SEPA
United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Research Triangle Park NC 2771 1
EPA-600/9-79-031a
August 19/9
Research and Development
Proceedings: Second
Conference on Waste
Heat Management
and Utilization
(December 1978,
Miami Beach, FL)
Volume 1
EPRI
FLORIDA POWER I LIGHT COMPANY
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the MISCELLANEOUS REPORTS series. This
series is reserved for reports whose content does not fit into one of the other specific
series. Conference proceedings, annual reports, and bibliographies are examples
of miscellaneous reports.
EPA REVIEW NOTICE
This report has been reviewed by the U.S. Environmental Protection Agency, and
approved for publication. Approval does not signify that the contents necessarily
reflect the views and policy of the Agency, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Information
Service, Springfield, Virginia 22161.
-------
EPA-600/9-79-031a
August 1979
Proceedings: Second Conference
on Waste Heat Management and Utilization
(December 1978, Miami Beach, FL)
Volume 1
S.S. Lee and Subrata Sengupta, Compilers
Mechanical Engineering Department
University of Miami
Coral Gables, Florida 33124
EPA Purchase Order DA 86256J
Program Element No. EHE624A
EPA Project Officer: Theodore G. Brna
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, NC 27711
Cosponsors: Department of Energy, Electric Power Research Institute, Environmental Protection
Agency, Florida Power and Light Company, Nuclear Regulatory Commission, and University of
Miami's School of Continuing Studies (In cooperation with American Society of Mechanical
Engineers' Miami Section)
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
ORGANIZING COMMITTEE
Dr. .lohn Nf?al
Department of Energy
Dr. Theodore G. Hrnn
Environmental Protection Agency
Mr. Frank Swanberg
Nuclear Regulatory Commission
Dr. John Maulbetsch
Electric Power Research Institute
Mr. Charles D, Henderson
Florida Power & Light Company
Dr. Samuel S. Lee
Conference Chairman,
University of Miami
Dr. Subrata Sengnpta
Conference Co-Chairman,
University of Miami
ADVISORY COMMITTEE
Dr. C. C. Lee
I'.-S. Environmental Protection Agency
Mr. Charles If. Kaplan
U.S. Environmental Protection Agency
Dr. Mostafa A. Shirazi
U.S. Environmental Protection Agency
Dr. Richard Dirks
National Science Foundation
Dr. Donald R. F. Harleman
Massachusetts Institute of Technology
Dr. Charles C. Coutant
Oak Ridge National Laboratory
Dr. C. S. Rodenhuis
Danish Hydraulic Institute, Denmark
Dr. It. Fuchs
Consulting Engineers Inc., Switzerland
Dr. P. F. Chester
Central Electricity Research Laboratory, England
CONFERENCE SUPPORT
Arrangements:
James Poi sant
Ruben Fuentes
The School of Continuing Studies
Special Assistant:
Sook Rhee
11
-------
ACKNOWLEDGEMENTS
The Conference Committee expresses its gratitude to
the Keynote Speaker, Dr. Eric H. Willis. It also greatly
appreciates the help of the Banquet Speaker, Dr.. William
C. Peters.
This Second Conference on Waste Heat Management has
been shaped with help from the Advisory Committee members
and the Session Chairmen. Their help is gratefully ack-
nowledged.
The numerous students and faculty who have helped as
Co-Chairmen of sessions and other organizational matters
were invaluable to the Conference Committee.
The sustained interest of sponsoring organizations
made this conference possible. The scientists and admini-
strators who have provided a leadership role in nurturing
this growing field of waste heat research deserve our sin-
cerest gratitude.
The participating scientists, engineers and admini-
strators have made this conference achieve the planned
objectives of technical interaction and definition of
future goals.
Conference Committee
Miami, December, 1978
111
-------
FOREWORD
The first conference on Waste Heat Management and Utilization held in
Miami during May 9-12, 1977 was a success in terms of participation,
comprehensive technical representation and quality. A questionnaire
submitted to the sponsors and participants at the meeting indicated a
strong interest in an annual or biannual mooting. In response to this
the second comprehensive conference in the subject area is being he]d during
December 4-6, 1978. This will estabish a biannual frequency and allow
significant progress during meetings.
A perusal of the table of contents will indicate that causes, effects.
prediction, monitoring, utilization and abatement of thermal discharges are
represented. Utilization has become of prime importance owing to increased
awareness, that waste heat is a valuable resource. Sessions on Co-generation
and Recovery Systems have been added to reflect this emphasis.
This second conference has working sessions covering important topics
in the subject area. This provides an interactive forum resulting in
relevant recommendations regarding research directions.
A well balanced Organizing Committee with an Advisory Board with
international composition has brought this conference to fruition. The
sponsoring organizations include governmental and private organizations
who are active in waste heat research and development.
Samuel S. Lee
Subrata Sengupta
IV
-------
CONTENTS
WASTE HEAT MANAGEMENT AND UTILIZATION CONFERENCE
OPENING SESSION
OPENING REMARKS
Samuel S. Lee, Conference Chairman, University of Miami
WELCOMING ADDRESS
Norman Einspruch, Dean of Engineering and Architecture,
University of Miami
KEYNOTE ADDRESS 1
Eric H. Willis, Deputy Assistant Secretary for Energy
Technology, Department of Energy, Washington, D.C.
PROGRAM REVIEW
Subrata Sengupta, Conference Co-Chairman, University of Miami
GENERAL SESSION
A WASTE HEAT UTILIZATION PROGRAM 13
J. Neal, Department of Energy, Washington, D.C.
W.F. Adolfson, Booz-Allen & Hamilton Inc,. Bethesda, MD
EPA PROGRAMS IN WASTE HEAT UTILIZATION 25
T. Brna, EPA, Research Triangle Park, NC
REVIEW OF EPRI PROGRAM 38
Q. Looney, J. Maulbetsch, Electric Power Research Institute,
Palo Alto, CA
THE ENERGY SHORTAGE AND INDUSTRIAL ENERGY CONSERVATION 39
E.H. Mergens, Shell Oil Company, Houston, TX
UTILIZATION I
USE OF SOIL WARMING AND WASTE WATER IRRIGATION FOR FOREST
BIOMASS PRODUCTION 66
D.R. DeWalle, W.E. Sopper, The Pennsylvania State University
POWER PLANT LAND AVAILABILITY CONSTRAINTS ON WASTE HEAT
UTILIZATION 76
M. Olszewski, H.R. Bigelow, Oak Ridge National Laboratory,
Oak Ridge, TN
-------
Page
COOLING PONDS AS RECREATIONAL FISHERIES - A READY MADE ?6
RESOURCE
J.H. Hughes, Commonwealth Edison Company, Chicago, IL
HEAT RECOVERY AND UTILIZATION FOR GREEN BAY WASTE WATER 96
TREATMENT FACILITY
R.W. Lanz, University of Wisconsin, Green Bay, WI
MATHEMATICAL MODELING I
WHY FROUDE NUMBER REPLICATION DOES NOT NECESSARILY ENSURE 106
MODELING SIMILARITY
W.E. Frick, L.D. Winiarski, U.S. Environmental Protection
Agency, Corvallis, OR
A CALIBRATED AND VERIFIED THERMAL PLUME MODEL FOR SHALLOW H4
COASTAL SEAS AND EMBAYMENTS
S.L. Palmer, Florida Department of Environmental Regulation,
Tallahassee, FL
FARFIELD MODEL FOR WASTE HEAT DISCHARGE IN THE COASTAL ZONE 129
D.N. Brocard, J.T. Kirby, Jr., Alden Research Laboratory,
Worcester Polytechnic Institute, Holden, MA
THERMAL CHARACTERISTICS OF DEEP RESERVOIRS IN PUMPED STORAG', 139
PLANTS
J.J. Shin, N.S. Shashidhara, Envirosphere Company, New Yor .,NY
ALGORITHMS FOR A MATHEMATICAL MODEL TO PREDICT ENVIRONMENTAL 150
EFFECTS FROM THERMAL DISCHARGES IN RIVERS AND IN COASTAL AND
OFFSHORE REGIONS
J. Hauser, Institut fur Physik, Germany
F. Tanzer, Universitat Giessen, Germany
EFFECT OF SALT UPON HOT-WATER DISPERSION IN WELL-MIXED 161
ESTUARIES - PART 2 - LATERAL DISPERSION
R. Smith, University of Cambridge, United Kingdom
MATHEMATICAL MODELING II
COST-EFFECTIVE MATHEMATICAL MODELING FOR THE ASSESSMENT OF 179
HYDRODYNAMIC AND THERMAL IMPACT OF POWER PLANT OPERATIONS
ON CONTROLLED-FLOW RESERVOIRS
A.H. Eraslan, K.H. Kim, University of Tennessee,
Knoxville, TN
HEAT LOAD IMPACTS ON DISSOLVED OXYGEN: A CASE STUDY IN 187
STREAM MODELING
A.K. Deb, D.F. Lakatos, Roy F. Weston, Inc.,
West Chester, PA
vi
-------
Page
A STOCHASTIC METHOD FOR PREDICTING THE DISPERSION OF THERMAL 199
EFFLUENTS IN THE ENVIRONMENT
A.J. Witten, Oak Ridge National Laboratory, Oak Ridge, TN
J.E. Molyneux, University of Rochester, Rochester, NY
A TWO-DIMENSIONAL NUMERICAL MODEL FOR SHALLOW COOLING PONDS 214
S. Chieh, A. Verma, Envirosphere Company, New York, ,NY
UTILIZATION II
WASTE HEAT FOR ROOT-ZONE HEATING - A PHYSICAL STUDY OF HEAT 225
AND MOISTURE TRANSFER
D. Elwell, W. Roller, A. Ahmed, Ohio Agricultural Research
and Development Center, Wooster, OH
BENEFICIAL USE OF REJECTED HEAT IN MUNICIPAL WATER SUPPLIES 236
R.W. Porter, R.A. Wynn, Jr., Illinois Institute of Technology
Chicago, IL
SUPER GREENHOUSE PROJECT UTILIZING WASTE HEAT FROM ASTORIA 6 246
THERMAL POWER PLANT
R.G. Reines, Cornell University, Ithaca, NY
EXPERIENCE WITH THE NEW MERCER PROOF-OF-CONCEPT WASTE HEAT 247
AQUACULTURE FACILITY
B.L. Godfriaux, Public Service Electric and Gas Company,
Newark, NJ. R.R.Shafer, Buchart-Horn: Consulting Engineers,
York, PA. A.F. Eble, M.C. Evans, T. Passanza, C. Wainwright,
H.L. Swindell, Trenton State College, Trenton, NJ.
UTILIZATION III
WASTE HEAT RECOVERY IN THE FOOD PROCESSING INDUSTRY 266
W.L. Lundberg, J.A. Christenson, Westinghouse Electric
Corporation, Pittsburgh, PA. F. Wojnar, H.J.Heinz Company,
Pittsburgh, PA.
GENERATION OF CHILLED WATER FROM CHEMICAL PROCESS WASTE HEAT 277
J. Entwistle, Fiber Industries, Inc., Charlotte, NC
THE SHERCO GREENHOUSE PROJECT: FROM DEMONSTRATION TO 286
COMMERCIAL USE OF CONDENSER WASTE HEAT
G.C. Ashley, J.S. Hietala, R.V. Stansfield, Northern States
Power Company, Minneapolis, MN
ANALYSIS OF ECONOMIC AND BIOLOGICAL FACTORS OF WASTE HEAT 296
AQUACULTURE
J.S. Suffern, M. Olszewski, Oak Ridge National Laboratory,
Oak Ridge, TN
VII
-------
ECOLOGICAL EFFECTS I
A QUALITATIVE/QUANTITATIVE PROCEDURE FOR ASSESSING THE
BIOLOGICAL EFFECTS OF WASTE HEAT ON ECONOMICALLY IMPORTANT
POPULATIONS
J.M. Thomas, Battelle Pacific Northwest Laboratories,
Richland, WA
A REVIEW OF STATISTICAL ANALYSIS METHODS FOR BENTHIC DATA 329
FROM MONITORING PROGRAMS AT NUCLEAR POWER PLANTS
D.H. McKenzie, Battelle Pacific Northwest Laboratories
Richland, WA
FURTHER STUDIES IN SYSTEMS ANALYSIS OF COOLING LAKES: -^4
HYDRODYNAMICS AND ENTRAINMENT
K.D. Robinson, R.J. Schafish, R.W. Beck and Associates,
Denver, CO. G. Comougis, New England Research, Inc.,
Worcester, MA.
SYNTHESIS AND ANALYSES OF EXISTING COOLING IMPOUNDMENT 353
INFORMATION ON FISH POPULATIONS
K.L. Gore, D.H. McKenzie, Battelle Pacific Northwest
Laboratories, Richland, WA
COOLING TOWER PLUMES
A SIMPLE METHOD FOR PREDICTING PLUME BEHAVIOR FROM MULTIPLE 357
SOURCES
L.D. Winiarski, W.E. Frick, U.S. Environmental Protection
Agency, Corvallis, OR
MODELING NEAR-FIELD BEHAVIOR OF PLUMES FROM MECHANICAL DRAFT 37 ?
COOLING TOWERS
T.L. Crawford, Tennessee Valley Authority, Muscle Shoals, AL
P.R. Slawson, University of Waterloo, Ontario, Canada
MECHANICAL-DRAFT COOLING TOWER PLUME BEHAVIOR AT THE GASTON
STEAM PLANT
P.R. Slawson, University of Waterloo, Ontario, Canada
CRITICAL REVIEW OF THIRTEEN MODELS FOR PLUME DISPERSION
FROM NATURAL DRAFT COOLING TOWERS
R.A. Carhart, University of Illinois, Chicago, IL
A.J. Policastro, Argonne National Laboratory, Argonne, IL
W.E. Dunn, University of Illinois, Urbana, IL
EVALUATION OF METHODS FOR PREDICTING PLUME RISE FROM
MECHANICAL-DRAFT COOLING TOWERS
W.E. Dunn, P. Gavin, University of Illinois, Urbana, IL
G.K. Cooper, Mississippi State University, Mississippi
viii
-------
ECOLOGICAL EFFECTS II
ENVIRONMENTAL COST OF POWER PLANT WASTE HEAT AND 461
CHEMICAL DISCHARGE IN TROPICAL MARINE WATERS
J.M. Lopez, Center for Energy and Environment Research
Mayaguez, Puerto Rico
THEORY AND APPLICATION IN A BIOLOGICAL ASPECT 468
T. Kuroki, Tokyo University of Fisheries, Tokyo, Japan
OCCURRENCE OF HIGHLY PATHOGENIC AMOEBAE IN THERMAL 479
DISCHARGES
J.F. De Jonckheere, Laboratorium voor Hygiene,
Katholieke Universiteit Leuven, Belgium
RELATION BETWEEN ZOOPLANKTON MIGRATION AND ENTRAINMENT 490
IN A 30UTH CAROLINA COOLING RESERVOIR
P.L. Hudson, S.J. Nichols, U.S. Fish and Wildlife Service
Southeast Reservoir Investigations, Clemson, SC
EFFECTS OF A HOT WATER EFFLUENT ON POPULATIONS OF 505
MARINE BORING CLAMS IN BARNEGAT BAY, NJ
K.E. Hoagland, Lehigh University, Bethlehem, PA
R.D. Turner, Harvard University, Cambridge, MA
COOLING TOWERS I
COLD INFLOW AJTO ITS IMPLICATIONS FOR DRY TOWER DESIGN 516
_F^K. Moore, Cornell University, Ithaca, NY
AN IMPROVED METHOD FOR EVAPORATIVE, CROSS-FLOW COOLING 532
TOWER PERFORMANCE ANALYSIS
K.L. Baker, T.E. Eaton, University of Kentucky, Lexington, KY
THE IMPACT OF RECIRCULATION ON THE SITING, DESIGN, 535
SPECIFICATION, AND TESTING OF MECHANICAL DRAFT COOLING
TOWERS
K.R. Wilber, Environmental Systems Corporation
A. Johnson, Pacific Gas & Electric Co-
E. Champion, Consultant
AN INVESTIGATION INTO THE MINERAL CONCENTRATION OF 547
INDIVIDUAL DRIFT DROPLETS FROM A SALTWATER COOLING TOWER
R.O. Webb, Environmental Systems Corporation, Knoxville, TN.
R.S. Nietubicz, State of Maryland, Department of Natural
Resources. J.W. Nelson, Florida State University, Tallahassee, FL
COGENERATION
COGENERATION TECHNOLOGY AND OUR TRANSITION FROM 548
CONVENTIONAL FUELS
J.W. Neal, Department of Energy, Washington, DC
ix
-------
Page
COGENERATION: THE POTENTIAL AND THE REALITY IN A 558
MIDWESTERN UTILITY SERVICE AREA
D.M. Stipanuk, Cornell University, Ithaca, NY
W.J. Hellen, Wisconsin Electric Power, Milwaukee, WI
ALTERNATIVE APPROACHES IN INDUSTRIAL COGENERATION SYSTEMS 572
J.C. Solt, Solar Turbines International, San Diego, CA
THE ENVIRONMENT FOR COGENERATION IN THE UNITED STATES 582
F.E. Dul, Envirosphere Company, New York, NY
FUEL COST ALLOCATION FOR THE STEAM IN A COGENERATION 595
PLANT
K.W. Li, and P.P. Yang, North Dakota State University,
Fargo, ND
COOLING SYSTEMS
APPLICATIONS OF MATHEMATICAL SPRAY COOLING MODEL 619
H.A. Frediani, Jr., Envirosphere Company, New York, NY
THE DEVELOPMENT OF ORIENTED SPRAY COOLING SYSTEMS 638
D.A. Fender, Ecolaire Condenser, Inc. Bethlehem, PA
T.N. Chen, Ingersoll-Rand Research, Inc., Princeton, NJ
ONCE-THROUGH COOLING POTENTIAL OF THE MISSOURI RIVER IN 651
THE STATE OF MISSOURI
A.R. Giaquinta, The University of Iowa, Iowa City, IA
T.C. Keng, Jenkins-Fleming, Inc., St. Louis, MO
A MODEL FOR PREDICTION OF EVAPORATIVE HEAT FLUX IN LARGE 663
BODIES OF WATER
A.M. Mitry, Duke Power Compnay, Charlotte, NC
B.L. Sill, Clemson University, Clemson, NC
WORKING SESSIONS - WORKSHOPS
(1) MANGEMENT AND UTILIZATION 677
(2) ENVIRONMENTAL EFFECTS 681
(3) MATHEMATICAL MODELING 682
(4) HEAT TRANSFER PROBLEMS IN WASTE HEAT MANAGEMENT AND 534
UTILIZATION
COOLING TOWERS II
THE CHALK POINT DYE TRACER STUDY: VALIDATION OF MODELS AND 686
ANALYSIS OF FIELD DATA
A.J. Poliscastro, M. Breig, J. Ziebarth, Argonne National
Laboratory, Argonne, IL
W.E. Dunn, University of Illinois, Urbana, IL
x
-------
Page
COOLING TOWERS AND THE LICENSING OF NUCLEAR POWER PLANTS 720
J.E. Carson, Argonne National Laboratory, Argonne, IL
A DESIGN METHOD FOR DRY COOLING TOWERS 732
G.K.M. Vangala, T.E. Eaton, University of Kentucky,
Lexington, KY
EVAPORATIVE HEAT REMOVAL IN WET COOLING TOWERS 742
T.E. Eaton, K.L. Baker, University of Kentucky,
Lexington, KY
COMPARATIVE COST STUDY OF VARIOUS WET/DRY COOLING CONCEPTS 772
THAT USE AMMONIA AS THE INTERMEDIATE HEAT EXCHANGE FLUID
B.M. Johnson, R.D. Tokarz, D.J. Braun, R.T. Allemann,
Battelle Pacific Northwest Laboratory, Richland, WA
UTILIZATION IV
ENVIRONMENTAL ASPECTS OF EFFECTIVE ENERGY UTILIZATION 805
IN INDUSTRY
R.E. Mournighan, U.S. EPA, Cincinnati, OH
W.G. Heim, EEA, Inc., Arlington, VA
WASTE HEAT RECOVERY POTENTIAL FOR ENVIRONMENTAL BENEFIT 817
IN SELECTED INDUSTRIES
S.R. Latour, DOS Engineers, Inc., Fort Lauderdale, FL
C.C. Lee, EPA, Cincinnati, OH
WASTE HEAT UTILIZATION AND THE ENVIRONMENT 830
M.E. Gunn, Jr., Department of Energy, Washington, DC
THERMAL STORAGE FOR INDUSTRIAL PROCESS AND REJECT HEAT 855
R.A. Duscha, W.J. Masica, NASA Lewis Research Center,
Cleveland, OH
PERFORMANCE AND ECONOMICS OF STEAM POWER SYSTEMS 866
UTILIZING WASTE HEAT
J. Davis, Thermo Electron Corporation, Walthatn, MA
COOLING LAKES
A ONE-DIMENSIONAL VARIABLE CROSS-SECTION MODEL FOR THE 878
SEASONAL THERMOCLINE
S. Sengupta, S.S.Lee, E. Nwadike, University of Miami,
Coral Gables, FL
HYDROTHERMAL STRUCTURE OF COOLING IMPOUNDMENTS 908
G.H. Jirka, Cornell University, Ithaca, NY
HYDROTHERMAL PERFORMANCE OF SHALLOW COOLING PONDS 90.
E.E. Adams, G.H. Jirka,- A. Koussis, D.R.F. Harleman,
M. Watanabe, M.I.T., Cambridge, MA
xi
-------
TRANSIENT SIMULATION OF COOLING LAKE PERFORMANCE UNDER
HEAT LOADING FROM THE NORTH ANNA POWER STATION
D.R.F. Harleman, G.H. Jirka, D.N. Brocard, K.H. Octavio,
M. Watanabe, M.I.T., Cambridge, MA
RECOVERY SYSTEMS
COMPARISON OF THE SURFACE AREA REQUIREMENTS OF A SURFACE 931
TYPE CONDENSER FOR A PURE STEAM CYCLE SYSTEM, A COMBINED
CYCLE SYSTEM AND A DUAL .FLUID CYCLE SYSTEM
M.H. Waters, International Power Technology
E.R.G. Eckert, University of Minnesota
UTILIZATION OF TRANSFORMER WASTE HEAT 960
D.P. Hartmann, Department of Energy, Portland, OR
H. Hopkinson, Carrier Corporation, Syracuse, NY
THE APPLICATION OF PRESSURE STAGED HEAT EXCHANGERS TO 980
THE GENERATION OF STEAM IN WASTE HEAT RECOVERY SYSTEMS
M.H. Waters, D.Y. Cheng, International Power Technology
HEAT RECOVERY FROM WASTE FUEL 1000
Y.H. Kiang, Trane Thermal Company, Conshohocke, PA
AQUATIC THERMAL DISCHARGES I
SURFACE SKIN-TEMPERATURE GRADIENTS IN COOLING LAKES 1011
S.S. Lee, S. Sengupta, C.R. Lee, University of Miami,
Coral Gables, FL
FOUR THERMAL PLUME MONITORING TECHNIQUES: A COMPARATIVE 1027
ASSESSMENT
R.S. Grove, Southern California Edison Company, Rosemead, CA
R.W. Pitman, J.E. Robertson, Brown and Caldwell, Pasadena, CA
EXPERIMENTAL RESULTS OF 'DESTRATIFICATION BY BUOYANT PLUMES 1028
D.S. Graham, University of Florida, Gainesville, FL
THREE-DIMENSIONAL FIELD SURVEYS OF THERMAL PLUMES FROM 1047
BACKWASHING OPERATIONS AT A COASTAL POWER PLANT SITE IN
MASSACHUSETTS
A.D. Hartwell, Normandeau Associates, Inc., Bedford, NH
F.J. Mogolesko, Boston Edison Company
SHORT-TERM DYE DIFFUSION STUDIES IN NEARSHORE WATERS 1Q57
D.E. Frye, EG&G, Environmental Consultants, Waltham, MA
S.M. Zivi, Argonne National Laboratory, Argonne, IL
EFFECTS OF BOTTOM SLOPE, FROUDE NUMBER, AND REYNOLDS 1Q69
NUMBER VARIATION ON VIRTUAL ORIGINS OF SURFACE JETS:
A NUMERICAL INVESTIGATION
J. Venkata, S. Sengupta, S.S.Lee, University of Miami
Coral Gables, FL ..
-------
Page
ATMOSPHERIC EFFECTS
METEOROLOGICAL EFFECTS FROM LARGE COOLING LAKES 1095
F.A. Huff, J.L. Vogel, Illinois State Water Survey, IL
COMPUTER SIMULATION OF MESO-SCALE METEOROLOGICAL EFFECTS 1104
OF ALTERNATIVE WASTE-HEAT DISPOSAL METHODS
J.P. Pandolfo, C.A. Jacobs, The Center for the Environment
and Man, Inc., Hartford, CT
A NUMERICAL SIMULATION OF WASTE HEAT EFFECTS ON 1114
SEVERE STORMS
H.D. Orville, P.A. Eckhoff, South Dakota School of Mines
and Technology, Rapid City, SD
ON THE PREDICTION OF LOCAL EFFECTS OF PROPOSED COOLING 1124
PONDS
B.B. Hicks, Argonne National Laboratory, Argonne, IL
AQUATIC THERMAL DISCHARGES II
MEASUREMENT AND EVALUATION OF THERMAL EFFECTS IN THE INTER- 1131
MIXING ZONE AT LOW POWER NUCLEAR STATION OUTFALL
P.R. Kamath, R.P. Gurg, I.S. Bhat, P.V. Vyas, Environmental
Studies Section, Bhabha Atomic Research Centre, Bombay, India
RIVER THERMAL STANDARDS EFFECTS ON COOLING-RELATED POWER 1146
PRODUCTION COSTS
T.E. Croley II, A.R. Giaquinta, M.P. Cherian, R.A. Woodhouse,
The University of Iowa, Iowa City, IA
THERMAL PLUME MAPPING 1160
J.R. Jackson, A.P. Verma, Envirosphere Company, New York, NY
THERMAL SURVEYS NEW HAVEN HARBOR - SUMMER AND FALL, 1976 1167
W. Owen, J.D. Monk, Normandeau Associates, Nashua, NH
BEHAVIOR OF THE THERMAL SKIN OF COOLING POND WATERS 1191
SUBJECTED TO MODERATE WIND SPEEDS
M.L. Wesely, Argonne National Laboratory, Argonne, IL
OPEN SESSION
ALTERNATE ENERGY CONSERVATION APPLICATIONS FOR INDUSTRY 1201
L.J. Schmerzler
MINERAL CYCLING MODEL OF THE THALASSIA COMMUNITY AS 1202
AFFECTED BY THERMAL EFFLUENTS
P.B. Schroeder, A. Thorhaug, Florida International University
Miami, FL
xiii
-------
Page
SYNERGISTIC EFFECTS OF SUBSTANCES EMITTED FROM POWER 1221
PLANTS ON SUBTROPICAL AND TROPICAL POPULATIONS OF THE
SEAGRASS THALASSIA TESTUDINUM; TEMPERATURE, SALINITY AND
HEAVY METALS
A. Thorhaug, P.B. Schroeder, Florida International University,
Miami, FL
WASTE HEAT MANAGEMENT AND UTILIZATION: SOME REGULATORY 1240
CONSTRAINTS
W.A. Anderson II, P.O. Box 1535, Richmond, VA
xiv
-------
KEYNOTE ADDRESS
Eric H. Willis
IT IS COMMONLY SUPPOSED., AND SOMETIMES QUITE APPROPRIATELY,
THAT THE GOVERNMENT COLLECTS MONEY BY TAXING THE DAYLIGHTS
OUT OF ITS CITIZENS, AND THEN PROCEEDS TO SPEND IT — PERIOD,
IT IS THE SPENDING PART WHICH I WANT TO ADDRESS THIS MORNING,
PARTICULARLY AS IT PERTAINS TO RESEARCH AND DEVELOPMENT IN
ENERGY SUPPLY TECHNOLOGIES, AND THE BALANCE AND EMPHASIS THAT
HAS TO BE PLACED UPON USING THOSE NEW, AND EXPENSIVE, SUPPLY
SOURCES SO THAT EVERY 3,T,U, IS MADE TO COUNT, HOW, AS IT
i
WERE, JO GET EVERY OUNCE OF JUICE OUT OF THE ORANGE, OUR
BUDGET IN THIS AREA FOR THE FISCAL YEAR JUST STARTED IS
$2,7/B, THIS IS A LOT OF MONEY — IT'S THE TAXPAYERS DOLLAR,
YOURS AND MINE, AND DESERVES BETTER THAN TO BE MERELY SPENT —
IT SHOULD BE PRUDENTLY INVESTED, JUST AS IN ANY OTHER BUSINESS
VENTURE, THE SIMPLE QUESTION THEN IS, "HOW CAN WE BEST
INVEST THE TAXPAYERS DOLLAR SO THAT IT EXERTS THE GREATEST
LEVERAGE AND THE GREATEST RETURN IN TERMS OF ASSURING THIS
NATION OF ALTERNATIVE AND RELIABLE ENERGY SUPPLIES, AND USING
THOSE SUPPLIES EFFICIENTLY,"
CLEARLY ONE MUST, AS IN ANY ENTERPRISE, HAVE AN
INVESTMENT STRATEGY IF ONE IS TO CIRCUMVENT THE ALL TOO
MANY RATHOLES THAT ARE WILLING RECIPIENTS OF MISDIRECTED
CASH,
-------
SINCE A STRATEGY IS IN A SENSE A ROAD MAP OF HOW YOU
GET THERE FROM HERE, ONE HAS TO BE ABLE TO PREDICT WITH
SOME DEGREE OF CONFIDENCE WHAT THE "THERE" LOOKS LIKE —
IT DEFINES THE GOAL, WITHOUT A GOAL, A STRATEGY IS A ROAD
MAP TO NOWHERE,
I RATHER LIKE THE STORY OF THE COUPLE OF FELLOWS GOING
TO THE COUNTRY FAIR, ONE WAS A BIG BURLY GUY AND THE OTHER
A WIZENED LITTLE SOUL WITH GLASSES LIKE THE BOTTOMS OF COKE
BOTTLES, THEY WENT INTO THE FAIR AND CAME UPON A SHOOTING
BOOTH, IN THE SHOOTING BOOTH WAS A PIPE, AND THE PIPE HAD
FIVE LITTLE HOLES IN IT, AND OUT OF THE FIVE LITTLE HOLES
CAME FIVE JETS OF WATER, ON THE FIVE JETS OF WATER DANCED
FIVE COLORED PING PONG BALLS, WITH FIVE SHOTS FOR 50 CENTS,
THE BIG GUY WENT UP FIRST, HE HAD HIS FIVE SHOTS, AND DOWN
CAME THREE PING PONG BALLS, THE LITTLE GUY STANDING BESIDE
HIM THEN TOOK THE GUN, AND THE BIG GUY, LOOKING AT HIM WITH
A GRIN, SAID "SEE WHAT YOU CAN DO," SO THE LITTLE GUY TOOK
AIM, THE GUN WENT BANG, AND LO AND BEHOLD ALL FIVE BALLS
FELL DOWN, THE BIG GUY WAS FLABBERGASTED, AND SAID, "MY
GOD, HOW DID YOU DO IT?" HE SAID, "iT WAS EASY, I JUST
SHOT THE GUY WHO PUMPS THE WATER," THE MORAL OF THE STORY
IS THAT YOU HAVE TO GO AFTER THE KEY ISSUES THAT DRIVE THE
SYSTEM,
-------
SO WHAT DRIVES THE SYSTEM? CLEARLY AT THE MOMENT IT IS
DOMINATED BY THE WORLD PRICE OF OIL, ALTHOUGH IT COULD AS
WELL BE DRIVEN BY AN INTERRUPTION IN SUPPLIES SUCH AS WE
EXPERIENCED IN '73, WHILE WE MAY LAY ASIDE THIS DISCONTINUITY
FOR A MOMENT, WE CANNOT IGNORE IT, THE RECENT IRANIAN OIL
WORKERS STRIKE COULD HAVE HAD WORLD WIDE REPERCUSSIONS,
SUPPOSING THERE IS NO SUCH DISCONTINUITY, IF WE LOOK INTO THE
CRYSTAL BALL TO SEE WHAT THE PRICE OF OIL IS LIKELY TO BE
IN THE FUTURE, WE AT ONCE HAVE A SENSE OF FRUSTRATION, TWO,
AT LEAST TWO, SCENARIOS TEND TO EMERGE, ONE SCENARIO SAYS
THAT ALTHOUGH THERE IS PRESENTLY A WORLD GLUT OF OIL TENDING
TO HOLD DOWN PRICES, THIS COULD GIVE WAY TO AN EXCESS OF
DEMAND IN THE '83-'85 TIME FRAME, RESULTING IN SIGNIFICANTLY
HIGHER PRICES, IF THE SOVIET UNION BECAME A NET IMPORTER
OF OIL IN THIS TIME FRAME, PARTICULARLY FROM THE MIDDLE EAST,
THE POSITION COULD SIGNIFICANTLY DETERIORATE,
A MORE OPTIMISTIC SCENARIO CONTINUES THIS SURPLUS OF
SUPPLY OVER DEMAND MUCH FURTHER THAN '85 WITH ONLY MODERATE
INCREASES IN PRICES, THE MEXICAN OIL RESOURCE PROSPECT
REINFORCES THIS, THERE IS TROUBLE EITHER WAY, FOR OIL
RICHES IN OTHER PEOPLE'S LANDS DO NOT ENDOW US WITH TITLE
TO THEM, THEY HAVE TO BE PAID FOR ~ WE ALREADY HAVE ABOUT
25% OF ALL THE OIL MOVING IN INTERNATIONAL TRADE, WHICH TURNS
OUT TO BE HALF OUR REQUIREMENTS, AND THE CONSEQUENCES ARE
-------
CLEAR TO ALL WHO READ NEWSPAPERS CR EVEN CARTOONS. THE
U,S, IS NOT IMMUNE TO THE CONSEQUENCES OF UNPAID DEBTS,
TO EXACERBATE THE PROBLEM, OUR DOMESTIC SUPPLIES OF OIL
PRESENTLY AT 8M B/D ARE DECLINING, EVEN WITH KNOWN FIELDS
AND CONVENTIONAL TECHNOLOGIES, AND INCLUDING ALASKA, 1990 MAY
SEE THIS AMOUNT HALVED, ADDITIONAL DISCOVERIES MAY ONLY
ADD ANOTHER 3M B/D, MAKING /M B/D IN ALL, BY PAYING WORLD
OIL PRICE, AND EXPANDED R&D, ENHANCED RECOVERY TECHNOLOGIES
MIGHT ADD AN EXTRA 3.3M B/D TO THIS, CLEARLY, NO SIGNIFICANT
INCREMENTAL DOMESTIC CONTRIBUTION TO OUR OIL REQUIREMENTS BY
1990 IS IN THE CRYSTAL BALL, ALSO CLEARLY, OUR TRANSPORTATION
SECTOR WILL STILL BE HEAVILY RELIANT UPON LIQUID FUELS —
ELECTRIC VEHICLES WILL ONLY PENETRATE THE MARKET WHEN THE
CHARGE DENSITY, WEIGHT AND DISCHARGE/RECHARGE RATES OF
BATTERIES ARE IMPROVED DRAMATICALLY,
LET US FOR A MOMENT EXAMINE OUR OPTIONS IN TERMS OF THE
CONSTRAINTS WHICH WE, AS SOCIETY HAVE PLACED UPON THOSE
OPTIONS, I WANT TO DEAL WITH THREE SUPPLY AREAS, FOSSIL
FUELS, NUCLEAR AND SOLAR, IN PARTICULAR, WHILE WE HAVE THF
PRIVILEGE OF LIVING IN A DEMOCRACY, WE ALSO HAVE THE
FRUSTRATIONS OF SPEAKING IN CONTRADICTIONS IN OUR DIFFERENT
ROLES AS CITIZENS, OUR FRIENDLY CITIZEN STARES WITH DISBELIEr
AT HIS ASTRONOMIC PUBLIC UTILITY BILL, CURSES THEM FOR WASTE
AND RIP-OFF, TURNS HIS THERMOSTAT DOWN TO A MERE 70°, GETS
IN HIS ONE-DRIVER CAR, COMPLAINS OF THE AIR POLLUTION AND
-------
THE EXCRESIENCES OF SMOKE STACKS, MENTALLY WRITES A STINGING
LETTER TO HIS CONGRESSMAN DEMANDING THAT THE GOVERNMENT DO
SOMETHING TO CLEAR IT ALL UP, AND FINALLY ENTERS THE QUIET
OF HIS OFFICE WHERE HE DASHES OFF YET ANOTHER IRATE LETTER
COMPLAINING ABOUT THE VICISSITUDES OF GOVERNMENT REGULATION
AND THE COST OF ENVIRONMENTAL COMPLIANCE,
IT IS IN THIS CLIMATE THAT WE SEEK TO DEVELOP OUR NATIONS
GREATEST RESOURCE ~ COAL,
THE REPLACEMENT OF OIL AND GAS BURNING FOR COAL IN
UTILITY AND INDUSTRIAL PLANTS IS AN IMPERATIVE, THIS WOULD
ALLOW OIL TO BE AVAILABLE FOR THE TRANSPORTATION SECTOR AND
GAS FOR THE MORE VULNERABLE DOMESTIC MARKET, THE DIRECT
COMBUSTION OF COAL IN AN ENVIRONMENTALLY ACCEPTABLE MANNER
POSES TODAY'S DILEMMA, THE NEW SOURCE PERFORMANCE STANDARDS
MAKE STRINGENT DEMANDS ON THE PERCENTAGE REDUCTION IN SULPHUR
DIOXIDE, NITROGEN OXIDES AND PARTICULATES, IRRESPECTIVE OF
WHETHER THE COAL IS HIGH-HEAT CONTENT, HIGH SULPHUR COALS
OR LOWER-HEAT CONTENT LOWER SULPHUR COALS, SIMILARLY,
THERMAL POLLUTION IS GENERATING INCREASING AWARENESS ~ SOME
ILL INFORMED, THE PHENOMENA INVOLVED HAVE TO BE BETTER
UNDERSTOOD AND THE ABATEMENT OF THERMAL DISCHARGES FULLY
PERSUED, I NOTE THE WELL MATCHED EMPHASES ON MODELLING,
ECOLOGY AND CONTROL TECHNOLOGY IN THE CONFERENCE PROGRAM,
-------
'WE EXACTING STANDARDS THAT SOCIETY DEH^iiCS CAN BE MHT
AT A PRICE, THE DAY OF RECKONING COMES WHEN THE PERCEIVED
BENEFITS ARE INSUFFICIENT TO JUSTIFY ADDITIONAL COSTS — WE
ARE NOT YET AT THAT POINT, BUT MAY BE QUITE CLOSE,
HOW SHOULD WE APPROACH OUR R&D INVESTMENTS INTO COAL
UTILIZATION? POLLUTANTS CAN BE TAKEN OUT AT THREE STAGES —
BEFORE, DURING AND AFTER COMBUSTION, PRIOR TO COMBUSTION
WE MAY CLEAN UP THE COAL (COAL BENEFICIATION) OR CONVERT IT
INTO A LOW-SULPHUR SOLID, SUCH AS THE SOLVENT REFINED COAL~I
PROCESS, WHICH MAKES A DEASHED AND DESULPHURIZED BOILER FUEL,
ALTERNATIVELY, COAL LIQUEFACTION PROCESSES CAN PROVIDE
DESULPHURIZED LIQUIDS FOR BOILER FUEL AND PROSPECTIVELY AS A
REFINERY FEED STOCK AS A SYNTHETIC SUCCESSOR TO CONVENTIONAL
OIL, REMOVAL OF MOST POLLUTANTS DURING COMBUSTION CAN BE
ACHIEVED IN THE ATMOSPHERIC FLUIDIZED BED PROCESS AND AFTER
COMBUSTION BY FLUE GAS DESULPHURIZATION IN SCRUBBERS AND
BAG HOUSES,
COAL GASIFICATION IS ALSO POTENTIALLY ATTRACTIVE,
PARTICULARLY HIGH BTU GAS WHICH PROMISES TO BE A SYNTHETIC
SUBSTITUTE FOR NATURAL GAS AND THUS A METHOD OF SUPPLEMENTING
DWINDLING SUPPLIES OF THAT PRECIOUS FUEL TO FILL THE ALREADY
EXISTING PIPE LINE SYSTEM AND DISTRIBUTION INFRASTRUCTURE
WHICH REPRESENT A LARGE CAPITAL INVESTMENT,
-------
AT THE MOMENT HOWEVER, NEITHER Sv.ilhLilC OIL Urt GAS Fr
-------
LAST FRANTIC MOMENTS OF THE 9bTH CONGRESS, TO OBTAIN A $3 PER
BBL, TAX CREDIT FOR PLANTS UP TO 10,000 BBL, PER DAY, FOR
ENCHANCED OIL RECOVERY, PRICE INCENTIVE ALSO SEEMS THE MOST
PRUDENT APPROACH, HERE IS AN INDUSTRY, VERY DIVERSE, A LARGE
NUMBER OF INDEPENDENTS, WITH A TREMENDOUS TRADITION OF
RESOURCEFULNESS, GOOD BUSINESS JUDGMENT, AND TECHNICAL KNOWHOW,
MY EXPERIENCE OF BUREAUCRATIC WASHINGTON MAKES ME SHUDDER AT
TELLING THIS COMMUNITY HOW TO DO ITS JOB, IT IS A COMMUNITY
USED TO RISKS AND REWARDS — WITH THE APPROPRIATE REWARDS,
RISKS WILL BE ACCEPTED, I MENTION THIS BECAUSE I BELIEVE
THERE ARE AREAS WHERE DIRECT GOVERNMENT INVOLVEMENT IS A
NECESSARY CONDITION FOR SUCCESS, AND OTHERS WHERE PRICE
INCENTIVES WILL INDUCE THE PRIVATE SECTOR TO MAKE THE FRONT
RUNNING, THE NATURE OF THE MIX IS VERY MUCH INVOLVED IN
DECIDING OUR R&D INVESTMENT STRATEGY ~ OUR RESOURCES ARE
NOT INFINITE, AND HARD CHOICES WILL HAVE TO BE MADE BETWEEN
COMPETING TECHNOLOGIES, IN THE PRESENT FISCAL YEAR, CONGRESS
HAS APPROPRIATED ABOUT $803 MILLION FOR FOSSIL FUEL TECHNOLOGY
DEVELOPMENT,
CLEARLY, WITH THE COST OF PRIME ENERGY RISING AS IT WILL,
WHATEVER THE PRECISE TIME SCHEDULE, THE COST PER USABLE ENERGY
UNIT, I,E,, WATT OR BTU HAS TO BE KEPT AS LOW AS POSSIBLE,
THIS SETS A PREMIUM UPON MORE EFFICIENT ENERGY CONVERSION
PROCESSES, BE THEY ENGINES OR COMBUSTION TECHNOLOGIES, "AND
ALSO FULLY UTILIZING THE HEAT LEFT OVER AFTER THOSE PROCESSES,
-------
I HE iHRUSi UF r AH i Or I H1 i> CGm- tK'LnCli iliUo, LOT-tLo 11\ i 0
PERSPECTIVE — IT IS UNCONSCIONABLE TO SQUANDER THOSE HARD
WON ENERGY SUPPLIES IN THE WAY WE DID WHEN WE ASSUMED,
INCORRECTLY OF COURSE, THAT ENERGY CAME FREE, OUR ECONOMY
WILL JUST NOT ACCOMODATE SUCH PRODIGAL ATTITUDES, THE MORAL
EQUIVALENT OF WAR MAY BE APPROPRIATE, BUT IT IS IN THE
POCKET BOOK AND IN OUR STANDARD OF LIVING THAT THE IMPACT
WILL BE FELT, I AM THUS GREATLY ENCOURAGED TO SEE SUCH
EMPHASIS ON COGENERATION AND RECOVERY SYSTEMS ON THE
CONFERENCE PROGRAM,
IN THE NUCLEAR ARENA, WE HAVE A DIFFERENT SET OF
CIRCUMSTANCES, THE CONSTRAINTS, ALTHOUGH HAVING A MAJOR
ENVIRONMENTAL COMPONENT, ARE HEAVILY SOCIETAL AND INSTITUTIONAL
IN NATURE, THE SEASONED PRACTITIONERS OF THIS TECHNOLOGY,
HAVING SPENT THEIR LIVES ON A NEW FRONTIER AND FOR A BRAVE
NEW WORLD, FIND THEMSELVES CHARACTERIZED AS THE ARCH APOSTLES
OF THE DEVIL HIMSELF, THIS WAS A TECHNOLOGY WHOSE DEVELOPMENT
WAS ALMOST ENTIRELY FINANCED BY AND CONDUCTED BY THE
GOVERNMENT ITSELF, BY DEFINITION IT WAS "GOOD" BECAUSE THE
END PRODUCT — CHEAP, RELIABLE AND ABUNDANT ELECTRICITY —
WAS "GOOD," IT is NOT TRUE TO SAY THAT POSSIBLE DETRIMENTAL
ISSUES WERE IGNORED, BUT THEIR EMPHASES WERE NOT THOSE OF
TODAY, CONSTRAINTS ON NUCLEAR POWER DEVELOPMENT TODAY CENTER
ON THREE ISSUES, THE PROLIFERATION POTENTIAL OF THE POSSIBLE
-------
DIVERSION OF PLUTONIUM FROM REPROCESSING PLANTS, REACTOR
SAFETY AND SITING, AND THE THORNY QUESTION OF WHAT TO DO WITH
THE NUCLEAR WASTE, OR, UNTIL WE GET A GOOD SOLUTION TO THE
REPROCESSING PROBLEM, THE STORAGE OF SPENT FUEL RODS, THIS
LATTER PROBLEM, WHETHER ANOTHER NUCLEAR POWER STATION IS
ORDERED BY A UTILITY OR NOT, HAS TO BE SOLVED MERELY TO TAKE
CARE OF THE RESIDUES FROM THOSE POWER STATIONS IN OPERATION
OR BEING BUILT,
ON THE NUCLEAR TECHNOLOGY SIDE, OUR THRUSi IS DIRECTED
TO EXTENDING THE FUEL LIFE OF THE EXISTING LIGHT WATER
REACOTRS AND CONDUCTING PRELIMINARY RESEARCH INTO PROLIFERATION
RESISTANT BREEDER TECHNOLOGIES, THIS IS A LONG RANGE, HIGH
RISK, HIGH PAY OFF VENTURE WHOSE COMMERCIALIZATION AT THE
EARLIEST WILL NOT BE BEFORE THE TURN OF THE CENTURY, AS IS
ALSO THE CASE WITH FUSION POWER,
SOLAR ENERGY is CURRENTLY EVERYONES DARLING, AND is IN
DANGER OF BEING THOROUGHLY SPOILED IF ITS DIRECTION BOWS TO
EVERY WHIM OF THE DAY, THE PRESIDENT IS JUST ABOUT TO RECEIVE
A DOMESTIC POLICY REVIEW ON SOLAR ENERGY WHICH WE HOPE WILL
PROVIDE A MUCH FIRMER BASE FOR PLANNING THAN WE HAVE HAD IN
RFCENT YEARS, THE PROBLEM AT THE MOMENT IS THAT WHILE THE
SUNSHINE COMES FREE, THE HARDWARE TO HARNESS IT DOES NOT,
ALSO THE GOOD LORD RATIONS IT ABOUT 2 TO 5 KW HOURS PER
SQUARE METER PER DAY,
10
-------
DIRECTLY FROM SUNLIGHT WITH ABOUT 10% EFFICIENCY, TO REPLACE
A 1000 MWE CENTRAL POWER STATION WITH THE EQUIVALENT OUTPUT
FROM PHOTOVOLTAIC CELLS REQUIRES A COLLECTING AREA OF SOME
TWENTY SQUARE MILES, WITH SOLAR PONDS AT 2% EFFICIENCY, ONE
WOULD NEED 200 SQUARE MILES!
AT PRESENT A PRICE OF 4 TO 6 CENTS PER KILOWATT HOUR OF
ELECTRICITY IS ABOUT THE UTILITY INDUSTRY NORM AND PHOTOVOLTAICS
AT THIS COST WOULD BE COMPETITIVE, TODAY'S COST OF PHOTOVOLTAIC
ELECTRICITY RANGES FROM 50 CENTS FOR THE MOST BASIC DC SYSTEM
TO $2 PER KWH FOR AN ON~THE-GRID RESIDENTIAL DESIGN, IF
PRESENT TRENDS CONTINUE WE EXPECT THIS WILL DROP TO THE RANGE
OF 10 TO 50 CENTS BY 1982, THIS 1932 PRICE DEPENDS ON ARRAY
COSTS' OF ABOUT $2 PER PEAK WATT, WE HAVE DESIGNED OUR PROGRAM
WITH THE GOAL OF BRINGING THIS COST DOWN BY AS EARLY AS 1986
TO THE RANGE OF 6-12 CENTS IN 1978 DOLLARS FOR POWER GENERATED
ON SITE AND USED BY CONSUMERS WITHOUT ON~SITE STORAGE,
HOWEVER; THERE IS A SIDE BENEFIT HERE WHICH COULD
SUBSTANTIALLY ALTER THESE ECONOMICS, SINCE, AS WE HAVE SEEN,
PHOTOVOLTAIC CELLS ARE ONLY 10-12% EFFICIENT AT THE MOMENT
IT FOLLOWS THAT 90% OF THE INCIDENT RADIATION IS NOT UTILIZED,
HERE SURELY IS A GREAT OPPORTUNITY FOR THE RECOVERY AND
UTILIZATION OF WHAT IS ESSENTIALLY WASTE HEAT, WE GRACE IT
WITH FANCY NAMES LIKE- TOTAL ENERGY SYSTEMS, BUT THE ABILITY
11
-------
TO GET THE MOST ENERGY, ELECTRIC AND THERMAL OUT OF SOLAR
INSTALLATIONS IS GOING TO HAVE A MAJOR IMPACT ON THE MARKET
PENETRATION OF SOLAR ENERGY, SOMEWHERE ALONG THE LINE, THIS
TOPIC MIGHT BE APPROPRIATE FOR DISCUSSION IN A CONFERENCE
SUCH AS THIS,
THUS, THE WHOLE THRUST OF OUR SOLAR ENERGY INVESTMENT IS
COST REDUCTION, TO GET THIS TECHNOLOGY ACCEPTED INTO THE
MARKETPLACE AND MAKE THE CONTRIBUTION WE HOPE WILL REQUIRE A
VERY DEFINITE CHANGE IN LIFE STYLES, IT WILL ONLY DO SO
WITH A MULTITUDE OF CHEAP, SOCIALLY ACCEPTABLE, DISPERSED
SYSTEMS, ONE HAS ONLY TO DRIVE THROUGH THE TOWNS AND VILLAGES
OF ISRAEL TO SEE THOUSANDS OF PRACTICAL DEMONSTRATIONS, THE
INVESTMENT FOR THIS WILL LARGELY BE PRIVATE CAPITAL, AND WAYS
HAVE TO BE FOUND OF FINANCING SOLAR APPLICATIONS,
IN CONCLUSION THEN, WE HAVE TO WALK BOTH SIDES OF THE
STREET — ENSURING RELIABLE ENERGY SUPPLIES IN ENVIRONMENTALLY
ACCEPTABLE WAYS, AND USING THOSE SUPPLIES IN THE MOST PRUDENT
MANNER. I CONGRATULATE THE ORGANIZERS OF THIS CONFERENCE
ON AN EXCELLENT PROGRAM ADDRESSING THOSE ISSUES, AND WISH
YOU ALL WELL IN YOUR DISCUSSIONS OF THE NEXT THREE DAYS,
12
-------
A WASTE HEAT UTILIZATION PROGRAM
3T. Neal
Division of Fossil Fuel Utilization
Department of Energy
Washington, D.C.
W.F. Adolfson
Senior Scientist
Booz-Allen & Hamilton Inc.
Bethesda, Maryland
ABSTRACT
The cornerstone of the National Energy Plan is conser-
vation. While "housekeeping" and modifications to existing
equipment can result in modest energy savings, major changes
could result in substantial energy savings.
New technologies which would utilize waste heat are one way
of reducing national energy needs, conserving valuable oil
and natural gas, and mitigating the severity of the impact
of shifting to coal and coal-derived fuels on a national
scale. The recoverable energy amounts to 20 to 30 percent
of the forecast national energy consumption.
This paper describes the potential for waste heat utilization
in the four market sectors: industry, residential/commercial,
transportation and utilities. There are, however, barriers
to using new waste heat recovery technologies. The Depart-
ment of Energy technological program, which is briefly
described, is one part of a Federal strategy to overcome
some of these barriers.
13
-------
INTRODUCTION
The Arab oil embargo of 1973-74 and the subsequent qua-
drupling of oil prices made our nation acutely aware of
its dependence on fuels in critically short supply—oil
and natural gas, which currently account for about 75
percent of domestic consumption. Moreover, over 25 percent
of this usage is supplied by imports and the supply shortfall
could increase by 50 percent by the year 2000, according
to statistics compiled by the Department of Energy. To
continue a dependence on such heavily used, fast diminishing,
and increasingly expensive energy resources could jeopardize
our national security, and risk economic, social, and poli-
tical dislocations in the relatively near future. To deal
with these risks the Administration and Congress are developing
a national energy policy with four major features: to enhance
conservation and lower the rate of growth of total U.S.
energy demand; to shift industrial and utility consumption
of natural gas and oil to other abundant resources; to
develop synthetic substitutes for oil and gas; and to re-
duce dependence on oil imports and vulnerability to inter-
ruptions of foreign oil supply.
The magnitude of effort required to induce industry and
utilities to convert from oil and natural gas, our favorite
fuels, to coal and other abundant fuels and to curb the
growth in energy demand is enormous, and the U.S. will
probably remain dependent on oil and natural gas for the
foreseeable future. Thus, conservation and fuel efficiency
become an essential step in the transition strategy.
Neither a complete solution nor a substitute for developing
new resources, conservation of existing supplies is often
cheaper than production of new resources, effectively
protects the environment, and moderates the impact of
rising prices.
Industry has in the past and probably will continue to
conserve energy as is economic, without Federal participation.
While "housekeeping" and modifications of existing equipment
can result in modest energy savings (10 to 15 percent),
major process changes and capital expenditures would be
needed for more significant savings (20 to 90 percent).
One way of actively pursuing a conservation strategy is to
develop the technology to recover waste heat from industrial
processes or electrical generation. Utilizing heat rejected
from a process at a temperature high enough about the lower
reservoir temperature (commonly ambient temperature) so that
additional work can be extracted from it can make a
14
-------
substantial contribution to reducing national energy con-
sumption. Opportunity exists in all four market sectors—
industry, residential/commercial, utility and transportation-
to employ waste heat recovery technologies.
However, there are many barriers hindering adoption of heat
recovery technologies, and the Department of Energy (DOE)
program is designed to mitigate the effects of some of
these impediments.
POTENTIAL FOR WASTE HEAT UTILIZATION
Studies have estimated that the overall economy of the U.S.
is less than ten percent energy efficient, which is based
on the second-law efficiencies of various energy consuming
activities (Figure 1). The potential for energy savings is
enormous. A modest improvement in efficiency would represent
a substantial savings in oil and natural gas usage. Recent
analysis has concluded that the U.S. could expend between
20 and 50 percent less energy and still maintain overall
economic growth into the 1990's. The National Energy Plan
is based on the calculation that enough waste energy exists
to allow the U.S. to cut the growth of its total energy use
to two percent or less per year, as compared to 4.8 percent
in 1976.
Many estimates have been made for the potential energy savings
which might accrue from successful commercialization of
advanced heat recovery technology. Bearing in mind the
difficulties in separating potential energy savings due to
component improvements and recovering waste heat, it is
estimated that about 30 percent of the forecast national
energy consumption could be recovered from fuel utilization
of waste heat recovery programs and concepts (Figure 2).
In the industrial sector, increased use of recuperators,
heat pumps, and cogeneration could significantly reduce
energy consumption. Space heating in residential/commercial
applications could be supplied by incorporating total energy
or integrated energy systems. By bottoming diesel and gas
turbine engines in the transportation sector, waste thermal
energy could be converted into additional mechanical energy
(Figure 3) .
The successful commercialization of ongoing waste heat re-
covery programs have the potential for reducing our annual
expenditures for imported oil by $15 billion in 1985 and by
$48 billion in the year 2000. Full utilization of all re-
coverable waste heat energy could reduce expenditures by
15
-------
$57 billion in 1985 and by $68 billion in the year 2000
(Figure 4). These savings would have a tremendous beneficial
impact on our balance of payments posture. In addition,
waste heat recovery maximizes desired output per unit of
fuel. This will tend to reduce energy costs to American
industry and the American public. It will also alleviate
the situation that occured in the winter of 1976-1977,
when plants were shut down because of fuel curtailments,
resulting in production disruptions and worker layoffs.
Waste heat recovery has a beneficial environmental impact.
Thermal pollution of atmosphere and waterways is reduced
when rejected heat is harnessed and used. Waste heat
recovery also reduces the quantity of fossil fuel that must
be burned to achieve a given performance level/'with a
consequent reduction in air pollution.
FEDERAL STRATEGY
Industry is affected uniquely by the various barriers to
employing new waste heat recovery technology. These
constraints are technological, financial and institutional
(Figure 5).
There are several Federal programs—technological programs,
incentives, regulations—that can help overcome these
barriers (Figure 6). To illustrate the DOE approach to
solving the technological problems, a strategy was formulated
to develop heat recovery technology as an alternative source
of energy by developing the necessary technology base for
recovering and using waste heat and by demonstrating the
technical and economic feasibility of the technological
components. These goals are being accomplished by focusing
on three activities: low grade heat recovery; high grade
heat recovery; and heat exchanger technology.
LOW GRADE HEAT RECOVERY
Temperatures below 200°F are categorized as low-grade heat.
Although there are large quantities of this source, much
can be recovered only at great energy costs. This low
quality heat can be best used for space heating and agri-
cultural applications, where it does not have to be pumped
very far.
16
-------
The low grade heat recovery activity has three objectives:
to reduce energy consumption by eliminating available energy
losses, to improve the efficiency of energy converters, and
to develop low-grade heat recovery technologies as alternative
sources of energy. Effort in this activity is directed at
development of low-temperature technology by evaluating con-
ceptual designs of selected novel heat engines and recovery
devices (e.g., elastomer heat engines, nitinol and other shape
memory alloy heat engines, and advanced industrial size heat
pumps) and by improving performance of low-temperature heat
exchangers and is directed at making use of rejected heat
from Federal facilities.
HIGH-GRADE HEAT RECOVERY
Temperatures above 200°F are classed as high grade heat. The
use of waste heat in this range offers the highest potential
for immediate benefit, since high grade heat is suitable for
various bottoming and topping cycles. Bottoming cycles cc
-------
vibration, fouling and corrosion, investigation of ceramic
materials including developing ceramic m-aterials and non-
destructive tests for heat exchangers and components and
developing the ceramic heat pipe; development of a heat
exchanger for low-grade heat applications; and development of
fluidized bed heat exchangers for applications to diesel
bottoming cycle boilers and residential heat pumps.
In summary, these efforts and related applications studies
in areas of cogeneration, district heating, total energy
systems, integrated energy systems, high temperature re-
cuperation and industrial heat pumps reflect an essential
step in the national energy strategy to transition to coal
and alternative energy sources. The Government is developing
and demonstrating new technologies that give industry the
opportunity to adopt equipment which could make substantial
improvements in energy savings.
18
-------
Eneryy Consuming Activity
(Current Technology)
Second Liiw Efficiency
(Percent
Industrial Sector
Process Steam Production
Steel Production
Aluminum Production . . .
Residential/Commercial Sector
Fossil Fuel-Fired Furnace
Electric Resistive Furnace
Air Conditioning
Gas Water Heating
Electric Water Heating
Refrigeration
Transportation
Automobile
Electric Power Generation . . .
33.0
23.0
13.0
5.0
2.5
4.5
3.0
1.5
4.0
9.0
33.0
Source: John A. Beltlmy, "Alternatives to Oil and Gas Thruutjh Eneryy Management "
(AIAA/EEI/IEEE 77-lonfil 7
/8
32 32
Figure 1
-------
Recoverable Energy from Full Utilization
of Waste Heat Recovery Programs
Energy Consuming Sector
| ] Industry
| | Residential/Commercial
Transportation
Electric Power Generation
7%
7% 7%
8%
1%
F!
iu]
f 16%
^14%
IT
i :
li
Year: 1985
Year: 2000
/H '08/I 19 32
Figure 2
-------
Total Waste Heat Recovery Summary
Sector
Industrial
• Ongoing Projects
• Total
Residential/Commercial
• Ongoing Projects
• Total
Transportation
• Ongoing Projects
• Total
Electric Power Generation
• Ongoing Projects
• Total
Four Sector Total
• Ongoing Total
• Total
1985
MBDOE
1.1
3.4
1.0
3.8
0.1
0.6
0.5
6.8
2.7
14.6
QUADS
2.2
6.8
2.1
7.7
0.2
1.2
1.0
13.5
5.5
29.2
2000
MBDOE
3.2
4.5
3.1
4.8
0.4
0.8
2.1
10.8
8.8
20.9
QUADS
6.3
9.0
6.1
9.6
0.7
1.6
4.2
21.7
17.3
41.9
78-5871 30-32
Figure 3
-------
NJ
N)
Category
Total Savings of Oil (MBDOE)
Total Savings of Oil (Quads)
% Reduction in Oil Imports
$/Yr. Savings on Oil Imports*
Estimated Energy Savings
Ongoing Projects
1985
2.7
5.4
23
14.8B
2000
8.7
17.4
76
47.68
Total Recoverable
Energy
1985
10.4
20.8
66
56.9B
2000
12.5
25.0
79
68.4B
'Based on an estimated value of $15 per barrel, which appears quite conservative for the 1985-
2000 time frame.
78-5871 31-32
Figure 4
-------
BARRIERS TO WASTE HEAT RECOVERY TECHNOLOGY
TECHNOLOGICAL CONSTRAINTS
New, Unproven Technology
Technical and Economic Risks to the Production Process
Concentration of R&D Funds Within a Few Companies
FINANCIAL CONSTRAINTS
Ki
CM
Shortage of Capital
Shortage of Discretionary Capital for Energy Conservation
INSTITUTIONAL CONSTRAINTS
Energy Cost is Low Percent of Total Cost of Sales
Security of Energy Supply More Important Than Price
Uncertainty Regarding Future Prices
Low Turnover Rate of Plant and Equipment
Resistance to Changes in Production Processes
Figure 5
-------
Federal Programs To Overcome Barriers
MAJOR CONSTRAINTS TO ENHANCED ENERGY CONSERVATION
KEY:
• > STRONG IMPACT ON OVERCOMING
CONSTRAINT
O «• MODERATE IMPACT ON OVERCOMING
CONSTRAINT
BLANK = LITTLE IMPACT ON OVERCOMING
CONSTRAINT
INSTITUTIONAL & OTHER
MAJOR STRATEGIC OPTIONS
TECHNOLOGY PROGRAMS
Figure 6
-------
EPA PROGRAMS IN WASTE HEAT UTILIZATION
by
Theodore G. Brna
U.S. Environmental Protection Agency
Office of Research and Development
Office of Energy, Minerals and Industry
Industrial Environmental Research Laboratory
Research Triangle Park, North Carolina
To be presented at:
The Second Conference
on
Waste Heat Management and Utilization
Miami Beach, Florida
December 4-6, 1978
25
-------
EPA Programs in Waste Heat Utilization
ABSTRACT
Waste heat utilization which supports the Environmental Protection Agency's
thermal pollution control activities is discussed. Waste heat utilization/re-
duction is classed into three options—utilization of waste heat, in-plant
electrical generation, and integrated energy production/use facilities.
Current projects with promising environmental benefits are presented with
particular emphasis on applications in agriculture.
INTRODUCTION
Waste heat has been defined as heat rejected by a process at a temperature
enough above the ambient temperature so that additional value may be extracted
from this heat (1). In addition, the sources of the waste heat are placed in
the high temperature range if above 698°C (1200°F) and in the low temperature
range if below 282°C (450°F).
The heat rejected by power plant cooling systems is often called waste
heat and would be placed in the low temperature range using the given criterion.
However, steam-electric power plant cooling systems usually reject heat at
temperatures between 5.5 and 22°C (10 and AO°F) above intake water temperatures;
consequently, this heat, which will receive particular attention here, has a
limited potential for beneficial use. In fact, many power plant engineers
would term "waste heat" an unfortunate choice of name for this rejected heat
as it is a consequence of the second law of thermodynamics and is held to the
lowest practical value in steam-electric power production.
Natural and man-made factors impact the magnitude and quality (as indicated
by temperature above ambient temperature) of waste heat from power plants.
Increasing demr.nds for electrical energy without corresponding gains in the
efficiency of electrical energy generation effect greater quantities of rejected
or waste heat. Measures to limit adverse environmental impacts through requir-
ing off-stream or closed cycle cooling in lieu of on-stream or open cycle
cooling at steam-electric generating stations lead to slightly higher tempera-
tures and heat rates and correspondingly higher heat rejection rates. This
higher quality waste heat, plus the Nation's commitment to greater energy
conservation, provides added impetus for seeking economical methods for using
waste heat beneficially.
The amendments to the Federal Water Pollution Control Act in 1972 estab-
lished zero discharge of pollutants for 1985 as a national goal and required
technology-based effluent limitations to be considered prior to granting water
discharge permits. Subsequent effluent guidelines, issued by EPA in 1974 and
based on the use of best available technology for dissipating the heat rejected
by steam-electric generating stations, resulted in increased use of closed
cycle cooling systems, particularly evaporative cooling towers. In addition
to compliance with these guidelines, encompassing both thermal and chemical
26
-------
effluents from power plants, cooling water shortages in some areas of the
United States are an additional factor for heat rejection at temperatures
above those characteristic of once-through or open cycle cooling. Currently,
these guidelines are being updated and are expected to lead to further progress
toward achieving the zero discharge goal. These guidelines will also address
the "best conventional pollutant control technology" (BCT) which is to be
implemented by July 1, 1984 according to the provisions of the Clean Water Act
of 1977, another amendment to the Federal Water Pollution Control Act.
In addition to the act cited, Congress has passed other legislation which
may limit effluent discharges from power plants. Table 1 shows EPA's legisla-
tive mandates and lists the Safe Drinking Water Act, the Toxic Substances
Control Act and the Resource Conservation and Recovery Act. Future guidelines
for implementing these acts could impact power plant cooling system choices,
practices, and heat rejection rates and temperatures. For example, limitations
on discharges from evaporative or wet cooling towers could lead to the use of
wet/dry or dry cooling towers, which would generally reject heat at higher
temperatures and rates than wet towers. Such guidelines could make the
benefits of utilizing waste heat from power plants more attractive.
WASTE HEAT UTILIZATION
The EPA program in waste heat utilization supports the reduction of
adverse environmental impacts. Environmental benefits enhanced by waste heat
utilization include (2):
1. Reduction of pollutant generation and release for a given useful
energy requirement
2. Cost savings for pollution abatement equipment
3. Conservation of energy
4. Potential generation of revenue to offset costs of pollution abatement
5. Elimination of adverse environmental impacts related to obtaining,
processing, and supplying the energy equivalent to the beneficially-
used waste heat
Funding for beneficial uses of waste heat is provided through EPA's
Office of Research and Development (ORD), which received its first separate
budget during Fiscal Year 1977. Table 2 gives ORD funding for 17 functions or
programs, the program receiving the greatest funding being Energy Conversion,
Use and Assessment for both FY 77 and FY 78 (3). Table 3 shows ORD funding
for 10 media, with Energy receiving around 40 percent for these fiscal years (3)
Percentage of ORD support via four funding mechanisms is shown in Table 4 (3).
Various ORD laboratories and functional elements support program? in waste
heat utilization, and consequently, the total level of this support is not
provided here. With regard to funding mechanism,the projects shown in Table 5
were supported via grants and interagency transfers.
27
-------
Numerous approaches to waste heat utilization are available, with waste
heat from power plant cooling systems being considered mainly for agricultural
and aquacultural applications. Projects supported through the Industrial
Environmental Research Laboratory-Research Triangle Park (IERL-RTP) have
focused on agricultural uses of waste heat as these applications reduce thermal
discharges to water bodies, while many aquacultural applications involve
natural water bodies heated by thermal discharges. Consequently, the latter
applications do not reduce thermal pollution although they make beneficial use
of heated waters. It is also noted that guidelines have been issued for the
discharge of specific pollutants, including waste heat, in a controlled manner
from a point source to an aquacultural project as a consequence of the 1972
and 1977 amendments to the Federal Water Pollution Control Act.
The EPA program for waste heat utilization/reduction has emphasized three
options because of their high potential application in supporting the Agency's
environmental mandates. These options are (A):
1. Utilization of waste heat discharged from industrial and utility
plants for agriculture and aquaculture
2. Generation of electricity in industrial plants
3. Development of integrated energy production/use facilities which are
more energy-efficient
The first option uses heat after its discharge from a process or facility,
while the others serve to reduce waste heat through optimized design and
management. The projects listed in Table 5 concern the first option and will
be emphasized here. Option 2, which refers to cogeneration, is a responsibility
of the Industrial Environmental Research Laboratory-Cincinnati (lERL-Ci), and
several studies supported by this Laboratory will be discussed in a later
session of this conference. The last option has received limited attention
from the waste heat utilization viewpoint.
Agricultural and Aquacultural Uses of Waste Heat
Most waste heat applications supported by EPA have concerned agriculture
as this option was believed to have the greatest potential for near-term
energy savings (through 1985) while reducing thermal discharges to water
bodies or air. This waste heat, appropriate generally to electric utility
cooling system sources, was at low temperatures and represented a very small
fraction of what would be available from a modern steam-electric generating
station and had seasonal use limitations.
Since previously completed studies concerning agricultural uses of waste
heat and supported by EPA have been reported elsewhere (2), the projects
discussed here will be those currently underway or completed very recently.
Most of the active projects pertain to greenhouse applications, which appear
to be economically attractive when selected flowers and vegetables are involved.
28
-------
A demonstration grant partially funded the greenhouse study at Northern
States Power Company's Sherburne County (Sherco) Power Plant in Becker, Minne-
sota. Warm condenser cooling water from the first 680-MWe unit is supplied
from the cooling tower main to the 0.2 hectare (0.5 acre) greenhouse. The
water, nominally at 29.5°C (85°F), provides both soil and air heating through
buried pipes and surface heat exchangers. The first 2 years of the demonstra-
tion, which was completed in May 1977, was successful and led to two commercial
greenhouses being constructed on the plant site. A third year of performance
data was funded with grant-derived income that was generated from the sale of
flowers and vegetables grown during the demonstration. The grant period was
recently extended to May 1980, and a study was begun to assess the feasibility
of using a waste heat greenhouse for fish hatchery operations. Currently,
three commercial firms are using waste heat at this plant site to heat about
0.7 hectare (1.7 acres) of greenhouse area. Further details on the economic
advantages and operating aspects of the greenhouse complexes at the Sherco
Plant will be presented by Ashley et al. (5) later during this conference.
Since fiscal year 1975, the Office of Energy, Minerals, and Industry in
EPA's Office of Research and Development has coordinated a Federal Energy/Environ-
ment Research and Development program that is conducted by 17 government
agencies. Under this program EPA and TVA have jointly supported studies by
TVA at its National Fertilizer Development Center in Muscle Shoals, Alabama.
Two major tasks are being addressed in these studies, both slated for completion
by the end of next year.
Soil heating to extend the growing season of field crops is being studied
using simulated condenser cooling water temperatures from two TVA plants. The
water is supplied to plastic pipes at various row spacings and depths beneath
the surface of the soil. Using several pipe arrangements as determined from
preliminary tests, soil heating was found to increase maturation and yields of
the crops studied (bush beans, cantaloupe, corn, okra, peanuts, squash, and
watermelon) during the spring months (up to mid-June) and to not change or
slightly decrease yields (squash and watermelon) during the summer. Although
spring-grown crops were available for the market a week or so earlier, more
data are needed to determine if soil heating is economically attractive in
this southern area. Preliminary indications are that soil heating is not now
economical for this area.
The other project with TVA treats the biological recycling of nutrients
in livestock wastes through use of waste heat. Currently, fish (tilapia and
silver and bighead carp) are being grown in tanks containing three different
concentrations of swine manure for growing algae at different water temperatures.
At different frequencies equivalent to water retention times of 5, 10, and 15
days, waste slurries from these tanks are supplied daily to Chinese water
chestnuts planted in sand beds. Tilapia weight increases from 50 to over
500 grams in 6 months have been obtained, while preliminary tests showed the
harvested water chestnuts and foliage to be of excellent quality for ruminant
forage. The water from these sand beds has been low in nutrients (nitrogen
and potassium) and has a greatly improved biological quality. Studies to
29
-------
optimize fish production are continuing as are greenhouse studies to optimize
duckweed and algae production in swine manure slurries warmed with simulated
waste heat. Future efforts to complement this nutrient recycling will include
biogas production from swine manure. Biogas generation does not remove major
plant nutrients (nitrogen, phosphorous and potassium), but can significantly
reduce the biological oxygen demand of the manure and the risk of oxygen
depletion in aquatic systems fertilized with digester effluent.
The use of waste heat from a nuclear plant for greenhouse agriculture has
caused some apprehension because of the possibility of radioactive contamination.
The potential impact of the Delaney Amendment to the Food, Drug and Cosmetic
Act on greenhouse food production using waste heat from the Vermont Yankee
Nuclear Plant was assessed, and a model,"for the New England marketing of
greenhouse-grown products at this site was developed. It was concluded that
the use of surface heat exchangers between the plant's condenser cooling water
and the greenhouse heating medium would essentially preclude potential radio-
active contamination of greenhouse products. It was also concluded that the
low temperature of the condenser water (23°C or 70°F) was too low to maintain
the required greenhouse temperature. Consequently, it was proposed to use a
biogas generation facility at a dairy farm near the plant so that waste heat
from the plant could enhance biogas generation and some of the gas generated
could serve for supplemental heating of the greenhouse.
The grant to Fort Valley State College supports comparison of foliage
plants, cut flowers, and bedding plants in a conventionally heated greenhouse
with one heated with waste heat and using an evaporative system which leads to
high humidity in the greenhouse. These greenhouses became operational last
September. The performance evaluation of the environmental control system in
the research greenhouse, techniques for controlling diseases related to high
humidity operation, and assessment of plant quality in the two greenhouses are
presently underway.
Efforts to identify, develop and demonstrate heat recovery within indus-
tries emitting thermal discharges to water and air streams are also being
made. Some applications have been identified and selected for demonstration
under the industrial energy conservation program and will be discussed by
Latour and Lee (6) during the "Utilization IV" session of this conference on
Wednesday. Also, during this session some environmental aspects of effective
energy utilization by industry will be presented by Mornighan and Heim (7).
Electrical Generation by Industry
Electrical generation by industry appears to have a high potential for
reducing waste heat for the period 1985-2000 (2). The use of the more efficient
field-erected boilers in lieu of currently used package boilers seems economi-
cally conducive to cogeneration applications encompassing both process steam
generation and electrical generation for in-plant uses. Other approaches to
promising cogeneration approaches will be discussed during this conference.
30
-------
Electrical generation by industry can reduce thermal and air pollutants
through fuel conservation, reduce the capital required for a given electrical
demand, and lower the cost of producing electrical power and processing steam.
These advantages become apparent when it is noted that process steam requires
about 20 percent of this Nation's primary fuel with less than a third of this
amount being used to generate electricity prior to its process use.
Integrated Energy Production/Use Facilities
This option appears attractive for waste heat utilization/reduction over
the long-term (beyond 2000) (2).. This application would have facilities which
are operationally compatible in terms of energy form, load characteristics,
and equipment lifetime. An increase in energy utilization of 10 percent or
more would be expected through joint utility-industry ventures.
Integrated energy facilities in the United States are inhibited by various
institutional problems, the incompatibility of production and user systems,
financial risk, lack of capital, and inappropriate planning. Candidates for
integrated energy facilities include district heating, power park complexes
with process heating and electrical power from the same source, surplus electri-
cal power from industries to electrical utilities, off-peak storage and on-
peak use of energy, and selected combinations from these candidates.
Although integrated energy facilities would serve to enhance environmental
quality and energy conservation, these factors have not yet spurred development
of such facilities in the United States. Hopefully, the recent passage of the
National Energy Act will stimulate industry, utilities, and government to
concerted action in the development of integrated energy facilities. Increasing
fuel costs and energy resource limitations may also hasten the day of integrated
energy facilities as the public becomes more convinced of the need for more
efficient utilization of energy.
SUMMARY
The effective utilization of rejected or waste heat contributes to improved
environmental quality and energy conservation. Consequently, the beneficial
use of waste heat supports several objectives of the Environmental Protection
Agency. Most EPA-sponsored projects have involved agricultural applications,
but the Agency is also supporting research and development in other promising
areas. Joint programs are a major mechanism for implementing ongoing agricul-
tural waste heat utilization projects. Integrated energy production/use
facilities have the potential for a major impact for the long-term utilization
of waste heat, and will require the cooperation of the private, public, and
governmental sectors if such facilities are to be viable.
31
-------
REFERENCES
1. Rohrer, W. M., Jr. and K. G. Krieder, "Sources and Uses of Waste Heat,"
_In Waste Heat Management Guidebook, NBS Handbook 121, U.S. Department of
Commerce, Washington, D.C. February 1977.
2. Graham, D. J., "EPA Views on Waste Heat Management and Utilization," _In
Proceedings of the Conference on Waste Heat Management and Utilization,
Miami Beach, FL, May 9-11, 1977, Volume I (University of Miami, Florida).
3. U.S. Environmental Protection Agency, "Research Highlights 1977," EPA-
600/9-77-044 (NTIS No. PB 281305), Office of Research and Development,
Washington, D.C., December 1977.
4. Christiansen, A. G., "Waste Heat Utilization/Reduction," In Proceedings
of the National Conference on Health, Environmental Effects, and Control
Technology of Energy Use, EPA-600/7-76-002 (NTIS No. PB 256845), U.S.
Environmental Protection Agency, Washington, D.C., February 1976.
5. Ashley, G. C., J. S. Hietala and R. V. Stansfield, "The Sherco Experience:
From Demonstration to Commercial Use of Condenser Waste Heat," to be
presented at the Second Conference on Waste Heat Management and Utilization,
Miami Beach, FL, December 4-6, 1978.
6. Latour, S. R. and C. C. Lee, "Waste Heat Recovery Potential for Environ-
mental Benefit in Selected Industries," to be presented at the Second
Conference on Waste Heat Management and Utilization, Miami Beach, FL,
December 4-6, 1978.
7. Mornighan, R. E. and W. G. Heim, "Environmental Aspects of Effective
Energy Utilization in Industry," to be presented at the Second Conference
on Waste Heat Management and Utilization, Miami Beach, FL, December 4-6,
1978.
32
-------
TABLE 1. EPA LEGISLATIVE MANDATES
Public Law No. Title Year Passed
1. 78-410 Public Health Service Act 1944
2. 89-272 Solid Waste Disposal Act 1965
3. 91-190 National Environmental 1969
Policy Act
4. 91-604 Clean Air Act Amendments 1970
5. 92-500 Federal Water Pollution 1972
Control Act Amendments
6. 92-532 Marine Protection, Research 1972
and Sanctuaries Act
7. 92-574 Noise Control Act 1972
8. 92-583 Federal Insecticide, Fungi- 1972
cide and Rodenticide Act
Amendments
9. 93-523 Safe Drinking Water Act 1974
10. 94-469 Toxic Substances Control Act 1976
11. 94-580 Resource Conservation and 1976
Recovery Act
12. 95-95 Clean Air Act Amendments of 1977
1977
13. 95-217 Clean Water Act 1977
33
-------
TABLE 2. ORD FUNDING BY FUNCTION OR PROGRAMMATIC CATEGORY
Funding (millions of dollars)
Function or Program Fiscal Year 1977 1978
Health Effects 28.2 35.4
Ecological Processes & Effects 17.9 21.3
Transport & Fate of Pollutants 13.3 13.5
Stratospheric Modification 2.5
Minerals Processing & Manufacturing 13.1 12.5
Renewable Resources 6.2 5.6
Waste Management 14.2 28.1
Water Supply 13.8 16.3
Environmental Management 1.4 1.6
Measurement Techniques & Equipment 4.0 3.8
Characterization/Measurement Methods 8.1 6.9
Quality Assurance 4.8 5.4
Technical Support 11.3 11.9
Energy Extraction & Processing 25.4 27.6
Energy Conversion, Use & Assessment 33.4 63.9
Energy-Environmental Effects 33.1 37.2
Other 7.0 7.4
TOTAL 237.7 298.4
34
-------
TABLE 3. ORD FUNDING BY MEDIA
Funding (millions of dollars)
Medium Fiscal Year 1977 1978
Air 45.0 42.3
Water Quality 42.6 56.8
Water Supply 13.8 16.3
Solid Waste 4.1 7.6
Pesticides 8.1 9-5
Radiation 0.8 0.8
Interdisciplinary 21.1 23.2
Toxics 1.4 3.5
Energy 93.8 131.0
Program Management 7.0 7.4
237.7 298.4
35
-------
TABLE 4. ORD FUNDING MECHANISMS
Funding (Percentage of Total)
Mechanism Fiscal Year 1977 1978
Contracts 33 38
Grants 20 21
Interagency Transfers 15 13
In-House Programs 32 28
36
-------
TABLE 5. ORD PROJECTS IN WASTE HEAT UTILIZATION
(Industrial Environmental Research Laboratory-Research Triangle Park)
Project Title
X. Beneficial Uses of Warm Water
from Condensers of Electric
Generating Plants
2. Soil Heating to ^tend Crop
Growing Season
3. Optimization of Biological
Recycling of Nutrients in
Livestock Wastes via
Utilizing Waste Heat
4. Horticulture Economic and
Quality Control Study
Potential Beneficial Use of
Industrial Waste Heat for
Greenhouse Production of Bed-
ding Plants, Cut Flowers, and
Foliage Plants
Grantee/Agency
Northern States
Power Company
Tennessee Valley
Authority
Tennessee Valley
Authority
Vermont Yankee
Nuclear Power
Corporation
Fort Valley
State College
37
-------
REVIEW OF EPRI PROGRAM*
Q. Looney, J. Maulbetsch,
Electric Power Research Institute,
Palo Alto, California
*This paper was not presented.
38
-------
THE ENERGY SHORTAGE AND INDUSTRIAL ENERGY CONSERVATION
E. H. Mergens, Manager Energy Conservation
Shell Oil Company
Houston, Texas U.S.A.
ABSTRACT
The industrial sector of the United States economy has been the mainstay
of the nation's energy conservation program. Two industries, Petroleum
Refining and Chemical and Allied Products, have made large contributions
to the reduction of waste heat and lower energy consumption. The perfor-
mance of these two industries are compared to targeted goals for energy
conservation. Specific areas of waste heat recovery are discussed and
examples of several actual industrial installations are presented.
INTRODUCTION
Good Afternoon ladies and gentlemen, it is my pleasure to come here and
join in a discussion of energy conservation in United States industry,
particularly in the petroleum and chemical industries.
The petroleum refining and the chemical industry are two of the largest
energy users in the country each using about (3) quadrillion BTU's of
energy, sometimes called energy quads. In total, industry uses about 37
percent of the nation's energy.
Energy conservation is not a new activity in industry. Indeed much of
the technological innovation in industry has been directed at using the
BTU of energy as many times as possible in the manufacturing process.
Analysis of Past Trends
In economic analysis of this trend toward increased energy utilization,
one of the commonly used indicators is the ratio of energy used to the
gross national product (Slide 1). The general slope in the last 25 to
30 years has been around (5) to (7) tenths of a percent per year reduction
in energy per constant dollar of gross national product. In using his-
torical GNP as a measure, care must be taken, however, to recognize that
the character of GNP has changed over the years. Government spending,
in the form of transfer payments, and a general increase in services
activity have distorted some of the real changes which have taken place.
39
-------
We believe it is better to analyze some individual components of the
economy to get a better picture of what is actually happening. The
next series of slides shown statistical data for various segments
of the economy. The years selected for comparison were 1968, 1972
and 1977.
Activity indicators such as population, commercial employment, vehicle
registrations and industrial production (as measured by the Federal
Reserve Board Index), all increased over the period. (Slide 2)
Energy use also grew as shown in the next slide (3) increasing from
63 quads to 76 quads. The growth was in residential, commercial and
transportation sectors, while industrial use stayed relatively
constant.
Comparing the percentage change of both the demographic items and
the energy consumption figures in the next two slides, reveals the
pattern of what is happening in the economy. Over the whole period,
(Slide 4) residential energy consumption rose four times as fast as
the activity indicators did, while commercial and transportation
consumption nearly matched activity increases. During the same
period, industrial energy consumption rose only one fifteenth as
fast as production increased.
Even after the Arab Embargo, (Slide 5) the residential energy used
increased 1.5 times the rate of the activity indicator, commercial
matched increases and transportation's rate of increase slowed
considerably, while industry produced more goods using less energy.
The reasons for these patterns are not hard to imagine. Industry is
price sensitive. In a competitive economy, the effect of the Arab
actions on crede oil price and, the effect of declining gas supplies
on the intrastate market price, have accelerated the activity of
industry's cost managers to accomplish energy conservation. Transpor-
tation energy consumption responds more slowly as equipment modifica-
tions and change-out are only part of the picture. Lower speed
limits, where enforced, have some effect, as does car-pooling, etc.,
but by and large American driving habits aren't changing rapidly.
Finally, in the residential and commercial area, the consumer is not
as rigorous a cost manager. Moreover, this sector of the economy
has been buffered by price control mechanisms such as those on
interstate natural gas or petroleum products. Until this consumer
gets the right price signals we can expect a continued disproportional
growth in energy consumption.
40
-------
Current Performance of Industry
From the foregoing, it is evident that industrial energy conservation
is a significant factor in the economy's improved energy utilization,
averaging about one percent per year since 1968 and 2.8 percent per
year in the post-Arab Embargo period.
The Federal Government has also reacted since the Arab Embargo
in the area of industrial energy conservation. The Energy Policy
and Conservation Act of 1975 required establishing voluntary goals
for the ten most energy intensive industries in the United' States.
The goal for petroleum refining was set at 20 percent reduction,
in energy per barrel intake, using 1972 as a base year and to be
accomplished by January 1, 1980. The chemical industrial goal was
a 14 percent reduction in energy per pound of product over the same
time frame.
Progress toward the goal has been steady over the last five years.
The petroleum industry is processing now nearly 20 percent more
crude while using about the same fuel as in 1972. (Slide 6A) After
making suitable corrections to today's operations versus 1972 there
is a net savings of 16.5 percent as reported by the American Petroleum
Institute. The chemical industry progress was equally impressive
producing over .3 trillion more pounds at comparable fuel usage.
Compared to 1972 the conservation reported by the Manufacturing
Chemists Association is 14.3%. (Slide 6B)
The goal for petroleum industry was established for the government
by the consulting firm, Gordian and Associates based in New York
City. The components for the target are shown in the next slide(7A).
For simplicity, the 19.4 percent was rounded up to 20 percent by
the government in setting the goal. The items enumerated are typical
areas for investment in improved energy efficiency.
The API recently conducted a survey of members, which got a partial
response, but gave us some insight into how the industry is actually
achieving their savings. Next slide (7B). It is interesting to
compare the actual with the predictions of the consultants in this
slide. Comparing with the Gordian estimate, we find that a fair
amount of activity occurred in all areas. Industry found more
insulating projects and waste heat/power recovery projects than
predicted. However, it looks like industry may find less potential
in heater and heat exchange investment than the consultant predicted.
41
-------
The industry is also respond i ng to the change in fuel, supply,
particularly the dec. 1 i ne i.n natural gas availability. The next
slide (8A) compares the average refinery fuel composition in 1972
and in 1977 as reported to the AI'l. Note how distillate, residual,
refinery gas, and electricity have all increased to make up for the
short fall in natural gas. A similar trend is shown in the chemical
industry slide (8B).
Shell Oil Program
Shell Oil Company participates in this voluntary reporting program
for manufacturing industries and reports its results through the
industry trade associations. (API in the case of refining and MCA
for chemical). Our company's program encompases more than just
manufacturing, however. The next slide (9) shows the organization
of our company energy conservation team of which I am chairman.
Through the efforts of functions represented on this team more than
11 million barrels per year fuel oil equivalent arc hcing saved in
Shell operations by actions taken since 1972. Last year, over 17
million dollars were invested in energy conservation projects, and
we continue to find additional opportunities for sayings. As of the
end of 1977, savings by the major sectors of the company amounted
to: 12 percent in the oil and chemical manufacturing operations; 15
percent in transportation and distribution of the product; and 16
percent in our exploration and production activitics. Manufacturing
plants and refineries play a significant role i 11 tlu'Se savings as
they use over 70 percent of the corporate energy, but contributions
have been made by each of the functions represented on this chart.
Employee carpools, heating and lighting reductions, more efficient
product transportation systems, and computer control in many areas.
In the final minutes of this talk, I'd like to show a few specific
examples of tilings we are doing i.n our manufacturing plants. First,
in several locations we carry out both oil and chemical, processing
at the same site. One. of the characteristics of oil processing
compared to chemical processing is shown in the next slide (10).
Oil operations are typically process heat related where we heat
oil to relatively high temperatures, while chemical operations are
steam heat related-distilling at lower temperatures. This allowed
us to do the following at our Norco Louisiana Manufacturing Complex.
This slide (11) shows a heat medium circulating loop. It takes
waste heat from oil operations -• coker, distiller, hydrocracker and
cat cracker and uses it to heat tit earn boiler feed water and operate
a chemical distilling column. This single project saves nearly a
half million barrels of fu.el per year, and the' loop spans more than
two miles as it loops around the processing complex.
42
-------
This is an example of the way high level heat can be cascaded down
to low level heat users. It is a demonstration of the 2nd Law of
Thermodynamics efficiency at work in a practical manner. The 2nd
Law consideration has been touted by theorists as a means of greatly
increasing our conservation progress. But as demonstrated here it
takes the right set of circumstances to make it practical. Other
physical limitations have to be met as well. The chemistry,
metallurgy, reaction kinetics, and even the geography involved, can
influence how closely we can come to a theoretical goal.
A similar caution should be expressed here concerning many of the
recent studies and the National Energy Plan approach to cogeneration.
The potential for cogeneration may have been greatly oversold as an
conservation tool. The potential for large amounts of byproduct
electricity while supplying industrial heat demand is not likely.
In the first place, there are few areas where large industry heat
demands are concentrated into the required geographical area. Secondly,
the estimators of this potential greatly underestimate the amount of
mechanical horsepower most large industrial complexes generated from
the steam raised. A recent survey indicates from two to four times
as much energy is used as mechanical horsepower rather than electrical
power from the steam generated in industry. Thus, most cogeneration
projects under consideration now by industry, while large in steam
generation capacity, provide only modest electbical generation (say
30 to 80 MW) which is mostly used up within the industry's own plant.
Finally in many industrial complexes, the installed capital and
existing plant utility balances preclude a radical change over to
cogeneration with current economics.
A conservation technique which can be applied in our operations,
is the use of a heat pump to help heat distillation columns. If
the right circumstances are available the heat pump to help boil
the liquids in a distillation column. The heat pump can reduce
the amount of energy required to boil the liquid by 15 to 30%.
The next slide (12) shows how an existing or new distilling column
can be equipped with this system. The main variables are the
temperature difference between the top and the bottom of the column,
the pressure level of the column, the size of the heat requirement,
and finally the temperature difference in the reboiler and the cost
of the horsepower needed to provide that temperature difference.
The major considerations to make the system economic, are the column
temperature difference and the size of the heat load. As can be
seen (Slide 13) this type of heat savings can be very attractive at
today's fuel costs. Shell has five such installations operating or
under construction now.
43
-------
As shown in a previous slide (10), a large portion of our fuel is
consumed to produce process heat by firing various furnaces. Today
we are fitting out many of these heaters with equipment which preheats
the combustion air making the fuel firing process much more efficient.
The schematic on the next slide (14) shows a before and after case,
which will save about 35,000 barrels per year of fuel while doing
the same processing job.
Another slide (15) shows a real example of such a combustion air
preheater on our crude oil heater furnace in our Martinez, California
Manufacturing Complex. This slide shows the elaborate duct work
needed to take hot stack gases out of the top of the heater and pass
it by fresh cold air coming into the heater with the fuel. This
project increases the firing efficiency from about 80 percent up to
91 percent and saves nearly 60,000 barrels of fuel per year. All
told we now have 8 recently completed air preheaters and several in
design which will save 450,000 barrels per year of fuel. This type
of installation, while common on most boilers has great promise in
the process heater application.
Finally, not to overlook the simplest way to conserve, we have even
put insulating covers over the clean out flanges of our hot oil heat
exchanges to save several thousand barrels per year of fuel. The
slide (16) depicts such an installation. A unique engineering problem
here had to do with the bolts which hold on the covers. When insulated,
the bolts got hotter and tended to elongate and this loosened the
cover and caused leaks. One of our engineers solved that problem by
using washers which were dish-shaped. The washers flattened out
when tightened. Then as the bolt lengthened with heat, the washer
flexed and kept the cover tight. So now we have heat recovery without
leaks. Of course, there are many additional items which comprise the
total company program and these are only a few examples.
Summary
In summary I'd like to reiterate several points. We have seen
that free market forces in the industrial area have been effective
in causing conservation. Industry and in particular the oil and
chemical industry have and will continue, to respond to the need to
conserve increasingly scarce fossil fuels. Considerable progress
has been made toward the 1980 goal and we expect to continue beyond
that. As we have in the past, we will operate our facilities as
efficiently as we know how applying theory where practical and using
the technical innovations that have characterized these industries
since their start.
Thank you for your attention and I'd be pleased to answer any
questions if I can.
44
-------
REFERENCES
American Petroleum Institute - Semi-Annual Energy Conservation Report
to Department of Energy - April 28, 1978
Battelle Columbus Laboratories - Draft Target and Support Document on
Developing an Maximum Energy Efficiency Improvement Target - SIC 28
Chemical and Allied Products - Washington, D.C. Federal Energy
Administration - July 1, 1976
Department of Energy - Voluntary Business Energy Conservation Program -
Progress Report No. 6 Department of Energy - Assistant Secretary for
Conservation and Solar Applications - Office of Business Assistance
Programs - April 1978
Department of Commerce - Volume Business Energy Conservation Program -
Progress Report A Office of Energy Programs - Federal Energy Adminis-
tration Office of National Energy Conservation Programs - November
1976
Gordian Associates, Inc. - An Energy Conservation Target for Industry
SIC-29 Washington, D.C. Federal Energy Administration - June 25, 1976
Gordian Associates, Inc. - The Data Base: Potential for Energy Con-
servation in Nine Selected Industries. Volume 2: Petroleum Refining
Conservation Paper 10 Washington, D.C. Federal Energy Administration -
June 1974
Grace, E. C., Born, Inc., Fidoe, H. L., Shell Oil - U. S. Refineries
Conserve Energy - Hydrocarbon Processing - May 1976, pp 120-121
Manufacturing Chemists Association - Semi-Annual Energy Conservation
Report by Department of Energy - May 1, 1978
Mergens, E. H., Shell Oil - How Shell Conserves Energy - Hydrocarbon
Processing - July 1977, pp 120-123
Pringle, W. H., Jr., Potential for Energy Conservation in Industrial
Operations in Texas Report NSF-RANN 74-231 NTIS Springfield, VA -
November, 1974
Sittig, Marshall - Practical Techniques for Saving Energy in Chemical,
Petroleum and Metals Industries - Noyes Data Corp., Park Ridge, NJ
Lib. Cong. No. 77-71855 - 1977
45
-------
U.S. ENERGY - GNP RATIO (1972$)
i
o
o
0
B
T
U
75
70
65
60
55
50
45
40
1945
I960
1955
1960
1965
TIME
1970
1975
1960
1985
1990
BCB
1-V78
CP/EE
-------
UNITED STATES
ACTIVITY INDICATORS
YEAR 1968 1972 1977
POPULATION, MILLIONS 200,7 708,8 216,9
COMMERCIAL EMPLOYMENT, MILLIONS 39,9 45,7 53,3
VEHICLF REGISTRATIONS, MILLIONS 101,0 118.5 143,0
INDUSTRIAL PRODUCTION, FRB INDEX 1,063 1,197 1,37]
47
-------
UNITED STATES
cNc: ;Y USE
QUADS '
YEAR 1968 1972 1977
RESIDENTIAL
COMMERCIAL
TRANSPORTATION
INDUSTRIAL
TOTAL
12,8
8,2
14,7
27,1
63,1
J5.4
9,0
17,3
29,5
71,2
17.3
10,7
20,0
78,0
76,0
1) QUADRILLION BTu'S
48
-------
UNITED STATES
•CHANGES
1968 vs 1977
RESIDENTIAL
COMMERCIAL
TRANSPORTATION
INDUSTRIAL
ACTIVITY
J NCR E AS£
8%
33%
m
29%
ENERGY
INCREASE
36%
,29%
36%
907
49
-------
UNITED STATES
CHANGES
1972 vs 1977
RESIDENTIAL
COMMERCIAL
TRANSPORTATION
:NDUSTRIAL
ACTIVITY
1RCREAS£
17%
21%
15%
ENERGY
:NCREASE_
16%
50
-------
REFINING INDUSTRY
OPERATING STATISTICS
1972 1977
REFINERY INTAKE, MMB/D 12.6 15.1
ENERGY USE, QUADS'!] 3.0 3.0
WEIGHTED AVG, MBTU/BBL 650 546
1] QUAD = BTU X 1015
SOURCE: API
51
-------
CHEMICAL INDUSTRY
OPERATING STATISTICS
1972 1977
PRODUCTION, POUNDS X 109 400 650
ENERGY USE, QUADS1-' 3<4 3'3
WEIGHTED AVERAGE, 1972
MBTU/LB 8,3 7.1
1] QUAD = BTU X 1015
SOURCE: MCA
52
-------
GORDIAN ASSOCIATES
ENERGY CONSERVATION MEASURES
SIC 2911
% OF ENERGY SAVINGS
CATEGORY 1980 VS 1972
PROCESS HEATERS 3.0
BOILERS 1.2
INSULATION 1.0
HEAT EXCHANGE 3.9
PROCESS CONTROL/REVISIONS 0.8
WASTE. HEAT/POWER RECY 0.8
LOSS CONTROL 1 -0
HOUSEKEEPING 7.1
OTHER . °-6
19.4
53
-------
MEASURES
SIC 2911.
SOURCE: PARTIAL SURVEY OF
API REPORTING GO'S. 54
% OF ENERGY SAVINGS
CATEGORY. .....1.97JLYLJ1Z1
PROCESS HEATERS J 2.9
BOILERS
INSULATION 1.5
HEAT EXCHANGE 1 .3
PROCESS CONTROL/INSTRUMENTS 0.9
WASTE HEAT/POWER RECY 2.3
HOUSEKEEPING 6.6
OTHER 1.1
16.5
-------
GORDIAN
PRORATED
3.6
0.9
3.3
0.7
1 .5
6.0
0.5
16.5
55
-------
FyEL_COMPOSITION
PERCENT
1972
1977
CRUDE
DISTILLATE
RESIDUAL
LPG
NATURAL GAS
REFY GAS
COKE
COAL
PURCHASED STEAM
PURCHASED ELECTRICITY
0.1
0.5
8.6
1 .4
31.4
35.5
14.6
0.1
1.1
6.7
0.1
0.9
10.1
0.9
24.9
3-9.7
14.2
0.1
1.3
7.8
SOURCE.1 AP
56
-------
CHEMICAL INDUSTRY
FUEL COMPOSITION
PERCENT
1972 1977
DISTILLATE 0,8 2.3
RESIDUAL 5,6 7.3
LP6 0.6 0.4
NATURAL GAS 47.3 39.5
PROCESS GAS 8.6 8,7
PROCESS LIQUID 1.7 3,5
PROCESS SOLIDS 0.4 0,2
COKE 0.2 0.2
COAL 9.4 8.7
PURCHASED STEAM 3.9 3.6
PURCHASED ELECTRICITY 21.5 25.6
SOURCE: MCA
57
-------
SHELL OIL COMPANY
CORPORATE ENERGY TEAM
EXPLORATION
AND
PRODUCTION
MANUFACTURING
OIL AND CHEMICALS
RESEARCH
EMPLOYEE
RELATIONS
PUBLIC
AFFAIRS
CORPORATE
ENERGY
CONSERVATION
TEAM
FINANCIAL
INFORMATION
AND
COMPUTER SERVICES
PURCHASING
AND
GENERAL SERVICES
LEGAL
REPORTS
TO
TRANSPORTATION
AND
DISTRIBUTION
MARKETING
GENERAL EXECUTIVE
58
-------
REFINERY/CHEMICAL ENERGY CONSUMPTION
PROCESS HEAT, STEAM GENERATION,
AND PURCHASED ELECTRICITY
D
LLJ
S
Z)
(/)
Z
O
o
CE
LU
Z
LU
O
h-
UL
O
UJ
O
<
Z
UJ
O
cc
70 r-
60
50
DU
40
30
20
10
PROCESS HEAT
Y7/
V
v
V
'/,
'//
X
Z
/
STEAM GENERATION
PURCHASED
ELECTRICITY
59
-------
HEAT TRANSFER LOOP
EXAMPLE OF MULTIPLE UNIT HEAT INTEGRATION
HEAT PICKUP
35 MMBTU/HR
( 53 MMBTU/HR
32 MMBTU/HR
82 MMBTU/HR
350 F
200 F
COKER PRODUCTS
330 F
200 F
CRUDE DISTILLATION PRODUCTS
AND REFLUX
330 F
200 F
HYDROCRACKER REFLUX
CO HEATER STACK GASES
BOILER FEED
WATER
CHEMICAL
PLANT
FRACTIONATOR
420 F
290 F
HEAT RETURN
129 MMBTU/HR 74 MMBTU/HR
60
-------
HEAT PUMP REBOIUNG
EXISTING
CONDENSER
•.>] TOPS £
IACCUM!
TOP
PRODUCT
FEED
EXISTING
REBOILER
I
BTM
PRODUCT
••».*••.-.».t.»» M». .*i 1.^4.
'COLUMN
PREHEATER
CONDENSER
COMPRESSER
H0
MOTOR!
REBOILER
CONDENSER
NEW
EX I STING
61
-------
60 |—
ECONOMIC RETURN
ON
REBOIL HEAT PUMPS
40
GO
01
cc
ID
h-
UJ
CC
LL
O
UJ
30
10
100MM BTU/HR
50MM BTU/HR
30
40
50
60
70
80
COLUMN TEMPERATURE DIFFERENCE. °F
62
-------
COMBUSTION
AIR PRE-HEAT
OIL
100°F
600° F
1000° F
A!//
v '/
AIR 70°F
OIL
600° F
OIL
100 F
FUEL
1000 BARRELS PER DAY
350° F
AIR 70°F
1050°F
OIL
600" F
AIR500'JF
FUEL
900 BARRELS PER DAY
63
-------
64
-------
-------
Use of Soil Warming and Waste Water Irrigation
for Forest Biomass Production
by
D. R. DeWalle and W. E. Sopper
School of Forest Resources
and
Institute for Research on Land and Water Resources
The Pennsylvania State University
University Park, PA, 16802, U.S.A.
ABSTRACT
The opportunities for increasing biomass production from hybrid poplar
plantations through the use of soil warming and waste water irrigation were
studied in central Pennsylvania. Plots were established to determine the
growth and development of hybrid poplar at three planting densities and
three treatments: (1) soil warming plus waste water irrigation, (2) waste
water irrigation only, and (3) no heat or irrigation. Soil heat was
supplied to the soil from a buried hot-water pipe network. Spray irrigation
of treated municipal waste water was conducted at a rate of 5 cm per
week. In each of the three treatments the growth and development of
sycamore, sweetgum, yellow-poplar and cottonwood were also evaluated at
the intermediate tree density. Biomass production of hybrid poplar was
doubled during the first growing season by waste water irrigation.
Sycamore and sweetgum showed a positive height growth response to soil
warming.
Paper delivered at the Waste Heat Management and Utilization Conference,
4-6 Dec., 1978, Miami Beach, FL.
66
-------
INTRODUCTION
Soil warming by circulation of condenser cooling water through pipe
networks buried in the soil has been proposed as a method for utilization
and dissipation of waste heat (DeWalle and Chapura, 1978; DeWalle, 1977). Con-
sidering the tremendous quantities of waste heat produced in power generation,
large acreages of land could be artifically heated in this manner. Although
yields of various vegetable and forage crops have been shown to increase
in response to soil warming, virtually no information exists on the response
of woody vegetation. Because of the potential usefulness of forest biomass
as an alternate, low-sulfur fuel source in electric power production, the
feasibility of using soil warming to stimulate the growth of forest biomass
was inves t igated.
Treated municipal waste water is another by-product of our modern
society which has proven useful in stimulating the growth of woody plants
(Sopper, 1975). Spray irrigation of treated municipal waste water permits
the natural recycling of nutrients through uptake by the biological system
and the recharge of potable water. If used in conjunction with soil warming
waste water irrigation could help to prevent excessive soil drying and
provides nutrients for accelerated plant growth. A moist soil also has a
higher thermal conductivity for more efficient heat dissipation from the
buried pipes.
Intensively-managed, short rotation, forest plantations can supply
significant amounts of biomass for fuel. Blankenhorn, et al. (1977) have
estimated that forest land within a 56.3 km radius of Renovo, Pennsylvania
could provide enough fuel to easily sustain a 100 MW electric generating
facility. Intensive management of this forest land could increase the
supply of biomass considerably. Bowersox and Ward (1976) have shown that
two harvests of hybrid poplar trees in seven years (poplar trees sprout
when cut) are capable of producing an average yield of 11.2 ovendry metric
tons of wood and bark per hectare per year in Pennsylvania. Even greater
yields of woody biomass are desirable to improve opportunities for fuel
production.
The results of a study of the effects of soil warming and waste water
irrigation on forest biomass production after the first growing season are
reported in this paper. Plots were established to determine growth and
development of hybrid poplar trees at three densities with 1) soil heat
plus waste water irrigation, 2) waste water irrigation only and 3) no heat
or irrigation. In addition, the effects of each treatment on sycamore,
cottonwood, tulip poplar and sweetgum trees were also evaluated at one
planting density.
STUDY AREA
The study was conducted 4 km north of The Pennsylvania State University
campus at State College in Centre County at the site of the soil warming
research facility. The site is located in the bottom of a small valley
where the soil is classified as a Chagrin gravelly, sandy loam. The climate
67
-------
of State College is a composite of a dry midwestern continental climate
and the more humid eastern coastal climate. Mean annual air temperature
is 10.0°C with a monthly average of -1.8°C in January and 22.2°C in July.
Average dates of the first and last freezing temperatures are 12 October
and 29 April, respectively. Annual precipitation averages 100 cm and is
evenly distributed throughout the year.
METHODS
Soil was artificially heated by continuously circulating hot water
through 5-cm diameter polyethylene plastic pipes buried at 26-cm depth
and 61-cm spacing. Circulating inlet water temperatures were 38°C in
June-August, 32°C in September-November and March-May and 27°C in December-
February. Average soil temperatures in each treatment at the 30.5- and
15.2-cm depths during the first growing season are shown in Figure 1.
Soil warming with waste water irrigation increased soil temperatures, while
waste water irrigation alone reduced soil temperatures relative to the
control area at both depths. Details of the soil warming and spray irri-
gation facilities have been previously described (DeWalle, 1977).
Waste Water Irrigation
Municipal waste water from State College, Pa. was sprayed above the
trees at a design rate of 5 cm per week with two applications per week.
A total of 86 cm of waste water was applied during the growing season from
7 June to 27 October, 1978. Irrigation was terminated on 2.1 October until
spring. Precipitation during this same period was 43 cm. Average concen-
trations of dissolved solid in the waste water for the growing season are
given in Table 1. Average pH of the waste water was 8.4. Based on the
amount of effluent applied and the average concentrations in Table 1, the
irrigation was equivalent to 144 kg/ha of nitrogen and 60 kg/ha of phosphorus
as P205.
TABLE 1
AVERAGE CHEMICAL COMPOSITION OF MUNICIPAL WASTE WATER
APPLIED DURING FIRST GROWING SEASON
., .. . Concentration Concentration
Constituent , ... Constituent , ...
(mg/£) (mg/£)
Ortho-P
Total-P
N03-N
NH4-N
Org-N
Total N
K
Cl
3.27
3.14
7.3
9.27
0.64
17.21
O.01
57.1
Hg
Cu
Zn
Cr
Pb
Co
Cd
Ni
0.6
0.07
0.13
0.02
0.06
0.01
0.001
0.03
68
-------
30
o
~ 25
UJ
(T
UJ
20
15
UJ
h-
=! 10
o
en
SOIL HEAT + EFFLUENT
CONTROL
15.2 cm DEPTH
(a)
MAY JUNE JULY AUG. SEPT. OCT.
o
o
UJ
o:
35
30
25
20
o:
UJ
a. 15
2 ID
UJ
H
_J 10
O
en
SOIL HEAT + EFFLUENT
30.5 cm DEPTH
(b)
MAY JUNE JULY AUG. SEPT. OCT.
1978
Figure 1. Comparison of soil temperatures among treatments at the
a) 15.2- and b) 30.5-cm depths.
69
-------
Soil moisture contents were increased substantially by the waste
water irrigation. Moisture contents at 15.2-cm and 30.5-cm depths were
gravimetrically determined on the day prior to irrigation at 14 locations
in each treatment (Table 2). Moisture contents on the irrigated plots
TABLE 2
AVERAGE SOIL MOISTURE CONTENT ON 28 SEPTEMBER, 1978
AT TWO DEPTHS FOR EACH TREATMENT
Soil Moisture Content (% wt.)
Treatment
15.2-cm depth 30.5-cm depth
Soil heat and
waste water 13.1 12.8
Waste water only 10.9 15.2
Control 8.0 8.2
were from 2.9 to 5.1% by weight greater at the 15.2-cm depth and from
4.6 to 7.0% by weight greater at the 30.5-i.iii depth than on the control area.
The heated area was drier at the 30.5-cm depth than the area receiving
effluent only due f^ '•he drying influence of the heated pipes. At the
15.2-cm depth rfr»d was reversed with the effluent-only having drier soil.
Plantation Establishment
Hybrid poplar plantations (Pooulus spp.) were established in 3 m
square subplots with 61 cm betv • ">"s and either 12, 24 or 48 cm spacing
between plants in a row. These sp gs gave 0.09, 0.18 and 0.36 m^ of growing
space per tree. Three replicate SUL Lots of each spacing were randomly
located in each treatment area. Hybrid poplar cuttings were planted during
the period 11-16 May, 1978. Cuttings were obtained from a Pennsylvania
state tree nusery and were a mixture of clones.
At the same time, single subplots were established in each treatment
area with cottonwood (Populus deltoides, Bartr.), yellow-poplar (Liriodendron
tulipifera L.), sweetgum (Liquidamber styraciflua L.) and sycamore (Platanus
occidentalis, L.) trees at the 24-cm spacing. Cottonwood was established
from cuttings obtained locally, but the sycamore, sweetgum and yellow-poplar
were planted as seedlings. Sycamore and tulip-poplar seedlings were purchased
from a Pennsylvania state nursery, while sweetgum seedlings were purchased
from a commercial nursery in Tennessee.
Competition from weeds was con'- oiled by Kv/wing between rows and
hand cultivation within rows. Sycamore -'.bplr-.s were sprayed with SEVIN
70
-------
to control a minor infestation by Japanese beetles.
RESULT:
Survival
Survival of the planted cuttings and seedlings as of 14 September
1978 was quite good, except for cottonwood (Table 3). Hybrid poplar
survival averaged 96% or higher on all subplots with no major trend due
to spacing or treatment apparent. Sycamore seedling survival was also
quite high (>97%) with the highest survival in the waste water only area.
A tendency for survival to be higher on the areas receiving waste water
only was even uiore noticeable with yellow-poplar and sweetgum. Poor
survival was only experienced with cottonwood due to the poor condition
of cuttings at planting. ;
i'ABLE 3
AVERAGE TREE SURVIVAL AT END OF FIRST GROWING SEASON
(14 September 1978)
Survival (%)
Species
Control
Yellow-poplar
Sycamore
Sweetgum
Cottonwood
71
98
71
49
Waste Water
Only
88
100
89
35
Soil Heat and
Waste Water
Hybrid poplar
12 cm plant spacing
24 cm plant spacing
48 cm plant spacing
97
97
98
97
98
98
99
96
98
86
97
80
49
Growth and Development
At the end of the first growing season, average stem height, mean
basal diameter and biomass production were all greater on the areas
receiving waste water only (Table 4). Areas receiving a combination of
soil heat and waste water irrigation were less effective in promoting
growth of hybrid poplar. The control areas receiving no waste water
and no supplemental soil heat had the lowest mean heights, basal diameters
and biomass production.
71
-------
Planting density did not show any consistent relationship to growth
among the treatments. In areas receiving heat and waste water the
greater mean height occurred at the 12 cm spacing; however, the 48 cm
spacing was associated with greater mean stem heights in the other
treatments. Mean basal diameters were greater with the 24 cm in-row
spacing in the heat plus waste water treatment, but in other treatments
greater basal diameters occurred with the 48 cm spacing. The explanation
of this inconsistency may be in the fact that the soil warming produced
more small sprouts per cutting, especially with the 48-cm spacing. Numbers
of stems on the areas receiving heat and waste water irrigation were 20%
and 10% greater than on the waste water only and control areas, respectively.
TABLE 4
MEAN HEIGHT, MEAN BASAL DIAMETER, AND TOTAL BIOMASS PRODUCTION
FOR ALL HYBRID POPLAR STEMS AFTER FIRST GROWING SEASON
Treatment
Stems-
Mean
Height
(cm)
21
Mean Basal-
Diameter
(cm)
Estimated-^/
Total Biomass
(metric tons/ha)
Soil heat and waste
water
12 cm in-row spacing
24 cm in-row spacing
48 cm in-row spacing
all spacings
Waste water only
12 cm in-row spacing
24 cm in-row spacing
48 cm in-row spacing
all spacings
643
365
204
1,212
638
353
170
96
91
84
92
111
108
115
1,161 111
0.78
0.80
0.74
0.78
0.89
0.93
0.97
0.91
1.42
2.14
Control
12 cm in-row spacing
24 cm in-row spacing
48 cm in-row spacing
all spacings
622
346
186
1,154
82
82
85
82
0.67
0.68
0.76
0-69
0.98
— Includes all multiple stems >30 cm height growing from cutting.
21
-Outside bark.
3/
— Based upon average stem diameter, height and number; tentative.
72
-------
Growth of hybrid poplar on the heated areas in June was similar to
growth on areas getting waste water only; however, after July 1 differences
in growth became apparent (Figure 2) Excessively high soil temperatures,
which exceeded 30°C at 15.2 cm during July and August (Figure 1), probably
caused reduced height growth on the heated areas. Height growth was
essentially complete by early to mid-September when soil temperatures began
to decline to more optimal levels.
160
140
120
§ 100
<2 80
LJ
X
60
40
20
I
FIRST GROWING
SEASON
LSD
EFFLUENT
SOIL HEAT
EFFLUENT
CONTROL
MAY JUNE JULY AUG. SEPT. OCT.
1978
Figure 2. Average height of hybrid poplar sprouts during first
growing season for each treatment. Each point represents
the average of 90 randomly selected trees.
73
-------
Height growth of other tree species tested showed variable response
to irrigation and soil warming (Table 5). Yellow-poplar and cottonwood
exhibited the greater mean height in areas receiving effluent irrigation
only, as did the hybrid poplar. However, both sycamore and sweetgum had
greater height growth in areas receiving a combination of soil warming
and waste water irrigation than in areas which were only irrigated or
held as controls.
TABLE 5
MEAN HEIGHT GROWTH OF SELECTED TREE SPECIES
AFTER FIRST CROWING SEASON
Height Growth (cm)
Species
_ .. Waste Water Soil Heat and
Control „ , r, ^ TT _
Only Waste Water
Hybrid poplar (24 cm spacing)
Yellow-poplar
Sycamore
Sweetgum
Cottonwood
82
45
65
25
45
108
64
122
38
73
91
60
126
47
62
CONCLUSIONS
During the first growing season, hybrid poplar with a combination of
soil warming and waste water irrigation grew less than with waste water
irrigation alone. High soil temperatures during July and August on the
heated plot may have caused stress in the young trees. Earlier planting
to stimulate earlier initial growth in the first growing season is
recommended for heated soils to compensate for any reduced growth in
mid-summer. Effects of soil warming on spring growth initiation will be
observed during the second growing season.
Height growth of both sycamore and sweetgum was enhanced by a
combination of soil warming and waste water irrigation during the first
growing season. A more intensive study of these species will be
conducted during the second growing season.
Waste water irrigation alone at rates of 5 cm per week doubled the
biomass production of hybrid poplar during the first growing season relative
to the control area. Waste water irrigation increased soil moisture contents,
increased available nutrients and reduced soil temperatures. Waste water
irrigation plus soil warming increased hybrid poplar biomass production by 45%
in the first growing season relative to the control.
74
-------
ACKNOWLEDGMENT
This report was prepared with the support of the Department of
Energy (DOE) Grant No. ET-78-G-01-3066. However, any opinions, findings,
conclusions, or recommendations expressed herein are those of the authors
and do not necessarily reflect the views of DOE.
LITERATURE CITED
Blankenhorn, P. R., T. W. Bowersox; J. Hillebrand and W. K. Murphey.
1977. Forest biomass evaluation procedure for consideration as a
fuel source for a 100 megawatt electric generating facility. Pa.
Science and Engin. Foundation, Final Report PSEF Grant 310, 63 pp.
Bowersox, T. W. and W. W. Ward. 1976. Growth and yield of close-spaced
young hybrid poplars. Forest Science 22(4):449-453.
DeWalle, D. R. 1977. Utilization and dissipation of waste heat by soil
warming. Univ. Miami, Dept. of Mech. Engin.; Proc., Conf. Waste
Heat Management and Utilization, Session VII A, 73-86.
DeWalle, D. R. and A. M. Chapura, Jr. 1978. Soil warming for utili-
zation and dissipation of waste heat in Pennsylvania. Nuclear
Technology 38(1):83-89.
Sopper, W. E. 1975- Wastewater recyclcing on forest lands, in Forest
Soils and Forest Land Management, edited by B. Bernier and
C. H. Winget, Laval Univ. Press, pp. 227-243.
75
-------
POWER PLANT LAND AVAILABILITY CONSTRAINTS
ON WASTE HEAT UTILIZATION*
M. Olszewski and H. R. Bigelow
Engineering Technology Division
Oak Ridge National Laboratory
Oak Ridge, Tennessee U.S.A.
ABSTRACT
An assessment of land available at nuclear power stations was performed in
an effort to determine the limitations land availability would impose on the
implementation of reject heat utilization systems. A waste heat utilization
factor was defined for all operating and planned nuclear power stations for
which a Preliminary Safety Analysis Report has been filed. This factor is
the percentage of the station's reject heat that could be utilized on the
land available for such use at the site.
The results indicate that reject heat from 115,000 MW(e) of generating
capacity could be utilized on the land which is available. The results
further indicate that about half of this potential implementation is at
stations that have enough land to accommodate waste heat utilization systems
sized to use all of the stations"s reject heat. Utilization of the remain-
ing 50% is about equally distributed among sites capable of using between
10 and 90% of their reject heat. Further, it seems reasonable that for many
applications an integrated waste heat complex, using several waste heat
technologies, will be required to avoid marketing problems. It also appears
that single application systems will be important for the sites that can
use only a small fraction of their total reject heat.
INTRODUCTION
Recent utility statistics [1] indicate that thermal power stations generated
about 1.6 x 1.012 MW(e)-hr of electrical power during 1975. Because of the
second law of thermodynamics, this resulted in about 11.0 x 109 GJ
(11 x 10 J 5 Btu) of low temperature heat being rejected to the atmosphere.
This low grade energy is equivalent to 4.4 x 108 m3 (2.8 x 10-' bbl) of oil
per year. Thus, if uses for this low grade heat could be identified and
implemented, a significant energy source could be added to the nation's
energy resources.
Various techniques have been proposed and studied to utilize this reject
heat, which is contained in power plant condenser cooling streams. These
*Research sponsored by the Advanced Systems and Materials Production
Division, Nuclear Energy Programs, U.S. Department of Energy under contract
W-7405-eng-26 with the Union Carbide Corporation.
76
-------
applications have focused primarily on agricultural (greenhouse and live-
stock facility heating, open field undersoil heating and spray irrigation)
and aquacultural applications. Most of the efforts in these areas have
been directed at developing the individual applications.
To date, however, there has been little information available concerning
possible implementation levels for these systems on a national scale. As
mentioned above the potential energy resource is large. However, since
the major uses for this heat appear to be in the agriculture and aqua-
culture areas, product marketing constraints may limit the use of reject
heat to only a small fraction of the available resource. A second con-
straint that may limit waste heat implementation is lack of available land
surrounding power stations. Since it is generally uneconomical to transport
the low grade reject heat any significant distance from the station, the
waste heat applications must be sited in close proximity to the power
station.
