for</strong>mance could be compared from one test to another.<br />
In the model calculations, a simplifying assumption was made that leakance was constant<br />
everywhere <strong>for</strong> any given model run. This simplification allowed a relative comparison of one test<br />
with another. It is known that leakance declines in a non-uni<strong>for</strong>m manner, with greater reduction<br />
near the pumped well and less reduction away from the well; however, the exact nature of<br />
leakance distribution is not known, and trying to estimate it with the available data would have<br />
been speculative. There<strong>for</strong>e, the assumption of a single leakance value was made. The resulting<br />
leakance value can be thought of as an “effective leakance” <strong>for</strong> comparison purposes.<br />
Throughout the period of testing, transmissivity remained constant while effective leakance<br />
declined steadily. Over time, the rate of leakance decline diminished and finally stabilized based on<br />
the last pumping test conducted in July 2002, after nearly 3 years of well operation. Measured<br />
leakance values, corrected <strong>for</strong> water temperature, are listed in Table 1.<br />
River scour associated with a bank full-flood event in Spring 2002 may have contributed to the<br />
apparent stabilization (actually, a slight increase in leakance). The average measured leakance in<br />
July 2002, corrected to the same water temperature as the original pumping test, was about<br />
0.18 inverse days, more than an order of magnitude lower than the original leakance value.
Table 1. Measured Leakance Values<br />
Date of Test Effective Leakance<br />
Data Acquisition (inverse days)<br />
October 1999 2.35<br />
March 2000 0.72<br />
October 2000 0.25<br />
April 2001 0.20<br />
September 2001 0.15<br />
July 2002 0.18<br />
The corresponding well capacity had declined by about a third in response to reduced river<br />
leakance.<br />
The reduction in river leakance is being evaluated on an ongoing basis, but is believed to be<br />
caused by clogging/compaction of riverbed sediments in response to collector well operation.<br />
Regular pumping tests and leakance evaluation will be continued in the future to further monitor<br />
the effects of riverbed clogging and periodic river scour associated with flood events on the Ohio<br />
River. The Louisville <strong>Water</strong> Company continues to do research in the area of riverbed clogging<br />
and leakance reduction.<br />
Hydrologist DAVID SCHAFER has over 30 years of experience in the groundwater industry.<br />
He has designed hundreds of high-capacity water supply wells, has analyzed hundreds of<br />
pumping tests, and is consulted regularly on well development and rehabilitation procedures.<br />
In addition, he has done extensive groundwater modeling using numerical models, analytic<br />
element models, and proprietary analytical models that he has developed. At present, he is<br />
President of his own company, David Schafer & Associates, which he founded in 1999.<br />
Schafer has published numerous articles on dewatering, well hydraulics, well development,<br />
and well rehabilitation, and contributed a portion of the well hydraulics section of Groundwater and Wells,<br />
Second Edition. Schafer received both a B.S. in Mathematics and an M.S. in Computer Science from the<br />
University of Minnesota.<br />
41
Session 3: Hydraulic Aspects<br />
Plugging in Riverbank-Filtration Systems:<br />
Evaluating Yield-Limiting Factors<br />
Stephen A. Hubbs, P.E.<br />
Louisville <strong>Water</strong> Company<br />
Louisville, Kentucky<br />
The <strong>RBF</strong> process involves the passage of river water containing dissolved and suspended solids<br />
through the sands and gravels of a connected aquifer. In this process, the removal of suspended solids<br />
involves dynamics that are similar to slow sand filtration, with both biological filtration and physical<br />
filtration acting to remove particles. If the entrained particles are not removed, the hydraulic<br />
conductivity of the riverbed will decrease due to particles plugging the pores of the aquifer.<br />
Work done at the Louisville <strong>Water</strong> Company in Louisville, Kentucky, and elsewhere have indicated<br />
that the bulk of particle removal takes place in the first few feet of the aquifer, which is consistent<br />
with behavior in a slow sand filter. The dynamics of how a riverbed restores its filtration capacity<br />
are not fully understood. Anecdotal in<strong>for</strong>mation from those operating large-capacity well fields<br />
indicates that these well fields require 3 to 5 years to “settle in” to their sustainable yield, which<br />
is usually within 50 to 75 percent of the initial capacity of the well field. Recent research at the<br />
Louisville <strong>Water</strong> Company has focused on developing a better understanding of the dynamics of<br />
riverbed plugging and predicting sustainable capacity from high-capacity <strong>RBF</strong> systems.<br />
Predicting Sustainable Yield<br />
The 20-MGD horizontal collector well at Louisville was constructed in 1999, and data on various<br />
parameters have been collected since that time. These data have been analyzed by traditional<br />
modeling techniques (Schafer, 2003) and, more recently, by regression techniques, with the<br />
specific interest of gaining more in<strong>for</strong>mation about the process of riverbed plugging.<br />
Data indicate a cyclic pattern in specific yield, assumed to be a function of water viscosity as<br />
impacted by water temperature. Data also indicate a decreasing trend with time <strong>for</strong> specific yield.<br />
The possible causes of this decrease include:<br />
• Plugging of the riverbed-aquifer interface.<br />
• Decreased conductance of the bulk of the aquifer.<br />
• Decreased conductance in the area of the well screen.<br />
Various parameters from the entire 4-year database at the Louisville <strong>Water</strong> Company were<br />
subjected to regression analysis, with specific yield as the dependent variable. Variables included:<br />
• River stage.<br />
• <strong>Water</strong> elevation in the collector caisson.<br />
Correspondence should be addressed to:<br />
Stephen A. Hubbs, P.E.<br />
Vice President, New Technology<br />
Louisville <strong>Water</strong> Company<br />
550 South Third Street • Louisville, Kentucky 40202 USA<br />
Phone: (502) 569-3675 • Fax: (502) 569-0813 • Email: SHubbs@lwcky.com<br />
43
44<br />
• River and caisson water temperature.<br />
• Time (linear), the square root of time, and natural log of time.<br />
The influence of the driving head from the river was factored into the specific yield value by<br />
subtracting river stage from water level in the caisson. This eliminated the need to carry a term<br />
<strong>for</strong> river level in the analysis. Thus, the term <strong>for</strong> specific yield was defined from the data as:<br />
(Well output in MGD)/(River stage – Caisson water level).<br />
The square root of time was selected as a parameter based on the hypothesis that:<br />
• Plugging of the aquifer was the result of the deposition of suspended sediment at the<br />
aquifer-riverbed interface.<br />
• The rate of this plugging process would decrease as the recharge area extended farther<br />
into the river.<br />
• A balance between plugging and riverbed scouring would eventually be established.<br />
Subsequent analyses indicated that a better regression fit was realized when the natural log of time<br />
was used as a parameter, as opposed to linear time or the square root of time.<br />
When the entire data set was analyzed, it was apparent that typical trends were skewed during<br />
times of flooding. The database was modified to exclude flooding periods and then re-analyzed.<br />
Results of this analysis are presented in Figure 1. The regression model provides a good fit with<br />
actual data, but varies the most at the initiation of pumping and during the period of Weeks 160<br />
to 180, following a significant flooding event.<br />
Specific Yield (normalized to mean of 0.491 MGD/ft)<br />
0.500<br />
0.400<br />
0.300<br />
0.200<br />
0.100<br />
0.000<br />
-0.100<br />
-0.200<br />
-0.300<br />
Ln = Natural log.<br />
Figure 1. Regression with Natural Log of Time (Cleaned Data)<br />
SpYield (MGD/ft) = (-0.013) + .0045 (Well Temp. F) + .0018 (River Temp F) - .095 (Ln (Time week))<br />
0 20 40 60 80 100<br />
Period (week)<br />
120 140 160 180 200<br />
Actual Data Predicted Data<br />
R = 0.96<br />
F = 583<br />
Specific Yield immediate following<br />
flood event is greater than predicted<br />
Period of flooding<br />
It is suggested that the parameter of natural log of time in this analysis is a good surrogate <strong>for</strong> the<br />
measure of plugging in the system. This analysis was extended to include an extrapolation of an<br />
additional 5 years of data, with the insertion of “jumps” representing the impact of periodic
iverbed scouring. This analysis shows the system stabilizing after 8 years of operation, with a<br />
specific yield cycling between 0.25 and 0.4 MGD/ft drawdown. This type of extrapolation is highly<br />
speculative, but fits anecdotal in<strong>for</strong>mation on sustainable yields from high-capacity <strong>RBF</strong> systems.<br />
Riverbed Shear Stress as an Indicator of Riverbed Scouring<br />
Riverbed scouring plays an essential role in determining the sustainable yield in <strong>RBF</strong> systems. The<br />
shear stresses exerted on the riverbed during periods of high flow provide <strong>for</strong> the transport of<br />
riverbed materials and the re-suspension and transport of particles trapped during the <strong>RBF</strong> process.<br />
The extent of riverbed scouring can be estimated as a function of riverbed shear stress exerted<br />
during high-flow events.<br />
Streams typically exert higher riverbed shear stresses near headwaters, with decreasing stresses<br />
exerted near the terminus of a large water body (such as an ocean). This implies that riverbed<br />
scouring will decrease near the terminus of a stream, consistent with the <strong>for</strong>mation of river deltas.<br />
Because of this tendency to deposit fine materials near the mouth of streams, these locations are<br />
typically not well suited <strong>for</strong> high-capacity <strong>RBF</strong> systems.<br />
Two techniques <strong>for</strong> estimating riverbed shear stresses were used in this analysis:<br />
• Measured velocity profiles across the river cross-section.<br />
• Stream-surface slope measurements.<br />
The shear stresses exerted on a streambed can also be inferred by looking at the riverbed material in<br />
transport during high-flow events, as indicated by the particle-size distribution of riverbed sediments.<br />
A compilation of particle-size data from the Ohio River in the Louisville area is presented in<br />
Figure 2. Data to the left represent suspended solids, while data to the center and to the right<br />
indicate bed material. The river at the point where these data were collected is on a bend, with<br />
Percent Passing<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
Figure 2. Particle Size Distribution Analysis – Riverbed and Suspended Solids<br />
(USGS Data, 1979–1982, Ohio River at Louisville)<br />
Flow= 239,000 cfs<br />
TSS=408-466 mg/l<br />
Suspended Solids Indiana side<br />
Riverbed<br />
Flow = 301,000 to 434,000 cfs<br />
TSS = 482 to 698 mg/l<br />
Main channel and Kentucky side Riverbed<br />
0<br />
0.001 0.01 0.1 1 10 100<br />
Particle Size<br />
SS -11/30/1979<br />
SS -6/10/1981<br />
SS-1/27/1982<br />
bed-1375' from KY bank<br />
bed-1650' from KY bank<br />
bed-1850' from KY bank<br />
bed-2000' from KY bank<br />
bed-2200' from KY bank<br />
bed-2500' from KY bank<br />
bed-2700' from KY bank<br />
bed-2900' from KY bank<br />
bed-3900' from KY bank<br />
bed-4120' from KY bank<br />
SS-6/2/82 mid-stream<br />
SS-6/2/82 Ind side<br />
TSS = Total suspended solids. cfs = Cubic feet per second. SS = Suspend solids. KY = Kentucky. Ind = Indiana.<br />
45
46<br />
the Kentucky side of the river on the outside of the bend. The prevalence of courser material in<br />
the riverbed on the outside of the bend reflects higher riverbed shear stress exerted at the outside<br />
face of the bend.<br />
These data indicate that the shear stresses exerted against the riverbed are capable of transporting<br />
riverbed media in the size range of 1 to 10 millimeters. According to available literature, this<br />
would require a streambed shear stress in the range of 5 Newtons per square meter.<br />
Stream velocity profiles and stream surface slopes of the Ohio River at high flow conditions<br />
(383,000 cubic feet per second) were also secured from USGS. Velocity profile data were reported<br />
down to about 4 ft, the approximate limit of the acoustic doppler current profiler technology used<br />
<strong>for</strong> this data collection ef<strong>for</strong>t.<br />
The velocity profiles used in this study <strong>for</strong> calculating boundary shear stress would be expected to<br />
reflect a maximum shear stress <strong>for</strong> any point in the river, as the velocity profile selected was the<br />
one with the highest velocity in the cross-section. The slope calculation technique, on the other<br />
hand, would be expected to reflect an average boundary shear stress across both the cross-section<br />
and length of river selected <strong>for</strong> calculation. Comparable data sets <strong>for</strong> these two methods of<br />
calculation revealed that the velocity profile (using the manipulation yielding the greatest shear<br />
velocity) provided a lower and less consistent estimate of boundary shear stress than did the<br />
method based on measured stream slope. These data are summarized in Table 1 from a 2000 USGS<br />
database:<br />
Table 1. Comparison of Shear Stress Calculation Methods at Three Locations<br />
in the Ohio River Near Louisville<br />
Location Profile-Calculated Shear Stress Slope-Calculated Shear Stress<br />
(N/m 2 ) (N/m 2 )<br />
12 Mile Island 1.63 3.21<br />
BEPWTP 1.68 5.33<br />
Zorn Avenue 5.79 5.62<br />
BEPWTP = B.E. Payne <strong>Water</strong> Treatment Plant. N/m 2 = Newtons per square meter.<br />
The inconsistency and lower results from the stream profile calculations may be the results of the<br />
higher degree of variability obtained from the acoustic doppler measurement technique <strong>for</strong><br />
velocity and, particularly, the inability <strong>for</strong> this technique to measure velocity at the lower portions<br />
of the stream. It is also noted that video images of the riverbed indicated surface irregularities (or<br />
“pocks” on the river bottom). These depressions would be expected to change the velocity profile<br />
as compared to a theoretical surface where friction is exerted by media roughness only.<br />
The slope-calculated shear stress values, on the other hand, were consistent and logical <strong>for</strong> all of<br />
the measured values. The higher stream flows were associated with higher shear stresses, and shear<br />
stresses tended to be greater where the river was deeper and narrower.<br />
The data from the Ohio River were compared to data from the Rhine River. The maximum<br />
boundary shear stress observed on the Ohio River during the high-flow event was<br />
7.5 Newtons per square meter. This compares to the average shear stress of the Rhine River of<br />
10 Newtons per square meter, as reported by Schubert (2002).
REFERENCES<br />
Schafer, D.C. (2003). “The Use of Aquifer Testing and Groundwater Modeling to Evaluate Changes in<br />
Aquifer/River Hydraulics at the Louisville <strong>Water</strong> Company.” Program and Abstracts, Second International<br />
Riverbank Filtration Conference, G. Melin, ed., <strong>National</strong> <strong>Water</strong> <strong>Research</strong> <strong>Institute</strong>, Fountain Valley.<br />
Schubert, J. (2002). “Hydraulic aspects of riverbank filtration: Field studies.” Journal of Hydrology, 266 (3-4):<br />
154-161.<br />
STEVE HUBBS is a Professional Engineer with 28 years of experience at the Louisville<br />
<strong>Water</strong> Company in Louisville, Kentucky. In the early 1980s, he began researching riverbank<br />
filtration as an alternate source of water <strong>for</strong> the Louisville <strong>Water</strong> Company, specifically<br />
looking at the reduction in disinfection byproduct precursors, river-borne organics, and<br />
mutagenicity in the riverbank-filtration process. His work continued in the 1990s with a<br />
focus on pathogen reduction, and his research is now being conducted on the hydraulic<br />
connection between the riverbed and the aquifer, with a focus on riverbed plugging<br />
dynamics and their influence on sustainable yields from high-capacity riverbank-filtration systems. Hubbs<br />
received an M.S. in Environmental Engineering from the University of Louisville, and is currently enrolled<br />
in the Ph.D. program in Civil and Environmental Engineering at the University of Louisville, focusing on<br />
the hydraulics of riverbank-filtration systems.<br />
47
Session 3: Hydraulic Aspects<br />
Application of Different Tracers to Evaluate<br />
the Flow Regime at Riverbank-Filtration Sites<br />
in Berlin, Germany<br />
Dr. Gudrun Massmann<br />
Free University of Berlin<br />
Berlin, Germany<br />
Dipl. Geol. Andrea Knappe<br />
Alfred-Wegener-<strong>Institute</strong> Potsdam<br />
Pottsdam, Germany<br />
Doreen Richter<br />
Free University of Berlin<br />
Berlin, Germany<br />
Dr. Jürgen Sültenfuß<br />
University of Bremen<br />
Bremen, Germany<br />
Prof. Asaf Pekdeger<br />
Free University of Berlin<br />
Berlin, Germany<br />
Introduction<br />
The inhabitants of metropolitan Berlin (Germany) rely on drinking water derived from groundwater<br />
within the City’s boundary. Production well galleries are generally located adjacent to the surfacewater<br />
system and artificial infiltration ponds, and around 70 percent of abstracted groundwater is<br />
estimated to originate from bank filtration and artificial groundwater recharge (Pekdeger and<br />
Sommer-von Jarmerstedt, 1998). Berlin’s water production is a semi-closed water cycle with an<br />
indirect potable reuse system: local wastewater treatment plants discharge treated effluent into the<br />
surface-water system (Figure 1). Induced by well abstraction, surface water infiltrates into the<br />
ground and is used <strong>for</strong> drinking-water production. The raw water requires only minimal treatment<br />
to ensure good drinking-water quality. After use, water is redirected into wastewater treatment<br />
plants and released into the surface-water system again after treatment. Hence, the quality of the<br />
raw water is influenced by a number of factors, including:<br />
• Temporal and spatial variations of surface-water quality, largely depending on location in<br />
relation to wastewater treatment plants.<br />
• Presence, permeability, and thickness of the colmation layer.<br />
• Lithology, permeability, and geochemistry of the aquifer sediment.<br />
• Nature of the wells (location, length, and depth of filter screens).<br />
Correspondence should be addressed to:<br />
Dr. Gudrun Massmann<br />
Scientific Colleague<br />
Free University of Berlin<br />
Hydrogeology Group • Malteserstr. 74-100 • 12249 Berlin Germany<br />
Phone: +49-30-83870472 • Fax: +49-30-83870742 • Email: massmann@zedat.fu-berlin.de<br />
49
50<br />
TS GWA Tegel<br />
TS Wannsee<br />
TS Tegeler See<br />
WW Spandau<br />
Spandau WW Jungfernheide<br />
Charlottenburg<br />
KW Ruhleben<br />
Unterschleuse<br />
WW Kladow<br />
Havel<br />
WW Stolpe KW Schönerlinde<br />
Tegeler See<br />
Spree<br />
WW Tiefwerder<br />
WW und PEA<br />
Beelitzhof<br />
Kleinmachnow<br />
KW Stahnsdorf<br />
WW Tegel<br />
Tegeler Flieβ<br />
PEA-Tegel<br />
input summer<br />
KW Ruhleben<br />
Telfowkanal<br />
Nordgraben<br />
Legend<br />
flow direction<br />
waterworks<br />
sewage treatment plant<br />
surface water treatment plant<br />
transect / field site<br />
weir<br />
lock<br />
Figure 1. Overview of the surface-water system, flow directions, waterworks, wastewater and surface-water<br />
treatment plants, and study sites in the western part of Berlin.<br />
• Hydraulic regime resulting from natural gradients and pumping per<strong>for</strong>mances.<br />
• <strong>Water</strong>-quality changes that occur during bank filtration.<br />
A major advantage of the application of bank filtration is the capability of the subsurface to<br />
remove contaminants during underground passage, either by physical filtration and/or (bio)chemical<br />
processes, such as adsorption, reduction, or degradation. In addition, the application of bank<br />
filtration spares natural groundwater resources, which inhibits the rise of deeper, more saline<br />
groundwater into local freshwater aquifers; however, since a proportion of the surface water in<br />
Berlin originates from treated effluent released by wastewater treatment plants, it is necessary to<br />
constantly monitor the fate of potential contaminants during bank filtration.<br />
Processes accompanying bank filtration and artificial recharge are currently studied in Berlin<br />
within a multidisciplinary cooperative project at the Berlin Centre of Competence called Natural<br />
and Artificial Systems <strong>for</strong> Recharge and Infiltration (NASRI) (KWB, 2002). The project focuses<br />
on the behavior and removal of, <strong>for</strong> example, pathogens, microcystins, and organic pollutants, as<br />
well as pharmaceutically active compounds, during underground passage. To interpret the<br />
behavior of these non-conservative water constituents, the local hydrogeological and hydraulic<br />
situation has to be well understood. Three field sites are the focus of current investigations, all in<br />
the western part of the City (see Figure 1). Two sites are located next to natural lakes (Tegeler<br />
See/Tegel Lake and Wannsee/Wannsee Lake) and one next to an artificial recharge pond (GWA<br />
Tegel). Only examples of the Wannsee Lake site will be discussed.<br />
Several tracers have been successfully applied in Berlin <strong>for</strong> various purposes. Some of these are<br />
useful to study water movement and derive mean residence times (e.g., delta deuterium [δD],<br />
delta oxygen 18 [δ 18 O], temperature [T], chloride [ C1 – ]), while others may be particularly useful<br />
in identifying the proportion of treated wastewater in surface water (e.g., boron [B], Cl – , δD, δ 18 O)<br />
or the proportion of either bank filtrate (e.g., ethylenediaminetetraacetic acid [EDTA], galdolinium<br />
[Gd], strontium [Sr]) or deeper saline groundwater (e.g., B, Cl-, sodium [Na+]) in raw water.<br />
Panke<br />
Landwehrkanal<br />
(KW Marienfelde)
Methodology<br />
Transects generally reach from the lake or artificial recharge pond to a production well parallel to<br />
the flow direction. These transects contain a number of observation wells, usually one below the<br />
lake, some between the lake and production well, and some beyond the well. A brief overview on<br />
the analysis discussed here is given in Table 1.<br />
The T-He age dating method uses the ratio of the concentration of radioactive tritium ( 3 H or T)<br />
derived from atmospheric nuclear bomb testing and its decay product, Helium ( 3 He), in<br />
groundwater to determine a groundwater age (i.e., the time passed since water had its last contact<br />
with the atmosphere).<br />
In case the abstracted water is a mixture of surface water and background groundwater only, the<br />
percentage of bank filtrate in the well (X) can be calculated as:<br />
X = [C w–C GW)/(C SW–C GW)]•100 [%]<br />
With C as the concentration of a suitable tracer in groundwater (C GW), well water (C w), or surface<br />
water (CSW).<br />
The simple mixing <strong>for</strong>mula can be used under the premise that the differences between groundwater<br />
and surface water are large and relatively stable, and well water is a mixture of two water components<br />
only.<br />
Results<br />
Surface-<strong>Water</strong> System<br />
Table 1. Overview of Analytical Methods<br />
Parameter Analytical Method Laboratory Details Provided In<br />
4 He, 20 Ne, Mass Spectrometry <strong>Institute</strong> of Environmental Physics, Sültenfuß et al.<br />
and 22 Ne Pfeiffer QMG 112 University of Bremen (in preparation)<br />
(Age Dating)<br />
3 He and 4 He Mass Spectrometry <strong>Institute</strong> of Environmental Physics, Sültenfuß et al.<br />
(Age Dating) MAP215-50 University of Bremen (in preparation)<br />
Cl – Photometry Free University of Berlin<br />
(Autoanalyser,<br />
Technicon)<br />
B ICP-OCE Free University of Berlin<br />
δ 2 H, δ 18 O Delta S MS Alfred-Wegener-<strong>Institute</strong>, Meyer et al. (2000)<br />
Potsdam<br />
EDTA DIN 38413-PO3 Berlin <strong>Water</strong> Works<br />
4 He = Helium 4. 20 Ne = Neon 20. 22 Ne = Neon 22. 3 He = Helium 3. 4 He = Helium 4. ICP-OCE = Ionic Coupled Plasma.<br />
The strain on surface water becomes larger as it flows through the City and is substituted by<br />
considerable amounts of treated wastewater. The wastewater influence provokes high<br />
concentrations of, <strong>for</strong> example, Cl – , B, anthropogenic Gd, high conductivities and temperatures<br />
(in winter), and more negative isotope signatures. The highest influence and largest fractions of<br />
treated wastewater can be seen in the Teltow Canal and Nordgraben (a ditch). The Havel River<br />
51
52<br />
contains the lowest concentration of wastewater indicators. For example, Figure 2 shows the<br />
B concentration and the proportions of treated wastewater in surface water in September 2002. The<br />
influence of the Nordgraben extends to transect Tegel Lake, while the field site Wannsee Lake is<br />
under the influence of the Teltow Canal. The percentage of treated wastewater is higher during<br />
the summer months, when the Ruhleben Treatment Plant discharges into the Teltow Canal.<br />
Figure 2. (a) Percentage of treated wastewater in surface water (calculated with discharge only; evaporation,<br />
storage changes, and precipitation neglected). (b) Boron concentration indicating the influence of<br />
the Teltow Canal in the south and Nordgraben in the northeast <strong>for</strong> September 2002.<br />
By combining the results of the discharge measurements with the concentrations of wastewater<br />
indicators, proportions of treated wastewater can be calculated with a mixing <strong>for</strong>mula <strong>for</strong> each<br />
sampling point. In Figure 3, results are shown <strong>for</strong> Tegel Lake in front of the transect. In October<br />
2001, a pumping device started operation, which pumps Havel River water into the PEA-Tegel, a<br />
phophate elimination plant (see Figure 1), when the natural discharge of the Nordgraben is very<br />
low in summer. This led to a strong decrease of treated wastewater in Lake Tegel near the transect<br />
(from 33 to 46 percent [Fritz, 2002] to an average of 11.7 +/– 1.9 percent). For surface water near<br />
transect Wannsee Lake, values of 20.1 percent in the summer (when Ruhleben discharges into the<br />
Teltow Canal) and 8.5 percent in the winter were calculated.<br />
Bank-Filtration System (Wannsee Lake)<br />
At Wannsee Lake, two transects exist, running perpendicular to the shore of Wannsee Lake<br />
between drinking-water Production Wells 3 (TS Wannsee 2) and 4 (TS Wannsee 1). The wells<br />
of the local gallery (Beelitzhof) have several filter screens in different porous aquifers separated by<br />
aquitards. A schematic overview is given in Figure 4. Also shown are the T and 4 He concentrations<br />
originating from hydrogen bomb testing in the 1960s, as well as uranium and thorium decay<br />
within the aquifer. In the two deeper aquifers, groundwater is considerably older than 50 years,<br />
since T is already decayed and the 4 He values are high. In contrast, the shallow wells reflect the<br />
present atmospheric concentrations of T, while 4 He could not be detected. The resulting age is less<br />
than the detection limit of 3 months. The production well is a mixture of all aquifers.
Figure 3. Percentage of treated wastewater calculated with B and Cl <strong>for</strong> surface water in front of transect<br />
Tegel Lake.<br />
W E<br />
3339<br />
3338<br />
3337<br />
BEE201OP<br />
BEE201UP<br />
3336<br />
3334<br />
3335<br />
Brunnen 4<br />
BEE200UP<br />
BEE200OP<br />
3332<br />
m below m above<br />
ground sea-level<br />
Figure 4. Tritium and terrigenic 4 He in different observation wells and Production Well 4 of TS Wannsee 1.<br />
Mean residence times are estimated with the help of (<strong>for</strong> example) δD and δ 18 O breakthrough<br />
curves. The surface water shows a clear seasonal signal, with more negative signatures in winter<br />
(Figures 5a and 6). The exemplary breakthrough curves of Observation Wells 3,337 and 3,335<br />
(see Figure 4) reflect the surface-water signal. Here, groundwater is made of bank filtrate only. The<br />
curves illustrate that residence times are rather low at this particular site. Travel times are around<br />
1 month on the distance from the lake to Observation Well 3,337, which is approximately<br />
two-thirds of the way to the production well. Well 4 does not show a seasonal signal and has a more<br />
negative signature.<br />
53
54<br />
δ 18 O [ 0⁄00 versus Standard Mean Ocean <strong>Water</strong>]<br />
EDTA [µg/L]<br />
–6<br />
–7<br />
–8<br />
–9<br />
7<br />
6<br />
5<br />
4<br />
3<br />
similar in the lake and Wells 3 and 5. Mixing calculations assuming a residence time of 1 to 2 months<br />
result in an average percentage of bank filtrate of 61 to 97 percent in Well 3, 10 to 47 percent in<br />
Well 4, and 88 to 96 percent in Well 5. The large variations are mainly due to the detection limit<br />
of 2-µg/L EDTA, which results in uncertainties in the actual concentrations of Well 4 and<br />
background groundwater. An evaluation of other tracers suggest that the proportion of bank<br />
filtrate in Well 4 is more likely to be in the order of magnitude of 10 percent.<br />
The differences between the water bodies become clear in the plot of δD versus δ 18 O (Figure 6),<br />
where Well 4 plots next to the deeper aquifer samples. In contrast, Well 3 and Well 5 plot within<br />
the surface-water samples (just like the shallow observation wells), but show less seasonal variation.<br />
δD [ 0⁄00 versus Standard Mean Ocean <strong>Water</strong>]<br />
–46<br />
–48<br />
–50<br />
–52<br />
–54<br />
–56<br />
–58<br />
–60<br />
–62<br />
–64<br />
–66<br />
–68<br />
–70<br />
Figure 6. δD versus δ 18 O at the Wannsee Lake field sites. Analysis of deeper aquifer samples was done from<br />
September 2000 to November 2001, with remaining data from May 2002 to March 2003.<br />
Conclusions<br />
Mar 03<br />
Jan 03<br />
Feb 03<br />
May 02<br />
Dec 02<br />
–9 –8 –7 –6 –5<br />
δD [ 0⁄00 versus Standard Mean Ocean <strong>Water</strong>]<br />
Jun 02<br />
Nov 02<br />
Sep 02<br />
Aug 02<br />
Jul 02<br />
The combination of different tracers enables the interpretation of the flow regime at sites where<br />
bank-filtration processes are currently studied in Berlin. With the help of T/He analysis, the ages<br />
of different water bodies can be estimated. The analysis of tracers showing distinct seasonal<br />
variations is used to estimate travel times, while water constituents — which are either mainly<br />
present in bank filtrate or background water — are used <strong>for</strong> mixing calculations. The results of the<br />
tracer studies can be used to interpret the fate and behavior of potential contaminants (e.g., drug<br />
residues, organic pollutants, etc.).<br />
55
56<br />
Acknowledgements<br />
We would like to thank the German <strong>Research</strong> Foundation, Veolia <strong>Water</strong>, and the Berlin <strong>Water</strong><br />
Company <strong>for</strong> financing the NASRI Project.<br />
REFERENCES<br />
Fritz, B. (2002). Untersuchungen zur Uferfiltration unter verschiedenen wasserwirtschaftlichen, hydrogeologischen<br />
und hydraulischen Bedingungen, Ph.D. Thesis, University of Berlin, Berlin, 203 pp.<br />
KWB (2002). NASRI Natural and Artificial Systems <strong>for</strong> Recharge and Infiltration, First Progress Report, reporting<br />
period May 2002 – December 2002.<br />
Pekdeger, A., and C. Sommer-von Jarmerstedt (1998). Einfluß der Oberflächenwassergüte auf die Trinkwasserversorgung<br />
Berlins, Forschungspolitische Dialoge in Berlin, Geowissenschaft und Geotechnik, Berlin, pp. 33-41.<br />
Meyer, T., L. Schönicke, U. Wand, H.W. Hubberten, and H. Friedrichsen (2000). Isotope studies of hydrogen<br />
and oxygen in ground ice – Experiences with the equilibration technique, Isotopes Environ. Health Stud., p 133-149.<br />
Sültenfuß, J., G. Massmann, and A. Pekdeger (in preparation). Datierung mit der He3-T-Methode am Beispiel<br />
der Uferinfiltration im Oderbruch.<br />
GUDRUN MASSMANN studied Geology at both the Universities of Bremen and<br />
Edinburgh, specializing in Hydrogeology. After receiving her Diploma, she worked as an<br />
occupational trainee at the Center <strong>for</strong> Groundwater Studies at the Commonwealth<br />
Scientific & Industrial <strong>Research</strong> Organization (CSIRO) Land and <strong>Water</strong> in Adelaide,<br />
Australia, gaining experience on Artificial Storage and Recovery. For the past 4 years, she<br />
has worked as a scientific colleague <strong>for</strong> the Hydrogeology Group of the Free University of<br />
Berlin. Her research interests include hydraulic modeling, hydrochemistry, tracer analysis,<br />
and isotopes. Massmann finished her Ph.D. in summer 2002 at the Free University of Berlin on a project<br />
dealing with evaluating and modeling hydraulic and hydrochemical processes accompanying bank filtration<br />
in a polder region in Germany. She then continued to word on bank filtration, but changed the field site to<br />
Berlin, where bank filtration plays an important role in drinking-water production.
Session 4: Siting<br />
Siting and Testing Procedures<br />
<strong>for</strong> Riverbank-Filtration Systems<br />
Samuel M. Stowe, P.G., CPG<br />
International <strong>Water</strong> Consultants, Inc.<br />
Columbus, Ohio<br />
Phased, multi-tasked investigations can be structured to collect pertinent data to evaluate yield,<br />
quality, and (ultimately) the design of <strong>RBF</strong> systems. The initial phase should:<br />
• Define project objectives (yield, quality, water use).<br />
• Identify project concerns (treatment and regulatory issues).<br />
• Collect available data <strong>for</strong> preliminary hydrogeological screening/feasibility.<br />
If the initial phase indicates a reasonable probability that potential favorable conditions exist<br />
(given project objectives), then subsequent phases can be designed to collect site-specific data <strong>for</strong><br />
comprehensive analysis. Subsequent phases should include:<br />
• Test drilling.<br />
• Hydraulic interval testing.<br />
• <strong>Water</strong>-quality screening.<br />
• Streambed characterization.<br />
• <strong>Water</strong>-level monitoring.<br />
• Detailed aquifer and water-quality testing.<br />
The key parameters to any <strong>RBF</strong> evaluation are aquifer transmissivity and streambed permeability.<br />
A good understanding of these parameters, along with the hydrogeological setting, will result in<br />
the proper evaluation of the expected yield and quality of a <strong>RBF</strong> system and will allow <strong>for</strong> the<br />
thorough evaluation of potential means of development. Understanding the ability of the aquifer<br />
to provide sufficient <strong>RBF</strong> to recharge water pumped from a <strong>RBF</strong> system is key to ensuring that<br />
long-term capacities can be sustained and that target water quality can be maintained through a<br />
balance of infiltrated surface water and groundwater.<br />
Two case studies are presented to provide examples of recent <strong>RBF</strong> investigations. One is <strong>for</strong> a<br />
low-yield system (1.0 MGD), with very difficult access conditions in New Mexico, where rafts,<br />
helicopters, and portable drilling equipment were used. The other is a high-yield system<br />
(50 MGD) along a large river in the Great Plains, where more standard access and drilling methods<br />
(rotary, rotosonic) were used.<br />
Correspondence should be addressed to:<br />
Samuel M. Stowe, P.G., CPG<br />
President<br />
International <strong>Water</strong> Consultants, Inc.<br />
6360 Huntley Road • Columbus, Ohio 43229 USA<br />
Phone: (614) 888-6263 • Fax: (614) 888-9208 • Email: smstowe@collectorwellsint.com<br />
57
58<br />
Phase 1 Investigations<br />
Once the project objectives have been adequately defined, existing data that pertain to the<br />
hydrogeology and hydrology of the region of interest should be collected, assimilated, and assessed.<br />
This research should focus on the feasibility of developing the required capacity from<br />
unconsolidated deposits. In<strong>for</strong>mation collected <strong>for</strong> evaluation would include:<br />
• Well logs.<br />
• Bridge boring logs.<br />
• Steam flow and quality data.<br />
• Records of production wells <strong>for</strong> other water utilities in the region.<br />
• Reports on file with state or local agencies.<br />
• Other in<strong>for</strong>mation that pertains to the hydrogeological setting of the area.<br />
Additionally, sites that are identified as potentially favorable should be visited and inspected.<br />
Phase 1 findings should include the following:<br />
• Listings of the available data reviewed and other resources used.<br />
• Discussions of the data and results of any preliminary analyses that were possible, with the<br />
confidence level of data obtained.<br />
• Identification of areas that appear to have a favorable potential <strong>for</strong> testing.<br />
• Preliminary rankings of prospective sites, with a discussion of relative advantages and<br />
disadvantages <strong>for</strong> siting.<br />
• Evaluations regarding water quality that may be expected from a <strong>RBF</strong> system located in<br />
this area, considering depth, proximity to the river, other regional water-quality problems<br />
experienced, and local features that may impact quality, etc.<br />
• Assessment of permits that may be required <strong>for</strong> installation and withdrawal using a<br />
<strong>RBF</strong> system.<br />
• Descriptions and budgets <strong>for</strong> work procedures recommended to be completed in<br />
subsequent phases.<br />
Phase 2 Investigations<br />
Phase 2 tasks would include drilling, sampling, and testing exploratory borings/observation wells<br />
at sites identified in Phase 1. Given logistics and site layouts, surface geophysical surveys (seismic,<br />
electrical resistivity, electromagnetic) could precede actual drilling. Geophysical surveys can be<br />
helpful in preliminary screening activities. These sites would be selected based upon access (land<br />
ownership, permission, etc.) and the potential <strong>for</strong> favorable saturated aquifer thicknesses near the<br />
river. The primary objective of this Phase is to locate the most favorable site(s) <strong>for</strong> detailed aquifer<br />
testing (Phase 3) and the collection of data <strong>for</strong> hydraulic and quality evaluations <strong>for</strong> site ranking.<br />
The selection of appropriate drilling and sampling methods are imperative during this phase.<br />
Drilling methods should consider:<br />
• Representative soil samples.<br />
• Expected drilling conditions (materials, depths).<br />
• Rate of advancement (timely drilling).<br />
• Cost-effectiveness.
• Potential <strong>for</strong> hydraulic testing and water-quality sampling.<br />
• Site access conditions.<br />
• Monitoring well installation.<br />
Potential drilling/sampling methods include:<br />
• Hollow stem auger with split-spoon sampling.<br />
• Cable tool with bailed samples.<br />
• Direct rotary with wash and split-spoon sampling.<br />
• Rotosonic with dual-tube continuous sampling.<br />
• Reverse rotary with wash samples.<br />
Generally, it has been found that rotosonic drilling provides the best overall results <strong>for</strong> representative<br />
soil sampling, speed, water sampling, and hydraulic testing. Drawbacks have been costs and<br />
access. In the rotosonic drilling method, a drill casing is advanced into the ground using<br />
rotary/vibrasonic techniques. This method does not require the use of drilling mud, so there is no<br />
mud to dispose of, and ground-surface disturbance is minimal. The rotosonic drilling method<br />
produces nearly continuous samples of the materials penetrated by the sample tube, and the<br />
method produces representative samples from unconsolidated, granular deposits.<br />
Sieve analyses should be per<strong>for</strong>med on selected lithologic samples collected from test borings to<br />
determine optimum well-screen design and to help evaluate hydraulic conductivity. Hydraulic<br />
interval testing can also be conducted on test borings to determine the hydraulic conductivity of<br />
selected intervals and to evaluate groundwater quality.<br />
Following the collection of all field data, the data are compiled and analyzed to determine the<br />
most favorable locations <strong>for</strong> detailed aquifer testing (Phase 3). This determination is based upon<br />
saturated thicknesses, hydraulic conductivities, and logistics. From Phase 2 results, it can be<br />
determined with a reasonable degree of certainty that one or more of the sites tested is capable of<br />
yielding the desired quantity of water.<br />
Phase 3 Investigations<br />
At the completion of Phase 2, if results are favorable, a site (or sites) <strong>for</strong> detailed aquifer testing is<br />
selected. Phase 3 studies generally include the installation of additional observation wells <strong>for</strong><br />
sampling subsurface materials, monitoring water levels, and conducting a detailed aquifer test,<br />
along with a test pumping well. Additionally, the evaluation of streambed conditions (width,<br />
depth, permeability) is generally conducted during this phase.<br />
Observation wells are positioned in a pattern around the test pumping well at appropriate locations<br />
and distances selected to facilitate data analysis, with emphasis on <strong>RBF</strong>. Following the installation<br />
of observation wells, a temporary test pumping well is installed. It is important that the test<br />
pumping well be constructed to produce enough water to adequately stress the aquifer. Well points<br />
can be installed in the river (conditions permitting) to further evaluate streambed permeability.<br />
The monitoring of water levels in the wells and river should be conducted prior to any pumping<br />
to evaluate antecedent trends and the correlation of stream level and groundwater levels, with<br />
levels converted to elevation datum <strong>for</strong> analysis. Following the collection of background water<br />
levels, a long-term (2- to 10-days) constant rate test is conducted. A general guideline is to run<br />
the test 24 hours after steady-state conditions are reached. At the conclusion of the test, the<br />
59
60<br />
pumping is discontinued and the recovery of water levels is monitored until at least 90-percent<br />
recovery is obtained in the observation wells. During the test, samples of the water pumped (and<br />
river) can be obtained <strong>for</strong> laboratory analyses. Additionally, the pump discharge should be<br />
monitored every 6 to 12 hours and river quality every 24 hours <strong>for</strong> field screening of temperature,<br />
pH, iron, hardness, turbidity, and specific conductance.<br />
All data from Phase 3 are analyzed to determine aquifer properties, streambed permeability, aquifer<br />
character, yield, anticipated quality, and conceptual <strong>RBF</strong> system design. Yield estimates should<br />
consider the impacts of river stage/flow and water temperature variations. Based upon the results,<br />
the preferred location of a <strong>RBF</strong> system can be recommended based upon an evaluation of benefits<br />
and drawbacks, including refinements in design. Phase 3 testing may also identify additional<br />
testing that may be necessary to finalize design.<br />
Case Studies<br />
Two case studies are briefly discussed to illustrate the phased approach to siting <strong>RBF</strong> systems and<br />
critical evaluation items. Case Study Number 1 was completed in north-central New Mexico along<br />
the Rio Grande (Figure 1), where difficult access conditions presented challenges <strong>for</strong> completion<br />
using conventional approaches. Project objectives were to locate a site that could yield at least<br />
1.0 MGD of infiltrated surface water from the Rio Grande <strong>for</strong> potable use (surface-water rights to<br />
1,200 acre-feet per year). Phase 1 studies (literature review, site reconnaissance) identified up to<br />
six potentially favorable sites within the canyon. Given difficult site access, the initial Phase 2 study<br />
involved surface geophysics (electrical resistivity) at the six locations, along with a control site.<br />
Access to the sites was obtained using an all terrain vehicle and rafts, with base camps set up <strong>for</strong><br />
overnight stays. Based upon the survey, two sites were selected <strong>for</strong> test drilling.<br />
Figure 1. Rio Grande Valley, New Mexico.<br />
Drilling was conducted using a portable modular core rig with wash samples and 2-inch<br />
outside-diameter split-spoon samples. At Site A, nine borings were drilled, six 1.5-inch diameter<br />
observation wells installed, and hydraulic testing (high-efficiency suction pumps) completed. As<br />
results were favorable, Phase 3 aquifer testing was completed with additional observation wells and<br />
a 66-hour constant rate test (0.144 MGD).<br />
Following the completion of testing at Site A, the equipment and base camp were mobilized to<br />
Site B using rafts and a helicopter. Phase 2 testing included the drilling of four borings, interval
testing, and a short constant rate test. As results here were not as favorable as at Site A, detailed<br />
aquifer testing was not completed.<br />
Case Study Number 2 was completed along the Missouri River in North Dakota (Figure 2), where<br />
more conventional investigative methods could be used. Project objectives were to locate a site<br />
that could yield up to 50 MGD of potable water from a <strong>RBF</strong> system. The study was being<br />
completed to evaluate the possibility of replacing an existing direct surface-water intake. Phase 1<br />
investigations (desktop study, site visit) determined that several sites existed within the area,<br />
which were promising. Phase 2 investigations were targeted at two sites, given the primary<br />
logistical concern of distance from existing treatment. Phase 2 included the drilling of three<br />
borings and conducting one hydraulic test using rotosonic methods.<br />
Figure 2. Missouri River Valley, North Dakota.<br />
Given favorable Phase 2 results, Phase 3 investigations were completed at one site with the<br />
installation of four additional observation wells and a test pumping well. Because of budget<br />
concerns, Phase 3 drilling was completed using mud rotary methods and a local drilling contractor.<br />
Testing was completed with the running of a 72-hour constant rate test (1.9 MGD). The analysis<br />
of data indicated a transmissive aquifer (40,000 square feet per day) in good communication with<br />
the river. Preliminary results indicate that the site will yield 30 MGD from a single horizontal<br />
collector well.<br />
SAM STOWE is President of International <strong>Water</strong> Consultants, Inc., a subsidiary of<br />
Collector Wells International, Inc. A hydrogeologist, Stowe has nearly 30 years of diverse<br />
experience in the groundwater industry. He has been in charge of projects involving<br />
aquifer-test analyses, riverbank filtration and recharge evaluation, groundwater quality,<br />
well design, groundwater management, numerical modeling, contamination investigation,<br />
and remedial action. He has also been involved in groundwater-supply projects <strong>for</strong> yields<br />
of nearly 100 million gallons per day and in contamination evaluations ranging from<br />
industrial organic pollution from landfills to the environmental effects of strip and deep-coal mining. In<br />
addition, Stowe is highly experienced in evaluations of induced infiltration potential from streams and<br />
oceans. He has completed hundreds of hydrogeological evaluations <strong>for</strong> horizontal collector wells in regard to<br />
yield, quality, and design, having been responsible <strong>for</strong> siting and designing many horizontal collector wells<br />
with individual yields ranging from 2 to 40 million gallons per day and is a recognized expert in their<br />
application. Stowe received a B.A. in Geology from Miami University and an M.S. in Geology from Ohio<br />
State University.<br />
61
Session 4: Siting<br />
<strong>Water</strong>-Quality Management <strong>for</strong> Existing Riverbank-<br />
Filtration Sites along the Elbe River in Germany<br />
Prof. Dr.-Ing. Thomas Grischek<br />
University of Applied Sciences Dresden<br />
Dresden, Germany<br />
Introduction<br />
In <strong>for</strong>mer times, the management of <strong>RBF</strong> sites in Germany focused on water quantity — how much<br />
water could be abstracted and what groundwater levels would result. Control measures were based<br />
on measurements of river and groundwater levels and pumping rates and on black-box models to<br />
estimate the portion of riverbank filtrate abstracted (e.g., Luckner and Nestler, 1982). The<br />
development of computer programs <strong>for</strong> groundwater flow and transport simulations resulted in<br />
increasing management applications <strong>for</strong> <strong>RBF</strong> sites (e.g., Koster et al., 1994; Heinzmann, 1998;<br />
Eckert et al., 2000).<br />
The number of published papers dealing with the complex management of water quality at <strong>RBF</strong><br />
sites is very low. Besides work done by Sontheimer (1991) and Schubert (1999) at sites along the<br />
Rhine River, a control and management concept <strong>for</strong> the Hengsen site on the Ruhr River<br />
(Schöttler and Sommer, 1992) must be highlighted. At the end of the 1980s, water-quality<br />
management also became a subject <strong>for</strong> <strong>RBF</strong> sites on the Elbe River due to problems with river and<br />
raw-water quality. Initial work done by Müller, Schwan, and others between 1985 and 1989 was<br />
continued by Nestler et al. (1998).<br />
<strong>Water</strong>-quality management must be based on detailed knowledge of groundwater flow conditions<br />
and monitoring measures to obtain sufficient data on water quality; however, this is not the case<br />
<strong>for</strong> every site. Of course, <strong>for</strong> small waterworks, such investigations and investments may be higher<br />
than the cost savings resulting from water-quality management. But, the optimization of water<br />
abstraction can have long-term effects on raw-water quality and treatment cost savings. <strong>Water</strong>quality<br />
management systems are well established at sites where problems with contamination have<br />
occurred (e.g., high concentrations of nitrate, organic halogens). In such cases, experiences may<br />
not be published to give the impression of a “no problem waterworks” or because a clear<br />
description is not easy due to the large amount of data needed to understand the complex system<br />
and site-specific boundary conditions.<br />
General <strong>Water</strong>-Quality Management Measures<br />
The first step in water-quality management is to clarify which advantage of <strong>RBF</strong> is the most<br />
important and to identify the main aims or problems. At one site, the nitrate concentration in the<br />
raw water should be decreased; at another site, the concentration of DOC, dissolved iron, or an<br />
organic contaminant should decrease.<br />
Correspondence should be addressed to:<br />
Prof. Dr.-Ing. Thomas Grischek<br />
Professor<br />
<strong>Institute</strong> <strong>for</strong> Geotechnics & <strong>Water</strong> Sciences<br />
University of Applied Sciences Dresden • Friedrich-List-Platz 1 • 01069 Dresden, Germany<br />
Phone: +49 351 4623350 • Fax: +49 351 4623567 • Email: grischek@htw-dresden.de<br />
63
64<br />
<strong>Water</strong>-quality management could include:<br />
• Optimization of the operation of existing abstraction wells.<br />
• Building additional abstraction wells.<br />
• Technical measures in the riverbed.<br />
• (Political) activities to improve river-water quality.<br />
• Contracts with farmers to change land use in a catchment zone.<br />
This paper focuses on the operation of existing wells affected by mixing processes in an aquifer,<br />
flow times and flow path lengths, and the catchment area <strong>for</strong> landside groundwater. Special<br />
measures to face problems resulting from contaminant shock loads in a river, as well as droughts<br />
or floods, are not covered because these are the subjects of other contributions in this volume.<br />
Strategies based on the operational control of existing abstraction wells could include:<br />
• Well operation to achieve a maximum volume of utilized aquifer (long flow paths and<br />
retention times).<br />
• Preferential operation of selected wells with best raw-water quality.<br />
• Operation of a limited number of wells with high abstraction rates to increase the<br />
proportion of bank filtrate in raw water.<br />
• Continuous operation of selected wells to meet the mean water demand and additional<br />
operation of other wells in peak periods.<br />
• Periodic operation of wells to increase the effects of dispersion and mixing in the aquifer.<br />
An additional technical measure could be a change of the well filter length and depth in an<br />
aquifer.<br />
Riverbank Filtration along the Elbe River<br />
At present, there are eight water facilities along the Elbe River that use <strong>RBF</strong> to supply more than<br />
1.5-million people with drinking water and industry with process water. Three of these sites have<br />
been chosen to carry out further investigations on water-flow and water-quality changes during<br />
<strong>RBF</strong>, especially due to the changing water quality of the Elbe River in the 1990s. Between 1992<br />
and 2002, different research programs focused mainly on the behavior of DOC, EDTA, sulfur<br />
organic compounds, and carbonic acids. Besides this specialized research, daily problems among<br />
water companies were investigated.<br />
All <strong>RBF</strong> sites along the Elbe River are operated under anoxic conditions. Reducing conditions<br />
result in the dissolution of iron and manganese along the flowpath between the river and<br />
abstraction wells. At some sites, denitrification occurring during <strong>RBF</strong> was found to compensate <strong>for</strong><br />
high nitrate concentrations in groundwater from other sources.<br />
Due to the closure of large industries and a new water pricing system after the reunification of<br />
Germany, water demand decreased significantly between 1989 and 1998. Lower abstraction rates<br />
<strong>for</strong> bank filtrate and an improvement in river-water quality allow <strong>for</strong> efficient groundwater<br />
management and the optimization of pumping schemes to obtain good raw-water quality.
Torgau Case Study<br />
The highly productive <strong>RBF</strong> scheme near Torgau in Saxony, eastern Germany, is situated within the<br />
Elbe River Basin. The basin is filled with Pleistocene deposits to a depth of 10 to 55 m that comprise<br />
interfingered glaciofluviatile sediments ranging from fine sand and silt to medium sand and gravel<br />
(Grischek et al., 1998). The deposits are overlain by Holocene river gravels (5- to 8-m thick) and<br />
meadow loam (2-m thick). The hydraulic conductivity values range from 0.6-2 × 10 –3 m per second.<br />
The Elbe River is in direct hydraulic contact with the aquifer. Riverbed clogging is low. The <strong>RBF</strong><br />
system consists of 42 vertical wells arranged in nine well groups (Figure 1) and has a capacity of<br />
150,000 m 3 /d. The distance between abstraction wells and the riverbank is about 300 m. The flow<br />
time of bank filtrate is between 80 and 300 days. The wells have a 20-m long filter placed in the<br />
lower layer of the aquifer at 30- to 50-m below ground surface. The site is well equipped with two<br />
sampling profiles along flowpaths between the river and abstraction wells, with monitoring wells<br />
in the landside catchment zone.<br />
5712<br />
5710<br />
5708<br />
5706<br />
2.3*<br />
2.0<br />
3.2<br />
2.6*<br />
3.2<br />
4.4<br />
4.1*<br />
1.7<br />
3.1<br />
I<br />
1.0<br />
1.0<br />
3.1<br />
II<br />
1.2<br />
1.9<br />
2.6<br />
2.7<br />
III<br />
1.3<br />
1.1<br />
3.0 1.0<br />
1.4 4.1<br />
0.8<br />
3.3 1.0<br />
0.9<br />
IV<br />
1.7<br />
3.3<br />
2.9<br />
2.8<br />
Mixed water<br />
DOC < 2.3 mg/L<br />
1.1<br />
0.9<br />
2.8<br />
0.8<br />
Lakes<br />
0.8<br />
1.5<br />
1.8<br />
3.3<br />
2.7<br />
4.6*<br />
2.8<br />
3.0<br />
5.1<br />
Mixed water 4.2<br />
DOC 2.3...3.0 mg/L<br />
Elbe River<br />
V VI<br />
VII<br />
1.6<br />
2.0<br />
4.3<br />
2.6<br />
5.2 5.0<br />
Mixed water 4.8<br />
DOC > 3 mg/L<br />
Legend<br />
Abstraction well<br />
3.7<br />
3.2<br />
3.0 3.3<br />
1.9<br />
1.7<br />
4571 4573 4575 4577<br />
Figure 1. Mean DOC concentration in milligrams per liter in groundwater in the catchment zone of the<br />
Torgau <strong>Water</strong>works from 1995 to 1997.<br />
The main aims of water-quality management include the maximum attenuation of organic<br />
compounds during aquifer passage and low concentrations of DOC, dissolved iron, and nitrate in<br />
raw water. Results from long-term monitoring programs and special field experiments provided an<br />
excellent database to draw conclusions on the effect of water-quality management measures. A<br />
detailed description of redox conditions and removal rates of organics is given in Grischek et al.<br />
(1998, 2000). Table 1 summarizes the effects of different management measures <strong>for</strong> the <strong>RBF</strong> site<br />
at Torgau.<br />
2.6<br />
3.2<br />
4.4<br />
4.4*<br />
Observation well with three<br />
sampling depths; mean DOC<br />
concentration in mg/L, 1995-97<br />
Affected by local infiltration<br />
of surface water<br />
VIII<br />
1 km<br />
IX<br />
1.5<br />
old branch<br />
65
66<br />
Table 1. Effect of Control Measures on Raw-<strong>Water</strong> Quality at the <strong>RBF</strong> Site in Torgau<br />
(Grischek, 2002)<br />
Control Measure Bank- Effect on Raw-<strong>Water</strong> Quality<br />
Filtrate DOC Trace NO3 – Fe 2+ NH4 + SO4 2–<br />
Portion<br />
Organics Mn 2+<br />
Continuous operation of wells<br />
with increased abstraction rate<br />
Continuous operation of wells<br />
with decreased abstraction rate<br />
Periodic operation of selected wells<br />
Preferential operation of selected<br />
wells with favorable catchment area<br />
Operation of every second or third<br />
well within a group of wells<br />
Technical measure<br />
Change of the well filter depth<br />
➚<br />
➚<br />
➚<br />
➚<br />
— — —<br />
➚<br />
➚ ➚➚➚<br />
➚ ➚ ➚<br />
➚<br />
➚<br />
➚➚➚<br />
➚<br />
— — — — —<br />
— Effect is negligible or only temporary. Concentration increase. Concentration decrease.<br />
—<br />
Readily degradable and highly adsorbable organic compounds are attenuated in the biologically<br />
active riverbed. Long flow paths and long retention times of the bank filtrate in the aquifer allow<br />
further attenuation of poorly degradable and some polar organics.<br />
A field test was done to study the effect of an increase in water abstraction resulting in a decrease<br />
in retention time of bank filtrate in the aquifer. Over a period of 1.5 years, water abstraction from<br />
five selected wells was increased by about 40 percent. Excluding some slight changes in water<br />
quality near the bank line, no significant effect due to the decrease in retention time was observed.<br />
DOC removal and denitrification along the whole flowpath were not affected (Grischek, 2002).<br />
A continuous operation of selected wells has the advantage of lower concentrations of dissolved<br />
iron in raw water. High concentrations of dissolved iron were not measured in the bank filtrate,<br />
but were measured in the landside groundwater. Switching off the abstraction wells results in<br />
groundwater flow towards the river and the transport of dissolved iron into the aquifer zone<br />
between the wells and river. When the pumps are switched on again, higher iron concentrations<br />
are observed in raw water. Furthermore, periodic well operation leads to higher well clogging.<br />
A groundwater flow and transport model has been used to simulate the change of the well filter<br />
depth and its effect on groundwater flow, especially groundwater flow from the opposite side of the<br />
river beneath the riverbed to the wells. Due to the long distance between the wells and bank line,<br />
the effect was found to be negligible. A change in the filter depth should only be considered if the<br />
aquifer consists of layers with very different hydraulic conductivities and if the well is located at a<br />
short distance from the riverbank (the distance is less than aquifer thickness).<br />
Surprisingly, the most important factor was the selection of abstraction wells according to their<br />
catchment zone <strong>for</strong> landside groundwater. Based on the dense net of observation wells, it was possible<br />
to identify zones with different concentrations of DOC and dissolved iron in landside groundwater.<br />
Figure 1 shows the range of DOC concentrations in groundwater in the catchment zones behind<br />
the wells.<br />
—<br />
—<br />
—<br />
—<br />
➚<br />
—<br />
➚ ➚➚<br />
➚ ➚<br />
➚<br />
➚<br />
➚ ➚➚➚
The mean concentration in the “mixed water” (see Figure 1) was calculated from the concentrations<br />
determined at different depths in the aquifer. Depth-dependent groundwater sampling was a<br />
key factor in understanding groundwater flow and quality changes in the catchment zone. The<br />
operation of Well Groups II to IV results in lower DOC concentration in the raw water compared<br />
to the operation of Well Groups VIII and IX. Fortunately, low DOC concentrations are associated<br />
with low dissolved iron concentrations. Thus, the advantage of appropriately selecting abstraction<br />
wells covers both parameters. Because the water-quality change <strong>for</strong> bank filtrate was very similar<br />
in all wells, an improvement of raw-water quality can be achieved mainly by the selection of wells<br />
abstracting the proportion of landside groundwater with the best quality. At the Torgau<br />
<strong>Water</strong>works, the preferential operation of these wells has already resulted in cost savings,<br />
especially <strong>for</strong> the removal of dissolved iron during the water-treatment process that requires iron<br />
sludge disposal.<br />
Summary<br />
The main aim at all <strong>RBF</strong> sites along the Elbe River is the attenuation of organic compounds and<br />
low concentrations of DOC, dissolved iron, and manganese in raw water. Long flow paths and<br />
retention times promote the attenuation of organics, but were found to have only a relatively small<br />
effect on iron and manganese concentrations. In general, the continuous pumping of selected<br />
wells should be preferred over periodic operation. At all sites, mixing ratios of bank filtrate and<br />
groundwater were found to be of main importance <strong>for</strong> the concentration of nitrate, sulfate,<br />
dissolved iron, and manganese in raw water. Due to different groundwater qualities within the<br />
whole catchment zone, a selection of wells having a catchment zone with good groundwater<br />
quality offered a significant improvement in raw-water quality. Thus, monitoring systems <strong>for</strong> <strong>RBF</strong><br />
sites should not only focus on bank filtrate, but should include observation wells landside of the<br />
abstraction wells. Due to site-specific boundary conditions, a detailed investigation of groundwater<br />
flow conditions and proportions of bank filtrate in raw water is very important <strong>for</strong> drafting<br />
effective water-quality management measures.<br />
REFERENCES<br />
Eckert, P., C. Blömer, J. Gotthardt, S. Kamphausen, D. Liebich, and J. Schubert (2000). “Correlation<br />
between the well field catchment area and transient flow conditions.” Proceedings, International Riverbank<br />
Filtration Conference, W. Jülich and J. Schubert (eds.), IAWR Rheinthemen 4, 103-113.<br />
Grischek, T., KM. Hiscock, T. Metschies, P.F. Dennis, and W. Nestler (1998). “Factors affecting denitrification<br />
during infiltration of river water into a sand and gravel aquifer in Saxony, Germany.” Wat. Res.,<br />
32(2): 450-460.<br />
Grischek, T., E. Worch, and W. Nestler (2000). “Is bank filtration under anoxic conditions feasible?”<br />
Proceedings, International Riverbank Filtration Conference, W. Jülich and J. Schubert (eds.), IAWR<br />
Rheinthemen 4, 57-65.<br />
Grischek, T. (2002). Zur Bewirtschaftung von Uferfiltratfassungen an der Elbe (Management of riverbank filtration<br />
sites along the River Elbe), Ph.D. thesis, Dresden University of Technology, <strong>Institute</strong> <strong>for</strong> Groundwater<br />
Management (in German).<br />
Heinzmann, B. (1998). “Beispiele für integriertes Management von Wasserressourcen in der Region Berlin-<br />
Brandenburg. (Examples of integrative management of water resources in the Berlin-Brandenburg region).”<br />
Wasserwirtschaft in urbanen Räumen. Schriftenreihe Wasser<strong>for</strong>schung, 3: 171-197 (in German).<br />
Koster, N., U. Willme, and K. Döhmen (1994). “Wasserwirtschaftliche Betriebsanalyse am Beispiel einer<br />
Wassergewinnungsanlage im Ruhrtal (Analysis of water management at a waterworks in the River Ruhr<br />
valley).” Wasser Abwasser Praxis, 5: 19-23 (in German).<br />
67
68<br />
Luckner, L., and W. Nestler (1982). “Zur Methodik des Aufbaus und der Nutzung von Kontroll- und<br />
Steuerungsprogrammen von Grundwasserfassungen (On the methodology of structuring and use of control<br />
programs <strong>for</strong> groundwater abstraction wells).” Wasserwirtschaft-Wassertechnik, 32(7): 219-223 (in German).<br />
Nestler, W., W. Walther, F. Jacobs, R. Trettin, and K. Freyer (1998). Wassergewinnung in Talgrundwasserleitern<br />
der Elbe (<strong>Water</strong> production in alluvial aquifers along the River Elbe), UFZ-<strong>Research</strong> Report 7 (in German).<br />
Schöttler, U., and H. Sommer (1992). “Optimierung einer Wassergewinnungsanlage mit Hilfe der Modellrechnung<br />
und ihre Auswirkungen auf die Grundwassergüte (Optimisation of a water abstraction system with<br />
the help of model calculations and the effects on groundwater quality).” Steuerung in der Wasserwirtschaft,<br />
DFG-Forschungsbericht, VCH-Verlagsges., Weinheim, 178-195 (in German)<br />
Schubert, J. (1999). “Riverbank filtration — Field studies, modeling, monitoring.” Proceedings, International<br />
Riverbank Filtration Conference, 4-6 Nov 1999, Louisville, Kentucky, 39-42.<br />
Sontheimer, H. (1991). Trinkwasser aus dem Rhein? Bericht über ein Verbund<strong>for</strong>schungsvorhaben zur Sicherheit<br />
der Trinkwassergewinnung aus Rheinuferfiltrat (Drinking water from the River Rhine? Report on a collaborative<br />
research project on safety of drinking water production from bank filtrate from the River Rhine), Academia Verlag,<br />
Sankt Augustin, Germany (in German).<br />
THOMAS GRISCHEK has 15 years of research experience in the field of riverbank<br />
filtration. He published eight papers on riverbank filtration in refereed journals as principal<br />
author and has contributed to more than 10 papers as co-author. His main research<br />
interests are the interaction of groundwater and surface water, especially riverbank<br />
filtration and artificial groundwater recharge, groundwater management, and diffusing<br />
groundwater pollution. Currently, he is Professor of <strong>Water</strong> Science at the University of<br />
Applied Sciences Dresden. Prior to this position, Grischek spent 8 years as a <strong>Research</strong><br />
Associate <strong>for</strong> the <strong>Institute</strong> <strong>for</strong> Groundwater Management at the Dresden University of Technology, as well<br />
as at the University of Applied Sciences Dresden. In addition, he also taught at the <strong>Institute</strong> <strong>for</strong> <strong>Water</strong><br />
Chemistry at the Dresden University of Technology. In 2003, he worked as a Referee at the Saxon State<br />
Agency <strong>for</strong> Environment and Geology. Grischek studied <strong>Water</strong> Management in the Department of <strong>Water</strong><br />
Sciences at Dresden University of Technology, where he graduated as a diploma engineer. He also received<br />
a Ph.D. in Environmental Engineering from the Dresden University of Technology in 2002, where he<br />
researched the “Management of Riverbank-Filtration Sites along the Elbe River.”
Session 5: Dynamics<br />
Using Models to Predict Filtrate Quality<br />
at Riverbank-Filtration Sites –<br />
What Is the Adequate Level of Modeling?<br />
Chittaranjan Ray, Ph.D., P.E.<br />
University of Hawaii at Mañoa<br />
Honolulu, Hawaii<br />
Henning Prommer, Ph.D.<br />
Delft University of Technology<br />
Delft, The Netherlands, and<br />
Center <strong>for</strong> Groundwater Studies<br />
Perth, Australia<br />
Prof. Dr.-Ing. Thomas Grischek<br />
University of Applied Sciences<br />
Dresden, Germany<br />
<strong>RBF</strong> is an efficient and low-cost treatment technology <strong>for</strong> drinking-water production that has<br />
been in use <strong>for</strong> more than a century in Europe and nearly half a century in the United States.<br />
There has been a recent surge in the design and construction of <strong>RBF</strong> facilities in the United States<br />
due to the potential to receive 0.5- to1.0-log removal credit <strong>for</strong> protozoa Cryptosporidium filtration<br />
under the upcoming Enhanced Surface <strong>Water</strong> Treatment Rule (USEPA, 2002). While most small<br />
water utilities use <strong>RBF</strong> as the sole treatment system (with the exception of disinfection), most<br />
medium to large utilities use <strong>RBF</strong> as a pretreatment system, which helps process per<strong>for</strong>mance in<br />
the final treatment system.<br />
The design and operation of <strong>RBF</strong> systems are facilitated by the use of flow and transport models<br />
<strong>for</strong> estimating yield and water quality of the filtrate; however, the level of the modeling ef<strong>for</strong>t can<br />
vary depending upon the utility, data availability, and problems to be solved. While water-quality<br />
modeling is the primary focus, flow simulation is the first step prior to undertaking transport<br />
simulations. In this paper, we present various scenarios involved in flow and transport modeling<br />
at <strong>RBF</strong> sites and discuss what level of modeling is adequate <strong>for</strong> what problems.<br />
Flow Simulations<br />
Most utilities undertake some sort of flow simulation to estimate yield from <strong>RBF</strong> wells. These<br />
simulations are broadly divided into (a) analytical and (b) numerical models. Analytical models<br />
(Ferris et al., 1962; Hantush, 1959) assume that the river fully penetrates the aquifer and there is<br />
no additional resistance to flow at the river-aquifer interface. Hantush (1959) also presented a<br />
simple procedure to handle partial penetration of the river by moving the recharge boundary some<br />
distance away from the actual location with respect to the pumping well. Other analytical models,<br />
Correspondence should be addressed to:<br />
Chittaranjan Ray, Ph.D., PE<br />
Associate Professor<br />
Department of Civil & Environmental Engineering<br />
University of Hawaii at Mañoa • 2540 Dole Street, 383 Holmes Hall • Honolulu, Hawaii 96822 USA<br />
Phone: (808) 956-9652 • Fax: (808) 956-5014 • Email: cray@hawaii.edu<br />
69
70<br />
as described by Dillon and Ligett (1983) and Wilson (1993), include transient effects and<br />
clogging. Their application is mostly limited to two-dimensional problems. Three-dimensional<br />
flow and hydraulic and material heterogeneity cannot be easily handled by these methods. Conrad<br />
and Beljin (1996) provide a result comparison from the application of analytical and numerical<br />
models <strong>for</strong> different sites.<br />
Nowadays, numerical models are most commonly used to simulate stream-aquifer interaction.<br />
MODFLOW (Harbaugh et al., 2000) and MODPATH (Pollock, 1994) are some of the commonly<br />
used models <strong>for</strong> estimating hydraulic heads and path lines of neutrally buoyant particles.<br />
MODFLOW can handle material heterogeneity and various boundary conditions. The River and<br />
Stream/Aquifer Interaction packages are typically used in MODFLOW to examine the interaction<br />
between surface water and groundwater. In the River package, the hydraulic heads in the aquifer<br />
and the stage of water in the river are prescribed. Certain block-centered grids serve as the river.<br />
If the hydraulic head in the aquifer is lower than river stage, then the flow is from the river to the<br />
aquifer and vice versa. The hydraulic conductivity of riverbed material controls the flow into or<br />
out of the river. In contrast to the River package, where the hydraulic heads are specified, the flow<br />
is routed through the channel in the Stream/Aquifer Interaction package using the Manning<br />
equation, channel geometry, roughness, and other parameters.<br />
MODPATH is used to conduct advective tracking of neutrally buoyant particles in the flow field using<br />
<strong>for</strong>ward or reverse particle tracking techniques. For example, reverse particle tracking from the screen<br />
zone of a well can determine the areas contributing to flow to a well. Similarly, <strong>for</strong>ward tracking of a<br />
set of particles from a riverbed would indicate if any of the particles will be captured by pumping wells<br />
located on riverbanks. Path lines can be delineated <strong>for</strong> steady and transient flow conditions.<br />
In MODPATH simulations, the hydraulic gradient and well location can play important roles. For<br />
example, in a stream-aquifer simulation, the path lines <strong>for</strong> a single well can be seen in Figure 1.<br />
This figure shows the impact of the placement of one or more wells on the flow field. In the left<br />
side of this figure, a portion of the flow to the pumping well originates from the river and the rest<br />
comes from upgradient locations; however, if a second well is installed to the northwest of this well<br />
(figure to the right), a hydraulic divide is created. As apparent, most of the flow to this new well<br />
comes from the aquifer.<br />
In most river-aquifer simulation models, the nearest riverbank or the mid-point of the river is often<br />
considered as the model boundary. This assumption may hold as long as the river is a significant<br />
Figure 1. Flow fields <strong>for</strong> one and two pumping wells located on the bank of the Illinois River.
source of induced infiltration to the well; however, if the stream bottom is constituted with low<br />
permeability material and the pumping rate is high, a portion of the pumped water could come<br />
from the other side of the river. Figure 2 shows a case study <strong>for</strong> the <strong>RBF</strong> wells of Meissen-<br />
Siebeneichen, Germany, in which the path lines go past the mid-point of the river and extend to<br />
the other side. Cross-sections A-A and B-B show that only particles starting in the upper layer of<br />
the aquifer actually reach the river. All other particles are turned away near the river, but travel<br />
towards the production borehole. Thus, high nitrate concentrations in the lower layer of the<br />
aquifer between the river and production borehole can be explained by groundwater flow from the<br />
opposite side of the river. Further, the impact of riverbed clogging on the proportion of pumped<br />
river water was investigated using the model. Under site-specific geological conditions, observed<br />
clogging of the river bed (1 × 10 –5 m per second of the 0.1-m thick clogging layer) causes a<br />
decrease in <strong>RBF</strong> by only about 5 percent, as compared to the “no clogging” case. In general,<br />
geological anisotropy was found to have a stronger effect on the proportion of groundwater flow<br />
beneath the riverbed than riverbed clogging. Even under conditions where riverbed clogging does<br />
not occur, groundwater can flow from the opposite side of the river beneath the riverbed towards<br />
the production boreholes (Grischek et al., 2002).<br />
Model Grid<br />
98.0<br />
98.1<br />
98.2<br />
A A<br />
98.3<br />
The Meissen-Siebeneichen example demonstrates that if groundwater on the other side of the river<br />
contained dissolved contaminants, the contaminants could possibly appear in the filtrate. Thus, while<br />
designing <strong>RBF</strong> systems, land use on the other bank of the river should be considered and potential<br />
pollution sources must be identified. The calibration of the flow model heavily depends on the<br />
accuracy of the hydraulic conductivity of riverbed material. The heterogeneity of riverbed material<br />
can produce uneven leakage through the riverbed. Further, hydraulic parameters of the riverbed<br />
material can change depending on the flow status of the river due to scouring and sedimentation<br />
process. To date, there is no reliable and verifiable method available to estimate riverbed hydraulic<br />
conductivity of large rivers in situ and to estimate its variability with the flow regime of the river. An<br />
overview on methods and recent developments is given by Macheleidt et al. (2002).<br />
Single-Species Transport Modeling<br />
To estimate the amount of river water entering a well, mass balance methods are generally<br />
employed. In such methods, the concentrations of a conservative chemical in the river, aquifer,<br />
and well are used to estimate the mass fractions coming from the river and aquifer:<br />
C well = xC river +(1 – x)C aquifer<br />
98.0<br />
98.1<br />
98.2<br />
98.4<br />
Flow Paths (Plan View)<br />
98.3<br />
Adjacent<br />
Granitic<br />
Rocks<br />
98.4<br />
(Equation 1)<br />
98.5<br />
98.5<br />
Elbe River<br />
Legend<br />
Production Well 500 m<br />
–98– Piezometric Contour<br />
in Meters Above Mean Sea Level<br />
98.6<br />
A Flow Path A<br />
(Side View)<br />
Well<br />
Figure 2. Location map of a <strong>RBF</strong> site of the Meissen-Siebeneichen <strong>Water</strong>works in Germany (after Grischek et al., 2002).<br />
71
72<br />
In Equation 1, the fraction of river water (x) can be calculated if the three concentrations are known.<br />
Similarly, if x is known from prior investigations, the concentration of a conservative chemical in<br />
the filtrate may be estimated; however, the equation is only valid <strong>for</strong> steady-state conditions. In other<br />
words, concentrations in the river and aquifer do not change with time. Also, it is assumed that<br />
infiltrating water from the river at a given moment in time is of the same quality as that reaching the<br />
well and that the quality of the bank filtrate and groundwater does not change over the whole<br />
thickness of the aquifer. Such assumptions hamper the use of Equation 1 <strong>for</strong> spill events, even <strong>for</strong><br />
conservative chemicals. Transient effects such as lag time (or travel time) issues are not considered.<br />
Further, Equation 1 cannot be used <strong>for</strong> non-conservative contaminants that are subject to<br />
degradation, sorption, or other reactions. For such contaminants, knowing the concentrations in the<br />
river and aquifer, as well as the mass fraction of the water derived from the river, one would not be<br />
able to predict the concentration at the well. For transient problems, the application of numerical<br />
models is typically required. In those, the advection-dispersion equation is solved with or without<br />
chemical reactions, depending upon the type of contaminants. The advection-dispersion equation<br />
can be solved analytically if the flow field is steady and reaction terms are linear. In cases where the<br />
use of three-dimensional numerical models is beyond the ability of water utilities, simple<br />
one-dimensional analytical models can be used to estimate chemical concentrations in pumped<br />
water, especially from shock events or chemical spills. Mälzer et al. (2003) used the analytical<br />
solution to the one-dimensional advection-dispersion equation to estimate the concentration of a<br />
chemical reaching a well. In their approach, they approximated the flow field between the river and<br />
aquifer using five flow tubes. For each tube, they solved the advection-dispersion equation with<br />
(sorption and first order degradation) to calculate the concentration at Observation Well M1<br />
(Figure 3); however, the assumption of steady flow through these flow tubes may not be valid <strong>for</strong> all<br />
cases. Further, estimating the concentrations <strong>for</strong> a contaminant at the pumping well may not be easy<br />
without knowing the amount of groundwater contribution to the well.<br />
Meters Above Sea Level<br />
35<br />
30<br />
25<br />
20<br />
15<br />
10<br />
Vertical Section at Wittlaer<br />
Rhine River<br />
Aquifer<br />
For transient three-dimensional simulations, MT3D/MT3DMS (Zheng and Wang, 1999) is<br />
probably the most common transport simulator that is used in conjunction with MODFLOW.<br />
MT3D uses the same grid of MODFLOW and solves advection-dispersion equation, along with<br />
sorption and degradation reactions, to estimate the concentration of a contaminant at a given<br />
location in the model domain. For estimating the concentration at the pumping well, a flow model<br />
must be built and calibrated to provide a good description of the velocity field. As mentioned<br />
earlier, the hydraulic conductivity of bed material at the river-aquifer interface can significantly<br />
23 m<br />
27 m<br />
Figure 3. Vertical section of the aquifer at Wittlaer between the Rhine River<br />
and Observation Well M1 and a simplified model assumption of five<br />
parallel tubes (after Mälzer et al., 2003).<br />
M1<br />
High<br />
Low
affect the amount of contaminant entering from the river to the aquifer and, eventually, to the<br />
pumping well(s). For flow fields affected by pumping wells, the assumption of equilibrium sorption<br />
(in contrast to kinetically controlled sorption) near the screen zone may not be valid due to fast<br />
travel rates of water particles. Ray et al. (2002) simulated the impact of riverbed/bank material<br />
properties on filtrate quality during a flood pass-through event at a <strong>RBF</strong> facility on the banks of the<br />
Illinois River. They used equilibrium sorption and first order degradation from literature-reported<br />
data. Also, in certain simulations, they considered the contaminant to be conservative. This<br />
assumption enveloped the two extremes. For atrazine, with a highly conductive bank, the<br />
concentrations in groundwater and the pumped well are slightly attenuated when the effects of<br />
sorption and degradation are considered (Figure 4A); however, without sorption and degradation,<br />
the concentrations can be close to that found in river water (Figure 4B).<br />
Atrazine Concentration, µg/L<br />
Atrazine Concentration, µg/L<br />
4<br />
3<br />
2<br />
1<br />
0<br />
4<br />
3<br />
2<br />
1<br />
0<br />
(A) with sorption and decay<br />
River water<br />
SP-1<br />
Caisson<br />
SP-3 is below detection limit<br />
0 20 40 60 80 100<br />
River water<br />
Days Since Start of Simulation<br />
(B) without sorption and decay<br />
SP-3<br />
SP-1<br />
Caisson<br />
0 20 40 60 80 100<br />
Days Since Start of Simulation<br />
Figure 4. Atrazine transport from the river to the aquifer and the pumping well<br />
with and without sorption and decay reactions <strong>for</strong> a highly conductive bed<br />
and bank (after Ray et al., 2002).<br />
73
74<br />
When the hydraulic conductivity of the bed and bank are reduced to moderate values (see sets<br />
A-A and B-B in Ray et al., 2002), there is a significant attenuation of atrazine at the collector well<br />
caisson (Figure 5). It is clear that the transient effects call <strong>for</strong> the use of complex three-dimensional<br />
models to account <strong>for</strong> the spatial and temporal variability of streamlines. Based on hydraulic head<br />
measurements and water-quality observations, the calibration of flow and transport models is<br />
essential <strong>for</strong> an accurate description and prediction of the fate of individual or multiple chemicals.<br />
The current version of the standard MT3DMS code (Zheng and Wang, 1999) accounts <strong>for</strong> the<br />
transport of multiple species and a small range of basic reactions; however, <strong>for</strong> cases where<br />
chemical species are assumed to interact in a more complex manner, MT3DMS-based packages<br />
such as RT3D (Clement, 1997) or PHT3D (Prommer et al., 2003a), which account <strong>for</strong> a greater<br />
variety of biogeochemical processes, are available.<br />
Atrazine Concentration, µg/L<br />
4<br />
3<br />
2<br />
1<br />
0<br />
River water<br />
Caisson (set B-B)<br />
Caisson (set A-A)<br />
Multi-Species and Multi-Component Transport Modeling<br />
No reaction – filtrate<br />
With reaction – filtrate<br />
River water<br />
0 20 40 60 80 100<br />
Days Since Start of Simulation<br />
Figure 5. Atrazine transport from the river to the aquifer and the pumping well <strong>for</strong> a case<br />
with low permeability riverbed/bank (after Ray et al., 2002).<br />
In those scenarios where the reaction progress of a dissolved chemical strongly depends on<br />
the concentration of one or more other dissolved species, reactive multi-species models can<br />
typically provide a better process description. This is particularly important if models are used in a<br />
predictive mode. Furthermore, if water-quality changes are additionally affected by water-sediment<br />
interactions, such as mineral dissolution/precipitation and/or ion-exchange reactions, a reactive<br />
multi-component transport model might need to be applied to explain specific field observations.<br />
Multi-component models typically use an extensive reaction database. Until recently, those<br />
databases were (in most cases) confined to the definition of thermodynamic equilibrium reactions,<br />
which made those models only applicable to systems where (all) reactions proceed relatively fast in<br />
relation to groundwater flow velocity (local equilibrium assumption); however, in most of the<br />
recently published models, a wide range of different kinetic reactions and processes can be defined.<br />
Models with those capabilities can be used to study complex process interactions that lead to non-
intuitive system behavior. In the case of <strong>RBF</strong>, such examples might be:<br />
• The reductive dissolution of iron oxides by DOC and the resulting release of sorbed<br />
heavy metals.<br />
• The assessment of pesticide mobility during flood events.<br />
Denitrification, degradation of pesticides, and dissolution of minerals are closely linked to DOC,<br />
a key component of river water whose concentration can vary depending on the season and flow<br />
events. Because of the reaeration process, river water is oxygenated compared to groundwater.<br />
During the induced infiltration process, oxygen-rich river water enters the aquifer, as does DOC.<br />
On its travel path, DOC from river water is oxidized and either mineralizes completely or is<br />
trans<strong>for</strong>med to intermediates through bacterial catalysis, together with the organic carbon that is<br />
perhaps naturally abundant in the aquifer as sediment-bound organic matter. Oxygen in<br />
the invading water is used as an electron acceptor in the process. Normally, there is sufficient<br />
carbon available <strong>for</strong> microbial use; however, oxygen can become in short supply along the flow<br />
path. Once microbes consume the oxygen, an anoxic zone develops where the nitrate of the<br />
infiltrating river water and groundwater is used as a substitute electron acceptor. This leads to the<br />
reduction of nitrate along the flow path. Once nitrate is also depleted, thermodynamically less<br />
favorable oxidized iron and manganese minerals and/or sulfate might act as alternative electron<br />
acceptors. Note that the simulation of the oxidation of one or multiple organic substrates using a<br />
sequence of electron acceptors is routinely applied in the field of bioremediation modeling, mainly<br />
where the transport and natural attenuation of oxidizable organic contaminants is simulated<br />
(Barry et al., 2002).<br />
The simultaneous simulation of <strong>RBF</strong>-typical denitrification and mineral dissolution reactions can<br />
be handled by a number of existing codes (<strong>for</strong> examples, see Table 1). One such code is EASY-<br />
LEACHER (Stuyfzand and Lüers, 2000), which was used to simulate reactions along a transect of<br />
the Torgau <strong>RBF</strong> site on the Elbe River in Germany. EASY-LEACHER is a two-dimensional<br />
reactive transport code in EXCEL spreadsheet, combining chemical principles with empirical<br />
rules in an expert system. The code was found useful to attain a first estimate of water quality in<br />
the production well <strong>for</strong> a <strong>RBF</strong> site where the operation of wells is started and <strong>for</strong> an easy<br />
calculation of different boundary conditions. In contrast to EASY LEACHER, which uses a<br />
collection of (one-dimensional) flow tubes to account <strong>for</strong> the transport of chemicals, the<br />
FEREACT model (see Tebes-Stevens et al., 1998; Tebes-Stevens and Valocchi, 2000) is a<br />
Table 1. Examples of Reactive Multi-Component Transport Models<br />
Model Reference<br />
CRUNCH Steefel (2001)<br />
EASY-LEACHER Stuyfzand and Lüers (2000)<br />
FEREACT Tebes-Stevens et al. (1998)<br />
PHT3D Prommer et al. (2003a)<br />
PHAST Parkhurst et al. (1995)<br />
MIN3P Mayer (1999)<br />
TBC Schäfer et al. (1998)<br />
HBGC123D Salvage and Yeh (1998)<br />
75
76<br />
two-dimensional, finite element-based transport model that can simulate biodegradation and<br />
geochemical reactions along, <strong>for</strong> instance, a vertical transect of the river-aquifer interface. Although<br />
the model cannot handle the true three-dimensional (and perhaps transient) flow dynamics<br />
experienced at a <strong>RBF</strong> site, it accounts <strong>for</strong> all typical geochemical and microbial reactive processes.<br />
An example <strong>for</strong> a fully three-dimensional model is the MODFLOW/MT3DMS-based code<br />
PHT3D (Prommer, 2002, Prommer et al., 2003a). It combines the previously mentioned<br />
MT3DMS (Zheng and Wang, 1999) with PHREEQC-2 (Parkhurst and Appelo, 1999), whereby<br />
the <strong>for</strong>mer solves the advection-dispersion equation and the latter accounts <strong>for</strong> all geochemical<br />
reactions. Prommer et al. (2003b) recently applied the model to simulate the transport and<br />
reactive processes that affect the fate of atrazine near a <strong>RBF</strong> scheme during a flood event.<br />
For that modeling study, the hydrogeological setting and hydrological characteristics described by<br />
Ray et al. (2002) were used to demonstrate the potential influence of a microbial lag effect and<br />
the redox-dependency of atrazine degradation on abstraction water quality during the flood event.<br />
They considered kinetically controlled mineralization of DOC, dissolution of sediment-bound<br />
organic matter, growth and decay of atrazine degraders, and microbially mediated atrazine<br />
degradation. During the flood event, DOC from the river water modifies the redox patterns in the<br />
aquifer along the flow path to the well screens. The simulations demonstrated that with the flood,<br />
the concentration of atrazine reaches a peak in a similar pattern to that found in floodwater and<br />
the concentration in groundwater remains high; however, once atrazine degraders are active, the<br />
concentration drops significantly (Figure 6, Case 1). On the other hand, if the DOC of the river<br />
water is somewhat increased, this leads to the <strong>for</strong>mation of an anoxic zone, which promotes<br />
denitrification, but inhibits atrazine degradation (Figure 6, Case 2).<br />
Head (m)<br />
142<br />
140<br />
138<br />
136<br />
134<br />
<strong>Water</strong> level<br />
Concluding Remarks<br />
River<br />
Obs. Pt.<br />
132<br />
0 50<br />
Time (days)<br />
100<br />
C (mol/l)<br />
C (mol/l)<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
x 104<br />
5<br />
4<br />
3<br />
2<br />
1<br />
x 107<br />
1<br />
Oxygen<br />
reactive case 1<br />
reactive case 2<br />
0<br />
0 50 100<br />
Atrazine<br />
conservative<br />
reactive case 1<br />
reactive case 2<br />
0<br />
0 50<br />
Time (days)<br />
100<br />
0<br />
0 50<br />
Time (days)<br />
100<br />
It is apparent that the level of modeling can range from a very simple mass balance to very<br />
complex studies that involve transient flow and biogeochemical reactions. The objective of the<br />
modeling study, as well as the type, quantity, and quality of data, will dictate what level of<br />
sophistication is needed. For example, if yield determination is the primary purpose, MODFLOW<br />
simulations will mostly be adequate. Preliminary investigations to delineate the sources of water<br />
1.5<br />
1<br />
0.5<br />
x Nitrate<br />
104<br />
8<br />
6<br />
4<br />
2<br />
0<br />
0 50 100<br />
x 108<br />
2<br />
reactive case 1<br />
reactive case 2<br />
Atrazine degraders<br />
reactive case 1<br />
reactive case 2<br />
Figure 6. Schematic representation of the 3-D model, hydraulic heads, and selected simulated concentrations<br />
of oxygen, nitrate, atrazine, and atrazine degraders.
entering the well can be carried out with MODPATH or comparable particle tracking tools;<br />
however, the accuracy of the simulation strongly depends on how well the true values of hydraulic<br />
conductivity of riverbed/bank materials are represented in the models. Especially <strong>for</strong> transient<br />
simulations, accurate estimates of riverbed and bank hydraulic conductivity are critical. For<br />
transport simulations, simple mixing models (as used by many utilities) may not provide good<br />
estimates of contaminant concentrations at <strong>RBF</strong> wells due to lag times involved, transient flow,<br />
and heterogeneity issues. Models using MT3D <strong>for</strong> simulating the transport of single or multiple<br />
contaminants may be adequate <strong>for</strong> specific problems (<strong>for</strong> example, worst-case estimates); however,<br />
the assumption of equilibrium reactions near well screens may not be valid. On the other hand,<br />
<strong>for</strong> kinetic reactions, the availability of appropriate rate parameters might not always be<br />
warranted. Finally, biogeochemical models such as PHT3D are most comprehensive and are ideal<br />
to study both steady and transient effects; however, the input data sets <strong>for</strong> such models can be<br />
extensive, and field data sets that underpin the simulations may not always be readily available.<br />
For such complex models, input parameters are rarely found in literature and are very much sitespecific.<br />
On the other hand, the determination of such parameters at the laboratory-scale requires<br />
both significant time and ef<strong>for</strong>t and, additionally, the usefulness of the results <strong>for</strong> field-scale<br />
simulations might still not be warranted due to scale-issues. Furthermore, the application of such<br />
models typically requires users with advanced modeling skills and experience in the areas of<br />
geochemistry and (groundwater) hydrology. After all, the complexity found in some of the models<br />
is only a reflection of the complexity and diversity found in nature and natural materials. To find<br />
and make use of adequate simplifications should be part of the development of the (site-specific)<br />
conceptual model. Each complex numerical model can also be used as a “simpler” model by fixing<br />
those parameters that were excluded in the corresponding simple model. Also, it must be noted<br />
that the parameters used in simpler models are by far not less site-specific than those of more<br />
complex models.<br />
A typical example of the “how much complexity is adequate” question is DOC, since it consists<br />
of thousands of single organic compounds with different biodegradation behavior. The question is<br />
how many different fractions (concerning biodegradability and sorption behavior) should be<br />
identified and used in the model to be adequate <strong>for</strong> the selected level of model discretization,<br />
reactions included, and kinetic parameters. Based on the application of different models <strong>for</strong><br />
chosen field-sites and qualified sensitivity analyses (benchmarking), a guideline should be<br />
developed to propose the right type of model <strong>for</strong> a specific task. The development of reduced<br />
models, such as the spreadsheet program EASY-LEACHER (Stuyfzand and Lüers, 2000), and the<br />
improvement of multi-species reaction models should be carried out simultaneously.<br />
Furthermore, <strong>for</strong> detailed modeling, the hydraulic component must be able to handle scour,<br />
deposition, high water levels, flooding, etc. For detailed transport and reaction modeling, varied<br />
properties of the riverbed material and aquifer material (e.g., organic carbon content, enzymatic<br />
activity) should be considered.<br />
Finally, we must mention that besides the widely known models mentioned in Table 1, there are<br />
many other models that have been applied successfully <strong>for</strong> the simulation of processes at <strong>RBF</strong> sites<br />
<strong>for</strong> the specific evaluation purposed.<br />
77
78<br />
REFERENCES<br />
Barry, D.A., H. Prommer, C.T. Miller, P. Engesgaard, and C. Zheng (2002). “Modeling the fate of oxidisable<br />
organic contaminants in groundwater.” Adv. <strong>Water</strong> Resour., 25: 945-983.<br />
Clement, T.P. (1997). RT3D - A Modular Computer Code <strong>for</strong> Simulating Reactive Multi-species Transport in 3-<br />
Dimensional Groundwater Aquifers, Battelle Pacific Northwest <strong>National</strong> Laboratory <strong>Research</strong> Report, PNNL-<br />
SA-28967.<br />
Conrad, L.P., and M.S. Beljin (1996). “Evaluation of an induced infiltration model as applied to glacial<br />
aquifer systems.” Wat. Resour. Bull., 32(6): 1,209-1,220.<br />
Dillon, P.J., and J.A. Liggett (1983). “An ephemeral stream-aquifer interaction model.” Wat. Resour. Res.,<br />
19(3): 621-626.<br />
Ferris, J.G., D.B. Knowles, R.H. Brown, and R.W. Stallmann (1962). Theory of Aquifer Tests, U.S. Geological<br />
Survey <strong>Water</strong> Supply Paper 1536-E, Government Printing Office, Washington, D.C.<br />
Grischek, T., D. Schoenheinz, and W. Nestler (2002). “Unexpected groundwater flow beneath the River Elbe at the<br />
bank filtration site Meissen, Germany.” <strong>Water</strong> resources and environment research, ICWRER 2002, Vol. 1: 116-120.<br />
Hantush, M.S. (1959). “Analysis of data from pumping wells near a river.” Jour. Geophys. Res., 64(11): 1,921-1,931.<br />
Harbaugh, A.W., E.R. Banta, M.C. Hill, and M.G. McDonald (2000). MODFLOW-2000, the U.S. Geological<br />
Survey modular ground-water model — User guide to modularization concepts and the Ground-<strong>Water</strong> Flow Process,<br />
U.S. Geological Survey Open-File Report 00-92, 121 p.<br />
Macheleidt, W., W. Nestler, and T. Grischek (2002). “Determination of hydraulic boundary conditions <strong>for</strong><br />
the interaction between surface water and groundwater.” Sustainable groundwater development, Geological<br />
Society, London, Special Publ. 193: 235-243.<br />
Mälzer, H., J. Schubert, R. Gimbel, and C. Ray (2003). “Effectiveness of riverbank filtration sites to mitigate<br />
shock loads.” Riverbank Filtration: Improving Source <strong>Water</strong> Quality, Kluwer Academic Publishers, Dordrecht,<br />
The Netherlands.<br />
Mayer, K.U. (1999). A numerical model <strong>for</strong> multicomponent reactive transport in variably saturated porous media,<br />
Ph.D. thesis, University of <strong>Water</strong>loo, <strong>Water</strong>loo, Ontario, Canada.<br />
Parkhurst, D.L., P. Engesgaard, and K.L. Kipp (1995). “Coupling the geochemical model PHREEQC with a<br />
3D multi-component solute transport model.” Fifth Annual V.M. Goldschmidt Conference, Penn State<br />
University, University Park Pennsylvania, USA, May 1995.<br />
Parkhurst, D.L., and C.A.J. Appelo (1999). User’s guide to PHREEQC - A computer program <strong>for</strong> speciation,<br />
reaction-path, 1D-transport, and inverse geochemical calculations: Technical Report 99-4259, U.S. Geological<br />
Survey <strong>Water</strong>-Resources Investigations Report.<br />
Pollock, D.W. (1994). User’s Guide <strong>for</strong> MODPATH/MODPATH-PLOT, Version 3: A particle tracking postprocessing<br />
package <strong>for</strong> MODFLOW, the U.S. Geological Survey finite-difference ground-water flow model,<br />
U.S. Geological Survey Open-File Report 94-464.<br />
Prommer, H. (2002). PHT3D: A Reactive Multicomponent Transport Model <strong>for</strong> Saturated Media, User’s Manual<br />
Version 1.0, Contaminated Land Assessment and Remediation <strong>Research</strong> Centre, The University of Edinburgh, UK,<br />
http://www.pht3d.org.<br />
Prommer, H., D.A. Barry, and C. Zheng (2003a). “PHT3D - A MODFLOW/MT3DMS based reactive multicomponent<br />
transport model.” Ground <strong>Water</strong>, 42(2): 247-257.<br />
Prommer, H., J. Greskowiak, P.J. Stuyfzand, and C. Ray (2003b). “Geochemical transport modeling of water<br />
quality changes during managed artificial recharge.” MODFLOW and More 2003: Understanding through<br />
Modeling, Proceedings of the International Ground <strong>Water</strong> Modeling Conference, Golden, Colorado USA,<br />
16-19 September 2003.
Ray, C., T.W. Soong, Y.Q. Lian, and G.S. Roadcap (2002). “Effect of flood-induced chemical load on filtrate<br />
quality at bank filtration sites.” J. Hydrol., 266: 235-258.<br />
Salvage, K.M., and G.T. Yeh (1998). “Development and application of a numerical model of kinetic and<br />
equilibrium microbiological and geochemical reactions (BIOKEMOD).” J. Hydrol., 209(1-4): 27-52.<br />
Schäfer, D., W. Schäfer, and W. Kinzelbach (1998). “Simulation of processes related to biodegradation of<br />
aquifers 1. structure of the 3D transport model.” J. Contam. Hydrol., 31(1-2): 167-186.<br />
Steefel, C.I. (2001). GIMRT, Version 1.2: Software <strong>for</strong> Modeling Multicomponent, Multidimensional Reactive<br />
Transport. Users Guide, Technical Report UCRL-MA-143182, Lawrence Livermore <strong>National</strong> Laboratory,<br />
Livermore, Cali<strong>for</strong>nia, 2001.<br />
Stuyfzand, P.J., and F. Lüers (2000). “Modelling the quality changes upon artificial recharge and bank<br />
infiltration.” Principles and user`s guide of EASY-LEACHER 4.5, KIWA-report SWI 99.199.<br />
Tebes-Stevens, C.L., and A.J. Valocchi (2000). “Calculation of reaction parameter sensitivity coefficients in<br />
multicomponent subsurface transport models.” Adv. <strong>Water</strong> Resour., 23: 591-611.<br />
Tebes-Stevens, C.L., A.J. Valocchi, J.M. VanBriesen, and B.E. Rittmann (1998). “Multicomponent transport<br />
with coupled geochemical and microbiological reactions: Model description and example simulations.”<br />
J. Hydrol., 209: 8-26.<br />
United States Environmental Protection Agency (2002). Long-Term 1 Enhanced Surface <strong>Water</strong> Treatment<br />
Rule, Final Rule. Federal Register, 67:9:1812 (January 14, 2002).<br />
van Breukelen, B.M., C.A.J. Appelo, and T.N. Oltshoorn (1998). “Hydrogeochemical transport modeling of<br />
24 years of Rhine water infiltration in the dunes of the Amsterdam water supply.” J. Hydrol., 209: 281-296.<br />
Wilson, J.L. (1993). “Induced infiltration in aquifers with ambient flow.” Wat. Resour. Res., 29(10): 3,503-3,512.<br />
Zheng, C., and P.P. Wang (1999). MT3DMS: A modular three-dimensional multispecies model <strong>for</strong> simulation of<br />
advection, dispersion and chemical reactions of contaminants in groundwater systems, Documentation and User’s<br />
Guide, Contract Report SERDP-99-1, U.S. Army Engineer <strong>Research</strong> and Development Center, Vicksburg, MS.<br />
CHITTARANJAN RAY is an Associate Professor in the Department of Civil &<br />
Environmental Engineering and an Associate <strong>Research</strong>er with the <strong>Water</strong> Resources<br />
<strong>Research</strong> Center at the University of Hawaii at Mañoa. His current research interests<br />
include riverbank filtration, pesticides in drinking-water wells, and the flow and transport<br />
of pathogens and chemicals in saturated/unsaturated media. Over the past 2 years, he has<br />
edited two books on riverbank filtration and written a monograph on pesticides in<br />
domestic wells. Among his honors, he is a recipient of the Fulbright faculty scholarship <strong>for</strong><br />
conducting riverbank filtration-related research in Nepal, India, and Bangladesh. Prior to joining the<br />
University, Ray worked as a staff engineer with the groundwater consulting firm of Geraghty & Miller, Inc.<br />
(now called Arcadis Geraghty & Miller) and in the Groundwater Section of Illinois State <strong>Water</strong> Survey,<br />
conducting various groundwater quantity and quality investigations, including bank filtration. Ray received<br />
a B.S. in Agricultural Engineering from Orissa University of Agriculture and Technology in India, an M.S.<br />
in Agricultural Engineering from the University of Manitoba in Canada, an M.S. in Civil Engineering from<br />
Texas Tech University, and Ph.D. Civil and Environmental Engineering from the University of Illinois at<br />
Urbana-Champaign.<br />
79
Session 5: Dynamics<br />
The 100-Year Flood of the Elbe River in 2002<br />
and Its Effects on Riverbank-Filtration Sites<br />
Dipl.-Ing. Matthias Krueger<br />
Fernwasserversorgung Elbaue-Ostharz GmbH<br />
Torgau, Germany<br />
Dipl.-Ing. Ingbert Nitzsche<br />
Fernwasserversorgung Elbaue-Ostharz GmbH<br />
Torgau, Germany<br />
Introduction<br />
To historians, Torgau, Germany, is mainly known as the <strong>for</strong>mer residence of the Saxon Electors<br />
and the cultural capital of sixteenth-century Saxony (Figure 1). A short time later, Martin Luther’s<br />
work gave the town on the Elbe the nickname, “Nursemaid of the Re<strong>for</strong>mation.” Finally, many<br />
people know Torgau as the historic site where the defeat of Nazi Germany was symbolized by the<br />
meeting of Soviet and American troops on the Elbe River on April 25, 1945.<br />
Figure 1. Air view of Torgau, Germany, 2 days be<strong>for</strong>e the flood peak.<br />
Torgau is the headquarters of the Fernwasserversorgung Elbaue-Ostharz GmbH, one of the largest<br />
drinking-water supply companies in Germany. The company was founded about 50 years ago and<br />
manages one waterworks at a reservoir in the Harz Mountains, five bank-filtration waterworks in the<br />
Elbe River Basin near Torgau, and a 700-kilometer long distribution network. In 2002, the<br />
drinking-water production was 76-cubic meters. About 3.5-million people were supplied with<br />
drinking water of high quality.<br />
Correspondence should be addressed to:<br />
Dipl.-Ing. Matthias Krueger<br />
Head of Laboratory<br />
Fernwasserversorgung Elbaue-Ostharz GmbH<br />
Naundorferstraße 46 • 04860 Torgau, Germany<br />
Phone: +49 3421 757511 • Fax: +49 3421 757522 • Email: matthias.krueger@fwv-torgau.de<br />
81
82<br />
Riverbank Filtration Near Torgau<br />
In this system, raw-water abstraction via bank filtration in the Elbe River Basin is a key element.<br />
The production boreholes are located in an alluvial sand and gravel aquifer with a thickness of<br />
40 to 60 m, covered by a 2- to 5-m thick layer of meadow loam. The meadow loam provides an<br />
important protection against pollution and infiltration during flooding. Due to erosive conditions,<br />
there is only low clogging of the riverbed. The good hydraulic connection between the Elbe River<br />
and the adjacent aquifer ensures stable water abstraction during low flow periods.<br />
Figure 2 shows the biggest waterworks, Torgau-Ost, which has a maximum capacity of<br />
120,000 m 3 /d. Raw water is abstracted from 42 production boreholes on the western bank of the<br />
Elbe River. At a distance of about 300 m between the production boreholes and the riverbank,<br />
the flow time of the bank filtrate is between 60 and 200-plus days.<br />
Figure 2. Torgau-Ost <strong>Water</strong>works and the Elbe River.<br />
The produced water has a good quality due to the natural purification ability of the aquifer. Two<br />
monitoring cross-sections positioned along assumed flowpaths to the production boreholes have<br />
been installed to control borehole operation and to guarantee flexibility in response to environmental<br />
impacts, both seasonal and long-term. These monitoring cross-sections include up to five<br />
observation stations between the production boreholes and the river. Each observation station<br />
consists of three to five monitoring points at varying depths.<br />
Raw water is purified by aeration and deacidification, pre-purification by 20 tube sedimentation<br />
basins, and fine purification by 16 open sand filters. Iron, manganese, and carbon dioxide are the<br />
main constituents of raw water that require treatment. These constituents are eliminated by<br />
adding hydrated lime and potassium permanganate. Prior to pumping into the public-water supply<br />
network, the water is further treated with small amounts of chlorine and chlorine dioxide to<br />
prevent bacteriological deterioration during the long transportation process to the consumer.<br />
The 100-Year Flood of the Elbe River in 2002<br />
In Torgau, the mean discharge of the Elbe River is about 330 cubic meters per second (m 3 /s). In<br />
August 2002, a so-called “Flood of the Century” occurred. The event was caused by a “Vb” (extreme)<br />
weather situation. Precipitation reached 300 millimeters per day, with 25 to 30 millimeters per hour<br />
in the Ore Mountains. This intensity had never been measured be<strong>for</strong>e in this region. In Dresden,<br />
the Elbe River exceeded the 1845 flood level of 8.77 m, hitting a record of 9.40 m on August 17,<br />