Information concerning the potential impact waste heat utilization can make
in the power industry would be important to power companies and research
organizations in determining research priorities in this area. If the poten-
tial exists to utilize a significant fraction of reject heat in the power
industry, then research efforts for the applications should receive priority
and demonstrations and commercial applications be developed as quickly as
possible. If, however, technical barriers to large-scale implementation are
identified, research can be redirected to remove these obstacles.
A recent Oak Ridge National Laboratory (ORNL) study [2] attempted to define
national implementation level for various waste heat utilization techniques
by examining economic and marketing constraints from the waste heat user's
perspective. The results of this study indicate that about 45% of the
reject heat from thermal power stations could be utilized. This study, how-
ever, did not account for land availability constraints associated with the
power plant sites. Since most of the reject heat uses currently being
examined require relatively large [on the order of 400 ha for a lOOO-MW(e)
plant] land areas to utilize a significant portion of the station's reject
heat, land availability at the power station may limit wide-scale implementa-
tion of such systems in the power industry.
The present study was initiated in an effort to determine if the implementa-
tion levels reported in the previous study would be achievable when power
plant site considerations were included in the analysis. The assessment
was performed by analyzing site sizes for operating and planned nuclear sta-
tions to determine the constraints land availability around these stations
could impose on implementation of waste heat use systems. Because relevant
site information was readily available for nuclear power stations, the study
was confined to this segment of the power industry.
ANALYSIS TECHNIQUE
Information concerning the power plant sites was taken,from Refs. 3, 4, 5
and 6. 'Since these sources used Preliminary Safety Analyis Reports (PSAR's)
77
-------
as their source, the sites under consideration were confined to existing or
planned nuclear power stations. Pertinent data concerning electrical out-
put, site size, exclusion distance, and utility ownership are detailed in
a recent ORNL report [7] for each station included in the study and will not
be repeated here.
The portion of the power plant site, available for waste heat utilization
facilities was estimated at 75% of the total site size. Initially, this
area was computed using the exclusion distance and the totaJ site size. In
this scheme, the exclusion distance radius was used to calculate a circular
exclusion area. This exclusion area was then subtracted from the site size
to give the area that could be used for waste heat utilization facilities.
However, for some sites this method yielded exclusion areas that were greater
than the total site area. This was typically true for less than 10% of the
sites, namely those that were located near bodies of water. When these sites
were excluded, it was found that the exclusion area generally occupied 25%
of the total site. Therefore, it was decided to uniformly use 75% of the
total site area as the area available for reject heat facilities.
The area available for waste heat utilization systems was then used to
determine what fraction of reject heat from the station could be utilized.
To perform this analysis, however, it was necessary to estimate the area
required by waste heat utilization systems to dissipate all of the reject
heat from the power station. This area is a function of the waste heat
utilization system being used and the site ambient weather conditions
during the summer (high ambient temperatures during the summer result in
minimum heat dissipation, hence, maximum system size). Such a detailed
analysis, however, was beyond the scope of this study. Therefore, an aver-
age required waste heat utilization system size was used for the analysis.
This average system size was estimated using the results of a previous
study [2] comparing various reject heat use alternatives. This study in-
cluded greenhouse, animal shelter, and aquaculture applications and, thus,
included the major waste heat use applications currently being examined.
These systems were designed to accommodate, if possible, the yearly cooling
needs of a lOOO-MW(e) power station in Portland, Oregon. The results of the
study indicated that an average area of 400 ha is required to utilize all
of the reject heat from a lOOO-MW(e) power station. Thus, for the purposes
of this study, it was assumed that 0.4 ha/MW(e) of Installed capacity was
required to utilize all the reject heat from a power plant.
The area available at nuclear power stations and the average area required
to utilize all of the reject heat from the station were then used to calcu-
late a waste heat utilization, factor (WHUF). The WHUF was calculated by
dividing the area available for reject heat utilization by the area required
to utilize all of the station's reject heat. The latter figure was obtained
by multiplying the average area requirement of 0.4 ha/MW(e) by the total
installed capacity of the station.
The WHUF thus indicates the percentage of the total reject heat of the
station that can be utilized if all of the unused area at the station
78
-------
was dedicated to a waste heat utilization system. (This system could con-
sist of a single application, such as greenhouses, or include a number of
applications.) A site having a WHUF of 100%, or greater, has enough land
available to utilize all of its reject heat. Thus land availability will
not constrain implementation of waste heat utilization systems at these
sites. Sites having a WHUF less than 100% have only enough land to utilize
a portion of their reject heat. Thus, for these sites, land availability
will restrict the use of reject heat.
The WHUF data were then used to obtain the distribution of stations, with
respect to their ability to use reject heat, by summing the number of
stations in each 10% range (0—9, 10-19, etc.) and plotting them in Fig. 1.
To obtain a better understanding of the relative importance of each WHUF
range, the total installed capacity distribution, corresponding to the site
distribution in Fig. 1, was plotted in Fig. 2.
The primary intent of this study was to determine the nuclear generating
capacity from which waste heat could be utilized. The distribution of this
capacity among the WHUF ranges is given in Fig. 3. These results were
obtained by multiplying the installed capacity values in Fig. 2 by the mid-
point WHUF for the range. The results in Fig. 3 thus indicate the generating
capacity whose waste heat could be utilized for the various WHUF ranges. For
example, at sites that can utilize 70 to 79% of their reject heat land avail-
ability considerations would allow reject heat from 8500 MW(e) of installed
capacity to be utilized [this is in contrast to the total installed capacity
of 11,500 MW(e) for this range from Fig. 2].
The results in Fig. 3 were then used to develop cumulative plots with respect
to the installed capacity from which reject heat could be utilized (Fig. 4)
and on a percentage basis (Fig. 5).
DISCUSSION OF RESULTS
The results illustrated in Figs. 1 and 2 indicate that, in terms of both
number of stations and generating capacity, sites that have enough land
available to utilize all of their reject heat (WHUF - 100%) provide the
greatest potential for implementing reject heat use systems. These sites
comprised 32% of the stations and 28% of the total generating capacity in-
cluded in the study and thus represented a large fraction of the nuclear
power generating capability.
The relative importance of these sites is further illustrated by considering
the amount of generating capacity from which reject heat can be utilized
(see Figs. 3-5). The results in Fig. 3 indicate that the waste heat from
53,000 MW(e) of generating capacity can be utilized at sites having a WHUF
of 100% or greater. The cumulative results plotted in Figs. 4 and 5 in-
dicate that these sites comprise about 50% of the total capacity that could
utilize reject heat. Thus, if reject heat utilization is to be implemented
at significant levels, much of this potential will have to be exploited.
79
-------
From Fig. 2 it would seem that the next most important group would be those
stations that have a WHUF in the range of 20-29% (i.e., could utilize between
20 and 29% of their reject heat). Hoxvever, the results in Fig. 3 indicate
that most of the remaining WHUF ranges are of about equal importance in
terms of capacity from which reject heat could be utilized. This is due to
the fact that, although the installed capacity represented by the power
stations in the lower WHUF ranges is greater than those in the higher ranges,
the fraction of this capacity which could utilize its reject heat is de-
creasing. Thus, the two trends offset each other and the capacity for which
reject heat could be utilized remains about constant. Therefore, the results
in Fig. 3 indicate that the sites in the WHUF ranges less than 100% are of
about equal importance in terms of implementation significance.
The relative importance of the various WHUF ranges is important when consid-
ering the type of waste heat utilization complex that would probably be
required to utilize the reject heat. Since the sites whose WHUF is 100% or
greater are about as important, in terms of implementation potential, as
those having a WHUF less than 100%, discussion of the makeup of the reject
heat complex will focus on these two applications. From Figs. 1 and 2 it
is apparent that the stations in the 100% or over WHUF groups are fairly
large stations. Dividing the total capacity [56,200 MW(e)] by the total
number of stations (36) yields an average station size of 1560 MW(e). A
waste heat utilization complex sized to utilize all the reject heat from a
station of this size would require about 600 ha. Since it is unlikely that
local conditions would permit the marketing of only one type of product, the
waste heat utilization system would be required to produce a number of
products. The required diversity of production could be accomplished in one
of two ways. One method could have the entire complex comprised of a single
application, which would market a number of products. For example, the
waste heat utilization system could be a large greenhouse complex producing
tomatoes, cucumbers, lettuce, and various floral crops. A second, and more
realistic, alternative would be an integrated system consisting of a variety
of applications. For example, the complex could include greenhouses, animal
shelter heating, aquaculture, and undersoil heating applications. Since each
of these applications markets a different product, this type of system would
offer greater design flexibility in sizing each individual application to
meet local marketing constraints.
For sites that utilize only a fraction of their reject heat, design of the
waste heat utilization system will depend upon specific site conditions.
Since most of the WHUF ranges are of about equal importance, some of the
systems will be large (for sites in the higher WHUF ranges) while others
will be smaller. The larger systems will face the same marketing con-
straints mentioned above and will probably include several applications.
The smaller systems, however, are not likely to face these constraints and
could include only one application.
Therefore, it appears that if reject heat utilization is to be implemented
at a significant level, integrated systems, involving several waste heat
technologies, will play an important role. It further appears that systems
comprised of only one application will also be utilized to a significant
degree.
80
-------
The results in Fig. 4, showing the cumulative capacity from which waste heat
could be utilized indicate that waste heat utilization systems could utilize
all the reject heat from 111,500 MW(e) of generating capacity, if land
availability at nuclear stations were the only criterion. Previous analy-
sis [2] of economic and marketing constraints showed that all the reject
heat from 180,000 MW(e) of generating capacity could be used if only eco-
nomic and marketing constraints were considered. Therefore, if the maximum
indicated potential of these systems at nuclear stations was achieved, about
60% of the projected maximum industry-wide implementation could be achieved.
Since nuclear power stations comprise only about 8% of the total thermal
generating capacity in this country, it is reasonable to expect that the
implementation levels based on economic and marketing constraints would not
be further limited by power plants site size considerations.
CONCLUSIONS
Analysis of power plant site size barriers to waste heat utilization has
indicated that enough land is available at nuclear stations to utilize the
reject heat from 111,500 MW(e) of generating capacity.
The results further indicated that sites that can utilize all of their
.reject heat make up approximately 50% of this capacity. The remaining
capacity is distributed fairly uniformly among the WHUF ranges below 100%.
This result indicates that integrated systems involving several waste heat
utilization technologies will be important for over 50% of the possible
implementation, while single application systems could be used for the sites
that are constrained to use smaller systems.
The results also indicate that economic and marketing criteria constrain
waste heat utilization implementation more severely than power plant land
availability considerations.
REFERENCES
1. Edison Electric Institute, Statistical Yearbook of the Electric Utility
Industry, Publications No. 75-39 (November 1975).
2. M. Olszewski, Power Plant Reject Heat Utilization: An Assessment of the
Potential for Wide-Scale Implementation, ORNL/TM-5841 (December 1977).
3. F. A. Heddleson, Design Data and Safety Features of Commercial Nuclear
Power Plants, ORNL/NSIC-55, Vol. 1-4 (1973-1976).
4. F. A. Heddleson, Design Data and Safety Features of Commercial Nuclear
Power Plants, ORNL/NSIC-96 (June 1976),
5. F. A. Heddleson, Design Data and Safety Features of Commercial Nuclear
Power Plants-, ORNL/NUREG/NSIC-136 (June 1977).
6. D. F. Cope and H, F. Bauman, Expansion Potential for Existing Nuclear
Power Station Sites, ORNL/TM-5927 (November 1977).
81
-------
7. M. Olszewski and H. R. Bigelow, Analysis of Potential Implementation
Levels for Waste Heat Utilisation in the Nuclear Power Industry^
ORNL/TM-6312 (to be published).
82
-------
00
CO
O
CO
DC
LU
5
O
CL
oc
LU
CD
ORNL-DWG 78-12465
40
30
20
10
0-9 10-19 20-29 30-39 40-49 50-59 60-69 70-79 80-89 90-99 >100
WASTE HEAT UTILIZATION FACTOR (%)
Fig. 1. Number of Power Stations that Can Utilize a Given Fraction of Their Waste Heat.
-------
llhNl OWl. /8 12406
60
50
40
< 30
? 20
10
n
n
09 10-19 20-29 3039 4049 5059 6069 7079 8089 9099 100
WASTt HEAT UTILI/IATION FACTOR l%>
Fig. 2. Total Installed Capacity of Stations that Can Utilize a Given
Fraction of Their Waste Heat.
OHNl OWCi ;8 12467
60
50 -
40
i $
o 5
30
20
0
o
z
n
n
09 10 19 20 29 30 39 40 49 50 59 (JO 09 70 79 80 89 90 99 -101)
WAS1L Ml AT UTIU^ATION (ACTOR i >..)
Fig. 3. Installed Capacity for Which Reject Heat is Utilized
84
-------
OHNl DWG /H 17468
Fig. 4.
120
o
ui
N
O
o
I
(—
(J
100
80
5 60
<
o
Q
tu
_J
_i
<
fe
40
20-
09 10 19 20-29 30-39 40-49 50 59 60 69 70 79 80 89 90 99 '100
WASTE HEAT UTILISATION FACTOR (%)
Cumulative Installed Capacity Utilizing Waste Heat Utilization.
3
o
I
5
o
120
100
80
40
20
n
09 10 19 70 29 30 39 4U 49 W> M 60 69 70 79 80 89
WASH Ht A I Ul 11 I/AT ION FACT OH (M
90 99
100
Fig. 5. Cumulative Percentage of Capacity Utilizing Reject Heat.
85
-------
COOLING PONDS AS RECREATIONAL FISHERIES
A READY MADE RESOURCE
J. H. Hughes
Commonwealth Edison Company
Chicago,, Illinois U.S.A.
ABSTRACT
The 1275 acre perched cooling pond constructed to enable the
thermal discharge from Dresden Nuclear Station to meet State
Water Quality Standards shows potential as being a recrea-
tional fishery and Illinois River hatchery.
This resource can be used in either or both of two ways.
The pond itself might be developed as a recreational fishery
by opening It to the public. The use of the cooling pond as-
a hatchery to supply the Illinois River is also a possibil-
ity.
Recently while analysing data collected to support our
§3l6(b) demonstration at Dresden Station we came to an
interesting conclusion: The perched cooling pond used for
condenser cooling has a large fish population.
Dresden is a nuclear generating station consisting of three
boiling water reactor units with a net capacity of 1795 mwe
Unit 1., a 207 mwe unit, began commercial service in I960.
It uses a once-through condenser cooling water system which
is separate from units 2 and 3- Units 2 and 3 each have a
capacity of 79^ mwe and were placed in service in 1970 and
1971- These two large units have a spray canal, cooling
pond system which can be operated as a closed cycle cooling
facility or as a condenser cooling water trimming facility
wherein all of the cooling water flows through the pond then
is discharged to the river. Any discharge flow in between
these extremes can also be obtained. See Figures 1-4.
All three units withdraw water from the Kankakee and Des
Plaines Rivers at their point of confluence. The cooling
water is discharged to the Illinois River which is formed by
86
-------
the combination of the two previously named rivers. The
temperature rise through the unit 2 and 3 condensers is
24°F maximum.
Open cycle operation of units 2 and 3 occurred for the first
year after unit 3 was placed in service while the cooling
pond was constructed. The cooling pond was an add-on feature
of the plant made necessary because of a reduction in thermal
standards of the Illinois River by the Illinois Pollution
Control Board. For the next three years the four mile long
spray canal and 1275 acre pond were operated in the once-
through mode.
Starting in 1977 a modified closed cycle operation has been
employed to avoid a severe megawatt derating of the station
due to high temperature of the water returning to the con-
densers. Under this plan the blowdown from the pond is
adjusted on a daily basis to obtain a condenser inlet tem-
perature no greater than 91«5°F. Blowdown rates vary from
50,000 gpm to 500,000 gpm under this mode.
While collecting impingement data for the unit 2-3 intake
structure we found that a considerable number of fish were
impinged. This prompted U.S. EPA to make a preliminary
determination that the intake structure did not confrom to
BAT criteria Note however that the water returning from
the pond in the closed cycle mode of operating also passes
through the intake. A close examination of the data coupled
with verification in the form of studies of fish passing
over the pond discharge spillway confirmed that most of the
fish impinged originated in the cooling pond., not in the
Kankakee River. Electroshocking studies of the pond have
shown that there is a large copulation of carp and gizzard
shad present with a small but significant number of small
mouth bass, bluegill, greensunfish and largemouth bass. We
also know from the spillway studies mentioned above that
large numbers of freshwater drum and channel catfish exist
in the pond.
The question arises, what conditions exist in the pond to
support these fish and how did they get there? Before the
cooling pond was constructed the underlying land contained
virtually no surface waters. All fish now present entered
by passing through the intake, circulating water pumps, con-
87
-------
densers and lift pumps either in the egg or larval stage.
Sufficient numbers survived to populate the pond. No inten-
tional stocking has ever occurred.
The water quality of the cooling pond has been found to be
superior to either of the two rivers which provide make-up
water. This is not an unexpected phenomenon, ponds have
played an important role in water treatment for years, pro-
viding an enhanced settling and oxidation environment for
pollutants. In addition the combined "heat treatment" of
the condensers, followed by aeration in the spray canals and
spillway improve the quality of the water beyond what would
be expected in a quiescent settling pond.
For years the belief that fish would not do well in an
environment heated to the extent of Dresden pond has been
widely accepted. However, the fish in the Dresden cooling
pond live in an environment that varies from about 113°F to
95°F. We have found that during the summer when the cooling
pond temperatures are highest that the fish avoid the
warmest part of the pond, but seem to have little trouble
surviving in the cooler pools which still exceed 90 F. This
conclusion is born out by extensive studies performed at our
Lake Sangchris which is an impounded lake constructed to
provide condenser heat dissipation at a coal fired gen-
erating station in Central Illinois. Lake Sangchris is a
productive lake for bass and is quite popular with fishermen
in water deficient central Illinois.
Table 1 indicates numbers of some of the species that left
Dresden Pond in the interval May through August 19?8. As
expected the majority of the fish exited in July and August
when temperatures in the lake are warmest and this can be
attributed to avoidance of warmer temperatures in the lake
by these species. We believe that it is remarkable that
such large numbers can even exist in the lake in light of
past thinking about heat, entrainment, and impingement.
These data suggest that more thought should be given by
regulators and fisheries management personnel to considering
cooling lakes as nursery areas for receiving wate , These
ponds would be operated in the closed cycle mode t. ^t of the
year and in the indirect open cycle mode in the summer.
This would maximize fish production in that optimal tetn-
88
-------
peratures would exist in some portion of the pond throughout
the year. It would also minimize entrainment effects (if
any) in the makeup and receiving water since, most fish eggs
and larvae are present in the spring and early summer.
You may recall from Table 1 that most of the fish that left
Dresden Pond were of a forage variety and few are sport fish,
Large numbers of sport species do not exist in Dresden Pond
and this has been attributed to a lack of suitable spawning
habitat. Dresden Pond is bermed with a shoreline consisting
almost entirely of rip rap with a 3-1 slope. No extreme
shallow areas for nest builders exist. However this defi-
ciency could have been corrected had it been considered be-
fore the lake was constructed.
There are a number of other methods available for making
cooling ponds suitable for spawning areas. Addition of long
canals provide spawning habitat for white bass as is ev-
idenced by large populations of this species in Lake
Sangchris and in another Commonwealth Edison cooling pond,
Collins pond. Also the presence of large areas' of rooted
macrophytes and bushy shoreline areas results in good year
classes of crappies at Lake Sangchris and Collins. Pond
levels can also be designed so that they can be manipulated
for drawdown in the summer to provide areas for establish-
ment of terrestrial vegetation and flooded again the follow-
ing spring. This provision will also provide additional
spawning'habitat, and it may be that species such as the
northern pike will .reproduce successfully under these condi-
tions. Collins Pond has a reasonably large population of
northern pike despite relatively high temperatures for this
cool water species. However, there has not been any success-
ful reproduction in this cooling pond.
Sport fish enhancement could also be undertaken by means of
small hatcheries. In this case desirable sport species
would be stocked as fry in small ponds adjacent to the
cooling ponds and allowed to grow through the summer. These
fingerlings in turn would be allowed to enter the lake by
simply pulling a standpipe and allowing the hatchery pond
water and fish to enter the cooling pond where abundant
forage exists. The forage and sport fish will concentrate
in the warmer water as winter approaches. ¥e consider
threadfin shad as the forage base for the stocked predators.
89
-------
This scheme will require very little effort since once
stocked in the hatchery ponds, no additional effort is
needed until fall. At that time the only effort needed is
in removing the hatchery pond standpipes.
In summary, data from Dresden Pond and other cooling ponds
and lakes leads us to believe that heat, entrainment and
impingement effects so vigorously pursued by regulatory
agencies are largely nonexistent. Compensation or the
ability to adjust to various population impacts is far
greater than has been considered in the past. This is not
to say that power plants should be sited just anywhere and
no impacts are occurring.
Furthermore, it is appropriate that the beneficial aspects of
waste heat should be considered in designing closed cycle or
partial closed cycle cooling systems. Cooling ponds and
lakes frequently offer substantial recreational benefits over
cooling towers and this alternative should be seriously con-
sidered by regulatory agencies when it is available. Also,
the concept of cooling ponds as nursery areas for receiving
waters should be addressed with detailed studies in the near
future.
The major advantage of using the warm cooling pond water in
this way is that little additional capital expenditure is
required. In addition, this use fills an important need for
the metropolitan Chicago area. There is a need for addi-
tional recreational fishing areas in this area, especially
geared to the desires of the casual fisherman who wishes to
have a resource nearby that he can visit for a short time
and be reasonably sure of a satisfying catch.
90
-------
TABLE 1 ESTIMATED NUMBERS OF SOM11 ... ISH SPECIES LOST
AT THE DRESDEN COOLING POND SPILLWAY,,
MAY THROUGH AUGUST, 1978
Species May
Gizzard shad 48,371
Bluegill 365
Channel Catfish 715
Freslr ter Drum 1,100
Green Sunfish 81
Smallmouth bass 20
Month
June July August Totals
224,423 188,271 329,734 790,799
1,227 1,040 620 3,252
1,296 1,461 1,069 4,542
1,390 1,462 979 4,931
1,017 342 313 1,753
29 67 78 194
91
-------
Fig. 1 Dresden Station and Cooling System
General Arrangement
92
-------
Fig. 2 Dresden Station Detail
93
-------
Fig. 3 Dresden Cooling Pond
94
-------
Fig. 4 Dresden Station and Spray Canals
95
-------
HEAT RECOVERY AND UTILIZATION FOR GREEN BAY'S WASTE WATER TREATMENT FACILITY
R. W. Lanz
University of Wisconsin - Green Bay
Green Bay, Wisconsin U. S. A.
ABSTRACT
V
A feasibility study to identify and evaluate commercially available heat
recovery systems for cost effective use in the three year old energy in-
tensive Green Bay Metropolitan Sewage Treatment Plant has been completed.
After studying all the systems at the facility, it was found that the li-
quid system showed the most potential for energy recovery and the fuel oil
fired hydronic heating system showed the greatest potential for energy
savings. It was determined that heat recovered from the liquid system
should be utilized in the heating system and that heat pumps would be the
"»ost applicable equipment for this heat recovery and utilization scheme.
Energy cost savings using heat pumps rather than the exclusive use of
fuel oil boilers to heat the facility should be $101,000 per year with a
capital equipment investment of about $425,000. The total annual savings
including maintenance and equipment amortization cost should be $58,000.
INTRODUCTION
A proposal was approved by the Commissioners of the Green Bay Metropoli-
tan Sewage District to have this investigator conduct a feasibility study
to identify and evaluate alternate energy sources or systems for cost ef-
fective use in the three year old Green Bay Metropolitan Sewage Treatment
Plant.
A first step in the study required a knowledge of the systems at the faci-
lity and the interrelationships between these systems. This knowledge
was gained by studying system schematics, conferring with plant personnel
who were involved with these systems and studying system components dur-
ing operating conditions.
Knowledge of the current energy consumption of particular systems at the
facility was also required before recommendations concerning conservation
of energy or alternate energy systems could be made. However, because
most systems were not monitored for energy consumption, the precise divi-
sion of the total energy consumption at the facility among its particular
systems was unknown. Therefore this division was determined by thermo-
dynamic calculations on these systems.
As a result of these calculations, it became obvious that recovering waste
96
-------
heat at the facility should take precedence over considerations of the gen-
eration of more heat. Therefore this study was expanded to include appli-
cability of commercially available heat recovery systems to the facility.
It is this aspect of the study which is presented in this paper.
EXISTING SYSTEMS AND THEIR ENERGY POTENTIAL
The systems at the facility were divided into two major catagories. The
first category, called the main systems, includes both the major energy
users and major energy transfer systems while the second catagory, called
minor systems, includes the minor energy users or energy transfer systems.
The main systems are the liquid system, solid system, process air system,
process boiler system, electric power system and hydronic heating system.
The minor systems are the control air system, cooling water system and the
chemical addition system.
The liquid system handles 18 to 20 million gallons per day of waste from
metropolitan Green Bay in addition to 16 to 18 million gallons per day
of waste from several paper mills in the Green Bay area. The energy con-
sumed by this system is used for the numerous electric motor driven pumps
to lift the waste from the interceptors for treatment and electric motor
driven a~rators to treat the waste in the contact basins. The mixture of
metropolitan and mill waste handled by the facility varies in temperature
from a minimum of 18 C (64 F) in January to a maximum of 31 C (88 F) in
July. The obvious resource in this system is the heat contained in 34 to
38 million gallons of liquid at these temperatures. Possibilities were
immediately visualized for extracting some of this heat, keeping in mind
that lowering the temperature of the liquid in the contact basins may im-
pair efficient sewage treatment.
The solids system transports the solids from various basins to the solids
handling building for the heat treatment and incineration process. Energy
consumed by this system is used for the electric motor driven pumps to
transport the solids. Little opportunity for energy recovery or utiliza-
tion was seen for this system.
The process air system provides compressed air for the aeration contact
basins used in the treatment process. Four air compressors, each driven
by a 2500 H. P. electric motor are the heart of the system. There have
been efforts by plant personnel to reduce the energy consumption of this
system by operating at a minimum air flow while maintaining adequate
treatment in the contact basins. As with the solids system, little op-
portunity for energy recovery or utilization was seen for this system.
The process boiler system supplies steam at 500 psi for the purpose of
heat treating the sewage sludge. The steam used in this process cannot
be reccndensed for boiler feed water because of the type of heat treat-
ment used. Therefore 100 percent boiler makeup water is required.
97
-------
Because of this characteristic, energy recovery from this system was not
evident.
The electric power system is the most nebulous of any system in the faci-
lity. It runs throughout the facility and is of utmost importance in the
operation of the plant. It provides energy for the hundreds of electric
motors needed to drive pumps, fans, compressor, agitators, and aerators
and for lighting, maintenance and control. Plant personnel have been re-
ducing the energy requirements of this system by using the most prudent
use of electric power in operating the plant.
The hydronic heating system for the facility uses a pair of fuel oil fired
hot water boilers located at one end of the facility. The heated water is
pumped through a 14 inch diameter pipe located in a tunnel system in a cir-
cumferential path around the facility. Secondary heating loops supplied
by the 14 inch line heat the seven individual buildings at the facility.
The water temperature at the cold end of the heating loop is controlled at
82 C (180 F) which produces temperatures at the hot end of about 91 C (196
F).
Since the facility is located in an extreme northern climate, it seemed
reasonable that heat recovered from the final effluent could be utilized
most of the time of the year in the heating system. However, the problem
of raising the temperature of large quantities of heat at a minimum efflu-
ent temperature of 18 C (64 F) to a temperature adequate for the heating
system had to be solved.
CURRENT ENERGY USE
The major forms of energy used at the plant are purchased electric power
and fuel oil. Although there is the capability of using natural gas,con-
sumption of this fuel will probably remain minor. Electric power consump-
tion at the facility, although of major importance, was not addressed in
this investigation. Fuel oil is burned for incineration, process steam
and space heating. Oil consumption is approximately 6500 gal/mo, for in-
cineration and 33,000 gal/mo, for process steam. Since the sewage treat-
ment process is essentially independent of the time of year, these values
are relatively constant throughout the year. The fuel oil consumption for
heating varies throughout the year.
The heating demand for the facility during the 1976-77 heating season is
shown in Table 1. As seen by the table, the peak heating load occurred in
January when 21.5 MM Btuh were needed, the minimum load occurred in Sept-
ember when 4.2 MM Btuh were needed. The average heat rate throughout the
season was 12.2 MM Btuh.
98
-------
HEAT RECOVERY SYSTEM
The nominal size of the heat recovery equipment for the facility was as-
sumed to have a capacity between 10 and 15 MM Btuh. Equipment smaller
than 10 MM Btuh would not produce energy savings that would justify the
investment in equipment. Equipment greater than 15 MM Btuh would tie up
heat recovery equipment that would not be used most of the year and great-
ly reduce the use of the facility's relatively new fuel oil boiler equip-
ment.
Five types of commercially available heat recovery systems were considered
for the facility. They were: heat pipes, liquid sorbent enthalpy ex-
change systems, total and sensible heat recovery wheels, coil energy re-
covery loops and heat pumps. Considering the magnitude of energy to be
recovered and considering that a liquid to liquid heat transfer device
would be used, it was determined that heat pump equipment would be the
best choice for this application.
HEAT PUMP CHARACTERISTICS
Heat pumps have been used for several years for air conditioning and home
heating in moderate climates. Their use in northern climates has not
been popular because of their high cost of operation. However, increas-
ing energy costs are precipitating an increase in heat pumps for heat
recovery applications even in northern climates.
Heat pumps employ a refrigeration cycle in their operation. The four
main components of a heat pump are the compressor, which compresses the
refrigerant; a condenser which rejects heat from the refrigerant; an
expansion device which expands the refrigerant to low temperature and
pressure; and an evaporator which receives heat from some medium and
thereby heats the refrigerant before it enters the compressor for re-
compression. The sum of all energy into the system must equal the sum
of all energy out of the system. Hence the rate of heat being rejected
by the condenser is numerically greater than the rate at which work is
delivered to drive the compressor because of the addition of heat into
the evaporator.
The useful effect desired from a heat pump is a maximum rate of heat re-
jection from the condenser for a minimum energy flow into the system to
drive the compressor. The ratio of heat rejected to the work input to
the system both expressed in the same thermodynaraic units, is called the
coefficient of performance or COP. The numerical value of this ratio is
always greater than or equal to one since the heat rejected is the sum
of the heat input to the system plus the work into the system. A COP
as large as possible is desired so that the multiplicative effect on the
work input by the heat pump gives the largest possible heat rejection to
the condenser.
99
-------
The COP is influenced in relatively minor ways by details of the heat
pump cycle design, the efficiency of the compressor, the design and
sizing of the evaporator and condenser, the choice of refrigerant, and
even the sizing of the refrigerant lines which interconnect all the com-
ponents. However, the most important factors which influence the COP
are the temperatures between which the heat pump must operate. The COP
is maximized by using the highest possible heat source temperature and
the lowest possible temperature at which heat is rejected. More speci-
fically, the greatest COP is obtained when the evaporator temperature is
as high as possible and the condenser temperature is as low as possible.
Because of the characteristics of the pressure enthalpy diagram for nor-
mal refrigerants, the COP is more sensitive to evaporator temperatures
than to condenser temperatures. Therefore fie primary consideration in
increasing the COP is to have the evaporate Derating at as high a temp-
erature as possible.
The cost of operation of the heat pump depends upon the COP and the cost
of electric power. The greater the COP and the lower the cost of electric
power, the cheaper the energy produced by the heat pump. If a heat source
at a relatively high temperature can be utilized by the heat pump thereby
assuring a reasonable COP along wilh a modest electric power rate, then
the heat pump compares favorably to fuel oil fired equipment.
HEAT PUMP APPLICATION TO THE FACILITY
Table 2 illustrates the application of a 10 MM Btuh heat pump to the faci-
lity. As seeu by the table, this heat pump would meet the heating load
for the facility for the months of September, October, April and May.
For the remai*>in0 months of the heating season, the heating boiler would
provide the additional heat needed. The totals in the table show that
the heat pump would provide approximately 69 percent of the heat required
by the facility.
Table 3 shows the application of a 15 MM Btuh heat pump to the facility.
As seen in the table, the heat pump would meet the total heating load
for the facility except for the months of December, January and February.
The totals show that the heat pump would provide 87 percent of the heating
load for the season.
Two industrial heat pump manufacturers in the United States claimed that
they could manufacture equipment to meet this load. One manufacturer
limits the size of their heat pumps to 5 MM Btuh, therefore two or three
of these machines would have to be used in series. The other manufacturer
is capable of building one machine to handle any load required by the
facility.
Three commercially available machines with the following characteristics
were considered. 1) A 10 MM Btuh machine with a COP of three receiving
100
-------
31,028 gpra of final effluent at 13 C (55 F) and discharging 2,300 gpra of
heating water at 71 C (160 F) with a capital cost of $300,000. 2) A 15
MM Btuh machine with a COP of three receiving 4,000 gpm of final effluent
at 13 C (55 F) and discharging 3,000 gpm of heating water at 71 C (160 F).
with a capital cost of $425,000. 3) A 15 MM Btuh machine with a COP of
two receiving 4,000 gpm of final effluent at 13 C (55 F) and discharging
1,650 gpm of heating water at 85 C (185 F) with a capital cost of $300,
000.
These choices represent equipment obtaining a maximum COP under the op-
erating temperature specifications. The machine with the lower COP
would produce heating water near the temperature previously used at the
facility. However, experimentation on the heating system during the
past winter showed that the facility could be adequately heated with
water temperatures as low as 71 C (160 F) thereby allowing the use of a
heat pump with a COP of three.
ECONOMIC ANALYSIS
Table 4 shows the energy costs for the existing heating system and the
three heat pumps described above. These values were calculated by using
the amount of heat produced by each type of equipment, shown in Tables
2 and 3, the current fuel oil price and electric power rate paid by the
district, the stated COP of each machine and a fuel oil boiler efficiency
of 70 percent. As shown by the table, maximum annual energy savings of
$101,450 would be realized by using the 15 MM Btuh machine with a COP
of three.
Total annual cost of each alternative was calculated by including main-
tenance, replacement and energy costs. Preventive maintenance cost was
quoted to be $750. per year by one of the major heat pump manufacturers.
Unscheduled maintenance was not estimated. The replacement cost was
calculated assuming a 20 year equipment life and an 8 percent return on
capital. These costs were found to be $30,570 per year for the $300,000
heat pump and $43,300 per year for the $425,000 heat pump. The instal-
lation cost for the equipment was not available at the time this paper
was written therefore the economic analysis does not include these costs.
Table 5 shows the total cost of the four alternatives and the savings
over the present system. As shown by the table, a maximum savings of
$58,000 per year is realized with the 15 MM Btuh heat pump with a COP
of three. It is shown that there is a net loss of $27,400 per year by
using the large heat pump with the low COP. The smaller heat pump
shows a savings of $49,230 per year which is less than the savings from
the larger heat pump with the same COP.
It should be noted that replacement and maintenance costs are not includ-
ed in the analysis for the fuel oil equipment currently being used.
101
-------
The reason is that the heat pump equipment would be installed in addi-
tion to and used in conjunction with the present system. Therefore
the heat pump replacement and maintenance costs are added costs that
would not exist if the heat pump was not added to the facility.
If the heat pump equipment was being installed in a new facility then the
savings over fuel oil equipment would be closer to the energy cost sav-
ings shown in Table 4 rather than the savings shown in Table 5. This is
because replacement costs would have to be included for the fuel oil
equipment as well as with the heat pump equipment.
CONCLUSIONS
This analysis has shown that sewage treatment plants offer a unique oppor-
tunity for economically recovering and utilizing waste heat with heat
pumps. All sewage treatment plants have an inexhaustible heat source in
relatively warm, clean final effluent. Unless the treatment facility is
located in a tropical climate, these treatment facilities could use some
portion of the heat for space heating. Therefore, if the economic in-
centive were presented, large savings to the taxpayers could be realized
in the utilization and recovery of waste heat at these facilities.
102
-------
TABLE 1
HEATING LOAD FOR THE FACILITY
DURING THE 1976-77 HEATING SEASON
Oil Consumption
Month (Gals.)
Sept.
Oct.
Nov.
Dec.
Jan.
Feb.
Mar.
Apr.
May
1976
1976
1976
1976
1977
1977
1977
1977
1977
31,000
72,800
96,600
150,200
163,700
121,400
86,000
56,300
18,100
TOTAL
Heating
Btu X 10 "9
3.05
7.13
9.47
14.72
16.04
11.90
8.43
5.52
1.77
78.00
Boiler Output
• MM Btuh
4.24
9.58
13.15
19.78
21.56
17.71
11.33
7.67
4.92
Month
Sept.
Oct.
Nov.
Dec.
Jan.
Feb.
Mar.
Apr.
May
TOTALS
TABLE 2
APPLICATION OF 10 MM Btuh HEAT PUMP
Heat Pump
(Btu X IP"9) (MM Btuh)
3.05
7.13
7.20
7.44
7.44
6.72
7.
5.
44
50
1.80
53.72
24
58
4
9
10.0
10.0
10.0
10.0
10.0
7.67
4.92
Heating Boiler
(Btu X 10"9) (MM Btuh)
2.27
7.26
8.63
5.18
.96
24.3
3.2
9.8
11.6
7.7
1.3
103
-------
TABLE 3
APPLICATION OF IS MM Btuh HEAT PUMP
Month
Heat Pump
(Btu X IP"9) (MM Btuh)
Heating Boiler
(Btu X IP"9) (MM Bt h)
Sept.
Oct.
Nov.
Dec.
Jan.
Feb.
Mar.
Apr.
May
TOTALS
3.05
7.13
9.47
11.2
11.2
10.1
8.4
5.5
1.8
67.8
4.24
9.58
13.2
15.0
15.0
15.0
11.3
7.67
4.92
—
—
3.57 4.8
4.80 6.6
1.81 2.7
10.2
104
-------
Alternative
TABLE 4
ANNUAL ENERGY COST ALTERNATIVES FOR HEATING
THE FACILITY
Energy Cost
Fuel Oil Electric Power Total Energy Savings Over
Cost Cost Cost Present System
Current
heating
system
10 MM Btuh
heat pump
COP of 3
15 MM Btuh
heat pump
COP of 3
15 MM Btuh
heat pump
COP of 2
$312,000
$ 97,200
$ 40,800
$40,800
312,000
$134,300 231,500 80,500
$169,750 210,550 101,450
$254,625 295,425 16,575
Alternative
No heat
pump
10 MM Btuh
heat pump
COP of 3
15 MM Btuh
heat pump
COP of 3
15 MM Btuh
heat pump
COP of 2
TABLE 5
TOTAL ANNUAL COST OF EACH ALTERNATIVE
Energy Maintenance Replacement Total
Cost Cost Cost Cost
Savings Over
Present System
312,000
231,500 700
210,000 700
295,400 700
312,000
30,570 262,770 49,230
43,300 254,000 58,000
43,300 339,400 -27,400
105
-------
WHY FROUDE NUMBER REPLICATION DOES NOT NECESSARILY ENSURE MODELING
SIMILARITY
W. E. Frick* and L. D- Winiarski
Corvallis Environmental Research Laboratory
U.S. Environmental Protection Agency
ABSTRACT
It is commonly assumed that Froude number replication ensures similarity
between fluid phenomena prototypes and their fluid models. The densimetric
Froude number typical in plume modeling is:
V
Fr = /—3——
/ Pa~P Dg
P
Where V is a characteristic velocity, Pa and P are the ambient and plume
densities respectively, D is a characteristic dimension and g is the accel-
eration of gravity.4 It is generally assumed that equivalent predictions
will result when equal Froude numbers and dimension!ess coordinates are
used. However, the nonlinear density behavior of many fluids as functions
of temperature, salinity or water vapor proves this assumption to be false,
and the discrepancy is not trivial. A buoyant plume in water may not
behave the same as a buoyant plume in air even if both have the same Froude
number as defined above. There can be a noticeable difference between
plumes in the same medium with the same Froude number but at different
temperature conditions. For example, a horizontal water plume at 40°C in
stagnant water of 0°C rises only briefly before sinking, while another
water plume at 40°C in ambient water of 20°C with the same Froude number
rises monotonically. The implications for various plumes and outfalls are
numerous. The explanation for this behavior is the subject of this paper.
INTRODUCTION
The most commonly accepted parameters in buoyant plume modeling are the
initial densimetric Froude number and the "initial velocity ratio. Some-
times the Reynolds number is also considered. Generally these are consi-
dered sufficient for chaYacterizing the plume in neutrally stable environ-
ments such as water with zero vertical density gradient or air with an
adiabatic temperature gradient. Otherwise a parameter reflecting the
stability of the ambient must be used (1,2). The reason the densimetric
Froude number works as well as it does as a similarity parameter in buoy-
ant plume modeling is that the term relates directly to the initial net
buoyant force per unit mass of plume material (i.e. the vertical acceler-
ation). As long as the behavior of the vertical acceleration throughout
* Now affiliated with the Oregon Dept. of Transportation, Salem, Oregon
106
-------
the plume trajectory can be uniquely characterized by Initial values, the
densimetric Froude number appears to be a useful similarity parameter.
Unfortunately this appears to be true only if we are dealing with a medium
that is neutrally stable and has a linear equation of state. Apparently
there appears to be an implicit assumption that nonlinear density effects
are insignificant in comparison to other effects. 'We will show that this
is frequently not the case.
A CATASTROPHIC EXAMPLE
As indicated, the equation of state is important in determining the value
of the buoyant acceleration. What can happen when two plumes have the
same densimetric Froude number but have different ambient and initial
plume temperatures? An extreme example is shown in Figure 1. The curves
are the trajectories of two horizontally discharged plumes in quiescent,
neutrally stabl^ ambient, however, they are not similar even though they
have the same ( nsimetric Froude number. One plume rises briefly before
sinking; its initial conditions are described in the lower right corner of
the figure. The other plume rises monotonically. The behavior of the
sinking plume can be understood when it is realized that the initial plume
temperature is 40°C and the ambient temperature is 0°C. Water happens to
have its maximum density at about 4°C. Figure 2 shows this graphically:
anywhere between 7.5°C and 0°C the density of water is greater than it is
at 0°C. Thus, when the plume temperature falls below 7.5°C, as it would
from mixing in the ambient water, the plume density becomes greater than
the ambient density and the plume becomes negatively buoyant.
IMPLICATIONS ON PHYSICAL MODELING
Another example will illustrate the problem of trying to simulate buoyant
air plumes in air with buoyant water plumes in water at the same densime-
tric Froude number. This is a common practice that needs to be critically
examined. Figure 3 makes such a comparison using a linear equation of
state instead of the ideal gas law. The ideal gas law:
P = p/(RT)
is, relatively speaking, considerably more linear at room temperature than
the corresponding case for water. However'? its curvature is opposite to
that of water (see Figure 4).
Returning to Figure 3, it shows four plume trajectories, two with a densi-
metric Froude number of 6/and two with 20. The plumes have an initial
temperature of 40°C and diameter of 2 meters. All are discharged horiz-
ontally into quiescent fluid at 2p°C with the velocities shown in the
figure. Those trajectories labelled LES are using a fluid with a linear
equation of state, the others are water. It can be seen that the predic-
tions are substantially different even though the same conditions exist.
A study of Figure 2 reveals why the trajectories are so different. The
densimetric Froude number only takes into account the initial buoyant
acceleration and says nothing of how the density behaves between the
107
-------
initial and the ambient temperatures. Suppose the plume at a particular
location has effectively mixed an equal amount of ambient into itself.
The average temperature i$ then 30°C at that point. The plume buoyancy
at that temperature using the equation of state for water can be seen to
be somewhat less than the plume buoyancy at the same temperature using
the linear equation of state. (To emphasize our point further, it is the
difference in density that is being discussed.) Thus the plume governed
by the linear equation of state always is more buoyant than the corres-
ponding water plume. At 22.5°C the ratio of buoyant accelerations (defined
in the figure) drops all the way to 0.74.
As can be seen from Figure 4, the ideal gas law is considerably more
linear than the equation of state for water at the same temperatures.
However, had the ideal gas law been used instead of the linear equation
of state the difference in trajectories would have been even greater.
THE RELATIONSHIP BETWEEN PLUME PROFILES AND BUOYANCY
So far we have not considered temperature or density distributions in
speaking about plumes. Many numerical models, including the one used in
this paper (3,4) make use of the "top hat" assumption. In such a model
the average temperature is often used to define the magnitude of the buoy-
ant acceleration. However, real plumes will exhibit temperature and
density profiles. This characteristic affects the overall buoyancy.
Using a water plume as an example (and assuming its temperature ranges
from 20 to 40°C across the plume section), the warmer portion of the
plume has more influence (making the plume less dense) than does the cooler
portion. Thus the plume is actually more buoyant than the "top hat" model
indicates. Using the linear equation of state in modeling will tend to
reduce the error that would otherwise occur by using the "top hat"
assumption but with a better equation of state. However, preliminary
integration shows that the magnitude of this effect is small in compari-
son.
CONCLUSION AND RECOMMENDATIONS
The densimetric Froude number is sometimes deficient as a similarity
parameter and must be augmented by knowledge of the equation of state.
This deficiency may impact the conduct of-physical modeling, and the
modeling results may require critical examination for invalid applications.
Also, it follows that accurate equations of state should be used in com-
puter models. Application of these recommendations in modeling will
improve research on plume phenomena by removing an otherwise misleading
source of error.
108
-------
REFERENCES
1. Shirazi, M.A. and Davis, L.R. Workbook of Thermal Plume Prediction1
Volume 1: Submerged Discharge* USEPA Environmental Protection
Technology Series, EPA-R2-72-005a, August 1972.
2. Winiarski, L.D. and Frick, W.E. Atmospheric Plume Nomographs with
Computer Model for Cooling Tower Plumes, USEPA Interagency Energy-
Environment Research and Development Series, manuscript.
3. Winiarski, L.D. and Frick, W.E. Cooling Tower Plume Model, USEPA
Ecological Research Series, EPA-600/3-76-100, September 1976.
4. Winiarski, L.D. and Frick, W.E. "Methods of Improving Plume Models,"
Cooling Tower Environment—1978 Proceedings, University of Maryland,
May 1978.
109
-------
30
20
Z/D
10
-10 —
Fr - 6.0
discharge =
2.06 m/s
TI = 40°C
Tamb -
D = 2 m
Fr = 6.0
discharge =
2.33 m/s
Ti = 40°C
Tamb = 0°C
D = 2 m
10
20
40
Figure 1.
Plume trajectories in quiescent ambient showing complet-
ely difterent behavior for two plumes in water with the
same Densimetric Froude Number.
110
-------
1000
s_
O)
•M
IT]
OJ
-o
eqn. of state
for water
density at 2Q°C
plume "buoyancy" \ plumf
"buoyancy"
at 30°C with LES
density
at 30°C
density at 30°C \
with LES \
Initial "buoyancy" \
on which Fr is based
density at 40°C
994 -
992
0
10 20
Temperature (°C)
30
40
Figure 2. Comparison of plume buoyancy in water using different
equation of state assumptions.
Ill
-------
30
20
Z/D
10
Fr = 6.0
discharge =
2.06 m/s
Tamb = 200C
D = 2 m
Fr = 20.0
discharge =
6.88 m/s
10
20
30
X/D
40
50
60
70
Figure 3. Plume trajectories in quiescent ambient showing the differences between the LES
assumption and the actual equation of state for water.
-------
co
E
CD
1.30
1.20
en
10
0)
•o
ideal gas
"1.10
270
280
290
300
310
320
Temperature (°K)
Figure 4. A relative comparison between the ideal gas equation
and the equation of state for water under an atmos-
phere of pressure.
113
-------
A CALIBRATED AND VERIFIED THERMAL PLUME MODEL
FOR SHALLOW COASTAL SEAS AND EMBAYMENTS
S. L. Palmer1
Florida Department of Environmental Regulation
Tallahassee, Florida U.S.A.
ABSTRACT
A numerical thermal dispersion model has been developed and linked to a
tidally driven, barotropic, hydraulic model of shallow coastal embayments.
The solution algorithm is multi-operational with the advective terms
treated implicitly and the diffusive terms treated explicitly. Analysis
of the computational dispersive and dissipative effects shows the model
to be similar in behavior to multi-operational implicit models. However,
the magnitudes of the amplitude and phase errors are smaller than for
either a purely implicit or explicit algorithm.
The model has been employed to predict the distribution of temperature
resulting from a once-through, seawater-cooled power generation facility
near Crystal River, Florida. Calibration was achieved by experimentally
determining the model parameters and coefficients for the impacted
basin. The model was then verified by comparing the position and shape
of the simulated plume to those of the observed plume. After simulation
of five real-time days (6/15/75 to 6/19/75) the simulated plume area was
within 15% of the observed pluxne area and the simulated mean plume
temperature differed from the observed value by only 0.05°C.
INTRODUCTION
Recent concern about the ecological future of "natural" environments has
precipitated a need for accurate and efficient methods of predicting the
impact of "man-made" stresses on environmental systems. Due to the
widespread use of natural water bodies as receiving basins for waste
products, it is of special importance that methods be developed for
predicting the effects of waste products on water quality.
A natural body of water may be described by a mathematical system composed
of interacting subsystems. The system is influenced by a variety of
phenomena such as wind, rainfall, solar radiation and runoff. The
geometric configuration and geomorphic composition of the basin will
also have a significant influence on the system. The response of the
system to these, influences determines the spatial and temporal distribution
of concentrations of various substances which affect water quality.
•'•The work on which- this paper is based was done while the author was at the
Department of Marine Science, University of South Florida, and was
supported by a grant from the Office of Water Research and Technology,
U. S. Department of the Interior.
114
-------
GENERAL MATHEMATICAL MODEL
Fundamental to the mathematical modeling of water quality is the concept
of constituent mass, balance which embodies a relationship among the
material change of the constituents, the diffusive flux and any local
sources and sinks. The general mass balance equation for a dissolved
constituent in an incompressible fluid is
DC = V- CKcVC) + r]C + r2, CD
Dt
where the material derivative (P/Dt)_ implies dependence upon the velocity
field. Under general conditions, a determination of the velocity field
involves the solution of a large set of complicated equations that de-
fine the relationships between mass, momentum, thermodynamic state,
salt, internal energy and entropy. However, several simplifying assump-
tions can be made that are applicable to many coastal environments.
Although the velocity field is three-dimensional, the fluid flow in a
shallow body of water is primarily horizontal. If it is assumed that in
shallow water bodies turbulence caused by frictional stress and vertical
eddy transfer uniformly mixes the fluid mass, the water quality model
can be reduced to a vertically independent, two-dimensional system by
integration from the bottom to the sea surface. Assuming that the pres-
sures are hydrostatic, that the mass, densities of the substances trans-
ported in the estuary are small compared with the density of the water,
that only shear stresses from horizontal components are important and
that the effects of small-scale velocity fluctuations are aggregated
into shear stress terms, the vertically averaged conservation of mass
and momentum of the fluid can be expressed as follows:
3t 3x 3y
CU _ fV + g H3_n + 1. (T£ - T|)- = 0> (3)
Dt 3x p
and
DV + f U + g H3n + i M? - ty}' = o ^
Dt 3y p
Under these assumptions: the constituent mass balance equation becomes
{2£L + (Rlc H
3x 3x 3y
This model, although greatly simplified, still does not lend itself
readily to analytic solution. The problem is further complicated when
the boundary and initial conditions are made to conform to estuaries
with irregular geometry and are driven by forces that are temporally and
spacially varying. The solution, however can be approximated by the
numerical technique of finite differences.
115
-------
In the finite-difference approximations of equations 2 through 5, the
discrete values of the variables were described on a space-staggered
grid as shown in Fig. 1. The water level n» the constituent con-
centration C, and mean sea level depth h were computed at integer values
of j and k. The mass transports U were computed at half-integer values
of j and integer values of k, while the mass transports V were computed
at half-integer values of k and integer values of j.
Solution of the momentum equations yields the velocity components required
for solution of the constituent balance equation. The finite-difference
approximations to the momentum equations were explicit in form and with
the exception of the Coriolis terms were the same as those used by Reid
and Bodine [l]. In his comparison of several finite-difference schemes,
Sobey [2] found that, within the range of numerical stability, Reid and
Bodine's algorithm held an advantage over the others in that their
scheme had the smallest numerical dispersion errors.
The finite-difference approximation of the constituent balance equation
was a space-centered, multi-operational system that treated the advective
terms implicitly and the dispersive terms explicitly. Computational
errors resulting from the finite-difference approximation to the continuum
equation can be classified as dissipative and dispersive. Numerical
dissipation results in a decrease in the amplitude of the solution wave
components without physical reason. Numerical dispersion causes the
components of the numerical solution (computed wave) to propagate at a
different speed than the components of the analytic solution (physical
wave >.
The ratio of the computed wave in amplitude and phase to the physical
wave after the time required for the physical wave to propagate over its
wave length provides a measure of the dissipation and dispersion associ-
ated with the computational procedure 13]. Analysis of the numerical
behavior of the model algorithm showed that the computational scheme
tended to be somewhat dissipative and highly dispersive whenever there
were fewer than ten finite-differences per wave length. However, as the
grid size became smaller the computational solution converged to the
physical solution in both amplitude and phase. The magnitude of the
dissipation and dispersion associated with this scheme was smaller than
for either a purely implicit or explicit scheme.
CRYSTAL RIVER MODEL
The water quality model has been applied to the estuary (Crystal River
Estuary; Fig. 2) adjacent to the Crystal River electric power generating
facility of the Florida Power Corporation (FPC). for the prediction of
thermal enrichment by heated effluent. The estuary is a gulf coastal
region of Northrcentral Florida encompassing the area in which the Cross
Florida Barge Canal encounters the Gulf of Mexico as well as the area in
Which the. generating facility discharges heated effluent. The estuary
is bounded on the. north and south by dredging spoils, on the east by a salt
JL16
-------
marsh, and is open to the Gulf of Mexico on the west. The estuary
itself is criss-crossed by several irregular strings of oyster bars
which form barriers to water flow, resulting in a basin with very complex
bathymetric features.
For the purpose of determining the dispersion of heated effluents in the
estuary, the estuary was modeled by a mesh compatible with the finite-
difference definitions of the previous section. A field measurement
program was then initiated for gathering data necessary for calculation
of bottom friction coefficients, eddy diffusion coefficients and metero-
logical exchange, parameters.
MODEL CALIBRATION
Friction Coefficients
In shallow estuaries, energy losses due to dissipative forces cause
significant sea surface slopes. The energy dissipation is brought about
by a combination of phenomena for which the individual components would
be almost impossible to determine. Thus for computational purposes all
of the losses were nerged into one "frictional" resistance force which
was assumed to be uniformly distributed throughout the region under
investigation.
The exact nature of frictional resistance in free surface flow is not
known. However, the phenomenon has been studied for channel flow by
Chezy and Manning [4] and was empirically found to be proportional to
the square of the flow velocity vector. An extension to two-dimensional
flow can be made by considering the components of the associated mass
transport vector.
Determination of friction coefficients for shallow estuaries from field
measurements can be extremely complicated. Dronkers 15] suggested a
method of calculating Chezy (friction) coefficients for a river section
using measurements of the vertical tide at both ends of the section and
the volume discharge through a cross-section as a function of time.
A relationship for two-dimensional flow that is analogous to Dronkers1
unidirectional formulation is
C - H|v*|//an/3x*, C6X
where V* is the vertically integrated water velocity and x* is a coordinate
axis in the direction of V*.
Since a resolution of the. sea surface slope, over a short distance is
difficult to ascertain even with intercalibrated tide, gauges, a method
for determining the surface, slope in terms of a measurable quantity was
devised. Several attempts were made to gather the necessary data to
impliment the technique, proposed by Palmer et^ a!L. [&] ; however, repeated
equipment malfunction prevented any degree of success.
117
-------
The sea surface, slope is. however, a necessary factor in determining the
Chezy coefficients. A limiting slope, for the. sea surface was calculated
from Airy wave, theory as.
3n = 1 3n.
8x* Cg 3t" (.7).
To provide a means of classifing frictional influences due to different
bottom types, the bottom composition of the estuary was surveyed. Each
grid in the estuary was then assigned one of eight bottom types: mud,
shell, limestone, grass, algae, sponge, patchy grass or sand.
In the "friction surveys", vertical velocity profiles were measured and
mean velocities were calculated for each of the bottom types. The velocity
profiles were measured at maximum ebb flow or maximum flood flow so that
the full effect of the friction was measiored. Values of the Chezy
coefficients were then calculated, converted and compared to the literature
values of the Manning number for each bottom type. As seen in Table 1
the values for channel flow were in good agreement with the values
calculated for the two-dimensional flow.
Dispersion Coefficients
Any dissolved constituent released into an estuary will undergo dispersion
due to advection by currents and large-scale turbulence and due to
diffusion by small-scale turbulence and molecular dynamics. However,
the molecular diffusion is small compared to the advective and large-
scale turbulent dispersion and is considered negligible.
A neutrally bouyant constituent released on the water surface should in
theory disperse such that isolines on the surface form a radially
symmetric patch about the center of mass. This center of mass, which is
the point of peak concentration within the patch, should coincide with
the point of injection. In fact such patches are rarely symmetric.
However, pertubations are random and on the average a radially symmetric
patch describes the spread by dispersion. Advection will then translate
the patch downstream and shear will cause the transport to be differential,
deforming the patch somewhat.
Kolmogorov [7] states that for vertically well mixed water bodies and
for scales of small magnitude, horizontal diffusion can be assumed
isotropic, thereby forming the radially symmetric patch. Thus, in a
frame of reference moving with the current, the dispersion of heat along
the surface can be described by an equation written in polar coordinates
with one coefficient Dr which is in the form
where 5 and q are empirically determined coefficients. The dispersion
equation becomes
-------
1£ = 1 1 (r D
"
Various values of q found by experimenters yield special cases of the
equation. For a diffusion coefficient constant with position within the
patch, q=*0, Dj^S, and
C « Co_ exp (-£?_}-. C1Q1
4-rrfit 46t
A diffusion coefficient varying directly with r, as reported by Joseph
and Sendner [8] requires that q=l, Bj^rfi, and
C - Co exp {-EL}. (.11).
21n&2t.2 St
Ozmidov [9] found that Dr varied with r4^3 so that q=4/'3, Dr=r3/45, and
C = 243C0 exp {-i£2/3}. (12)
3irt3 4- and (i2^3-* ^1
119
-------
which, according to the solutions, vary linearly with A(r) , A(x )_,
A(x / )L, respectively. The results are shown in Fig. 3. Clearly, only
the first order pairing approximated a linear relationship. Here t=129.
seconds, and the slope was .05.3 /ft. The slope equals 1/6 t, and so q=l
and 6=.0"45 m/sec.
Since heat can be expected to diffuse in the same manner as the dye in a
vertically well mixed region, the results of the above experiment were
used to determine dispersion coefficients, for the model. At each time
step, the diffusion from one model grid square to the next was calculated.
The distance r (center of one grid and center of adjacent grid), will
always be the length of the side of a grid, or 200 m; thus r=200 m, q=l,
6=.045 m/sec and Dr=6 r*5=9 m^/sec. This corresponds reasonably well
with the values found by various experimenters and tabulated by
Kullenberg [10] .
Heat Budget
The heat budget of an estuary is affected by the prevailing meteorological
conditions as well as by direct input by the discharge of heated water.
In any given time period the temperature at some fixed point in an
estuary will increase, decrease, or remain constant depending on the
flux of heat through the water surface and on the mixing with surrounding
water that occurs during that period.
The temperature change due to mixing is determined by solution of the
constituent mass balance equation. The heat flux through the water
surface, on the other hand, can be expressed as
where
qsn = net solar radiation flux;
qatm= net atmospheric radiation flux; Q.3 )_
qw = water surface radiation flux;
qc = sensible heat flux;
qe = evaporative heat flux.
The "energy received" terms qsn and qatm can be measured together as
total net incoming radiation qR. These two terms are indpendent of the
water surface temperature and were determined by field measurement.
Since a single— layer , two-dimensional model was employed, all net incoming
radiation was assumed to be. absorbed and distributed uniformly throughout
the water column. This allowed calculation of the resultant temperature
change CAT!.. The relationship between AT and the surface, heat flux is
AT = Aq
Cptt. (14).
The details: of the calculation of these, terms, are. reported By Klausewitz
et al. [11] .
120
-------
As a verification technique for the heat exchange model, temperature
data collected on June 15, 1975 were used as initial conditions.
Meteorological data collected between June 15 and June 19. were then used
in the model to predict the temperature at each of the test stations
during the survey conducted on June 19-
A comparison of the model predictions and the observed data are presented
in Table 2. The temperature values from the model after five days of
simulation had a mean value 0.5°C greater than those from the field.
These values were reasonably close considering the number of factors
affecting water temperature. Which factors were responsible for this
increase was not determined.
However, the net radiometer malfunctioned and values of total incoming
solar radiation were used instead of net incoming solar radiation.
Since the model requires a measure of net solar radiation which takes
into account reflected light from the sea surface and the shallow bottom,
this is- the most likely source of the increase in the ambient temperature
of the model over that found in the field.
MODEL VERIFICATION
A five day survey of the Crystal River Estuary was conducted from June 15,
to June 19, 1975. During that time continuous records of meterological
parameters, tide heights and effluent characteristics were maintained.
These records were transcribed into numerical form for use as driving
functions in the model. A salinity, temperature, depth (STD) survey on
June 15 provided the initial temperature distribution seen in Fig. 4.
The five day period was then simulated with the model and the predicted
thermal plume (Fig. 5) was compared to the results of an STD survey
conducted on June 19 (Fig. 6).
The general shape and location of the modeled plume were similar to
those of the field results with the major discrepency being the isotherms
near the Cross Florida Barge Canal. Apparently more water was exchanged
between the discharge basin and the Cross Florida Barge Canal than was
considered by the model. Also, the model plume contained higher tempera-
tures near the outfall than were found in the field data. This could
have been caused by the 3/4 mile long discharge canal and the associated
atmospheric cooling effects which were not included in the model due to
computer storage limitations. Also, the discharge basin was required to
be at least two feet deep in order that the hydraulics model remain
stable (ho bottom exposure, was allowed during low tidal conditions)...
Since in fact most of the salt marsh and mudflat region north of the
canal is much shallower than two feet, significantly more nocturnal
cooling can occur than the model simulated. However, if the simulated
and observed thermal pltjmes> coyer the same regions of a vertically well-
mixed estuary then the ratio of the surface areas of the two plumes
provides- a quantitative measure, of their relative, "closeness in extent".
Since the field data points did not cover the entire Crystal River
estuary, only those regions of the estuary cove-red by the field survey
171
-------
can be compared. The portions of the estuary considered are. within the.
region marked by the dotted lines on Figs. 5 and 6. The areas containing
water with temperatures between the various isotherms are listed in
Table 3.
Taking into account the fact that the ambient water temperatures resulting
from the model simulations were about 0.5°C warmer than they should be,
a second table was created as shown in Table 4. In this case the modeled
and observed areas with similar temperatures were much closer. The mean
plume temperatures were 32.10°C for the observed plume and 32.05°C for
the calculated plume.
CONCLUSIONS
The similarity in shape and location between the modeled and field
plumes after simulation of five days indicates that the hydraulics
portion of the model was generally adequate to describe the water movements
in this highly complex estuary.
The overall plume size (area with temperatures >31°C) calculated by the
model was within 15% of the field plume size once corrections for ambient
temperatures were made. Also the mean calculated plume temperature was
32.10°C, compared to a mean observed plume temperature of 32.05°C.
However, the model exaggerated the size of the warmer portions of the
plume, due to ba thyme trie approximations in the model and the absence of
heat loss to the atmosphere in the discharge canal.
Predictions of the plume size and configuration over a tidal cycle have
also been made for the FPC Anclote power plant [12]. Comparison of
model results with field data indicates that the thermal prediction
model is readily adaptable to other estuaries that are similar to the
Crystal River Estuary.
REFERENCES
1 Reid, R. and B. Bodine, 1968. Numerical Model for Storm Surges
In Galveston Bay. Proc. J. of Waterways and Harbors Div.,
Km. Soc. Civil Engr., Vol. 94, No. WW1, pp. 33-57.
2 Sobey, R. J., 1970. Finite-Difference Schemes Compared for Wave-
Deformation Characteristics in Mathematical Modeling of Two-
Dimensional Long-Wave. Propagation. Technical Memorandum
No. 32, U.S. Army Corps of Engineers Coastal Engineering
Research Center, Washington, B.C., 22 pp.
3 Leendertse, J. J., 1967. Aspects of a Computational Model for Long
Period Water Wave Propagation. Memorandum RM-5294-PR, the Rand
Corporation, Santa Monica, California. 164pp.
4 Chow, Ven Te, 19S9-. Open Chanel Hydraulics. wcGraw-Hill Book
Company, New York, New York. 680 pp.
122
-------
5 Dronkers, J. J., 1964. Tidal Computations in Rivers and Coastal
Waters. North Holland Publishing Company, Amsterdam. 518 pp.
6 Palmer, S. L., K. L. Carder, B. A. Rodgers and P. J. Behrens, 19.75.
Calibration of a Thermal Enrichment Model for Shallow,
Barricaded Estuaries.; Annual Report 1974-1975. Dept. of
Mar. Sci., Univer. of So. Fla., St. Petersburg, Florida.
141 pp.
7 Kolmogorov, A. N.f 19.41. The local structure of turbulence in
an incompressible viscous fluid for very large Reynolds
numbers. C. R. Acad. Sci., USSR 30, 301.
8 Joseph, J. and H. Sendner, 1962. J. Geophys. Res., 67:3201.
9 Oamidov, R. V., 1965. Izv. Atmos. Oceanic Phys. Ser., 1:257.
10 Kullenberg, G., 1974. An Experimental and Theoretical Investigation
of the Turbulent Diffusion in the Upper Layer of the Sea,
Report #25, Kobenhavas Universitet Institut for Fysisk
Oceanografi.
11 Klausewitz, R. H. S.L. Palmer, B.A. Rodgers and K.L. Carder, 1974.
Natural Heating of Salt Marsh Waters in the Area of the
Crystal River Power Plant. Independent Environmental Study
of Thermal Effects of Power Plant Discharge, Technical
Report #3, Dept. of Mar. Sci., Univer. of So. Fla.,
St. Petersburg, Florida, 31 pp.
12 Palmer, S. L., 1975. Predicted Summer Isotherms for Anclote
Ancharge after Power Plant Initialization. In: Anclote
Environmental Project Report 1974. G. F. Mayer and V. Maynard
(eds.J Dept. of Mar. Sci., USF, St. Petersburg, Fla. pp. 36-65.
NOTATION
c Chezy coefficient
Cg shallow water wave speed
C Constituent concentration
C0 Initial concentration
Cp Specific heat capacity
Dc Radial dispersion coefficient
f Coriolis parameter
g Gravitational acceleration
H: Total water depth
%ax Mgy'Tif" expected water depth
KC Dispersion coefficient
q Empirical coefficient
Q Water mass source/sink.
r radial distance
T! Constituent reaction rate
*2 Constituent source/sink
R! Vertically integrated reaction
rate coefficient
R2 Vertically integrated source/
sink
U x directed mass transport
V y directed mass transport
6 Empirical coefficient
Aq Surface, heat flux
As Finite-difference space increment
At Finite-difference, time increment
AT Temperature change
n Sea surface variation
p Sea water denisty
TS Surface stress
18 Bottom stress
123
-------
N>
Bottom type
Mud
Shell
Limestone
Grass
Algae
Sponge
Patchy Grass
Sand
Literature n
0.035
0.035
0.015
0.022
0.025
0.032
0.027
0.030
Calculated n
0.
0.
0.
0.
0.
0.
0.
0.
0358
0306
0043
0244
0295
0335
0244
0292
n
-0.0008
0.0094
0.0107
-0.0024
-0.0045
-0.0015
0.0026
0.0008
Sta. No.
19
18
16
17
1
T
0
Field
30
30
30
30
30
30
0
Results
.1
.3
.5
.3
.6
.36
.19
Model
31.
30.
30.
31.
30.
30.
0.
Results
3
8
7
2
3
86
40
t
+1
+
+
+
-
0.
0.
(°c)
.2
.5
.2
.9
.3
50
59
CO
Table 1 Comparison of calculated and literature
values for the Manning number.
Table 2 Comparison of ambient field temperatures
(June 19, 1975) with ambient model tem-
peratures after four days of simulation.
(See Fig. 6 for Station Locations)
-------
Interval
(°c)
>34
33-34
32-33
31-32
Field
(km2)
.24
.42
.56
2.62
Model
(km2)
.36
.75
1.89
4.02
Ratio
.67
.56
.30
.65
Interval
(°C)
> 34
33-34
32-33
31-32
Field
(km2)
.24
..42
.56
2.62
Model
(km2)
.22
.51
1.01
2.77
Ratio
1.09
.82
.55
.95
NJ
tn
Table 3 Comparison of field and Model plume areas. Table 4 Comparison of field and adjusted model
plume areas.
-------
Ax
ON
k i 1 - -
k
y
K k-l --
1
•
1
I
M
^ X
1
1
1
. 1
1
1
•~
i
f
- -H -
1
1
i
j
1
I
— 1- _
1
1
— 1 —
1
1
- H--
1
1
i
M
r\
—
—
— • «-
, h
V
U
I-H
M
I
Co
^^
Cl
Fig. 1 Finite-difference grid scheme.
Fig. 2 Location of Crystal River Estuary
-------
O
i i i
10 a(f
i t i i
100 ym 300 4on 900 coo TOO «o<) eoo 1000 tiuo 1200 tioo 1400
Fig. 3 Dispersion data compared to empirical
relationships.
C\ ,- -ZS
\ I . , 32.0
Ebb Tide STO
Temperature at Of I
June 15.1975
C.I. • 5°C
Fig. 4 Initial temperature distribution
-------
K)
00
Modol Tcmporaturo Results
For Eorly Flood Tldn
Juno 19, I97!5
Region compared with
(laid results.
FLOOD TIDE STD
loiriporaluro
-------
FARFIELD MODEL FOR WASTE HEAT
DISCHARGE IN THE COASTAL ZONE
D. N. Brocard and J. T. Kirby, Jr.
Alden Research Laboratory
Worcester Polytechnic Institute
Holden, Massachusetts, USA
ABSTRACT
Waste heat discharges in coastal areas with alternating tidal currents may
lead to a background temperature rise due to insufficient flushing of the
area. In this case, the discharge dilution water, and possibly also the
intake water, will have a rise above ambient.
This long term heat buildup cannot be predicted by the types of models typi-
cally used for nearfield temperature calculations; either physical models
or integral type jet models. Large scale circulation models using finite
differences or finite elements numerical methods are capable of predicting
this heat buildup. These models, however, remain costly to use and do not,
as yet, alleviate the need for separate nearfield studies.
The model presented in this paper is designed to permit rapid evaluation of
long term heat buildup near a heat discharge. Isotherm shapes are assumed
and their areas are calculated as a function of surface heat flux, turbu-
lent diffusion, and net current speed (flushing). Combined with nearfield
temperature rise predictions, the background temperature rise can be
estimated. A case study is presented as well as dimensionless plots of
results.
INTRODUCTION
Coastal areas are widely used for the disposal of waste heat from power
generation. Typical disposal schemes involve rapid mixing of the heated
effluent with ambient water intended to limit the size of the higher tem-
perature rise isotherms. This mixing is achieved by using turbulence
associated with the initial momentum and buoyancy of the discharge. These
forces also transport heat away from the discharge point. The discharge
momentum, however, decays due to interfacial or bottom friction and the
discharge buoyancy decreases with plume dilution so that beyond a "dis-
charge controlled zone," these mechanisms are no longer effective in
transporting heat away from the discharge area. Ambient currents, surface
heat transfer, and turbulent diffusion become the dominant heat transport
mechanisms. In coastal areas with alternating tidal currents, the net
ambient current which provides flushing may be only a fraction of the
maximum observed current. With little flushing, it is possible that a
long term heat buildup will develop in the general area of the discharge.
This heat buildup will result in a background temperature rise, raising
the temperature of the ambient water used for nearfield dilution and
possibly also the temperature of the intake water.
129
-------
The design of discharge structures is based on compliance with nearfield
thermal criteria and uses models for the prediction of thy nrarfield dis-
tribution of temperature rises. These models are eithei ydraulic scale
models or mathematical models for jets and plumes. Neither of these
allows the prediction of the background temperature rise which is actually
independent of the discharge scheme. Large scale circulation models using
finite differences or finite elements numerical methods are capable of
predicting the long term heat buildup. These models, however, remain
costly to use and do not, as yet, alleviate the need for separate near-
field studies.
The model presented in this paper calculates the farfield distribution of
temperature rises due to a source of heat of given intensity as a function
of net current, surface heat flux, and turbulent diffusion characteristics.
This model allows a rapid evaluation of the farfield temperature distri-
bution and, if combined with nearfield results, gives an estimate of the
background temperature rise.
MODEL DEVELOPMENT
Consider a power plant cooling system with intake and discharge structures
located on, or a short distance off, a straight shore. For the farfield
analysis, this system can be idealized as a source of heat on the frontier
of a semi-infinite body of water. If the net longshore current speed, V ,
is small compared to the maximum tidal current speed, the heat will spread
in the offshore direction as well as alongshore. Self similar temperature
rise isotherms are, therefore, assumed. For simplicity and flexibility,
the isotherm shape is selected as rectangular with an aspect ratio "a",
as shown in Figure 1. The aspect ratio, as yet undetermined, is expected
to increase with increasing net current speed, V , which stretches the
patch of heat in the alongshore direction.
With these isotherms, the rate of heat transfer across the water surface
between the AT and AT + dAT isotherms, located at x and x + dx is:
d $ = 2 K a x dx AT (1)
s
where K is the surface heat flux coefficient.
For ocean applications, the vertical diffusion of heat is negligible
compared to lateral diffusion, and a constant thickness, H, for the
heated layer can be assumed [1], as shown in Figure 1. With vertically
uniform temperatures in this layer, the net rate of head advection
(flushing) out of the volume bound by the AT isotherm is:
$f = pC H x AT Vn (2)
The rate of heat transport out of this same volume by turbulent diffusion
is:
= -PC D H (a + -) x —^ (3)
p a dx
130
-------
in which D is a diffusion coefficient. At steady state, the rate of heat
transport out of the volume enclosed in the AT isotherm should equal the
rate of heat input into this volume. The latter is equal to the plant heat
discharge rate; pC QD ATD> where Q and AT are the plant discharge flow-
rate and temperature rise. The following hee
heat balance equation results:
I
2K a x dx + pC H x AT V - pC DH (a + —) x -=^ = pC Q AT (4)
Q P n p a' dx K p XD D
This integral equation is rewritten in dimensionless form as follows:
f
' n
2R nAT*dn + F^ AT* - r\ —- = 1 (5)
'o
where TI = x/H, and
„ „ V H H D(a + —)
R = . F = n AT* = n AT a AT (6)
pC D (a + -) D(a + -) WD D
^ p a a'
Differentiating Equation (5) with respect to r\ leads to the following second
order differential equation:
J A m J* -rt
(7)
Two conditions are required to solve this equation. One is given by Equa-
tion (5) while the other is that AT should go to zero as r\ goes to infinity.
In the general case, solving the above boundary value problem requires a
numerical approach. However, Equation (5) shows that when r) goes to zero,
AT will go to infinity. While this result is clearly in contradiction with
physical reality, it will also lead to numerical difficulties. At T\ - 0,
which can be interpreted as the location of the discharge, turbulent jet
mixing (not included in the above derivation), becomes dominant. This
mixing could be modeled here by artificially increasing the dispersion co-
efficient, D, as T] goes to zero. For the purpose of the farfield analysis,
however, it is sufficient to limit the calculations to a region outside of
the discharge nearfield, taken at TI = n • In the nearfield region (n < r\ ) ,
the temperature will be assumed constant so that Equation (5) applied at
ri gives:
Rr]2AT*+ FT) AT*-n
00 ° °
- 1 (8)
in which AT * is the nearfield temperature. This equation replaces Equation
(5) as boundary condition to the problem. The size, r) , of the nearfield
region could be estimated based on nearfield temperature studies. The value
of n , however, was shown to have little effect on the farfield results for
practical values of the parameters.
131
-------
Special Case; No Current (V = 0)
In the special case when the net current velocity would be equal to zero,
(V = 0 -> F = 0), Equation (7) can be transformed to the following form:
where £ = r)/2R. This equation is a modified Bessel equation of order zero.
The solution of Equation (9) with the boundary conditions discussed above
is:
AT* = - - - (10)
Rn 2 K (n /2R) + n *^2R K, (n /2R)
o o o o 1 o
where K and K, are modified Bessel functions or order zero and one.
Numerical values of these functions are available in Tables [2] and in
standard computer subroutine packages [3]. For r) = 0, the above equa-
tion simplifies to
AT* = KO (n *R) (11)
For practical applications, numerical values given by Equations (10) and
(11) are very close.
MODEL RESULTS
In the general case of F / 0, Equation (7) was solved numerically. An
exponential transformation was done on r| to properly treat the boundary
condition at n -> °° while keeping small discretization steps near TI = r\
where the other boundary condition is specified. For F ** 0, Equation
was used. Solutions are presented in Figures 2 to 4 in terms of the dimen-
sionless parameters defined in Equation (6) . The plots of AT* versus r) are
equivalent to plots of temperature rise versus offshore distance. Values
of the flushing parameter, F, and of surface heat loss parameter, R, span
the range' of practical applications.
Figures 2 and 3, which have constant values of F and several values of R,
show the effect on the temperature of changes in the surface heat flux co-
efficient, K. It is interesting to note that the reduction of temperatures
obtained for a given increase of K is smaller when the net flushing current
is larger.
Figure 4 shows the effect of varying the flushing parameter, F, for a given
value of the heat flux parameter. As expectable, temperatures decrease for
increased flushing.
132
-------
MODEL APPLICATION
For a practical application, and once the necessary coefficients have been
determined, the graphs in Figures 2 to 4 can be used to obtain estimates of
the extend of the farfield temperature rise isotherms. If nearfield results
are available, these graphs can also give the background temperature rise to
be added to the predicted nearfield values. Such an application is examined
in the following section.
The estimation of some of the parameters involved in this model requires
further attention, in particular, the diffusion coefficient, D, the iso-
therm aspect ratio, a, and the depth of the heated layer, H.
The diffusion coefficient can be evaluated using the following "4/3 law"
first proposed by Richardson T4]:
D = A CJ4/3 (12)
where a is the standard deviation of the patch of heat and A is a coeffi-
cient which has the dimension of length to the 2/3 power per unit of time.
Okubo [5] showed that the coefficient, A, actually varies to a small extent,
with the dimension of the patch of heat, represented here by cr. For O
between 2000 ft and 10000 ft, a value of A = 0.002 to 0.003 ft2'3/sec is
appropriate. The standard deviation, 0, should be established by trial and
error.
The aspect ratio of the isotherms, a, can be estimated as follows: The
distance needed for the momentum of the discharge plume to be dissipated
by friction was shown by Lee et al [6] to be approximately 8H/f where f is
the applicable friction factor (interfacial or bottom). This distance is
approximately the distance that the plume will'travel in the offshore direc-
tion (if the initial discharge is directed offshore). The distance travelled
alongshore is approximately equal to the tidal excursion, i.e., Tu /IT +
TV /2, taking the net current into account. T is the tidal period and UT
the longshore tidal velocity amplitude. An estimate of the isotherm aspect
ratio, therefore, is:
UT
T^
a = —i_ (13)
The depth of the heated layer, H, is best estimated based on nearfield
results. In the absence of such results, judgement is required. It should
be noted that a complete finite difference or finite elements model of the
area would also require the input of a layer depth.
CASE STUDY
Consider a proposed generating plant which will discharge 4.4 x 10 BTU/sec
near a straight shore. Field studies indicate that the net current speed
in the area varies from 0.0 to 0.2 ft/sec and using Equation (12) with
a = 10,000 ft, a value of D = 500 ft2/sec is obtained. A representative
133
-------
value of the surface heat loss coefficient for summer conditions is K -
200 BTU/ft2/day/°F. In addition, nearfield studies showed that the thick-
ness of the heated layer in the dispersion region is 15 ft. Finally, the
procedure outlined above leads to an isotherm aspect ratio a - 1. With
these values, the flushing and surface heat loss parameters are F = 0 to
1.2 x 10-3 and R = 2.2 x 10~7.
The areas of the temperature rise isotherms, as calculated by the above
farfield model, are plotted in Figure 5 for net current speeds of 0.0, 0.1,
and 0.2 ft/sec. Also, an isotherm aspect ratio a = 2 was considered and
the corresponding isotherm areas are plotted in dotted lines.
On the same plot, results of a nearfield thermal study are indicated. These
results are averages of nearfield results at different times during the tide
cycle. Temperatures predicted by the nearfield model are significantly
lower near the discharge point than those predicted by the farfield model.
As already pointed out, this is due to the fact that nearfield turbulent
mixing is not included in the farfield model. The results of the two models
should, however, match at an intermediate point. This can be obtained by
adding to the nearfield temperatures a constant background temperature rise.
The point where the two models should match is not clearly defined. This
point should correspond approximately to the distance from the discharge
where the initial momentum is lost by friction. The distance was earlier
seen to be 8H/f; here approximately 12,000 ft if f is taken equal to 0.01,
a typical value for interfacial friction. The corresponding area is 3,300
acres. With this value, and an isotherm aspect ratio a = 1, the background
temperature rise would be zero for V =0.2 ft/sec and 0.7°F for V = 0.0
ft/sec. A larger aspect ratio, which is probable for V =0.2 ft/sec,
would lead to background temperature rises about 0.2°F larger than the
previous values.
In the particular case study considered here, the near and farfield results
are almost parallel near the matching point. Therefore, a change in the
location of the matching point would not result in significantly different
values of the background temperature rise.
REFERENCES
1. Csanady, G.T., "Turbulent Diffusion in the Environment," D. Reidel
Publishing Company, 1973.
2. Abramowitz, M., and Stegum, I.A., editors, "Handbook of mathematical
Functions," U.S. Department of Commerce, National Bureau of
Standards, Applied Mathematics Series 55, 1964.
134
-------
3. IBM Scientific Subroutine Package.
4. Richardson, L.F., "Atmospheric Diffusion Shown in a Distance
Neighbor Graph," Proceedings of the Royal Society of London,
A110, 1926.
5. Okubo, A., "Oceanic Diffusion Diagrams," Deep Sea Research, Vol. 18,
1971.
, i
6. Lee, J.H., Jirka, G.H., and Harleman, D.R.F., "Modelling of Uni-
directional Thermal Diffusers in Shallow Water," Report No. 228,
R.M. Parsons Laboratory for Water Resources and Hydrodynamics,
MIT, 1977.
135
-------
FIGURE 1 FARFIELD MODEL ISOTHERMS
136
-------
AT*
1.5 -
1.0 -
0.5 -
100
3 4 56789 10000
2 3 456789 1000 2
TJ = x/H
FIGURE 2 FARFIELD MODEL RESULTS FOR F = 0 AND A RANGE OF R
AT*
i i i i i i i
o l_
3 456789 10000
2 3 456789 1000 2
7} = x/H
FIGURE 3 FARFIELD MODEL RESULTS FOR F = 6 x 1
-------
AT*
2.0 -
1 1 \ 1—I I I I
1.5 -
1.0 -
0.5 -
0
100
3 4 56789 10000
2 3 456789 1000 2
TI = x/H
FIGURE 4 FARFIELD MODEL RESULTS FOR R = 1 x 10'7 AND A RANGE OF F
AT*
\ VN = 0.0 ft/sec
FARFIELD
MODEL
RESULTS
10 20 40 60 80 100 200 400 600 1000 2000 4000
ISOTHERM AREA (ACRES)
10000
FIGURE 5 MATCHING OF NEARFIELD AND FARFIELD RESULTS FOR CASE STUDY
138
-------
THERMAL CHARACTERISTICS OF DEEP RESERVOIRS
IN PIMPED STORAGE PLANTS
J J Shin and N S Shashidhara
Envirosphere Company
A Division of Ebasco Services Incorporated
New York, New York, USA
ABSTRACT
Temperature distribution of the upper and lower reservoirs of pumped stor-
age hydroplants is governed by (1) the pumping characteristics of the re-
servoirs, (2) heat transfer through air-water interface and (3) tempera-
ture and flow rates of water flowing into and out of the reservoirs.
The differences in the thermal characteristics of the upper and lower re-
servoirs would depend upon the relative elevations of inlet and outlet
ports and the rates of discharge during "generating cycle" and "pumping
cycle".
Mathematical models are presented in this paper to estimate the tempera-
ture distribution in the upper and lower reservoirs of a pumped storage
project, with an assumed schedule during "generating and pumping cycle".
The results of the mathematical modela re utilized also to establish the
temperature distribution characteristics in deep reservoirs without a
pumped storage plant. The mathematical formulations take into account the
influential effects of solar radiation, natural heat convection due to den-
sity differences, forced heat convection by inflow and outflow, heat diffu-
sion, surface heat losses by evaporation and conduction, and surface heat
gain by long wave radiation. Reservoir discharge temperatures with and
without pumping from the lower reservoirare also provided.
INTRODUCTION
A pumped storage plant is an arrangement whereby electric power is generated
during peak load periods of an electric system by utilizing water previously
pumped into a storage reservoir during off-peak periods, utilizing excess
energy from the same system. Pumped storage plants besides having low
operating/maintenance costs and outage rates have also a low unit invest-
ment cost. They are able to transpose low-value off-peak energy into high-
value on-peak energy. In pumped storage plants-, the energy required to
pump water during off-peak periods helps to maintain a steady load on the
thermal plants.
However, installation of pumped storage facilities on existing reservoirs
or natural water bodies can alter their thermal structure significantly.
Such thermal alterations are of great importance to the existing biologi-
cal communities and species. In deep reservoirs, strong thermal stratifi-
139
-------
cation may develop during .certain seasons, and the isotherms are horizontal
during most of the year. W ™ The pumped storage operation between two
existing deep reservoirs disturbs their thermal stratification. This dis-
turbance affects the life support system of many forms of aquatic life in
the pumped storage project.
This paper establishes the thermal stratification in pumped storage re-
servoirs, with and without pumping. Variation of outlet/tailrace tempera-
ture from the lower reservoir due to pumped storage operation is also ana-
lyzed.
DEEP RESERVOIR THERMAL MODEL
The thermal model used in this study is based on the deep reservoir, ±em]jera-
ture prediction model developed by Harleman and his associates. "/ I ' vv
The deep reservoir thermal stratification process is governed by a heat ba-
lance involving solar radiation, natural heat convection due to density dif-
ference, forced heat convection by inflow and outflow, heat diffusion, sur-
face heat losses by evaporation, back radiation and conduction, and surface
heat gain by long wave radiation.
Some important assumptions have been made to set up the governing equations,
such as:
1) The isotherms in a stratified reservoir are horizontal.
2) The existence of horizontal isotherms in a reservoir suppress the
vertical motion to the extent that turbulent transfer of momentum or
heat can be neglected.
3) Solar radiation is transmitted in the vertical direction only.
4) Heat transfer through the sides and the bottom of the reservoir is
negligible.
5) Specific heat at constant pressure and the heat diffusion coeffi-
cient are constant through all space and time concerned. In addition,
the Boussinesq approximation for the density is also made.
The conservation of heat energy equation becomes:
dt A(y) ay dr fCpA ay
(1)
(Ui(y,t)Ti(y,t)-Uo(y,t)T)
where:
T(y,t) temperature at elevation y and time t
140
-------
y elevation
t time
D heat diffusion coefficient
A(y) cross-sectional area of reservoir at elevation y
/> density
Cp specific heat at constant pressure
0(y,t) heat flux of internally abosrbed radiation at elevation y and
time t
V(y,t) vertical velocity component at elevation y and time t
Ui(y,t) inflow velocity at elevation y and time t
Uo(y,t) outflow velocity at elevation y and time t
•t • 7
Ti(y,t) inflow water temperature at elevation y and time t
Equation (1) is a transient, one-dimensional, second order partial differ-
ential equation. Therefore, two boundary conditions and one initial con-
dition are necessary in order to formulate a solution.
The thermally homogeneous state of a reservoir in the early spring season
is provided as the initial condition.
(2) T - T at t - 0 for all y
o
The heat energy balance at the reservoir surface is achieved between the in-
coming radiation 0 , atmosphere radiation 0 , surface heat loss 0, and the
amount of heat diffused into the reservoir from the water surface.
(3) ^ Ty" y=ys " *0o + 0a - 0L
The heat transfer through the bottom of the reservoir is negligible.
a T
(4) -?-i . o at y-yb for all t
A numerical scheme is employed to solve the governing equation. Harleman's
approach was written in an explicit and finite element schematization.
The numerical stability criteria, owing to an explicit scheme, are:
(5) D
141
-------
(6) V -~ < 1
v ' Ay -
where At and Ay are time and elevation intervals.
MODEL VERIFICATION
The mathematical model discussed above was verified using field measure-
ments.
Since direct measured data for solar radiation and long wave radiation were
not a vail at
method. (5)
not available, they were calculated from a modification to Kennedy's
Prior to temperature prediction analysis, it was necessary to examine the
validity of the basic model. Comparison of the predicted reservoir tem-
perature profile (using the model) with actual measurements made in the
field was performed.
Figure 1 shows the measured and the predicted reservoir temperature pro-
files for the upper reservoir. W An examination of Figure 1 indicates
that the predicted and measured data agree well.
TEMPERATURE PREDICTION DUE TO PUMPED STORAGE OPERATION
The modified pumped storage thermal model was utilized to establish the
thermal structure of the upper reservoir about 90 m deep and lower reser-
voir about 70 m deep. The outlet temperature of the lower reservoir is
also presented. Reservoir temperature profiles are predicted and compared
for two modes of operation: with and without pumping. It is seen that
the temperature profiles of the two reservoirs are closely related to
pumping schedules, inflow and outflow rates and levels of pumping ports.
The existing penstock of the upper reservoir is located at 25 m depth from
the surface and the new pumping port is assumed to be located at about 13 m
depth. Similarly for the lower reservoir, the outlet/tailrace port is located
at 28 m depth and pumping port is located at 4 m depth. The ports are lo-
cated such that during the (power) generation mode, the outlet water from
the upper reservoir enters subsurface layers of the lower reservoir. During
the pumping mode, the surface water of the lower reservoir is pumped to the
upper reservoir.
The pumping schedule was assumed such that the upper reservoir loses two
feet in elevation per day during the week (cumulative). Water was pumped
back during the weekend (approximately 15 hour generation and 9 hour pum-
ping during a weekday and 24 hour pumping a day through the weekend).
During the generation mode,equal amounts of waterwere assumed to be dis-
charged through the existing and the new generation (pumping) outlets.
When inflow from other sources into the upper reservoir is greater than the
outflow, no pumping was done. For the lower reservoir, it was assumed
that the minimum outflows could always be maintained throughout the genera-
142
-------
tion and pumping modes. This is in order to maintain minimum flow require-
ments in the river on the downstream side.
DISCUSSION OF RESULTS
Predicted reservoir temperature profiles with and without pumping are
plotted in Figure 2 for the upper reservoir and Figure 3 for the lower
reservoir for different monthly conditions. Site specific meteorological
data for the year 1961 has been adopted in this analysis. However, plant
operational details assumed in the analysis are preliminary. Temperature
profiles for both upper and lower reservoirs are presented for the same
time period.
For the upper reservoir (Figure 2), surface temperature is not affected
significantly by pumped storage operation; however, temperature near the
pumping port region (13 m depth) is affected by pumping and shows consider-
able increase when compared to similar temperatures without pumping
operation. This is primarily the result of the location of the new
pumping and discharge outlets. During generation mode, relatively cooler
water from the upper reservoir is discharged through the new outlet to the
lower reservoir while during pumping mode, the upper reservoir receives
relatively warmer water from the surface layer of the lower reservoir.
As a result, the upper reservoir continuously gains thermal energy through
pumped storage operation and temperature increase is as much as 5 C at
13 m depth. On the other hand, temperature of the 25 m depth region
(existing outlet position) decreases only slightly by pumped storage
operation. This is because of the reduced mixing between the cooler deep
waters (below 25 m depth) and warmer surface waters. Water temperature at
depths below 25 m remains essentially unchanged.
Figure 3 shows four different temperature profiles with and without pumping
for the lower reservoir. Temperatures at depths greater than 28 m remain
unchanged while surface temperatures show a significant decrease (up to
8 C) due to pumping. This result is discussed here. During the generation
mode, relatively cool water enters from the upper reservoir. However,
during the pumping mode, warm surface water from the lower reservoir is
pumped to the upper reservoir before surface heat can be transferred to
the lower layers. As the seasonal effect of solar radiation on the water
surface becomes stronger in late spring, more heat is transferred to the
lower layers. As a result, water temperature in the lower layers increases
with time, until early fall (September). The surface layer of the lower
reservoir gains thermal energy during daytime generation mode, but this
warm water will be pumped back to the upper reservoir during nighttime
pumping mode. Thus the lower reservoir continuously loses thermal energy
due to pumped storage operation. This exchange of thermal energy between
1:wo reservoirs affects the outlet/tailrace temperature of the lower
reservoir as discussed below.
Figure 4 shows a comparison of lower reservoir outlet/tailrace
143
-------
temperatures with and without pumping for March through December 1961. In
general, it is seen that the outlet/tailrace temperature is always lower
with pumping when compared to the same without pumping. However, the
temperatures are almost identical in March. From April through the rest
of the year outlet/tailrace temperature with pumping is 2 to 4 C lower
when compared to the outlet/tailrace temperatures without pumping. Outlet/
-tailrace temperatures with pumping remain almost unchanged until May,
while for the same period (April and May) outlet/tailrace temperatures
without pumping shows an increasing trend. Starting in June, temperature
both with and without pumping shows an increasing trend until October
followed by a decreasing trend from October through December. The reasons
as to why temperatures with pumping show an increasing trend from June,
while without pumping temperatures show an increasing trend in March/April
itself is explained here. As explained earlier, during daytime the
surface layer of the lower reservoir gains thermal energy mainly from
radiation. However, during nighttime this warm surface layer will be
pumped up before the surface heat can be transferred to the lower layers.
As time passes and the effect of radiation gets stronger, more of surface
heat is transferred to the lower layers even though most of it is lost due
to pumping. As a result, outlet/tailrace temperature eventually becomes
affected by radiation and starts to increase after May. With no pumping
the same effect is felt earlier.
During the fall months, heat gain within the reservoir surface is reduced
and the temperature profile becomes more uniform prior to onset of
"reservoir turnover" which happens during early winter. Temperature
difference between the surface and the outlet layers further decreases
and outlet/tailrace temperatures with and without pumping become almost
identical by mid winter. When the reservoir "turnover" occurs, outlet/
tailrace temperatures with and without pumping would be identical. The
new warming cycle begins again in April and the annual cycle is repeated.
Thermal energy of the upper reservoir gained due to pumped storage
operation will be lost to the atmosphere during winter months. Because
of this additional thermal energy in the upper reservoir, the fall
turnover of the upper reservoir will be delayed, while the lower reservoir
turns over earlier than in a typical reservoir.
CONCLUSIONS
Based on one specific pumping schedule, and one set of specific inlet,
outlet and pumping port positions, thermal structure has been established
for two deep reservoirs of the pumped storage plant. When pumping
schedules and port positions are changed, the temperature details will
also change accordingly.
For the specific case analyzed here, some interesting conclusions may be
drawn. The upper reservoir gains thermal energy by releasing cooler
water to the lower reservoir and receiving warmer water. On the contrary
the lower reservoir loses thermal energy by releasing warmer water to the
144
-------
upper reservoir and receiving cooler water. Thus, for the upper reservoir,
temperature increases with pumping while for the lower reservoir the same
is true when there is no pumping. This behavior in temperature pattern is
more significant during late spring and summer months. Outlet/tailrace
temperatures of the lower reservoir are generally lower with pumping.
Temperatures (with or without pumping) gradually increase with time.
.However, at the start of the spring cycle, there is a two month time lag.
That is, the temperatures without pumping exhibit an increasing trend
starting in March/April while the same is true with pumping in May/June.
After this, the two temperatures continuously increase at the same rate
until the end of the year, while maintaining a difference of 2 to 4°C
between them at all times. After the fall turnover the two temperatures
will be identical. Temperatures will remain identical until March of the
following year and then the new cycle starts again.
As an outcome of these considerations, it is possible to minimize the
thermal impact by installing the pumping ports of the two reservoirs at
the same depths from the surfaces so that the same amount of water with
similar temperature characteristics would be exchanged between the two
reservoirs. It should also be possible to control the reservoir and
outlet temperatues by providing an adjustable outlet/tailrace. Engineering
details of such a scheme is beyond the scope of this paper.
REFERENCES
1 Huber, W C, D R F Harleman and P J Ryan, "Temperature Prediction in
Stratified Reservoirs", Journal of the Hydraulics Division, Proceed-
ings of the American Society of Civil Engineers, April 1972.
2 Shin, J J, "The Alteration of Thermal Stratification of the Deep
Reservoir Due to the Operation of a Pumped Storage Hydropower Plant",
Engineering Foundation Conference, August 1975.
3 Ryan, P J and D R F Harleman, "Temperature Prediction in Stratified
Water Mathematical Model - User's Manual", MIT Ralph M Parson's Lab,
Supplement to Report 16130 DJH01/71, April 1971.
4 Parker, F, B A Benedict and C Tsai, "Evaluation of Mathematical
Models for Temperature Prediction in Deep Reservoirs", Environmental
Protection Agency, EPA-660/3-75-038.
5 Kennedy, R E, "Computation of Daily Insolation Energy", Bulletin,
American Met Society, Volume 30, Number 6, June 1949.
6 Huber, W D and D R F Harleman, "Laboratory and Analytical Studies
of the Thermal Stratification of Reservoirs", MIT Ralph M Parson's
Lab, TR 112, October 1968.
145
-------
UJ
10 12 14
AUG U S T
TEMPERATURE °C
16 18 20
FIGURE I MEASURED AND PREDICTED TEMPERATURE PROFILES FOR THE
UPPER RESERVOIR
PREDICT ED
MEASURED
146
-------
WITH PUMPING
WITHOUT PUMPIN6 • • • • • •
10
20
•
X
I- 40
0.
w
0 SO
60
I— '
70
4
/
r
•
/
/
/
^^
•'"
^M^
/
*
^
•
•
*
•
•
•
y
•
>
/
/
1 6 8 10 12 14 16 18 20 4 6 8 10 12 14 1
JUNE JULY
T
i —
-H
^
•
6 18 20 22 4 6 1
*
•
•
«^"
•
•
^
•
•
t
/'
1'
^
•
•
^
•
•
^
*
•
t
^
S^
•
•
i
)
,.
20
30
40
90
ftO
70
1 10 12 14 16 18 20 4 6 8 10 12 14 16 18
AUGUST SEPTEMBER
FIGURE 2 -PREDICTED RESERVOIR TEMPERATURE PROFILES WITH AND WITHOUT PUMPING FOR THE UPPER RESERVOIR
-------
0
10
20
•
* 30
X
•- 40
a.
u
0 50
h-
oo
WITH PUMPI NG
WITHOUT PUMPING
•
/
/
/ •
•>
/
•
*
•
-
•
•
•
•
•
»
»
/
/
/
/
1
1
•
/
•
•
•
•
•
•
•
•
•
f
1
S
4 6 8 10 12 14 16 18 20 4 6 8 10 12 14 16 18 20 22 4 6 I
JUNE JULY
1
\
J*
/
•
•
•
•
•
•
*
•
•
1
•
•
/
/
S
1.
•
•
•
•
•
•
•
•
•
•
•
I 10 12 14 16 18 20 4 6 8 10 12 14 16 18 20
A U G U S T SEPTENB E R
FIGURE 3 -PREDICTED RESERVOIR TEMPERATURE PROFILES WITH AND WITHOUT PUMPING FOR THE LOWER RESERVOIR
-------
16
12
o
o
IU
K
\
UJ
I-
MAR APR
MAY
JUN
JUL
AUG SEPT
OCT
NOV
DEC
WITHOUT PUMPING
— — — — - WITH PUMPING
FIGURE 4-PREDICTED LOWER RESERVOIR OUTLET TEMPERATURE PROFILES WITH AND WITHOUT
PUMPING
-------
ALGORITHMS FOR A MATHEMATICAL MODEL TO PREDICT ENVIRONMENTAL
EFFECTS FROM THERMAL DISCHARGES IN RIVERS AND IN COASTAL, AND
OFFSHORE REGIONS
J. Hauser F. Tanzer
Institut fur Physik 1. Phys. Institut
GKSS Universitat Giessen
2O54 Geesthacht/FRG 63OO L-Giessen/FRG
ABSTRACT
The ecological response of aquatic biota is a very important factor to be
considered in assessing the consequence of thermal effluents. Therefore a
model is necessary to predict the temperature as a function of time and
position in the specified region and then/ to predict the accompanying
temperature ecological effects. The corresponding transport processes are
described by partial differential equations which are solved numerically.
In the numerical simulation of these processes for a flow region with com-
plex shoreline geometries four main problems arise, viz. (a) the descrip-
tion of the boundary (b) the determination of all discrete elements lying
within the solution area together with their corresponding shapes (c) the
treatment of boundary conditions, and (d) the numerical solution of the
transport equation.
The present paper describes the development of algorithms for the solution
of all points (a) through (d) for two-dimensional multiconnected regions.
150
-------
DESCRIPTION OF THE ALGORITHMS
Mathematical models are of particular interest to assess the environmental
impact of power plant operations in tidal estuaries, coastal regions, and
rivers including the effects of recirculation and reentrainment of the
heated discharge water. The ecological response of the aquatic bioata to
the discharged cooling water is a very important factor, and a model pre-
dicting the consequences of thermal effluents will be of great use.
In Germany, at present [l] , the following standards are being applied to
once- through river cooling: maximum temperature at the end of the con-
denser discharge canal 3O °C, maximum temperature in surface water after
mixing 28 °C, maximum temperature increment in surface water after complete
mixing 5 K. The above temperature limits are valid only for waters of
sufficiently good quality. In any case, dissolved oxygen should not fall
below 5 mg/£.
In situations of practical relevance the specified river, coastal or off-
shore region will be an area, which is irregularly shaped. Hence, a com-
plete solution of the simulation problem requires algorithms for the
following steps:
(a) description of the boundary,
(b) determination of all discrete elements lying within the solution
area together with their corresponding shapes,
(c) treatment of boundary conditions,
(d) numerical solution of the transport equation.
The solution of the points (a) through (d) is performed by algorithms
SPINE, INAREA, SHAPE, and SOLVE [2].
As tidal effects play an important role the integral form (see Eq. 1 below)
of the transport equation is preferred to the differential equation because
of its conservative property; i.e. in order to allow adaptation to flow
situations with a time and spatially varying water depth. For the purpose of
this paper the transport equation is considered to be of the form
-- I y(x,t)6.V + j> Y(x,t)v • dA = $ v
-------
affect the temperature distribution. In order to avoid these errors it is
necessary to accurately model the specified region. Hence, our program
allows completely arbitrary cross-sectional areas for discrete elements,
a denotation used in [3, 4].
The user wishing to describe the boundary of his solution area (possibly
multiconnected) does this by specifying a set of closed curves. Each curve
consists of a number of segments each formed by a set of splines which
describe the boundary curve between two successive data points. At the
intersection of two segments the boundary curve need not be differentiable
as indicated in Fig. 2. In general the closed boundary curves cannot be
described as a function y = f(x), so we use a parameter representation
x(t), y(t).
Together with the data points the initial and final slopes of the tangent
for each segment have to be specified. The complete input of a sample boun-
dary curve, representing the Lower Elbe geometry between river-kilometers
641 and 667, is shown in Table 1, and Fig. 1 depicts the graphical output,
constructed by SPLNE, for this region including the grid structure utilized.
The method used in this algorithm, is a cubic spline interpolation in that
different cubic polynomials x = Xj_(t), y = y^(t) represent the boundary
curve between different pairs of data points (x.j_, yj_) - (xi+j, yi+l)- The
coefficients are chosen such as to ensure continuity of xj_(t), and yftt)
as well as the first and second derivative at each data point. However, no
system of linear equations is solved, rather the coefficients are determined
by matrix multiplication. Here it is interesting to note that the same cubic
splines are used for the spatial approximation of Eq. (1).
After the construction of the boundary all mesh points lying within the
solution area (interior mesh points) must be determined. This is performed
by algorithm INAREA. INAREA relies heavily on the spline representation,
discussed above.
A special difficulty for the algorithm arises, when the intersection point
of two segments coincides with an intersection point of a boundary curve
and a line x = x^ (Fig. 2). In order to determine whether an interval I
on that line, which normally contains several mesh points, lies within the
specified region it is necessary to know the slope of the tangent vectors
of the two intersecting segments. Here it should be noted that the boun-
dary curve - for the sake of simplicity we consider only a single curve -
is followed in a clockwise direction such that the interior of the curve
lies to the right. We now shift the origin of the coordinate system so
that its new origin coincides with the coordinates of the afore mentioned
intersection points; i.e. the interval I lies always on the negative
y'axis (Fig. 2).
Since the solution area lies to the right of the curve, we can state the
following rule (Fig. 2).
Assuming the tangent of the second segment to be fixed, and
rotating the tangent of the first segment counter-clockwise
(positive sense) until they coincide, an interval I belongs
to the solution area if the negative y'axis is intersected.
152
-------
Fig. 3 shows three sample cases and Fig. 4 shows the computer printout for
the Lower Elbe geometry. An asterisk indicates an interior mesh point.
One of the fundamental advantages of the employed method is its complete
flexibility in considering discrete elements of any arbitrary geometry.
Although the general formulation uses a Cartesian index (i, j) the method
does not require the approximation of a boundary by rectangular elements.
SHAPE constructs discrete elements exactly matching an irregular boundary.
Here, again, the spline representation is essential. The enclosure surface
of each discrete element may be computed using the spline representation.
From the algorithm INAREA we know all interior mesh points. To each interior
mesh point we associate a discrete element, here called an incomplete dis-
crete element and denoted by IDEj^j. However, the area covered by all these
incomplete discrete elements is less than the solution area due to the
omission of the boundary areas BA-j^ j (Fig. 5) . We wish to extend and reshape
all incomplete discrete elements out to the boundary curve, thus construct-
ing a set of variable-sized, irregular-shaped discrete elements that span
the specified region. •
A boundary area is associated to an incomplete discrete element if their
common side length has the maximal value out of all IDEs also adjacent to
this boundary area. The principle of this relation is shown in Fig. 5. Since
a discrete element is three-dimensional we denote its cross-section, as
depicted in Fig. 5, by SDEj^ j which stands for surface discrete element.
Since the solution algorithm used reduces Eq. (1) to a system of coupled,
ordinary first-order equations in time, it is necessary to specify initial
conditions at all discrete elements. On the other hand, according to the
formulation of the method , there are no boundary conditions in the sense
of boundary value problems, since the conservation laws are integrated over
each discrete element or its enclosure surface; i.e. the effects of the
boundary are imposed on the half-point surfaces of the discrete element or
on the irregular-shaped boundary of a boundary discrete element. For com-
putational simplicity we assume that a fluid boundary coincides with lines
x = Xj+]/2 or y = yj+i/2- This is in no way a restriction, since a fluid
boundary can be freely chosen. Values on the half -point boundary, however,
cannot be approximated by second upwind differencing, since the correspond-
ing adjacent discrete element may lie outside the solution area. In the
case of temperature we obtain the values for the temperature and the normal
derivatives on half-point surfaces by a simple Taylor expansion having
second order accuracy. Eqs. (2) then represent the general mathematical
formulation of our system.
3Y. . (t)
For the spatial approximation of the right-hand side of Eqs. (2) , performed
by algorithm SOLVE, one can use the donor cell method or second upwind dif-
ferencing which is both conservative and transportive. However, this method
is only of second order accuracy when employed to the approximation of half
point values, since the mathematical forms are based on linear averaging
153
-------
between two discrete elements. Considering the approximation of derivates
gw SV-,
hr- i T—j these are approximated by finite forms which are only of first-
order accucary.
It is, however, desirable, as is pointed out in [5], to improve the spatial
approximation, since this will lead to considerably increased time steps.
Reducing the mesh size from AXj to Ax2 in a two-dimensional space, and
using the FTCS explicit method for the diffusion problem, increases the
computer time by a factor (Ax,/Ax2)'+. Hence, it is clear that methods with
improved accuracy allowing greater mesh sizes are highly desirable. For
this purpose we again use the cubic spline representation to improve the
accuracy for the approximation of half-point values and derivatives. Dis-
crete points of support are the center-points (mesh points) of the discrete
elements. To obtain computational simplicity through decoupling of coordi-
nates a cubic spline function is constructed separately for each row and
column.
The final result is a numerical procedure having fourth-order accuracy for
the approximation of half-point values and third-order accucary for approxi-
^m J^VI/..
mation of derivatives (—, —} , that is the truncation error is of degree
two higher than for the donor cell method.
In conclusion, the above mentioned algorithms SPLNE, INAREA, SHAPE, and
SOLVE can be used for all transport processes including complex shoreline
geometries. The detailed results and mathematical formulations of this
extensive study are being prepared for presentation.
ACKNOWLEDGEMENTS
We would like to thank A. MULLER, GKSS for a number of valuable suggestions.
We also would like to acknowledge the assistance provided by B. MITTELSTAEDT,
GKSS in the computation of the various figures and tables.
154
-------
REFERENCES
[1] LAWA
[2] HAUSER, J.
TANZER, F.
[3] ERASLAN, H.A. -
[4] ERASLAN, H.A.
KIM, K.H.
HARRIS, J.L.
[5] ROACHE, P.
Grundlagen fur die Beurteilung der Warmebelastuiig
von Gewassern, 2. verbesserte Auflage 1977, pp.116
Algorithms for the Different Steps in tht= Numex al
Solution of Partial Differential Equations and
Application to the Design of an Ion Extraction System,
IEEE International Conference on Plasma Science,
15-17 May 1978, Monterey, California
A Transient, Two-Dimensional, Discrete-Element,
Far-Field Model for Thermal Impact Analyses of Power
Plant Discharges in Coastal and Offshore Regions.
Part 1: General Description of the Mathematical Model
and the Results of an Application, ORNL-494O, Feb.
1975, Oak Ridge
Systematic Application of Transient, Multi-Dimensional
Models for Complete Analysis of Thermal Impact in
Regions with Severe Reversing Flow Conditions,
Waste Heat Management and Utilization, 9-11
May 1977, Miami Beach, Florida
Computational Fluid Dynamics, Hermosa Publishers,
1976
155
-------
* H £ T R A N -GiSS- 79/03/27/
* LOWER ELgf Pl.-'FR FR'1" STRfAM-KM
DATA TQ SPFC'rY
AREA. THE FIRS? !••:•. MRcR u F
CAR!) INDtCATFS A SCLlD (1)
OR A FLU;i) (0! R"",,M)ARY.
UnUflOARV C.'NPI T !r~\JS,THcN,CALCUl. ATI D
pkOM TMESF INFORMATIONS.
TMF FOLL:'«JNG THREF NUMBERS MAY
BF USED TO STATh SPFCUl
C3NDITIUMS FHR rFMPHRATURF AND
VEU1CITY CHMPONEMTS.
THE FOLLOWING TWn NljMRFRS SPECIFY
X AND Y CJOROINATEStRESPECTIVELY.
EACH CURVF IS RUN ALONG IN A
CLPCKWISE DIRECTION.
777 END OF S
888 END OF BOUNDARY CURVF.
999 END HF C3GRDINATE INPUT
1. CURVE
1. SEGMENT
1 101 101 101 0. 15.3
1 101 101 101 7.1 18.9
1 1. SEGMENT PINNAU
0 inl 101 101 29.4 16.6
777
* 6. SEGMENT
1 101 101 101 31.6 16.4
1 101 101 IPl 34.0 12.5
1 101 101 101 38.2 10.9
1 101 101 101 42.1 10.4
1 101 101 101 45.7 11.1
777
* 7. SEGMENT
1 101 101 101 45.7 10.6
777
* 8. SEGMFNT
1 Ifll 101 101 43. S 8.8
1 101 101 101 44.4 7.6
1 101 101 101 44.6 5.7
777
* 9. SEGMENT
1 101 101 101 45.2 4.7
777
* 10. SEGMENT
1 101 101 101 53.2 7.3
6S.4
60.9
57.5
56.3
53.8
49.6
47.8
44.5
39.2
5.3
2.2
2.0
1.4
1.4
1.7
0.6
0.4
2.1
1 Iftl 1H 101 59.6 8.4
777
* 11. SEGMENT
1 101 101 131 59.4 10.2
777
+ 12. SEGMENT
1 iol 101 m 67.3 i?,3
1 l.ll 1»1 101 75.8 1?.7
1 101 101 101 80.0 16.0
777
* 13.
0 101 101 101 80.0 8.0
777
* 14. SEGMENT
1 101 101 101 78.7 7.7
I 101 101 101 73.4 7.5
1 101 101 101 70.3 6.3
777
* is. SEGMENT LUEHE
0 101 101 101 70.0 6.3
777
* 16. SEGMENT
1 101 101 101
1 101 101 101
1 101 101 101
1 101 101 101
1 101 101 101
1 101 101 101
1 181 101 101
1 101 101 101
1 101 101 101
777
* 17. SEGMENT SCHWINGE
0 101 101 101 39.0 2.0
777
* 18. SEGMENT
1 101 101 101 34.0 3.0
1 101 101 101 29.9 4.7
1 101 101 101 24.5 6.5
1 101 101 101 20.1 8.0
1 101 101 101 17.7 8.3
1 101 101 101 15.5 8.6
1 101 101 101 9.1 8.0
1 101 101 101 0.6 7.2
777
* 19. SEGMENT
0 101 101 101 0.0 15.3
• 88
* 2. CURVE PAGENSANO
* 20. SEGMENT
1 |01 101 101 11.5 16.8
1 101 101 101 18.4 16.4
1 |01 101 101 21.7 16.3
1 101 101 101 22.8 15.8
1 101 101 101 23.8 15.4
1 101 101 101 26.0 13.4
1 101 101 101 29.4 10.2
1 \01 101 101 19.4 14.2
1 101 101 101 13.7 16.1
1 101 101 101 11.2 16.2
1 101 101 101 11.5 16.8
• 88
* 3. CURVE
* 21. SEGMENT
AUGBERG'OROMMEL
Table 1: Input data for the Lower Elbe geometry
156
-------
1
1
1
1
1
777
*
1
777
*
1
1
1
1
1
1
88*
I'M
101
101
ioi
ioi
101
101
im
101
mi
ioi
101
KM
101
KM
KM
I'M
22.
101
23.
101
I'M
I'M
KM
KM
I'M
1' I
I'M
Kl
I'M
I'M
SFC,".
101
S '•;-,->
l'H
KM
KM
Ml
11!
1 M
52
37
42
43
•«4
E'JT
43
ENT
41
38
35
33
35
32
. ;
.4
. o
.9
. 4
.6
.5
.?
.7
j
. 1
-»
1
9
3
0
ft
5
(S
f*
-7
9
L
;
' .4
.2
.5
.C
.6
.5
.9
.7
.4
-1
J. I
7 .4
* 4. njww ; I in HE SAND
*
1
1
1
I
1
1
1
888
*
*
1
1
1
101
101
101
101
101
101
ioi
101
m
UM
1 I'M
1
777
*
1
777
*
1
1
1
1
999
*
*
101
I'M
1.11
101
KM
101
I HI
SE3
?«.
101
101
1M
101
101
liM
101
5. C
25.
I'M
1-1
I'M
101
I'M
26.
KM
27.
101
I'M
1CI
I'M
Tlil.
MEMT
sr ;>•
I'M
IM
1-il
10 I
I'M
KM
101
U '-1 V ^
Silo"
1 M
1"!
KM
101
1% 1
S! G"
1 M
Ssf,«
i :< i
KM
1J1
IM
A it-
Mr
FMT
53
57
•)9
ft3
4
53
. 0
.1
.T
.3
.1
. 1
.f.
'i
4
5
5
3
2
3
.4
.7
. I
.3
.3
.8
.4
-ANSKALRSANr
h \' 1
76
7ft
7V
77 ,
•10
FM
30
E'MT
M
77
76
76
: !'•
GKF^
* MEASURED PRDM T
_ 3
. q
.2
s
.0
.0
.3
.1
.4
.3
,.
S I
iK
0
!;
')
10
1
I
1
9
9
9
S
• .'i
.5
.8
.9
. ;
,7.6
1.5
1.^
.6
.3
.5
LHPP
•^GL?S
X-AX!S
FOR
ARE
i
POSIT!'
CnJNTER-CLICKWTSf
27 is
F;,CH
:S ARE
Sj
E IN
HRECTI.-1N.
0
78
-93
-74
-6
-85
44
121
18
91
'l
14
-90
-IfS'.
•168 -
-168
-169 -
171 -
90
3
-34
-U*
14h
13
34
-TO
-Ml
0
12
-*3
-7*
10
-?5
96
m
9
M i
' O
37
-
-------
Fig. 1: Top view of the Lower Elbe River between river-kilometers 641
and 667 including the grid structure utilized
Fig. 2: Determination of interior mesh points. The shaded area denotes
the solution area S
158
-------
Segl
I--S
I&S
IfiS
Fig. 3: Sample cases for the location of the solution area with respect
to interval I. The shaded areas denote the solution areas
m .
1 -
Fig. 5: Construction of discrete elements DE. .. For instance,
SDE = IDE0 U BA U BA and SD"E = IDE u BA
£ , £ £. , /. , 3 ->»-> -> i 4
159
-------
*•«•*»»*»•••
Fig. 4: Computer printout for the Lower Elbe geometry. An asterisk marks an interior mesh point
-------
EFFECT OF SALT UPON HOT-WATER DISPERSION IN WELL-MIXED
ESTUARIES. PART 2. LATERAL DISPERSION
R. Smith
Department of Applied Mathematics and Theoretical Physics,
University of Cambridge, U.K.
ABSTRACT
The dispersion of heat in shallow water differs from that of
passive contaminants in that the buoyancy can modify the flow
and hence can effect the dispersion. This has the serious
implication that on-site experiments with dyes are of limited
use in helping to predict the thermal impact of any proposed
industrial developments. For estuaries the task of modelling
the dispersion of heat is further complicated by the fact that
there are already buoyancy-driven currents due to the salinity
•distribution. Thus, not only does the heat change the flow
directly, but also it changes the flow indirectly by modifying
the salinity distribution. Here it is shown that in a
vertically well-mixed estuary the interactions are such as to
disperse heat preferentially towards the shoreline for an ebb
tide and away from the shoreline for a flood tide. In extreme
cases the asymmetry in the lateral dispersion coefficient
can be several hundred percent.
INTRODUCTION
From dye experiments in shallow water there is ample evidence
of the diffusion-like character of the dispersion process Ql^J.
Thus the natural mathematical model with which to describe
dispersion is a linear diffusion equation with an appropriately
chosen value of the effective diffusivity. Indeed, it has
become conventional to quantify dispersion in terms of this so-
called dispersion coefficient.
The explanation as to why a diffusion equation should arise,
rather than some more complicated aquation, dates back to a
classic paper by G.I. Taylor in 1921 [2]]. However, it was
not until several decades later that he was able to give a
quantitative theory for the particular case of a neutrally
buoyant contaminant in pipe flow £3, ^3 • This work revealed
that the dispersion coefficient depends upon effects, such
as diffusion perpendicular to the dispersion direction, which
are negligible as regards to other aspects of the flow.
Furthermore, contrary to intuition, the dispersion coefficient
161
-------
reduces as the transverse diffusion increases.
It is this dependence upon readily overlooked effects that
has made it so difficult for engineers to conjecture simple
empirical formulae for dispersion coefficients {[5J • For
example, Fischer's £6j calculated and observed values for the
longitudinal dispersion coefficient in non-rectangular
channels can be several orders of magnitude greater than
Elder's [7J results for channels of constant depth.
Buoyancy effects
Another facet of this sensitivity of the dispersion process
is that the dispersion of a buoyant contaminant, such as hecit
or fresh water, is different from that of a passive contaminant.
Close to an outfall the induced vertical stratification
-suppresses the turbulence and can delay considerably the
achievement" of complete vertical mixing [8j . Then further
away from an outfall the difference in hydrostatic pressure
between the centre of a buoyant plume and the surrounding
denser water drives a secondary flow (see Fig . l) which
accelerates the lateral dispersion £9] . Finally, in confined
flows when cross-sectional mixing is not quite complete, this
same secondary flow augments the transverse turbulent diffusion
r T
and leads to a reduced rate of longitudinal dispersion L10J .
The net result is that on-site experiments with dyes are of
limited use in helping to predict the thermal impact of any
proposed industrial development.