2002. The towns of Meissen, Torgau, Wittenberg, and Dessau were also partially flooded.
In Torgau, the water level of the Elbe River increased within 10 days by about 8 m up to 9.47 m<br />
on August 19, a level that has never been observed be<strong>for</strong>e. The peak discharge of the Elbe River<br />
in Torgau was estimated to have been 4,295 m 3 /s. The City of Torgau had to be evacuated<br />
completely. Technical staff, soldiers, and helpers built anti-flood barricades with thousands of<br />
sandbags.<br />
To ensure a safe drinking-water production, the water company <strong>for</strong>med an emergency task <strong>for</strong>ce<br />
on August 13 to organize necessary measures.<br />
The Effects of the Flood<br />
Most bank-filtration waterworks near Torgau are located at elevated levels at a sufficient distance<br />
to the river and are normally not affected by flooding. Nearly all production boreholes were flooded,<br />
but still functioned due to their special construction. A problem arose with the power-supply system<br />
<strong>for</strong> the boreholes. Trans<strong>for</strong>mer stations are located behind the dikes. But, due to the high water<br />
level and risk of dikes breaking, the stations had to be extra protected; there<strong>for</strong>e, the stations were<br />
housed-in using sheet piling and equipped with pumps <strong>for</strong> dewatering and additional emergency<br />
power generators (Figure 3).<br />
Figure 3. Protected trans<strong>for</strong>mer station behind a dike.<br />
As a result of these measures, the power supply was maintained at most locations. Only two small<br />
waterworks in the north of Torgau had to be abandoned due to water influx at trans<strong>for</strong>mer stations.<br />
Besides the aspect of water quantity, there is water quality. As a result of <strong>for</strong>mer bank-filtration<br />
research programs at the Torgau site, there is an excellent knowledge of flow processes and<br />
water-quality changes along the flowpath. Based on this knowledge, no significant change in<br />
raw-water quality was expected, mainly because the flow time of the bank filtrate ranges between<br />
60 and 300-plus days. For control purposes, the continuous monitoring of raw-water quality was<br />
intensified during the flood, and a special groundwater-monitoring program <strong>for</strong> the cross-sections<br />
was set up after the flood.<br />
All results of measurements at the outflow of the waterworks proved that drinking-water quality<br />
was not at risk at any time. The quality of the production water met the German standards <strong>for</strong><br />
83
84<br />
drinking-water quality. Drinking-water disinfection was done by 0.2 mg/L chlorine dioxide (ClO 2)<br />
and 0.6 mg/L chlorine (Cl 2). There was no problem with bacterial contamination.<br />
But what about raw-water quality? The flood caused a strong increase in turbidity, organic carbon<br />
concentration, and number of microorganisms in river water. The increase in DOC from 5 mg/L<br />
to more than 10 mg/L in river water did not affect the DOC concentration in the raw water<br />
(Figure 4). The increase in DOC in river water was mainly caused by an increase of the<br />
biodegradable fraction of DOC, which is consumed along the flowpath of the bank filtrate. Thus,<br />
biodegradation and mixing in the aquifer prevented an increase in DOC concentration in the raw<br />
water and an increase in DBPs.<br />
40<br />
35<br />
30<br />
25<br />
20<br />
15<br />
10<br />
Increases in concentrations of organic trace compounds such as polynuclear aromatic hydrocarbon,<br />
chlor-organics, and pesticides were not observed in Elbe River water during the flood. Despite<br />
many reported inputs of contaminants, the huge discharge caused an effective dilution and<br />
minimized the risk. After the flood, concentrations were found to be around the mean annual<br />
values. Despite these facts, measurements of the sum of adsorbable organic halogenated<br />
compounds (AOX) were used to check <strong>for</strong> the potential contamination of river water and raw<br />
water. The AOX concentration in the raw water remained at a low level of less than 30 µg/L,<br />
giving no indication of contamination (see Figure 4). Measured DOC and AOX concentrations<br />
in bank-filtrate samples were found to be within long-term concentrations ranges.<br />
Increased heavy metal concentrations in the river water also did not affect raw-water quality.<br />
Conclusions<br />
5<br />
0<br />
Elbe River<br />
4/29/02<br />
Elbe River<br />
8/20/02<br />
Borehole 22<br />
9/4/02<br />
Raw water<br />
10/7/02<br />
AOX (µg/l)<br />
DOC (mg/l)<br />
Figure 4. DOC and AOX concentrations in Elbe River water and raw water<br />
The 100-year flood of the River Elbe in August 2002 became the greatest challenge <strong>for</strong><br />
Fernwasserversorgung Elbaue-Ostharz GmbH in its history. Maintaining the power supply <strong>for</strong><br />
pumps in the production boreholes was the most important factor, whereas the quality of<br />
abstracted bank filtrate was never at risk due to the design of bank-filtration sites. Based on the<br />
measures taken and management of efficient distribution network, the water company ensured a
stable water supply <strong>for</strong> millions of people during the flood without restrictions to quantity and<br />
quality. The available technology <strong>for</strong> drinking-water treatment was successful in producing<br />
high-quality drinking water.<br />
Since 1990, MATTHIAS KRUEGER — a process engineer and chemist — has worked <strong>for</strong><br />
the Fernwasserversorgung Elbaue-Ostharz GmbH, a water company that provides drinking<br />
water <strong>for</strong> over 3.5-million people in Germany. Early on, he was responsible <strong>for</strong> water<br />
treatment technology, then became Head of Laboratory in 2001. Under his leadership, the<br />
Laboratory has recently been involved in large research projects on physical and chemical<br />
processes during bank filtration. Prior to joining Fernwasserversorgung Elbaue-Ostharz<br />
GmbH, Krueger was an Engineer at Galvanotechnic in Leipzig, Germany, <strong>for</strong> 7 years. His<br />
areas of interest include flow path, residence times, and redox conditions in the behavior of pollutants during<br />
riverbank filtration. Krueger received a diploma (Dipl.-Ing.) in Chemistry from the Technical University<br />
Ilmenau and completed post-graduated studies in analytics and spectroscopy at Leipzig University.<br />
85
Session 5: Dynamics<br />
Temporal Changes of Natural Attenuation Processes<br />
During Bank Filtration<br />
Paul Eckert, Ph.D.<br />
Stadtwerke Düsseldorf AG<br />
Düsseldorf, Germany<br />
Rudolf Irmscher, Ph.D.<br />
Stadtwerke Düsseldorf AG<br />
Düsseldorf, Germany<br />
<strong>RBF</strong> is a well-proven natural purification step in water supply. Sustainable water management<br />
should be based specifically on natural purification methods, as declared by the International<br />
Association of <strong>Water</strong>works in the Rhine Catchment Area, in their 2003 memorandum. To<br />
achieve this aim, a profound knowledge of the purification capacity of bank filtration is essential.<br />
At the Düsseldorf <strong>Water</strong>works in Germany, the influence of long-term — as well as periodic —<br />
changes of both hydraulics and river-water quality on natural attenuation processes were<br />
investigated.<br />
The improvement of Rhine River water quality over the last 30 years enabled the Düsseldorf<br />
<strong>Water</strong>works to reduce their technical treatment expenses. Temperature variations throughout the<br />
year and flood events significantly influenced the purification capacity of bank filtration. This<br />
rein<strong>for</strong>ces the need <strong>for</strong> flexible technical treatment methods capable of adapting to changing rawwater<br />
quality.<br />
Even though the complete replacement of subsequent technical treatment st<strong>eps</strong> might be seen as<br />
an unreachable vision, the substantial knowledge we have acquired on the purification capacity<br />
of bank filtration enables the design of tailor-made treatment methods.<br />
Site Description and Treatment Concept<br />
The City of Düsseldorf is situated in northwestern Germany, in the lower Rhine Valley (Figure 1).<br />
The Düsseldorf <strong>Water</strong>works supply 600,000 inhabitants with treated bank filtrate. A multiprotective<br />
barrier concept ensures the constant production of high-quality drinking water (Figure 2).<br />
Natural attenuation processes during bank filtration <strong>for</strong>m the first and most efficient protective<br />
barrier. The subsequent protective barrier is raw-water treatment, including ozonation, biological<br />
active filtration, and active carbon adsorption.<br />
The Rhine River has a length of 1,320 kilometers and a catchment area of 185,000 square kilometers;<br />
it is the third biggest river and the largest source of drinking water in Europe. The mean discharge<br />
of the Rhine at Düsseldorf is 2,200 m 3 /s, while the <strong>Water</strong>works use less than 2 m 3 /s. During flood<br />
events, discharge increases up to 9,900 m 3 /s.<br />
Correspondence should be addressed to:<br />
Paul Eckert, Ph.D.<br />
Head of the <strong>Water</strong> Management Department<br />
Stadtwerke Düsseldorf AG<br />
Abt. Wasserwirtschaft • Höherweg 100 • 40233 Düsseldorf, Germany<br />
Phone: +0211/8218359 • Fax: +0211/821778359 • Email: peckert@swd-ag.de<br />
87
88<br />
Protection Zones<br />
II<br />
IIIa<br />
IIIb<br />
Local Frontier<br />
0 1 2 3 4 5km<br />
Rhine<br />
Krefeld<br />
The production wells, situated at a distance of 50 to 350 m of the riverbank, discharge water from<br />
a sandy and gravel aquifer. Depending on the hydraulic situation, the residence time of bank<br />
filtrate in the aquifer varies between weeks to several months.<br />
From a water-resources perspective, bank filtration is characterized by an improvement in water<br />
quality (Kühn and Müller, 2000). The most important effects of bank filtration include:<br />
• Removal of particles and turbidity.<br />
Duisburg<br />
Neuss<br />
Figure 1. Geographic location of the Düsseldorf <strong>Water</strong>works.<br />
Corg<br />
Inactivation<br />
of virusis and<br />
pathogena<br />
CO 2<br />
Staad <strong>Water</strong>works<br />
Düsseldorf<br />
Lörick <strong>Water</strong>works<br />
Flehe <strong>Water</strong>works<br />
Auf dem Grind Well Field<br />
Ozonation<br />
NBG GmbH<br />
Figure 2. The multi-protection barrier concept <strong>for</strong> drinking-water production.<br />
Mettmann<br />
Mettmann<br />
Rhine<br />
N<br />
Biological active filtration<br />
Activated carbon adsorption<br />
upper layer<br />
lower layer
• Equalization of fluctuating concentrations in river water.<br />
• Removal of biodegradable compounds.<br />
• Removal of bacteria, viruses, and parasites.<br />
The efficiency of these natural attenuation processes is influenced by river-water quality and the<br />
hydraulic situation. Continuous and periodic changes of water quality must be considered together<br />
with fluctuating river-water levels.<br />
Temporal Changes Affecting Bank Filtration<br />
During the last 30 years, Rhine River water quality has improved significantly. Many actions to<br />
reduce nutrients and pollutants were necessary. These measures comply with the best available<br />
technology in production, as well as wastewater treatment along the Rhine. The return of salmon<br />
in the year 2000 is a visible sign of the success of this program.<br />
These quality improvements were accompanied by a decrease of ammonia and DOC in river<br />
water. Oxygen concentrations reached saturation. The higher oxidation capacity of bank filtrate<br />
is linked to more efficient natural attenuation processes within the aquifer. This enabled the<br />
<strong>Water</strong>works to reduce treatment expenses.<br />
Natural attenuation processes are affected periodically by flood events. The water level increases<br />
up to 5 m, which leads to a higher influx of river water into the aquifer. Thus, oxygen and organic<br />
carbon flux also increase by a factor of 3 to 4. The aerobic microbes adapted to the lower flux are<br />
not able to degrade the entire organic carbon during flood events. Another decrease of natural<br />
attenuation processes becomes evident in the removal of bacteria. While mostly raw water already<br />
fulfills the European Drinking <strong>Water</strong> Standard, higher colony counts are observed in the<br />
production wells following flood events (Irmscher and Teermann, 2002; Schubert, 2002). These<br />
findings explain the need of subsequent purification processes: the so-called “second protective<br />
barrier” (see Figure 2).<br />
Another effect on natural attenuation processes during bank filtration is caused by changing water<br />
temperatures throughout the year. As the temperature of the river rises quickly in the spring and<br />
summer, the temperature of the water in the aquifer also rises, but not as quickly due to a certain<br />
time delay. Microbial activity related to temperature leads, there<strong>for</strong>e, to a higher degradation rate<br />
of organic carbon. The organic carbon is removed more efficiently; however, it must be considered<br />
that, under aerobic conditions, the bank filtrate becomes more aggressive versus calcite. A flexible<br />
treatment step must be adaptable to these changing conditions.<br />
Conclusions<br />
To reduce technical treatment expenses, river water of high quality, together with a profound<br />
knowledge of the natural attenuation processes within the aquifer, is essential. At the Rhine River,<br />
successful ef<strong>for</strong>ts have improved water quality significantly. More efficient natural attenuation<br />
processes accompanied this improvement during bank filtration, which enabled the <strong>Water</strong>works<br />
to reduce expenses in water treatment.<br />
Temporal changes of river-water quality and hydraulics still influence natural attenuation processes<br />
during bank filtration. They have to be well-understood to design adequate treatment st<strong>eps</strong> and<br />
to define specific target values on river-water quality. The multi-protective barrier concept,<br />
including both natural and technical purification, is still necessary to ensure drinking water of<br />
continuously high standards.<br />
89
90<br />
REFERENCES<br />
Irmscher, R., and I. Teermann (2002). “Riverbank filtration <strong>for</strong> drinking water supply – A proven method,<br />
perfect to face today’s challenges.” <strong>Water</strong> Science and Technology, 2(5-6): 1-8.<br />
Kühn, W., and U. Müller (2000). “Riverbank filtration.” Journal AWWA, 92(12): 60-69.<br />
Schubert, J. (2002). “<strong>Water</strong>-Quality improvement with riverbank filtration at Düsseldorf <strong>Water</strong>works in<br />
Germany.” Riverbank Filtration: Improving Source-<strong>Water</strong> Quality, C. Ray, G. Melin, and R.B. Linsky (eds.),<br />
Kluwer Academic Publishers, Dordrecht.<br />
PAUL ECKERT has more than 10 years of professional experience in water-management<br />
projects. He was involved in the project <strong>for</strong> bank filtration (risk assessment of shock loads)<br />
funded by the Federal Ministry <strong>for</strong> <strong>Research</strong> and Technology (BMFT). Currently, as Head<br />
of the <strong>Water</strong> Management Department, he is responsible <strong>for</strong> protection zone management<br />
at Düsseldorf <strong>Water</strong>works, where special investigations are per<strong>for</strong>med in the field of bank<br />
filtration and groundwater remediation. A hydrogeochemist, he specializes in<br />
hydraulic/hydrochemical modeling, the design of field studies, and the assessment of<br />
groundwater-quality problems. Extended experience includes the application of a geographic in<strong>for</strong>mation<br />
system database <strong>for</strong> groundwater. Eckert received both an M.S. (Diplom) and Ph.D. from the University of<br />
Bochum in Germany, where he researched the efficacy of enhanced natural attenuation processes in a BTEXcontaminated<br />
aquifer with nitrate.
Session 5: Dynamics<br />
An Update of the City of Guelph’s Response<br />
to Regulation 459/00: Effective Natural In Situ<br />
Filtration of Several Groundwater Under the Direct<br />
Influence of Surface-<strong>Water</strong> Supplies<br />
Dennis E. Mutti, P.E.<br />
Associated Engineering Limited<br />
Toronto, Ontario, Canada<br />
Peter Busatto<br />
City of Guelph <strong>Water</strong>works Division<br />
Guelph, Ontario, Canada<br />
Douglas H. Stendahl<br />
City of Guelph <strong>Water</strong>works Division<br />
Guelph, Ontario, Canada<br />
Caroline Korn, P.E.<br />
Associated Engineering Limited<br />
Toronto, Ontario, Canada<br />
Elia Edwards, P.E.<br />
Associated Engineering Limited<br />
Toronto, Ontario, Canada<br />
Associated Environmental Limited developed a protocol <strong>for</strong> determining GWUDI status to assist<br />
the City of Guelph in Ontario, Canada, in assessing treatment requirements <strong>for</strong> their municipal<br />
water supply. This protocol is based on protocols established by the United States Environmental<br />
Protection Agency (USEPA) and further defined in American <strong>Water</strong> Works Association <strong>Research</strong><br />
Foundation Project #605. The protocol entails three stages of assessment. Stage 1 is a review of<br />
existing in<strong>for</strong>mation to characterize the well as a true groundwater or GWUDI. This data includes<br />
well construction and maintenance records, sanitary condition of the well, hydrogeological data<br />
including time-of-travel estimates (well-head delineation), geological characteristics, and waterquality<br />
data. Stage 2 is initiated if there is insufficient data to make a characterization of the nature<br />
of the well in Stage 1 and consists of a period of data collection that will allow the characterization<br />
to be completed. Wells that are characterized as GWUDI are then subject to a Stage 3 assessment,<br />
which consists of a data collection period followed by an assessment of the level of natural or in situ<br />
filtration that is occurring. If sufficient evidence of in situ filtration is shown in the Stage 3<br />
assessment, then a case can be made <strong>for</strong> waiving the chemically assisted filtration component of the<br />
minimum treatment requirement <strong>for</strong> GWUDI in the Province of Ontario. The minimum treatment<br />
requirements in the Province of Ontario <strong>for</strong> GWUDI sources with effective in situ filtration is 3-log<br />
inactivation <strong>for</strong> Giardia cysts and 4-log inactivation <strong>for</strong> viruses from disinfection alone.<br />
Correspondence should be addressed to:<br />
Dennis Mutti, P.E.<br />
Senior <strong>Water</strong> Treatment and Supply Engineer<br />
Associated Engineering Limited<br />
525-21 Four Seasons Place • Toronto, Ontario M9B 6J8 Canada<br />
Phone: (416) 622-9502 • Fax: (416) 622-6249 • Email: muttid@.ae.ca<br />
91
92<br />
The City of Guelph’s supply system consists of a network of 21 groundwater wells and a spring<br />
collector system, augmented by artificial recharge. The Stage 1 assessment was completed <strong>for</strong> all<br />
of the City’s wells as a part of the provincially mandated Engineers’ Report. Two bedrock wells<br />
(Burke Well and Downey Well) and an overburden well system (Carter Wells 1 and 2) were<br />
identified as requiring a Stage 2 assessment due to insufficient existing data. The collector system<br />
(Arkell Spring and Glen Collector) was identified as GWUDI by definition and required a Stage 3<br />
assessment. The City of Guelph decided to expedite the process and collect data required <strong>for</strong> a<br />
Stage 3 assessment <strong>for</strong> the three wells, as well as <strong>for</strong> the surface-water recharge and collector system<br />
in case the wells were determined to be GWUDI.<br />
The results of the GWUDI assessment were presented at a stakeholders meeting with key<br />
provincial regulators. Consensus was reached as follows:<br />
• Burke Well and Downey Well are representative of true groundwater supplies and, as such,<br />
must provide disinfection as mandated by the Drinking <strong>Water</strong> Protection Regulation<br />
(2.0-log virus inactivation).<br />
• The treatment requirement to provide chemically assisted filtration <strong>for</strong> the Carter Wells,<br />
Arkell Recharge System, and Glen Collector System should be waived as effective<br />
natural in situ filtration is provided through the subsurface. As such, the treatment<br />
requirements <strong>for</strong> Giardia cysts (3-log inactivation) and viruses (4-log inactivation) may<br />
be provided by disinfection only.<br />
The City of Guelph is currently negotiating treatment requirements, as well as the final wording<br />
in its Consolidated Certificate of Approval, with the Ontario Ministry of the Environment.<br />
DENNIS MUTTI has 13 years of progressive experience as a Project Manager and Process<br />
Engineer, the last 5 of these years with Associated Engineering, which provides environmental<br />
engineering consulting services to the Ontario, Canada, market. His focus has been on<br />
process evaluation and development, and the design, implementation, and optimization of<br />
water supply and treatment systems. Mutti has worked on all types of water-supply and<br />
treatment projects, including water-supply master plans, water treatment plant designs and<br />
upgrades, groundwater supplies, automation and Supervisory Control and Data Acquisition<br />
(SCADA) system implementation, and start-up and commissioning. He received both a B.S. in Chemical<br />
Engineering and an M.S. in Civil Engineering from the University of <strong>Water</strong>loo.
Session 5: Dynamics<br />
On Bank Filtration and Reactive Transport Modeling<br />
Dr.-Ing. Ekkehard Holzbecher<br />
Humboldt University<br />
Berlin, Germany<br />
Prof.-Dr. Gunnar Nützmann<br />
Humboldt University<br />
Berlin, Germany<br />
Modeling can enhance the understanding of the influence of hydraulic, transport, and biogeochemical<br />
processes <strong>for</strong> the development of bank filtration as an applied technology. Because a<br />
coupled three-dimensional approach, including all biogeochemical trans<strong>for</strong>mations, speciation,<br />
and kinetics, is not feasible nowadays in applied projects, a simpler procedure is proposed. In<br />
the first step, flow is calculated using analytical solutions or numerical models in two- or<br />
three-dimension. In the second step, reactive transport is simulated in one-dimension along<br />
flowpaths. How these two st<strong>eps</strong> are per<strong>for</strong>med in a project depends on the chosen software <strong>for</strong> flow<br />
and reactive transport. A simple approach is demonstrated that uses analytical solutions <strong>for</strong> flow and<br />
the pH redox equilibrium equation (PHREEQC) <strong>for</strong> reactive transport. This approach is appropriate<br />
<strong>for</strong> understanding processes in porous media in the direct vicinity of a surface-water body.<br />
Flow Modeling<br />
The simulation of reactive transport at a field site is always based on a flow model. Flow models<br />
can account <strong>for</strong>:<br />
• Inhomogeneities.<br />
• Anisotropies.<br />
• Site-specific design of the well system.<br />
• Irregular (usually) setting of the boundaries.<br />
Rather complex two-dimensional or three-dimensional flow patterns emerge.<br />
Although the capability of software has increased in conjunction with the power of hardware,<br />
there are still limiting factors <strong>for</strong> solving general three-dimensional reactive transport models.<br />
Transport models require a small spatial resolution (e.g., a high number of nodes or blocks) in<br />
which dependent variables are computed. Biogeochemistry requires a small temporal resolution to<br />
capture the changes of some variables in response to changes in the system.<br />
Numerical errors and their propagation when solving huge linear and/or nonlinear systems are<br />
hard to predict. If some well-known constraints (grid Péclet-number criterion, Courant-number<br />
and Neumann-number criteria) are not fulfilled, numerical errors increase dramatically. The<br />
Correspondence should be addressed to:<br />
Dr.-Ing. Ekkehard Holzbecher<br />
Senior Scientist<br />
Humboldt University Berlin<br />
<strong>Institute</strong> of Freshwater Ecology and Inland Fisheries • Müggelseedamm 310 • 12587 Berlin, Germany<br />
Phone: +0049-30-64181 667 • Fax: +0049-30-64181 663 • E-Mail: holzbecher@igb-berlin.de<br />
93
94<br />
resulting numerical dispersion or oscillations disturb results in that the results cannot be used <strong>for</strong><br />
comparisons with observations.<br />
To couple the result of a flow model with a general geochemical approach, as given by the<br />
PHREEQC code, the flow system must be reduced to one-dimension. Flowpaths can be taken from a<br />
two- or three-dimensional model and be used <strong>for</strong> a one-dimensional reactive transport simulation.<br />
The procedure is applicable <strong>for</strong> bank-filtration systems, as concentration gradients in transverse<br />
directions can be assumed to be much smaller than in longitudinal directions.<br />
In the first step towards such a simulation, flowpaths are taken from analytical solutions <strong>for</strong><br />
idealized well gallery systems. MATLAB (2002) software is applied <strong>for</strong> the analytical solution.<br />
Systems with different numbers of wells, different distances from the surface-water body, and<br />
different pumping rates can be investigated. It is also possible to introduce a groundwater base<br />
flow. Moreover, different types of boundary conditions at the bank can be considered.<br />
Figures 1 and 2 show the results of a simulation <strong>for</strong> two wells. The calculations with analytical<br />
solutions, as well as the graphical representation, were made using MATLAB.<br />
Figure 1. Hydraulic potential φ (m 3 /s) <strong>for</strong> two wells with different pumping rates and distances from a<br />
surface-water body near a gaining stream.<br />
Some basic results can be obtained <strong>for</strong> a single well near a gaining stream. There is a critical<br />
pumping rate, Q crit , below which only groundwater is pumped. By increasing the pumping rate<br />
above Q crit , the share of surface water (filtrate) increases. An equal share between groundwater<br />
and surface water is given <strong>for</strong>:<br />
Q = 6.1878 •Q crit<br />
(Equation 1)
Figure 2. Streamlines <strong>for</strong> two wells with different pumping rates and distances from a surface-water body near<br />
a gaining stream.<br />
We intend to further explore the use of analytical solutions. A numerical algorithm, which<br />
calculates flowpaths and travel times from the surface-water body to the wells, is currently being<br />
tested. If combined with geochemical computations, the approach can provide a highly important<br />
tool <strong>for</strong> the design of well galleries.<br />
Travel times along flowpaths are obtained by numerical integration.<br />
Reactive Transport Modeling<br />
With the outlined strategy, the well-established PHREEQC code (new version: PHREEQC2) can<br />
be applied <strong>for</strong> reactive transport modeling to reduce the number of spatial dimensions. The<br />
PHREEQE code (which is the origin of PHREEQC) was mainly a geochemical speciation code<br />
and originally not intended to be combined with a transport model. Velocity is not an input to be<br />
specified. Instead, lengths (∆x), time step (∆t), cells (number of blocks), and shifts (number of time<br />
st<strong>eps</strong>) must be given as parameters (Parkhurst, 1995). Nevertheless, the simulated velocity v<br />
results from the input parameters:<br />
ν = ∆x / ∆t<br />
(Equation 2)<br />
The so-called “mixing cell approach” <strong>for</strong> advection, which uses the operator-splitting technique, is<br />
combined with the finite difference method <strong>for</strong> diffusion and proves to be competitive with advanced<br />
numerical techniques. Errors from the discretization advection term are successfully suppressed.<br />
PHREEQC can also be used <strong>for</strong> variable velocity along flowpaths. The time step in Equation 2 is<br />
fixed, but grid spacing varies locally, taking velocity changes into account. Small velocities result<br />
in small ∆x and high velocities in large ∆x.<br />
95
96<br />
As an alternative to PHREEQC <strong>for</strong> certain tasks, a new model based on MATLAB is currently in<br />
development. The model is being tested in a number of benchmark studies. Figure 3 shows the<br />
results of the MATLAB model <strong>for</strong> a test of redox sequences, as compared with PHREEQC output.<br />
Concentration [mol/L]<br />
5.00E-03<br />
4.50E-03<br />
4.00E-03<br />
3.50E-03<br />
3.00E-03<br />
2.50E-03<br />
2.00E-03<br />
1.50E-03<br />
1.00E-03<br />
5.00E-04<br />
CH20 MATLAB<br />
N(0) PHREEQE<br />
S-2 MATLAB<br />
02 PHREEQC<br />
CH20 PHREEQC<br />
N(-3) MATLAB<br />
S(-2) PHREEQC<br />
pH MATLAB<br />
N(5) MATLAB<br />
N(-3) PHREEQC<br />
HCO3 MATLAB<br />
pH PHREEQC<br />
N(3) PHREEQC<br />
S(6) MATLAB<br />
HCO3 PHREEQC<br />
pe MATLAB<br />
Figure 3. Reactive transport (redox) test case: Redox zoning after a 1-year simulation timeframe.<br />
N(0) MATLAB<br />
S(6) PHREEQC<br />
O2 MATLAB<br />
pe PHREEQC<br />
0.00E-00<br />
–6.00<br />
0.00 10.00 20.00 30.00 40.00 50.00<br />
x [m]<br />
60.00 70.00 80.00 90.00 100.00<br />
In the MATLAB model, valence electron balances are obtained using an adequate conceptualization<br />
of operational valence electron balancing. Redox equilibrium modeling, including kinetics<br />
<strong>for</strong> organic carbon biodegradation and operational valence electron balancing to the equilibrium<br />
module, yields automatically to a correct choice of electron acceptor.<br />
For non-redox equilibrium systems, which incorporate proton and component mass balances and<br />
mass action equations, the operational valence electron balance and the electron (e – ) as <strong>for</strong>mal<br />
species entering mass action equations — derived from redox half reactions — are considered.<br />
Since free electrons are generally not observed in aqueous solutions (Thorstenson, 1984),<br />
operation valence electron balancing has to be carried out by adapting the share of different<br />
valence states of heterovalent components (in a sense, valence electron “bookkeeping”) (Appelo<br />
and Postma, 1996).<br />
In inorganic redox systems, the sum of mobile operational valence electrons is an additional<br />
transport species. In the presence of biodegradation redox reactions (which supply additional<br />
carbonate species to the system), the actual operation electron balance has to be corrected by the<br />
carbonate species arising from biodegradation, as pointed out by Brun and Engesgaard (2002).<br />
A test case including hydrodynamic transport to the redox system was set up using MATLAB. At<br />
the inlet of the model column, a solution that is rich in mobile organic matter infiltrates into an<br />
aquifer void of organic matter, but contains oxygen, nitrate, and sulphate as electron acceptors.<br />
Unlike Redox Test Case 1, ammonium (N[–3]) is included in the equilibrium reaction framework.<br />
During infiltration, a steady-state redox zone is simulated similar to the redox zone observed at<br />
16.00<br />
14.00<br />
12.00<br />
10.00<br />
8.00<br />
6.00<br />
4.00<br />
2.00<br />
0.00<br />
–2.00<br />
–4.00<br />
pH. pE [–]
well galleries near the Unterhavel (Berlin). The distance from the inlet at which the redoxcline<br />
(a large jump in the redox state) is expected depends on:<br />
• Hydrodynamic parameters in relation to the first-order organic matter decay constant.<br />
• Initial redox state (operational valence electron balance).<br />
• Initial electron-acceptor concentrations.<br />
• Availability of reactive organic matter.<br />
These are the main sensitivity parameters that determine the extent of redox zoning.<br />
Figure 3 shows the simulated steady-state species and pH/oxidation-reduction (redox) potential (pE)<br />
distribution along the model transect. The simulation reached steady-state concentrations after a<br />
150-day simulation time. The pE profile shows the typical variation caused by aerobic respiration<br />
(pE buffering at +15), denitrification (pE buffering at +12 to +14), and sulphate reduction<br />
(pE buffering at –3). The profiles of oxygen (O 2) concentrations, nitrate (N[5]), and sulphate<br />
(S[6]) concentrations also are figured so that the changes of pE are explained straight<strong>for</strong>wardly.<br />
Because ammonium (N[–3]) is incorporated into the speciation, elemental nitrogen (N[0]) will<br />
not be the end step of the nitrogen species, but will be reduced to ammonium in conjunction with<br />
the sulphate-reduction step.<br />
Outlook<br />
The outlined coupling strategy <strong>for</strong> flow and reactive transport is intended to be applied on real<br />
sites. Flow simulation will be done using MODFLOW (Harbaugh et al., 2000) or FEFLOW (2002)<br />
<strong>for</strong> site-specific models and by analytical solutions <strong>for</strong> conceptual models concerning well-gallery<br />
design. For reactive transport, there is the choice between PHREEQC and MATLAB.<br />
REFERENCES<br />
Appelo, C.A., and D. Postma (1996). Geochemistry, groundwater and pollution, Balkema, Rotterdam.<br />
Brun, A., and P. Engesgaard (2002). “Modeling of transport and biogeochemical processes in pollution<br />
plumes: literature review and model development.” Journ. of Hydrology, 256: 211-227.<br />
FEFLOW (2002). Finite element, subsurface flow & transport simulation system, WASY, Berlin, Germany.<br />
Harbaugh, A.W., E.R. Banta, M.C. Hill, and M.G. McDonald (2000). MODFLOW-2000, the U.S. Geological<br />
Survey modular ground-water model — User guide to modularization concepts and the Ground-<strong>Water</strong> Flow Process,<br />
U.S. Geological Survey Open-File Report 00-92, 121p.<br />
MATLAB (2002). MATLAB the language of technical computing, The MathWorks, Inc., Natick, MA.<br />
Parkhurst, D.L. (1995). PHREEQC: A computer program <strong>for</strong> speciation, reaction-path, advective transport, and<br />
inverse geochemical calculations, U.S. Geological Survey, <strong>Water</strong> Res. Invest. Report 95-4227, Lakewood, 143p.<br />
Thorstenson, D.C. (1984). The concept of electron activity and its relation to redox potentials in aqueous<br />
geochemical systems, Open-File Report 84-072, U.S. Geological Survey Denver, Colorado.<br />
97
98<br />
Mathematician EKKEHARD HOLZBECHER has worked the field of groundwater<br />
science and applications since 1984. Most of his projects are related to numerical<br />
modeling. Past projects include investigating the safety of nuclear waste repositories in<br />
Germany, saltwater intrusion in the Nile-Delta in Egypt, geothermal flow in Japan, and<br />
abandoned industrial sites in Germany. Currently, he is investigating various problems<br />
regarding eco-hydrology. Since 2002, Holzbecher has been a Senior Scientist with<br />
Humboldt University in Berlin, where he teaches groundwater and transport modeling<br />
courses, and participates in the Natural and Artificial Systems <strong>for</strong> Recharge and Infiltration project. He was<br />
also involved in the international Groundwater Hydrology Modeling Strategies <strong>for</strong> Per<strong>for</strong>mance Assessment of<br />
Nuclear Waste Disposal (HYDROCOIN) project and is a member of the United Nations Educational,<br />
Scientific and Cultural Organization (UNESCO) working group <strong>for</strong> the Development and Calibration of<br />
Coupled Hydrological/Atmospheric Models. Holzbecher received a Ph.D. from the Civil Engineering<br />
Department at the Technical University Berlin and the German degree of Habilitation (qualification <strong>for</strong> a<br />
tenure professorship) at the Faculty of Geosciences at the Free University Berlin, where he is now a<br />
Privatdozent.
Dinner Presentation<br />
Hydraulic Sensitivities and Reduction Potential<br />
Correlated with the Distance Between<br />
the Riverbank and Production Well<br />
Bernhard Wett, Ph.D.<br />
University of Innsbruck<br />
Innsbruck, Austria<br />
Objectives<br />
Beginning with a detailed case study of a <strong>RBF</strong> system at the alpine river, Enns, in Austria, some<br />
aspects of the question, “In what ways do specified hydraulic parameters affect water quality (e.g.,<br />
what is the appropriate well distance from the riverbank)?” are elaborated. In addition to<br />
monitoring data, numerical sensitivity analyses will be applied to display interactions between<br />
well distances, filtrate production, and reduction processes.<br />
Methodology and Site Description<br />
The investigated <strong>RBF</strong> well is situated about 50 m from the bank of the oligotrophic alpine river,<br />
Enns, at the beginning of the 5-kilometer long reservoir of the HPP Garsten power plant (Figure 1).<br />
At this particular location, the river stage (as determined by a dam) is 302.0 m above sea level and<br />
showed only minor elevations when discharge varied from about 70 to 540 m 3 /s during a 1-year<br />
measurement period. Since the riverbed is cut into the dense flysch zone, river water infiltrates<br />
almost exclusively through the bank and not the bottom. Between the river and well, aquifer<br />
thickness is about 5 m, and the total thickness of the gravel layer is 15 m (Ingerle et al., 1999;<br />
Wett et al., 2002).<br />
Organic loading is very low (the DOC concentration varies between 1 and 2 mg/L), and river<br />
water is saturated with oxygen (greater than 10 mg/L). Groundwater quality reflects the trends of<br />
land use and the intensity of agriculture along the river. The further downstream of the river, the<br />
higher the nitrate and pesticides concentrations are in groundwater. In the region of the filtration<br />
site, groundwater shows nitrate concentrations of about 50 mg/L. The enrichment of groundwater<br />
by river filtrate (less than 5 mg/L nitrate) offers a solution to obtain nitrate concentrations in<br />
agreement with drinking-water standards.<br />
A high-grade steel box with windows <strong>for</strong> visual observation and video recording of riverbed clogging<br />
was installed in the riverbank. One set of five probes was installed at the upstream side of the box in<br />
natural sediment (multi-level Probe F at depths between 0.1 and 0.9 m beneath the riverbed surface),<br />
and two sets of five probes (D and E) were installed downstream in specified filter sand (0 to 4 mm)<br />
to measure hydraulic heads and obtain water samples. <strong>Water</strong> levels in the probes correspond with the<br />
levels of graduated glass pipes within the steel box. An analysis of hydraulic head data and relative<br />
variations of head differences allows <strong>for</strong> filter velocity calculations. Higher hydraulic conductivity in<br />
Correspondence should be addressed to:<br />
Bernhard Wett, Ph.D.<br />
Scientific <strong>Research</strong>er and Lecturer<br />
<strong>Institute</strong> of Environmental Engineering<br />
University of Innsbruck • Technikerstr. 13 • A-6020 Innsbruck, Austria<br />
Phone: +43 512 507 6926 • Email: Bernhard.Wett@uibk.ac.at<br />
99
100<br />
Figure 1. Schematic overview and cross-section of the <strong>RBF</strong> site on the Enns River, a tributary of<br />
the Danube River.<br />
the filter sand causes a significantly higher filter velocity than in natural sediment (the mean filter<br />
velocity is 1.99 × 10 –5 m per second and 0.83 × 10 –5 m per second, respectively).<br />
Results<br />
In general, microbial metabolism and degradation processes require both electron donators and<br />
acceptors. Depending on the site and its inherent boundary conditions, an oligotrophic river<br />
would have carbon as a limiting factor. The lack of electron donators can be compensated <strong>for</strong> by<br />
long hydraulic retention in organic-rich sediment layers at high water temperatures. A long<br />
retention time (represented by low infiltration rates in Figure 2) in the most biologically active<br />
zone near the water-sediment interface drives the balance between the availability of electron<br />
donators and acceptors towards higher oxygen consumption. The arrow indicating the influence<br />
of hydraulic parameters refers to the distance d from the river to the production well, hydraulic<br />
Figure 2. Limitations of microbial metabolism in the riverbank and their balancing factors.
conductivity k (especially of the riverbed), aquifer thickness H, and production rate Q BF.<br />
Additionally, an increase in temperature causes a decrease in water viscosity and, there<strong>for</strong>e, creates<br />
higher hydraulic conductivity (this is a 50-percent increase in the presented case study).<br />
It is a known fact that particulate organic matter deposited in the riverbed represents an important<br />
carbon pool (Bretschko and Moser, 1993). The metabolized mass of particulate organic carbon can<br />
be estimated from the respiration of the sediment community to identify the dominant carbon<br />
source. The difference between total hyporeic respiration and DOC respiration, as depicted in<br />
Figure 3, describes a substantial gap in the mass balance between carbon fluxes in and out of the<br />
riverbed. In the case of increased DOC loads of the filtrate, the microbial community can adapt<br />
as long as electron acceptors are available (e.g., basin maturation <strong>for</strong> soil-aquifer treatment of<br />
percolated secondary effluent occurs within a few months) (Quanrud et al., 2003). Reactive<br />
transport models can describe the relatively quick development from electron donator limitation<br />
towards electron acceptor limitation (Lensing et al., 1994).<br />
DOC immobilization / microbial C respiration<br />
(µ mol C per L)<br />
DOC immobilization C respiration<br />
Station C Station D<br />
Station F Station E<br />
month (1997/1998)<br />
(non-DOC)-C respiration<br />
Figure 3. Seasonal variation of DOC immobilization (∆DOC), total hyporheic carbon respiration (HCR),<br />
and the difference between ∆DOC and HCR in the riparian zone (0.9 to 1.0 m from the sedimentwater<br />
interface, with Stations C and F in natural sediment and Stations D and E in specified filter<br />
sand) (Brugger et al., 2001).<br />
Figure 4 demonstrates influences on microbial reduction processes both by temperature and flow<br />
velocity: the oxygen and nitrate of the electron acceptors show the highest peaks during the winter<br />
season (the oxygen concentration approaches saturation at water temperatures of 2-degrees Celsius<br />
in February) and drop to low points during the summer season (8.5-mg/L oxygen [O 2] in Station E<br />
and 4.5-mg/L O 2 in Station F in July, when water temperature reaches 15-degrees Celsius). This<br />
period of increased oxygen depletion corresponds with concentration peaks of iron and manganese.<br />
The mean oxygen depletion during the summer season in the riverbed achieves 4.0-mg/L O 2 at<br />
Bore F and 2.8-mg/L O 2 at Bore D. These respiration rates include a portion <strong>for</strong> DOC immobilization<br />
of 25 and 19 percent, respectively (see Figure 3). Hence, both DOC immobilization and<br />
(non-DOC)-C respiration<br />
(µ mol C per L)<br />
101
102<br />
Figure 4. Courses of oxygen, nitrate, iron, and manganese concentrations at Station E in riverbed zones with<br />
low hydraulic conductivity and at Station F with higher conductivity during the 1-year monitoring<br />
period (0.9 to 1.0 m from the sediment-water interface).<br />
total hyporheic carbon respiration are significantly higher in Infiltration Area F, where the filter<br />
velocity is 58 percent lower than at Bore D. These results suggest that monitored degradation<br />
processes are limited by available retention time and that the hydrolyses of particulate organic<br />
matter is the time-limiting process step.<br />
The spatial distribution of oxygen, as presented in Figure 5, clearly shows the influence of<br />
retention time on reduction processes. During the migration time in the first meter of the flowpath<br />
in natural sediment, the oxygen concentration drops to a minimum level and the subsequent<br />
recovery is attributed to mixing with oxygen-enriched water.<br />
Figure 5. Oxygen concentration profiles along the flowpath from the river to the well,<br />
starting from four different infiltration areas (Stations D and E in natural<br />
sediment, July 27, 1998).<br />
In the following section, a calibrated model of the presented site serves as the initial situation (plot<br />
and cross-section in Figure 6) of a systematic parameter investigation. The linear sensitivity<br />
analysis uses a function, δ v,p , to quantify the sensitivity of the model response to a unit change in
parameter value:<br />
Figure 6. Initial conditions of numerical sensitivity analyses (MODFLOW software).<br />
dv/ v<br />
δv,p = where p = parameter and v = output variable (model response).<br />
dp/ p<br />
The larger the value of the function, the more significant the specific parameter is <strong>for</strong> model<br />
behavior. The applicability of this <strong>for</strong>m is limited to linear cause-and-effect relationships or small<br />
parameter variations. The results in Table 1 outline the major influence of riverbed clogging<br />
k A/k BF on filtrate production Q BF and well distance d on the total migration time HRT and<br />
maximum infiltration rate v max.<br />
Table 1. Comparison of Parameter Sensitivities to Hydraulic Properties of the Considered <strong>RBF</strong> Site<br />
d v,p k A / k BF k A Q Prod. d H<br />
HRT [d] 0.24 0.33 –1.01 1.89 1.20<br />
Q BF [%] –0.25 –0.15 0.03 –0.18 0.17<br />
v max [m/h] –0.44 –0.44 0.89 –0.76 –0.89<br />
HRT = Hydraulic retention time. Q BF = Filtrate portion. v max = Maximum infiltration rate.<br />
103
104<br />
Conclusions<br />
Biologically mediated degradation processes mainly occur at the first meter of the flowpath from<br />
the river to the well. Hydraulic retention in this layer and infiltration rates, respectively, are<br />
crucial parameters of additional particulate carbon respiration. Infiltration velocity, in turn,<br />
depends on the site-specific hydraulic setting, especially on production rate, well distance, and<br />
aquifer thickness. The appropriate selection of these parameters, there<strong>for</strong>e, should aim towards<br />
achieving preferable hyporheic conditions.<br />
Acknowledgements<br />
Basic results presented in this paper have been achieved by close interdisciplinary cooperation in the<br />
course of a research project coordinated (Ch. Hasenleithner) and funded by the Ennskraft Company.<br />
REFERENCES<br />
Brettschko, G., and H. Moser (1993). “Transport and retention of matter in riparian ecotones.” Hydrobiologia,<br />
251: 95-102.<br />
Brugger, A., B. Wett, I. Kolar, B. Reitner, and G.J. Herndl (2001). “Immobilization and bacterial utilization<br />
of dissolved organic carbon entering the riparian zone of the alpine Enns River, Austria.” Aquatic Microbial<br />
Ecology, 24(2): 129-142.<br />
Ingerle, K., A.P. Blaschke, A. Brugger, C. Hasenleithner, G.J. Herndl, H. Jarosch, I. Kolar, H.J. Lensing, S.<br />
Pöschl, N. Quéric, B. Reitner, F. Schöller, R. Sommer, and B. Wett (1999). Forschungsprojekt Uferfiltrat<br />
(<strong>Research</strong> project bank filtration), <strong>Research</strong> Initiative Verbund, Vienna, 60, 43-78.<br />
Lensing, H.J., M. Vogt, and B. Herrling (1994). “Modeling of biologically meditated redox processes in the<br />
subsurface.” J. of Hydrology, 159: 125-143.<br />
Quanrud, D.M., J. Hafer, M.M. Karpiscak, J. Zhang, K.E. Lansey, and R.G. Arnold (2003). “Fate of organics<br />
during soil-aquifer treatment: sustainability of removals in the field.” Wat. Res., 37: 3,401-3,411.<br />
Wett, B., H. Jarosch, and K. Ingerle (2002). “Flood induced infiltration affecting a bank filtrate well at the<br />
River Enns, Austria.” J. of Hydrology, 266(3-4): 222-234.<br />
BERNHARD WETT has been a Scientific <strong>Research</strong>er at the Department of Environmental<br />
Engineering at the University of Innsbruck in Austria <strong>for</strong> 9 years. His recent scientific<br />
work has focused on the separate biological treatment of rejection water — a major<br />
interaction between the wastewater and the sludge lane of a treatment plant. With regard<br />
to riverbank filtration and its relation to flooding events, he has conducted an<br />
interdisciplinary project together with various companies and universities over the past<br />
several years. The central question of these investigations was the vulnerability of<br />
riverbank-filtration waterworks, in regard to both hydraulic and water-quality aspects. Wett participates in<br />
working groups of national water associations (Wastewater Technology Association, Austrian <strong>Water</strong>-and<br />
Waste Management Association) and in the EU-COST-624 Action. He lectures in various fields of<br />
environmental engineering, including numerical methods of groundwater flow modeling. Wett received an<br />
M.S. in Civil Engineering and a Ph.D. in Engineering Science-Environmental Engineering from the<br />
University of Innsbruck.
Keynote Presentation<br />
Riverbank Filtration: The European Experience<br />
Prof. Dr.-Ing. Martin Jekel<br />
Technical University of Berlin<br />
Berlin, Germany<br />
Prof. Dr.-Ing. Thomas Grischek<br />
University of Applied Sciences<br />
Dresden, Germany<br />
Introduction<br />
The first historical citation of a small bank filtration system <strong>for</strong> a village dates back to 1798 in a<br />
German text entitled, About <strong>Water</strong>. It describes a construction procedure <strong>for</strong> artificial bank<br />
filtration into a dug well, about 2 m from a turbid creek. The intermediate space is filled with sand<br />
to remove pollution during gravity flow into the well. The sand is replaced twice a year to<br />
guarantee sufficient yield. Thus, we interpret this treatment as a combination of <strong>RBF</strong> and slow<br />
sand filtration, two techniques still widely used and known as effective physical and biological<br />
filters with many similarities.<br />
In Europe, more than 130 years of experience exist in the O&M of large bank-filtration schemes.<br />
The oldest operating <strong>RBF</strong> waterworks were founded in the 1870s in Germany and The Netherlands<br />
during the introduction of centralized water supply in urban areas. The most important sites are<br />
situated along the Danube, Rhine, and Elbe rivers, and the lakes in the City of Berlin area. A large<br />
variety of schemes has been designed and operated according to site-specific conditions. Many<br />
research projects have been conducted within the last 20 years to study attenuation processes<br />
during bank filtration.<br />
Relevance of Riverbank Filtration in Europe<br />
Groundwater derived from infiltrating river water provides 45 percent of drinking-water supplies in<br />
Hungary, 16 percent in Germany, 5 percent in The Netherlands, and 50 percent in the Slovak<br />
Republic. In Germany, the City of Berlin depends on bank filtration <strong>for</strong> about 60 percent of its<br />
drinking water. The City of Düsseldorf, situated on the River Rhine in Germany, is entirely supplied<br />
with drinking water derived from bank filtration. Budapest, the capital of Hungary, also depends<br />
100 percent on bank filtration. <strong>Water</strong>works in many other cities (e.g., Cologne, Dresden, Zurich,<br />
Lindau, and Maribor) rely on bank filtrate as an important resource. In Finland, 217 waterworks use<br />
bank-filtration techniques as part of their water-treatment process (Kivimäki, 1995).<br />
Besides known bank-filtration sites, there are many groundwater works where bank filtration is<br />
not planned and is unintentional, especially within periods of high surface-water levels during<br />
floods.<br />
Correspondence should be addressed to:<br />
Prof. Dr.-Ing. Martin Jekel<br />
Professor, Department of <strong>Water</strong> Quality Control<br />
Technical University of Berlin<br />
Sekr. KF 4 • Strasse des 17. Juni 135 • 10623 Berlin, Germany<br />
Phone: +(30) 314 - 25058 • Fax: +(30) 314 - 23313 / 23850 • Email: jekel@itu201.ut.TU-Berlin.DE<br />
105
106<br />
Geohydraulic Characteristics<br />
Grischek et al. (2002) provide an overview on the geohydraulic characteristics of different<br />
bank-filtration sites in Europe. Typical aquifers used <strong>for</strong> bank filtration consist of alluvial sand and<br />
gravel deposits and have a hydraulic conductivity higher than 1 × 10 –4 m per second. The<br />
thickness of the exploited alluvial aquifers covers a wide range. Along the Danube River in the<br />
Slovak Republic and Elbe River in Germany, aquifer thickness exceeds 50 m. Along the Lot River<br />
in France and Neckar River in Germany, boreholes are operated in shallow aquifers of about 5-m<br />
thickness. The distance between production wells and the bank of the river or lake is, in general,<br />
more than 50 m. Excluding some sites in The Netherlands and along the Danube River in the<br />
Slovak Republic, travel times are mostly between 20 and 300 days.<br />
In some countries, a travel time minimum of 50 days is proposed because an accepted rule of<br />
thumb assumes that about 50 days are sufficient to obtain pathogen-free water. But, recent findings<br />
underline that other factors, such as surface area (flowpath length), riverbed properties, and redox<br />
conditions, play an important role in removing pathogens.<br />
At most sites, vertical wells are in operation. At sites with a high aquifer thickness (<strong>for</strong> example,<br />
Kalinkovo in the Slovak Republic), vertical wells have screen lengths of 40 m, allowing high<br />
abstraction rates. Older systems often include siphon pipes connected to vertical well galleries<br />
with low abstraction rates (<strong>for</strong> example, Karany in the Czech Republic and Dresden-Tolkewitz in<br />
Germany). Horizontal wells are used at sites with high abstraction rates, such as in Düsseldorf,<br />
Germany, and Budapest, Hungary.<br />
Riverbeds at <strong>RBF</strong> sites are normally cut into an underlying sand and gravel aquifer, resulting in<br />
direct hydraulic contact of the river and aquifer. At many <strong>RBF</strong> sites, erosive conditions in the river<br />
limit the <strong>for</strong>mation of a siltation layer. Detailed research to understand the specific mechanisms<br />
controlling clogging has been undertaken on the Rhine River (Schubert, 2001), Enns River<br />
(Ingerle et al., 1999), and Elbe River (Heeger, 1987). At most rivers, infiltration occurs in the<br />
areas between the bank adjacent to the wells and the middle of the river, sometimes over the<br />
whole width of the river. At the Rhine River, partial clogging of the riverbed limits infiltration<br />
near the bank adjacent to the wells. The riverbed infiltration zone is fairly clean, but other areas<br />
are coated with a layer of about 1-millimeter thickness. Sand ripples also develop. Under rapidly<br />
changing river levels, the riverbed is cleaned (Schubert, 2000). The other case is observed at<br />
impounded river lakes in Berlin, where infiltration occurs mainly via the bank because thick<br />
organic mud layers at the bottom of lakes have low hydraulic conductivity.<br />
Surface-<strong>Water</strong> Quality<br />
In general, organic carbon concentrations (TOC) in river waters used <strong>for</strong> <strong>RBF</strong> are between 1 and 6 mg/L.<br />
The lakes of Berlin are higher in TOC, up to 10 mg/L, due to natural fulvic acids, effluent organic<br />
matter, and algal growth. In Finland, where NOM concentrations contribute to TOC values of<br />
4 to 15 mg/L, problems are associated with natural humic substances and DBP <strong>for</strong>mation.<br />
Temperature variations in most rivers range from zero to 25-degrees Celsius, and the pH is between<br />
6 and 8.<br />
Since the late 1950s, the water quality of large rivers in Europe began to deteriorate. High<br />
wastewater input threatened the use of bank filtrate. Furthermore, spectacular spills (<strong>for</strong> example,<br />
the Sandoz accident on November 1, 1986 [Sontheimer, 1991]), underlined the need <strong>for</strong><br />
sanitation measures and pollution control. The activities of waterworks and water-industry<br />
associations, authorities and industries, transborder programs, and the closure of industries all
esulted in a significant improvement of river-water quality (e.g., the Rhine River since 1980 to<br />
1985 and the Elbe River since 1990). Historic water-quality records show a strong decrease in the<br />
concentrations of phosphate, polynuclear aromatic hydrocarbon compounds, pesticides such as<br />
atrazine and mecoprop, and biodegradable organic carbon.<br />
Raw-<strong>Water</strong> Quality<br />
A summary evaluation of elimination processes at bank-filtration sites in The Netherlands showed<br />
an effective, sustainable, and redox-independent elimination of polycyclic aromatic hydrocarbons,<br />
polychlorinated biphenyls, chloroorganic pesticides, bromo<strong>for</strong>m, dichloroaniline, nitrobenzene,<br />
nitrotoluene, and chlornitrobenzene (Stuyfzand, 1998). Anoxic conditions cause an effective<br />
elimination of pesticides, such as atrazine, diurone, and simazine, which are less (or not) degraded<br />
under aerobic conditions.<br />
During bank filtration at the Rhine River, the concentrations of DOC, AOX, and sulphur<br />
compounds in Rhine water are reduced by more than 50 percent during a mean retention time of<br />
the infiltrate in the aquifer of 6 to 8 weeks (Brauch and Kuehn, 1988). Concentrations of<br />
malodorous compounds such as geosmin and monoterpenes decrease through degradation<br />
processes during bank filtration by 80 to 99 percent (Juettner, 1995). During bank filtration at<br />
Schlachtensee, Wannsee, and Tegeler See lakes in Berlin, Germany, algae-born terpenes causing<br />
malodor of lake waters were not found in bank filtrate (Chorus et al., 1992). Complex studies on<br />
biogeochemical reactions are reported by Bourg and Bertin (1993), Sontheimer (1991), and<br />
Nestler et al. (1998).<br />
The long-term monitoring of water quality has demonstrated the effective removal of a range of<br />
contaminants, including ammonium and polar compounds. Results from monitoring programs at<br />
sites on the Rhine, Elbe, and Danube rivers and at Tegel Lake have been recently summarized by<br />
Kuehn and Mueller (2000), Brauch et al. (2000), Grischek (2002), Mucha et al. (2002), and Fritz<br />
(2002), respectively.<br />
Drinking-<strong>Water</strong> Treatment<br />
In the 1960s and 1970s, one of the major problems was bad odor of drinking water derived from<br />
bank filtrate. The main ways to solve this problem were a reducing the organic load of the river<br />
water and using GAC filtration as a post-treatment step (Sontheimer, 1980).<br />
At some sites, advanced technologies such as ozone treatment, biological filtration, or GAC<br />
adsorption were established to treat the pumped infiltrate (e.g., along the Rhine River). At other<br />
sites, simple treatment technologies, such as pH-adjustment, aeration with sand filtration <strong>for</strong> iron<br />
and manganese removal, or disinfection by chlorine, are sufficient <strong>for</strong> meeting today’s drinkingwater<br />
standards (e.g., along the Elbe River). The necessary treatment st<strong>eps</strong> are still in discussion,<br />
especially due to recent findings of persistent organic trace compounds in bank filtrate.<br />
In Germany, different tests have been developed to classify compounds into compound classes that<br />
are not removed during aquifer passage and not removed by common drinking-water treatment<br />
(Sontheimer, 1991; Mueller et al., 1993). This concept has been used to identify compounds that<br />
are problematic in terms of drinking-water quality and to initiate source control measures <strong>for</strong><br />
reducing the input of these mobile substances into receiving rivers.<br />
107
108<br />
Problems<br />
At several sites (e.g., Budapest, Hungary and Dresden, Germany, or at the lower Rhine River), the<br />
quality of backside groundwater beneath the river floodplain, especially regarding the nitrate<br />
concentration, is becoming a greater problem than bank-filtrate quality.<br />
Problems with persistent organic compounds (e.g., pharmaceuticals) are reported <strong>for</strong> bank<br />
filtration in Berlin, Germany, where the percentage of surface waters comprised of municipal<br />
sewage effluents is relatively high (Ziegler et al., 2000, 2002). Recent work has shown that<br />
pharmaceuticals, with sources including feed additives, animal drugs, and human medical care,<br />
can be persistent in the aquatic environment. Trace-level analysis shows the presence of drug<br />
residues such as clofibric acid (a metabolite of a blood lipid regulator in medical care) in municipal<br />
sewage effluents and in receiving surface waters. As polar compounds, clofibric acid and several<br />
other pharmaceuticals, such as diclofenac, propyphenazone, carbamazepine, and primidone, are<br />
highly mobile (Heberer, 2002). Findings of pharmaceutical residues in drinking water are not<br />
desirable from a hygienic point of view; nevertheless, from today’s knowledge, these low<br />
concentrations do not have any toxicological impacts on human health. A limited removal of the<br />
polar drug residues is possible using activated carbon filtration. In summary, polar-peristent<br />
synthetic organic contaminants such as pharmaceuticals and industrial compounds require greater<br />
attention to assess the possibility, if any, of attenuation during bank filtration.<br />
This topic is one of the major issues in a large-scale research program of the new Berlin Centre of<br />
Competence on <strong>Water</strong>, called NASRI, which commenced in May 2002. <strong>Research</strong> activities include<br />
several field-site investigations with bank filtration and recharge, as well as semi-technical and<br />
laboratory studies on hydrogeology, modeling, algal toxins, viruses and bacteria, natural and<br />
anthropogenic organic compounds, and the clogging processes. Results will be presented in this<br />
survey, as well as in other presentations out of the NASRI-group at this conference.<br />
Conclusions<br />
For the future, bank filtration is viewed as a promising approach to increase limited groundwater<br />
resources and to provide a sustainable pretreatment step with multiple-barrier functions regarding<br />
chemical and microbial parameters.<br />
A new period in the exchange of worldwide experience started in November 1999 with the<br />
Inernational Riverbank Filtration Conference in Louisville, Kentucky (United States), followed by a<br />
workshop on the Attenuation of Groundwater Pollution by Bank Filtration in June 2000 in Dresden,<br />
Germany, an International Riverbank Filtration Conference in November 2000 in Düsseldorf, Germany,<br />
and a NATO Advanced <strong>Research</strong> Workshop on Riverbank Filtration on Understanding Contaminant<br />
Biogeochemistry and Pathogen Removal, which was held in September 2001 in Tihany, Hungary.<br />
REFERENCES<br />
Bourg, A.C.M., and C. Bertin (1993). “Biogeochemical processes during the infiltration of river water into<br />
an alluvial aquifer.” Environ. Sci. Technol., 27(4): 661-666.<br />
Brauch, H.-J., and W. Kuehn (1988). “Organische Spurenstoffe im Rhein und bei der Trinkwasseraufbereitung.”<br />
gwf Wasser Abwasser, 129(3): 37-44 (in German).<br />
Brauch, H.-J., U. Mueller, and W. Kuehn (2000). “Experiences with riverbank filtration in Germany.”<br />
Proceedings, International Riverbank Filtration Conference, IAWR, 4: 33-39.<br />
Chorus, I., G. Klein, J. Fastner, and W. Rotard (1992). “Off-flavors in surface waters - How efficient is bank<br />
filtration <strong>for</strong> their abatement in drinking water?” Wat. Sci. Technol., 25(2): 251-258.