In an important paper, Erdogan and Chatwin Fll] extended
Taylor's analysis (jjQ to apply to buoyant contaminants in pipe
flow. They found that the dispersion is still governed by a
diffusion equation, but with the significant difference that
±he dispersion coefficient is a nonlinear function of the den-
sity gradient. For reasons of (vertical) symmetry the
dispersion coefficient is an even function. Thus, for weak
gradients the contaminant distribution c(z,t) satisfies the
^'Erdogan-Chatwin" equation
C4- ~ I \uo + Coccz) ua ) C,L (1)
t \ *.*/«. /2.
Here D0 is the dispersion coefficient for passive contaminants,
o( is the constant of proportionality between the density
change and the concentration c, and D is a constant.
The same equation (1 ) was derived by Prych [9] for the lateral
dispersion of buoyant contaminants in open-channel flow, t
being the advection time downstream of the outfall and z being
the cross-stream coordinate. In that context there is no
need for the restrictive assumption that the gradients be weak.
162
-------
Unfortunately, there is poor agreement between Prych's theory
and his experiments. Smith [ia| showed that this is remedied
when Prych1s "laminar" calculation of the D2 coefficient is
replaced by a more appropriate "turbulent" calculation (see
Fig. 2). The revised and much increased estimate for Dp is
(a)
where h is the water depth, g the acceleration due to gravity,
.* the friction velocity, and k the von Karman constant
; typically .4). Prom a wide range of experimental results
Ijfj a reasonable formula for Do is
v.-.
I
Heat and Salt
Por a thermal dispart •» situated in an estuary there is
dispersion of two contaminants, heat and salt, both of which
change the density. If tHe turbulent intensity were weak,
then the different Lap 'iffusivities of heat and salt
could give rise t«j th cacular double-diffusive phenomena
described in cha1 . Turner's monograph £l43 . Here we
shall restrict : tontion to vertically well-mixed estuaries,
where by implica ~oi, the flow is highly turbulent and the eddy
diffusivities for Leau and salt are approximately equal. Thus,
although there are interactions between the heat and salinity
distributions, the v a.-ie not so severe as to lead to the
steepening of temj ;rature gradients.
The assumption in this paper is that the dispersion takes
place predominanatlj; across the estuary, the case of longi-
tudinal dispersion Having been addressed in part 1 yL5j| •
Thus, here .we are concerned with the middle field, where
vertical mixing has been achieved but the hot-water plume
extends only a small way across the estuary. Other implicit
assumptions are that the estuary is sufficiently straight
and free from jetties for the primary flow to be unidirectional.
The mathematical task therefore is to extend equation (1 ) to
incorporate a second contaminant and to investigate the nature
of the solutions.
A striking feature of the solutions obtained below is that
the interactions between heat and salt lead to the heat's being
preferentially dispersed towards the saltier region of the
estuary. To a first approximation the salinity distribution
in a wide estuary is carried along with the tidal current.
Thus, at high tide the fluid in the faster-moving, deeper
163
-------
central region of the estuary originates further out to sea,
and is saltier, than the fluid close to the shore. This means
that at high tide a hot-water plume in a wide estuary would
tend to be deflected away from the shoreline. At low tide
the opposite is the case. For narrower estuaries (less than
about 200m wide) the more complete cross-sectional mixing of
salt means that the lateral salinity gradient depends upon
the tidal direction rather than the tidal height L6J • Thus,
in a narrow estuary heat is dispersed preferentially towards
-the shoreline when the tide is going out.
QUALITATIVE DEDUCTIONS
Although the heat and salt both modify the flow, there is
never- the -less a single flow which disperses the two contam-
inants. Thus, if the turbulent eddy diffusivities for the
two constituents are equal, then it follows that the two
dispersion processes are dynamically identical , and that there
is a common dispersion coefficient for both heat and salt. In
view of the results (1,2,3) f°r a single buoyant contaminant,
we infer that the common dispersion coefficient is simply a
function u.f the total density gradient
The opposite signs of the p and oi. terms allow for the fact
that the salinity s and the temperature 0 have opposite
«ffects upon the water density.
Even without any calculations we can now infer that heat is
preferentially dispersed across the estuary in the direction
of increasing salinity. The effect of heat is to reduce the
•density locally; this results in a locally reduced density
gradient on the fresh side of the plume and an increased
gradient on the salty side. The dispersion coefficient is a
xapidly increasing function of the density gradient. Thus,
the dispersion of heat is reduced on the fresh side of the
plume and is increased on the salty side.
Modified salinity distribution
A similar argument enables us to infer that the middle of the
plume become saltier with a compensating reduced salinity at
the tails of the temperature distribution. Upstream of the
tmtfall the salinity distribution across the estuary is in
•equilibrium, the cross-stream flux of salt being balanced by
the non-uniform longitudinal advection |[6] • This equilibrium
is disturbed by the hot-water plume. At the fresh side of the
plume the lateral dispersion of salt is reduced. Thus the
164
-------
flux of salt; from tlie centre of the plume is less than that
needed to maintain the equilibrium of the salinity distri-
bution, and the water is therefore even fresher than it would
have been in the absence of the plume. At the salty side of
the plume the flux of salt towards the centre of the plume is
in excess of the equilibrium flux, and there too the water is
fresher than it would have been in the absence of the plume.
The missing salt accumulates in the central part of the plume
(see Fig.3).
It is this perturbation of the salinity distribution that is
the distinctive feature of hot—water dispersion in estuaries.
Not only is the lateral dispersion coefficient (4) changed by
the heat directly, but also it is changed indirectly by the
thermally-induced salinity perturbation. Thus, we cannot
study the temperature distribution without simultaneously
determining the salinity distribution.
DISPERSION EQUATIONS
In order to focus attention upon the tuermally-induced
changes we regard .the pre-existing (locally constant) salinity
gradient 1T^ as being known. It then follows that the
salinity anomaly s1 and the temperature © satisfy the coupled
nonlinear diffusion equations
et -
<7>
The form (7) of the dispersion coefficient makes it convenient
to change the dependent variables from s' and Q to ^>x and
© , where o' is the density perturbation. Thus equation
(5) is replaced by
i \
(8)
where the "pseudo-dispersion coefficient" D1 for the composite
quantity p' is given by
165
-------
The virtue of this change of variables is that there is no
coupling between the 0 distribution and the nonlinear diffus-
ion equation for p' .
We note that D' is always positive. This makes^it possible
for us to infer that for a buoyant discharge ^ will be
everywhere negative. The asymmetry of D1 means that pertur-
bation density gradients, with the same sense as the pre-
existing density gradient a , will be dispersed most
rapidly. Consequently, the $ distribution is skew (see
Fig.3) and the centroid moves towards the salty side of the
plume. On average D' exceeds the dispersion coefficient D
for heat. Thus the temperature distribution is narrower and
more highly peaked than the o' distribution. The physical
reason for this difference is that in the centre of the plume
the accumulation of salt counters the density-lowering effect
of heat, while at the edges of the plume the reduced salinity
ajod thfe heat both tend to lower the density.
With o' and 0 as the dependent variables, the dispersion
coefficient D becomes independent of 0
i.e.
Thus, once equation (8) has been solved for the density
perturbation p' , we can evaluate D and equation (6) becomes
a. linear equation for © . The only difficulty is that created
by the dispersion coefficient's being non-constant. For
example, if in the absence of heat the salinity-driven second-
ary flow doubles the lateral dispersion coefficient (i.e.
~ LX ) and if the presence of the plume leads
_ O
to perturbation density gradients of i-sPv » then D varies by
a factor of 2.6. *\*-
SERMITE SERIES REPRESENTATION
Formally the equation (8) for o only differs from the Erdogan-
Chatwin equation (1 ) in that tlie dispersion coefficient has
linear as well as quadratic dependence upon the perturbation
density gradient. Thus, we can use to advantage the mathemat-
ical methods recently us<=>d by the author to study the Erdogan
Chatwin equation
Extremely far downstream we can expect the concentration
gradients to become weak, and equations (6,8) both to reduce
to constant coefficient diffusion equations. In turn this
Implies that the concentration profiles eventually become
166
-------
Gaussian (assuming that the shorelines are distant). A
natural way of representing the evolution towards normal
distributions is by means of Hermite series
-i^ . (12)
Here S and Q are the rates at which salt and heat are being
discharged from the outfall, h is the water depth and u the
locally constant tidal velocity. By definition the leading
coefficients a0 , bo are equal to 1. Also, if O"(t) and Uj(t)
are chosen to be the exact standard deviations then ag and b
are both zero. the formulae for the first few Hermite poly-
nomials Hen(x) are
*- 1
(13)
Representing tlie nonlinear terms
Our object is to replace all the terms in the dispersion
equations for the density anomaly and for temperature by
Hermite series of the forms (ll) and (12) respectively. It is
only the G/ dependence of the dispersion coefficients which
presents any difficulties. For the ^^ term this entails the
introduction of the constants G(i,j;k,n) such that
r -, n
y.2J . Since JJ has either of the values UJ /0" or 1 , we
shall simply denote the corresponding constants by G and G1 .
For the o^ terms in DT and D we introduce the further cons
tants F1 and F where
-i! MS F(C;k,n)Hcn(x) 5)
167
-------
Jin l2 it is explained how G can be evaluated from Mehler1 s
formula
The same methods permit F to be evaluated. The properties
which we shall subsequently make use of a~"e that
(17)
(18)
(19)
and that F, G are zero when the sum of their arguments is odd.
Proceeding as in |l2]] , we achieve the desired Hermite series
representations of the dispersion equations (8) and (6).
From the Hen coefficients we obtain equations for the rates
of change of an(t) and bn(t):
L
h u
"' k ^.r.k.n-Oa^q.
- • a
, ,
'
(2l)
168
-------
*Qf n!Y
* »
TX
•»t •«./ A " ~" ^ Jl>" V •' ^i^j "•»_! "Vl
„
(22)
recall -that P1 , G' are used to denote tlie case i/ = 1 , while
F, G denote the case JJ-^/
-------
For practical purposes it often suffices to know the simplest
global properties of the contaminant distribution, such as the
centroid and the standard deviation. Thus, if a2 and b2 are
chosen to be zero, then the required information is contained
in the four quantities a-| , O"1, b-| , W* • . With the particular
.truncation described above, the n = 1, 2 equations involve just
"these key variables :
A -n"1 1 H u o-3 { 3)
,
hu
-------
rate of spreading of a plume [16] , but also their interactions
.make the plume gr ^ even faster.
ILLUSTRATIVE EXAMPLE
Equations (23-26) represent a considerable simplification
from the coupled dispersion equations (5-7), which in turn
are vastly more simple than the full equations of motion for
two buoyant contaminants in a turbulent flow. Furthermore, as
is evidenced by the results shown in figure 2, there is not an
undue loss of accuracy. The simplifications result more from
focussing attention upon what is regarded as being the
primary information, rather than from the approximations which
are made. The limiting factor to this process of simplifi-
cation is the intrinsic complexity of the interactions between
lie at and salt.
Equations (23-26) do not seem to admit of exact analytic
solutions. Also, the large number' of degrees of freedom (e.g.
lateral salinity gradient, source strength, initial standard
deviation) precludes any simple non-dimensional classification
of numerical results. Instead, we shall illustrate the effects
of salt upon the dispersion of heat for an arbitrarily selec-
ted situation.
Figure 4 shows the calculated shape of the temperature plume
(i.e. one standard deviation either side of the centroid) for
a 2OM¥ heat discharge with initial standard deviation 5m when
the estuary conditions are:
I = I %o km"1.
There is a ten-to-one exaggeration of the plume width. For
comparison purposes the plume shapes are also shown for an
untie a ted discharge, and for a heated discharge in the absence
of any salinity gradient.
Several general features are worthy of comment. First, close
to the outfall the rate of spreading, but not the sideways
displacement, is independent of the salinity gradient. Second-
ly, at moderate distances the rate of spreading is significan-
tly greater than that in either of the two limiting cases of
Jio heat or of no salinity gradient. Thirdly, far from the
outfall the plume evolves as per a dye plume emitted into the
estuary from a virtual source far upstream of the actual source
position and displaced into the saltier water. In the
particular case shown in figure 4, the virtual source would be
Dearly 500m upstream and displaced by 9m away from the outfall
into the saltier water.
171
-------
CONCUODING REMARKS
The parameter values in the illustrative example are quite
modest, yet the effect of salt upon heat dispersion remains
very noticeable even a kilometre from the outfall. Nuclear
power stations can have cooling requirements several orders
of magnitude larger than the example [ifj. Thus the effect
discussed in this paper would remain significant at much
greater distances from the outfall.
Two methods which have been used to predict the thermal impact.
of hot—water discharges in estuaries are: on -site dye
experiments, and hydraulic scale models. The first of these
neglects any effects due to the buoyancy of the discharge,
while the second method neglects any effects of the salinity
distribution (due to scaling difficulties). The present work
reveals that both of these methods err on the side of safety
ant? under-estimate the overall dispersion. Furthermore, the
effects of salt and heat are not merely additive, their
interactions also contribute to the rate of spreading of the
plume. In the proximity of a shoreline, another important
aspect of the effect of salt upon dispersion of heat is that
the plume tends to be deflected towards the saltier side.
Depending upon the state of the tide, this can lead to higher
shoreline temperatures than might have been expected.
ACKNOWLEDGEMENT
The author would like to thank the Central Electricity
Generating Board for financial support through the award of a
research fellowship, and for a travel grant to attend this
conference.
APPENDIX - CONCENTRATION PROFILES
Although the position and width of the plume are the most
important and useful items of information, it is never-the-
less of interest to determine the concentration profiles in
more detail. For any particular initial discharge profile this
could be done by solving numerically the nonlinear diffusion
equation (8) for the density anomaly p' , and the coupled
linear diffusion equation (6) for the temperature 9 . How-
ever, if we are to distinguish general features from the
particular, then solutions would need to be computed for a
range of situations. Here we take a simpler course of action
and seek instead approximate analytic solutions valid far
downstream of the outfall (assuming, as above, that shorelines
are distant).
177
-------
/dlcr1 -i- ^OQ0 = (F'terms)/cra + (&
For the density anomaly ^ , the starting point for our
calculations is the equation (21 ) for the Hermite coefficients
Qft . If we take CT to be the exact standard deviation,
•then a2 = 0 and the n = 2 equation can be used to replace t by
CT as the independent variable. The resulting equations for
a , a , ai • • • • take the form
rms)/^ (27)
where the right-hand-side terms involve summations of double
and triple products of the Hermite coefficients. These terms
are even more cubersome than those in equation (21 ) due to the
change of independent variable.
.An order-of -magnitude estimate of the solutions of equations
{27) suggests that an decays at the same rate as the larger of
•O" and the nonlinear forcing. In particular, ao = 1 (by
definition), O~QA tends to a constant, and the higher-degree
icoefficients are, at most, of order- o-""3t
These qualitative results, together with the explicit formulae
^18-20} for the constants F' , permit us to evaluate the domin-
ant contributions to the forcing terms in equation (27). It
is then straightforward to solve for the asymptotic approxim-
ations to the Hermite coefficients a^ , a^ et seq.
,
(28)
3 «
We observe that the odd coefficients decay at the rate 0~ ,
while the even coefficients decay at the faster rate O"~^ .
Hence, sufficiently far downstream the concentration profile is
determined by just the odd coefficients. Furthermore, from
equation (28) we see that these coefficients depend only upon
such bulk properties as the buoyant output. Thus, the shape
of the asymptotic concentration profile is a general feature,
and for a given flow situation the width scale varies as the
square-root of the buoyancy output.
Prom equation (29) we see that for the even Hermite coefficients
the influence of the detailed conditions close to the outfall
are contained in the single factor (CTC^). Thus, to ipiprove
upon the asymptotic concentration profile it suffices to
173
-------
determine this one extra piece of information. More crucially,
it is the even n = 2 equation which determines the standard
deviation of the plume :
Thus, the asymptotic influence of the buoyancy upon the plume
width is proportional to ( O~QA). Hence, it is not possible
to calculate O~ a accurately without also calculating (crQ^).
This is the motivation for the two—term approximation used in
the main body of the paper. However, the range of validity
of equations (23-26) used above is not necessarily restricted
to asymptotically large distances downstream of the outfall.
For the temperature 0 , the starting point for the calculations
is equation (22) for the Hermite coefficients b^ . The
pattern of calculations is as described above, the presence
of the 0~* terms being dealt with simply by noting that
asymptotes to a constant. The resulting asymptotic approxim-
ations to the Hermite coefficients are that
tant , \ = 0 , <31>
The corresponding asymptotic approximation to the equation for
the standard deviation of the temperature distribution is
174
-------
u
(35)
Arguing as above, we assert that it is not possible to calculate
accurately the standard deviation vU of the temperature distri-
bution without also calculating the centroid position ( W b^ )
and the standard deviation and centroid of the density pertur-
bation. Hence the need for the four equations (23-26) to give
a quantitative mathematical description of the effects of salt
upon the lateral dispersion of hot water.
REFERENCES
1. J.W. Talbot and G.A. Talbot "Diffusion in shallow seas
.and in English coastal and estuarine waters". Rapport.
P.—v. Reunion,Conseil international pour 1'Exploration
de la Her 167 (197*0 93-110.
2. G.I. Taylor. "Diffusion by continuous movements".
Proc. Lond. Math. Soc. Ser. 2. 20 (l92l) 196-212.
3. G.I. Taylor. "Dispersion of soluble matter in solvent
flowing slowly through a tube". Proceedings of the Royal
Society A 219 (1953) 186-203.
•4. G.I. Taylor. "The dispersion of matter in turbulent flow
through a pipe". Proceedings of the Royal Society A 223
(1954) 446-468.
5« H. Liu. "Predicting dispersion coefficient of streams".
J. Environ. Eng. Div. A.S.C.E. 103 (1977) 59-69.
6. H.B. Fischer. "The mechanics of dispersion in natural
streams." J. Hyraul. Div. A.S.C.E. 93 (1967) 187-216.
7. J.W. Elder. "The dispersion of marked fluid in turbulent
shear flow." J. Fluid Mech. 5_ (1959) 544-560.
b. G.C.C. Parker. "Study of factors affecting sea water
temperatures at Sizewell power station - An appraisal of
the 1975 Sizewell hydrographic survey data." C.E.G.B.
report: GDCD (1977).
9. E.A. Prych. "Effects of density differences on lateral
mixing in open channel flows." Keck. Laboratory of Hydr-
aulics and Water Resources, California Institute of Techn-
ology Report KH-R-21 (1970).
175
-------
10. R. Smith "Longitudinal dispersion of a buoyant contaminant
in a shallow channel." J. Fluid Mech. 78 (l9?6) 677-689.
11. M.E. Erdogan and P.O. Chatwin. "The effects of curvature
and buoyancy on the laminar dispersion of solute in a
horizontal tube." J. Fluid Mech. 29 (1967) 465-484.
12< R. Smith "Asymptotic solutions of the Erdogan-Chatwin
equation." J. Fluid Mech. 88_ (1978) 323-337.
13- H.B. Fischer. "Longitudinal dispersion and turbulent
mixing in open channel flow." Ann. Rev. Fluid Mech. 5_
(1973) 59-78.
14. J.S. Turner. Buoyancy Effects in Fluids. Cambridge
University Press (1973).
15« R. Smith "Effect of salt upon hot-water dispersion in
well-mixed estuaries. Part 1. Longitudinal dispersion."
Estuarine and Coastal Marine Science 6^ (1978).
S.M. Sumer "Transverse dispersion in partially stratified
tidal flow." Hydraulic Engineering Laboratory, California
Berkeley, Report WHM-20 (1976).
176
-------
DENSE
A////////////////////////////////
Figure 1. Cross section illustrating the secondary flow.
30
10
8 6
2 3
c
o
1 '
1
5 0-6
0-3
0-1
I '
60 100 300 600 1000 3000 6000 10000
Dimensionless source strength
Figure 2. Comparison between Smith's \12] theory and
Prych's [9] experimental results for the excess variance
due to buoyancy as a function of the strength and width
of the source.
177
-------
DENSITY
CHANGE
P'
Figure 3. Distributions of salinity, temperature and density
anomaly across a hot-water plume in an estuary.
AOm-r
SALTIER WATER
500m
1km
25m L FRESHER WATER "•
Figure 4. Plume shapes for a hot-water discharge in an
estuary.
178
-------
COST-EFFECTIVE MATHEMATICAL MODELING FOR THE ASSESSMENT
OF HYDRODYNAMIC AND THERMAL IMPACT OF POWER PLANT
OPERATIONS ON CONTROLLED-FLOW RESERV-JIM'S *' *
3 > U 5
A. H. Eraslan and K. H. Kim
The University of Tennessee
and
i
Oak Ridge National Laboratory
The operation of a large power plant can substantially alter the
natural h'ydrodynamic and thermal conditions in a controlled-flow
reservoir depending on the reversing flow conditions, and the condenser
flow rate and the condenser temperature rise of the cooling water
system. The realistic assessment of the thermal impact usually requires
long-term (days) real-time simulations of the hydrodynaraic and thermal
cnditions in the reservoir. The direct applications of the transient,
multi-dimensional, coupled hydrodynamic and thermal mathematical models
(without exception) require large amounts of computer time (dictated by
the computational stability requirements of the hydrodynamic solution)
in order to be able to produce results with acceptable quality of
resolution in sufficiently large regions^ In the licensing of a
proposed power plant, the assessment of its potential thermal impact is
usually based on analyses which must consider a multitude of scenarios
for natural and plant operational conditions. Hence, from computer cost
considerations, the repetative applications of the multi-dimensional,
coupled, hydrodynamic and thermal models generally become extremely
costly (usually prohibitive) for simulating the required number of
different scenarios in the assessment of the thermal impact.
An alternate approach to the assessment of the thermal impact of
power plant operations on controlled-flow reservoirs is to employ the
recently-developed zone-matching methodology [1] based on the systematic
applications of different transient, multi-dimensional, discrete-element
hydrodynamic and thermal transport models [2-91- The application of the
Research supported by:
U.S. Nuclear Regulatory Commission, Health and Environmental Research
2Branch, Office of Nuclear Regulatory Research
U.S. Department of the Interior, National Power Plant Team, Fish and
3Wildlife Service, Office of Biological Services
Professor and Director, Environmental Impact Project, Department of
Engineering Science and Mechanics
Consultant and Project Director, Unified Transport Approach for the
5Assessment of Power Plant Impact, Energy Division
Associate Director and Research Associate Professor, Environmental
Impact Project, Department of Engineering Science and Iv-chanics
179
-------
zone-mathcing methodology enables the implementation of the complex
multi-dimensional hydrodynamic models only as needed and only for
relatively short periods of real-time simulations of the necessary
general class of flow conditions. The results of the hydrodynaraic"
simulations can be used as the flow input data to the transient,
multi-dimensional discrete-element thermal transport models for the
necessary long-term (days) simulations of the temperature conditions in
controlled-flow reservoirs.
The systematic zone-matching methodology was applied to the
assesseraent of the hydrodynamic and thermal impact of the operation of a
typical power plant (Peach Bottom Atomic Power Station) [10] on a
typical controlled-flow reservoir (Conowingo Reservoir) [10] where the
flow conditions in the particular section of the Susquehana River are
controlled by the upstream and downstream hydro-electric power plants
(at the Holtwood and Conowingo dams, respectively) and by the
intermitantly operating purnp-storage facility (Muddy Run) .
The transient, one-dimensional, discrete-element hydrodynamic and
thermal transport model (ESTONE) [2, 3] was applied to simulate 2? days
of continuous flow and temperature conditions in the reservoir during
the period 25 June 1974 - 21 July 1974. The results of the computer
simulations and the field-measured data indicated excellent agreement
during the period of the simulations (see Fig. 1).
The two-dimensional simulations of the quasi-steady, plant-induced
flow conditions were obtained by the applications of the fast-transinet,
two-dimensional hydrodynamic model (SLOFLO) [4, 5] as shown in Fig. 2.
The dam-controlled and pump-storage-controlled natural flow conditions
in the reservoir were simulated for relatively short periods (24 hrs) by
the application of the fast-transient, two-dimensional, multi-layer
discrete-element hydrodynamic model (HYDROS) [6, ?]• The results of the
two- and three-dimensional hydrodynamic simulations for the short
periods (24 hrs) were extended, by the applications of the zone-matching
methodology and the results of the one-dimensional hydrodynamic
simulations to construct the necessary approximate flow fields for the
long-term (10 days) termperature simulations with the fast-transient,
two-dimensional (including vertical variations) discrete-element thermal
transport model (TEMPER) [8, 9] as shown in Fig. 3. The results of the-
computer predictions for the two-dimensional multi-layer distributions
of temperature conditions (depth-averaged, surface and bottom) agreed
very well with the available field-measured data (within the limitations
of the accuracy of the measurements) throughout the 10-day period of the
simulations (see Fig. 4).
The results of the study concluded that the application of the
zone-matching methodology is significantly more efficient in computer
time and cost than the applications of transient, multi-dimensional
coupled hydrodynamic and thermal transport models for repetative
long-term simulations of the hydrodynamic and temperature conditions
that are usually needed for the assessment of the Ujt.rmal impact of
power plant operations on controlled flow reservoirs.
180
-------
REFERENCES
1. A.H. Eraslan, W.L. Lin, and K.H. Kira\ "A Systematio-
Near-Field/Far-Field Zone-Matching Methodology Based on Uniformly-Valid
Singular Perturbation Theory for Generating Complete, Two-Dimensional
Natural and Plant-Induced Unstratified Flow Conditions in Lakes,
Estuaries, and Coastal Regions" ORNL/NUREG-41, (in press).
2. A.H. Eraslan and K.H. Kim, "A Fast-Transient, One-Dimensional,
Discrete-Element Transport Model for Predicting Hydrodynamic, Thermal
and Salinity Conditions in Controlled Rivers and Tidal Estuaries for the
Assessment- of the Impact of Multiple Power Plant Operations"
ORNL/NUREG-21, (in press).
3. A.H. Eraslan, K.H. Kim, S.K. Fischer, R.D. Sharp, and M.R.
Patterson, "ESTONE: A Computer Code for Simulating Fast-Transient,
One-Dimensional Hydrodynamic, Thermal and Salinity Conditions in
Controlled Rivers and Tidal Estuaries for the Assessment of the Impact
of Multiple Power Plant Operations" ORNL/NUREG-22, (in press).
l|. A.H. Eraslan and K.H. Kim, "A Transient, Two-Dimensional
Approximate Hydrodynamic Model for Simulating Natural and Plant-Induced
Flow Conditions in Lakes, Estuaries, and Coastal Regions" ORNL/NUREG-M2,
(in press) .
5. A.H. Eraslan, K.H. Kim and A.K. Atakan, "SLOFLO: A Computer Code
for Approximate Simulations of Transient, Two-Dimensional Natural and
Plant-Induced Flow Conditions in Lakes, Estuaries, and Coastal Regions"
ChNL/KuREG-^3, (in press).
6. A.H. Eraslan, W.L. Lin and K.H. Kim, "A Fast-Transient,
Two-Dimensional, Multi-Layer Discrete-Element Hydrodynamic Model for
Predicting Flow Conditions in Estuaries, Lakes, and Coastal Regions"
USDI Report (in press).
7. A.H. Eraslan, W.L. Lin and K.H. Kim, "HYDROS: A Computer Code
for Simulating Fast-Transient, Two-Diraensional, Multi-Layer Flow
Conditions in Estuaries, Lakes, and Coastal Regions" USDI Report (in
press).
8. A.H. Eraslan and K.H. Kim, "A Fast-Transient, Two-Dimension^,
Discrete-Element Thermal Transport Model for Predicting Temperature
Distributions in Lakes, Estuaries and Costal Regions for the Assessment
181
-------
of the Thermal Impact cT Dower Plant Operations" ORNL/NUREG-23, (in
press).
9. A.H. Eraslan, K.H. Kim, S.K. Fischer, R.D. Sharp and M.R.
Patterson, "TEMPER: A Computer Code for Simulating Fast-Transient
Two-Dimensional Temperature Distributions in Lakes, Estuaries and
Coastal Regions for the Assessment of the Thermal Impact of Power Plant
Operations" ORNL/NUREG-2M, (in press).
10. A.H. Eraslan and K.H. Kim, "A Comprehensive Mathematical Modeling
and Computer Simulation Study of the Hydrodynamic and Thermal Conditions
in the Conowingo Reservoir and Comparisons of the Results with
Field-Measured Data in the Vicinity of the Peach Bottom Atomic Power
Station" ORNL/NUREG-TM-32, (in press).
182
-------
OO
90 \-
cb
SflS
25 JUNE 1374 - ST1 JULY 1974
ESTONE SJMULBTiCr',
FIELD DPTfl
-- CRILY PYERRDE
DRILY nnnmjri
DfllLY HRX1MUM
i t I i i i
0 1 2 3 4 5 6 7 fi fl 10 II 12 12 14 15 16 17 18 19 20 21 22 23 24 25 26 27
TIME (DRY6)
Fig. 1. Comparison of Computer Simulation Results and Field-Measured Data for Temperature
Conditions at the State Line in the Conowingo Reservoir during 25 June 197A - 21 July 197A.
-------
00
1 .0
A
X
I
S
1 0
0.5
0.0L
0.3
0
1.00 FT/SEC
20
30
35
t f
17
0.8
X AXIS * 104
1 .3
1 .6
Fig. 2. Two-dimensional computer simulation results of the plant-induced flow conditions in the
Conowingo Reservoir in the vicinity of Peach Bottom Atomic Power Station (exploded view of the sub-
region in the vicinity of the power plant).
-------
00
tn
1.00 FT/SEC
0 20
61 GT 8*1
8JLJS4—"^
0.0
0.5
1.5 2.0 2.5
X AXIS * 104
3.0
3.5
Fig. 3. Two-dimensional computer simulations (with self-similar vertical variations) of the
velocity and temperature conditions in the Conowingo Reservoir in the vicinity of Peach Bottom Atomic
Power Station at 11:00 hr on 15 July 1974. ' '
-------
93
li-
es
V
-a
vx
LJ
a:
UJ
cu
x:
LU
85
83
75
TEMPER SIMULATION
SURFACE
DEPTH-AVERAGED
BOTTOM •
FIELD-MEASURED DATA
O SURFACE
A DEPTH 5 ft
O DEPTH 10 ft
0.0
2.0 4.0 6.0 8.0 10.0
DISTANCE ALONG NORMAL TO SHORELINE Cthousand ft}
12.0
Fig. A. Comparison of Computer Simulation Results and Field-Measured Data for Temperature
Conditions along a Line Norma] to the Shoreline at 1200 ft Downstream of the Discharge of Peach
Bottom Atomic Power Station in the Conowingo Reservoir at 12:00 hr on 18 July 1974.
-------
HEAT LOAD IMPACTS ON DISSOLVED OXYGEN:
A CASE STUDY IN STREAM MODELING
A.K. Deb, Ph.D., P.E. and D.F. Lakatos
Weston Environmental Consultants-Designers
West Chester, Pennsylvania, U.S.A.
ABSTRACT
The impacts of heat loads from a power generating station on the receiving
waters are two fold: 1) raise the temperature, and 2) depress the dis-
solved oxygen of the receiving water. Both parameters are sensitive to
the biological life of the receiving waters.
The objective of this study is to find the impact of heat loads from Pic-
way Generating Station of Columbus and Southern Ohio Electric Company on
the dissolved oxygen and temperature of the Scioto River. The study was
primarily comprised of two parts: l) temperature prediction using the
one-dimensional COLHEAT computer model; and 2) dissolved oxygen analysis
of temperature effects using the QUAL II model. A linear relationship of
changes in AT was obtained.
INTRODUCTION
The Picway Generating Station, located on the east bank of the Scioto River
approximately 10 miles south of Columbus, Ohio, has three coal-fired gen-
erating units, all of which use once-through cooling. Two of the units,
numbers 3 and k, rated at 30 MWe each, are scheduled to be removed from
service in 1980. Unit 5 is a cycling unit rated at 100 MWe. The Scioto
River is the source of cooling water for unit 5. The maximum withdrawal of
water is 156 cfs.
The Scioto River is grossly polluted for a distance of ^0 to 50 miles by
sewage treatment plant effluents entering the river upstream of the Picway
Station. This study was conducted for a 3l6(a) demonstration analysis for
unit 5 and included existing and predicted temperature and dissolved oxygen
(DO) distributions under low-flow conditions in the Scioto River.
The thermal effluent from the Picway Generating Station raises water temp-
erature in the Scioto River, and there is considerable concern about the
impact of this heat load on biological life and dissolved oxygen levels in
the river, particularly in view of the relatively poor overall quality of
the river water. Two large municipal wastewater treatment plants, the
Jackson Pike STP and the Southerly STP, which are upstream of the Picway
Station (see Figure 1), will decrease their monthly average effluent BOD
concentration to 8 mg/L in 1981 or 1982. The Waste Load Allocation Report
187
-------
(WLAR) for the Scioto River includes a statement to the effect that the re-
duction of these loads will significantly improve the water quality of a
^0- to 50-mile stretch of river below Columbus. The Picway Station is
located approximately 10 miles downstream of the Jackson Pike STP and l.*j
miles downstream of the Southerly STP. It is of concern that the projected
improvement in dissolved oxygen levels in this stretch of the Scioto River
resulting from the reduction in BOD load may be compromised by the heat
load from the Picway Station.
This study was conducted to identify and evaluate the impacts of higher
river temperatures on the biological life and dissolved oxygen levels in
the stream below the Picway Station. Since conditions of low river flow
and high heat load are infrequent and thus cannot readily be measured in
the field, use of predictive mathematical models has been recommended.
This study was conducted in two phases: 1) temperature prediction in the
downstream side of ,the Scioto River; and 2) dissolved oxygen prediction in
the Scioto River in the upstream and downstream sides of the power plant.
It was intended that the results of the first phase modeling study, i.e.,
the predicted temperatures under low river flow and high-heat load condi-
tions, will be used in the second phase to predict the depression of dis-
solved oxygen as a result of these heat loads.
THERMAL MODELING
The model to be used for the thermal modeling study was to effectively sim-
ulate the stream characteristics of the Scioto River as well as respond to
various waste heat loadings, river flows, and atmospheric conditions.
During low-flow conditions, the Picway Station discharges a heat load which
occupies the entire width of the river and mixes completely (top to bottom
and side to side) with the river water. Under these conditions, only one-
dimensional models (which assume complete mixing) are appropriate.
The COLHEAT model, developed by the Hanford Engineering Development Labora-
tory, was used in this study to predict the mixed water temperature of the
Scioto River downstream of the power plant.
COLHEAT Model
The COLHEAT River Simulation model is a far-field, one-dimensional model
designed to simulate temperatures in the portions of the river in which
complete mixing can be assumed.
The river is divided into various segments of fixed volume, through which
water is transported. The water temperature is modified by the local at-
mospheric environment, which is introduced into the model by means of a
heat budget. The advected heat, if any, in a particular segment is intro-
duced and mixed completely with the river water of that segment. The
188
-------
temperature rise of the mixed water of the segment caused by the introduc-
tion of advected heat and atmospheric heat exchange is calculated and as-
sumed to be uniform throughout the whole segment. The model's routine
provides temperatures at an upstream location and simulates the subsequent.
temperature changes at one or more downstream locations.
To evaluate the far-field temperature distribution due to thermal dis-
charges into a stream, the following information is required as input to
COLHEAT:
River dimensions, reduced to equivalent trapezoidal cross-sections.
Daily average water temperatures at the upstream end of the reach
under study.
Da ily r iver flows.
Meteorological data, including daily average values of wind velocity,
air temperature, dew point, sky cover, and incoming short-wave
radiat ion.
Daily advected heat discharge information.
Tributary flows and temperatures.
COLHEAT uses a heat budget to simulate the effect of the local atmospheric
environment on the rate of heat exchange between the water body and the
atmosphere. The process used in COLHEAT for heat exchange between air and
water can be written as:
Net heat added = (net insolation) - (back radiation) - (evaporation)
+_ (conduction) + advection
Convergence of the Model
COLHEAT's transport mechanism uses the relative magnitude of the flow out
of a river segment to determine the average water temperature. If the time
step is not selected properly, the water from one river segment can travel
through several segments in one time step. In selecting the time step and
the length of each segment of the river, it is necessary to insure that the
travel time of the water through each segment is greater than the time step
used in the computation. This can be expressed as
. - Min Vol (1)
At >r: Ti
Max F1ow
Where:
At = Time step used in the model
Min Vol = Volume of smallest river section
Max Flow = Maximum of flow rate during period of simulation
189
-------
If the convergence criterion set up in equation (1) is not satisfied,
the numerical solution will not converge to the correct solution and could
lead to instability in the computed temperatures.
Wind -Speed Function
In evaluating the surface heat exchange caused by evaporation and conduc-
tion, the COLHEAT model utilized the Lake Hefner wind-speed function. In
their summary of the results of various studies on wind-speed function (as
shown in Figure 2), Edinger, Brady, and Geyer [1, 2] indicated that the
Lake Hefner wind-speed function equation underpredi cts the wind-speed func-
tion by about 50 percent at low speeds and by about 30 percent at higher
wind speeds. These underpredict ions lead to an underpredi ct ion of heat
transfer by evaporation, which in turn leads to an overest imat ion of summer
temperatures and underestimation of winter temperatures. Figure 2 shows
the considerable discrepancies among the various wind-speed function
formulat ions.
Brady, Graves, and Geyer [3] in their study pointed out that if a water
surface is warmer than its overlying atmosphere, "...the air adjacent to
the water surface would tend to become both warmer and more moist than that
above it, thereby becoming less dense. The resulting vertical convective
air currents. . .might be expected to achieve much higher rates of heat and
mass transfer from the water surface (even in the absence of wind) than
would be possible by molecular diffusion alone."
They developed the following wind-speed function:
f<»>- 73.1,3* 0.71, »2, in (_
w = Wind speed in mph
In the Scioto River downstream of the Picway discharge, the water surface
temperature under low-flow conditions in summer is high in relation to the
air temperature. The average wind speed during July and August is about
2.6 mph. Under such conditions, there will be a considerable amount of
evaporative loss as a result of convective air currents above the water
surface, which is neglected in the Lake Hefner wind-speed function. There-
fore, it would be more appropriate to use the wind-speed functions of Brady,
et al .
In the present study the COLHEAT model was modified by replacing the Lake
Hefner wind-speed function with that of Brady, et al., to calculate heat
loss by evaporation and conduction. A duplicate set of computer runs, one
with the Lake Hefner function and one with that of Brady, et al., shows
that the Brady, et al. , function results in lower absolute temperature and
lower temperature rises (AT) (Figure 3).
The COLHEAT model was run to simulate the temperature rise of the Scioto
River as a result of heat load under various low-flow and high-heat load
190
-------
conditions. The results were used to predict the dissolved oxygen depres-
sion of the Scioto River downstream of the power station as a result of
heat load.
DISSOLVED OXYGEN PREDICTION MODELING
The objective of the dissolved oxygen modeling study was to identify and
evaluate the impacts of higher river temperatures on the dissolved oxygen
levels in the stream below the Picway Station using mathematical modeling.
There are many models available for simulating dissolved oxygen in streams.
The model which was selected for this study was to effectively simulate the
dissolved oxygen characteristics of the Scioto River and be capable of re-
sponding to varying flows, wasteloads from sewage treatment plants, and
heat loads from the Picway Station.
The QUAL II model was used in this study primarily because it has been
applied and tested on many rivers throughout the United States and has per-
formed reliably and accurately. It is a simple, steady-state, stream-
quality-analysis computer model, and has been recognized by the U.S. Envi-
ronmental Protection Agency (EPA) as being capable of predicting water
quality (particularly dissolved oxygen) with a relatively high degree of
accuracy.
The QUAL II model basically integrates the advection-dispersion mass-
transport equation for all the water quality constituents to be modeled.
This equation includes the effects of advection, dispersion, individual
constituent changes, and sources and sinks.
The QUAL II model is very comprehensive and can simulate biochemical oxygen
demand (BOD) and dissolved oxygen in streams, as wel1 as the dynamic be-
havior of conservative minerals, nitrogen, and nine other parameters. The
program determines the interactions of the nutrient cycle, algae production,
benthic oxygen demand, and carbonaceous oxygen uptake, and their effects on
the behavior of dissolved oxygen.
APPLICATION OF QUAL II TO THE SCIOTO RIVER
Adapting the QUAL II model to function as a predicting mechanism for water
quality conditions in the Scioto River around the Picway Station required
considerable acquisition and evaluation of data, and calibration of the
model for a selected set of representative river conditions.
The scope of the data-acquisition effort in this study essentially was de-
fined by the input data requirements of QUAL II:
Identification and description of the stream reaches involved.
Temperature profile of the stream reaches.
191
-------
Headwater data.
Inflow data (freshwater and wastewater).
Coefficients and exponents for hydraulic equations relating
river flow to the depth and linear velocity.
Various reaction coefficients for in-stream reactions.
Two very important input data items that must be identified to accurately
apply QUAL I I for this type of analysis are:
1. Temperature Profile of Stream - This is used in calibrating the
model, and establishes the initial condition of the stream system
being modeled. This data was available from sampling stations on
the Scioto River, along with the results of the COLHEAT study pre-
v iously d i scussed.
2. Model Coefficients - The particular coefficients of interest for
the QUAL II model are the rate coefficients and the hydraulic co-
efficients and exponents. Data on hydraulic coefficients and ex-
ponents were derived from the C&SOE Thermal Discharge Study and
from a stream survey conducted specifically for this study. How-
ever, certain limitations of the available data indicated the
need for development of assumptions of conditions at various points
where the data were missing. These assumptions, based on detailed
information concerning flow-depth-velocity relationships at the
U.S. Geological Survey stream-gauging stations (the upper and
lower boundaries of the river stretch under study), were the basis
for interpolated values for the intermediate reaches.
Initial values for rate coefficients for biochemical oxygen demand (BOD) and
stream reaerati'on were based on generally-accepted values in the technical
literature. During the calibration of the model, these coefficients were
adjusted on successive runs to provide a better match of the simulated and
measured data.
CALIBRATION OF THE MODEL
For calibration purposes, the QUAL II model was set up to simulate the 25-
mile stretch of the Scioto River on a particular day for which detailed flow
monitoring and water quality sampling and analyses were readily available.
Since the critical condition for evaluating the impact of the generating
station's heat load on dissolved oxygen is that of low river flow, the date
selected from those available was one on which the river flow was compara-
tively low. The heat load from the Picway Station was introduced as input
in the form of the temperature rise of the mixed water downstream, as
192
-------
derived from the COLHEAT model results. The proper boundary and initial
conditions were established by using the flow and quality data of the head-
waters, and the temperature profile of the Scioto River throughout the study
stretch.
During the calibration process, the rate coefficients for different reaches
of the stream were adjusted (within accepted limits) to match the model-
predicted dissolved oxygen values with the corresponding measured values.
Figure k, a graphical summary of the calibration results, shows that the
model-predicted dissolved oxygen values are generally somewhat lower than the
corresponding measured values, but the differences are small. The coeffi-
cients used in the model were all within the scientifically acceptable
limits, were internally consistent, and consequently it was felt that they
could be used with confidence for accurate prediction of dissolved oxygen
behavior.
MODEL APPLICATION
The stream characteristics (i.e., physical characteristics and model reac-
tion coefficients) developed for calibration were used for dissolved oxygen
simulations under several different stream-flow and heat-load conditions.
Low flows, such as those which typically occur during the summer months,
are critical to the dissolved oxygen regime. Therefore, four low-flow
conditions were analyzed, and two heat-load situations were simulated. The
first (with heat load added) was the historical seven-day peak load of the
three generating units at the Picway Station. The second (without heat
load added) represented the situation where the station was not in operation.
In developing the required input data for the modeling runs, values for
initial temperature conditions in each study reach had to be determined.
The flows selected for the model runs are those for which temperature pre-
dictions were generated by the COLHEAT model. The COLHEAT temperature
predictions were therefore used in the QUAL II modeling runs.
DISCUSSION OF MODELING RESULTS
The results of the QUAL II model runs for the Scioto River in the vicinity
of the Picway Station were presented in the form of plots of dissolved
oxygen versus stream length. Examples of these plots for both the low- and
high-flow runs are given in Figures 5 and 6. This type of plot was presented
for all flow and heat load situations that were analyzed.
In these figures, two curves are plotted downstream of the Picway Station:
the solid line shows dissolved oxygen when the Picway Station is operating,
while the broken line represents dissolved oxygen without the Picway heat
load. The difference in ordinate between these two curves represents the
effect of the Picway heat load on dissolved oxygen, as determined by QUAL II.
193
-------
The model-predicted dissolved oxygen plots show that under the lower flow
conditions the dissolved oxygen concentration of a considerable length of
the Scioto River below Columbus will be zero, and that the water quality
of this segment will not improve very much with the projected reduction
of wasteloads from the upstream sewage treatment plants. This lack of
improvement is attributed to the fact that the entire river flow at low
flow is essentially sewage effluent.
In order to find a relationship between the maximum temperature rise at
the Picway discharge and the maximum reduction in dissolved oxygen accruing
from the heat load, Figure 7 has been plotted from computer outputs of
various runs. This figure illustrates the primary result of the QUAL II
modeling study of the river, and shows a relationship indicating the antici-
pated change in the dissolved oxygen characteristics of the stream (at its
most critical location) for a wide range of heat waste loads that could be
introduced to that stream.
ACKNOWLEDGEMENTS
Much of the material in this paper was developed during the preparation of
a thermal discharge study for the Picway Generating Station of the Columbus
and Southern Ohio Electric Company. The help of Mr. Henry L. Schulte, Jr.
of Columbus and Southern Electric Company and Thomas Johnson of Roy F. Weston,
Inc. is gratefully acknowledged.
REFERENCES
Edinger, J.E., Brady, O.K. and Geyer, J.C. "Heat Exchange and Transport
in the Environment," Electric Power Research Institute Report No. 12,
2. Edinger, J.E., Brady, D.K. and Geyer, J.C. "Heat Exchange and Transport
in the Environment," Electric Power Research Institute Report No. ^9,
197*4.
3. Brady, O.K., Graves, W.L. and Geyer, J.C. "Surface Heat Exchange at
Power Plant Cooling Lakes," Cooling Water Discharge Project Report
No. 5, Edison Electric Institute, Publication No. 69-901, New York, 1969.
194
-------
COLUMBUS
* ; j
\ L.I""
-" -'JACKSON
PIKE ST P
1-270 (RM: 121
SHAOEVILLE (RM: 1170)
(RM:1160)
MAIN STREET
MOUND STREET
\^
GREEN LAWN
PICWAY GENERATING
STATION
COMMERCIAL POINT (RM: 112.5)
SOUTH BLOOMFIELD (RM 107.5)
RED BRIDGE (RM: 100.0)
MONITORING STATIONS
RM- 1000 RIVER MILES ABOVE CONFLUENCE
' ~': ' OF THE SClOTO AND OHIO RIVERS
Figure 1 Schematic Diagram of the Scioto River System from
Columbus, Ohio to Circleville, Ohio
195
-------
50i
r
4 6 8 10 13 14
WINDSPEED, W (m/s)
SOURCE: EDINGER, BRADY AND GEYER 11974)
Figure 2 Comparison of Empirical Formulae for the Wind Speed
Function, from Edinger, Brady and Geyer, 1974
eo r
5 0
40
30
20
1 0
LaKe Hefner Wind-Speed Function
Brady, et al . Wind-Speed Function
34567
Distance From Source Heat Load — Miles
10
Figure 3 COLHE AT Temperature Prediction at Low Flow (150 cf s)
with Unit 5 Operating at the Load Exceeded 10 Percent of
Time (80 MWe)
196
-------
— JO-
z
O 40-
>•
s
O
530
O
1
Q
:> o OBSERVED
CO*II>UTEO
RIVER MILES-
SCIOTO RIVER
Figure 4 Results of Calibration of QUAL II Using 22 June 1977
— PtCWAY STATION IN OPERATION
— FHCWAY STATION NOT IN OPfRATtON
RIVER MILES-
SCIOTO RIVER
Figure 5 "Model" Prediction of Dissolved Oxygen in the Scioto
River With a River Flow at the Picway Generating Station
of 150 cfs and Existing Upstream Sewage Treatment
Plant Loads
197
-------
7.0-,
~ t.o-
O 40-
s
o
PICWAV iTATW* W O*CH*TIO«
MC*AV STATION HOT M OPCflATIOM
\
\
\
\
\
\
\
\ J
RIVER MILES-
SCIOTO RIVER
Figure 6 "Model" Prediction of Dissolved Oxygen in the Scioto
River With a River Flow at the Picway Generating Station
of 534 cfs and Existing Upstream Sewage Treatment
Plant Loads
5 miles
AT OF RIVER WATER AT PICWAY STATION ( F)
Figure 7 Sensitivity of Changes in Dissolved Oxygen to Changes
in Temperature (A T) Downstream of the Picway
Generating Station
198
-------
A STOCHASTIC METHOD FOR PREDICTING THE
DISPERSION OF THERMAL EFFLUENTS IN THE ENVIRONMENT
Alan J. Witten
. Energy Division
Oak Ridge National Laboratory*
Oak Ridge, Tennessee 37830
uohn E. Molyneux
-versfty of Rochester
.nester, New York 14627
Article prepared for publication in the
Proceedings of the Waste Heat Management
and Utilization Conference
By acceptance of this article, the publisher or
recipient acknowledges the U. S. Government's
right to retain a nonexclusive, royalty-free
license in and to any copyright covering the
article.
*0perated by Union Carbide Corporation for the U. S. Department
of Energy under Contract No. W-7405-eng-26.
199
-------
A STOCHASTIC METHOD FOR PREDICTING THE DISPERSION OF
THERMAL EFFLUENTS IN THE ENVIRONMENT
Alan J. Witten
ENERGY DIVISION
and
John E. Molyneux
UNIVERSITY OF ROCHESTER
ABSTRACT
The equation governing the distribution of waste heat in a receiving
medium is the advective'diffusion equation. Historically, this equation
has been studied determkiistically employing either analytical methods
requiring unrealistic simplifications, or more realistic numerical
schemes v/hich are to costly for extended simulations. An attractive
alternative method to deterministic techniques is the stochastic approach.
Here the same advective diffusion equation is considered, however, one
or more of the coefficients is assumed to be random. The appeal of the
method is two-fold. First it provides a computationally efficient
scheme for realistic long-term predictions by taking advantage of the
reduced input data requirements, and second, it inherently has the
capability to deal with accidental releases where ambient conditions are
unknown.
This problem has been formulated in one-dimension with the advection
velocity being a random function of time. If it is assumed that the
time integral of this ra'ndom velocity is gaussian, then all the excess
temperature statistics can be expressed in terms of the mean flow and
the autocorrelation of the random fluctuations. This assumption is not
particularly restrictive, since it is imposed on the integral of the
random variable, and is'necessary to make the problem tractable. Using
this method, general expressions for all the moments of the excess
temperature have been derived. In addition, a number of closed-form
special cases have been solved for the probability distribution.
Results to date have proved to be quite interesting. In particular
this work demonstrates the nonlinear dependence of the excess tempera-
ture on the advective f-low. Therefore, using a long-term averaged flow
does not produce the same result as solving the transient problem and
then averaging. In fact, if the rms velocity fluctuations are large, a
simple solution usinq the mean flow could underestimate the excess
temperature by a factor of two.
200
-------
I. THE SHORT-TERM RELEASE
In this section, we model the short-term release of a passive
contaminant into a river. This model has application to the problem of
an accidental release or other situations in which the river flow is
unknown. The. model assumptions are listed below:
1. The short-term release of a contam,.,ant is represented by an
instantaneous discharge.
2. The receiving river has a. constant cross-sectional area.
3. All quantities are averaged over river cross-section.
The mathematical formulation is now given by the initial value
problem
a(x = ± »; = 0, eft = 0; = c 6(x)
o
This equation can immediately be integrated
, _,, " -2--
c(x,t) =
however, this result is of little value since the river flow u(t) is,
by design, unknown. To extract information from equation (2) we let
u(t) = UQ + u'(t) where u is the zero-mean random fluctuations. In equation
(2), the river flow appears only as an integral over time. It, is therefore,
appropriate to introduce a new stochastic function
-i-
(3)
201
-------
and introduce the final model assumption by allowing the statistics
of 9 to be Gaussian. This assumption is not particularly restrictive
since it is imposed upon £ rather than u It can be argued, by
virtue of the Central Limit Theorem (Batchelor, 1953), that the
statistics of u' are still quite general. With this assumption,
equation (2) can be solv&d for all the one-point statistical moments
of the concentration.
The first statistical moment is the mean and this can be written
t)2/2(2Kt+
and second, using the time-averaged input data u does not produce
the same result as using the transient input data u(t) and then
averaging. Through this result, the non-linear dependence of the
concentration on the advecting flow is quite evident.
To present the higher order statistical moments, we introduce
a change of variables which renders the results more compact. The
new independent variables^ are
a = a
=
VQt)
202
-------
Here a is a relative magnitude of the concentration, b is the
ratio of the effective diffusivity to the actual diffusivity and
5 is a nondimensional distance relative to the position of the mean
advective front scaled by the diffusion length. Now the one-point
Nth order statistical moment is given by
Nb)
The most useful result for predictive applications is the probability
distribution. This distribution is related to the statistical moments
by the equation
oo
/ (xtt) =J cN fl(o;xit) do , (8)
o
where f\ is the probability distribution formally defined as Prob
{o <_o(x3t) < o + da}. Equation (8) can be inverted to obtain an
expression for fj. This is
fife; x,t) = c-\*b lnra/c;f «"
o < o < a (9)
This equation is the priricipal result of this section.
This result is shown graphically in Figures 1 and 2. Figure 1
is a series of plots of the produce af\ versus o/a for several values
of the nondimensional position 5. Moving away from the position of
the mean advection front £ = 0, the peak of the probability curve
moves towards lower concentration while the curve becomes broader
203
-------
and more symmetric. For increasing distances, the curve again becomes
skewed as the peak approaches zero concentration. Although it was
assumed that the stochastic function e is Gaussian it is clear from
this figure that the dependent variable a, the concentration, is not
Gaussian. This further demonstrates the nonlinear relationship
between the advecting flow and the concentration.
Figurp 2 shows plots of the most .probable concentration versus
position for two values of the randomness parameter b(t), a.long with
a plot of the mean concentration versus position. This figure demonstrates
the importance of higher order statistical moments by illustrating the
inadequacy of the mean in characterizing the concentration field. For
both values of b, the mean underpredicts the concentration for small
distances and overpredicts the concentration for large distances.
A more complete discussion of this initial value problem is given
in Molyneux and Witten (1-979).
II. THE CONTINUOUS RELEASE
In this section, we model the continuous release of a passive
contaminant into a river. This model can be applied to problems with
unknown river flow. In addition, it is particularly useful for
problems involving the.lo/ig-term Characterization of the contaminant
concentration resulting from a continuously discharging source. In
this case the time history of.the flow may be specified, however,
treated as .an. unknown for .computational,.efficiency., Rather than.
solving the transient problem deterministically, a steady-state
stochastic approach- caul.d ,provide th.e il.on.grterm concentration statistics
204
-------
at a minimal cost. The model assumptions here are similar to the short-
term release with the exception that the source is continuous with time
and a dalta function in sjpace.
The model equation is o
subject to the boundary and initial conditions
o(x = ±°°; = o(t = Q) = 0 (11)
Again we take the advecting flow to be the sum of a mean plus zero mean
random fluctuations given by
u(t) = U + u'(t) ' (12)
o
The solution to equation (10) subject to the specified conditions is
(13)
t.
o(x t)-S f /*<
c*,t) =_^y daj<
By defining a new random;function'
f
9 = J u'(t)dt ,
205
-------
the mean concentration can be written
t [x-UQ(t-t')]2
- —- I dt' [(< + < )(t - t')^ e ^<^e^^~^'} (14)
t
5 f
o I
f\ y~" I
0
where the effective diffusivity K can be .written
&
2(t
The effective diffusivity defined above is time dependent. If 9 is
assumed to be stationary, equation (15) asymtotically approaches
(16)
for large time, where T is the correlation time. This is the result
C?
obtained by Taylor (1954) for turbulent longitudinal dispersion in
pipes. Although the effective diffusivity becomes a constant for large
time, as given by equation (16), it is clear from equation (14) that
the mean concentration is a function of the time-history of the advective
diffusion process and, therefore, the transient behavior of K cannot
e
be ignored even in steady-state applications.
Higher order one-point statistical moments have been derived from
equation (13). The nth order moment in nondiaiensional form is
07)
206
-------
where the vectors, matrices, and functions in this equation are defined by
y = [ a; - nu x -T\2l..., x - i\nf - Distance (18)
D. . = 2n-6- . — Diffusion
^3 i> 1 3
f- _'t
Qij = P / dti / '
' T
- Random advection
U 2t
P = -2 — 2-- Non-random Peclet No.
_
- - Random Peclet No.
r <
-Autocorrelation function for u'(-\)
Equation (17) has minimal computational value since the numerical
determination of the nth order moment requires the inversion of an
n x n matrix and then an n-fold numerical integration. However, by
virtue of the fact that the nonrandom Peclet No. P is large for most
environmental applications, the integrand of equation (17) can be
asymptotically represented and n—1 integrals can be evaluated analytically,
without loss of accuracy. Now the determination of any one-point statistical
moment has been reduced to a single numerical integration and the formula
for the nth moment is given by
nP
4 (r\+nPJ>)
& *
2 ' TT J (r\^nP^)'
nP r
r
n + P
\ ,n—\ )
hr-2-;
207
-------
where
n
= /"(
•'rt
- o) R(o)do (20)
Equation (19) has been used to generate several plots of the steady-
state mean concentration as a function of position for four values of the
ratio of the random Peclet No. to the nonrandom Peclet No. Pr/P, cind
these are shown in Figure '3. The curve for Pr/P = 0 corresponds to the
deterministic nonrandom result. For larger values of P /P, the mean
concentration becomes greater. This feature demonstrates that transients
in the flow field cannot, in general, be ignored even in state-state
applications. It is important to reiterate the well known fact that the
state-state concentration is independent of any constant value of the
longitudinal dispersion coefficient. The only way that transient flows
can properly be modeled is. by solving the transient problem deterministicall
or by using a time-varying diffusivity.
208
-------
III. ADDITIONAL REMARKS
In this paper, two simple stochastic models have been presented.
Several of the simplifying assumptions made in the mathematical formulations
can be, to some extent,- relaxed. The point source assumption was purely
a mathematical rnnvenience\ The problems can now be solved without
additional difficulty for any Fourier transformable source distribution.
It is also possible to allow the river cross-section to1 vary with
downstream position and'allow the river flow to vary with position so
as to conserve mass. Ths can be accomplished by formulating the problem
using the volume of river between the source and some arbitrary downstream
location x as the spatial variable rather than simply the distance. This
will result in a Fourier transformable differential equation only if
diffusion is neglected. The neglect of diffusion is justified since the
effective diffusivity resulting from the random velocity fluctuations
dominates any small-scale .diffusion except at very short times. The
presence of diffusion at short times after the source is turned on is
only necessary with a point source. In this case, the diffusion is
required for initial smoothing of the singularity. However, this.
difficulty can be avoided by considering the more realistic case of
a distributed source. As was previously mentioned, such a source
can readily be included.
Finally, the models presented assumed the release of a passive
contaminant. This again was a mathematical convenience. It is possible
to introduce a simple sink or first order reaction rate. In this case,
a chance of variables will result in this term being transformed out of
the problem producing the identical differential equations considered
in the previous sections.
209
-------
REFERENCES
Batchelor, G. K., The Theory of Homogeneous Turbulence. Cambridge
University Press, Cambridge, England, 1953.
Molyneux, J. E. and A.*J. Witten, Diffusion of a Passive Scalar with
Random Advection, submitted to Water Resources Research, 1979.
Taylv/r, G. I., The Dispersion of Matter' in Turbulent Flow Through a
Pipe, Proc. Roy. Soc. A223. 446, 1954.
210
-------
3.0J.
0\ 1 1 » * »lf"T I
3.0 4-
(b) 5 = 0.5
2.0 4
i.o +
(3 .
1.0
Figure 1. Probability as a function of concentration for several positions
relative to the mean advection front.
-------
I
distance
Figure 2. Plot of mean concentration versus p'osition, and plots of most
probable concentration versus position for two values of the
randomness parameter.
212
-------
P /P = 1.0
P£/P =0.5
P /P = 0.1
•Vl'
= 0.0
0 0.2 0'.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Figure 3. Plots of mean concentration versus position for several values
of Pr/P.
213
-------
A TWO-DIMENSIONAL NUMERICAL MODEL FOR SHALLOW COOLING PONDS
S Chieh and A Verma
Envirosphere Company
A Division of Ebasco Services Inc
New York, New York USA
ABSTRACT
A two-dimensional time dependent finite-difference numerical model is
developed to describe the thermal performance of a shallow cooling pond.
The terms in the governing heat transport equation include convection,
dispersion and non-linearized form of surface heat loss to the atmosphere.
The water body is grided in a finite number of cells and heat balance is
satisfied exactly for each of these cells. The flow distribution in the
pond caused by the circulating water flow is determined by a linearized
formulation of the hydrodynamic equations of motion and continuity. This
approach is reasonable for a low velocity regime (away from the outlets)
normally encountered in typical cooling ponds associated with power plant
discharges.
The model is applied to a real cooling pond situation with variable depth
and irregular boundary. Based on the circulation pattern in the pond,
several zones of flow can be discerned. Model results indicate that in
relatively stagnant zones the role of lateral dispersion is predominant and
its neglect would result in unrealistic temperature gradients. To verify
this conclusion, the results of the present model were compared with those
of a Langrangian type model which considers only the convection and heat
loss terms in the heat transfer equation.
INTRODUCTION
A numerical model for predicting the transient temperature distribution in
an irregular shaped shallow cooling pond is proposed. The model yields
vertically averaged velocity and temperature distributions thus providing
a tool in evaluation of cooling pond performance. This model represents
an improvement over Yeh, Verma and Lai's [l] shallow cooling pond model
by incorporating the longitudinal and lateral dispersive heat transfer in
the equation of heat transport. Results of the proposed model are compared
with those of the non-dispersive model of Yeh, et al to highlight the role
of dispersion on cooling pond temperature distribution and intake water
temperature predictions.
HYDRODYNAMIC MODEL
As input to the convective-dispersion equation of heat transport, velocity
214
-------
distribution caused by the heated discharge in the cooling pond has to be
determined. Yeh, et al [1J have developed a mathematical model to describe
the steady state flow field in a shallow pond with irregular boundary and
variable depth. Three major assumptions, upon which their model is based
are: (1) convective acceleration terms in the equation of motion are small
in comparison to friction and pressure terms, (ii) bottom stresses are
expressible as linear functions of velocity, and (iii) pressure is hydro-
static. After vertically averaging the continuity equation and the
simplified equations of motion, they obtained:
-8
-g OVdy + -Jg - ^ = 0 (Ib)
d(uh) d(vh) = 0 (Ic)
dx ^y
where: P = density of fluid, K = friction factor, U = a scaling velocity,
g = acceleration due to gravity, >? = surface displacement, Txs and
Tys are surface stresses and x, y are spatial coordinates, u and v
are averaged velocity components in x, y directions and h = water
depth.
Introducing a stream function ^ (x,y) such that:
1
u = -
h (x,y) £y """ h (x,y)
The continuity equation (Ic) is automatically satisfied and equations (la)
and (Ib) can be combined to give:
& A z2* 2
_. + _a = _ __ __
Equation 3 is solved by over relaxation techniques subject to shoreline
values of ^, prescribed by the locations of intake and discharge points.
Once 4> (x,y) has been obtained, the flow field is determined by the use of
Equation 2.
TWO-DIMENSIONAL DISPERSIVE THERMAL MODEL
Assuming no internal sinks or heat exchange across solid boundaries the
vertically averaged thermal transport equation can be written as:
215
-------
(M) - (hul) +
where: D , D = thermal dispersion coefficients along the x and y axis,
m respectively
Q = rate of heat exchange per unit area between the water
surface and the atmosphere
P = specific weight
c = specific heat
A difference formulation, based on the"cell method", is proposed. The
procedure generates a set of algebraic equations which satisfy the
homogeneous, conservative and transportive properties for each cell of a
spatial grid network. In the context of this thermal transport model, the
homogeneous property assumes an average temperature condition within a
cell, the conservative property requires that the heat flux of any cell is
balanced exactly, and the transportive property ensures that a temperature
is advected in the direction of flow.
Figure 1 illustrates a typical network of equally sized rectangular cells.
During any time interval At, the heat balance for cell (i, j) can be
expressed as:
At • Hc + At • Hd + At ' Hg = Ace (5)
Net Heat Transport + New Heat Transport + Net Heat Transfer = Accumula-
3y Convection By Diffusion at the Water tion
Surface
Each of the heat terms, H , H., H and Ace are expanded below:
COS
Accumulation
Ace = PC
(Tlfj - Tlfj) Ax Ay hi,j (6)
where: T^,. = temperature at t + At
T.,. = temperature at t
"i,j = average water depth of the cell i, j
Ax = cell length
Ay = cell width
216
-------
Heat Transfer at the Water Surface
Hg = Q Ax Ay (7)
Q is the net heat flux and includes energy exchanges due to the various
types of incoming and c :ng radiation and losses due to evaporation
and convection.
Q can be written as:
Q = <1-7S) Hs + (l-Ta) H^ - Hbr - Hc - He
where Hg is the short wave r - - radiation, H the atmospheric long wave
radiation, H. r the back radi from wfffcerbody, H the sensible heat
flux due to convection and H i_i.e latent l.eat of evaporation. 7 and*y
are reflection coefficients for short and " ~ng wave radiations. For
expressions of these heat transfer terms, re^or to Edinger and Geyer[2]and
Ryan and Harleman [3].
Dispersion
H,
d ~~
pc i+i
+ Ay h^
+ Ax h.+^
+ Ax h.
DX Ti+1 . - T. .
Ax
D T. , . - T. .
x 1-1,1 1,1
AX
D T
y i, i+1 - Ti , i
Ay
D T. T. .
V 1,1-1 1,1
(8)
Ay
where h.., etc represent the average d^nth along appropriate faces of the
cell i,j and D , D are dispersion coefficients in x, y directions.
Convection (9)
Hc = Hcr + Hcl ' Hct + Hcb
where: H = heat flux by convection
r,l,t,b° = subscripts which indicate the right, left, top and bottom
faces, respectively, of an individual cell
From Equations (2), the flow rates through a cell face can be expressed as:
217
-------
qx = h (x,y) Ay us -
and q = Ai^_
y *
Hence, the following difference expressions can be written for terms on
the right hand side of Equation 9:
H . .x f •
£_ = *i,J- - M,J T^. i i,j-l
P •
H
cr - <*i+l,J - *i+l,J-l) T. if
, J-l) T^l.J if ^1+1, J >*i+l, J-l
"cb = (^i+l,j-l - ^i,j-l) T . if
T. . if
1 > J
Hct = (^i,j - ^i+l,j) T if
1 > J
if *i,J < ^1+1 ,J (10)
Model Equation
Substituting Equations 6-9, into Equation 5 yields the following difference
expression for the heat transport equation:
T T At r *
i,j = i»J + r^—^ Q Ax Ay + H,
pc AX Ay h- • L d
P ^-jJ
Thus, knowing the flow field (l^(i,j)), Equation 11 can be integrated in
time to find the temperature distribution in a cooling pond as a function
218
-------
of time for changing meteorological conditions,
MODEL APPLICATIONS
A proposed closed cycle (recirculating) cooling pond to receive 1585 cfs
.of cooling water with a condenser rise of 20.8°F (reject heat of 7.4 x
10* BTU/hr) is investigated. The pond has an average depth of 12.68 feet
and a surface area of 2500 acres and has been created by diking the
periphery of a natural terrain.
Figure 2 shows the shape of the pond and the predicted flow pattern in
terms of stream lines-^ = constant curves. The stream functions have been
normalized by the total flow rate. Thus 90% total flow is contained with
^ = 0.05 and t = 0.95 curves.
Although the thermal model is formulated and is capable of solving the
unsteady temperature distribution with varying meteorological conditions,
a steady state example is only presented for the present discussion.
Figure 3 shows the isothermal distribution resulting from the given flow
rate of 1585 cfs, condenser rise of 20.8°F, and the following assumed steady
state meteorological parameters:
H (solar radiation) = 2714 BTU/ft2/day
s
Ta (Air temperature) = 88.4°F
Dew point = 70.2°F
W (Wind speed) = 7.9 mph
r ficients were arbitrariiv chosen as: !)__= 1.0 rr
D. = 1.0 ft2/s.
Dispersion coefficients were arbitrarily chosen as: DX= 1.0 ft /s and
Superimposed on the same Figure 3, are the temperature distribution
predicted by the Lagrangian approach of Yeh, et al based on a non-
dispersive model (D = D =0). This comparison between the results of
the present and Yeh* et ^1 models illustrates the role of dispersion on
the temperature distribution for cooling ponds of irregular shapes.
The intake water temperature calculated by the dispersive (present) and
non-dispersive (Yeh, et al) models are 93°F and93.5T respectively, which
agree closely. However, the isotherms predicted by the two models within
the pond are quite different. An immediate implication of this result is
that the non-dispersive Lagrangian model is adequate to predict the intake
water temperature. However, a two-dimensional dispersive model should be
used to determine the temperature distribution more realistically.
Examination of the temperature distribution (Fig 3) reveals that:
(1) Near the discharge, the temperature predictions by the dispersive and
219
-------
non-dispersive models agree closely. This is because this region is
convection dominated and the effects of dispersion is not significant.
(2) In the stagnant zones of the pond (located near the left and bottom
right corner) the dispersion is dominant and the temperature predic-
tions differ by more than 2°F.
(3) In the vicinity of the intake, the isotherms predicted by the non-
dispersive model become almost parallel to the stream lines and the
lateral temperature gradients are very large. The dispersive model
on the other hand removes this anomaly.
CONCLUSIONS
A method for determining temperature distribution in shallow cooling ponds
by numerical modeling techniques has been presented. The "cell method"
was applied to solve the two-dimensional vertically-averaged diffusive
heat flux equation. Although the model in its present form has not been
verified, it has been applied to study the role of dispersion on thermal
distribution cooling ponds. Numerical experiments indicate that the non-
dispersive Lagrangian model is probably sufficient to predict condenser
intake temperature in irregular shaped ponds. However, significant
difference was found for the temperature distribution predicted by the
two-dimensional dispersive and non-dispersive Lagrangian models.
220
SC
-------
REFERENCES
1. Yeh, G T, Verma, A P and Lai, F H, "Unsteady Temperature Prediction
for Cooling Ponds," Water Resources Research, Vol 9, No. 6 (1973).
2. Edinger, J R and Geyer, J C, "Cooling Water Studies for Edison
Electric Institute, Project No. RP-49 - Heat Exchange in the Environ-
ment," The John Hopkins University, June 1, 1965.
3. Ryan, P J and Harleman, Dr F, "An Analytical and Experimental Study of
Transient Cooling Pond Behavior," Technical Report No. 161,
Ralph M Parsons Laboratory for Water Resources and Hydrodynamics,
Dept of Civil Eng, MIT, January 1973.
221
-------
Vii
.
TI-I.J
»
.
TiHi
»i,J-l
TI,M
*i+l,H
t
A,
1
Fig 1 - NETWORK OF RECTANGULAR CELLS
222
-------
INTAKE
FEET
Fig. 2 .- COOLING POND SHAPE AND STREAMLINE PATTERN
-------
INTAKE
FEET
2-D PRESENT DISPERSIVE MODEL
2-D YEH ETAL'S NON-DISPERSIVE MODEL
Fig 3 - TEMPERATURE DISTRIBUTION IN THE COOLING POND
-------
WASTE HEAT FOR ROOT-ZONE HEATING - A PHYSICAL STUDY
OF HEAT AND MOISTURE TRANSFER
D. L. Elwell, W. L. Roller and A. Ahmed
Department of Agricultural Engineering
Ohio Agricultural Research and Development Center
Wooster, Ohio 44691
ABSTRACT
A study of the use of power plant waste heat for root-zone heating in green-
houses is being undertaken. As the first part of this study, the physical
properties of a system that supplies heating water through buried pipes
and irrigation water through a constant level water table have been charac-
terized. Three different soils (sand, peat-vermiculite and silt-loam)
have been investigated at heating water temperatures of 25, 30, 35 and 40°C.
The resultant heat and moisture transfer figures are internally consistent
and in good agreement with comparable results of other investigators, where
available. The water table irrigation system is capable of maintaining
suitable soil moisture content for plant growth up to the highest tempera-
tures investigated, and the soil heating system can supply up to 30% of
the total, seasonal heating load of a greenhouse while producing root-zone
temperatures up to 32°C (89°F). These results will now serve as a base
for continuing study with lettuce growing on the system.
INTRODUCTION
The fact that rejected, or waste, heat from electric power generating plants
is basically a low temperature thermal resource leads to both physical
and economic restrictions on the practical utilization of this large and
potentially valuable resource. One possible application that shows consi-
derable promise for overcoming these difficulties is greenhouse heating.
Space heating requirements can be met from a low temperature resource.
If, in addition, at least some of the heat is applied through the soil of
the greenhouse, thus raising the temperature of the root-growth-zone, then
the economic benefits can be twofold. First, overall heat load require-
ments can be partially met and second, plant growth and yield can be en-
hanced if other soil criteria can be maintained. Because of these poten-
tial advantages both total greenhouse heating [l] - [3], and soil heating
applications of waste heat [4] - [?] are being studied by various inves-
tigators at this time.
The present study is directed toward the evaluation of heat and moisture
transfer in heated soils of various types (as described below) that are
irrigated from below, through the maintenance of a water table at an appro-
priate depth. It is based on the previous work of Shapiro [5], It is part
of a three-year study, supported by EPRI (Contract No. RP1110-1), and is
designed to provide detailed information about the effects of soil hea-
ting and sub-surface irrigation on soil conditions and about plant growth
under these conditions.
Authorized for publication as document No. 149-78 of The Ohio Agricultural
Research and Development Center.
225
-------
EXPERIMENTAL
One bay (5.5 x 29m (18.5 x 96 ft.)) of a two bay, double plastic greenhouse
was separated into an independent heating chamber by dividing the green-
house with a sheet of transparent plastic. Independent gas consumption
records were maintained for the heating units in each bay. This bay was
equipped with four small soil plots designed for the proposed heating and
irrigation studies of the experiment. Three of these plots were equipped
with sub-soil heating and sub-irrigation systems while the fourth plot was
unheated and surface-irrigated to serve as a control for comparison pur-
poses. The three heated plots contained "soils" that consisted of Wooster
silt-loam, a peat-vermiculite mixture, and sand, respectively and the un-
heated, control plot contained Wooster silt-loam. Each of these plots was
2.9 m (9.5 ft.) wide by 3.0 m (10 ft.) long and was sealed at a depth of
0.6 m (24 in.) by means of a plastic liner as shown in Fig. 1. The bottom
0.1 m (4 in.) of each of these plots was filled with pea gravel topped with
coarse sand, to allow for rapid, horizontal, water table equalization. A
constant water table was maintained at a depth of 0.5 m (20 in.) by an auto-
matic irrigation system.
Constant-temperature heating water, heated to 25, 30, 35 and 40°C for suc-
cessive intervals, was supplied to the heated plots by means of a mixing
valve controlled by a Honeywell air pressure regulator. This heated water
was circulated through each plot at a nominal rate of 0.16 liters/sec
(2.5 gal/min) by means of 0.1 m (4 in.) i.d. headers that supplied 0.025 m
(1.0 in.) i.d. ABS plastic pipes as shown in Fig. 2. These pipes passed
through the .soil at a depth of 0.3 m (12 in.) and with a separation of 0.3 m
(12 in.). The plots were insulated on all four sides with 0.05 m (2 in.)
of styrofoam insulation in order to minimize edge effects in heat flow in
the soil.
Temperatures throughout the soil of the heated plots were recorded from
thermocouples at the locations indicated in Figs. 1 and 2. A similar, but
less-extensive, set of thermocouples existed in the control plot. A Kaye
Instruments System 8000 recorder was used to read selected thermocouples
every 30 minutes continuously. All thermocouples were read at least once
a day, and all readings of both types were fed to a teletype paper-tape-
punch unit for transfer to a computer for analysis.
Ten-junction thermopiles were connected across the inlet and outlet sides
of the heating water circulation system for each plot, and regular read-
ings from these thermopiles were combined with heating water flow infor-
mation to provide total rates of heat input at each heating level. Radio-
maters, both inside and outside the greenhouse, were used to record light
levels at various times throughout the day. And moisture distribution in-
formation was obtained by taking core samples at various locations in the
plots.
RESULTS
The soils used in the plots had the following properties. The Wooster
-------
silt-loam soil was composed of 25% sand (primarily very fine sand - 0.1 to
0.05 mm), 60% silt and 15% clay, had a dry density of 1.32 g/cm3, and had
a ptfrosity (based on saturated water content) of 0.44. The sand "soil"
was composed of 99+% quartz with approximately 55% of the particles being
in the medium sand range (0.5 to 0.25 mm) and the remaining 45% being fine
>attd (0.25 to 0.1 mm), had a dry density of 1.63 g/cm3, and had a porosity
of 0.38. And the peat-vermiculite "soil" was composed of roughly 50% (by
volume) commercially available sphagnum peat moss (no further analysis was
attempted) and 50% vermiculite particles, had a dry density of 0.16 g /cm3,
and had a porosity of 0.81.
Moisture
the system is designed so that all moisture supplied to the water table
in each plot must, at equilibrium, be balanced by an equivalent amount of
evaporation from its surface. Irrigation water supply rate is shown in
Pig. 3. It should be noted that 1.0 liters/day (0.26 gal/day) is equiva-
lent to 1.13 x 10~4 m/day (0-031 in/wk) of water level change so that even
at the highest consumption rates shown the irrigation demand is modest.
The values for the silt loam soil are in excellent agreement with the pre-
vious work of Sepaskhah et al. [8] but the present values for moisture
consumption in sand are approximately fifteen times smaller than is re-
ported there. This difference may be readily attributed to the differen-
ces in the moisture application techniques involved which result in great
loss of water to the subsoil in the case of Sepaskhah, due to high hy-
draulic conductivity of sand.
Moisture content at various depths in each type of soil was determined by
taking core samples from two locations in each plot at regular intervals
throughout the experimental period. The resultant moisture profiles are
shown in Fig. 4. There was a distinct tendency for the upper layer of
the peat-vermiculite to lose moisture as solar radiation increased. This
loss amounted to a 50% decrease from 0.32 g/cm3 water content in January
to 0.16 g/cm in May. There was no detectable change in this period in
the upper layers of the other two soils.
effect of heating on moisture content in the region of the hot water
pipas was examined carefully. There was no detectable drying in the sand
at any of the temperatures used in this experiment . There was also no
drying in the peat-vermiculite through the 35°C heating level, but there
was limited drying - from 0.58 to 0.51 g/cm3 - in the region of the heat
Ing pipes when the heating water was held at 40°C. Similarly, there was
no drying in the Wooster silt-loam at 25 or 30°C, but there was some dry-
ing at the higher temperatures - from 0.39 to 0.35 g/cm3 and 35°C and
from 0.39 to 0.30 g/cm3 at 40°C. In both of these cases, the drying, was
no longer detectable within two weeks after heating had been terminated.
It should be noted that these drying effects seemed to apply equally to
core sar.ples that were taken midway between two heating pipes and those
that were taken as close to a heating pipe as possible.
The moisture consumption (Fig. 3) and moisture profile (Fig. 4) results
227
-------
for sane1 have been compared with the work of deVries [9J and of Jury
and Miller [10]. In the notation of the latter paper, total steady-state
moisture flow can be represented by
Qm - -Lmm <6>T> $L -LmT (9,T) dT _K(e,T) (D
dZ dZ
where Qm is total moisture flow (cm3/cm2 sec), K is hydraulic conductivity
(era/sec), 9 is volumetric moisture content (cm3 I^Q/cm3 soil), T is temper-
ature (°C), and 1^ and Lmf are transport coefficients. The most useful
comparison is in terms of these coefficients, and while"'the present experi-
ment was not designed for this particular calculation and did not inde-
pendently obtain hydraulic conductivity values, ihe general agreement is
quite good. The only significant difference lies in the fact that the
steep portion of the sand moisture profile occurs at somewhat lower mois-
ture contents (9 = .10 to .20) than would be expected.from Jury and Miller's
(10) data (9 = .14 to .24). No suitable comparisons were obtained for
either of the other two soils. A crude calculation, based on a value of
K = 1.7 x 10~5 ,cm/sec obtained from Van Wijk and deVries [ll], yielded for
the silt-loam at 0 = 0.35; L^ = 6.1 x 10~3 cm2/sec, and L^ = 4.5 x 10~6
cm2/sec °C. These values are roughly a factor of 40 less than corres-
ponding values for sand and should be •- ed only r ".rude estimates.
Wet bulb and dry bulb air temperatures were recorded at various locations
in the greenhouse and no systematic effect of the irrigation system on
humidity was detected.
Heat
Fig. 5 gives the rate of heat absorption from the hot water pipes in each
of the soils. The values shown are averages of daily values that were
calculated from thermopile and flow rate records, and they have been cor-
rected for losses in f-^e headers. The daily values show r.m.s. deviations
from the average that ai_ less than 10%, and the calculated uncertainty
in these averages is + 5%.
The largest temperatu :»^ drops recorded acrosi the plots were 1.46, 1.10,
find 2.43°C for the sarc peat-vermiculite and silt-loam respectively.
These drops occurred during a two week interval when flow rates were al-
lowed to drop be •' desirable levels, and in general, temperature drops
were less than h, hese amounts. Thus, the heating pipes were approxi-
mately isothermal ' lateral temperature variations could be neglected.
Fig. 6 shows a porLxOn ">f one of the weekly computer plots of temperatures
recorded from the thermo^ uples. Days 99 through 102 are April 9 to Ap-
ril 12, 1978 at which time ~he heating water was at T = 40°C, and channels
20 and 22, and 30 and 3"? are records from the 0.1 m (4") and 0.3 m (12")
depths in the sand and p^at-vermiculite soils respectively. Fig. 7 shows
one of the temper; ture profile plots drawn by the computer from all of
228
-------
the thermocouple readings recorded. This particular one is for the sand
plot at 8:38 AM on April 9, 1978. Both of these figures represent very
typical data.
The temperature profiles, with compensation for diurnal variations obtained
from the weekly temperature plots, were analyzed for heat flow using ther-
mal conductivities given by deVries [12] and Nakshabandi and Kohnke [13].
The values thus calculated were then compared with the heat absorption
values given in Fig. 5. For the sand (using X = 4.1 and 6.1 x 10" 3 cal/cm
sec °C at average values of 9 = 0.09 and 0.28 respectively, and using
AA/°C = 2.3 x 10~5 cal/cm sec °C/°C) the agreement was within + 3% in all
cases, and it was further determined that 57 + 3% of the heat coming from
the pipes was carried upward to the greenhouse air while the remaining 43%
was carried downward to the deep soil. For the silt-loam (using X = 2.9
and 3.2 x 10~3 cal/cm sec °C at average values of 6 = 0.35 and 0.43 re-
spectively) the agreement is equally good at 25°C, and the rate of increase
of thermal conductivity with temperature (which cannot be determined from
available references) was found to be 4.4 x 10"^ cal/cm sec °C/°C for an
agreement to within + 7% at the higher temperatures. In this case 62 + 3%
of the heat was carried upward. Finally, for the peat vermiculite the
closest comparison available is for a peat soil given by deVries (12) .
Use of the thermal conductivities for this peat yields a total heat flow
that is 21% less than that obtained experimentally. Thus it must be con-
cluded that X = 0.95 and 1.27 x 10~3 cal/cm sec °C at 6 = 0.52 and 0.71
respectively in this peat-vermiculite sample. These values yield agree-
ment with the total heat flow data to within +_ 5% and there is no indi-
cation of a significant increase in thermal conductivity with temperature
rise. And, in this case 50 +_ 3% of the heat was carried upward.
The thermal conductivities obtained above were used to calculate penetra-
tion of diurnal temperature variation into the soil following the methods
of Wierenga et al. [14]. The calculated penetration (corrected for non-
uniformity in moisture content, thermal conductivity and specific heat)
at 0.3 m (12") was compared with actual penetrations into the sand and
silt-loam soils. In both cases the calculated values were roughly 8%
higher than the observed values, but there was an uncertainty of at least
+ 10% in these experimental values and therefore there is at least margin-
al agreement at this point. No useful comparison could be made in the case
of the peat-vermiculite due to excessive damping at the relatively low con-
ductivities involved.
The effect of the heating on the temperature of the soil in the three plots
was determined from the various computer drawn temperature profiles.
The smallest amount of warming occurred in the peat-vermiculite where the
average root-zone temperature increase varied from A = 5°C at a heating
temperature of 25°C to A =10°C at 40°C. The warming in the silt-loam
was in each case, from 2 to 2 1/2°C greater than these values, and the
warsing in the sand was a similar increment greater still. Thus, in gen-
eral terms, average root-zone temperature of the heated soils varied
from a minimum of 17°C (62°F) to a maximum of 32°C (89°F) which is an ideal
range for future plant growth studies.
229
-------
The maximum heating requirement for the greenhouse in which the experi-
ments were carried out was determined both from calculations and from
gas consumption records to be 215 watts/m2 (1650 BTU/ft2/day) . At this
high level the heat transfer from the soil heating pipes to the greenhouse
a5r represents a maximum, for the various heating levels, of 4.1 to 7.1%
ot the total heat load for the peat-vermiculite; of 8.7 to 17.7% for the
silt-loam; and of 9.3 to 15.3% for the sand. Thus, on a lower, season
average basis, a root-zone heating scheme of the same configuration as the
one studied and utilizing actual power plant cooling water should be able
to supply from 10 to 15% of total heat load when peat-vermiculite is used
as the growth medium, and from 20 to 30% when either sand or silt-loam
are used.
It should be noted that the three soil heating plots have a total area
that is only 16% of the area of the experimental bay of the greenhouse and
that, therefore, even at the highest heating levels only a small percen-
tage of the heat load was being met through soil heating. This percentage
was not directly detectable from gas consumption records.
DISCUSSION
From the results given above, the peat-vermiculite soil was a relatively
poor medium both in terms of its heat transfer properties and on the basis
of root-zone temperature increase. Both the sand and silt-loam soils were
better than twice as effective heat transfer agents, and while the sand
reached somewhat higher root-zone temperatures, both achieved the full
range of soil temperatures (up to 30°C (85°F)) that is desirable for plant
growth studies. In addition, in none of the cases, as was expected, were
the heat transfer rates sufficient to meet the full heating load. How-
ever, in the two more favorable media at least the heat transfer rates
were high enough to suggest that root-zone heating combined with nighttime
insulation techniques and reduced air temperatures could substantially re-
duce aerial heating requirements. Thus root-zone heating must presently
be evaluated primarily in terms of increased plant productivity but in the
future could also become a major factor in heat load considerations.
Both moisture transfer and general soil moisture content were lowest in
the sand. It remains for the next portion of the present study to indi-
cate whether or not these moisture levels will be adequate for plant
growth. The peat-vermiculite and the silt-loam soils both showed limited
drying at high temperatures. However, this drying was not sufficient to
alter the generally favorable moisture levels, and in these two soils at
least, the present irrigation system appears, from a physical standpoint,
to be a very suitable, one for maintaining excellent growing conditions.
Thus, the basic physical characteristics of the three diverse but repre-
sentative soils used have been characterized in considerable detail for
the irrigation and soil heating scheme that is under investigation and
for a range of temperatures appropriate to power plant waste heat appli-
cation. This information will now serve as a base for continuing studies
of the system with lettuce growing on it, and it will be determined both
230
-------
how the system parameters serve to modify the growth of the lettuce and
how this growth in turn affects the physical characteristics of the system.
REFERENCES:
1. Boyd, L.L., R.V. Stansfield, G.C. Ashley, J.S. Hietala, and T.R.D.
Toukinson. Greenhouse Heating With Warm Water from Electric Genera-
ting Plants. A Demonstration Project. Proceedings of an Internation-
al Symposium on Controlled Environment Agriculture: 169-183.
The University of Arizona, Tucson, Arizona, 1977.
2. Bond, B.J., E.R. Burns, R.S. Pile and C.E. Madeweli- The Use of Waste
Heat in Greenhouse Agriculture. Proceedings of an .International Sym-
posium on Controlled Environment Agriculture: 151-168. The University
of Arizona, Tucson, Arizona, 1977.
3. Rotz, C.A. and R.A. Aldrich. Feasibility of Greenhouse Heating in
Pennsylvania with Power Plant Waste Heat. American Society of Agri-
cultural Engineers. Paper No. 77-4530. Chicago, Illinois.
4. Boersma, L., L.R. Davis, G.M. Reistad, J.D. Ringle, and W.E. Schmisseur.
A Systems Analysis of the Economic Utilization of Warm Water Discharge
from Power Generating Stations - Final Report.
Bulletin No, 48, Oregon State University. 1974.
5. Shapiro, H.N, Simultaneous Heat and Mass Transfer in Porous Media
with Application to Soil Warming with Power Plant Waste Heat. Ph.D.
Thesis. The Ohio State University. 1975.
6. Sondern, J.A. Soil Warming in the Open. Research Report 77-5.
Institute of Agricultural Engineering, Wageningen, The Netherlands.
7. Wells, L.G., A.D. Ward, J.N. Walker, and J.W. Buxton. Heat Loss From
Heated Greenhouse Soil Beds. American Society of Agricultural Engi-
neers. Paper No. 77-4529. Chicago, Illinois. 1977.
8. Sepaskhah, A.R., L. Boersma, L.R. Davis, and D.L. Slegel. Experimen-
tal Analysis of a Subsurface Soil Warming and Irrigation System Util-
izing Waste Heat. American Society of Mechanical Engineers.
Paper No. 73-WA/HT-ll. Detroit, Michigan.
9. deVries, D.A. Simultaneous Transfer of Heat and Moisture in Porous
Media. Trans. Amer. Geophysical Union, 39:909-916. 1958.
10. Jury, W.A. and E.E. Miller. Measurement of the Transport Coefficients
for Coupled Flow of Heat and Moisture in a Medium Sand. Soil Sci.
Soc. Amer. Proc. Vol. 38:551-557, 1974.
11. Van Eijk, W.R. and D.A. deVries. The Atmosphere and the Soil, Pages
17-61 in: W.R. Van Wijk (ed.). Physics of Plant Environment.
North Holland Publishing Company, Amsterdam. 1963.
12. deVries, D.A. Thermal Properties of Soils. Pages 210-235 in: W.R.
van Wijk (ed.) Physics of Plant Environment. North Holland Publish-
ing Company, Amsterdam. 1963.
13. Nakshabandi, G.A. and H. Kohnke. Thermal Conductivity and Diffusivity
of Soils as Related to Moisture Tension and Other Physical Properties.
Agr. Meteorol. 2:271-279. 1965.
14. Wierenga, P.J., D.R. Nielsen, and R.M. Hagan. Thermal Properties of
a Soil Based Upon Field and Laboratory Measurements. Soil Sci. Soc.
Amer. Proc., Vol. 33:354-360. 1969.
231
-------
B
:i
t
1
;
• •-*-•" -
C
D
Ex, •
i
Fig. 1. Vertical section of an experimental soil warming plot showing
the depths at which the thermocouples (') are located. The hatch
marks represent styrofoam insulation. (A) is the supply header,
(B) is the return header, (C) is one of the ten heating pipes,
(D) is a stand pipe for controling the water level, (E) is two
inches of fine sand, (F) is three inches of pea gravel and
(G) is an impermeable membrane.
i
I
•
i
p
4
L
^ S
X !
Q >l
/•
/
«•
1
/
•*
/
-• /^V .
OD
** **?
•
^ •
H
J
3
XXXX\\X\\\ \\VX\X XX \
f
/
/
f
Fig. 2. Horizontal view of a plot showing two sets of thermocouples
on the pipes and two sets between the heating pipes. The labeling
is the same as in Fig. 1.
232
-------
Q 20
U
tu
H
oc
o
E
o:
l9
10
+ WOOSTER SILT-LOAM
K PEAT- VCRMICULITE MIXTURE
•SAND
UNHEATED 29 30 35 40
TEMPERATURE,°C
Fig. 3. Irrigation water supply rate at the different heating levels.
u
»*
•
K
O
V)
111
o
Ul
cc
3
V)
o
0.8
07
0.6
0.5
0.3
+ WOOSTER SILT-LOAM
x PEAT- VERMICULITE MIXTURE
•SAND
x
X
° 0.2
O.I
O.I 0.2 0.3 0.4
DEPTH , METERS
0.5
Fig. 4. Moisture Profiles. The circled points indicate the drying
effects that are discussed in the text.
233
-------
x • o to »o tr
Fig. 5. Rate of heat absorption in each of the soil plots.
1S7B 99 8 38
O Pi
8
S
LJ
V
DEPTH*. r/F.TF.RS
Fig. 7. Computer plotted temperature profile,
F'LOT A
X
x
i-o
234
-------
V
II
[H
2
*
1!
I
~
*
II
R
g
,*
?
o
2
LLJ
~y
2
5
1
j
*
,i*
" * * ^
*
*• * '
1
/
\
\
4
i
\
j
{
j
A
£
A*
A
i 4
v\
* -
A
1
;
1 "•
,
•
f
J
1
k
v *
..
\ *
*v "" *
*
J*
^h
j
A
A
&
=•
* ;
\
•"L **
j #
*
*
*
A
1
* ;
\
V
J
HI
i
§
St
:> i *T'
I
^«
*•
V
V
cs
f *<•
•s :*
j; *
v .*
,1
M
V
s
J
». J*
;.* » ^
i
|
i
5 /
. };*
;
V ^
I »
J
.5
i
— -_ — ^_
.,» "
s,
••f
t
/
^v
(,
*
.»
*
f
*%,
*
*
#-
L •
„.-•
" •» .
'.,B
x
\
i
,/
'
* m H
^
»
'-.,
""'•
>
/
^ '
|
}
»,
* » .
i
^
^
•f
f*
" "« ,
• »
I
/
»
M *
-|
J
I
-O
-
.10
- &
-co
_
-
-o
u-UJ
H
: s
—
-CD
-
-
-S
- 8
"
-tc
-
;
-a
- s
-CD
-
BE ' sk- " *€ " a oe 62 92 *a a
D 030
g. 6. Computer plot of various thermocouple readings versus time.
See text for discussion.
235
-------
BENEFICIAL USE OF REJECTED HEAT
IN MUNICIPAL WATER SUPPLIES
R. A. Wynn Jr. and R. W. Porter
Illinois Institute of Technology
Chicago, Illinois U.S.A.
ABSTRACT
The relatively low temperature of thermal discharges from
steam-electric power plants makes waste-heat utilization
difficult without modification of the power cycle and atten-
dant reduction in electrical-energy generating efficiency.
The present paper concerns in situ beneficial use of waste
heat by direct once-through condenser discharge into a
municipal water supply. Computations are presented regarding
the matching of flow rates, heat losses in distribution and
energy savings. A number of benefits and penalties are also
assessed qualitatively including legal and operational aspects
and reliability. Especially attractive are improvements in
generating efficiency, alleviation of water-pipe freeze up,
savings in water-heater energy, and improvement in wastewater
treatment effectiveness. Disadvantages include the need for
careful control of condenser water quality and the question
of public acceptance of water heated by about 13 C. Two
cases with operating experience are briefly discussed,
although the installations were not developed for energy
conservation purposes.
INTRODUCTION
It is well known in power engineering that for every unit of
electrical energy generated in a steam-electric power plant,
approximately two units of heat energy must be rejected to the
environment. Nearly 30% of the total U.S. energy consumption
is for generating electricity, and almost all of this in
thermal plants [1]. Consequently, about 20% of the total
U.S. energy consumption goes into waste heat of electric
generating stations. The possibility of recovering even a
small portion of this energy, perhaps the largest locally
concentrated source of waste heat, is attractive.
Unfortunately, the relatively low temperatures required in
conventional steam condensers preclude many utilization
schemes. Cooling water is typically heated by only about 14 C
in most plants. The present paper concerns the possibility
236
-------
of alleviating these problems, while still deriving benefits,
by discharging heated once-through cooling water from condensers
directly into a municipal supply. The principal direct
beneficial use of waste heat in this scheme is the savings in
water-heater energy input. Perhaps equally important is the
energy savings in using once-through cooling of power-plant
condensers rather than closed-cycle cooling towers or cooling
ponds which result in higher turbine back pressures, reduced
thermal efficiency and greater fuel consumption. By-product
benefits include alleviation of water-pipe freeze up in winter,
and slightly elevated wastewater temperatures, which will
reduce the cost of sewage treatment. Multiple use of water
resources and the elimination of evaporative consumption in
power-plant cooling are attractive in the larger view of
conservation. Also significant is the elimination of local
environmental effects of heat and humidity discharges to the
atmosphere near the plant site.
Potential Energy and Water Conservation
About 4% of the U.S. energy consumption is due to residential
and commercial water heating [2-4J. Overall efficiencies
are currently 50-62% [5]. In newer gas units, the efficiency
is about 72% less 1%/hour of storage while for electric it is
981 less 4%/hour [6.7]. The average per capita water
consumption is 568 L/day [8]. Based on 33% thermal efficiency,
1 kw(e) per capita electrical consumption, and a 14-C rise in
cooling water, the per capita condenser" flow is 2990 L/day
which is about 5 times the water supply rate. Thus only
.plants near filtration plants or water conduits need be
utilized to preheat much of the water-supply capacity.
Considering the increasing requirements of the Federal
Water Pollution Control Act [9], increases in electrical
demand and shortages of water [10, 11], multiple use of
water resources is attractive.
Legal and Operational Problems
There appears to be no legal temperature restriction on
discharges from power plants into municipal water supplies [12].
However, there are limits on Ph, residual chlorine, rust
inhibitors, etc. as governed by the Safe Drinking Water
Act [13]. Recent developments in anti-surface-tension agents
allow the necessary chlorine dosages to be reduced to
extremely low levels and still prevent biological fouling[14].
Chlorine is also added during water-supply treatment, and it
is necessary to maintain a residual concentration throughout
the distribution system for sanitary .purposes. However, in
the event of accidents involving contamination, provisions
should be made to alert the filtration plant in order to adjust
treatment or switch to an alternate supply. Indeed, the
presence of a full-capacity back-up cooling system and an
237
-------
alternate water intake for the municipality would be highly
desirable in order to avoid a combined shut down. According
to Oliker [15], a nuclear plant in Europe is currently
providing district heating, and more such plants are under
design and construction.
Typical power plants operate at 70-80% capability [16] whereas
municipal supplies are seldom out of service. The flow in
a municipal supply may peak on a daily basis at 150% of the
annual average and range on a hourly basis from 25-300% of the
daily average [4, 17]. While there are similarities in power
and water demand patterns, alternate stand-by systems would
be useful in matching requirements, in addition to providing
reliability.
Obtaining the necessary cooperation between the power and
water utilities would be facilitated in municipalities where
they are commonly owned and, ideally, commonly located.
If mechanical closed-cycle cooling is employed, such as
mechanical-draft towers and atmospheric sprays, operating
costs would be reduced for the private or public utility. In
any case, once-through cooling will increase thermal efficiency
and avoid de-ratings under extreme environmental conditions.
Therefore, there may be an economic advantage to the electric
utility to finance the venture. The advantage to the water-
supply system is essentially one of alleviating pipe freeze
up in winter in northern regions. No difficulty is anticipated
in handling water elevated by only 14 C or less, at least
in the northern U.S. The direct benefit to consumers is
reduced energy consumption for hot-water heating. Electric
utility rates could also be reduced, preferably in those areas
where the system is actually employed in order to further
offset the inconvenience of heated water in summer. Bottled-
water companies indicate that a family of four uses about 38
liters per month of water for drinking purposes which
could be refrigerated or simply left to stand in order to cool.
THERMAL EFFECTS
The present analysis is not all inclusive and many simplifying
approximations have been made. Further details are given in
Ref. 18.
Power Plant Efficiency
For the purpose of the evaluation, two modern double lOOO-Mw(e)
boiler-turbine combinations were assumed, one nuclear [19]
and one fossil [20] fueled, since there is considerable
difference in thermal characteristics. The steam condensers
238
-------
were analyzed using the NTU method and recommended design
criteria of the Keat Exchanger Institute 121]. The sizes
were fixed with a 13.9-C rise in cooling water temperature
at the design point and required 2.23 and 4.41 ML/min
(million liters/minute) of water coolant for fossil and
nuclear fuel, respectively. Alternative atmospheric-spray
systems, evaporative cooling towers and cooling ponds were
sized according to simple one-parameter models [22, 23, 24]
and meterological conditions [25,26] not exceeded more than 5%
of the time during summer in Chicago. All three systems behaved
similarly, and are represented here by a single hypothetical
cooling system, the tabulated heat-rate data and the steam-
condenser and cooling-sytem equations were solved iterativelv
[27] . . *
Results are summarized in Table 1. The major interest for
energy conservation is the advantage in heat rates of once-
through cooling over that of closed-cycle systems. The
savings range from 0.6-1.5% for fossil fuel and 1.5-2.5% for
nuclear with the most advantage in summer. The annual savings
is 1% for fossil fuel and 1.9% for nuclear, which is larger
due to its lower efficiency. However, fossil fuel is usually
far more expensive.
Transmission
If the plants are not co-located, transmission could be by
buried pipe or open channel. Heat transfer from buried pipe
is complex involving possible mass transfer due to thermal
and moisture gradients and capillary action, which are all
affected by the type of soil [28]. The heat transfer rate
per unit length from a buried pipe of diameter d, depth D
to the top of the pipe, temperature T, ground surface
temperature Tr and thermal conductivity K is ideally -S k (T-T }
where S is the shape factor and 2ir/S = arc cosh (2D/d + 1) [29].
The thermal conductivity for soil typically ranges from
0.35 to 2.6 w/m-C [30]. Here kff = 0.87 w/m-C is used.
Therefore, an energy balance may be performed on the
flowing water. The result is cooling of the mass flow rate
m from temperature T, to T2 over a length L. Thus
(T2 - TQ)/(T1 - TG) = exp(-S kG L/(m GW) ) (1)
where c is specific heat of water (4.19 kJ/kg -C).
\w
In order to estimate the importance of the loss between the
power plant and filtration plant, we assume average conditions
in the Chicago area [17] where there are two filtration plants.
239
-------
Each is assumed to receive half the total Chicaao flow with
TI = 26.7 C, TG = 12.8 C, ft = 1.94 x 1CT L/d, d'= 6.10 m,
D = 3.05 m,and therefore S = 3.57. For example, the resulting
temperature drop T. - T2 = 0.07 C over a length L = 160 km.
If open channel flow were used, the losses would be much
greater. If each channel were 24 m wide, and we assume a surface
exchange coefficient [24] of 35.5 w/m - C and a natural
equilibrium temperature of 12.8 C, the temperature drop over
only 16 km is 1.9 C. In the remaining analyses, we assume
that the plants are either connected with large buried pipes
or are co-located, such that no significant interconnection
loss is present.
Filtration_
Renn [31] has summarized the effects of increased temperature
on municipal supplies. The rates of organic decomposition
and activity of disinfectants are increased while pathogenic
organisms survive shorter periods. The cost of all chemicals
except alum used for coagulation is about $0.30/year on a
per capita basis, or $1,350,000 for the entire Greater Chicago
system [17]. Evaluation of possible savings is complicated
by the effect of pH which often simultaneously changes with
temperature. In the case of chlorine, the chemical dosage
needed for disinfectant at the plant decreases but dosage
to maintain a residual concentration throughout the system
increases. The production of odorous blue-green algae is
increased. The optimal temperatures of blue-green algae
are 16-27 C while Chicago water is currently about 2-21 C.
Applying a 14-C rise brings the Chicago range to 16-35 C, and
odors would be favored in winter rather than summer. The
effect of temperature on density and viscosity is such to favor
settling and permit more rapid filtration. A detailed
experimental study by Camp, Root and Bhoata [32] noted that
the alum coagulant requirement at optimal pH was. virtually
unchanged with temperature. If so, no advantage in alum
dosage can be claimed. Further, no advantage can be gained
for the overall size of facilities because of the need for
full capacity in the event of power-plant outages.
Distribution
Steady state operation is assumed, and the calculations are
performed for average summer, spring-fall and winter
conditions. It is assumed that the power plant elevates the
ambient water by either 13.9 (high AT) for perfectly matched
flows with maximum condenser heating or 8.3 C (low AT) which
jnight represent conditions where there is a 60% load factor
or a 40% dilution with ambient water. Ambient (T.) and heated
water (T. + AT) temperatures are listed in Table i. For the
present calculations, the flow rates and distribution config-
240
-------
uration of the City of Chicago were utilized [17]. The average
Chicago system_ flow rate is 3.876 x 10 L/d with a seasonal
adjustment, m/A(annual), shown in Table 2. Consideration of the
City of Chicago system as a balanced sexies-parallel network
resulted in the overall SL = 12,943 km. Temperature drops to
the end of the system were computed to be nearly uniform at 0.7
and 0.4 C for high and low AT, respectively (Table 2).
Computations were also performed on a quasi-unsteady basis
where steady conditions at 150% of the average steady flow
rate were first analyzed. Then, water was presumed to stand for
8 hours (nightime) and the heat transfer was computed.
Finally flow at 150% of the average rate was assumed for 16
hours. The system residence time is about 4 hours. The
reduced temperature change during the periods of increased
flow very nearly compensates for the transient heat transfer
at night under stagnation conditions. The net reduction in
elevated temperature ranged from only 0.3-0.8 C on the daily
average [18]. The neglected heat capacity of the ground
and pipes should reduce this effect even further.
Water Heater-Energy Savings
In the present section, we use one half the maximum distrib-cion-
system losses in order to simulate the average system condition.
First, we estimate the losses in the home from the intake
point to the water heater. It is assumed for the sake of the
present calculations that a single line carries the water to
be heated over substantially the entire distance involved.
For the purposes of estimating the above losses, we assume
a 7.6 m long, 2.5-cm outside-diameter line carrying the water
to be heated in the home. The standard correlation for
horizontal pipes was employed [29] in order to determine the
film coefficient for natural convection for the expected
laminar-flow regime. Because most cold-water lines are presently
insulated to avoid condensation and because reducing losses
is favorable using preheated water, the use of insulation
is assumed for both unheated and heated water.
The heat input to the water heater used directly for
elevating water temperature is essentially proportional to
the water temperature rise. This includes certain losses
but not those associated with maintaining water temperature
levels during storage. For the present calculations we use an
average water-heater temperature setting of 62.8 C as given
in the proposed U.S. water-heater test standards (Federal
Register, October 4, 1977). The percentage savings in heat
energy required to raise the water temperature (but not
maintain it) by using preheated water are shown in Table 2.
The annual average savings is 24% for the high AT and 14.5%
241
-------
for the low AT. However, as discussed previously, there is an
additional heat-energy input which is independent of preheating.
The ratio of heat input including certain losses used to
elevate water temperature (but not maintain it) to useful energy
rise which actually increases the energy of water is 1/n where,
according to the aforementioned Rheem Company data,nQ = 0.72
and 0.98 for gas and electric units, respectively. The overall
heat input ratio is related to the service efficiency which
includes a decay in efficiency due to losses during storage.
Thus for a residence time t, the ratio of total heat input to
useful energy rise is 1/n + 0.5t dn/dt on the average. Using
the referenced rate of chinge of efficiency_with storage
residence time t, dn/dt =-0.04 and -0.01 hr~ for new gas and
electric units, respectively. Assuming a standard 151-liter
tank and the flow rates of Table 2, the seasonal flow residence
times t are 12.8, 15.6 and 16.3 hrs. for summer, spring-fall
and winter, respectively. As a result, the ratio of heat-energy
input required to elevate water temperature to total heat-energy
input for service requirements, is found to be 0.64, 0.57 and
0.55 for gas in summer, spring-fall and winter, respectively,
and about 0.93 for electric. The difference between gas
and electric units is due to electric heaters being nearly
100% efficient in directly heating the water, whereas gas
units have stack and other losses. The resulting savings in
total heat input are given in Table 2. The annual average
savings are reduced to 14% and 22%, for gas and electric units,
respectively, with the high AT and 8%
-------
corresponding user discharge temperatures were found to be quite
uniform at 22.8, 22.8 and 21.6 C, respectively. Indeed, the
data seem to show that the water comes to essentially
equilibrium at the user environmental temperature. It is
not clear how much the water would have to be preheated
before being discharged at elevated temperatures. Based on
the temperature data of Table 2 for the average intake to
the user f and based on 22% of the flow being heated to
62.8 C in ?he water heater [4], the resultant mixed flow of
heated and unheated water apparently was cooled 7.2 C in
summer and was heated 1.8 and 6.6 C, respectively, in spring-
fall and winter. These increments can be approximated by a
straight line versus the average intake water temperature TO>
The computations were repeated using preheated water and
the same function,resulting in annual average increases in
wastewater temperatures of 0.8 and 0.2 C, respectively, for
high and low AT.
CASF. STUDIES
At Henderson, Kentucky [34], located on the Ohio River, local
conditions made construction of new water intakes very expensive
for a new filtration plant. Instead, the discharge from an
existing municipal power plant on the same site was used
successfully. Serving a 17,000 population, the filtration
plant employs walking beam flocculators and rake-type sludge
collectors. Chemicals added include hydrated lime and dry
aluminum sulphate for coagulation and Color and turbidity
removal. Chlorine for disinfection and taste and odor control,
activated carbon for taste and odor control, and sodium
silicofluoride for fluoridation are added at one or more points
in the system. At Edmonton, Alberta (Canada) a power-plant
discharge at an average of 18 C is blended with 0.6 C river
water in order to produce an intake averaging 13 C all winter
[35]. An average flow of 227 ML/da serves a population of
550,000. The use of heated water has minimized the occurence
of water-main breaks and virtually eliminated domestic freeze-
UDS.
ACKNOWLEDGEMENTS
This work was supported by U.S. Department of Energy, Office
of Energy Technology, under contract EC-77-S-02-4531.
REFERENCES
1. "Energy in Focus, Basic Data", Federal Energy Administration,
Washington, DC, May 1977.
2. Behrens, C. , "AKAK Offers Energy Savings to the Public", Appliance
Manufacturer, Vol. 22, No. 10, p. 80-82, October 191i.
3. Rimaero, D. , Utilization of Haste Heat iror Power Plants; Noyes
Data Corp-, ParX Ridge, NJ, 1974.
243
-------
4. Kammerer, J. C., "Water Quality Requirements for Public Supplies
and Other Uses", Handbook of Water Resources and Pollution Control,
Van Nostrand Reinhold Co., New York, 1976.
5. Babbitt, H., Plumbing, McGraw-Hill, New York, 1960.
6. Rheem Manufacturing Co., Private Communication to R. Wynn by
Mr. Haag, Engineer, Chicago, December 29, 1977.
7. Consdorf, A. P. and Behrens, C., "Appliance Energy Efficiency:
The Trials of a Test", Appliance Manufacturer, Vol. 25, No. 3,
40-59, March 1977.
8. Beall, S., "Uses of Waste Heat", ASME Paper 70-WA/Ener-10, 1970.
9. "Federal Water Pollution Control Act Amendments of 1972", United
States Code, Public Law 92-500, October 18, 1972.
10. Parker, F. L. and Krenkel, P. A., Physical and Engineering
Aspects of Thermal Pollution, CRC Press, Cleveland, 1970.
11. Norton, R. C., Westre, W. J. and Larsen, C. L. , "Dry Cooling
Design Characteristics of a Large Power Plant", Proceedings
of American Power Conference, Vol. 37, p. 591-597, 1975.
12. Lum, J., Private Communication to R. Wynn, U.S. Environmental
Protection Agency, January 10, 1978. See also Federal Register,
Vol. 39, No. 196, pp 36186-36207, October 8, 1974.
13. "Safe Drinking Water Act", United States Code, Public Law 93-523
(1974). See also Interim Primary Regulations in Federal Registers
of March 14, 1975, December 24, 1975, July 9, 1976 and
July 11, 1977.
14. Grier, J. C. and Roensch, L. F., "The Achievement of Slime Control
in Utility Condensers without Impairing Discharge Water Quality",
Proceedings of American Power Conference, Vol. 39, 1978 (to be
published).
15. Oliker, I., "Cogeneration Power Plants Serve District Heating
Systems", Mechanical Engineering, Vol. 100, No. 7, p. 24-29,
July 1978.
16. Spencer, D. and Gildersleeve, 0., "Market Potential for New Coal
Technologies", EPRI Journal,Vol. 3, No. 4, p. 19-26, May 1978.
17- Pavia, R. A., "Annual Report 1976 Operating Statistics", City of
Chicago, Department of Water and Sewers, Chicago, 1977.
18. Wynn, R. A. Jr. and Porter, R. W., "Beneficial Use of Waste Heat
in Municipal Water Supplies", IIT Waste Energy Management Report
(in preparation).
19. "Heat Rates for Fossil Reheat Cycles Using General Electric
Steam Turbine-Generators 150,000 kw and Larger", General Electric
Stean Turbine-Generator Products Division, Schnectady, NY,
February 1974.
20. Ortolano, L. and Smith, F. A., "Costs of Thermal Effluent Standards
for Power Plants", Journal of the Power Division, ASCE, p. 15-31, July 1974.
21. Steam-Its Generation and Use, Babcock and Kilcox Co., New York,
p. 6-18, 1972.
22. Porter, R. W. and Chaturvedi, S., "Atmospheric Spray-Canal Cooling
Systems for Electric Power Plants", Proceedings of Waste Heat
Management and Utilization Conference, Vol. 2, University of Miami,
p. IV-C-121 to 161, May 1977.
23. Fraas, A. P. and Ozisik, M. M., Heat Exchanger Design, Wiley, NY,
p. 241-255, 1965.
24. Lorenzi, D. L. and Porter, R. W.,"Simplified Analysis of Surface
Energy Exchange from Heated Bodies of Water", ASME Paper, WAM,
December 1978.
25. Bennett, I., "Monthly Maps of Mean Daily Insolation for the
United States", Solar Energy, p. 145-158, July-Sept. 1965.
26. Ruffner, J., The Weather Almanac, Gale Research, Detroit, 1974.
27. Rao, D. and Porter, R. W., "Effect of Alternate Cooling Systems
and Beneficial Use of Waste Heat on Electric Power Plant Performance,
IIT Waste Energy Management Report (in preparation).
28. Baladi, J. Y., Schoenhals, R. J. and Ayers, D. L.,"Transient Heat
and Mass Transfer in Soils", ASME Paper 78-HT-31, May 1978.
29. Holman, J., Heat Transfer, McGraw-Hill, NY, 1976.
30. Kreith, F., Principles of Heat Transfer, Intext, NY, 1973.
31. Renn, C. E. , "Warm Water Effects on Municipal Supplies", Journal
of the American Water Works Association, Vol. 49, p. 405-413,
April 1957.
32. Camp, T. R., Root, D. A. and Bhoota, B. V., "Effects of
Temperature on Rate of Floe Formation", Journal of the American
Water Works Association, Vol. 32, 1913-1927, November 1940.
33. Lue-Hing, C., Metropolitan Sanitary District of Greater Chicago,
Private Communication to A. Obayashi, April 1978.
34. Highland, J. T., "Power Plant Cooling Water Provides Domestic
Supply", Public Works, p. 101-104, November 1962.
35. Gyurek, L., Private Communication to R. Porter, The City of
Edmonton, Alberta, Canada, Water Treatment Plants, January 6,1978.
244
-------
Table 1, Comparison of Seasonal Performance of Typical
2000 Mw(e) Power Plants with Alternate Cooling near Chicago.
Fossil Fuel
Once- Closed-
through cycle
Summer
Condenser Intake (C)
Turbine Back Pres. (k P )
Thermal Eff iciency (%) a
Savings in Heat Rate(%)
Spring Fall
Condenser Intake (C)
Turbine Back Pres. (k P )
Thermal 'Eff iciency (%)
Savings in Heat Rate(%)
Winter
Condenser Intake CC)
Turbine Back Pres. (k P )
Thermal Eff iciency (%) a
Savings in Heat Rate(%)
23.2
7.6
43.4
1.5
13.3
4.2
43.6
0.9
1.4
3.4
43.7
0.6
29.9
10.5
42.8
25.7
8.5
43.2
22.5
7.3
43.4
Nuclear Fuel
Once- Closed-
through cycle
23.2
7.6
31.7
2.5
13.3
4.2
32.1
1.8
1.4
3.4
32.2
1.5
29.9
30.9
25.7
8.5
31.5
22.6
7.3
31.7
Table 2. Steady-State Data for Unpreheated and Preheated
Kater Supply Based on Chicago Configuration. (Bracketed data
refer to high initial AT = 13.9 C (Upper) and low initial
AT = 8.3 C (lower). )
Summer -Spring-Fall Winter
"Unheated Heated Dnheated Heated Dnheated Heated
Ambient Water T.(C)
Intake T^ + AT(C)
Ground T_(C)
fa/fa (annual)
Supply Discharge T (C)
o
Distribution Max
AT(C)
Overall Max AT(C)
Avg. User Intake
T (C)
WaSer-Heater Inlet
T (C)
VJa?er-Heater Flow
(L/d)
Useful-Energy Savings
(%)
Gas-Heater Savings (%)
Electric -Heater
Savings (%)
User Discharge (C)
Run-Off (C)
m (run-off) /ft (supply)
Initial Sewer (C)
21.1
21.1
20.0
1.17
21.1
0.0
21.1
[35.0]
L2 9 . 4'
20.0
1.17
p4.J
]29. Q_
~-0. T
L-0.4
0.0
21.1
21.1
285
-
-
_
22.8
22,8
0.27
22.9
'13.2-
7.9
F34.?
29.2
9.4 9.4 0.6
0.6
9.4 [23.3 0.6 0.4.31
!l7.8' 8.9l
10.6 10.6 2.2
0.96 0
96 0.92
2.2
0.9_2
9.5 P22.6, 0.7 J13.7
U7.3_ !
8.5
0.1 1-0.7 0.1 —0.7T
•-0.5 L:
-0.4_
0.1 n.3.2- 0.1 713.1"
L7-9_:
f
"33.?
28.5.
285
~30~!
: 18'
rif;
LIU
128-
Ll7j
\~24.B-
24. 2
22. 8
0.27
1 24.5"!
L24.0J
Final Sewer (C)
22.8
."24.31
_23.8j
9.4 r2
1
2.9^ 0.6 j
7.61
7.9_
14 .f]
8.7!
10.7 T22.7' 2.9 [14.9]
U7.2: Lio.ij
233 233 223
f2
jl
"1
"2
_1
22.8 [2
L2
3" - r
4 :
3~ - [
81
r - i
3.
3.5" 21.6
2.6_
10.6 10.6 0.6
0.40
0.40. 0.47
19.3 |19.8 , 14.9
'-18.9J
18.9 (19.4 | 14.4
U8.6J
223
20"i
12 :
11]
7j
"is]
n:
22.5"
21. 9!
0.6
0.47_
15. 6 1
,15.2]
15.1
J4.«.
245
-------
Title: Super Greenhouse Project Utilizing Waste Heat from Astoria #6
Thermal Power Plant*
Author: R. G. Reines
Abstract:
The energy demands for heating greenhouses in the Northeastern Region
of the United States has increased significantly in the past 3 years. This
has created interest in means to both conserve energy in greenhouses as well
as search for alternative sources of heat. Presently, greenhouse heating
systems convert high quality fossil energy to provide the temperature
portion of the energy requirements for photosynthesis. The notion that
discharge temperatures from thermal power plants are relatively coincident
with botanical temperature requirements.creates a potential large scale use
of low quality energy now discarded at all thermal power plants in New York
State. Traditional utilization of waste heat for greenhouses has been from
thermal power plants which rely on cooling towers rather than once through
cooling. Little work has been done on the investigation of using once
through power plants as potential sources of heat for greenhouses due to
low temperature heat that is available. The purpose of this discussion
paper is to present a conceptual design of a super greenhouse utilizing
waste heat from a thermal power plant, Astoria Unit #6, in Astoria, New York.
This paper will include considerations in economic and energy require-
ments of new low cost energy efficient greenhouse design and new low cost
heat exchanger configurations which would utilize this low quality heat.
In addition, potential energy and economic impact and scenarios of
implementation of large scale utilization of super greenhouses utilizing
waste heat in New York State will be discussed.
Author Affiliation:
Department of Agricultural Engineering, Cornell University
(now with ILS Laboratory, Tijeras, New Mexico)
*This paper was not presented.
246
-------
EXPERIENCE WITH THE NEW MERCER
PROOF-OF-CONCEPT WASTE HEAT AQUACULTURE FACILITY
Bruce L. Godfriaux,(1)Robert R. Shafer(2)
Albert F. Eble,*3*Mark C. Evans(3)Tom
Passanza,^3*Connie Wainwright,(3)and
Hal. L. Swindell,(3)
ABSTRACT
At the First Waste Heat Management and Utilization Conference,
a paper was given that summarized the results of our pilot
waste heat aquaculture research program and explained the
concept of sequential (diseasonal) aquaculture. The design
of a proposed proof-of-concept aquaculture facility was also
discussed. This design was subsequently modified.