Fritz, B. (2002). Uferfiltration unter verschiedenen wasserwirtschaftlichen, hydrogeologischen und hydraulischen<br />
Bedingungen. (Investigations of river bank filtration influenced by different hydrogeological, hydraulic and water<br />
management systems), PhD thesis, Free University of Berlin (in German).<br />
Grischek, T. (2002). Zur Bewirtschaftung von Uferfiltratfassungen an der Elbe. (Management of bank filtration<br />
sites along the River Elbe), PhD thesis, Dresden University of Technology (in German).<br />
Grischek, T., D. Schoenheinz, E. Worch, and K. Hiscock (2002). “Bank filtration in Europe - An overview<br />
of aquifer conditions and hydraulic controls.” Management of aquifer recharge <strong>for</strong> sustainability, P. Dillon (ed.),<br />
Swets & Zeitlinger, Balkema, Lisse, 485-488.<br />
Heberer, T. (2002). “Occurrence, fate, and removal of pharmaceutical residues in the aquatic environment:<br />
a review of recent date.” Toxicology Letters, 131: 5-17.<br />
Heeger, D. (1987). Investigations on clogging of river beds, Unpublished PhD-thesis (in German).<br />
Ingerle, K., A.P. Blaschke, A. Brugger, C. Hasenleithner, G.J. Herndl, H. Jarosch, I. Kolar, H.J. Lensing, S.<br />
Pöschl, N. Queric, B. Reitner, F. Schoeller, R. Sommer, and B. Wett (1999). Forschungsprojekt Uferfiltrat<br />
(<strong>Research</strong> Project Bank Filtration), <strong>Research</strong> Initiative Verbund, Vienna 60 (in German).<br />
Kivimäki, A.L. (1995). “Production of artificially recharged groundwater using bank filtration” Publication of<br />
the <strong>Water</strong> and Environment Administration 573, Helsinki, Finland (in Finnish).<br />
Kuehn, W., and U. Mueller (2000). “Riverbank filtration: An overview.” Journal AWWA, 92, (12): 60-69.<br />
Mucha, I., D. Rodak, Z. Hlavaty, and L. Bansky (2002). “Groundwater quality processes after bank<br />
infiltration from the Danube at Cunkovo.” Riverbank Filtration: Understanding Contaminant Biogeochemistry<br />
and Pathogen Removal, C. Ray (ed.), Kluwer Academic Publishers, The Netherlands, 177-219.<br />
Mueller, U., B. Wricke, and H. Sontheimer (1993). “Wasserwerks- und trinkwasserrelevante Substanzen in<br />
der Elbe.” Vom Wasser 81, 371-386 (in German).<br />
Nestler, W., W. Walther, F. Jacobs, R. Trettin, and K. Freyer (1998). “<strong>Water</strong> production in alluvial aquifers<br />
along the River Elbe.” UFZ-<strong>Research</strong> Report 7, 203 p. (in German).<br />
Schubert, J. (2001). “How does it work? Field studies on riverbank filtration.” Proceedings, International<br />
Riverbank Filtration Conference, 2-4 Nov. 2000, Düsseldorf, Germany, IAWR-Rheinthemen 4, 41-55.<br />
Schubert, J. (2000). “Entfernung von Schwebstoffen und Mikroorganismen sowie Verminderung der<br />
Mutagenität bei der Uferfiltration (Removal of suspended matter and microorganisms and reduction of<br />
mutagenity during bank filtration).” gwf Wasser Abwasser, 14(1/4): 218-225 (in German).<br />
Sontheimer, H. (1980). “Experience with riverbank filtration along the Rhine River.” Journal AWWA,<br />
72: 386-390.<br />
Sontheimer, H. (1991). Drinking water from the River Rhine? Academia Verlag, Sankt Augustin (in German).<br />
Stuyfzand, P.J. (1998). “Fate of pollutants during artificial recharge and bank filtration in the Netherlands.”<br />
Artificial Recharge of Groundwater, Balkema, Rotterdam, 119-125.<br />
Ziegler, D., C. Hartig, S. Wischnack, and M. Jekel (2000). “Behaviour of dissolved organic compounds and<br />
pharmaceuticals during lake bank filtration in Berlin.” Proceedings, International Riverbank Filtration<br />
Conference, IAWR, 4: 151-160.<br />
Ziegler, D., C. Hartig, S. Wischnack, and M. Jekel (2002). “Organic substances in partly closed water cycles.”<br />
Management of aquifer recharge <strong>for</strong> sustainability, P. Dillon (ed.), Swets & Zeitlinger, Balkema, Lisse, 161-167.<br />
109
110<br />
<strong>Water</strong> chemist and treatment engineer MARTIN JEKEL has about 28 years of experience<br />
in the field of water and wastewater treatment, as well as in water-quality analysis. Since<br />
1988, he has been a full Professor <strong>for</strong> water-quality control at the <strong>Institute</strong> <strong>for</strong><br />
Environmental Engineering of the Technical University of Berlin. His research interests<br />
include the development of the “Muelheim Process” <strong>for</strong> drinking-water treatment, basic<br />
and applied studies on the preoxidation mechanisms in coagulation of water treatment,<br />
advanced treatment processes <strong>for</strong> indirect potable water reuse, a new adsorption technique<br />
with Granular Ferric Hydroxide <strong>for</strong> solving the worldwide problem of arsenic in groundwater, and the<br />
analysis and fate of new organic trace organics, such as iodinated X-ray contrast agents. During the last<br />
5 years, he has conducted several studies on the processes of organics removal in lake bank filtration in the<br />
Berlin area, and he is Scientific Coordinator of Natural and Artificial Systems <strong>for</strong> Recharge and Infiltration,<br />
one of the largest bank-filtration and recharge research programs worldwide. Jekel received a diploma<br />
(M.Sc.) in Chemistry, a Ph.D. in Chemical Engineering, and the German degree of Habilitation<br />
(qualification <strong>for</strong> a tenure professorship) at the University of Karlsruhe, Germany.
Session 6: Microorganisms<br />
Using Microscopic Particulate Analysis<br />
<strong>for</strong> Riverbank Filtration<br />
Jennifer L. Clancy, Ph.D.<br />
Clancy Environmental Consultants, Inc.<br />
St. Albans, Vermont<br />
William D. Gollnitz<br />
Greater Cincinnati <strong>Water</strong> Works<br />
Cincinnati, Ohio<br />
Microscopic Particulate Analysis has been used since the mid-1980s as a tool to examine the<br />
nature of biological particulates in water. This methodology is used <strong>for</strong> two purposes:<br />
• To examine groundwater <strong>for</strong> the presence of biological surface-water indicators. These<br />
species may indicate that the groundwater source is GWUDI and, hence, subject to the<br />
requirements of the Surface <strong>Water</strong> Treatment Rule.<br />
• To determine filtration efficiency by comparing the types of particulates in raw and<br />
finished waters. The log removal of particulates is calculated to determine the efficiency<br />
of the filtration process. Both full-scale and pilot processes can be evaluated using<br />
Microscopic Particulate Analysis.<br />
Microscopic Particulate Analysis samples are collected by filtering a relatively large volume of<br />
water through a cartridge filter. The sample is generally several hundred gallons, but volume can<br />
be adjusted depending on the type of sample. The filter is shipped overnight on ice to the laboratory.<br />
The filter is cut apart and the fibers washed or backwashed directly (Envirochek capsule) to<br />
remove particulate matter. The particulates are concentrated into a pellet, a portion of which may<br />
be subjected to buoyant density gradient centrifugation to separate biological particulates from<br />
heavier debris. This material is examined in two ways:<br />
• Direct microscopic observation using brightfield and phase or differential interference<br />
contrast to identify, enumerate, and classify biological particulates.<br />
• A portion of the sample may be stained using fluorescent monoclonal antibodies specific<br />
to Giardia cysts and Cryptosporidium oocysts. The sample is examined using epifluorescence<br />
microscopy to detect the presence of these parasites.<br />
The data are reported as the numbers and types of biological particulates. Categories include, but<br />
are not limited to, algae, diatoms, rotifers, crustaceans, insects, protozoa, plant debris, Giardia cysts,<br />
and Cryptosporidium oocysts. The log reduction of each bio-indicator category is determined by<br />
comparing the numbers of particulates in raw and finished water samples. A filtration per<strong>for</strong>mance<br />
rating is determined, and the significance of any additional data or observations is explained.<br />
Other in<strong>for</strong>mation that can be obtained includes the presence of alum/polymer floc, carbon fines,<br />
Correspondence should be addressed to:<br />
Jennifer L. Clancy, Ph.D.<br />
President<br />
Clancy Environmental Consultants, Inc.<br />
P.O. Box 314 • St. Albans, Vermont 05478 USA<br />
Phone: (802) 527-2460 • Fax: (802) 524-3909 • Email: jclancy@clancyenv.com<br />
111
112<br />
and other material in the finished water. The overall microscopic quality of the water can be readily<br />
assessed using Microscopic Particulate Analysis. If a surface-water sample has been analyzed<br />
simultaneously, a comparison of particulates in surface-water and groundwater samples can be made.<br />
The rationale in examining groundwater using Microscopic Particulate Analysis is that if<br />
surface-water indicators are observed, then the source may be at risk <strong>for</strong> contamination by Giardia<br />
cysts and Cryptosporidium oocysts. Groundwaters that are at risk of surface-water influence are<br />
considered to be surface waters under the Surface <strong>Water</strong> Treatment Rule and are required to<br />
comply with filtration and disinfection requirements. An obvious extension of this technique is to<br />
examine water samples in systems using <strong>RBF</strong>. Similar to conducting a filtration plant efficiency<br />
study, a river or source sample can be compared to samples from the groundwater collection<br />
device, and a determination of the level of in situ natural filtration occurring through the aquifer<br />
can be made. This idea of natural filtration to improve water quality is not new. The concept of<br />
using the aquifer as a natural filtration device was termed “<strong>RBF</strong>” over 100 years ago, although the<br />
process has been used <strong>for</strong> centuries. <strong>RBF</strong> is a standard surface-water treatment practice in Europe.<br />
In 1997, the authors first proposed considering natural filtration when per<strong>for</strong>ming GWUDI<br />
evaluations of collection devices located in porous media aquifers. In a long-term study of<br />
groundwater collection devices in the North Platte Alluvial Aquifer in Casper Wyoming, the<br />
authors found that natural filtration through the aquifer produced finished water of superior<br />
quality to that produced by the water filtration plant, both using the North Platte River as a<br />
source. Surface-water indicators (algae and diatoms) were found in some of the groundwater<br />
collection devices, and always at significantly lower concentrations than in surface water. Even<br />
when algae and diatoms were present, cysts and oocysts were never observed in any of the<br />
groundwater samples. The reduction of algae and diatoms through the aquifer ranged from 4.2 to<br />
greater than 5 log, while the removal range in the plant, which consistently met all compliance<br />
requirements, ranged from 0.3 to 1.5 log.<br />
The key to providing adequate <strong>RBF</strong> is the aquifer type. Natural filtration through porous media<br />
aquifers, particularly those with a matrix in the sand- and gravel-size range, is very effective <strong>for</strong><br />
particulate removal. Natural filtration is similar to engineered filtration in that we expect to see a<br />
reduction of particulates through the process; the end result is not necessarily the observation of<br />
no surface-water indicators, but adequate removal so as to minimize the risks to public health. In<br />
the Surface <strong>Water</strong> Treatment Rule, this is defined as 2-log removal of Giardia cysts through<br />
filtration, and in the Long Term 2 Enhanced Surface <strong>Water</strong> Treatment Rule (LT2ESWTR), 3-log<br />
Cryptosporidium removal through some combination of processes, including <strong>RBF</strong> removal credit.<br />
This paper examines the filtration efficiencies of six surface water treatment plants and six systems<br />
using in situ natural filtration or <strong>RBF</strong>. Each system was monitored regularly using Microscopic<br />
Particulate Analysis <strong>for</strong> a minimum of 1 to 11 years. Samples were collected in pairs (raw and<br />
treated surface water or river water and groundwater collection device) and analyzed by a single<br />
group of analysts, minimizing some of the inherent variability noted between laboratories and<br />
analysts. Sample sites are located in the United States and Canada, and the selected sites represent<br />
a broad geographic region (Cali<strong>for</strong>nia to Connecticut). Over 1,400 samples were analyzed in this<br />
comparison.<br />
As noted in previous studies, surface-water samples showed the largest diversity and concentrations<br />
of microorganisms. Algae and diatoms occurred in the highest concentrations, ranging from l0 2 to<br />
greater than 10 9 per 100 gallons. The reduction of indicators through engineered filtration ranged<br />
from no reduction to greater than 6 log 10. Log 10 reduction of biological indicators varied within<br />
treatment plants, and poorer filtration per<strong>for</strong>mance noted using Microscopic Particulate Analysis
was independent of finished water turbidity. All plants in the study were considered well-operated<br />
and met the requirements of the Surface <strong>Water</strong> Treatment Rule. In some cases, the levels of algae<br />
and diatoms in finished water were actually higher than that observed in raw water. This may be<br />
due to sampling anomalies, growth of algae in the filters, poor filtration per<strong>for</strong>mance, or vagaries<br />
associated with the Microscopic Particulate Analysis method itself.<br />
For <strong>RBF</strong> groundwater collection devices, log 10 reduction of algae and diatoms ranged from about<br />
4 log 10 to greater than 7 log 10. Reductions across the aquifer through natural filtration were far<br />
more consistent, and the minimum level of reduction was always greater than that required <strong>for</strong><br />
filtration per<strong>for</strong>mance under the Surface <strong>Water</strong> Treatment Rule (2-log 10 Giardia and 3-log 10<br />
Cryptosporidium). No Giardia cysts or Cryptosporidium oocysts were noted in any of the <strong>RBF</strong><br />
groundwater collection device samples.<br />
While all samples met turbidity requirements, the biological water quality measured using<br />
Microscopic Particulate Analysis was far superior in the <strong>RBF</strong> groundwater collection device<br />
samples than that noted in the treatment plant effluents. This same observation has been noted<br />
when comparing surface-water treatment plant effluents and groundwater samples in previous<br />
studies. <strong>RBF</strong> is a more effective process than engineered filtration <strong>for</strong> removing biological particles<br />
from surface-water sources.<br />
JENNIFER CLANCY is a microbiologist with 25 years of experience in environmental<br />
microbiology, focusing on water. She is currently President of Clancy Environmental<br />
Consultants, Inc., which provides microbiological testing and consulting services to the<br />
water and wastewater industries. Clancy has spent years studying the issue of groundwater<br />
under the direct influence of surface water in systems in the United States and Canada.<br />
Prior to <strong>for</strong>ming Clancy Environmental Consultants, Inc., in 1994, she was the Director<br />
of <strong>Water</strong> Quality at the Erie County <strong>Water</strong> Authority in Buffalo, New York, and was<br />
responsible <strong>for</strong> administration of the <strong>Water</strong> Quality Department. Clancy received a B.S. in Microbiology<br />
from Cornell University, an M.S. in Microbiology and Biochemistry from the University of Vermont, a Ph.D.<br />
in Microbiology and Immunology from McGill University, and an M.S. in Environmental Law from Vermont<br />
Law School.<br />
113
114
Session 6: Microorganisms<br />
Transport and Removal of Cryptosporidium Oocysts<br />
in Subsurface Porous Media<br />
Menachem Elimelech, Ph.D.<br />
Yale University<br />
New Haven, Connecticut<br />
Garrett Miller<br />
Yale University<br />
New Haven, Connecticut<br />
Zachary Kuznar<br />
Yale University<br />
New Haven, Connecticut<br />
Cryptosporidium parvum contamination is considered one of the most important water-supply<br />
problems today. Disinfectants commonly used by water treatment plants are ineffective at reducing<br />
Cryptosporidium parvum risk, and deep-bed filtration represents one of the main barriers to oocyst<br />
contamination. <strong>RBF</strong>, which is gaining popularity in the United States, presents an alternative<br />
method of pretreating water to remove oocysts. Very little research has been done so far to<br />
understand the mechanisms controlling the transport and filtration behavior of Cryptosporidium in<br />
saturated porous media under chemical and physical conditions relevant to <strong>RBF</strong>. The objective of<br />
this study was to elucidate the role of electrostatic double layer interactions in the attachment and<br />
transport of Cryptosporidium in flow through saturated porous media. Well-controlled column<br />
experiments were carried out using ultra-clean quartz sand and Cryptosporidium parvum oocysts.<br />
Complimentary experiments using a stagnation point flow system have been conducted under<br />
identical chemical and hydrodynamic conditions. Initial bacterial cell deposition rates are compared<br />
with column breakthrough curves, and the results are used to highlight the controlling mechanisms<br />
of Cryptosporidium parvum adhesion and transport.<br />
MENACHEM ELIMELECH is the Llewellyn West Jones Professor of Environmental<br />
Engineering and Director of the Environmental Engineering Program at Yale University.<br />
His research interests center on problems involving physicochemical and colloidal<br />
processes in engineered and natural systems, and he has worked on the dynamics of colloid<br />
transport and deposition in porous media, transport and fate of microbial particles (viruses,<br />
bacteria, and Cryptosporidium) in the subsurface environment, and contaminant removal<br />
by membrane processes. Elimelech is the author of over 90 journal publications and is the<br />
principal author of Particle Deposition and Aggregation (1995). Among his recent honors, he was the recipient<br />
of the Association of Environmental Engineering and Science Professors (AEESP) Outstanding Paper Award<br />
in 2002 and the AEESP Doctoral Dissertation Award, with his graduate student Eric M.V. Hoek, in 2002.<br />
He also has served on the Editorial Advisory Boards of Environmental Science & Technology, Environmental<br />
Engineering Science, Desalination, and Journal of Colloid and Interface Science. Elimelech received both a B.S. in<br />
Soil and <strong>Water</strong> Sciences and an M.S. in Environmental Science and Technology from the Hebrew<br />
University in Jerusalem and a Ph.D. in Environmental Engineering from The Johns Hopkins University.<br />
Correspondence should be addressed to:<br />
Menachem Elimelech, Ph.D.<br />
Llewellyn West Jones Professor of Environmental Engineering<br />
Department of Chemical Engineering • Environmental Engineering Program<br />
Yale University • P.O. Box 208286 • New Haven, Connecticut 06520-8286 USA<br />
Phone: (203) 432-2789 • Fax: (203) 432-2881 • Email: menachem.elimelech@yale.edu<br />
115
116
Session 6: Microorganisms<br />
Laboratory and Field Strategies <strong>for</strong> Assessing<br />
Pathogen Removal by Riverbank Filtration<br />
Monica B. Emelko, Ph.D.<br />
University of <strong>Water</strong>loo<br />
<strong>Water</strong>loo, Ontario, Canada<br />
Mark T. Watling<br />
University of <strong>Water</strong>loo<br />
<strong>Water</strong>loo, Ontario, Canada<br />
Martin M. Côté<br />
University of <strong>Water</strong>loo<br />
<strong>Water</strong>loo, Ontario, Canada<br />
Introduction<br />
<strong>RBF</strong> systems can significantly reduce the concentrations of many surface-water pollutants<br />
(Shubert, 2000; Wang et al., 2000; Kuehn and Mueller, 2000); however, predicting and<br />
quantifying those reductions is difficult. An understanding of subsurface pathogen transport,<br />
particularly Cryptosporidium and viruses, is critical <strong>for</strong> utilities faced with potential GWUDI of<br />
surface-water wells. In the United States, the LT2ESWTR prescribes Cryptosporidium removal<br />
credits (0, 0.5, or 1.0 log) based on aquifer grain-size distribution and the distance between the<br />
well and riverbed (USEPA, 2001). Regulations in Ontario, Canada, require utilities to<br />
demonstrate in situ filtration per<strong>for</strong>mance using particle counts (less than 100 particles greater<br />
than or equal to 10 micrometers per milliliter) and qualitative assessments of potential temporal<br />
changes in well effluent particle counts and raw-water quality (Ontario Ministry of the<br />
Environment, 2001). While these approaches represent an attempt at demonstrating the efficacy<br />
of <strong>RBF</strong> as a treatment technology <strong>for</strong> reducing pathogen concentrations in drinking-water sources,<br />
it is generally acknowledged that they do not provide adequate assessments of pathogen removal<br />
by <strong>RBF</strong>.<br />
Assessments of pathogen removal by <strong>RBF</strong> are difficult. Although column studies may approximate<br />
some <strong>RBF</strong> per<strong>for</strong>mance, they are of limited value because they often fail to adequately represent<br />
temporal changes in regional groundwater conditions or microbiological activity, geologic<br />
heterogeneity, and physical scale (filtration length) of the true system. While full-scale<br />
investigations are possible and have been conducted, they are costly and plagued by analytical<br />
limitations such as low indigenous pathogen concentrations (that preclude representative<br />
sampling) and, in the case of Cryptosporidium, unreliable analytical methods with low and highly<br />
variable recoveries (Nieminski et al., 1995; Clancy et al., 1999). Full-scale investigations are<br />
further complicated by the use of more readily analyzed, but unproven, surrogates (which is<br />
Correspondence should be addressed to:<br />
Monica B. Emelko, Ph.D.<br />
Assistant Professor<br />
Department of Civil Engineering<br />
University of <strong>Water</strong>loo • <strong>Water</strong>loo, Ontario N2L 3G1 Canada<br />
Phone: (519) 888-4567 ext. 2208 • Fax: (519) 888-6197 • Email: mbemelko@uwaterloo.ca<br />
117
118<br />
necessitated because of the a<strong>for</strong>ementioned analytical limitations). As a result, the outcomes of<br />
full-scale per<strong>for</strong>mance or surrogate validation studies are difficult to interpret and compare,<br />
hindering the development of widely accepted protocols <strong>for</strong> assessing pathogen removal by <strong>RBF</strong>.<br />
Objectives<br />
In this paper, we highlight some recent laboratory and analytical advances that are useful <strong>for</strong><br />
improving assessments of pathogen removal by <strong>RBF</strong>. The specific objectives of this work include:<br />
• Investigating the impact of column orientation (and the associated impacts of<br />
gravitational <strong>for</strong>ce) on pathogen removal in laboratory studies investigating <strong>RBF</strong><br />
per<strong>for</strong>mance.<br />
• Quantitatively demonstrating the impact of the actual number of discrete particles<br />
(e.g., pathogens or surrogates) counted from a sample on the uncertainty of inferences<br />
drawn from experimental and monitoring data.<br />
• Proposing pathogen or surrogate count targets so that reasonably certain conclusions can<br />
be drawn from experimental or monitoring data.<br />
Methodology<br />
Column Studies<br />
It has been suggested that column orientation (horizontal versus vertical versus angled) may be an<br />
important parameter in mimicking subsurface transport and filtration due to the impacts of<br />
sedimentation (Harvey, 1997). To investigate the impact of the gravitational <strong>for</strong>ce on pathogen<br />
removal during column studies, four 10-cm filter columns with 6-cm internal diameters were<br />
packed with sieved aquifer material and were operated at conditions typical of those that may be<br />
encountered during <strong>RBF</strong> (flow of ~1 milliliter per minute). Two of the columns were operated in<br />
a horizontal mode and two of the columns were operated in an upflow vertical mode. Preliminary<br />
investigations were conducted at water-quality conditions that do not favor pathogen removal<br />
(i.e., low ionic strength).<br />
After the columns were flushed with degassed water <strong>for</strong> several hours, the influent reservoir —<br />
which fed the four columns simultaneously — was spiked with both Cryptosporidium oocysts and<br />
oocyst-sized polystyrene microspheres. The influent oocyst and microsphere concentrations were<br />
determined from influent reservoir samples. Effluent samples were collected at intervals of<br />
0.25-pore volumes until at least 6-pore volumes of influent had passed through the columns.<br />
Microsphere removals from samples collected during the passage of 0.3- to 5.2-pore volumes<br />
through the columns are discussed herein.<br />
Polycarbonate membranes (25-millimeter, 0.40-micrometer nominal porosity) were used to filter<br />
the samples. Samples were filtered directly on a manifold with a vacuum of ~5-inches of mercury.<br />
Microsphere enumeration was per<strong>for</strong>med at 100× and 400× magnification (Nikon Labophot 2A,<br />
Nikon Canada Inc., Toronto, Ontario, Canada). The filtered sample volumes were selected to<br />
yield between 10 and 1,000 microspheres per membrane.<br />
Data Reliability and Interpretation<br />
The occurrence of waterborne pathogens such as Cryptosporidium is difficult to measure precisely<br />
because they are present in varied concentrations, often at concentrations so low that detection<br />
is difficult (Lisle and Rose, 1995). Not surprisingly, current methods <strong>for</strong> quantifying
Cryptosporidium are unreliable, laborious, and expensive. Analytical recoveries are often low and<br />
highly variable (Nieminski et al., 1995; Clancy et al., 1999).<br />
Nahrstedt and Gimbel (1996) proposed a model <strong>for</strong> the process of estimating the concentration<br />
of Cryptosporidium oocysts in a body of water; it addresses uncertainty associated with sampling and<br />
enumeration. The authors used:<br />
• A Poisson distribution to model true sample counts (representative sampling).<br />
• A binomial distribution to model the recovered fraction of oocysts (random analytical error).<br />
• A Beta distribution to describe non-constant analytical recovery.<br />
The authors provided only limited in<strong>for</strong>mation on how to apply their model to obtain<br />
concentration estimates and to quantify the uncertainty in these estimates.<br />
To obtain both concentration and removal estimates and to quantify the associated uncertainties,<br />
Emelko (2001) presented a Bayesian analysis of the Nahrstedt and Gimbel (1996) model and<br />
utilized the Gibbs sampler to provide easy access to a wide variety of estimated quantities<br />
(e.g., point estimates, probability [confidence] intervals, probability density functions, etc.). In<br />
brief, this approach results in a joint posterior probability density function <strong>for</strong> the true<br />
concentration of pathogens in the water body (c), the probability of pathogen recovery from the<br />
water sample (p), and the number of pathogens in the water sample (N) given the observed<br />
pathogen count (X). The uncertainty of recovery is accounted <strong>for</strong> by the Beta distribution’s shape<br />
parameters (a and b). Replicate sampling is incorporated, if applicable (n replicate samples). The<br />
posterior probability density function, after simplification and suppressing constant multipliers, is:<br />
Df [(c,p,Ni|Xi),i = 1,…, n] ∝ c –1+ Σ N i×e –c+Σ V i×p a–1+Σ X i×(1–p) b–1+Σ(N i –X i ) ×Π<br />
Since the parameter of interest is the true concentration c, the remaining (nuisance) parameters<br />
may be integrated out (Box and Tiao, 1973). This operation finds the mathematical expectation<br />
of c over all values of the nuisance parameters. The posterior distribution given by Equation 1 is<br />
very difficult, as it stands, to integrate or to use directly to make inferences about c. The Gibbs<br />
sampler, a Monte Carlo method that produces an indefinitely long Markov chain of vectors of<br />
parameter values (Geman and Geman, 1984), can handle these problems with relative ease.<br />
Applicable to all discrete particles, this technique was used in the present investigation; details<br />
regarding the development of this approach were provided in Emelko (2001).<br />
The impact of the actual number of pathogens or surrogates counted from a sample on uncertainty<br />
(confidence interval <strong>for</strong> pathogen concentration) is demonstrated herein using a concentration of<br />
one pathogen per 100 liters. Counts between one and 1,000 pathogens (and the associated sample<br />
volumes that would yield a concentration of one pathogen per 100 liters) were utilized. The recovery<br />
data and associated Beta parameters (a = 28.12, b = 8.43) used in this assessment correspond to oocyst<br />
recoveries ranging from 69 to 85 percent and average 77 percent, as reported by Emelko (2001).<br />
Results<br />
Column Studies<br />
n<br />
i=1<br />
n<br />
i=1<br />
Preliminary investigations of Cryptosporidium oocyst-sized polystyrene microsphere removal by<br />
horizontal and upflow vertical columns are summarized in a box and whisker plot (Figure 1). This<br />
figure illustrates that a consistent level of microsphere removal was achieved during the period<br />
n<br />
i=1<br />
n<br />
i=1<br />
n<br />
i=1<br />
Vi N i<br />
(N i –N i)<br />
(1)<br />
119
120<br />
Microsphere Removal (log 10 )<br />
5<br />
4<br />
3<br />
2<br />
1<br />
0<br />
Vertical #1 Vertical #2 Horizontal #1 Horizontal #2<br />
Figure 1. Cryptosporidium-sized microsphere removal by 10-centimeter columns operated in upflow vertical<br />
and horizontal orientations.<br />
that 0.3- to 5.2-pore volumes of water passed through the columns. In addition, the data suggest<br />
slightly higher removal of oocyst-sized microspheres by the horizontal columns. Moreover,<br />
excellent reproducibility between duplicate columns was achieved; this is also evident in the<br />
breakthrough curves (not shown). It should be noted that some microsphere loss occurred in the<br />
sample line prior to reaching the columns. The overall microsphere removals are, there<strong>for</strong>e,<br />
somewhat high; however, given that the columns received the same influent water (losses<br />
occurred be<strong>for</strong>e the feed lines were split), the relative relationship between the columns should be<br />
unaffected. Continued experiments will be useful in determining whether these differences in<br />
microsphere removal as a function of column orientation are statistically significant.<br />
Data Reliability and Interpretation<br />
n=8 n=8 n=8 n=8<br />
Column<br />
The impact of the actual number of pathogens or surrogates counted from a sample on uncertainty<br />
(confidence interval <strong>for</strong> pathogen concentration) was demonstrated using a concentration of<br />
one pathogen per 100 liters (Figure 2). Figure 2 illustrates that the uncertainty associated with<br />
pathogen concentration data can be considerably reduced by increasing the number of pathogens<br />
that are counted in a sample; naturally, this can be achieved by increasing the processed sample<br />
volume. Similar to the present analysis, Emelko and Reilly (2002) demonstrated the same impact<br />
on uncertainty when examining confidence intervals <strong>for</strong> pathogen removal (as opposed to<br />
concentration) data. To avoid confidence intervals that extend beyond an order of magnitude,<br />
counts of approximately 10 or more pathogens (preferably, approximately 50 pathogens) are<br />
required from an individual sample. Only limited improvements in uncertainty are observed at<br />
counts greater than 50 pathogens per sample.<br />
It is important to note that the outcomes in Figure 2 are associated with the recovery profile of<br />
one analytical method. Somewhat different outcomes with respect to the range of the confidence<br />
intervals would be expected with different analytical methods (and their associated recovery<br />
profiles); however, the same general trend of increased counts resulting in decreased uncertainty
Pathogen Concentration (pathogens/100 Liters)<br />
10.<br />
1.<br />
0.1<br />
0.01<br />
0.001<br />
Figure 2. Ninety-five percent confidence intervals obtained <strong>for</strong> pathogen concentrations using the<br />
Cryptosporidium recovery profile of Emelko (2001).<br />
would be expected. This type of statistical analysis <strong>for</strong> already reported data can be somewhat<br />
difficult as adequate recovery data (actual counts) are often lacking from the reported literature,<br />
underscoring the need <strong>for</strong> guidance on how to adequately report pathogen recovery in<strong>for</strong>mation.<br />
The analysis in Figure 2 does not address the issues of replicate sampling or pooled data. Emelko<br />
and Reilly (2002) briefly discussed this issue and demonstrated that the largest improvements<br />
associated with decreasing data uncertainty were associated with increased total counts, regardless<br />
of the number of samples from which they were derived. The impact of pathogen recovery profiles<br />
on this relationship remains to be fully demonstrated.<br />
Conclusions<br />
1 2 5 10 50 100 500 1000<br />
Pathogen Count in Sample<br />
Several preliminary conclusions have resulted from this work. They include the following:<br />
• Pathogen removal appears to be slightly higher in horizontal laboratory columns as<br />
compared to upflow vertical columns. As more data are collected, the statistical significance<br />
of this outcome must be determined.<br />
• Column studies that are used to investigate <strong>RBF</strong> should be conducted using a column<br />
orientation that is as representative of the site as possible; in most cases, this orientation<br />
should be at or near horizontal.<br />
• To minimize the uncertainty associated with pathogen data so that it ranges over less<br />
than an order of magnitude, total counts of between 10 and 50 pathogens are required.<br />
This can be achieved via increased sample volume processing or increased replication.<br />
121
122<br />
REFERENCES<br />
Box, G.E., and G.C. Tiao (1973). Bayesian Inference in Statistical Analysis, Addison Wesley Publishing<br />
Company, Reading, MA.<br />
Clancy, J.L., Z. Bukhari, R.M. McCuin, Z. Matheson, and C.R. Fricker (1999). “USEPA Method 1622.”<br />
Journal AWWA, 91(9): 60-68.<br />
Emelko, M.B. (2001). Removal of Cryptosporidium parvum by granular media filtration, Ph.D. Dissertation,<br />
University of <strong>Water</strong>loo, <strong>Water</strong>loo, Ontario, Canada.<br />
Emelko, M.B., and P.M. Reilly (2002). “Reporting and Regulating Cryptosporidium Concentrations and<br />
Removals.” Proceedings, AWWA WQTC, AWWA, Denver, CO.<br />
Geman, S., and D. Geman (1984). “Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration<br />
of Images.” IEEE Trans. Pattn. Anal. Bach. Intell., 6: 721.<br />
Harvey, R.W. (1997). “In Situ Laboratory Methods to Study Subsurface Microbial Transport.” Manual of<br />
Environmental Microbiology, C.J. Hurst, G.R. Knudsen, M.J. McInerney, L.D. Stetzenbach, and M.V. Walter<br />
(eds.), ASM Press, Washington, D.C., p. 586-589.<br />
Kuehn, W., and U. Mueller (2000). “Riverbank filtration: An overview.” Journal AWWA, 82(12): 60-69.<br />
Lisle, J.T., and J.B. Rose (1995). “Cryptosporidium Contamination of <strong>Water</strong> in the U.S.A. and U.K.: A Mini<br />
Review.” Jour. <strong>Water</strong> SRT–Aqua, 44(3): 103.<br />
Nahrstedt, A., and R. Gimbel (1996). “A Statistical Method <strong>for</strong> Determining the Reliability of the Analytical<br />
Results in the Detection of Cryptosporidium and Giardia in <strong>Water</strong>.” Jour. <strong>Water</strong> SRT – Aqua, 45(3): 101.<br />
Nieminski, E.C., F.W. Schaefer, and J.E. Ongerth (1995). “Comparison of two Methods <strong>for</strong> Detection of<br />
Giardia Cysts and Cryptosporidium Oocysts in <strong>Water</strong>.” Appl. and Envir. Microbiol., 61(5): 1714.<br />
Ontario Ministry of the Environment (MOE) (2001). Terms of Reference, Hydrogeological Study to Examine<br />
Groundwater Sources Potentially Under Direct Influence of Surface <strong>Water</strong>, Ontario Ministry of the Environment.<br />
Shubert, J. (2000). “How does it work? Field Studies on Riverbank Filtration.” Proceedings, International<br />
Riverbank Filtration Conference, Düsseldorf, 2-4 November 2000, Internationale Arbeitsgemeinschaft der<br />
Wasserwerke im Rheineinsugsgebiet (IAWR), Düsseldorf.<br />
USEPA (2001). <strong>National</strong> Primary Drinking <strong>Water</strong> Regulations: Long Term 2 Enhanced Surface <strong>Water</strong> Treatment<br />
Rule, 40 CFR Parts 9, 141 and 142, United States Environmental Protection Agency.<br />
Wang, J.Z., R. Song, and S.A. Hubbs (2000). “Particle removal through riverbank filtration process.”<br />
Proceedings, International Riverbank Filtration Conference, Düsseldorf, 2-4 November 2000, Internationale<br />
Arbeitsgemeinschaft der Wasserwerke im Rheineinsugsgebiet (IAWR), Düsseldorf.<br />
MONICA EMELKO has research expertise in microbial pathogen and surrogate transport<br />
and removal by porous media, developing analytical methods <strong>for</strong> enumerating<br />
microorganisms during porous media investigations, and developing statistical tools <strong>for</strong><br />
describing analytical uncertainty. She was awarded the American <strong>Water</strong> Works<br />
Association Academic Achievement Award <strong>for</strong> her doctoral dissertation, and has over<br />
25 publications on pathogen transport and filtration in porous media systems. Emelko has<br />
been an advisor in the development of the Long-Term 2 Enhanced Surface <strong>Water</strong><br />
Treatment Rule and serves on several American <strong>Water</strong> Works Association <strong>Research</strong> Foundation advisory<br />
committees <strong>for</strong> projects focused on characterizing filter effluents and filtration per<strong>for</strong>mance; she is also the<br />
Vice-Chair of the Particulate Contaminants <strong>Research</strong> Committee <strong>for</strong> the American <strong>Water</strong> Works Association.<br />
At present, she is an Assistant Professor in the Department of Civil Engineering at the University of<br />
<strong>Water</strong>loo in Canada. Emelko received B.S degrees in Chemical Engineering and Environmental Engineering<br />
from the Massachusetts <strong>Institute</strong> of Technology, an M.S. in Civil Engineering from the University of<br />
Cali<strong>for</strong>nia, Los Angeles, and a Ph.D. in Civil Engineering from the University of <strong>Water</strong>loo.
Session 6: Microorganisms<br />
Fate of Disinfection Byproduct Precursors<br />
and Microorganisms During Riverbank Filtration<br />
W. Joshua Weiss<br />
The Johns Hopkins University<br />
Baltimore, Maryland<br />
Edward J. Bouwer, Ph.D.<br />
The Johns Hopkins University<br />
Baltimore, Maryland<br />
William P. Ball<br />
The Johns Hopkins University<br />
Baltimore, Maryland<br />
Charles R. O’Melia, Ph.D.<br />
The Johns Hopkins University<br />
Baltimore, Maryland<br />
Kellogg J. Schwab, Ph.D.<br />
Bloomberg School of Public Health, The Johns Hopkins University<br />
Baltimore, Maryland<br />
Binh T. Le<br />
Bloomberg School of Public Health, The Johns Hopkins University<br />
Baltimore, Maryland<br />
Ramon Aboytes, D.V.M., Ph.D.<br />
American <strong>Water</strong>, Belleville Laboratory<br />
Belleville, Illinois<br />
Experience with <strong>RBF</strong> in Europe and more recently in the United States has demonstrated<br />
significant improvements in raw-water quality, including the removal of NOM, biodegradable<br />
compounds, pesticides, microbes, and other water-quality contaminants and compensation <strong>for</strong><br />
shock loads of chemical contaminants (Kuehn and Mueller, 2000; Ray et al., 2002a and 2002b;<br />
Tufenkji et al., 2002; Weiss et al., 2003a and 2003b). Because of these potential improvements,<br />
regulators and utilities in the United States have recently looked more strongly at <strong>RBF</strong> as a means<br />
<strong>for</strong> providing high-quality sources <strong>for</strong> drinking water; however, little data are available to compare<br />
the per<strong>for</strong>mance of <strong>RBF</strong> with that of conventional drinking-water treatment processes more<br />
commonly used in the United States (e.g., coagulation, flocculation, sedimentation) from identical<br />
river-water sources, especially with regard to the removal of organic DBP precursor material. In<br />
addition, little is known about the extent to which <strong>RBF</strong> may serve to reliably remove Giardia,<br />
Cryptosporidium, and other pathogens (e.g., bacteria, viruses) from river water. In particular, data<br />
are needed on the transport of microbial pathogens through riverbank systems relative to that of<br />
more easily measured indicator parameters, such as particles and coli<strong>for</strong>m bacteria.<br />
Correspondence should be addressed to:<br />
W. Joshua Weiss<br />
Ph.D. Student/<strong>Research</strong> Assistant<br />
Department of Geography and Environmental Engineering<br />
The Johns Hopkins University • 3400 N. Charles Street • Baltimore, Maryland 21218 USA<br />
Phone: (410) 516-6220 • Fax: (410) 516-8996 • Email: jweiss@jhu.edu<br />
123
124<br />
In the above context, research was conducted to document water-quality benefits during <strong>RBF</strong> at<br />
three major river sources in the Midwestern United States (the Ohio River at Jeffersonville,<br />
Indiana; Wabash River at Terre Haute, Indiana; and Missouri River at Parkville, Missouri),<br />
specifically with regard to reducing DBP precursor organic matter and microbial pathogens.<br />
Specific objectives were to:<br />
1. Evaluate the merits of <strong>RBF</strong> <strong>for</strong> removing organic DBP precursor material.<br />
2. Evaluate whether <strong>RBF</strong> can improve finished drinking-water quality by removing and/or<br />
altering NOM in a manner that is not otherwise accomplished through conventional<br />
processes of drinking-water treatment (e.g., coagulation, flocculation, sedimentation).<br />
3. Evaluate changes in the character of NOM upon ground passage from the river to wells.<br />
4. Evaluate the merits of <strong>RBF</strong> <strong>for</strong> removing pathogenic microorganisms.<br />
5. Compare the transport of microbial pathogens with that of some potential surrogate or<br />
indicator parameters (e.g., particles, turbidity, coli<strong>for</strong>ms, aerobic and anaerobic spores,<br />
diatoms, bacteriophage).<br />
To address Objectives 1, 2, and 3, samples of river source waters and bank-filtered well waters from<br />
the three study sites were analyzed <strong>for</strong> a range of water-quality parameters, including:<br />
• TOC.<br />
• DOC.<br />
• UV-absorbance at 254-nm (UV254). • Biodegradable dissolved organic carbon.<br />
• Biological AOC.<br />
• Inorganic species.<br />
• DBP <strong>for</strong>mation potential.<br />
In the second year of the project, river waters were subjected to a bench-scale conventional<br />
treatment train consisting of coagulation, flocculation, sedimentation, glass-fiber filtration, and<br />
ozonation. The treated river waters were compared with the bank-filtered waters in terms of TOC,<br />
DOC, UV 254, and DBP <strong>for</strong>mation potential. In the third and fourth years of the project, NOM<br />
from the river and well waters was characterized using the XAD-8 resin adsorption fractionation<br />
method (Leenheer, 1981; Thurman and Malcolm, 1981). XAD-8 adsorbing (hydrophobic) and<br />
non-adsorbing (hydrophilic) fractions of the river and well waters were compared with respect to<br />
DOC, UV 254, and DBP <strong>for</strong>mation potential to determine whether <strong>RBF</strong> alters the character of the<br />
source-water NOM upon ground passage and, if so, which fractions are preferentially removed.<br />
The ongoing research to address Objectives 4 and 5 consists of:<br />
• Field studies at the three study sites to document actual changes in microorganism<br />
concentrations upon subsurface travel between the rivers and wells.<br />
• Parallel laboratory column studies with riverbank aquifer media to provide insights into<br />
process mechanisms so that reliable treatment credits <strong>for</strong> pathogen removal can be<br />
established and the suitability of using indicator parameters <strong>for</strong> pathogens that are<br />
difficult to measure in these systems can be determined.<br />
The results of the DBP-precursor study demonstrate the effectiveness of <strong>RBF</strong> at removing organic<br />
precursors to potentially carcinogenic DBPs. When compared to a bench-scale conventional
treatment train optimized <strong>for</strong> turbidity removal, <strong>RBF</strong> per<strong>for</strong>med as well as treatment at one of the<br />
sites and substantially better than treatment at the other two sites in terms of removing organic<br />
carbon and DBP-precursor material. Removals of TOC and DOC upon <strong>RBF</strong> at the three sites<br />
generally ranged from 30 to 70 percent, compared to 20- to 50-percent removals upon bench-scale<br />
treatment of river waters. Reductions in precursor material <strong>for</strong> a variety of DBPs (THMs, HAAs,<br />
haloacetonitriles, haloketones, chloral hydrate, and chloropicrin) upon <strong>RBF</strong> ranged from<br />
50 to 100 percent using both the <strong>for</strong>mation potential (FP) and uni<strong>for</strong>m <strong>for</strong>mation conditions<br />
(UFC) tests (Standard Methods, 1998; Summers et al., 1996), while reductions upon bench-scale<br />
treatment were generally in the range of 40 to 80 percent. The substantially higher reductions of<br />
the DBP precursors relative to those of TOC and DOC indicate a preferential reduction upon<br />
ground passage in the NOM that reacts with chlorine to <strong>for</strong>m DBPs.<br />
Upon both bench-scale conventional treatment and <strong>RBF</strong>, a shift was observed in DBP <strong>for</strong>mation<br />
from the chlorinated to the more brominated species due to the removal of DOC relative to<br />
bromide upon treatment or <strong>RBF</strong>. Brominated THMs have greater toxicity than chloro<strong>for</strong>m, so the<br />
shift from chlorinated to brominated DBP species means that the reduction in risk is not directly<br />
proportional to the reduction in DBP <strong>for</strong>mation. Risk calculations demonstrated the ability of<br />
<strong>RBF</strong> in all cases to reduce the theoretical excess cancer risk due to THMs <strong>for</strong>med upon<br />
chlorination, and with substantially better per<strong>for</strong>mance than the bench-scale treatment train.<br />
The results of the NOM characterization study indicate that <strong>RBF</strong> appears to be equally capable of<br />
removing material of different character. The different removal mechanisms in the subsurface<br />
(e.g., sorption, biodegradation, filtration) combine to provide similar removal of the operationally<br />
defined hydrophilic and hydrophobic fractions of organic material upon ground passage. Thus, the<br />
observed reductions in DBP <strong>for</strong>mation upon <strong>RBF</strong> were largely the result of a decrease in the NOM<br />
concentration rather than a consistent change in NOM character.<br />
Field monitoring of a number of microorganisms on a tri-monthly basis between 1999 and 2000<br />
(Table 1) indicated that <strong>RBF</strong> may also serve as a significant barrier <strong>for</strong> removing microbial<br />
contaminants, including human pathogens. The monitoring data demonstrated:<br />
• Greater than 3-log removal of Clostridium spores.<br />
• Greater than 2.5-log removal of E. coli C bacteriophage (somatic phage host).<br />
• Greater than1.9-log removal of E. coli Famp bacteriophage (male-specific host).<br />
Log removals were calculated using “average” concentrations determined <strong>for</strong> each organism as the<br />
sum of the counts over all sampling rounds divided by the sum of the volumes sampled over all<br />
sampling rounds (Parkhurst and Stern, 1998); non-detects were treated as being zero (in the event<br />
of no detects over all sampling rounds, the average concentration is reported as being less than<br />
one count divided by the sum of the volumes sampled over all sampling rounds). Assuming that<br />
these indicator organisms can be used as surrogates <strong>for</strong> Giardia cysts and human enteric viruses,<br />
<strong>RBF</strong> at the three study sites met or surpassed the per<strong>for</strong>mance requirements in the United States<br />
<strong>for</strong> conventional coagulation, sedimentation, and filtration (i.e., 2.5-log removal <strong>for</strong> Giardia cysts<br />
and 2.0-log removal of viruses). More recent monthly field-monitoring results (Table 2) indicate<br />
greater than 0.8-log reduction of Bacillus, greater than 3-log reduction of Clostridium, greater than<br />
1.8-log reduction of bacteriophage MS-2, and greater than 3.7-log reduction of bacteriophage<br />
φX174 concentrations in bank-filtered waters relative to river waters. Cryptosporidium and Giardia<br />
were occasionally detected in river waters and were below the detection limits during all monthly<br />
sampling events in well waters.<br />
125
126<br />
Table 1. Results from 1999 to 2000 Tri-Monthly Field Monitoring<br />
(Seven Total Sampling Rounds) a,b<br />
Clostridium E. coli C<br />
Bacteriophage<br />
c E. coli F-ampc Average counts/<br />
100 mL<br />
Indiana American <strong>Water</strong> at Jeffersonville, Indiana<br />
pfu /100 mL pfu /100 mL<br />
Ohio River 122 49 12<br />
Well 9 3.2] 2.8] 0.09 [2.1]<br />
Well 2 3.2] 2.8] 2.2]<br />
Indiana American <strong>Water</strong> at Terre Haute, Indiana<br />
Wabash River 183 147 13<br />
Collector Well 0.07 [3.4] 3.3] 2.3]<br />
Well 3 3.4] 3.3] 2.3]<br />
Missouri American <strong>Water</strong> at Parkville, Missouri<br />
Missouri River 143 31 6<br />
Well 4 3.3] 2.6] 1.9]<br />
Well 5 3.3] 2.6] 1.9]<br />
a<br />
Concentrations calculated as Σ (counts <strong>for</strong> all sampling rounds)/Σ (volume sampled <strong>for</strong> all sampling rounds); see text.<br />
b<br />
Log removals shown in brackets.<br />
c<br />
E. coli C and E. coli F-amp are the host organisms.<br />
pfu = Plaque-<strong>for</strong>ming unit.<br />
Because of low and variable concentrations of pathogens such as Cryptosporidium in river waters,<br />
it is difficult to evaluate log removals <strong>for</strong> such parameters. Laboratory-scale column studies with<br />
riverbank media are currently underway to compare the behavior of pathogens (viruses Poliovirus<br />
and Feline Calicivirus; bacteria E. coli; and protozoans Giardia and Cryptosporidium) with that of<br />
a number of potential surrogate parameters. The potential surrogate or indicator parameters being<br />
studied include:<br />
• Bacteriophage MS-2 and PRD-1 (<strong>for</strong> human viruses).<br />
• Particles (latex and native river particles).<br />
• Turbidity.<br />
• Aerobic and anaerobic spores.<br />
Column studies are being conducted under a variety of physical/chemical conditions, with pH,<br />
ionic strength, multi-valent cation concentration (Ca 2+ ), NOM concentration, and flow rate as the<br />
variables, with laboratory-prepared water and river water collected from full-scale <strong>RBF</strong> systems.<br />
Column experiments indicate that bacteriophage can travel through riverbank sediment columns<br />
to a greater extent than Poliovirus, suggesting that bacteriophage may be useful as conservative<br />
indicators of human viruses. Continuing studies will further explore this relationship under a<br />
variety of physical/chemical conditions. Similar studies will explore the relationship between the<br />
transport of particle counts, turbidity, and aerobic and anaerobic spores with that of E.coli,<br />
Cryptosporidium, and Giardia.
Acknowledgements<br />
We gratefully acknowledge the support of the USEPA Office of <strong>Research</strong> and Development<br />
(Project CR-826337; Thomas F. Speth, Project Officer), USEPA Science to Achieve Results<br />
(STAR) program (Grant #R82901101-0; Angela Page, Project Officer), and the utility<br />
subsidiaries of American <strong>Water</strong>.<br />
REFERENCES<br />
Table 2. Results from January to December 2002 Monthly Field Monitoring<br />
(12 Total Sampling Rounds) a,b<br />
Bacillus Clostridium Total E. coli Bacterio- Bacterio- Crypto- Giardia<br />
(cfu/L) (cfu/L) Coli<strong>for</strong>ms (MPN/L) phage phage sporidium (cysts/L)<br />
(MPN/L) MS-2 PhiX174 (oocysts/L)<br />
(pfu/L) (pfu/L)<br />
Indiana-American <strong>Water</strong> Company at Jeffersonville, Indiana<br />
Ohio River 8.9 × 10 4 3.8 × 10 2 9.2 × 10 4 2.6 × 10 3 6 × 10 1 1.7 × 10 3 2.5 × 10 –2 3 × 10 –2<br />
Well 9 2.5 × 103 0.5]<br />
Well 2 1.3 × 104
128<br />
Ray, C., G. Melin, and R.B. Linsky, editors (2002b). Riverbank Filtration: Improving Source-<strong>Water</strong> Quality,<br />
Kluwer Academic Publishers, Dordrecht.<br />
Standard Methods <strong>for</strong> the Examination of <strong>Water</strong> and Wastewater, Twentieth Edition (1998). APHA, AWWA,<br />
& WEF, Washington.<br />
Summers, R.S., S.M. Hooper, H.M. Shukairy, G. Solarik, and D.M. Owen (1996). “Assessing DBP Yield:<br />
Uni<strong>for</strong>m Formation Conditions.” Journal AWWA, 88(6): 80.<br />
Thurman, E.M., and R.L. Malcolm (1981). “Preparative Isolation of Aquatic Humic Substances.” Envir. Sci.<br />
& Technol., 15(4): 463.<br />
Tufenkji, N., J.N. Ryan, and M. Elimelech (2002). “The Promise of Bank Filtration.” Envir. Sci. & Technol.,<br />
36(21): 423A.<br />
Weiss, W.J., E.J. Bouwer, W.P. Ball, C.R. O’Melia, M.W. LeChevallier, H. Arora, and T.F. Speth (2003a).<br />
“Riverbank Filtration: Fate of Disinfection By-product Precursors and Selected Microorganisms.” Journal<br />
AWWA, in publication (expected October, 2003).<br />
Weiss, W.J., E.J. Bouwer, W.P. Ball, C.R. O’Melia, H. Arora, and T.F. Speth (2003b). “Riverbank Filtration:<br />
Comparison with Bench-Scale Conventional Treatment <strong>for</strong> NOM and DBP Precursor Reductions.” Journal<br />
AWWA, in publication (expected December, 2003).<br />
JOSH WEISS is a <strong>Research</strong> Assistant in the Department of Geography and Environmental<br />
Engineering at The Johns Hopkins University, where he is currently investigating waterquality<br />
improvements during riverbank filtration at three Midwestern drinking-water<br />
utilities. He expects to receive a Ph.D. in 2004 after the completion of his thesis,<br />
“Reduction in Disinfection Byproduct Precursors and Microorganisms During Riverbank<br />
Filtration.” Weiss is the co-author of seven publications — including a chapter in Riverbank<br />
Filtration: Improving Source-<strong>Water</strong> Quality (2002) — and several conference proceedings,<br />
and was the 2000 recipient of the Chesapeake Section American <strong>Water</strong> Works Association Student Paper<br />
Award. Prior to attending Johns Hopkins, he researched the geochemical impact of proposed developments<br />
on a Florida barrier island <strong>for</strong> the Georgia <strong>Institute</strong> of Technology and investigated the bioremediation of sites<br />
contaminated by nitroaromatic compounds <strong>for</strong> Argonne <strong>National</strong> Laboratory. Weiss received a B.S. in Civil<br />
Engineering from the Georgia <strong>Institute</strong> of Technology and a M.S. in Environmental Engineering from The<br />
Johns Hopkins University.