In April, 1978, construction of the modified Mercer Proof-of
Concept Aquaculture Facility was completed. Facility process
water can be derived wholly or in part from five sources:
generating station discharge water, ambient river water,
well water, tempering pond (reservoir) water, and recirculated
facility process water. The operation of the overall system
is discussed.
Results through the use of this system for the rearing of
rainbow trout, Salmo gairdneri, (Richardson), completed on
June 6, 1978, in addition to the results to date (August,
1978) for the other species presently being cultured at the
facility are discussed. These species include the American
eel, Anguilla rostrata,(Lesueur) and channel catfish, Ictalurus
punctatus (Rafinesque). Projected harvest densities for the
latter two species are briefly outlined.
^Public Service Electric and Gas Company, Newark, NJ
f 2)
'Buchart-Horn: Consulting Engineers, York,PA
(3)Trenton State College, Trenton, NJ
247
-------
INTRODUCTION
The proposed new proof-of-concept waste heat aquaculture fac-
ility was described in [1] given at the First Conference on
Waste Heat Management and Utilization in May 1977. Since
then the Mercer Proof-of-Concept has been built and operated
for some five months as of August 1978.
Several modifications have been made to the original proposed
design due to monetary constraints and site specific charac-
teristics which were only discovered after aguaculture facility
construction commenced. These modifications are briefly des-
cribed later in the "Introduction" section of this paper.
The Mercer Aquaculture Facility (Fig. 1) at the Public Service
Electric and Gas Company Mercer Generating Station is four
miles south of Trenton, New Jersey, on the Delaware River. The
station which is coal-fired, generates some 600MW of power and
discharges approximately 1.7 million liters per minute of
heated discharge water at a maximum of 6°C above the ambient
Delaware River water temperature.
The recently constructed Mercer Proof-of-concept Aquaculture
Facility is shown in Fig. 2, with piping layout. Changes
in the constructed facility as compared to the proposed fac-
ility as described in [1] include the elimination of two
outdoor concrete raceways 30.5 m long x 2.4 m wide by either
2.4m or 1.2 m deep and the substitution of six 12.2 m long
x 1.8 m wide x 1.2 m deep nursery raceways by two nursery
raceways 18.3 m long x 1.8 m wide x 1.2 m deep. Also, a
recirculation chamber was added to the temperature moderation
pond (Fig. 2). The chamber is really the "heart" of
the new facility and allows aquaculture personnel tremendous
flexibility in using and blending different water sources.
The function and operation of this structure will be described
in greater detail later in the paper and in [2].
Other modifications in the constructed facility from the pur-
posed facility include the elimination of separate piping
for well water to raceways/ponds and the elimination of the
aeration lagoon in the waste treatment system for the process-
ing of raceway/pond cleaning wastes. One of the originally
purposed 5,700 liters/min discharge canal intake pumps was
also eliminated, since two of the originally proposed main
grow out raceways were dropped. The aforegoing modifications
were required when the bids on the proposed facility were
higher than anticipated.
After the wells were drilled, it was discovered that the well
water contained extremely high levels of iron (17 to 23 ppm)
248
-------
and manganese (3-5 to 3-8 ppm) . This proved toxic to trout,
and plans to set up a trout hatchery have been postponed un-
til such time or a suitable treatment method for the removal
of iron and manganese from the well water can be determined.
Without the well ater, trout cannot be reared all year round
at the Mercer Aquaculture Facility. During the summer months
the only cool water source for trout culture was well water.
DESIGN, LAYOUT AND OPERATION OF THE MERCER PROOF-OF-CONCEPT
AQUACULTURE FACILITY
The facility is a combination of previously existing pilot
facilities as described by Godfriaux et. al. [3] and shown in
Figure 2 by hatching, and recently completed proof-of-concept
facilities.
DESIRED OPERATIONAL FEATURES
From the experience of a pilot research program in the utili-
zation of heated condenser cooling water in aquaculture, it
was discovered that when the generating station cooling water
was chlorinated, it proved toxic to rainbow trout. Subse-
quently, all aquaculture facility intake pumps withdrawing
station discharge canal water were stopped automatically
during this period.
This discontinuous aquaculture process flow proved to be the
limiting factor in obtaining high density trout production
due to the rapid depletion of dissolved oxygen in the pro-
cess water. This problem was accentuated since discharge
water temperatures and the trout biomass increased through
the spring season.
For the present proof-of-concept facility, a continuous
process flow was desired through the chlorination cycles of
the generating station. This could be achieved by recir-
culating the existing aquaculture facility process water
entirely during these periods, adding additional non-
chlorinated heated discharge water to the aquaculture
facility from a storage reservoir area (tempering pond -
see Fig. 2) or by adding ambient river or well water (now
not feasible) during the station chlorination period or any
combination of the above.
RECIRCULATION CHAMBER
This was designed to provide a continous flow of process water
to the aquaculture facility even during generating station
chlorinations periods. In addition, the recirculation chamber
allows aquaculture facility personnel to use singly or in
249
-------
combination generating station discharge water,ambient temp-
erature Delaware River water, aquaculture facility process
water (through recirculation) or tempering pond (reservoir)
water.
The design and operation of the recirculation chamber is shown
in Figure 3. Under normal operations, water from the river,
station,discharge canal and well is pumped into the tempering
pond for mixing and blending. The tempering pond overflows
the distribution chamber inlet weir. The water then flows
by gravity to the various aquaculture facility raceways, ponds
and laboratories. When flow into the tempering pond is less
that the outflow from the distribution chamber, the water
level begins to fall in the distribution chamber upending
the float switches (Fig. 3). This triggers the operation of
the submersible, recirculation pumps. If the water level of
the distribution chamber falls below the level of the bypass
chamber, them a checkvalve opens allowing the effluent aqua-
culture process water to flow into the distribution chamber
for recircultion. At the same time, a restricted flow of new
water continues to flow from the tempering pond through the
tempering pond drain into the distribution chamber even
though the tempering pond water level is below the level of
the distribution chamber inlet weir (Fig. 3).
Once normal flow is restored to the tempering pond, the temp-
ering pond water level rises and eventually overflows the
distribution chamber inlet weir. The water level within the
distribution chamber rises and the checkvalve to the bypass
chamber closes. Next,the float switches are immersed in the
water and this turns off the recirculation pumps. Water
flow to the aquaculture facility then returns to normal op-
eration by gravity flow. For further detailed information
on the disign and operation of the recirculation chamber,
see [2].
FACILITY PUMP AND PIPING SYSTEM
Figure 4 shows a simplified flow schematic of aquaculture
facility process water system. Most facilities only re-
ceive the blended water from the tempering pond. However,
the nursery raceways not only receive blended water, but
also river ambient, well water (not being used at the present
time) and supplementary heated water from the heat exchanger.
These additional water sources were provided to the nursery
raceways, since they would often be rearing juveniles of
species that would be"out-of-season" in the main grow out race-
ways and would require water sources with different tempera-
ture regimes. Therefore,trout fry and fingerlings would require
cool water during the summer months and American eels, shrimp
250
-------
Macrobrachium rosenbergii and catfish would require warm, heat
exchanger water during the winter months for growth.
All waste aquaculture facility process water is routed back
to the bypass chamber of the recirculation chamber for poss-
ible re-use before being discharged to the generating station
discharge canal.
AERATION SYSTEM
Subsequent to the writing of the first paper [1], an aeration
system with two Gardner-Denver positive displacement blowers,
each capable of delivering 5.23 m 3/min of air at 9 psig were
installed as part of the proof-of-concept aquaculture facility.
A supplementary of pure oxygen bubbling system was also in-
stalled for use during emergencies. Gaseous oxygen is intro-
duced into the raceways via Schramm bioweve diffusers fitted
to a liquid oxygen container. Further, the regulator of the
oxygen container was fitted with a solenoid valve that was
opened by the same pulse that triggered the generating station
chlorine program.
By November 1978,we expect also to have installed a sidestream
oxygen injection system. The reason this system will be needed
is to supply additional oxygen when catfish and rainbow-trout
start approaching densities of 190 kg/m^. The water flow alone
to the main grow out raceways will not supply sufficient dis-
solved oxygen at these rearing densities.
With the sidestream oxygen injection system, a predetermined
quantity of water from the supply line,feeding each raceway
is further pressurized, thereby increasing the water dissolved
oxygen carrying capacity. Gaseous oxygen from a nearby liquid oxy-
gen storage container is injected into the pressurized water
which is then injected at different points in the rearing
ponds and raceways.
WATER MONITORING AND ALARM SYSTEM
An alarm system is used to provide notifications to the aqua-
culture project manager, the generating station control oper-
ator (in the power plant control room), and outside aquacul-
ture personnel in the event of any abnormal environmental con-
ditions in the project. The various parameters are being
monitored (water temperature, water pressure), by Asco temp-
erature and pressure switches. When a reading is out of
bounds, they activate electrical switching devices. The elec-
trical output of these devices is sent to a common control
relay. This relay is used to initiate an electronic tele-
phone dialer system. The telephone dialer uses a pre-recorded
251
-------
magnetic tape cartridge that contains telephone numbers and
a message. It is used to call and inform outside people that
an alarm condition exists. The dialer system is a modifica-
tion of a commercially available system made by Dryton, Inc.
for fire and security uses in residential and industrial
applications.
AQUACULTURE WASTE WATER TREATMENT
As stipulated in the EPA Draft Development Document for Pro-
posed Effluent Limitations Guidelines and New Source Perform-
ance Standards for the Fish Hatcheries and Farms (1974) [4],
process water flow may be discharged directly back into the
water source. However,water ^used for raceway cleaning should be
settled (particulates removed) before effluent is returned to
receiving waters. The waste settling basin shown in Figure
2 was constructed for removing particulates from raceway clean-
ing waters.
CHLORINE CONTROL SYSTEM
The Fisher-Porter chlorine control equipment has been modified
to include an electrical interlock with the heated discharge
water intake pumps. The operation of the interlock system
depends on the opening of the chlorine dilution water valve.
Upon activation of the control relay for -':the dilution water
valve, an auxiliary relay causes the heated discharge water
pump contactor to open thereby stopping the pumps. An addi-
tional feature of this system is the ability to operate only
one pump at a time or to override the system completely in the
manual mode.
SEQUENCE OF CULTURE OPERATIONS
Due to the large seasonal temperature fluctuation in the Del-
aware River from 0°C to 31°C in addition to a maximum of 6°C
thermal increment from the Mercer Generating station, both a
warm and cold water fish species were selected to be reared
during the warmer and colder seasons of the year. This new
aquaculture concept is called"diseasonal aquaculture". The
initial species selected for culture were rainbow trout Salmo
gairdneri (cold water species) and the freshwater shrimp
(Macrobrachium rose bergii).
However, a venture analysis 15] indicated that the production
level from Macrobr..chium rosenbergii would not be sufficient
to cover operating expenses. Therefore, starting in July 1978,
channel catfish Ictalurus punctatus is being experimentally
252
-------
cultured at high densities to determine suitability both bio-
logically and economically as the replacement warm water
species.
Figure 5 shows the sequence of rainbow trout and catfish cul-
ture operations through the year. This schedule is idealized
at the present time, since the facility well water can not
be used due to its high iron and manganese content. There-
fore trout culture during the summer months can not take
place at the Mercer Aquaculture Facility. However, this cycle
as shown would be followed during the warmer months for trout
at any off-site trout hatcheries which might be developed in
the near future.
CULTURE RESULTS WITH NEW AQUACULTURE FACILITY
RAINBOW TROUT AND CHANNEL CATFISH
From the beginning of April to August 16, 1978, the recirc-
ulation chamber has been fully operational. During this period
rainbow trout and channel catfish were being reared in two
concrete raceways, each 30.5m x 3.6m by either 1.2m or 1.8m
deep. These raceways were divided longitudinally and rainbow
trout were placed in one side of each raceway. With catfish
the raceways were divided transversely and the catfish were
placed in the upper end of each raceway. The deep raceway
was supplied with a process water flow rate of 5,700 liters/
min. (1,500 GPM) and the shallow raceway received 2,600
liters/min. (700 GPM). Density of rainbow trout during this
two month period varied from 26.7kg/m3 (1.7 lbs./ft.3) at
stocking to 51.1kg/m3 (3.2 Ibs./ft3) at harvest. Catfish
density to date has varied between 13.2kg/mJ (0.8 lbs./ft.3
to 14.8kg/m3 (0.9 lbs./ft3).
Initially, at the beginning of April, when rainbow trout were
cultured," ambient river water temperatures were sufficiently
low that only heated discharge water from the generating sta-
tion was used (T&3LE 1). During the chlorination period, pro-
cess effluent water from the aquaculture facility was recir-
culated with the addition of new, fully oxygen saturated water
via the 25cm. line from the adjacent temperating pond (Fig.
3) . This arrangement kept dissolved oxygen values above 5
ppm at the discharge end of both raceways during this period.
By the beginning of May (TABLE 1), the ambient water temper-
atures had risen to a level that temperature moderation of the
station heated discharge water was required for trout culture.
During this period 7,200 liter/min. (1,900 GPM) of water
from the station discharge canal was blended with 1,900 liters/
min. (500 GPM) of ambient river water. During station chlor-
ination periods the discharge canal pumps were shut down and
25:
-------
a 5,700 liter/min. (1,500 GPM) recirculation pump was automat-
ically activated. During the chlorination periods the 1,900
liters/min. ambient river pump was left running. Since part
of the process flow rate was recirculated and the flow rate was
reduced during the recirculation period, pure oxygen was
bubbled in through a 2cm diameter PVC pipe alone the bottom of
each raceway.
Slightly later in May (not shown in TcA'BL-E 1) , another 1,900
liters/min. ambient river pump was added to further dilute the
heated discharge water from the Mercer Generating Station.
By May 15th, with increasing Delaware River ambient water
temperatures, the heated discharge water intake pump was
shut down and only the ambient river pumps (3,800 liters/min.)
were left running. A 5,700 liter/min, recirculation pump
was also brought into service and left running constantly un-
til trout harvest on June 5 and 6, 1978. TABLE 1 indicates
that with a combination of fresh ambient river water blended
with recirculated process water, dissolved oxygen values were
maintained above 5.8 ppm at the discharge end of each raceway
during this period before harvest . During this period, how-
ever, the lower two-thirds length of each raceway was contin-
uously aerated by a 7.5cm diameter PVC pipe supplied with
air at 9 psig from 185 ft.^/min. Gardner-Denver positive
displacement blower (Model 5PDR6).
It was also noted that as the effluent water from each race-
way dropped over the weir boards, the water picked up between
two and three ppm of dissolved oxygen.
On July 11, 1978, some 60,000 17 to 20cm long catfish fing-
erlings were delivered to the Mercer Aquaculture Facility
from Arkansas. Approximately 3,000 were lost due to the hand-
ling stress of the long journey. There have been some dis-
ease problems including infestations of Epistylis. On July
28, 1978 during a formaldehyde treatment for Epistylis ap-j
proximately 7,000 additional catfish were killed. The reasons
for this were not readily apparent. The treatment was eff-
ective in eliminating the Epistylis infestation. However,
we are now having bacterial infections of Aeromonas and Pseud-
omonas which is being treated with Terramycin mixed in the
catfish feed.
TABLE.2 shows operational data for the aquaculture facility
during the catfish culture period to date (August 16, 1978).
The reason that the catfish biomass has not increased over the
one month culture period is that there has been mortalities of
approximately 11,000 catfish of 60,000 originally stocked.
254
-------
One reason for the higher incidence of disease among catfish
than we originally expected is that since July we have on a
partial water recirculation mode. This has been due to two
reasons: the ambient river water temperature plus thermal
increment from the generating station has resulted in the
station discharge water being higher that the catfish optimam
growth temperature, and our new 5,700 liter/min. (1,500 GPM)
discharge canal intake pump unexpectedly broke down. Both
ambient river and remaining discharge canal water flows are
not sufficient to meet agriculture process water flow^equire-
ments. Thus, partial recirculation of aquaculture process is
required .
STRIPED BASS AND EELS
Approximately 6,000 5cm long striped bass fingerlings weigh-
ing 3.2kg were picked up from Edenton National Fish Hatchery
in Edenton, North Carolina on June 28, 1978. Remaining fishasof
August 8, 1978 now weigh 12.7kg and vary in length from 6cm
to 10cm. Mortality rate to date has been 2,562 fish or 42.7%.
All mortalities have been attributed to fungal infections of
Saprolegnia. Treatments to control these infestations have
been with Formaldehyde at 125 ppm for one hour. Feeding of
striped bass has consisted of 35% protein trout feed four times
per day every two hours at 6.9% of their total body weight.
On April 4, 1978, some 3,000 to 3,500 early juvenile eels lg
to 2g in weight were received at the Mercer Aquaculture Fac-
ility. An additional 15,500 juvenile eels were delivered to
the facility on May 27, 1978 and placed in the nursery race-
ways (Fig. 2) for grow out. At the present time (August 15th)
the juvenile eels weigh on average approximately lOg each.
Projected goals for the grow out of eels are to realize harvest
densities of 35 to 50kg/m2 in the nursery racex^ays.
We hope to harvest 8,250 eels at 200g each over a time period
of 8 to 10 months with a cumulative mortality of 50-55%
DISCUSSION AND CONCLUSIONS
The new Mercer Proof-of-Concept Aquaculture Facility has
given aquaculture personnel great versatility in maintaining
various operational modes at the installation. This operation-
al flexibility of the Mercer Proof-of-Concept Aquaculture
Facility is made possible by the recirculating chamber. Over
the first five months of use, the recirculation chamber has
allowed aquaculture facility personnel to choose between a
wide variety of operating options by selecting the blend
and source of aquaculture facility process water. This has
allowed for the maintenance of water temperatures as near
optimum as possible for the culture of species being reared
255
-------
at the time. These operating options are listed below.
1. Use of heated discharge water only
2. Use of ambient river water only
3. Blend (in varying proportions) of ambient and station
discharge waters
4. Recirculation of any of the above waters
5. Recirculation of any of the above waters with addition-
al water from the adjacent tempering pond which acts
as a temporary water storage reservoir
6. Partial recirculation of process water with balance of
required facility process flow coming from one or
both of the above water sources (discharge or ambient
river water)
Another attractive feature of the proof-of-concept aquacul-
ture system is that it can maintain process flow by gravity
for 30 to 60 minutes by using the tempering pond water supply
in the event of an electric power failure. We temporarily
lost electric power to the aquaculture facility for 20 min-
utes. This feature was greatly appreciated at the time.
Results with rainbow trout culture which achieved densities
up to 50.9kg/m3 indicate that water flow rates alone will
not supply sufficient dissolved oxygen if trout and catfish
harvest densities are nearly quadrupled to 178.2kg/m3. ^
supplementary oxygenation through a sidestream oxygen injec-
tion system will be required.
To date channel catfish have shown greater susceptability
to disease than rainbow trout with consequent higher mort-
ality rates. The initial long truck journey from Arkansas
to New Jersey no doubt severly stressed these fish. It can
not be determined at the present time whether transportation
stress is responsible for a good portion of the disease
outbreaks with catfish.
256
-------
ACKNOWLEDGMENTS
This research is funded by NSF/RANN Grant ENV 76-19854 and
PSE&G Authorization RD-443. Other contributions to the
/project have been made by Long Island Oyster Farms, Trenton
State College and Rutgers University.
Many persons from outside the project have provided valuable
suggestions and help to this work. We would like to acknowl-
edge in particular the help provided by: Dr. E. H. Bryan
(Program Manager, Sanitary/Environmental Engineering, NSF-
ASRA/PFRA), Mr. R. Pitman (General Manager, LIOF), Mr. R. A.
Huse (General Manager, R&D, PSE&G) and Mr. J. Morrison
(Manager, Mercer Station, PSE&G). In addition, we would
like to express our appreciation for the help and support
that the Maintenance, Performance and Yard Departments of
the Mercer Generating Station have given us since the
beginning of the project.
257
-------
REFERENCES
1. Guerra, C. R., B. L. Godfriaux and C. J. Sheahan 1977:
Utilization of waste heat from power plants by sequential
culture of warm and cold weather species in (Lee, S. S.
and S. Sengupta (Eds.) Proceedings of the Conference
on Waste Heat Management andUtilization, Vol. 2,
May 9-11, 1978 in Miami Beach, Florida. pp. V-C-213
to V-C-232.
2. Godfriaux, B. L. and R. R. Shafer in press: A method
of maintaining aquaculture process flow during chloriria-
tion of electric generating station cooling water, in
(Godfriaux, B. L. et al, eds.) Proceedings of Power
Plant Waste Heat Utilization in Aquaculture - Workshop II,
March 29-31, 1978, Rutgers University.
3. Godfriaux, B. L., H. J. Valkenburg, A. Van Riper and
C. R. Guerra 1975: Power plant heated water use in
aquaculture in Proceedings of the Third Annual Pollution
Control Conference of the Water and Wastewater Equipment
Manufacturer's Association, April 1-4, 1975. pp. 233-50.
4. Schneider, R. F. 1974: Development document for proposed
effluent limitations guidelines and standards of per-
formance for the fish hatcheries and farms. Prepared
by the Environmental Protection Agency, Office of Enforce-
ment, National Field Investigations Center,- Denver,
Colorada. 237 p.
5. Godfriaux, B. L., C. R. Guerra and R. E. Resh 1977:
Venture analyses for intensive waste heat aquaculture.
in (Avault, J. W ., (eds) Proceedings of the Eighth
Annual Meeting of the World Mariculture Society,
January 9-13, 1977, San Jose, Costa Rica. pp. 707-22.
258
-------
TARLR I
IO
OPERATIONAL DATA FOR AQUACULTURE FACILITY PROCESS WATER
DURING THE CULTURE OF RAINBOW TROUT SALMO GAIRDNERI
Dissolved Oxygen
(ppm)
Raceway
Shallow
Shallow
Deep
Deep
Shallow
Shallow
Deep
Deep
Shallow
Deep
Raceway
Date Inlet
4/7/78
4/7 /78a
4/7/78
4/7 /78a
5/6/78
5/6 /78a
5/6/78
5/6 /78a
5/2 1/7 8b
5/2 1/7 8b
10.4
10.2
10.4
10.4
10.2
8.7
10.2
8.4
9.6
9.6
Raceway
Outlet
7.4
8.9
8.0
9.5
6.5
6.1
4.5
4.4
5.8
5.8
Process Water Source liters/min. (GPM) Trout Biomass
Discharge Canal Ambient River
9,100 (2,400)
-
9,100 (2,400)
-
7,200 (1,900) 1,900 (500)
1,900 (500)
7,200 (1,900) 1,900 (500)
1,900 (500)
3,800 (1,000)
3,800 (1,000)
Recirculation kg/m3
33.0
5,700 (1,500) 33.0
35.0
5,700 (1,500) 35.0
, 43.9
5,700 (1,500) 43.9
43.2
5,700 Jl, 500' 43.2
5,700 (1,500) 50.9
5,700 (1,500) 51.1
lbs./ft.3
2.0
2.0
2.2
2.2
2.7
2.7
2.7
2.7
3.1
3.2
Process Flow
Temper, are
13.9
13.9
14.0
14.0
15.0
15.0
15.0
15.0
17.4
17.5
aPure oxygen was bubbled through 2cm diameter PVC pipe at the bottom
of each raceway during water recirculation periods.
bThe lower two-thirds length of each raceway was continuously aerated
by a 7.5cm diameter PVC pipe supplied with atmospheric air at 9 psiq
from 5.25m3/min. positive displacement blower.
69-09
-------
TABLE 2
OPERATIONAL DATA FOR AQUACULTURE FACILITY PROCESS WATER
DURING THE CULTURE OF CATFISH ICTALURUS PUNCTATUS
Raceway
Date
Dissolved Oxygen
(ppm)
Shallow 7/12/78
Deep 7/12/78
Shallow 8/1/78a
Deep 8/2/78a
Shallow 8/17/78a
Deep 8/17/
Raceway
Inlet
7.2
7.2
8.4
7.5
1 7.4
1 7.2
Raceway
Outlet
6.8
6.6
7.4
6.2
5.9
4.5
Process Water Source liters/min. (GPM)
Catfish Biomass
Discharge Canal Ambient River Recirculation kg/n>3 lbs./ft.^
5,700 (1,500)
5,700 (1,500)
3,000 (800)
3,000 (800)
3,000 (800)
3,000 (800)
3,800 (1,000) 11,300 (3,000) 7.4 0.4
3,800 (1,000) 11,300 (3,000) 14.5 0.9
3,800 (1,000) 11,300 (3,000) 9.2 0.6
3,800 (1,000) 11,300 (3,000) 14.7 0.9
3,800 (1,000) 11,300 (3,000) 9.7 0.6
3,800 (1,000) 11,300 (3,000) 18.9 1.2
Process Flow
Temperature
28.0
28.0
25.0
24.0
29.0
29.0
N>
O>
O
aPure oxygen was bubbled through 2cm diar..et- PVC
of each raceway.
-*: the bottom
69-74675
-------
MERCER
GENERATING
Figure 1. Map of the Delaware River Drainage Basin Showing
the Location of the Mercer Generating Station
261
-------
/>
Figure 2- Scheme of
Renovated Aquaculture Facilitiefc,
Mercer Generating Station,
Trenton, NJ
Ni
previously
existing
pilot aqua-
culture facilities
1. Lime* from Well 1 and Well 2 to laboratories and raceway*
2. Line from Delawire River to Aquaculture Syatem
3. Line from Mercer Station discharge canal to Temperature-
Moderation Pond
-------
GRAVITY DRAIN
FROM ALL
RACEWAYS.
POND,:*
DORSAL VIEWS OF RECIRCULATION CHAMBER
A— SUBMERSIBLE PUMPS
WOODEN WE&
FOR WATER~~~
LEVEL CON-
TROL
DRAIN
TO RIVER
VALVE FROM
RECIRCU-
LATINO
POND
LOWER CHECK
Tf
I
JLJ
VALVE TO RACEWAYS, POND, G.H.
FLOATATION ON
AND OFF SWITCHES
WOODKBHEIB
GRAVITY
DRAIN FROM
ALL RACE-
WAYS, POND,
fc G.H.
ON
WATER FROM SUBMERSIBLE PUMPS
TO RACEWAYS, POND, & G.H.
LATERAL VIEW OF SYSTEM
TOP VIEW OF PIPING
ON RECIRCULATING
CHAMBER
TEMPERING POND
.TEMPERATURE MODERATION
POND DRAIN
DRAIN TO
RIVER
VALVE TO REGULATE
WATER FROM RECIRCULATING POND
CHECK VALVE, PIPE LEADS TO
RACEWAY, POND, G.H.
SUBMERSIBLE PUMPS
TOTAL GPM
POSSIBLE
5,100 GPM'
CANAL
WATER
,100
.GPM
TOTAL
RIVER WATER ' WELL WATER
1,000 GPM TOTAL 1,000 GPM
TOTAL
DISTRIBUTION
CHAMBER
Figure 3.. Operation of Tempering Pond - Recirculation Chamber System
-------
WELLS
3,800 1/min.
DISCHARGE
CANAL
11,700 1/min
RIVER
AMBIENT
3,800 1/min.
BY-PASS
Recirculation Ch
\
Temp.ering
Pond
imber
O
,O
"cula
tion
Chamber
Pumps
O
O
tt O
Pep
O
I -H
a -P
o>
-------
MONTHS
ACTIVITY
Ictalurus punctatus
1. Brood Stock Spawned
2. Eggs Hatched & Larval Cycle
Completed
3. Early Fry Reared
U. Fry-Early Fingerling
Grov Out
5. Fingerling Grow Out
6. Final Grow Out & Harvest
7. Brood Stock Recruited &
Maintained
LOCATION
J F M A M J J A S 0 N D
Laboratory I
Laboratory I
Laboratory I
Laboratory II
Insulated Nursery Raceways
Outdoor Ponds & Raceways
Laboratory II
Salmo gairdneri
1. Eyed Eggs Received
2. Eggs Incubated & Hatched
3. Early Fry Reared
U. Fry Grow Out (9 cm at
stocking)
5. Fingerling Grow Out
(20 cm at stocking)
6. Harvest 30 cm Trout
Laboratory II
Laboratory II
Laboratory II
Nursery Raceways
Outdoor Ponds & Racewavs
Outdoor Ponds & Raceways
Figure 5, Sequence of Shrimp and Trout
Culture Operations at the
Mercer Aquaculture Facility
S = Start, H = Harvest
-------
WASTE HEAT RECOVERY IN THE FOOD PROCESSING INDUSTRY
W. L. Lundberg and J. A. Christenson
Westinghouse Electric Corporation
Advanced Energy Systems Division
F. Wojnar
H. J. Heinz Company
U.S.A. Division
Pittsburgh, Pennsylvania U.S.A.
ABSTRACT
A project is described which evaluated the potential for waste heat recovery
in the food processing industry. The work was performed by Westinghouse
Electric Corporation under a Department of Energy contract and with the
cooperation of the H. J. Heinz Company. In particular, applications of
thermal storage were sought. However, the project was not restricted to
those applications, and attention was also given to satisfying immediate
energy needs with recovered waste heat. The primary purpose of the work
was the study of waste heat recovery systems and methods that could have a
significant impact upon the nation's energy consumption if the food industry
applied them on a large-scale basis. The paper discusses the technical
aspects of potential waste heat recovery systems and the economics of in-
stalling them at selected survey factories.
INTRODUCTION
Westinghouse recently completed a nine month study project to assess the
potential for economical waste heat recovery in the food industry and to
evaluate prospective waste heat recovery systems. The project was funded
by the U.S. Department of Energy (DOE) and was conducted by Westinghouse
with the cooperation of the H. J. Heinz Company. Heinz arranged access
to two of their manufacturing plants within the Heinz U.S.A. Division and
permitted Westinghouse personnel to analyze factory and food system
operations. At each factory, a waste heat availability study was performed
and the resulting data were then applied as the basis for recovery system
conceptual design. The system studies focused on thermal performance and
economic evaluations and in this paper, the results of that effort and
project conclusions regarding feasibility are presented.
SURVEY SITES AND SELECTION RATIONALE
The Heinz U.S.A. Division operates eleven factories engaged in a variety
of thermal food processes. Two of those factories,located in Pittsburgh
and Lake,City, Pennsylvania,were selected as survey sites for our project.
The Pittsburgh Factory is a large multi-building facility employing over
2,000 production people and specializing in foods processed with steam and
hot water. The factory's main products are baby foods and juices, canned
soups and canned bean products. At the Lake City Factory, approximately
150 employees are engaged in the preparation of dessert products that are
quick-frozen as the last step in the process. Lake City operations are
conducted in a modern, single-story building having a total floor area of
approximately 70,000 ft2.
266
-------
The products manufactured by the Pittsburgh Factory place it in the Canned
Specialties (SIC 2032) industry but the processes used are also common to
Canned Fruits and Vegetables (SIC 2033) as well. Similarly, the Lake City
products would place that factory, strictly speaking, within the Ice Cream
and Frozen Desserts (SIC 2024) industry. However, similar food processes
are also employed in the Frozen Specialties (SIC 2038) and Frozen Foods
and Vegetables (SIC 2037) industries. Based upon food industry data, the
annual energy consumption by these five industries (SIC 2032, 2033, 2024,
2038, and 2037), which are represented by our Pittsburgn and Lake City
survey sites, totals to nearly 120 x 1012 Btu (Canned to Frozen Ratio
%60/40) and accounts for 13% of the total food industry (SIC 20) energy
usage. If waste heat recovery systems were implemented to reduce energy
consumption in those five industries by 10%, the annual savings, nation-
wide, would amount to 0.012 quad or approximately 1.3% of the current SIC
20 usage. This potential industry impact is significant and therefore it
appeared that survey and conceptual system design work at the Pittsburgh
and Lake City Factories would be very worthwhile. This was the main
reason for selecting those particular factories for use as survey sites.
WASTE HEAT AVAILABILITY
At each factory, operations were analyzed and waste heat production rates
were estimated. At the Pittsburgh Factory, hot waste water is the main
waste heat source and the estimates were based, in the case of several con-
tributing systems, upon actual measurements of waste water flow rates and
temperatures. Where only temperatures could be conveniently measured, the
flow rates were estimated by applying the temperature data and known values
of product temperatures and flow rates to a system energy balance. In
several cases, neither waste water flow nor temperature could be determined
conveniently and in those situations, the systems were modeled mathemati-
cally to produce estimates of the required parameters.
At the Lake City Factory, the refrigeration system condensers represent the
waste heat source of most interest during this project. For conceptual
system design purposes, average refrigerant flow rates and condenser heat
dissipation rates were predicted based upon estimated compressor electrical
loads and assumed refrigerant cycle state points. In this section, the
food processing systems at the two survey sites are described and the
results of the waste heat availability study at each site are summarized.
Pittsburgh Factory
The Pittsburgh Factory operates several food systems that produce hot waste
water. To eliminate the possibility of product contamination, waste water
from the processes is not recycled nor is waste heat recovery attempted.
Instead, the waste water streams are collected by a drain system that
transports them from the factory via clear water and sanitary sewer systems.
The various waste water sources evaluated at the Pittsburgh Factory and
waste water conditions for individual units are identified in Table 1.
To assess the total waste heat availability, the unit data from Table 1
were combined with operational data (cycles per shift, cycle times, number
of units in operation, etc.) supplied by Heinz U.S.A. The resulting
information was compiled by shift and on a building-by-building basis and
267
-------
is summarized in Table 2. The table entries apply to those systems
selected for waste heat recovery. Systems not represented in Table 2 were
neglected due to low temperature and/or low volume reasons and to high fouling
potential in the case of the blowdown heat source.
Lake City Factory
The Lake City Factory has two major refrigeration requirements - blast
freezing and cold storage. Pie and cake products complete the preparation
process on a conveyor line that transports them through a blast freezer
maintained at -35°F. When they exit the freezer in 20-40 minutes, the
frozen products are placed in a storage freezer (-5°F) to await shipment.
The refrigeration effect for both freezers is provided by an ammonia system
driven by ten rotary and reciprocating compressors. On the discharge
side of the compressors, the refrigeration system is arranged in two
separate parts. Eight compressors (five blast freezer, two storage and
one ice bank unit)* discharge to six manifolded condensers while the two
remaining compressors (Group 2) are staged units that serve a storage
freezer and discharge to their own condenser. Since blast freezer opera-
tion only occurs during the one or two production shifts of each day, waste
heat rejection at the first condenser group peaks during those shifts and
declines to a minimum (imposed by the ice bank and storage load) during
the night time hours. Heat rejection at the independent condenser, however,
is more uniform over the entire day due to its storage function. The
point of this discussion is that two independent sources of refrigeration
waste heat exist at the factory that behave differently with time. There-
fore, they could be linked with two independent waste heat applications
whose energy demands also follow different time patterns.
At the compressor discharge, the assumed ammonia vapor conditions were 185°F
at 150 psia. An average refrigerant flow rate for each compressor was
calculated based upon factory electrical data, compressor on-time estimates
and nameplate horsepower, assumed state-points and a compressor mechanical
efficiency of 0.85. The resulting refrigerant flow rate for the Group 1
compressors during production hours is 7,390 Ib/hr and for Group 2, 1,860
Ib/hr. On the basis of these flow rates and other system data, the annual
heat dissipation at the condensers totals 4.0 x 1010 Btu. For comparison,
this quantity is larger than the factory's annual total energy consumption
(gas and electricity combined) by approximately 50%.
WASTE HEAT APPLICATIONS
Pittsburgh Factory
Three applications of waste heat were selected for study at the Pittsburgh
Factory. They involve the heating of boiler make-up water, fresh water
for food processes and factory clean-up water. The main method of heating
process water at the factory is by the direct injection of cold water with
steam. Since process water is not recycled, condensate losses are high
and must be matched by make-up. By Heinz U.S.A. estimates, nearly two-
thirds of the boiler feedwater flow is composed of make-up water and
*Referred to subsequently as the Group 1 compressors.
268
-------
preheating that water would be an excellent use of waste heat. In this
case, conventional heat exchange could be used and the timing is opportune
since the highest demand for make-up occurs during production when waste
heat availability also peaks. A similar application exists in the case of
food processing water.
Production area clean-up is a daily operation that requires significant
energy input. While spot clean-ups occur as needed during the production
hours, the major clean-up effort is performed during third shift when
production hardware is disassembled and washed or cleaned in place.
Heinz U.S.A. estimates that the factory requires 200,000 gal. of heated
clean-up water (150-180°F) per day and meeting that demand is responsible
for 3-4% of the plant's total energy consumption. Waste heat could be
used to satisfy a portion of the required energy. However, since this
particular need and the waste heat supply are not in phase, thermal energy
storage would be required.
Lake City Factory
The Lake City Factory also has a daily need for heated clean-up water
(8,000 to 10,000 gal. at 150°F) and waste heat from the refrigeration
system could satisfy it. Again, thermal storage would be needed since
the waste heat supply and the demand for energy in this case are not in
phase.
A second waste heat application involves the heating of air which is blown
beneath the freezer floors. A potential problem with ground floor freezers
is that the earth beneath the floor will freeze eventually and cause the
floor to heave. To eliminate or minimize the problem, freezer plants
usually install floor insulation and warm the earth beneath the floor
electrically or by circulating a warm non-freezing liquid through a piping
grid. A third method to prevent freezing is to blow warm air through ducts
installed below the floor. This method is employed at Lake City. Outdoor
air is drawn in through a roof-mounted system composed of a vaneaxial
fan and an air temperature controlled gas-fired heater. The modulated
heater is activated whenever the outdoor temperature falls below 7p°F.
The system is rated at 7000 scfm and the annual air heating load is 1.4 x
109 Btu or approximately 5% of the factory's total energy need.
WASTE HEAT RECOVERY SYSTEM DESIGN AND EVALUATION
Pittsburgh Factory
Several recovery system concepts to satisfy the waste heat applications
discussed above are being considered for use at the Pittsburgh Factory.
Features of the reference system, shown schematically in Figure l,will be
discussed herein. In that concept, waste water streams at various temp-
eratures are collected from processes occurring in three production build-
ings. The high temperature streams (i.e., those above 140°F) are collected
in the high temperature accumulator (HTA) while all low temperature streams
are channeled to a low temperature accumulator (LTA). The LTA and HTA
will be insulated and they serve as surge tanks between the waste water
269
-------
source and application points. All of the waste water sources identified
in Figure 1 produce continuous waste water streams except the retorts.
Retort operation occurs on a batch basis and the temperature of the waste
stream emitted by a retort varies with time. It is intended, therefore,
that the retort drain system will be equipped with temperature-sensing,
three-way valves. The valves will direct waste water to the HTA during
the high temperature part of the cool down and to the LTA when the waste
water temperature falls below a set-point value. In studies performed
to date, the set-point has been 140°F. The purpose in separating the
waste water streams by temperature as opposed to mixing them pertains to
the storage part of the system. Using high temperature waste water to
prepare the highest temperature stored water decreases the cost/benefit
ratio for storage and causes it to approach that for the entire system.
From the HTA, waste water flows on demand to the high temperature heat
exchanger (HTHX) located in the Power Building. At the HTHX, heat is
transferred to incoming fresh water as it flows to various food processes.
When those demands diminish while hot waste water is still available at
the HTA, the system will automatically divert the flow of heated fresh
water to storage. Water that accumulates in storage during the production
period would then be used during third shift for clean-up purposes.
Waste water collected in the LTA will flow to two low temperature heat
exchangers (LTHX) also Icoated in the Power Building and mounted in para-
llel. Waste heat will be applied at that location to preheat fresh
water for food processing and for boiler make-up. The parallel heat
exchanger arrangment is necessary since the food processing and make-up
applications require water from two different sources — on-site wells and
the city water main, respectively.
Estimated materials and installation costs for the reference waste heat
recovery system are identified in Table 3. Allowing an additional 15%
for engineering brings the total system cost to $413,300. At current fuel
prices, Heinz U.S.A. estimates the delivered value of energy at $3.05 per
million Btu's and in one year's time (220 production days, 2 shifts per
day, 7 hours per shift), it is further estimated that the reference system
described above will recover approximately 7.0 x 1010 Btu thereby reducing
factory energy consumption by 5-6%. Thus, the projected dollar savings
produced annually by the system are $214,000 which would return the capital
investment at the rate of 35% per year. The return on investment (ROI)
calculation is based upon the Heinz ten-year, discounted cash flow method
and assumes a 10% investment tax credit, a 12% depreciation rate and a
50% income tax. The calculation made no allowance for fuel price escala-
tion.
Lake City Factory
For the Lake City Factory, two independent systems have been evaluated.
The first system, shown schematically in Figure 2, will apply refrigeration
waste heat from the Group 1 compressors to heat fresh water for later use
during third shift clean-up operations. The system will be located in an
existing water distribution system at a point between the softener exit
and the first clean-up station take-off. The thermal storage tank and its
270
-------
circulation loop will be solid at water main pressure with a storage
capacity of approximately 6,000 gallons which would be sufficient to supply
the factory's daily clean-up water needs. System control will center on
the storage tank temperature and the refrigerant temperature at the heat
exchanger exit. For tank temperatures less than a set point value (say
150°F), the centrifugal circulation pump will energize provided the re-
frigerant temperature is greater than a set point value representing its
saturation temperature. With those conditions satisfied, the temperature
sensing flow control control valve will open and water will circulate from
storage to the heat exchanger and back to storage. The water will enter
the thermal storage tank via a low velocity distribution ring near the tank
top and horizontal baffling arranged along the height of the tank will pro-
mote stratification. The circulating pump will take its suction from the
low temperature zone at the bottom of the tank and water to be delivered
to the clean-up stations would be withdrawn at the top.
The system heat exchanger will operate as a desuperheater and under normal
conditions, vapor will exit the unit with 5-10°F of superheat still re-
maining. The vapor will then flow to the existing condensers to complete
the heat rejection process.
Table 4 presents materials and installation cost estimates for the water
heating system. The total system cost, including engineering, would be
$36,900 and the system would displace natural gas at the rate of nearly
1900 MCF annually. This reduction in fuel consumption is valued, at
current fuel prices, at $3,870 per year and it would reduce the factory's
total energy input (gas and electricity) by 7% and its natural gas con-
sumption by 13%. Again based upon the Heinz project evaluation method
and assumptions, the annual ROI for the water heating system will be 8%.
A second waste heat recovery system for the Lake City Factory is shown
schematically in Figure 3 and would apply refrigeration waste heat to
warm freezer floor air. As the figure indicates, outdoor air is presently
warmed by an existing gas-fired heater and distributed to floor ducts by
a fan. The fan operates continuously and air temperature control is
achieved by regulating the flow of natural gas to the heater. Waste heat
from the Group 2 compressors could be utilized to preheat or completely
heat floor air by installing an air-cooled condenser at the heater inlet.
Refrigerant vapor from the storage freezer system would be routed through
the condenser on its way to the existing evaporative condenser. The
system control would cause the incoming air to bypass the new condenser
in any proportion to maintain the gas heater inlet tmeperature at a set-
point value. Thus, the fraction of the refrigerant condensing load handled
by the new condenser would also be a variable ranging from zero during
warm weather operation to nearly one-fourth under design, cold-weather
conditions. It is noted that the function of the recovery system would be
to supplement the existing gas-fired system. Thus, any portion of the
air heating load not satisfied by waste heat will be supplied automati-
cally by the gas heater which will continue to operate with its own
control system and independent of the waste heat recovery system.
271
-------
Costs to procure and install an air heating system are identified in Table
4 and the total estimated cost, with engineering, is $17,700. The system
would displace natural gas at the yearly rate of 1390 MCF (valued at
$2,880) and reduce the factory's natural gas and total energy consumption
by 10% and 5%, respectively. The air heating system ROI is estimated at
14%.
CONCLUSIONS
Based upon our study, it is concluded that waste heat recovery from certain
food processes is feasible and can be performed economically using avail-
able, off-the-shelf hardware. This conclusion is certainly valid for the
large Pittsburgh Factory where both the anticipated ROI and the predicted
fuel and energy savings are especially attractive. In fact, the predicted
performance of the Pittsburgh Factory system is sufficiently attractive
that plans are being developed to design and install an instrumented
demonstration system at the factory. The purpose of the DOE supported
project will be to evaluate actual hardware performance, to optimize system
design and to determine actual costs and benefits resulting from the waste
heat recovery system. The results of the project will then be publicized
to encourage the application of similar waste heat recovery concepts within
the food industry.
While the predicted economic performance of the Lake City systems is less
than desired, it is believed that the project results do warrant additional
system studies. This is especially true in the system scaling area and
the results of the study should be extrapolated to refrigeration plants of
other sizes. In fact, a similar effort should also be undertaken for
the hot water systems of the Pittsburgh Factory type. Through such efforts,
it will likely become evident that retrofit projects within a certain heat
recovery range are more feasible than others and plant sizes and con-
ditions most appropriate for such projects could therefore be identified.
272
-------
TABLE 1. INDIVIDUAL UNIT WASTE WATER CONDITIONS — PITTSBURGH FACTORY
Source
Continuous Can Washer
Continuous Bottle Washer
Continuous Cooker/Cooler
Horizontal Stationary Retorts
Canned Products
Glass Products
Continuous Pasteurizer
Continuous Cooler
Horizontal Rotary Retorts
Type 1
Type 2
Continuous Boiler Slowdown
Flow Rate (gpm)
12
7
60
100
100
80
55
10
30
12
Temperature (°FJ
175
175
140
100
140
170
120
135
125
212
TABLE 2. WASTE WATER SUMMARY — PITTSBURGH FACTORY
Power Building
Continuous Cooker/Coolers
Can Washers
Meat Products Building
Horizontal Stationary Retorts
(Glass Products)
Can Washers
Bottle Washers
Continuous Pasteurizer
Bean Building
Can Washers
Continuous Coolers
Total Flow Rate & Average
Temperature
Average Flow Rate & Temperature
First Shift Second Shift
120 gpm-140°F
37 gpm- 175°F
316 gpm-140°F
25 gpm-175°F
7 gpm-175°F
80 gpm-170°F
25 gpm-175°F
110 gpm-120°F
720 gpm- 145°F
60 gpm-140°F
12 gpm-175°F
316 gpm-140°F
12 gpm-175°F
7 gpm-175°F
80 gpm- 170°F
12 gpm-175°F
55 gpm-120°F
554 gpm-145°F
273
-------
TABLE 3. ESTIMATED COSTS — PITTSBURGH FACTORY SYSTEM
Item Materials ($) Installation ($)
Tanks
Heat Exchangers
Pumps
Strainers
Valves
Piping
Instrumentation/Controls
90,000
44,400
13,400
10,000
15,500
27,300
17,800
6,700
6,000
4,000
12,800
92 ,200
19,300
218,400 141,000
Total Materials & Installation $359,400
TABLE 4. ESTIMATED COSTS — LAKE CITY WATER HEATING SYSTEM
Item Materials ($) Installation ($)
Thermal Storage Tank
(6000 gal)
Heat Exchanger
Pump
Water Flow Control Valve
Temperature Sensor & Transmitter
Piping
8,300
3,500
500
1,100
3,500
2,300
3,000
1,200
500
200
400
5,000
19,200 10,300
Total Materials & Installation —$29,500
TABLE 5. ESTIMATED COSTS — LAKE CITY AIR HEATING SYSTEM
Item Materials ($) Instal lation ($)
Condenser
Piping
Air Ducting
Controls
Fan Motor
2,800
600
} 3,000
500
1,800
1,100
1,200
800
6,900 4,900
Total Materials & Installation —$11,800
274
-------
N)
--J
Cn
MEAT PRODUCTS BUILDING
HIGH TEMPERATURE WASTE SYSTEM ( >I40 F)
LOW TEMPERATURE WASTE SYSTEM (< 140°F)
THERMAL STORAGE SYSTEM
HOT PROCESS WATER
•BOILER MAKEUP
HTA-HIGH TEMPERATURE ACCUMULATOR
HTHX-HIGH TEMPERATURE HEAT EXCHANGER
HTb - HIGH TEMPERATURE SUMP
LTA-LOW TEMPERATURE ACCUMULATOR
LTHX-LOW TEMPERATURE HEAT EXCHANGER
Fig. 1. Waste Heat Recovery System - Pittsburgh Factory
-------
<
DE3iJPE?,HEAT;'!G
HEAT EXCHANGES
S5'F
90 "F
CONDENSERS
(EXIST)
i 1 7pm
185°F
7390 Ib/hr
Condensate Refrigerant
PJeturn Vapor [n
(Group I)
X
ISO'F
FKtorv Roof
City /
-------
GENERATION OF CHILLED WATER FROM CHEMICAL PROCESS WASTE HEAT
Jack Entwistle
Fiber Industries, Incorporated
Charlotte, NortbfcCarolina U.S.A.
ABSTRACT
Recovery of waste heat from a chemical process column, and the
conversion of waste heat into chilled water, was accomplished at the
Palmetto, S. C. plant of Fiber Industries, Inc. by the installation
of a waste heat boiler/absorption chiller system. Overhead vapors,
at 40 pslg, from two columns,provide the heat source. Two waste heat
boilers were installed adjacent to the columns. Each waste heat boiler
is capable of generating 13900 pounds of steam per hour at 16 psig
when the plant is operating at maximum production. The absorption
chiller has a nominal capacity of 1140 tons and at maximum rating
requires 21000 pounds of steam per hour at 12 psig. Lithuim bromide
is the absorbent.
INTRODUCTION
Expansion of plant production facilities required an additional chiller
to supply 45°F water to air conditioning units. Two alternatives were
evaluated.
1. Installation of an additional electrically driven centrifugal chiller,
similar in design to existing equipment.
2. Installation of a waste brt*;boiler/absorption chiller system
utilizing waste heat in the overhead vapors of a plant process
column.
The study estimated that the waste heat boiler/chiller installation would
cost $600,000, the centrifugal chiller installation would cost $300,000
and the annual operating costs would be $100,000 lower, at existing
electrical rates, if the absorption chiller was installed. A present-
worth analysis determined that the absorption chiller was the best
economic choice.
The actual installation cost was $450,000 and electrical rate increases
have paralled the original estimate so that project economics have
proved to be better than the study predicted.
277
-------
SYSTEM DESCRIPTION
General
Prior to Installation of the waste heat recovery system, 28 million
Btu/hour of energy was wasted In the condensation of process column
overhead vapors, using cooling tower water as the method of removing
heat. The waste heat recovery system makes a large fraction of the
energy contained 1n these column vapors available for generation of
chilled water 1n the absorption chiller. Because of the fear of con-
tamination either by or to the process 1t 1s not practical to use these
vapors as a direct makeup to the plant steam system.
The system recovers this energy by use of the two waste heat boilers
located adjacent to the columns. Vapors from the column overheads
exit the column at 40 pslg and 289°F. These vapors are piped to the
channel (tube) side of the waste heat boiler, where they are condensed
by boiler feed water surrounding the tubes (relatively colder at 252°F).
The latent heat released by the condensation 1s transferred to the
surrounding water, and 16 pslg steam Is produced. The condensed
overheads are then returned to the column overhead piping, upstream
of the water-cooled condenser. The water level outside the tubes within
the waste heat boiler 1s maintained by use of centrifugal pumps which
transfer feed water from the utility boiler deaerator system. Solids
control within the waste heat boiler 1s achieved by manual Inter-
mittent blowdown from the boilers.
Steam is piped to the lithium bromide cycle absorption chiller which 1s
located alongside the centrifugal chillers in the plant utilities
building. This location enabled convenient connections to be made to
the existing cooling water and chilled water systems. Figure 1 shows
the arrangement of the complete waste heat recovery system.
Waste Heat Boilers
The waste heat boilers recover most of the heat content of the column
overhead vapors and convert the energy Into a usable form. i.e. low
pressure steam. One waste heat boiler has been provided for each
column in order to preserve the Integrity of the self-contained con-
cept of the process equipment. Each waste heat boiler 1s a large U-
tube kettle type reboller with condensate sump on the tube side, and
mist eliminator head on the shell side. Table 1 summarizes the design
parameters.
The design maximum steam rate for each unit 1s 13,900 pounds/hour.
However, the actual steam rate which can be delivered by each waste
heat boiler 1s a function of the following:
(a) Percentage of column overheads flow bypassed directly to the
condenser for pressure control.
(b) Plant production rate.
(c) Column reflux ratio.
278
-------
Initially the control system was set up to bypass about 10% of the
vapor flow directly to the condenser so that precise column pressure
control could be maintained by varying the flow rate to the condenser.
This method protects the production process from column upsets but
also reduces the amount of recovered energy. After three months of
satisfactory operation it was found that the column could be protected
from adverse pressure fluctuations, even with 100% vapor flow to the
boiler, by controlling the steam pressure.
Steam Piping
The steam generated by the waste heat boilers is transported some 400
feet to the plant utility building in a 12 inch diameter pipe. Since
the rate of steam generation is a function of plant production, and at
maximum production the rate of steam generation is greater than the
absorption chiller demand, piping connections were made to the plant
low pressure steam piping system. Thus total plant low pressure steam
requirements can be balanced.
To ensure that the absorption chiller was placed in service without
production upsets a piping connection was made to the 300 psig utility
boilers. The chiller was (initially operated with steam supplied from
the utility boilers. After chiller operation was satisfactorily estab-
lished this connection was permanently closed.
Absorption Refrigeration Machine ^
The absorption chiller is a lithium bromide cycle York, model EMI 130,
machine. Design conditions are summarized in Table 2. The chiller
uses the combined output of both waste heat boilers as its source of
steam. The operating characteristics of the process column and the
absorption machine were analyzed to derive the relationship of refri-
geration tonnage to steam output.
T = (Si + S?) - 2620
13785
Where T •» Net refrigeration, tons
Sj = Steam rate, waste heat boiler #1, pph
S2 « Steam rate, waste heat boiler #2, pph
A graphical representation of this relationship is shown in Figure 2.
Operation of an absorption refrigeration machine is based on four factors.
1. A refrigerant, in this case water, which boils at a temperature
below that of the liquid,also water, being chilled. Since water
is used as a refrigerant and a coolant the refrigerant evaporator
must be at a pressure of 5 to 10 mm Hg for the boiling temperature
to be low enough to chill the coolant.
2. An absorbent, in this case lithium bromide, which possesses a great
affinity for the refrigerant. This affinity enables the refri-
gerant vapor to be converted to the liquid phase while still at
low pressure.
279
-------
3. A source of energy, steam from the waste heat boilers, to drive
an endothernrlc reaction.
4. A source of cooling water to remove heat from an exothermic
reaction.
A schematic representation of the cycle 1s shown 1n Figure 3.
Water 1s pumped over the chilled water coll and 1s evaporated by the
heat 1n the chilled water; this transfer of heat reduces the temperature
of the chilled water. Water vapor from the evaporator 1s absorbed by an
intermediate concentration of L1Br thus diluting the solution. Since
this is an exothermic reaction heat must be removed by cooling water.
The dilute solution Is pumped to the generator, via a heat exchanger,
where the water and L1Br are separated. This Is an endothermic reaction
so that heat, in the form of steam from the waste heat boilers, is re-
quired. The water vapor 1s condensed by cooling water and flows back
to the evaporator. The strong solution of lithium bromide flows through
a heat exchanger back to the absorber. The heat exchanger transfers
heat from the LiBr leaving the generator to the L1Br being pumped from
the absorber to improve cycle efficiency. A vacuum pump purges the
unit of non-condens1ble gases and a de-crystallization pipe prevents
the formation of solid LIBr.
COOLING AND CHILLED HATER
The existing cooling water pumps and towers were capable of handling
the machine cooling requirements. A cooling water recirculation
pump was installed to maintain a minimum cooling water inlet tempera-
ture of 75°F. If inlet water temperature drops below this value
crystallization of the LiBr solution may occur.
One additional chilled water pump was installed in the existing chilled
water sump. Existing pumps provide back up capability in case of
pump failure.
CONCLUSION
Although the thermodynamic performance of an absorption refrigeration
machine is not very high, three very real practical reasons supported
its selection;
1. Availability of an un-utilized machine from a sister plant.
This obviously reduced the capital cost of the installation.
2. Compatabi11ty with the production process. When tying together
production and utilities functions it is essential that con-
straints are not placed on the operation of production facilities.
The absorption machine has a large turn-down ratio, and an ability
to operate over a 10 psig range of steam supply pressure, there-*
fore it operation Imposes no constraints on production require-
ments .
280
-------
Compatability with existing utility systems. The absorption
chiller operates in parallel with the centrifugal chillers,
is located alongside them in the utility building and is con-
venient to cooling water, chilled water and electrical systems.
Although the machine has a complex thermodynamic cycle the
operating techniques are relatively simple. This latter point
is very Important to the successful implementation of an energy
management program. Improving energy efficiency or recovering
waste heat in an industrial plant generally adds equipment to
the plant facilities. Ensuring that complexities are minimized
is of paramount importance. Operating personnel must be well
trained, but successful operation is only assured if they are
not overwhelmed by complex equipment and controls.
281
-------
TABLE 1. WASTE HEAT BOILER DESIGN PARAMTERS
T.E.M.A. Designation
T.E.M.A. Class
Design Press./Temp.
Material of Constr.
Total design heat transfer rate
Condensing fluid,tubes
Condensing rate, max.
Boiling fluid, shell
Boiling Rate, Max.
Design solIds content 1n
boiler water
Design moisture content 1n
steam, max.
Operating Temp, shell In/out
Operating Press., shell
Operating temp., tubes In/out
Operating press., tubes
Maximum shell diameter
Straight length of tubes
32/50-214 BKU
C
75 ps1g/600°F
C.S. shell/304 S.S.
13,798,500 BTU/Hr.
Process Vapors
15,000 pounds/hour
Water
13,900 pounds/Hr.
10 ppm by wt.
1% by wt.
202°F/2520F
16 pslg
2890F/287°F
40 pslg
50 Inches
214 Inches
Tubes
TABLE 2. ABSORPTION CHILLER DESIGN PARAMETERS
Design capacity
Steam Consumption
Steam Pressure
Steam control valve pressure drop
Chilled water temp.,Inlet
Chilled water tempv outlet
Chilled water flow rate
Chilled water pressure drop
Condenser water temp., Inlet
Condenser water temp., outlet
Condenser water flow rate
Condenser pressure drop
Maximum chill water A T
Maximum steam -pressure
Minimum chill water flow
Cooling water temp.(minimum)
1,000 tons (1140 maximum)
18,500 Ib/hr. (21000 Ib/hr. max.)
12 psig
3 psi
52° F
42°F
2,400 GPM
35 ft. (15.2 psi)
850F
98.3°F
4,500 GPM
24 ft. (10.4 psi)
140F
15 psig
890 GPM
750F
282
-------
NJ
00
•
«jH£i?j£*
v tai'i **MM
j
**
1
t
|V
-------
FIGURE 2. STEAM CONSUMPTION VS NET REFRIGERATION
8 ::
it;
:3^
o
Ul
I Q"
., ..h ::;!:. j . i ,. 1 !| t..;,....:, -;..|.:.:,. i .;. i ...:,.;. ; ;. t .-/.,:.
. . i !.:.: . :., i :...: .., ;.. I: ,.;.,.:....; , i . ;. i . . . j ..,.::,; i i i/i , ... i
. j ! | ,. ; , I- 4 )., r.4 4..J j •; | . ...| .!.,..... j j j , .. . , , .,..:.. ... i i /•-; t | . I-
' f i ! i • ' i i i i i • • ! i I' i " i * ' i i . ' ,;,,.,;. Y . i i ..'
, j • • , < - • | ; " ' '!''*. ' " * I "l ' ,
:i his::: !!!:l!:J;!.!;:i-:!!-hi :•:!!:::!!
...LI..:
•i
-i. L J. L,;.
;..j.r.;-u.:.....-4 i ,.u.;
.:. j I...).., .. r_i..4. ;.;r .j j.., „.!...-.:. i
|..;......, .i...;.4.y ,.i._.-...-|..H..4..j
t-'-!-!
.... ----- . .
4. I.-.-4 -
:-l ^ 4-
'---: 1 r-
.. i_i.4-f : . rl -.
j.i^...i,44-f J-j
H-
-r -•
f t-;-Ji-i T TT !' : ' ' ' I
-~f-' f^-i-j-4-i.j..ii ,-. .
- 1 t +— • ' •'••' -i-r"}"' • i'; f-i-i-••-:••• • ••!-•-•!••. -J J !••*•! / \ < •
.:.!....... .i. 1.1...; T : .,... .;....,..1^ ...:.:... .1 +.,.,..t , ,-i/...i i .
...'_.]....'._ • i ' i j : 1 . ! " :.:...:•/ i l . .
I -
. ., ._.,.. .... *. i- .---»••) •-' •••-.-.-'••.- i-t_.-r-j-..1 -) :/, .a.».i.
, ,j ..^.^ ,....,.) .. 1.1. .l_,-_. . j i-^. ;.!_!.*_)._; .j...J.{,-(/^ .,-i-j-i (....a--.
: M 11:: i.rl 11 i:l"|T;'!in'ni'j ].; yf.b"|TjJ.'i);! 11
rh'.rf i 'iiH trziziJ"iit:nnTJ-^U'i4-i4"^t^ivi •"!-]'
.. ....; , . .4.-;-r..i.-H-^
.- -..». ..:.»._.j_L4.*4- ! -f -f-i-t
,.., 4 ,. i ..|_:_U4. 4- .4 ;--;
-
. }..;.,.;
...,.:, r..._.J_^.._i.L-_;._i.|-..1|_i., ,_,.:..: ...,
r • • i • J-' -/ :- • --i |»r- --<• «-f- I'" ' 1 !- i M • • '
'/, ••••'. 1 ' i ' l ' i
. j ;-;-* . 4, . i . ;T.-/' i-v, r ---.—-i •- r.i-:..|-4-.Uj i.; i : ..)...
jTjri.fiTi.'!""; ].T:/rj"±|"tJd^:;i:::'i:7.T,l.1'Hi'iii"": - iT.1
^ ;'.i!'} I H hy!_'' L'HilU '• i'i; !''•'' T' I j I i-;•' t
ljJl.U-1 MT..^—...;-..!..^
H.44.;.,.;,./ ..., ! ..._,.4-U-^-. I:.:.j.j_qxi .LLiTT
i : t_,_|.4., Hv...i.._!_._:-j.r-f-4_r._u,_i ..;_;—a.L,-4.' ,..; J.4 4-4-1-..
.-I ;-f4 ./ Jj-^ , f-r-.^ J..1.. 1.4.I....I.4 J .1 ..I.; ... .......
-.!.4,..., ./,.,... .:.!.,.. : . ., .|..j.|_i. T , .. . , ..,. , ...J J ] • . |.4 -'.".
..,.;...,.,,. r1, ..... ,.,..,., ,--—4.^, ,..|..J.|...}.jrLj., . 4J..
i -I ••!•«••' i ' i i • • i . •• '-1 . !•-;•' •*-*-•• . . , j ! i .,..4-1 -4- I ... ,..:..! .
1 '• I i-H ' • i : v -' ' ; ' " ' * ' • • •• i :
i; :o;
M,.,! .... t r,i<...!».. iiiiii.l... ii.
-t-j ;-^yj : • r-
—t . .1 . . , |........
.....J . ft .- i 4.^.4.
i 4 .J ., ,,;.;.;. *., 4. .
.... : ,.4.|. .......i .: .
H- : --.J-r-j-H 1
i" I t . T ' ' 1 :
:- : .j . , ;.(-(.}-4 i
. .4 . ... | | ,.j . !..
- - • -4 H-l I ••!-
r-H4!~rH-fl-
~'!.'\ •'! iili I",".!"
:::±:ij^4.:l:L!T'
,...;... ^..^l-j-i.^.
--i-i-i-i r.:.....-i-...r.
.: i
i-jtii;-ijj;
"T"'!'llT.Ti-.r
. 4..i .. !^ . - | J
. 1 i ; . ; j J 4--i
,.l .l.\ | | L.J-. j
I : . I
284
-------
FIGURE 3 SCHEMATIC OF ABSORPTION CYCLE
GENERATOR
J'l
SOLUTION — —
HEAT
EXCHANGER
CONDENSER
WATER
- - SOLUT'OX
SENERATCR
PU'.'P
- Rtl r-ic-ri,.M;T
ABSORBER
PUMP
T\ A A
REFRICERAMT
PUMP
VACUUM
PUMP PURGE
285
-------
THE SHERCO GREENHOUSE PROJECT: FROM
DEMONSTRATION TO COMMERCIAL USE OF CONDENSER
WASTE HEAT
G. C. Ashley
Northern States Power Company
Minneapolis, Minnesota U.S.A.
J. S. Hietala
Northern States Power Company
Minneapolis, Minnesota U.S.A.
R. V. Stansfield
Northern States Power Company
Minneapolis, Minnesota U.S.A.
ABSTRACT
Northern States Power Company's Sherburne County Plant produces both elec-
tricity and waste heat for commercial sale. As a result of nearly ten
years of research, development and demonstration, it is now technically and
economically feasible to utilize condenser reject heat for commercial
greenhouse heating in Minnesota. A three-year demonstration project
jointly funded and sponsored by Northern States Power Company, the Univer-
sity of Minnesota, and the U.S. Environmental Protection Agency has lead
the way for commercial adoption of the concept. Experience during the
demonstration project proved that condenser waste heat available at
approximately 85°F was suitable to maintain a greenhouse growing environ-
ment of 55 to 60°F when outside air temperatures fell as low as -43°F.
During the first year of operation of the pipeline system serving waste
heat to commercial greenhouse customers, an overall availability of service
of 97% was achieved. The savings in heating costs to commercial operators
using waste heat have amounted to nearly $5,000 an acre year compared to
conventionally heated greenhouses. While there are presently three commer-
cial operators with 1.7 acres in production, the experiences of these
operators have been sufficiently satisfactory that future expansion of
waste heat service at the Sherburne County Plant site is expected.
INTRODUCTION
Since 1970, Northern States Power Company (NSP) has been investigating
beneficial uses of the heat energy available in condenser cooling water at
electric generating plants. The concept of utilizing condenser cooling
water for agricultural heating applications, proposed initially by re-
searchers at Oak Ridge National Laboratory (ORNL) in the late 1960's, was
of particular interest to NSP; since new generating plants were then in
the planning stages that would utilize closed-cycle cooling systems.
Closed-cycle cooling system designs result in condenser outlet water temp-
eratures of generally not less than 85°F during the winter heating season.
286
-------
With this knowledge, NSP approached the University of Minnesota Agricul-
tural Experiment Station requesting their assistance to determine the best
potential application for utilizing condenser cooling waters in agricul-
ture. After extensive investigation into the field of waste heat utiliza-
tion and several visits to waste heat related projects, the University of
Minnesota advised that the concept of utilizing this waste heat for green-
house heating appeared promising.
Based on the University of Minnesota's recommendations and research efforts
with evaporative pad heating systems conducted at ORNL and also at TVA, NSP
decided to test the concept of greenhouse heating in a 2,000 ft2 commer-
cially operated greenhouse in Minneapolis during 1974 and 1975. Encour-
aging results from this research effort eventually lead to an award from
the U.S. Environmental Protection Agency (EPA) to demonstrate the technical
and economic feasibility of using power plant waste heat in a larger
scale, one-half acre greenhouse. The Sherco Greenhouse, as it i1^ now
known, was a cooperative effort of NSP, the University of Minnesota, and
EPA. The EPA grant was for a two-year period completed in May 1977; but
because of the success and the interest of EPA in commercial adoption of
the concept, the project was extended through June, 1978.
The Sherco Greenhouse was built in the Fall of 1975 and heated during 1975-
76 with simulated waste heat, since the power plant was still under con-
struction. In September 1976, the Sherco Greenhouse began using condenser
reject heat from Sherburne County (Sherco) Unit I condenser cooling water.
The Sherco Greenhouse was heated from the waste heat source in the 1976-77
heating season and in the 1977-78 heating season. Due to the success and
publicity of the Sherco Greenhouse, a commercial florist and a commercial
vegetable operator approached NSP in 1977 requesting suitable sites and
warm water service to heat commercially operated greenhouses. Therefore,
in the Summer of 1977 NSP decided to provide warm water from the Sherco
Plant on a commercial service basis to greenhouse operators. Presently
warm water service is being provided to three commercial operators with a
total acreage in production of about 1.7 acres. This paper describes only
the experience and economic projections of NSP in providing waste heat
service in the form of warm water to commercial greenhouse operators. A
more complete description of the Sherco Greenhouse Project can be found in
references 1 and 2.
EXPERIENCE WITH COMMERCIAL GREENHOUSES
As a result of the successful experience of the Sherco Greenhouse project,
NSP was approached by commercial operators in the Spring of 1977 and asked
to provide a site and warm water service to a one-acre commercial floral
operation and to a .2 acre commercial vegetable operation. Both commercial
facilities began construction in 1977 and warm water was first sold commer-
cially to the one-acre floral operation in November, 1977. The smaller,
.2 acre, vegetable operation did not require warm water service until
February, 1978.
287
-------
Description of Facilities
The one-acre commercial floral greenhouse is an arch roof, gutter connected
plastic covered structure like the Sherco Greenhouse; however, it has a bay
width of 17.5 feet and length of 144 feet with 17 gutter-connected bays to
provide a total enclosed area of 42,840 ft , or just less than one acre.
(See Figure l). The greenhouse was erected in mid-summer 1977 and initial-
ly planted partially in roses and chrysanthemums in August, 1977. Follow-
ing the first grow out of the mums, the greenhouse was fully cropped in
roses and is now producing only roses, the heating systems for the
commercial greenhouse facility include both soil heating and air heating
of a design similar to that used in the Sherco Greenhouse. The commercial
floral operation began taking warm water service for greenhouse heating in
November, 1977 and has now experienced an entire heating season utilizing
warm water from the operating power plant.
The commercial vegetable producer's facilities are two quonset style green-
houses that are covered with a double layer of polyethylene. Each unit
measures 30 feet wide by 150 feet long. The commercial vegetable producer
is using a hydroponic culture method and the first crops produced were
lettuce and spinach. The heating system employed in the vegetable opera-
tion is conceptually the same as that used in the Sherco Greenhouse in that
it utilizes dry fin tube heat exchangers and packaged air handling units.
However, the air distribution system is under the growing beds rather than
above the growing beds as in the Sherco Greenhouse and the commercial rose
range.
In order to serve warm water reliably, NSP decided in the Summer of 1977 to
make an interconnection between the two generating units so that waste heat
could be provided from either of the units. In order to do this, a tie
into the Unit II riser pipe near the cooling tower was made and a pipeline
was constructed to interconnect to the previously existing pipeline from
the Unit I cooling tower. It was deemed necessary to have an interconnec-
ted pipeline system from the two units to provide sufficiently reliable
warm water service to maintain commercial interest in condenser waste heat
for greenhouse heating. The pipeline installed in 1977 was an 18" diam-
eter cast iron pipe that was uninsulated and buried to a nominal depth of
about 5 feet. The 18" diameter pipeline was intentionally over-sized to
allow for future expansion of commercial warm water service. The pipeline
chosen is capable of delivering sufficient flow to heat up to 14 acres of
greenhouses at the Sherco site.
Harm Water Service Operating Experience
Warm water service first became commercially available for greenhouse heat-
ing on November 2, 1977. From that time until the end of May, 1978 the
warm water service system operated with an overall availability of 96.6%.
On a weekly basis during the first heating season, there were only four
weeks out of thirty when warm water was unavailable for part of the time.
Unavailable service to the commercial greenhouse operations was primarily
288
-------
due to pipeline failures. There were two pipeline failures experienced
during the heating season; one was the failure of a service line from the
main pipeline into the commercial floral operation; the other failure was
the failure of a main 12" diameter plastic pipeline. In the former case,
a four-day outage occurred; in the latter case a 26-hour outage occurred.
The only time that a two-unit generator outage made warm water unavailable
was during May when Unit I was scheduled out for a month for annual main-
tenance; and Unit II was simultaneously scheduled out for weekend mainte-
nance. Thus, for a period of 56 hours, warm water was unavailable due to
a simultaneous two-unit outage. Fortunately, this occurred in May when
the heating requirements of the greenhouses were not great.
While the warm water was available 96.6% of the time, it was not always
available at temperatures high enough for maintaining the greenhouse
environment desired by the commercial operators. This was due primarily
to reduction in electrical load on the two-unit station over night and
also, to some extent, resulted from frozen coal problems experienced at the
power plant. Figure 2 indicates the effect of unit loading on warm water
temperature. It can be seen that as the generator unit load varies from
about 200 MW to 700 MW, which is the normal load range, the condenser out-
let water temperature can vary over a range of about 20°F. Due to the
nature of the electrical system loads of NSP; a unit like Sherco, designed
to load follow, reduces load over night which reduces the condenser outlet
water temperature. Even so, over the first heating season, the waste heat
was available at the condenser outlet above 85°F for 83% of the time.
Figure 3 indicates the measured warm water temperature distribution for
all of January, 1978. As indicated, 73% of the time the temperature was
greater than 85°F; at least entering the pipeline. After taking into
account pipeline heat losses for delivery to the customer, the temperature
available at the customer point of use was reduced 3°F.
Supplemental Heat Energy
The commercial greenhouse operations utilizing warm water heating also
have a backup heating system in the event of pipeline system failure or a
two-unit generating station outage. While the backup heating system is
primarily designed for standby and emergency heating, it can also be used
to supplement heat energy available from the warm water source. The back-
up heating systems used by the commercial operators at the Sherco site are
propane fired unit heaters. Propane was selected as the fuel over oil,
partly because propane would be required in any event for C02 generation;
and, partly because the cost of propane fired heating equipment is less
than the cost of oil-fired heating equipment.
Even though the warm water was available to the customers 97% of the time
during the first commercial heating season, the one-acre floral range
consumed 8,700 gallons of propane. While about 1,700 gallons of this pro-
pane consumption was for C02 generation, the balance was used to provide
standby heat during pipeline failures and a two-unit outage; as well as
supplemental heat at times when the warm water temperature was too low to
289
-------
maintain the desired greenhouse temperature. The total propane consumed
represented less than 10% of the annual heat requirements of the green-
house and that used only for supplemental heat amounted to 5%. It is
expected that as the pipeline system is operated in the future, the operat-
ing availability of the pipeline system will improve and the supplemental
and standby heat energy requirements will be diminished.
Electrical Energy Consumption
Because of the relatively low temperature difference between the heating
source and the desired space temperature, a significant amount of elec-
trical energy is required by the fans and pumps to produce the required
heat transfer rate. For example, in the case of the one-acre commercial
operation, the heating system fans have a total motor horsepower of 85,
while the liquid pumping requires 10 Hp. Thus, a total of 95 Hp or approx-
imately 70 KW of electrical load results from the full load operation of
the heating system. Because the heating fans cycle on and off by thermo-
static control, the actual daily utilization of energy is significantly
less than the full load operating requirement. The total measured elec-
trical consumption in the one-acre commercial floral range during the
first heating season was 237,000 KWH. It is estimated that about 85% of
this electrical consumption was directly attributable to the operation of
the warm water heating system. Based on NSP electrical rates, the cost of
electric energy to drive the warm water heating system is one of the most
significant operating cost items. The others, of course, are the cost of
delivering the warm water to the customer and the cost of supplemental
heating fuel. Presently the cost of electric energy to operate a warm
water heating system for a typical acre is about $7,000 per year; while
the cost to deliver warm water is about $8,000 per acre-year. Though the
electric energy requirement to operate the warm water heating system has a
significant cost impact, compared to a heat pump system extracting heat
from the same low temperature water source, the greenhouse direct heat
transfer heating system is more than three times as efficient, utilizing
one-third as much electric energy as a heat pump system.
Economic Feasibility
The economic feasibility of utilizing condenser reject heat for greenhouse
heating is dependent upon many variables. The most important of these are
site specific variables such as distance from the waste heat source to the
greenhouse acreage served, climatic conditions, electric rate structures,
land cost, and distance to market. Because of these highly site specific
variables, generalizations about the economic feasibility of utilizing
waste heat for greenhouse heating are of limited value.
However, it has been found that at least at the Sherco site, conditions
are such that the concept appears to be economically feasible. For example,
the most recent cost estimates to install the necessary pipeline services
to serve a 14 acre greenhouse complex located within roughly one-half mile
of the waste heat source, reveal a pipeline investment of about $35-40,000
290
-------
per acre of greenhouse served. Depending on cost of money and other
considerations, the investments in pipelines generate a requirement for a
return of about $6-8,000 per acre per year to recover the investment.
In addition to the capital cost of pipelines, there is an operating cost
associated with pumping the warm water from the waste heat source to the
user; and, there is a cost associated with the overall operation and main-
tenance of the pipeline network. At the present time, NSP estimates that
the total variable operating cost will run roughly U/thousand gallons of
water delivered to the user. At this rate, a typical one-acre greenhouse
operation in Minnesota would realize an operating cost of $1,500 per acre-
year.
The approximate total capital and operating cost associated with delivery
of waste heat at the Sherco site is now $8,000 an acre-year. NSP has
adopted a philosophy of providing waste heat on a cost-of-service basis
with a return on investment that is equivalent to the rate of return on
investment for other utility services provided by the Company. That is to
say, there is no energy value placed on waste heat. No portion of the
fuel cost of the power station of any capital investment normally associa-
ted with a power plant are allocated to the cost of serving waste heat.
The cost of serving waste heat is based on incremental investment and
incremental operating costs only. This cost-of-service pricing philosophy
may or may not be unique in the utility industry.
Since a major portion of the cost of serving waste heat is related to
investments in pipelines, it is possible to offer long-term fixed-price
contracts to pay for the fixed cost associated with the waste heat service.
So in much the same way as sewer and water are assessed to individual users
in municipal water systems, the commercial greenhouse operators are essen-
tially assessed their portion of the cost of the pipeline network and then
allowed to pay for it over a period of 20 years. The variable cost of
operating, maintaining, and pumping the water in the pipeline are subject
to yearly price changes based on actual operating experience. The effect
of the long-term fixed-price contract for waste heat energy is to make the
waste heat system more and more attractive in the future as other energy
costs escalate.
In addition to the investments incurred by the utility to supply warm water
to a commercial greenhouse, the greenhouse operator incurs a significant
investment in the warm water heating system; the coils, fans, and pumps re-
quired to transfer sufficient heat to the greenhouse. For the Minnesota
climatic conditions, the estimated installed cost of a complete one-acre
warm water heating system including controls, electrical work, and a back-
up heating system"is about $85,000 per acre. This compares to convention-
al greenhouse heating systems that, on the same basis, might cost about
$50,000 per acre.
The comparative operating costs of utilizing waste heat versus conventional
heating are indicated in'Table 1. The figures presented are based on a one
291
-------
acre, double plastic greenhouse located at the Sherburne County Plant site.
The total operating costs on a per acre year basis for the waste heat
system in 1978 are estimated to be $17,000, which includes $8,000 for
waste heat, $7,000 for electricity, and $2,000 for standby propane. A
typical conventional greenhouse using No. 2 fuel oil has an estimated heat-
ing cost of $28,000 per year.
There is an apparent savings of $11,000 per acre year by utilizing the
waste heat system in 1978 dollars. But, the big advantage of the waste
heat system is more apparent in the future as fuel costs escalate. The
waste heat system would realize an escalation in the cost of propane and
electricity; but, the waste heat costs would not increase very much due to
the fact that the pipeline investment had been made 20 years previously.
In 1998, at a compound escalation of 6% for all utilities and variable
operating costs, the waste heat system shows a cost advantage of $51,000
per acre-year. These comparative operating costs are projections based in
part on operating experience and anticipated future operating experience at
the Sherco plant site. The actual operating costs for all utilities for
the one acre commercial floral range totaled 37<£/f't^ for the period Novem-
ber-May. This compared to 65
-------
station is so great as to impact local property tax mill rates to the
extent that property taxes tend to be low in areas in which power plants
are located. This can have a beneficial financial effect on the commercial
greenhouse operation.
While it appears economically feasible to provide waste heat service to
greenhouse customers and it appears reasonable for the greenhouse customers
to make additional investments in heating systems to save in future heat
energy costs, it is still too early to predict whether this trend will hold
true at many power plant sites in the U.S. or whether this will prove to be
a relatively unique situation.
CONCLUSIONS
Based on the experiences over a three-year period with the Sherco Green-
house, it is apparent that condenser reject heat, available at the rela-
tively low temperature of 85°F, is a wholly satisfactory heat energy source
for greenhouse heating in northern climates. The concept of utilizing
this waste warm water has been accepted and adopted by commercial green-
house operators in Minnesota. Though the commercial acreage in production
now is only 1.7 acres, depending upon future fuel cost escalation and mar-
ket conditions, an expected initial expansion of up to 14 acres at the
Sherburne County Plant site is reasonable. Whether the concept demonstra-
ted can be expanded throughout the U.S. will depend on a myriad of site
specific and market related variables, but at this time the prognosis is
good.
REFERENCES
1. Ashley, G. C. and Hietala, J. S. "The Sherco Greenhouse: A Demonstra-
tion of the Beneficial Use of Waste Heat". Proceedings of the Waste Heat
Management and Utilization Conference. May 9-11, 1977.
2. Boyd, L. L. et. al. "Use of Waste Heat from Electric Generating
Plants for Greenhouse Heating". Proceedings of the ASAE 1977 Winter Meet-
ing. Paper No. 77-4531. December 13-16, 1977.
293
-------
TABLE I
COMPARATIVE OPERATING COSTS OF WASTE HEAT
VERSUS CONVENTIONAL HEATING
Fuel Cost
Standby Propane
Electricity
Waste Heat Cost
Total Per Acre
Per Year
- 1978 Costs -
Waste Heat Conventional
- 1998 Costs-
Waste Heat Conventional
2,000
7,000
8,000
27 ,000
1,000
6,000
22,000
10,000
86,000
3,000
17,000
28,000
38,000
89,000
Basis: #2 Fuel Oil @ $.384/gal
Propane Cost @ $.43/gal
Electricity Cost @ 3.5<£/kwh
Fig. 1 Commercial Floral Range, Vegetable Range, and Sherco
Greenhouse in Foreground. Power Plant in Background.
294
-------
u.
UJ
cc
5
cr
UJ
a.
2
UJ
(K
UJ
1
a:
<
120-
110-
100-
90-
80-
70-
60-
50-
T
T
T
100 200 300 400 500
GENERATOR UNIT LOAD ~ MW
Fig. Z TYPICAL WARM WATER TEMPERATURE
VARIATION WITH UNIT LOAD
600
700
50
60
70
80 90
TEMPERATURE fn|r}
100
110
130
Fig. 3 CONDENSER OUTLET WATER TEMPERATURE DISTRIBUTION
JANUARY, 1978
COMPOSITE OF UNIT I AND UNIT II
295
-------
ANALYSIS OF ECONOMIC AND BIOLOGICAL FACTORS OF
WASTE HEAT AQUACULTURE
J. S. Suffern and M. Olszewski
Oak Ridge National Laboratory*
Oak Ridge, Tennessee 37830
ABSTRACT
A waste heat aquaculture system using extensive culture techniques is
currently under investigation at the Oak Ridge National Laboratory. The
system uses nutrients in waste water streams to grow algae and zooplankton
which provide feed for fish and clams. A tilapia polyculture association
and the freshwater clam Corbicula are the animals cultured in the system.
The investigations detailed in this study have been performed to determine
the economic and biological feasibility of the system and examine energy
utilization. A net energy analysis identified the energy saving potential
for the system. This analysis included all energy costs (both direct and
indirect) associated with building and operating the system.
The economic study indicated that fish production costs of $0.55/kg
($0.24/lb) were possible. This cost, however, depends upon the fish pro-
duction rate and food conversion efficiency and could rise to as much as
$1.65/kg ($0.75/lb).
The biological studies have examined growth relationships and production
potential of the cultured organisms. In the laboratory, growth-tempera-
ture optima have been defined (32°C, with good growth rates between 26 and
34°C) for tilapia hybrids. In addition, growth rate acceleration experi-
ments have been carried out, developing techniques which yield 40% higher
growth rates in experimental fish as compared with controls. Using cage
culture techniques in sewage oxidation ponds, we have obtained production
estimates in excess of 50,000 kg/ha/yr (50,000 Ib/acre/yr).
*
The energy utilization study indicated that, when all energy costs are
included, fish from the aquaculture system may require only 35% of the net
energy now required for fish products from the ocean. However, the energy
requirements also depend on system parameters and could be as large as the
energy required for ocean caught products.
*Research sponsored by the Department of Energy under contract with Union
Carbide Corporation.
296
-------
The analyses indicate that the system is economically feasible. They also
indicate that significant energy savings are possible if waste heat aqua-
culture products replace ocean caught products.
INTRODUCTION
A waste-heat utilization strategy must meet basic feasibility criteria in
order to be considered viable. [1] At ORNL, we have been carrying out
multidisciplinary studies to evaluate the economic and technological feasi-
bility of aquaculture as a waste heat utilization strategy, and have reached
some very interesting conclusions. This paper presents the results of our
studies to date, outlining the conceptual design we have found to be opti-
mal, and the biological, economic and energetic analyses we have carried
out.
SYSTEM DESIGN
The basic precept of current aquaculture research at ORNL is that produc-
tion systems should be designed to produce low-cost protein. Our basic
thinking is that, although protein supplies are adequate in the United
States at present, a large part of the world is already protein poor, and
conditions of surplus are not likely to last much longer. The design of
such a system is a radical departure from the capital-intensive mono-
culture techniques in vogue in the United States today. First, for reasons
of system efficiency and stability, polyculture, that is, growing several
species at-once, should be used. Second, instead of using expensive high-
protein feed, waste streams, such as sewage, animal wastes, or food proces-
sing wastes, could provide nutrient input.
Based on these ideas, a preliminary system concept was developed and sub-
jected to extensive engineering and economic analyses. The current concep-
tual design (Fig. 1) is a sequential pond system, involving a succession
of ponds dedicated to different cultures: algae, mixed species of fish,
crustaceans, or rooted vegetation, into which controlled amounts of heated
water and waste-stream nutrients are fed. Although details may change as
a result of future research, the basic concept is expected to remain.
BIOLOGICAL STUDIES
Several different species arrays may be used in the system, [2], and of
these, one based on tilapia appears most promising. They are excellent
food fish and are readily marketable. Second, they have been cultured
extensively around the world and enjoy a reputation of having few disease
problems. Third, as a result of worldwide culture activities, a good data
base exists on many aspects of their biology. Fourth, literature reports
on several species indicate growth-temperature optima from 25 to 35 C.
297
-------
This is substantially above that recorded for temperate-latitude fish, a
factor that suits the genus well to waste heat systems where relatively
high temperatures might be expected. Also, all tilapia species h,ive about
the same body shape and size, allowing some standardization of harvest and
processing techniques.
Tilipia have one major drawback as a food culture organism: they become
reproductively active at a small (<100 g) size. This could result in a
crop of fish too small to market as human food, and is a problem which
must be solved if the fish are to be practical for use in a commercial
operation. At present, there are two ways around this dilemma. The first
is sex reversal by hormone injection. This technique involves the adminis-
tration of androgenic hormones to young fish, accomplishing sex reversal
with about 95 to 99% effectiveness. The second method is hybridization.
Several of the species, when crossed, produce monosex hybrids. Done care-
fully, this technique is 100% effective in some crosses.
We chose to use a monosex hybrid, the all-male cross of $ Tilapia hor-
norum by? Tilapia mossambica, in our initial experimental program. This
hybrid is known to be a plankton feeder, and appeared to be a promising
candidate to fill that feeding niche in a polyculture array.
Our experimental efforts have been designed to answer questions which are
fundamental to commercialization of the waste nutrient-waste heat aqua-
culture concept. A primary bit of needed information is the thermal regime
necessary to maintain maximum growth rates in cultured organisms. In order
to assess the optimum growth temperature of Tilapia mossambica ff x Tilapia
hornorum 6 hybrids, fish between 45 and 55 mm standard length were divided
into sixteen groups of ten fish each. Two groups of ten fish each were
placed in each of eight glass aquaria, 45 x 89 x 38 cm, and separated by
wire mesh partitions. Each aquarium was maintained at a different tempera-
ture: 20, 23, 26, 28, 30, 32, 34, or 37°C, with a continuous flow-through
of about 3.5 liters/minute. Purina trout chow (#2) was fed to the fish ad
libidum and the tanks were vacuum-cleaned daily.
At the initiation of the experiment, tilapia were acclimated to test
temperatures for one week. Laboratory lighting throughout the experimental
period was maintained at a seasonal (summer) photoperiod; activated by an
outdoor photocell. All fish were weighed and measured biweekly after
anesthesia in tricane methanesulfonate. The experiment ran for 39 days.
Growth at all temperatures was approximately linear throughout the experi-
ment (Fig. 2). The largest increase in biomass occurred with those fish
maintained at 32°C with the fish at 26, 28, 30 and 34°C also showing sub-
stantial growth. Even though the fish at 32°C produced the most biomass,
the confidence intervals on percent increase in weight per day (Fig. 3)
indicate that there is not a statistically significant difference between
the growth rates at 26, 28, 30, 32 and 34°C. Statistical significance may
not be the most important criterion for selecting aquaculture temperatures,
however, when the trend of the results is so clear.
298
-------
In order to test the utility of our waste-stream nutrient concept, the
Oak Ridge National Laboratory (ORNL) sewage oxidation ponds were used as
an experimental facility. It was felt that they were an excellent simula-
tion of a waste-nutrient-loaded pond system, and would produce results
similar to a system devoted to aquaculture.
The two ORNL sewage ponds are sequential. Raw sewage flows from a pre-
liminary mechanical chopper into the first pond, resides there for about
six days, then flows into the second pond, is retained for another six days,
then is discharged (receiving chlorination during the process) to a stream.
Table 1 lists some salient characteristics of this pond system during the
experimental period. The sewage ponds are rectangular, and are lined with
polyvinyl chloride. To facilitate the oxidation process each pond is
equipped with large bubble aerators, designed and installed to produce a
short turnover time. This aeration system is very effective; in several
months of observation, dissolved oxygen readings did not drop below 5.5 ppm.
During the experimental period (July and August) average daily temperatures
were 26-34°C, found to be favorable in the laboratory growth-temperature
experiments. Due to the rapid turnover caused by the aeration system, the
ponds were isothermal.
3
The tilapia were stocked in the ponds at a density of 53/m in floating
cages made of 6 mm mesh Vexar netting and wood frames, and were weighed and
measured biweekly after anesthesia in tricane methanesulfonate.
The plankton assemblages in the two ponds were strikingly different. The
first pond, into which raw sewage was discharged from the preliminary treat-
ment facility, was almost a monoculture of E_ug_l_ena. for the duration of the
experiment. This condition appears to be the "natural" equilibrium situa-
tion. The second pond was a fairly stable zooplankton culture, dominated
by a large strain of Daphnia pulex. Numbers of this organism declined
during the peak temperatures in mid-July, but became re-established as
temperature began to fall in August, again dominating the plankton assem-
blage.
Change in mean weight of the tilapia is shown in Figure 4. As can be seen
by inspection, average weight gain was essentially linear over time; about
7.5 g/week (~ 1.07 g/day). Although the differences are not statistically
significant, it appears that the fish in pond 2 (the one dominated by
zooplankton) grew faster than those in pond 1. Lodge, et al. [3] have
recorded growth rates between 1.7 and 2.0 g/day with tilapia weighing 60-
150 g in the same system.
The production potential of a waste-nutrient system was calculated from
the results, and is quite large. From these experimental data, we calcu-
late production rates of 397 g/m2/wk or 3970 kg/ha/wk or 206,440 kg/ha/yr.
This production estimate must be viewed with caution, as it is based on
extrapolation from experimental results, not the operation of a full-scale
system. Several assumptions underlie the estimate, some of which make it
299
-------
excessive. The first is that the entire pond surface is assumed to be
covered with cages. This is unreasonable, as it allows no room for harvest-
ing equipment (boats). Assuming one half of the pond surface to be open
allows room for boats and reduces the production estimate by a factor of
two. We also assumed that the fishes' daily growth increment remained con-
stant up to marketable size (300 g). Our experimental data indicate no
decrease in growth increment as fish size increases up to around 150 grams,
but we feel that such a decrease is likely in larger fish, and that produc-
tion estimates should be reduced by a factor of two to account for this.
We have assumed that the system operates near the optimum growth tempera-
ture for the fish year-round, a situation which is feasible in temperate
latitudes only with the addition of heat for much of the year, making our
production estimates applicable only to waste heat aquaculture systems (any
other source of heat, excepting geothermal, would be too expensive).
It is probably not necessary to reduce production estimates to account for
density-dependent effects if pond design and stocking densities are similar
to those discussed in this paper. Density-dependent effects may alter
production rates in two basic ways. First, fish crowding may lower grov/th
rates. This is probably not the case at the stocking densities postulated
here; experiments carried out by the senior author (Suffern, unpublished)
indicated no difference in growth rate between fish stocked in the sewage
ponds at 10/m^ and those stocked at 53/m3. A second density dependent
effect on production is the effect of fish feeding on food organisms. If
consumption by the fish exceeds food organism production rates, fish produc-
tion rates will decline. Assuming Golueke, et al.'s [4] reported produc-
tion rates for algae in hypereutrophic, intermittently aerated ponds, a
conversion effeciency from algal biomass to fish biomass of 10%, rapid
mixing, half the surface area covered with cages, and a pond depth of about
2.5 meters, fish production between 25,000 and 50,000 kg/ha/yr could be
supported. This is probably a low estimate of supportable production
because Goleuke's estimates were based on ponds which were aerated only
during the winter as versus all-year round aeration in our proposed sys-
tem. Also, the support estimate does not account for aufwuchs utilization
by the fish, a component of their ration in this cage culture system.
Given all these considerations we feel it is reasonable to expect produc-
tion on the order of 50,000 kg/ha/yr from a waste-heat waste-nutrient aqua-
culture system.
We carried out growth acceleration experiments to explore ways of boosting
production rates in waste-heat aquaculture systems. There is evidence
that changes in growth rates in some animals can be caused by exposure to
moderate doses of hard radiation. [5] This effect was tested on tilapia
hybrids.
Twenty-five young tilapia hybrids (mean weight 2,7 g, mean standard length
4.60 cm) were exposed to 500 rads of gamma radiation in a cobalt^0 source.
From previous experiments at ORNL with channel catfish (Ictalurus puncta-
tus) (Blaylock, unpublished) there was evidence that this dosage might
enhance growth rates. The irradiated tilapia were placed in a flow-through
300
-------
tank kept at 30 C. A control group of 25 fish was kept in an adjacent tank
with identical flow and temperature conditions. Both groups were fed ad
libidum with Purina #2 trout chow.
The results of this experiment are shown in Figure 5. The mean weights of
the fish for the control and treated groups are tabulated in Table 1. The
data clearly show that up to week 8 there was no difference in the growth
patterns of the two groups. After that time, however, one observes a
possible pattern with the irradiated group gaining faster than the control
group.
Using the Subhatme d-statistic [6] we find the mean weights significantly
different at the 0.05 level at fourteen weeks; an indication of real diver-
gence. This trend continues through week 52, at which time the irradiated
fish average about 40% heavier than controls.
At the present time, detailed autopsies are being performed on the fish in
an attempt to ascertain the mechanism whereby gamma radiation causes growth
rate changes. Our hypothesis is that the administered radiation caused
gonadal atrophy resulting in growth rate changes parallel to, for example,
those seen in cattle which have been made steers.
Several aspects of this system's biology remain to be examined. For
example, heavy metal accumulation, pathogen transmission and effluent
characteristics are all areas of potential concern to commercialization.
Fitzgerald and Suffern [7] have examined heavy metal uptake in sewage pond
raised fish. While results are still being analyzed, preliminary indica-
tions are that heavy metal accumulation may not be a major problem.
Lodge, et al. [3] have examined nutrient processing and effluent charac-
teristics of a sequential pond cage culture system, and find dramatic
reductions in nutrient levels in successive ponds. Pathogen transmission,
a subject of concern with the FDA, has yet to be explored adequately.
ECONOMIC ANALYSIS
The previous discussion of the system's operation indicates that the fish
and clam production systems can be considered to be independent systems.
Results from the preliminary analysis [8] of the system indicated that the
clam system was the major income center for a combined system. However,
the experimental efforts described above have indicated that the fish
production system has the potential to match the clam system profitability.
Therefore, this report will be limited to an economic feasibility analysis
of the fish production system.
System Size
Analysis of the system was performed for an 0.4 ha (1 acre) fish-pond. To
determine if an auxiliary plankton production pond was required it was
301
-------
necessary to estimate fish production (and corresponding food intake) and
plankton growth rates.
Fish production rates for tilapia polyculture systems of this type have
not been firmly established. Annual yields of 7378 kg/ha (6500 Ib/acre)
have been reported for tilapia grown in sewage enriched ponds in Africa.
[9] The investigations at ORNL suggest that an annual yield of up to 85,
125 kg/ha (75,000 Ib/acre) is possible for cage culture in aerated ponds
if sufficient algae is provided. [10] Because of these fish production
uncertainties, the system was analyzed over this range.
Food conversion rates are similarly unknown. However, it is generally
accepted that conversion efficiencies of 10% (wet weight to wet weight)
are typical when moving from one trophic level to the next. Assuming
the algal feed is approximately 803Twater and using a 10% conversion
efficiency yields a food conversion ratio of 2:1 [2 kg (dry weight) algae
converted to 1 kg (wet weight) of fish]. Since there are a number of
uncertainties in arriving at this ratio, the system was analyzed using
food conversion ratios of 2.5:1 and 5:1.
From values reported in the literature [4, 11] it appeared that average
algal production rates of 10 to 20 g/m2-day (90 to 180 Ib/acre-day) are
achievable in ponds using animal manures as their source of nutrients.
Since it was assumed that sufficient nutrients would be available to sus-
tain this production, the system design was based on an average daily
algal production of 15vgr (dr wgt)/m2 (135 Ib/acre).
Based on the above assumptions, the required algal pond sizes were computed
for an 0.4 ha (1 acre) fish growth pond. These results are shown in
Table 2.
The capital cost for construction of ponds were based on constructing them
by "ising bulldozers to push up levees. Costs for this type of construction
range from $2625 to $3250 per hectare. ($1050 to $1300 per acre) [12, 13,
14]. Therefore, a cost of $3000/ha ($1200/acre) was used.
The fish production rates obtained in our studies were obtained using cage
culture techniques in an aerated pond. Thus, equipment costs for the fish
pond system included aerators and cages in addition to the usual fish
handling equipment.
It was assumed that cylindrical cages constructed from Conwed fabric were
used. The cost for these cages, including a flotation system, was esti-
mated at 10<£/ft3. [15] For design purposes it was assumed that half the
surface area of the pond was taken up by cages. The aerator costs were
based on data available in commercial fish farming journals. [16] Assuming
a shipping cost of 5% of the purchase price, a cost of $300 was used for
a (350 gpm) aerator. The power required by the aerator is 0.8 kw.
302
-------
The number of aerators required depend upon the volume of the fish pond and
the pond circulation rate desired. The pond circulation rate is defined
as the average time required for the entire pond volume to be pumped
through the aerators. Based on an 0.4 ha (1 acre) surface area and a 2 m
(6 ft) pond depth the aerator requirements over a range of pond circula-
tion rates is given in Table 3.
A summary of the fish pond capital costs is given in Table 4. It was
assumed that the ponds were constructed to utilize gravity flow so large
pumps for emptying ponds were not necessary. The equipment costs include
fish handling equipment and ohter miscellaneous equipment needs.
The capital costs for the auxiliary algal pond are also given in Table 4.
To prevent light transmission problems and maintain high algal productivity
these ponds are only 0.6 m (2 ft.) deep. Small circulation pumps are also
provided to ensure mixing of the water.
Since the algal ponds are not as deep as the fish ponds, they are less ex-
pensive to construct. Based on crawfish pond (usually 1 m deep) con-
struction costs [17] experience, algal pond construction costs were
estimated at $1600/ha ($400/acre).
A plastic heat exchanger was .designed to transfer 8790 kw (30 x 10 Btu/hr)
to the ponds. In sizing the heat exchanger an overall heat transfer con-
ductance of 369 W/m2-°C (65 Btu/hr-ft2-°F) and a log mean temperature
difference of 5.6°C (10°F) were used. The heat exchanger material is a
polypropelene copolymer extruded in the form of a honeycomb with the flow
channels in the interior of the plate. The material cost is $1.90/m2
($0.24/ft2). The fabricated heat exchange cost was estimated to be
$8.07/m2 ($0.75/ft2) of heat exchange area.
The heat loss for an acre of pond area varies with climatic conditions and
temperature of the pond. Experimental data from Oregon [18] indicates
heat losses of 5.8 MW (20 x 106) Btu/hr) from an acre algal pond main-
tained at 35°C (95°F). Accounting for cooler climates a heat loss of
8.8 MW (30 x 106 Btu/ha) per acre was used for this study. This results
in a capital cost of $10,800 ha ($27,000/acre) of pond surface area for
the heat exchanger.
These capital costs were annualized using a fixed charge rate (FCR) of_
15%. The fixed charge rate includes minimum return on investment, capital
item depreciation, taxes, insurance, project lifetime and a number of
other items. Using a FCR of 15% for a project with a 20-year lifetime
yields a return on investment of 8%. For the purposes of this study, tax
credit considerations were not included in the FCR. Including an invest-
ment tax credit of 10% (which is typical) yields a return on investment of
about 14%. Thus, neglecting the tax credit results in a conservative
economic analysis. A summary of the annual fixed charges for each pond
is given in Table 5.
303
-------
A summary of the operating costs for the system components is given in
Table 6.
The total annual fish production costs were then obtained by summing the
annual fixed charges and the annual operating costs from Tables 5 and 6.
These costs were then applied to the system sizes given in Table 2 to
yield the total annual fish production cost. The production cost was
divided by the fish production rate to yield the unit production cost.
These results are plotted in Figures 6 and 7, and indicate production
costs between $0.55/kg ($0.24/lb) and $1.65/kg ($0.75/lb); depending on
production rates and food conversion efficiencies.
NET ENERGY ANALYSIS
The energy required, directly and indirectly, to produce fish by the
aquaculture method described has been estimated and compared with the
energy used in harvesting an equal amount of fish by conventional fishing.
The electrical energy required for operating the pumps and aerators is
estimated directly from the projected water flow rates and pressure heads.
Other direct and indirect energy requirements, such as those for pond
construction, heat exchanger fabrication and installation, and labor, are
calculated using tabulated net energy intensities. The energy intensity,
defined as the energy required to make a dollar's worth of products or
services, has been calculated from economic census data and tabulated for
each major economic sector. [19] For example, in 1967 a dollar's worth
of concrete products required 117.2 MJ (117.234 Btu). This includes all
energy expended from the time when the constituents were in the ground as
minerals through the final fabrication. Energy embodied in machinery or
facilities used, in supplies consumed, in transportation, etc., is in-
cluded. The energy required for each portion of the aquaculture installa-
tion has been calculated using the energy intensity for the broad economic
sector which corresponds best to it.
A summary of the energy requirements determined in this manner is given
in Table 7. The net energy of the feed stream and the discharge water
stream are assumed to be free inputs and are not included. In addition,
each acre of pond requires 1908 GJ (1908 million Btu) for the heat
exchanger. The use of fingerling stocking fish has been assumed; an
additional cost of 1675 KJ/kg for stocked fish.
The energy requirements in Table 7 are used with the pond areas in Table 2
to calculate the total energy requirements for fish production. The
results are plotted in Figure 8. The energy needed to obtain an equal
amount of fish by conventional fishing, estimated at 52.9 GJ/kg (24,000
Btu/lb), is also shown. The results plotted in Figure 8 give the ratio of
the net energy required by extensive aquaculture to that required by con-
ventional fishing. These numbers suggest that, for large enough produc-
tion rates, aquaculture may produce fish at considerably lower net energy
expenditures than fishing.
304
-------
REFERENCES
1. Gross, A. C. and M. C. Cordaro. 1977. Waste Heat Utilization from
a Utility Standpoint. The Problem of Implementation. In: The
Proceedings of the Conference cm Haste Heat Management and Utilization,
S. S. Lee and S. Sengupta, eds. p9A-29-39.
2. Beall, S. E., C. C. Coutant, M. Olszewski and 0. S. Suffern. 1977.
Energy from Cooling Water. Industrial Water Engineering 14(6): 8-14.
3. Lodge, D. M., J. S. Suffern and S. M. Adams. In preparation. Growth
of all-male hybrid Tilapia in sewage pond cage culture. Oak Ridge
National Laboratory, Oak Ridge, Tennessee.
4. Golueke, C. G., W. J. Oswald, G. L. Duan, C. E. Rixcord and S. Scher.
1973. Photosynthetic reclamation of agricultural solid and liquid
wastes. Environmental Protection Agency Ecological Research Series,
Report #EPA-R3-73-031. 83 pp.
5. Donaldson, L. and K. Bonham. 1964. Effects of low-level chronic
irradiation of chinook and coho salmon eggs and alevins. Trans. Am.
Fish. Soc. 93(4): 333-341.
6. Finney, D. J. 1971. Statistical Method in Biological Assay. Griffin,
London. 668 pp.
7. Fitzgerald, C. and J. S. Suffern. In preparation. Trace metal
analysis of a sewage-aquaculture system. Oak Ridge National Labora-
tory, Oak Ridge, Tennessee.
8. Olszewski, M. 1977. The Potential Us_e of Power Plant Reject Heat
in Commercial Aquaculture, ORNL/TM-5663". Oak Ridge National Labora-
tory, Oak Ridge, Tennessee.
9. Anon. 1974. Sewage for Tilapia, Fish Farming Int. 2: 108-109.
10. Suffern, J. S., et al. 1978. Growth of Monosex Hybrid Tilapia in
the Laboratory and Sewage Oxidation Ponds. In: Symposium cm Cul-
ture of Exotic Fishes, R. 0. Smitherman, W. L. Shelton and J. H.
Grover, eds. Fish Culture Section, American Fisheries Society.
Pp. 65-82.
11. Boersma, L., et al. 1974. Arnmal_ Waste Co_nyersion Systems Based on
Thermal Discharges, Special Report 416, Agricultural Experimental
Station, Oregon State University.
12. Robert Snow Means Company, Building Construction Cost Data 1978,
p. 19.
305
-------
13. Dupree, H. 1978. Fish Farm Experimental Station, Stuttgart
Arkansas, personal communication with J. S. Suffern, Oak Ridge
National Laboratory, January 1978.
14. Crawford, K. 1978. Personal communication with J. S. Suffern,
Oak Ridge National Laboratory, January 1978.
15. Sipe, M. 1978. Personal communication with J. S. Suffern, Oak
Ridge National Laboratory, January 1978.
16. Commercial Fish Farmer and Aquaculture News 4(4), p. 21 (May 1978).
17. Bean, R. 1978. Louisiana State University, personal communication
with J. S. Suffern, Oak Ridge National Laboratory, January 1978.
18. Gasper, E., et al. 1975. Utilization of Waste Heat in a Biological
System . The Conversion of Swine Manure to Single Cell Protein
and Methane Gas, presented at the American Society of Agricultural
Engineers, Chicago, Illinois, December 15-18, 1975.
19. Bullard, C. W., P. S. Penner, and D. A. Rilati. 1976. Energy
Analysis: Handbook for Combining Process and Input-Output Analysis,
ERDA-77-61.
306
-------
Table 1
Height Gain of Irradiated (500 rads 60Co gamma)
and Control Tilapia
Irradiated Control
Week Mean Weight, Std. Dev. Mean Weight, Std. Dev.
0
2
4
6
8
14
20
48
52
2.70 +
5.19 +
12.60 +
19.29 +_
22.38 +
59.48 +
72.57 +
218.55 +
238.06 +
0.54
0.9
2.54
4.59
5.46
8.79
10.18
17.34
22.27
2.78 +
3.94 +
11.48 +
15.28 +
21.46 +
51.80 +
58.32 j^
144.91 +
161.47 +
.36
.87
2.59
3.25
5.50
12.99
15.92
22.71
22.01
Table 2
Algal Pond Sizes for 0.4 ha (1 Acre) Fish Pond
Food Conversion Ratio
Fish Production Rate •
(Kg/yr) 5:1 2.5:1
Algal pond size (ha)
2,948
4,536 0.004
11,340 0.6168 0.1098
22,680 1.639 0.616
34,020 2.6629 1.129
307
-------
Table 3
Pond Aerator Requirements
Circulation Rate
(hours)
24
12
6
3
1.5
0.75
Req. Aerator Flow Rate
(gpm)
1,350
2,700
5,400
10,800
21,600
43,200
# Aerators
4
8
16
31
62
124
Cost
($)
1,200
2,400
4,800
9,300
18,600
37,200
Power
(Kw hr/day)
77.2
154.4
308.8
598.3
1196.6
2393.2
Table 4
Pond Capital Cost Summary
Item
Fish Pond Cost
($/acre)
Algal Pond Cost
($/acre)
Pond Construction
Equipment
Cages
Aerators
Misc. Equipment (Pumps)
Subtotal
Contingency (9 25%)
Total
1,200
850
13,100
See Table 2
200
15,350 + Aerators
3,840 + (0.25) Aerator
19,190 + (1.25) Aerator
400
100
200
700
175
875
308
-------
Table 5
Summary of Annual Fixed Charges for 0.4 ha Pond
System
Fish Pond 12 hour Aerator Circulation
Fish Pond 3 hour Aerator Circulation
Algal Pond
Heat Exchanger
Annual Fixed Charges
$ 3,338
4,631
130
4,050
Table 6
Summary of Annual Operating Costs
Operating Cost ($ 1 acre-year)
Item
Management & Labor
Materials
Electricity
@ 2.5£/km hr
Stocking
Total
Fish Pond Algal Pond
3 hour 12 hour
750
100
5,460
($0.03)
6,312
+ stocking
750 100
100 100
1,410 60
(Stocking
rate)
2,260
+ stocking 260
Heat Ex.
150
150
309
-------
Table 7
Net Energy Requirements for 0.4 ha Ponds
Item
Energy Required (GJ)
3 hour
Fish Pond
12 hour
Fish Pond
Algal Pond
Capital Items
Construction
Pumps
Equipment
Cages
Total Capital
Operating Items
Electricity
Labor
Total Operating
27
5
385
1589
2006
14741
20
14761
27
5
127
1589
1748
3831
20
3851
27
5
33
30
20
50
310
-------
WATER FLOW
ORNL-DWG77-13474
VIA HEAT EXCHANGES
*
PONDI
ALGAE, BACTERIA,
200PLANKTON
HEATED WATER
*
POND II
FISH (SEVERAL
DIFFERENT KINDS)
1
L
+ t
POND III
ALGAE, BACTERIA,
ZOOPLANKTON (AGAIN)
-
POND IV
CLAMS
CRAYFISH
-
POND V
ROOTED
AQUATIC
VEGETATION
I
WASTE STREAM
NUTRIENTS
(SEWAGE, CATTLEWASTES,
PROCESS WASTES HIGH IN
N&P, ETC.) + DILUENT
Figure 1. Conceptual design of the ORNL waste-heat aquaculture system.
-------
ORNL-DWG 77-18577
720
640 -
560
480 -
400 -
C/)
IT)
Q 320
co
240 -
160
80 -
0
0
3 4
TIME (weeks)
Figure 2. Growth of tilapia hybrids at different temperatures in the
laboratory.
312
-------
ORNL-DWG 77-20817
OUTFLOW SIDE OF TANK
INTAKE SIDE OF TANK
ERROR BARS: 95% C.I
20 22
26 28 30 32 34
TEMPERATURE (°C)
36 38 40
Figure 3. Tilapia growth versus
temperature after 39 days.
313
-------
ORNL-DWG 77-20818
100
90
80
70
60
50
40
30
20
1O
0
POND 1
POND 2
COMBINED
(ERROR BARS:± 1
STANDARD DEVIATION)
STOCKING DENSITY : 53/m3
I
468
TIME (weeks)
10 12
Figure 4. Growth of tilapia hybrids in sewage pond cage culture.
314
-------
ORNL-DWG 78-20758
tn
270
240
210
180
LU
cc
UJ
150
120
90
60
30
0
o IRRADIATED (500 rods OF 60Co GAMMA)
• CONTROLS
(ERROR BARS'* 1 STANDARD DEVIATION)
I
12
1
42
48
18 24 30 36
TIME (weeks)
Figure 5. Growth of irradiated and control tilapia hybrids in the
laboratory.
54
-------
240
200
ORNL OWG 78-2O795
-co-
160
c/)
o
0 120
o
ID
O
O
cc
o_
80
40
UNLINED AERATED FISH POND
ALGAL PRODUCTION : 135 Ib/acre-day
FCR: 15%
FOOD CONVERSION: 2.5'1
12
I
25,000 50,000
FISH PRODUCTION (Ib/acre-yr)
Figure 6. Production costs for fish assuming a
2.5:1 food conversion ratio.
75,000
-------
.o
\
-co-
2.80
2.40
2.00
ORNL-DWG 78-20796
h-
en
O
O
§ 1.60
£ 1-20
Q
O
or
°- 0.80
0.40
UNLINED, AERATED FISH POND
ALGAL PRODUCTION: 135lb/acre-day
FCR: I57o
FOOD CONVERSION: 5M
Figure 7.
25000 50000
FISH PRODUCTION (Ib/acre-yr)
Production costs for fish assuming a
5:1 conversion ratio.
75000
-------
ORNL DWG 78-20797
oo
0
Figure 8.
28,000 55,000 85.00O
PRODUCTION RATE (Kg/ha-yr)
Ratio of energy required for aquaculture to that required by
ocean fishing.
-------
A QUALITATIVE/QUANTITATIVE PROCEDURE FOR
ASSESSING THE BIOLOGICAL EFFECTS OF WASTE
HEAT ON ECONOMICALLY IMPORTANT POPULATIONS*
John M. Thomas
Ecosystems Department
Pacific Northwest Laboratory
Operated by Battelle Memorial Institute
Richland, WA 99352
ABSTRACT
Recent research suggests statistical procedures which can be
used to detect changes in populations of aquatic biota due
to the operation of nuclear power plants. However, the
effect of such changes on future populations must be assessed,
I have devised a procedure in which a probable model, not
necessarily a mathematical model coded for a computer, for
economically important populations is derived and modified
in conjunction with field monitoring studies, beginning in
the preoperational period.
The advantages of my proposal are fourfold:
•;
a) Field studies will be terminated (i.e.,
good quality field work is rewarded)
after an agreed time.
b) Imaginative scientists, currently doing
routine monitoring, can design and
carry out studies which investigate key
questions in population ecology.
c) Both statisticians and modelers have
the opportunity to suggest the directions
of, or institute, new field work.
d) The adversary mode used to establish
and assess the adequacy of current
monitoring programs (i.e., involving
litigation) is bypassed.
*Based on work done for the U. S. Nuclear Regulatory Com-
mission and the U. S. Department of Energy under Contract
EY-76-C-06-1830.
319 JMT
-------
Various features of the plan are amplified in this paper and
the specific portions discussed include: 1) the amount of
population decline which might be ecologically acceptable, 2)
the precision needed for measuring crucial parameters, 3) the
interpersonal problems which may result during frequent
meetings of a "negotiating team" over many years, and 4)
equilizing monitoring costs among power plant sites.
INTRODUCTION
Recent research (Eberhardt [1], Thomas [2,3], Thomas et al.,
[4], Gore et al., [5,6], Murarka et al., [7], Adams et al.,
[8], McKenzie [9], McKenzie et al., [10], and McCaughran
[11]) has suggested several possible methods for field designs,
statistical analysis, and useful field techniques necessary
for the conduct and assessment of biological effects derived
from monitoring studies at nuclear power plants. Thomas and
Thomas et al., [3,4] suggested that sampling all trophic
levels diluted efforts to assess population effects on
economically important species and devised a scheme whereby
monitoring studies would be stopped after a finite time.
The purpose of this paper is to amplify portions of the plan
and to discuss certain aspects in more detail. Additional
features discussed include possible interpersonal conflicts
resulting from the use of a negotiating committee charged to
direct studies during the course of a long term (up to eight
years) program and cost equitability among monitoring studies.
RECOMMENDED CHANGES FOR CURRENT MONITORING PROGRAMS
Currently, intervenor charges of population impacts on
economically important fish and shellfish has resulted in the
construction and refinement of computer simulations of
mathematical models for most life stages of species of
concern. These models are impossible to defend (as correct)
on a scientific basis partly because: 1) they employ estimates
of parameters and either no data or data poorly measured data
perhaps obtained at some other site, usually ecologically
dissimilar, and 2) a generally limited understanding of
ecosystems. To circumvent these difficulties, I advocate
that a representative model, not necessarily a model constructed
to run on a computer, be derived and modified in conjunction
with field studies beginning in the preoperational period.
The results and understandings gained from the model can then
be used to design additional field studies to obtain either
information identified as needed or better estimates of
previously studied parameters (i.e., number currently in
320
-------
various life stages, age specific mortalities, etc.). Since
the basic plan is designed to replace current monitoring
programs, I envision that the resources expended now can be
devoted to implementing the current ideas. If monitoring
requirements under NEPA are changed by regulatory agencies in
the future (i.e., less monitoring effort), then my proposal
must be reevaluated in terms of cost. One idea might be to
require "pool" funding by nuclear and nonnuclear power plant
operators.
Clearly, the studies outlined above will be expensive so most
monitoring efforts on lower trophic levels should be aban-
doned. If circumstances arise whereby monitoring lower trophic
levels is mandated, then only outfall areas should be con-
sidered (with appropriate matched control stations [9]). The
latter action is not without justification (Thomas and Thomas
et al., [3,4]). In the second paper [4] we have proposed that
such efforts must be terminated in a finite time both as an
incentive for plant owners and to ensure continuing and
adequate financial support. In addition, I propose that
expenditures should be of about the same magnitude at most
power plant sites. Power plants sited in locations where no
economically important species reside probably should con-
tribute to studies at other sites to attain approximate cost
sharing. Clearly, the public pays (via higher power bills)
for studies as currently conducted as well as under this
plan. Since the population effects issue is universal, it
seems reasonable to expect the public to share the costs.
This is necessary because the answer to the question, "What
are the effects on future biological populations?" will not
likely result from studies at any single site but instead
will take many years of sustained effort at many sites
utilizing diverse ecological studies and new field techniques.
This is due, in part, to current lack of adequate field
procedures to obtain data on ecological parameters in situ
and because we cannot usually separate power plant effects
from other life span insults (i.e., oil pollution, changed
fishing pressure, carrying capacities), and sometimes our
inability to define the size and range of spawning populations.
Thus, we must develop improved understandings of population
ecology and much better field procedures. Given enough time,
both objectives can emanate from my suggested monitoring
program because: 1) power plant operators are rewarded by
terminating monitoring in a known time, 2) ecological scien-
tists are freed from routine monitoring currently required,
and 3) quantitative scientists will have a continuing input
into the design and day-to-day operation of field programs.
321
-------
THE SUGGESTED MONITORING SCHEME
Figure 1 contains an expanded version of the plan first
proposed in Thomas et al [4]. While the example is oriented
toward a fishery where several years may elapse before progeny
such as salmon and herring return to spawn; an extension to
shellfish or other species could readily be made. Studies
may be necessary to approximate numbers and assess parameters
relevant to population dynamics of all life stages of the
species of concern when a projection of future population
numbers is to be made. Thus, while I do not advocate studies
of lower trophic levels in general, these will be necessary
when immature forms of the species of concern are a portion
of such trophic levels. In addition studies of other lower
trophic level species may be necessary because of their
particular influence(s) on the population of concern. The
logical plan shown in Figure 1, which includes negotiation to
define the content of the monitoring, discrete field studies
and the number of years of study, constitutes the basis for a
stopping rule for power plant ecological impact studies.
The first feature deserving special attention is the pre-
liminary assessment during the first preoperational year.
Since this is a "guessing stage," considerable credance
should be given to local experts, such as fish and game
personnel and local ecologists in order to guesstimate future
effects. In addition, some attention needs to be given to
emotional concerns of persons considered something less than
experts (i.e., fishermen) as a means to anticipate possible
future intervenor questions. For example, if catch effort
data are available for prior years, it should be pointed out
that wide fluctuations have occurred in the past, for which
we may or may not have any clues as to causes. In constrast,
the local experts should be used to provide insight for
a word model (box and arrow diagram?) of the population of
concern; useful input may also come from fishermen. Some
examples of guesstimates needed are included in Figure 1.
During the second preoperational year the full negotiating
committee should be established. The exact number of members
and their professional training may be partially reflective
of a site specific problem. However, the suggested membership
in Figure 1 should perhaps be a minimum. The use of con-
sultants should reflect the committee's particular needs
based on their collective technical inadequacies. However,
they should have the power to employ specialists as needed,
independent of any power company influence. As soon as
organizational matters are decided preliminary limits, both
statistical and biological, that the initial studies can be
expected to attain should be negotiated. The reality of the
322
-------
expectations will be assessed and reevaluated during the
third preoperational year. Further, during the third pre-
operational year both the size of the population change(s)
and the statistical precision desired for essential parameters
needed to project changes should be promulgated, even though
these may still be partly educated guesses.