Session 6: Microorganisms<br />
Assessment of the Microbial Removal Capabilities<br />
of Riverbank Filtration<br />
Vasiliki Partinoudi<br />
New England <strong>Water</strong> Treatment Technology Assistance Center<br />
Department of Civil Engineering, University of New Hampshire<br />
Durham, New Hampshire<br />
M. Robin Collins, Ph.D., P.E.<br />
New England <strong>Water</strong> Treatment Technology Assistance Center<br />
Department of Civil Engineering, University of New Hampshire<br />
Durham, New Hampshire<br />
Aaron B. Margolin, Ph.D.<br />
New England <strong>Water</strong> Treatment Technology Assistance Center<br />
Department of Civil Engineering, University of New Hampshire<br />
Durham, New Hampshire<br />
Larry K. Brannaka, Ph.D., P.E.<br />
New England <strong>Water</strong> Treatment Technology Assistance Center<br />
Department of Civil Engineering, University of New Hampshire<br />
Durham, New Hampshire<br />
Introduction<br />
Riverbank filtrate includes both groundwater and river water that has percolated through the<br />
riverbank or riverbed to an extraction well. One of the primary objectives of this study was to<br />
assess the microbial removal capabilities of <strong>RBF</strong> independent of any groundwater dilution (i.e., a<br />
worse-case scenario). This study monitored total coli<strong>for</strong>ms, E. coli, and aerobic spore-<strong>for</strong>ming<br />
bacteria along with other water-quality parameters over a 12-month period in the following<br />
locations:<br />
• Pembroke, New Hampshire.<br />
• Mil<strong>for</strong>d, New Hampshire.<br />
• Jackson, New Hampshire.<br />
• Louisville, Kentucky.<br />
• Cedar Rapids, Iowa.<br />
This study also monitored the removal of male-specific coliphage achieved by <strong>RBF</strong> in Louisville,<br />
Kentucky, and Cedar Rapids, Iowa. Samples were collected at both of these sites <strong>for</strong> the detection<br />
of enteric viruses.<br />
Correspondence should be addressed to:<br />
Vasiliki Partinoudi<br />
Graduate Student<br />
Environmental Technology Building • Department of Civil Engineering • Environmental <strong>Research</strong> Group<br />
University of New Hampshire • 25 Colovos Road • Durham, New Hampshire 03824 USA<br />
Phone: (603) 862-1197 • Fax: (603) 862-3957 • Email: Vp4@cisunix.unh.edu<br />
129
130<br />
Objectives<br />
The overall study objectives were to:<br />
• Quantify the contributions of river water and groundwater to <strong>RBF</strong>-extraction water.<br />
• Assess <strong>RBF</strong> as a viable treatment and pretreatment option.<br />
• Compare <strong>RBF</strong> to slow sand filtration in regards to particulate, organic precursors, and<br />
microbiological removal capabilities expressed in terms of log-removal credits.<br />
The focus of this paper will be on microbial removals achieved by <strong>RBF</strong>.<br />
Characterization of Sampling Sites<br />
Five <strong>RBF</strong> sites (three in New Hampshire, one in Kentucky, and one in Iowa) were chosen that<br />
possessed different characteristics, in terms of the degree of hydraulic connection between the<br />
river and <strong>RBF</strong> extraction well, pumping rates, well construction criteria, geology, water source and<br />
quality, and distance of the <strong>RBF</strong> extraction well from the river. The sites were chosen to enable<br />
the researchers to assess the abilities of <strong>RBF</strong> technology in different settings. The study sites in<br />
Louisville, Kentucky, and Cedar Rapids, Iowa, were chosen because these particular sites have<br />
been characterized in detail in previous studies.<br />
Pembroke, New Hampshire: The well site lies within a stratified-drift aquifer located within the<br />
Soucook River Valley. The eastern aquifer boundary is delineated by contact with glacial till. The<br />
aquifer consists of layers of fine to course sand of varying thickness grading with fine gravel to<br />
boulders in some locations. In the area of the <strong>RBF</strong> extraction well, aquifer thickness ranges from<br />
9.7 to 18.9 m, with a saturated thickness of 7.6 to 17.4 m. The distance between the <strong>RBF</strong><br />
extraction well and river is approximately 55 m. The <strong>RBF</strong> extraction well has a diameter of<br />
30.5 centimeters and is 17-m deep. The subsurface material at that depth was dense silt and very<br />
fine sand.<br />
Mil<strong>for</strong>d, New Hampshire: The Mil<strong>for</strong>d State Fish Hatchery is located along the Souhegan River<br />
near its confluence with Purgatory Brook. The Souhegan River Valley contains rich deposits of<br />
glacial sand, gravel, and silt. The well is a 61-centimeter diameter gravel-pack well, constructed<br />
to a depth of 19.8 m. The screen is 50.8 centimeters in diameter and the length is 4.6 m. The<br />
aquifer at the well site consists mainly of sand and gravel. The distance between the river and the<br />
well is 23 m. The well is pumped continuously and delivers 1.6 MGD.<br />
Jackson, New Hampshire: The Jackson <strong>Water</strong> Precinct infiltration gallery was constructed in the<br />
early 1980s and is located on the banks of the Ellis River. Five infiltration galleries are connected<br />
to the river to provide adequate flow to the <strong>RBF</strong> extraction well. The five infiltration gallery<br />
intakes are each 6.1-m long, 1.2-m deep, and 1.2-m wide, and are located underneath the riverbed.<br />
They were placed 1.2-m apart, thus covering a total area of 10.8 m. A 12.2-m long polyvinyl<br />
chloride pipe connects each gallery to an intake pipe leading to the 20.3-centimeter diameter,<br />
7.3-m deep <strong>RBF</strong> extraction well equipped with a submersible pump. A new infiltration gallery is<br />
now under construction.<br />
Louisville, Kentucky: The construction of the collector well was completed in March 1999. The<br />
Louisville <strong>Water</strong> Company has extensively characterized this site. The <strong>RBF</strong> extraction well is on<br />
the bank of the Ohio River, about 30.5 m from the river. The depth of the well is 12.2-m below<br />
the river bottom and the pumping rate is 75.7-million liters per day. The well has seven horizontal<br />
laterals, which are each 73.1-m long and constructed of 30.5-centimeter stainless steel, with well
screens along the entire lateral length. Of the seven laterals, L4 is the lateral underneath the river<br />
separated by at least 40 ft of aquifer material; L1 is the furthest (perpendicular) to the river; and<br />
L2 is parallel to the river.<br />
Cedar Rapids, Iowa: The Cedar River is a meandering steam that has cut into the bedrock<br />
surface, exposing steep valley walls in the study area (Schulmeyer, 1995), where a total of<br />
53 vertical and horizontal wells have been installed. Seminole Valley Park Well Number 22 is<br />
located on the banks of the Cedar River, and is 19.5 m from the crest of the riverbank. The <strong>RBF</strong><br />
extraction well was drilled in 1991 to a total depth of 17.4-m below grade. The borehole diameter<br />
was 10.7 centimeters to depth, and a 7.6-centimeter screen and casing were installed.<br />
Microbial Analyses<br />
Total Coli<strong>for</strong>ms and E. coli: The IDEXX Quanti-Tray/2000 method was used <strong>for</strong> this analysis.<br />
Aerobic Spore-Forming Bacteria: Procedures were followed as outlined by Ballester and Margolin (2000).<br />
Virus Indicators: Male-specific and somatic coliphage viruses were monitored intensively <strong>for</strong><br />
2 weeks in December 2002 at Louisville, Kentucky, and Cedar Rapids, Iowa, using a single agar<br />
overlay (Method 1602, 1999) and a two-step enrichment method. Antibiotic-resistant strains<br />
E. coli F-amp (resistant to Streptomycin and Ampicillin) and E .coli CN-13 (resistant to Nalidixic<br />
Acid) were used as hosts <strong>for</strong> F-specific coliphage and somatic coliphage, respectively. Analyses<br />
followed the method as outlined by Ballester and Margolin (2000).<br />
Viruses: The intensive coliphage monitoring was followed by the collection of samples to detect<br />
human enteric viruses (Adenovirus type 40 and 41, Astrovirus, Poliovirus, Coxsackie virus,<br />
Rotavirus, and Echovirus) using positive microwound filters. The virus samples were analyzed<br />
using the Integrated Cell Culture-Reverse Transcription-Nested Polymerase Chain Reaction<br />
(ICC-RT-nPCR) method, due to its high specificity and sensitivity (Chapron et al., 2000).<br />
Problems with Assessing <strong>RBF</strong> Removal Capabilities<br />
It is essential to understand the extent and magnitude of river-aquifer interaction to address waterquality<br />
and supply issues, as well as to ensure the health of ecosystems (Winter et al., 1998). There<br />
are many questions to be answered as to how <strong>RBF</strong> works, how to establish travel time, and how to<br />
assess the mixing ratio of river water to groundwater in an <strong>RBF</strong> extraction well. These questions<br />
need to be answered to determine the removal efficiency of the <strong>RBF</strong> process independently of<br />
groundwater dilution.<br />
What Is the Exact Travel Time from the River to the Well?<br />
Table 1 provides a summary of travel time determinations <strong>for</strong> river water to reach the <strong>RBF</strong><br />
extraction well and the selected method used <strong>for</strong> each site in this study.<br />
How Much Removal Is Due to Filtration and How Much Is Due to Dilution with Groundwater?<br />
A variety of water-quality parameters (Table 2) were analyzed to assess the contribution of river<br />
and groundwater to <strong>RBF</strong>-extracted water. Knowledge of the mixing ratio would be necessary to<br />
determine removals achieved solely by subsurface filtration. The degree of mixing depends on<br />
factors such as hydraulic head gradient in the aquifer, aquifer properties, and hydrostatic pressure<br />
in the river, as well as the degree of connectivity between the river and parts of the aquifer.<br />
131
132<br />
The contribution of river water and groundwater to each of the <strong>RBF</strong> extraction wells was<br />
evaluated based on a different parameter at each site due to different types of in<strong>for</strong>mation available<br />
at each site. The average source contributions to the <strong>RBF</strong>-extracted water during this study period<br />
is summarized in Table 3.<br />
Results and Discussion<br />
Table 1. Travel Times from the River to the <strong>RBF</strong> Extraction Well<br />
Sampling Site Travel Time Evaluation of Travel Time<br />
(from river<br />
to <strong>RBF</strong> well)<br />
Pembroke, 5 days Darcy’s Law in terms of seepage velocity<br />
New Hampshire<br />
Mil<strong>for</strong>d, 1 day Darcy’s Law in terms of seepage velocity<br />
New Hampshire<br />
Jackson, 1 day Infiltration gallery. Estimated from in<strong>for</strong>mation based<br />
New Hampshire on the construction details of the gallery<br />
Louisville, 1 day Estimated from in<strong>for</strong>mation provided by the<br />
Kentucky Louisville <strong>Water</strong> Company (initial pumping test data)<br />
Cedar Rapids, 5 days Estimated from in<strong>for</strong>mation provided by the<br />
Iowa City of Cedar Rapids (Schulmeyer, 1995)<br />
Table 2. Sampling Parameters Used to Assess the Mixing Ratio in the <strong>RBF</strong> Extraction Well<br />
Sulfate Color Hardness<br />
(true)<br />
Chloride UV 254 absorbance Alkalinity<br />
pH Particle counts Redox potential<br />
Temperature Selected radioisotopes (222Ra) Conductivity<br />
<strong>RBF</strong> reduces DOC by as much as 82 percent and reduces temperature spikes by more than<br />
46 percent in the summer and 96 to 97 percent in the winter months. Turbidity was reduced to<br />
well below 1 ntu (80- to 86-percent removal). The reductions were computed from field values<br />
presented in Table 4.<br />
Typical river-water concentrations ranged between:<br />
• Below detection limit to 24,192 colony-<strong>for</strong>ming units (cfu)/100 milliliters <strong>for</strong> total coli<strong>for</strong>ms.<br />
• Below detection limit to 1,031 cfu/100 milliliters <strong>for</strong> E. coli.<br />
• Eighty-four to 145,000 cfu/100 milliliters <strong>for</strong> aerobic spore-<strong>for</strong>ming bacteria (see Table 4).<br />
All three of these microbial concentrations were below detection limit (less than 1 cfu/100 milliliters)<br />
in <strong>RBF</strong>-extraction well water, even after eliminating the “dilution” effects with groundwater.
Table 3. Average Distribution of River <strong>Water</strong> and Groundwater to <strong>RBF</strong>-Extracted <strong>Water</strong><br />
Percent of Percent of Parameter<br />
River <strong>Water</strong> Groundwater Upon Which<br />
in <strong>RBF</strong> Wells in <strong>RBF</strong> Wells the Ratio Is Based<br />
Pembroke, 40.7 ± 3.7 59.3 ± 3.7 Conductivity<br />
New Hampshire<br />
Mil<strong>for</strong>d, 40.8 ± 6.4 59.2 ± 6.4 Sulfate<br />
New Hampshire<br />
Jackson, 100 0 Infiltration gallery receiving only<br />
New Hampshire river water (due to an impermeable<br />
barrier that prevents groundwater<br />
from entering the well)<br />
Louisville, 78.1 ± 4.4 21.9 ± 4.4 Hardness<br />
Kentucky<br />
Cedar Rapids, 70 30 Based on groundwater flow modeling<br />
Iowa<br />
Parameter/<br />
Site<br />
Pembroke,<br />
New Hampshire<br />
Table 4. Typical Ranges of Selected Analytes of Interest<br />
Mil<strong>for</strong>d,<br />
New Hampshire<br />
Jackson,<br />
New Hampshire<br />
Louisville,<br />
Kentucky<br />
Cedar Rapids, Iowa<br />
Date 8/01 to 11/03 11/01 to 11/02 5/02 to 11/02 9/01 to 12/02 9/02 to 1/03<br />
River <strong>RBF</strong> GW River <strong>RBF</strong> GW River <strong>RBF</strong> River <strong>RBF</strong> GW River <strong>RBF</strong> GW<br />
pH 6.3–7.3 6.1–6.6 5.6–6.2 6.3–7.6 5.8–6.6 5.9–6.7 6.3–7.3 6.0–7.5 NA NA NA 8.2–8.5 7.6–7.7 7.3–7.4<br />
DOC (mg/L) 1.7–6.6 0.3–1.2 0.3–0.7 1.6–6.4 0.5–1.3 0.3-0.8 2.4–2.7 1.7–1.8 3.0–5.0 1.7–2.0 1.3–1.8 2.8–4.3 1.5–2.5 0.1–0.3<br />
Temperature (°C) 0.3–25.7 8.5–13.8 8.3–11.8 0.4–26.1 7.3–16.2 6.7-15.3 10–21.1 9.5–11 3.6–29.7 13.3–26.415.5–16.6 0.4–20.6 8–22.9 9.4–13.1<br />
Conductivity 42–150 193–269 294–395 119–214 107–152 113-194 23–83 52–69 NA NA NA 454–666 473–624 468–658<br />
(µS/cm)<br />
Total 141– BDL BDL 423– BDL BDL 52–1,356 16–160 10– BDL BDL 75– BDL BDL<br />
Coli<strong>for</strong>ms 1,399 2,431 24,192 2,572<br />
(cfu/100 mL)<br />
E. coli 4–108 BDL BDL 6–119 BDL BDL 2.0–20 2.0–10 2–1,031 BDL BDL 23–56 BDL BDL<br />
(cfu/100 mL)<br />
Aerobic 84– BDL BDL 124-1,093 BDL BDL BDL BDL 3,500– BDL BDL 39–217 BDL BDL<br />
Spore-Forming 1,997 145,000<br />
Bacteria<br />
(cfu/100 mL)<br />
NA = Not available. µS/cm = MicroSemens per centimeter. GW = Groundwater. BDL = Below detection limit.<br />
Typical concentrations of aerobic spore-<strong>for</strong>ming bacteria, total coli<strong>for</strong>ms, E. coli, and male-specific<br />
coliphage in river water and <strong>RBF</strong>-extracted water can be seen in Table 5. The total removal is<br />
calculated by subtracting the concentration of an analyte in the river from that in the <strong>RBF</strong><br />
extraction well. Total coli<strong>for</strong>ms and E. coli were reduced on average by at least 2.6- and 0.8-log removal<br />
credits, respectively. These values were considered artificially low due to low numbers observed in<br />
river water. <strong>RBF</strong> achieved a 4.4-log credit reduction of total coli<strong>for</strong>ms on May 20, 2002, in<br />
133
134<br />
Louisville, Kentucky. Aerobic spore-<strong>for</strong>ming bacteria were removed by at least 1.9-log credit in<br />
Pembroke, New Hampshire, and up to 5.2-log credits in Louisville, Kentucky. These values are in<br />
line with those reported <strong>for</strong> Louisville, Kentucky, by Wang et al. (2002).<br />
The male-specific coliphages ranged between 3,453 and 4,622 plaque-<strong>for</strong>ming units (pfu)/100 mL<br />
in river water. The concentration of the male-specific coliphages was reduced at least 80 percent<br />
by riverbank passage at all the study sites, as is indicated in the reduction summary presented in<br />
Table 5. Although there was a high concentration of male-specific coliphages in the river water<br />
and <strong>RBF</strong>-extracted water, the virus samples collected in Cedar Rapids in December 2003 and<br />
Louisville in March 2003 were negative (ICC-nPCR method) <strong>for</strong> the viruses of interest.<br />
Conclusions<br />
At each study site, the water quality of the source water, aquifer material, distance from the river<br />
to the <strong>RBF</strong> extraction well, travel time, and mixing ratio between river and groundwater were<br />
evaluated and found to differ from site to site. Travel times and mixing ratios were evaluated by<br />
different methods based on the in<strong>for</strong>mation available at each site. <strong>RBF</strong> was found to be a sitespecific<br />
process and should be treated as such. Based on differences between sites in source-water<br />
quality, aquifer material, and other parameters, it will be difficult to develop guidelines applicable<br />
to all <strong>RBF</strong> sites.<br />
In summary, the sites evaluated in this study indicated the conservative effectiveness of <strong>RBF</strong> in<br />
removing bacteria and virus indicators (any groundwater dilution with <strong>RBF</strong> extract should<br />
contribute to even lower microbial concentrations). Aerobic spore-<strong>for</strong>ming bacteria, total<br />
coli<strong>for</strong>ms, E. coli, and male-specific bacteriophage were removed by at least an average of 1.9, 1.0,<br />
0.3 and 0.2 logs, respectively. Based on this study, <strong>RBF</strong> shows potential to be a viable pretreatment<br />
and treatment process and warrants additional study.<br />
Acknowledgements<br />
• USEPA <strong>for</strong> funding this project through the New England <strong>Water</strong> Treatment Technology<br />
Assistance Center at the University of New Hampshire.<br />
• Jackson, New Hampshire <strong>Water</strong>works.<br />
• Nicola A. Ballester and Justin H. Fontaine of the University of New Hampshire.<br />
• Mil<strong>for</strong>d, New Hampshire Fish Hatchery personnel.<br />
• Cedar Rapids <strong>Water</strong> Department, Iowa.<br />
• Pembroke, New Hampshire <strong>Water</strong>works.<br />
• Louisville <strong>Water</strong> Company, Kentucky.<br />
• Melisa A. Smith of the University of New Hampshire.
Location Number<br />
of<br />
Sampling<br />
Events [1]<br />
Table 5. Average Reductions Achieved by <strong>RBF</strong><br />
River [2] <strong>RBF</strong> [2] Background<br />
Well [2]<br />
Total<br />
Log<br />
Removal<br />
Average<br />
Percent<br />
Removal<br />
Achieved by<br />
Subsurface<br />
Filtration<br />
Average<br />
Percent<br />
Removal<br />
Achieved by<br />
Groundwater<br />
Dilution<br />
Aerobic Spore Forming Bacteria (cfu/100 mL)<br />
Pembroke, 19 502 BDL BDL >1.9 100% 0%<br />
New Hampshire ± 11<br />
Mil<strong>for</strong>d, 13 673 BDL BDL >2.1 100% 0%<br />
New Hampshire ± 55<br />
Jackson,<br />
New Hampshire<br />
3 32 ± 1 BDL NA BDL 100% 0%<br />
Louisville, 11 33, 404 BDL BDL >3.5 100% 0%<br />
Kentucky ± 5,672<br />
Cedar Rapids,<br />
Iowa<br />
5 101 ± 8 BDL BDL >2.6 100% 0%<br />
Total Coli<strong>for</strong>ms (cfu /100 mL)<br />
Pembroke, 19 521 BDL BDL >2.1 100% 0%<br />
New Hampshire ± 63<br />
Mil<strong>for</strong>d, 13 1,091 BDL BDL >2.6 100% 0%<br />
New Hampshire ± 114<br />
Jackson, 3 504 85 NA >0.5 100% 0%<br />
New Hampshire ± 42 ± 12<br />
Louisville, 11 3,921 BDL BDL >1 100% 0%<br />
Kentucky ± 182<br />
Cedar Rapids, 5 1,391 BDL BDL >1.4 100% 0%<br />
Iowa<br />
E. coli (cfu /100 mL)<br />
± 25<br />
Pembroke,<br />
New Hampshire<br />
19 27 ± 4 BDL BDL >0.6 100% 0%<br />
Mil<strong>for</strong>d, New<br />
Hampshire<br />
13 55 ± 10 BDL BDL >0.8 100% 0%<br />
Jackson,<br />
New Hampshire<br />
3 30 ± 1 6.5 ± 1 NA >0.4 100% 0%<br />
Louisville,<br />
Kentucky<br />
11 5 ± 2 BDL BDL >0.3 100% 0%<br />
Cedar Rapids,<br />
Iowa<br />
5 16 ± 1 BDL BDL >0.7 100% 0%<br />
Virus Indicators (Male-Specific Coliphage) (pfu/100 mL)<br />
Louisville, 4 4,342 3,703 3,402 >0.2 80.2% 19.80%<br />
Kentucky ± 24 ± 22 ± 18<br />
Cedar Rapids, 5 3,438 753 BDL >0.7 100% 0%<br />
Iowa ± 21 ± 9<br />
Where: [1] = One sampling event includes the collection of a river water, groundwater, and <strong>RBF</strong>-extracted water sample.<br />
The <strong>RBF</strong>-extracted water sample was sampled with travel time taken into consideration.<br />
Where: [2] = This value shows the average concentration of the microorganism of interest throughout the study ± analytical error.<br />
BDL = Below detection limit. NA = Not available.<br />
135
136<br />
REFERENCES<br />
Ballester, N.A., and A.B. Margolin (2000). Detection of Bacillus Spores in Environmental Samples, University<br />
of New Hampshire <strong>Water</strong>borne Disease Laboratory SOP Manual.<br />
Chapron, C.D., N.A. Ballester, J.H. Fontaine, C.N. Frades, and A.B. Margolin (2000). “Detection of<br />
Astroviruses, Enteroviruses, and Adenovirus types 40 and 41 in Surface <strong>Water</strong>s Collected and Evaluated by<br />
the In<strong>for</strong>mation Collection Rule and an Integrated Cell Culture-Nested PCR Procedure.” Applied and<br />
Environmental Microbiology, 66(6): 2,520-2,525.<br />
Method 1602 (1999). Male-specific (F+) and Somatic Coliphages in <strong>Water</strong> by Single Agar Layer Procedure.<br />
United States Environmental Protection Agency-821-R-00-00X. Draft: July 1999.<br />
Schulmeyer P.M. (1995). Effect of the Cedar River on the quality of the groundwater supply <strong>for</strong> Cedar Rapids,<br />
Iowa, U.S. Geological Survey <strong>Water</strong> Resources Investigations Report 94-4211.<br />
Wang J.Z., S.A. Hubbs, and R. Song (2002). Evaluation of riverbank filtration as a drinking water treatment<br />
process, American <strong>Water</strong> Works Association <strong>Research</strong> Foundation and American <strong>Water</strong> Works Association.<br />
Winter T.C., J.W. Harvey, O.L. Franke, and W.M. Alley (1998). Groundwater and surface water – A single<br />
resource, U.S. Geological Survey Circ. 1139, 79 pp.<br />
VASO PARTINOUDI is a graduate student at the University of New Hampshire who is<br />
working on her Master’s thesis, “Riverbank Filtration as a Viable Treatment and<br />
Pretreatment Process.” She has worked as a <strong>Research</strong> Assistant at the New England <strong>Water</strong><br />
Treatment Technology Assistance Center (WTTAC) at the University of New Hampshire<br />
and has participated in WTTAC projects, per<strong>for</strong>ming various tasks such as installing and<br />
maintaining water monitoring equipment and pilot slow sand filters, and sampling and<br />
per<strong>for</strong>ming various water-quality tests in the laboratory and in the field. She has presented<br />
her thesis-related work at the New England <strong>Water</strong> Works Association conference in 2002 (paper titled,<br />
“Riverbank Filtration as a Viable Treatment and Pretreatment Method”) and in 2003 at the American<br />
Geophysical Union conference in Nice, France (poster presentation titled, “Assessment of the microbial<br />
removal capabilities of Riverbank Filtration”). Partinoudi has received a B.S. in Civil Engineering from the<br />
University of Brighton, United Kingdom, and an M.S. in European Construction Engineering from the<br />
University of Coventry, United Kingdom.
Session 7: Organics Removal<br />
Riverbank Filtration: A Very Efficient Treatment<br />
Process <strong>for</strong> the Removal of Organic Contaminants?<br />
Dr.-Ing. Heinz-Jürgen Brauch<br />
DVGW-Technologiezentrum Wasser<br />
Karlsruhe, Germany<br />
Dr. rer. nat. Frank Sacher<br />
DVGW-Technologiezentrum Wasser<br />
Karlsruhe, Germany<br />
Prof. Dr. Wolfgang Kühn<br />
DVGW-Technologiezentrum Wasser<br />
Karlsruhe, Germany<br />
Introduction<br />
In Germany, <strong>RBF</strong> has been used <strong>for</strong> more than 100 years as a natural treatment process <strong>for</strong> the<br />
production of drinking water (Sontheimer, 1980 and 1991; Kühn and Müller, 2000; Sacher and<br />
Brauch, 2002). Nowadays, the major raw-water resource <strong>for</strong> drinking-water supplies in Germany<br />
is groundwater (about 64 percent), whereas bank-filtrated (or infiltrated) water is about 16 percent<br />
(Sacher and Brauch, 2002; Brauch et al., 2001). Compared to this, the direct abstraction of river<br />
water is of minor importance (less than 1 percent). In many cases (and mostly along larger rivers),<br />
a clear distinction between bank-filtrated water and groundwater is difficult, and the raw water<br />
used <strong>for</strong> drinking-water production is bank-filtrated water blended with groundwater.<br />
Bank-filtrated water as a source of raw water provides high-quality river water, and its subsoil passage<br />
guarantees an efficient and lasting removal of suspended matter, microorganisms, and organic<br />
micropollutants. Hence, the occurrence and fate of organic contaminants in river water and bankfiltrated<br />
water are of great concern <strong>for</strong> water suppliers worldwide using surface water or artificial<br />
groundwater as a drinking-water resource. Due to the huge number of possible contaminants in river<br />
water, a necessary restriction has to be made on organic substances that may be relevant to drinkingwater<br />
production (Sacher and Brauch, 2002 and 1999; Sacher et al., 2001a). These target<br />
compounds are characterized by criteria like microbial biodegradability, adsorbability onto activated<br />
carbon and onto soil material, behavior versus oxidation agents, bioaccumulation, and groundwater<br />
mobility, as well as specific data about production and consumption quantities. Toxicological data, if<br />
available, are also important, but are not the main criterion.<br />
Methodology<br />
Within the last few years, the behavior of selected hydrophilic organic micropollutants during<br />
<strong>RBF</strong> was studied in waterworks on the lower Rhine River, as well as by laboratory-scale<br />
Correspondence should be addressed to:<br />
Dr.-Ing. Heinz-Jürgen Brauch<br />
Head of the Analytical Department<br />
DVGW-Technologiezentrum Wasser (TZW)<br />
Karlsruher Strasse 84 • D-76139 Karlsruhe, Germany<br />
Phone: +49(0)721/9678-150 • Fax: +49(0)721/9678-104 • Email: brauch@tzw.de<br />
137
138<br />
experiments. Most of the hydrophilic compounds under investigation are industrial chemicals<br />
with high production volumes and a broad range of applications; there<strong>for</strong>e, they are likely to enter<br />
river water if they are not totally removed in industrial or municipal wastewater treatment plants.<br />
In this paper, data on their fate during <strong>RBF</strong> will be presented, whereby the results of both lab-scale<br />
experiments and long-time measurements of river water and bank-filtrated water will be given.<br />
For the simulation of microbial degradation during <strong>RBF</strong>, a so-called “test filter” is used. This<br />
closed-loop apparatus was developed and used by Sontheimer and Völker (1987) <strong>for</strong> the<br />
characterization of fractions of surrogate parameters from wastewater effluents. In the last few<br />
years, the method was adjusted and optimized to study the biodegradation of single compounds at<br />
concentration levels relevant <strong>for</strong> the environment (Karrenbrock et al., 1999; Knepper at al.,<br />
1999). Details of the experimental set-up are given elsewhere (Karrenbrock et al., 1999).<br />
Several waterworks along the lower Rhine River were selected to measure the behavior of organic<br />
micropollutants during <strong>RBF</strong>. All use bank-filtrated water from the Rhine River as raw water <strong>for</strong><br />
drinking-water production (Schubert, 2000; Denecke, 1997; Brauch et al., 2000), whereby their<br />
wells are situated in a distance of 30 to 50 m from the Rhine River. In all cases, the raw water is a<br />
mixture between bank-filtrated water and groundwater, whereby the groundwater fraction ranges<br />
between 5 and 40 percent (i.e., the raw water is predominantly bank-filtrated water from the<br />
Rhine River). Regular measurements were per<strong>for</strong>med over a time period of several years to attain<br />
reliable and significant data on the removal of organic compounds during underground passage<br />
(Brauch et al., 2000).<br />
Results<br />
Complexing Agents<br />
Aminopolycarbonic acids like nitrilotriacetic acid (NTA), EDTA, or diethylenetrinitrilopentaacetic<br />
acid (DTPA) are used as chelating agents in detergents and industrial cleaners, as well as in the<br />
photo, textile, and pulp- and paper-making industries. Due to their widespread use, NTA and<br />
EDTA are permanently found in the Rhine River in concentration levels of more than 1 µg/L<br />
(ARW and AWBR annual reports; Sacher et al., 1998). Besides these compounds, other<br />
complexing agents like ß-alaninediacetic acid (ADA) or 1.3-propylenedinitrilotetraacetic acid<br />
(PDTA) are used <strong>for</strong> special applications or as substitutes <strong>for</strong> EDTA. To test the biodegradation of<br />
these synthetic compounds, test-filter experiments were per<strong>for</strong>med in which water from the Rhine<br />
River at Karlsruhe was spiked with NTA, EDTA, DTPA, PDTA, and ADA at a concentration<br />
level of 10-µg/L each. These experiments showed that the concentration of NTA decreased quite<br />
rapidly, indicating a fast microbial degradation of this complexing agent. The concentration of<br />
ADA decreased much slower and, after 35 days, about 30 percent of the initial concentration was<br />
still present. The concentrations of EDTA, PDTA, and DTPA seemed to be more or less constant,<br />
indicating that these complexing agents are persistent under the conditions of the test-filter<br />
experiment.<br />
Looking at the concentrations of NTA, EDTA, and DTPA in the Rhine River and in the raw<br />
water of the waterworks, which consists of at least 90-percent bank-filtrated water from the Rhine<br />
River, it is clear that in correspondence to the results of the test-filter experiments, NTA is nearly<br />
totally removed during <strong>RBF</strong> and is only sporadically found in raw water. On the other hand,<br />
EDTA proved to be recalcitrant and was present in the raw water under investigation. DTPA was<br />
found only once in raw water, but concentrations in the Rhine River are quite near to the limit<br />
of determination (which was 2 µg/L until 1996 and is currently 1 µg/L), and mixing with some
uncontaminated groundwater could lead to an apparent removal of DTPA; there<strong>for</strong>e, based on<br />
environmental data, no definite statement can be given on the behavior of DTPA during <strong>RBF</strong>.<br />
ADA is only sporadically found in the Rhine River, and PDTA could not be detected in the Rhine<br />
River (but is often found in the Neckar River, <strong>for</strong> instance).<br />
Aromatic Sulfonates<br />
Aromatic sulfonates are the corresponding bases to sulfonic acids and, due to their permanent<br />
negative charge, are highly soluble in water. Benzenesulfonates are mainly used as intermediates<br />
during the manufacture of azo dyestuffs, optical brighteners, ion-exchange resins, plasticizers, and<br />
pharmaceuticals. Amino- and hydroxynaphthalene-sulfonates and anthraquinonesulfonates are<br />
important building blocks <strong>for</strong> azo dyestuffs. A major source of sulfonated stilbenes is the production<br />
of fluorescent whitening agents <strong>for</strong> laundry products and paper. Naphthalenesulfonates and their<br />
condensates with <strong>for</strong>maldehyde are large-scale products that have widespread applications,<br />
including paper chemicals, superplasticisers <strong>for</strong> concrete, textile auxiliaries, and synthetic leather<br />
tanning agents (Redin et al., 1999).<br />
Test-filter experiments with 2-naphthalene-sulfonate and 1,5-naphthalenedisulfonate showed<br />
that the behavior of the 2-naphthalene-sulfonates is different. Whereas 2-naphthalene-sulfonate<br />
degraded very fast and, after 2 days, could not be detected in water, 1,5-naphthalenedisulfonate<br />
proved to be persistent and no change in concentration could be observed even after 30 days. The<br />
behavior of 1,5-naphthalenedisulfonate is characteristic <strong>for</strong> many two- or threefold sulfonated<br />
naphthalene compounds. Naphthalene-sulfonates with two sulfo groups in alpha position seem to<br />
be especially persistent (Lange et al., 1995).<br />
As could be expected from the results of the test-filter experiments, 1,5-naphthalenedisulfonate,<br />
which is almost always present in the Rhine River, is not removed during <strong>RBF</strong> and was always<br />
found in the raw water of the waterworks under investigation. The same behavior was found <strong>for</strong><br />
2-amino-1,5- and 2-amino-4,8-naphthalene-disulfonate, <strong>for</strong> 1,3,5- and 1,3,6-naphthalenetrisulfonate,<br />
<strong>for</strong> cis-4,4’-dinitro-2,2’-stilbene-disulfonate, and <strong>for</strong> 8,8’-methylenebis-2-naphthalenesulfonate<br />
(Lange et al., 1995).<br />
Pharmaceutical Compounds<br />
Drugs were produced, prescribed, and used in quantities up to some hundred tons per year (in<br />
Germany). Due to an incomplete elimination in wastewater treatment plants, residues of<br />
pharmaceutical products have recently been found in surface waters and groundwaters (Ternes,<br />
1998; Heberer, 2002; Sacher et al., 2001b). In the Rhine River, compounds like diclofenac (an<br />
antirheumatic and analgesic), carbamazepine (an antiepileptic that is also used as antidepressant),<br />
and clofibric acid and bezafibrate (two lipid-regulating agents) are most often found with<br />
concentrations in the 10- to 100-nanograms per liter range. To study their behavior during <strong>RBF</strong>,<br />
test-filter experiments were per<strong>for</strong>med, yielding that only bezafibrate is biodegradable under the<br />
conditions of the test-filter experiment. The concentration of carbamazepine decreases as a<br />
function of time, but even after 30 days, it is present in the test-filter system. The concentrations<br />
of diclofenac and clofibric acid are more or less constant, indicating that the respective<br />
compounds are not easily biodegradable. Monitoring data on the behavior of diclofenac and<br />
carbamazepine during <strong>RBF</strong> under environmental conditions show that that diclofenac was never<br />
detected in the raw water under investigation, although it was nearly always found in the Rhine<br />
River. Qualitatively, the same results were found <strong>for</strong> bezafibrate and clofibric acid (even if clofibric<br />
acid was only found a few times in the Rhine River). For bezafibrate, this finding is in accordance<br />
with the results of the test-filter experiments. For diclofenac and clofibric acid, there is a<br />
139
140<br />
contradiction to the results of the test-filter experiment, indicating that these two compounds are<br />
not biodegradable during underground passage. A satisfying explanation <strong>for</strong> the more effective<br />
elimination of diclofenac and clofibric acid under environmental conditions cannot be given.<br />
Furthermore, it can be seen from the monitoring data that carbamazepine is always present in the<br />
raw waters of all waterworks taking bank-filtrated water from the Rhine River, confirming the<br />
poor biodegradability of this compound found in the test-filter experiment.<br />
Conclusions<br />
Many organic micropollutants present in the Rhine River (and other river waters) are eliminated<br />
during <strong>RBF</strong>. Test-filter experiments prove that this elimination is mainly due to a microbial<br />
degradation of the compounds. Until now, less is known about metabolites and their behavior in<br />
the environment and during treatment processes in waterworks. Only few compounds are not<br />
totally removed during <strong>RBF</strong> and enter the raw waters used <strong>for</strong> drinking-water production. For the<br />
waterworks under investigation at the Rhine River, the compounds found in raw water could be<br />
totally removed by subsequent treatment st<strong>eps</strong> like ozonation or GAC filtration, in most cases.<br />
Hence, the use of bank-filtrated water instead of river water is advisable if industrial or municipal<br />
wastewaters may affect river-water quality. The use of bank-filtrated water, however, cannot<br />
replace further treatment st<strong>eps</strong>.<br />
REFERENCES<br />
Annual reports of Arbeitsgemeinschaft Rhein-Wasserwerke e.V. (ARW) and Arbeitsgemeinschaft Wasserwerke<br />
Bodensee-Rhein (AWBR).<br />
Brauch, H.-J., F. Sacher, E. Denecke, and T. Tacke (2000). “Wirksamkeit der Uferfiltration für die Entfernung<br />
von polaren organischen Spurenstoffen.” gwf-Wasser/Abwasser, 141: 226-234.<br />
Brauch, H.-J., U. Müller, and W. Kühn (2001). “Experiences with riverbank filtration in Germany.”<br />
Proceedings, International Riverbank Filtration Conference, Rheinthemen 4, 33-39.<br />
Denecke, E. (1997). “Auswertung langzeitlicher Messreihen zur aeroben Abbauleistung der Uferpassage einer<br />
Wassergewinnungsanlage am Niederrhein. Z.” Wasser-Abwasser-Forschung, 25: 311-318.<br />
Heberer, T. (2002). “Occurrence, fate, and removal of pharmaceutical residues in the aquatic environment:<br />
A review of recent research data.” Toxicology Letters, 131: 5-17.<br />
Karrenbrock, F., T.P. Knepper, F. Sacher, and K. Lindner (1999). “Entwicklung eines standardisierten<br />
Testfilters zur Bestimmung der mikrobiellen Abbaubarkeit von Einzelsubstanzen.” Vom Wasser, 92: 361-371.<br />
Knepper, T.P., F. Sacher, F.T. Lange, H.-J. Brauch, F. Karrenbrock, O. Rörden, and K. Lindner (1999).<br />
“Detection of polar organic substances relevant <strong>for</strong> drinking water.” Waste Management, 19: 77-99.<br />
Kühn, W. and U. Müller (2000). “Riverbank filtration — An overview.” Journal AWWA, 92: 60-69.<br />
Lange, F.T., M. Wenz, and H.-J. Brauch (1995). “The behaviour of aromatic sulfonates in drinking water<br />
production from River Rhine water and bank filtrate.” Analytical Methods and Instrumentation, 2: 277-284.<br />
Redin, C., F.T. Lange, H.-J. Brauch, and S.H. Eberle (1999). “Synthesis of sulfonated naphthalene<strong>for</strong>maldehyde<br />
condensates and their trace-analytical determination in wastewater and river water.” Acta<br />
Hydrochim. Hydrobiol., 27: 136-142.<br />
Sacher, F., E. Lochow, and H.-J. Brauch (1998). “Synthetische organische Komplexbildner - Analytik und<br />
Vorkommen in Oberflächenwässern.” Vom Wasser, 90: 31-41.<br />
Sacher, F., and H.-J. Brauch (1999). “Bewertung organischer Einzelstoffe im Hinblick auf ihr Verhalten bei<br />
der Wasseraufbereitung.” Veröffentlichungen aus dem Technologiezentrum Wasser, 7: 111-127.
Sacher, F., H.-J. Brauch, and W. Kühn (2001a). “Fate studies of organic micropollutants in riverbank<br />
filtration.” Proceedings, International Riverbank Filtration Conference, Rheinthemen, 4: 139-148<br />
Sacher, F., F.T. Lange, H.-J. Brauch, and I. Blankenhorn (2001b). “Pharmaceuticals in groundwaters —<br />
Analytical methods and results of a monitoring program in Baden-Württemberg, Germany.” J. Chromatogr.,<br />
A 938: 199-210.<br />
Sacher, F., and H.-J. Brauch (2002). “Experiences on the fate of organic micropollutants during riverbank<br />
filtration.” Understanding contaminant biogeochemistry and pathogen removal, C. Ray (ed.), Kluwer Academic<br />
Publishers, The Netherlands, p. 135-151.<br />
Schubert, J. (2000). “Entfernung von Schwebstoffen und Mikroorganismen sowie Verminderung der<br />
Mutagenität bei der Uferfiltration.” gwf-Wasser/Abwasser, 141: 218-225.<br />
Sontheimer, H. (1980). “Experiences with riverbank filtration along the Rhine River.” Journal AWWA, 72: 3,386-3,392.<br />
Sontheimer, H., and E. Völker (1987). Charakterisierung von Abwassereinleitungen aus der Sicht der Trinkwasserversorgung.<br />
Veröffentlichungen des Bereichs und Lehrstuhls für Wasserchemie am Engler-Bunte-Institut der<br />
Universität Karlsruhe 31.<br />
Sontheimer, H. (1991). Trinkwasser aus dem Rhein? Bericht über ein Verbund<strong>for</strong>schungsvorhaben zur Sicherheit<br />
der Trinkwassergewinnung aus Rheinuferfiltrat, Academia Verlag, Sankt Augustin.<br />
Ternes, T.A. (1998). “Occurrence of drugs in German sewage treatment plants and rivers.” Wat. Res.,<br />
32: 3,245-3,260.<br />
HEINZ-JÜRGEN BRAUCH has over 20 years of experience in resolving water-quality<br />
problems, with special emphasis in drinking-water production. He has conducted many<br />
research projects on the development and optimization of analytical determination<br />
methods <strong>for</strong> organic micropollutants, including emerging contaminants and fate studies on<br />
ground surface and drinking water, as well as the design and implementation of monitoring<br />
strategies <strong>for</strong> controlling hazardous substances. In addition, he has completed several<br />
projects in cooperation with water utilities to improve and optimize the treatment<br />
techniques <strong>for</strong> removing organic substances. Brauch has published numerous articles and papers on modern<br />
analytical techniques <strong>for</strong> detecting organic micropollutants, fate and behavior of organics in drinking-water<br />
treatment, occurrence and distribution of organic substances between water, and sediment and soil, as well<br />
as on water-quality problems in surface water and groundwater. Brauch has been the Head of the Analytical<br />
Department of DVGW-Technologiezentrum Wasser (Technology Center <strong>for</strong> <strong>Water</strong>) in Karlsruhe, Germany<br />
since 1990, and is the Chairman or Member of many national and international working groups in this field.<br />
He received a Ph. D. in Chemical Engineering from the University of Karlsruhe.<br />
141
142
Session 7: Organics Removal<br />
Organics Removal by Riverbank Filtration<br />
at the Greater Cincinnati <strong>Water</strong> Works Site<br />
Jeffrey Vogt<br />
Greater Cincinnati <strong>Water</strong> Works<br />
Cincinnati, Ohio<br />
William Fromme<br />
Greater Cincinnati <strong>Water</strong> Works<br />
Cincinnati, Ohio<br />
Bruce Whitteberry, P.G.<br />
Greater Cincinnati <strong>Water</strong> Works<br />
Cincinnati, Ohio<br />
William D. Gollnitz<br />
Greater Cincinnati <strong>Water</strong> Works<br />
Cincinnati, Ohio<br />
Objective<br />
With the upcoming promulgation of the Stage 2 Disinfectants and Disinfection By-Product Rule,<br />
water utilities are required to reduce DBPs in their finished water (USEPA, 2003). The<br />
concentration of NOM in source waters is directly related to the concentration of DBPs <strong>for</strong>med<br />
in finished waters (Owen et al., 1993). The challenge <strong>for</strong> water utilities is to reduce precursors and<br />
NOM prior to disinfection. The natural process of <strong>RBF</strong> may be an effective method to remove<br />
NOM. As part of a research study, TOC, UV 254, and THM <strong>for</strong>mation potential were used to<br />
evaluate the organic removal capabilities of <strong>RBF</strong>. This project was a cooperative study between the<br />
USGS, Miami University in Ox<strong>for</strong>d, Ohio, and the Greater Cincinnati <strong>Water</strong> Works. This was a very<br />
large project with many objectives. The objective of this paper is to evaluate the removal of NOM<br />
by <strong>RBF</strong>.<br />
Background<br />
The Greater Cincinnati <strong>Water</strong> Works owns and operates a 40-MGD drinking-water treatment<br />
plant located in southwestern Ohio. The source water <strong>for</strong> the plant comes from 10 production<br />
wells located in an alluvial aquifer system adjacent to the Great Miami River. The aquifer is a<br />
mixture of sand and gravel and has a hydraulic conductivity of 200 to 300 ft (62 to 91 m) per day.<br />
Five monitoring wells were drilled adjacent to the river and close to a production well (CMB1).<br />
Each monitoring well was drilled at varying depths in relation to the production well. Well 1A is<br />
the shallowest vertical well and is the nearest to the river. Well 1D is the deepest well and closest<br />
to CMB1. Wells 1B and 1C were placed at intermediate depths between Wells 1A and 1D.<br />
Correspondence should be addressed to:<br />
Jeffrey Vogt<br />
Chemist<br />
Greater Cincinnati <strong>Water</strong> Works<br />
5651 Kellogg Ave • Cincinnati, Ohio 45228 USA<br />
Phone: (513) 624-5624 • Fax: (513) 624-5670 • Email: Jeff.Vogt@gcww.cincinnati-oh.gov<br />
143
144<br />
Well 1C is located at the top of the production well screen, and Well 1D is located at the bottom<br />
of the production well screen. To identify discrete sections of the aquifer, the monitoring wells<br />
were developed with 2-ft (0.6 m) screened interval. An inclined well (Well 1I) was drilled from<br />
the top of the riverbank at a 20- to 30-degree angle from the horizontal plane. This well allowed<br />
the study team to collect samples approximately 5- to 10-ft (1.5- to 3.1-m) below the streambed.<br />
Samples were collected and analyzed on a weekly basis and, later, at monthly intervals from<br />
September 1999 until May 2001.<br />
Materials and Methods<br />
TOC and UV 254 were used to evaluate NOM removal. TOC analyses were analyzed with a<br />
Tekmar/Dohrman Phoenix 8000 using the UV-Persulfate Method. UV absorbance at 254-mn<br />
wavelength was per<strong>for</strong>med with a Hach DR4000. UV absorbance is primarily related to the humic<br />
fraction of NOM. UV at a wavelength of 254 nm is absorbed by double bonds and/or aromatic<br />
structures mostly produced from the breakdown of plant and animal matter (Owen et al., 1993).<br />
THM <strong>for</strong>mation potential were used to evaluate the removal of DBP precursors. The concept<br />
behind the analysis is to maximize the THM <strong>for</strong>mation reaction. Conditions (chlorine dose, pH,<br />
hold time, and high temperature) are set so that the reaction is “pushed” to <strong>for</strong>m THMs. A dose<br />
of 15-mg/L free chlorine was applied to the samples using a sodium hypochlorite solution. The<br />
sample pH was adjusted to 9.5 with a borate buffer and then incubated at 35-degrees Celsius <strong>for</strong><br />
7 days. After incubation, residual chlorine was measured. Samples were poured off and analyzed<br />
<strong>for</strong> THMs with a Varian 3400 Gas Chromatograph using USEPA Method 502.2.<br />
Results and Discussion<br />
All three of the parameters demonstrated the ability of <strong>RBF</strong> to remove organic material. Table 1<br />
represents water-quality data obtained from the river, inclined well, monitoring wells, and production<br />
well. The removal values represent removals from the Great Miami River through the wells. The<br />
average river results show relatively high concentrations of NOM as compared to drinking-water<br />
standards. Organic material was removed through the streambed (Well 1I), and is continually<br />
reduced as it migrates downward into the aquifer. Peaks in organic concentrations in the Great<br />
Miami River correlated with peak concentrations in Well 1I, yet at a reduced concentration (Figure 1).<br />
TOC is reduced 37 percent from the Great Miami River to Well 1I. An additional 20 percent of<br />
TOC was removed to Well 1C. This produces a total reduction in TOC of 59 percent. UV and<br />
THM <strong>for</strong>mation potential demonstrated even better reductions of, respectively, 63 and 72 percent.<br />
Table 1. Average Values and Percent Removals from the Great Miami River<br />
Over the Entire Study Period<br />
Great Well 1I Well 1A Well 1B Well 1C Production Well 1D<br />
Miami Well<br />
River CMB1<br />
DEPTH (ft) 5 to 10 31 46 60 60 88<br />
Avg. Avg. % Avg. % Avg. % Avg. % Avg. % Avg. %<br />
TOC 5.48 3.36 39 2.50 54 2.32 58 2.27 59 1.17 79 0.63 89<br />
UV 254 0.147 0.092 37 0.063 57 0.057 61 0.055 63 0.023 84 0.007 95<br />
THMFP 924 404 56 277 70 267 71 260 72 131 86 55.3 94<br />
THMFP = THM <strong>for</strong>mation potential.
TOC (mg/L)<br />
9<br />
8<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
1<br />
Site 1 Total Organic Carbon<br />
GMR FP1i FP1A FP1B FP1C CMB1 FP1D<br />
0<br />
9/1/99 12/2/99 3/3/00 6/3/00 9/3/00<br />
Date<br />
12/4/00 3/6/01 6/6/01 9/6/01<br />
Figure 1. TOC <strong>for</strong> the Great Miami River and well samples <strong>for</strong> the entire study period.<br />
Figure 2 presents UV 254 data <strong>for</strong> all study locations. The bulk of UV reduction is observed between<br />
the Great Miami River and Well 1I (37 percent). An additional 20-percent reduction is seen<br />
between Wells 1I and 1A. The reduction between Wells 1A and 1B is about 4 percent. Between<br />
Wells 1B and 1C, there is about 2-percent reduction. The downward trend of UV reduction<br />
continues through all of the locations, but the zone surrounding Wells 1A, 1B, and 1C<br />
demonstrates minor change. The similarities within that zone can be observed in the maximum,<br />
average, and minimum values <strong>for</strong> those wells. Additional reductions are seen at Well 1D. Results<br />
<strong>for</strong> Well 1D showed very low organic concentrations in all measured parameters when compared<br />
to others wells. It is likely that this water is a combination of regional groundwater and water with<br />
a much longer flow path to the production well. This downward trend or pattern <strong>for</strong> reduction can<br />
be observed in data plots of all three parameters.<br />
CMB1 has a 30-ft (9-m) intake screen. It draws in water from a much larger capture zone than the<br />
monitoring wells due to a much longer screen and significantly higher pumping rates. The water<br />
entering the production well is a mixture of water passing through Wells 1D and 1C, as well as<br />
other parts of the aquifer. The results <strong>for</strong> CMB1 fall between the results <strong>for</strong> both Wells 1C and 1D<br />
in three parameters.<br />
CMB1 showed the least amount of variation in the data, as compared to Wells 1A, 1B, and 1C in<br />
Figure 2. UV results of the samples collected at CMB1 showed a maximum value of 0.029 cm –1 ,<br />
an average value of 0.023 cm –1 , and a minimum value of 0.016 cm –1 . Figure 2 shows the UV results<br />
<strong>for</strong> CMB1 plotting very close together. The calculated standard deviation <strong>for</strong> CMB1 UV data is<br />
0.003 cm –1 . Comparing that value to the standard deviation of Well 1C (0.009 cm –1 ), it is<br />
observed that CMB1 shows little variation in its results. This trend of consistent results between<br />
the data points can be seen in the THM <strong>for</strong>mation potential and TOC results <strong>for</strong> CMB1. This is<br />
likely due to the large capture zone normalizing the concentration.<br />
145
146<br />
Absorbance (cm –1 )<br />
0.30<br />
0.25<br />
0.20<br />
0.15<br />
0.10<br />
0.05<br />
0.00<br />
UV 254 Absorbance<br />
Maximum Average Minimum<br />
GMR FP1i FP1A FP1B FP1C CMB1 FP1D<br />
Location<br />
Figure 2. Maximum, average, and minimum UV values at sample locations.<br />
Looking again at Figure 1, all of the wells except <strong>for</strong> the inclined well demonstrated very<br />
consistent results. The average results <strong>for</strong> Wells 1A, 1B, and 1C are similar <strong>for</strong> all of the parameters<br />
(see Figure 2). These data demonstrate that <strong>RBF</strong> is very effective in reducing any organic shock<br />
load from the river. High concentration peaks in the river and inclined well are unnoticed in the<br />
remaining wells. Others (Kuehn et al., 2000) have also demonstrated this process when using <strong>RBF</strong>.<br />
THM <strong>for</strong>mation potential can be used as a surrogate <strong>for</strong> THM concentrations in finished water<br />
(Owen et al., 1993). The ultimate goal of the Stage 2 Disinfectants and Disinfection By-Product<br />
Rule is to reduce THMs in finished waters produced by water treatment utilities across the<br />
country. UV and TOC measure bulk concentrations of organic matter, but THM <strong>for</strong>mation<br />
potential is a measurement of actual THM concentrations. The organic matter, which directly<br />
reacts with the chlorine to <strong>for</strong>m THMs, was observed to be the most reduced through the aquifer.<br />
Fifty-six percent of that matter was reduced at Well 1I, with an additional reduction of 21 percent<br />
through the 1A, 1B, and 1C zone. At CMB1, the matter was reduced to a total reduction of<br />
86 percent. These results suggest that <strong>RBF</strong> does an excellent job in reducing organic matter that<br />
<strong>for</strong>ms THMs.<br />
The data from this project also supports the concept that the natural process of <strong>RBF</strong> produces<br />
consistent water quality through time. Through this study period, there was no apparent breakthrough<br />
of organic material. This would indicate that the processes, which remove NOM in the<br />
aquifer, do not reach a saturation point, as would be expected in an engineered system, such as<br />
GAC.