Current experience in ecology indicates that coefficients of
variation (standard deviation divided by the mean expressed
as a percentage) can vary from 20 to over 100 percent and
that some variables may be nearly impossible to measure in
situ (i.e., natural mortality of fish eggs or larvae) except
in a crude way. Because of these factors, I propose that
even crude estimates of the size of confidence intervals for
key parameters and the size of population change that the
monitoring programs might detect should be deferred to the
third preoperational year. Thus, second year efforts should
be designed to assess what is possible. Studies instituted
during the third preoperational-year should mimick those for
operational years. Clearly, if adults of the species of
concern migrate and return more than one year later, addi-
tional preoperational years of study may be required. In
such cases, the financial resources should be - spread out and
some ancillary studies considered essential may have to be
cancelled. Thus, the size of the difference to be detected
between operational and preoperational years will be wider
and possible site specific effects may be more difficult to
assess. However, since monitoring results from many sites
collected during many years will be necessary to address the
population effects issue, the data, techniques'and experience
will make a contribution to the overall assessment.
I advocate the cost spreading above, so all sites incur costs
of a similar amount, only in cases where power plants are not
sited in an ecologically unsuitable location. Otherwise,
particular power companies should be assessed the increased
cost when additional preoperational data are needed. One
function of the negotiating committee might be to make the
initial assessment of site suitability.
In the absence of any experience with the plan outlined in
Figure 1, a discussion of how it should function in opera-
tional years may be premature but the features given can be
an adequate starting point. However, some realistic limits
for the size of population changes to be detected and confi-
dence bounds for key parameters must be an integral feature
and need to be based on preoperational studies. Since there
may not be enough adequate data from preoperational studies,
I suggest that the population projection made using some
representation of the system (i.e., model) should result in
323
-------
an estimate at least one order of magnitude below that presumed
to be ecologically detrimental. In light of the fact that the
confidence with which we can expect to measure field para-
meters is dependent on many factors, any suggested a priori
recommendations about their possible precision may be pre-
mature. However, the negotiating committee should insist on
replicated studies which at least correspond to state-of-the-
art, and in some cases, should require research which will
illow improving the reliability of field measurements. Finally,
che committee must define the length of the operational
study. For lack of other guidance, I suggest one complete
reproductive cycle for migratory species with a long repro-
ductive cycle, of three to four years or more, and a three
year period for others with shorter cycles. Additional years
should be required if inadequate studies are conducted and
ameliorative schemes maybe instituted if damage to the
population is judged unacceptable.
It should be noted that while at times the negotiating
committee may operate in somewhat of an adversary mode, that
decisions about the overall conduct of the program and
important questions needing further research, will be the
result of a consensus of views. Currently, one often hears
utility, intervenor, and regulator alike state, "If we could
just get away from the lawyers, lock the people with genuine
concerns and the scientists in a room, we could hammer out
(negotiate) a reasonable program to assess impact." The
proposal depicted in Figure 1 goes a long way toward attaining
that goal.
POTENTIAL PROBLEMS OF A FUNCTIONING NEGOTIATING COMMITTEE
It is essential that the negotiating committee function in a
manner whereby decisions taken have not emanated from divisive,
as opposed to "considered", deliberations. Since the members
will have diverse backgrounds, biases, and expectations and
because at times they will need to operate in a collective
bargaining atmosphere, strong differences of opinion can and
will develop. Figure 2 illustrates some "people problems"
such a committee may encounter and offers the suggestion that
a return to the goalsetting and mutual understanding phases
may be necessary since the period of interpersonal involvement
may be up to eight years. To aid in recognizing phases where
harmful disagreement is either beginning or has become divi-
sive, I suggest that a disinterested leader or committee
chairman be appointed to direct committee business. The sole
function of this individual will be to recognize the points
of possible friction shown in Figure 2 which could delay
attaining the goal of finishing a quality study in Y-years.
324
-------
Gore, K. L., Thomas, J. M., Kannberg, L. D., Watson, D.
G. (1977c). Evaluation of nuclear power plant environ-
mental impact prediction, based on monitoring programs.
BNWL-2152, NRC-1, Battelle, Pacific Northwest Labora-
tories, Richland, Washington. 31 pp.
6. Gore, K. L. , Thomas, J. M. and Watson, D- G. (1978).
Quantitative evaluation of environmental impact assess-
ment based on aquatic monitoring programs at three
nuclear power plants. J. Environ. Mgt. 6, in press.
7. Murarka, I. P. Policastro, A. J., Ferrante, J. G.,
Daniels, E. W. and Marmer, G. P. (1976). An evaluation
of environmental data relating to selected nuclear power
• plant sites. ANL/EIS-1 (Kewaunee), ANL/EIS-2 (Quad-
Cities), ANL/EIS-3 (Duane Arnold), ANL/EIS-4 (Three Mile
Island), ANL/EIS-5 (Zion), ANL/EIS-6 (Prairie Island),
ANL/EIS-7 (Nine Mile Point). Argonne National Laboratory,
Argonne, IL.
8. Adams, S. M., Cunningham, P. A., Gray, D. D. , Kumar, K.
D. and Witten, A. J. (1977). A critical evaluation of
the nonradiological environmental technical specifica-
tions. ORNL/NUREG/TM-70 (Surrey Power Plant Units 1 and
2); ORNL/NUREG/TM-71 (Peach Bottom Atomic Power Station
Units 2 and 3); ORNL/NUREG/TM-72 (San Onofre Nuclear
Generating Station Unit 1). Oak Ridge National Laboratory,
Oak Ridge, Tennessee.
9. McKenzie, D. H. (1978). A review of statistical analysis
methods for benthic data from monitoring programs at
nuclear power plants. Proceedings of this conference.
10. McKenzie, D. H., Kannberg, L. D., Gore, K. L., Arnold,
E. M., and Watson, D. G. (1977). Design and analysis of
an aquatic monitoring program at nuclear power plants.
PNL-2423, NRC-1. Battelle, Pacific Northwest Labora-
tories, Richland, Washington. 125 pp.
11. McCaughran, D. A. (1977) . The quality of inferences
concerning the effects of nuclear power plants on the
environment. In Proceedings of the Conference Assessing
Effects of Power-Plant Induced Mortality on Fish Popula-
tions, Gatlinburg, Tennessee, May 3-6, 1977. (W. Van
Winkle, ed.) pp. 229-242. Pergamon Press.
325
-------
ACKNOWLEDGMENTS
Dr. Dan McKenzie contribured several technical suggestions
for improving the manuscript and Mrs. Judy Helbling provided
editorial comments.
REFERENCES
1. Eberhardt, L. L. (1976). Quantitative ecology and
impact assessment. J. Environ. Mgt. 4,27-70.
2. Thomas, J. M. and Eberhardt, L. L. (1976) . Ecological
impact assessment. In Proceedings of the Conference to
Computer Support of Environmental Science and Analysis.
Albuquerque, New Mexico, July 9-11, 1976. (S. Fernbach
and H. M. Schwartz eds.) pp. 181-197. CONF-750706, U.S.
Energy Research and Development Administration, Washington,
D.C.
3. Thomas, J. M. (1977). Factors to consider in monitoring
programs suggested by statistical analysis of available
data. In Proceedings of the Conference for Assessing
Effects of Power-Plant-Induced Mortality on Fish Pop-
ulations, Gatlinburg, Tennessee, May 3-6, 1977. (W. Van
Winkle, ed.) pp. 243-255. Pergamon Press.
4. Thomas, J. M. (1978). Statistical methods used to
assess biological impact at nuclear power plants. J.
Environ. Mgt. 6, in press.
s~"
( Gore, K. L., Thomas, J. M. Kannberg, L. D., and Watson,
D. G. (1976). Evaluation of Monticello Nuclear Power
Plant, environmental impact prediction, based on moni-
toring programs, BNWL-2150, NRC-1, Battelle, Pacific
Northwest Laboratories, Richland, Washington. 127 pp.
Gore, K. L., Thomas, J. M., Kannberg, L. D., Mahaffey,
J. A. and Watson, D. G. (1977a). Evaluation of Haddam
Neck (Connecticut-Yankee) Nuclear Power Plant, environ-
mental impact prediction, based on monitoring programs.
BNWL-215-, NRC-1, Battelle, Pacific Northwest Labora-
tories, Richland, Washington, 181 pp.
Gore, K. L., Thomas, J. M., Kannberg, L. D. and Watson,
D. G. (1977b). Evaluation of Millstone Nuclear Power
Plant, environmental impact prediction, based on mon-
itoring programs. BNWL-2152, NRC-1, Battelle, Pacific
Northwest Laboratories, Richland, Washington. 120 pp.
326
-------
FIGURE 1. A PROCEDURE FOR A QUANTITATIVE/QUALITATIVE
ASSESSMENT OF THE EFFECTS OF WASTE HEAT ON A Fl SHERY.
FIRST
PREOPERATIONAL
YEAR
A WORD MODEL IS PROPOSED
ASSESS AVAILABLE DATA
HYPOTHESIZE POSSIBLE EFFECTS
PRFIIMINARY ASSESSMENT
• POWER COMPANY
• GOVERNMENT REGULATORS
• IOCALEXPE.RTIECOLOGIST?!
BEST GUESSES
• DO WE SUSPECT COMPENSATION?
RESEARCH?
• WHAT SIZE POPULATION CHANGES CAN WE CONCEIVABIY MEASURE?
NEW RESEARCH ON FIELD METHODS?
• WHAT ARE THE TOTALLY UNKNOWN FACTORS?
OCEAN MORTAIITY?
RESEARCH?
• CAN DATA FROM OTHER LOCALES BE USED?
FUND RESEARCH El SEWHERE?
SECOND
PREOPERATIONAL
YEAR
SECOND ASStSSMFNT
• POWER COMPANY
• GOVERNMENT REGUIAIORS
• IOCAL EXPERTS
ESTABLISH NEGOTIATING COMMITTEE
POWER COMPANY
BIOLOGIST (Si
GOVERNMENT REGUIATORS
POSSIBLE INTERVENORS
PUBLIC
QUANTITATIVE ECOLOGIST
PANEL OF CONSULTANTS
NEGOTIATION
SET SIZE OF COPULATION CHANGE TO DETECT
SET STATISTICAL 1IMITS FOR KEY VARIABLES
• AGE STRUCTURE
• LARVAL MORTAL ITY
• FECUNDITY
• ETC.
PRELIMINARY
•INSTITUTE PRELIMINARY RESEARCH & MONITORING
•REFINE MODEL (NOT NECESSARILY CODED FOR A COMPUTER!
THIRD
PREOPERATIONAL
YEAR
EXPAND MONITORING STUDIES
DEFINE NEEDED PARAMETERS & ADDITIONAL RESEARCH
SET TIME LIMITS FOR OPERATIONAL STUDY
RENEGOTIATE STATISTICAL & BIOLOGICAL LIMITS
LIST STRATEGY FOR UNANTICIPATED EVENTS (I.E. WEATHER!
Y-OPERATIONAL
YEARS
• CONTINUE FIELD STUDIES
• EVALUATE RESULTS
"RUN MODEL"
NEGOTIATING COMMITTEE
• REVISE FIEID STUDIES OR RESEARCH
• LEAVE "AS IS"
• SUGGEST NEW STUDIES
1
YES
FOR Y OPERATIONAL YEARS
• ARE PROJECTED POPULATION EFFECTS WITHIN AGREED
UPON BOUNDS & BEIOW ECOIOGICAI.LY DETRIMENTAL
LEVELS? ONE ORDER Of MAGNITUDE'
• ARE FIELD STUDIES PROGRESSING?
« ARE RE Sf ARCH GOAI S RUNG Ml I?
I NO
IAVE Y-YEARS PASSED
1
YES
1
STOP
PUBIISH RESULTS ~
NEGOTIATING COMMITTEE
• REVISE GOAI S
• REDESIGN PLANS
• PENALIZE POWER CO. Y-MORE YEARS
ANY AMELIORATIVE PLANS?
327
-------
FIGURE 2. POSSIBLE RESOLUTION OF "PEOPLE PROBLEMS" ENCOUNTERED
BY A NEGOTIATING COMMITTEE FUNCTIONING FOR Y-YEARS
to
oo
PLAN FAILS
WILLINGNESS TO SHARE INFORMATION BY MEMBERS
EXPECTATION OF VARIOUS MEMBERS
BIASES OF VARIOUS MEMBERS
EDUCATIONAL LEVEL 0"F MEMBERS
CREDIBILITY OF CONSULTANTS
Y-YEARS
STUDIES
TERMINATE
UNDERSTANDING OF EACH MEMBERS ROLE
VARIOUS DEGREES OF COMMITTMENT
• STABLE FIRST FEW YEARS
DISAGREEMENT-PROBLEMS
(RESOLVABLE WITHOUT GREAT FRICTION)
• TOTAL DISAGREEMENT-FRICTION
[HARDTO RESOLVE.PRESSURE (OUTSIDE PRESSURE?)]
• STUDIES GO POORLY
• COMMITTEE BEGINS TO LOSE MEMBERS
• AMBIGUITY-UNCERTAINTY
-------
A REVIEW OF STATISTICAL ANALYSIS METHODS
FOR BENTHIC DATA FROM MONITORING PROGRAMS
AT NUCLEAR POWER PLANTS*
D. H. McKenzie
Ecosystems Department
Pacific Northwest Laboratory
Battelle Memorial Institute
Richland, WA 99352
United States of America
ABSTRACT
Recent reviews of the monitoring programs at nine
attempted to evaluate the aquatic impacts thought
charges. These reviews have provided an opportun
some of the statistical approaches used a_ p_os_teri
such programs. This paper addresses three major"
be answered when attempting to analyze these data
to assess impact? What statistical analysis tech
and What is an appropriate experimental unit and
rates? Each of these areas is discussed and eval
data collected at the nine power plant sites.
nuclear power plants have
to be due to thermal dis-
ity to compare and evaluate
P.ri to analyze data from
areas or questions that rnusl
: What data do we analyze
nique should we employ?
how do we select error
uated based on the benthic
base
In general, considerably more taxonomic detail was present in the data
than was utilized in the statistical analyses and recommendations are
developed that will increase the cost effectiveness in these areas. The
advantages and disadvantages of three statistical approaches, analysis of
variance, regression, and time series, are discussed and illustrated with
examples from the nine power plant sites. The selection of statistical error
rates and experimental units is discussed and arguments presentee! for estab-
lishing appropriate values.
INTRODUCTION
The primary objective of an environmental monitoring program can be simply
stated as the measurement of the environmental changes induced by the opera-
tion of the power plant. A recently concluded effort by the staff of three
laboratories [Argorme National Laboratory (ANL), Oak Ridge National Labora-
tory (CRNL), and Pacific Northwest Laboratory (PNL)] examined nine nuclear
power plant monitoring programs and evaluated the degree to which the above
objective was met. The ANL group evaluated the environmental monitoring data
for Zion [1], Prairie Island [2], and Nine Mile Point [3] Nuclear Power
Planes. The ORNL group examined the results'of the Surry [4], Peach Bottom
Based on work done for the U.S. Nuclear Regulatory Commission
329
-------
[5], and San Onofre [6] Nuclear Power Plants' environmental monitoring pro-
grams. The monitoring programs conducted at Monti cello [7], Haddam Neck [8],
and Millstone [9] Nuclear Power Plants formed the basis for the PNL evalua-
tions. These reviews provide an opportunity to compare and evaluate the
statistical methods and approaches used to analyze data from such programs.
While it is tempting to ascribe state-of-the-art precedent to the techniques
used in these reviews, the temptation should be resisted due to the a^
posteriori nature of these efforts. The techniques chosen by each of the
groups was primarily dictated by the availability or lack of data. In
general, the three groups concluded that the ability of the monitoring pro-
gram to measure impact is dependent upon many things, but probably of great-
est importance is the design or planning approach taken when establishing
the monitoring program.
This paper examines the results of the three laboratories and addresses what
would be appropriate if the analysis was not constrained by the a posteriori
approach. Three major areas or questions are discussed that should be
answered during the design or planning stages for a monitoring program. The
first question is: What data are appropriate for the assessment of impact?
The second important area is the selection of the statistical analysis
technique(s). The third area concerns the appropriate experimental unit and
selection of error rates. Each of these areas is discussed and evaluated
based on the benthic data presented in the nine power plants reviewed by the
three laboratories.
Data Collected
There is considerable uncertainty or at least lack of consistency concerning
what abiotic and biotic data are to be recorded as exemplified by the nine
monitoring programs. The same uncertainty goes unresolved through the
analysis and interpretation stages. Uniformity in the entire assessment pro-
gram will increase the effectiveness of the overall efforts.
The approach to resolving the issue should be to look at the information
needed to assess impact and not what can be measured. In general , the
reviewed studies appear to have been based on the "measure everything"
approach. However, the analysis and interpretations can be based on only a
small subset of that data.
The biotic characteristics that were examined in the nine impact evaluations
were the numbers of benthic organisms per area in one or more of the follow-
ing categories: total numbers, phylum, class, order, genus, species and
total biomass. No rationale was presented to support the selection of the
specific level of taxonomic identification chosen. In addition, the results
of the individual analyses of the various categories were given equal value
in the interpretation of impact. Thus, there does not appear to be a con-
sensus among the reviewers on the level of taxa identification needed for
impact assessment (or perhaps the Approach was to try everything and hope
that something would "turn-up"). However, the reviewers were constrained
by the a £oj^ej2_o_ri_ nature of their studies and had to use what was available,
not necessarily what was needed, or most desirable for the impact assessment.
330
-------
The following scheme is recommended for establishing the level of taxononn'c
identification, analysis and interpretation. First, it should be realized
that identification to the species level of all benthic organisms is unwar-
ranted. The maximum number of analyses of b&nthic categories was seven arid
only five were fully discussed and interpreted [8]. The majority of the
nine benthic program analyses examined total organisms and one or two sub-
divisions. Thus, it appears that the authors from the three laboratories
considered this level of analysis to be appropriate for the assessment of
impact on the benthos. While it is impossible to conclude that what the
three laboratories used is ideal, it does appear to be consistent with the
general state-of-the-art of impact assessment for benthic communities.
However, a priori establishment of an overly simplified or restrictive number
of taxonomic units for recording the data is probably unwise. This number
will be site specific and will depend on the complexity of the benthic corn-
muni ty, its economic values, and the amount and detail of the information
that exists during the design phase.
Although site specific characteristics may indicate a need for adjustments
in the general rule, it is recommended that not more than five to seven sub-
divisions of the benthic taxa be recorded, analyzed and interpreted. This
appears to be consistent with the general state-of-the-art of impact assess-
ment for benthic communities.
Another biotic parameter that was considered in three of the nine studies
was the biomass of the benthic organisms. The Peach Bottom benthic data
analysis and impact assessment were conducted solely on biomass data [5].
Changes in methods restricted the usefulness of Surry benthic biomass data
to comparisons among stations within individual years [4]. Gore et al. [8]
concluded on the basis of a correlation analysis of the Haddam Neck benthic
data that counts, wet and dry weights contained much the same information
and analyzed only the count data. Thus, it appears that count data is to be
preferred over biomass data, and that when both are available, the biomass
data add very little to the analysis. However, this must be balanced by the
relative costs of obtaining the two types of benthic data. In general, bio-
mass measurements are less costly and can be obtained more quickly than count
data; dry weights are usually preferred over wet weights because their
measurement endpoint can be more precisely obtained.
Another factor that warrants consideration during the design and establish-
ment of the benthic monitoring program is the expected frequency of zero
organism counts in the samples. These should be avoided when possible by
increasing the area sampled or decreasing the number of taxonomic categories
or both. The reasoning behind this recommendation is primarily statistical,
although the biological interpretation of the data base is also complicated
by increasing frequencies of zero counts. The usual underlying statistical
assumptions that are made in order to analyze the data include the assump-
tion of a normal distribution or alternatively that the data car be trans-
formed into the normal distribution. Data sets with substantial numbers of
zero counts cannot be assumed or transformed to follow the normal distribu-
tion and alternative techniques (often more complicated or less powerful)
331
-------
must be utilized. Thus, it is desirable to avoid for statistical reasons,
whenever possible, data sets containing substantial numbers of samples with
zero organisms. If the true density of organisms is such that zero count
samples occur frequently, it follows that there is also a low probability of
detecting a plant impact that reduces the number of these benthic organisms.
Abiotic measurements are frequently made along with the collection of benthic
data and are usually based on point estimates, e.g. grab samples of water
or sediment. Only the review of Haddam iieck Nuclear Power Plant [8]
attempted to use these measurements in the data analysis. Their conclusions,
based on correlations with water and sediment temperatures and dissolved
oxygen, indicate that, these measurements were of little value in explaining
changes in the benthic communities. However, this does not provide good
evidence that abiotic data should not be collected. To be of greatest use
in assessing impacts on the biota, physical and chemical measurements should
be made on a very frequent or continuous basis in order to define the varia-
tions in these factors and to obtain an integrated measure of potential
stressors. The expense and difficulties in maintaining the instrumentation
necessary for the continuous monitoring of these parameters will preclude
their measurement at more than a very few stations. Physical and chemical
measurements also are necessary in plant monitoring to obtain evidence that
effluent discharge to surface waters are within applicable state and Federal
water quality standards.
REVIEW OF STATISTICAL TECHNIQUE
of Variance
The reviews by both AMI and PNL relied heavily upon analysis of variance
techniques to test the hypothesis of no impact from the operation of the
power plant. In general, time, space, stations, and operating status were
included as fixed effects in the model. The data were assumed to be log-
rormal and were transformed prior to the analysis. The major term of inter-
est was the interaction term representing the operating status x sites
(control, treatment, or exposed). Multiple observations at a station taken
at the same time were considered to be replicates. Descriptions of the
models, assumptions, and test results are presented [1,8].
This approach is probably the best available methodology when faced with data
bases that contain the following types of problems: missing observations,
unbalanced sample sizes, confounding of effects and no a^ £rip_ri_ design. Often
the required underlying assumptions are not fully satisfielf7~wTth the inves-
tigator relying upon the robustness of the technique and the acknowledgment
that the tests are only approximate. However, if we look beyond the problems
of analyzing "inadequate" data bases and consider this analysis of variance
approach to well-designed monitoring programs, some items for consideration
emerge.
The first item to be considered is the manner in which the analysis tests
the impact hypothesis. The effect of the plant cannot be represented in the
332
-------
factorial design. The hypothesis testing occurs by examining the significance
of the interaction term, operating status x sites. The interaction for the
determination of environmental changes can be depicted as follows:
Stations
Control Effected
Status
Preoperational M,, M,?
Operational M2] M22
•where M-JJ is the mean of all observations taken with operating status i and
at station category j. The interaction hypothesis that is tested is Mi] -
M]2 = Hie - ^22 and Is performed with a single degree of freedom test [10,11].
A disadvantage of this test is the uncertainty that is introduced by the non-
normal underlying probability distributions. The problem of nonnormal data
is generally recognized but a satisfactory solution has not been achieved.
McCaughran [11] points out that interaction effects are influenced by the
scale of measurement; thus, transformations may introduce or remove signifi-
cant interaction effects. He recommends that the data not be transformed,
therefore relying on the robustness of the analysis to mitigate the effects
of the nonnonnal data distribution. Thomas [12] takes the alternative
approach and reasons that the effects from nonnormality are more important
and transforms the data prior to analysis. He acknowledges that this may
alter the significance of the interaction terms. Thus, the judgment of a
significant effect (impact) may be dependent upon the choice^of scale adopted
by the investigator.
A second and major disadvantage is the interpretation of a statistically
significant interaction effect. One of the underlying causes for a signifi-
cant interaction effect can be the impact resulting from the power plant
operations. However, there are also numerous other causes that can give rise
to a significant interaction term. In general, the interaction term may be
significant whenever the main factors, operating status and sampling sites,
are not strictly additive. The following quote from Winer [13] summarizes
the problem:
In some cases there is an intrinsic interaction between the fac-
tors which cannot be considered a function of the choice of the
scale of measurement. These cases are not always easily distin-
guished from cases in which the interaction is essentially an
artifact of the scale of measurement.
Thus, a test for significance of the interaction term may not be a true test
of impact.
As pointed out by McCaughran [11] the above problems disappear when a one-
way analysis of variance design is used. In two-way designs, the problems
333
-------
are formidable but tractable. The majority of the design? considered by
Gore et al. [7,8] and Murarka et al. [1] were on the order of three- to
five-way analyses, often with & nesting of factors. These designs, there-
fore, include higher order interaction terms that can be statistically
significant and, thereby, confound the interpretation of the interaction
effect of interest. Thus, the interpretation of a significant interaction
term as a significant power plant induced impact does not have the desired
quantitative statistical foundation that is to be preferred for making envi-
ronmental decisions.
An additional problem associated with using analysis of variance techniques
to evaluate monitoring data is sufficient computer resources. When the
model contains several main effects and several levels for each effect, the
number of cells required to estimate the complete model can easily surpass
the available memory of most computers. A major contribution to the
increased number of cells results from using each station as a level. Two
methods have been used in the analysis of monitoring data by Gore et al. [7,8]
and Murarka et al. [1] to circumvent the problem. Gore et al. [7,8] suggests
that the monthly effects be treated by separate analyses. They constructed
12 analysis of variance tables, one for each month, and tested for impact
within each month to reduce computer core requirements. The major drawback
appears to be that the number of hypotheses and tests escalates beyond
reasonable limits. For example, in analyzing the Haddam Neck benthic data,
4 species and 3 phylum groups were analyzed for 11 monthly data sets, a total
of 77 analysis of variance tables. Thus, the solution to the computer space
problem has created a problem in controlling the experimentwise error rates
(i.e., the probability levels for the F-tests).
Murarka et al. [1] resolved the problem of limited computer core by reducing
the number of factors in the model. He omitted month, depth zones and
repeated samples at the same stations and greatly reduced the number of cells
and, thus, computer core requirements. This technique requires the additional
assumption that years represent "random" observations cf the preoperational
and operational periods. The consequences of this approach, mainly the effect
on the estimate of the error variance, need to be further explored and tested
before the usefulness of omitting factors can be established.
Regression Techniques
Regression techniques were used by Adams et al« [4,5] to test hypotheses of
impact from the power plant. In general, they attempted to estimate the
relationship between the control and treatment area station densities by
fitting a linear regression line to the paired observations. Murarka et al.
[14] proposed a multiple regression model in which data from all the control
stations were used to predict the density at the treatment station with the
plant operating.
The assessment of a plant impact with the Surry data was severe,y constrained
by the lack of useable preoperational monitoring data [5]. The authors
attempted to fit the following general model to the operational data:
334
-------
R- = "c (RJ « n (f ) pj
s c j=1 j
where Rs and Rc are the density at the stressed and control stations, respec-
tively, k and a are coefficients that estimate the relationship between the
two stations, the fj's are the stress factors, and 3j the associated coeffi-
cient: The model is linearized by a logarithmic transformation and the
resulting model fit to the benthic data.
RS = ^n k + a In RC + B1 6Q + 32 6£
where 60 is the ln (Ts/Tc) at lag zero and 62 is In (TS/TC) at lag two; where
TS and Tc represent temperatures at the discharge and control areas, respec-
tively. The units for the time lag or the reasons for omitting a lag one
term are not given by the authors. The model is fit by ordinary least-square
methods and the coefficients 3] and B£ are interpreted as indicating delete-
rious effects when they are negative. We believe such an interpretation can
lead to serious errors. The problems with the above procedure are twofold,
one with the application of the methodology and the other with the interpre-
tation of the statistical analysis.
The major exception to the above procedure is the interpretation of the model
parameters to indicate cause and effect (i.e., impact). Adams et al. [4]
imply that the inclusion of the term, a In Rc, accounts for the natural vari-
ability in the system and that if 3i is statistically nonzero, then a signifi-
cant effect due to f-j is present. Such an interpretation is unwarranted based
on this statistical method. The model is fit using least squares, which uses
all the independent variables simultaneously, and with equal weight. The
method does not. attempt to first explain as much variability as possible
using only the control station density, i.e., account for naturaly variation,
and then attempt to explain the pure noise component. The contribution to
the linear regression of the individual variables cannot be adequately
assessed by using this technique. Also, it is unclear what effect the inclu-
sion of the lagged terms had and what criteria were used to select the lag 0
and lag 2 terms. A method which examines the correlation and partial corre-
lation coefficients, for example a stepwise regression procedure, is needed
to evaluate the relative importance of the individual terms in the model. In
addition, the authors do not establish that a deleterious effect can be
expected to produce a negative coefficient (31 or 32) within the multiple
regression equation. Thus, I am not convinced that an assessment of impact
can be based on the sign of 3] or 32-
The use of the above linear regression models is dependent upon assumptions
about the distributions and associated error structures of both the X(RC)
and Y(RS) variables. Snedecor and Cochran [15] list the following three
assumptions that are made about the relationship between X and Y for ordi-
nary linear regressions:
1) For each selected X there is a normal distribution of Y from which the
sample vafue ofT is drawn at random.
335
-------
2) The population values of Y corresponding to a sel_e_cte_d X has a mean y
that lies on the straight line y = a + 3X.
3) In each population, the standard deviation of Y about its mean a + 0X
has the same value, assumed constant as X varies (emphasis added).
These assumptions are likely to be satisfied by the monitoring data, except
that values of X were not selected. The values of X are subject; to the ssme
errors of measurement and inherent variability as the values of Y. Ricker
[16] reviews a number of different applications of linear regression in
fisheries research and concludes that estimates based on ordinary predictive
regressions are biased to some degree. In addition, he gives a number of
the rel event statistical procedures that can be applied to obtain unbiased
estimates under certain other assumptions. Thus, although these problems are
often ignored by many authors, we believe that the resultant analyses can be
biased and additional efforts are required to establish their validity.
Murarka et al . [14] proposes a multiple regression model of the form:
where the subscript j indicates a specific time point in the operational
period at station i. The model was fit to preoperational data to estimate
6, tj> and the variance a^. The test of impact is performed by calculating
the predicted value of V-jj based on the operational period observations at
X-j j and comparing these to the observed data YJJ. While this approach is
seemingly attractive, it has two major drawbacks. As pointed out by Murarka
[10] further testing and evaluation of the problems associated with the fact
that both the X and Y variables are subject to error must be accomplished
before accepting the technique. Additional problems may arise in those cases
where the operational control station densities are considerably beyond the
range of the preoperational densities. In these cases, the variance, which
depends on the difference between the operational and the preoperational
mean density, may become large with a resulting decrease in sensitivity of
the tests.
Another linear regression approach is suggested by Adams et al . [5] in their
analysis of the data from the Peach Bottom Atomic Power Station. Their
approach considers the following model for the preoperational and operational
data sets for each pair of treatment-control stations:
T = C3 or In T = B In c
where T and C are the densities at the treatment and control stations, respec-
tively, and 3 is slope of the regression line. Thus, the data are transformed
by logarithms and a zero intercept regression line fit by least squares tech-
niques. The hypothesis that the regression coefficient, Sp (preoperational)
and BO (operational) are equal is tested to evaluate the impact of the power
336
-------
plant operation. This method contains the same uncertainty as the previous
regression methods because both T and C are measured with approximately the
same error structures.
In addition, no explanation or rationale is presented for choosing the zero
ir -rcept models in logarithmic units. Therefore, we are left to speculate
abtut the underlying assumptions. A model which is slightly less restrictive
could be written as T = a CB. The above formulation assumes that a = 1 for
all cases. The model transformed by logarithms would then be of the form
In T = In a + 3 In C, and a test of the hypothesis that a = 1 could be made.
This preliminary test should be made because, although it may be logical to
assume that whenever In C approaches zero, In T should also approach zero,
it is not clear that it is a straight line relationship throughout the entire
range of In C and In T values. As pointed out by Snedecor and Cochrai; [15]:
This model (zero-intercept) should not be adapted without careful
inspection of the data, since complications can arise. If the
sample values of X are all some distance from zero, plotting may
show that a straight line through the origin is a poor fit,
although a straight line that is not forced to go through the
origin seems adequate. The explanation may be that population
relation between X and Y is curved, the curvature being marked
near zero but slight in the range within which X has been
measured. A straight line of the form (a + bx) will then be
a good approximation within the sample range, though untrust-
worthy for extrapolation.
This advice seems appropriate in the present situation. It would seem
reasonable to expect that the relationship between T and C could well be dif-
ferent as C approached zero. An additional implied caveat is that extrapola-
tion beyond the observed range of prooperational densities may lead to
extraneous conclusions.
Another problem arises when we consider the experimentwise error rate of the
above procedure. For example, the benthic monitoring data was collected at
one control and ten treatment stations at the Peach Bottom facility [5].
Subsequently, 18 3 values are estimated and 8 comparisons of 3p = 30 are
made. Since a Type I error rate of 0.10 or 0.05 for all of these tests
should be maintained, the individual tests should be conducted at an adjusted
a level. These authors combined the preoperational and operational data sets
and estimated a single 3 for stations where 3D = 3o- However, it would seem
more beneficial to combine stations with similar preoperational and opera-
tional relationships, i.e., where 3pi = 3P1 and 301 = 8oi. These combined
data sets would then increase the sensitivity of the test of interest,
Bp = 30 in addiLi on to reducing the overall number of tests.
Time Series Analysis
The use of time series analysis technique was proposed and utilized by
Hurarka et al. [1,2] and Murarka [3] as an alternative to the linear models
337
-------
approach. Lettenmaier and Hurry [17] also investigated the applicability of
intervention analysis, a method based on the time series approach, to the
monitoring of nuclear power plant environmental impacts. On the surface,
these models appear to provide a very useful technique for examining the
impact hypothesis.(a) Instead of assuming that the data meet the underly-
ing assumptions for the classical analysis of variance approach, the data
are treated as a time series and the integrated moving average process given
by Box and Tiao [19] is applied.
The major disadvantage of the approach appears to be the requirement for
moderate to large data sets. The length of the data set as well as the fre-
quency of observation required to successfully apply the method with suffi-
cient precision severely limit its application to monitoring data. With
regards to the frequency of sampling, Box and Jenkins [18] suggest that the
interval should be fairly short compared with the time constants expected
for the system. The time constants are a measure of the speed with which
the system approaches a new state following a shift in the input conditions.
Since it is likely that many of the aquatic organisms respond fairly rapidly
to certain kinds of changes, a sampling interval of daily or every few days
appears.to be needed. With the possible exception of impingement and
physical-chemical monitoring, sampling at this level is both economically and
biologically not feasible.
Another aspect of the method that needs further investigation before the time
series approach can be fully evaluated is the modeling of the impact itself.
Murarka et al. [1] investigated a model which assumed that impact could be
represented by an immediate and constant effect upon the time series. Undoubt-
edly this represents an over simplification of the actual impact function and
additional work and investigations are needed to define more realistic
formulations.
STATISTICAL CONSIDERATIONS
Error Rates
There are three important factors to consider when setting out to establish
the number of samples to collect on a monitoring program. The first of these
is to evaluate the consequences of concluding that an impact exists, when in
actuality none exists, and to assign an acceptable probability to this happen-
ing. The second factor is the reverse of the first, that is, to assign an
acceptable probability to concluding that no impact exists, when one actually
is present. The third factor that must be considered is the size or magni-
tude of an impact or change that the monitoring program should be capable of
detecting. With these three factors and an estimate of the variance or coef-
ficient of variation, it is possible to answer the often asked question:
How many samples should be collected?
(a) A complete review of time series is beyond the scope of this paper, the
interested reader is referred to Box and Jenkins [18], Box and Taio [19,20].
338
-------
An alternative way to establish sample sizes, and w suspect the most univer-
sal, is to start with a level of effort or price range and calculate how many
samples can be collected within those constraints. This approach often relies
heavily upon approximating the expenditures of previous monitoring prog rem-
and what the market will bear. If the previous studies were well-designed
and conducted at an appropriate level and were similar to the current situa-
tion, them reasonable results can be anticipated. However, when using this
method, the program designer should be aware of the resultant probability
levels and detectable changes that the program is capable of producing.
The selection of the a error rate for testing hypothesis has routinely been
set at 0.05 by scientists and engineers. However, by comparison those
experiments generally were characterized by: 1) controlled experimental con-
ditions; 2) consequences of a wrong decision easier to quantify; 3) signifi-
cance end interpretation of statistical differences uncomplicated; 4) greater
certainty about the underlying distributions; 5) relatively small coefficients
of variation; 6) sampling relatively easy and cheap. The area of testing
impact hypotheses has few, if any,, of these characteristics. Therefore, we
recommend that a slightly larger a error rate of 0.10 be used for testing
impact hypotheses, this also provides a better balance between a and 3 error
rates within the limited range of feasible sample sizes.
Additional consideration should be given to the problems of controlling the
experiment^se error rates. If we assume that each taxoncmic group represents
the experimental unit, then the above error rates are appropriate for each
analysis. However, if the experimental unit is assumed to be a trdphic level
assemblage that potentially will be impacted, then the error rates of the
analysis for each of the individual taxoncmic groups should be adjusted to
control the overall experimental error rates. Thus, if more than two.> taxono-
mic groups are tested, the investigator should be concerned about the
experimentwise error rates.
An example of the consequences of using taxa level error rates when a trophic
level error rate is appropriate will illustrate the problem. If we were to
use an a level of 0.10, 10% chance of making a Type I error, BQ% probability
of not making a Type I error on e_ach_ analysis of a single taxa, then the
formula: p = 1 - (1 - a)n where a = Type I error rate and n = number of
analyses, can be used to calculate the probability of making at least one
Type I error for the n taxa groups. Thus, if a = 0.1 and n = 6, then p = 0.49
or stated simply, we have approximately a 50£ chance of concluding there is
a significant impact on at least one taxa group when, in fact, there is no
impact. A similar line of argument can be made for the probabilities asso-
ciated with making at least one Type II error. That is, if we have a power
of 0.80 on the above six individual analyses, we are likely to fail to
detect a significant impact, i.e. make at least one Type II error, from the
power plant approximately 7K of the time. Thus, whenever the trophic level
is the appropriate experimental unit the investigators must adjuct the error
rates on the individual analyses to achieve acceptable experimentwise error
rates.
-------
The magnitude of the 3 error rate or power levels for testing impact hypothe-
sis was not addressed in any of the nine case studies conducted by ANL, ORNL
and PNL. In general, acceptable power levels have-not been established and
are generally ignored in the evaluation of impact hypotheses. McCaughran [11]
infers by example that a power of 0.90 is to be preferred. Nearly equal pre-
ference is given to power of 0.70, 0.80 and 0.90 by Thomas [12], It would
appear reasonable to assume that the consequences of making either a Type I
or II error are potentially of similar magnitudes. When we make a Type I
error, the power customers pay for unneeded environmental controls at the
nuclear power plant. On the other hand, if we make a Type II error, the
general public incurs an added loss due to the undetected environmental
impact. If we guard against serious Type II errors by requiring continuing
monitoring programs, then perhaps we can allow increased Type II error rates.
However, it would not seem advisable to make decisions based on tests that
have a power of less than 80%. I recommend that the power of the tests impact
hypothesis be 0.80. The selection of the power for testing the impact hypo-
thesis includes the following considerations: 1) consequences of making a
3 error; 2) the uncertainity of the experimental data; 3) cost of the field
program; and 4) our ability to reevaluate the decision periodically during
the life span of the power plant. The g error rates that are greater or less
than this appear to require excessive sample sizes or are too insensitive to
provide a sound basis for decision making. While the selection of a pov/er
level of 0.80 is qualitative, I believe it will aid in focusing attention on
this statistical attribute of impact assessment.
The third consideration in establishing sample sizes is the magnitude of
change to detect that which is appropriate for impact assessment. As was
pointed out to each of the three laboratories' studies, ANL, ORNL and PNL,
research is needed to provide information on the biological significance of
an observed change and what constitutes a significant change. Thus, while
no site specific or generic value can be recommended, it appears that the
detection of 50% changes in density for the data sets reviewed in this report
is a reasonable level.
SUMMARY AND CONCLUSIONS
The review of the nine power plant environmental impact assessment programs
has indicated that there is very little uniformity or consistency in the
quantitative approach. This extends beyond the site specific requirements
imposed by the habitat and ecosystem. In addition, a difference exists
between the objectives as established for the monitoring program and the
inferences that can be based on the data actually collected. Three aspects
of the impact quantification and assessment were examined and the following
conclusions reached for the benthic populations.
The identification of benthic organisms to the "lowest possible taxon" is
generally unwarranted for the current state-of-the-art of impact assessment.
Rather, organisms should be classified into a maximum of five taxonomic or
340
-------
functional groups. In addition, collecting data on abiotic water quality
parameters solely to establish ecological relationships is unrealistic for
the scope of a monitoring program.
Of the three analytical methods reviewed in dttail it appears that the analysis
of variance approach offers the best available framework for testing impact
hypothesis. This technique, although likely requiring further refinement,
offers the greatest probability of correctly identifying impacts.
The testing of impact hypothesis should be dons with known error rates,
including a definition of the experimental unit and the probabilities of
making Type I and Type II errors. It is recommended that benthis studies be
designed to detect at least 50% changes in population density with an a rate
of 0.10 and a 6 rate of 0.20. These three parameters should be evaluated and
clearly stated in the environmental impact statement.
REFERENCES
1. Murarka, I. P., A. Policastro, E. Daniels, J. Ferrante and F. Yaslow.
1 9 76 . An E v a ]v a t i on o f E n v i r onrn s ntaj _Data_Re la ti_n_g to J5_eJ^ct.edJ'|ucJ_ea_r
Power Plant Sites; _ The_ Zi i'ojlJju_cTe_a r J!pw^_Jt£ti on '"Sl't.e . AlC/ E 1 S -:5 ,
Argonne National Laboratory, Argorme, fL."
2. Murarka, I. P., J. G. Ferrante, E. W. Daniels, E. E. Pentecost. 1976.
An Evaluation of Environmental Pat a_ R e 1 atin g_ t o_Se Tect e d _N ucj ear P owerr
"
_ _ _ _ _
Plant Sites: Prairie Island N uj^arjfc ene j^aVm g_ PjTp nt "Site . A N L/ E I S ~ G ,
Argonne National Laboratory, 'Argonne, TC7
3. Murarka, I. P. 1976. An Evaluation of Environmental Pat a__R e ]_a_ t ing t o
Selected ul£ar Power Plant Sites : 1 !J
___ _ _._
Station Si te. ANL/EfS-7, Argonne National Laboratory, "Argonne, 1L.
4. Adams, S. M. , P. A. Cunningham, D. D. Gray and K. D. Kumar. 1977. A_
Critical Evaluation of the Monradiol o_g_i c a l__Eny v^onjTirji ja 1 J'e ch n i ca 1
Specifications, Vol. 2, Surry Power 'Plants Unfts'l jnd 2. ' ORNL/NUREG/TM-/
Oak Ridge National Laboratory, Oak ""Ridge, TN.
5. Adams, S. M. , P. A. Cunningham, D. D. Gray, K. D. Kumar and A, J. Witten.
1977. ACritic3l Evaluaton jf the Nonradjqlocn'cjjl EiTvironmental
'
_
Technical ~Spec ii]_ca_ti_o_n£,__\/pj . 3, Peach ~P£f ^_
Units 2 and 3'." OR"NL/NUREG/TM-7T Dak 'Ridge "i^t For, aT Laboratory,
Oak Ridge, TN.
6. Adams, S. M. , P. A. Cunningham, D. D. Gray and K. D. Kumar. 1977. A
Criti cal Evaluation of t hej2pj]ra_di^ljD£i£aJ_Ejri^^
%?^9J3i-^aJL1 9Di.i_y£] '_i> ~San_On_o^r^iJu£]£ajM5ene^
ORNL7NTJREG7TM- 72". 0"ak"Ridge' National Laboratory, Oak "Ridge, TN.
341
-------
7. Gore, K. L. , J. M. Thomas, L. D. Kannberg and D. 6. Watson. 1976.
U(2J.8JL Po'-'/er_ P ljmt^,_ JEnvi roninental Impact
BNWL -2T50/NRC-1. Battelle,
_
Richfand"," WA. '
8. Gore, K. L., J. M. Thomas, L. D. Kannberg, J. A. MahafFey and D. G.
Wa tson . 1 976 . iYl]j^tioji_ojM^
Power Plant, _Erwi fo_nji}enta_l_Iniig_a_ct j:'redi5:t;i_o_n_,_Ba_sed on Mom' to ring
Programs': Bl^L~Fi W^TT" B at te Ti e ~ R i c h 1 a n d , WA". "
9- Gore, K. L., J. M. Thomas, L. D. Kannberg and D. G. Watson. 1977.
Evalua t i_c n_of_M j 11s t one Nuclea r_P o wer _P lant ,_ E nv i jonm_ent a _]__Impa _ct
'P red f c t, i on , Ba s_ed olfMon 1 1 oruig J''rug_rajJK BHWL^l^/TJRC-'l . Battelle,
Richland," WA.
10. Murarka, I. P. 1977. Statj's^icaj^.n? lysis of the D. C. Cook Preopera-
t_i onal En vi ronmental Moni tori ng Program . ANL/EIS-9. Argonne National
Laboratory, Argonne, IL." " "*
11. McCaughran, D. A. 1977. The quality of inferences concerning the effects
of' nuclear power plants on the environment, pp. 229-242. In: Proceedings
of_ t_h_e Conference 9P^s_s_essj>,a_the Effects of_Power PJent Induced Mor-
t.a 1 T ty o n " F'i s h P b pill a f 1 o n s~. "Van "WThk Te Tedl , Pergiin;on "Press, MY.
12. Thomas, J. M. 1977. Factors to consider in monitoring programs sug-
gested by statistical analysis of available data, pp. 243-255. In:
Proceed inns f the Conforene on Assessm the Effects of Power Pant
_ ^ _
j n du ce d l^tVrTtfy^Jjri^TilsTr PopiTI aTTonsJ Van WThkfe (ed ) . Perge/rnori Press,
NY.
13. Winer, B. J. 1971. Stati sti ca 1 Pri nci pi es In Experinient^l Des jgji _.
Second Edition. McGraw-Hill , NY',"' 907 "pp"."'
14. Murarka, I. P., A. J. Policastro, J. G. Ferrante, E. W. Denials and
G. J. Manner. 1976. An_ Evaluation of Environmental Data Role ting to
Selected Mud ear Power P lantrTH's'sT" A S'ynthesTs and sTJmMry7~Wi th
Recomnierida t'i ons . AKL/ E I S-8~. _" Argonne "Natfonal Laboratory, "Argonne, IL.
15. Snedecor, .G. W. and W. G. Cochraru 1967. Statistical Methods, Sixth
Edition. Iowa State University Press, IA., 593 pp.
16. Ricker, W. E. 1973. Linear regressions in fishery research. J. Fish.
Res. Bd. Can. 30:409-434.
17. Lettenmaier, D. D. and L. C. Murry. 1977. Des i qn of ' JNonradi_q1 ogjcal
Aguati_c_S_aiiip_l ijig J'_ro2_ram^_fo_r Nu^leajM^^ej^^l_airt_ I in pa c ~t Asses s'men t~
Using IritervetvLi onTli i_d lysis . GW-fiKC-6. University of Washington,
Seattle, WAT
18. Box, G. E. P. and G. M. Jenkins. 1970. J_irne_ .Series Analysis Forecasting
and Control . Holden Day, CA., 553 pp.
342
-------
19. Box, G. E. P. and G. C. Tiao. 1965. A change in level of a nonstationary
time series. Biometrika 52:181-192.
20. Box, G. E. P. and G. C. Tiao. 1975. Intervention analysis with applica-
tions to economic and environmental problems. J. Am. Stat. Assoc. 70:70-79,
343
-------
FURTHER STUDIES IN SYSTEMS ANALYSIS OF
COOLIi'G LAKES: HYDRODYNAMICS AND ENTRAINMENT
Kenneth D. Robinson and Robert 3. Schafish
R.W. Beck and Associates
Denver, Colorado
George Camougis
New England Research, Inc.
Worcester, Massachusetts
ABSTRACT
The effects of entrainment on semi-closed ecosystems in cooling lakes are not clearly
understood. A methodology is presented wherein entrainment assessment is based on
power plant induced hydrodynamics and biological population dynamics. Entrainment
probability distributions are developed for important areas (habitat) in cooling lakes.
These are compared with regeneration times and critical development times for
specific biota which allows a semi-quantitative evaluation of the effects of entrain-
ment on the entire ecosystem.
INTRODUCTION
Entrainment is a factor that must be accounted for in the overall biological evaluation
of cooling lakes. Rather than consider entrainment as a separate issue in cooling lake
ecosystem analysis, that evaluation must include the relationships among entrainment
effects, lake hydrodynamics and thermal effects. A systems approach to cooling lake
analysis provides an important assessment tool. This paper is a further development of
systems concepts applied to cooling lakes which was presented as a paper at the 1977
Waste Heat Management and Utilization Conference (1).
Federal 316 (b) guidelines to meet the requirement that intake structures reflect the
best technology available for minimizing adverse impact, classify potential impact
designs according to the volume of intake flow relative to the volume of the source
water body (2). The guidelines further define a zone of potential involvement for
entrainment as that portion of the source water body that is likely to be drawn into an
intake structure. These concepts are inappropriate for intakes in closed ecosystems
such as those found in cooling lakes. Hydrothermal analysis shows that the entire
volume of a moderately loaded cooling lake can be entrained, even in lakes with
control arms.
Mortality resulting from entrainment can be viewed as a predatory process in the
cooling lake ecosystem. How this predation affects the various trophic levels in the
lake is a function of intake variables such as location, cooling water flow rate, intake
configuration and design. It is also a function of the cooling lake hydrodynamics which
dictate how the prey arrives in the vicinity of the predator. A more complete
understanding of the relationship between hydrodynamics and entrainment shows that
there is not a uniform progression of the entire lake contents toward the intake.
344
-------
Rather there is a very wide variation in the residence time of any particle of water
which is dependent upon the lake hydrodynamics. This variation can be described and
analyzed.
COOLING LAKE ECOSYSTEMS
As with most lacustrine systems, cooling lakes develop a complex biological commu-
nity consisting of various trophic levels and biotic groups. The final community
depends upon the physical and chemical factors in the lake ecosystem.
Representative Biota
No single method has been accepted universally as the procedural approach to
assessing the aquatic biota in environmental studies. However, based upon general
ecological principles by Odum(3) and the useful manual by Weber (4) and the pertinent
guidelines from the USEPA (5) it has become customary to assess the following aquatic
biota in environmental studies of cooling lakes:
(1) phytoplankton
(2) periphyton
(3) macrophyton
(*) zooplankton
(5) benthic macroinvertebrates
(6) fishes
The trophic relationships among these various biotic groups are generally well
understood. Also the major genera and species can be predicted with some degree of
confidence, although each ecosystem has its own unique characteristics. Cooling lake
ecosystems may differ from those of natural lakes because of power plant induced
flow patterns and thermal structures. Heat addition can increase productivity due to
longer growing seasons while entrainment can act upon certain biota in a predatory
way.
Habitat Assessment
Of great practical importance is the assessment of the various habitats in the cooling
lake ecosystem. For example, the physical conditions such as lake morphometry,
depth, light, temperature, dissolved oxygen and other factors greatly influence the
types of biota that grow in the various locations within the lake. The importance of
habitat assessment to arrive at some estimate of biological importance is discussed in
the USEPA guidance document (5) on entrainment studies. An important concept is
that some aquatic biota (e.g., phytoplankton) tend to be homogenous in their
distribution throughout the water mass; in contrast, other biota (e.g., macrophyton,
fishes) tend to have more localized habitats.
Thus, as examples, the shallow areas with macrophytes, the deeper areas of the lake,
and the areas with different substrates can all be identified and mapped. This permits
a graphic presentation of specific habitat zones. Finally, the types of biota (especially
fishes) that are likely to use the habitat zones can be identified. Some estimate of the
probability of entrainment of the various biota within the various habitat zones can
345
-------
then be made by correlating habitat, regeneration capacity and the influence of the
cooling lake hydrodynamics.
Regeneration Capacity
Another important factor in the assessment of entrainment is the regeneration
capacity of the various biotic groups. Organisms representative of all biotic
categories may be entrained, including phytoplankton, zooplankton, larval forms of
benthic macroinvertebrates, and fish eggs and larvae. The microbiota (e.g., phyto-
plankton) have short regeneration times, and the effects of entrainment are generally
not critical to the maintenance of viable populations. On the other hand, the
entrainment of fish eggs and larvae with longer incubation and development times may
be potentially more adverse to fish populations, especially where spawning generally
occurs on a yearly basis. However, population dynamics are also based on fecundity
and density-dependent factors, and are important compensation considerations. Con-
sequently, any realistic assessment of regeneration capacity must consider these other
factors along with the regeneration times. The following table presents some typical
times for various reproduction parameters for major biotic groups (Table 1).
TABLE 1
Biotic Various
Groups Reproduction Parameters Time
Bacteria Typical Doubling Time Less than 1 hr
Phytoplankton Typical Doubling Time 10 hr - 100 hr
Zooplankton Brood Interval 2-5 days
Maturation Time 5 days
Fish Eggs Typical Incubation Time 2-10 days
Fish Larvae Typical Development Time for More than one
Larval Stages month
Juvenile Fish Typical Maturation Time More than one
year
It is important to recognize that cooling lakes function as closed systems with respect
to entrainment, in contrast with riverine systems where the through-put is large.
Because the entire contents of the cooling lake may be subject to entrainment which
can result in high mortality, the concepts of regeneration time or critical development
time for various biotic components are especially important in assessing the effects of
entrainment. Should a particular group of organisms be subject to entrainment with a
frequency close to or greater than regeneration time (or critical development time),
the effects on the balance of the ecosystem could be significant.
346
-------
COOLING LAKE HYDRODYNAMICS
Cooling lakes are hydrothermally distinct from natural lakes or reservoirs because of
the relatively large volumes of heated water which are circulated through them. In
natural impoundments, seasonal variation causes cyclical stratification and destratifi-
cation due to density differences between warm and cool water. Cooling water
discharge from a power plant superimposes additional thermal conditions and creates
unique flow patterns which can significantly affect water quality and influence aquatic
habitat. The pumping-induced circulation can also affect the way in which organisms
are entrained by the intake structure.
Residence Time Distribution
Residence time is generally regarded as the time for water to flow from the discharge
to the intake. An average residence time for the discharge flow is calculated on a
flow-through basis (lake volume/flow rate). For purposes of entrainment assessment
this concept can be expanded to include the time for water to flow to the intake from
any specific area (habitat) in a lake.
The flow dynamics in a cooling lake can be very complex; because of the induced
hydro-thermal structure, an average residence time may not be meaningful since the
discharge flow will more likely be re-entrained over a range of residence times.
Depending on the lake configuration, part of the discharge flow could be "short-
circuited" with a re-entrainment time much shorter than the average residence time.
Conversely, another part of the discharge could take much longer. For this reason it is
more realistic to consider a residence time distribution over a range of entrainment
time intervals in order to account for the actual flow patterns.
A residence time distribution is equivalent to an entrainment probability distribution
for water (or organisms) in the vicinity of the thermal discharge. Using the same
circulation patterns, entrainment probability distributions can also be developed for
any area (habitat) in a cooling lake. Thus, for aquatic organisms in a particular
location with respect to the cooling water intake, an entrainment probability distribu-
tion can be compared with regeneration times for biota such as zooplankton or
phytoplankton, or for critical development times in the case of fish eggs and larvae.
Cooling Lake Model
Hydrothermal analysis can be used to predict the complex flow structures in cooling
lakes (1). Circulation patterns can range from simple plug flow in long, narrow lakes to
complex, multilevel circulations in wide, strongly stratified lakes. Studies have shown
that, with proper design, cooling water discharge from a power plant will tend to form
a distinct, heated surface layer (due to density differences) which will remain intact
and tend to spread over the entire surface of a cooling lake.(6)
Ryan and Harleman (7) have developed a mathematical model for an idealized cooling
pond which separates the water body into a discharge mixing region, a warm surface
layer and a cooler sublayer as shown in Figure 1. This constitutes the primary flow
pattern in many cooling lakes and forms the basis for calculating residence time
distributions. In the model, heated discharge water enters the surface layer where it
entrains cooler subsurface water and then flows to the far end of the reservoir. At the
347
-------
end of the surface layer, destratification causes down-welling of surface water to the
sublayer. The sublayer can be stratified due to heat input from the down-welling and
also from solar radiation not absorbed in the surface layer.
The actual detailed behavior of flows and temperatures in such lakes will be somewhat
more complex than in the idealized mathematical model. Some of the variations can
be accounted for by considering important secondary flow patterns. Research has
shown that a buoyancy-driven circulation pattern can be established in dead ended
sidearms causing warmer surface water to flow into the arms.(8) As the surface flow in
the sidearm cools, the more dense water sinks to the sublayer and causes a return flow
to the main water body. This creates a flow residence time within the sidearm which
becomes one component of the residence time distribution for the cooling water flow
in the entire lake. Watershed runoff, pumped makeup water and wind-driven currents
may create other secondary flow patterns which can add components to the residence
time distribution.
Model Application
The analysis described above was applied to a 1500-acre cooling lake (shown in Figure
2) to calculate probabilities of entrainment at specific locations within the lake. The
calculations were based on a plant capacity of 1000 MW, a condenser cooling water
flow of 800 cfs and a condenser temperature rise of 27.5 F. Based on analysis, the
discharge will produce a heated surface layer 3 feet thick. The intake, located on the
lower side of the lake, draws water from both the surface layer and the sublayer. This
produces a complex circulation pattern in which only a portion of the surface layer is
entrained. The remainder of the surface flow continues to the lower end of the lake
where down-welling to the sublayer occurs. Part of this sublayer flow will return to
the intake, while the remainder will flow in the sublayer to the upper end of the lake
to be re-entrained in the discharge mixing region. Other secondary flows such as
sidearm circulation will add components to the flow patterns.
For a reservoir volume of 14,000 acre-feet, the discharge flow will have an average
residence time in the reservoir of about nine days on a flow-through basis. However,
because of the complex flow pattern, some of the discharge water will flow via the
surface layer to the intake while the remainder may follow any one of the other
patterns discussed above. Using the hydrothermal analysis discussed above, represen-
tative entrainment probability distributions were developed for three locations in the
lake: the discharge area, a midlake area and a side arm (as shown in Figure 2). Because
of the simplifications necessary for the analysis, these distributions are somewhat
idealized. They are, nevertheless, useful for evaluating the potential impact of
entraiment on the ecosystem of the cooling lake. As the figure shows, the area near
the discharge has a relatively high probability of entrainment within 1 to 5 days. This
is in contrast to the areas near the end of the lake which have the residence time
distributions displaced towards the longer residence times.
ENTRAINMENT ASSESSMENT
The integration of the entrainment probability distributions with regeneration and
development times can provide a semi-quantitive basis for assessing the effects of
entrainment. Since the mortality from entrainment predation can be very high, the
348
-------
regeneration times for various biota are important. Comparison of regeneration times
for the smaller organisms (e.g. bacteria, plankton) with residence time distributions as
presented in the illustrative example can indicate whether or not entrainment will
adversely affect populations. The same interpretations can be applied to fish eggs and
larvae. However, for these organisms, the entrainment probabilty is further complica-
ted by factors such as spawning habitat, egg buoyancy, mobility and preferred habitat
of immature forms. Fish eggs and larvae are not always uniformly distributed in the
water column. Thus correlation of habitat with the entrainment probability of the
adjacent water mass becomes an important assessment tool.
The use of this technique can be illustrated by considering Table 1 and Figure 2. The
probability of entrainment within 1-5 days for the upper end of the lake is high
compared with the lower end of the lake. This may have some effect on populations of
biota with short regeneration times. However, there is still adequate regeneration
time available to assure that these biota can sustain viable populations, perhaps at
somewhat reduced levels. Also, the analysis implies that there is a considerable
portion of the lake where these organisms will be subject to a low probability of
entrainment for the duration of their reproductive time span. Figure 2 shows that the
probability of water mass entrainment within 30 to 90 days is quite high for all
locations within the lake. This may have significant implications to biota with
reproductive paramenters within that time range. However, as discussed above, it is
necessary that other factors (e.g. spawning habitat, egg buoyancy) be considered in the
overall assessment of the probable effects of entrainment.
CONCLUSIONS
The overall approach of this analysis has the advantage of looking at a cooling lake as
an integrated system. Residence times correlated with habitat zones can provide
realistic estimates of the probability of entrainment of the various biota and thus
permits an assessment of the significance of entrainment to the lake ecosystem as a
whole. \
This approach to entrainment evaluation in cooling lakes has implications in several
areas of cooling lake ecosystem analysis. A thorough understanding of lake hydrody-
namics in the early planning stages can allow selection of the design features with the
purpose of controlling entrainment effects. An intake system could be designed as a
selective predator by integrating hydrodynamics, habitat of important biota and
cooling water intake location and design. In this way, entrainment mortality for
certain biota could be selectively increased and could actually benefit the lake
ecosystem. This approach is divergent from the present regulatory framework for
analysis of entrainment effects which requires that the intake system be selected so as
to minimize environmental effects. It is apparent that guidelines developed for
entrainment evaluation in cooling lakes should be re-evaluated to reflect these
concepts.
349
-------
REFERENCES
1. Robinson, K.D., Schafish, R.3. and Camougis, G. 1977. A Systems Approach to
Biological and Thermal Considerations in Cooling Lake Analyses. Proceedings of
the First Annual Waste Heat Management and Utilization Conference, Miami
Beach. May 1977.
2. U.S. Environmental Protection Agency (USEPA). 1976. Development Document
for Best Technology Available for the Location, Design, Construction and
Capacity of Cooling Water Intake Structures for Minimizing Adverse Environ-
mental Impact. Effluent Guidelines Division, Office of Water and Hazardous
Materials. Washington, D.C.; Government Printing Office.
3. Odum, E.P. 1971. Fundamentals of Ecology. Third Edition. Philadelphia: W.B.
Saunders Company.
^. Weber, C.I. (ed.). 1973. Biological Field and Laboratory Methods for Measuring
the Quality of Surface Waters and Effluents" National Environmental Research
Center, Office of Research and Development. Cincinnati: U.S. Environmental
Protection Agency. (EPA 670/4-73-001).
5. U.S. Environmental Protection Agency (USEPA). 1977. Guidance for Evaluating
the Adverse Impact of Cooling Water Intake Structures on the Aquatic Environ-
ment; Section 316 (b), P.L. 92-500 (draft).
6. 3irka, G.H.; Abraham G.; and Harleman, D.R.F. 1975. An Assessment of
Techniques for Hydrothermal Prediction. Ralph M. Parsons Laboratory for
Water Resources and Hydrodynamics Report No. 203. Cambridge: The
Massachusetts Institute of Technology.
7- Ryan, P.3. and Harleman, D.R.F. 1973. An Analytical and Experimental Study of
Transient Cooling Pond Behavior. Ralph M. Parsons Laboratory for Water
Resources and Hydrodynamics Report No. 161. Cambridge: The Massachusetts
Institute of Technology.
8. Brocard, P.; Jirka, G.H.; and Harleman, D.R.F. 1975. Buoyance-Driven Circula-
tions in Side Arms of Cooling Lakes. Presented at American Society of Civil
Engineers' Convention, Denver, Colorado, November 1975.
350
-------
DISCHARGE
.MIXING REGION
SURFACE LAYER
SUBLAYER
1
INTAKE
IDEALIZED COOLING POND MODEL
FIGURE I
-------
Q.LU
1-5 5-10 10-20 20-30 30 +
RESIDENCE TIMES-DAYS
DISCHARGE
PLANT
og60
>-i- 50f
ZJ540' ' MV
K • _ I
£g,o III
1-5 5-10 10-20 20-30 30+
RESIDENCE TIMES-DAYS
1-5 5-10 10-20 20-30 30 +
RESIDENCE TIMES-DAYS
PROBABILITY DISTRIBUTIONS
FOR
ENTRAINMENT IN A COOLING LAKE
352
FIGURE 2
-------
SYNTHESIS AND ANALYSES OF EXISTING COOLING
IMPOUNDMENT INFORMATION ON FISH POPULATIONS*
K. L. Gore and D. H. McKenzie
Battelle Memorial Institute
Pacific Northwest Laboratory
Richland, Washington U.S.A.
ABSTRACT
This paper presents the results of a literature review and assessment
study undertaken to examine the effects of a once-through cooling mode
of power plant operation on small, essentially closed aquatic ecosystems,
represented by cooling impoundments. Fourteen cooling impoundments were
selected based on physical criteria and availability of suitable ecologi-
cal references. No major detrimental effects appeared from power plant
operation on fish populations inhabiting the cooling impoundments. Some
direct qualitative effects were indicated among fish inhabiting areas
that received heated effluents, such as earlier seasonal spawning or
faster growth rates. Increased water velocity associated with effluent
discharge apparently had more controlling influence on fish distribution
and abundance than did temperature, particularly during spawning seasons.
Statistical analyses were performed on data sets of gill net catch data.
Comparisons between control impoundments and heated impoundments were
based on correlation matrices and/or Wilcoxon's Signed Rank Test.
Results indicated there were no statistically significant differences in
fish populations between heated and control impoundments attributable to
plant operation.
INTRODUCTION
It is the common practice of the National Environmental Policy Act of
1969 to require assessment and evaluation of the potential impacts of
power plant operation. In addition, power plants must also abide by the
provisions of Sections 316(a) and 316(b) of the Federal Water Pollution
Control Act Ammendments of 1972, which are concerned with the attainment
of best available technology of plant design to minimize adverse environ-
mental impacts attributable to thermal effluents, impingement, and
entrainment.
The traditional way of disposing of waste heat from power operation has
been direct discharge (once-through) into various aquatic ecosystems.
However due to the above legislative actions, artificial cooling impound-
ments are feasible alternatives to the problem of waste heat disposal.
*Work performed under Contract Nc. 2311202956 for the Electric Power
Research Institute
353
-------
The objectives of our study were (a) to synthesize existing fish popula-
tion information from cooling impoundments, (b) to analyze the data to
extract rational conclusions on fish population effects of condenser
cooling systems, and (c) to identify effects that cannot be assessed
with existing data.
For the purpose of this paper, a cooling impoundment was identified as a
semi-confined water body used to provide recirculation water to cool
condensers of electric power plants in freshwater ecosystems. Load
ratio was defined as impoundment surface area in acres divided by rated
plant generating capacity in MWe. A load ratio of ten or less was
arbitrarily selected as the upper limit where impacts at the ecosystem
level would be expected to occur. In some cases where suitable ecolog-
ical data from cooling impoundments were available, sites with load
ratios greater than ten were included.
Site Selection Criteria
Selection of sites for ecological assessment required: (1) identification
of power plants utilizing freshwater cooling impoundments in their
generating cycle, (2) determination of impoundments that were essentially
closed systems and the surface area of each, (3) compilation of basic
plant operational data, and calculation of load ratios, (4) referencing
available ecological publications for each impoundment, and (5) initial
scrutiny of references to determine their value in our synthesis and
assessment program.
Initial screening and evaluation of about 135 electric power plants were
done by completing a data form for each prospective site. Evaluation of
completed forms and the accumulated list of publications pertaining to
aquatic ecological studies at each site enabled us to narrow the original
list. Eventually, 14 sites were selected for intensive data synthesis
and analysis (Table 1).
Data from publications acquired for each site were evaluated for complete-
ness with ecological matrices designed to cover a wide range of various
abiotic and biotic parameters. Among the options and considerations used
in final site selection were: (1) impoundment load ratio, (2) number of
publications available for each site, (3) quantity and quality of avail-
able ecological data, (4) extent of literature coverage of fish popula-
tions at each, site, (5) geographical location, and (6) whether or not
"control situations" existed.
ANALYSIS AND DISCUSSION
Analysis of potential impacts on fishery resources in cooling impoundments
are divided into two main themes. The first covers what we have deter-
mined from the data collected from 14 sites in a generic qualitative
manner. The second covers independent qualitative and semi-quantitative
analyses of selected data sets. In this latter part, comparison is made
354
-------
of two cooling impoundments with three nearby control reservoirs, in
which parameters such as game to rough fish ratios, fish condition
factors, and catch per unit effort are treated.
Qua! itajbive Assessment
There are apparently no major detrimental effects from power plant opera-
tion on fish populations inhabiting closed system -cooling impoundments.
However, this conclusion i: accompanied by several limitations.
First and most important, most available studies were highly qualitative
because they were based on nonquantitative evidence. This was due mainly
to sampling design deficiencies.
Second, most available studies were research projects undertaken by
students to complete requirements for advanced degrees. Many projects
were narrow in scope (e.g., dealing with only one aspect of the life
history of a single fish species), and provided little ecological informa-
tion on generic fish population parameters in relation to the ecosystem.
Many available studies had a duration of one year or less, and did not
examine annual seasonal variations in population abundance. Reports
prepared by fish and wildlife agencies were usually not designed to
detect or measure power plant effects on fish populations, but to obtain
data on which to base recommendations for fish management and enhancement
of the sport fishery.
Third, available data outside the southwestern United States was scanty.
The majority of cooling impoundments, including the 14 sites chosen for
our evaluation, were located in Texas where ambient water temperatures
are historically high [1], Fish living in Texas waters have high tempera-
ture tolerances [2]. Thus, conclusions reached concerning the effects of
thermal additions on fish populations in Texas cooling impoundments did
not necessarily apply to .other geographical areas.
A fourth limitation to complete evaluation of possible power plant effects
on fish populations was the absence of data measuring impingement,
entrainnient, and/or chlorination impacts. At the 14 sites we assessed, no
indication was given whether these potential perturbations existed or-not.
The question of thermal effects on reproductive capacity of fish popula-
tions was also not addressed quantitatively, but the topic was occasion-
ally dealt with briefly in a qualitative sense.
Any ecological study concerning the effects of stress on fish populations
must determine if the majority of adults reproduce normally, if a high
percentage of the eggs hatch, and if young fish have a "typical"
survival rate the first year of life. This type of study is important
because sound judgments on whether power plant operations are affecting
the fish populations 'in d cooling impoundment are difficult without
basic information on reproductive success and progeny survival. Survival
and growth of stocked fish is an incomplete answer.
355
-------
We evaluated the appropriateness of each statistical test applied by
various authors in individual studies. We looked to see: (1) if a
correct test was applied to the correct data set, (2) if there were
deficiencies in sampling design, (3) if there was a better way to test
the data, and (4) if it would be beneficial to retest the data. In most
cases the appropriate statistical test was applied. However, since
fishery statistical methods have evolved rapidly, in some cases a more
appropriate test was available for application.
We usually did not analyze the data available in various reports again
for two reasons. First, a better or newer statistical test, in our
judgment, would probably not have altered the results due to limitations
imposed by deficiencies in sampling design. Second, in order to make
quantitative assessments, several assumptions were required that out-
weighed in a negative sense the weight and power of the quantitative test.
Therefore, we attempted to statistically evaluate the original data, or
analyzed data not previously tested statistically, in only a few cases.
Analysis of Specific Data Sets
We obtained fishery data sets for two heated reservoirs, Lake Colorado
City and Lake Nasworthy, that could be compared with nearby unheated
reservoirs. Lake Colorado City was compared with Champion Creek Reservoir
(control). Lake Nasworthy was compared with Twin Buttes Reservoir and
San Angelo Reservoir (controls).
Data sets for 8 to 11 years were assembled for each comparison, covering
the period from 1960 to 1975. Semiquantitative (correlation matrices)
and nonparametric (Wilcoxon's Signed Rank Test) statistics were employed
for data evaluations. Data at each location were presented as averages
of one to four sampling trips per year.
A correlation matrix was used to evaluate associations among the variables
for the combined data set, which included data from heated and control
reservoirs as appropriate. Variables tested included: (1) game fish by
number, and weight; (2) rough fish by number and weight; (3) channel cat-
fish and largemouth bass by number, percent number, weight, percent
weight, average weight, and condition factor; (4) game fish to rough fish
ratio by number and weight; (5) average weight of game and rough fish;
and (6) percent of game fish by number and weight.
Variables showing a high correlation coefficient were investigated further
by plotting individual data. Plots were made for the three following
variables in the Lake Colorado City-Champion Creek comparison: (1) game
to rough fish ratio by weight, to number of largemouth bass; (2) game to
rough fish ratio by weight, to percent game fish by weight; and (3) game
to rough fish ratio by weight, to average weight of channel catfish.
For the Lake Nasworthy - Twin Buttes and San Angelo Reservoir comparison,
the following eight variables showed a high correlation value: (1) average
number of game fish, to total number of game fish, (2) average weight of
rough fish, to total rough fish weight per set, (3) average weight of
356
-------
rough fish, to total rough fish weight per set; (3) average weight of
rough fish, to average weight of channel catfish; (4) average weight of
game fish, to average weight of largeinouth bass; (5) average weight of
rough fish, to number of gizzard shad per set; (6) average'weight of
rough fish, to percent shad by number; (7) average weight of rough fish,
to average weight of shad; and (8) average weight of game fish, to
percent number of river carpsucker.
The important relationship that the plots depicted was the comparative
distribution of data points from heated and control impoundments. The
distribution in most plots (Figures 1 and 2) indicated that there were no
major differences in fish population parameters between heated and control
impoundments. Therefore, the underlying relationships were apparently
similar. We did not attach probability levels to the observed correlation
coefficients because probability levels are heavily influenced by the
frequency of zero values in the data sets, which did occur to some degree.
Thus, our correlation matrices approach was general. However, if large
differences between variables actually existed, the correlation matrix
would demonstrate them. We conclude from this analysis that there were no
significant differences between fish populations in heated and control
impoundments.
The correlation matrix evaluation was taken one step further. We decided
that even though two variables did not correlate initially, it was still
possible that a real difference in fish population parameters between the
two water bodies might occur. We chose variables that should show real
differences if they were actually occurring, and conducted additional
tests.
As expected, the variables did not correlate. Furthermore, dispersal of
data points indicated that there were no large differences in fish popu-
lations between the two impoundments. If differences were to occur, a
clumping of data points for each reservoir would be expected. Therefore,
no detectable differences existed for noncorrelated variables.
The final approach we employed in analyzing fish population data was a
nonparametric statistic. Wilcoxon's Signed Rank Test [3] was used to
examine possible differences between variables. Parameters tested were
those where differences due to power plant operation were most likely to
occur. The following parameters were tested between heated and control
reservoirs: (1) game fish number per gill net set, (2) rough fish number
per gill net set, (3) game/rough fish ratio by number per gill net set,
(4) channel catfish "K" factor (5) largeinouth bass :'K" factor, (6)
channel catfish number per gill net set, and (7) largemouth bass number
per gill net set.
No significant differences were found between Lake Nasworthy, Twin Buttes
and San Angelo Reservoirs (Table 2). Significant differences in large-
mouth bass condition factors could not be established because data were
lacking.
357
-------
For the Lake Colorado City - Champion Creek Reservoir comparison, several
significant differences were found (Table 3). The number of rough fish
per gill net set was significant (95% level), indicating a larger rough
fish population in Lake Colorado City. However, the number of game
fish per gill net set between the sites was not significantly different.
Since the rough fish population was larger in Lake Colorado City, the
game to rough fish ratio by number was also significant (99% level).
The channel catfish condition factor was significantly higher at
Champion Creek Reservoir, but total number per gill net set of channel
catfish was significantly higher for Lake Colorado City. Thus, Lake
Colorado City apparently supported a larger population of channel catfish,
but they weighed less for a given length than channel catfish from
Champion Creek Reservoir. Largemouth bass populations, total game fish
per gill net set, and bass condition factors between the impoundments were
not significantly different. Consequently, both reservoirs appeared to
support basically the same viable fish populations although Lake Colorado
City had the larger population of rough fish. According to Texas Parks
and Wildlife Department reports, Lake Colorado City has received heavy
fishery pressure for at least 15 consecutive years, yet sustains a pro-
ductive fishery with very little supplemental stocking.
In summary, we conducted independent statistical evaluations of fish
population data. Available data was first analyzed from a general
perspective (correlation matrix), then investigated that approach in
more detail. We finally employed a specific test to data sets where
expected differences from power plant operation should be readily
revealed. However, the data analyzed were yearly averages of the
variables tested. Fish were often collected just one time annually,
while several collections were made in other years. Moreover, sampling
seasonality varied from year to year and between impoundments. Thus,
the data are not as representative as desired of the fish populations
in each impoundment. Gill net catches over a 15 year period provided
the best available data set, which did allow us to investigate some
potential differences in fish populations between heated and control
impoundments.
On the basis of independent statistical analyses, and the qualitative
judgments and limiting qualifications mentioned earlier, we conclude
that no statistically significant differences existed-between fish
populations of the cooling impoundments and control sites.
Based on generic information reviewed for 14 sites, it appeared that
cooling impoundments represented functioning ecosystems. Fish popula-
tions are apparently maintained without the aid of management stocking
programs in some cases; whereas, fish populations in other cooling
impoundments needed annual supplemental stockings. However, stocking
usually reflected intense fishing pressure where carrying capacity of the
water body was exploited other than deficiencies in that water body for v
normal fish reproduction.
Whether the need for supplemental stocking in cooling impoundments was
caused by extensive fishing pressure, perturbations from the power plant
358
-------
or any of a number of possible other reasons could not be determined.
Stocking needs were probably due to other causes, rather than power
plant induced. Other lakes (without power plants) also required
supplemental stocking to maintain game fish populations, sometimes when
subject to less fishing pressure.
Increased water velocity associated with discharge of heated effluent
apparently influenced fish distribution and abundance more than did
increased temperature, especially during the spawning season, for most
fish species. Fish were free to seek or avoid discharge areas as their
preferences dictated, regardless of whether the controlling factor was
temperature or current. The increased circulation in cooling impound-
ments caused by the power plant discharge was generally considered to
benefit aquatic life, but there was no evidence provided to prove or
disprove this hypothesis.
Without quantitative evidence to judge whether detrimental effects were
occurring in cooling impoundments, our best qualitative assessment told
us that there were no overriding perturbations on fish populations from
power plant operations in cooling impoundments. However, this conclusion
was reached without quantitative evidence. Until studies are designed and
implemented to produce data amenable to quantitative assessment, only
qualitative interpretations can be made.
359
-------
REFERENCES
1. Drew, H. R. and J. E. Til ton. 1970. Thermal requirements to protect
aquatic life in Texas reservoirs. J. Wat. Pol 1r Contro1 Fed. 42:
562-572.
2. Strawn, K. and J. E. Dunn. 1967. Resistance of Texas salt and
freshwater marsh fishes to heat death at various salinities.
Texas J. Sci. 19:57-76.
3. Hollander, M. and D. A. Wolfe. 1973. Nonparametric Statistical
Methods. Wiley and Sons, New York, 503 pp.
360
-------
TABLE 1
RESULTS OF SCREENING POTENTIAL SITES
FOR ECOLOGICAL DATA SYNTHESIS AND ANALYSIS (1)
Generating Plant
(Utility
Asheville
Carolina Power & Light
Big Brown
Texas Power & Light Co.
V. H. Braunig
San Antonio Public Service Bd.
Eagle Mountain
Texas Electric Service Co.
Handley
Texas Electric Service Co.
Lake Catherine
Arkansas Power & Light Co.
Morgan Creek
Texas Electric Service Co.
North Lake
Dallas Power & Light Co.
San Angel 9
West Texas Utilities
Sim Gideon
Lower Colorado River Authority
Striker Creek
Texas Power & Light Co.
Sundance/Wabamun
Calgary Power Ltd.
Thomas Hill
Assoc. Electric Coop.
Tradinghouse Creek
Texas Power & Light Co.
Load Ratio
Cooling (Surface)
Impoundment Acres/MWe
Lake Julian
Fairfield Lake
Braunig Lake
Eagle Mountain
Lake
Lake Arlington
Lake Catherine
Lake Colorado
City
North Lake
Lake Nasworthy
Lake Bastrop
Striker Creek
Reservoir
Lake Wabamun
Thomas Hill
Reservoir
Tradinghouse Creek
Reservoir
0.8
2.0
1.5
13.0
4.4
2.6
1.9
1.2
12.0
1.6
3.4
23.2
9.6
1.5
No. of
Reports
13
6
9
20
16
6
20
13
28
10
6
17
18
7
I)
v,, .„»., ,u for site slection include: (a) freshwater environment;
(b) low load ratio, usually <">0; (c) semienclosed water bodies;
(d) once-through cooling cycle; and (e) quality and quantity of
ecological information.
361
-------
TABLE 2
RESULTS OF WILCOXON'S SIGNED RANK IEST APPLIED TO FISH DATA FROM
LAKE NASWORTHY (HEATED) AND TWIN BUTTLS RESERVOIR AND SAN ANGELO RhSERVOIR (CONTROLS)
Gamefish by- number per set
Rough fish by number per set
to
Year
1961
1962
1963
1964
1955
1966
1967
1963
1969
1970
1971
Year
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
Lake Nasworthy
5.13
2.20
9.53
12.33
10.00
18.33
19.33
25.06
17.33
19.78
Game/rough
Lake Nasworthy
0.24
0.28
0.34
2.96
0.21
0.35
0.48
0.79
0.33
0.58
Twin Buttes
—
--
—
15.50
12.00
7.12
12.28
15.17
9.17
9.50
fish ratios
Twin Buttes
--
--
—
0.69
0.35
0.19
0.46
0.47
0.20
0.25
Rank
--
—
--
-2
-1
7
3
5
4
6
NS
by numbers
Rank
—
--
—
7
-3
4
1
5
2
6
NS
San Angelo
23.20
2.20
5.30
—
9.00
3.17
14.00
10.78
24.33
8.00
San Angelo
0.48
0.14
0.29
—
0.37
0.12
0.56
0.31
0.57
0.96
Rank
-9
1
3
2
8
4
7
-5
6
NS
Rank
-5.5
3
1
—
-4
5
-2
9
-6.5
-8
NS
Lake Nasworthy
21.77
8.00
28.37
4.17
47.25
51.67
40.67
31.72
51.75
33.94
Lake Nasworthy
0.18
0.40
0.21
1.82
0.40
0.51
0.80
0.50
0.44
0.54
Twin Buttes
--
--
—
23.75
34.67
37.12 -
26.61
32.50
44.50
37.38
Game/Rough Fish
Twin Buttes
—
--
—
0.07
0.20
0.40
0.44
, 0.76
0.89
0.37
Rank
--
—
--
-7
4
6
5
-1
3
-2
NS
Ratios by
Rank
--
--
--
7
3
1
5
-4
-6
2
NS
San Angelo
18.20
15.20
18.27
—
24.25
25.33
25.00
34.50
42.33
3.33
Weight
San Angelo
0.24
0.09
0.42
—
0.30
0.21
0.43
0.33
0.17
1.68
Rank
-9
-2
4
--
6
7
5
-1
3
8
NS
Rank
-1
7
-4
--
2
6
8
3
5
_q
NS
-------
TABLE 2 (CONTINUED)
1966
Ye'ar Lake Nasworthy Twin Buttes Rank San Angelo Rank Lake NasworthyTwin ButtesRankSan AngeTo Rank
1961 — - — 1.56 — 2.25 C C
1962 1.04 -- -- 1.46 -8 2.90 -- a 2.52 a
1963 1.55 — — 1.60 -1.5 2.68 — n 2.90 n
1964 n n
1965 1.92 2.08 -3 -- -- 2.49 2.87 o — o
1.81 1.56 5 1.35 1 — 2.30 t — t
1967 1.92 1.74 4 2.06 4.5 2.68 2.58
1968 2.12 1.74 7 1.87 7 2.89 2.85 t 3.45 t
1969 1.90 1.62 6 1.82 3 2.68 e 3.09 e
1970 1.93 1.86 1 1.79 4.5 2.65 2.25 s 2.35 s
1971 1.89 2.01 -2 1.94 -1.5 2.91 3.36 t — t
-------
TABLE 3
RESULTS OF WILCOXON'S SIGNED RANK TEST APPLIED TO FISH DATA FROM
LAKE COLORADO CITY (HEATED) AND CHAMPION CRtEK RESERVOIR (CONTROL)
Game fish by_number per set
Rough fish by number per set
Year
I960
1961
1962
1953
1964
1965
1966
1967
1963
1969
1970
1971
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
Champion Cr
?.9
1.7
8.5
13.2
26.5
13.2
20.4
9.2
6.3
Channel
1.82
—
1.71
1.88
2.30
2.03
1.93
2.04
1.94
eek Lake Colorado
4.7
16.3
12.7
10.8
12.5
9.8
12.4
13.4
8.3
catfish "K" factor
1.75
1.47
1.52
1.69
1.65
1.77
1.85
1.68
1.76
City Rank
-1
-9
-4
3
8
6
5
-7
-2
NS
1
—
4.5
4.5
8
6
2
7
3
**
Champion Creek
2.4
2.5
1.8
2.3
5.0
5.2
11.4
23.2
14.9
Largemouth bass "K"
2.94
1.97
2.02
2.62
2.13
2.47
2.17
2.34
2.84
Lake Colorado City
4.0
8.0
25.7
12.8
23.5
13.4
17.3
19.1
15.6
factor
2.74
2.56
2.28
2.48
2.47
2.57
2.88
2.54
2.76
Rank
-2
-2
-4
-9
-7
-8
-6
-5
3
-1
*
4.5
-8
-6
3
-7
-2
-9
-4.5
1
NS
* = Significant at 953 level
** = Significant at 99% level
364
-------
TABLE 3 (CONTINUED)
Catfish by number per set Bass by number per set
Year Champion Creek
1960
1951
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
0.23
—
0.17
0.75
1.67
1.11
1.65
2.00
1.19
Lake Colorado City Rank
1.58
4.56
4.25
2.00
1.50
2.61
3.65
4.25
2.20
-4
--
-8
-3
1
-5
-6
-7
-2
**
Champion Creek
2.15
0.50
1.33
0.25
1.33
0.67
0.59
0.50
0.06
Lake Colorado City Rank
6.33
5.00
2.62
0.25
1.33
0.44
0.47
0.55
0.56
8
-9
-7
1.5
1.5
5
4
-3
-6
NS
Game/rough fish by number
Year Champion Creek
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
0.6
1.3
2.3
5.2
2.1
1.4
1.4
1.0
1.2
Lake Colorado
0.8
0.6
0.4
0.6
0.4
0.6
0.6
0.5
0.3
City Rank
-1
3
8
9
7
4.5
4.5
2
6
365
-------
O
O
3.60
2.70
1.80
3 0.90
Cd
LU
0.00
D
AAD A A
DO •
a A p o ®
a • •
0.0
D
6.0 12.0 18.0 24.0
GAME FISH BY NUMBERS
30.0
Figure 1
Average Weight of Game Fish Plotted Against Game Fish by Total Numbers for
'lake Nasworthy (0), Twin Buttes Reservoir (A) and San Angelo Reservoir (D).
X
v>
£ 2.40
o
1
i 1.60
0
LU
^
LU
< 0.80
a:
LU
<
0.00
A
©
D A
e
D
a
•
a a
a
- A
A A • •
• D
_
1 1 1 1 1
0.0
6.0 12.0 18.0 24.0
AVERAGE WEIGHT CATFISH
30.0
Figure 2.
Average Weight of Rough Fish Plotted Against Average Weight Catfish for
Lake Nasworthy (0), Iwin Buttes Reservoir U) and San Angelo Reservoir (D)
366
-------
A SIMPLE METHOD OF PREDICTING PLUME
BEHAVIOR FROM MULTIPLE SOURCES
L.D. Winiarski, W.E. Frick*, and A. Carter
Corvallis Environmental Research Laboratory
U.S. Environmental Protection Agency
Corvallis, Oregon 97330 U.S.A.
ABSTRACT
A method for predicting the behavior of plumes from multiple sources with
the aid of a computer program or nomograph is illustrated and compared with
field data. Sources close together are treated as an equivalent single
source which is designed to have the same efflux of mass, momentum and
energy. The perimeter of the equivalent single source is taken to be the
same as the total perimeter of the multiple sources. The apparent area of
the actual sources as projected normal to the wind is used with the equiva-
lent source to determine the wind-induced entrainment.
INTRODUCTION
Complex plume behavior can be predicted with simple, logical numerical
procedures. The key to accurate yet inexpensive computation lies in the
ability to model the mass entrainment and horizontal momentum transfer
between the wind and the plume. The simple assumption that all the wind
impinging on the plume (i.e. passing through the wind projected area) is
entrained and adds its momentum to the plume has proved to be a useful
concept for single source plume predictions (1,2). It is logical to assume
that a similiar concept could be extended to deal with multiple source
plumes.
ASPECT FACTOR
The results that follow are based on two hypotheses: (1) the main difference
between the multiple plumes and a hypothetical single "equivalent plume" with
the same mass, momentum and energy flux is that the projected area and
peripheral area of the single, equivalent plume differs from that of the
multiple tower arrangement, and (2) after the plumes merge they can be approxi-
mated by an axi-symmetric equivalent plume.
A single plume model can be used to approximate multiple plume behavior by
introducing an Aspect Factor (AF) to account for the differences in pro-
jected area computation before the plumes have merged.
AF _ actual projected cylindrical area of multiple plumes
projected area of equivalent single plume
Now affiliated with the Oregon Department of Transportation, Salem.
367
-------
Aspect Factor is used as a coefficient multiplying the projected area of
the cylindrical part of the equivalent plume. This allows the computation
of the projected area to approximate that of the actual multiple source
arrangement. Although AF changes along the trajectory, consideration of the
initial geometry and wind direction often indicates how this variation can
be approximated. Once the relationship between AF and equivalent radius is
determined, it can be input as a Itable to the single plume model. The follow-
ing example will illustrate this.
EXAMPLE
The model predictions will be compared with data from three Neurath natural
draft cooling towers (3, 4). Each tower radius-is 22.4 m. The centers of
the towers form an equilateral triangle 114 m on a side (Figure la). An
equivalent source, radius 38.8 m, is also shown with its center at the
centroid of the three towers. The calculation shows AF = 1.73 initially.
Aspect Factor is constant until the upwind plume begins to "shadow" the
downwind plumes. This occurs when the individual plume radius is about 28.5 m.
At that point the equivalent plume radius is 49.35 m (Figure Ib). Once
shadowing occurs, the value of AF changes. The next reference point occurs
when the plumes begin to merge. From the geometry of Figure Ic this would
occur when the individual plume radius is about 57 m (equivalent radius
98.7 m). Aspect Factor is calculated to be 1.15 at this point. A linear
variation of AF with equivalent radius is assumed between points b and c.
The last significant point (AF = 1.0) is where the plumes have merged into
a single, approximately round plume. The precise point is difficult to de-
fine. In this case a simple approximation would be to consider an extrapola-
tion until AF = 1.0, which occurs at an equivalent radius of about 111.6 m.
A more general prescription for estimating this last point where AF = 1.0
is to calculate the radius where the equivalent plume would just encompass
the plume of the tower furthest from the centroid. However, in this case
there is no practical advantage in using this procedure rather than the
simple extrapolation shown, because AF would be so close to 1.0 in this
region.
RESULTS
The model predictions are compared with data in Figures 2 through 8. An
improvement has been made in the numerical procedure for calculating the
psychrometric variables otherwise the model is the same as the one discussed
in Reference 2. Except for high humidity cases, the new procedure yields
practically the same results as presented previously (2). When the humidity
is high, the point of saturation (visibility limit) is extremely sensitive
to the calculation procedure. This can have a noticeable effect on visible
plume length.
368
-------
NOMOGRAPHS
In order to make estimates of plume behavior as well as see the overall
effect of plume variables we have prepared a series of nomographs (6) like
that shown in Figure 9. In addition to the aspect factor, the key para-
meters are
Vft
velocity ratio, K =
Froude number, Fr =
wind speed
V
Pa - P Dg~
V
Neutrality number, N = - -
VA6 g Df
AZ T
where V is the initial velocity, p and p are the ambient and plume densi-
ties respectively, D is the source aiameter, g is the acceleration due to
gravity, A6/AZ is the potential temperature gradient, and T is the ambient
temperature. The neutrality parameter is a new dimensionless parameter
we have introduced to account for non-adiabatic conditions.
Large values of N (e.g., T? ~ 0) mean nearly adiabatic (neutral) conditions.
Generally, nondimensional trajectories are found to be the same if N, Fr
and K are the same. This is subject to the condition that density-tempera-
ture relationships of the fluids involved behave similiarly within the per-
tinent temperature ranges. The advantage of achieving approximate similarity
is considerable, but one must be aware of potential problem areas (6).
To illustrate, the parameters of the Neurath run N49 (Figure 5) were Fr = .5,
V = 3.5 m/s, N = 3.2. The nomograph (Figure 9) parameters approximate
tnese. Because of similarity it can be compared to run N49 even though it
is generated from a different set of conditions. A plume trajectory for
K = .5 (based on an average value of the extrapolated wind speed) gives a
rise of about 3 equivalent diameters (230 m) at a distance of 15 diameters
(1160 m) which corresponds approximately with Figure 5.
The nomographs show the general trend of the important variables. For a
given ambient temperature gradient and set of tower exit conditions, each
K curve represents a center!ine plume trajectory at a given ambient wind
speed. Relatively speaking, K = 10 is a low wind condition, K = .5 is
high. In this example, the atmospheric temperature gradient is strongly
stable (i.e. N is small) so the plume reaches an equilibrium height in a
relatively short distance even in a low wind situation.
369
-------
SUMMARY
Three significant points of this study are:
(1) The aspect factor concept appears to be a simple, yet effective method
of taking into account multiple source geometry with a single source
plume model.
(2) A similarity parameter (N), akin to the Froude number, can be used to
account for non-adiabatic atmospheres.
(3) Used judiciously, nomographs can be very helpful in obtaining approxi-
mate answers to atmospheric plume problems;-
REFERENCES
1. Winiarski, L.D. and Frick, W.E. Cooling Tower Plume Model, USEPA Eco-
logical Research Series, EPA-600/3-76-100, September 1976.
2. Winiarski, L.D. and Frick, W.E. "Methods of Improving Plume Models,"
Cooling Tower Environment--1978 Proceedings Water Resources Research
Center, University of Maryland, May 1978.
3. Vereins Deutscher Ingenieure. "Untersuchungen an einem Naturzug -
NasskUhlturm." Fortschritt-Berichte der VDI Zeitschriften. Reihe 15.
Nr. 5. Juli 1974.
4. Policastro, A.J., Carhart, R.A. and Devantier, B. Validation of Selected
Mathematical Models for Plume Dispersion from Natural-Draft Cooling
Towers. Presented at the Waste Heat Management and Utilization Conference,
Miami Beach, 9-11 May 1977.
5. Winiarski, L.D. and Frick, W.E. Atmospheric Plume Nomographs with Computer
Model for Cooling Tower Plumes, USEPA Interagency Energy-Environment
Research and Development Series, manuscript.
6. Frick, W.E. and Winiarski, L.D. Why Froude Number Replication Does not
Necessarily Ensure Modeling Similarity. Second Waste Heat Management and
Utilization Conference, Miami Beach, 4-6 December 1978.
370
-------
a) Initial separate plumes
equivalent plume,
WIND
2.4m
\
—-1 T3
Q ~I.15
(initial radius)
b) Shadowing starts
c) Merging begins
x*
57.0m
3x2x28.5_
2x49.35
(a) (b)
38.8 49.35
(c)
98.7 111.6
— Equivalent radius(m) —
Figure 1. Approx. change in Aspect Factor as Neurath plumes grow.
371
-------
o
•z.
1
o
UJ
>
o
CD
§100
0
LEGEND
*— A
of actual plume outline
Predicted visible plume
Predicted invisible plume
Predicted centerline
humidity (0-1.0)
.50
1.0
Wind speed
0
100 400
DISTANCE FROM TOWER (m)
Figure 2. Neurath run N15.
0(m/s) 10 20
i i i
280(°K)285 290
.50
1.0
200
1
cs
100
0 100
DISTANCE FROM TOWER (m)
200
0(m/s) 5 10
280(°K)285290
Figure 3. Neurath run N34.
372
-------
LEGEND
Trace of actual plume outline
500
*—* Predicted visible plume
• • • Predicted invisible plume
Predicted center!ine
x Ambient humidity (0-1.0)
s Ambient temperature
*• Wind speed
o
C£
c?
1001-
0 400
DISTANCE FROM TOWER (m)
Figure 4. Neurath run N37.
2000 r
1500
1000
500
100
O
C£
CD
o
- CQ
:r
i—f
UJ
0 400 900
DISTANCE FROM TOWER (m)
1900
.50
1.00
•0(m/s) 5 10
235(°K) 290 295
.70
1.00
0(m/s) 12 24
t i i
270(°K)272.5 275
Figure 5. Neurath run N49.
373
-------
LEGEND
Trace of actual plume outline
o 500
o
CD
-------
LEGEND
-x-
1500
o
z
§1000
U-l
>
o
CO
Trace of actual plume outline
Predicted visible plume
Predicted invisible plume
Predicted centerline
Ambient humidity (0-1.0)
•Ambient temperature
•Wind speed
I-
C3
tmm4
Ul
X
500
100
.50
1.0
400 900 1400
DISTANCE FROM TOWER (m)
1900
0(m/s) 15 . 30
280
Figure 8. Neurath run N67.
375
-------
CKI
ASPECT F
- A8/AZ-.
V0= 3 sti/s
89 12 13 14 15 16 17
Figure 9. Nomographs for atmospheric plumes.
-------
MODELLING NEAR-FIELD BEHAVIOUR
OF PLUMES FROM MECHANICAL DRAFT
COOLING TOWERS
T.L. Crawford and P.R. Slawson
Mechanical Engineering Dept.
University of Waterloo
Waterloo, Ontario, Canada
ABSTRACT
For the purpose of analysing plume behavior from cooling towers it is con-
venient to distinguish two principal regions: the near-field and the far-
field. In the former the plume is strongly influenced by the source geo-
metry, effluent release height, local terrain and the presence of nearby
structures of towers making mathematical description extremely complex.
By comparison, in the far-field the plume behavior is relatively insensi-
tive to details of its origin and mathematical formulation is much better
understood and developed. The near-field behavior of plumes from mechani-
cal draft cooling towers is particularly complex and often dramatically
affected by the proximity of adjacent towers. This results in frequent
downwash of the tower effluent which then leads to re-ingestion of a sig-
nificant fraction of its own effluent (recirculation) or of the effluent
of an upwind tower (interference) thus affecting tower performance.
Single camera photographic observation of the near-field behavior of plumes
from the twin mechanical draft cooling towers at plant Gas ton were made.
A three-dimensional numerical model of the near-field plume behavior has
been developed. A general overview of the model and a comparison of model
prediction on the near-field plume behavior with an observed time-mean
plume are given.
INTRODUCTION
Physically realistic vapor and drift plume models are required in the en-
vironmental assessment of heat dissipation technologies such as mechanical
and natural draft cooling towers, spray ponds, and spray canals. In mo-
deling vapor and drift plume behavior for these evaporative cooling de-
vices it is convenient to distinguish two principle regions, the near-
field and the far-field. The near-field plume (within about five charac-
teristic lengths downwind) is complex in behavior because of the strong
influence of source geometry, effluent release height, wind shear, local
terrain, and nearby structures on the flow field. These influences disturb
the near-field flow to a turbulent state that is poorly understood. As
a result of the near-field complexities, development of a realistic mathe-
matical modeling technology has-been slow. In contrast, the far-field
* Presently with the AIR QUALITY BRANCH of the Tennessee Valley Authority
377
-------
plume (beyond about 10 characteristic lengths downwind) is relatively well-
behaved and relatively insensitive to details of its origin. The mathe-
matical formulation of far-field plume behavior is much better understood
and developed. In the intermediate region, the complex near-field plume
evolves to the simpler, self-similar far-field plume. The boundary condi-
tions for far-field plume models depend on the integral effect of the near-
field detail.
The importance of the near-field region to far-field plume behavior varies
greatly and depends on the extent of the near-field disturbances. Recent
full-scale studies of various cooling tower complexes have illustrated the
importance of the near-field disturbances on far-field plume behavior.
Slawson et al. (1975 and 1978) showed that, for a rather simple natural
draft cooling tower geometry, up to a 50-percent depression in far-field
plume trajectories resulted from near-field tower downwash. A recent study
of twin 9-cell mechanical draft tower plumes illustrated a dynamically more
complex near-field plume behavior with subsequent severe effects on far-
field plume behavior. A mathematical model of near-field plume behavior
would be useful for specifying boundary conditions for far-field plume mo-
dels and for assessing the effects of recirculation and interference on
tower performance.
Modelling Considerations
To study the highly complex behavior of such near-field flows, a computer
model capable of solving the time-dependent, three-dimensional equations
governing fluid motions greatly distorted by structures and none-passive
plumes was developed (Crawford 1977). The validity of the model was estab-
lished by comparison with a variety of physical modeling studies. Presented
herein is a comparison of initial model calculations with one experiment from
the full-scale study of twin 9-cell mechanical draft cooling towers. A
brief overview of the modeling methodology is presented here and further de-
tails may be found in Crawford (1977).
The computer model solves the full nonlinear equations that specify the mean
transport and diffusion of fully turbulent, three-dimensional incompressible
flows. These equations are, the mass conservation equation,
3U
3X.
= 0
the energy or diffusion equation
39
3(eu_.)
3X.
j
3X.
378
-------
and the 3 component momentum equations
K
g - K
•
3* 3X.. p 9Xi T • 3X. 8X.
J J
where the Einstein subscript notation is used. In these equations, the den-
sity p, hydrostatic pressure P, potential temperature 6, temperature T, and
velocity components % may be functions of the position coordinates X± and
of time t. The parameters Km and K^ are the eddy exchange coefficients for
heat and momentum which were specified as a function of the vertical coor-
dinate Xs. (K varied linearly with height from similarity theory.)
In order to solve the above governing equations for a particular flow, it
is necessary to incorporate the pertinent initial and boundary conditions
which make the problem determinate. The correct specification of these
can become a significant part of the overall solution problem.
The numerical solution of the governing equations is accomplished by app-
lying finite difference approximations to them. The solution method is
fully implicit, employing velocity components and pressure as the main de-
pendent variables. These variables are computed for one of four staggered,
interlacing grid systems which fill the domain of interest. A hybrid cen-
tral-upwind difference scheme is employed to improve accuracy and aid con-
vergence. The solution of the flow dynamics proceeds in a successive
guess-and-correct fashion. In the first part of each cycle, intermediate
velocities are calculated from the momentum equations using the pressure
distribution from the previous cycle, where a uniform pressure distribution
is assumed initially. The second step adjusts these velocities and press-
ures such that mass conservation is satisfied. Following this balance of
mass, other variables of interest are subject to their respective conser-
vation equations.
Although it is possible to include the balance equations for liquid water
and water vapor in the above set of equations, in order to predict the
extent of the vapor plume. They were not ix: luded in the present model for
the sake of simplicity. Since the near-field upwind, cavity and wake re-
gions are strongly elliptic in nature, further simplifications of the
equations and the subsequent solution procedure was avoided.
The computer program load module, including plotting routines and array
storage for the 30 x 13 x 16 grid used, requires 386 K bytes of storage.
Storage of the grid required about 292 K of this 386 K. The execution
time for simultaneous solution of the flow field and the energy equation
about the two towers and plotting of all results was 416 seconds on an
IBM 270/165. The solution time for this simulation is increased due to
the strong coupling of equations by the buoyancy force term.
379
-------
Gaston Mechanical Draft Cooling Tower Study
A number of full-scale experiments investigating near- and far-field con-
densed plume behavior from twin mechanical draft cooling towers were con-
ducted by Slawson and Crawford (1976). Some details of the experimental
work was given by (Champion et al, 1977). Only near-field observations
are pertinent to the present work. Subsequent experimental details may be
found in a limited circulation report soon to be published.
The twin 9-cell cross-flow towers studied,serve an 800-mw unit of the
Gaston generating station in Wilsonville, Alabama. Each tower has an over-
all length of 100 m, width of 22m, and height of 17m. The towers are ori-
entated one behind the other with parallel axes spaced about 100 m apart.
Each of the nine cells within a tower has a 200-hp motor driving a 9 m dia-
meter fan to give a plume exit velocity of about 8 m/s.
The objectives of the field experiment were to (1) quantitatively document
visible plume behavior, (2) measure appropriate meteorological variables,
and (3) measure tower fluxes of heat and mass. The visible or condensed
plume behavior was quantitatively documented by a single camera photographic
technique following that given by Halitsky (1961). The meteorological data
consisted of: temperature profiles (dew point and dry bulb) from an in-
strumented aircraft, wind speed and direction profiles from dual-theodolite
tracking of 30-gm pilot balloons and wind and temperature data from instru-
mented on-site micrometeorological towers. Although a variety of source
measurements were taken, only exit velocity and temperature measurements
are used herein. Exit temperature was obtained from thermistor traverses
of cell exits. Normal to the line of towers the terrain is relatively flat
with variations of about 10 meters.
Figure 1 illustrates the observed wind speed and temperature profiles for
the morning of January 15, 1976. The wind direction was normal to the to-
wers and the observed profile was characterized by a UA of 1.07 m/s and Zo
of 1.47 m. At 0845 hours the meteorological tower AT indicated a transi-
tion from slightly stable to slightly unstable stratification. The dash-
line plotted in the symmetry plane of figure 2 illustrates the observed
time-mean near-field condensed plume boundary.
Initial and Boundary Conditions
The inflow velocity profile (log-law) was defined in the usual manner for a
neutral atmospheric boundary layer characterized by UA = 1.07 (m/s) and
a roughness length of ZQ = 1.47 m. Table 1 summarizes the initial and boun-
dary conditions for flow about the twin towers. Simulations were run with
a 30 x 13 x 16 grid which was refined about the towers. The towers were
located about the Y+ grid boundary which was defined as a symmetry plane.
Use of a symmetry plane in this manner significantly reduces the grid sto-
rage requirement. The grid extended upwind of the first tower X/L = 5,
downwind of the second tower X/L = 6.6, above Z/H =5.25 and to the side
Y/L = 3.6, where L is the tower half-length and H is its height.
380
-------
The primary difficulty of this problem is its three-dimensional nature and
the difficulty in applying boundary conditions. Boundary conditions about
the towers were incorporated by significant software modifications in the
main solution control subroutine. See Crawford (1977) for further detail.
Figures 3 and 4 illustrate the significance of observed and simulated tower
downwash and flow disturbance. The coarse nature of the grid used is also
illustrated by the "+" marks of figure 4. Four horizontal and three ver-
tical grid lines intersect each tower. Little refinement in the grid is
possible with present generation computers since three-dimensional grid
refinement is proportional to the cube root of grid points added. Thus
at present is is difficult to significantly extend the physical scale of
the computer simulation in order to model more of the near field plume.
One possible method of extending the modelling domain would be to hybri-
dize a three-dimensional numerical model and the one-dimensional integral
plume theory. This could be done by solving the one-dimensional simulated
flow-field as a boundary condition.
Comparison of Numerical Simulation With Observations
Figure 3 compares the near-field observed plume boundary and simulated tem-
perature distribution about the twin towers. This comparison is primarily
qualitative since temperature contours cannot be rigorously compared to
plume condensation boundaries with the present program. Correct computation
of the condensation boundary would require adding conservation equations
for vapor and liquid water as indicated earlier. The condensed boundary is
usually defined where the liquid water content reaches some small value or
zero. A simplified method for estimating this boundary from the computer
simulated plume temperature field is as follows.
Figure 2 illustrates the saturation curve at a given pressure for air over
a range of air temperatures. The point T represents the condition of air,
assumed to be saturated, leaving the cooling tower, while point A is the
ambient condition of the final condition of the plume after infinite di-
lution. The line T-A is the assumed intermediate plume conditions at vary-
ing levels of dilution. At plume temperatures between T and C the plume is
supersaturated or condensed, and between C and A "dry". The visible bounda-
ry is assumed to occur at C corresponding to 2° C which agrees qualitatively
well with the 2° C isotherm of figure 3 predicted by the computer model.
381
-------
BIBLIOGRAPHY
Champion, E.R., C.H. Goodman, and P.R. Slawson, "Field Study of Mechanical
Draft Cooling Tower Plume Behavior". Waste Heat Management and Utilization
Conf., (1977).
Crawford, T.L., "Numerical Modeling of Complex Two- and Three-Dimensional
Flow andDiffusion Problems in the Natural Air Environment, Ph.D. thesis,
Mechanical Engineering Department, University of Waterloo, (1977).
Halitsky, J., "Single Camera Measurement of Smoke Plumes", Int. Journal
of Air and Water Pollution. V. 4, No. 3/4, pp 185-198 (1961).
Slawson, P.R. and J.H. Coleman, "Natural-Draft Cooling-Tower Plume
Behavior at Paradise Steam Plant", Waste Heat Management and Utilization
Conf., (1977).
Slawson, P.R. , J.H. Coleman, and J.P. Blackwell, "Natural Draft Cooling
Tower Plume Behavior at Paradise Steam Plant", (Part I), E-AQ-76-1-
Tennessee Valley Authority, Chattanooga, Tennessee, August (1975).
Slawson, P.R., J.H. Coleman, and J.W. Frey, "Natural Draft Cooling Tower
Plume Behavior at Paradise Steam Plant", (Part II), Tennessee Valley
Authority, Chattanooga, Tennessee, (1978).
Slawson, P.R., T.L. Crawford, C.H. Goodman, E.R. Champion, Jr., "Plume
Behavior From Mechanical Draft Cooling Towers at Plant E.G. Gaston",
Part I," to be published.
382
-------
TABLE I
SUMMARY OF BOUNDARY CONDITIONS
FOR FLOW ABOUT TWIN MECHANICAL DRAFT COOLING TOWERS
Boundary
Type
Inflowb (X_)
Outflow (X+)
Freestream (Y_)
Symmetry (Y )
Ground (Z_)
Freestream (Z )
Tower Faces
Dependent Variable
U
U.K ln(Z/Z )
* o
dU _ n
dX~ °
U
GO
dU - 0
dY ~ °
0°
U
00
d
V
0
dV _
dX" °
dV _
dY ~ °
0
oc
V
oo
0
w
0
dw - n
dX ~ °
dW.O
dY U
dw - n
dY ~ °
0
w
oo
d
T3
0
41-n
dX " °
0
41=0
dY U
41=0
dZ
dT _ n
dZ " °
d
Notes:
a. The towers were a significant source of thermal energy.
b. The inflow boundary condition is also the initial condition for the
30 x 13 x 16 grid. The velocity profile was logarithmic and character-
ized by V. = 1.07 m/s and Z = 1.47 m.
x o
c. The boundary condition for U and V at the ground was set with a wall
function having a roughness coefficient equivalent to Z .
d. Tower inlet and exit conditions were set to match observed momentum
and energy fluxes.
383
-------
Figure 1 OBSERVED TEMPERATURE AND WIND SPEED PROFILES ON JANUARY 15, 1976, AT WILSONVILLE, ALABAMA
300
250
200
150
100
50
X 0
DEW POINT
nx
A OX
A >B
_tFCEOT)
Q 0832
X 0850
A 0905
A UX
A x a
A »
DRV BULB
»A
* A
MA
xn A
nx A
D X A
.12 -10 -8
-6 -4 -2
TEMPERATURE (C°)
300
2JQ
200
150
o 100
50
/
T/TT
1
*/
I
LEGEND
a 0835 PILOT BALLOON DATA
a 0850 PILOT BALLOON DATA
X 0905 PILOT BALLOON DATA
T TOWER DATA
5 6 7
WIND SPEED (ra/s)
TLC
-------
Figure 2 ESTIMATION OF CONDENSED PLUME BOUNDARY
25
20
O
0 15
SATURATED AIR
0 5 10 15 20 25
TEMPERATURE (°C)
30
385
-------
Figure 3 COMPARISON OF OBSERVED PLUME BOUNDARY AND SIMULATED TEMPERATURE
DISTRIBUTION ABOUT GASTON'S WIN MECHANICAL DRAFT COOLING TOWERS
\A\A\\\\\\\\\\\.\\\\\\ \\A\\\\\\\\\\\\\\\\ \\A
X-Y PLRNE TEMPERflTURE (D-C) flT 22M
\\\\\\\N\N\\\\\
\ \ \ \ \ XX \ \ NX \ \\ \
\\\\\X\\\\\\\X N
1-1 PLASE TEMPEMTURE (0-C)
D-D CROSS SECTION
\ \\ \ \ \\\\\\\\\\
386
-------
Figure 4 SIMULATED VECTOR FLOW FIELD ABOUT GASTON'S TWIN MECHANICAL DRAt,
COOLING TOWERS
Symmetry Plane
-V. X \ * +
A \ \ 4 f
^ n / j;
r/ / / / / '
/ / / s '
]\
/ /
f
t
/
2 B/3
Ground
Plane
»'
I
J-
-#• ^
f
I
J- .
u
CROSS SECTION A-A
\ \ \\ \ \ \ \
CROSS SECTION C-C
X \ \
SCALE
50 METERS
CROSS
t
\
\
\
X
>.
>K
~*v
-H^,
-4-
-«-
\ \
CROSS
Tt
X
\
X
>v
-*.
-*.
•>*.
-t-~
^J^
SECTION B-B
t t
\
I
"^*.
"~^->_
*+-
X \ \
\
\
X
X
^
--*^.
*^~*
•^r
* * ' Mil
• - 1 f '• i
^ III
' "*• * *— **^ jff /,, jt
^^4_u^r:^: $ \
~~*r~\ ~t;:\~-*T- * \
-+- — 1~ — *- -»- -i- + i
\ \ \ N N \ N
1 m/s
SECTION D-D |
\
X
\
X
X
\
^g
"^+-^
---^r
T
\
X
X
X
•— »-,
-i^r~*'
^ * * * * t 1
\V-4-Sffl
Nv.-*. -+. * / / '
^-1 — t-^f * i\
H ~| 1 .^ ^ /
T*^fc -r*w r~-^~ jr f 1
•^-1 _^- | jj | .,_ | " jf f *
~^S^-U^^ / '
h-eg t • t " 1 " ^+-* > 1
\ \
387
-------
MECHANICAL-DRAFT COOLING-TOWER PLUME BEHAVIOR
AT PLANT GASTON
P.R. Slawson
Mechanical Engineering Department
University of Waterloo
Waterloo, Ontario, Canada
ABSTRACT
Field studies on the twin mechanical-draft cooling-tower plume behavior at
the Gaston Steam Plant, Willsonville, Alabama, were conducted during Febru-
ary, 1973 and January-February 1976. Details on the experimental technique
and measurements of some of the observed far-field time-mean condensed (vi-
sible) plumes are presented with corresponding source parameters and am-
bient air data. Predictions of far-field plume behavior and visible plume
length using a slightly modified version of the closed-form model after
Slawson and Coleman [1] Slawson [2] are compared with observations. The
effects of elevated inversions, severe vertical wind shear and the nature
of the source geometry on the observed plume behavior and subsequent pre-
dictions by the simple closed-form model are discussed.
INTRODUCTION
It is fairly well accepted that a thorough understanding of cooling tower
plume behavior (both near-field and far-field) is required in order to de-
velop mathematical models of visible plume length, trajectory and salt or
other effluent deposition for purposes of environmental impact assessment.
The majority of work on cooling tower plume behavior both experimental and
theoretical has been associated with natural draft cooling towers. Conse-
quentely, little comprehensive field data exists for model development on
the behavior of plumes from mechanical draft cooling towers. Recent work
by Slawson et al [3], [4], has shown that plume downwash around natural
draft cooling towers can significantly alter far-field plume trajectory.
Thus, due to the usual more complex source geometries of reduced tower
height, multiple fans, and presence of adjacent towers, the near-field
plume behavior from mechanical draft cooling towers can be expected to
dramatically affect both operating performance through plume recirculation
and interference as well as far-field plume behavior.
Southern Company Services Inc. has conducted two full-scale field study
observational programs on the mechanical draft cooling tower plumes at
Plant Gaston, Willsonville, Alabama, during February 1975 and January-
February 1976. The objective of the field studies was to observe by pho-
tographic means the near-field and far-field behavior of the condensed
(visible) portion of the cooling tower plumes. In addition, detailed
388
-------
measurements on source and ambient parameters were made. Some details
of the experimental technique may be found in a paper by Champion et al
[5]. An attempt at modelling the observed near-field plume behavior is
given by Crawford and Slawson [6].
The objective of the work outlined in this paper was to determine whether
the simple integral vapor plume model of Slawson [2] could be modified or
site tuned to produce consistant and reasonable agreement between the ob-
served and theoretical far-field plume trajectories and lengths for the
twin mechanical draft cooling towers at Plant Gaston.
EXPERIMENTAL DESCRIPTION
The twin Ecodyne concrete cross-flow mechanical draft cooling towers at
Plant Gaston each service an 880 MWe coal-fired steam electric generating
unit. Each tower has an overall length of 98.8 m width of 22.2 m and
height of 16.9 m. There are 9 cells (9 fans) per tower. Each fan has a
diameter of 9.2 m and rotates at 117 rpm to produce an average exit velo-
city of 8.14 meters per second. The towers are parallel to one another and
separated by a distance of approximately 100 meters. They are aligned
along a north-west south-westerly direction, and are located on the north-
east side of the generating station. Measurements were made of tower exit
temperature and velocity along with ambient measurements of dry bulb and
dewpoint temperatures, wind speed and direction from ground level to above
plume top. The source and ambient data were obtained bracketing the time
period when the visible plume was being photographed from far-field and
near-field camera stations. The visible plume photographs (one every min-
ute during a given one to two hour period) were subsequentely reduced to
time-mean visible plume outlines covering a period over which ambient con-
ditions (particularly wind) were considered reasonably steady. The near-
field visible plume top and bottom were much more qualitative in nature
than the far-field due to the strong three-dimensional nature of the plume
near the source.
MODELLING CONSIDERATIONS
As indicated above, the basis for the vapour plume model discussed here is
the closed form or analytical solution of the integral equations governing
moist plume behavior given by Wigley and Slawson [7], Slawson [2]. If one
assumes that the plume is well bent-over and that it is rising in a uni-
form wind field these equations reduce to:
_ o M
a H
dM
389
-------
dx
U = - MG (4)
d£ = (5)
dx U w
where M - UR W, F - UKb, H = UR Aq are the fluxes of vertical momentum,
buoyancy and excess specific humidity respectively. A buoyant accelera-
tion per unit mass is given by b » pAT/Ta. The plumes vertical transport
velocity and wind speed are given by W and U respectively. The specific
humidity gradient is given by G = dqa/dz, and the Vaissalla frequency
(for stable stratification) by N ~(gl^& 38/8z)%. Further details on the
development and solutions to the above equations may be found in papers
by Wigley and Slawson [7], [8], [9], Slawson and Coleman [1] and Slawson
[2].
Following the work of Slawson and Csanady [10] the plumes growth or en-
trainment of ambient air is considered to occur in two phases. In the
first phase the self -gene rated turbulence of the plume is the dominant
mechanism for growth while in the second or atmospheric phase, atmospheric
turbulence is responsible. Here, we are mainly concerned with plume be-
havior in a stable atmosphere and following Slawson and Coleman [1] assu-
me that the atmospheric phase of plume growth starts at the point of maxi-
mum rise. Also, for modelling purposes we assume that plume growth in the
atmospheric phase is described by the Pas quill sigma's.
The vapour plume model as outlined here and described in detail in the
above mentioned references had been previously used for describing the
behavior of single natural draft cooling tower plume.
As noted previously, the additional complexities associated with the plume
dynamics of mechanical draft cooling tower plumes must somehow be accoun-
ted for if the model was to be modified in order to adequately describe ,
the far-field plume behavior at Plant Gaston. It was thought that the
modifications to the natural draft cooling tower plume model must at least
account for, (a) the existence in this case of two sources of non-circular
initial exit geometry (long thin rectangular sources), (b) the stronger
wind speed shear layer associated with plumes closer to the ground and (c)
wind direction relative to the long axis of the towers. The detailed des-
cription of near-field plume behavior is out of the realm of my inte-
gral model and is thus the subject of the paper by Crawford and Slawson
[6]. Here we attempted to account for the effects (a), (b) and (c) above
by empiricle or s end -empi ride methods.
390
FRS
-------
Hie thin rectangular geometry of the mechanical draft cooling tower source
was initially incorporated into the model by simply considering a plume
element to be of near-rectangular cross-section, rather than circular.
The source shape was also altered to account for the projection of the
real tower source into a plume normal to the wind direction. With this mo-
dification to the model it became apparent from test calculations that the
rectangular source geometry (aspect ratio) of the Gaston Towers had little
effect on plume behavior in the far-field (ten tower lengths from the
source). Thus, for the Gaston site at least the model used a circular
plume but incorporated merging of two plumes, wind speed shear and wind or
plume direction effects.
The plumes from the two towers were considered to have merged together com-
pletely at the point where the edges of the two plumes were calculated to
meet. At this merge point where the plumes just touched, a new plume cross-
sectional area (Am) was calculated equal to the sum of the areas of the two
individual plumes. From this new area a new effective plume radius was de-
termined. The new (merge point) momentum buoyancy and excess specific hu-
midity fluxs were determined by combining the individual fluxes in propor-
tion to their respective plume cross-sectional areas, thus,
M
m
F
Mm - < + > A
2
F - (-r± + 7=) A (7)
m A. A~ m
H H
and H = (-r± + -r^) A (8)
m AI A_ m
The model also alters the trajectory from the single plume to the merged
plume at the merge point. The trajectory of the merged plume starts at an
elevation based on a weighted average of the previous two plume elevations
at the merge point. Subsequentely, virtual source variables are calcu-
lated for the now merged plume and the model re-runs as a single plume mo-
del.