Conclusions<br />
<strong>RBF</strong> is very effective in reducing organic matter in source waters. The reduction of NOM was<br />
demonstrated using TOC and UV 254 absorbance. THM precursor removal was demonstrated<br />
using THM <strong>for</strong>mation potential.<br />
• TOC reductions ranged from 39 to 79 percent between the Great Miami River and<br />
CMB1.<br />
• UV254 absorbance reductions ranged from 37 to 84 percent between the Great Miami<br />
River and CMB1.<br />
• THM <strong>for</strong>mation potential demonstrated the greatest range of reductions between the<br />
Great Miami River and CMB1, with values of 56 to 86 percent.<br />
• <strong>RBF</strong> is very effective in reducing organic shock loading. The high concentrations seen in<br />
the river were not seen in the 1A, 1B, or 1C monitoring wells. During the study period,<br />
there was never a breakthrough of organic matter. The results <strong>for</strong> the three parameters<br />
remained consistent when compared to varying concentrations in the Great Miami River.<br />
• Production well (CMB1) data remained very consistent in all of the measured parameters.<br />
REFERENCES<br />
Owen, D.M., G.L. Amy, and Z.K. Chowdhury (1993). Characterization of Natural Organic Matter and Its<br />
Relationship to Treatability, American <strong>Water</strong> Works <strong>Research</strong> Foundation, Denver, Colorado.<br />
USEPA (2003). Stage 2 Disinfectants/Disinfection By Product Rule, United States Environmental Protection<br />
Agency, Washington, D.C.<br />
Kuehn, W., and U. Mueller (2000). “Riverbank filtration: An Overview.” Journal AWWA, 92(12): 60-69.<br />
JEFFREY VOGT is a Chemist with the Greater Cincinnati <strong>Water</strong> Works, where he has<br />
worked since 1992. For the last 5 years, he has been involved in riverbank-filtration<br />
issues/research and was largely involved in the 3-year Flowpath Study. His responsibilities<br />
during the study involved planning, sampling, analyses, data evaluation, and interpretation<br />
of the results. In 2002, he presented and published at the American <strong>Water</strong><br />
Resources Association’s Groundwater/Surface <strong>Water</strong> Interactions Conference in<br />
Keystone, Colorado. Some of his current responsibilities include managing continuing<br />
riverbank-filtration issues, parasite-monitoring program <strong>for</strong> the Long Term 2 Enhanced Surface <strong>Water</strong><br />
Treatment Rule at the Greater Cincinnati <strong>Water</strong> Works’ Richard Miller Treatment Plant, and treatment<br />
studies <strong>for</strong> the Greater Cincinnati <strong>Water</strong> Works’ three treatment plants. Vogt holds a Class 3 <strong>Water</strong><br />
Treatment license in the State of Ohio and received a Science degree from the University of Cincinnati.<br />
147
148
Lunch Presentation<br />
Potential Uses of Riverbank Filtration<br />
<strong>for</strong> Regulatory Compliance<br />
Stig Regli<br />
United States Environmental Protection Agency<br />
Washington, D.C.<br />
<strong>RBF</strong> may have its most general application <strong>for</strong> systems seeking compliance with the LT2ESWTR.<br />
On August 11, 2003, the USEPA proposed the LT2ESWTR and included provisions by which<br />
<strong>RBF</strong> could be used as one of the compliance options <strong>for</strong> providing Cryptosporidium removal credits<br />
(USEPA, 2003). While the USEPA has previously recognized (through guidance implementation<br />
decisions) that <strong>RBF</strong> is a technology that can achieve pathogen removal, the LT2ESWTR is the<br />
first United States drinking-water regulation that specifically recognizes <strong>RBF</strong> as a compliance<br />
technology option.<br />
Under the proposed LT2ESWTR, filtered systems must monitor source water <strong>for</strong> Cryptosporidium<br />
to determine what source-water bin concentration category it belongs in and whether additional<br />
treatment is required. As part of this determination, the USEPA provides a “toolbox” of technologies<br />
by which systems can assess their total removal/inactivation credits <strong>for</strong> Cryptosporidium.<br />
The proposed LT2ESWTR recognizes <strong>RBF</strong> as a “toolbox” pretreatment technique that can<br />
provide a system 0.5- or 1.0-log additional pretreatment credit, if it meets specified design criteria<br />
and monitoring criteria.<br />
For <strong>RBF</strong> to be eligible <strong>for</strong> credit as a pretreatment technique, the following proposed criteria must<br />
be met:<br />
• Wells must be drilled in an unconsolidated, predominantly sandy aquifer, as determined<br />
by grain-size analysis of recovered core material — the recovered core must contain<br />
greater than 10-percent fine-grained material (grains less than 1.0-millimeter diameter)<br />
in at least 90 percent of its length.<br />
• Wells must be located at least 25 ft (in any direction) from the surface-water source to be<br />
eligible <strong>for</strong> 0.5-log credit; wells located at least 50 ft from surface water are eligible <strong>for</strong><br />
1.0-log credit.<br />
• The wellhead must be continuously monitored <strong>for</strong> turbidity to ensure that no system<br />
failure is occurring. If the monthly average of daily maximum turbidity values exceeds<br />
1 ntu, the system must report this finding to the State. The system must also conduct an<br />
assessment to determine the cause of high turbidity levels in the well and consult with<br />
the State to determine whether the previously allowed credit is still appropriate.<br />
Systems using <strong>RBF</strong> as pretreatment to a filtration plant at the time that the system is required to<br />
monitor <strong>for</strong> Cryptosporidium must sample the well effluent <strong>for</strong> the purpose of determining the bin<br />
Correspondence should be addressed to:<br />
Stig Regli<br />
Environmental Engineer<br />
United States Environmental Protection Agency<br />
OGWDW (4607M) • 1200 Pennsylvania Avenue NW • Washington, D.C. 20460 USA<br />
Phone: (202) 564-5270 • Fax: (202) 564-3767 • Email: regli.stig@epa.gov<br />
149
150<br />
classification. Where bin classification is based on monitoring the well effluent, systems are not<br />
eligible to receive additional credit <strong>for</strong> <strong>RBF</strong>. The rationale <strong>for</strong> the above proposed criteria and<br />
opportunity <strong>for</strong> public comment are described in detail in the Federal Register (68FR47692) and<br />
are available on the web at http://www.regulations.gov/fredpdfs/03-18295.pdf.<br />
<strong>RBF</strong> also provides the opportunity <strong>for</strong> directly reducing organic DBP precursor levels or indirectly<br />
facilitating the application of advanced precursor removal technologies, such as nanofiltration.<br />
The USEPA is proposing the Stage 2 Disinfection By-Product Regulation to mitigate concerns <strong>for</strong><br />
the potential risk of developmental and reproductive effects from DBPs. Since the compliance<br />
dates of the LT2ESWTR will coincide with those of the Stage 2 Disinfection By-Product<br />
Regulation, there may be opportunities <strong>for</strong> utilities to use <strong>RBF</strong> <strong>for</strong> helping to achieve compliance<br />
with both regulations.<br />
The simultaneous reduction of other regulated contaminants by irreversible adsorption, biodegradation,<br />
dilution with groundwater, or attenuation mechanisms is also possible with <strong>RBF</strong>. This is<br />
clearly a site-specific issue whose success cannot be assumed without adequate testing/monitoring.<br />
REFERENCE<br />
USEPA (2003). <strong>National</strong> Primary Drinking <strong>Water</strong> Regulations: Long Term 2 Enhanced Surface <strong>Water</strong> Treatment<br />
Rule; Propose Rule, Federal Register, 68(154): 47,691-47,696.<br />
STIG REGLI is an Environmental Engineer <strong>for</strong> the Office of Ground <strong>Water</strong> and Drinking<br />
<strong>Water</strong> of the United States Environmental Protection Agency. He has been with the<br />
United States Environmental Protection Agency since 1979 and is involved with<br />
developing national drinking-water regulations <strong>for</strong> public water systems. His major focus<br />
has been as a Regulation Manager (1985 to 1996) and, more recently, as a Senior Science<br />
Advisor pertinent to the control of pathogens and disinfection byproducts. He has also<br />
been on extended leave to work on drinking-water related projects in Somalia and<br />
Thailand. Prior to working at the United States Environmental Protection Agency, he taught environmental<br />
engineering courses as a Peace Corps volunteer at Kabul University in Kabul, Afghanistan. Regli received<br />
both a B.S. in Mechanical Engineering and an M.S. in Civil Engineering from Duke University.
Session 8: Emerging Contaminants Removal<br />
Transport and Attenuation of<br />
Pharmaceutical Residues During Bank Filtration<br />
Andy Mechlinski<br />
<strong>Institute</strong> of Food Chemistry<br />
Technical University of Berlin<br />
Berlin, Germany<br />
Thomas Heberer, Ph.D.<br />
<strong>Institute</strong> of Food Chemistry<br />
Technical University of Berlin<br />
Berlin, Germany<br />
Bank filtration and artificial groundwater recharge are important, effective, and cheap techniques<br />
to treat surface water and remove microbes and inorganic and (some) organic contaminants;<br />
however, the purification capacity of these techniques varies and is limited in the removal of some,<br />
but not all, potential impurities. Thus, bank-filtration research began investigating pharmaceutically<br />
active compounds (PhACs) when a number of groundwater samples from bank filtration sites<br />
positively detected these compounds. To date, the mechanisms <strong>for</strong> removing impurities and<br />
chemical reactions of the water components have not sufficiently been understood. These subjects<br />
are currently addressed in a new research project called NASRI. In this interdisciplinary project,<br />
the fate and transport of some new emerging contaminants during bank filtration are investigated<br />
at Tegel and Wannsee Lakes and at a groundwater replenishment infiltration pond in Berlin,<br />
Germany. The locations of the field-sites are shown in Figure 1.<br />
The field sites are equipped with different types of monitoring wells that screen at various depths<br />
and are drilled between the infiltration bank and water-supply wells, as well as behind the watersupply<br />
wells. An example of the transects is shown in Figure 2.<br />
These transects are sampled monthly, which allows the fate and transport of PhACs during<br />
groundwater recharge to be monitored. The samples are analyzed by solid phase extraction,<br />
chemical derivatization, and gas chromatography-mass spectrometry (GC/MS) applying selected<br />
ion monitoring. Two novel analytical methods detect PhACs even in complex environmental<br />
samples (Reddersen, 2003). Additionally, some other PhACs (such as antibiotics and estrogenic<br />
steroids) are analyzed by high-pressure liquid chromatography mass spectrometry (HPLC-MS/MS).<br />
In Berlin surface waters, PhACs are found up to the microgram-per-liter level as highly persistent<br />
residues. Six PhACs (Figure 3), including the analgesic drugs diclofenac and propyphenazone, the<br />
antiepileptic drugs carbamazepine and primidone, and the drug metabolites clofibric acid and<br />
1-acetyl-1-methyl-2-dimethyl-oxamoyl-2-phenylhydrazide (AMDOPH), were found to leach from<br />
contaminated watercourses into groundwater aquifers. They also occur at low concentrations in<br />
receiving water-supply wells. Bank filtration was, however, also found to decrease concentrations<br />
(e.g., of diclofenac) or even remove some PhACs (bezafibrate, indomethacine, antibiotics, and<br />
estrogens). Other PhACs (such as carbamazepine and, especially, primidone) were identified as<br />
being excellent tracers of sewage contamination in surface water and groundwater.<br />
Correspondence should be addressed to:<br />
Andy Mechlinski<br />
Graduate Student<br />
<strong>Institute</strong> of Food Chemistry<br />
Technical University of Berlin • Sekr. TIB 4/3-1 • Gustav-Meyer-Allee 25 • 13355 Berlin, Germany<br />
Phone: +49 (30) 314-72267 • Fax: +49 (30) 314-72823 • Email: andymechlinski@web.de<br />
151
152<br />
Acknowledgements<br />
The authors would like to thank Veolia <strong>Water</strong> and the Berlin <strong>Water</strong> Company <strong>for</strong> financing the<br />
NASRI project.<br />
REFERENCE<br />
TS Tegeler See<br />
GWA Tegel<br />
3339<br />
3338<br />
Sickerbecken<br />
TS Lieper Bucht<br />
Unterhavel<br />
TEG365 TEG248<br />
TEG364<br />
TEG247<br />
3337 3335<br />
4<br />
3336 3334<br />
TS Wannsee<br />
Galerie Wannsee<br />
3333<br />
3332<br />
3311<br />
3310<br />
Br.20<br />
Tegeler See<br />
3301<br />
3302<br />
3309<br />
3308<br />
3307<br />
3306 12<br />
3303 13 3305<br />
14<br />
Bernauerst.<br />
3304<br />
Galene West<br />
WW Kladow<br />
Testfeld<br />
Marienfelde<br />
WW Spandau<br />
Spandau<br />
Havel<br />
WW Stolpe<br />
Kleinmachnow<br />
KW Stahnsdorg<br />
Tegeler See<br />
PEA-Tegel<br />
WW Tegel<br />
WW Jungfernheide<br />
Spree<br />
Langsamsandfilter<br />
Uferfiltration<br />
8 Peilrohre<br />
Tegeler Fließ<br />
Charlottenburg<br />
KW Ruhleben<br />
Unterschleuse<br />
WW Tiefwerder<br />
WW und PEA<br />
Beelitzhof<br />
Einleitung<br />
KW Ruhleben<br />
Teitkowkanal<br />
Messstellen<br />
KW Schönerlinde<br />
(KW Marienfelde)<br />
KW Waßmannsdorf<br />
KW Münchehofe<br />
WW Friedrichshagen<br />
Reddersen, K., and T. Heberer (2003). “Multi-methods <strong>for</strong> the trace-level determination of pharmaceutical<br />
residues in sewage, surface and ground water samples applying GC-MS.” J. Sep. Sci. (in press).<br />
Nordgraben<br />
Panke<br />
Speicherteich<br />
Mühlendamm<br />
Oberschleuse<br />
Landwehrkanal<br />
Neukölln<br />
Spree<br />
Panke<br />
Tettowkanal<br />
MHG<br />
KW Falkenberg<br />
WW Wuhlheide<br />
Wuhle<br />
WW Johannisthal<br />
TS Müggelsee E.<br />
Dahme<br />
Erpe<br />
Wasserwerk waterworks<br />
Klärwerk sewage treatment plant<br />
Phospat-Eliminationsanlage<br />
surface water treatment plant<br />
TS Transsekte Testfield<br />
Wehr weir<br />
Schleuse lock<br />
Nottekanal<br />
Neue Mühle<br />
Fredersdorfer<br />
Fließ<br />
Wernsdorf<br />
© Free University of Berlin<br />
Federal Environmental Agency (UBA)<br />
TS Müggelsee C.<br />
Flakenfließ und<br />
Löcknitz<br />
Figure 1. Locations of the Tegel transect, Wannsee transect, and groundwater recharge area in Berlin, Germany.<br />
SW<br />
? ?<br />
existing observation wells<br />
Pegel Tegler See<br />
3310 3309<br />
3311<br />
?<br />
3301 3308<br />
TEG371OP<br />
TEG371UP<br />
3302 3307<br />
approximately 120 m<br />
TEG372<br />
3303<br />
3306<br />
new observation wells (08/2002)<br />
glacial till<br />
Br 13<br />
NE<br />
3304 3305<br />
Figure 2. Profile of the Tegel transect with 14 observation wells and <strong>Water</strong> Works Well Br. 13.<br />
Woltersdorf<br />
Große Tränke<br />
depth below<br />
ground [m]<br />
0 m<br />
5 m<br />
10 m<br />
15 m<br />
20 m<br />
25 m<br />
30 m<br />
35 m<br />
40 m<br />
45 m<br />
veritas<br />
iustitia<br />
libertas<br />
Freie Universität Berlin
H 3C O O N<br />
N<br />
H3C N CH3 Cl<br />
AMDOPH Bezafibrate<br />
O<br />
H 5C 6<br />
H 5C 2<br />
Cl<br />
CH 3<br />
H<br />
N<br />
O<br />
CH 3<br />
Clofibric Acid<br />
NH<br />
CH 3<br />
C C COOH<br />
CH 3<br />
N<br />
O<br />
C<br />
N<br />
HOOC<br />
Primidone Propyphenazone Indometacine<br />
Figure 3. Structures of some detected PhACs.<br />
NH-CH 2CH 2<br />
O<br />
ANDY MECHLINSKI is a Ph.D. student at the <strong>Institute</strong> of Food Chemistry of the<br />
Technical University of Berlin in Germany. His research interests include the analysis of<br />
pharmaceutically active compounds by gas chromatography-mass spectrometry and the<br />
investigation of the environmental fate and transport of pharmaceuticals during<br />
groundwater recharge at bank-filtration sites in Berlin. In 2001, he took the first part of<br />
the final examination covering food chemistry at the Technical University of Berlin. After<br />
this, he prepared a diploma thesis concerning the transport and attenuation of pharmaceuticals<br />
during bank filtration at two field sites in Berlin. From 2002 until 2005, he will be involved in the<br />
interdisciplinary Natural and Artificial Systems <strong>for</strong> Recharge and Infiltration project. Mechlinski recieved<br />
his undergraduate degree at the <strong>Institute</strong> <strong>for</strong> Food Chemistry of the Technical University of Berlin, where he<br />
investigated the behavior of several pharmaceuticals and other contaminants during riverbank filtration at<br />
two field sites in Berlin.<br />
H 3C<br />
CH 3<br />
C<br />
CH 3<br />
O<br />
CH 3<br />
CH 2<br />
H<br />
N<br />
Cl<br />
Diclofenac<br />
N<br />
O<br />
Carbamazepine<br />
Cl<br />
C<br />
2H 2<br />
CH 3<br />
N<br />
C CH 2<br />
O<br />
COOH<br />
Cl<br />
153
154
Session 8: Emerging Contaminants Removal<br />
Attenuation of Pharmaceuticals<br />
During Riverbank Filtration<br />
Traugott Scheytt, Ph.D.<br />
<strong>Institute</strong> of Applied Geosciences<br />
Technical University Berlin<br />
Berlin, Germany<br />
Petra Mersmann<br />
<strong>Institute</strong> of Applied Geosciences<br />
Technical University Berlin<br />
Berlin, Germany<br />
Marcus Leidig<br />
<strong>Institute</strong> of Applied Geosciences<br />
Technical University Berlin<br />
Berlin, Germany<br />
Thomas Heberer, Ph.D.<br />
<strong>Institute</strong> of Food Chemistry<br />
Technical University Berlin<br />
Berlin, Germany<br />
Objective<br />
Occurrences of PhACs from human medical care are reported in groundwater not only from the<br />
Berlin (Germany) area (Heberer et al., 1998; Scheytt et al., 1998), but also from several places<br />
worldwide (Heberer, 2002). These findings initiated more specific investigations concerning the fate<br />
and transport of those pharmaceuticals under water-saturated and unsaturated conditions (Scheytt<br />
et al., in preparation). Field investigations at the bank infiltration site at Lake Tegel (Berlin,<br />
Germany) show the presence of several classes of pharmaceuticals, such as antirheumatics<br />
(e.g., diclofenac), analgesics (e.g., propyphenazone), and blood lipid regulators (clofibric acid) in<br />
both surface water and groundwater (Verstraeten et al., 2003). At Lake Tegel, clofibric acid was found<br />
at concentrations up to 290 nanograms per liter and propyphenazone up to 250 nanograms per liter,<br />
whereas concentrations of diclofenac were around detection limit.<br />
These results will be compared to preliminary results from an ongoing study at the Santa Ana<br />
River in Orange County, Cali<strong>for</strong>nia. Using its <strong>for</strong>ebay facilities, the Orange County <strong>Water</strong> District<br />
Correspondence should be addressed to:<br />
Traugott Scheytt, Ph.D. (after December 2003)<br />
Department of Civil and Environmental Engineering<br />
609 Davis Hall • University of Cali<strong>for</strong>nia • Berkeley, Cali<strong>for</strong>nia 94720 USA<br />
Phone: (510) 642-0151 • Fax: (510) 642-7483 • Email: scheytt@ce.berkeley.edu<br />
Traugott Scheytt, Ph.D. (until December 2003)<br />
Associate Professor <strong>for</strong> Hydrogeology<br />
<strong>Institute</strong> of Applied Geosciences<br />
Technical University Berlin • Ackerstr. 71-76 • 13355 Berlin, Germany<br />
Phone: +49 30 314-72417 • Fax: +49 30 314-25674 • Email: traugott.scheytt@tu-berlin.de<br />
155
156<br />
recharges surface water from the Santa Ana River into the Orange County Groundwater Basin.<br />
The fact that a high amount of the Santa Ana River’s flow is of wastewater origin has prompted<br />
concern that PhACs that survive secondary and tertiary treatment and groundwater transport may<br />
end up in the aquifer.<br />
To understand the transport and attenuation processes of those substances, laboratory column<br />
experiments were conducted (Mersmann et al., 2002). Based on these results, it is possible to<br />
identify and predict the transport and fate of PhACs during bank infiltration and to identify the<br />
pharmaceuticals, which may enter the aquifer via bank infiltration.<br />
Materials and Methods<br />
Pleistocene sediments from a well drilling campaign carried out by the Berlin <strong>Water</strong> Works were<br />
used to prepare the soil column <strong>for</strong> the column leaching experiment. The sediment was sampled<br />
at a depth of approximately 60-m below ground level and consisted of medium-grained sand. The<br />
sediment was manually packed into a stainless steel column measuring 35 × 14 centimeters (inner<br />
diameter). A gauze net and 0.5-diameter globes were placed at both the top and bottom of the<br />
column to prevent soil particles from leaching. The column was pre-wetted and, afterwards,<br />
equilibrated with groundwater originating from the same location and depth like the sediment<br />
itself. Groundwater was led through the column with a flow rate of about 0.3 m per day and a<br />
bottom-to-top flow direction to ensure saturated conditions in the column.<br />
Equilibration of the column with pure groundwater took about 5 days (4.83 pore volumes) be<strong>for</strong>e<br />
pharmaceutical compounds were applied. The groundwater used <strong>for</strong> the column experiment<br />
represents a typical groundwater from the Berlin area not contaminated by any PhAC residues.<br />
Lithium chloride (LiCl) (used as tracer) and the pharmaceuticals were applied to the same<br />
groundwater, which was then passed through the column <strong>for</strong> approximately 10 days. Beside the<br />
tracer and pharmaceutical compounds, all other parameters (e.g., flow rate) were kept the same<br />
during all three phases of the study. The experiments took place at a room temperature of<br />
approximately 20-degrees Celsius and all parts of the column experiment, including the tank, were<br />
protected against exposure to light. The eluted liquid was collected in fractions of approximately<br />
25 milliliters and analyzed <strong>for</strong> contents of anions, cations, pharmaceutical chemicals, and the<br />
lithium chloride used as a tracer.<br />
The concentration of the lithium chloride tracer was 10 mg/L, and the pharmaceuticals had a<br />
concentration of 10 µg/L in the spiked water. Physico-chemical parameters (redox potential, pH,<br />
temperature, oxygen saturation, specific conductance) were measured every 10 minutes using<br />
respective electrodes coupled to a data logger. Lithium chloride was chosen as a tracer because the<br />
background concentration of lithium was definitely below detection limit in all sediment and<br />
groundwater samples used <strong>for</strong> the experiments. Lithium also shows a transport behavior comparable<br />
to a nonreactive tracer and can be analyzed in a rapid and cost-effective manner.<br />
For the analysis of pharmaceutical compounds, water samples were adjusted to a pH of 2 and then<br />
extracted by solid-phase extraction using a non-endcapped reversed phase adsorbent<br />
(RP-C18 Bakerbond Polar Plus). Then the analytes and surrogate standard were derivatized,<br />
making them amendable to gas chromatographic separation (Heberer et al., 1998).<br />
Two microliters of the sample extracts (100 microliters <strong>for</strong> each sample) were analyzed by capillary<br />
GC-MS with selected ion monitoring. Depending on the sample volume (100 to 1,000 milliliters),<br />
the limits of determination were between 1 and 10 nanograms per liter, and the limits of quantitation<br />
were between 5 and 25 nanograms per liter. The analytical recoveries range between 80 and 120 percent.<br />
For further analytical details, refer to Heberer et al. (1998).
Results and Conclusion<br />
The Santa Ana River study will provide in<strong>for</strong>mation about the concentration of some high-volume<br />
pharmaceuticals from human medical care in the Santa Ana River, especially acetaminophen,<br />
carbamazepine, diclofenac, gemfibrozil, ibuprofen, metoprolol, naproxen, primidone, and propranolol.<br />
Laboratory experiments show that clofibric acid exhibits no degradation and almost no<br />
retardation (Rf = 1.1). After spiking groundwater with lithium chloride and diclofenac, the<br />
movement of diclofenac within the column is much slower than the movement of the lithium<br />
tracer (Figure 1), leading to a retardation factor of Rf = 2.6 <strong>for</strong> diclofenac. Propyphenazone<br />
(Rf = 2.0) is also retarded, whereas significant degradation was not observed <strong>for</strong> both pharmaceuticals,<br />
diclofenac and propyphenazone, under prevailing conditions in the soil column. Ibuprofen<br />
is degraded under aerobic conditions, whereas only little degradation was observed under<br />
anaerobic conditions. The retardation factor <strong>for</strong> ibuprofen was extrapolated to be 4.0.<br />
Carbamazepine shows no degradation in the soil column experiments, but significant retardation<br />
(Rf = 2.8) under prevailing conditions.<br />
C/C 0<br />
1.2<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0.0<br />
0.0 5.0 10.0 15.0<br />
Pore Volume<br />
20.0 25.0 30.0<br />
Lithium Diclofenac<br />
Figure 1. Concentration of lithium and diclofenac at the outflow of the soil<br />
column; results from a single-substance laboratory soil column experiment.<br />
It was concluded that at least clofibric acid, propyphenazone, and carbamazepine are recalcitrant<br />
under groundwater conditions and will be transported within the aquifer at the Berlin bank-filtration<br />
site. Additionally, due to anaerobic conditions in the deeper part of the aquifer, diclofenac and<br />
ibuprofen may also occur in groundwater and are transported if these compounds are not completely<br />
degraded in surface water or in the aerobic part of the aquifer.<br />
Compared to the bank-infiltration site at Lake Tegel in Berlin, the situation is quite different at<br />
the Santa Ana River, especially in respect to the climate and hydrogeological setting. Among the<br />
already mentioned pharmaceuticals of interest at the Santa Ana River, acetaminophen is<br />
prescribed in high amounts, but it is expected that this compound will be attenuated during bank<br />
infiltration due to its high biodegradability. Additionally, due to mostly aerobic conditions in the<br />
aquifer at the Santa Ana River, diclofenac and ibuprofen will be degraded under these aerobic<br />
conditions. Among the pharmaceuticals that might be expected in groundwater are carbamazepine,<br />
primidone, gemfibrozil, metoprolol, naproxen, and propranolol.<br />
157
158<br />
Acknowledgments<br />
Deutsche Forschungsgemeinschaft has funded parts of this work, and the <strong>National</strong> <strong>Water</strong> <strong>Research</strong><br />
<strong>Institute</strong> is funding the research on attenuation of pharmaceuticals during recharge at the Santa<br />
Ana River. The authors appreciate the cooperation of the Berliner Wasserbetriebe in gaining<br />
access to waterworks wells and the help of the Orange County <strong>Water</strong> District, which has greatly<br />
contributed to the study.<br />
REFERENCES<br />
Heberer, T. (2002). “Occurrence, fate, and removal of pharmaceutical residues in the aquatic environment:<br />
A review of recent research data.” Toxicology Letters, 131: 5-17.<br />
Heberer, T., K. Schmidt-Bäumler, and H.-J. Stan (1998). “Occurrence and distribution of organic contaminants<br />
in the aquatic system in Berlin. Part I: Drug residues and other polar contaminants in Berlin surface<br />
and groundwater.” Acta Hydrochimica et Hydrobiologica, 26: 272-278.<br />
Mersmann, P., T. Scheytt, and T. Heberer (2002). “Column experiments on the transport behavior of<br />
pharmaceutically active compounds in the saturated zone.” Acta Hydrochimica et Hydrobiologica,<br />
30(5-6): 1-10.<br />
Scheytt, T., T. Mersmann, M. Leidig, A. Pekdeger, and T. Heberer (in preparation). “Transport of<br />
pharmaceutically active compounds (PhACs) clofibric acid, diclofenac, and propyphenazone under water<br />
saturated conditions.” Ground <strong>Water</strong>, Westerville, OH.<br />
Scheytt, T., S. Grams, and H. Fell (1998). “Occurrence and behavior of drugs in groundwater.” Gambling with<br />
groundwater – physical, chemical, and biological aspects of aquifer-stream relations, J.V. Brahana, Y. Eckstein,<br />
L.K. Ongley, R. Schneider, and J.E. Moore, eds., IAH/AIH Proc., St. Paul., MN.<br />
Verstraeten, I.M., T. Heberer, and T. Scheytt (2003). “Occurrence, characteristics, transport, and fate of pesticides,<br />
pharmaceutically active compounds, and industrial products and personal care products at bank-filtration<br />
sites.” Riverbank Filtration: Improving Source-<strong>Water</strong> Quality, C. Ray, G. Melin, and R.B. Linsky, eds.,<br />
Kluwer Academic Publishers, Dordrecht.<br />
TRAUGOTT SCHEYTT has been an Associate Professor <strong>for</strong> Hydrogeology in the<br />
Department of Civil Engineering and Applied Geosciences at the Technical University<br />
Berlin since 1996. His teaching duties include courses on groundwater chemistry,<br />
groundwater flow, and transport of organic contaminants. Current research areas include<br />
the transport and attenuation of pharmaceutically active compounds, natural attenuation<br />
of contaminants emanating from landfills, and prediction of transport and degradation of<br />
organic contaminants in groundwater. Presently, he is on sabbatical at University of<br />
Cali<strong>for</strong>nia, Berkeley, and will stay in Cali<strong>for</strong>nia until January 2004 to conduct research on the transport and<br />
attenuation of pharmaceutically active compounds during riverbank filtration at the Santa Ana River in<br />
Orange County, Cali<strong>for</strong>nia. Scheytt received an M.S. and Ph.D. in Geology from Christian-Albrechts-<br />
University Kiel in Germany.
Session 8: Emerging Contaminants Removal<br />
The Fate of Bulk Organics and Emerging Contaminants<br />
During Soil-Aquifer Treatment<br />
Dr. Jörg E. Drewes<br />
Colorado School of Mines<br />
Golden, Colorado<br />
Dipl.-Ing. Tanja Rauch<br />
Colorado School of Mines<br />
Golden, Colorado<br />
Introduction<br />
Wastewater reuse employing soil-aquifer treatment is becoming an increasingly important strategy<br />
<strong>for</strong> many utilities in the United States and abroad to augment local drinking-water sources where<br />
supplies are limited. One major water-quality issue associated with soil-aquifer treatment leading<br />
to indirect potable or nonpotable reuse of wastewater effluents is the fate and transport of organic<br />
constituents. Effluent-derived bulk organic matter can impair the quality of recovered water by<br />
interfering with post-treatment processes, such as coagulation, adsorption, or membrane<br />
treatment. Bulk organic matter is also a known precursor <strong>for</strong> DBPs. The presence of bioavailable<br />
organic carbon after soil-aquifer treatment could also increase the microbial regrowth potential in<br />
distribution systems. In addition, BOM might be comprised of organic micropollutants, which<br />
survive during soil-aquifer treatment and which are associated with potential adverse human<br />
health effects. The objective of this study was to investigate the fate and transport of bulk and<br />
trace organics present in reclaimed water during soil-aquifer treatment leading to indirect potable<br />
reuse. The American <strong>Water</strong> Works Association <strong>Research</strong> Foundation and USEPA funded this study.<br />
Methodology<br />
The fate of organics during soil-aquifer treatment was investigated using controlled biodegradation<br />
studies in adapted soil-columns and full-scale infiltration facilities in Arizona and Cali<strong>for</strong>nia. The<br />
study design followed a watershed-guided approach considering hydraulically corresponding<br />
samples of drinking-water sources, soil-aquifer treatment-applied wastewater effluents, and<br />
subsequent post-soil-aquifer treatment samples representing a series of different travel times in the<br />
subsurface. Extensive characterization of organic carbon in the different samples was per<strong>for</strong>med<br />
using state-of-the-art analytical techniques (such as size-exclusion chromatography with online<br />
DOC and UV absorbance detection, carbon-13 nuclear magnetic resonance spectroscopy, Fourier<br />
trans<strong>for</strong>m infrared spectroscopy, and elemental analysis) and biomass/bioactivity measurements<br />
(dehydrogenase activity; total viable biomass through phospholipid extraction; substrate induced<br />
respiration) to distinguish between primary removal mechanisms (biodegradation versus adsorption).<br />
The mechanisms contributing to bulk organic matter removal during initial recharge were<br />
Correspondence should be addressed to:<br />
Dr. Jörg E. Drewes<br />
Assistant Professor<br />
Environmental Science & Engineering Division<br />
Colorado School of Mines • Golden, Colorado 80401-1887 USA<br />
Phone: (303) 273-3401 • Fax: (303) 273-3413 • Email: jdrewes@mines.edu<br />
159
160<br />
identified by isolating three bulk fractions from treated wastewater effluents:<br />
• Hydrophilic carbon (HPI).<br />
• Hydrophobic acids (HPO-A).<br />
• Colloidal organic matter.<br />
HPI and HPO-A were isolated from a domestic secondary effluent by XAD-8 fractionation.<br />
Colloidal organic matter was operationally defined as organic matter in the size range of<br />
approximately 6,000 Dalton to 1 micrometer and was isolated using dialysis diffusion after a preconcentration<br />
using large volume rotary evaporation. Trace organics selected <strong>for</strong> this study were<br />
PhACs and personal care products, as well as selected endocrine disrupting compounds<br />
(17β-estradiol, estriol, and testosterone).<br />
Pharmaceutical compounds were selected <strong>for</strong> this study based on their production, per-capita<br />
consumption, and occurrence in domestic effluents and surface waters in the United States and<br />
Middle Europe. Caffeine has been a well-known PhAC of wastewater origin <strong>for</strong> more than<br />
20 years. In the category of analgesics/anti-inflammatory drugs, diclofenac, ibuprofen, ketoprofen,<br />
naproxen, fenoprofen, propyphenazone, meclofenamic acid, and tolfenamic acid were identified.<br />
Carbamazepine and primidone were identified as antiepileptic drugs. Pentoxifylline represented a<br />
common blood viscosity-affecting agent. Gemfibrozil, clofibric acid, and fenofibrate were selected<br />
to represent blood lipid regulators. Organic iodine was used as a surrogate parameter <strong>for</strong> the<br />
detection of X-ray contrast agents (triiodinated benzene derivates such as iopromide and<br />
diatrizoate) and their halogenated metabolites.<br />
Results<br />
Data derived from several soil-aquifer treatment field sites consistently indicate a substantial<br />
removal of DOC during travel through the subsurface. At one site employing tertiary treatment<br />
prior to recharge, average DOC concentrations of 5.6 mg/L in the tertiary effluent decreased from<br />
5.6 mg/L to approximately 1.45 mg/L in groundwater monitoring wells after traveling 6 to 12 months<br />
in the subsurface. At the same time, the specific UV absorbance increased, indicating a preferred<br />
removal of aliphatic compound groups within bulk organic matter. Further DOC removal<br />
occurred during long-term travel in the subsurface of 1 to 2 years from 1.45 mg/L to approximately<br />
1.1 mg/L. At another site employing only secondary treatment without nitrification prior to<br />
recharge, the DOC concentration was efficiently reduced to approximately 2 mg/L after<br />
percolating through the vadose zone. The specific UV absorbance increased only slightly from<br />
1.7 liters per milligram and meter (L/mg m) in the secondary treated effluent to 2.0 L/mg m in<br />
groundwater samples. These findings pointed to a fast and substantial removal of biodegradable<br />
organic carbon during the initial phase of groundwater recharge. The results of spectroscopic<br />
investigations generally demonstrated that naturally derived and effluent-derived organic matter<br />
after SAT overlap extensively in molecular weight distribution, amount and distribution of<br />
hydrophobic and hydrophilic carbon fractions, and chemical characteristics based on elemental<br />
analysis and structural analysis; however, the residual portion of DOC contained both<br />
effluent-derived organic matter and NOM.<br />
Results of this study indicate that during vadose zone treatment, HPI is removed by a combination<br />
of physical and biological mechanisms. HPO-A is least efficiently removed of all fractions during<br />
soil infiltration. Both adsorption and biotrans<strong>for</strong>mation seem to contribute to this removal. Low<br />
biomass activity responses in soil-columns fed with HPO-A indicated that these substances are<br />
refractory in nature. The fate of effluent-derived colloidal organic matter during groundwater
echarge is not well understood. Our studies give reason to believe that colloidal organic matter<br />
can be efficiently removed during initial soil recharge, and that this removal is based on the<br />
filtration mechanism and biodegradation. Colloidal organic matter stimulated high biomass<br />
activities, but was not fully mineralized. Remobilization and the breakthrough of colloidal organic<br />
matter was observed in column experiments fed with colloidal organic matter under higher<br />
infiltration rates. The removal of colloidal organic matter might, there<strong>for</strong>e, depend upon hydraulic<br />
regimes in the aquifer during recharge operation. The majority of biological removal in the HPI<br />
and HPO-A fractions occurred in the first 30-centimeters of soil infiltration.<br />
Findings of this study demonstrated that the dominating removal mechanism <strong>for</strong> steroids during<br />
soil-aquifer treatment is adsorption, although biodegradation is also taking place and is important<br />
<strong>for</strong> a sustainable operation avoiding compound accumulation in the system. The study showed<br />
that steroid removal was not dependent upon the type of organic background matrix present (HPI,<br />
HPO-A, colloidal organic matter) or redox and flox conditions (aerobic versus anoxic; saturated<br />
versus unsaturated). 17β-estradiol, estriol, and testosterone were efficiently removed from initial<br />
concentrations as high as 500 nanograms per liter after only 5.2 hours of contact with subsurface<br />
media in the presence of soil microbial activity.<br />
In addition, the study revealed that the stimulant caffeine, analgesic/anti-inflammatory drugs, and<br />
blood lipid regulators were efficiently removed to concentrations near or below the detection limit<br />
of the analytical method after retention times of less than 6 months during groundwater recharge.<br />
The antiepileptics carbamazepine and primidone were not removed during groundwater recharge<br />
under either anoxic saturated or aerobic unsaturated flow conditions during travel times of up to<br />
8 years. Organic iodine showed a partial removal only under anoxic saturated conditions (as<br />
compared to aerobic conditions) and persisted in recharged groundwater.<br />
Conclusions<br />
Findings of this study demonstrated that soil-aquifer treatment represents an efficient treatment<br />
step to reduce bulk organic carbon and to remove trace pollutants of concern present in reclaimed<br />
water.<br />
JÖRG DREWES is an Assistant Professor of Environmental Science and Engineering at<br />
the Colorado School of Mines. He has been actively involved in research in the area of<br />
water reclamation, water reuse, and groundwater recharge <strong>for</strong> more than 11 years. His<br />
research interests include water and wastewater treatment engineering; potable and nonpotable<br />
water reuse (soil-aquifer treatment and microfiltration/reverse osmosis); state-ofthe-art<br />
characterization of natural and effluent organic matter; contaminant transfer<br />
among environmental media; and the fate of endocrine disrupting compounds and<br />
pharmaceuticals in natural and engineered systems. Drewes has published more than 50 journal papers, book<br />
contributions, and conference proceedings. Among his honors, he was awarded the Willy-Hager Award in<br />
1997, Quentin Mees <strong>Research</strong> Award in 1999, and Dr. Nevis Cook Excellence in Teaching Award in 2003.<br />
Drewes received both an M.S. and Ph.D. in Environmental Engineering from the Technical University of<br />
Berlin, Germany.<br />
161
162
Session 8: Emerging Contaminants Removal<br />
Ethylenediaminetetraacetic Acid Occurrence and<br />
Removal Through Bank Filtration in the Platte River,<br />
Nebraska<br />
Jason R. Vogel, Ph.D.<br />
United States Geological Survey<br />
Lincoln, Nebraska<br />
Larry B. Barber, Ph.D.<br />
United States Geological Survey<br />
Boulder, Colorado<br />
Tyler B. Coplen, Ph.D.<br />
United States Geological Survey<br />
Reston, Virginia<br />
Ingrid M. Verstraeten, Ph.D.<br />
United States Geological Survey<br />
Baltimore, Maryland<br />
Thomas F. Speth, Ph.D., P.E.<br />
United States Environmental Protection Agency<br />
Cincinnati, Ohio<br />
Jerry Obrist, P.E.<br />
City of Lincoln <strong>Water</strong> System<br />
Lincoln, Nebraska<br />
The USGS, USEPA, and City of Lincoln <strong>Water</strong> System (Nebraska) have conducted a study to<br />
determine the occurrence and removal of EDTA, NTA, and nonylphenol monoethoxycarboxylate<br />
to nonylphenol pentaethoxycarboxylate (NP1EC-NP5EC) in the hydrologic system at the City of<br />
Lincoln well field. The objective of the study is to evaluate the occurrence and removal of EDTA,<br />
NTA, and total nonylphenolpolyethoxycarboxylate (NPEC) by bank filtration at the City of<br />
Lincoln well field. This presentation will discuss removal during two sampling periods — May and<br />
August 2002 — based upon surface-water fractions in the collector well calculated using stable<br />
isotope ratios of hydrogen and oxygen.<br />
The rationale <strong>for</strong> selecting compounds evaluated in this study was based on the hierarchical<br />
analytical approach and includes a range of compounds covering a spectrum of uses and effects.<br />
For example, EDTA is a low-toxicity, high-production volume chemical used in multiple<br />
domestic, commercial, and industrial applications to <strong>for</strong>m stable, water-soluble complexes with<br />
trace metals. Because of its uses and chemical characteristics, EDTA occurs at relatively high<br />
concentrations and can persist in the aquatic environment (Barber et al., 1996; Barber et al., 2000;<br />
Leenheer et al., 2001; Barber et al., 2003).<br />
Correspondence should be addressed to:<br />
Jason R. Vogel, Ph.D.<br />
Hydrologist<br />
United States Geological Survey<br />
Room 406, Federal Building • 100 Centennial Mall North • Lincoln, Nebraska 68508 USA<br />
Phone: (402) 437-5129 • Fax: (402) 437-5139 • Email: jrvogel@usgs.gov<br />
163
164<br />
Site Description<br />
There are 40 active production wells at the City of Lincoln well field. Two of these wells are<br />
horizontal collector wells screened in alluvial sand and gravel approximately 26-m below the<br />
Platte River; they provide approximately 50 percent of the municipal water during most times of<br />
the year. The remaining 38 wells are vertical production wells that are developed in alluvial<br />
sediments mainly consisting of sand and gravel. At this location, the quality of the river water has<br />
a large effect on the quality of the bank-filtered water obtained from the collector wells. Similarly,<br />
the water quality of the groundwater in the vertical production wells directly corresponds to the<br />
distances of the wells from the river. Because of the direct link between the collector wells and<br />
river, the collector well is usually turned off during the month of May and first part of June to avoid<br />
the flush of herbicides associated with spring planting in this agricultural area. After collection,<br />
well water is treated by ozonation, filtration, and chlorination be<strong>for</strong>e distribution.<br />
Sampling<br />
Representative surface-water samples were collected quarterly from the Platte River at the wellfield<br />
site using equal width-increment, flow-weighted sampling. Groundwater samples were also<br />
collected quarterly from one of the collector wells and from two of the vertical groundwater wells.<br />
One groundwater well was located relatively close to the river (within 100 m), with another<br />
located away from the river (1,000 m). All samples were filtered through 0.7-m glass fiber filters<br />
and collected in pre-cleaned amber glass bottles. Samples <strong>for</strong> EDTA, NTA, and NPEC analyses<br />
were preserved with 2-percent by volume <strong>for</strong>malin. This presentation will discuss the results of<br />
samples from May and August 2002.<br />
Analysis<br />
EDTA, NTA, and NP1EC-NP5EC were measured using a modification (Barber et al., 2000) of<br />
the method of Schaffner and Giger (1984). Samples (100 milliliters) were evaporated to dryness,<br />
acidified with 5-milliliter 50-percent (volume per volume) <strong>for</strong>mic acid/distilled water, and evaporated<br />
to dryness. Acetyl chloride/propanol (10-percent volume per volume) was added, the sample<br />
heated at 90-degrees Celsius <strong>for</strong> 1 hour to <strong>for</strong>m the propyl-esters, and the propyl-esters were<br />
extracted into chloro<strong>for</strong>m. The chloro<strong>for</strong>m extracts were evaporated to dryness and re-dissolved<br />
in toluene <strong>for</strong> analysis by GC/MS, as described below.<br />
The propyl-ester extracts were analyzed by electron-impact GC/MS in both the full-scan and<br />
select ion monitoring modes. The general GC conditions were:<br />
• Hewlett Packard (HP) 6890 GC.<br />
• Column-HP Ultra II (5-percent phenylmethyl silicone), 25 m × 0.2 millimeters,<br />
33-micrometer film thickness.<br />
• Carrier gas, ultra-high purity helium, with a linear-flow velocity of 27 centimeters per second.<br />
• Injection port temperature, 300-degrees Celsius.<br />
• Initial oven temperature, 50-degrees Celsius.<br />
• Split vent open, 0.75 minutes.<br />
• Ramp rate, 6-degrees Celsius per minute to 300-degrees Celsius.<br />
• Hold time, 15 minutes at 300-degrees Celsius.
The MS conditions were as follows:<br />
• HP 5973 Mass Selective Detector.<br />
• Tune with perflurotributylamine.<br />
• Ionization energy, 70 electron volt (eV).<br />
• Source pressure, 1 × 10 –5 torr.<br />
• Source temperature, 250-degrees Celsius.<br />
• Interface temperature, 280-degrees Celsius.<br />
• Full scan, 40 to 550 atomic mass units at one scan per second.<br />
Concentrations were calculated based on select ion monitoring data using diagnostic ions <strong>for</strong> each<br />
compound. Each compound was identified based on the matching of retention times (±0.02 minutes)<br />
and ion ratios (±20 percent) determined from the analysis of authentic standards. An eight-point<br />
calibration curve (typically ranging from 0.01 to 50 nanograms per microliter) and internal<br />
standard procedures were used <strong>for</strong> calculating concentrations. Surrogate standards were added to<br />
the samples prior to extraction and derivatization to evaluate compound recovery and whole<br />
method per<strong>for</strong>mance.<br />
Results<br />
In general, based upon results from May and August 2002, measured EDTA concentrations <strong>for</strong><br />
surface-water samples were larger than <strong>for</strong> groundwater samples. In addition, EDTA concentrations<br />
decreased and total NPEC concentrations increased in the groundwater as the distance of<br />
the well from the river increased. NTA was only detected in one surface-water sample at very low<br />
levels and not at all in groundwater samples.<br />
Using surface-water fractions in the collector well determined from deuterium and oxygen-18 ratios,<br />
the transport of EDTA was nearly conservative during these two sampling periods. Total NPEC<br />
concentrations were lower than predicted. Further analysis will be <strong>for</strong>thcoming in the presentation.<br />
REFERENCES<br />
Barber, L.B., G.K. Brown, and S.D. Zaugg (2000). “Potential endocrine disrupting organic chemicals in<br />
treated municipal wastewater and river water.” Analysis of Environmental Endocrine Disruptors, L.H. Keith,<br />
T.L. Jones-Lepp, and L.L. Needham, eds., American Chemical Society Symposium Series 747, American<br />
Chemical Society, Washington, DC, p. 97-123.<br />
Barber, L.B., E.T. Furlong, S.H. Keefe, G.K. Brown, and J.D. Cahill (2003). “Natural and Contaminant<br />
Organic Compounds in Boulder Creek, Colorado under High-Flow and Low-Flow Conditions, 2000.”<br />
Comprehensive water quality of the Boulder Creek <strong>Water</strong>shed, Colorado, under high-flow and low-flow conditions,<br />
S.F. Murphy, L.B. Barber, and P.L. Verplanck, eds., U.S. Geological Survey <strong>Water</strong> Resources Investigations<br />
Report 03-4045.<br />
Barber, L.B., J.A. Leenheer, W.E. Pereira, T.I. Noyes, G.A. Brown, C.F. Tabor, and J.H. Writer (1996).<br />
“Organic contamination of the Mississippi River from municipal and industrial wastewater.” U.S. Geological<br />
Survey, Circular 1133, p. 114-135.<br />
Leenheer, J.A., C.E. Rostad, L.B. Barber, R.A. Schroeder, R. Anders, and M.L. Davisson (2001). “Nature and<br />
chlorine reactivity of organic constituents from reclaimed water in groundwater, Los Angeles County,<br />
Cali<strong>for</strong>nia.” Environmental Science and Technology, 35: 3,869-3,876.<br />
Schaffner, C., and W. Giger (1984). “Determination of nitrilotriacetic acid in water by high-resolution gas<br />
chromatography.” Journal of Chromatography, 312: 413-421.<br />
165
166<br />
Hydrologist JASON VOGEL has been with the United States Geological Survey in<br />
Lincoln, Nebraska, <strong>for</strong> a little over a year. During that time, he has been the Project Chief<br />
<strong>for</strong> a bank-filtration study in cooperation with the United States Environmental Protection<br />
Agency and the City of Lincoln <strong>Water</strong> System, and is also Lead Scientist of the agricultural<br />
chemical transport team in the Nebraska District. Be<strong>for</strong>e joining the United States<br />
Geological Survey, Vogel was a research engineer in the Biosystems Engineering Department<br />
at Oklahoma State University <strong>for</strong> 5 years. He has published articles on a wide variety of<br />
topics, including geostatistics and stochastic design, vadose zone transport, and microbial transport in<br />
groundwater, and co-authored the chapter on Geostatistics in Statistical Methods in Hydrology, Second Edition,<br />
with C.T. Haan. Vogel received a B.S. in Biological Systems Engineering from the University of Nebraska,<br />
an M.S. in Agricultural Engineering from Texas A&M University, and Ph.D. in Biosystems Engineering from<br />
Oklahoma State University.
Session 9: Public Policy and Regulatory<br />
Riverbank Filtration as a Regional Supply Option<br />
<strong>for</strong> the United States<br />
Leo Gentile, P.G., CPG<br />
Jordan Jones & Goulding<br />
Norcross, Georgia<br />
David Haas, P.E.<br />
Jordan Jones & Goulding<br />
Norcross, Georgia<br />
D. Joseph Hagerty, Ph.D., P.E.<br />
University of Louisville, Kentucky<br />
Louisville, Kentucky<br />
Peggy H. Duffy, P.E.<br />
Hagerty Engineering<br />
Jeffersonville, Indiana<br />
Introduction<br />
<strong>RBF</strong> has been used on the lower Rhine River <strong>for</strong> over 130 years and, today, comprises 15 to 20 percent<br />
of the water supply in Germany as a whole (Schubert, 2000). In the United States, the majority<br />
of larger water systems (serving 10,000 or more) use surface water (Wang et al., 2002). Smaller<br />
systems supplying 10,000 or less often rely on groundwater <strong>for</strong> supply because of reduced treatment<br />
requirements. <strong>RBF</strong> is employed in communities in the United States such as Cedar Rapids, Iowa;<br />
Lincoln, Nebraska; Louisville, Kentucky; and Sonoma County, Cali<strong>for</strong>nia. Some communities<br />
such as St. Helens, Oregon, and Sioux Falls, South Dakota, essentially use <strong>RBF</strong> by drawing water<br />
from collector wells constructed adjacent to rivers, although this is usually recognized as GWUDI<br />
(Wang et al., 2002). Still, <strong>RBF</strong> is an underutilized supply whose benefits are likely available to<br />
more communities to help provide better and more uni<strong>for</strong>m raw-water quality. <strong>RBF</strong> potential is<br />
greatest <strong>for</strong> communities in the Midwestern United States that are located where glacial valleyfill<br />
aquifers occur that are interconnected to adjacent surface-water bodies.<br />
Conditions<br />
The need <strong>for</strong> water is the basic condition <strong>for</strong> <strong>RBF</strong> or any other supply source. The magnitude of<br />
the demand and physical conditions dictates what supply source is most feasible. If a community<br />
is located along a major river or stream, then surface-water intake is a relatively simple solution;<br />
however, when water treatment issues and costs are considered, then groundwater — which may<br />
require no more treatment than disinfection — is the next logical choice. <strong>RBF</strong> is an alternative<br />
Correspondence should be addressed to:<br />
Leo Gentile, P.G., CPG<br />
Senior Project Hydrogeologist<br />
Jordan, Jones & Goulding, Inc.<br />
6801 Governors Lake Parkway • Norcross, Georgia 30071 USA<br />
Phone: (678) 333-0148 • Fax: (770) 455-7391 • Email: LGENTILE@JJG.com<br />
167
168<br />
supply option when these conditions are not available that provides typically larger quantities<br />
than wells, and more uni<strong>for</strong>m and higher quality raw water compared to surface-water intake.<br />
Aquifer Thickness and Extent: To supply the millions of gallons a day needed <strong>for</strong> a typical<br />
municipal water supply, the aquifer needs to be sufficiently thick and extensive to sustain such a<br />
yield without depleting the aquifer. If the aquifer is thin (less than about 100-ft thick), the cone<br />
of depression and amount of drawdown from wells penetrating the aquifer will be narrow and<br />
steep, and the well will likely not be efficient in producing the required yields. Excessive<br />
drawdown indicates an inefficient well and/or an over-drafted aquifer. Constructing collector wells<br />
with multiple horizontal lateral collector galleries will help increase well efficiency and create a<br />
broader cone of depression with less drawdown. Still, if the aquifer is relatively thin and limited<br />
in aerial extent, the sustained (or safe yield) may not be sufficient to meet demand. With <strong>RBF</strong>,<br />
conventionally constructed wells, collector wells, or specially designed tunnels overcome the<br />
physical shortcomings of the aquifer. The leakance from an adjacent river or lake recharges the<br />
aquifer, augmenting the safe yield (Figure 1). Thus, a relatively thin and aerially limited aquifer<br />
can yield the water quantities needed <strong>for</strong> municipal supplies. <strong>RBF</strong> also yields water of more<br />
consistent and better overall quality.<br />
River or<br />
Recharge Pond<br />
<strong>Water</strong> Table<br />
A B<br />
NOT TO SCALE<br />
River or<br />
Recharge Pond<br />
NOT TO SCALE<br />
Pumping Well<br />
<strong>Water</strong> Table<br />
Modified from Gallaher and Prize, 1966<br />
EXPLANATION<br />
Figure 1. Groundwater flow recharge to a river (A) and an <strong>RBF</strong> well (B) in a sand and gravel aquifer (from<br />
Lloyd and Lyke, 1995).<br />
Aquifer Characteristics: Glacial and alluvial sand and gravel deposits are some of the most<br />
productive aquifers in the world (Todd, 1980). The hydraulic conductivities and amount of water<br />
in storage are often high. The recharge potential from adjacent streams, rivers, or lakes is also<br />
considered highly favorable as natural or induced vertical gradients, coupled with good hydraulic<br />
conductivity, allows <strong>for</strong> leakance from the surface-water body to the aquifer. Pleistocene-age sand<br />
and gravel deposits have been used <strong>for</strong> well fields and <strong>RBF</strong> sites. By comparison, high hydraulic<br />
conductivity and yields can also occur in bedrock aquifers with fracture or karst porosity; however,<br />
storage capacity may be limited. Furthermore, groundwater flow through fractures or karst porosity<br />
is essentially conduit flow. Recharge from adjacent rivers or lakes would be minimally filtered,<br />
reducing the benefit from <strong>RBF</strong>.<br />
Other Considerations: There must be effective interconnection between the aquifer and surface-water<br />
body. Occasionally, an aquifer is confined or semi-confined by a layer of finer sediments that<br />
reduces the potential <strong>for</strong> leakance from the surface to the aquifer (Figure 2). <strong>RBF</strong> sites are more<br />
efficient where there is good interconnection (over 90 percent of withdrawn water can be bank<br />
filtrate [Eckert et al., 2000]). Clogging of the interconnection through <strong>RBF</strong> use, or the natural<br />
accumulation of fines, also reduces the efficacy of <strong>RBF</strong>. The removal of streambed fines by water<br />
flow from naturally high-gradient streams and rivers or during floods will help restore infiltration.
FEET<br />
1,000<br />
900<br />
800<br />
700<br />
600<br />
VERTICAL SCALE GREATLY EXAGGERATED<br />
DATUM IS SEA LEVEL<br />
Modified from Gann and others, 1973<br />
0 1 2 MILES<br />
Figure 2. Glacial valley-fill aquifer overlain by fine grained sediments in the Great Miami River Valley, Ohio<br />
(from Miller and Appel, 1997).<br />
German researchers experimented with excavating a “window” by dredging accumulated<br />
sediment. While the measure increased <strong>RBF</strong> efficiency, it was considered temporary, lasting only<br />
weeks be<strong>for</strong>e silting over (Schubert, 2000). Rivers whose flow has been altered <strong>for</strong> flood control<br />
and navigation improvement, such as the Ohio River, are particularly prone to clogging. Steep<br />
gradients and floodwaters that may have scoured the riverbed in the past have been eliminated.<br />
Wells or specially designed tunnels used to extract groundwater and bank filtrate must also be<br />
properly located to maximize system efficiency. Rivers and streams that occur with the thumbprint<br />
of the most recent continental glacial period in the United States (e.g., Wisconsin ice age) have<br />
a similar terraced profile. The unaltered modern river channel carries normal stage flows within<br />
the primary terrace. This terrace corresponds to a 100-year flood plain. Where significantly altered<br />
<strong>for</strong> flood control or navigation, the river may now have inundated the primary terrace extending<br />
from bank to bank up to the secondary terrace. The secondary terrace extends outward to the<br />
500-year flood plain. This channel was cut into the bedrock by glacially fed rivers. The rivers were<br />
large braided streams laden with sand and gravel. As the glacial outwash diminished, the scour<br />
channels filled with sand and gravel to become aquifers. The tertiary terrace is the remnant of<br />
peak stage flooding and its resultant scouring into the bedrock. Relatively thin and finer grained<br />
sediments were deposited as the flood stage fell (see Figure 2). These sediments may have <strong>for</strong>med<br />
local, relatively less-productive aquifers that are not connected to the modern river channel. The<br />
secondary terrace is the most productive place to locate <strong>RBF</strong> wells (e.g., out of main navigation,<br />
interconnected to the river, relatively thicker aquifer, greater potential <strong>for</strong> effective filtration).<br />
Potential <strong>RBF</strong> Regions<br />
Grand River<br />
B<br />
0 1 2 KILOMETERS<br />
The greatest potential <strong>for</strong> <strong>RBF</strong> — and, in fact, where <strong>RBF</strong> has been implemented — in the United<br />
States is in the Midwestern states. Figure 3 shows the location of mid- to large-sized communities<br />
in the United States that are located along rivers and on alluvial aquifers. Due to site conditions,<br />
<strong>RBF</strong> is a viable water supply option <strong>for</strong> many, but not all, communities.<br />
Midwest stream-valley aquifers in Midwestern states west of the Mississippi River are important<br />
sources of water <strong>for</strong> communities and industries (Figure 4). These aquifers are unconsolidated sand<br />
and gravel deposits that are thicker, greater in extent, and more productive in the major river<br />
valleys of the region. The stream-valley aquifers are unconfined and in direct connection to adjacent<br />
streams and rivers. Average yields from wells range from 100 to 1,000 gallons per minute, with some<br />
wells yielding over 3,000 gallons per minute (Olcott, 1992). <strong>RBF</strong> opportunities may be available<br />
Buried Valley<br />
B<br />
EXPLANATION<br />
Clay, silt, and fine-grained sand<br />
Coarse-grained sand and gravel<br />
Sand, gravel, and boulders<br />
Till<br />
Bedrock<br />
169
170<br />
EXPLANATION<br />
Dune Sand<br />
NOT TO SCALE<br />
Recharge<br />
Discharge<br />
South<br />
N<br />
Figure 3. Potential <strong>RBF</strong> areas in the United States (from USGS, 1965).<br />
Evaporation<br />
Arkansas<br />
River<br />
Evapotranspiration<br />
<strong>Water</strong> Table<br />
Stream to<br />
Aquifer Leakage<br />
Aquifer<br />
Bedrock<br />
East<br />
using stream-valley aquifers that are located along major rivers such as the Arkansas, Des Moines,<br />
Grand (Michigan), Mississippi, Platte, Missouri, and Wisconsin rivers, plus numerous mediumsized<br />
streams that course the region (Miller and Appel, 1997).<br />
West<br />
Precipitation<br />
Canal Diversion System<br />
Irrigation<br />
Well<br />
Crop Consumptive Use<br />
Deep Percolation<br />
of Precipitation and<br />
Applied Irrigation <strong>Water</strong><br />
Pumpage<br />
Modified from Barker and Others, 1983<br />
North<br />
Figure 4. Midwestern stream valley site favorable <strong>for</strong> <strong>RBF</strong> (Arkansas River Valley) (from Miller and Appel,<br />
1997).<br />
Legend<br />
Metropolitan Areas with<br />
Bank Infiltration Potential<br />
Population 250,000–1,000,000<br />
Population 1,000,001–2,500,000<br />
Population greater than 2,500,000<br />
Unconsolidated Aquifers<br />
Alluvial aquifers suitable <strong>for</strong> bank infiltration<br />
Sand and gravel<br />
Consolidated Rock Aquifers<br />
Sandstone: includes some unconsolidated sand<br />
Carbonate rock: limestone and dolomite;<br />
and in Texas and Oklahoma, some gypsum<br />
Sandstone and carbonate rocks<br />
Volcanic rocks, chiefly basalt<br />
Crystalline rocks, igneous and metamorphic<br />
Withdrawals from wells
The surficial aquifers in Midwestern states that border the Ohio River and are east of the<br />
Mississippi River are similar to those described previously, but are primarily of glacial origin. These<br />
aquifers are very productive (1,000 gallons per minute) and supply almost 50 percent of the fresh<br />
groundwater produced in the region. The course-grained aquifers are divided into two categories:<br />
• Deposits at or near the land surface occurring in stream and river valleys.<br />
• Deposits buried by a layer of fine-grained material that occur in <strong>for</strong>mer river valleys cut<br />
into bedrock and filled with coarse-grained glacial outwash.<br />
The sands and gravels range in thickness from less than 100 to over 600 ft in some buried bedrock<br />
valleys. Large yields are possible from wells completed in the glacial outwash aquifers that are<br />
hydraulically connected to streams, rivers, and lakes, and the wells are sufficiently close to the surfacewater<br />
body (Lloyd and Lyke, 1995). Communities located near rivers such as the Illinois, Kaskaski,<br />
Wabash, White, Kankakee, Maumee, Great Miami, and Scioto rivers could benefit from <strong>RBF</strong>.<br />
Valley-fill glacial aquifers also occur in the Northeastern United States. While extensive, the<br />
aquifer thickness and productivity is less than those occurring in Midwestern states (Olcott,<br />
1995). In general, the valley-fill aquifers are less common along major streams and rivers, so the<br />
potential <strong>for</strong> <strong>RBF</strong> is considered lower than in the Midwestern United States. Alluvial aquifers<br />
occur along streams and rivers in the Northwestern states, but sparse population and relatively low<br />
water demands can be met through either surface-water intakes or groundwater wells. In the arid<br />
Southwest, few major rivers or streams are perennial, thereby limiting <strong>RBF</strong> potential. <strong>Water</strong><br />
supplies in Gulf Coast states are met through groundwater withdrawals from productive regional<br />
aquifers and surface water. In the Piedmont region, water needs are met through surface water, as<br />
rainfall is usually adequate to meet demands and aquifers are fractured crystalline rock, which are<br />
not considered viable <strong>RBF</strong> areas.<br />
REFERENCES<br />
Eckert, P., C. Blomer, J. Gothardt, S. Kamphausen, D. Liebich, and J. Schubert (2000). “Correlation between<br />
the Well Field Catchment and Transient Flow Conditions.” Proceedings, International Riverbank Filtration<br />
Conference, November 2 - 4, Dusseldorf, Germany.<br />
Lloyd, O.B., and W.L. Lyke (1995). “Ground <strong>Water</strong> Atlas of the United States, Segment 10 - Indiana,<br />
Illinois, Kentucky, Ohio.” U.S. Geological Survey Hydrologic Investigations Atlas 730 -K.<br />
Miller, J.A., and C.L. Appel (1997). “Ground <strong>Water</strong> Atlas of the United States, Segment 3- Kansas,<br />
Missouri, Nebraska.” U.S. Geological Survey Hydrologic Investigations Atlas 730 -D.<br />
Olcott, P.G. (1995). “Ground <strong>Water</strong> Atlas of the United States, Segment 12 - Connecticut, Maine,<br />
Massachusetts, New Hampshire, New York, Rhode Island, Vermont.” U.S. Geological Survey Hydrologic<br />
Investigations Atlas 730 -J.<br />
Olcott, P.G. (1992). “Ground <strong>Water</strong> Atlas of the United States, Segment 12 - Iowa, Michigan, Minnesota,<br />
Wisconsin.” U.S. Geological Survey Hydrologic Investigations Atlas 730 -M.<br />
Schubert, J. (2000). “How Does it Work: Field Studies on Riverbank Filtration.” Proceedings, International<br />
Riverbank Filtration Conference, November 2 - 4, Dusseldorf, Germany.<br />
Todd, D.K. (1980). Groundwater Hydrology, John Wiley & Sons, New York.<br />
United States Geological Survey (1965). Productive Aquifers in the Conterminous United States, United States<br />
Geological Survey.<br />
Wang, J.Z., S.A. Hubbs, and R. Song (2002). Evaluation of Riverbank Filtration as a Drinking <strong>Water</strong> Treatment<br />
Process, American <strong>Water</strong> Works Association <strong>Research</strong> Foundation and American <strong>Water</strong> Works Association.<br />
171
172<br />
LEO GENTILE is Senior Hydrogeologist and Group Manager with Jordan, Jones &<br />
Goulding in Atlanta, Georgia. He has a broad range of experience in applied geology and<br />
hydrogeology in the United States. His background includes over 20-years of experience<br />
in technical support and management on a broad range of projects, including waterresources<br />
development, remedial investigations, and feasibility studies at Comprehensive<br />
Environmental Resource Compensation and Liability Act sites; Resource Conservation<br />
and Recovery Act facility investigation and closure; investigation and remediation of<br />
agricultural chemical sites; and assessments and reclamation <strong>for</strong> mining and petroleum-related sites. Gentile<br />
received a B.S. in Geology/Mineralogy from Ohio State University, an M.S. in Petroleum Geology from<br />
Oklahoma State University, and an MBA in Enterprise Risk Management/Finance from Georgia State<br />
University. He is a Registered Professional Geologist in seven states and Puerto Rico, and is a Certified<br />
Professional Geologist by the American <strong>Institute</strong> of Professional Geologists.
Session 9: Public Policy and Regulatory<br />
Application of the Long Term 2 Enhanced Surface<br />
<strong>Water</strong> Treatment Rule Microbial Toolbox at Existing<br />
<strong>Water</strong> Plants<br />
Richard A. Brown<br />
Environmental Engineering and Technology, Inc.<br />
Newport News, Virginia<br />
All surface-water utilities in the United States will be required to comply with specific Cryptosporidium<br />
removal/inactivation targets in the LT2ESWTR, depending upon their bin classification as a<br />
result of raw-water Cryptosporidium occurrence levels. The “microbial toolbox” is intended to<br />
provide utilities with a range of treatment options <strong>for</strong> meeting LT2ESWTR compliance<br />
requirements.<br />
This presentation will include a general discussion of LT2ESWTR requirements, including two of<br />
its main components:<br />
• Cryptosporidium sampling.<br />
• Treatment credits from the microbial toolbox.<br />
In particular, this will include a discussion of how these requirements relate to systems with <strong>RBF</strong><br />
or other groundwater wells that are GWUDI. LT2ESWTR requirements <strong>for</strong> <strong>RBF</strong> are different <strong>for</strong><br />
systems in place at the time the Rule is published versus well systems installed later. New systems<br />
are eligible <strong>for</strong> treatment credits if they meet USEPA-defined requirements <strong>for</strong> media gradation,<br />
separation distance from the associated surface-water source, and turbidity monitoring. Existing<br />
<strong>RBF</strong> and GWUDI wells are not eligible <strong>for</strong> any direct credits, but will produce indirect credits if<br />
Cryptosporidium bin assignment samples are collected from well effluent. According to the<br />
LT2ESWTR, these Cryptosporidium samples must be collected from well effluent if well water is<br />
sent to a subsequent treatment process be<strong>for</strong>e being delivered to consumers. Public water-supply<br />
GWUDI or <strong>RBF</strong> wells that provide water directly to customers without subsequent physical<br />
treatment must collect Cryptosporidium bin assignment samples from the surface-water source.<br />
Additional benefits of <strong>RBF</strong> (removal or degradation of other contaminants and microorganisms,<br />
dampening of water quality and temperature spikes, etc.), as well as potential issues of concern<br />
(clogging of pores, scour associated with flooding/high stage events), will also be outlined.<br />
Other microbial toolbox alternatives will be outlined, though in less detail, and cost comparisons<br />
<strong>for</strong> different treatment strategies will also be discussed. Strategies <strong>for</strong> source-water sampling and<br />
the preparation of contingency plans will also be addressed.<br />
Correspondence should be addressed to:<br />
Richard A. Brown<br />
Engineer<br />
Environmental Engineering and Technology<br />
712 Gum Rock Court • Newport News, Virginia 23606 USA<br />
Phone: (757) 873-1534 • Fax: (757) 873-2392 • Email: rbrown@EETinc.com<br />
173
174<br />
RICHARD BROWN, an Environmental Engineer at Environmental Engineering and<br />
Technology, Inc. in Newport News, Virginia, has almost 20 years of experience working<br />
on surface-water and groundwater quality, regulatory compliance, and water-treatment<br />
projects. Recently, this has included extensive involvement with the American <strong>Water</strong><br />
Works Association, Association of Metropolitan <strong>Water</strong> Agencies, and various United<br />
States drinking-water utilities during negotiations with the United States Environmental<br />
Protection Agency regarding the Stage 2 Disinfection By-Products Rule and Long Term 2<br />
Enhanced Surface <strong>Water</strong> Treatment Plant. Prior, Brown worked <strong>for</strong> 11 years at the Los Angeles County<br />
Sanitation Districts, per<strong>for</strong>ming and directing numerous projects involving the investigation and<br />
interpretation of groundwater chemistry at six municipal solid waste landfills. The American <strong>Water</strong> Works<br />
Association <strong>Research</strong> Foundation study that is the subject of Brown’s presentation during the Second<br />
International Riverbank Filtration Conference will be published in the American <strong>Water</strong> Works Association’s<br />
e-Journal in September 2003. Brown received a B.S. in Civil Engineering from Purdue University and an<br />
M.S. in Environmental Engineering from the University of North Carolina at Chapel Hill.
Session 9: Public Policy and Regulatory<br />
Draft Protocol <strong>for</strong> the Demonstration of<br />
Effective Riverbank Filtration<br />
William D. Gollnitz<br />
Greater Cincinnati <strong>Water</strong> Works<br />
Cincinnati, Ohio<br />
Verna J. Arnette, P.E.<br />
Greater Cincinnati <strong>Water</strong> Works<br />
Cincinnati, Ohio<br />
Bruce L. Whitteberry<br />
Greater Cincinnati <strong>Water</strong> Works<br />
Cincinnati, Ohio<br />
Introduction<br />
<strong>RBF</strong> is a natural in situ filtration process by which microbial pathogens in surface water are<br />
removed via the porous media of the streambed and aquifer under induced infiltration conditions<br />
created by a pumping well. With the promulgation of the LT2ESWTR, the USEPA will allow<br />
0.5- and 1.0-log treatment credit <strong>for</strong> Cryptosporidium removal using <strong>RBF</strong> (USEPA, 2003).<br />
Based upon a review of data from several <strong>RBF</strong> systems in the United States, it is not uncommon<br />
<strong>for</strong> <strong>RBF</strong> to achieve a removal per<strong>for</strong>mance better than the proposed 1.0-log reduction credit. The<br />
0.5- and 1.0-log credits are based on general conservative requirements obtainable by most <strong>RBF</strong><br />
systems; however, there are still several concerns with the effectiveness of <strong>RBF</strong> during worst-case<br />
hydrologic and water-quality conditions. For example, high flow events may significantly scour a<br />
streambed, thus reducing its filtration capabilities. To date, there has been a reluctance to provide<br />
additional credit because an acceptable procedure has not been developed that will allow <strong>for</strong> the<br />
demonstration of consistent removal during these periods of high risk.<br />
The Greater Cincinnati <strong>Water</strong> Works has developed the following draft protocol <strong>for</strong> demonstrating<br />
<strong>RBF</strong> pathogen removal. The draft protocol will:<br />
• Allow utilities a method arguing <strong>for</strong> the re-designation of a groundwater from GWUDI<br />
back to “groundwater.”<br />
• Allow <strong>for</strong> demonstration of 2.0-log Cryptosporidium removal in lieu of engineered filtration<br />
under the Interim Enhanced Surface <strong>Water</strong> Treatment Rule.<br />
• Allow <strong>for</strong> greater than 1.0-log treatment credit under LT2ESWTR.<br />
• Provide consistency <strong>for</strong> evaluating <strong>RBF</strong> sites.<br />
The draft protocol is based upon a review of several regulatory guidance publications dealing with<br />
GWUDI evaluations and “alternative filtration technology”(USEPA, 1991; USEPA, 1992;<br />
Correspondence should be addressed to:<br />
William D. Gollnitz<br />
Supervisor of Treatment, <strong>Water</strong> Quality & Treatment Division<br />
Greater Cincinnati <strong>Water</strong> Works<br />
5651 Kellogg Ave • Cincinnati, Ohio 45228 USA<br />
Phone: (513) 624-5657 • Fax: (513) 624-5670 • Email: William.Gollnitz@gcww.cincinnati-oh.gov<br />
175
176<br />
Wilson et al., 1996; USEPA, 2001). In addition, the Greater Cincinnati <strong>Water</strong> Works has<br />
reviewed data from an extensive flowpath study conducted in cooperation with the USGS at the<br />
Greater Cincinnati <strong>Water</strong> Works’ Charles M. Bolton Well Field (Sheets et al., 2002; Gollnitz et<br />
al., 2003), as well as published data from other <strong>RBF</strong> sites such as Louisville, Kentucky (Wang et<br />
al., 2002); Casper, Wyoming (Gollnitz et al., 1997); and others (Cote et al., 2002). The protocol<br />
is in response, in part, to requests <strong>for</strong> comments on additional <strong>RBF</strong> credit in the LT2ESWTR. It<br />
is hoped that the protocol will provide initial guidance to the USEPA, primacy agencies, and<br />
water utilities. The draft <strong>for</strong>mat of the protocol will facilitate comment and modification by<br />
regulatory and utility personnel.<br />
The <strong>RBF</strong> process is very complex and is still not fully understood; however, our knowledge to date<br />
does pinpoint several controlling factors that are primary with respect to the removal of microbial<br />
pathogens and surrogates. These factors are identified in Cote et al. (2002) and include:<br />
• Assessment of river and groundwater quality.<br />
• Flow velocity through the streambed and aquifer.<br />
• Dilution with regional groundwater.<br />
The <strong>RBF</strong> protocol has, there<strong>for</strong>e, been designed around:<br />
• Characterizing the aquifer and its capability as a <strong>RBF</strong> system.<br />
• Identifying conditions <strong>for</strong> testing during periods of high flow velocity.<br />
• Testing during periods of minimal dilution with regional groundwater.<br />
The latter two items involve identifying periods of maximum induced filtration during high stage<br />
events (as related to storm or reservoir release events) and maximum groundwater pumping. Theoretically,<br />
these two factors would correspond with periods of high velocity and minimum dilution.<br />
Pre-Demonstration Evaluation<br />
The purpose of the pre-demonstration evaluation is to characterize the aquifer and collection<br />
devices with respect to hydrology and water quality. This includes a determination of the worst-case<br />
conditions <strong>for</strong> minimal natural filtration of pathogenic protozoa. This evaluation will then identify<br />
the type of demonstration project needed <strong>for</strong> determining treatment credit. The pre-demonstration<br />
evaluation will organize and evaluate existing data and will identify what additional data is needed<br />
to complete this first phase. The pre-demonstration project will specifically look at the following<br />
important aspects of the natural filtration process:<br />
1. Surface-water and groundwater quality.<br />
• Filtration capability.<br />
• Contaminant and surrogate concentrations.<br />
• Continuous monitoring of turbidity.<br />
2. Potential infiltration rate variability.<br />
3. Time of travel between surface-water and groundwater collection devices.<br />
The first aspect of Item 1 is to characterize source-water quality with respect to its filtration<br />
capability. It is recommended that the utility per<strong>for</strong>m analyses to determine the ionic strength of<br />
both surface water and groundwater. The ionic strength may provide an indication of the filtration<br />
capability of the sediments. Cote et al. (2002) and others have suggested that a high ionic strength<br />
improves the removal of particles; however, its role is not yet fully understood. It should be noted
that multiple samples should be evaluated. Ionic strength may vary between low (primarily<br />
groundwater inflow) and high (primarily runoff) river flow conditions.<br />
The second aspect of Item 1 is to determine the level of contaminant (Giardia, Cryptosporidium)<br />
and surrogate (algae, diatoms, spores, particle counts) concentrations and their seasonal<br />
variability. If this data does not already exist, it is recommended that the utility per<strong>for</strong>m monthly<br />
Giardia/ Cryptosporidium monitoring using Method 1623 (the same as required under<br />
LT2ESWTR), monthly Microscopic Particulate Analysis using the USEPA Consensus Method<br />
(1992), and weekly endospore analysis (Rice, 1996). Another optional surrogate parameter is<br />
particle counts. Particle counts should be in the two size ranges <strong>for</strong> Giardia (7 to 10 micrometers)<br />
and Cryptosporidium (3 to 5 micrometers). Data collection <strong>for</strong> all parameters selected should be<br />
representative of one full hydrologic cycle (e.g., 1 year). The concentrations of pathogens,<br />
primarily Cryptosporidium, will determine the level of treatment needed. Because Giardia and<br />
Cryptosporidium concentrations are typically low in surface waters, combined with detection<br />
limitations during sample analysis, they cannot be used to determine log-reduction credit. Log<br />
reductions should be based upon levels of algae, diatoms (from Microscopic Particulate Analysis),<br />
endospores, and particle counts; however, prior to per<strong>for</strong>ming the demonstration project,<br />
measurable concentrations need to be established. Particle counts may be used; however, they<br />
should be considered secondary due to the fact that they detect sediment or natural microbial<br />
growth (e.g., iron bacteria) in groundwater from wells. Typically, log reductions using particle<br />
counts are lower as compared to algae and spores due to the measurement of other inert and<br />
microbial particles.<br />
The third aspect of Item 1 is the continuous monitoring of turbidity. Turbidity monitoring is a<br />
requirement <strong>for</strong> engineered filtration systems, and has been included as a requirement on each<br />
collection device <strong>for</strong> <strong>RBF</strong> credit under LT2ESWTR <strong>for</strong> the period that the device is in use.<br />
Groundwater from <strong>RBF</strong> sites should be consistently low (0.01 to 1.0 ntu). Significant excursions<br />
above the normal level, especially after high infiltration periods, should be investigated. As with<br />
particle counts, turbidity also measures inert geologic material and other non-pathogenic microbials.<br />
As part of the pre-demonstration phase, it is important to estimate the range of potential unit<br />
infiltration rates out in the active area of the streambed, articularly to estimate the highest value<br />
under conditions of high-river stage and high groundwater pumping during periods when river<br />
water is warm. Typically, infiltration rates at <strong>RBF</strong> sites are lower than slow sand filters, but should<br />
be investigated to see if they are higher. Infiltration rates can be calculated using Darcy’s Law, with<br />
input variables being river stage, groundwater elevations below the river stage, streambed permeability,<br />
streambed thickness, and surface-water temperature (affecting viscosity) (Gollnitz et al.,<br />
2002). River-stage data may be obtained from a local United States geological gaging station. If<br />
this data is not available, the maximum stage elevation may be estimated from the depth of the<br />
river channel. The groundwater elevation should be measured below the active infiltration area<br />
of the streambed using piezometers in the streambed or monitoring wells on each side of the river.<br />
Streambed permeability can be estimated using pump test analysis (preferred, but costly), seepage<br />
meters, temperature probes, and piezometers (each method has inherent problems and should be<br />
considered carefully be<strong>for</strong>e selection). Streambed thickness can be estimated by visually<br />
inspecting sediments from shallow excavations of the streambed during low flow. Temperature can<br />
be measured directly in surface water. Surface-water temperature typically reflects the average<br />
daily ambient air temperature. If this data is not available, it may be necessary to collect it during<br />
the pre-demonstration period. Another important aspect is to look at the frequency of the<br />
occurrence of high infiltration events. Frequency analysis requires long-term river stage data,<br />
which may be limited <strong>for</strong> most <strong>RBF</strong> sites.<br />
177
178<br />
<strong>Water</strong>-quality monitoring during the demonstration project should consider the travel time<br />
between the river and collection device. Travel time should be calculated <strong>for</strong> the shortest flowpath<br />
(e.g., fastest potential time of travel) during continuous pumping of the collection device. Time of<br />
travel can be estimated using various techniques ranging from tracer studies (conductivity,<br />
temperature, and chloride), analytical flow calculations, and groundwater flow modeling. Each<br />
method has inherited limitations that must be considered. It is recommended that the utility<br />
determine a sampling window based upon a realistic range of travel times using two or more methods.<br />
Another important consideration is the travel time of particles. Larger particles such as algae may<br />
have longer travel times due to retardation in the aquifer. Sampling windows should be long<br />
enough to take these into consideration. For example, Table 1 lists various travel-time estimates<br />
<strong>for</strong> Production Well 1 at the Charles M. Bolton Well Field, including limitations. The fastest<br />
theoretical time is 1 day; however, a more accurate estimate using conductivity indicates that the<br />
travel time of groundwater ranges from 10 to 14 days (Sheets et al., 2002). Interestingly, algae<br />
appear to arrive at the production well much later (approximately 38 days). In this case, a high<br />
infiltration-event monitoring sampling window may be from 1 to 60 days after the event peak. It<br />
is recommended that 10 to 15 samples be collected during this time period, with increased<br />
sampling frequency around the anticipated period <strong>for</strong> possible breakthrough of microbial surrogates.<br />
Table 1. Time of Travel Estimates between the Great Miami River and Production Well 1<br />
at the Charles M. Bolton Well Field<br />
TOT<br />
Methodology (Days) Comments<br />
Calculated Fixed Radius 1 Short TOT; considers only horizontal flowpath;<br />
no gradient<br />
Groundwater Flow Model 8 Limitations with model design and scale<br />
Conductivity 10 to 14 Real time measurement; range based upon multiple<br />
measurements; considered most accurate method<br />
Event Temperature Lag 26 Real-time measurement; problems due to<br />
thermodynamics<br />
Algae Event Peak 11 to 38 Limited data; considers flow characteristics of algae<br />
TOT = Time of travel. CMB = Charles M. Bolton Well Field.<br />
Demonstration Evaluation<br />
The demonstration evaluation is broken down into three options: 1-year monitoring, multi-event<br />
monitoring, or seasonal monitoring. One of these options is selected based upon the results of the<br />
pre-demonstration evaluation. Log reduction <strong>for</strong> all three options should be based upon the<br />
average of Microscopic Particulate Analysis and endospore data.<br />
The 1-year monitoring option is <strong>for</strong> systems with little or limited infiltration rate and water-quality data.<br />
It is recommended that utilities monitor river stage, groundwater elevations, and surface-water<br />
temperature on a daily basis. Daily calculations can be made with Darcy’s law using conservative<br />
values of permeability (e.g., highest measured value) and thickness (lowest measured value).<br />
These calculations can be easily made using a computer spreadsheet (Gollnitz et al., 2002). With<br />
respect to water quality, it is recommended that utilities collect weekly samples of turbidity and<br />
endospores from surface water, along with monthly Giardia, Cryptosporidium, and Microscopic<br />
Particulate Analysis. Each collection device being evaluated should be continuously monitored <strong>for</strong>
turbidity. Endospores, Giardia, Cryptosporidium, and Microscopic Particulate Analysis data should<br />
be sampled at each collection device on a period and frequency determined from the time-of-travel<br />
data. During the 1-year period, an attempt should be made to identify conditions <strong>for</strong> high<br />
infiltration and to monitor one or more events, if possible.<br />
The multi-event monitoring option is provided <strong>for</strong> systems with infiltration rates that are<br />
significantly high, as compared to engineered filtration systems, during storm events, upstream<br />
reservoir releases, and during periods of heavy groundwater pumping. <strong>Water</strong>-quality monitoring<br />
should include multiple turbidity, Giardia, Cryptosporidium, Microscopic Particulate Analysis, and<br />
endospore data collected from the river during and shortly after the river stage peak. As with all<br />
options, groundwater turbidity should be continuously monitored. Groundwater samples <strong>for</strong> the<br />
other parameters should be collected within a defined sampling window using the time-of-travel<br />
data. Endospores and particle counts should be collected more frequently due to their lower cost.<br />
A major advantage to this option is that monitoring resources can be concentrated during these<br />
periods, which in turn may reduce overall costs. Infiltration rate parameters should be monitored<br />
to determine the magnitude of the event.<br />
Option three is seasonal monitoring. Pre-demonstration evaluations may identify if a system is at<br />
risk during a 2- to 4-month period of the year. For example, a controlled river in the arid West<br />
may have a high stage only during the summer months, when water is released from reservoirs <strong>for</strong><br />
irrigation. <strong>Water</strong>-quality monitoring should be concentrated during this high-risk period, taking<br />
time of travel into consideration, as well as estimating the magnitude of infiltration.<br />
REFERENCES<br />
Cote, M.M., M.B. Emelko, and N.R.Thomson (2002). “Factors Influencing Prediction of Cryptosporidium<br />
Removal in Riverbank Filtration Systems: Focus on Filtration.” Proceedings, American <strong>Water</strong> Works Association<br />
WQTC Conference, Denver, CO.<br />
Gollnitz, W.D., J.L. Clancy, and S.C. Garner (1997). “Reduction of Microscopic Particulates by Aquifers.”<br />
Journal AWWA, 89(11).<br />
Gollnitz, W.D., J.L. Clancy, B.L. Whitteberry, and J.A. Vogt (2003). “Riverbank Filtration as a Microbial<br />
Treatment Process.” Journal AWWA (in press).<br />
Gollnitz, W.D., B.L. Whitteberry, and J.A. Vogt (2002). “Induced Infiltration Rate Variability and <strong>Water</strong><br />
Quality.” Proceedings, SW-GW Interactions Conference, American <strong>Water</strong> Resource Association, Keystone, CO.<br />
Rice, E.W., K.R. Fox, R.J. Miltner, D.A. Lytle, and C.H. Johnson (1996). “Evaluating Plant Per<strong>for</strong>mance<br />
with Endospores.” Journal AWWA, Denver, CO.<br />
Sheets, R.A., R.A. Darner, and B.L. Whitteberry (2002). “Lag Times of Bank Filtration at a Well Field,<br />
Cincinnati, Ohio, USA.” Journal of Hydrology, 233: 162-174.<br />
USEPA (1991). Guidance Manual <strong>for</strong> Compliance with the Filtration and Disinfection Requirements <strong>for</strong> Public<br />
<strong>Water</strong> Systems using Surface <strong>Water</strong> Sources, American <strong>Water</strong> Works Association, Denver, CO.<br />
USEPA (1992). Consensus Method <strong>for</strong> Determining Groundwaters under the Direct Influence of Surface <strong>Water</strong><br />
using Microscopic Particulate Analysis (MPA), Port Orchard, WA.<br />
USEPA (2001). Draft, Considerations <strong>for</strong> Evaluating the Effectiveness of Alternative Filtration at Central Wyoming<br />
Regional <strong>Water</strong> System (CWRWS), Region 8, Denver, CO (unpublished letter to CWRWS).<br />
USEPA (2003). <strong>National</strong> Primary Drinking <strong>Water</strong> Regulations: Long Term 2 Enhanced Surface <strong>Water</strong> Treatment<br />
Rule, 40 CFR 141 and 142, Washington, D.C.<br />
Wang, J.Z., S.A. Hubbs, and R. Song (2002). Evaluation of Riverbank Filtration as a Drinking <strong>Water</strong> Treatment<br />
Process, Tailored Collaboration with Louisville <strong>Water</strong> Co., American <strong>Water</strong> Works Association <strong>Research</strong><br />
Foundation, Denver, CO.<br />
Wilson, M.P., W.D. Gollnitz, S.N. Boutros, and W.T. Boria (1996). Determining Ground <strong>Water</strong> under the Direct<br />
Influence of Surface <strong>Water</strong>, American <strong>Water</strong> Works Association <strong>Research</strong> Foundation Project 605, Denver, CO.<br />
179
180<br />
BILL GOLLNITZ is a Supervisor of Treatment <strong>for</strong> the Greater Cincinnati <strong>Water</strong> Works,<br />
where he is responsible <strong>for</strong> water quality and treatment <strong>for</strong> both the Charles M. Bolton and<br />
Mason Ground <strong>Water</strong> Facilities. Gollnitz has been in the water supply industry <strong>for</strong> over<br />
30 years, having previously managed surface-water and groundwater utilities in New York<br />
and Rhode Island. He has also had extensive experience in water-supply protection and<br />
surface-water/groundwater interactions. In addition, he was a Project Manager and co-author<br />
on the American <strong>Water</strong> Works Association <strong>Research</strong> Foundation’s project “Determining<br />
Ground <strong>Water</strong> under the Direct Influence of Surface <strong>Water</strong>.” Gollnitz has completed groundwater under the<br />
direct influence of surface water evaluations in Connecticut, New York, Ohio, Rhode Island, and Wyoming,<br />
and he has been published in the Journal American <strong>Water</strong> Works Association and elsewhere. In 1992, he was<br />
honored with an American <strong>Water</strong> Works Association Best Paper Award. Gollnitz received a B.S. in Biology<br />
from Mount Union College in Alliance, Ohio, and a M.S. in Environmental Science-<strong>Water</strong> Resources from<br />
the State University of New York, College of Environmental Science and Forestry in Syracuse, New York.
Session 9: Public Policy and Registration<br />
Source <strong>Water</strong> Protection and Riverbank Filtration<br />
in the Dyje River Basin<br />
Prof.-Dr. Petr Hlavínek<br />
Brno University of Technology<br />
Brno, Czech Republic<br />
Prof.-Dr. Jaroslav Hlavác˘<br />
Vodárenská Akciová Spolec˘nost a.s.<br />
Brno, Czech Republic<br />
<strong>Water</strong> is not a commercial product like any other but, rather, a heritage that must be protected,<br />
defended, and treated as such. Achieving regional water-quality goals often involves substantial<br />
capital investments and changes in public attitudes concerning resource management. Economic<br />
impacts include:<br />
• The cost of facilities designed to reduce the discharge of contaminants into natural waters<br />
or to improve the quality of waste-receiving waters.<br />
• Limitations on economic activities and economic development in a particular region or<br />
river basin.<br />
Those responsible <strong>for</strong> the <strong>for</strong>mulation and adoption of water-quality plans and management<br />
policies must have a means of estimating and evaluating the temporal and spatial economics,<br />
environmental, and ecologic impacts of these plans and policies. This need has stimulated the<br />
development and application of a wide range of mathematical modeling techniques <strong>for</strong> predicting<br />
the impact of alternative pollution control plans.<br />
The Dyje River is one of the biggest boundary streams in the Czech Republic. It is located in<br />
southern Moravia, crosses the state boundary between Austria and the Czech Republic several<br />
times, and creates the state boundary (with a total length of approximately 50 kilometers). There<br />
are 86 urban areas (agglomerations) larger than 2,000 population equivalent, with a total of more<br />
than 1.2-million inhabitants in the area of interest. The area of the Dyje River Basin is<br />
approximately 12,000 square kilometers, and the monitored river network is approximately<br />
1,000-kilometers long (Figure 1). There is number of <strong>RBF</strong> plants along the Dyje River basin that<br />
are, in the long run, affected by river-water quality.<br />
The Dyje Project deals with the areas that do not yet meet European Union legislation and<br />
correspond with the Instrument <strong>for</strong> Structural Policies <strong>for</strong> Pre-Accession (ISPA) definition of<br />
agglomerations.<br />
Correspondence should be addressed to:<br />
Prof.-Dr. Petr Hlavínek<br />
Professor of Civil Engineering, Faculty of Civil Engineering<br />
Brno University of Technology<br />
Zizkova 17 • 602 00 Brno, Czech Republic<br />
Phone: 420-541147733 • Fax: 420-543245032 • Email: hlavinek.p@fce.vutbr.cz<br />
181
182<br />
Figure 1. Location of the Dyje River Basin in the Cezch Republic. A= Austria. SK = Slovakia.<br />
PL = Polland. D = Germany.<br />
The goal of the project is to ensure that:<br />
• The quality and volume of water collected from polluters and treated in wastewater<br />
treatment plants complies by the year 2005 and is within the area managed by Union<br />
water supply and sewer systems with the values set down by Directive No. 91/271/EEC.<br />
• All wastewater discharged to the sewer is collected.<br />
• Within any drained area serviced by a wastewater treatment plant, all wastewater is<br />
carried to that wastewater treatment plant and treated in compliance with the values set<br />
down by Directive No. 91/271/EEC.<br />
• <strong>National</strong> parks and landscape-protected areas within the region are protected in terms of<br />
surface-water and groundwater quality.<br />
A group of related projects covers the following priority measures, as classified below:<br />
Category A.1: Reconstruct and upgrade existing wastewater treatment plants in municipalities<br />
with populations over 2,000 (or equivalent).<br />
Category A.2: Refurbish existing large wastewater treatment plants (<strong>for</strong> populations over 10,000<br />
[or equivalent]), including equipment <strong>for</strong> the removal of nitrogen compounds and phosphorus.<br />
Category A.3: Reconstruct existing sewer systems connected to existing wastewater treatment<br />
plants to provide sufficient capacity and a high level of treatment.<br />
Category A.4: Complete sewer systems and connect them to existing wastewater treatment plants<br />
to provide sufficient capacity and a high level of treatment.<br />
Category A.5: Construct wastewater treatment plants in municipalities with populations over<br />
2,000 (or equivalent).<br />
Category A.6: Complete sewer systems and wastewater treatment plants in municipalities with<br />
populations over 2,000 (or equivalent).
For the first phase of the Dyje Project, 10 agglomerations were chosen based on area investigations<br />
and comparisons with European Union standards. A technical solution was designed <strong>for</strong> those<br />
locations. Wastewater treatment and collection needs were identified and their standards<br />
compared with European Union standards. See Figure 2 and Table 1.<br />
1. Dyje – Moravská Dyje<br />
2. Jihlava – lower part<br />
3. Jihlava – upper part<br />
4. Oslava<br />
5. Bobrava<br />
6. Svratka – lower part<br />
7. Svratka – upper part<br />
8. Litava<br />
Figure 2. Sub-catchments 1 through 8 of the Dyje River.<br />
Table 1. Basic Data of the Dyje River Catchment Area<br />
Increase in Connected Inhabitants PE 203,606<br />
Increase in Removed Pollution PE 154,905<br />
Total Capacity of New and Upgraded WWTP PE 350,133<br />
Increase in Removed Pollution – SS ton/year 3,110<br />
Increase in Removed Pollution – BOD 5 ton/year 3,392<br />
Increase in Removed Pollution – COD ton/year 6,785<br />
Increase in Removed Pollution – N-NH4 ton/year 622<br />
Increase in Removed Pollution – Total ton/year 141<br />
Increase in Treated Wastewater m 3 /year 9,103,587<br />
Wastewater Treated on WWTPs m 3 /year 20,975,687<br />
Total Length of New and Upgraded WWTP kilometers 601<br />
WWTP = Wastewater treatment plant. SS= Suspended solid. BOD 5 = Biochemical oxygen demand.<br />
COD = Chemical oxygen demand. N-NH4 = Ammonia nitrogen. PE = Population equivalent.<br />
183
184<br />
As a result of improvements in pollution source-management at the Dyje River Basin, mathematical<br />
modeling techniques were used to assess stream-water quality. The study aimed to evaluate<br />
the effect of improvements on stream-water quality at main streams in the Dyje River basin, with<br />
special respect to the Dyje River transboundary profile (downstream of the City of Br˘eclav). The<br />
results were used as bases in the decision-making process. By doing this, financial sources can be<br />
distributed more efficiently in the basin and the application of remediation measures would better<br />
improve water quality while considering the whole spectrum of requirements connected to the<br />
process of approaching the European Union.<br />
There are a number of <strong>RBF</strong> plants along the Dyje River Basin. After decades of safe operation,<br />
some came under heavy pressure due to increasing river-water pollution. As an example, the <strong>RBF</strong><br />
plant at Vojkovice is discussed. See Figures 3 and 4.<br />
Concentration mg/L<br />
2.5<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
0<br />
1979<br />
1981<br />
1983<br />
Average<br />
1985<br />
1987<br />
1989<br />
Date<br />
Maximum<br />
Figure 3. Maximum and average concentration of N-NH4 at the <strong>RBF</strong> plant at Vojkovice from 1979 to 2000.<br />
In 1967, 170,000 tons of biochemical oxygen demand (BOD 5) per year were discharged into rivers<br />
in the Czech Republic. From 1957 to 1970, 800 new wastewater treatment plants were built, and the<br />
load decreased to 142,000 tons of BOD 5 in 1975. Between 1970 and 1980, the construction of<br />
wastewater treatment plants was stifled; there<strong>for</strong>e, the load increased to 198,000 tons of BOD 5 in<br />
1980. In 1990, the load decreased to 148,000 tons of BOD 5. Due to the massive construction of<br />
wastewater treatment plants after 1990, the load decreased to 67,000 tons of BOD 5 in 1995 and<br />
to 65,000 tons of BOD 5 in 2000.<br />
Gradual improvements in quality are expected after the construction and completion of the<br />
project, “Protection of <strong>Water</strong> in the Dyje River Basin.” It is expected that a number of <strong>RBF</strong> plants<br />
will be immediately affected (Ivancice, Moravské Bránice), while others will be gradually affected<br />
(Lomnicka, Omice, Vojkovice).<br />
1991<br />
1993<br />
1995<br />
1997<br />
1999
Concentration mg.l 1<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
1979<br />
1981<br />
1983<br />
1985<br />
Average<br />
1987<br />
1989<br />
Date<br />
Maximum<br />
Figure 4. Maximum and average concentration of N-NO 3 in the <strong>RBF</strong> plant at Vojkovice from 1979 to 2000.<br />
The simulation models are a relatively crude approximation of interactions among various<br />
constituents that occur in water bodies. The best water-quality simulation model is the simplest<br />
one that will adequately predict water-quality impacts within a particular water body associated<br />
with a particular water-quality management policy. Yet, in spite of current limitations, simulation<br />
models are the only reasonable means available <strong>for</strong> predicting surface-water quality. The state-ofthe-art<br />
in water-quality modeling and the understanding of physical, chemical, and biological<br />
processes that affect water quality are improving rapidly. The model provides preliminary<br />
estimates; a more precise tool should be used after collecting all necessary data (i.e., especially<br />
more accurate pollution data, corresponding real discharge data at the time of sampling, more<br />
accurate flow hydrodynamics calculation, etc.). The activities mentioned are quite time-consuming<br />
and expensive, yet necessary when improvements to the receiving waters must be more precisely<br />
assessed and quantified.<br />
Regional projects <strong>for</strong> water protection are complex, complicated, multidisciplinary, and long-term.<br />
They are politically important, with complicated relations among authorities both at the local and<br />
international levels. Close cooperation of all involved parties and companies is absolutely<br />
necessary. The Dyje Project was established as a pilot project <strong>for</strong> a regional solution in the Czech<br />
Republic. The project, “<strong>Water</strong> Protection of the River Dyje Basin,” has been accepted <strong>for</strong><br />
financing from the ISPA fund.<br />
1991<br />
1993<br />
1995<br />
1997<br />
1999<br />
185
186<br />
PETR HLAVÍNEK has 20 years experience in wastewater treatment and water quality. He<br />
has 10 years of experience as a lead designer in the design company, HYDROPROJECT.<br />
At present, he is Professor of Civil Engineering at Brno University of Technology, Faculty<br />
of Civil Engineering. He also acts as Vice-President of the Czech Wastewater Treatment<br />
Experts Association, as well as Deputy Head of the <strong>Institute</strong> of Urban <strong>Water</strong> Management<br />
and as Member of the governing board of the Faculty of Civil Engineering at Brno<br />
University of Technology. Hlavínek has completed more than 200 research reports, expert<br />
opinions, and reports dealing with water quality and wastewater treatment. He has published more than 120<br />
articles dealing with wastewater treatment and water quality, and is author and co-author of six books,<br />
including Industrial Wastewater Treatment, Upgrading of Wastewater Treatment Plants, Hydraulics of WWTP,<br />
Wastewater Treatment-Examples of Calculation, Drainage of Urban Areas — Conceptual Approach and <strong>Water</strong><br />
Structures, and <strong>Water</strong> Structures. Hlavínek received an M.S. at Brno University of Technology, a Dipl. in<br />
Sanitary Engineering at IHE Delft Netherlands, and a Ph.D. at Brno University of Technology.
Session 11: Case Studies “Lessons Learned”<br />
Greater Cincinnati <strong>Water</strong> Works Flowpath Study<br />
Field Design: Methodology and Evaluation<br />
Bruce Whitteberry, P.G.<br />
Greater Cincinnati <strong>Water</strong> Works<br />
Cincinnati, Ohio<br />
William D. Gollnitz<br />
Greater Cincinnati <strong>Water</strong> Works<br />
Cincinnati, Ohio<br />
Jeffrey Vogt<br />
Greater Cincinnati <strong>Water</strong> Works<br />
Cincinnati, Ohio<br />
Objective<br />
The Greater Cincinnati <strong>Water</strong> Works, USGS, and Miami University of Ox<strong>for</strong>d, Ohio, entered<br />
into a joint research study to evaluate the effectiveness of <strong>RBF</strong> on an unconsolidated sand and<br />
gravel aquifer at the Greater Cincinnati <strong>Water</strong> Works’ Charles M. Bolton Well Field. One of the<br />
goals of this study was to better understand the aquifer’s effectiveness at reducing biological<br />
drinking-water contaminants. The purpose of this presentation is to describe and evaluate the<br />
study’s field design to determine its effectiveness <strong>for</strong> achieving study goals. This presentation will<br />
also discuss water-quality anomalies in the observed data and how they relate to field design.<br />
Methodology<br />
The Charles M. Bolton Well Field is located within the Great Miami Buried Valley Aquifer, which is<br />
composed of sand and gravel. The well field consists of 10 production wells located 50 to 600 ft<br />
from the Great Miami River. The well field is situated along the southern edge of the Great Miami<br />
Buried Valley Aquifer and is bounded by the bedrock valley wall to the south and the Great Miami<br />
River to the north. Due to this configuration, the well field is dependent upon the river as a major<br />
source of induced recharge or infiltration.<br />
Two study sites were chosen within the well field. The sites are adjacent to Production Well 1<br />
(referred to as Site 1) and to Production Well 8 (referred to as Site 8). At each site, four vertical<br />
monitoring wells with 2-ft long well screens were installed along a theoretical flowpath between<br />
the river and production well. Each well was installed progressively deeper, moving away from the<br />
river and toward the well. The two deepest wells were installed to correspond to the top and<br />
bottom of the production well screen. In addition, an inclined well was installed at each site. The<br />
inclined wells were installed at 20- to 30-degree angles from the horizontal and drilled beneath<br />
Correspondence should be addressed to:<br />
Bruce Whitteberry, P.G.<br />
Hydrogeologist<br />
Greater Cincinnati <strong>Water</strong> Works<br />
5651 Kellogg Ave • Cincinnati, Ohio 45228 USA<br />
Phone: (513) 624-5611 • Fax: (513) 624-5670 • Email: Bruce.Whitteberry@gcww.cincinnati-oh.gov<br />
187
188<br />
the river. This allowed water-sample collection from approximately 5- to 10-ft beneath the Great<br />
Miami River.<br />
Each of the vertical flowpath wells was constructed using 4-inch diameter polyvinyl chloride wells<br />
with 2 ft of 10-slot (0.010-inch) screen. The lengths of the screens were designed to provide the<br />
shortest interval necessary to accommodate a sample pump and downhole data.<br />
Once the wells were installed and developed (to clean the well of construction-induced silt), a<br />
data sonde was installed in each well. The data sonde was used to record temperature, conductivity,<br />
pH, dissolved oxygen, and water levels. Each well was fitted with a bladder pump to collect<br />
water samples. By using dedicated equipment <strong>for</strong> sampling, the possibility of introducing contamination<br />
into the well was minimized, sampling was more consistent and efficient, and eliminating<br />
equipment blank samples from the quality control program reduced analytical costs. The inclined<br />
wells were constructed similarly to vertical wells, but with a 6-inch diameter polyvinyl chloride<br />
well casing and 5 ft of 10-slot (0.010-inch) well screen. A submersible sampling pump and data<br />
sonde were installed in each inclined well.<br />
To monitor the river, USGS installed a stream gage station at Site 1. The gage continuously<br />
measured and recorded river stage. It was also fitted with an automated sampling pump and data<br />
sonde to collect readings of the same parameters measured in the wells.<br />
Prior to each sampling event, the monitoring wells were purged using low-flow (minimal<br />
drawdown) purging methodology. Each well was pumped at a rate of less than 1 liter per minute<br />
(1 to 3 liters per minute <strong>for</strong> the inclined wells), while minimizing drawdown in the wells (usually<br />
less than 0.02 ft). Temperature, pH, specific conductance, and dissolved oxygen were measured<br />
during purging, and the well was sampled when the parameters stabilized according to USGS<br />
<strong>National</strong> <strong>Water</strong>-Quality Assessment Program Sampling Protocol (Koterba et al., 1995). With<br />
low-flow techniques, sampling-induced turbidity problems can sometimes be minimized. Because<br />
the goal of this study was to sample discrete portions of the aquifer in close proximity, this<br />
technique was also desirable to minimize the mixing from portions of the aquifer above or below<br />
the screened area.<br />
Anomalous Data<br />
Because the purpose of this study was to evaluate the effectiveness of the aquifer in acting as a<br />
filtration system, significant anomalies in the data were of concern because of their potential to<br />
represent a preferred flowpath through which water is not effectively filtered. One anomaly<br />
observed was the concentrations of TOC in Well FP8B. Because this well is of intermediate depth,<br />
the TOC concentration was expected to be between the concentrations of shallower (FP8A) and<br />
deeper (FP8C) wells. This is the trend seen at Site 1; however, during the first 2 months of the<br />
study, FP8B had TOC concentrations several times higher than the river (Figure 1). Throughout<br />
the first 2 months of monitoring, TOC dropped steadily and stabilized, with concentrations<br />
between those of Wells FP8A and FP8C.<br />
Another anomaly identified in the study was high particle counts in Wells FP8B and FP8D. Well<br />
FP8B particle counts were lower than the river, but higher than FP8A (a shallower well closer to<br />
the river). Particle counts in Well FP8D, the deepest monitoring well, were higher or very near<br />
the levels of particles in the river (Figure 2).