A trajectory constant was used in the model of Slawson [2] to account for
the effect on plume trajectory of plume downwash around the natural draft
cooling tower. In the modified version used for the Gaston plumes the
trajectory constant was made an empiricle function of the observed wind
speed shear and the wind angle relative to the long axis alignment of the
towers. It was known from the numerical integration model of Slawson [2]
that a principal effect of vertical wind speed shear was to lower the plume
trajectory. Also, physical modelling studies have shown that aerodynamic
wakes and recirculating flows are affected by the orientation of rectangu-
lar bluff bodies to the principal flow direction. It was also felt that
the entrainment of ambient air into the plume, which in the model is incor-
porated through the value of the entrainment constant a, would depend on the
391
-------
amount of wind direction shear present over the plume rise region. Thus
the entrainment constant was made an empiricle function of the amount of
wind direction shear.
It is obvious from the previous discussion that a great deal of rather hea-
vy handed empiricism was required in order to modify the simple vapour plume
model so that it would apply to the twin mechanical draft cooling tower
plumes at Plant Gas ton. The need for these or similar modifications drama-
tically emphasize the shortcomings in the physical basis of the, simple va-
pour plume model when applied to mechanical draft cooling tower plume
behaviour.
RESULTS AND DISCUSSIONS
Three far-field time-mean plumes have been selected from the data base as
examples of the observations and subsequent model comparison. Since the
model has been developed to include considerable empiricism from a limited
data base it certainly cannot in its present form be considered as a pro-
ven predictive tool. Table 2 lists the source data on tower exit velocity,
temperature (assumed at saturation) and surface pressure corresponding to
the observations given in figures 1 to 6. The ambient data on temperature
and wind profiles illustrated in these figures represents the average for
the time period corresponding to the plume observational period. Further
details on the model and specific measurements will be given in a final
report by Slawson et al [11].
In figure 1 the model trajectory well approximates the observed but the
final plume, length appears to be overpredicted by some 33%. However, in
this case the visible end of the plume was obscured by cloud.
In figures 3 and 5 model trajectories and plume lengths are in good
agreement with the somewhat shorter observed plumes. During all three of
the plume observational periods the ambient profiles of wind speed and
temperature are complicated by the presence of severe "kinks" in the pro-
files. The simple top-hat distribution of plume parameters used in the
model may be too severe a restriction for such complex profiles of ambient
parameters. However, the model uses the average ambient parameter over
the predicted depth of the plume when calculating plume parameters in an
attempt to account for variability across the plume. A significant diffe-
rence between the observations of figure 1 and those of figures 3 and 5 is
the wind and or plume direction relative to the long axis alignment of the
twin mechanical draft cooling towers. In figure 1 the wind is blowing
nearly parallel to the long axis while in figures 3 and 5 it is normal to
the long axis of the towers. Again, it is doubtfull that the simple empi-
ricism developed for the present model to account for wind orientation ef-
fects on plume near-field behaviour is established well enough on the basis
of such a limited data set. One soon realizes both from observations on
near-field plume behaviour and from the complexities and uncertainties as-
sociated with modelling such behaviour, as Crawford and Slawson [6] illu-
strate, that any empiricism purporting to account for such behaviour is at
least site specific as well as denpendent on the plume model itself.
392
-------
In addition to the problems of incorporating the effects of near-field
plume dynamics and complex time-mean profiles of ambient parameters one
must in the present model decide on the point of onset as well as a des-
cription of plume spread in the atmospheric diffusion phase. In cases
of very long plumes in stable atmospheres, the atmospheric phase may be-
come very important to predicting the observed plume length. Thus the
choice of stability category and subsequent sigma's becomes important in
the present model.
CONCLUSIONS
Based on the data used here we conclude that it is very difficult to model
even far-field plume behaviour (to the extent of these observations) with
a simple integral vapour plume model. Since considerable empiricism app-
ears to be required in order to make the simple model work, then a much
larger data base than that at our disposal is required to firm-up the func-
tional form of any empiricism introduced to the model for mechanical draft
cooling tower plume applications. It is hoped that perhaps a numerical
integration model which can at least account for vertical wind speed shear
may better approximate the real physics of plume behavior in the far-field.
ACKNOWLEDGEMENT
The author gratefully acknowledges the continued support for this research
given by Southern Company Services Inc. Appreciation is also expressed
to the Electric Power Research Institute for its support and interest in
the model development.
REFERENCES
1. Slawson, P.R., and Coleman, J.A. (1977): "Natural draft cooling tower
plume behavior at Paradise Steam Plant", Proc. Waste Heat Management
and Utilization Conf., Miami, Florida, May 9-11.
2. Slawson, P.R. (1978): "Observations and Predictions of Natural Draft
Cooling Tower Plumes at Paradise Steam Plant", Atmos. Env., 12,
1713-1724.
3. Slawson, P.R., Coleman, J.H., and Frey, J.W. (1975): "Natural Draft
Cooling Tower Plume Behavior at Paradise Steam Plant (Part I)",
Tenn. Valley Auth., Div. of Env. Plan. E-AQ-76-1, 145 pp.
4. Slawson, P.R., Coleman, J.H., and Frey, J.W., (1977): "Natural Draft
Cooling Tower Plume Behavior at Paradise Steam Plant (Part II)",
Tenn. Valley Auth., Div. of Env. Plan. TVA/EP-78/01.
393
-------
5. Champion, E.R., Goodman, C.H., and Slawson, P.R., (1977): "Field Study
of Mechanical Draft Cooling Tower Flume Behavior", Proc. Waste Heat
Management and Utilization Conf. Miami, Florida., May 9-11.
6. Crawford, T.L., and Slawson P.R., (1978); "Modelling Near-Field Be-
haviour of Plumes from Mechanical Draft Cooling Towers", Proc. Waste
Heat Management and Utilization Conf. Miami, Florida, Dec. 3-6.
7. Wigley, T.M.L., and Slawson, P.R., (1975): "The Effect of Atmospheric
Conditions on the Length of Visible Cooling Tower Plumes",
Atmos. Env.. 9, 439-445.
8. Wigley, T.M.L., and Slawson, P.R., (1971): "On the Condensation of
Buoyant Moist, Bent-Over Plumes"", J. Appl. Meteor.. 10, 253-259.
9. Wigley, T.M.L., and Slawson, P.R., (1972): "A comparison of Wet and
Dry Bent-Over Plumes", J. Appl. Meteor., 11, 335-340.
10*. Slawson, P.R., and Csanady, G.T., (1967): "On the Mean Path of Buoyant
Bent-Over Chimney Plumes", J. Fluid Mech.» 28, 311-322.
11. Slawson, P.R., Crawford, T.L., Goodman, C.H., and Champion, E.R. (1978):
"Plume Behavior from Mechanical Draft Cooling Towers at Plant E.G.
Gaston". To be published.
394
-------
TABLE 1
DATA ON SOURCE PARAMETERS AND ATMOSPHERIC SURFACE PRESSURE
DATE TIME Wo Tpo Po
(m/s) (°C) (nb)
12/2/75 0717-0757 9.64 32.24 1013.9
13/2/75 0655-0745 9.64 25.94 1023.3
15/1/76 0647-0745 9.74 28.14 1029-4
395
-------
LoJ
=>
O
O
oo
<
UJ
X
800
600
400
200
DATE: 12-2-75
TIME : 0717-0757 hrs
OBSERVED TIME-MEAN VISIBLE PLUME
MODEL VISIBLE PLUME CENTRELINE
TRAJECTORY
MODEL PREDICTED
RE-EVAPORATION ZONES
MODEL PLUME ENDS AT 2260 (m)
I
I
200
400
600 800 1000 1200
DISTANCE DOWNWIND (m )
1400 1600 1800
Figure 1 Comparison of Predicted and Observed
Time-Mean Plumes 12/2/75.
-------
500
~
0
z
D
o
K 300
(D
LJ
>
§200
^
H
I
o 100
LU
0
DATE : !2-2-'75
TIME : 0717-0757 hrs
- \ \
\ \
\ \
- \ \
\ \
\ \
\ \
\ \
\ \
\
\
\
\
\
\
\
\
- ^^
^^s^
^^^ s
.^^ S
jfS^ Jf \
0246
WINDSPEED
I I 1
10 II 12 13
/.
\
%
s
\
\
/
I
i / 1 1
8 10 12
( m/s)
i ! 1
14 15 16
DEW POINT (°C)
i i t
10 II 12 13
l i i
14 15 16
- DRY BULB (C)
Figure 2 Observed Ambient Temperature and Wind
Speed Profiles, 12/2/75, 0717-0757 Hours.
397
-------
UJ
O
o: 375
UJ
> 250
GO
125
H
O
UJ
X
DATE : 13-2-75
TIME : 0655-0754 hrs
OBSERVED TIME-ME AN VISIBLE PLUME
MODEL VISIBLE PLUME CENTRELINE TRAJECTORY
0
125
250 375 500 625 750
DISTANCE DOWNWIND (m)
875
1000
Figure 3 Comparison of Predicted and Observed
Time-Mean Plumes 13/2/75.
-------
500
^ 400
o
z
3
0
X
o 300
LJ
O
2 200
^
H
I
O
u 100
I ,
0
DATE : 13-2-75
- TIME : 0655- 0745 hrs
"
—
I
0 2
I
-5 -4
«•••
) i
__
0
""-.^
^•^•^^
*^^ .— — *
— « --* -^" "
r
i
_ __ — - •""""""""""
..— -*•**
'""^^^
i
./
^^^
vX^
>^
.s
s\ 1 I
4 6 8 10 12
VA/IMH opr~Fn ^
WIIMU or LtLU \ 1
m/s)
1 1 1
-3 -2-10 1
DEW POINT CO
i i t
12345
• DRY BULB (°C)
Figure 4 Observed Ambient Temperature and Wind
Speed Profiles, 13/2/75, 0655-0745 Hours.
399
-------
O
O
LJ
O
400
o
CO
UJ
300
O
QQ
200
UJ
X
100
OBSERVED TIME-MEAN VISIBLE PLUME
MODEL VISIBLE PLUME CENTRELINE
TRAJECTORY
DATE : 15-1-76
TIME ! 0647-0745 hrs
0 120 ^u ot>u 48° 60° 72°
DISTANCE DOWNWIND (m)
Figure 5 Comparison of Predicted and Observed
Time-Mean Plumes, 15/1/76.
-------
500
E
-400
D
z
D
S 300
o
Id
>
8 200
h
I
2 100
LJ
I
0
DATE : 15-1-76
TIME : 0647-0755 /
hrs /
-4
468
WIND SPEED (m/s)
10
12
i
i
-10
-9 -8 -7
DEW POINT (°C)
i i I
-6
-5
-3
-2
-I
- DRY BULB ( C)
Figure 6 Observed Ambient Temperature and Wind
Speed Profiles, 15/1/76, 0647-0745 Hours.
401
-------
CRITICAL REVIEW OF THIRTEEN MODELS FOR
PLUME DISPERSION FROM NATURAL DRAFT COOLING TOWERS
R. A. Carhart*, A. J. Policastro, S. Ziemer, and K. Haake
Division of Environmental Impact Studies
Argonne National Laboratory
Argonne Illinois, USA
ABSTRACT
Thirteen models for natural-draft cooling-tower (NDCT) plume rise are
evaluated theoretically and tested with 39 sets of visible plume field data
from three sites. The models evaluated have been employed or developed for
use in the environmental-impact evaluation process for nuclear power plants.
Two separate theoretical approaches represent the models tested: semi-
empirical and integral. The ten models using the integral approach treat
the same basic phenomena through the use of conservation equations: entrain-
ment; buoyancy; pressure drag; heat, vapor and liquid water dispersion.
Each model simulates these phenomena in a slightly different manner.
It is found that a wide range in predictions occurs among the models.
Models which compute plume bending by entrainment alone predict too much
dilution when the plume trajectory is predicted correctly. The most
successful models employ additional means for plume bending which does not
result in more entrainment. Among the methods used is to add a pressure
drag to bend the plume, or to treat a smaller spreading rate for the
moisture portion of the plume, or to provide a rapid plume bendover in the
zone for flow establishment. The correctness of each of the latter assump-
tions is yet to be determined.
Model performance may be divided into three categories: competitive,
average, and below average. The competitive models [Winiarski-Frick, Stone
§ Webster, Slawson (Closed Form) and Hanna] satisfy the following criterion:
the models can predict visible plume length within a factor of 2.5 (predicted
visible plume length is 0.4 to 2.5 of observed) and visible plume height by
a factor of 2 (predicted visible plume height is 0.5 to 2.0 of observed) in
50% of all cases tested. Seven models predict with varying lesser degrees
of performance with two models (ORFAD and Frick) predicting consistently
poorly. Some of the model/data discrepancies can be traced to data errors
and uncertainties in the measurement of tower exit and ambient conditions.
However, the models have all shown a consistent performance with the
visible plume data over the 39 data sets and have presented a unified
picture of their performance. Improved predictions can be made by improving
model assumptions and by calibrating the improved model to the field data.
*Visiting Scientist.Permanent Address: Department of Physics. University
of Illinois at Chicago Circle.
402
-------
INTRODUCTION
The Environmental Protection Agency under the authority of the Federal
Water Pollution Act Amendments of 1972 mandates the use of closed-cycle
cooling for nearly all new electrical power plants. At present, utilities
normally choose to install evaporative natural-draft and mechanical-draft
cooling towers. Criteria used for siting new plants include considerations
of tower environmental impact. The towers must be designed to minimize
interference and recirculation for economic reasons, and to minimize off-
site fogging, icing, shadowing and drift deposition for environmental
reasons. Predictions of these environmental impacts must be presented by
the utility to state and federal regulatory bodies in the licensing process.
Thus the availability of reliable techniques for predicting atmospheric and
ground-level disturbances produced by the tower plumes is essential.
A number of mathematical models have been developed and are in use for
predicting tower environmental impacts, but none of these models has been
validated using a wide range of experimental data. The Nuclear Regulatory
Commission recently sponsored a program at Argonne National Laboratory to
validate the popular and promising models currently available. The final
report of the study is in preparation, [1] and this paper focuses on only
some of the results obtained. It should be read with reference to our
earlier paper for additional details not repeated here [2].
In this paper the formulations of thirteen single-tower natural-draft
cooling tower plume models is intercompared in a common format along with
each model's degree of success in predicting the visible plume outlines for
39 single-tower plumes from three sites. Model performance is traced to
specific assumptions made in model formulation.
PHYSICAL AND MATHEMATICAL ASPECTS OF NDCT PLUME DISPERSION
The emission from a natural-draft cooling tower into an ambient environment
resembles the classic problem of a jet in a free environment. More precisely,
it is a problem of a vertical jet of finite size with initial momentum and
buoyancy being dispersed in a crosswind. It is reminiscent of the emission
of stack gases into the atmosphere from a fossil-fired plant. Unlike the
stack, however, the cooling tower has a much larger exit diameter, a much
smaller temperature difference with the ambient environment (i.e., less
buoyancy) and a smaller exit velocity (less momentum per unit mass). As
the plume disperses, it entrains ambient air which has physical properties
varying with height. (Wind speed, wind direction, dry-bulb temperature,
and relative humidity are usually functions of height above the ground in
the natural atmosphere). The smaller exit velocity (actually densimetric
Froude number) of the outlet for the cooling tower sometimes causes downwash
conditions for the plume under moderate and high winds. In such case, the
cavity generated behind the cooling tower due to the crosswind interaction
with the tower will provide a low pressure field below the plume which has
the effect of pulling the plume downward and increasing the rate of entrainment
(mixing) of the plume. The effect of the tower structure thus complicates
403
-------
the free-jet problem. A second complication to the free jet problem is the
presence of moisture in the plume. Due to thermodynamic processes present
inside the tower, the plume is generally saturated at exit and contains
some liquid recondensate. Additional liquid water can form from condensation
when the warm moist plume mixes with the cooler ambient air, and when the
plume cools during adiabatic ascent. This condensation releases latent
heat warming the plume further. Later in the plume history, the plume
moisture tends to disperse below the saturation level due to ambient mixing
but because of evaporation of accumulated liquid water (absorbing heat),
the saturation state of the plume is maintained. A common assumption is
that the plume is visible whenever liquid water is present in the plume.
As long as liquid water is present, the plume state is assumed to remain on
the saturation curve of the X-T psychrometric diagram, where X is the plume
mixing ratio and T, the plume temperature. It should be noted that the
treatment of condensation/evaporation energetics in the plume provides
feedback to the plume temperature, increasing it during condensation and
decreasing it during the evaporation phase. Generally, however, moisture
thermodynamics significantly affects the dynamics of the plume only under
extreme conditions such as very cold and/or very humid ambient atmospheres.
In our plume model validation study, thirteen models [3-15] for NDCT plumes
were evaluated. Ten employ the integral formulation in which a set of con-
servation equations are solved to predict plume dispersion. These con-
servation equations can be reduced to a set of coupled nonlinear ordinary
differential equations which are solved by a variety of numerical methods.
From each model we have extracted eight differential equations for the
following plume properties: R(plume radius), vx(horizontal plume velocity),
w(vertical plume velocity), Tp(plume temperature), Xp(plume specific humidity),
a(plume liquid water content), x(downwind coordinate), and z(vertical
coordinate). The first three are dynamic equations, the next three are
thermodynamic equations, and the last two are kinematic equations. A brief
review of each equation follows.
1. Mass Flux Equation: In this equation, the mass flow into the
plume from the ambient environment is defined. If m is the plume mass
flux through a plane normal to the centerline (divided by TT) and s is the
distance along the centerline from the tower exit plane to this plane, then
the fractional entrainment rate, u, (called simply the entrainment rate in
this work) is defined by
»p*m (i)
The rate of entrainment of ambient air into the plume is usually expressed
by means of an entrainment velocity, ve. In top-hat models (i.e., models
that assume uniform lateral profiles of plume variables) with circular
plume cross-sections, there is a relationship between u and ve based on
conservation of mass:
ff *pa've _ ambient flux into jet _
y __ .plume flux along jet centerline
*pp*
where v is the total plume velocity along the centerline (v2 = vx^ + w^
404
-------
For each of the ten integral models the entrainment assumption is given in
Table 1. It is defined for each model in terms of a functional expression
for v or ve. The different assumptions for y represent major differences
among the models. To obtain the relationship in Eq. (2), one must view ve
as uniform around a given plume circumference and directed radially inward.
2. Horizontal Momentum Equation: The initially zero horizontal
momentum of the jet is increased through addition of the entrained momentum
of plume air directed downwind and the action of a pressure (drag) force
arising from the pressure difference between windward and leeward sides of
the jet. Each model presents an equation for dvx/ds, where vx is the total
horizontal plume velocity. (Note that vx • v cos 9, where e is the angle
of the centerline above the horizontal). Major differences among the
models exist in terms of their assumptions on the pressure force (its
presence or not and its functional form). Some models make the bent-over
plume assumption in which the plume is assumed to accelerate immediately
upon exit to the wind speed. In such a case vx = u and the horizontal
momentum equation is redundant. The equations used by the models are
summarized in Table 2.
3. Vertical Momentum Equation: The initial vertical momentum of the
plume as it exits the tower is augmented by the plume buoyancy directed
upwards and the vertical component of the plume drag force (if present)
directed downwards. The buoyancy force includes the additional effect of
moisture in the plume (moist air is lighter than dry air) and the negative
buoyancy effect of the weight of liquid water in the plume. Every model
yields an equation for dw/ds, where w is the plume vertical velocity, w = v
sin 9. See Table 3 for the assumed form of this equation in each model.
4. Equation for Plume Enthalpy: The plume temperature decreases with
downwind distance due to the interplay of (a) mixing with ambient air, (b)
adiabatic expansion and cooling as the plume rises into lower pressure
regions, and (c) the energetic effects of net condensation or evaporation
of liquid water. From each model, one can extract the equation used for
dTp/ds, the change of plume centerline temperature with centerline distance.
Table 4 lists the version of this equation used in each model.
5. Moisture Dispersion Equation: The plume mixing ratio Xp changes
due to mixing with ambient air naving different moisture content. As
generally drier ambient air is entrained and mixes with plume air, liquid
water can condense or evaporate affecting the moisture content of the
plume, so long as the plume remains at saturation at plume temperature and
pressure. After liquid water has been exhausted by evaporation, further
mixing will cause Xp to drop below saturation. Condensation and evaporation
change the plume temperature which indirectly affects plume mixing ratio.
Every model yields an equation for dXp/ds, the change of plume mixing ratio
with centerline distance. In Table 5 is given the appropriate equation for
each model. Differences among models exist in the treatment of moisture
thermodynamics; for example, some models treat condensation alone and
neglect evaporation of liquid water while some others neglect liquid water
entirely.
6. Liquid Water Dispersion Equation: The plume liquid water content
begins initially as the condensate present within the plume at exit. This
initial condensate is dispersed over a larger plume volume due to plume
405
-------
mixing and is increased or decreased as a result of the condensation or
evaporation that takes place in the plume. As a result, the plume liquid
water content, a, in kg/kg will vary with centerline distance. From each
model, we extracted the equation chosen for da/ds, the change of plume
liquid water content with centerline distance. Table 6 lists these equations
by model.
7,8. Plume Trajectory Equations: Two kinematic equations are written
to relate plume centerline distance s and the downwind and vertical coordinates
x and z.
In interpreting Tables 1-6, several points must be borne in mind. Whenever
a conflict arose between intentions presented in the model's theoretical
description and the actual equations we derived by taking the infinitesimal
limit of the steps followed in the computer code, the equations implied by
the latter were chosen. This procedure guarantees that Tables 1-6 reflect
the actual equations used to produce all model predictions for those models
whose source codes were available to us. In all cases we ran the models
with very small step sizes and in double precision to minimize truncation
error. In the two models which differentiate types of liquid water (cloud-
water, small drops; hydrometeor, larger drops) and include effects of
freezing for low plume temperatures; i.e., the Hanna [9] and Lee (NUS) [12]
models, the equations in Table 6 apply to a as the sum of all liquid water
in the plume, and apply strictly when the plume is above the ice-nucleation
point. (In our studies, we find freezing effects to be small.) Also, the
small effects of rainout and evaporation of hydrometeor water are neglected
in the equations presented for the Hanna and Lee (iNUS) models. Finally, to
facilitate the comparisons, some other small terms in the model equations
presented in Tables 1-6 have been dropped whenever they contributed a mere
few percent relative to the terms which have been retained. No significant
differences in model performance can be traced to those terms and their
omission from the tables serves to focus the comparison on the essential
variations among model formulations.
It is important to recognize the several most controversial areas in the
modeling of single-tower NDCT plumes. Perhaps the most important is the
choice of entrainment function as mentioned above since the models are very
sensitive to the entrainment theory used. The presence or not of pressure
drag effects on the plume and the method by which it is modeled (including
the alternative of the bent-over plume assumption) is a second major area
of controversy. Pressure drag affects mainly plume bending and, as a
result, brings the plume in contact with different air masses depending on
the local height of the plume.
Plume thermodynamics is treated differently in the models depending on the
modeler's expectation of the importance of condensation and/or evaporation
effects. Some models treat freezing of liquid water in the plume on very
cold days and the possibility of rainout. Also, different criteria are
used to define the boundary of the visible plume. A common criterion is that
the visible part of the plume is defined as that region where .the local
liquid water content is greater than zero.
406
-------
The treatment of the final atmospheric diffusion phase of plume dispersion
also differs among the models. Most models ignore it while others recognize
the need to treat atmospheric-turbulence-induced dispersion beyond maximum
rise. A difficult phenomenon to model is tower-or building-induced downwash.
The effect of wind interaction with solid bodies upwind of the plume is to
increase plume bending and increase entrainment from below the plume. The
treatment of such boundary-induced effects is really beyond the integral
approach. Although no fundamental treatment exists, one model [4] attempts
to treat it through empirical equations. Most models ignore the problem.
It is the combination of alternative assumptions which distinguishes one
model from the other.
METHODS OF EVALUATING NDGT PLUME MODELS
The good method to evaluate the performance of NDCT plume models is to
•compare model predictions to field data encompassing profiles of temperature,
specific humidity, and velocity at numerous transects through the plume.
In this way, a three-dimensional view of the plume in terms of its gross
characteristics (spreading, trajectory, etc.) and its microphysics (3D
temperature, moisture and velocity) can be attained. Unfortunately, no
such data were available at the time of our study. Some measurements [16]
of this type at Neurath and Meppen in West Germany have been made recently
for a number (usually three) plume cross sections downwind. The data have
just now been made available to us and will be useful in future studies.
A large body of data exists [17-21] for visible plumes at several American
and European sites. A typical data set encompasses visible plume outlines
(usually time averaged over a 20 minute to 1 hour period), ambient profiles
of dry-bulb temperature, relative humidity, and wind speed (sometimes a
single profile or at times several profiles averaged over the period of
visible plume photographs) and tower exit conditions such as tower exit
temperature, updraft velocity, relative humidity, and sometimes liquid
water content. The better data in that category involves time-averaged
measurements of tower exit conditions, ambient profiles, and visible plume
outlines. The disadvantage of such data is that only the visible portion
of the plume is being compared to model predictions. The invisible portion
can be much larger and extend far downwind. Also, airplane measurements
have revealed that the visible portion of the plume generally occupies the
top portion of the plume with the bottom invisible. However, the visible
plume does give a measure of the plume trajectory and spreading of the
saturated moisture portion of the plume. We have accumulated 39 data sets
for single NDCT plumes. Our comparison of the models to this large data
base have revealed distinct trends in model performance which helped us
locate model flaws from the model/data comparisons. It is the consistent
performance of each model to our visible plume data base which has proven
the value of our model/data comparisons to this body of data. Of course,
comparisons of model predictions with in-plume data and with single-phase
data (dry plumes in a dry environment) should complement the results we
have obtained to date.
407
-------
The scope and nature of the plume data sets have been described previously [2].
The 39 single-tower cases are listed in abbreviated form in Table 7, which
gives averages of ambient dry-bulb temperature, relative humidity, wind
speed, and average lapse rate for each case over the observed visible plume-rise
region. These average values allow a comparison of the environmental and
tower conditions for each single-tower plume case, and illustrate the range
of conditions found in our data base. Our model calculations, however,
employed the full ambient profiles measured.
The 12 Lunen cases represent winter conditions for the cooling tower plume
from a 335 MWe fossil-fired plant located inland in West Germany. The
Liinen tower represents a small source of heat and moisture release. The 14
Chalk Point data sets include 7 December cases and 7 June cases from a
tower for a 630 MWe fossil-fired plant located on an estuary in Maryland.
The Chalk Point tower represents a moderate source of heat and moisture
release. From Paradise, Kentucky, we have 13 data sets taken in the winter
at a single tower rejecting heat from a 1100 MWe fossil-fired plant. The
Paradise tower represents a large source of heat and moisture release.
Consequently, a balance of source sizes is represented in the field data.
Figures 1-21 summarize selected model/data comparisons for 20 sets of
visible plume data. On the left-hand side of each figure is a two-dimensional
presentation of the visible-plume outline. The abscissa is distance from
the tower centerline along the direction of prevailing wind, and the ordinate
is height above ground level. The vertical distributions of ambient conditions
are given in the right-hand side of each figure; the symbols represent
points of actual measurement. Note that the scales, both horizontal and
vertical, are varied to illustrate the great disparity in the extent of
visible plume predicted by the models.
To aid in quantifying model reliability, we have listed in Table 8 some
relatively simple statistical indices of model performance, obtained from
the 39 single-tower cases listed in Table 7. These indices are all based
on the ratio of the predicted to observed length and height of the visible
plume, which we denote by p. Excluded from the averages are those cases
in which PJ_ > 5 or pj_ < 0.2 so as to minimize the impact of a few very poor
predictions. We have instead tabulated the number of cases that are within
a factor of 5 in the column of Table 8 labeled %. Also tabulated is the
number of times the prediction falls within a factor of 2 and 2.5 (for visible
plume length). In interpreting the columns N£ and Nc, one should note that the
total number of single-tower cases was 39. The p ^ distribution is further char-
acterized by its range, its simple arithmetic mean, and 10 raised to a power equal
to the average of the absolute values of the logarithms of the p^'s. (See Table
8). This latter average was used since it handles overprediction and under-
prediction equally and since it weighs values of pi near 1 more heavily than
those far from 1. These simple statistical measures are not the only sensible
choices, but do give important insight.
A critique of the formulation of each model and the discussion of our
model/data comparisons is summarized on a model-by-model basis below.
408
-------
DISCUSSION OF MODEL FORMULATIONS AND ANALYSIS OF MDDEL/DATA COMPARISONS
The formulations of the 13 models will be sunmarily reviewed with our
critical comments. Selected comparison graphs of the more than 215 available
in our complete study are presented to show the significant features of
each model's performance with the data. Finally, each model's performance
based on several criteria using all 39 data sets is discussed in the final
section of the paper.
Slawson-Wigley Model [3]
The general version of this model assumes a plume of circular cross-section
with plume variables (defined here as total values rather than excess above
ambient) distributed as top hat profiles. The general .version does not make the
bent-over plume assumption and allows for immediate condensation of liquid water
in the plume whenever supersaturation conditions are computed. The energetics
of evaporation of liquid water is, however, not treated. The computer code
for the model is set up to allow for several simplifying assumptions to be
applied to the general theory. Of those permitted, Slawson and Wigley
suggest that, best results can be obtained by making two simplifying assumptions
defined upon input: (a) the bent-over plume assumption, and (b) the liquid
water content of the plume at any cross-sect ion is zero. The authors
recommended the use of these options within the code for our validation
work.
It should be noted that the assumption of no liquid water in the plume
implies a very simple treatment of plume thermodynamics. Plume temperature
and plume mixing ratio are varied due only to adiabatic mixing with ambient
air, and to cooling with plume rise at the dry adiabatic lapse rate (see
Tables 4-6). However, if adiabatic mixing produces water vapor above the
local saturation value, then Xp is allowed to exceed the saturation value;
i.e., the plume relative humidity.is permitted to exceed 1001 with a
remaining at zero. This plume relative humidity increase above 100% will
often occur in the region of the plume just beyond the exit plane. The
plume relative humidity begins at 100% (saturation) at the tower exit,
builds up to values exceeding 100%, and then due to further entrainment
during plume dispersion the plume relative humidity falls below 100%, at
which time the plume is assumed to lose visibility. Thus this version of
the model conserves the flux of water vapor by allowing plume supersaturation
instead of employing the common device of creating a liquid water reservoir
which is increased or decreased during plume dispersion. This treatment of
plume thermodynamics is simplified because it ignores the energetic effects
of condensation heating and evaporation cooling. The Lee-Batty model (to
be discussed later) also assumes o = 0, but limits Xp to the saturation
value and loses water by discarding all condensate as it forms. Another
difference between the two models is that the Lee-Batty model includes
condensation heating while the simplified Slawson-Wigley model does not;
both models do not account for evaporation cooling.
409
-------
Slawson and Wigley treat entrainment by using the assumption on the entrain-
ment velocity that ve = a|w|, where a = 0.3. This assumption is the most
common one used for models with the bent-over plume assumption, but Slawson
and Wigley employ it for the generalized version of the model as well. It
should be noted that Fan's assumption [22] reduces to v = a|w| when a
bent-over plume is assumed. The model does not make the Boussinesq approximation.
The Slawson-Wigley model employs a separate formulation for the diffusion
phase of plume dispersion; i.e., the region in which plume momentum and
buoyancy have been dissipated and ambient turbulence predominates. Slawson
and Wigley employ the criterion that when the plume velocity drops below
1/10 of the wind speed, ambient turbulence dominates plume dispersion. At
that plume cross-sectional location, the plume's vertical velocity is set
to zero yielding a mild discontinuity in plume slope. This cross section
is used to match the initial phase of plume growth (from the integral
model) to a Pasquill Gaussian diffusion model representing atmospheric
turbulence. The matching is accomplished by conserving mass, momentum,
energy, and moisture fluxes at that cross section. Further plume dispersion
is computed with the same equations as the integral model except that the
plume radius is determined from the vertical and lateral widths of the
Gaussian distribution which themselves grow with distance downwind at a
rate dependent upon the ambient stability according to Pasquill [23] curves.
The model predictions show a uniform trend with trajectories rising faster
than the observed ones. This behavior seems to be due primarily to the
absence of a vertical "pressure" or "drag" force. (Note that the bent-over
plume assumption is equivalent only to a very strong horizontal force
proportional to (vx-u)). Because of the vx = u assumption, the behavior of
this model's trajectories emphasizes a trend we have noted. All models in
our study which do not augment entrainment with some additional momentum-
transfer mechanism, predict trajectories lying above the observed ones.
Absence of downwash effects, as with all other models studied, also con-
tributes to the trend, especially for high wind, speeds. A study of the
plume data [24] suggests that downwash effects are noticeable whenever
u0/w0 > 1.0. Also, omission of the weight of liquid water in the vertical
velocity equation contributes slightly, but noticeably, to the trend of
higher plume trajectories which can be seen particularly for visible plumes
which level off before disappearing such as the one in Fig. 1.
The predicted visible plumes in this model seem to be generally short and
somewhat low. In fact, for 21 of our 39 cases, the predicted visible plume
length was more than 50% shorter than the observed length. The absence of
thermal inertia in the plume seems to be partly responsible, but the major
effect seems to be that 0.3 for a is too large as an entrainment coefficient,
and generates too great a dilution. Our experience with this form of
entrainment assumption suggests lower values for better model-data agreement.
Fig. 10 illustrates one of these short predictions. We found that for" the
cases with a low ambient humidity such as the Chalk Point cases for June,
where essentially no condensation occurs, the Slawson-Wigley model shows no
visible plume. On the other hand, when ambient relative humidity exceeds
410
-------
901, the model]s predicted plumes are very long. No clear physical reason
for this behavior is evident. For other comparison graphs for this model,
see Figs. 2 and 11.
Slawson (Closed-Form) Model [41
This model is a version of the Slawson-Wigley theory [3] which contains
sufficient simplifying assumptions to provide a closed-form solution to the
governing equations. The assumptions made are (a) bent-over plume assumption,
(b) no liquid water in the plume (a=0), (c) the Boussinesq approximation,
and (d) simplified 'representation of ambient meteorology including uniform
wind speed, constant temperature lapse rate and constant dew point lapse
rate. These four simplifications are sufficient to allow an exact integration
of model equations resulting in a set of algebraic equations for plume
temperature, mixing ratio, vertical velocity, radius, and plume center line
height as a function of downwind distance.
The method used by Slawson to define ambient parameters is worth reviewing.
The uniform wind speed (equal to plume horizontal velocity) is taken as the
average value of the horizontal wind speed over the plume rise region.
This average value is estimated by successive iteration. First, the tower
top wind speed is used to predict maximum rise above the tower. Once this
rise is obtained, a linear least-squares fit is made to the wind profile
over this height and the value at half maximum is computed. Then the rise
is predicted using the improved value. Two iterations are specified to
avoid the possibility, which occurs in some cases, of indefinite oscillations
in the value of the average wind speed calculated. Slawson also assumes
that the dew point lapse rate and temperature lapse rate are constant with
height. The appropriate average value is obtained by iteration and linear
least-squares fitting as done for wind speed after an initial estimate is
made.
The Slawson model is the only model reviewed in this study which attempts
to account for plume downwash effects. This is accomplished through an
empirical modification of the closed-form expression for the plume center-
line trajectory. The empirical modification of the trajectory equation was
fixed through fitting to 13 plume trajectories from the Paradise plumes.
The expression effectively reduces the height of the plume at any downwind
distance under downwash conditions. The empirical modification causes the
plume height to decrease more rapidly than usual with increasing tower-top
wind speed and vertical wind shear for downwash conditions. The increased
effect of downwash on entrainment is not considered by Slawson. Slawson
employs the same entrainment function as used in the Slawson-Wigley theory.
It should be pointed out that this closed-form model has received more
attention by Slawson than the numerical Slawson-Wigley model in terms of
calibration to field data. The Slawson model has been used rather extensively
for environmental impact calculation due to its simplicity and to the fact
that the environmental data usually available is not very detailed and is
411
-------
thought to be adequately represented by the above simplified treatment.
The closed-form model shows better results in comparison to the field data
than does the Slawson-Wigley model. This is true in spite of the more
accurate treatment of ambient meteorology and plume dynamics employed in
the Slawson-Wigley approach. We believe that the source of improvement
seen in the Slawson model is in the calibration to actual field data of the
entrainment function and downwash coefficients. [The entrainment coefficient
a = 0.3 used in the Slawson-Wigley model was actually obtained through
calibration of the Slawson model to field data. The appropriateness of
that coefficient to the Slawson-Wigley equations needs to be tested.]
The performance statistics of the Slawson model is on a par with the most com-
petitive models. The model performs better than 9 of the other 12 models
tested. The model yielded predictions for all 39 single-tower cases, yet
predicted no visible plume for one summer Chalk Point case. Predicted
visible plume lengths tended to be short, consistent with the behavior of
the Slawson-Wigley model. Visible plume rise predictions were more evenly
distributed with both over and under predictions occurring. Trajectory
predictions of the Slawson model were improved over those of the Slawson-
Wigley model predictions in cases with large observed downwash (see Fig.
3). The added entrainment probably present with downwash is not represented
in the model, which may help explain the length overprediction seen in this
Figure. Plume trajectory predictions were generally good for the duration
of observed plume visibility, and some impressive fits were achieved, as
shown in Fig. 4. Further comparisons may be examined in Figs. 16-21.
Weil [5]
The model assumes a circular plume cross-section with top-hat profiles of
plume variables, although a more general theoretical formulation is presented
as well. Weil also introduces the Boussinesq approximation. In the buoyancy
term (see Table 3) of the vertical momentum equation, several small approximations
are made. The model assumes an entrainment rate proportional to g|w|
with 3 = 0.55. This value of entrainment coefficient is the largest value
chosen for this parameter among the models reviewed that use this entrainment
equation. Weil introduces no pressure or drag forces in the model and, as
a result, only entrainment affects the change in the horizontal plume
velocity, and only buoyancy augments entrainment in the vertical momentum
equation.
The Weil model fully represents the major thermodynamic processes in the
plume. The equation for temperature change in the plume (Table 4) correctly
includes the effects of adiabatic mixing and the energetics of evaporation
and condensation. The treatment of moisture thermodynamics leads to the
factor (1+a)"1 in the equation for dTp/ds (see Table 4); the term -
is the saturated adiabatic lapse rate. Note that the factor (1+a)
affects the other two terms on the right-hand side of the dTp/ds equation
as well. In the model, liquid water is allowed to build up in the initial
stages of plume dispersion due to condensation, heating the plume. As the
412
-------
plume disperses, the liquid water is allowed to evaporate, cooling the
plume until the plume mixing ratio drops below its saturation value. From
that point on, simple mixing and adiabatic cooling govern the plume dispersion.
A comparison of model predictions to field data reveals that the Weil
plumes typically rise above the observed plumes and lose visibility rather
rapidly. The explanation for this is partially that the model has no
pressure or drag forces to augment entrainment and buoyancy in the transfer
of momentum and the consequent bending of the plume. Fig. 5 illustrates a
typical Weil prediction. Due to the large entrainment in the model, the
plumes spread rapidly and disperse the moisture rapidly leading to a shorter
visible plume. For other plume predictions of the model, see Figs. 6, 7,
and 9.
The model does not compare favorably with the better models as may be seen
from the model's performance statistics given in Table 8. In fact 20 of 39
visible plume rise predictions have p^ values below 0.5 and 33 have values
below 1.0. In terms of visible plume length predictions, 34 of 39 ratios
Pi lie below 1.0 and 31 of 39 are less than 0.5. The model's basic formulation
appears to be sound except for the items mentioned above and perhaps the
need to add an atmospheric diffusion phase to the model. The model should
give better predictions when corrected and calibrated to the data to determine
6.
Frick Model [6]
The philosophy of the Frick model rests heavily on the assumption of an
elliptical cross section at each downwind location. Several experimental
observations reported in the literature indicate that single vertical jets
in a crossflow have cross sections that are horse-shoe-shaped (elliptical
as an approximation) rather than circular. Altering the round-plume hypothesis
in favor of an elliptical plume cross section directly impacts entrainment
in the Frick model since Frick assumes that entrainment is directly proportional
to the area of the projection of the (elliptical) cross section on a plane
normal to the wind direction. Use of an elliptical cross section will
provide a larger entrainment into the plume than the corresponding round-
jet assumption.
The Frick model is set up for top hat profiles of plume variables. The
Boussinesq approximation is used. The semi-major axis of the plume is "a"
(in the horizontal) and the semi-minor axis is "b" (in the vertical).
While important in determining the actual value of the entrainment rate,
the equations for the separate growth rates of a and b with centerline
distance are not detailed here. The main effect of the ellipticity can be
seen in the Frick model entrainment rate formula (see Table 1) where y
contains factors such as (a/b)(V2) which increases the entrainment rate
with increasing ellipticity. The model's equation for u contains no adjust-
able constants since it is based instead on an assumption that for any
given plume slice, the ambient air which iimpinges on the portion of that
slice which is exposed to the oncoming ambient wind is fully entrained.
413
-------
Since the area exposed depends on the growth of a and b, and since that
growth is, in turn, dependent on y , one must solve Prick's set of equations
in a consistent manner to obtain p. This explains the complicated form of
the velocity-dependent factor in a(s) for the Frick model in Table 1. That
factor, {1 - (uvx)/(v2){l - (uvx)/(2v2)}}-! will vary from 1.0 initially to
(a) values slightly below 1.0 if uvx/v^ > 2.0, (i.e., if the windspeed
greatly exceeds the plume exit velocity) and to (b) a value of 2.0 for a
fully bent-over plume. During the rise of the plume, the term in the
entrainment velocity ir'l /a/b (u/v) |w| tends to dominate especially at
first. The resulting values of y are considerably larger than for other
models initially and throughout the region where the plume is bending over
strongly.
Our model statistics summary (Table 8) and Figures 6-7 show that this model
predicts very short visible plumes compared to observed. (See also Figs. 5
and 9.) Final length predictions average 1/3 of observed values. The
predicted centerline trajectories are reasonably close to the observed ones
over the predicted visible portion of the plume. As expected, the model
also consistently predicts a lower final visible plume rise in accord with
the short final length trend. The Frick equation for vertical velocity
variation is common, accounting for the expected velocity decline due to
entrainment, and containing the usual buoyancy term. The model assumes no
vertical drag force.
The treatment of moisture thermodynamics does not follow common practice.
The model documentation suggests that the standard equation for dTp/ds is
used, allowing for (a) adiabatic mixing with ambient air, (b) adiabatic
expansion with plume rise and (c) latent heat either released to (con-
densation) or absorbed from (evaporation) the plume. (Frick neglects the
variation of saturation mixing ratio with pressure). The usual equation
for this is
dT
^ ' (1+a)'1 {-Yd £ - yCCTp - Ta) + L. (Xp - Xa))} , (3)
where a = (L/cp)(dQs/dT) evaluated at T = Tp and Q? is the saturation
mixing ratio at plume temperature. Eq. (3) is typically applied in computer
codes in a sequence following the steps (a), (b), and (c) above. Step (c)
is carried out to return the plume state to the saturation curve, if indeed
the plume is saturated, and is usually carried out through an iteration
procedure to properly locate the plume on the saturation curve. However,
in the actual coding of the iteration for step (c) the equation used by
Frick is, instead,
dT ,
a/ --*d7-" ( (V V (4)
The second term in Eqn. (4) is the limit as L -»• « of the term proportional
to u in Eqn. (3), and the first term is the limit as L -»• 0 of the saturated
adiabatic lapse rate, (1+a)"1 Yd- Since a ranges between 0.3 and 4.0 over
414
-------
normal plume temperature limits, the differences between Eqn. (3) and Eqn.
(4) are substantial, and use of the limits on L do not provide an accurate
approximation. The plume is forced to remain at Xp = Qs(Tp) by the coded
model logic, (see Table 5), but Tp will drop too fast and consequently Xn
will drop too fast when Eqn.~(T) is used. Effectively, then, water vapor
is not conserved and disappears from the system. This effect, while
secondary to the large entrainment rate in magnitude, also serves to shorten
the predicted visible plume. In fact, the equation for liquid water flux
effectively implied by the Frick model equations is
dX
« . c- ' ts
The terms containing u are normally negative throughout the dispersion of
most plumes, except for a short initial region where adiabatic mixing
usually causes additional condensation. (Derivation of Eqn. (5) is obtained
directly from the Frick model equations in Tables 4-6). Using the correct
equations for Tp, Xp and a which conserve total water will lead to Eqn. (5)
with only the first term on the right-hand side.
Recall that the Frick model ignores the pressure dependence of the saturation
mixing ratio, using a standard value of pressure of 1020 mb. Since the
errors in computing Xp and Xa have the same trend, this approximation has
little effect on the predictions.
The trajectories predicted from the Frick model all lie above the observed
trajectories; i.e., the predicted plumes bend over too slowly. This
behavior seems to be a persistent feature of all models which do not use a
strong "pressure" force or bent-over plume assumption, and instead modify
the plume's horizontal momentum only by means of the added momentum of
entrained ambient air.
Winiarski-Frick Model [7]
This model is similar in structure to that of Frick and is a later version
of the Winiarski-Frick model presented in Ref. 25. The plume is assumed to
have a circular cross-section and top-hat profiles of plume variables, at
least in the initial phase. The Boussinesq approximation is not used.
In terms of plume dynamics, equations for the entrainment rate, horizontal
velocity, and vertical velocity appear in Tables 1-3.
The model allows for two possible entrainment mechanisms: (1) impingement,
in which the momentum of all of the ambient air which is directed towards
the "exposed" portion of the plume surface for any given cross-sectional
slice is transferred to the plume, and (2) aspiration, in which ambient is
drawn into the plume as a result of the turbulence induced by the difference
between the plume centerline velocity and that component of the ambient
velocity parallel to the plume centerline, |y-u cos e|, representing a
415
-------
shearing action. Winiarski and Frick believe that the impingement mechanism
is predominant in moderate to large wind situations and the aspiration
mechanism is important in very weak wind conditions. Actually the mechanisms
will overlap and it is difficult to ascertain how much entrainment is due
to each one. Their numerical studies have indicated that the most fruitful
assumption is that the larger of the two mechanisms at any downwind distance
is to be used to represent the total entrainment mechanism. An anomaly
exists in the functional form for the impingement entrainment function.
The entrainment rate due to impingement near the tower exit becomes infinite
when the wind speed is approximately /T times the exit velocity and becomes
negative when the wind speed exceeds this value. The negative value does not
cause any computational problems since the maximum value of entrainment and
impingement is used at any cross -section. The physical validity of the
functional form of entrainment by impingement (see Table 1) is, however,
questionable.
It is interesting to compare the magnitude of the Entrainment by impingement
in the Winiarski-Frick model with the entrainment rates in other models.
Rewriting the Winiarski-Frick impingement- entrainment velocity as aw, we
may compare the value of the complex function that defines a with the
constant values of 0.3 and 0.55 as chosen by other modelers. Sample numerical
evaluations of the Winiarski-Frick a reveal it to be much larger than the
range 0.3-0.55. As a result, Winiarski-Frick plume predictions should be
short due to excessive dilution with centerline trajectories lying somewhat
above observed ones. This is based also on the fact that this version of
the Winiarski- Frick model does not include any drag force mechanism to
augment plume bending and does not include the "virtual mass" concept used
earlier [25] which would cut down the effect of plume buoyancy. As a
result, we would expect to characterize the model's predicted plumes as
very short and consequently low as compared to observed plumes.
However, an additional feature of the model serves to lengthen the predicted
plumes by a large amount. The top hat assumption requires the moist plume
to be fully-mixed laterally so that the temperature and humidity are uniform.
This complete mixing laterally may imply that the plume is below saturation
at a particular cross- section. Experimental data have shown that plume
variables follow more bell- shaped distributions than top-hat profiles.
However, distributing the same amount of energy and moisture so that peak
values occur at the center with minimum values along the edge may result in
a condensed visible core with an invisible, subsaturated outer region.
This smearing of the top- hat distribution into a bell -shaped distribution
will clearly lengthen plume predictions. Winiarski and Frick chose the
following profiles for excess temperature and water vapor,
AT(s,r) = ATc(s) COS ; AQ(s,r) = AQJs)
where s is the centerline distance from the tower, r is the lateral distance,
AQc(s) and ATc(s) are the centerline (peak) values of the new distributions
AT(s,r) and AQ(s,r) defined above ambient values.
416
-------
These profiles apply only for downwind distances beyond the point where the
plume has lost visibility (o - 0) with the top-hat differential equations.
Winiarski and Frick define at this evaporation point the lateral profile of
AT and AQ by conserving energy and moisture, (i.e., area under AT and AQ
curves match the corresponding area obtained from the top-hat profile).
Interestingly, ATC and AQC become 3.37 times the corresponding top-hat
values leading to a supersaturated central region of the plume which extends
laterally to 63% of the computed top-hat radius. The original equations
with the top-hat profile assumption are employed to calculate the invisible
radius R with downwind distance. This R also is used to characterize the
growing width of the two cosine functions above. The amount of lengthening
is very significant in most cases. The Winiarski-Frick equations imply that
the mass flux through the plume cross section where the cosine distribution
predicts loss of visibility contains precisely 3.37 times the mass flux in
the plume at the point of evaporation of the initial top-hat plume. Another
limitation of the model is that ambient conditions at the point of evaporation
are used in the mixing relationships defining plume visibility when the
cosine distributions are operational, rather than local ambient conditions
at the cross section under study. To us, the use of a bell-shaped distribution
after visibility has ended, according to the top-hat differential equations,
is primarily a method to extend plume length and represents part of the
calibration procedure of the model to data. It should be noted that Winiarski
and Frick calibrated the model to data not by adjusting coefficients but by
testing numerous different physical assumptions until a combination which
seems to agree with data was determined.
Model/data comparisons for the Winiarski-Frick model appear in Figs. 5-7 and
9. Note the discontinuity in the visible plume outline. The discontinuity
is located at the point where visibility ends for the top-hat plume. This
point of evaporation results from solving the full thermodynamic and dynamic
equations describing plume development. Up to this point, plume equations were
solved using the top-hat assumptions only; beyond that point, the cosine
distribution for plume temperature and moisture was employed from which a
smaller visible radius of the supersaturated core becomes the visible radius.
The model's performance places it in the group of the four most competitive
models. Its rise predictions are the best of all models studied, and the
length predictions, while usually short, place the model among the top three
for visible plume length predictions. This seems surprising in view of the
fact that the equations governing the visible core, which normally accounts
for most of the predicted visible plume length, omit a number of physical
effects usually considered important to include in predicting lengths.
Recall that no liquid water reservoir or energetic effects of condensation
and evaporation are involved in the use of the bell-shaped temperature and
moisture distributions. However, as mentioned above, Winiarski and Frick
have calibrated the model to a larger data base than other modelers have
used, including single-phase data and some Lunen and Chalk Point data.
Also, the calibration has not been done by selection of values for adjustable
constants, but by varying the model's physical assumptions, in ways illustrated
by the choice to use entrainment by impingement or aspiration, but not
both.
417
-------
QRFAD (Oak Ridge Fog and Drift Model) [8]
In the ORFAD model, the concentration of water vapor emitted by the cooling
tower is assumed to be dispersed in the atmosphere in a Gaussian manner
about a centerline trajectory computed from a Hanna-Briggs plume rise
formula. The spreading rate of the Gaussian distribution depends on the
Pasquill stability class estimated from local meteorological parameters.
[For the model/data comparisons at Paradise, the appropriate meteorological data
were not available and consequently we obtained the lapse rate from on-site
soundings.] ORFAD calculations are traditionally made with reliance solely
upon ground-level data for the state of the ambient atmosphere. Unfortunately,
no correction of the measured ground-level wind speed is made in order to
estimate its value at plume height for input into the Hanna-Briggs trajectory
formula. As a result, the use of these smaller ground-level wind speeds
leads to predicted centerline trajectories which rise much above the observed
trajectories. Use of a power-law wind speed profile to estimate elevated
winds from ground-level measurements should provide some improvement in the
predicted traj ectories.
Two additional theoretical weaknesses exist in the ORFAD model. First, in
considering the effect of the ground as a boundary to the dispersion of
water vapor, ORFAD does not reflect the Gaussian tail of the plume water
vapor from the ground, but merely doubles the plume water-vapor concentrations
at all heights. More importantly, the model assumes that the ambient
concentration (gm/m^) of water vapor at the ground is the same for all
heights; thus, when dTa/dz < 0, ORFAD predicts a "cloud layer" or fog at
and above the height where the water-vapor density obtained at ground level
begins to saturate the ambient air. When natural cloud layers were reported
as present in some data cases, this procedure was seen to underestimate
their heights severely, and it often gave cloud-layer predictions on clear
days. When dTa/dz < 0, the visible plume predicted by ORFAD can rise into
this cloud layer without ever terminating. The ORFAD plume terminates only
if the lapse rate is positive or if conditions are very dry or windy. In
only 12 of 39 test cases did the model give definite predictions. These 12
cases normally had either inversions or were very stable with low humidity.
The twelve definite predictions can be classed as uniformly long and high
at the point of disappearance, in accord with the trajectory trend noted
above and the overestimation of plume water vapor. One of these plume
predictions is shown in Fig. 8. The overall predictive ability of this
model for visible plume outline is the weakest of all models tested. Our
major criticism of the model formulation is that the simulation of the
plume as a spreading Gaussian distribution of fog about a centerline is
physically incorrect. The plume spreads due to the coupled effect of plume
buoyancy, momentum, interaction with the ambient wind, and plume moisture
thermodynamics. The ORFAD model basically assumes that ambient turbulence
is spreading the plume moisture about a pre-determined plume centerline.
An additional problem with the Gaussian approach is that it ignores the
possibility of natural adverse gradients of water vapor concentration near
the ground that would prevent the downward dispersion of plume vapor to the
ground. To us, the physical picture used in ORFAD of the dispersion of
moisture is incorrect as a "basis for plume or, ultimately, ground-fogging
predictions.
418
-------
Hanna [9]
The framework of the Hanna model is similar to many other models. However,
a few interesting modifications have been made. Hanna recognizes the
apparent paradox that a model, through entrainment alone, cannot predict
the correct trajectory, spreading, and dilution simultaneously. Hanna,
therefore, employs Briggs' suggestion [26] as to a possible resolution of
the apparent paradox. Briggs argues that the change of the momentum flux
with height depends partly on the fact that ambient air above the rising
plume must also be accelerated. With this reasoning the effective radius
of the "momentum plume" is therefore larger than the radius as determined
from temperature differences. Hanna follows Briggs who suggests that for
bent-over plumes aRm/az = 0.6 for the momentum plume (% = momentum radius)
and 3Rt/3z =0.4* for the temperature plume (Rt = temperature radius).
Consequently, the ratio, %, of the effective momentum flux to the momentum
flux within the temperature plume approaches (0.6/0.4)2 or 2.25. Hannafs
second major modification relates to a separate radius for the moisture
plume. Hanna claims that observations by Slawson et al. [27] and Meyer et
al. [28] indicate that visible plume lengths are consistently underpredicted
by basic plume models, due to the fact that inhomogeneities in the plume
can result in locally saturated spots, even though the plume may be unsaturated
on the average. So the use of a smaller P^ than Rt and %, is meant to
represent the inhomogeneity of the plume, where only part of the disturbed-
flow region is pictured as being occupied by saturated, heated plume air.
Hanna simulates this by assuming that the ratio, E^, of the cross-sectional
area of the moisture plume to the temperature plume approaches 1/2. In
terras of the rate of change of the radius with height, this assumption can
be written 3Rw/3z = 0.71 (3Rt/3z) where RW is the radius of the moisture
plume.
Hanna assumes top hat plume profiles for all variables. The plume cross
section is taken to have a central circular core R 1 RW and two annular
regions Rw
-------
2(0.4)w/(Rtv). Since (0.6)/% -»• (0.4)/Rt as s becomes large, these two
entrainment rates are comparable in effect. The entrainment rate for the
"moisture plume" is (as seen from the (Xp - Xa)-term in Table 4) 2(0.2)w/(Rwv),
Thus, the dilution of plume moisture and liquid water occurs at about half
the rate as that of the dilution through mixing and lowering of plume
temperature. As a corollary, due to the different radii, the effects of
condensation or evaporation on plume temperature are reduced as seen in the
Ew contribution to the factor (l+Ewa)'1 for dTp/ds, (Table 4). Otherwise
the term would have represented the saturated adiabatic lapse rate of
(l+a)"l Yd5 thus, in this model, a smaller central region of moisture is
contributing energy effects over a larger region of temperature elevation.
It should be noted that the actual form of Hanna's equations as used and
shown in the Tables only satisfy the conservation laws for large s, since
his entrainment factors are all the constant large-s limits of the variable
ones implied by the radial growth assumptions, (dR^^/dz) = (.71) (dRt/dz) and
(dRjn/dz) = (1.5) (dRt/dz). It would be relatively simple to modify the code
to satisfy the conservation laws exactly throughout plume rise.
The model generally predicts accurate plume trajectories, with a slight
tendency for predicted trajectories to lie above the observed ones. (See
Figs. 1,10). The visible length predictions are among the best of all
models with an absolute log-mean average length ratio of 1.57, (see Table
8). Very few predictions lie below 0.2 of the observed values or above 5.0
of the observed values. Predictions of visible plume rise are likewise
among the best with an absolute log-mean average rise ratio of 1.67, and
only 4 of 39 predictions are incorrect by more than a factor of 5. A
drawback of the model as formulated is the lack of any logic to follow
plume dispersion beyond maximum rise, leading to no definite plume length
predictions in 14 of 39 single-tower cases included in this study. As can
be seen from Fig. 1, illustrating one of these ambiguous predictions, an
added atmospheric diffusion phase could improve overall model performance
by providing predictions in all cases. The model predicts long and high
plumes for high wind cases; i.e., above 9 m/s average wind speed, probably
due to omission of the effects of tower downwash. Also, in cases of high
ambient relative humidity; i.e., above 85%, especially in neutral-to-
unstable conditions, a problem exists in the performance of this model, as
it tends to overpredict plume length considerably in these cases; (see Fig.
11, another noh-terminating case).
It is interesting that the Hanna model is among the best performers.
Undoubtedly the choice of three different radii for the plume is an important
component. Theresis experimental evidence from studies of jet mixing in air
that the turbulent spreading rate for momentum is greater than the correspon-
ding spreading rate for heat, which supports the assumption % > Rt= But,
in reviewing, our model/data comparisons for all models, there is no trend
to show that predicted plume lengths are shorter than observed. This would
weaken the motivation for the assumption that PV < Rt=
420-
-------
Tsai-Huang (Stone and Webster) Model
The Tsai-Huang model assumes a circular cross-section. The model is one of
the two which assumes Gaussian profiles for all plume variables. The
equations for the plume are actually applied after the plume has passed
through a zone of flow establishment in which jet profiles change from
uniform at the tower exit to Gaussian distributions at the end of the ZFE.
Tsai and Huang obtain characteristics for the ZFE from the data of Fan [22]
(centerline length of ZFE and trajectory angle at end of ZFE). However,
the actual equation of the plume trajectory is not known; consequently,
Tsai and Huang assume a parabolic equation for the plume trajectory from
which the coordinates at the end of ZFE can be determined. As a result of
the parabolic assumption, the empirical bending used by Tsai and Huang is
made somewhat arbitrary. The density deficit at the end of the ZFE is
taken as one-half the initial value. On the other hand, the plume velocity
and mixing ratio are taken to be unchanged from the tower exit to the end
of the ZFE. These then establish the initial conditions for the integration
of the model's differential equations for ZFE which are listed in Tables 1-
6.
The zone-of-established flow portion of the model uses Abraham's [29]
modification of Fan's theory[22] of vertical jet discharges in a crossflow.'
In this theory, entrainment is defined as the sum of two terms, the first
(predominant in the very near field) developed from the theory of nonbuoyant
jets in stagnant ambients and the second (predominant as jet momentum
subsides) taken from the theory of buoyant thermals in stagnant fluids.
The model also considers a drag force operating normal to the jet centerline.
The model does not consider condensation, evaporation or liquid water
effects in plume dynamics. The discharged moisture mixes with the ambient
air by simple mixing as in the Slawson-Wigley model. The plume is assumed
to be saturated as it leaves the tower and no longer remains visible when
the mixing of the plume wi th colder ambient air .reduces the temperature to
a point at which the moisture at the plume centerline is just below saturation
level for that temperature. The thermodynamics of the plume are thus
greatly simplified.
It is useful to compare the Tsai-Huang equations to the more traditional
equations; e.g., those of Slawson and Wigley. Examining Tables 4 and 5,
the following differences occur: (a) Tsai and Huang replace the dry adiabatic
lapse rate in the dTp/ds equation by the actual ambient lapse rate, and (b)
Tsai and Huang add (w/v) dXa/ds to the equation for dXp/ds. Under near-
neutral and well-mixed conditions these terms will give the same effects as
those present in the Slawson-Wigley model. However, under stable conditions,
the plume temperature will decrease too slowly and the plume mixing ratio
will behave in some unpredictable fashion. The Slawson-Wigley equations
represent the usual theory for a model, such as Tsai-Huang, which assumes
absence of liquid water and the neglect of condensation/evaporation energetics,
Predictions of this model tend to have lower final rise values than do the
421
-------
observed plumes. Of the predicted-to-observed rise ratios, 28 out of 39
lay below 1.0. The empirical trajectory used to define the zone for flow
establishment seems to be responsible. Most predicted plumes show a rapid
ber.Jover, followed by an artificial "lift-off" perhaps due to residual
buoyancy in the plume (see Fig. 12). Apparently the slope of the centerline
at the start of the established flow region given by the parabolic trajectory
is not consistent with the actual residual buoyancy and horizontal and vertical
velocities of the plume. Thus, the overall plume rise would be artificially
lowered. In addition, ZFE relationships are being used here outside their
range of established validity, especially for small values of UO/WQ,
as represented by the case shown in Fig. 13.
An interesting feature of the Tsai-Huang model is that the predicted
plume tends to be very long whenever it remains visible as it reaches fully
bent-over conditions. When the slope of the plume centerline is small, the
entrainment becomes nearly zero. (One term in the entrainment function
•drops out and the other is very 'small). Plumes will travel long distances
downwind with very little entrainment. This tendency of very reduced
entrainment is underscored in high-humidity situations (Liinen S-6, for
example), leading to extremely long visible plumes. When the plume levels
out, entrainment vanishes and since y •* 0 the temperature equation becomes
dTp/ds = dTa/ds. Thus, the predicted plumes never show vertical oscillations
at the Brunt-Vaisala frequency as they level off but rather extend to
extremely great lengths with no apparent change in plume variables. Seven
of the 39 predicted plumes for this model were of this type, with the plume
failing to terminate at any reasonable downwind distance.
The model's length predictions were among the best, especially when both
the pi-ratio in Table 8 and the model's degree of reliability are jointly
considered. The rise predictions should also be considered state-of-the-
art as seen from the Table. For the model to be useful for environmental
impact studies, the stage" of the plume when it is fully bent-over needs to
be treated correctly with introduction of a non-zero entrainment velocity
after rise ceases.
Lee-Batty Model [11]
This model is similar in formulation to that of Weil and Slawson-Wigley.
The Batty-Lee and Slawson-Wigley models both use the bent-over plume assumption.
The Lee-Batty modifications of the Slawson-Wigley model are in these major
areas. First, an elliptical cross section is used after maximum plume rise
whenever the ambient stability class is not neutral; otherwise a circular
cross-section is assumed. The eccentricity of the elliptical shape depends
on the ambient stability class. Second, the entrainment velocity can
differ from that used in the Slawson-Wigley model, depending on the value
of certain ambient parameters -- namely, the wind speed, ambient lapse
rate, and eddy-dissipation rate. Different functional forms of the entrainment
velocity are used, depending upon whether neutral or non-neutral ambient
stability prevails, but these are only important at or near the point of
final rise.
422
-------
Of the 39 single-tower plume length predictions given by this model, 27
were less than 50% of the observed values. We attribute this behavior
primarily to overly-rapid dilution of plume air by ambient air, in view
of the rather large value chosen for a of 0.4 (see Table 1). Because the
model plume trajectories are also higher than the observed ones, the visible
plume rise predictions, while low, are not as far below the observed values
as one might expect. Figures 12 and 13 serve to illustrate typical predictions
of this model.
One other major feature of model formulation contributes substantially, and
sometimes predominantly, to early disappearance of the predicted visible
plumes, and to the elevated trajectories. In this model, no liquid water
is allowed to build up in the plume, (see Table 6), in spite of the fact
that (a) the saturated adiabatic lapse rate is introduced in the dTp/ds
equation, cf. Table 4, and (b) Xg is forced to remain at saturation with
associated condensation heating during the first stages of plume dispersion
(see Tables 4 and 5). So when additional condensation of plume moisture is
expected; i.e., (Tp - Ta) - B'1 (Xp - Xa) is positive or, equivalently, the
plume state after simple adiabatic mixing with entrained ambient fluid is
supersaturated on the X-T diagram, then no liquid water is added to a
"reservoir" for later evaporation. (The expected plume condensation
should occur early in the plume dilution in most cases, when adiabatic
mixing causes additional condensation of plume moisture.) Then, as soon as
(Tp - Ta) - B~l (Xp - Xa) drops below zero, the model plume immediately is
assumed to be subsaturated. At this point, the appropriate dry adiabatic
lapse rate is used in the dTp/ds equation, and the plume temperature and
mixing ratio begin to change due only to simple adiabatic mixing. Instead
of this procedure, collected liquid water should be allowed to evaporate
when the plume state would otherwise fall below the saturation curve. The
predicted plumes thus have no evaporative phase with attendant cooling of
the plume air. Clearly, the Lee-Batty procedure fails to conserve water by
a substantial amount.
In most models, the evaporative phase (during which the previously built-up
liquid water gradually disappears reducing plume buoyancy) occupies the
major portion of visible plume dispersion. Also, this model's visibility
criterion demands that plume relative humidity falls below 97% before
disappearance of the visible plume. This serves to lengthen the plumes
somewhat relative to other models. Both the high trajectories and the
extreme shortness of the predicted plumes can be traced in part to the omission
of a liquid water reservoir in the plume and the absence of an evaporation
phase.
Lee (NUS Corp.) Model [12]
The computer code for this model is proprietary to the NUS Corporation yet
the major details on the model development are in the open literature. We
were provided with an object deck of the computer code to run at ANL. We
could not therefore verify the coding of the model, which limits the confidence
we can place in results of our search for major causes of the model's
predictive behavior.
423
-------
The model accounts for entrainment from the theories of Morton, Taylor, and
Turner [30] and Briggs [31]. A drag force is assumed to act normal to the
jet centerline of the plume; the bent-over plume assumption is not made
while the Boussinesq assumption is followed. Kessler's theory [32] of
microphysics is applied to partition the total liquid water in the plume
into cloudwater and hydrometeor components. Rainout from the plume is
treated.
Visible plume rise predictions of this model rank among the top three or
four (cf. Table 8). However, the plume length predictions are usually-
short, with 28 of 38 length predictions less than 50% of observed. Visible
plume trajectories generally lie above the observed trajectories with the
greatest deviations near the tower.
We converted the model's equations for plume temperature, mixing ratio, and
liquid water into our standard form; these equations for plume thermodynamics
appear in Tables 4-6. The equations simulate the energetics of both condensation
and evaporation of liquid within a treatment of conservation of total water
in the plume. After the plume has lost its liquid water through dispersion
and evaporation, simple mixing equations are employed in the region of
plume subsaturation.
Lee employs a common form for the vertical velocity equation (see Table 3),
but an uncommon form for the horizontal momentum equation (see Table 2).
The horizontal velocity equation retains a fairly weak "pressure-drag"
force which increases Vx and an entrainment term, -uvx, which actually acts
to decrease vy. The usual expression derived for a Gaussian model (such as
the Lee (NUS) Model) is instead +y (u/2 - vx) where the plume downwind
coordinate satisfies dx/dt = 2vx. The effect of using -yvx instead of
y(u/2 - vx) should be to limit v^ to small values and prevent it from
approaching the wind speed. We attempted to test this expectation through
examination of the computer output for a number of cases. Although the
printout of the object code does not give vx, we were able to infer values
of vx from the plume slope and vertical velocity which were presented. We
found that the horizontal velocity of the plume does not remain at small
values and does indeed approach the wind speed but does so more slowly than
most other models.
Two representative model predictions are given in Figs. 14 and 15, one in
low winds and the other in high winds and both in cold temperatures. The
model has no diffusion phase which would allow the plume to disperse beyond
the first point where the updraft velocity becomes zero (maximum rise).
However, it was only 4 of 39 cases in which the plume was still visible at
maximum plume rise. Thus, in only those four cases, was visible plume
length not predicted.
In 5 of the 39 cases, the predicted plumes showed no signs of termination
after downwind integration of 5-10 km, and are regarded as yielding indefinite
results. These anomalous cases occurred when the predicted plume had nearly
leveled off at a height where the ambient air was saturated; the long
424
-------
plumes actually result from inaccurate trajectory predictions in those
cases.
Calabrese - Halitsky - Woodard (Pickard - Lowe - Garrick Inc.) Model [15]
The computer code for this model is proprietary although the details of the
model formulation are in the open literature. We were not able to examine
the computer code for the model, but instead we arranged to send input data
for our 39 single-tower test cases to K. Woodard at Pickard-Lowe-Garrick,
who ran the cases and provided us with model output. He chose not to use
all of the ambient profile data for the computer runs, but instead extracted,
from the data only those values which would have been obtained from an on-
site meteorological tower, as they would do in an assessment study for a
new power plant.
The model employs a simple form of the Briggs-Hanna plume rise formulas.
We do not provide details here since the proprietary values of the empirical
constants are not available. A power-law extrapolation scheme is adopted
to estimate elevated wind speed from ground-level values. The model uses
Halitsky's empirically-derived formulas [33] for a flow establishment phase
and an established flow phase of plume dispersion. The established flow
formulas lead to the choice of widths for the final Gaussian diffusion
phase. In the diffusion phase, where the plume has reached final rise,
both excess enthalpy and excess water vapor (above ambient) follow Gaussian
distributions which spread dependent upon the Pasquill stability class.
The tail of the Gaussian distribution which would continue below the ground
is treated by a method of reflection, which results in additional concentration
added to the above-ground distribution. The model formulation is similar
in spirit to the ORFAD model, but embodies a number of the improvements
suggested earlier for the ORFAD treatment. The model is not formulated to
provide predictions at wind speeds lower than about 1 m/s. Such plumes
from NDCT's are considered by the modelers to have low environmental impact.
The predictive performance of this model was the, best among the three semi-
empirical models (ORFAD, Saame, and Calabrese-Halitsky-Woodard) and compares
favorably with several of the integral models. The model performance is
remarkable in view of the reliance on surface weather data.
The predicted plumes followed no clear trends with respect to the observed
plumes, with the exception that the trajectories lay uniformly above the
observed ones. This random behavior may be due to uncertainties in the
extrapolation of surface weather data. Fig. 16 illustrates the inherent
limitations of wind-speed extrapolation for a particular case showing an
incorrect trend on plume trajectory. The 9 cases where complete predictions
were not made is implied by the value of Np in Table 8. These cases fell
into two classes. First, there were extremely long predictions, where the
outputs we received simply didn't include the end of the visible plumes;
expanded runs would probably define these plumes. (Fig. 17 shows the model
prediction for a very long case). Second, there were cases where no predicted
plumes were shown at all: in summer on dry days, and in very low winds.
425
-------
It should be noted that the first output point for the model prediction is
40 m downwind, and a finer grid would probably further define presently
ambiguous cases. Thus, the model's overall reliability for giving complete
(though not necessarily correct) predictions is probably better than the 9
missed cases would suggest.
Stephen - Moroz [14]
This model assumes a circular plume cross section with top-hat profiles
and does not adopt the Boussinesq approximation. The bent-over plume assumption
is made but the plume horizontal velocity is kept fixed at an average wind
speed rather than being allowed to vary according to the ambient wind speed
profile. The average windspeed, Q, is taken as the ambient wind speed at
tower top.
The model's entrainment assumption is the most complicated of any studied
to date. It attempts to include the additional effects of atmospheric
turbulence by means of an empirical parametrization suggested by Briggs [32].
The entrainment function also includes a "cloud-phase" entrainment rate
when the plume reaches heights in excess of 600 m above ground. The full
degree of complexity of the entrainment assumption is apparent from Table
1.
The model adopts a Lagrangian approach to formulate the plume equations by
following a plume parcel of length S into which ambient air mixes. To
follow a fixed mass of (diluted) plume air, one must allow S to vary.*
In the model, only the vertical mass flux, or the mass flux through a
horizontal plane is conserved. A more correct choice would have been to
take S/v to be a constant, where v = (w2 + a2)1/2. Thus, the mass flux
through a plane normal to the plume centerline would be taken as conserved.
In the model, then, as the plume vertical velocity tends to zero, so will
S, and excessive radial growth of the plume is needed to "conserve" mass
flux.
It is simple to translate the model's equations from a Lagrangian picture
to the Eulerian picture adopted in our presentation of model equations.
However, the reformulation of plume equations in terms of total mass flux,
VR2pp, rather than mass flux through a horizontal plane, necessitates
solving a system of implicit equations for the variation of primitive plume
variables with centerline distance. The resulting complications in the
model equations shown in Tables 1-6 may be a source of some of the unusual
plume behavior predicted by the model. It should be noted that this model's
equations do not lead to conservation of the usual fluxes. Further, the
additional factors and terms do not lead uniformly to identifiable trends
and effects. Thus, the discussion of this model's predictive characteristics
must be more descriptive without much motivation from the manner of treatment
of the physical effects.
*
This is effected by taking S/w to be a constant throughout plume development.
Thus as the plume vertical velocity tends to zero, so will S, and excessive
radial growth is needed to "conserve" mass flux.
426
-------
Notably, the model's radial growth is unrelated to any actual dilution of
plume air taking place as in other models which define an entrainment
function in an Eulerian scheme; rather, it is defined by the behavior of
the vertical velocity and the conservation equation of mass flux through a
horizontal plane. When the plume is still visible as w approaches zero,
extreme radial growth results. See Fig. 17 for an example of this effect.
However, the model does well for plumes rising in low winds, as shown in
Figs. 18 and 19, as one might expect since the conservation of plume mass
flux becomes better approximated by a vertical conservation equation and
therefore the model treatment is more appropriate for these low wind situations,
The model trajectories tend to lie well above the observed ones, due
mainly to the use of the tower-top wind speed as the average value; this
tower top value is usually less than the average wind speed over the observed
plume-rise region. The predicted plumes have a slight tendency to be long.
This trend is strengthened by examination of the 13 out of 39 cases where w
became zero before the predicted plume disappeared. Since the model makes
'no provision, as presently formulated, for plume dispersion after final
rise is attained, these plumes can be considered as longer still. The
predicted plumes also tend to have final rise values above the observed
ones, for reasons mentioned above concerning the trajectories and horizontal
lengths. It is interesting that the model's values for the absolute-log-
mean ratio of predicted-to-observed lengths are among the best when one
includes only the 23 of 39 cases where those ratios lie between 0.2 and
5.0. Also, on all measures of visible plume-rise, the model performed
comparably with the best-performing models, although 28 of 39 of those
ratios were above 1.0. This would seem to suggest that plume dynamics in
the vertical direction are better represented by the model equations than
for the horizontal direction.
Saarne Model [15]
The Saame model is the third semi-empirical model tested in this study.
Each of those models adopts a similar general approach. Water vapor and
enthalpy are Gaussian-distributed about an empirically determined centerline.
Ground- level atmospheric data are utilized and extrapolated upward as
needed. Tower thermodynamics are also simulated to obtain initial water
vapor and updraft velocity for the plume. These models, then, are set up
to be directly useful in environmental impact assessment. However, the use
of data at the ground to extrapolate to plume height and the simulation of
tower flow behavior can not be as accurate as the direct use of measured
values when available. Thus, much of the predictive inaccuracy of these
three models, can be traced to the use of degraded input data. An interesting
further study would be to test a modification of the models which would
employ the best measured data for test cases, while at the same time,
preserving the basic empirical formulations. This exercise would aid in
discovering the inaccuracies in the formulations themselves.
In Saame's model, the usual stability-dependent Gaussian widths for the
•427
-------
water vapor distribution are introduced, as functions of downwind distance.
Their form is based, again, on Pasquill's curves. A virtual origin upwind
of the tower is located, such that the centerline water vapor concentration
of the distribution through a vertical plane though the tower center equals
the average water vapor concentration leaving the tower. The effective
release height of the diffusing vapor is computed from the Carson-Moses
formula [34] using a wind, speed at tower top inferred from that at ground by
a power-law assumption. The model does not attempt to follow the plume's
trajectory from the tower exit to the effective stack height, but instead
the plume outline begins and is directed horizontally at the effective
height. Such an approximation is normally reasonable for calculation of
pollutant dispersal far downwind, but corresponds poorly with visible
outlines of relatively short plumes, which become invisible while still
rising (cf. Fig. 12). The model uses the actual ambient relative humidity
profile to delineate the region where plume water vapor saturates the
ambient.
The predicted visible plume outlines in this model occur at heights uniformly
above the observed outlines, even when the observed plume has reached final
rise. Of the 39 ratios of predicted-to-observed visible plume rise, only
7 are less than 1.0 in value. Use of an average wind speed over the plume-
rise region, rather than a tower-top value would generally improve this
behavior. Also, the predicted visible plumes seem usually to be shorter
than the observed plumes, with only 4 of 39 predicted-to-observed length
ratios exceeding 1.0 in value. Fig. 20 shows one of the predicted plumes
which is longer than the corresponding observed plume. The model fails to
make a prediction in two very low wind conditions, but otherwise provides
complete predictions. The model's performance statistics in Table 8 are
competitive with the more inaccurate of the integral models. One of its
better plume rise predictions is shown in Fig. 21.
DISCUSSION OF MODELS PERFORMANCE
Table 8 summarizes the performance statistics of each model. Numerous
criteria may be established to determine which models perform best. Each
criterion clearly is arbitrary to some degree. For example, we might
choose the better models by the number of cases the models predicted plume
characteristics within a factor of 2. On the other hand, one might want to
choose a model that gave the fewest complete misses (i.e., a large %
count). A criterion which seems to separate the better performing models
(trajectory, spreading, visible plume properites) is the following: model
predictions are within a factor of 2 for visible plume height and a factor of
2 1/2 for visible plume length and maintain these factors 501 of the time. The
models of Winiarski-Frick, Hanna, Slawson (Closed Form) and Tsai-Huang
(Stone § Webster) satisfy this criterion. Their performance for the prediction
of visible plume outlines is also best upon examination of our complete set
of graphs similar to those of Figs. 1-21. The graphical (and performance
statistics) results of the Frick and ORFAD models shows an overall poor
performance. All other models tested provide comparatively fair predictions.
428
-------
Several additional items should be noted. The four better-performing
models can themselves benefit from improvement, e.g. the Hanna model can be
improved with a diffusion phase added to the model along with a treatment
of downwash. Each model can benefit from a calibration of its experimental
coefficients to the full collection of available field data. Second, not
all model/data discrepancies are caused by model errors. Uncertainties in
tower exit and ambient measurements along with the transient nature of
plume dispersion make the acquisition of field data difficult. Each data
set contains some experimental errors in the measured model input conditions as
well as in the determination of visible-plume outlines. The disappearance of the
visible plume into natural cloud layers makes the plume hard to define in
some cases. Also, when the plume appears as a series of puffs, the precise
definition of a steady-state outline is difficult to determine. Also, time-
averaging of a series of time-lapse photos inevitably induces some error.
In spite of the experimental difficulties and experimental error in each
data set, the models performed very consistently with the data base as a
whole and this performance allowed us to trace the major model/data discrepancies
to model assumptions. It should also be recalled that the visible portion
of the plume is often only a small fraction of the entire plume and generally
represents only the topmost portion of the plume. Consequently, our model/data
comparisons test only the initial part of the plume. Only recently have
some European data on in-plume measurements of the visible and invisible
portions of the plume been made available which allow us to test, in a
limited manner, model predictions further from the tower. Such data complements
our large visible plume data base and provides additional information for
model validation and model improvement.
ACKNOWLEDGMENTS
This work was funded by the Nuclear Regulatory Commission. The authors
also wish to express their appreciation to the modelers whose work was
utilized and for their cooperation.
REFERENCES
1. A. J. Policastro, R. A. Carhart, W. E. Dunn, S. Ziemer, B. Devantier,
and K. Haake. Evaluation of Mathematical Models for the Prediction of
Plume Dispersion from Natural-draft Cooling Towers. Division of
Environmental Impact Studies. Argonne National Laboratory, (in
preparation). 1978.
2. A. J. Policastro, R. A. Carhart, and B. Devantier. Validation of
Selected Mathematical Models for Plume Dispersion from Natural-Draft
Cooling Towers. Paper presented and published in Proceedings of Waste
Heat Management and Utilization Conference. Miami Beach, Florida.
May 9-11, 1977.
3. P. M. Wigley and P. R. Slawson. "The Effect of Atmospheric Conditions
on the Length of Visible Cooling Tower Plumes." Atmospheric Environment.
9. 1975.
429
-------
4. P. R. Slawson and J. H. Coleman. "Natural-Draft Cooling-Tower Plume
Behavior at Paradise Steam Plant." Waste Heat Management and Utilization
Conference. Miami Beach. May 1977.
5. J. Weil. "The Rise of Moist Buoyant Plumes." Journal of Applied
Meteorology. 13. p. 435-443. June 1974.
6. W. E. Frick. TTThe Influence of Stratification on Plume Structure."
Master's Thesis. Atmospheric Sciences Department. Oregon State Univ-
ersity. June 6, 1975.
7. L. Winiarski and W. Frick. Methods of Improving Plume Models. IN:
Cooling Tower Environment -- 1978 Proceedings. A Symposium on Environmental
Effects of Cooling Tower Emissions. May 2-4, 1978. Power Plant Siting
Commission. Maryland Department of Natural Resources. Report PPSP-
CPCTP-22. WRRC Special Report No. 9. May 1978.
8. M. E. LaVerne. "Oak Ridge Fog and Drift Program (ORFAD) User's Manual."
ORNL/TM-5021. 1976.
9. S. R. Hanna. "Predicted and Observed Cooling Tower Plume Rise and
Visible Plume Length at the John E. Amos Power Plant." Atmospheric
Turbulence and Diffusion Laboratory. ATDL Contribution File No.
75/21. Oak Ridge, Tennessee. November 1975.
10. Y. J. Tsai and C. H. Huang. "Evaluation of Varying Meteorological
Parameters on Cooling-Tower Plume Behavior." Symposium on Atmospheric
Diffusion and Air Pollution. AMS Publication. Boston, Massachusetts.
p. 408-411. 1972.
11. K. B. Batty. "Sensitivity Tests with a Vapor Plume Model Applied to
Cooling Tower Effluents." Master's Thesis. Department of Meteorology.
Penn State University. November 1976.
12. J. L. Lee. "A Numerical Simulation of Atmospheric Convection Caused
by Heat Dissipation at Large Power Centers." American Meteorological
Society. Third Symposium on Atmospheric Turbulence, Diffusion and Air
Quality. October 26-29, 1976.
13. R. V. Calabrese, J. Halitsky, and K. Woodard. "Prediction of Temperature
and Moisture Distributions in Cooling Tower Plumes." Pickard-Lowe-
Garrick Inc., Washington, D.C. 1974.
14. D. W. Stephen and W. J. Moroz. "Plume Rise from Wet Cooling Towers in
Strong Winds." Engineering Research Bulletin B-107. Center for Air
Environment Studies. Penn State University, University Park, Pennsylvania.
September 1972.
15. J. Saame. "An Analytical Investigation of the Formation and Dispersion
of Fog Plumes from a Natural Draft Water Cooling Tower for Various
Meteorological Conditions." Master's Thesis. Department of Mechanical
Engineering. University of Pittsburgh. 1971.
16. H. Fortak. Use of a Power Glider Equipped with Instruments for Three-
Dimensional Structural Study of Man-Made Thermal Convection. Translated
from German. Report for the Deutschen Forschungsgemeinschaft. ORNL-
tr-4101. April 1974 - May 1975.
17. Arbeitsgruppe uber die meteorologischen Auswirkungen der Kuhlturme
"Bericht uber die meteorologische Messerie am Standort Liinen vom
27.11.1972 bis 1.12.1972 Zur Teiluberpru'fung and Verfeinerung des
numerischen Modells SAUNA." Dienst fur Luftreinhaltung der Schweizerischen
Meteorologischen Zentralanstalt Payerne. 17 Mai 1973.
430
-------
18' iJ?ii??S; :?halk Point Surface Weather and Ambient Atmospheric
?^?iS S ? f J Intensive Test Period, December 15-19, 1975 (Revision)"
Applied Physics Laboratory, The Johns Hopkins University. December
1976; Environmental Systems Corporation. "Environmental Systems
°at n °Tes Da
Data for the Peri<*i °f December 15-19,
19. J. H. Meyer and Jenkins, W. R. "Chalk Point Surface Weather and Ambient
Atmospheric Prof lie Data, Second Intensive Test Period, June 14-24,
I976' A£Plied physics Laboratory, The Johns Hopkins Univeristy.
January 1977; Environmental Systems Corporation. "Environmental
Systems Corporation Seasonal Test Data for the Period of October 1,
1975 - June 30, 1976." April 1977. (Chalk Point Cooling Tower Projects
Documents PPSP-CPCTP-11 and PPSP-CPCTP-12. Volume 2 of 2).
20. P. R. Slawson and J. H. Coleman. "Natural Draft Cooling Tower Plume
Rise Behavior at the Paradise Steam Plant, Part II." Tennessee Valley
Authority. Division of Environmental Planning. (Draft). 1976.
21. P. R. Slawson. Personal Communication. Department of Mechanical
Engineering University Waterloo, Waterloo Ontario, Canada, 1977.
22. L. N. Fan. "Turbulent Buoyant Jets into Stratified or Flowing Fluid."
California Institute of Technology Report No. KH-R-15. 1967.
23. F. Pasquill. "Atmospheric Diffusion". Van Nostrand Monographs.
London. 1962.
24. A. J. Policastro, W. E. Dunn, R. A. Carhart, M. Breig, S, Ziemer. and
J. Ziebarth. Progress Report to.-EPRI: Model Improvement Program for Cooling
Tower Plume Rise and Drift Deposition Models. January- March 1978.
Argonne National Laboratory. Argonne, Illinois.
25. L. Winiarski and W. E. Frick. "Cooling Tower Plume Model." Thermal
Pollution Branch. PNWRL. NERC EPA. Corvallis, Oregon. 1975.
26. G. A. Briggs. Plume Rise Predictions. IN: Proceedings of Workshop
on Meteorological and Environmental Assessment. American Meteorological
Society. 45 Beacon Street. Boston, Massachusetts. 1975.
27. P. R. Slawson and J. H. Coleman. "Natural -Draft Tower Plume Behavior
at Paradise Steam Plant, Part II." Tennessee Valley Authority.
Division of Environmental Planning, (to be published).
28. J. H. Meyer, T. W. Eagles, L. C. Kbhlenstein, J. A. Kagan, and W. D.
Stanbro. "Mechanical-Draft Cooling Tower Visible Plume Behavior -
Measurements, Models, Predictions" IN: Cooling Tower Environment -
1974. Eds. Steven Hanna and Jerry Pell. U.S. Energy Research and
Development Administration. 1975.
29. G. Abraham. "The Flow of Round Buoyant Jets Issuing Vertically into
Ambient Fluid Flowing in a Horizontal Direction." Delft Hydraulics
Laboratory Publication No. 81. The Netherlands. March 1971.
30. B. Morton, G. Taylor, and J. Turner. "Turbulent Gravitational Convection
from Maintained and Instantaneous Sources." Proc. of Royal Society of
London. Ser A. Vol. 234. p 1-23. 1956.
31. G. A. Briggs. Plume Rise. AEC Critical Review Series. TID-25075.
82 pp. 1969.
32. E. Kessler. On the Distribution and Continuity of Water Substance in
Atmospheric Circulations. Meteorological Monographs. Vol. 10. 84
pp. American Meteorological Society! Boston. 1969.
431
-------
33. J. Halitsky. International Journal of Air and Water Pollution. Vol.
10. pp 821-843. 1966.
34. J. E. Carson and H. Moses. "Calculation of Effective Stack Height."
Argonne National Laboratory Report ANL - 7220. pp 96-99. January
1967.
432
-------
=rick
Tible 1. Modal fetminwit Assumptions for
1 <*»«
»_ ar • *» **• **» I'*™
linUrski-Frick
" " ff if v1 • ve • «ax (vel, V(1A), s - 0.1
vel " «(»)« -Vs 'v-ucos 81
Manna'1'
0.6
If • ve " 6i (»•" =<» 8) • S,u cos 9 sin 9
- .(». 62 . .354, r - WT, » - u cos 9 . ^
'Lee-Battv(c)
• 0.1S
Stephen-Moroz
-093-
V » T:u:)1/2 , (w > l.OlSu, : <
, (w < l.OlSu, stable, : < 600-H)
vu 1/6
|.285 C^j) , fw < l.OlSu, unstable and neutral
: : < 600-H) j
(0.1).
min (z . H, 305)
600-H)
(a) This is the entraiiment rate for the "mcmentm plune" the largest cross-
section with radius %. The factor Em is the area of the cross-section
of the momentum plume divided by the area of the cross-section of the
temperature plume. &n is 1 initially and grows to :.:s late in plume
rise. R,, is the tower exit radius. The initial mas,s flux As increased
at the start of calculations by a factor of [1 » WQ-/^-]1'-.
(b) This model assumes Gaussian plume profiles and should be considered as
having a special set of initial conditions at the end of an empirical
";one for flow establishment". These special conditions provide initial
conlitions for the dynamic and themodvnamic equations solved in the model.
The updraft velocity at the end of IFE is 0.5 x0i the corresponding temper-
ature excess ana mixing ratio excesses are both halved; bp .; • R^; ami
v1 is the excess plume velocity along the trajectory at the centerline.
(.c) The ve equation is modified near or at maxima rise, depending on ambient
stability, and the cross-section may become elliptical. The equation
given above governs the major portion of plume rise.
433
-------
Table 2. Model liquations for Change of Horizontal Velocity with Center line
Distance.
Table 3. Model Hquations for Change of Vertical Velocity with Centerline
Distance
Model
Slawson-Wigley
Weil
I'rick
Winiarski-I-'rick
t lamia
Tsai -Huang
I.ee-Batty
l.ee
Stephen-Moroz
liquation for dv /ds
-T-— = -r- • v " u
ds ds ' x
dv
dT1 = "l"-V
Jv
jgi . „ (U - VJ
UVjc - ii(u-v )
X till
3s- = 3? • vx = "
dvx , Cj, UW3
Js x R n ^ ' 1)
dvx
Ts- ' " w '*
Jw
ils~ M
dw
3s"
u" u"
(fsr- - ")
u t = 1(1 + 0.31x), .22 -: i:w/l.m <_ l.U.
10 T. is a reference temperature, the ambient dry-bull) tunpcruture at tlic
lower exit.
-------
Ol
! 4. Model li|u:iliuiis lor IlKingc til' I'liiiiL- uni|M.kriiture with (V
Ilistui^e M!H;II IMufle is Satunittxi.
Neil
•rick
liqiutiuii lur d'l Als, fStitufatiHl IMiuifl
- •'«> *i (X,, - X,MI
p
Same as IT ick.
^'-M;-;-ii""'i.'p-y
II
T 1.711 „- IV \ l|. > " II. I
Jj. ' .1. ..
Js
" *" ,11 ll • - " "'Al'' '• "''"''' " "'i4' "'"'"" "
!>!.-• 1/2, I > llk/K -• «.7I; .IN - iHiiMH". l.ili:,'.
Tuble 5. Model Equations for Qiange of Plume Mixing Ratio with Centerline
Distance when Plume is Saturated.
Model
ilawson-Wigley
Veil
Hrick
Viniarski-Frick
llanna
Tsai -lluang
l.ee- Batty
Lee
Stephen -Moroz
Equation for dX /ds, (Saturated Plume) (a)
2t--"
-------
Table 6. Model liquations for Change of Liquid Water with Centerline Distance.
CM
Model
ilawson-Wigley
Veil
Frick
Winiarski-l-'rick
Hanna(b)
rsai-lluung
Lee-Batty
.00
Stephen-Moroz
liquation for do/Js'a'
E • °- ° • °
£ - -po * pd+a)-1 (Y(J S + „ [(Tp - Ta) - H'1 (Xp - Xa)|)
3§ --MO*^ l(Tp- V -r(Xp-xa)),
)
Same as Trick
£ . -B(I * V,-' {-,d S - jj_ (0.4W) lflp- V
^ <°-7" £ j£*,i| i;w -» 1/2, i i i\/u - i//z
Site
>,
C
IO
E
5
01
3
c
c
;^3
^
u
3
•M
C
S
,
2.29
-0.11
-0.91
-0.33
-0.94
-O.JO
-0.86
-0.42
-0.85
-0.95
-0.83
-1.44
-0.35
-0.79
-0.94
-1.50
-0.90
-0.93
-1.01
-0.82
-0.79
-1.06
-1.02
-1.13
-------
tD
V)
2
i
s
Model
Slawson-Wigley
Slawson (closed form)
Weil
Frick
Winiarski-Frick
ORFAD
Hanna
Tsai-ftjang
(Stone § Webster Engr. CorpJ
Lee-Batty
Lee(NUS)
Calabrese-Halitsky-Woodard
(Pickard-Lowe-Garrick Inc.)
Stephen-Maroz
Saame
Range of oi
0.0-9.3
0.0-16.8
0.12-6.1
0.0-3.83
0.13-8.6
O.S6-7.9
0.23-11.3
0.01-10.6
0.11-10.0
0.0-13.0
0.0-10.6
0.4-20.5
0.0-9.7
N2
19
23
IS
16
30
5
26
24
19
27
17
24
17
N5
33
32
32
36
36
11
35
32
32
35
28
35
36
NF
2
1
0
0
0
27
0
0
1
1
9
0
2
•-5&1
1.42
1.26
0.80
O.hi
0.83
2.34
1.39
0.92
0.92
1.18
1.54
1.44
2.08
-i
1.30
1.10
0.78
0.59
0.37
1.29
1.02
0.49
0.75
0.68
1.02
0.97
0.94
P2 - lf«t
1.88
1.72
2.13
2.19
1.50
2.2"
1.67
1.52
1.88
1.59
1.86
1.68
2.01
tog P^
":
0.22
0.19
0.19
0.23
0.13
0.19
0.18
0.14
0.23
0.17
0.17
0.19
0.17
S
c
i
^
0)
•H
•H
Model
Slawson-Wigley
Slawson (closed form)
Weil
Frick
Winiarski-Frick
ORFAD
Hanna
Tsai-Huang
(Stone 5 Webster Engr. Corp.)
Lee-Batty
Lee(NUS)
Calabrese-Hal i tsky - Woodard
(Pickard-Lowe-Garrick Inc.)
Stephen-Mjroz
Saame
Range of o^
0.0-48.9
0.0-299
0.03-24.3
0.07-2.51
0.08-16.0
"2
7
13
4
4
22
0.32-16.2
0.24-2.90
0.01-4.9
O.OS-3.9
0.0-5.3
0.0-5.4
0.1-3.8
0.0-24.4
0
18
19
8
5
13
16
11
N2.S
12
19
4
4
26
0
19
20
13
7
18
18
18
S5
23
26
18
IS
32
2
23
27
24
20
26
23
28
NF
4
•1
0
o
0
17
15
2
6
9
15
2
-•&>i
0.91
1.19
0.52
0.36
0.79
2.49
1.19
1.64
0*85
0.42
1.21
1.41
0.73
"1
1.02
1.09
0.56
0.13
0.50
2.17
0.72
1.08
0.90
0.22
0.91
0.98
O.S1
,, = io:/n£'los *i'
2.25
2.07
2.94
2.99
1.79
3.'9
1.57
1.79
7 ~>~
2.68
1.88
1.60
2.21
3:
0.18
0.17!
0.19
0.13
U.iS
I
o.ogi
0.16
0.19
0.15
0.20
0.17
0.19
0.15
Notes:
Pi is defined as the ratio of predicted to observed (either length or height as indicated)
N? is the number of times the prediction is xirhir. J factor of 2, i.e., 0.5 '• ^ < 2.0
\'~ , is the number of times the prediction :s kiti-.in .1 factor of 2.5, i.e., 0.4 ;^ - -\D
N'S is the number of times the prediction is within a factor of 5, i.e., 0.2 < ii ' 3.0
Np is the number of failures of the model in 39 data sets
oj is the standard deviation of the pi distribution
a2 is the standard deviation of the |log p;j distribution
Table 8. Performance Statistics for Thirteen Single-tower Models for Predictions of Visible Plume Rise and
Visible Plume Length.
437
-------
PARADISE—FEBRUARY 3.1973 (CASE P2-3)
Fig. 1.
OBSERVED PLUME
HANNA
SUW90N-WICLZT
ORFAD
UCEND
c = DRT BULB TEMP.
o=REl_ HUMIDITT
4 - WIND SPEED
WIND SPEED (M/3)
Comparison of Plume-model Predictions of Hanna, Slawson-Wigley
and ORFAD to Observed Visible-plume Outline at Paradise: February
3, 1973.
LLTNEN—RUN SS9—DECEMBER 224972 (1200 HRS.)
OBSERVED PLDH2
LESBMD
o - DRT BULB TEMP.
o- REL. HTjmm'r i
^ — wufD SP&KU
III
i
•I-
s*1
DISTANCE ?BOM TOWEB (METERS)
WIND SPEED («/3)
DRY BULB TEMP. (C)
o to «o ao ao uio
RELATIVE HUMIDITY (r.)
Fig. 2. Comparison of Plume-model Predictions of Hanna, Slawson-Wigley and
ORFAD to Observed Visible-plume Outline at Lunen: December 22 1972
(1200 Hrs.).
438
-------
LUNEN—RUN 34 NOVEMBER 30J972 (1000 WKS.)
puna
ajae. ram
PICX.-LOWI-GARJUCX
0-MTT BULB TIMP.
o- ML BUMBITT
> (M/S)
t • » u m i
PRES3UM (MB)
mat Town (MZTIM)
« • ,—rtt-i—,
DRT BULB TJMf.
ULATTVI aUMIDITT (X)
Fig. 3. Comparison of Plume-model Predictions of Stephen-Moroz, Slawson
(Closed Form) and Pickard-Lowe-Garrick to Observed Visible-plume
Outline at Liinen: November 30, 1972 (1000 Hrs.).
PARADISE- FEBRUARY 94973 (CASE PZ-t)
OBSERVED PLUM1
9TEPHEM-MOROZ
suAitsoH cua roui
MCK- LOWe-GAJUUCX
LEGEND
D - DRY BULB TEMP.
3 » REL. Hinnomr
A « WIND SPEED
WWD SPEED (M/S)
an *w HD uoe tj» um 011
DISTANCE PROM TOWER (METERS)
-
i-*'
-u -10 -«-«-'
DRY BULB TEMP. CC)
RELATIVE HflUDITY <^l
4. Comparison of Plume-model Predictions of Stephen-Moroz, Slawson
(Closed Form) and Pickard-Lowe-Garrick to Observed Visible-plume
Outline at Paradise: February 9, 1973.
439
-------
DECEMBER 16.1975 (133O-1310 HRS.)
OBSERVED puna
WKIL
nuac
a - ORT BULB TEMP.
O-RXL Himmmr
& - WIND SPEED
*-PRESSURE
DISTANCE FROM TOWER (METERS)
(M/3)
PRESSURE (MB)
DRY BULB TEMP,
RELATIVE HUMIDITY
Fig. 5. Comparison of Plume-model Predictions of Weil, Frick, and Winiarski-
Frick to Observed Visible-plume Outline at Chalk Point: December
16, 1975 (1230-1310 Hrs.).
PABADISE—FEBRUARY 34973 (CASE P2-5)
OBSERVED PLUME
LEGEND
3 * DRT BULB TEMP.
o - REL. mJMromr
t> . WtND SPEED
WD)D SPEED (M/S)
• • 10
*
t o
tti
KB HO 300 Ot
DISTANCE PROM TOWER (METERS)
t
i
t
DRT BULB TEMP. CC)
RELATIVE HUMTOITY
Fig. 6. Comparison of Plume-model Predictions of Weil, Frick, and Winiarski-
Frick to Observed Visible-plume Outline at Paradise: February
5, 1973.
44Q
-------
1
CHALK POINT I DECEMBER 18.1975 (1002-1113 HRS.)
OBSERVED PLUM! LEGEND
s:8T~
FRICK a - WL. _.....
wmuRSja-FRiac *'PREBURI
WIND SPEBD (MS)
PBEBURE (MB)
:||1
! §
iElJ
it
DISTANCE FROM Town OUTERS)
DRY BULB TEMP. (ER (METERS)
DRY BULB TEMP, (t)
RELATIVE HUMIDITY (7.)
Fig. 8. Comparison of Plume-model Predictions of Hanna, Slawson-Wigley
(no plume), and ORFAD to Observed Visible-plume Outline at Chalk
Point: June 24, 1976 (0940-1200 Hrs.).
441
-------
LUNEN—RUN S6 DECEMBER 14972 (0900 HHS.)
(M/3>
OBSERVED puna
WED.
nucK
WINlARSEI-nUCX
LEGEND
a - DRT BULB TEMP.
o> REL. HUMIDITY
a - WIND SPEED
+ -PRESSURE
PRESSURE (MB)
DISTANCE FROM TOWER (METERS)
-3-11 J 5 t 9
DRY BULB TEMP. CC)
RELATIVE HUMIDITY
Fig. 9a. Comparison of Plume-model Predictions of Weil, Frick, and Winiarski-
Frick to Observed Visible-plume Outline at Liinen: December 1, 1972
(0900 Hrs.) . . . small scale.
LUNEN—RUN 36 DECEMBER 1J.972 (0900 HRS.)
i
ial-j
111
•\\\
•in
iv.
\\\
OB3ERVED PLUME
mat
FDICE
a - DRT BULB TEMP.
o =» REL. HUUUJi'I'T
a =- WIND SPEED
+ -PRESSURE
WWD SPEED (M/S)
PRESSURE (MB)
1MB 1900 i«0 2KB MOO MOO t*00 MOO MOO
DISTANCE FROM TOWER (METERS)
DRY BULB TEMP. CC)
RELATIVE HUMIDITY (5!)
Fig. 9b. Comparison of Plume-model Predictions of Weil, Frick, and Winiarski-
Frick to Observed Visible-plume Outline at Liinen: December 1 1972
(0900 Hrs.) . . . large scale.
442
-------
CHALK POINT I DECEMBER 17.1973 (0928-1048 HRS )
si-
> >
OBSERVED PLUME
HAXMA
SLAW SON-WIG LEY
ORFAD
j - REL HUMIDITY
> - WD»D SPEED
DISTANCE mow Town (UTTERS)
DDT BULB TOO.
RELATTVE HUMmrnr
Fig. 10. Comparison of Plume-model Predictions of Hanna, Slawson-Wigley, and
ORFAD to Observed Visible-plume Outline at Chalk Point: December
17, 1975 (0928-1048 Hrs.).
LUNEN — RUN SSI7 — JANUARY 2+J973 (12OO HRS.)
OBSERVED PLUME
HAltNA
SLAWSON-WIGLET
a - REU HUMIDITY
.:. - WIND SPEED
SPEED <-/3)
DCTtANCZ FROM TOWER (METZRS)
-< -I • 1 I l
DRY BULB TEMP, fc)
HUMmrrr (%)
Fig. 11. Comparison of Plume-model Predictions of Hanna, Slawson-Wigley and
ORFAD to Observed Visible-plume Outline at Lunen: January 24, 1973
(1200 Hrs.).
443
-------
PARADISE—FEBRUARY 1U973 (CASE P2-12)
a
>
OBSERVED PLUMB
LEt-BATTY
STONE—WEBSTER
LEGEND
a - DRY BULB TEMP.
o- REL. HUMIDITY
A - WIND SPEED
WIND SPEED (M/S)
DISTANCE FROM TOWER (METERS)
-* -+ -3 -* -I 0 I 2
DRY BULB TEMP. PC)
RELATIVE HUMIDITY (X)
Fig. 12. Comparison of Plume-model Predictions of Lee-Batty, Stone § Webster
and Saame to Observed Visible-plume Outline at Paradise: February
11, 1973.
LUNEN—RUN SS7—DECEMBER 20.1972 (1200 MRS.)
OBSERVED PLUME
LEE-BATTY
STONE-WEBSTER
SAAME
LEGEND
' DRY BULB TEMP.
= REL. HUMIDITY
• WBID SPEED
WIND SPEED (M/S)
DISTANCE FROM TOWED (METERS)
DRY BULB TEMP. PC)
RELATIVE HUMIDITY (7.)
Fig. 13. Comparison of Plume-model Predictions of Lee-Batty, Stone $ Webster
and Saame to Observed Visible-plume Outline at Liinen: December
20, 1972 (1200 Hrs.).
444
-------
PARADISE—FEBRUARY 34873 (CASE PZ-3)
OBSERVED runa
L8SEHP
O-DFT BULB TEMP.
0-MB- HUtflPITT
a- WDfD SPEED
(M/a)
DISTANCE FROM TO*E8 (METERS)
DRY BULB TEMP
RELATIVE HUMIDITY (7.)
Fig. 14. Comparison of Plume-model Predictions of Lee (NUS) to Observed
Visible-plume Outline at Paradise: February 3, 1973.
CHALK POINT I DECEMBER 154973 (0903-1146 HRS.)
OBSERVED PLUME
LEE MODEL
LECBHD
O - DRT BULB TEMP
o > REL HUMIDITY
4 . WD«D : ~~
UBJMKI FROM TOWB OOTERS)
WWD SPEED (M/S)
DRT BULB TEMP. CO
KBJOTVI HUMUMTT (X)
Fig. 15. Comparison of Plume-model Predictions of Lee (NUS) to Observed
Visible-plume Outline at Chalk Point: December 15, 1975 (0902-
1146 Hrs.).
445
-------
PARADISE—FEBRUARY 3,1973 (CASE P3-3)
OBSERVED PLUME
STEPHEN-MOROZ
SLAWSON CLOS FORM
PICK-LOWE-GARR1CK
LEGEND
a -, DRY BULB TEMP.
o - REL HUMIDITY
» o WIND SPEED
WIND SPEED (M/3)
no uoo IMO ITDO am 2300 an awo
DISTANCE FROM TOWER (METERS)
DRY BULB TEMP. CC)
RELATIVE HUMIDITY
Fig. 16. Comparison of Plume-model Predictions of Stephen-Moroz, Slawson
(Closed Form) and Pickard-Lowe-Garrick to Observed Visible-plume
Outline at Paradise: February 3, 1973.
LUNEN—RUN 3S11-JANUARY 10.1973 (1200 MRS.)
OBSERVED PLUME
STEPHEN-MOHOZ
SLAWSON CLOS. TORM
PICK.-LOWE-GARR1CK
LEGEND
- = DRY BULB TEMP
; = REL HUMIDITY
a = WWD SPEED
wmo SPEED IM s)
J-
II-
DISTANCE PROM TOWER (METERS)
DRY BULB TEMP. (
-------
LUMEN—RUN 339—DECEMBER 220872 (1200 BBS.)
I
§1
gl
3
O
5!
H
i.
!•
i-i
O-DKT BUUD TEMP.
O* ML HUMPIrT
a- vutu
DI3TAMCI niOM TOwra (UTTERS)
-« -J 0 1 *
DRY BULB TEMP. (TT)
RELATIVE HUMUHTT (-.)
Fig. 18. Comparison of Plume-model Predictions of Stephen-Moroz and Slawson
(Closed Form) to Observed Visible-plume Outline at Liinen: December
22, 1972 (1200 Hrs.).
CHALK POINT U JUNE 234976 (1030-1230 HRS.)
I (M/3)
CUB. ram
SrSESajfiT ~W<&
ll
r
i]
11
'
I-I,
'-PBB8URI
DISTANCE FROM TOWKR (METERS)
DRY BULB TEMP. (TT)
RELAITVS
Fig. 19. Comparison of Plume-model Predictions of Stephen-Moroz and Slawson
(Closed Form) to Observed Visible-plume Outline at Chalk Point:
June 23, 1976 (1030-1230 Hrs.).
447
-------
PARADISE—FEBRAURY 5,1973 (CASE P3-5)
OBSERVED PLUME
LEE-BATTY
STONE-WEBSTER
SAAME
LEGEND
c = DRY BULB TEMP.
; = HEL. HUMIDITY
^ = WIND SPEED
WIND SPEED (M/S)
Il
111
i
goo TOO ooo uoo LJOO iaco ITOT LDOO 2100
DISTANCE FROM TOWER (METERS)
DRY BILB TEMP.
30 » 00 « TO 73
RELATIVE HUMIDITY (£)
Fig. 20. Comparison of Plume-model Predictions of Lee-Batty, Stone § Webster
and Saame to Observed Visible-plume Outline at Paradise: February
5, 1973.
CHALK POINT I DECEMBER 18,1975 (1003-1113 HRS.)
OBSERVED PLL'ME
LEE-BATTY
STONE-WEBSTER
SAAME
LEGEND
z = DRY BULB TEMP.
-: = REL. HUMIDITY
J = WIND SPEED
- = PRESSURE
WIND SPEED (M/S)
PRESSURE (MB)
600 TOO OOO 000 1000 UOO
a
M =
DISTANCE PROM TOWER (METERS)
-a -ao -is -10 -5 o
DRY BULB TEMP. CC)
1020304430907880
RELATIVE HUMIDITY (7.)
Fig. 21.
Comparison of Plume-model Predictions of Lee-Batty, Stone § Webster
and Saame to Observed Visible-plume Outline at Chalk Point: December
18, 1975 (1002-1113 Hrs.).
448
-------
EVALUATION OF METHODS FOR PREDICTING
PLUME RISE FROM MECHANICAL-DRAFT
COOLING TOWERS
W. E. Dunn, G. K. Cooper* P. M. Gavin
University of Illinois at Urbana-Champaign
Urbana, Illinois U.S.A.
ABSTRACT
This paper evaluates various methods commonly used to predict the length
and height cf the visible plume produced by an array of mechanical-draft
cooling towers by comparing predictions with observational data from the
Benning Road Power Station and from a smaller array of towers at the
Purdue University Power Plant. Four different approaches - empirical,
integral, cloud-physics and finite-difference - are examined. Statistical
estimates of predictive capability are given. Problems inherent in the
application of these approaches are discussed. Observations concerning
areas of weakness and thus areas of potential improvement are made.
INTRODUCTION
Disposing of large quantities of low grade heat has long been a major
problem for industry and, in particular, the electric utility industry.
Historically, the waste heat naturally generated in the conversion of
thermal energy to mechanical energy has been rejected to nearby lakes and
rivers. In recent years, however, the shortage of available once-
through sites combined with stringent federal regulations on the use of
natural waterways has dictated the adoption of alternative methods of
cooling. The most popular alternatives, at present, are natural-draft
and mechanical-draft cooling towers.
These systems which transfer large quantities of thermal energy and
moisture to the atmosphere may also produce adverse environmental impacts.
Most of these impacts are directly or indirectly related to the vapor
plume formed in the atmosphere by the thermal energy and moisture
release. Thus, for the assessment of potential adverse effects, know-
ledge of the way in which the plume disperses in the atmosphere is
essential.
In this paper, we evaluate several different methodologies commonly used
to predict the length and height of the visible plume produced by
*Now at Mississippi State University.
449
-------
proposed mechanical-draft cooling tower installations by comparing predic-
tions with observational data from the Benning Road Power Station and
from a smaller array of towers at the Purdue University Power Plant. This
work provides a measure of the reliability with which predictions can now
be made, but more importantly provides valuable insights into the weak-
nesses of various approaches and into methods of improvement.
THE MATHEMATICAL MODELS
The methods tested represent four different types of approach - empirical,
integral, cloud-physics and finite-difference.
To some extent all models are empirical. In this context, however, an
empirical model is one based solely on observational data with little or
no theoretical basis. Such methods are clearly undesirable from a
fundamental viewpoint, but are often the only recourse in situations for
which the physics is very poorly understood. Moreover, extensive data
analysis may uncover basic scaling relationships that are otherwise
obscured by the complexity of the problem. Unfortunately, the one purely
empirical model tested did not perform well with data from a higher power
site indicating that the essential scaling relationships were not
captured.
The integral and cloud-physics methods are virtually identical and it may
be argued should not be distinguished. Both are one-dimensional analyses
in which the dominant physical process is entrainment; the principle dif-
ference lies in their historical origins and consequent emphases. Integral
models, evolved from the theory of jets, generally have no or only a
simple treatment of thermodynamics. The cloud-physics models, on the
other hand, include more detailed treatments of cloud processes such as
condensation and evaporation and, in general, consider several different
forms of liquid water.
Finite-difference models attempt to treat the physics of the problem on
a microscopic level. These models, although more difficult and expensive
to use, offer the potential of handling more complex tower configurations
in which factors such as downwash and merging are significant problems.
This potential cannot be realized, however, if the processes themselves
are poorly understood or if unacceptable simplifications are necessary
for computational tractability (assuming a principle direction of flow,
for example). Since these models are in an early stage of development,
what can now be said is subject to revision as new and better techniques
become available.
The most common approach to plume modeling is based on the plume-rise
theory put forth by Briggs [1] and subsequently extended by Hanna [2,3].
This theory is itself an extension of the now-classic integral analysis
of jets presented principally by Morton, Taylor, and Turner [4]. Simple
450
-------
algebraic relationships derived from this theory are in widespread use
under the names "Briggs," "Briggs-Hanna," and "Hanna-Briggs" plume-rise
formulae. These formulae represent closed-form solutions to the integral
conservation equations once a great many simplifying assumptions are made.
For practical application of the formulae as a predictive tool, much
schematization of the problem is necessary. The ambient must be idealized
in terms of a single wind speed, humidity, lapse rate, and reference tem-
perature. The complex thermodynamic nature of the plume must be repre-
sented in terms of a single dynamic parameter, the initial buoyancy flux.
Owing to the large number of ways of implementing these plume-rise
formulae, all of which appear at least superficially to be reasonable,
the variation among ostensibly identical formulations can be great.
Moreover, the empirical coefficients inherent in these formulae can be
varied on the basis of many different justifications which merely multi-
plies the number of predictions one can obtain.
The usual recommendation is that ambient conditions be "averaged" over
the height of the "plume." Of course, when the formulae are being used
in a predictive sense, the height of the plume is unknown. One common
alternative is to average over some fixed height such as 500 m. A more
involved procedure is to determine ambient conditions iteratively, i.e.,
by alternately calculating average ambient conditions and plume height
until convergence is achieved. This latter approach appeared to us to be
preferable and was thus used. Interestingly, this iterative procedure
failed to converge in a large number of cases. Rather the calculated
plume height would begin to flip-flop between two different values. In
such cases, we took the simple arithmetic average of the two values as
the predicted plume height. In addition, we found the common practice of
using the wind speed at tower top gave very poor predictions.
Determination of atmospheric stability (stratification) can also be a
source of difficulty. In general, an integrated or weighted average of
temperature lapse rate over the height of the plume was found superior to
the usual two point method, although the latter is typically satisfactory
owing to a weak or missing dependence of predictions on stability. More-
over, use of the moist lapse rate was found to be undesirable since it
made unstable some cases which were stable based on the dry lapse rate.
Exit buoyancy flux F is defined as
where g is acceleration of gravity, WQ is plume exit speed, p is ambient
density, R is plume °xit radius, and Ap is the density difference
between the ambj_ _ and the exiting plume.
In the computation of Ap, it has been suggested by Hanna and others that
some account should be made of the latent heat of the plume. This is
usually done by assuming that some fraction A of the latent heat is
451
-------
converted into sensible heat. Thus, the effective plume exit temperature
TjJ is given by
To =To + A(hfg/Cp)Au,
where T is the actual exit temperature, h is latent heat of vaporiza-
tion, C is specific heat of air, and Aw is the mixing ratio of the exiting
plume llss that of the ambient.
We found that inclusion of latent heat gave too much buoyancy making the
plume too high and short. Of course, one might argue that the entrainment
coefficient should also be modified for moist buoyant plumes.
Plumes from several closely grouped sources must be treated in terms of a
Single source as supposed in the theoretical development of the plume-
rise formulae. One extreme assumption is to simply merge all plumes
instantly upon exit and sum all fluxes. A second approach is presented
by Briggs [5]. He analyzes plume merging in terms of an enhancement
factor which treats, at least heuristically, the relative distance
between sources.
A simple computer code was written to facilitate intercomparison of the
various plume-rise formulae and the comparison of plume height and length
predictions with experimental data. In addition, computer codes were
acquired for the models presented by Eagles and Kohlenstein [6], McVehil
[7], and Commonwealth Associates [8]. These models are based in large
part on the Hanna-Briggs plume-rise theory, but were modified and/or
calibrated for predicting mechanical-draft tower plumes.
In addition to the several integral models tested, one model from each
of the other three categories was examined. May, et al. [9], have
adapted their one-dimensional cloud model to prediction of mechanical-
draft cooling tower plumes. This model integrates conservation equations
for vertical momentum, energy, and water content vertically using a
classical entrainment hypothesis and the assumption that the plume moves
horizontally at the local wind speed. Thus, this model is quite similar
to the "bent-over" plume analysis of Briggs, yet handles an inhomogenous
ambient and contains a more thorough thermodynamic analysis.
The model by Smith and Agee [10] was developed for Indianapolis Power
and Light Company from four sets of winter plume observations. This
model was included in the study since it represents an empirical model
developed from mechanical-draft cooling tower data.
Compared to the above closed-form models, the two-dimensional finite-
difference model presented by Taft [11] is considerably more difficult
and expensive to use. Although the computer code was obtained from the
author, a considerable amount of programming time was expended converting
certain machine-specific operations, reworking the input format to be
452
-------
compatible with the standardized format of the data and algorithmatizing
and programming calculation of the "equivalent" initial conditions
required by the model, a procedure originally done by hand. A typical
production run of this program requires from 10 to 20 minutes of execution
time on our CDC CYBER-175 and occupies about 80,000 words of memory.
Owing to the significant cost involved, only a few such runs were made
for comparison with available data.
THE OBSERVATIONAL DATA
A significant obstacle to the validation and improvement of mathematical
models for mechanical-draft cooling tower plumes is the scarcity of
observational data. For this study, we had only two sets of data avail-
able to us, one of which may be too flawed to give meaningful results.
The better of the two is the set of data collected by Meyer, et al. [12],
at the Benning Road Power Plant near Washington, D. C. during the period
from October 25, 1973 through February 26, 1974. The plant with a
generation capacity of only 289 MWe has two linear rectangular mechanical-
draft units of 8 cells each. The towers are situated on relatively level
terrain near the Anacosta River. The only obstruction is a nearby
incinerator. The cells are 21.3 meters high with an exit radius of 4.7
meters. The axes of the two tower structures lie in the east-west direc-
tion. The field study conducted by Meyer, et al., included measurements
of (a) entrance and exit temperatures of the cooling water, plant load,
and cooling water flow rate; (b) exit dry-bulb and wet-bulb temperatures
for a single instrumented cell; (c) exit air speed profiles taken for the
instrumented cell from which volumetric flow rates were calculated; (d)
Rawinsonde profiles of ambient temperature, water vapor density, and
wind speed; (e) surface-level wind direction and pressure; (f) cloud
height and stability class; and (g) height and length of the visible
plume obtained from photographs taken normal to the wind direction.
This set of data, composed of over 50 observations, is. reasonably com-
plete and is tabulated in a form that is fairly convenient to use. One
deficiency is the lack of visible plume outlines by which to test addi-
tional aspects of the models.
The second set of data employed in this study was gathered by Westlin [13]
at the Purdue University Power Plant during the period between July 9,
1971 and October 10, 1971. This plant has a single tower 18.1 m high
consisting of 3 cells, each with an exit radius of 4.6 m. Although the
terrain is reasonably flat, the power plant which is considerably taller
and longer than the tower is located directly east of the tower. Each of
the 30 observations includes (a) difference between entrance and exit
temperatures of the cooling water; (b) exit temperature of the plume;
(c) volume flux calculated from the manufacturer's fan ratings; (d) near
surface measurements of temperature, relative humidity, pressure and wind
speed (with a threshold of 1 m/s); and (e) visible plume heights and
453
-------
lengths extracted from photographs of the plume. Again, visible plume
outlines are missing and, in addition, no ambient profile data are avail-
ajle. Documentation is also less complete compared with the Benning Road
data. Most distressing is the fact that for many of the cases in which
the wind speed was reported to be below the 1 m/s threshold of the
anemometer, the length to rise ratio of the plume is much larger than in
those cases for which the wind speed is reported to be greater than 2 m/s.