Concentration (mg/L)<br />
35<br />
30<br />
25<br />
20<br />
15<br />
10<br />
5<br />
0<br />
Aug-99 Nov-99 Feb-00 May-00 Sep-00 Dec-00 Mar-01 Jul-01<br />
Figure 1. Total organic carbon at Site 8.<br />
Counts/100 ml<br />
10000000<br />
1000000<br />
100000<br />
10000<br />
1000<br />
100<br />
Discussion<br />
Production Well 8 FP8A FP8B FP8C River<br />
Date<br />
Production Well 8 FP8A FP8B FP8D River<br />
10<br />
Aug-99 Dec-99 Mar-00 Jun-00 Oct-00 Jan-01 Apr-01 Jul-01<br />
Date<br />
Figure 2. Particle counts (3 to 5 micrometers in size) at Site 8.<br />
In general, the field design was appropriate <strong>for</strong> this project. Several monitoring wells were intentionally<br />
placed close together to evaluate the aquifer in detail. Although appropriate <strong>for</strong> this study,<br />
the design was costly. Similar designs with several monitoring wells and intensive water-quality<br />
analyses may be prohibitively expensive <strong>for</strong> many utilities. Based on the results of this study,<br />
natural filtration at similar sites could be evaluated using fewer wells. With fewer wells, the wells<br />
can be designed with longer screens to accommodate higher pumping rates <strong>for</strong> analyses, such as<br />
189
190<br />
Microscopic Particulate Analysis. In addition, longer screen lengths may be less susceptible to<br />
local aquifer heterogeneities because water will be contributed from a larger portion of the aquifer.<br />
For other study designs, using data sondes that measure only temperature and specific conductance,<br />
as well as eliminating the inclined wells (depending on the goals of the study), would reduce costs.<br />
The construction of a simple stilling well fitted with a transducer, data sonde, and sampling pump<br />
would be adequate <strong>for</strong> short-term river monitoring.<br />
In general, the monitored parameters showed the most significant drop in concentration from the<br />
river through the riverbed to the inclined wells. Further reductions of concentrations continued<br />
(although not as dramatically) as the water traveled from below the streambed to the production<br />
well. Another significant drop in concentrations occurred from the “C” wells (nearest the<br />
production wells) to the production wells. This is believed to be caused by the dilution effect of<br />
regional groundwater as a result of the large capture zone of the production well.<br />
The high TOC concentrations observed in Well FP8B during the first 2 months of the study<br />
(described earlier) may be due to the growth of indigenous aerobic bacteria after oxygen was<br />
introduced during well installation and development. Once oxygen was consumed, the aerobic<br />
bacteria likely died off and TOC dropped to natural levels. This indicates that this particular well<br />
intercepted a zone of the aquifer with different biological properties than other parts of the aquifer.<br />
The presence of strong odors of hydrogen sulfide, indicative of sulfur-reducing bacteria, and the<br />
excessive corrosion of metal pump fittings are also indicative of biological activity in this well. Had<br />
these high concentrations been due to a preferential flow path, TOC concentrations would not<br />
have exceeded the levels of the river and would not have been expected to decline.<br />
If the high particle counts in Wells FP8B and FP8D were a result of poor filtration, other constituents<br />
also show poor reduction from the river to the well. Aside from the initial high concentrations in<br />
FP8B, TOC concentrations were between those of Wells FP8A and FP8C. Additional parameters<br />
such as UV 254, nitrate, and chloride (not shown) also reflected effective filtration through the<br />
aquifer. This demonstrates that the high particle counts in Well FP8B originated in the aquifer and<br />
were not a result of a preferential flowpath. In the case of FP8D, the particle levels are above the<br />
counts in the river, indicating a secondary source of particles. As in FP8B, other parameters in FP8D<br />
indicated the high particle counts were not a result of preferential flow.<br />
It is widely understood that even a relatively homogeneous aquifer contains localized heterogeneities<br />
in particle size, in situ biology, and water chemistry. These heterogeneities, however, do<br />
not necessarily compromise filtration through the aquifer. Some anomalies can be recognized and<br />
evaluated by incorporating sufficient monitoring points, lengthening well screens to intercept<br />
larger sections of the aquifer, and monitoring <strong>for</strong> sufficient parameters so that anomalies can be<br />
recognized and evaluated.<br />
Acknowledgement<br />
Funding <strong>for</strong> this project was provided by a grant from the Ohio <strong>Water</strong> Development Authority, a<br />
grant from USGS, and funds from the Greater Cincinnati <strong>Water</strong> Works. The authors thank each<br />
of these agencies <strong>for</strong> their contributions and support throughout this project.
REFERENCE<br />
Koterba, M.T, F.D. Wilde, and W.W. Lapham (1995). Ground-water data-collection protocols and procedures <strong>for</strong><br />
the <strong>National</strong> <strong>Water</strong>-Quality Assessment Program: Collection and documentation of water-quality and samples and<br />
related data, U.S. Geological Survey Open File Report 95-399.<br />
BRUCE WHITTEBERRY has been a Professional Hydrogeologist <strong>for</strong> Greater Cincinnati<br />
<strong>Water</strong> Works <strong>for</strong> the past 5 years. In addition to <strong>RBF</strong> research, Whitteberry oversees the<br />
wellhead protection programs <strong>for</strong> two well fields and addresses various groundwater<br />
quantity and quality concerns <strong>for</strong> the Greater Cincinnati <strong>Water</strong> Works. He is one of the<br />
Greater Cincinnati <strong>Water</strong> Works’ representatives <strong>for</strong> the Hamilton to New Baltimore<br />
Ground <strong>Water</strong> Consortium, and he also oversees the Consortium’s regional groundwater<br />
monitoring program, as well as serves in an advisory capacity on groundwater issues. Prior<br />
to joining the Greater Cincinnati <strong>Water</strong> Works, he spent several years in the groundwater industry, both in<br />
consulting and in a regulatory capacity. Whitteberry received a B.S. in Geology from Olivet Nazarene<br />
University and an M.S. in Hydrogeology from Wright State University.<br />
191
192
Session 11: Case Studies “Lessons Learned”<br />
The Hungarian Experience with Riverbank Filtration<br />
Ferenc Laszlo, Ph.D.<br />
<strong>Institute</strong> <strong>for</strong> <strong>Water</strong> Pollution Control<br />
<strong>Water</strong> Resources <strong>Research</strong> Centre<br />
Budapest, Hungary<br />
The water-supply systems of Hungary provide drinking water <strong>for</strong> 9.5-million inhabitants<br />
(95 percent of the population) in 2,100 settlements of the country. About 88 percent of the total<br />
volume supplied originates from groundwater sources. About 30 percent of this (approximately<br />
800,000 m 3 /d) is abstracted from <strong>RBF</strong> water resources.<br />
There is an additional 3-million m 3 /d free <strong>RBF</strong> capacity along the Hungarian section of the<br />
Danube River.<br />
The largest <strong>RBF</strong> system in Hungary provides drinking water <strong>for</strong> 2-million inhabitants in Budapest.<br />
The present capacity is about 900,000 m 3 /d. Approximately 850 <strong>RBF</strong> wells are located on two<br />
islands (Szentendre Island and Csepel Island) of the Danube River. Some parts of this <strong>RBF</strong> system<br />
have operated <strong>for</strong> more than 100 years (Laszlo and Homonnay, 1986). The hydraulic conductivity<br />
of the gravel terraces, the composition of the filtration layer, and the water quality of the river<br />
enables the production of high-quality drinking water.<br />
In recent years, considerable ef<strong>for</strong>ts were focused on investigating and developing methods of<br />
practical application <strong>for</strong> the reliable utilization and protection of bank-filtered water resources.<br />
The activities were mainly directed toward evaluating pollution impacts and exploring the<br />
processes and causes of water-quality changes in <strong>RBF</strong> systems.<br />
Studies in an <strong>RBF</strong> pilot area on Szentendre Island included:<br />
• An extended field investigation to provide adequate data <strong>for</strong> a comprehensive<br />
water-quality evaluation and <strong>for</strong> modeling activities.<br />
• Adsorption experiments in the laboratory.<br />
• Application and development of numerical models to assess and predict water-quality<br />
changes in the <strong>RBF</strong> system that are induced by various potential changes in the<br />
conditions of the river and filtration media.<br />
Numerical modeling was applied to:<br />
• Calibrate hydraulic and transport parameters.<br />
• Calculate the transport of various pollutants from the Danube River to the wells.<br />
• Assess quality changes in the water of the production wells.<br />
• Predict the consequences of accidental pollution in the Danube River.<br />
Correspondence should be addressed to:<br />
Ferenc Laszlo, Ph.D.<br />
Director, <strong>Institute</strong> <strong>for</strong> <strong>Water</strong> Pollution Control<br />
<strong>Water</strong> Resources <strong>Research</strong> Centre<br />
H-1095 Budapest • Kvassay ut 1., Hungary<br />
Phone: +36 1 215 9045 • Fax: +36 1 216 8140 • Email: laszloferenc@vituki.hu<br />
193
194<br />
During the field investigation program, hydrogeological, hydraulic, hydrological, chemical,<br />
physico-chemical, biological, and microbiological data were collected and evaluated. The results of<br />
the water-quality studies revealed characteristic quality changes in the filter media: the chemical<br />
oxygen demand concentration gradually decreased from the river towards abstraction sites,<br />
indicating the decomposition of organic pollutants. Hydro-biological studies indicated that the<br />
Danube River had considerable fecal coli<strong>for</strong>m pollution, but that bacteria removal during <strong>RBF</strong> was<br />
very efficient. Coliphages used as indicators of virus pollution were also measured. The results of<br />
these tests verified that coliphages were not able to pass through filter media.<br />
The retardation of specific organic micropollutants is very different depending on their biodegradability,<br />
sorption affinity, etc. The mobile, persistent organic micropollutants (e.g., atrazine,<br />
simazine, trichloroethene, and benzene) have low removal efficiency (less than 30 percent); on<br />
the other hand, some pesticides (e g., terbutrin, 2,4-D, and carbaryl) and petroleum hydrocarbons<br />
are removed substantially. The retardation or remobilization of some heavy metals is influenced<br />
by redox conditions.<br />
Adsorption experiments in the laboratory provided in<strong>for</strong>mation on the partition coefficients of<br />
various micropollutants between spiked Danube water and the material of filtration medium taken<br />
from <strong>RBF</strong> sites (Laszlo and Literathy, 2002). Two riverbed materials were used: one represented the<br />
sandy alluvium having low organic material; the other was taken from a sedimentation zone of the<br />
riverbed where river water infiltrates through sediment having high organic material content and fine<br />
particle size. The partition coefficient varied widely <strong>for</strong> the different organic micropollutants and<br />
heavy metals, depending mainly on the character of the pollutants. The type of bed material was also<br />
important: the partition coefficients were higher in silty riverbed material than in the sandy matrix<br />
<strong>for</strong> all pollutants.<br />
Numerical transport modeling — based on measured hydraulic, hydrogeological, and physicochemical<br />
parameters — concluded that relatively short-duration accidental pollution has limited<br />
risk <strong>for</strong> <strong>RBF</strong> schemes along the Danube River due to a wide range of travel times from the river to<br />
abstraction wells. Higher risk would be long-lasting, continuous pollution of the Danube.<br />
Another concern is the impacts of river training and gravel dredging on the quality of riverbankfiltered<br />
water (Laszlo et al., 1990). Training structures (groynes, parallel dykes) and dredging<br />
operations affect the hydraulic conditions in the river that are conducive to silting in areas with<br />
reduced flow velocities. Adverse hydrochemical changes occur in the silted filter layer, especially<br />
the dissolution of iron and manganese, and higher concentrations of ammonium are observable.<br />
Bed degradation owing to dredging causes changes in the inflow ratio to the abstraction wells,<br />
increasing the proportion of polluted groundwater from the background areas in the wells.<br />
Conclusions<br />
<strong>RBF</strong> has been an effective method <strong>for</strong> drinking-water supply in Hungary <strong>for</strong> a long time.<br />
The sustainability of <strong>RBF</strong> abstraction schemes depends on several factors, including the quality of<br />
river water, characteristics of the filtration media, retention time within the aquifer, and quality<br />
of the groundwater adjacent to the extraction site.
REFERENCES<br />
Laszlo , F., and Z. Homonnay (1986). “Study of effects determining the quality of bank-filtered well water.”<br />
Conjunctive <strong>Water</strong> Use, IAHS Publ. No. 156, 181-188.<br />
Laszlo, F., Z. Homonnay, and M. Zimonyi (1990). “Impacts of river training on the quality of bank-filtered<br />
waters.” Wat. Sci. Tech., 22 (5): 167-172.<br />
Laszlo, F., and P. Literathy (2002). “Laboratory and field studies of pollutant removal.” Riverbank filtration:<br />
Understanding contaminant biogeochemistry and pathogen removal, C. Ray, ed., Kluwer Academic Publishers,<br />
Dordrecht.<br />
FERENC LASZLO is Director of the <strong>Institute</strong> <strong>for</strong> <strong>Water</strong> Pollution Control of the <strong>Water</strong><br />
Resources <strong>Research</strong> Centre in Hungary. A chemical engineer, Laszlo has 30 years of<br />
experience in research related to water pollution and aquatic chemistry. His research<br />
interests include the protection of riverbank-filtration systems, investigation of<br />
water-quality changes in different water resources, study of micropollutants, simulation of<br />
the fate and transport of contaminants in the aquatic environment, development and<br />
operation of water-quality monitoring system, and accident emergency warning systems.<br />
In addition, he has provided short-term consulting and expert services to the World Health Organization and<br />
United Nations Development Program. Laszlo received an M.S. in Chemical Engineering and a University<br />
Doctorate Degree in Aquatic Chemistry from Veszprem University in Hungary.<br />
195
196
Session 11: Case Studies “Lessons Learned”<br />
Nitrate Pollution of a <strong>Water</strong> Resource – 15 N and 18 O<br />
Study of Infiltrated Surface <strong>Water</strong><br />
Frantisek Buzek, Ph.D.<br />
Czech Geological Survey<br />
Prague, Czech Republic<br />
Renata Kadlecova<br />
Czech Geological Survey<br />
Prague, Czech Republic<br />
Miroslav Knezek, Ph.D.<br />
Prague, Czech Republic<br />
This study was undertaken to determine the effects of agricultural and human activities at the<br />
village of Karany near Prague, the Czech Republic, on the quality of water resources. Three groups of<br />
wells use bank infiltration water from the Jizera River to produce about 1,000 liters per second of<br />
quality drinking water <strong>for</strong> Prague. Some wells have exhibited a steady increase in nitrate content<br />
in recent years (Figure 1), although river-water quality remains good. A question arises of the<br />
origin of nitrate contamination and its transportation to wells, and to what extent bank-filtered<br />
water contributes to produced water. Also of interest are other contributions and how long it takes<br />
<strong>for</strong> these contributions to reach the wells.<br />
Figure 1. Nitrate content (in milligrams per liter, 6-months average value) of contaminated capture wells.<br />
Correspondence should be addressed to:<br />
Frantisek Buzek, Ph.D.<br />
Head of Stable Isotope Lab<br />
Czech Geological Survey<br />
Geologicka 6 • 152 00 Prague 5, Czech Republic<br />
Phone: 420 251085345 • Fax: 420 251818748 • Email: buzek@cgu.cz<br />
197
198<br />
Site Description<br />
The study area is <strong>for</strong>med by Cretaceous sandstones and marlstones (the Jizera <strong>for</strong>mation) of Turonian<br />
age. The Jizera River represents a base <strong>for</strong> local drainage. A shallow aquifer is situated in Late<br />
Pleistocene and Holocene alluvial deposits. They consist of sands and sandy gravel 7- to 9.5-m thick.<br />
The Quaternary aquifer is 2.5- to 5-m thick. It is hydraulically connected with the Jizera River<br />
stream. The Quaternary sediments have a high degree of transmissivity and high permeability. The<br />
Cretaceous deposits have an average transmissivity and free groundwater surface.<br />
On the right bank of the Jizera River, where the Turonian aquifer has a sandy development, its<br />
transmissivity is only about 1 order inferior to that of the alluvial plain deposits. On the left bank<br />
of the Jizera River, the permeability is lower due to a higher proportion of marly component in the<br />
Turonian sediments. Contamined water Wells 227 to 299 (Skorkov capture wells) are located<br />
along the Jizera River at a distance of 200 to 300 m from the right bank (Wells 227 to 270) and<br />
left bank (Wells 271 to 299) (Figure 2). Sojovice capture wells are located on the left bank only;<br />
one part is recharged by artificial infiltration (Wells 120 to 187), another part is traditional bank<br />
infiltration at a distance 300 m from the river (Wells 188 to 226). The river terraces are used by<br />
local farmers <strong>for</strong> the cultivation of corn, potatoes, and maize. Local villages have poor sewer<br />
systems, and there are several landfills located in the area.<br />
Methods<br />
From 1999 to 2000, we took water samples at various time intervals from the river, precipitation<br />
in the recharge area, precipitation at the site, and wells <strong>for</strong> the oxygen isotope composition of<br />
water (δ 18 O - H 2O), and river and wells <strong>for</strong> the isotope composition of nitrogen in nitrate<br />
(δ 15 N-NO 3). 1<br />
δ 18 O data were used to:<br />
• Describe the river system.<br />
• Specify groundwater recharge.<br />
• Calculate the contribution of infiltrated river water to well production.<br />
• Model infiltration water flow and transportation time to the well.<br />
• Specify additional water contributing to well water.<br />
δ 15 N-NO 3 data were used to:<br />
• Specify the origin of nitrate and possible mixtures.<br />
• Trace seasonal trends in nitrate content.<br />
• Follow reactions in the groundwater system (denitrification).<br />
River System<br />
The mean residence time T equals 7.2 months, as calculated from variations of δ 18 O values of the<br />
river and precipitation in the recharge area. An average contribution of direct precipitation to<br />
runoff is about 13 percent, with a maximum of 36 percent during the highest flood event.<br />
1 Isotopic composition is measured in delta units, δ(in ‰) = Rx /R s -1) × 1000, where R denotes the ratio of the heavyto-light<br />
isotope (e.g., 18 O/ 16 O or 15 N/ 14 N), and R x and R s are the ratios in the sample and standard, respectfully. As<br />
international reference standard is used, Standard Mean Ocean <strong>Water</strong> (SMOW) stands <strong>for</strong> water and atmospheric N 2<br />
<strong>for</strong> nitrogen.
Figure 2. A sketch of the study area.<br />
Key: 1 = Forested area.<br />
2 = Jizera-Elbe watershed.<br />
3 = Capture well.<br />
4 = Groundwater flow.<br />
5 = Deep well.<br />
6 = Landfill.<br />
Nitrate content and δ 15 N-NO 3 of the Jizera River follow seasonal variations. δ 15 N-NO 3 varies from<br />
8 ‰ in the winter to about 1.5 ‰ (base flow only). The highest nitrate content (about 25 mg/L)<br />
was measured during snowmelt. Base flow nitrate content is very low (about 8 to 10 mg/L).<br />
Additional nitrate to base flow originates mostly from drained precipitation events (short time or<br />
fast component of runoff), flushing fertilizers bringing mostly organic nitrogen in the spring<br />
(manure applied in the winter period), and inorganic nitrogen during the growing season<br />
(fertilizers applied on leaves).<br />
199
200<br />
The nitrate content and δ 15 N-NO 3 of the Jizera River do not follow the same seasonal variation.<br />
δ 15 N-NO 3 varies from 8 ‰ in winter to about 1.5 ‰ (base flow only). The highest nitrate content<br />
(about 25 mg/L) was measured during snowmelt. Base flow nitrate content is very low (about 8 to<br />
10 mg/L). Figure 3 shows the nitrate content and isotopic composition of the Jizera River.<br />
Figure 3. Nitrate content and isotopic composition of the Jizera River.<br />
As the response of river runoff to precipitation is quite fast (from 3 to 5 days after a storm), a<br />
simplified two-component model of runoff generation is applicable. The discharged water consists<br />
of groundwater and short residence-time components. Short time components are <strong>for</strong>med by<br />
drained precipitation events flushing fertilizers — predominantly the organic nitrogen applied in<br />
winter — and by inorganic nitrogen applied during the growing season. Increasing δ 15 N-NO 3<br />
levels during the autumn corresponds to residual nitrate levels from the unsaturated soil zone<br />
below root level. A similar seasonal variability was observed during the next year as well.<br />
Infiltration System<br />
<strong>Water</strong> Balance<br />
The actual and modeled δ 18 O data were compared during a flood event and steady-state<br />
conditions. At flood conditions, the contribution of infiltrated river water increased up to<br />
100 percent, with a transit time of only about 5 days. At steady-state conditions, infiltrated river<br />
water <strong>for</strong>ms about 60 percent of infiltrated water, with a transit time of 28 days. The additional<br />
40 percent of infiltrated water cannot be explained by local groundwater only. It must originate<br />
from a shallow aquifer. Fast transit times of this recharging water were confirmed by a variability<br />
of δ 18 O in the wells, which is higher than in the river. Additional recharging water can be<br />
modeled by local precipitation input, with a delay in the order of weeks.<br />
Nitrate Contamination of Skorkov Wells<br />
From comparing the high nitrate content of contaminated wells and their seasonal δ 15 N-NO 3<br />
variability, it is obvious that:<br />
• Nitrate originates mostly from water other than infiltrated water.<br />
• Sources of nitrate are very similar to river sources (i.e., similar sequences of nitrogen<br />
inputs during the year.<br />
Wells have a higher concentration of nitrates than river water.
By analyzing the δ 15 N-NO 3 of single wells along the contaminated part of bank infiltration, we<br />
identified the possible sources of nitrate. Besides the seasonal application of fertilizers and manure,<br />
village sewer systems and landfills in the recharge area also contribute to water pollution in the<br />
wells (Figure 4).<br />
Figure 4. Nitrate content and isotopic composition of contaminated water (Skorkov single wells).<br />
The importance of local precipitation <strong>for</strong> recharging wells is obvious from the long-time variation<br />
of nitrate content of contaminated wells (see Figure 1). Minimal values on an otherwise<br />
continuously increasing plot correspond to years with very low precipitation input (25-percent less<br />
than average). With low precipitation input, wells are recharged preferentially by deeper<br />
groundwater with a lower nitrate content. Within a high precipitation period, wells are recharged<br />
by fast infiltrated local precipitation washing out nitrate in the unsaturated zone.<br />
Nitrate Contamination of Sojovice Wells<br />
Sojovice capture wells have two parts: one part is recharged by artificial infiltration (Wells 120 to<br />
187), another part is traditional bank filtration at a distance of 300 m from the river (Wells 188<br />
to 126). Contrary to the Skorkov system, the Sojovice wells are not affected directly by<br />
precipitation (Figure 5). Besides, both the bank infiltration and artificial infiltration parts are<br />
partially recharged by denitrified groundwater. From a time sequence of nitrate content and its<br />
δ 15 N-NO 3 on single wells, we could separate three components of the recharge:<br />
• River water.<br />
• Shallow groundwater with a nitrate source typical <strong>for</strong> infiltration in the left bank.<br />
• Deep groundwater with a reduction zone and denitrification.<br />
Flow paths of groundwater components are different — perpendicular to river flow (shallow<br />
component) and along the river (deep component).<br />
Conclusions<br />
Nitrate contamination of the bank infiltration systems originates in local sources of nitrate,<br />
including inappropriate uses of manure/fertilizers and leaking village sewers. A considerable<br />
201
202<br />
Figure 5. Nitrate isotopic composition of contaminated water (Sojovice single wells).<br />
increase of nitrate in the 1980s originated both because of the change in agriculture practices (due<br />
to an increased proportion of maize cultivation) and the change in hydrology dynamics of the<br />
system (due to the extensive use of water resources, new infiltration pathways were developed that<br />
recharge more shallower groundwater than be<strong>for</strong>e).<br />
FRANK BUZEK has worked <strong>for</strong> the Czech Geological Survey <strong>for</strong> about 25 years, and is<br />
Head of the Stable Isotope Group. His primary research interest includes the application<br />
of isotope geochemistry to environmental problems, hydrology, and organic geochemistry.<br />
As an all-round geochemist, he was responsible <strong>for</strong> stable isotope data on carbon and<br />
nitrogen cycling in <strong>for</strong>est soils (European Union international projects CANIF, NIPHYS,<br />
and FORCAST) and tracing atmospheric emissions and groundwater pollution<br />
(International Atomic Energy Agency project on isotope technique in groundwater<br />
pollution). Most of the recent case studies solved by Buzek and his team have dealt with nitrate in water<br />
resources. Buzek received an M.S. in Physical Chemistry from the <strong>Institute</strong> of Chemical Technology and a<br />
Ph.D. in Applied Surface Chemistry from the <strong>Institute</strong> of Chemical Processes Fundamentals at the Czech<br />
Academy of Sciences in Prague.
Session 11: Case Studies “Lessons Learned”<br />
Microbial Growth in Artificially Recharged Groundwater:<br />
Experiences from a 4-Year Project<br />
Ilkka T. Miettinen, Ph.D.<br />
<strong>National</strong> Public Health <strong>Institute</strong><br />
Kuopio, Finland<br />
Markku J. Lehtola, Ph.D.<br />
<strong>National</strong> Public Health <strong>Institute</strong><br />
Kuopio, Finland<br />
Prof. Terttu Vartiainen<br />
<strong>National</strong> Public Health <strong>Institute</strong><br />
Kuopio, Finland<br />
Prof. Pertti J. Martikainen<br />
Kuopio University<br />
Kuopio, Finland<br />
Introduction<br />
There are regions in Finland where large groundwater aquifers are not available and where the<br />
artificial groundwater recharge technique is an important alternative <strong>for</strong> drinking-water production.<br />
In Finland, 9 percent of drinking water is produced using the artificial recharge of groundwater or<br />
bank-filtration techniques. The production of artificially recharged groundwater has increased<br />
during the last decade. The high content of organic carbon (humus) present in surface water is a<br />
serious problem when this water is used to produce drinking water.<br />
Organic carbon can react with chlorine during disinfection, <strong>for</strong>ming halogenated byproducts. The<br />
high availability of organic carbon may also cause microbial problems in distribution networks.<br />
The fraction of organic carbon that microbes can use <strong>for</strong> growth (i.e., AOC) is considered to be<br />
the most important nutrient affecting microbial growth in drinking waters (van der Kooij, 1992).<br />
In boreal areas, phosphorus — in addition to organic carbon — has been shown to regulate<br />
microbial growth in drinking waters (Miettinen et al., 1996).<br />
Objectives<br />
A 4-year project examined how the artificial recharge of lake water affects the following:<br />
• Molecular weight distribution of organic matter.<br />
• AOC content.<br />
• Microbially available phosphorus (MAP) content.<br />
• Microbial growth potential in water.<br />
Correspondence should be addressed to:<br />
Ilkka Miettinen, Ph.D.<br />
<strong>Research</strong>er, Laboratory of Environmental Microbiology<br />
<strong>National</strong> Public Health <strong>Institute</strong><br />
Department of Environmental Health • P.O. Box 95 • FIN-70701 Kuopio Finland<br />
Phone: (358) 17-201371 • Fax: (358) 17-201155 • Email: Ilkka.Miettinen@ktl.fi<br />
203
204<br />
Methodology<br />
Experimental Sites<br />
Changes in water quality during artificial groundwater recharge were studied in five water works.<br />
All water works (<strong>Water</strong> Works A, B, C, D, and E) are located in esker areas, where the soil consists<br />
mainly of sand and gravel. Lake water is filtrated into the ground by basin infiltration in <strong>Water</strong><br />
Works A, C, D, and E. <strong>Water</strong> Works A, B, and E use sprinkling infiltration through <strong>for</strong>est soil,<br />
while <strong>Water</strong> Works B and C use the same lake as their raw-water source.<br />
<strong>Water</strong> Analyses<br />
The quality of raw-water samples and water samples from different monitoring wells at different<br />
filtration distances (raw water, the first monitoring point after infiltration, and the last monitoring<br />
point after filtration) were analyzed from 1998 to 2001. Filtration distance varied from 10 to 280 m in<br />
the first monitoring point and from 390 to 1,200 m in the last monitoring point (artificially recharged<br />
groundwater). The quality of organic carbon was analyzed in terms of its molecular weight distribution<br />
using the high-pressure size-exclusion chromatography (HPSEC) method (Vartiainen et al., 1987).<br />
AOC was measured with a modified bioassay (Miettinen et al., 1999) originally presented by van der<br />
Kooij et al. (1982), and MAP was analyzed by bioassay (Lehtola et al., 1999). Microbial growth<br />
potential (heterotrophic growth potential) was tested using a method by Miettinen et al. (1997).<br />
Results and Discussion<br />
Organic matter in lake waters consisted mainly of high and intermediate molecular-size fractions. These<br />
fractions decreased rapidly in the filtration process. In contrast, the lowest molecular weight fractions<br />
of organic matter decreased only slightly during filtration. The strongest reduction in the content of<br />
AOC occurred in the beginning of filtration. Later on, there was only a slight decrease in AOC<br />
content. On average, there was a 53-percent reduction in the concentration of AOC during the<br />
artificial recharge of groundwater (Table 1). Changes in MAP concentrations during filtration varied<br />
among water works. MAP content decreased strongly (89- to 99.9-percent removal) during filtration<br />
in <strong>Water</strong> Works A, B, and D. <strong>Water</strong> Works E was the only exception where MAP content increased<br />
during the filtration process, indicating that phosphorus was dissolving from soil into water (see Table 1).<br />
Table 1. Concentrations (Range and Mean Value in Parentheses) of AOC and MAP<br />
in Raw <strong>Water</strong> and Artificially Recharged Groundwater. Number of Observations: 4 to 11.<br />
<strong>Water</strong> AOC: µg Acetate eq. C/L MAP: µg MAP-P/L<br />
Works RW ARW RW ARW<br />
A 54 to 366 (134) 17 to 132 (46) 0.98 to 9.90 (3.84) 0.22 to 0.34 (0.27)<br />
B 15 to 120 (56) 15 to 73 (46) 1.79 to 5.31 (2.57) 0.15 to 0.37 (0.28)<br />
C 39 to 144 (78) 27 to 90 (49) Not Determined Not Determined<br />
D 27 to 288 (151)
Microbial growth was weak in all lake-water samples, despite the fact that these waters had high<br />
AOC and MAP contents; however, there was strong microbial growth in artificially recharged<br />
groundwater. Lower microbial growth in surface water could be associated with the grazing activity<br />
of protozoa (Hahn and Hofle, 2001). The microbial growth potential in groundwater was strongest<br />
in the first sampling point and weakened during filtration (Figure 1). The AOC content did not<br />
correlate with microbial growth potential; however, the maximum microbial counts correlated<br />
with the MAP content (r = 0.70, p = 0.000, n = 26) when raw-water samples and samples with<br />
MAP over 5 µg/L (no phosphorus limitation) were excluded from the data.<br />
Figure 1. Increase in microbial cell numbers (R2A plate counts) with time when water samples were<br />
incubated in the laboratory. The figure shows mean numbers in samples taken from raw water or<br />
artificially recharged groundwater of <strong>Water</strong> Works A, B, D, and E. Symbols: RW = Raw water.<br />
1.point = The first monitoring point after infiltration. 2. Point = The last monitoring point after<br />
filtration). Number of observations: 18 to 21.<br />
Conclusions<br />
• The quality of organic matter changed during artificial recharge: high and intermediate<br />
molecular weight organic compounds were removed from infiltrated lake water.<br />
• The decrease in AOC content was strongest in the beginning of infiltration. An increase<br />
in filtration distance had a minor effect on AOC removal.<br />
• MAP removal depended on soil characteristics. In most of the studied water works,<br />
removal was good.<br />
• Microbial growth was higher in artificially recharged groundwater than in raw water<br />
(surface water). The microbial growth potential decreased during filtration.<br />
• Because the removal efficiency <strong>for</strong> phosphorus was higher than that <strong>for</strong> carbon, phosphorus<br />
became the main nutrient regulating microbial growth in artificially recharged<br />
groundwater.<br />
205
206<br />
REFERENCES<br />
Hahn, M.W., and M.G. Höfle (2001). “Grazing of protozoa and its effect on populations of aquatic bacteria.”<br />
FEMS Microbiol. Ecol., 35: 113-121.<br />
Lehtola, M., I.T. Miettinen, T. Vartiainen, and P.J. Martikainen (1999). “A new sensitive bioassay <strong>for</strong><br />
determination of microbially available phosphorus in water.” Appl. Environ. Microbiol., 65(5): 2,032-2,034.<br />
Miettinen, I.T., T. Vartiainen, and P.J. Martikainen (1996). “Contamination of drinking water.” Nature,<br />
381: 654-655.<br />
Miettinen, I.T., T. Vartiainen, and P.J. Martikainen (1997). “Microbial growth and assimilable organic<br />
carbon in Finnish drinking waters.” Wat. Sci. Techn., 35(11/12): 301-306<br />
Miettinen, I.T., T. Vartiainen, and P.J. Martikainen (1999). “Determination of assimilable organic carbon<br />
(AOC) in humus-rich waters.” Wat. Res., 3(10): 277-2,282.<br />
Vartiainen, T., A. Liimatainen, and P. Kauranen (1987). “The use of size exclusion columns in determination<br />
of the quality and quantity of humus in raw waters and drinking waters.” Sci. Total Environ., 62: 75-84.<br />
van der Kooij, D., W.A.M. Hijnen, and A. Visser (1982). “Determinating the concentration of easily<br />
assimilable organic carbon.” Journal AWWA, 74: 540-545.<br />
van der Kooij, D. (1992). “Assimilable organic carbon as an Indicator of Bacterial Regrowth.” Journal AWWA,<br />
84(2): 57-66.<br />
ILKKA MIETTINEN has been a researcher <strong>for</strong> the <strong>National</strong> Public Health <strong>Institute</strong> in<br />
Finland since 1987. Currently, he serves as <strong>Research</strong>er of the Laboratory of Environmental<br />
Microbiology at the <strong>Institute</strong>, as well as <strong>Research</strong>er <strong>for</strong> Kuopio University. For the past<br />
3 years, his research has focused on the role of phosphorous in microbial growth potential<br />
and biofilm <strong>for</strong>mation in distribution networks. A recent project of his is called,<br />
“Surveillance and control of microbiological stability in drinking-water distribution<br />
networks.” Past research has focused on chemical and microbiological quality in humusrich<br />
lake water during bank filtration, as well as the effects of strong oxidants on the quantity and quality of<br />
organic matter and microbial regrowth in treated waters. Miettinen received a B.S. in Biochemistry, M.S. in<br />
Biotechnology, and Ph.D. in Environmental Sciences at the University of Kuopio, where he currently is<br />
Docent of Environmental Sciences.
Session 11: Case Studies “Lessons Learned”<br />
Evaluation of the Existing Per<strong>for</strong>mance of Infiltration<br />
Galleries in Alluvial Deposits of the Parapeti River<br />
Dip.-Eng. Alvaro Camacho<br />
Bolivian Association of Sanitary Engineers<br />
La Paz, Bolivia<br />
Introduction<br />
The Parapeti River is part of the drainage system of the Parapeti-Izozog watershed, which covers<br />
about 52,000 square kilometers and is the sole water resource <strong>for</strong> supplying drinking water to the<br />
City of Camiri, Bolivia. The strata of the basin consist of Quaternary deposits and the permeability<br />
is good; however, precipitation is low (only 700 to 800 millimeters) and, as this zone is located in<br />
the watershed between the Amazon and La Plata river systems, existing groundwater quantity is<br />
very low. The aquifer depth is between 100 to 200 m in the vicinity of Camiri (yield capacity is<br />
less than 1.0 liters per second). The City is located in the region of Santa Cruz, Bolivia, at an<br />
altitude of 810-m above sea level, in the southeast, and has a population of 25,000 inhabitants.<br />
The average temperature is 25-degrees Celsius, with a maximum of 38-degrees Celsius in the<br />
summer and a minimum of 11-degrees Celsius in the winter.<br />
The City’s water demand is supplied by five infiltration galleries buried below the riverbed<br />
(Figure 1), on the bedrock, at a depth of about 4 to 5 m. The galleries are interconnected through<br />
Gallery<br />
1<br />
Pumping<br />
Station<br />
Collector<br />
Well<br />
Parapeti River<br />
Gallery<br />
2 Gallery<br />
5<br />
Figure 1. Location of the study area.<br />
Correspondence should be addressed to:<br />
Dip.-Eng. Alvaro Camacho<br />
Consultant Engineer<br />
Bolivian Association of Sanitary Engineers<br />
Casilla 9348 • La Paz, Bolivia<br />
Phone and Fax: (591-2) 241- 6283 • Email: alcamachog@unete.com<br />
Study<br />
Area<br />
Camiri<br />
Location of the Infiltration Galleries<br />
Collector<br />
Well<br />
Collector<br />
Well<br />
Gallery<br />
3 Gallery<br />
4<br />
Collector<br />
Well<br />
Collector<br />
Well<br />
207
208<br />
a network and linked to a collector well sited on the riverbank. From there, the collected water is<br />
pumped to a disinfection tank prior to delivery to customers.<br />
The infiltration galleries are channels with a 1.0-m × 1.0-m section or drains of 12-inch covered<br />
by four layers (Table 1) of a filter medium (Figure 2)<br />
4 to 5 m<br />
Permeable<br />
Alluvial<br />
Deposits<br />
Figure 2. Infiltration gallery system.<br />
Table 1. Artificial Filterbed Layers of Infiltration Galleries<br />
Bed Layers Type/Size Height<br />
First Layer (Top) Coarse Sand 1.75 m<br />
Second Layer Gravel, 1 to 2 inches 1.0 m<br />
Third Layer Gravel, 2 to 3 inches 1.0 m<br />
Fourth Layer (Bottom) Small Stone, 3 to 6 inches 1.0 m<br />
Natural Sand River<br />
Section 1 m × 1 m<br />
<strong>Water</strong> Surface<br />
Gravel 1 to 2 inches<br />
Gravel 2 to 3 inches<br />
Gravel 3 to 6 inches<br />
Bed Rock<br />
River Bed<br />
In this area of the Parapeti River, where the galleries are in operation, the riverbed consists of<br />
permeable alluvial material, sand, and gravel of about 4- to 5-m deep, which creates favorable<br />
conditions <strong>for</strong> groundwater flow through the unconfined aquifer (Huisman, 1978). Because the<br />
movement of groundwater develops small velocities in granular material, the flow is laminar (Harr,<br />
1990) and this is the driving <strong>for</strong>ce <strong>for</strong> the sedimentation process in the sand and gravel layers of<br />
the river. The Parapeti is a mountain river whose water flow is subject to dramatic variations, both<br />
in quality and quantity, during the dry season (May to October) and rainy season (November to<br />
April). With a relatively steep slope (0.5 to 1 percent), intensive sediment transport rates occur<br />
in the rainy season (the maximum flow exceeds 1,000 m 3 /s, as compared to 5 m 3 /s in the dry<br />
season). Regarding water quality in the river, turbidity levels in the rainy season have been reported<br />
at about 15,000 ntu and suspended solid concentrations at 30,000 mg/L (Binnie & Partners<br />
Consultants, 1982).<br />
In spite of Parapeti’s water quality, infiltration galleries have worked continuously <strong>for</strong> more than<br />
15 years, producing water of good quality and using disinfection as the only treatment process,<br />
with a discharge of 48 to 50 liters per second. The only annual O&M consists of removing the<br />
finest particles of sand that have reached the bottom of the galleries.
Objectives<br />
The aim of the study was to evaluate the existing per<strong>for</strong>mance of infiltration galleries to improve<br />
on the water quality of surface waters. The study was oriented to answer the following question:<br />
“To what extent do infiltration galleries work to remove contaminants from surface-water supplies?”<br />
Materials and Methods<br />
A sampling program was implemented during the dry and rainy seasons. Two complete sets of water<br />
samples were collected and analyzed between 2002 and 2003 (Schubert, 2002). The first set of samples<br />
was taken in the dry season, from early October to November 2002, when suspended-solid<br />
concentrations in the river are low. The second set of samples was collected during the rainy<br />
season, from February and April, when the river carries higher suspended-solid concentrations. In<br />
both periods, water samples were taken over a period of 41 to 45 days. Grab samples were collected<br />
three times per day in the Parapeti River and in two of the five infiltration galleries (Numbers 1<br />
and 4). Raw-water samples collected in the river were taken from a depth of 0.30-m below the<br />
surface and 100-m upstream of the galleries. At the same time, water samples were taken from the<br />
effluents of the two infiltration galleries (Numbers 1 and 4) at the inlet point of the disinfection tank.<br />
Eight water-quality parameters were chosen: temperature, color, turbidity, suspended solids, total<br />
dissolved solids, pH, conductivity, and fecal coli<strong>for</strong>m counts. Grab samples <strong>for</strong> temperature and pH<br />
were collected in glass bottles and analyzed on site. Random samples <strong>for</strong> color, turbidity, suspended<br />
solids, total dissolved solids, and conductivity were collected in glass bottles <strong>for</strong> analysis in a<br />
laboratory. Microbiological samples were collected in sterile 50-mL Pyrex glass <strong>for</strong> analysis in a<br />
laboratory within 24 hours. The test was conducted using a portable OXFAM-DEL AGUA water<br />
test kit (membrane filtration technique). A duplicate was prepared from each sample. All the<br />
experimental tests were based on the “Standard Methods <strong>for</strong> the Examination of <strong>Water</strong> and<br />
Wastewater” (1975).<br />
In addition, exploration holes were dug into different layers of the coarse material to take samples from<br />
different filter media that cover the galleries. Each layer of the filter media, from Galleries 3, 4, and 5,<br />
was washed with distilled water to per<strong>for</strong>m a column-settling test. This experiment was conducted to<br />
determine the frequency distribution of settling velocities and particle diameters. Also, each sample<br />
was subject to physical analysis, such as mass density, porosity, and granular grain-size distribution.<br />
Results and Discussion<br />
Figures 3 to 5 illustrate seasonal variations in water quality in the Parapetí River and infiltration<br />
galleries. For the purposes of this paper, only two indicators of water quality were chosen: suspended<br />
solids and fecal coli<strong>for</strong>m.<br />
During the rainy season, raw water in the Parapeti River had suspended-solid levels ranging from<br />
45 to 4,850 mg/L, with a mean value of 618 mg/L. These figures differ substantially in comparison<br />
to the dry season, where the average value was 72 mg/L and the maximum was 410 mg/L. In the<br />
rainy season, because the river’s flow is high (from 5 to 1,000 m 3 /s, with a mean value of 30 m 3 /s),<br />
the elevated suspended-solid concentration is caused by erosion that occurs on the steep slopes of<br />
watershed highlands. As shown in Figures 3 to 8, the infiltration galleries that were evaluated<br />
during the same period had very low levels. The values were ranged from 0 to 7 mg/L, with an average<br />
of 2 mg/L far below the common standards. From this evidence, it has been demonstrated that the<br />
per<strong>for</strong>mance of the infiltration galleries reached a removal ratio of more than 90 percent. Turbidity<br />
of 5 ntu or less in 98 percent of the daily samples was reported in the effluents (see Figure 5).<br />
209
210<br />
6045<br />
5045<br />
4045<br />
3045<br />
2045<br />
1045<br />
45<br />
8<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
1<br />
0<br />
River<br />
Gal. 1<br />
Gal. 4<br />
Suspended Solids (mg/L)<br />
Feb-03 Feb-03 Feb-03 Mar-03 Mar-03 Mar-03 Mar-03 Mar-03 Mar-03 Apr-03 Apr-03<br />
Figure 3. Raw water and infiltration gallery water quality.<br />
70150<br />
60150<br />
50150<br />
40150<br />
30150<br />
20150<br />
10150<br />
150<br />
200<br />
150<br />
100<br />
50<br />
0<br />
River<br />
Gal. 1<br />
Gal. 4<br />
Fecal Coli<strong>for</strong>ms (FC/100 mL)<br />
Feb-03 Feb-03 Feb-03 Mar-03 Mar-03 Mar-03 Mar-03 Mar-03 Apr-03<br />
Figure 4. Raw water and infiltration gallery water quality.<br />
% Cummulative Distribution<br />
120%<br />
100%<br />
80%<br />
60%<br />
40%<br />
20%<br />
Faecal Coli<strong>for</strong>ms in Galleries. Frequency Distribution.<br />
Rainy Season (2003)<br />
0%<br />
0 20 40 60 80 100 120 140 160 180<br />
NTU<br />
Figure 5. Quality of water produced in the infiltration galleries.<br />
Gallery 1<br />
Galeria 4
6000<br />
4000<br />
2000<br />
Suspended Solids (mg/L)<br />
0<br />
Minimum Mean Maximum<br />
Gallery 1 0 1 5<br />
Gallery 4 0 2 7<br />
River 45 618 4850<br />
Figure 6. A comparison of raw water and water produced by the galleries.<br />
80000<br />
60000<br />
40000<br />
20000<br />
FE Faecal RMS (FC/100 mL)<br />
0<br />
Minimum Mean Maximum<br />
Gallery 1 1 0 20<br />
Gallery 4 1 32 169<br />
River 160 8958 60685<br />
Figure 7. A comparison of raw water and water produced by the galleries.<br />
Microbiological water-quality parameters in the Parapeti River show poor quality with a wide range<br />
of 160 to 60,000 fecal coli<strong>for</strong>m counts/100 milliliters, with an average value of 9,000 fecal coli<strong>for</strong>m<br />
counts/100 milliliters. The infiltration gallery data illustrate that in one gallery (Gallery 1), the<br />
mean value was zero fecal coli<strong>for</strong>m counts/100 milliliters, with a maximum of 20 fecal coli<strong>for</strong>m<br />
counts/100 milliliters achieving a removal ratio of more than 90 percent. Three fecal coli<strong>for</strong>m<br />
counts /100 milliliters or less in 98 percent of the samples was found. In the other gallery (Gallery 4),<br />
the mean value was a concentration of 32 fecal coli<strong>for</strong>m counts/100 milliliters, with a maximum<br />
of 169 fecal coli<strong>for</strong>m/100 milliliters representing a removal ratio of more than 90 percent. In this<br />
gallery, 120 fecal coli<strong>for</strong>m counts/100 milliliters or less in 98 percent of the daily samples was found.<br />
These differences may be attributed to variations in gallery O&M.<br />
211
212<br />
Cummulative Distribution<br />
100%<br />
90%<br />
80%<br />
70%<br />
60%<br />
50%<br />
40%<br />
30%<br />
20%<br />
10%<br />
0%<br />
Conclusions<br />
Suspended Solids in Galleries. Frequency Distribution.<br />
Rainy Season (2003)<br />
0 1 2 3 4 5 6 7 8<br />
NTU<br />
Figure 8. Quality of water produced by the galleries.<br />
Gallery 4<br />
Gallery 1<br />
Infiltration galleries built in the permeable alluvial deposits of the rivers can be suitable systems<br />
<strong>for</strong> improving water quality from surface waters. Removal ratios of more than 90 percent <strong>for</strong><br />
turbidity, suspended solids, and fecal coli<strong>for</strong>m have been recorded (during this research project).<br />
Infiltration galleries produce high-quality and stable volumes of water independent of seasonal<br />
variations in the quality and quantity of raw-surface stream, minimizing environmental effects.<br />
It seems that there is a natural and self-cleaning filtration process at work here. The permeable<br />
deposits of the Parapeti River <strong>for</strong>m a highly efficient system <strong>for</strong> the pretreatment of water that<br />
contains high concentrations of solids.<br />
Moreover, the filtration qualities of the deposits are constantly maintained by the action of the<br />
river itself: during periods of flooding, the top layer of filtration material — the coarse sand — is<br />
stirred-up and cleaned by the sheer <strong>for</strong>ce of water flow. As flooding subsides, the clean, loose sand<br />
re-settles on the riverbed, thus preventing the natural process of clogging and hardening, which<br />
would otherwise impair its efficacy as a filter. It is, there<strong>for</strong>e, possible that the natural composition<br />
of the river deposits, assisted by the cycle of the river itself, are acting as an effective and<br />
self-cleaning filter of the solids contained in the river.<br />
With proper studies and depending on local conditions and the characteristics of raw-water<br />
quality, infiltration galleries are a suitable alternative <strong>for</strong> supplying drinking water to small and<br />
medium communities, with minimum capital costs and lower O&M costs. The system in Camiri<br />
has been running since the early 1980s, and no critical O&M problems have been observed<br />
(particularly clogging). It is essential that galleries must have easy access to facilitate the periodic<br />
cleaning of sediments from conduits (the minimum channel section should be 1.0-m × 1.0-m or<br />
more <strong>for</strong> manual cleaning). The infiltration gallery system in Camiri has shown that the amount<br />
of maintenance required on the galleries is very small indeed. For this reason, as well as the<br />
benefits of improved or maintained water quality, infiltration galleries have a significant advantage<br />
over other conventional systems.<br />
Further research is still needed to <strong>for</strong>m a complete understanding of the whole process and the<br />
mechanisms that are involved in removing contaminants at higher concentrations.
REFERENCES<br />
American Public Health Association (1975). Standard Methods <strong>for</strong> the Examination of <strong>Water</strong> and Wastewater,<br />
Fourteenth Edition, American Public Health Association, New York.<br />
Binnie & Partners Consultants (1982). Final Report on the design of the surface water treatment plant in Camiri,<br />
Binnie & Partners Consultants, Santa Cruz, Bolivia.<br />
Harr, M.E. (1990). Groundwater and Seepage, General Publishing Company, Toronto, Ontario.<br />
Huisman, L. (1978). Ground <strong>Water</strong> Recovery, International <strong>Institute</strong> <strong>for</strong> Hydraulic and Environmental<br />
Engineering, Technical University of Delft, The Netherlands.<br />
Schubert, J. (2002). “German Experience with Riverbank Filtration Systems Ray.” Riverbank Filtration<br />
Improving Source-<strong>Water</strong> Quality, C. Ray, G. Melin, and R.B. Linsky, editors, Kluwer Academic Publishers,<br />
Dordrecht.<br />
Since 2002, ALVARO CAMACHO has been a consultant engineer <strong>for</strong> the water and<br />
sanitation sectors, including the Bolivian Association of Sanitary Engineers. Prior, he was<br />
the Director General of <strong>Water</strong> Supply and Sanitation <strong>for</strong> the Ministry of Housing Services<br />
of the Republic of Bolivia, a Project Leader in the <strong>Water</strong> and Sanitation Program <strong>for</strong> rural<br />
areas in Bolivia <strong>for</strong> the World Bank, and Guest Lecturer of Sanitary Engineering at the<br />
Universidad Mayor de San Andrés in Bolivia. He is also the author of several reports on<br />
topics such as water-treatment plants designed within the multi-stage filtration concept,<br />
developing the water and sanitation sector in Bolivia, technical standards <strong>for</strong> the design of unconventional<br />
sewer systems in Bolivia, and guidelines <strong>for</strong> the design of water-treatment plants in small and medium<br />
communities. Camacho received a degree in Civil Engineering from the Universidad Mayor de San Andrés<br />
in Bolivia and a Dipl.-Engineer in Sanitary Engineering from the International <strong>Institute</strong> <strong>for</strong> Hydraulics and<br />
Environmental Engineering, in conjunction with the Technical University of Delft, The Netherlands. He is<br />
a M.S. researcher at the Universidad del Valle in Cali, Columbia, where he is conducting a field study on<br />
infiltration galleries <strong>for</strong> water supply.<br />
213
214
Session 11: Case Studies “Lessons Learned”<br />
Sensitivity and Implications of Microscopic Particulate<br />
Analysis – A Collector Well Owner’s Perspective<br />
Barry C. Beyeler<br />
City of Boardman<br />
Boardman, Oregon<br />
The City of Boardman, Oregon, owns two horizontal collector wells situated along the banks of<br />
the Columbia River, which provides the City’s water supply. The horizontal collector well system<br />
was chosen in 1975 <strong>for</strong> its unique ability to produce high volumes of high-quality water. The<br />
horizontal collector wells fill a role of vital importance in the City by addressing the balance of<br />
plentiful water supplies <strong>for</strong> industrial development and the need <strong>for</strong> high-quality drinking water<br />
<strong>for</strong> the citizens of the community.<br />
The City has been confident of the water quality produced by the collector and the filtration<br />
effectiveness it possesses; however, under provisions of the Surface <strong>Water</strong> Treatment Rule, the<br />
potential water-quality effects of the hydraulic connectivity to the Columbia River came under<br />
question. The City chose to per<strong>for</strong>m Microscopic Particulate Analysis on the collector and Columbia<br />
River in the attempt to indicate and quantify the filtration effectiveness of the horizontal collector<br />
well system.<br />
Per<strong>for</strong>mance of Microscopic Particulate Analysis in the field presented many challenges that<br />
needed to be addressed to ensure that a quality representative sample was taken. There are many<br />
ways this sample can be compromised, which could produce non-representative or inaccurate<br />
results. Understanding the sensitivity of the Microscopic Particulate Analysis process from start to<br />
finish, combined with an assessment of what the indicated results would trigger under the Safe<br />
Drinking <strong>Water</strong> Act concerning additional treatment and subsequent associated costs, required<br />
significant care and diligence.<br />
As a result of the sampling and analysis that the City per<strong>for</strong>med, the State of Oregon Department<br />
of Human Resources Drinking <strong>Water</strong> Program has determined that the City is operating a<br />
“groundwater” system. This is based on the quality of the water and assurances provided by<br />
accurate and representative data obtained through Microscopic Particulate Analysis. As the City<br />
brings its second collector online, the coordination of the Microscopic Particulate Analysis<br />
sampling ef<strong>for</strong>ts are being discussed with the Department of Human Resources Drinking <strong>Water</strong><br />
Program to ensure that this collector per<strong>for</strong>mance can be quantified appropriately. The Microscopic<br />
Particulate Analysis process will be accomplished over the next year to provide the in<strong>for</strong>mation<br />
to make appropriate decisions. The City will follow the same process of sampling the river and<br />
collector to indicate filtration effectiveness.<br />
In retrospect, there have been numerous other benefits to the process of Microscopic Particulate<br />
Analysis sampling. The increased knowledge of how the system works has been invaluable in<br />
Correspondence should be addressed to:<br />
Barry C. Beyeler<br />
Community Development Director<br />
City of Boardman, Oregon<br />
202 N Main Street • P.O. Box 229 • Boardman, Oregon 97818 USA<br />
Phone: (541) 481-9252 • Fax: (541) 481-3244 • Email: bbeyeler@cityofboardman.com<br />
215
216<br />
<strong>for</strong>ums concerning Endangered Species Act listings of salmon and steelhead populations in the<br />
Columbia and Snake River basins. As proposals <strong>for</strong> river operations are brought <strong>for</strong>ward, the City<br />
of Boardman can more accurately respond to what the potential impacts may be. The increased<br />
knowledge and communication between the City and State and Federal agencies has also been<br />
beneficial in allowing the City to get through permitting processes associated with this water right.<br />
Although there is no direct method to measure the impact Microscopic Particulate Analysis has<br />
had in this regard, there is no doubt that it has played a role in improving communication linkages<br />
and the level of understanding between the agencies and public involved.<br />
Since April 2003, BARRY BEYELER has been Community Development Director <strong>for</strong> the<br />
City of Boardman, Oregon, where he has been employed <strong>for</strong> over 22 years. Among his<br />
responsibilities, he coordinates and reviews all land-use activities within the City and its<br />
urban growth boundary. He also oversees all planning and land-use issues relating to<br />
utilities operations (water, wastewater, storm drainage, and traffic infrastructure), natural<br />
resource impacts and environmental policy, wellhead protection, and public education.<br />
Among his professional affiliations, Beyeler is a member of the Oregon <strong>Water</strong> Resources<br />
Department Ground <strong>Water</strong> Advisory Committee, where he is currently serving a third 4-year term and is<br />
<strong>for</strong>mer Committee Chair. Beyeler received the Northeastern Oregon Sub-Section American <strong>Water</strong> Works<br />
Association Activities Award in both 1992 and 1995.