Sadly, the original records and photographs were lost.
COMPARISON OF MODEL PREDICTIONS WITH OBSERVED PLUME HEIGHTS AND LENGTHS
A few simple statistics will be used to gage model performance. They are
6 = ratio of predicted to observed plume parameter (either
length or height as indicated by subscript),
N = number of observations in data set,
6 = maximum value of 6 for given data set,
6 . = minimum value of 6 for given data set,
Y = log-mean value of 6 defined as
1 N
Y = exp - jZj
f = fraction of cases (in percent) for which 0.5 <_ 6 <_2.Q,
fl = fraction of cases (in percent) for which 0.2 _< 6 _< 5.0,
f~ = fraction of cases overpredicted, i.e., for which 6 > 1.0,
and
f = fraction of cases for model failed to give a prediction
for any reason.
Tables 1-3 summarize these statistics for the several models tested. An
explanation of the motivation for using these statistics is in order.
The maximum and minimum values of the 6 ratio are meant to indicate
worst case performance. In contrast, the log-mean ratio Y is meant to
characterize model performance in an average sense but one in which
overprediction and underprediction are treated on equal footing and one
in which an error by a factor of 4 is treated as being twice as bad as
an error by a factor of 2. The fraction of cases for which the predic-
tion is within a factor of 2 of the observation is meant to gage overall
model performance taking into account the limitations of the data.
Based on typical uncertainties for the measured parameters and baring
major blunders on the part of the field workers, the range in predictions
one might typically expect is roughly a factor of 2. Of course, model
predictions are most sensitive to uncertainty in the ambient values for
high relative humidity predictions. The fraction within a factor of 5
indicates the percentage of time that the model prediction is totally out
of bounds as compared with the data and thus also aids in assessing worst
case performance. Finally, the fractions of overprediction and model
454
-------
failure are important for identifying systematic behavior in model/data
discrepancies.
One may argue that these statistics do not reflect the weighting one might
use in the analysis of potential environmental impact where, for example,
the overprediction of short plumes by a factor of 2 is not nearly as
undesirable as is underprediction of long plumes by a factor of 2. Here,
however, we are most interested in determining to what extent the models
reproduce the basic scaling relationships of the physical problem. In
this sense, it is equally important to predict both short and long plumes
accurately.
For the Benning Road data set, overall model performance is good. Only
the model of Smith-Agee gives systematically poor predictions being con-
sistently too short and low by a factor of roughly 5. Among the different
models based on the Hanna-Briggs analysis, the assumption of fully merged
plumes and calculation of buoyancy flux without latent heat additions
proved superior. In general, we found that formulae developed and cali-
brated using dry-plume data perform better than those developed in an
attempt to include thermodynamic effects. There may well be several
compensating influences at work.
There is virtually no significant distinction among the various models
(except for those already noted). Predictions are with a factor of 2
roughly 80% of the time with a mean error factor of 1.5 to 1.7. It must
be pointed out that all of these models were at least partially cali-
brated against the Benning Road data; data from a second larger installa-
tion would be most valuable in testing model capabilities.
The model by Taft was consistently too short and low in its. predictions
by a factor of roughly 4 in rise and 6-10 in length for the three com-
parisons that were made. An explanation for this behavior is presently
lacking.
Model/data discrepancies are considerably greater for the Purdue data.
All of the models tend to overpredict both plume rise and length. Only
the cloud-physics model of May e_t_ al. performed well with this data.
These results must be viewed cautiously owing to the uncertainties
in the data and to the fact that most plumes were quite short so that
large factors of overprediction may not be very serious. The reason(s)
for the good performance of the cloud-physics model are not presently
known. The model of Smith and Agee did better with this data.
SUMMARY AND CONCLUSIONS
This paper considers and evaluates by comparison of predictions with
observation data four different approaches to plume modeling. The
results of this study are: (a) Among the integral formulations based
455
-------
on the results of Briggs and Hanna, the "dry-plume" formulae give better
predictions of plume rise and length than do the formulae which attempt
to treat thermodynamic effects. (b) The integral models predicted plume
character very well at the Benning Road Station but not as well at the
smaller Purdue Power Plant. One explanation is perhaps that several of the
models were at least partially calibrated with the former data. (c) The
cloud-physics model did well at both sites although the reason for this
result was not determined. (d) The empirical model tested did not do
well at the Benning Road site suggesting that it did not conatin the cor-
rect basic scaling relationships, (o") The one finite-difference model gave
plumes consistently too short and low. A poor understanding of the micro-
scale physics treated by this model is a possible explanation, (f) Treat-
ments of plume merging and downwash are poor. However, the common approach
of instantaneous merging did well with the Benning Road data perhaps again
because of favorable calibration.
456
-------
REFERENCES
1. G. A. Briggs, Plume Rise. AEC Critical Review Series, Report TID-25075,
1969.
2. S. R. Hanna, Predicted and Observed Cooling Tower Plume Rise and Visible
Plume Length at the John E. Amos Power Plant, AE, 1£, 1043-52 (1976).
3. S. R. Hanna, Meteorological Effects of the Mechanical-Draft cooling
Towers of the Oak Ridge Gaseous Diffusion Plant, Cooling Tower
Environment - 1974, CONF-740302, 1975.
4. B. R. Morton, G. I. Taylor, and J. S. Turner, Turbulent Gravitational
Convection from Maintained and Instantaneous Sources, Proc. Roy. Soc.
(London), Ser A, 234" ;-23 (1956).
5. G. A. Briggs, Plume Rise from Multiple Sources, Cooling Tower
Environment - 1974, CONF-740302, 1975.
6. T. W. Eagles and L. C. Kohlenstein, A Cooling Tower Visible Plume
Prediction Model Based on Measurements, presented at Fifty-fifth
Annual Meeting of American Geophysical Union, held in Washington, D.C.
1974.
7. G. E. McVehil, Personal Communication, Boulder, Colorado, May, 1977.
8. Y. H. Huang, Personal Communication, Commonwealth Associates,
Jackson, Michigan, January, 1977.
9. L. E. May, H. D. Orville, and J. H. Hirsch, Application of Cloud Model
to Cooling Tower Plumes and Clouds, Institute of Atmospheric Sciences,
South Dakota School of Mines and Technology, Rapid City, South Dakota,
September, 1977.
10. P. J. Smith, and E. M. Agee, Empirical Equations for Mechanical Draft
Cooling Tower Plumes, prepared for Indianapolis Power and Light Company,
Indianapolis, Indiana, 1969.
11. J. Taft, Numerical Model for the Investigation of Moist Buoyant
Cooling Tower Plumes, Cooling Tower Environment - 1974, CONF-740302,
1975.
12. J. H. Meyer> T. W. Eagles, L. C. Kohlenstein, J. A. Kagan, and W. D.
Stanbro, Mechanical Draft Cooling Tower Visible Plume Behavior:
Measurements, Models, Predictions, Cooling Tower Environment - 1974,
CONF-740302, 1975.
13. P. R. Westlin, A Field Study and Analysis of Cooling Tower Plumes,
M. S. Thesis, Purdue University, 1972.
457
-------
TABLE 1. MODEL PERFORMANCE STATISTICS FOR THE
BF.NNING ROAD PLANT
Haima- Briggs/S*
Hanna-Briggs/S
Hanna-Briggs/SL
Hanna-Briggs/M
Hanna-Briggs/M
Hanna-Briggs/ML
Eagles-Kohlenstein
Eagles-Kohlenstein
Smith-Agee
Smith-Agee
May, et al.
May, et al.
McVehil
McVehil
Commonwealth
Commonwealth
Parm
H
L
L
H
L
L
H
L
H
L
H
L
H
L
H
L
0.18
0.17
0.10
0.45
0.28
0.17
0.35
0.46
0.03
0.08
0.54
0.40
0.36
0.32
0.20
0.23
2.5
6.5
1.8
0.8
3.0
1.3
4.4
3.6
0.6
0.9
2.9
10.
4.0
3.2
4.2
9.3
I '
2.4
2.7
4.5
1.7
1.8
2.6
1.6
1.4
7.6
4.1
1.5
1.6
1.6
1.5
2.0
1.8
f,
42
22
2
84
74
22
90
94
8
8
86
88
70
88
50
84
96
96
100
48
100
94
100
100
40
68
100
98
100
100
100
96
percent
10
4
0
56
18
4
42
36
0
0
74
46
30
24
46
46
0
0
0
0
0
0
0
0
4
4
0
0
0
0
0
0
"S =
single cell; M = multiple cell plumes merged instantly; L
heat added to buoyancy flux
= latent
458
-------
TABLE 2.
MODEL PERFORMANCE STATISTICS FOR THE
PURDUE PLANT
Hanna-Briggs/S*
Hanna-Briggs/S
Hanna-Briggs/SL
Hanna-Briggs/M
Hanna-Briggs/M
Hanna-Briggs/ML
Eagles-Kohlenstein
Eagles-Kohlenstein
Smith-Agee
Sraith-Agee
May, et al.
May, et al.
McVehil
McVehil
Commonwealth
Commonwealth
Parm
H
L
L
H
L
L
H
L
H
L
H
L
H
L
H
L
percent
rain
max
0.03
0.16
0.98
0.59
3.5
0.53
4.3
1.1
2.2
5.8
2.4
2.4
34.
65.
16.
8.6
I '
6.6
2.8
2.4
11.5
3.4
2.4
5.4
2.2
2.5
2.5
1.8
1.6
10.0
4.8
8.4
2.7
1 *.
2
0
44
63
0
38
63
6
63
63
50
38
50
0
25
0
44
fe
5
50
88
81
13
81
88
44
81
81
88
69
69
13
63
13
75
f
0
100
88
44
100
88
63
100
50
44
31
63
56
100
88
100
100
0
0
0
0
0
0
0
0
0
0
31
31
0
0
0
0
single cell; M = multiple cell plumes merged instantly; L
heat added to buoyancy flux
latent
459
-------
TABLE 3. MODEL PERFORMANCE STATISTICS FOR THE BENNING
ROAD AND PURDUE PLANTS COMBINED
Hanna-Briggs/S*
Hanna-Briggs/S
Hanna-Briggs/SL
Hanna-Briggs/M
Hanna-Briggs/M
Hanna-Briggs/ML
Eagles-Kohlenstein
Eagles-Kohlenstein
Smith-Agee
Smith-Agee
May, et al.
May, et al.
McVehil
McVehil
Commonwealth
Commonwealth
Farm
H
L
L
H
L
L
H
L
H
L
H
L
H
L
H
L
min
'max
0.35
0.46
0.03
0.08
0.98
0.40
0.36
0.32
0.20
0.23
14.
24.
2.2
5.8
2.9
10.
34.
65.
16.
9.3
Y
3.1
2.7
3.9
2.7
2.1
2.6
2.1
1.6
5.8
3.6
1.6
1.6
2.5
2.0
2.8
2.0
£
*2
32
27
17
64
65
32
70
86
21
18
74
79
53
73
38
74
fs
85
94
95
40
95
93
86
95
50
73
92
91
79
91
79
91
percent
f
0
32
24
11
67
35
18
56
39
11
8
71
48
47
40
59
59
F
0
0
0
0
0
0
0
0
3
3
10
10
0
0
0
0
single coll; M • multiple coll plumos merged instantly; L
heat added to buoyancy flux
latent
460
-------
ENVIRONMENTAL COST OF POWER PLANT WASTE HEAT
AND CHEMICALS DISCHARGE IN TROPICAL MARINE WATERS
J.M. Lopez
Center for Energy and Environment Research
Mayaguez, Puerto Rico
ABSTRACT
A1125 MW, oil-fired power plant has been discharging once-through cooling
water in Guayanilla Bay, Puerto Rico for several years. The heated effluent
enters an enclosed cove connected to the eastern portion of the bay by a
narrow mouth. This Thermal Cove and surrounding area once supported lux-
uriant growth of mangroves, seagrasses and the associated plant and animal
life of these ecosystems. In this study, a critical review was made of pre-
vious investigations concerned with various aspects of the intake and dis-
charge area environments. An assessment of environmental degradation due to
the power plant activity is offered which considers losses in the major eco-
system components and the accumulation of hazardous chemicals in the receiv-
ing system.
INTRODUCTION
The South Coast electric power plant complex, located in Guayanilla, Puerto
Rico is comprised at present of six oil-fired generating units with a total
generating capacity of 1125 MW. Once-through cooling water is withdrawn
from Guayanilla Bay at a rate of 1,597,882 1pm (604,800 gpm). A 10°C temp-
erature ri'se is imposed on the cooling water which is discharged through a
100 m canal into a nearly enclosed lagoon 900 m long by 200 m wide. The
heated lagoon, referred to as Thermal Cove, in turn discharges into Guaya-
nilla Bay through its 30 m wide mouth. In addition'to heated seawater, the
power plant effluent contains chemical contaminants such as petroleum hydro-
carbons, trace metals (Cu, Cr, Hg, Zn and Ni) and chlorine emanating either
from the power plant or from otherindustries that discharge wastewaters in
Guayanilla Bay. The ecosystem in the area affected by this heated effluent
is composed of seagrasses, mangroves and soft bottom communities. Seagrasses,
predominantly Thalassia testudimum communities, occur in shallow flats (1 m)
immediately outside of the mouth of the Thermal Cove. Red mangroves, Rhizo-
phora imangle are the dominant trees in the community that fringes the entire
eastern shore of the Thermal Cove. Similar communities are found in the
cooling water intake region in Guayanilla Bay. Several investigations have
been conducted at this laboratory into the nature and extent of the ecosystem
response to this heated effluent. For these, comparative ecological studies
were performed using the cooling water intake region as a comparison site.
The configuration of the receiving area, being nearly enclosed, is such that
most of the impact of the heated effluent on the environment is localized (Fio 1)
461
-------
This allows for an assessment of ecological losses and consequences to the
environment within distinct physical bounds. In this paper I review the ex-
isting information on losses in the biota of the Thermal Cove and present
recent data on the levels of hazardous chemicals accumulated there as a re-
sult of the plant's operation over the years. The ecological losses or
changes and the accumulation of chemicals and their significance represent
an estimate of the environmental cost of discharging waste heat and chem-
icals associated with the generation of electricity in tropical marine waters.
ECOLOGICAL STUDIES
Kolehmaiisen et .a1!. (1) summarized results of studies performed by Puerto Rico
Nuclear Center scientists in the Guayanilla Bay Thermal Cove. About 95% of
zooplankters entering the cooling system were killed in the condensers and
in the discharge canal. Within 100 m of the end of the discharge canal (8°C
above ambient) further zoopankton mortality was observed. However, in areas
at 4°C and 6°C above ambient, species diversity and biomass were higher than
in areas at ambient temperature. Most of the mortality apparently occurred
in the long exposure (14 min.) to high temperatures in the discharge canal
with only 10% of the mortality occuring in condensers and pumps.
Benthic macroanimals were absent from the area within the Thermal Cove (1).
This can be the combined effect of erosion by the strong effluent stream,
high temperature and chemical toxicity. The heated effluent exits the Ther-
mal Cove as a surface layer leaving the benthos unaffected outside of the
Thermal Cove. Loss of benthic macroanimal communities extends to the area
occupied by this lagoon.
About 19 ha of fringing mangroves (predominantly Rhizophora mangle) line the
eastern shore of Guayanilla Bay including the Thermal Cove. Trees nearest
the discharge canal and at the mouth of the lagoon were killed by sediment
erosion from around.their roots caused by strong currents generated by the
power plant effluent (1). Seedlings from red mangrove trees in the Thermal
Cove are smaller than those from areas not affected by the heated effluent
and showed very low probability for survival and growth (2). Banus and
Kolehmainen (2) found no young rooted and growing seedlings in the Thermal
Cove. In 1974 they found most trees showing visible signs of stress such as
lesser and smaller leaves and few and small seedlings.
Kolehmainen et al. (3) found that mangrove root communities in the Thermal Cove
decreased in species diversity as temperatures increased above 34°C. In
Thermal Cove mangrove roots a species of barnacle and the tree oyster made up
most of the biomass. The lowest biomass occurred near the canal discharge
where currents and temperatures are greatest.
462
-------
Fish fauna within the heated lagoon (28 species) was less diverse than in
the intake area (53 species). The dominant species in the Thermal Cove were
the mojarras and the sea bream (1). Recent fishing efforts by gill nets and
trawling has yielded little or no fish while some success is obtained at the
mouth of the lagoon. No significant fishery appears to exist in the Thermal
Cove as a result of the power plant discharge.
Schroeder (4) determined that seagrass (Thalassia) beds occuring near the
mouth of the Thermal Cove and receiving" tnlTKeated effluent contained less
plant material (leaves and root biomass) per unit area than comparison areas
within Guayanilla Bay.
CONTAMINATION BY CHEMICALS
Chemical contaminants moving through the power plant cooling system may be
augmented by corrosion losses of trace metals and petroleum spills or oil
and grease discharges. Chlorine may react with organic substances at the in-
creased temperature of this effluent to form substances that are more hazard-
ous. The potential toxicity of this discharge, however, has not been deter-
mined. It is conceivable that synergism may exist in the combined effects
of chemicals and heat. Bottom sediments of natural water systems are the
repository of substances occurring in the overlying waters and their compo-
sition reflect the longterm chemical regime of the said waters. The chemi-
cals discharged by the power plant accumulate in the sediments of the Thermal
Cove.
Surface sediment samples were obtained from the Thermal Cove and intake areas.
Portions were dried and ground and wet digested with modified aqua regia and
hydrogen peroxide prior to atomic absorption analysis for Cd, Cu, Ni, and Zn
and Cr. Mercury determinations were performed by flameless atomic absorption.
The sediments were also soxhlet-extracted with a benzene-methanol mixture
and content of petroleum hydrocarbons was determined gravimetrically after
saponification.
Results (shown in Table I) demonstrate that the Thermal Cove sediments have
accumulated significant amounts of Cd, Cu, Ni, Zn, and petroleum hydrocarbons
containing increased levels relative to the intake and adjacent areas. The
Thermal-Cove sediments in fact, showed the highest conentration of Zn, Cu,
Cd, Ni and petroleum hydrocarbons of any other area in Guayanilla Bay. These
data imply an enrichment by the power plant of the load in the waters of
Guayanilla Bay. The Thermal Cove acts as a catchment basin and sink of po-
llution. The increase from intake to discharge area in ug/g is from 1.6 to
2.3 for Cd, from 47 to 117-for Cu, from 20 to 47 for Ni and from 0.19 to 0.34%.
No appreciable enrichment was observed for Cd, Cr and Hg although it is evi-
dent that at least translocation of chemicals freely occurs. The actual
chemical form of these substances or whether the power plant chemically
changes them is not known.
463
-------
The accumulated chemicals are likely partially responsible for the lack of
benthic organisms. In addition, this pool of hazardous substances can be
potentially available for assimilation in the biospere. These can become
biologically available when associated to organic matter such as detrital
particles of mangroves or zooplankton which abound in the area. Through that
route the hazardous substances may become magnified in the marine food web
that eventually reaches the human consumer. In effect, other research at
this laboratory (5) has shown that trace metals become available to turtle
grass and mangrove tissue in Guayanilla Bay. In particular, mercury accumu-
lation and cycling by mangroves can be substantial compared to known sources
to the bay. Mercury is actually permeating the biospere and levels in some
top predators, consumed by humans, show biomagnification to unacceptable
levels.
CONCLUSIONS
The environmental costs of producing electricity using coastal tropical waters
for cooling can be approximated by considering losses to the biota of the
Thermal Cove in Guayanilla Bay. Studies have shown that the area of the
Thermal Cove (900 m X 200 m) is essentially denuded of life or severely im-
paired as a habitat. That area is lost for shelter, feeding or breeding
grounds of most animals. It is lost for fishing also. Several hectares of
mangrove forests are destroyed or severely deteriorated. Losses in the
overall efficiency of the Thalassia ecosystem outside of the cove are also
in evidence where lower biomass of these important plants occur in the affec-
ted areas. These effects on the aquatic environment do not consider possible
environmental cost due to possible air pollution resulting from combustion
of petroleum by the power plant.
One important impact of the heated effluent discharge that is often over-
looked is the addition of chemical contaminants such as trace metals and pe-
troleum hydrocarbons to the receiving system. The data presented, demonstrate
a net accumulation of these hazardous substances in the sediments confined to
the Thermal Cove area. This is a cost to the environment since the area is
rendered virtually a biological desert partially due to chemicals and heat. The
chemical accumulation can have far reaching consequences as it may provide
a long lasting pool of biologically available hazardous substances. This
can lead to widespread bioaccumulation that can affect human health and also
impair commercial fisheries through fish-flesh tainting by hydrocarbons,
toxicity and biomagnification to unacceptable levels of substances such as
mercury, as has been shown to be the case in Guayanilla Bay. The environ-
mental cost of the Costa Sur plant waste heat and chemical discharge includes
the area of the Thermal Cove and the ecosystems it represents. This cost,
however, goes beyond that because of the remote, long-lived effects of the
accumulated chemicals.
464
-------
ACKNOWLEDGMENTS
Sampling and chemical analyses were performed by J.A. Ramirez Barbot, L.L.
Cruz, S. de la Rosa and D.D. de Caro. This work was supported by the U.S.
Department of Energy.
REFERENCES
1. Kolehmainen,S.E., F.D. Martin and P.B. Schroeder "Thermal Studies on Tro-
pical Marine Ecosystems in Puerto Rico" In:"Environmental Effects
of Cooling Systems at Nuclear Power Plants" IAEC, Vienna (1975)
2. Banus, M.D. and S.E. Kolehmainen "Rooting and Growth of Red Mangrove Se-
edlings from Thermally Stressed Trees" In: 'Thermal Ecology II,
Esch, G.D. and McFarlane, R.W. (eds.) ERDA (197F]
3. Kolehmainen, S.E., T. Morgan and R. Castro "Mangrove Root Communities
in a Thermally Altered Area in Guayanilla Bay, Puerto Rico" In:
Thermal Ecology, Gibbons, J.W. and Sharitz, R.R. (eds.), USAEC,
(1974)
4. Schroeder, P.B. "Thermal Stress in Thai assia Testudinum" Ph. D. Disserta-
tion, The University of Miami (1975)
5. Lopez, J.M. and H.J. Teas "Trace Metals Cycling in Mangroves" Symposium
on Trace Metals Cycling in Coastal Plant Ecosystems, American Bota-
nical Society, Virginia Polytechnic Univ. (1978)
465
-------
TABLE I
RELATIVE CONCENTRATION OF CHEMICALS
IN SEDIMENTS OF INTAKE AND THERMAL COVE
Cadmium, ug/g
Copper, ug/g
Chromium, ug/g
Nickel, ug/g
Mercury, ug/g
Zinc, ug/g
Petroleum hydrocarbons, % dry wt.
INTAKE
1.6
47
34
20
0.46
57
0.19
THERMAL COVE
2.3
117
37
47
0.50
91
0.34
466
-------
GUAYANILLA BAY
500
FIGURE I
LOCATION OF STUDY ARFA IN GUAYANILLA BAY
467
-------
THEORY AND APPLICATION IN A BIOLOGICAL ASPECT
T. Kuroki
Tokyo University of Fisheries
Minato-ku, Tokyo, JAPAN
ABSTRACT
From a view-point of bio-thermodynamics, the author expounds
the theory of thermal stimulus on aquatics. The strength of
thermal stimulus depends upon temperature change (8), but not
upon temperature (T) itself in a poikllothermal animal. For
the purpose of good management and utilization of waste heat,
data should be retreated along the understanding of stimulation
(0) in a short period response of animal and of the stimulation
quantity (Q=6-t) in a long period response. These Interpre-
tation under the relations between responses of aquatics and
ln(9) (or ln(Q)) are shown in several examples. A new plank-
ton collector (MERI-0 type) which is to record environmental
factors in real time will be introduced.
INTRODUCTION
For a nice management to protect a marine ecosystem against
waste heat and for an effective utilization of waste heat to
aquaculture, exact informations about a thermo-dynamical phys-
iology and a temperature conditional ecology of aquatics is
essential. In general, the body temperature of poikilother-
mal animal is almost equal to the temperature of surrounding
water under a steady condition for about 20 minutes or more.
So, the water temperature (T°K) itself is only a physiological
condition for fish, it can not be stimulus for fish-body.
The thermal stimuli for fish should be the temperature change
(inceasings or decreasings ±AT°K). Then, the thermal irri-
tability (Weber ratio) of aquatic animal can be expressed as
"A9/9", not as "AT/T"; where AT or 6 is a temperature differ-
ence from the initial state, and A9 is a differential value of
9 by small space or by short time.
Tamiya(1952)^ had discussed about the threshold curve(Welss' *
hyperbola) that the susception at the part of longer duration
of stimulus on this curve follows to "Weber's Law" (A9/9Con-
stant) and that, in the part of shorter duration, the suscep-
tion for stimulus follows "Quantity Law" (AQ/Q;constant, where
Q=9-t and AQ=A9-t or AQ=0-At). In Tamiya's theory, the order
of length (short or long) of period of stimulus duration should
be determined relatively with the length of period in order to
468
-------
recognize the response of living body change. Therefore, for
the plans of management and utilization of waste heat, it is
yery important to select the scale order of space or time
length best fitted to the biological aspect. In this paper,
only aquatic animal(fish) is studied; but the theory and its
develop^c*an be applied also to other animals, even to insects
or vegetables.
THERMAL SUSCEPTION OF FISH
Two examples of thermal susception of fish under various water
temperature are shown here. One of them is an electro-physi-
ological experiment by Kuroki^Land the other is a study of e-
cological migration by HanaokasJ.
Physiological aspect
First, electrifying threshold curves(Welss' hyperbola) of carp
were obtained in various water temperature (5-35°C) as Fig. 1
of example at 15°C. From these curves, rheobase (q) and
chronaxie (p) at various temperature were determined as Fig. 2.
Here, it should be recognized that the rheobase q is the minimum
strength of stimulus in electrifying threshold; on the other
hand , chronaxie p is the stimulus duration where the quantity
of electric energy in stimulus for threshold is minimum. The
value of p shows the shortest at 26°C and that the value of q
shows the least at 13°C. That is, fish selects a zone of
water temperature being nearer to one of the two temperatures
according to physiological needs.
Ecological aspect
Hanaoka (1972) ^had proposed the "migration triangle" to show
the environmental conditions for fish schools under their mig-
ration in the fishing ground. In Fig. 3» the relations between
water-temperature and salinity (osmotic pressure) along the
migration course of fish schools are drawn as triangle shapes
ckockwise or anti-clockwise with months elapsed. There is a
certain range of moderate water temperature for fish, and fish
schools migrate from a higher temperature zone to a lower one
in a stage of their life cycle, and they migrate by otherwise
despositlon in the other stage. It should be noted that fish
does not always select a single optimum temperature, but that
they will select various temperatures in a certain range.
Thermal stimulation on fish
Usual kinds of fish are poikilothermal animals, therefore their
body temperatures are almost equal to the surrounding water
'temperature. If there would be any difference between two temper-
469
-------
atures of body and water, it should be only ±0.2°C; + means
body temperature is higher than water temperature, perhaps, in
an exothermal state of fish and - means lower body temperature
in an endothermal state. ^ As a special case, several kinds
of fish (tuna and shark) have a system of keeping warmness, so
called "wonderful net". 9 In case when fish school run into
the water mass where a temperature differs withAT0 (= 90 ) from
the other water mass at time t = -t0, after a delay t0 of heat
transmission to the sensor of fish, the strength of thermal
stimulation is exponential to t, (cf. Fig. 4).
e=e.-e'et CD,
where, Q ; temperature difference at t=t
0C; temperature difference at t=-t0~0
Q j exponential constant
(2).
Here, several assumptions can be developed as follows:
sensitive organ, there is a certain reversible system;
In a
(3),
where k and k' in the relation are rate constants in the pro-
cess^indicated by arrows. When the system receives a stimulus
of thermal strength 8 , following change of the reaction occurs
with an additional rate constant Ks6(proportional to 9 ),
k'
Dennoting (P1), the concentration of P' will increase as fol
lows;
470
-------
A certain parameter G may be assumed, which is quantitatively
related with the occurrence of response as a sign of irrita-
bility. A small definite value of AG corresponds to a defi-
nite magnitude of response. The significance of this para-
meter is that it is quantitatively related to the free energy
(P) determined by the concentration of P' in the following way;
constant )
The relation between F and the concentration of P1 is given by
the Second Law of thermodynamics, thus
( R ; gas constant )
( T ; absolute temperature ).
From the equations (6) and (7) ,
dG _<*RT dlP"
d6 P' d9
(8).
From the equations (2) and (5),
(k»ks6XP]-klPl
-ce C9)-
Substituting (9) to (8)
471
-------
For simplicity, let us introduce K,
(ID.
Integrating the equation do) under the condition that T and K
are nearly Independent of 9,
/3 >
or
G= *RCTK
*y*£ ; constant ).
v»
The equations (12) and (13) show that the value of response G
has linear relation with logarithmic value of 9. Further-
more , from the equations (10) and (11), the Weber ratio in
thermal stimulation follows;
A9 AG-C
~ =
-------
(12) and (13), and obtained Fig. 6. In this figure, K > 0 or
k'[P'] > (k+ks9)[P], in other words, leftward reaction is
stronger than rightward reaction at all time in the equation
(l|); and k'[P'] is quite higher than (k+ks9)-[P] in summer (B-
line in Pig. 6), and not so higher in winter (D-line).
Thermal motive power for upstream or downstream movements of
rainbow trout fry from the data of Northcote (1962)u is shown
in Fig. 7. The number of movement is seemed to be propor-
tional to a certain quantity of temperature chage. This quan-
tity is found to be the sum of two decreases between the maximum
temperature of one day(afternoon) and the minimum temperature of
the nex day(early morning) in 3 consecutive days. The results
of these examples prove the equations (12) or (13); that is, the
magnitude of response G have the proportional relation with the
logarithm of temperature change 9.
INVESTIGATION OF THE INFLUENCES OF WASTE HEAT
We can measure the temperature continuously at any point in the
mixing area of waste warm water and surrounding cool water, but
it is very difficult to measure quantitatively the responses
(various changes of behavior or shape) of aquatic animals in the
same area.
Generally, it is very difficult to select biological factor to
measure, because the relations between the response of aquatic
animals and mixing water are compQicated and abstruse. In the
physiological or ecological states of poikilothermal animal,
"warmness" is essential for the animal life, and "coolness" is
also as the same important as "warmness"; but the necessary
period of wamness or coolness should be quite different in vari-
ous stages. Some proposals about methods to investigate the
influences of waste heat on aquatic animals will be descibed in
this section:
In case of relatively long duration stimulus
When the response of aquatics G can be measured in the intervals
of several seconds, the thermal stimulus 6 (or temperature
change in mixing zone of waste heat) will be relatively long
duration for aquatics, because the 9 must be transmitted to sen-
sors of fish-body in scores of seconds. In this case, the in-
fluence of water temperature on aquatics can be considered along
the law of animal irritability in the equation (14). If the
minimal quantity of response AG is constant, then the equation
(1*0 is equal to Weber's Law (A9/9 = constant). And, it will
be proper to measure the water temperature with the Eulerian
method in this case; that is, measuring apparatus can be set on
fixed points in the mixing zone.
In the case relatively short duration stimulus
473
-------
The thermal stimulus 6 should be transmitted to the fish-body
in the time length of several scores seconds just same as the
above case, but the stimulus duration is relatively shorter
than the period of response in this case. For example, the
response period of aquatics is the time length of hours or days,
and the stimulus is transmitted twice or more times, consecutive
ly or interruptedly, during the period. In this condition,
the influence of temperature on aquatic animals should follow
the "Quantity Law" of vegetable irritability;
(15)-
Tamiya had proved theoretically the equation (15). If AG is
set constant, the value of AQ/Q will be constant as of Weber's
Low in a broad sense. To investigate the phenomena of "Quan-
tity Law", it is necessary to apply a Lagrange's method; for
example, a telemetry apparatus to measure continuously the posi-
tion of fish and the surrounding water temperature at the same
time.
Practical period and distance to measure in the sea.
It takes only a few seconds to recognize the response AG of a
swimming behavior of fish under a sharp temperature gradient.
But it may take several days or months to recognize the response
AG or G such as vernalization and maturity of aquatics. When
there is a spatial temperature gradient of about 0.02°C per
meter in the sea and if fish swim with the speed 0.1~2.0 m/sec
the water temperature should be measured once per 10 seconds ~
twice per one second, because the threshold of fish against tem-
perature change is estimated about ±0.02°C . Then measuring
apparatus must be set at every one meter distance. For an-
other example, usually plankton and fish-egg can be drifted with
vortices or currents at velocity range 1~100 cm/sec. If the
threshold of plankton is assumed to be about ±0.5°C, then the
measuring period should be once per 25 sec. in high velocity
current (lm/sec) and may be once per 2500 sec. (about 42 minutes)
in low velocity (Icm/sec.).
CONCLUSION
One should know the role of thermal stimulus on the sense of
aquatic animals. When treating the data for utilization of
waste heat, the strength of thermal stimulation 0 and the quan-
tity of stimulation Q ( = 6-t ) must be understood in the strict
sense. The apparatus to measure the influences of waste heat
on the poikilothermal animals should be designed under the full
474
-------
understanding of 9 and Q . in Japan, a special purpose ap-
paratus to collect planktons and fish-eggs in the sea off the
nuclear power stations is under design and will be set up at
the Marine Ecology Research Institute (Pig. 8).
REFERENCE
1. H. Tamiya (1952): "A New Interpretation of the Weber Law and
the Weiss Law"; Cytologia, Vol. 17, No. 3, pp. 2*13
-269 .
2. T. Kuroki (1954): "On the Relation between Water Temperature
and the Response for Stimuli"; Mem. Fac. Fish. Ka-
goshima Univ. , Vol. 3, No. 2, pp. ig3^.
3. T. Hanaoka et al. (1972): "Studies on the Osmotic Pressure
of Environmental Media and Body Fluid of Marine
Fish (I)"; Bull. Jap. Soc. Sci. Fisheries, Vol. 38,
pp. 1351-135~£7~
4. T. Kuroki (1967): "Thermal Stimulation on Fish"; Bull. Jap.
Soc. Sci. Fisheries. Vol. 33, pp.264-27^
5. F. G. Carey (1966): "Fishes with Warm Bodies"; Sci. American,
Vol. 212, No. 4, pp. 36-44.
6. H. Hirata (I960): "Diural Rhythm of the Feeding Activity of
Goldfish with Special Reference to the Inflection
point of Temperature change"; Bull. Jap. Soc. Sci.
Fish. Vol. 26, pp. 783-791.
7. T. G. Northcote (1962); "Trout Migration at Loon Lake"; Jour.
Fish. Res. B. Canada, Vol. 19, No. 2, pp. 201-270.
•475
-------
at 15 °c
Fjg.i. Example of Weiss-curves Fig. 2. Chronaxie and Rheobase
Water temperature °C
35 30 25 20 15 10 5
iii ^"T^^TT™^™"T"
Vvolt
5
. 2
I*~
I 1
a —o—-snout to
snout to e
msec
1.5
1.0
0.8
0.6
0.4
volt
2.0
1.5
0 0.2
-------
Fig.5. Feeding activity of gold fish. (HiraU, 1960)
9 15
Sept. 20
21
3 9 15 21
Sept. 21 hour
Fig. 6.
L.-1
en
•cK
104
Log. relations .§
04-
between temp.
increasing and {3 0.2
feeding acti- £
vity.
Range
A;May(17- 23'c)
B;Aug(18-25°c)
C;Sep.(13- 18 °c)
D;Nov.(11- 14 dc)
Fig.7.
Downstream
and upstream h
movement of -8
rainbow trout
fry E
(After Northcote^
50 100 150
Feeding frequency
h'1
50 100
Number of fry/ trap(4~6 hours)
477
-------
oo
©General measure
(thermister,
salinometer,
turbidity-meter)
:reening mesh
Fig.8. "MERI-0" plankton collector and measuring system.
Plan
• ^w* W»* ••• IU I I *«• «^f | '
(with a mechanism.
of back flushing)
) Rotary main net
and precise counter
of plankton (with
size recorder )
® Suction pumps
Section
Side view
( Depressor and stabilizer, not illustrated)
-------
OCCURRENCE OF HIGHLY PATHOGENIC AMOEBAE
IN THERMAL DISCHARGES
J. F. De Jonckheere
Laboratorium voor Hygiene
Katholieke Universiteit Leuven, Belgium
ABSTRACT
Primary amoebic meningoencephalitis and amoebic meningoencephalitis are
human diseases caused by free-living amoebae Naegleria fowleri and
Acanthamoeba spp. respectively.
In a region with temperate climate, N. fowleri is present in its infec-
tive stage in warm discharges of different industries. In control sur-
face waters this pathogen could not be demonstrated. The incidence of
high concentrations of pathogenic N. fowleri in warm discharges stresses
the need of controlling this agent by disinfection of the cooling water
or the need to prevent its dispersal in nature by using closed-cycle
cooling systems. The use of saltwater would also prevent its growth.
Also the number of pathogenic Acanthamoeba spp. can be increased in warm
discharges, thereby contributing to the abundance of this pathogen in
the environment. Moreover Aoanthamoeba spp. are very resistant and
therefore very difficult to eliminate. The epidemiology of Acanthamoeba
infections is however still obscur.
INTRODUCTION
Some 20 years ago, free-living amoebae of the genus Aoanthamoebae were
isolated from tissue cultures and recognized for the first time as
potentially pathogenic [1]. It was suggested that free-living amoebae
might produce disease in man.
The first human cases of infection by free-living amoebae were reported
in 1965 in Australia and the disease was thought to be caused by Aoantha-
moeba sp. [2]. In 1966 three cases of infection by free-living amoebae
were reported in the USA [3] and the disease was named primary amoebic
meningoencephalitis (PAM) to differentiate it from secondary infections
of the brain caused by Entamoeba histolytica.
Only in 1968 the causative organism of PAM was recognised as Naegleria
sp. [4, 5}. This pathogenic Naegleria was in 1970 identified as dif-
ferent from N. gruberi, a common amoeboflagellate found in water and
soil, and it was named N. fowleri [6] in honor of Dr. Malcolm Fowler,
who first recognized the disease. The separate species identity of N.
479
-------
fowleri, has since been confirmed by different techniques.
After the first reports in Australia and the USA, cases of PAM were recog-
nized throughout the world ; for a review see Willaert [7].
Pathology
The majority of human infections was caused by N. fowleri. To differen-
tiate PAM from infections caused by Acanthamoeba spp. the latter was
called amoebic meningoencephalitis (AM) [8], PAM usually affects young,
previously healthy people. It has an acute course and the incubation
time is probably 3 to 7 days. Diagnosis is usually made postmortem.
Almost all known cases had a history of recent swimming or water contact,
and the amoeba is entering through the nose.
Infections occurred after swimming in chlorinated swimming pools in Belgium
[9] and Czechoslovakia [10], in lakes in Virginia [11] and Florida [3], in
hot springs in California [12] and New-Zealand [13], in thermal polluted
waters in Belgium [14] and Czechoslovakia [15], and after contact with
chlorinated tap water in Australia [16].
AM caused by various species of Acanthamoeba occurs in debilitated indi-
viduals. It is usually a more chronic disease resulting in death after
several weeks or months [8.1. The actual site of invasion by Acanthamoeba
is not known. Strains of Acanthamoeba have also been identified in
chronic diseases of the human eye in the USA [17] and England [18].
^
Ecology of the amoebae
Free-living amoebae are widespread in nature, especially strains of
Naegleria and Acanthamoeba can be found in every natural water.
Reports on the isolation of pathogenic N. fowleri- and pathogenic Acantha-
moeba spp. from the environment have been very scarce after the discovery
of human infections by these amoebae.
Pathogenic N. fowleri. were reported to be isolated in 1972 in Australia
from tap water [16] and in 1973 from soil [19], in India in 1972 [20]
and in the USSR in 1973 [2l] from sewage sludge samples, and in 1974 in
Poland [22] and in 1975 in Belgium [23] from thermal polluted water.
Isolations of pathogenic N. fowleri from thermal polluted waters have
since been confirmed in Belgium [24], Poland [25], Florida and Texas
[26], and from sewage in Korea [27 ]. Pathogenic N. fowleri- were also
isolated from freshwater lakes in subtropical regions [28| and on one
occasion from the nose of a boy who had been swimming previously in a
lake [29].
While Acanthamoeba spp. can be isolated from almost every environment,
pathogenic Acanthamoeba strains have been isolated from tissue cultures
in the USA [I, 30] and France [31], from sewage sludge in India [20],
from hot water springs in New Zealand [32], from swimming pools in
480
-------
Czechoslovakia [33], Belgium [34] and France [35], from salt water in the
USA [36] and from lakes in Poland [25].
SAMPLING OF THERMAL POLLUTED WATERS IN BELGIUM
Because of the importance of water with high temperatures in the infection
with N. fowleri and the ability of pathogenic N. fowleri and pathogenic
Acanthamoeba spp. to grow at higher temperatures than nonpathogenic
strains [37], we have investigated the impact of thermal discharges on the
occurrence of pathogenic Naegleria and Acanthamoeba in a region with tem-
perate climate.
Screening for pathogenic N. fowleri (Fig. 1) was started in 1974 while a
selective screening for pathogenic Aoanthamoeba (Fig. 2) in thermal
discharges was started beginning of 1978, although pathogenic Aoanthamoeba
had been infrequently identified in swimming pools and aquarium samples
previously [38].
N. fowleri investigations
Regular sampling for pathogenic N. fowleri were initiated in May 1974 and
extended till August 1977. During this period the warm discharges of 39
different factories have been investigated in Belgium (Fig. 3). Part of
these were sampled at intervals of 2 to 6 months. In three warm water-
bodies the occurrence of pathogenic N. fowleri was also determined
quantitatively- Isolations were performed during summer as well as
during winter. The isolation procedure has been published [39], while
the method for quantitation is in Press [40].
In eight of these factory discharges and the surrounding surface water,
pathogenic N. fowleri were found on several occasions. These pathogenic
strains were highly virulent, killing all mice within 5 days when in-
stilled intranasally. That these sites may be dangerous for human health
is shown by the occurrence of PAM after swimming in one of these thermal
polluted waters [14].
During almost every sampling of these eight factories, nonpathogenic
variants of N. fowleri [41] were isolated. It was first thought that
these are related very closely to the pathogens and that they could become
virulent under certain circumstances [41]. Evidence has now accumulated
that these isolates are much more different [42] and a new species is
being created (in preparation). This nonpathogenic species seems however
to have the same ecological preference as the highly pathogenic N. fowleri
but is more widespread. Apart from being isolated from discharges with
pathogenic N. fowleri, it was isolated from the warm discharges of 6
other factories. Apart from 2 aquaria we have never isolated in Belgium
a nonpathogenic variant from waters others than thermal discharges.
Pathogenic N. fowleri strains were only found in thermal polluted water.
481
-------
Of the eight tactories with pathogenic N. fowleri, four are metallurgical
factories, three chemical plants and one an electricity power plant. The
6 with only nonpathogenic variants, were 4 metallurgical factories, one
chemical plant and one cooking plant. Since it is believed these variants
are indicatororganisms for pathogenic N, fowleri, the use of a more
selective isolation method [39] may prove the presence of pathogenic
strains at these 6 factories.
In all cases of positive factories, be it with pathogenic or nonpathogenic
strains, the factories were rather old, indicating that these Naegleria
spp. need a long time to become established in a favorable warm environ-
ment, in a country with a temperate climate where it can normally not
survive in nature [43].
The quantitative studies have shown that pathogenic strains can be isola-
ted from samples as small as one ml, thus proving to be very hazardous to
human health. Furthermore the dispersal of high concentrations of
pathogenic organisms in surfacewater can pose serious problems to
recreational waters in the neighbourhood. Also the water distribution
works have been very concerned by the presence of highly pathogenic
amoebae in this water which is used, after treatment, as drinking water
for a large population of our country.
However, it is shown that these amoebae do not withstand low chlorine
concentrations [44] and that their cysts are effectively distroyed by
chlorine levels commonly used in our country by drinking water
distributors [34].
Aoanthamoeba spp. investigations
While preparing this text, the ecology of pathogenic Aoanthamoeba spp. in
warm discharges has been investigated only since a short time.
To be able to look for any relation between the results with Aoanthamoeba
and the ecology of N. fowleri, investigated previously, the same warm
water discharging factories were chosen.
The approach for the isolation of Aoanthamoeba was different than for
Naegltii'La because of the biological differences between these genera and
because Aoanthamoeba strains do grow much slower than other common
amoebae. The exact isolation procedure will be published elsewhere [in
preparation].
From a total of 9 places sampled, three were control waters and 6 thermal
discharges. In all control waters Acanthamoeba strains were found while
in three thermal discharges this genus was not isolated. In the other 3
discharges there was a definite increase in number of Acanthamoeba
compared to the controls. The Acanthamoeba isolates were tested for cyto-
pathic effect in Vero cell cultures and for virulence in mice. Since
these experiments are still being performed and the results therefore in-
complete, it is difficult to draw definite conclusions as to the impact
of higher water temperature on the prevalence of pathogenic Acanthamoeba
482
-------
strains. It is noted however that part of the Acanthamoeba strains in
control waters and in thermal waters are pathogenic. As the total number
of Aoanthamoebais increased in some thermal waters, also the number of
pathogenic strains was higher. During this study it became also obvious
that a high proportion of the Acanthamoeba strains obtained after incu-
bation at 37 °C, are pathogenic for mice. Any local increase in
Acanthamoeba number should therefore be prevented as it allows the
dispersal of pathogenic strains. Although only few human cases of
Acanthamoeba infections are known [8] and its epidemiology is unknown,
any enhancement of Auanthamoeba growth and dispersal should be prevented.
This in particular as the cysts of this genus are highly resistant against
chlorine, even at the highest concentrations used for preparing drinking
water [34].
CONCLUSIONS
It has been shown that Escherich-ia col-i dies off more quickly at high
temperatures than at lower temperatures [45] thereby establishing a
beneficial effect of the cooling waters. We would emphasize that E. coli.
is not the only pathogen to man present in water. We have found that the
total number of bacteria is raised when the water comes out of the cooling
system. These bacteria were isolated at 37 °C, the temperature of the
human body.
Most importantly, we showed that amoebae, highly pathogenic to man, have
found a favorable habitat in cooling water. Besides the high temperature,
which makes it possible for them to compete with other more common
protozoa, the higher supply of bacteria, which act as their food, is a
factor in their favor.
As N. fouleri does not tolerate salt concentrations of 0.5 % [6] and is
never demonstrated in sea water, while pathogenic Acantfaimuvba spp.
tolerate high salt concentrations and have been isolated from brackish
and salt water environments [36], the use of seawater for cooling purposes
could only prevent the spread of N. fovlevi. As the occurrence of the
disease produced by N. fowlei'i is directly related with water contact,
this would however already be beneficial.
The only effective measure to prevent dispersal of both pathogens in the
environment is the use of closed-cycle cooling systems.
Now that emphasis is given to the efficient use of waste heat as an
energy-saving measure, it should be realized that such water should be
treated effectively to prevent an increase of fatal human diseases.
AKNOWLEDGEMENT
This research is supported by grant 258-77-7 ENV B of the European
Economic Community.
483
-------
REFERENCES
1. CULBERTSON, C.G. , J.W. SMITH, H.K. COHEN and J.R. MINNER. Experimen-
tal infection of mice and monkeys by Aoanthamoeba. Am. J. Path. 35,
185-197, 1959.
2. FOWLER, M. and F. CARTER. Acute pyogenic meningitis probably due to
Aoanthamoeba sp. : a preliminary report. Brit. Med. J. 2, 740-742,
1965.
3. BUTT, C.G. Primary Araebic Meningoencephalitis. Engl. J. Med. 26,
1473-1476, 1966.
4. CULBERTSON, C.G., P.W. ENSMINGER and W.M. OVERTON. Pathogenic
Naegleria sp. - Study of a strain isolated from human cerebrospinal
fluid. J. Protozool. 15, 353-363, 1968.
5. CARTER, R.F. Primary amoebic meningoencephalitis. Clinical, patho-
logical and epidemiological features of six fatal cases. J. Path.
Bact. 96, 1-25, 1968.
6. CARTER, R.F. Description of a Naegleria sp. isolated from two cases
of primary amoebic meningoencephalitis, and the experimental patho-
logical changes induced by it. J. Path. 100, 217-244, 1970.
7. WILLAERT, E. Primary amoebic meningoencephalitis. A selected
bibliography and tabular survey of cases. Ann. Soc. beige Med.
trop. 54, 416-440, 1974.
8. MARTINEZ, A.J., C. SOTELO-AVILA, J. GARCIA-TAMAYO, J. TAKANO MORON,
E. WILLAERT and W.P. STAMM. Meningoencephalitis due to Aoanthamoeba
sp. Pathogenesis and Clinico-Pathological study. Acta neuropath.
(Berl.) 37, 183-191, 1977.
9. JADIN, J.B., J. HERMANNS, G. ROBYN, E. WILLAERT, Y. VAN MAERCKE and
W. STEVENS. Trois cas de meningo-encephalite amibienne primitive
observes a Anvers (Belgique). Ann. Soc. beige Med. trop. 51,
255-266, 1971.
10. CERVA, L. and K. NOVAK. Amoebic meningoencephalitis : sixteen
fatalities. Science 160, 92, 1968.
11. DUMA, R.J., W.I. ROSENBLUM, R.F. MCGEHEE, M.M. JONES and E.G.
NELSON. Primary amoebic meningoencephalitis caused by Naegleria.
Two new cases, response to Amphotericin B, and a review. Ann.
Intern. Med. 74, 861-869, 1971.
12. HECHT, R.H., A. COHEN, J. STONER and C. IRWIN. Primary amebic
meningoencephalitis in California. Calif. Med. 117, 69-73, 1972.
484
-------
13. CURSONS, R.T.M., T.J. BROWN, B.J. BRUNS and D.E.M. TAYLOR. Primary
amoebic meningoencephalitis contracted in a thermal tributary of the
Waikato River-Taupo : A case report. N.Z. Med. J. 84, 479-481, 1976.
14. VAN DEN DRIESSCHE, E., J. VANDEPITTE, P.J. VAN DIJCK, J. DE JONCKHEERE
and H. VAN DE VOORDE. Primary amoebic meningoencephalitis after
swimming in stream water. Lancet ii, 971, 1973.
15. CERVA, L., M. FERDINANDOVA, K. NOVAK, V. PTACKOVA, M. SCHROTTENBAUM
and V. ZIMAK. Meningoencephalitis durch Amoebida Naegleriidae.
Einer weiteren Fall in der Tschechoslowakei. Isolierung des Erregers.
Munch, med. Wschr. Ill, 2090-2094, 1969.
16. ANDERSON, K. and A. JAMIESON. Primary amoebic meningoencephalitis.
Lancet i, 902-903, 1972.
17. NAGINGTON, J., P.G. WATSON, T.J. PLAYFAIR, J. Me GILL, B.R. JONES and
A.D. Me G. STEELE. Amoebic infection of the Eye. Lancet ii, 1537-
1540, 1974.
18. JONES, D.B., G.S. VISVESVARA and N.M. ROBINSON. Aoanthamoeba poly-
phaga keratitis and Aoanthamoeba uveitis associated with fatal
meningoencephalitis. Trans, ophthal. Soc. UK. 95, 221-232, 1975.
19. ANDERSON, K. and A. JAMIESON. Primary amoebic meningoencephalitis.
Lancet ii, 379, 1972.
20. SINGH, B.N. and S.R. DAS. Occurrence of pathogenic Naegleria aerobia,
Hartmannella culbert'soni. and H. rhysodes in sewage sludge samples of
Lucknow. Curr. Sci. 41, 277-281, 1972.
21. GORDEEVA, L.M. Isolation and cultivation of Limax amoebae capable to
grow at 37 °C. Progress in Protozoology, Abstr. IV Congr. Int.
Protozool. 159, 1973.
22. KASPRZAK, W. and T. MAZUR. Small free-living amoebae isolated from
"warm" lakes : investigations on epidemiology and virulence of the
strains. 3rd. Int. Cong. Parasitol. Munich Abs. ICP III, 1, 188-189,
1974.
23. DE JONCKHEERE, J., P. VAN DIJCK and H. VAN DE VOORDE. The effect of
thermal pollution on the distribution of Naegleria fowleri. J. Hyg.
(Camb.) 75, 7-13, 1975.
24. DE JONCKHEERE, J. and H. VAN DE VOORDE. The distribution of
Naegleria fowleri in man-made thermal waters. Am. J. Trop. Med.
Hyg. 26, 10-15, 1977.
25. KASPRZAK, W. and T. MAZUR. Environmental strains of Naegleria
IV Int. Congr. Parasitol. Warsazawa Sect A. 29, 1978.
485
-------
26. STEVENS, A.R., R.L. TYNDALL, C.C. COUTANT and E. WILLAERT. Isolation
of the etiological agent of primary amoebic meningoencephalitis from
artificially heated waters. Appl. Environ. Microbiol. 34, 701-705,
1977.
27. SOH, C.T., K.I. IM, B.P. CHANG and H.K. HWANG. Pathogenicity of free-
living amoebae isolated from various places in Seoul. V Int. Congr.
Protozool. Abst. 412, New York, 1977.
28. WELLINGS, P.M., P.T. AMUSO, S.L. CHANG and A.L. LEWIS. Isolation and
identification of pathogenic Naeglevia from Florida lakes. Appl.
Environ. Microbiol. 34, 661-667, 1977.
29. CHANG, S.L., G.R. HEALY, L. Me CABE, J.B. SHUMAKER and M.G. .SCHULTZ.
A strain of pathogenic Naeglevi-a isolated from a human nasal swab.
Hlth. Lab. Sci. 12, 1-7, 1975.
30. TYNDALL, R.L., E. WILLAERT, A. STEVENS and A. NICHOLSON. Isolation
of Aaanthamoeba from cultured tumor cells. J. Cell. Biol. In press.
31. PELOUX, Y., A. NICOLAS, L. PEYRON and J. BEURLET. Problemes poses
au laboratoire par les 'kmibes libres". A propos d'un cas de conta-
mination de cultures cellulaires par une amibe du genre "Hart-
mannella". Path. Biol. 22, 587-592, 1974.
32. JAMIESON, J.A. Studies of amoebae of the genus Naeglena. M.S.
Thesis. University of Adelaide, 48 p. , 1975.
33. CERVA, L. Studies of Limax Amoebae in a Swimming Pool. Hydro-
biologia 38, 141-161 , 1971.
34. DE JONCKHEERE, J. and H. VAN DE VOORDE. Differences in destruction
of cysts of pathogenic and nonpathogenic Naegleria and Aaanthamoeba
by chlorine. Appl. Environ. Microbiol. 31, 294-297, 1976.
35. DERR-HARF, C. Etude des amibes libres dans les eaux de Strasbourg.
Doctoral Thesis, Strasbourg, 141 p., 1977.
36. SAWYER, T.K., G.S. VISVESVARA and B.A. HARKE. Pathogenic amoebas
from brackish and ocean sediments, with a description of Acantha-
moeba hatchetti, n.sp. Science 196, 1324-1325, 1977.
37. GRIFFIN, J.L. Temperature tolerance of pathogenic and nonpathogenic
free-living amoebas. Science 178, 869-870, 1972.
38. DE JONCKHEERE, J. Identifikatie, ekologie en destruktie van
Naegler-ia fowlevi, Carter 1970. Doctoral thesis, Leuven, 201 p.,
1977.
486
-------
39. DE JONCKHEERE, J. Use of an axenic medium for differentiation
between pathogenic and nonpathogenic Naegleria fowleri isolates.
Appl. Environ. Microbiol. 33, 751-757, 1977.
40. DE JONCKHEERE, J. Quantitative study of Naegleria fowleri in surface
water. Protistologica. In press. 1979.
41. DE JONCKHEERE, J. and H. VAN DE VOORDE. Comparative study of six
strains of Naegler-ia with special reference to nonpathogenic variants
of N. fowleri. J. Protozool. 24, 304-309, 1977.
42. STEVENS, A.R., E. GALLUP, J. DE JONCKHEERE and E. WILLAERT. Dif-
ferences in ultrastructure and lectin sensitivity of pathogenic and
nonpathogenic Naegler-ia fowleri. IV Int. Congr. Parasitol. Sect. A,
31-32, 1978.
43. CHANG, S.L. Resistance of pathogenic Naegleria to some common
physical and chemical agents. Appl. Environ. Microbiol. 35, 368-375,
1978.
44. DERREUMAUX, A.L., J.B. JADIN, E. WILLAERT and R. MORET. Action du
chlore sur les amibes de 1'eau. Ann. Soc. Beige Med. Trop. 54, 415-
428, 1974.
45. BROCK, T.D. High temperature systems. Ann. Rev. Ecol. System. 1,
191-220, 1970.
487
-------
Fig. 1. Naegleria sp. with differential interference contrast (1,750 X)
a. Trophozoite with bulging pseudopodia and prominent karyosome
in nucleus
b. Typical round cyst
.•ig. 2. Acanthamoeba sp. with differential interference contrast (1,750 X
a. Trophozoite with typical filliform pseudopodia and many
vacuoles
b. Two typical polyhedral cysts, with outer wall loosely applied
to the inner cyst wall.
488
-------
-t*.
oo
31 39
Fig. 3. Map of Belgium with factories sampled. Sites positive for pathogenic N. fallen are circled.
The broken lines are limits of the different provinces ; thick lines are rivers ; thick lines
with cross-lines are canals.
-------
RELATION BETWEEN ZOOPLANKTON MIGRATION AND
ENTRAINMENT IN A SOUTH CAROLINA COOLING RESERVOIR
P. L. Hudson and S. J. Nichols
U. S. Fish and Wildlife Service
Southeast Reservoir Investigations
Clemson, South Carolina U.S.A.
ABSTRACT
Knowledge of the vertical and areal distribution of zooplankton could be
used to influence the siting and design of intake structures for
condenser cooling water systems of steam-electric and hydroelectric power
stations. We studied the vertical distribution of zooplankton in Keowee
Reservoir, South Carolina, a cooling reservoir for a 2580-MW nuclear
power plant. The nuclear plant withdraws hypolimnetic water from depths
of 20-27 m, beneath a surface skimmer wall for once-through condenser
cooling. Diel and seasonal variations in vertical distributions of
zooplankton influenced the degree of entrainment at the power plant.
For about 9 months, zooplankton was surface oriented, and only relatively
small amounts were entrained. During July, August and September,
however, most zooplankters were bottom-oriented during the day, and
large numbers were entrained whereas at night most moved to or near the
surface and few were entrained. The danger of entrainment during these
months was thus restricted largely to the 12 to 13 daylight hours.
INTRODUCTION
Entrainment in power plant water intake systems and impingement on
intake screens cause substantial losses of plankton and fish [l]. Wide
variations in the magnitude and causes of these losses, however,
preclude reliable prediction of the effects of planned power plants.
Knowledge of zooplankton distributional patterns could facilitate the
placement and design of water intake structures to control or reduce
entrainment.
Zooplankton, an essential component of the food chain in reservoirs,
is readily entrained in the condenser cooling water systems of steam-
electric power stations. However, various species are often confined to
relatively restricted zones, and their uneven vertical distribution
varies diurnally and seasonally. Typically, zooplankton migration is
upward at night and downward during the day. Although, the diurnal
behavior of zooplankton has been summarized and documented by
Hutchinson [2], seasonal variation in migration remains obscure. The
ideal zone for withdrawal of cooling water would be that at which
plankton population density is lowest over a 2^-hour period.
490
-------
Skimmer walls have been used to a limited extent to enable the withdrawal
of cooling water from selected depths. A skimmer wall is a concrete
barrier extending from just above full surface water elevation to a depth
that allows only hypolimnetic water to pass into the intake structure.
Skimmer walls increase generating efficiency by providing a year-round
supply of cool water. Skimmer walls may also reduce the entrainment of
larval fish [3] and zooplankton [U,5]. The objectives of the present
study were to (l) determine the diel and seasonal vertical distribution
of zooplankton in Keowee Reservoir, South Carolina; (2) measure the
entrainment of zooplankton beneath a skimmer wall; and (3) evaluate the
effects of power plant operation on the zooplankton community in the
discharge area.
RESERVOIR DESCRIPTION
Keowee Reservoir was filled between 1968 and 1971 to provide cooling
water for Duke Power Company's Oconee Nuclear Station. The 7^+35 ha
reservoir has a mean depth of 15.8 m at full pool (2^3.7 m above mean
sea level). The reservoir has two sections (south and north) of nearly
equal area (Fig. l). The 2580-MW Oconee Nuclear Station is located
between the two sections; cooling water is withdrawn from the south
section and discharged into the north section. Cooling water withdrawn
from the south reservoir passes under a skimmer wall (Fig. 2) 19-8-
27.^ m below full pool surface level, flows over a submerged weir
(the top of which is 9-1 m below the surface), and then travels 1.5 km
down a canal to the plant. Water from the plant is discharged into the
north section of the reservoir, 9-12 m below the full-pool surface level.
Maximum cooling water flow is 13^ m3/s, and mean flow is 99 m3/s at Q0%
generating capacity. Maximum condenser cooling water flow is about four
times the average outflow from Keowee Reservoir and could theoretically
circulate the entire volume of the reservoir through the station in U.5
months.
METHODS
Zooplankton abundance was estimated monthly at five sites (A,B,C,D, and
X, Fig. 1.) in Keowee Reservoir, from July 1973 to November 1977-
Samples were collected at each site by taking four 5-min oblique (15 m to
surface) tows with a high speed Miller sampler [6] with a 0.156-mm-mesh
net. The data obtained from these monthly samples indicated a need for
a study of the vertical distribution and entrainment of zooplankton.
Only monthly data from September 1976 through August 1977 are used in
this paper.
Vertical distribution and entrainment were measured by sampling for U-day
periods in September 1976 and February, May, and August 1977- On the
first day of each period, a series of horizontal tows were taken
immediately outside the skimmer wall (Fig. 1, solid arrows) at mid-day
(1200-lUOO) and at night (2^00-0200). Duplicate tows were made with the
491
-------
Miller sampler at 5 m depth intervals from the surface to 20 ra (Fig. 2).
On the second and third day, four oblique tows from the bottom of the
intake canal to the surface at two stations (Fig. 1, solid arrows) were
taken at 3-hour intervals for 2h hours. On the Hth day duplicate
oblique tows from the bottom to the surface were taken at eight stations
(1.0 km apart) along a transect extending Q.h km from the nuclear plant
discharge cove up the north basin of the reservoir (Fig. 1, dashed lines)
Each zooplankton sample was concentrated or diluted to a fixed volume,
depending on the abundance of organisms, and four 1-ml subsamples were
withdrawn and placed in a rotary counting chamber. All organisms in the
subsample were counted and identified to species. This subsampling
procedure was repeated three times. Because of their large size, all
Chaoborus and Leptodora in the entire sample were counted.
RESULTS AND DISCUSSION
Vertical Distribution
Vertical distributions of zooplankton in Keowee Reservoir were similar in
February and May (Fig. 3) and represent the type of distributions that
were typical during periods when water temperatures from the surface to
a depth of 10 m were relatively homothermous and lower than 20 C (mid-
October to June). In February and May, zooplankton was concentrated in
water less than 15 m deep during the day and migrated upward about 5 ni
during the night. The vertical patterns shown in Figure 3 reflect the
distribution of the dominant species—Bosmina longirostris, Diaptomus
mississippiensis , and Diaphanosoma branchyurum—which accounted for 96%
(by number) of the zooplankton population in February and TQ% in May..
Less abundant species, such as Mesocyclops edax, Cyclops vernalis,and
Tropocyclops prasinus, had vertical distributions similar to the common
species in May, but during the daytime in February, most were 20 m below
the surface. Holopedium amazonicum was not collected in February but
maintained a population maximum at 5 m in May, with no evidence of
diurnal migration. Chaoborus punctipennis was benthic in February but
planktonic in May; peak numbers were at the 15 m depth contour. Most of
the zooplankton was above the top of the skimmer wall opening during
February and May, and thus not subject to entrainment (Fig. 3).
As thermal stratification developed in the summer and surface water
temperatures rose above 20 C, zooplankton populations concentrated at
greater depths during the day but migrated toward the surface at night
as indicated in the August 1977 and September 1976 samples (Fig. 3).
The depth of maximum zooplankton numbers was difficult to assess because
sampling below 20 m was made impractical by bottom obstructions; however,
a daytime sample taken in August at a depth of 23 m indicated that
densities continued to decline with increasing depth. Since no samples
were taken below 20 m in September, day-night comparisons of zooplankton
densities were restricted to 0^-to 20-m zone. Since zooplankton moved
upward at night, any sizable population movement into the 0-to 20-m zone
492
-------
from below 20 m would substantially increase population densities at
night. Abundance of most species increased only slightly at night, but
the^densities of Mesocyclops and the semi-planktonic Chaoborus doubled,
indicating that a large proportion of their populations were below 20 m
during the day. Woodmansee and Grantham [?] reported similar behavior
for these two species in a small Mississippi reservoir. Because dominant
species in Keowee Reservoir (Bosmina, Diaptomus, and Diaphanosoma) did
not increase in numbers in the 0-to 20-m zone at night in September, we
assumed that the maximum population density during the day was in the
vicinity of the 20-m depth contour. All species migrated toward the
surface at night during August and September except Holopedium, which
maintained a maximum density at 5 m, as in May.
The vertical distribution we observed generally paralleled that described
by Hutchinson [2]. He described various nocturnal migration patterns,
which are slight deviations from the simplest type where the organisms
start moving upward before or shortly after sunset and reach the upper
layers sometime before midnight. They remain near the surface for
several hours and descend at dawn. Since we did not sample at close time
intervals for 2^-hour periods at the skimmer wall, we could not determine
the specific kind of nocturnal migration. However, our samples at noon
and midnight should approximate the distribution extremes over a 2U-hour
period and our 20-m-deep daytime sample should yield a suitable estimate
of the maximum zooplankton population subject to power plant entrainment.
Few investigations have been made of daytime vertical distribution of
zooplankton through a season or over a year. Plew and Pennak [8] found
a slow seasonal drift of the zooplankton population downward in spring
and upward in autumn. Langford [9] found that zooplankton populations
moved downward during the day as summer progressed, and Hutchinson [2]
presumed this to be due to increasing surface temperature. To place
temporal bounds on our daytime summer vertical distribution, we
determined the mean temperature where the zooplankton population density
was greatest in the water column. The maximum densities in May, July,
and August were at depths where temperatures ranged from 19 to 23 C.
Water of these temperatures were at the depth of the skimmer wall opening
from mid-July to late August which was the time period of maximum
entrainment.
Entrainment
Migration patterns of zooplankton differ according to species, sex, and
age of individuals. However, since the dominant species behaved
similarly in Keowee Reservoir, this section deals mainly with total
zooplankton.
The 2i»-hour pattern of zooplankton abundance in the intake canal during
February and May (Fig. U) reflected the differences in day-night
densities at the skimmer wall opening. In May, day and night densities
at 20 m near the skimmer wall opening (20-m depth contour) were about
I/liter and the densities in the canal fluctuated from 0.8 to 1.I/liter
493
-------
over the 2lt-hour period (Fig. 3 and U). In February, day and night
differences at the skimmer wall opening (3.0 vs. 1.5/liter) produced a
slight diel periodicity in the canal (Fig. h). Overall mean population
densities in the canal during February and May (3.0 and 1.0/liter,
respectively) vere low compared with surface populations at the control
stations (D and X, Fig. l) at that time (6.0-13.0/liter).
In August and September the distinct diel fluctuation in zooplankton
densities in the canal (Fig. k) correlated with the day-night population
differences at the skimmer wall opening (Fig. 3). Maximum densities of
zooplankton occurred at the skimmer wall opening during daylight, but
this timing did not correspond with peak densities in the canal (Fig. 3
and k). The difference in the time of peak abundance was the result of
travel time (which is a function of volume of water in the intake canal
and pumping rate at the Oconee Station) between the skimmer wall sampling
station and stations in the canal, a 6- to 9-hour time lag. Zooplankton
entrained at the skimmer wall at midday (1^00 hours) reached the canal
sampling stations between 2000 and 2300. The different density patterns
in the canal in August and September were probably due to different
pumping rates and diurnal migration patterns at the skimmer wall. The
water level in the canal was lower and pumping rates were higher during
August (110 m-Vs) than in September (So.** m /s), which could have
produced a 2- to 3-hour difference in zooplankton travel time.
Furthermore, the amount of zooplankton and extent of time spent at the
skimmer wall opening could vary due to differences in depth distribution
(Fig. 3) and migration patterns between August and September, thus
shifting the density peaks by as much as 2 to 3 hours. Overall, maximum
and minimum densities at the skimmer wall opening were similar to the
values in the canal, and entrainment varied diurnally with 12 hours of
high values followed by 12 hours of low values.
Reservoir Zooplankton
High zooplankton densities in the discharge cove were followed by
declines in densities 1 km from the cove based on transect data (Fig. 5).
Zooplankton densities usually were lowest during all sample periods 2-^4
km away and either recovered (Fig. 5 A,B,D) or continued to decline
(Fig.. 5C). The density levels at both ends of the transect had little
relation to entrainment levels, whereas those near the middle (2-k km)
of the transect were similar to the mean 2U-hour density in the canal
(Table 2).
Zooplankton densities in the discharge cove were usually higher than the
maximum concentrations observed in the intake canal, except in the
August samples. For example in February 1977, the maximum density was
U.5/liter in the canal and based on transect samples nearly 12/liter in
the discharge cove. Zooplankton densities from the monthly sampling in
the discharge cove (Station C) were sometimes the highest in the
reservoir in 1976-77 (Table l). Since about 9 months of the year these
large numbers of zooplankters were not passing through the plant, the
high densities in the discharge cove probably resulted from concentrating
494
-------
currents. Preliminary dye and drogue studies in the cove indicated the
presence of counter currents and back eddies. The densities of Bosmina
in a surface sample taken in the cove were 12 times those in the canal.
Bosmina is a small, weak-swimming cladoceran that tends to air-lock (an
air bubble forms under the carapace) easily [lO]. Passage through the
condenser tubes could cause air-lock, and as the plankters float to the
surface, back eddies or counter currents could concentrate sizable
populations in the discharge cove. A concentration of zooplankton in a
discharge area also occurred in Lake Monona, Wisconsin [l.l]. However,
high concentrations in Keowee Reservoir were localized, since the
discharge cove represents less than L% of the reservoir surface area.
Zooplankton numbers 1 to k km from the discharge cove should be similar
to densities noted in the intake canal (Table 2). Pumping rates during
this study ranged from 88 to 110 m^/s and only 12 days were required to
replace the water in an area 1 to k km from the discharge structure. The
densities at 1 to U km were close to the mean densities in the canal
during October and August but were higher in May and lower in February
(Table 2).
The reduction in zooplankton abundance 1 to k km from the discharge cove
may have resulted from latent mortality due to plant passage. Duke
Power Company [5] estimated zooplankton mortality of 0 to 33% (annual
mean 2.%) after passage through Oconee Nuclear Station; maximum mortality
was in July. Additional delayed mortality could have occurred further
upstream from the discharge cove, but our data did not show this trend.
Bosmina numbers in the area 1 to H km from the discharge structure were
20 to 63% lower than the mean numbers in the canal, whereas Diaptomus
densities were usually higher. Numbers of zooplankters entrained from
mid-October to June were substantially lower than the numbers normally
found in the water column (Table 1 and 2), resulting in an estimated
area of low zooplankton density covering about 15* of the area of the
reservoir. Therefore the overall effect of entrainment in this 1 to H
km area was apparently dilution.
Zooplankton densities in late July (Fig. 5C) showed different horizontal
distributions. Low concentrations of organisms were not observed, but
there was a steady decrease in numbers upstream from the discharge cove.
During these months, maximum concentrations of zooplankton were found at
the skimmer wall opening (20 m deep); this high concentration, combined
with the long photoperiod (ik hours), resulted in the entrainment of
maximum numbers (8/liter). During August intake canal zooplankton
densities were at the highest levels (Fig. h). Consequently,
inordinately large numbers of plankton were discharged into the north
basin, compared with the average standing crops in other parts of the
reservoir.
The zooplankton distribution pattern along the transect, 1.0-8.5 km
from the discharge cove, probably varied independently of entrainment.
Populations in this area were affected by reproduction and repopulation
from other areas of the reservoir. Variation in zooplankton abundance
495
-------
in this section of the reservoir depended on environmental factors, and
possibly on the operational level of the Jocassee Pumped-Storage Station,
located 15-7 km upstream from the Oconee Station. On any given date
densities in the U.0-8.5 km area were as high as (Fig 5D) , lower than
(Fig. 5C), or equal to (Fig. 5A,B) entrainment levels.
CONCLUSIONS
Zooplankton diel migration in Keowee Reservoir during most of the year
placed most of the population above the skimmer wall opening during the
day and virtually always at night. This behavior resulted in seasonal
and diurnal variations in entrainment. Selective withdrawal of water
with low zooplankton densities, coupled with latent mortality, created
an area of low standing crop covering 15$ of the reservoir surface for
about 9 months. Maximum numbers of zooplankton during the day in July
and August followed the 20 C isotherm downward as the surface waters
warmed. During the day, concentrations at the skimmer wall opening were
high and large numbers were entrained and discharged into the north basin.
Because entrainment was reduced by the presence of the skimmer wall,
there was a dilution of zooplankton in most of the outfall area (l to
k km up north basin) and a possible scarcity of food for higher aquatic
animals.
Entrainment does not appear to be a major factor in determining
reservoir-wide zooplankton abundance. The use of a skimmer wall and the
withdrawal of hypolimnetic water for steam electric plants' cooling
systems appears highly desirable in systems such as the Keowee Reservoir.
Oconee Nuclear Station has maintained state water quality temperature
standards, the cool water has increased thermal efficiencies for power
plant operation, and entrainment of zooplankton is relatively small over
most of the year.
Knowledge of zooplankton behavior could be used to predict the zones of
influence around steam electric and hydroelectric plant cooling water
intakes. Three-dimensional velocity profile models superimposed on 2k
hour movement patterns of zooplankton would allow prediction of
entrainment levels. At Multi-level inlet-outlet hydroelectric stations,
this knowledge could reduce entrainment or facilitate the transport of
excess production from a reservoir to a tailwater system.
496
-------
REFERENCES
1. Boreman, J. 19TT- Impacts of power plant intake velocities on fish.
Topical Briefs: Fish and Wildlife Resources and Electric Power
Generation, No. 1. U. S. Fish and Wildlife Service. Ann Arbor,
Michigan. 10 pp.
2. Hutchinson, G. E. 1967. A treatise on limnology. Vol. II. Wiley
and Sons, Inc. New York, N. Y. 1115 pp.
3. Ruelle, R., W. Lorenzen, and J. Oliver. 1977. Population dynamics
of young-of-the-year fish in a reservoir receiving heated effluent.
Pages U6-67 in_ W. Van Winkle, ed. Proceedings of a Conference on
Assessing Effects of Power-Plant-Induced Mortality on Fish Populations,
Pergamon Press, N. Y.
14. Davies, R. M. , and L. D. Jensen. 197^. Zooplankton entrainment.
Pages 162-172 in L. D. Jensen, ed. Environmental responses to thermal
discharge from Marshall Steam Station, Lake Norman, North Carolina.
Electric Power Research Institute, Research Project RP-^9, Pala Alto,
Calif.
5. Duke Power Company. 1977. Zooplankton. Pages 271-33^ in_ Oconee
Nuclear Station environmental summary report, 1971-1976. Duke Power
Company, Charlotte, N. C.
6. Miller, D. 1961. A modification of the small Hardy plankton sampler
for simultaneous high-speed plankton hauls. Bull. Mar. Ecol. 5:
165-172.
7. Woodmansee, R. A., and B. J. Grantham. 196l. Diel vertical
migrations of two zooplankters (Mesocyclops and Chaoborus) in a
Mississippi lake. Ecology h2: 619-628.
8. Plew, W. F. , and R. W. Pennak. 19^9. A seasonal investigation of
the vertical movements of zooplankters in an Indiana lake. Ecology
30: 93-100.
9. Langford, R. R. 1938. Diurnal and seasonal changes in the
distribution of limnetic Crustacea of Lake Nysissing, Ontario. Univ.
Toronto Stud. Biol. Ser. No. ^5 (Publ. Ontario Fish Lab. 56). 1*2 pp.
10. Martin, D. B., and J. F. Novotny. 1975- Studies to determine
methods for culturing three freshwater zooplankton species.
Ecological Research Series, U. S. Environmental Protection Agency.
Corvallis, Oregon. (EPA-6660/3-75-010). 33 pp.
11. Brauer, G. A., W. H. Neill, and J. J. Magnuson. 197^- Effects of
a power plant on zooplankton distribution and abundance near plant's
effluent. Water Research 8: U85-U89-
497
-------
TABLE .1. MEAN MONTHLY ZOOPLANKTON DENSITIES (NO./LITER) IN KEOWEE RESERVOIR, SEPTEMBER 1976 TO
AUGUST 1977. STATION LOCATIONS ARE SHOOT IN FIGURE 1. (STANDARD ERRORS IN PARENTHESES).
1976
STATIONS
SEPT.
A 5-78
(0.33)
B 3.4l
(0.22)
C 11.88
(0.25)
S D ^-07
CO
(0.15)
X 7.08
(0.55)
OCT . NOV .
5-
(0.
3.
(0.
4.
(0.
3.
(0.
10.
(o.
57 1.51
4o)(0.l6)
08 1.01
29)(0.09)
00 2.63
26)(0.13)
88 1.04
17)(0.22)
38 2.13
33)(0.2l)
DEC.
2.40
(0.03)
1.31
(0.11)
4.60
(0.41)
1.27
(0.16)
2.29
(0.05)
1977
JAN.
3.98
(0.08)
2.25
(0.27)
10.55
(1.30)
4.84
(0.31)
6.32
(0.28)
FEB.
3.05
(0.14)
1.69
(0.21)
5.95
(0.54)
5.87
(0.20)
7.16
(0.93)
MAR.
4.34
(0.16)
1.50
(0.08)
4.67
(0.21)
11.13
(0.62)
28.61
(2.42)
APR.
3.85
(0.26)
1.76
(0.02)
2.64
(0.24)
12.27
(0.75)
7.78
(0.21)
MAY JUNE
8.04 5
(0.15)(0
3.27 4
(0.08)(0
5.12 5
(0.30)(0
12.54 5
13.09 6
(1.43)(0
.20
.35)
.26
.12)
.14
.17)
.98
.20)
.83
.44)
JULY
3.86
(0.06)
4.08
(0.18)
3.15
(0.20)
3.04
(0.29)
3.10
(0.30)
AUG.
5.40
(0.84)
4.81
(0.66)
3.50
(0.14)
2.64
(0.14)
3.38
(0.31)
-------
TABLE 2. MEAN ZOOPLANKTON DENSITIES (NO./LITER) AT THE SKIMMER WALL OPENING,
IN THE INTAKE CANAL, AND 1-U KM UP THE NORTH BASIN FROM
DISCHARGE COVE, KEOWEE RESERVOIR, SEPTEMBER 1976 to AUGUST 1977.
27 SEPTEMBER- 7-11 FEBRUARY 9-13 MAY 1-5 AUGUST
LOCATION
1 OCTOBER 1976 1977 1977 1977
SKIMMER WALL 6.25 2.21 1.08 3.00
INTAKE CANAL 2.93 3.03 0.96 3.71*
NORTH BASIN 3-53 2.06 2.17 3.20
499
-------
x Discharge Try^f Keowee
,7 Dam
*. -Oconee V /
/
Fig. 1. Zooplankton sampling stations in Keowee Reservoir,
South Carolina. Letters A,B,C,D, and X indicate
regular zooplankton monitoring stations. Solid lines
with arrows indicate sampling stations outside and
inside intake canal. Dashed line indicates location
of transect.
500
-------
full pond el 244m
: I o w
Ski m m e r
wall
el. 224 m
submerged
xv e i r
reservoir bottom el. 216m
Schematic Profile >P the Intake Hkimmer ".•/•ill
"uhrner
'>.-owee R(.-:-,ervoir, "outh Carolina.
501
-------
NUMBER PER LITER
2 4 6 8 10 12 14
2 4 6 8 10 12 14
5
I
H
Q.
UJ
O
Cn
O
Q.
UJ
O
5-
10.
15.
20.
25.
\\
5.
10.
15,
20.
25.
day
night
skimmer
wall
o pen i ng
February
8 10 12 14
Augus t
5.
10.
15.
20.
25.
May
101214
10.
15.
20.
25.
I
September
3. Vertical Distribution of Total Zooolankton in Front of the
Skimmer Wall (1200 to 1410 hours) and (2400 to 0200 hours)
on February 9, Mav 11, August 3, 1977 and September 29. 1976.
-------
6
5.
4.
3.
2.
1.
noon 3
February
darkness
mn
3.0
6 9 noon
darkness
noon 3 6 9 mn 3 6
May
x-1.0
9 noon
tn
O
noon 3 6 9 mn 3
August
X-3.7
9 noon
September
noon 3
9 mn 3
Fig. 4. Total Zooplankton Densities in the Intake Canal During a
Twenty-four Hour Period on February 7-3, 1977, May 9-10,
1977,"August 1-2, 1977 and Sentember 23-29, 1976.
X-2.9
9 noon
-------
B
k
0)
«"
_l
w
a
In
0)
E C
3
Z
10
8
6.
4
2
0.
February 1977
6.
4.
2.
0,
8.
6.
4,
2.
0
May 1977
July 1976
10.
8.
6.
4.
2.
October 1976
Distance (km)
Fig. 5. Total Zooplankton at Eight Stations 1 km Apart
Along 3.4 km Transact from Discharge Cove up
North Basin; February 10, 1977, Mav L2, 1977,
July 30, 1976 and October 1, 1976.
504
-------
EFFECTS OF A HOT WATER EFFLUENT
ON POPULATIONS OF MARINE BORING CLAMS IN BARNEGAT BAY, NEW JERSEY
K. E. Hoagland
Lehigh University
Bethlehem, Pennsylvania U.S.A.
and
R. D. Turner
Harvard University
Cambridge, Massachusetts U.S.A.
ABSTRACT
The Oyster Creek Nuclear Generating Station began operation in December 1969.
By 1971, damage to wooden structures from marine boring clams was evident in
Oyster Creek and adjacent parts of Barnegat Bay. Increased salinity and
temperature in Oyster Creek, brought about by the Station's cooling system,
allowed shipworms to develop large breeding populations. Actions taken in
1974-1976 to reduce the environmental impact of the cooling system included
reduction of the temperature of the effluent and removal of infested wood
from Oyster Creek. The shipworm population declined, but primarily because
of cold winter and spring temperatures coupled with prolonged winter and
spring outages of the Generating Station in 1976 and 1977. No outage
occurred in the winter-spring of 1978, and there was an outbreak of ship-
worms in the summer, 1978. The dominant shipworm in Oyster Creek is now
a subtropical species. Population structure of this shipworm differs from
that of native species in unaffected portions of Barnegat Bay.
INTRODUCTION
The Oyster Creek Nuclear Generating Station at Forked River, New Jersey
(Fig. 1) began operation in December 1969. It is owned by the Jersey Central
Power and Light Company (J.C.P.& L.). It uses Forked River as a source of
cooling water and Oyster Creek as a discharge canal for its once-through
cooling system. When the station is pumping water, Forked River flows from
Barnegat Bay upstream into the South Branch of Forked River [1], [2].
Salinity in both creeks is now that of adjacent portions of Barnegat Bay.
Characteristics of Barnegat Bay have been reviewed elsewhere [3].
Docking facilities have been in Oyster Creek since 1944, but not until the
summer of 1971 did the property owners become aware of large-scale damage
to wooden structures caused by marine boring clams, the Teredinidae (often
called shipworms). Because the creek formerly had been of low salinity,
"...freshwater to about 2,500 feet downstream of US Route 9 ..." [4], it
had been thought safe for untreated wood, despite the fact that shipworm
attacks have occurred from time to time in Barnegat Bay proper [5]. Under
new conditions caused by the Generating Station cooling system, the Oyster
505
-------
Creek marinas were providing massive amounts of wood for shipworm infes-
tation.
We began research into the attack of Teredinidae during the summer of 1971
at the request of three marina owners. Early work by one of us [1] indi-
cated that the Generating Station was responsible for the attack by pulling
larvae of the teredinids into Forked River and Oyster Creek, by increasing
the salinity of both creeks such that teredinids could breed there, and by
substantially increasing the temperature of Oyster Creek.
The study was expanded in Sep, Briber, 1976, when the U.S. Nuclear Regulatory
Commission (NRC) funded the study. In 1974-1975, the NRC required the
J.C.P.& L. Company to buy the Oyster Creek marinas, remove untreated wood
from the creek, and reduce the temperature of the effluent by increasing
the pumping of unheated "dilution" water from Forked River.
This paper reviews the data on plant operations and species distribution
and population growth of Teredinidae in the vicinity of Oyster Creek. The
purpose is to see if J.C.P. & L.'s actions reduced the shipworm infestation.
We comment on the siting and operation of plants producing waste heat, with
regard to potential shipworm attacks.
METHODS
Untreated straight-grain white pine panels, 2 x 9 x 21 cm, are used to
collect populations of Teredinidae. The panels are weighed, soaked in
artificial sea water for 2 weeks, attached to aluminum racks, and set into
the water vertically, resting about 15 cm above the water-sediment inter-
face. All panels are aligned similarly with respect to currents and depth.
They are placed along creek or bay shoreline off docks or bulkheads, 0.8 to
2.0 meters deep. There are 19 stations (Figs. 1 & 2). Stations 1, 2, 16,
and 17 are inshore bay control stations. Stations 3 and 7 are creek
controls; 4-6 and 9 record shipworms in the south branch of Forked River;
10 - 13 are in Oyster Creek. Stations 8, 14, and 15 are bay stations having
a slight thermal influence. Stations 18 and 19, not figured, are offshore
bay controls located on Long Beach Island near Barnegat Light. Due to space
constraints, we present data from stations 1,2,3,4,5,8,11,12,11 and 18; the
remainder are discussed elsewhere [6].
Data are reported from 2 types of panels. Cumulative panels are deployed
in sets of 12 each May, and are removed one each month to record cumulative
attack of the shipworms. When removed, a cumulative panel is replaced with
a panel that is then left in the water for 12 months this is called a
yearly panel. The yearly panels provide information on species composition,
age structure, size of individuals, and population density.
Temperature and salinity are monitored monthly, except at stations 1,5,11,
and 14, where constant-recording instruments are located. In the laboratory
the shipworms are removed from the panels, measured, and identified to
species. Analysis of reproductive condition of the shipworms is performed,
506
-------
.but is not reported here. Once dissected, the remaining wood chips are
treated with acid to remove calcium carbonate fragments, dried to constant
weight, and the weight is recorded. Comparison of the final weight with
the pre-submergence weight tells the percentage of wood destroyed by marine
boring organisms.
RESULTS
The complete data are available in our reports to the NRC [6]. Salinities
in Oyster Creek and Forked River (TABLE I) are the same as in Barnegat Bay
and are higher than control creeks: (eg, station 3). Power plant actions
to reduce the shipworm population have not included reduction in salinity.
In the period October, 1971 to May, 1974, temperatures in Oyster Creek were
as high as 12° C above ambient water temperatures and usually ran 4 to 10°
•C above. Since then, the volume of water pumped has increased (TABLE II),
and temperatures have been between 2 and 7° C above ambient (TABLE III).
Constant recording thermometers have shown temperatures in Forked River to
be 1 to 2° C above ambient [6], suggesting that recirculation of the heated
effluent occurs [3].
TABLE IV lists major outages of the Generating Station. In 1974 - 1977,
there were lengthy outages in spring and/or winter. In 1978, the annual
refueling outage did not occur until September. J.C.F. & L. removed some
of the untreated pilings and trash wood from Oyster Creek over the period
January - March, 1976. Therefore a part of the shipworm breeding population
and its habitat was destroyed. No wood was removed from Forked River.
TABLE V summarizes wood weight loss for the 1977-1978 cumulative panel
series. The 1976 series showed significant damage (greater than 30% loss
of wood weight) at stations 1, 2, 4, 5, 8, and 11, with more than 70% loss
at stations 1, 2, and 8. Damage by type of station was: Northern bay ^
Oyster Creek and the mouth of Forked River > South Branch of Forked River >
Southern bay > Creek controls. The 1978 series, after 3 months, showed
significant damage only in Oyster Creek.
TABLE VI reviews the species composition and total number of shipworms
found in cumulative and yearly panels, (1) in 1973, after the Generating
Station began operating but before corrective action was taken; (2) 1976
and 1977, after water temperature was reduced and wood was removed from
Oyster Creek; and (3) in 1978, after the plant remained in operation over
one continuous winter and spring period. Panels deployed in Oyster Creek
or at the mouth of Forked River before September, 1975, became riddled with
shipworms after a few months. Between Fall, 1975, and Winter, 1977-1978,
settlement was moderate in Oyster Creek and Forked River (less than 30
animals per panel). However, a new outbreak of shipworms occurred in the
summer of 1978.
Teredo navalis has always dominated the Long Beach Island area (station 18),
and Bankia gouldi, the inshore bay areas (stations 1 and 2). Between 1971
507
-------
and 1974, B_. gouldi and a few T. navalis occurred in Oyster Creek and
Forked River. In 1974, the semi-tropical Teredo bartschi and Teredo
furcifera were found in Oyster Creek. Soon T_. furcifera was found at
other stations (TABLE VI), but we found no T^. bartschi in 1976 or 1977,
and assumed that it had been eliminated by the cold waters of the winters
of 1975 and 1976. The outbreak of shipworms in 1978 was especially
interesting in that it was composed entirely of Teredo bartschi. Of the
areas suffering severe attack in 1977-1978, stations 2, 11-12, and 18,
each area is dominated by a different species of shipworm.
DISCUSSION
The heavy attack of shipworms at some bay control stations (#2 and 18) but
not others (#17) may result from local water current patterns; shipworms do
poorly in regions of poor circulation. Pollution from marinas, competition
for space with other fouling organisms such as Limnoria and Hydroides, and
shellfish disease are other factors that we [6] and others [7] have
implicated. Oyster Creek and the South Branch of Forked River should be
compared with creek stations 3 and 7, and these have had only light attack
since our study began. These stations are comparable in water flow, tem-
perature, and salinity to Oyster Creek and the South Branch of Forked
River before the Generating Station began operations. Outbreaks of ship-
worms on Long Beach Island, either before or during our study, are not
relevant to the situation in Oyster Creek, for the outbreaks are of the
non-estuarine species T_. navalis, in an oceanic environment.
The change in species composition in Oyster Creek acts to prolong the
breeding season, because Teredo species are capable of producing larvae
later into the Fall than the native estuarine species, B_. gouldi. T_.
navalis, also native to New Jersey, is normally found offshore, but is more
common in the Generating Station's cooling system creeks than in natural
tidal creeks. The result is that four species of the Teredinidae, 2 native
and 2 introduced, can live in Oyster Creek, increasing the likelihood that
the proper physical conditions will trigger an outbreak of at least one of
them. The subtropical T_. bartschi releases its young as pediveligers that
settle in aggregations adjacent to the parent. Out laboratory studies show
that individuals mature in 6 weeks or less, depending on water temperature;
2 to 3 generations can easily occur in one year. This life history pattern
led to the outbreak of shipworms in 1978 in Oyster Creek. So far, only 3
T. bartschi have been found in the mouth of Forked River; spread of this
species must take place by movement of infested wood from one creek to the
other unless currents are swift and distance is short, because the pedi-
veligers remain free-living for only a few hours or days. The fact that
Clapp Laboratories [8] reported a few T. bartschi from Oyster Creek during
1976-1977 when we found none indicates that a localized few surviving
adults are responsible for the Oyster Creek outbreak. The outbreak can be
correlated with the lack of a plant shutdown in 1978 until September,
allowing the proper conditions for the subtropical T_. bartschi to breed.
We are now conducting laboratory studies of this species to delineate its
physiological tolerances and preferences.
508
-------
CONCLUSIONS
The shipworm outbreak in Oyster Creek and the South Branch of Forked River
is due to: 1) increased salinity allowing entry of 3 species of Teredo.
and more consistent reproduction of the estuarine species B.. gouldi;
2) higher temperatures, allowing faster growth and reproduction of native
shipworms and survival and breeding of subtropical species; 3) rapid water
flow; and 4) the presence of untreated wood. Reduction of temperature and
of the amount of wood can reduce the shipworm populations, but if any sig-
nificant amount of penetrable wood is present, an outbreak can occur, given
favorable temperatures and salinities.
At Oyster Creek, control of the cooling system temperature necessitated
circulation of saline water into 2 low salinity creeks; this alters the
local biota. We believe that siting of a cooling system such that it puts
bay water into an estuary is a poor strategy. Given present conditions at
Oyster Creek, we recommend that schedulable plant shutdowns occur in the
Spring when the bay is undergoing its natural warming cycle. The tempera-
ture of receiving waters should remain less than 9° in winter to prevent
growth and breeding of shipworms all year around. Any marine vessels
coming to the area of thermal addition from sub-tropical or tropical waters
should be treated to kill boring and fouling organisms that might other-
wise be introduced.
Finally, there is need to study population genetics of the introduced
shipworms to see if they are being selected to be more resistent to cold
temperatures. If so, they might eventually spread throughout Barnegat Bay.
REFERENCES
1 Turner, R.D., 1974. Bull. Amer. Mai. Union 39:36-41.
2 Young, J.S. & A.B. Frame, 1976. Int. Revue ges. Hydrobiol. 61: 37-61.
3 Kennish, M.X. & R. K. Olsson, 1975. Environ. Geol. 1:41-64.
4 Directorate of Licensing, U.S. A.E.G., 1973. Docket No. 50-219, Chapt.
5, p. 17, 115.5.21.
5 Nelson, T.C., 1922. New Jersey Agric. Coll. Exp. Sta. Report. State of
New Jersey Publ.
6 Hoagland, K.E., R.D. Turner, M. Rochester, &/or L. Crocket, 1977-1978.
Analysis of populations of boring and fouling organisms in the vicinity
of the Oyster Creek Nuclear Generating Station. 8 reports to the U.S.
Nuclear Regulatory Commission.
7 Hillman, R.E., 1978. Journ. Invert. Pathol. 31:265-266.
8 Richards, B. et al, Annual Report to J.C.P.&L., No. 14819, June 1, 1976 •
Nov. 30, 1977.
509
-------
TABLE I
A SUMMARY OF SALINITY DATA SELECTED SHIPWORM STATIONS
tn
i—i
o
1972
Month:
Station
1
2
3
4
5
8
11
12
17
1
6.0
_
6.6
_
_
.
13.0
13.0
-
8*
18.5
_
17.0
_
_
.
23.5
25.0
-
1973
1 7
2.5 10.0
_ .
2.0 8.1
_ '„
-
-
19.0 19.0
21.0 19.2
-
1974
2
20.0
_
10.0
_
_
-
22.0
22.0
-
8°
17.0
_
14.0
_
_
-
20.0
22.0
-
1975
1 7
15.0 14,0
_ -
16.0 11.0
_ _
- 18.0
21.0 19.0
19.5 19.0
20.0 19.0
_
1976
1*
15.8
-
2.8
_
18.4
18.6
16.8
18.4
-
7
#
19.8
22.5
29.0
29.0
28.5
28.5
28.0
32.0
1977
1 7*
20.4 25.0
10.0 28.0
18.7 20.0
22.3 28.0
25.5 27.5
23.9 28.0
24.0 27.0
24.0 28.0
26.7 32.0
1978
1
14.2
14.8
10.5
17.1
16.6
20.6
16.2
15.1
18.6
7
16.5
11.0
8.5
24.0
22.0
24.5
21.5
23.5
37. 0#
TABLE III
Month:
StatTon
1
2
3
4
5
8
11
12
17
AT
1
2.5
_
2.5
_
_
_
13.9
14.2
_
11.7
1972
8*
23.5
.
25.5
_
_
_
26.5
29.0
.
5.5
A SUMMARY
1973
1 7
2.2 27.0
-
4.4 27.7
_
- -
_ ~
11.1 30.0
.12.2 31.1
_
10.0 4.1
OF TEMPERATURE
2
2.8
-
3.9
_
.
_
10.0
11.1
-
8.3
1974
8°
26.6
"-
28.3
-
-
_
32.2
32.2
-
5.6
DATA SELECTED SHIPWORM STATIONS
1975
1 7
4.4 28.0
-
6.6 30.0
.
- 27.0
6.6 29.0
10.5 30.0
11.1 31.0
-
6.7 3.0
1*
3.0
-
4.5
-
3.5
3.0
1.5
1976
7
27.0
27.0
30.0
27.5
28.0
29.0
30.5
2.8 30.3
-
*
24.5
3.5
1977
1 7*
0.2 23.9
0.1 24.4
0.1 25.5
0.0 25.0
0.3 25.0
1.8 26.1
3.5 25.8
4.9 26.1
0.0 25.0
4.7 *
1
1.0
2.5
#
0.0
1.4
1.8
5.5
6.0
3.4
1978
7
23.3
21.6
27.2
22.2
22.8
24.4
25.5
25.0
26.1
5.0 2.2
*6enerat1ng Station not operating
jUMIssIng or suspect data
-No station yet established
"Increased dilution pumping from June onward
Note: Temperatures are In °C; salinities are In °/oo
-------
TABLE II
VOLUME OF WATER PUMPED, OYSTER CREEK NUCLEAR GENERATING STATION*
YEAR
Mo.
1
2
3
4
5
6
7
8
9
10
11
12
70
D1
0.0
0.0
0.0
0.0
0.0
0.2
0.8
1.0
0.4
0.3
0.0
0.0
71
D
0.0
0.0
0.0
0.0
0.0
0.3
1.0
1.0
0.4
0.0
0.3
0.5
72
C2
3.0
2.9
3.0
2.0
0.6
1.2
4.0
3.4
4.0
3.9
3.6
3.7
D
0.1
1.0
1.0
1.0
0.1
0.3
1.0
0.8
1.0
1.0
0.3
0.0
73
C
4.0
4.0
4.0
4.0
2.4
3.8
4.0
4.0
3.0
3.7
4.0
4.0
D
0.0
0.1
0.0
0.4
1.0
1.0
1.0
1.0
1.0
0.9
0.9
1.0
74
C
3.2
3.9
3.1
1.5
0.3
4.0
4.0
4.0
4.0
2.6
3.3
3.6
0
1.1
1.0
0.9
0.6
0.7
0.0
1.4
1.9
1.3
1.5
1.6
2.0
75
C
4.0
3.4
3.8
0.0
0.8
3.2
4.0
3.1
2.4
2.7
2.4
2.5
0
2.0
1.9
1.8
1.8
1.0
1.8
2.0
1.9
1.1
1.6
1.9
1.7
76
C
0.0
0.0
2.8
4.0
4.0
3.9
4.0
4.0
4.0
3.9
4.0
3.8
D
1.8
1.9
1.6
2.0
.1
.6
.8
.8
.4
.5
.9
.8
77
C
3.2
3.5
4.0
3.8
0.0
0.0
1.0
4.0
4.0
3.9
4.0
4.0
D
1.0
1.4
1.8
1.6
0.9
1.0
0.7
1.8
1.4
1.6
1.9
1.9
78
C
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0
D
1.9
1.8
1.9
1.9
1.6
1.8
1.9
1.9
*Data courtesy of J. C. P. & L. Co.
Average number of dilution pumps operating per month. 260,000 gal/m1n per pump.
o
Average number of circulation pumps operating per month. 115,000 gal/mln per pump. No circulation
pump data were made available for 1970-1971.
-------
TABLE IV
OYSTER CREEK NUCLEAR GENERATING STATION
OUTAGE DATES*
1970
1/31 - 2/12
#4/19 - 5/21
10/16 - 10/29
1971
1/25 - 1/28
2/12 - 2/19
3/3 - 3/5
#9/18 - 11/11
11/16 - 11/21
1973
1/1
#4/14
7/21
#9/8
1/13
6/4
7/25
10/5
1972
1/28 -
#5/1 -
8/9 -
11/11 -
2/2
6/20
8/15
11/13
1974
1/12 -
3/7 . -
#4/13 -
10/8 -
11/11 -
1975
1/20
3/11
7/3
10/15
11/15
1976
#1/1 - 3/11
7/28 - 7/31
1977
#4/23 - 8/4
10/21 - 10/22
11/14 - 11/15
12/3 - 12/4
1978
6/14 - 6/16
12/29 - 12/31
2/4 - 2/9
#3/29 - 5/26
6/13 - 6/15
8/27 - 9/3
9/24 - 10/3
10/5 - 10/6
11/25 - 12/1
12/19 - 12/21
#12/27 - 12/31
*Courtesy of J.C.P. & L Company.
Load reductions not Included.
#Major outage for refueling or repair.
TABLE V
Percent Wood Weight Loss
Cumulative Panels Submerged May 27, 1977
Months Submerged 123
Station
1
2
3
4
5
8
11
12
14
18#
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
0
0
11
5
56
2
1
2
2
4
10
0
25
4_
5
59
1
8
15
0
21
7
0
70
5
0
60
0
7
7
16
21
15
3
79
6
4
60
4
6
11
10
36
19
4
75
I
6
54
2
8
5
13
19
20
0
75
8
26
53
1
5
5
4
48
17
1
*
9
*
*
*
12
5
6
32
19
*
64
10
1
58
0
3
1
19
56
10
10
70
11
0
49
3
2
15
4
29
9
3
73
12
7
*
0
21
15
*
45
17
11
70
*Pane1 lost, or unretrievable due to ice cover.
#Station new in 1977.
512
-------
TABLE VI
SPECIES COMPOSITION AND NUMBER OF SHIPWORMS IN CUMULATIVE PANELS
AFTER ONE SUMMER'S EXPOSURE, AND YEARLY PANELS REMOVED IN AUGUST
SERIES BEGUN:
Apr. 30, 1976
May 27, 1977
May 31, 1978
tn
M
1/4
PANELS REMOVED:
Species:
Station
1
2
3
4
5
8
11
12
17
18
PANEL DEPLOYED:
PANEL REMOVED:
Species:
Station
|
2
3
4
5
8
11
12
17
18
_BJL
21
92
51
21
42
1
3
Bg
-30
-20
R = panel riddled;
Aug. 8, 1976
Tn Tf Tb Tsj>
4 6
1 1
2 2
1
3
Aug. 18, 1972
Aug. 8, 1973
Tn Tf Tb Tsp
Total
21
102
0
53
25
43
1
3
3
-
Total
-30
-
-20
-
-
-
R
R
-
"
more than 100 shlpworms.
# - July. panels; August panels were
t = more than
100;
missing.
Aug. 8,
B£ Tn Tf Tb
1
113
3
4
1
7
Aug. 8,
Aug. 8,
Bj. Tn Tf Tb
9
106 1
6 2
9
13
3
2
Mostly B. gouldl
1977
Tsp
1
2
4t
1976
1977
Isp.
; some
Total
1
113
0
4
6
1
0
7
0
4t
Total
9
107
0
8
9
lost
13
3
2
"
T. naval 1s.
Bg
6
1
3
1
Bg
3
42
5
4
17
Aug.
Tn Tf
1
1
1
2
5
Aug.
Aug.
Tn Tf
2
1
1
1
loot
8, 1978
Tb Tsp
46 8
80
8, 1977
8, 1978
Ik Tsp
1
417 36
180 28
Total
1
6#
0
2
4
0
57
80
0
5
Total
3
42#
0
6
18
454
208
0
loot
too small and numerous for an exact count.
- • no station yet established.
Key to the species: Bg = B_. gouldl; Tn. = I- naval 1s; Tf s !• furdfera; Tb = J_. bartscM; Tsp = J_. sp. unknown.
-------
Static
Figure 1: The Relative Locations of Stations 1-13, the Oyster Creek
Nuclear Generating Station, and its Cooling System
514
-------
New Jersey
Figure 2: The Relative Locations of Stations 14 - 17.
515
-------
COLD INFLOW AMD ITS IMPLICATIONS FOR DRY TOWER DESIGN
F.K. Moore
Cornell University
Ithaca, New York U.S.A.
ABSTRACT
The phenomenon of cold inflow near the exit of a cooling tower is
analyzed in terms of the possibility that the free plume may begin
inside the tower at some location determined by compatibility of
the plume with the parameters of the tower. It is shown that a
dimensionless group which compares the momentum flux of the entrained
flow and the buoyancy governs the problem, and critical values
of this quantity are found. Results show that conventional natural-
draft dry towers should not be greatly affected by cold inflow,
but "low" designs may require turbulence generators to enhance
entrainment rate at the tower top. In any case, the usual flare
near the tower top should be eliminated or reversed. Low towers
will also require careful heat-exchanger design to limit sensitivity
to wind.
INTRODUCTION
A natural-draft cooling tower contains a mass of slowly-moving air which is
slightly buoyant relative to the surrounding ambient air. If it were not for the
air's motion, instability would obviously exist, with cold outside air spilling
into the tower at its top outer edge. When draft is strong, as in tall chim-
neys, this cold-inflow tendency is not evident, but it usually occurs in
natural-draft evaporative cooling towers, with their lower velocities and
wide exits [1], [2]. Under wind, when the plume is laterally displaced, cold
inflow can be especially severe at the upwind lip of the tower. In a certain
range of wind speeds, inflow may be periodic, in a "puffing" mode [1].
Figure 1 shows a sequence of 3 photos taken about 5 seconds apart of the
towers at Didcot, England [3]. The plumes do not always..fill the exit area,
clearly suggesting a periodic inflow. Experimentally, Jorg and Scorer [4]
have studied cold inflow, but in too low a Reynolds number range to be helpful
for the study of cooling towers.
Obviously cold inflow, if it occurs, will degrade the performance of a cooling
tower by causing a premature separation of the buoyant flow from the tower
wall, and an effective narrowing of the exit flow area [5], Possible plume
boundaries of this sort are sketched in Figure 3c. From a design point of
view, concern about cold inflow conflicts with a desire to lower the draft
height of towers for cost and aesthetic reasons, and the concern must be
especially acute for dry natural-draft towers. Not only is the incentive
strong to develop low towers, but it is known that dry towers are severely
affected by wind [5], and cold inflow may be implicated.
516
-------
It would be important to understand how premature separation, or cold inflow,
is related to tower design parameters. As an obvious extreme case, if a
weak flow of buoyant air were introduced over a very wide area, with no
side walls, one would expect the same plume development as observed for a
large fire [6], namely, an initial buoyant acceleration characterized by
a convergent upward flow, this followed by a spreading and dilution of the
plume due to turbulent entrainment of ambient air at the plume's boundaries.
As we shall see, such a large plume will therefore have the hourglass shape
shown in Figure 2. Now, if a low tower wall should be erected around the
base of such an outflow, it would have no effect; the converging lower part
of the plume will not interact with the wall. Only when the wall is high
enough to "catch" the diverging, entrainment-dominated part of the plume
can the wall begin to take effect, and begin to cause draft. It would
seem that the extreme wide, low tower proposed by Roma [7] would fail for
this reason. Figure 3c contains sketches which illustrate these ideas.
In this paper, we will develop quantitative estimates of these effects,
beginning with an approximate analysis of a buoyant plume with entrainment,
and then matching those results to calculated dry-cooling-tower configura-
tions. For the latter purpose, we will use certain previous calculations
[8] which happen to have varied the tower parameters of interest in a
systematic way, based on a particular heat-exchanger surface.
We shall be especially interested to know where the plume may start in a
tower assuming draft and plume flow rate are properly related. Again,
referring to Figure 3c, are any or all of the sketched results possible?
The effect of entrainment rate will be explored (presumably, a high entrain-
ment rate will favor the plume being "caught" by the wall), and means of
increasing that rate will be discussed.
Finally, we will suggest design parameter limitations for the avoidance of
cold inflow, and relate these results to certain new estimates [5] of the
sensitivities of tower thermal performance to various loss mechanisms.
These results will suggest the proper directions for efforts to develop
towers of low draft requirement.
Although our chief concern is with dry, natural-draft towers, most of our
results (up through Figure 5) are applicable to wet towers as well.
PLUME ANALYSIS
General Solution
We begin with the classical one-dimensional plume analysis of Morton,
Taylor and Turner [9]. Although that theory is generally used to describe
plumes emanating from a virtual point source, it is possible to solve
their equations for an area source, which reveals the initial convergent
behavior of interest here. Morton [10] has made such an extension of [9],
but not in a form suitable for our purposes. The basic equations are
517
-------
(b2u)' = Zebu
(bV)1 = b2ga
2
irb Ua = Q/g
(a = Ap/p)
(1)
(2)
(3)
which express, respectively, that volume
flow in the plume increases with distance
by entrainment according to the coefficient
e (a in [9]), that momentum flux increases
by buoyancy, and that heat flux is con-
stant along the plume. The sketch
illustrates notation. Equations (1-3)
assume one-dimensional flow with a
"top-hat" velocity profile. The inherent
limitations of a one-dimensional flow
model near the beginning of the plume (at (1
would not warrant refining the profile
assumption.
The solution of eqs. (1-3) which envisions a
certain flow through level (T) (see sketch) is
U = 0.8286 (-
plume
boundary
b = 1.099
1/5 3/10
) Cn
U
2b,
1.6u2E
(4)
(5)
\
where u
of x:
and b are dimensionless functions of a dimensionless version
J/5
I'
c^ge
bM
x = 0.3433 (-L-L4)
c 3/10 *
C] x
(6)
The functions u* and b* are displayed on Figure 2. The dashed and dot-
dashed lines^correspond to assumptions of pure entrainment (with a virtual
source at x = -2.27) or pure buoyancy. The plume radius b has a "throat"
with b* = 1-400 when x* = 0,641.
fixed relative to the tower until
We note that the coordinate
u and b are specified at
x is not
518
-------
Hatch to the Tower.
The next step is to match Equations (4) and (5) to tower conditions. The
first, and simplest test is to ask whether a plume assumed "anchored" at the
tower top would first converge upward, as in Figure 3c or diverge as in
Figure 3a. If the former, then we should consider the possibility that it
"falls down" into the tower (Figure 3c). If the latter, the plume is secure
and no inflow is possible. The dividing condition is a vertical exit
(Figure 3b), when the throat of b (Figure 2) is precisely at the tower exit.
We may differentiate Equation (5) to find when that happens, and the result is
U?e
Where for future reference, we assign the symbol F to the dimensionless
group. Divergence (Figure 3a) corresponds to F > 1/8. That is, if the
momentum of the entrained flow (u«epu) is large enough compared to buoyancy,
cold inflow cannot occur. A typical tower might have u » 4 mps, Ap/p = 0.1
(air 30°C warmer than ambient), and e is normally about 0.1 without wind.
In such a case, the exit diameter DE would have to be less than 13m to
be entirely proof against inflow. This is about 5 times less than the usual
DE for natural draft, and so we must investigate further to find whether
cold inflow should be expected in such cases.
We note that chimneys, with diameters smaller than 13m (and velocities higher
than 4 mps) should never experience inflow, and that mechanical-draft towers,
which have exit velocities up to 10 mps and quite small diameters (13m is
not unreasonable) should also be immune to cold inflow.
We now go the next step and ask, if F < 1/8, how far the plume may fall into
the tower. Referring to the notation sketch, we should compare the possible
YI to Y, the tower height. A simple draft equation would say that
-, nf\i v ^ — ' n^ r fo\
°-|9(Y " Yi / 2" u] CL \°'
where Ci is the overall pressure-loss coefficient of the tower. Chiefly, it
reflects the pressure drop through the bundles, and C, = 16 would be a
representative figure. One may introduce Equation (8) into Equations (4) and
(5) and find a set of solutions, depending on the choice of e, for which the
plume just grazes the tower exit. Figure 4 shows the results for Ap/p = 0.1,
CL = 16, DE = 61 m and Y = 110 m. In that case, Equation (8) says that if
cold inflow does not happen (YI = 0), the exit velocity is u-j = 3.7 m/s.
These parameters would represent a dry cooling tower. For a wet tower,
Ap/p and CL would both be smaller.
Figure 4 suggests that for this rather ordinary tower, the plume may tend to
"fall in" unless e is greater than about 0.5. If e = 0.2, the plume could
begin about half-way, and the exit flow would be only about 0.72 of the
519
-------
full-tower value, 3.7 nips, and a corresponding loss of cooling performance
would be anticipated. Actually, except under wind, mixing coefficients
should not even be as high as 0.2.
Clearly, these results are too pessimistic; cooling towers evidently do not
suffer from cold inflow to the degree suggested by Figure 4. In the next
section, we shall see how to make more realistic estimates. However,
Figure 4 is useful merely to illustrate the idea of a plume "falling in" a
cooling tower.
EFFECTIVE ENTRAPMENT RATE
Restricted-Flow Model.
On reflection, one sees that the grazing plumes of Figure 4 are impossible,
because there is no way to supply the entrainment
flow to the plume boundary. If a plume "falls in",
it may go only so far; the sketch shows how the
entrainment required by the plume boundary must
be supplied through the gap (d) between plume and
lip, and it must be supplied in proportion to the
penetration depth Y-].
Let us estimate the cold inflow velocity ue.
First, we note that separation from the lip should
be expected, and classical.free-surface theory [11]
would narrow the gap to Jrd, as sketched. We
note, too, that the downward velocity ue might
remain constant because pressure tends,to be con-
stant along a (dashed) free streamline. Assuming
that u inside the plume is nearly constant, the
velocity difference across the plume boundary
is u + ue. It is this relative velocity which
would govern turbulent entrainment. That is,
v« * e(u + ue). The entrainment volume flow
e(u + ue)Y] must be provided by that over the
lip, ued/2. Equating these two expressions,
one finds that
u
u"
(9)
For use in calculations which express entrainment as eu, it will be appro-
priate to define an effective entrainment rate ee such that eeu = e(u + ue),
which is the true rate. Equation (9) then provides that
J2
e
(10)
520
-------
where N = 2 according to Equation (9), because a free-surface separation
was assumed in the model sketched. Actually, the separated zone would proba-
bly become fully mixed and ue would be reduced accordingly. It would be
reasonable (and conservative) to take N = 1. Obviously, the effective
entrainment rate ee may be many times the physical rate e, and we note that
penetration depth cannot exceed (l/Ne)d.
Using Equation (10) in the combined tower-internal plume analysis (Equations
(4-6, 8), one finds a set of possible solutions for each case. In each set,
there is solution with maximum penetration (Yi). These "worst-inflow"
solutions yield Figure 5. Clearly, if the usual value e = 0.1 applies,
effects of cold inflow should be expected for towers less slender in shape
than about 1.8, which is typical of current practice [5], However, if the
basic entrainment coefficient e can be increased to 0.4, very low shapes
will "work." Perhaps experiments will show that turbulence generators near
the top of the tower (on the inside surface) can substantially increase e.
We should note that, under wind, plumes develop above a tower as if the
average E is about 0.4; however, this effect would perhaps not be felt
inside the tower, and no doubt, circumferential variations of e are impor-
tant. Thus, one should not rely on wind to provide the e needed to prevent
cold inflow.
It has been proposed by Professor Ernst of Karlsruhe University [3] that
cold inflow may be prevented by turning the shell inward slightly at the
tower top, causing a local overpressure to oppose inward flow. From the
point of view of this paper, such an inward curl of the shell may indeed be
helpful if it corresponds to the desired initial plume angle (Figure 3c);
the plume which would otherwise "fall in" might thereby be stabilized at
the top against small (but not large) disturbances.
In any case, it is obvious that the usual slight flare given to the exit portion
of the tower would encourage cold inflow by allowing room for entrainment
flow to supply a"fallen-in" plume. Such flares should be eliminated, and
replaced by a slight convergence, over the upper third (say) of the tower.
Finally, we may recall the definition of F in Equation 7, and ask how that
number compares with the value 1/8 for vertical exit flow, for the conditions
assumed in Figure 5. They were a-, = 0.1 and C, = 16. For no inflow, and
e = 0.1,
8F = 14.18 |£= 0.089 s (11)
LL
Inasmuch as inflow seems not serious if s > 2, one concludes that the value
of 8F should generally be kept greater than about 0.18, or, in round
numbers, 0.2. In other words, a substantial deqree of initial convergence
can occur without significant inflow, for typical cooling-tower parameters,
521
-------
IMPLICATIONS FOR LOW-TOWER DESIGN
Cold Inflow as a Function of Dry-Tower Design.
Figure 5 has shown how one variable, tower shape(s), can affect prospects
for cold inflow. Other variables, such as bundle resistance (Ci ), will also
be important, presumably. In [5], simple formulas were developed to evaluate
the effects of a wide variety of parameter changes for dry towers, and the
same approach may be applied here. We will consider F to be a design
parameter, in effect, and assume that it should be kept above the value
8F = 0.2 mentioned in the previous section.
Using [5], one may show that
8F = 14.18 ^ (12)
In effect, CL in Equation (11) is replaced by B5 . Losses outside the heat
exchanger are represented by & as a coefficient multiplying exit dynamic
head, and the value of B is typically about 1.4. The quantity £ is the
ratio of actual exit area to that which would be needed if there were no
heat-exchanger flow resistance [5,8]. It is an optimizing parameter; small
C means a small tower but a large bundle surface area, and large ? means
the reverse. Thus, c would be selected according to the designer's relative
weighting of tower and bundle costs. A common choice would be 5 = 2.5.
Figure 6 illustrates this relationship for a particular type of bundle and
tower shape, as calculated in [8].
Equation (12) in conjunction with Figure 6 shows that the designer potentially has
the difficulty that the need for a large enough F might compel him to adopt
a design with an uneconomically large heat exchanger. With this question in
mind, a number of different cases treated in [8] were examined to see how F
varies with parameter changes. Figure 7 shows the results.
The left-most pair represent two choices of 5, within the customary range.
Even the larger produced an 8F of 0.28, well above the value expected to
be critical for inflow. Next, the effect of changing to a single-pass bundle
is considered. Again, 5 = 2.5 yields 8F = 0.28, but a larger tower and a
smaller heat-exchange area (Aa) are required. Clearly, for a given rela-
tion of tower size and Aa, fewer passes tend to help increase F. Doubling
heat-rejection rate produces the next example, and a simple economy of scale
is evident for both tower size and Aa.
The most interesting case is the last, which considers a low tower shape
(s = 1 instead of 1.7). Here, for the same 8F, a smaller E, (and larger
Aa) is needed. Suppose we go to £ = 2.5; then the tower is still low,
compared to the first more conventional case, but Aa is now reduced to a
reasonable level. However, we must be able to tolerate 8F = 0.16, which
Figure 5 suggests is marginal. No doubt, special design features will be
needed to make this last case secure against cold inflow, but the requirement
seems not to be extreme or discouraging.
522
-------
Sensitivities to Loss Mechanisms.
Figure 7 also mentions two parameters S6 and Sw. These quantities may
be written as follows, for the present purpose:
TTD ' ^T1* (const.) \ = (const. )Sg (13)
= (const.) (_!E)2/3 s (const.)10-2s (14)
9w Q£s
They are derived in [5], and represent the proportional changes of ITD
(overall cooling performance) as a loss mechanism other than bundle resistance
changes (Equation (13)), and the change as wind-connected loss changes
(Equation (14)). The constant is unknown or uncertain in both cases. In
both cases, cold inflow might be implicated. In any event, these expressions
can describe on a relative basis the susceptibilities of tower performance
to aerodynamic losses.
Returning to Figure 7, the two most noticable trends of these susceptibility
factors are, first, that large scale reduces the danger of performance loss
due to wind, and second, that low shape (small s) greatly increases that
danger. Because performance loss due to wind may be traced with some confi-
dence to the behavior of air flow as it enters the bundle face [3,5], great
care will be needed in designing heat-exchanger arrangements in the
tower .
The Approach to Low Dry-Tower Design.
With the potential siting and cost advantages jof a low shape for natural -draft
dry cooling, we may recapitulate the results of this discussion, first admit-
ting that the theory of cold inflow advanced here is rough, and requires
verification.
1. Cold inflow seems not to be a fundamental bar to the maintenance of
draft in towers down to, or even below, Y/D^ = 1, provided certain improve-
ments in customary practice are made:
(a) Instead of the usual outward flare of the upper third of the tower,
it should converge slightly, also over the upper third of the tower.
(b) Care should be taken to promote vigorous flow near the tower wall.
(c) Introduce turbulence generators just inside the tower exit.
The foregoing suggestions are applicable to wet as well as to dry towers.
2. Low towers will be more sensitive to performance loss due to wind
than more slender towers. Although there is no doubt a connection with cold
inflow here (a tower precarious in that regard will no doubt be more influenced
by wind), here the chief concern must be the heat exchanger and its arrange-
ment in the tower. This, matter is discussed in [12].
523
-------
3. Any developments which reduce the cost of heat-exchange surface
(e.g., plastic tubing [7]) will permit lower values of £ and hence,
more security against inflow; however, sensitivity of tower performance to
wind losses would be higher (see Figure 7, case s = 1). Whether such a
change would be helpful in low-tower design is therefore uncertain.
ACKNOWLEDGEMENT
This work was supported by US DOE and by NSF (Heat Transfer Program).
REFERENCES
1. G. Ernst, "Cooling Towers, Design and Plume Behavior," Seminar of
International Centre Heat and Mass Transfer, Dubrovnik, 1975.
2. E. Baer, G. Ernst, and DA Wurz, "Untersuchung zur Schwadenstrbmung in
den Kronen von Naturzugkuhltiirmen," VDI Kuhlturmseminar, Dusseldorf,
May 25, 1977. Inst. f. Technische Thermo., U. Karlsruhe.
3. F.K, Moore and K.E. Torrance, "Air Flow in Dry Natural-Draft Cooling
Towers Subject to Wind," Report to ERDA, Cornell University, Dec. 8, 1977.
4. 0. Jorg and R.S. Scorer, "An Experimental Study of Cold Inflow into
Chimneys," Atmos. Environment, v. 1_, p 645, 1967.
5. F.K. Moore, "Effects of Aerodyanmic Losses on the Performance of Large
Dry Cooling Towers," ASME Publ. 78-WA/HT-18, Winter Annual Meeting,
Dec. 10-15, 1978.
6. J.S. Adams, D.W. Williams, and J. Tregellas-Williams, "Air Velocity,
Temperature, and Radiant-Heat Measurements within and Around a Large
Free-Burning Fire," 14th Symposium (International) on Combustion,
p 1045, The Combustion Institute, 1973.
7. C. Roma, "An Advanced Dry Cooling System for Water from Large Power
Station Condensers," Proc. Am. Power Conf., v. 35_, p 626, 1973.
8. F.K. Moore and C.C. Ndubizu, "Analysis of Large Dry Cooling Towers with
Power-Law Heat-Exchanger Performance," Journal of Heat Transfer,
Aug. 1976, p. 345.
9. B.R. Morton, G.I. Taylor, and J.S. Turner, "Turbulent Gravitational
Convection from Maintained and Instantaneous Sources," Proc. Roy. Soc.,
Ser. A, No. 234, p. 1.
10. B.R. Morton, "Forced Plumes," Jour. Fluid Mech., v. 5_, p. 151, 1959.
11. H. Lamb. "Hydrodynamics." 6th Ed., para. 73, Dover, New York (1945).
12. F.K. Moore, "Aerodynamics of the Heat Exchangers and their Arrangement
in Large Dry Cooling Towers," ASME Publ. 78-WA/HT-19, Winter Annual
Meeting, Dec. 10-15, 1978.
524
-------
Figure 1. Sequence of photographs, about 5 seconds apart,of the cooling towers
at Didcot, England. Flow separation is especially evident in the
discharge from the middle tower.
525
-------
UJ
O
cn
o
UJ
cn
1
en
UJ
DIMENSIONLESS VELOCITY, u*
AND RADIUS, b*
Figure 2. Dimensionless plume velocity (u*) emd radius (b*) as functions of
dimensionless vertical distance (x ), from Equations (4-6). Dashed
lines refer to a pure entrainment or point-source extreme, and the
dot-dash lines refer to a buoyant plume with no entrainment.
526
-------
tn
plume
f
'— exit
tower
(a) PLUME DIVERGENT (b) PLUME VERTICAL
AT EXIT ( #>'/8) AT EXIT
(c)PLUME CONVERGENT
AT EXIT (#< !/8)
Figure 3. Sketches of possible exit plume shapes, depending on F (Equation (7)).
If plume is initially convergent (c), the plume may perhaps originate
within the tower; three possibilities are shown.
-------
€ = 0.56
u,
"u,"
Figure 4. The set of plume boundaries which just graze the tower top, for
Ap/p = 0.1, CL = 16, DE = 61m, Y = 110m, and various entrainment
rates e. Also indicated are the reduced exit flow velocities
which result from the "fallen-in" plumes. Solutions are possible
only in the range 0.20 < e < 0.56.
528
-------
tn
0.3 0.4 0.6 0.8 I 1.5 2
TOWER SLENDERNESS, ssY/yf Dt
3 4
Figure 5.
Calculated depths of deepest penetration (Y-|) of plume into the
tower, and the corresponding decrease of through-velocity (u]),
for dry towers with Ap/p = 0.1, Ci = 16, and two values of e,
as functions of tower shape. Results are independent of scale.
-------
EXIT AREA, ipD!(loV)
Figure 6. Possible combinations of air-side heat-exchange area (Aa) and
tower exit area (-rDF2). as specified by the parameter 5. From
[8], for spine-fin4surface, heat rejection of 608 GJ/hr, and ITD
of 22°c.
530
-------
150 -
On
1
-
-
mm
mmt
1
Q=608GJ/h
2 passes
-^^ s=l.7 ^
£ - 2.5 2.0
8# = .28 .44
Sp = .16 .25
Sw = -32 .34
IO-5A0 = 2.0m2 2.1
1 pass
2.5
.28
.16
,41
1.4
•
1216 GJ/h
2.5
.28
.16
.24
3.5
, 125-
scale,
m.
100"
75-
1
s^ S = 1 ^^r
2.5 2.0
.16 .26
.16 .25
.68 .73
2,4 2.6
5O-
25-
Figure 7. Display of calculations based on [5] and [8] of tower size, inflow
parameter (8F), loss sensitivity (S ), wind sensitivity (Sw), and
heat-exchange area, showing effects of (from left) number of passes,
tower scale, and low shape, each changed one at a time. Height
and exit diameter are measured on the scale given.
-------
- DRAFT -
AN IMPROVED METHOD FOR EVAPORATIVE,
CROSS-FLOW COOLING TOWER PERFORMANCE ANALYSIS
by
Kenneth L. Baker
Thomas E. Eaton
Mechanical Engineering Department
University of Kentucky
The analysis of cross flow cooling towers may use one of
several methods. For example, Zivi and Brand [1] or Baker and
Mart [2] use enthalpy as the driving force for heat transfer
while other analyses use different quantities for calculating
heat transfer potentials [3]. This paper presents a more exact
method for the analysis of evaporative, cross-flow cooling towers.
A new method of calculating cooling tower performance in
a cross-flow tower was developed using the same basic equations
as Nahavandi [4] extended to a two-dimensional analysis, The
analysis incorporates variations of water and air properties
inside the cooling tower fill.
Several tower design and operating parameters are required
for the analysis: tower-on water temperature, water loading,
air loading, dry bulb temperature, relative humidity, and tower
packing hydraulic diameter. The Number of Transfer Units (NTU)
is calculated but is not required for this analysis.
A Fortran program, XFCTWET, for analyzing linear (in-line)
cross-flow towers has been written. The tower fill is divided
into a mesh or grid; the tower analysis proceeds using finite
difference techniques. A set of twenty-six (26) equations is
iteratively solved for each grid in the tower fill. A typical
mesh of 25 X 25 provides accurate results and reasonable computer
costs. The variations of all parameters inside the cooling, fill
532
-------
may be output from the program, e.g., local water and air tempera-
tures, relative humidity, and evaporation rates.
References
[1] Zivi, S. M., and Bruce B. Brand, "An Analysis of the Cross-
Flow Cooling Tower", Refriger^ation_Engirie^ri.ng, Vo 1. 64 „
No. 8, August, 1956, pp. 31-34, and 90:91.
(21 Baker, Donald R., and Leon T. Mart, "Cooling Tower Charac-
teristics as Determined by the Unit-Volume Coefficient",
Refrigeration Engineering, Vol. 60, No. 9, 'September, 1952,
pp. 965^971.
[3] Eaton, T. E., "Evaporative Cooling Tower Performance: A
Comprehensive Bibliography", Industrial Heat Rejection
Project Report, Mechanical Engineering Department, Univer-
sity of Kentucky, May 1978.
[41 Nahavandi, Amir N., and Johann J. Oellinger, "An Improved
Model for the Analysis of Evaporative Counterflow Cooling
Towers", Nuclear Engineering and Design, Vol. 40. 1977,
pp. 327-336.
533
-------
INTRODUCTION
Closed cycle evaporative cooling systems are being used extensively at
present and are being considered, more than not, for condenser cooling at
future sites of steam electric power plants. With the ever increasing
emphasis on the efficient use of fuels, it is important that these cooling
systems be properly specified, designed, constructed and tested. For the
mechanical draft cooling tower, this process carries the burden of another
variable to be determined, that of the impact of recirculation on the thermal
performance of the cooling equipment and the operating efficiency of the
power plant.
Recirculation is defined as the entraintrient of the warm exhausted cooling
tower plume into the inlet air flow of the tower. The result is an increased
inlet air wet-bulb temperature (over the prevailing ambient wet-bulb tempera-
ture) and an associated increase in the thermal equilibrium of the cooling
tower over that level it would have obtained in the absence of recirculation.
If a power plant is already operating near its design thermal equilibrium,
the addition of cooling tower plume recirculation may result in inefficient
plant operation or in more drastic cases may force a reduction in plant load.
Despite data from field measurements made over twenty years by the Cooling
Tower Institute and modeling studies presented to the American Society of
2 3
Mechanical Engineers in 1972 ' , neither group has made, until recently, a
concerted move to account for recirculation in their codes and associated
test practices. In fairness to both groups, it was not clear for some
time that instrumentation, sophisticated enough for cost-effective acqui-
sition of multiple-point (including multiple heights) inlet wet-bulb
534
-------
THE IMPACT OF RECIRCULATION ON THE SITING,
DESIGN, SPECIFICATION, AND TESTING
OF MECHANICAL DRAFT COOLING TOWERS
K. R. Wilber1
R. D. Moore2
D. E. Wheeler3
A. E. Johnson
ABSTRACT
Significant recirculation of the warm exhausted plume of a mechanical draft
cooling tower can occur as a result of wind induced pressure gradients
around the cooling tower structure. Despite data and studies (although
both somewhat limited) indicating an improved methodology was required for
assessment of the contribution of recirculation to thermal performance
degradation of these towers, it was only recently that this methodology was
developed and demonstrated. This paper presents some of the first field
data acquired using this improved assessment methodology. Plume recircula-
tion data from three separate rectangular mechanical draft cooling towers
are presented with specific emphasis on entering air wet-bulb temperature
profiles and associated performance reductions. For each data set corre-
lations are made with wind speed and the component of wind direction
perpendicular to the tower's major axis. Finally, the importance of
recirculation on tower specification, operation and design is discussed
along with general recommendations for a consistent and repeatable accept-
ance testing methodology.
1. K. R. Wilber is Assistant Director, Cooling Systems Projects, Environ-
mental Systems Corporation.
2. R. D. Moore is a Project Manager, Cooling Systems Projects, Environ-
mental Systems Corporation.
3. D. E. Wheeler is an Engineer, Cooling Systems Projects, Environmental
Systems Corporation.
4. A. E. Johnson is a Mechanical Engineer, Department of Engineering
Research, Pacific Gas and Electric Company.
535
-------
4
temperatures, was available. Other modeling and dimensional analyses have been
conducted since that period, but the need for field data still existed. In
5
1976, the Electric Power Research Institute funded a study examining
specific aspects of cooling tower testing methodology, which addressed, among
other areas, instrumentation and instrumentation deployment for reasonable
acquisition of inlet air wet-bulb temperature. The approach that evolved
proved to be a dramatic improvement over the CTI work which has been criti-
cized for inaccuracy and inconsistency since some heat balance calculations
o
showed errors of 50 percent or more . This work also lacked data on the
larger towers characteristic of today's power plants. Finally, resultant
correlations of the data could only be made with tower length, while factors
such as exhaust velocity, fan stack length, wind speed and direction,
intuitively more responsible for exhaust plume dynamics and degree of
recirculation were non-correlatable.
With this background, field data acquired on 3 different size rectangular
mechanical draft cooling towers under widely varying meteorological con-
ditions are presented. A preliminary correlation (in lieu of more data and
more exhaustive data examination) is presented tying increased entering wet-
bulb to prevailing wind speed and direction. Further, the impact of
recirculation on tower sizing is discussed. Finally, the accuracies and
repeatability of the testing methodology are addressed with recommendations
for specification and acceptance test approach.
536
-------
DATA ACQUISITION, PRESENTATION AND ANALYSIS
General Approach to the Data Acquisition
To assess the degree of recirculation on the crossflow mechanical draft
cooling towers tested, inlet air wet-bulb temperature measurements were
made using a network of nine sensors (three vertically at three positions
horizontally along the tower) on each side of the tower within approxi-
mately 2 meters of the air inlet. Each sensor was placed at the center
point of imaginary equal area segments of inlet air. (More detail
regarding the background of this approach is given in References 3 and
5.) Supplementing the inlet wet-bulb sensors were typically 3 wet-bulb
sensors located upwind of the tower consistent with the approach specified
in ASME PTC-23, 1958, test code for Atmospheric Water Cooling Equipment.
With this requirement for near-simultaneous acquisition of 21 wet-bulb
sensors, not to mention system hot and cold water temperature measurements,
it was determined that the use of a data acquisition system would provide
the most effective measurements approach. Details of the particular
system are given in References 5 and 6. Supplementing these temperature
measurements, additional performance parameters including waterflow rate,
fan horsepower, wind speed and wind direction were acquired and data were
analyzed in a fashion generally consistent with existing test code stipu-
7 8
lations ' . In two of the campaigns, on-line magnetic tape systems
interfaced with the temperature data acquisition system facilitated
storage and subsequent analyses of the data.
Data Presentation
Performance data were acquired on three separate cooling towers with design
537
-------
specifications shown in Table I below.
TABLE I
GENERAL COOLING TOWER DESIGN INFORMATION
FOR THE TOWERS TESTED
Tower A Tower B Tower C
Water Flow Rate (gpm) 192,700 372,000 275,000
Hot Water Temperature (°F) 116.0 100.0 115.7
Cold Water Temperature (°F) 94.0 82.0 84.0
Wet-Bulb Temperature (°F) 79.0 70.0 55.0
Number of Cells 9 13 16
Fan Horsepower 187 189 160
Fill Height (ft) 42 36 40
Fill Length (ft) 324 469 596
A minimum of seven separate tests, conducted under different meteorological
conditions on each of the towers listed in Table I, were performed. An
example of the type of inlet wet-bulb distribution resulting from the pre-
vailing winds at Tower A, is shown in Figure 1. It is noted that the winds
are well within the constraints of the test codes for thermal acceptance
testing, yet a difference of nearly five percent results depending on
whether one calculates capability based on inlet air wet-bulb temperature
or ambient (upwind) wet-bulb temperature.
As a result of the acquisition of inlet wet-bulb temperature profiles on
the 3 towers, it was noted that recirculation can be significant and the
538
-------
FIGURE 1. EXAMPLE OF INLET MEASURED WET-BULB TEMPERATURE
DISTRIBUTION UNDER A SPECIFIC WIND CONDITION
North Side
V
t
f
South Side
Avg. Wind Speed 1.9 M/Sec ( 4.2 roph)
PLAN VIEW
Average Measured Temperatures (°C)
79.3
75.9
76.5
80.9
77.1
76.3
79.7
76.2
75.8
North Side (Facing South)
75.6
74.9
74.2
74.4
74.6
76.2
76.5
76.6
77.1
South Side (Facing North)
Average Hot Water Temperature (°C) 115.9
Average Cold Water Temperature (°C) 94.6
Circulating Water Flow Rate (gpm) 204.670
Average Fan Horsepower 184
Average Ambient Wet-Bulb Temperature (°C) 75.0
Average Inlet Wet-Bulb Temperature (°C) 76.5
Ratio of Tower Capability Based on Ambient Wet-Bulb Temperature
to Tower Capability Based on Inlet Wet-Bulb Temperature 0.954
539
-------
resultant impact on capability substantial. This fact is shown in Table
II, which includes both the increase in the average inlet wet-bulb as well
as the associated reduction in tower capability. Recalling that Tower C
was the longest of the 3 towers (see Table I), it is not clear, as was
suggested by Reference 1, that recirculation is a function of tower length
for towers of these proportions; on the contrary, the data seem to indicate
that recirculation is inversely proportional to length for winds across the
tower, but this conclusion would be premature without additional analyses.
The data do suggest that a combination of wind speed and wind direction
with respect to the tower's major axis may be more directly responsible for
recirculation and degradation of performance. The next section addresses
itself to examining the potential for this correlation.
Data Analysis and Discussion
The influence of wind speed and wind direction on increased inlet air
wet-bulb temperature was determined using standard correlation techniques.
It was found based on 42 cases that individually neither parameter cor-
related well with increasing wet-bulb (r = 0.18 and r = 0.02 respectively).
It was determined that the component of the wind perpendicular to the
tower's major axis was significantly correlated with the increase in wet-
bulb over ambient (r = 0.81). Using the equation:
A = a + b (u sine)
where A = the difference between ambient and inlet air wet-bulb
temperature (°F)
a,b = constants
u = ambient wind speed
G = the angle of the prevailing wind with respect to the
major axis of the cooling tower
540
-------
TABLE II. REPRESENTATIVE RECIRCULATION DATA AND
THE RELATED IMPACT ON TOWER CAPABILITY
Wind
Test Speed
Tower No. (mph)
1 8
2 6
A 3 5
4 2
5 7
1 20
2 20
B 3 20
4 30
5 35
1 9
2 8
C 3 9
4 6
5 9
Wind
Direction Increase in
(With Respect
to the
Tower's Major
Axis)
65
85
75
85
35
0
30
30
75
75
20
40
30
60
30
Average
Inlet Wet-Bulb
Temp
(°F)
4.5
3.9
2.8
1.6
2.3
-0.1
1.7
1.4
7.9
8.0
0.3
1.1
1.6
1.0
0.5
Decrease
in
Capability
(%}
8
6
5
4
6
0
5
6
15
11
1
2
3
1
1
541
-------
Figure 2 shows a "best fit" line plotted using a = 0.13 and b = 0.24 and
associated representative data points. Although there is some scatter in
the data, it is clear that even relatively light (i.e., less than 10 mph)
wind conditions (with e greater than about 30°) can cause increases in
entering wet-bulb in excess of 2°F. Further, it is clear that ground-
level measurements made upwind of the tower cannot assess the degree of
recirculation existing during a test.
From a design and specification standpoint, if elevated inlet wet-bulb
temperatures of over 2°F greater than ambient represents a characteristic
situation for a specific tower and site meteorological situation, a
reduction in capability on the order of 5-10 percent may be expected. In
this situation, the addition of another cell of cooling tower may be
required to meet the required cold water return temperatures under demand-
ing meteorological conditions.
Finally, tower capabilities calculated based on multiple tests, using
inlet wet-bulb temperature were typically more repeatable under
varying meteorological conditions than those using ambient wet-bulb
temperatures. As an example for seven tests on Tower B, where
substantial recirculation was measured, the standard deviations for
capabilities determined using inlet wet-bulb and ambient wet-bulb were
2.76 and 5.77% respectively. Additionally, the average reduction in
capability was six percent. The fact that the capabilities are more
repeatable using the inlet wet-bulb approach is expected since, the
manufacturers performance curves, on which the capabilities are based,
542
-------
tn
U.
o
t-
UJ
CO
i
K
U
Z
II
0
A - TOWER A
B - TOWER B
C - TOWER C
2466
22 24 26 28 30 32 34 36 38 40
U SIN 0 (mph)
Figure 2: Correlation of Increase in Wet-bulb Temperature
with the Component of Wind Perpendicular to the
Cooling Tower's Major Axis
-------
are constructed using inlet wet-bulb. Despite this, the use of inlet
wet-bulb as the basis for performance and specification has met with
periodic resistance. This resistance is fueled by the false hope that
the manufacturer will assume the responsibility for recirculation or
has some control over the wind-induced eddies on the lee side of the
tower and the resultant rise in inlet wet-bulb temperature. The data
acquired in this study strongly suggest that neither is the case.
544
-------
CONCLUSIONS
Measurements of plume recirculation were made on three large rectangular
crossflow mechanical draft cooling towers. Based on these measurements,
the following conclusions are made:
1. Significant wind-induced recirculation can occur if towers are not
properly oriented. When proper orientation cannot be effected, considera-
tion in the tower design must be made to increase its capacity to offset
expected performance reductions which result from recirculation.
2. A reasonable correlation exists with the component of wind perpendic-
ular to the major axis of the tower and the increase in entering air
wet-bulb temperature over the prevailing ambient wet-bulb temperature.
It is recognized, however, that more data need to be acquired to refine
this correlation.
3. The inlet wet-bulb measurement approach provides a more repeatable
result in the determination of cooling tower capability when tests are
taken under different wind conditions.
545
-------
BIBLIOGRAPHY
1. "Technical Sub-Committee #2, Report on Recirculation," Cooling Tower
Institute Bulletin PFM-110, July 24, 1958.
2. Reisman, J. I., "A Study of Cooling Tower Recirculation," presented
at the American Society of Mechanical Engineers' 93rd Winter Annual
Meeting, November 26-30, 1972.
3. Reisman, J. I., and Ovard, J. C., "Accurate Performance Testing of
Crossflow Cooling Towers," presented at the American Society of
Mechanical Engineers' 93rd Winter Annual Meeting, November 26-30,
1972J
4. Kennedy, J. F. and Fordyce, H., "Plume Recirculation and Interference
in Mechanical Draft Cooling Towers," presented at the Cooling Tower
Enviroment - 1974, University of Maryland, March 4-6, 1974.
5. Environmental Systems Corporation, "Examination of Specific Aspects
of Cooling Tower Testing Methodology, Research Project RP 905-1-2,
prepared for the Electric Power Research Institute, August 1978.
6. Wilber, K. R. and Maulbetsch, J. S., "Field Examination of Cooling
Tower Testing Methodology," presented at the Cooling Tower Institute
Annual Meeting, January 31 - February 2, 1977.
7. The American Society of Mechanical Engineers', "Atmospheric Water
Cooling Equipment," Power Test Code No. 23, 1958.
8. The Cooling Tower Institute, "Acceptance Test Code for Water Cooling
Towers, ATC-105, February 1975.
546
-------
An Investigation into the Mineral Concentration
of Individual Drift Droplets from a Saltwater
Cooling Tower
by
Ronald 0. Webb1
Richard S. Nietubicz2
John W. Nelson3
A common assumption in wet cooling device drift studies and drift transport
modeling is that the drift droplets carry the same mineral concentration as
the circulating water. Using this assumption, cooling device mineral emis-
sions calculated from liquid emission data rarely agree with direct measurements
of mineral emission. Ratios of measured to calculated mineral emission values
are generally greater than one and sometimes range up to a factor of ten depend-
ing on the cooling device and operating conditions. This difference has been
attributed to an increase in the bulk concentration of the droplets due to the
evaporative cooling process.
This paper presents results of a study to examine residues from individual
drift droplets collected near the exit plane of a large natural draft salt
water cooling tower located at the Chalk Point steam electric generating plant
in southern Maryland. The droplets ranged in size from 10 ym to over 1000 pm
in diameter and were collected on a filter paper which yields a stain when a
droplet impacts upon it. The stains were individually examined using nuclear
analysis techniques to determine the mineral mass residue in each stain. Mineral
concentration in the original droplets were then calculated from this data.
Results of the measurements are presented together with a description of the
examination techniques and sensitivities.
Environmental Systems Corporation
2State of Maryland, Department of Natural Resources
3Florida State University
547
-------
COGENERATION TECHNOLOGY AND OUR
TRANSITION FROM CONVENTIONAL FUELS
J. W. Neal
Acting Assistant Director
Heat Engines and Heat Recovery
Division of Fossil Fuel Utilization
Department of Energy
Washington, D.C.
ABSTRACT
Cogeneration — the simultaneous production of either electricity or
mechanical work along with useful thermal energy has long been used
to provide industrial and residential/commercial users with on-site
energy needs. Also known as total energy systems, integrated energy
systems, dual-purpose powerplants, and waste-heat utilization systems,
cogeneration provided over 15 percent of the nation's electric energy
in 1950. But after 1950, as a result of what appeared to be a limitless
supply of cheap energy resources, cogeneration declined. Today, it
accounts for only 4 percent of our electric energy.
The oil embargo of 1973-74 and the subsequent quadrupling of oil prices
stimulated a new national interest in conservation as an important
means of more efficiently using our energy resources. The National
Energy Plan submitted by President Carter to the Congress in April of
1977 embodies conservation as its cornerstone. One of the principal
strategic components is that industries and the utilities using oil
and natural gas should convert to coal and other abundant fuels.
Another important element of the plan is that we will endeavor to
recover and utilize waste heat to help offset our growing dependence
on critical petroleum and natural gas fuels.
This paper deals with the technology for cogeneration, particularly
emphasizing the importance of cogeneration power systems being adaptable
to a transition from petroleum and natural gas to alternate fuels by
the turn of the century. Waste heat is introduced as one of our
nearest term alternate fuels where cogeneration has proven to be one
of the major applications for its effective use. The technical
limitations of representative current technologies for cogeneration
are discussed; the need for advanced technology in cogeneration is
established. Concluding is the overall strategy for advanced cogen-
eration technology development as seen by the Department of Energy.
548
-------
INTRODUCTION
Today, only about one-half of the fossil fuel energy consumed by industry
is effectively used and only about one-third of the fuel consumed by
utilities is actually converted into electricity. Most of the unused
energy is wasted as it escapes through smokestacks or is dumped into
rivers. If it was recovered industrial and utility waste heat could
be used for further electrical generation, industrial processes, or
space heating.
Some of this waste heat can be recovered and re-used by combining
presently-independent processes and using the reject heat from one
process to provide part of the heat needed for another process, a
concept known as thermal cascading. One of the most common methods of
recovering and effectively using waste heat in this manner is cogeneration.
In cogeneration applications, mechanical work or electricity is
simultaneously generated along with useful thermal energy. Since
waste heat is recovered and utilized as an alternate fuel, the actual
fossil fuels required to produce both the electrical and thermal needs
are often no more than required to produce the electrical energy alone.
This cogeneration capability has existed for over 50 years. In 1975 it
provided over 15 percent of the nation's electric energy; today however
it accounts for only 4 percent. But as the demand grows for heavily
used, fast diminishing, and increasing more expensive oil and natural
gas, cogeneration can become a valuable energy conserving option. For
cogeneration to become a significant form of power generation depends
on overcoming barriers due to fuel supply and price uncertainties,
technological risks, marginal economics, environmental controls, and
fear of regulation on the part of non-utility power generators.
Government involvement through DOE's cogeneration program addresses
removal of many of these impediments to reverse the decline in
cogeneration, along with a strong technology development- effort to
ultimately permit the substitution of non-critical fuels for oil and
gas in cogeneration systems.
One of the more important near term non-critical fuels in this regard
is waste heat. In the category of fossil fuels between now and 1985,
residual petroleum products can be considered less^- critical than
light distillate and natural gas. For the late 1980's we hope to see
coal-derived liquids become commercially available displacing significant
quantities of petroleum. And if we accomplish a major shift of the
industrial and utility stationary power sector to the direct use of
coal, integrated fluidized bed combustion systems and gasification combined
cycle systems can both become important options for its environmentally
acceptable use in areas previously restricted, for environmental reasons,
to oil and gas fired systems.
^•Although not necessarily much cheaper, residuals are projected to be
more readily available for power generation than distillate fuels through
1985.
549
-------
All of these fuels are of importance to cogeneration systems. This paper
attempts to place in perspective the technology requirements for cogeneration
as we see this change in the predominate fuel types between now and the
turn of the century. A major emphasis is placed on the importance of
using waste heat in the near term as an alternate fuel. This can buy
us time for solving the technological and environmental problems associated
with using coal and alternate fossil fuels derived from coal to displace
oil and gas.
COGENERATION PERMITS THE USE OF WASTE HEAT AS FUEL
As previously introduced, cogeneration is one means of using waste heat
as an alternative to additional fossil fuels. This is brought about
by recovering reject heat from a process or from the generation of
electricity and using that in lieu of supplemental fossil fuel to
supply additional energy requirements.
There are two basic applications for cogeneration; industrial cogeneration
which supplies process heat or steam along with the industry electrical
needs, and residential/commerical cogeneration supplying both the
electricity and thermal energy needs of a community.
To further distinguish the two, (Figure 1) cogeneration powerplants
supplying industrial process heat are generally located on the plant
site, or, in the case of clustered industries near the load center to
avoid thermal losses in distributing the high temperature energy to the
process. There are two common arrangements for supplying process heat,
topping systems where electricity is first generated from the combustion
source and the reject heat then supplied to the process; and bottoming
systems where process heat is first supplied by the combustion source
and electricity then generated from the reject heat.
Residential/commercial cogeneration powerplants, on the otherhand, are
predominately central station utilities. Heat rejected as a byproduct
of electrical generation is recovered and distributed to customers
supplying their space heating, hot water, and cooling needs. Heat is
distributed through insulated pipes, usually not over about 25 miles
from the utility.
If we were to embrace cogeneration as a major energy option for both
the industrial and residential/commercial sectors, substantial
energy savings could be achieved between now and the turn of the century
as shown in Table I. By effectively using waste heat as an alternate
fuel in lieu of addition fossil fuel, we could accumulate up to 56
quadrillion BTU's of energy savings by the year 2000. As shown, the
annual savings could amount to over 1.5 quads by 1985 and almost 6.5
quads in 2000. This equates to about three quarters of a million
barrels of oil per day in 1985 and over 3 million barrels per day in
2000. Savings of this magnitude are significant when one considers that
our daily oil import level now in 1978 is around 8 million barrels per
day.
550
-------
LIMITS OF CURRENT TECHNOLOGY
There are three principal technologies currently available for cogeneration.
These are illustrated in Figure 2 and include steam systems for both topping
and bottoming, gas turbine topping systems with waste heat boilers, and
diesel topping with waste heat recovery.
Because it is externally fired, extraction or non-condensing steam
turbines are the only technology available today for cogeneration which
has multifuel capability to include coal and a variety of alternate
fuels such as industrial wastes. The extraction steam turbine is most
commonly configured for the topping of process steam. In this arrangement,
a fossil fuel-fired boiler is used to generate high pressure steam which
is then expanded through a steam turbine-generator set to produce
electricity. Low pressure steam is then extracted at a suitable
state point in the expansion step for use in the industrial process.
Considering limitations for steam topping applications, most industrial
plants use cheaper, easier installed packaged boilers rather than
field installed boilers. Packaged boilers are not ideally suited for
steam turbine topping because the generated pressure and temperature
are generally too low; they also cannot be economically converted to
burn coal.
From the standpoint of heat recovery applications, steam bottoming
systems in the power ranges required for cogeneration are also not
an optimum means of recovering low temperature heat (400°-1000°F).
Steam turbines, below 6000 hp, are generally less advanced than larger
multi-megawatt turbines. Their efficiencies are lower and they cost
more per kilowatt than the larger steam turbines. Also at these low
temperatures steam pressures are low, requiring costly multi-stage
turbines to extract work at high efficiency levels. Additionally,
because of the large heat of vaporization the phase change of water
into steam does not permit a good match with low temperature sensible
heat sources, which tends to lower the efficiency of the heat recovery
system.
Gas turbine and diesel topping systems offer two distinct advantages
over steam turbines— higher efficiency and greater electrical-to-thermal
output. Electrical generation efficiency of the prime mover is very
important in achieving maximum energy savings through cogeneration
as illustrated in Figure 3. Since basic efficiencies reach approximately
35 percent for diesels and almost 30 percent for gas turbines, overall
energy savings can be up to nearly three times greater than those
savings for typically 20 percent efficient extraction steam turbines
used in cogeneration. In addition, topping of process steam or process
heat with more efficient gas turbine or diesel prime movers provides
higher electrical-to-steam ratios as shown in Table II. Electrical
generation per unit of process heat is four to eight times greater for
gas turbines and diesels, respectively, than steam turbines. Moreover,
equipment is available in a range of sizes from 100 KW to over 40 MW
for diesels and up to 75-100 MW for gas turbines.
551
-------
The fundamental problem in basing advanced cogeneration systems on a mix
of today's higher efficiency diesels and gas turbines is their requirement
for high quality distillate petroleum and natural gas fuels. These are
in critically short supply, and while encouraging limited diesel and gas
turbine based cogeneration will promote more efficient use of these fuels,
a more significant shift to such systems could actually aggravate the
shortage of these fuels and defeat the national thrust toward utilizing
coal and coal-derived fuels.
ROLE OF ADVANCED TECHNOLOGY
It is our future fuels for cogeneration along with environmental con-
siderations which establish a need for advanced technology. Our
national program in advanced cogeneration technology must provide us
with three important capabilities.
The first is achievable in the early 1980's. That is to shift this
critical petroleum based fuel burning capacity in the industrial and
utility sector over to the environmentally acceptable use of lower
grade, more available, residual petroluem fuels. This will have an
immediate effective conservation impact by lessening the competition
for higher quality fuels thus making them more available for the
transportation and home heating sector where the concept of low grade
fuel switching is less feasible. Moreover, since economically
competitive future coal-derived fuels will be of a much lower quality
than todays distillate, it will be a much easier step for stationary
power equipment to adapt from residuals to coal-derived liquids than
from clean distillate to coal-derived liquids.
The second important capability is that of being able to convert this
low grade oil burning capacity to coal-derived liquid fuels when they
become commercialized in the late 1980's.
Thirdly, and a longer term effort, is to develop new environmentally
acceptable means of utilizing a maximum range U. S. coal directly in
cogeneration applications which would otherwise be restricted to oil
and gas fuels due for emissions regulations.
COGE.NERATION TECHNOLOGY DEVELOPMENT STRATEGY
DOE's cogeneration technology development strategy has been developed
around the evolution in fuels as seen between now and the turn of the
century. The overall strategy is illustrated in Figure 4.
Two generations of technology are anticipated by then in order to promote
a smooth transition from todays predominately oil and gas fueled equipment
to future systems using domestic coal fuel resources.
The second generation technologies shown are intended to help us make
the transition to alternate fuels with maximum expediency. The importance
of the second generation technologies to a smooth evolutionary fuel
552
-------
transition strategy is evident. As shown our nearest term alternate fuel
option is waste heat from various sources including power generation.
This is also the most readily adaptable since environmental impacts
are minimal. The major technical requirements are for cost effective
heat recovery and utilization technology that could be commercialized
by the early 1980's.
*
More difficulty, and requiring longer time is the adaptation of stationary
heat engines such as diesels and gas turbines to alternate, predominately
lower grade fuels from coal. This transition is seen on Figure 4 via
the development of engine technology for the environmentally acceptable
use of lower grade residual petroleum. Such a step permits definition
of minimum acceptable fuel specifications based on best available new
engine technology. These specifications would be used by the fuel
processing industry to tailor acceptable synthetic liquid coal fuel
products for displacing petroleum fuel in the second generation
stationary powerplants.
It should be emphasized that the second generation engine technologies
are not necessarily optimized for coal-derived fuels. Rather they are
proven on low grade petroleum fuel to the extent that they can be
adapted to coal fuel products that are essentially no worse than
residual petroleum. This is an important first step since it can permit
the use of coal liquids which are minimally processed and thus substantially
more economical than a fully refined coal fuel.
The optimum economical use of coal is dependent on the successful
development of third generation technologies along with the synergistic
benefits of the second generation technologies, and the experience
gained during the initial introduction of coal fuels in that equipment.
That experience can substantially accelerate the commercial availability
of technology for the general use of a wide range of coal in its
most economic forms for a particular applications - either directly,
gasified, or as raw non-processed liquids. The third generation tech-
nologies, designed and developed specifically to use coal and coal-
derived fuels in the most efficient, environmentally acceptable, and
cost effective manner could, with knowledge gained from second generation
technologies begin market entry before 1995.
To evaluate the potential of various technology options shown, the
Department of Energy is identifying opportunities for advanced cogen-
eration in six of the more energy-intensive industries in the United
States (paper and pulp, chemical, petroleum refining, steel, textiles,
and food and kindred products). They currently consume approximately
70 percent of the nation's industrial-thermal energy.
This effort, the Cogeneration Technology Alternatives Study (CTAS)
was initiated in late 1977 and is scheduled for completion in 1979.
During the first year, industrial-process energy requirements are
being defined. Concurrently, the characteristics of advanced
cogeneration powerplants such as those based on open and closed
gas turbines, diesel, Stirling engines, steam turbine, organic
Rankine cycles, and fuel cells will be evaluated along with ancillary
553
-------
components such as heat pumps, and energy storage to determine optimal
configurations. The powerplants operational characteristics will be
matched with the industrial energy requirements in future months to
assess the technical impact of these advanced systems on industrial
cogeneration installations in the 1985 through 2000 time frame.
A similar but smaller effort is underway to define the most appropriate
advanced technology options for total energy system applications in
the residential/commercial sector. This is known as the Total Energy
Technology Assessment Study (TETAS). It was begun in early 1978 and
is also scheduled for completion in 1979. Future support by DOE for
the development of new power systems designed specifically for cogen-
eration applications will be strongly influenced by the results of
the CTAS and TETAS studies.
SUMMARY
The magnitude of a complete transition to coal-burning capacity in
stationary power equipment servicing the residential/commercial and
industrial sectors by the year 2000 is staggering, when one considers
that it has taken the U. S. an average of 60 years for each of our
previous two fuel transitions — from wood to coal and then from coal
to oil and gas. In this regard the early use of waste heat as an
alternate fuel source can buy much needed lead time for developing
the technologies needed for the transition to alternate fossil fuels
derived from coal.
DOE's heat recovery component technology program (Figure 5) is
structured to provide that capability. It addresses heat recovery
in two categories, low grade heat which is below about 200°F, and
high grade heat which is in excess of 200°F. The third area of
research is heat exchanger technology. The heat recovery program is
discussed in greater detail in a separate paper at this conference.
A clarification of its relation to the overall cogeneration technology
development effort has been presented here, along with the overall
importance of heat recovery in a fuel transition strategy.
554
-------
en
en
en
Table I
Estimate Cogeneration Energy Savings
Total
Estimated Savings (Quads)
Systems
Integrated and total energy
Industrial park energy
Industrial process heat
General waste heat recovery
Market Application
Residential/commercial
Industrial/commercial
Industrial
All sectors
Annual
1985
0.38
0.16
0.57
0.47
Annual
2000
1.52
0.85
2.64
1.44
Cummulative
2000
12.0
7.0
24.0
13.0
1.58
6.45
56.0
Prime Mover
Table II
Electric-to-Thermal Characteristics of Cogeneration Powerplants
Electric Power/Process Steam
(KWH 1010 BTU)
o Steam Turbines
o Gas Turbines
o Diesels (with recovery of heat from
jacket coolant)
o Diesels (without recovery of heat
from jacket coolant)
50
200
300
400
-------
Figure 1
Typical Cogeneration Powerplant Arrangements
Industrial Cogeneration — Process Heat and Steam
ElactrlcltyO . -- '!-, |- ELctricHvO
CT 1.- -- tra D ' •. H
Plan!
^\ I S»».m""n/'/S "~~ r^V^X
' ^ • '-.jf E /V /V /
Fu»l|> | Bollst
Fuol|} i Boll..
F«nc« IMa--t •:•!:: (
Fronl End/Topping
Cog«nemtion
IPrimnrv tmphasisl
Back End/Bol1omtng
Coqeneration
Residential/Commercial Cogeneration - Space Heating and Cooling
_ Electricity [>
District Heating
Energy Path,,
Community
Systems,Etc
Extrnttlon
Stenm Topping
Cycli.
Staant Rankina
Bottoming Cycla
Figure 2
Technologies Currently Available for Cogeneration
Diesel Engine
Topping Cycle
cn
Ul
Figure 3
Maximum Fuel Savings in Cogeneration Occurs
with High Efficiency Prime Movers
35,
'Diesel
Turbine
urbine
10 20 30 40
Prime Mover Electrical Efficiency
-------
Figure 4
Advanced Degeneration Technology Development Strategy
Current Technologies
• Steam Turbine*
• Gas T urbmes
• Dwiels
2nd Generation Technologies
• Heavy Fuel Gas 1 urbme
• Heevy Fuel Diesel
• Phoiphoric Acio Fuel Cell
• Organic Rank inf Cycles
• High EM Small Sieam
• Low Temp Hprii Recovery
Thermal Activated Cooling
Heal Pumps
Low Temp Heal Engines
Improved Heal Each
Predominate Fuels
• Near Term
Waste Heal
Drstillaie Petroleum
Natural Gas
• Mid Term
Coal Derived Liquids
Longer Term
Coal Derived Liquids
- Coal Derived G200°F)
• Rankine Bottoming
• Thermionic Converter Topping
Heat Exchanger Technology
• Fouling/Corrosion/Vibration
• Improved Mat'ls/Temp Capability
• Reduced Cost
557
-------
COGENERATION: THE POTENTIAL AND THE REALITY
IN A MIDWESTERN UTILITY SERVICE AREA
D. M. Stipanuk, Cornell University, Ithaca, NY
W. J. Hellen, Wisconsin Electric Power, Milwaukee, WI
ABSTRACT
A study was performed by Wisconsin Electric Power of the magnitude and
location of major boiler fuel users within its service territory. Cost
estimates were obtained for coal fired cogeneration units and for steam
distribution systems. Steam and electric costs from cogeneration units
serving three (3) areas of significant steam load were calculated. The
distance between steam loads and the annual steam load factor were shown
to significantly affect the cost of steam. Electricity costs were
relatively unaffected by the levels of steam sales made in the areas
studied.
INTRODUCTION
Wisconsin Electric Power Company (WE) is the largest electric utility
in the state of Wisconsin. The WE service territory consists of a band
of land approximately 40-50 miles wide along Lake Michigan from the
Illinois State Line north about 100 miles which contains the city of
Milwaukee. In addition there are small areas served in east central
and northern Wisconsin and in the upper penisula of Michigan. The
total population served by WE is in excess of 2 million-
The spring of 1977, WE began a study to assess the potential market for
steam sales, the cost and operating characteristics of cogeneration units
and the costs of steam distribution piping. This report provides a
summary of part of this study.
THE EXISTING WISCONSIN ELECTRIC COGENERATION ACTIVITIES
The existing WE cogeneration system provides district heating service
for about 600 customers in an area of approximately two square miles in
downtown Milwaukee. Historically the system was divided by the Milwaukee
River into a Eastside and a Westside system. This division still exists
for the low pressure system. The customers served are primarily offices
and public buildings. The customers are supplied through a network of
high pressure (180 psig) and low pressure (15 psig) steam mains. The
largest high pressure steam mains are 30" (O.D.) and the smallest are 4".
The smallest low pressure mains consistent with current design philosophies
are 6". No condensate is returned from customers.
The primary steam supply point for the district heating system is the
Valley. Power Plant. The Valley plant is composed of two (2) turbines
with two (2) boilers for each turbine. This coal fired cogeneration
558
-------
plant is capable of a continuous steam sendout in excess of 1,000,000
Ibs/hr and up to 1,700,000 Ibs/hr for limited periods. The continuous
sendout at maximum capability is presently limited by the makeup water
(demineralizer) capacity.
The two (2) turbines in the Valley plant are controlled extraction units
rated at 140 Mw each with no steam extraction. With maximum extraction
of 850,000 Ibs/hr, each unit drops to 65 Mw resulting in the plant
capacity dropping from 280 Mw to 130 Mw.
Additional steam for backup and peaking is available from the East Wells
and Commerce Street Plants. The East Wells Plant is coal fired and
can continuously supply 132,000 Ibs/hr east of the Milwaukee River.
Commerce Street is an oil fired unit with a continuous capacity of approxi-
matley 200,000 Ibs of steam per hour to the westside. On a first
contingency basis the system can supply a peak demand at 1,600,000 Ibs/hr.
The existing WE district heating system experiences large variations in
steam sales from summer to winter. TABLE 1 shows the monthly variations
in steam sendout averaged for the years of 1971 through 1976. Steam
sales vary even more significantly than does steam sendout since the
system losses represent a large percent of the summer sendout and a
much smaller percent of the winter sendout.
In two (2) of the past seven (7) years, the winter electrical peak for
Wisconsin Electric and the district heating system peak occurred on the
same day. In a third year, a single day separated the two peaks.
COGENERATION UNITS - DESIGN AND COST
WE retained Gilbert Associates, Inc. (Gilbert), a division of Gilbert/
Commonwealth Engineers and Consultants, to supply capital cost estimates
and unit heat rate data for coal fired steam cogeneration units in the
600, 300, 200, 100 and 80 Mw sizes. These cogeneration units were to
have steam export flows of 250,000 and 500,000 Ibs/hr at 250 psig.
Capital cost and heat rate data for coal fired steam only units providing
250,000 and 500,000 Ibs/hr were also provided.
Unit Design
The unit designs for 600, 300, and 200 Mw units indicated a capability
of extracting at most 25% of their cold reheat steam flow. Cold reheat
extraction required throttling to achieve the desired supply conditions
of 250 psig. The 200 Mw unit cannot supply the full 500,000 Ib/hr flow
since 25% of its flow is approximately 350,000 Ib/hr. All three (3) of
these units (600, 300 and 200 Mw) experience a decrease in their
electrical capability of approximately 1 Mw for each 10,000 Ib/hr of
steam extracted.
559
-------
Gilbert also provided designs for an alternate 200 Mw unit and for
100 and 80 Mw units with controlled extraction turbines. For the 100
and 80 Mw units the electrical output decreased as the amount of steam
exported increased. The alternate 200 Mw unit was additionally modified
with an oversized high pressure turbine and increased boiler capacity to
supply the additional heat input requirements. To improve reliability
of steam service the single boiler was changed to two half size boilers.
These modifications allowed the 200 Mw unit to maintain its 200 Mw rating
at 500,000 Ib/hr steam extraction.
UNIT COST ALLOCATIONS
In the development of capital costs and their allocation a two (2) unit
site concept was used. This provided for backup steam supply since only
one unit was designated to provide steam. Since unit electrical capability
declined with increasing steam sendout, a capital cost demand charge was
included to provide the required makeup electrical power. The additional
cost to maintain fully rated output was approximately equal to the
incremental cost per Mw for units from 300 to 600 Mw.
Since a cogeneration uni.t requires less capital investment than separate
steam only and electric only units it was necessary to devise a means
of allocating the capital costs of the cogeneration unit. The
allocation method used assumed the savings in captial cost for cogeneration
construction (including the additional cost to maintain rated output)
versus separate steam and electric facilities would be shared equally
between steam and electric customers. Costs of the steam only and
electric only plants were reduced by the capital savings calculated and
these reduced costs were used as the capital costs assigned to steam
and electric. The results of these allocations as a function of unit
size are shown on TABLE 2.
Energy cost allocations assumed no subsidization of steam cost would
be made by the electric customers. For the units studied, this means
the amount of fuel charged to electric generation in a cogeneration
case will be no more than it would be in an electric only case. The
amount of fuel saved in a cogeneration case is allocated equally between
the steam and electric units resulting in reduced fuel inputs for each.
TABLE 3 illustrates this procedure for the 600 Mw unit.
FIGURE 1 combines the capital and fuel cost allocations discussed previously
for the cost of 500,000 Ibs/hr of steam exported from cogeneration units,
a steam only plant providing 500,000 Ibs/hr and for electricity
from cogeneration and electric only plants. Annual capacity factors of
70% were assumed for both steam and electric. Significant economies
of scale are shown to exist for electric power.
560
-------
STEAM DISTRIBUTION SYSTEMS - DESIGN AND COST
Gilbert provided WE with design and cost data for steam distribution
systems with condensate return. Supply lines were of carbon steel'and
return lines were stainless steel due to the corrosive nature of
condensate. Drains, traps, vents, expansion joints and valves were
included as needed. From 3" to 4" of insulation was provided on supply
lines and 2" on return lines. Assumed supply conditions for the steam
are 460° F and 250 psig with returned condensate at 120° F.
Costs were developed for both above and below ground installation. The
above ground installation would most likely be used in rural areas or
for installations made in industrial parks. Below ground installations
would be required in urban areas and for the crossing of roads and other
obstacles. Below ground installations were placed in prefabricated,
reinforced concrete trench boxes with top slabs of reinforced concrete
with manhole access for customer service laterals or for required access
to valves etc. Below ground installations with single supply pipe and
with dual supply were considered. The reliability required by process
steam customers requires the ability to remove a line from service
for maintenance and still maintain service. The geographical limitations
of many areas do not lend themselves to a loop type of distribution
system. The placing of two (2) supply mains in a single trench box
provides the required reliability at minimum cost.
Two (2) types of piping runs were assumed for the study: a straight pipe
run intended for use as a main supply line or for branch runs of a
relatively simple nature; and a complex run was intended for urban
areas where obstacles would be present to straight line piping runs
and where service laterals to customers were more frequent. TABLE 4
shows installed costs for various sizes of steam distribution piping
for the types of piping runs considered.
IDENTIFYING THE POTENTIAL MARKET FOR STEAM
The approach taken by WE to assess the potential market for steam was
based on the premise that the potential market consisted of those
institutions and industries already established in the WE service
territory and presently having a significant demand for process or
space heating steam. No considerations were given to speculative
situations such as the location of new, large steam using facilities
(i.e. refineries or paper mills) within the WE service territory.
To identify institutions and industries currently using steam, data
from several sources was used. The Department of Natural Resources
CbNR) of the State of Wisconsin annually complies an emmissions
survey which lists all major boiler fuel users, their fuel consumption
and boiler sizes. The DNR survey does not cover governmental units
using boiler fuel. The DNR survey was checked against natural gas
561
-------
usage data from the two natural gas utilities serving the WE territory.
Data from a period prior to widespread gas curtailments was used.
In this way the report was able to include large governmental boilers
and a check was available to insure the DNR data was accurate and
complete. Several small errors were noted requiring direct contact
with fuel users to verify the correct fuel usage levels.
TABLE 5 contains a summary of the survey data on a county by county
basis for the WE service area. A lower end cut off point of 1 x 1011
BTU (eq) per year of boiler fuel was used to reduce the task of dealing
with many small users. The lower level fuel cut off chosen corresponds
to an average hourly usage of about 10,000 Ibs/hr. Two (2) counties,
Milwaukee and Winnebago, account for almost 70% of the total fuel
consumed in industrial and institutional boilers.
Following the compilation of the data for TABLE 5 a more detailed review
was made of the potential steam users in the following counties.
Fond du Lac Milwaukee Racine Winnebago
Kenosha Outagamie Sheboygan
This review required the use of county maps on which the location of
major users was plotted. Likely groupings of potential steam customers
were explored to determine if users were in close enough proximity
to one another to warrant a distribution network. No consideration
was given to determining sites for the cogeneration plant itself.
The. detailed review using county maps resulted in the selection of
three (3) areas as good potential markets for steam sales. This
selection was primarily based on the proximity of the steam users
to each other, the size of the estimated steam loads and, to a lesser
degree, the nature of the land near the potential users. In one
instance, the area to be served was more rural than others and it was
believed the reduced cost of above ground distribution would justify
the slightly longer pipe runs.
Since the boiler fuel used by the potential steam users was consumed
for both process and space heating needs, a computer program was devised
to estimate the peak steam load for the potential steam users. The
program utilized data gather from operating experience with WE's own
district heating system and degree day data for the WE service
territory.
The peak steam loads were used to estimate pressure drops in piping runs.
Pressure drop calculations were used to size the steam lines used for
the distribution system. The peak load information could also be used
to determine the smallest cogeneration size which could provide the
required steam output. This was beyond the scope of this study.
562
-------
The three (3) areas selected as providing good potential markets for
steam have been designated as WE-S, WE-C and WE-N to maintain the
anonymity of the companies included. These areas are shown on TABLES
6, 7, and 8. Information is also contained on these TABLES regarding
peak loads, size of pipes and length of pipe runs.
COSTS OF STEAM AND ELECTRICITY FROM COGENERATION UNITS SERVING
POTENTIAL MARKET AREAS
Following the selection of potential market areas WE-S, WE-C and WE-N
and the selection and costing of pipe runs required, it was necessary
to use the previously established allocation method for the capital,
fuel and operating expenses of the cogeneration unit.
The cogeneration unit used in the cost calculations for each potential
market area was the 600 Mw unit located in pairs on site. Selection of
the same unit size allowed WE to analyze only the effect of steam load
variations and distribution system differences on costs of delivered
steam. The effect of unit size had been analyzed earlier and it was
shown that distinct economies of scale existed as unit size increased.
TABLE 9 indicates the annual cost of energy associated with service to
the three (3) areas selected. The cost of electricity did not vary
significantly between any of the potential steam market areas. The
cost of steam varied considerably from area WE-S to areas WE-C and WE-N.
WE-S has a steam cost of $14.98 per 100 Ibs while WE-C and WE-N were
$5.93 and $5.07 per 1000 Ibs respectively. The reasons for the large
variations in steam cost are found by referring to the cost data of
TABLE 9 and to TABLES 6, 7 and 8 where the system peak loads, annual
steam usage and distribution system sizes are shown.
Area WE-S has a peak steam load of about 300,000 Ibs/hr with an annual
steam load factor of only 31%. The loads in area WE-S are predominantly
space heating even though the steam is supplied to several large
manufacturing plants. In addition, large distances are involved in the
distribution system. The identified customers represent the largest
steam users in the area. The area near these loads is somewhat rural
and land may be available for power plant siting. However, the steam
loads do not appear to be favorable for consideration of a cogeneration
installation.
Area WE-C has a peak steam load of 494,000 Ibs/hr with a relatively high
annual load factor of 53%. The loads are a mixture of process and space
heating. The identified customers are located in the heart of the down-
town Milwaukee area. Distribution distances are relatively small but
uncertainly associated with the congested area below urban streets may
result in higher than estimated distribution system expenses.
Area WE-N has a peak steam load of 811,000 Ibs/hr with an extremely high
annual load factor of 67%. The loads are primarily process loads with
563
-------
a small amount of space heating. The identified customers are located
along a river in a fairly dense urban setting. Distribution costs
may be higher than those estimated since problems may exist in the
area due to below street congestion and additional costs for crossing
the river.
The effect of load factor and distribution system costs on the annual
cost of steam for the three (3) potential markets is illustrated by
the cost information below. Operating costs per 103 Ib of steam are
essentially the same in each area.
WE-S WE-C WE-N
Cogeneration Unit Annual
Fixed Charges ($/103lbs of steam) 7.37 3.59 2.41
Distribution system Annual
Fixed Charges ($/103lbs of steam) 5.82 .56 .94
In summary, steam costs were highly sensitive to the distance between
loads due to the expense of distribution systems. Areas with high load
factors absorbed the distribution system costs and the cogeneration
unit fixed charges in large quantities of steam resulting in lower costs
per Ib of steam. Dense service areas with high load factors (i.e.
process loads) are most likely to result in lower prices for steam
relative to areas with poor load factors such as result from space
heating loads.
564
-------
TABLE 1
EXISTING WE DISTRICT HEATING SYSTEM
MONTHLY AVERAGE STEAM SENDOUT
1971 - 1976
Jan.
Feb.
Mar.
Apr.
May.
June
July
Aug.
Sep.
Oct.
Hov.
Pec.
TOTAL
Unit
Size
(Mw)
600
300
200
100
80
MONTHLY AVG. STEAM SENDOUT
1st. Unit
355,149
217.401
152,062
92,470
78,085
106LBS
430
407
362
298
197
98
56
58
60
111
214
337
2628
TABLE 2
COGENERATION UNIT
COMPLETION COSTS W
Additional 1st Unit Cost
To Maintain Rated MH(2)
23,814
23,475
23,631
23,631
23,631
X of Total
16.3
15.5
13.8
11.3
7.5
3.7
2.2
2.2
2.3
4.2
8.2
12.8
100.00
2nd Unit
331,773
206.264
140,792
84.678
72,457
Sulphur Coal with S02 removal - 103 June 1977 dollar*.
500,000 lbi/hr steam export.
565
-------
TABLE 3
ENERGY ALLOCATION FOR 600 MW UNIT
EXPORTING 500,000 LB STEAM PER HOUR
No Extraction Reduced Net Equivalent No
Heat Rate x Electric Output = Extraction Heat
(BUT/Net KW) at 500,000 Ib/hr To Electric
Extraction (Kw) MBTU/HR
9891 x 546,904 - 5409
Actual Unit Equivalent No Additional Heat
Heat Input - Extraction Heat = Input Due to
MBTU/HR To Electric Steam
MBTU/HR BTU/LB Steam
5944 - 5409 = 1068
Steam Only Additional Heat Heat Savings
Unit Heat Rate - Input Due to Steam = For Cogeneration
(BUT/LB Steam) (BTU/LB Steam) (BTU/# Steam)
1404 - 1068 (1) - 366
Credit To 366 BTU/LB(1) 168 BTU/LB
Electric Per - 2 -
LB Steam
Revised Steam - 1068 + 168 ' - 1236 BTU/LB
Heat
Allocation (BTU/LB)
«
See calculation one (1) line above for derivation.
566
-------
TABLE 4
STEAM DISTRIBUTION SYSTEM COSTS
SINGLE MAIN
Size
Steam
Supply
30"
20"
18"
12"
10"
6"
Rural
Straight
804
591
568
407
351
278
Below Grade
Urban
Straight
889
654
632
469
413
334
Urban
Complex
1129
786
745
539
468
365
Above
Rural
Straight
377
275
258
181
149
119
Grade
Rural
Complex
612
404
367
247
200
146
DUAL MAIN
t
30"
20"
18" ,
12"
10"
6"
Below
Rural
Straight
1562
1148
1089
776
670
531
Ground
Urban
Straight
1629
1222
1171
849
740
593
Urban
Complex
2108
1486
1397
989
850
655
Excluding: 1) Land and Land Rights
2) Rerouting of Traffic
3) Interruption of Other Services
4) Permits and Fees
5) Drainage from remote locations
Note: All cost $/ft. in June 1977 dollars
567
-------
TABLE 5
SUMMARY OF
MAJOR BOILER FUEL USERS
BY COUNTY
(1)
County
Fond du Lac
Dodge
Jefferson
Kenosha
Milwaukee
Outagamie
Ozaukee
Racine
Sheboygan
Shawano
Walworth
Washington
Waukesha
Waupaca
Winnebago
Total
No. Total
of Boiler Capacity
Firms (106BTU/hr)
1 135
1 *
1 1.3
5 1066
48 7155
13 3109
1 *
10 1000
3 1127
2 *
1 300
4 70
4 68
2 *
14 2020
110 16051
Total
Space Heating
Fuel in 1975
(1012 BTU)
.02
*
-0-
1.83
8.24
1.86
*
2.60
.12
*
.34
.05
.34
*
.64*
16.04*
Total Fuel Consumed
in 1975 (1012 BTU)
2.62
.11
.18
2.44
20.47
2.44
.10
3.87
2.00
.55-
.40
.37
.58
.29
10.85
47.27
(2)
(1) All firms using in excess of 1 x 1011 BTU in 1975 excluding
electric utilities
(2) Combination of DNR data and information from gas companies
*Data incomplete due to exclusion of certain firms from DNR
data
568
-------
TABLE 6
STEAM SERVICE AREA WE-S
(Supply)
Drawing Not to Scale
Distribution System
From
(Supply)
1
8
8
3
3
4
5
Length
To (Ft.)
8 0
2 500
3 11,600
7 9,000
4 10,700
5 1,780
6 6,500
Diameter
(In.)
N/A
12
20
10
18
10
18
Peak Load
(103 #/hr)
307
109
198
44
154
43
111
Est. Annual Steam Usage
Customer
A
B
C
D
Total
Usage
289
119
325
108
841
TABLE 7
STEAM SERVICE AREA WE-C
I 1 (Supply)
Drawing Not To Scale
Distribution System
Length Diameter
frsa To (Ft.) (In.)
(Supply)
2
3
4
0
0
5700
N/A
N/A
18
Eat.
Peak Load
494
205
289
Est. Annual Steam
Usage
Usage
1,001
1.307
2,308
Customer
A
B
Total
569
-------
TABLE 8
STEAM SERVICE AREA WE-N
Drawing Not To Scale
Distribution System
Est.
From To Length Diameter Peak Load
(ft.) (in.) (103#/hr)
811
102
709
498
120
378
72
246
11
110
60
Est. Annual Steam Usage
106 f
Usage
Supply 1
1 2
1 3
3 4
4 5
4 6
4 7
7 10
7 9
10 11
9 8
7 12
0
1700
3800
1900
500
500
4400
1000
1000
1000
1000
4000
N/A
10
30
20
6-
6-
30
10
18
6
10
10
Customer
A
B
C
D.
E
F
G
H
I
TOTAL
515
1406
588
570
-------
Ui
4-1
CO
O
O
00
Vi
-
32--
31--
30-
29--
28"
27--
264-
FIGURE 1
ANNUAL COST COMPARISON
AT SITE BOUNDARY
SEPARATE VS. COGENERATION FACILITIES
70% ANNUAL CAPACITY FACTOR
HIGH SULFUR COAL WITH S02 REMOVAL
JUNE 1977 $
Separate Electric
Cogeneration Electric
0-0-
Separate Steam
0
-D
Cogeneration Steam•
-T- 20
. . 19
-18
. 17
, . 16
.- 15
. - 14
- 13
.- 12
- 11
•- 10
'- 9
• - 8
- - 7
•- 6
• - 5
-U 4
o
o
o
CO
o
o
0)
4J
cn
80 100
200
300
400
500
600
-------
ALTERNATIVE APPROACHES IN INDUSTRIAL
COGENERATION SYSTEMS
J. C. Solt
Solar Turbines International
An International Harvester Group
San Diego, California
ABSTRACT
This paper traces the development of cogeneration systems in
industry, and discusses some early applications. The effect of
changing markets and economic conditions is evaluated and
specific examples are presented to illustrate the increasingly
favorable climate for cogeneration system applications in
today's energy conscious society.
BACKGROUND
In the early 1960's, many people became aware of the fact that
it was possible to generate electric power at the site where it
was to be used and also to produce process heat from the same
system. This allowed generation of electric power less expen-
sively than at a central generating plant, and the savings
could be applied against the capital cost of the equipment. The
common generic term for this type of system was "total energy."
With few exceptions, early cogeneration or "total energy"
systems had no interconnect with electric utilities. Their real
challenge was to maintain the appropriate relationship between
demand for electricity and heat. But isolation from electric
utilities generally meant that equipment was run to meet
electrical demand with no regard for heat demand. Excess heat
was simply wasted, or any short-fall in heat production was
made good by using auxiliary equipment. In either case, eco-
nomic and thermal efficiency was impaired.
As a result, early cogeneration attempts usually proved dis-
appointing. Industrial process installations were generally
more successful than commercial installations because indus-
trial users had higher annual utilization rates for both
electricity and heat, and greater maintenance capabilities.
Commercial users frequently based economic calculations on
ideal conditions which were not present year-round.
Another factor limiting early attempts at cogeneration was
that, without a utility tiein, individual users had to come up
with their own solutions to peaking and standby problems. A
user with a peak electrical use twice that of minimum use would
have to buy three units: one to generate minimum load, one to
572
-------
supply peak power, and a third as a standby. In this example
average use rates could be less than 50 percent, substantially
diluting any savings offered by onsite cogeneration.
One solution would have been to retain a utility tie and
generate only a portion of the electrical demand onsite with
the utility supplying the shortfall. The system could then be
operated to meet heat demand only. Even with the generator
operating at less than 100 percent, more attractive efficiency
figures would be obtained because of the more complete exhaust
heat utilization. However, in the past, the average electrical
utility regarded onsite generation as competition and was
unlikely to agree to participation in a "share-the-load" type
arrangement.
Concern about fuel availability and cost in the seventies has
created new possibilities for successful application of Co-
generation. It should be obvious, however, that cooperation
between electrical utilities and process heat users is needed
if the potential represented by cogeneration is to be realized.
ALTERNATIVES FOR UTILITIES
There are basically two different ways in which an electrical
utility and an industrial user can participate in a cogen-
eration activity. Each alternative has advantages and dis-
advantages and it is important that both be considered for
each individual application to be certain that the utility
and the industrial customer take advantage of the appropriate
arrangement.
The first alternative is for the electrical utility to own and
operate the cogeneration facility. The electrical generation
equipment is then tied directly into the utility grid and the
equipment can be operated to satisfy the heat requirements of
the industrial customer. The industrial customer then continues
to buy electrical power in the usual fashion, but he also
purchases the process heat (usually in the form of steam) from
the utility. The industrial customer typically will be able to
purchase this process heat at a lower cost than if he produced
it himself, and he does not bear the capital cost of the
boilers.
The second alternative is for the electrical utility to furnish
an intertie to a system owned and operated by the industrial
heat user. As previously stated, this intertie would improve
both the capital cost to the industrial user and allow him to
operate at an optimum cycle efficiency. The electrical utility
benefits by getting additional generating capacity without the
problems associated with raising capital, siting the system,
and overcoming environmental objections.
573
-------
SOLAR'S EXPERIENCE
Solar Turbines International has been furnishing cogeneration
systems to the process industry for more than 17 years. There
are over 280 Solar turbines in this type of service and they
have accumulated over eight million operating hours. Some of
these packages have operated in excess of 100,000 hours. Our
largest installation includes 12 turbine generator sets at a
single location. Another single-site installation has two
generator sets, two natural gas compressor packages, and three
mechanical drive refrigeration packages.
HOW A TURBINE WORKS
A gas turbine engine consists of a compressor, a combustor, and
an expansion turbine (Fig. 1). Ambient air is drawn in through
the compressor and flows steadily into the combustor. Fuel is
fired in the combustor and the resulting, high-energy hot gas
passes through the expansion turbine. The energy extracted
through the turbine section is used to drive the compressor and
the load. \
Burning is continuous in the combustor, so temperatures in the
turbine section are maintained by controlling the fuel-to-air
ratio. Most industrial gas turbines use from three to four
times as much air as is required to combust the fuel. Conse-
quently, the exhaust stream contains 15 to 20 percent oxygen,
so it is possible to burn additional fuel in the exhaust stream
to increase the amount of thermal energy available. Since the
exhaust is clean and dry it can also be used directly as hot
air for drying or curing.
Solar manufactures three industrial turbine engines which can
be used in cogeneration topping cycles (Fig. 2). These engines
are available in complete generator, compressor or mechanical
drive packages capable of driving any type of customer speci-
fied equipment.
EXHAUST HEAT UTILIZATION
There are three distinct methods of utilizing gas turbine
exhaust heat. Fuel burned in almost any industrial process can
be credited totally or partially by one of these methods:
• The most common procedure is to pass the exhaust flow
through a heat exchanger (or boiler) and transfer its
heat to a process fluid or gas such as HgO, air, or oil
(Fig. 3). Since the exhaust gas contains over 16 percent
oxygen, additional fuel can be fired in the exhaust flow
to achieve higher initial temperatures or increased
energy as required by the particular process. A number of
574
-------
manufacturers offer these types of heat exchangers and
boilers as standard product lines.
• A second common method is to use the exhaust flow, which
is relatively clean and dry, as a direct-heating and/or
drying medium (Fig. 4). The exhaust flow can be diluted
with ambient air to achieve any lower temperature and
increased volume flow desired. If the process back pres-
sure is low, this can be done with a simple inductor. At
higher back pressures, a fan is required to provide the
dilution air. As in the case above, additional fuel can
be burned in the exhaust prior to process use to match
any temperature/energy requirement.
• A third method, which is frequently utilized in the oil
and gas industry, is to use the exhaust flow as highly
preheated combustion air in devices such as boilers,
space heaters, oil heaters, and hot gas generators
(Fig. 5). This is the most efficient method of exhaust
utilization since all of the exhaust energy above ambient
can be credited to fuel saved in the combustion device.
GENERAL PERFORMANCE
The values shown in all three examples (Figs. 3, 4, and 5) are
based on the following assumptions:
Ambient conditions
Fuel
Load
Steam data:
Condensate return
Steam conditions
Boiler efficiency
and 60°F
sea level
liquid or gas
100 percent
- 200°F
- dry and saturated
- 80 percent
For altitudes other than sea level, read the correction factor
from the following table, and multiply power output, fuel
consumption, and mass flow by the factor. For part-load con-
ditions and other ambient requirements, please contact Solar.
The constants shown in these examples can be used to perform a
similar preliminary energy balance for other types of systems.
Altitude
(ft)
0
2,000
4,000
6,000
8,000
10,000
Factor 1.000
0.930 .
0.864
0.801
0.743
0.688
575
-------
EVALUATING THE BENEFITS OF A SOLAR COGENERATION SYSTEM FOR YOUR
APPLICATION
Return on investment in a cogeneration system can be determined
by considering four primary and two secondary variables. The
primary variables are:
• Utility Electrical Rate ($/kWh)
This should be a composite rate including energy charge,
demand charge, and taxes as applicable. If a declining
rate structure is in effect, it should be taken into
account. This information can be obtained from your
utility bill and a rate schedule available from the
utility company.
• Annual Utilization (%)
Like any other capital investment, return improves with
higher utilization. Maximum utilization can be achieved
by generating the required base load while in parallel
with the electrical utility, which would provide peaking
and standby power as required.
A i TT.,- • i • 4.- Annual Hours of Operation
Annual Utilization = - -,„_-. . - ; — - -
8760 hours/year
• Fuel Cost - $/Million Btu ($/mm Btu)
The net cost of onsite generated electricity will be
directly proportional to cost of fuel used. Cost is
expressed in $/million Btu as a convenient form for the
necessary calculations. It can be determined as follows:
Number 2 Diesel
Heating Value = Approximately 130,000 Btu/gal
Cost = 0.363/gal
$/mm Btu = (0.363) = $2.79/mm Btu
Natural Gas
One therm = 100,000 Btu
Cost = $0.18/therm
$/mm Btu = 10 x 0.18 = $1.80 /mm Btu
576
-------
• Installed Cost of Cogeneration System ($/kW)
Precise estimates of system installed cost must be made
on a case-by-case basis because of variations in such
factors as site conditions, local labor costs, and heat
recovery equipment used. However, for purposes of an
initial economic analysis, the following average values
of total installed cost can be used.
Saturn $480/kW
Centaur $400/kW
Mars $300/kW
The secondary variables are:
• Net Fuel Rate for Onsite Generated Electricity
Fuel rate for a generator is expressed in terms of
Btu/kWh, i.e., the number of Btu's of fuel input required
to produce one kilowatt for one hour. In order to find
the incremental fuel required to produce the electricity
above that fuel required to satisfy the process heat re-
quirements, we credit the fuel used by the gas turbine
with the fuel that would have had to be used to satisfy
the process heat requirements. (In calculating fuel
credit, one must take into account such factors as boiler
efficiency so that actual fuel required to satisfy the
process requirements is determined.) When this value of
incremental fuel is divided by the number of kilowatts
produced, the result is a quantity defined as Net Fuel
Rate (NFR). In equation form:
NFR =
/turbine] _ /fuel value of \
\ fuel / \heat recovered/
kW output
Net fuel rate will vary as a function of the type of heat
recovery system and average kW output, but typical values
will range from 4000 to 6000 Btu/kWh, which is less than
50 percent of the fuel required by the utility. For an
initial evaluation, an average value of 4500 Btu/kWh is
satisfactory.
• Maintenance Cost
Maintenance cost should be considered in calculating net
cash flow, but has a relatively minor impact. Typical
maintenance costs vary in the range of 0.18£/kWh, depend-
ing primarily on the operating cycle. Frequent starts and
stops will result in higher maintenance cost while
continuous operation will result in lower maintenance
cost.
577
-------
Application and Evaluation
The application and evaluation procedure can be best illus-
trated with a specific example.
Step
Select a generator set (or sets) which meets the desired
heat load by referring to Fig. 6, which shows the recov-
erable heat or approximate steam production capacity of
Solar's three generator set sizes.
Check that the kW rating of the unit(s) selected matches
favorably with the plant's electrical demand. The system
could be sized to provide all or only part of the plant
requirements, of either the steam or power, with other
equipment or with the utility providing the difference.
Assume that a Centaur generator set is selected to
provide part of the plant's heat and electrical require-
ments. List the four primary and two secondary economic
variables previously described:
Utility composite rate + 0.036/kWh
Annual utilization for three shifts, six days, 50
weeks =
3 x 8 x 6 x 50 _ aoa
8760 ~ S2%
Fuel cost (No. 2 Diesel) = $2.79/million Btu
Installed cost = $400/kW
Net fuel rate (average value) = 4500 Btu/kWh
Maintenance cost = $0.0018/kWh
Calculate cost of electricity generated:
Cost = fuel rate x fuel cost + maintenance
Cost = (4500) (2.79)/106 + 0-0018 = $0.0144/kWh
Calculate saving/kilowatt hour:
utility rate - cost of electricity generated
= 0.036 - 0.0144 - $0.021/kWh
Read annual saving/kilowatt installed from Fig. 7:
Savings = $150/kW installed
578
-------
(2) Calculate simple payback:
_ installed cost/kW .__ ,_. „ „
annual savings/kW = 40° - 15° = 2.7 years
(D Read after-tax discounted cash flow rate of return from
Fig. 8:
= 21%
Either rate of return or payback can be used to evaluate the
investment. This example shows a very attractive return on
investment and it is clear that the application with these
conditions would be pursued in almost any industry. Your
company's financial guidelines should indicate whether the
return for your application is practical.
CONCLUSION
Solar has used gas turbine .engines to provide site-generated
electric power in almost every conceivable application. There
are certain industries, however, that present much more favor-
able economics than others. In general, process industries are
excellent candidates for cogeneration type systems because they
have a high usage factor (three-shift operation), a high ther-
mal requirement on a 24-hour a day basis, and they normally
have a good understanding of the sophisticated mechanical
equipment described in this paper. These process industries
include:
Natural gas processing
Petrochemical and refining
Paper and pulp
Food processing
Textiles
Clay, cement, and glass
Lumber and wood products
Metals
Solar's equipment can be used for either total or partial
electrical power generation, refrigeration, and as direct drive
for mechanical equipment in almost any of these applications.
579
-------
GENERATOR
/
/
NLET .
"**"—-— 4
'
____ — •
r COMBUSTOR \
>> t
POWER TRANSFER DIAGRAM
Fig. 1. Basic Simple Cycle Gas Turbine
SATURN 800 kW
CENTAUR 2700 kW
MARS 7400 kW
Fig. 2. Solar Gas Turbine Generator Sets
Suck T«mp«r«tuK 1° PI
Steam Output lib/or)
EinBUBt Temperatun I°F)
Pud Input (million Btu/orl
Electric*! Output IkW)
Saturn
300
6425
848
13.22
BOO
49.20
6524
Centaur
300
17.886
836
39.84
2628
140.0
6685
Mara
300
33,112
781
80.52
6865
288.9
5723
Fig. 3. Unfired Turbine Exhaust Used to Produce 15 psi Steam
Elwtrk.1 Output (kW)
Air Ma*i Flow
(tbouuDda Ib/kr)
Fig. 4. Turbine Exhaust Used Directly as Hot Air Source
Suck Temperatun (°F)
SU*m Output (Ih/hr)
Fuel lopot InUlioa
Btti/hr)
Electric^ Output (kW)
Air MM* Flo*
ItbouMfld Ib/hrl
N«t Fud IUW
(Bta/kW)
Saturn
9.14
848
13.22
800
49.20
4361
Ceataur
27,27
836
39.84
2628
140-0
4783
Mat*
52.28
781
80.52
6865
288.9
4113
Saturn
300
48.050
49.2
848
13.22
800
49.20
4361
Ceol-ur
300
135.508
139
836
39.84
2628
140.0
4783
Mara
300
276.359
287
781
80.52
6865
288.9
4113
Fig. 5. Turbine Exhaust Used as Combustion Air in 200 psi Boiler
580
-------
STEAM PRESSURE (pile)
g § s i i 8
1
iATURt-
(BOOkV
\T
0
\
\
\
^
v
V
X
ENTAUF
•670KW
i
\
\
\
^
IF EXHAUST IS USED DIRECTLY
AS HOT AIR SOURCE, USE
THE FOLLOWING VALUES FOR
RECOVERABLE HEAT (MMBtu/hr)
\
MARS
(7400 h
^
SATU
CENT
MAR,
• PEF
H 75
EC
W)\
•^^^^^^
RN 9.E
AUR 21
X
FORM*
sig IS V
DNOMIZ
\
k-~.
NCE BE
VITHOU
ER
^
"^^
LOW
r AN •
**-^
•**
RECOVERABLE HEAT (xlO* Slu/hr)
OR APPROXIMATE STEAM CAPACITY [x!03 ib/hr)*
•ASSUMING LATENT HEAT OF VAPORIZATION EQUALS 1000 Bio/It
Fig. 6. Cost of Generating Po
ng Power
20 4Q 60 80 100 120 140 I 160 180
220 240 260 280 300
ANNUAL SAVINGS/KILOWATT IS/KW)
These curves shnw annual savings befort taxes
ys(*m tilt was tisumed lo be 15 years
Lrsitfht line depreciation wu used
ft]vulalions w«r« bawd on after list discounted
«sh How
ment tax credit wai included
as used
prn-eni nprrilmg time and percent load when
r ire included in the
Fig. 7. Economic Evaluation
581
-------
THE ENVIRONMENT FOR COGE!-ERATION IN THE UNITED STATES
ENVIROSPHERE COMPANY
A DIVISION OF EBASCO SERVICES INCORPORATED
B. R. Rossi, J. R. Jackson and E. F. Dul
Presented at
Second Conference on
WASTE HEAT MANAGEMENT AND UTILIZATION
MIAMI BEACH, FLORIDA
DECEMBER, 1978
582
-------
TABLE OF CONTENTS
Page
A. INTRODUCTION 1
B. RESULTS OF CURRENT COGENERATION
STUDIES/EFFORTS 3
C. CURRENT FEDERAL STATUS 5
D- LEGISLATIVE STATUS - CALIFORNIA 6
E. SUMMARY ANALYSIS OF LEGISLATIVE THRUST
AND IMPLICATIONS ... ..... 7
F. CONCLUSIONS AND RECOMMENDED ACTIONS 9
583
-------
A. INTRODUCTION
The national thrust toward energy conservation has directed
significant attention toward cogeneration. This attention
has taken three primary forms to date:
Specific individual utility actions that could be
considered to be the vanguard of the "second
generation" of cogeneration activities in the nation.
Legislative and administrative action at the federal
and state levels designed to ultimately encourage
constructing cogeneration facilities.
Federal funding of cogeneration research.
Additionally, the electric utility industry, the academic
community, and the consulting engineering and industrial
sectors have undertaken various degrees of analysis together
and separately to evaluate cogeneration opportunities.
Based on this, there is little question from the national
perspective that in many circumstances cogeneration is an
effective conservation method, can be attractive to the
utility industry, and will contribute to improving the
nation's energy self-sufficiency. It will provide
opportunities for:
More cost-effective and efficient energy use.
Reduction in national expenditures for energy
facility construction and operation.
Reduction in the degree of environmental effects
involved in the current mode of separate development
of electric generation facilities and process steam
production facilities.
Movement toward an integrated, total energy economy
that focuses on the gross energy cycle rather than
the individual sector domains (i.e., oil and gas
industry, process steam industry, electric utility
industry).
. Assisting utilities to meet future load requirements
in view of delays that may be encountered in getting
large generating facilities on line.
584
-------
Given the foregoing, the questions which must be raised
at this time are: "How is the national effort towards
cogeneration supported by the activities of the electric
utility industry?" and "What direction should the utility
industry assume in this effort?". Related questions are:
. Who is considered as the focal point for implementing
cogeneration?
Are stimuli being provided and in what form?
What are the formal requirements on the electric
utility industry in establishing (or not.establish-
ing) cogeneration facilities in the United States?
What roles are available to the electric utilities
in the United States' cogeneration effort?
What experiences do we have to date and what do they
teach us?
In order to answer these questions, this paper is divided
into six parts that:
1. Cite the major results of various current studies/
efforts concerning the engineering/economical/
licensing opportunities of cogeneration.
2. Identify the current status of federal legislation
fostering cogeneration.
3. Identify the current status of state activities
affecting cogeneration.
4. Analyze the targets of the various legislative
efforts and the position of the electric industry.
5. Assess the implications to the electric utility
industry inherent in the success or failure of the
national cogeneration effort.
6. Identify what actions the electric utility industry
should take to aggressively and effectively
participate in the national thrust toward cogenera-
tion, based on our current experiences.
585
-------
B. RESULTS OF CURRENT COGENERATION STUDIES/EFFORTS
There have been numerous recent studies performed by all
sectors on the issue of cogeneration and cogeneration
ventures.
Included among the major cogeneration studies performed to
date are:
. Energy Industrial Center Study, performed by
Dow Chemical Company for the National Science
Foundation in 1975.
A Study of Inplant Electric Power Generation in the
Chemical Petroleum Refining, and Pulp and Paper
Industries, performed by Thermo-Electron
Corporation for the Federal Agency Administration
in 1976.
. The Potential for Cogeneration Development in Six
Major Industries by 1985, performed by Resources
Planning Associates, Inc. for the Department of
Energy in 1977.
. National Coal Policy Project, sponsored by
Georgetown University and performed this year.
The premier results of these studies include the following:
Nationwide, current installed cogeneration capacity
is approximately 4000 MW and represents about four
percent of the total U.S. electrical power
generation.
Without government action, cogeneration capacity
from process steam can reach 10,000 MW by 1985.
With government action, cogeneration capacity from
process steam can reach 12,000-16,000 MW by 1985.
Widespread application of cogeneration utilizing
coal could save up to two million barrels of oil
per day by the year 2000.
Various mixes of industrial cogeneration activities
and industry-utility joint ventures can save the
electric utility industry §2 to $5 billion dollars
per year in capital investment requirements by
1985.
586
-------
These savings will ease the construction financial
burden of the electric utility industry in a period
of time when financing remains difficult to obtain.
The additional fuel required to produce both power
and process steam at a common site is about one-
half that required for generating electric power
only, resulting in a net reduction in air pollutant
emissions per unit of energy.
The maximum potential for cogeneration lies in such
process steam industries as: chemical, textile,
food processing, pulp and paper, steel, and
petroleum refining, with opportunities in cement,
beet processing, canning, and minerals processing.
These studies also are very clear in identifying that,
without governmental action, these potential benefits
could become lost opportunities.
Premier cogeneration projects that currently involve
significant utility industry participation should, for
logical consideration, be separated into "first" and
"second" generation experiences. The first generation
experiences, to cite a few, include those by:
Gulf State Utilities and Exxon Corporation at
Gulf States' Louisiana Station Unit No. 1
originated about 1930.
Houston Lighting & Power Company and Champion Paper
Company at HL&P's Deepwater Station initiated about
1928.
Delmarva Power & Light and Getty Oil's (formerly
Tidewater) arrangement at Delaware City, Delaware
initiated about 1956.
Public Service Electric and Gas Company of New
Jersey and Exxon Corporation at PSE&G's Linden
Station initiated about 1957.
Second generation experiences, or those that have occurred
since the total energy hiatus of the 1960's and early
ly70's, include:
587
-------
Southwestern Public Service Company's activities
which were overviewed by its Mr. Cliff Mlinar (Vice
President-Marketing) at the 5th Energy Technology
Conference and exposition in Washington, B.C. in
February, 1978. Its aggressive activities over the
past two years have accounted for opportunities that
negate the need for about three to four percent of
previously planned new plant construction.
The 50 MW joint venture between the Eugene Water and
Electric Board of Oregon and the Weyerheauser
Corporation in Springfield, Oregon.
The current activities of the City of Modesto,
California concerning the development of a 65 MW
unit to supply process steam to local canneries.
The recent activities (during part of 1976 and 1977)
of Southern California Edison Company which resulted
in the identification of 39 projects with a potential
of 612 MW of cogeneration.
C. CURRENT FEDERAL STATUS
Federal legislative action affecting cogeneration is
incorporated into that part of the National Energy Act
entitled, The Public Utility Regulatory Policies Act of
1978. Concerning cogeneration, the principal points are:
Federal rules will be established to be followed
by State PUCs.
Rates will be "just and reasonable" (no definition
supplied) and shall not discriminate against
cogenerators and small producers.
The rate at which utilities will buy power from
cogenerators will be at the incremental costs of
power produced by the utility.
All cogenerators and small power producers (under
30 MW or 80 MW for biomass) will be exempt from the
Federal Power Act, if FERC decides it is necessary
for purposes of the ACT, and will be exempt from
state laws and the Public Utility Holding Act.
Federal licensing will still be required, but with
expressed congressional intent that licensing
applications be expedited.
588
-------
In addition, other points impacting on cogeneration, are as
follows:
interconnections - A new provision, Subsection G to
Section 202 of the Federal Power Act, will permit FERC
to order the physical connection of transmission
facilities if it will encourage the overall conservation
of energy or capital and promote the efficient use of
facilities and resources.
Wheeling - A new provision to Section 202 of the Federal
Power Act, Subsection H, will permit FERC to order
wheeling of power if it will promote the efficient use
of electric generation and transmission facilities and
conserve a significant amount of energy.
D. LEGISLATIVE STATUS - CALIFORNIA
In order to gain a perspective on the status of state
legislation affecting cogeneration, administrators in
various states were surveyed to determine the extent of
cogeneration activities/regulations within each respective
state. It was found that the states vary widely regarding
cogeneration experience and activities. California, as
discussed below, has progressed the furthest in the area of
cogeneration:
California - Currently there is much activity in the
state regarding cogeneration. Governor Brown has stated
that he supports growth in geothermal, solar and
cogeneration technologies. The Energy Resources
Conservation and Development Commission (Energy
Commission ) and the PUC bcch have taken a lead role
in fostering cogeneration activity in the state. The
"Staff Report on California Cogeneration Activities",
published on January 17, 1978, outlines such actions
as:
Assembly Bill No. 2046 (Calvo) concerning geo-
thermal power and cogeneration was introduced on
August 8, 1977 and passed the Assembly with no
amendments on January 23, 1978. Although the Calvo
Bill was withdrawn from session, another bill
(SB 1805, Holmdahl) incorporating similar
provisions has been approved by the Governor
(September 20, 1978). The bill concerns the
regulation and review of such facilities as
follows:
589
-------
a. Revises the existing two-step facility siting
process before the Siting Board of the Energy
Commission, which normally requires about 36
months, for cogeneration applications for
plants not exceeding 300 MW to complete the
process in 12 months.
b. Exhaust steam, waste steam, heat, or resultant
energy produced in connection with cogenera-
tion technology would be exempted from state
sales taxes.
E. SUMMARY ANALYSIS OF LEGISLATIVE THRUST AND IMPLICATIONS
It is apparent from the previous discussions that Federal
and State Governments are actively promoting cogeneration
as a means to conserve energy. In addition, the thrust of
the legislation is such that the process industries, as
opposed to the electric utility industry, are the focal
point for implementing cogeneration. However, the utility
industry will likely be faced with having to make decisions
regarding the following activities:
Providing interconnections with industrial
cogenerators
Purchasing excess industrial generated power
Wheeling industrial generated power
Developing rate schedules for purchase and standby
power to encourage cogenerators and small producers
Using the crude "carrot and stick" analogy, significant
effort appears to be directed at offering the "carrot" to
the process industries. However, no "stick" is obvious
or probably even possible. Although this is not a problem
per se, our concern is that, if the "carrot" is not taken
up by the process industries, will the "stick" be brought
down on the utility industry?
To be practical, we believe it is fair to state that a
number of "negative" factors could be expected by the
utility industry, particularly if cogeneration is pursued
without adequate planning. These factors include:
590
-------
Loss of major base load demand
Higher consumer rates in the short term because of
fewer customers for existing generating capability
^ . Maintenance of large bulk capacity standby reserves
System load management difficulties associated witn
the schedule, capacity and reliability of purchased
and wheeled industrial power
System planning difficulties associated with the
duration and reliability of industrial cogeneration
activities
Siting and licensing (air quality) concerns
associated with industry-utility colocation ventures
The above concerns have been expressed frequently by utility
planners and are among the major reasons why cogeneration
currently contributes only about four percent of the total
U.S.-generated electrical power. Based on the legislative
thrust at federal and state levels toward efficient energy
use, conservation, and an integrated total energy economy;
however, the implications to the electric utility if the
process industries do not pursue cogeneration probably are
more significant. Potentially, these include:
Delays in licensing hearings related to determination
of the extent of utility-encouraged conservation
practices in demonstration of need proceedings. (On
this point we would like to note that at a recent
rate hearing, a utility was questioned by the state's
energy commission representatives about the need for
power and site selection process associated with a
proposed coal-fired facility. Whereas the utility
clearly indicated the factors utilized in establish-
ing the need and the site, the energy agency
essentially brought the proceedings to a halt when it
was identified that no consideration was given to at
least 15 potential cogeneration opportunities within
10 miles of the proposed site which offered sub-
stantial opportunity for electrical generation. This
may well become a ".stock" issue at other hearings in
the future.)
591
-------
Increasing financial burden caused by continuing
growth in demand, rising fuel costs, environmental
controls and inflation.
Increasing demand for suitable generating sites at a
time when large land tracts and water resources are
becoming scarce and more valuable; air quality
regulations for fossil plants are becoming stricter;
local opposition to nuclear plants is rising; and the
licensing process is becoming longer and more
complicated.
Risk of further government intervention if cogenera-
tion opportunities are not pursued by the process
industries.
F. CONCLUSIONS AND RECOMMENDED ACTIONS
Because the national effort toward energy conservation
initiated by President Carter's National Energy Plan is
directing significant attention towards cogeneration, the
concern of those involved in the electric utility industry
should be as follows:
Federal and State Governments are looking toward the
process industries as the focal point for implement-
ing cogeneration.
The federal and state thrust toward fostering
industrial cogeneration is through encouragement
rather than regulation.
The role of the utility industry in this national
effort appears to be considered as one of "accommoda-
tion" rather than prime responsibility.
If cogeneration is pursued without adequate planning,
it will adversely impact the electric utility
industry.
If the process industries do not follow through on
cogeneration activities, the utilities may, by
default, be held responsible because they will be
required to demonstrate in rate and environmental
licensing hearings specific actions taken to promote
energy conservation, including cogeneration.
592
-------
In overview then, the electric utility is in the apparent
position of being responsible for fostering and accounting
for industrial cogeneration activities without any formal
means, including federal and state regulatory measures, to
influence industrial decisions. The solution to this
potential dilemma, as we see it, is for the electric utility
industry to take the initiative in the form of positive
actions.
The following are typical actions that could be pursued by
an electric utility to effectively and aggressively
encourage cogeneration activities within its service area.
The intent of this section is not to suggest a specific
course of action, but rather to outline various actions that
reasonably appear positive in nature. Whether a utility
employs some or all of these comments or develops its own,
the key point to remember is this: it is up to the electric
utility industry to define its role in cogeneration before
that role is defined by others.
Identify cogeneration potential within the service
area
a. Institute studies of major industries to
determine potential cogeneration opportunities
b. Enter into discussions with industrial
candidates about their planned or proposed
cogeneration activities.
c. Investigate in conjunction with industry on a
case-by-case basis the alternative benefits
associated with:
Industrial byproduct generation
Industrial on-site generation with utility
management
- Utility supplied process steam and power
- Industry-utility joint venture
Institute provisions for cogeneration activities
a. Develop costs for standby services, energy
purchased from cogeneration and interruptible
service
593
-------
b. Develop rates for standby services, interrupt-
ible services and price guidelines for the
purchase of energy that reflect the values to
the utilities in the rate proposal
c. Assist customers in reviewing projects to
establish criteria which will expedite cogenera-
tion efforts
Document all activities in reports to the Public
Utilities Commission for subsequent rate, siting
and licensing hearings.
Support federal and state cogeneration studies and
the EEI Task Force on On-Site Generation.
Actively pursue having the role and responsibility
of the electric utility industry understood and
defined in all emerging federal and state
legislation.
Ensure federal and state regulatory understanding
of the implications to the electric utility
industry if the process industries do not actively
pursue cogeneration incentives and opportunities.
594
-------
FUEL COST ALLOCATION FOR Tlft. STEAM IN A COGENLRATION PLANT
K. W. Li and P. P. Yang
North Dakota State University
Fargo, North Dakota U.S.A.
ABSTRACT
The paper is intended to propose a method which would allocate the fuel
cost to the process steam produced in a cogeneration system. The proposed
method is established on the basis of the Second Law of Thermodynamics
involving a use of the concept of thermodynamic availability. Several
penalty terms are taken into consideration to produce a realistic fuel
cost chargeable to the process steam users. An illustrative example is
used to show the application of this method to a cogeneration system.
INTRODUCTION
The production of electricity from the fossil fuels is a very inefficient
process. The efficiency of the modern power plants ranges from 32% for a
simple steam plant to 43% for an advanced, sophisticated steam-gas turbine
plants. This means that currently between 57% and 68?'. of the heating
value of the fuel is being discharged as waste heat. Some of this waste
heat is unrecoverable, such as radiation and stack loss from the boiler.
But much of this heat can be used for the industrial or agricultural
applications, or for the heating of building. The use of this waste heat
is called waste heat utilization. Through the utilization of waste heat,
the overall conversion of the fuel to useable energy can be as high as 80/o.
Cogeneration plant is a system which would produce electricity as well as
process steam. There are many system arrangements currently in use [1-8].
In one arrangement, steam is first generated and partly expanded in a
turbine for production of electricity. Then, the used steam is sent out as
the process steam. In another arrangment, steam from the boiler is expand-
ed in a turbine. The process steam is extracted from the turbine at some
points. The steam not extracted will continue its expansion process and
eventually condense in a condenser. In general, the cogeneration plant
has an overall efficiency around 80% [8]. The combined heat and electric
power production was often ignored in the past. This was due,in part,
to the institutional constraints and the economic distortions it produced.
It was also due,in part, to the technical problems. But in recent years
there is an increasing awareness that these problems are not unsurmount-
able. At the time when the energy conservation program is high in the
national priorities, the cogeneration system has become increasingly
important.
595
-------
mis paper is intended to propose a method which would allocate the fuel
cost to the process steam produced in a cogeneration system. This is not
a simple matter because steam first serves as a working substance in a
turbine and then, serves as the process steam. In the past fuel alloca-
tion to these two functions is either arbitrary or based upon BTU content
[8]. Because of these, there have been some confusions and distortions
in the evaluation of cogeneration systems.
The proposed method has been established on the basis of the Second Law
of Thermodynamics. The thermodyriamic availability of steam was used to
indicate the values of steam. The higher the thermodynamic availability
of steam, the more valuable the steam would be. In this paper the avail-
ability of steam in a system is first calculated. Then the fuel cost is
allocated to the steam according to its availability. The method is
straightforward. An illustrative example is used to show its application
to a cogeneration system.
THERMODYNAMIC AVAILABILITY OF STEAM
The thermodynamic availability of the working substance including steam
is defined by
a= (h-ho) -T0(s-so) (1)
Thermodynamic availability is the property which measures the potential
of working substance to do work under certain surrounding conditions.
The change of the availability in a process means a change in the capac-
ity of a working substance to do work. In a work producing process the
decrease of availability is generally larger than the useful work
produced. The difference between these two is defined as the irrevers-
ibility of the process [10]. In the process where there is no work
involved, the irreversibility measures the loss of availability due to
the friction or due to the heat transfer through a finite temperature
difference.
To measure the process charateristies, the term "process effectiveness"
is usually applied. The general definition of the process effectiveness,
denoted by e, is [10].
Increase in the availability of desired output
e Decrease in the availability required
or
" Aoutput / Ainput
596
-------
For a work producing process the effectiveness defined in L:q. (2) can be
easily interpreted as the ratio of the actual useful work obtained to the
maximum useful work obtainable. From the therrnodynamic viewpoints, all
reversible processes between any two given end states are equally effec-
tive in producing the useful work, so that the effectiveness of any
reversible process is unity. The effectiveness of an actual process,
however, is always less than unity. In other words, the term, 1 - e,
is a measure of the potential for improvements.
For a heat reception process, the effectivness is the ratio of the
increase of availability in the working substance to the decrease of
availability in the substance from which the heat is transferred.
Similiar interpretations can be given to the processes of heat rejection
(cooling) and v/ork absorption (pumping).
ij
To illustrate the calculations of the thermodynamic availability and the
process effectiveness, a typical conventional steam turbine system was
examined. Fig. 1 shows the turbine system arrangement and its heat
balance under full-load conditions. With Eq. (1), the thermodynamic
availabilities under various steam conditions were calculated. The
results are shown in Fig. 2. It is seen that the steam availability
is a function of steam temperature and pressure. The steam with a higher
availability is more valuable than that with a lower availability.
Table 1 summarizes the calculated results of the process effectiveness.
Also included in Table 1 are the results for A 4-DU4-> ^-inout and Aloss
In this study, the process availability loss per unit time A, is
defined as difference between these two terms A. . and A t . By the
nature of the engineering process, this term must be positive or zero in
the limiting case. In this investigation the results of a steam genera-
tor analysis (boiler and reheater) and the thermodynamic availability of
coal used for plant operation were also obtained. Details of the anal-
ysis can be found in the reference [9].
It is interesting to notice from Table 1 that most availability losses
takes place in the steam generator (3143 MBtu/hr out of the total loss
3725 MBtu/hr in the plant) and the process effectiveness of this device
is only in the neighbourhood of 50%. The condenser is also relatively
ineffective. But the availability loss in the condenser is approximately
216 MBtu/hr which is much less than that in the steam generator. These
observations are obviously contradictory to those made by the First Law
of Thermodynamics. It is believed that the Second Law analysis is very
much needed when the real inefficient components are to be identified in
the plant system [10].
597
-------
FUEL COST OF UNIT AVAILABILITY
The cost of steam generated in a conventional power plant or a cogenera-
tion plant mainly consists of two terms: fuel cost and equipment cost.
In a cogeneration plant steam is partly used to produce electricity arid
partly used as process steam. Thus, there arises the question how should
the production cost of steam be distributed to these two functions. In
this study attention will be focused to the fuel cost allocation.
To determine the fuel cost (C) of steam (or water) under various condi-
tions in a power plant or the process steam in a cogeneration plant,
one must know the fuel cost per unit availability (C,). The relationship
between C and C
1
C = C
1
(4)
The fuel cost of unit availability, as shown later in this paper, is
determined by the method of steam production and the cost of the primary
fuel such as coal, oil and nuclear energy. Eq. (4) indicates that for
a given C,, the fuel cost of the steam would increase as the thermo-
dynamic availability of the steam increases. In other words, the steam
would be more valuable and costs more if it has a high availability,
that is a high potential of producing the useful work. Since the steam
is generated continuously in the system under consideration, it is quite
convenient to use the term, fuel cost rate, which is defined as below,
or
C2 = C] * a * w
C2 = C * w
(5)
(5a)
It is seen in Eq. (4) and Eq. (5) that the terms C, C-j, C2 are interre-
lated. Once one of them is known, the other two terms can be easily
determined. The basic procedures used in this studyare first to deter-
mine C2 and C-, and then, to calculate C by Eq. (4) or Eq. (5).
Fig. 3 indicates three major devices frequently encountered in a power plant
or a cogeneration plant. They are the boiler, the turbine and the heat
exchanger. Also included in the diagram are their corresponding fuel cost
balance equations and the indications of whether or not there is any
change in the fuel cost per unit availability (C,). It is seen that in
the operation of each device, the fuel cost is transferable by 100%. It
means that the fuel cost of device output is exactly equal to the fuel
cost of the device input. For example, the fuel cost of the steam at the
boiler outlet must be equal to the sum of fuel costs for the primary fuel,
hot air and water.
598
-------
The fuel cost of unit availability (C,) may or may not change in the
process, depending upon whether or not the substance receives additional
useful heat (availability) from an external source. In the steam
generator where substance receives additional useful heat (availability)
from the combustion, the term C, changes as indicated in Fig. 3a. In the
turbine process, the term C-, would remain unchanged because the substance
does not receive any additional useful heat (availability). Similarly,
in a heat exchanger, the hot fluid would have unchanged fuel cost of unit
availability while the cold fluid has changed its fuel cost of unit
availability.
Some explanations may be needed for these C0 f and C0 3 in Fig. 3a. The
<- ,< L ,a
term C2 ^ is the cost of the primary fuel consumed per unit time.
Mathematically it is:
2,f = Cl,f *af *wf
or
jf = (COST)f * (HV)f * wf (7)
Evidently, the terms C, v and C2 f can be easily calculated if the fuel
consumption, price and heating value are known. The term C0 „ is the
<~ ,a
fuel cost associated with the hot air prepared by the extraction steam.
In a mathematical form, it is:
C9 = C, * a * w. /ox
c ,a i ,a a a (o)
For determination of the fuel cost component of steam (or water), the
primary procedures are to set up the cost balance equations for each
element in the system and to solve these equations simultaneously. To
be more specific, one should first prepare a cost balance for each
element (such as the boiler, turbine, and reheater) in terms of the fuel
cost of feedwater at the boiler inlet. This feedwater fuel cost is then
calculated by taking a cost balance on the control volume which consists
of the whole feedwater heater train including the feedwater and conden-
sate pumps.- In general, the feedwater fuel cost at the boiler inlet is
the sum of the fuel costs of all extracted steam and water entering the
feedwater heater train, minus the fuel cost associated with the drain
directed back to the condenser.
To illustrate the use of procedures just described, the conventional
power plant shown in Fig. 1. will be used. With the following input
information, the fuel costs of steam under various conditions were
calculated.
599
-------
coal consumption, wf = 475,000 (Ib/hr)
coal heating value, (HV)f = 13,297 (Btu/lb)
coal availability, af = 13,186 (Btu/lb)
coal price, (COST)f = 0.8 ($/10cBtu of heating value)
The results are shown in Fig. 2. For simplicity, only the term C is
indicated in the diagram. Since the availability (a) and flow rate (w)
are indicated at various locations in the diagram, the term C-, and Cp can
be easily calculated with Eq. (4) and Eq. (5).
It is interesting to note that the fuel cost of steam shown in Fig. 2
varies with the thermodynamic availability of steam. As the availability
of steam decreases, the fuel cost of steam would decrease. For instance,
the steam has a fuel cost 1.0852 dollars per thousand pounds at the HP
turbine inlet conditions. But, the fuel cost of steam would drop to
0.8132 dollars per thousand pounds at the HP turbine exhaust end. Also
shown in Fig. 2 are the fuel costs for the boiler feedwater under various
conditions. At the condenser outlet where the availability of substance
is low (ie. 1.351 Btu/lb), the condensate has a fuel cost a 0.00226
dollars per thousand pounds. But at the boiler inlet, the fuel cost of
boiler feedwater would increase to 0.255 dollars per thousand pounds.
In the power plant under consideration, an auxiliary turbine is used to
drive the feedwater pump and the extraction steam is used to operate the
air preheater. Fig. 2 indicates steam fuel cost, steam flow rate, and
other information for these equipments.
FUEL COST CHARGEABLE TO STEAM
In a cogeneration plant electricity and process steam are produced
simultaneously. Fig. 4 indicates a typical cogeneration system. In the
system the steam of 3515 psia and 1000 F enters the turbine at the rate
of 5,492,903 Ib/hr. At the extraction point where the pressure and
temperature are respectively 153 psia and 634.2 F, the steam is withdrawn
for the process use at the rate of 426,539 pounds per hour. A complete
heat balance of this cogeneration system is shown in Fig. 4.
To determine the fuel cost of process steam, one must calculate the
availability of steam and the steam fuel cost for unit availability such
as those shown in Fig. 5. The procedures are identical to those for a
conventional power plant. As shown in Fig. 5, the fuel cost of steam at
the extraction (153 psia and 634 F) is 0.716 dollars per thousand pounds.
With consideration to the value of the return process water, (0.104
dollars per thousand pounds), the fuel cost of the process steam should
be 0.612 dollars per thousand pounds. In a general form, the equation is
600
-------
TFCST = c . ~
To have a more accurate and
alty terms to be included.
1.
2.
3.
realistic fuel cost, there are several pen-
These penalty terms are related to,
the process steam F,.
the feedwater heater' drain ?„.
the exhaust steam from an auxiliary
the exhaust steam from an air pre-
The equations used for these penalty calculation are summarized in Table
2. The derivations are straight forward. First, the loss per unit mass
of process steam is calculated. Second, the fraction of the loss (DF)
shared by the process steam must be determined. In this study the term
DF is simply
condenser
condenser
condenser
turbine,
condenser
heater, F
1
1
1
F3
1
A '
OSS
OSS
OSS
OSS
associated
associated
associated
associated
wi
wi
wi
wi
th
th
th
th
A
DF1 =
ps
Aps + Atl
(10)
if the loss is associated with the process steam, or
V * At
(11)
if the loss is associated with steam from other sources. With all these
penalty terms, the equation for the fuel cost chargeable to the process
steam would become,
PCS = FCS
(12)
It should be pointed out__that the penalty terms are generally small as
compared with the term FCS". Using the numerical information shown in
Fig. 4 and Fig. 5, the fuel cost of the process steam (P = 153 psia,
T = 634 F) was calculated and found at 0.617 dollars per thousand pounds,
Table 3 indicates all calculations involved.
FUEL COST CHARGEABLE TO ELECTRIC POWER
The procedures for the determination of the fuel cost chargeable to the
electric power generation are similar to those for process steam. First,
the consumption of steam availability for each turbine cylinder is
601
-------
calculated. Then, this term is multiplied by the unit cost of availa-
bility to obtain the fuel cost rate of this particular turbine. In a
mathematical form, it is
C2,t = Cl,t* (fi *wi -fe*we) (]3)
To determine the average fuel cost in terms of rnills/kwhr, one should
calculate the sum of the fuel cost rates for all turbine cylinders and
then, divide the resultant expression by the total turbine output. That
is
"(TCPT =
(C
+ C
+ C
A
(14)
gen
Like the case of determining the fuel cost chargeable to the process
steam, the term TFCF) should be modified to reflect the additional costs
against the electric power generation. These costs are related to,
1. condenser loss associated with the process steam G, .
2. condenser loss associated with the feedwater heater drain, Gp.
3. condenser loss associated with the exhaust steam from an auxiliary
turbine G,,.
4. condenser loss associated with the exhaust steam from an air pre-
heater, G..
5. condenser loss associated with the exhaust steam from the LP turbine,
G5-
The equations for the calculation of these terms are summarized in
Table 2. It is seen that these terms are generally obtained by multiply-
ing the loss per unit mass of process stearn by the distribution factor
(DF) and the ratio of the process steam output to the electric power
output (R). Evidently, the distribution factor for the calculation of
fuel cost chargeable to the power users is either 1.0 - DF, or 1.0 - DFp.
In this study, the ratio (R) is not dimensionless , but has a unit of
pound per hour per kilowatt. With these corrections, the fuel cost
chargeable to the power would become,
FCP = TFCP7 + (61 + G2
+ G5) * 1000
(15)
For illustration, the numerical information in Fig. 4 and Fig. 5 was used
and the fuel cost chargeable to the electric power generation was
calculated. It was found that the electric power was generated at the
fuel cost of 6.832 (mi 1 Is/kwhr) . Table 4 summaries all calculations
involved.
602
-------
DISCUSSION
The method proposed here for the prediction of the fuel cost chargeable
to the process steam was established on the basis of the Second Law of
Thermodynamics. To be more specific, the concept of therinodynamic avail-
ability was utilized. This approach is-more reasonable arid realistic.
The value of steam (or water) does not depend upon its energy content
but its thermodynamic availability which represents its capability of
producing useful work. A process in which the steam is throttled to the
atmospheric pressure from a high pressure is an example. In this proc-
ess, the steam loses little of its energy and its temperature remains
almost the same. Experiences indicate that after throttling, the steam
loses almost all of its potential of producing useful work. From an
energy viewpoint, the process efficiency is high because little energy is
lost. However, from the work producing viewpoint, the process effi-
ciency is almost equal to zero. Since the availability is a measure of
work-producing potential, a use of this concept becomes necessary in
assessing the value of steam (or water).
The proposed method is general is nature. It is applicable to the util-
ity company owned cogeneration plant where the fuel cost chargeable to
the process steam is desired. Also, it is equally applicable to the
industrial plant where the fuel cost chargeable to tiie power generation
is needed. In this method, the users (power users or steam users) have
to pay for their consumption of the steam availability on the same basis.
This may be the underlining principle developed in this study. The
following case is used as an illustration. The steam acquires a change
of availability 100 units at a fuel cost X dollars per pound. In a
certain process, 80 units of the availability are consumed for the
electric power generation while 20 units of the availability are consumed
by the process steam users. By the proposed method, the power and steam
users would be required to pay 80% and 20% of the total fuel cost re-
spectively. It is seen that the proposed method gives no preference to
either power users or steam users. For some reasons this method may not
become acceptable in commercial practice, but it surely provides the
fairest means for the distribution of fuel cost at^least from the
thermodynamic view point.
In practice the BTU method has been frequently adopted for allocating
the fuel cost in a cogeneration plant [8]. By this method the steam
and power users are charged according to the amount of heat they take.
Little attention is given to the thermodynamic steam conditions under
which the heat is consumed. Because of these, this BTU method generally
charges the steam users much more than that by the availability method
proposed in this paper.
Another method of allocating the fuel cost is the adiabatic heat drop
method [8]. In this method an arbitrary condenser pressure is first
chosen as a reference pressure. The adiabatic heat drop from the initial
state to the reference pressure is debited to the electric power
603
-------
'ic heat drop from the inlet conditions of process
steam to the reference pressure is also debited to the heating process
and, at the same time, credited to the electric power generation. This
adiabatic heat drop method is in some sense sirniliar to the availability
method, because both methods charge the power and steam users in accord-
ance with the consumption of work-producing potentials. However, in the
availability method, there is no arbitrary assumption of reference pres-
sure. Also, the assumption of adiabatic heat drop in the real and
fictitious turbines would become unnecessary.
In this paper a method has been proposed for the allocation of fuel cost
to the process steam and to the electric power generation in a cogenera-
tion plant. No attention was given to the distribution of equipment
cost. The systems under consideration are representative of those cur-
rently used by utility companies. These are generally more complex than
those used in the industrial plant. The proposed method is believed to
be equally valid for all these cogeneration systems.
In summary, the study has generated a fair means of allocating fuel cost
between the electric power generation and the process steam. The pro-
posed method should provide an additional approach to the problem of
sterm costs which has had considerable controversies in practice.
REFERENCE
1- J^lJLyi5L»_A-_> "Economic Aspects of the Combined Generation of Heat
and Electricity". Brown Boveri Review, January 1978.
2. Schwarzinbach, A., "Economic Design of District-Heating Power
Plants". Brown Boveri Review. September 1977.
3. Muhlhauser, H., "Steam Turbines in Conventional Combined District
Heating and Power Stations". Brown Boveri Review, September
1977
4. Schwarzinbach, A^jind^Frutschl. H. U., "Gas Turbines for District
Heating Power Plants". Brown Boveri Review. September 1977.
5. Hohl, R., "District Heating in Switzerland" Brown Boveri Review.
1973.
6. Kad-rnozka, J'J. andI Jlanus L., "Optimation of heat and power plants"
8'tirwbrTcf En'eTgyTo'iTf. 1970. Div. 4. paper 4.2-34.
7. Schuller, K. H., "Heat extraction for district heating from Nuclear
power plant" Brown Boveri Review, 1976.
8. Diamant, R. M. E_._, Total Energy, Pergamon Press 1970.
9. YangT Peter, "Computer Simulation of Steam Turbine System and Steam
" Cost "Estimates" M. S. Thesis, Dept. of Mechanical Engineering,
North Dakota State University, 1978.
10. Obert_,_E_-_f:_» "Concepts of Thermodynamic" McGraw-Hill Book Co. 1960.
11. Coijitoj_s_,__W_.__H_:_,"What is the true Cost of Electric Power and Steam
~~ from a Cogeneration Plant" presented in the American Power
Conference, Chicago, 1978.
604
-------
NOMENCLATURE AND SYMBOLS
a = Availability of a substance (Btu/lb)
•
Agen = Generator output (kw)
Ainput = Tota1 avaiAbility input to a device or process (Btu/hr)
^output = To1:al availability gained in a device or process (Btu/hr)
Aloss = The d'ifference between A. . and A . . of a device or
process (Btu/hr)
A = Steam availability consumed in the industrial process (Btu/hr)
r
At = Steam availability consumed by the turbine (Btu/hr)
A,, = Total availability of the process steam consumed in the
11 turbine (Btu/hr)
C = Fuel cost per unit mass (Dollars/lb)
C-, = Fuel cost per unit availability (Dollars/Btu)
Cp = Fuel cost rate (Dollars/hr)
(COST)f = The cost of fuel (Dollars/106 Btu of heating value)
F = Cost penalty chargeable to the process steam (Dollars/lb)
FCP = Fuel cost chargeable to the electric power (mills/kw-hr)
FCP = Basic fuel cost chargeable to the electric power (nii 11/kw-hr)
FCS = Fuel cost chargeable to the process steam (Dollars/lb)
FCS~ = Basic fuel cost chargeable to the process steam (Dollars/lb)
G = Cost penalty chargeable to the electric power (Dollars/kw-hr)
h - Enthalpy of a substance (Btu/lb)
h - Enthalpy of a substance under ambient conditions (Btu/lb)
(HV)f = Heating value of fuel (Btu/lb)
iti = Mass flow quantity (Ib/hr)
"'drain = Heater drain flow quantity that goes to condenser (Ib/hr)
R = Ratio of the process steam output to the electric power
output (Ib/kw-hr)
s = Entropy of a substance (Btu/lb-R)
s = Entropy of a substance under ambient conditions (Btu/lb-R)
T = Ambient temperature (R)
o
w = Flow quantity (Ib/hr)
c = Device effectiveness (dimensionless)
605
-------
SUBSCRIPTS:
a = air
as = Steam for air preheater
at = Auxiliary turbine
c = Condenser
drain = Feedwater heater drain that goes to condenser
e = Exit
f = Fuel
HP = High pressure steam turbine
i = inlet
IP = Intermediate Pressure steam turbine
o = outlet
p = Power
ps = Process steam
s = Steam
t = Turbine
606
-------
TABLE 1 SECOND LAW ANALYSIS OF A MODERN STEAM TURBINE CYCLE
Item Description of component
1. Steam turbine cycle
2. Steam generator
a. Comb us tor
b. Heat exchanger
c. Thermal stack gas loss
d. Diffusion loss
3. HP turbine
4. IP turbine
o a. Section 1
"-1 b. Section 2
5. LP turbine
a. Section 1
b. Section 2
c. Section 3
d. Section 4
e. Section 5
6. Condenser
a. Heat transfer loss
b. Effulent loss
Aipput
(106 BTU/HR)
6264.99
6273.2
806.81
817.07
474.40
342.67
1171.61
332.29
171.08
151.72
217.28
299.24
216.65
output
(10bBTU/HR)
2519.3
3129.4
730.84
765.87
451.44
314.43
1022.59
309.81
152.29
371.84
192.09
230.55
less
(105BTU/HR)
3725.69
3143.81
1632.77
1331.26
74.6
105.14
75.97
51.20
22.96
28.24
148.91
22.48
18.79
13.88
25.19
68.59
216.65
102.15
104.50
Device
Effectiveness
40.2
49.93
90.58
93.70
95.16
91.76
87.30
93.24
89.02
90.85
88.14
77.05
52.85*
-------
TABLE 1 (continued)
Item Description of component
7. Air preheater
a. Heat transfer loss
8. Auxiliary turbine
9. Feedwater heater train
a. Heater No. 1
b. Heater No. 2
c. Heater No. 3
g d. Heater Ho. 4
oo e. Heater No. 5
f. Heater No. 6
g. Heater No. 7
10. Feedwater pump
11. Condensate pump
Aipput
(105BTU/HR)
37.87
98.48
592.72
36.64
42.72
42.35
40.08
107.59
97.72
225.62
79.56
2.17
Aoutput
(10bBTU/HR)
8.21
79.56
537.93
28.44.
33.70
37.01
36.42
104.46
82.91
215.00
72.72
1.20
Aloss
(10%TU/HR)
29.66
•"•'r
18.92
54.79
8.20
9.02
5.35
3.66
3.13
14.82
10.62
6.84
0.97
Device
Effectiveness
(*)
21.7
81.30
90.75
77.65
78.88
87.35
90.85
97.10
84.84
95.30
91.40
55.30
Note; Calculations were made with the input information, af » (BTU/LB of coal),
wf = 475,000 (LB of coal/HR)
* The condenser effectiveness is calculated by treating the effulent loss as the output
of the condenser.
-------
TABLE 2 SUMMARY OF EQUATIONS FOR CALCULATION OF COST PENALTY TERMS
Item
1 . Condenser loss
associated
with the process
steam
2. Condenser loss
associated with
the feedwater
heater drain
Associated con-
denser loss per
unit mass of
process steam,
CLOSS, ($/lb)
ps.o ~ c,o
drain, o~ c,o
0
' drain, o
Cc,o^ * mdrain
/ m
PS
0
Cost penalty
terms for pro-
cess steam,
F, (S/LB)
F] • DF] * CLOSS
F] = DF1 * CLOSS
FI - o
F2 = DF2 * CLOSS
F2 = 0
Cost penalty
terms for elec-
tric power,
G, (S/KW-HR)
G1 = (1-oFj) * R
* CLOSS
G1 = (l-flFj) * R
* CLOSS
V°
G2 = (1 - DF)* R
* CLOSS
G2 = 0
Conditions
Process steam re-
turned to conden-
ser directly
Process steam re-
turned to feed-
water heater and
eventually drain-
ed to condenser
Process steam not
drained to conden-
ser
Heater drain direct-
ed to condenser
No heater drain
directed to conden-
ser
o
-------
TABLE 2 (continued)
Item
•
; 3. Condenser loss
associated with
the exhaust
steam of auxi-
liary turbine
4. Condenser loss
associated with
exhaust steam
from air pre-
heater
5. Condenser loss
associated with
LP turbine ex-
haust steam
Associated con-
denser loss per
unit mass of
process steam
CLOSS, (S/lb}
< Cat,o- Cc, J~
* -v7 V
' drain, o~
Cc,o} * V*PS
< Ca,o- Cc,J
* YV
0
(CLP,o-Cc,o>
* *LP,o/ApS
Cost penalty
terms for pro-
cess steam,
F, (S/lb)
F3 = DF2* CLOSS
F4 C DF2* CLOSS
F4 = DF2* CLOSS
F4 = 0
F5 = 0
Cost penalty
terms for elec-
tric power, G,
(S/KW-HR)
G3 = (1-DF2) * R
* CLOSS
G4 - (1-DF2) * R
* CLOSS
G4 = (1-OF2) * R
* CLOSS
G4 = 0
G5 = CLOSS * R
Conditions
Exhaust steam
directed to
condenser
Exhaust steam or
water returned to
heater and even-
tually drained to
condenser
Exhausted to con-
denser directly
No exhausted steam
or water returned
to condenser
Exhaust steam
directed to
condenser
-------
TABLE 3 FUEL COST CALCULATION FOR THE PROCESS STEAM
Availability Flow
Consumption Quantity
Quantity (LB/HR)
(106 BTU/HR)
A__ 156.22
.ps
Atl 149.25
At 2944.36
mas 113740.0
iti
ps 426539.0
m . 303119.0
at
""drain 660328.0
c 1
C '
s,o
C
a,o
Cat,o
r
drain, o
C „
c,o ,
F1 x 103
f9 x 103
t -j
F3 x 103
F4 x 103
FCS
Cost of
Fuel
($/103ln)
0.716
0.104
0.104
0.119
0.005
0.002
0.000
0.001
0.004
0.001
0.611
FCS
0.617
-------
TABLE 4 FUEL COST CALCULATION FOR THE GENERATOR OUTPUT
Availability Flow
Consumotion Quantity
Quantity (LB/HR)
(106 BTU/HR)
A e 156.22
.ps
Atl 149.25
At 2944.36
mas 113740.0
mps 426539.0
*al. 303119.0
at
rtidrain 660328.0
mLp 2868686.0
CLP,o
drain, o
r
'3,0
c
at.o
C
c,o 3
G, * NT
G2 * 103
G3 * 103
G4 * 103
G5 * 103
TCP
Cost of Generator
Fuel , Output Cost
($/10 LB) (Mills/ KW-HR)
0.1113
0.0056
0.1043
0.1192
0.0023
0.000
0.003
0.043
0.017
0.403
6.366
FCP
6.832
612
-------
. 3515
I' 1COO
OTHER
P
?
^-^
^^
74^4 p
550.2 -f
1275.4 H
7TJ 5*TD
W : Flow Rate (LS/KR)
P : Pressure (?5IA)
F : Tor.pcraturo (F)
H : 2r.±halpy (3TU/L3)
1. "et Heat
.Rate: 7326
2. !.ct Generator
Output: 725Q95
0)7.5H
(XV) 4235 P
Ii-15.93
Fig. 1. Heat Balance Diagram of a Conventional Power Plant
-------
o\
ioov.57 V
(.•-".6J A
1.0352 C
KP
30ILSR
JS17.092 A
C
IP
j;f-.&*.i I ,^ 7^;0.17;S[-i. j^jj
" ' *"
C
.79 A
52:- c
231 />(?<» •;.'
4ZC.73 A
0.717 C
103/^17
275."G3 W
1™ "i ^ " 1
J IJ '-
5, r* •^'
0.7IV
V : new Rat- (U/KR)
A : A«il=b:iity
C : Co;-;
.i-o'.2 C Q.''•,„•; C 0.
:2.'t7 A
3.113 C
U.OjI-i C U.UJ'i C 0.010 C 0.0036 C
1. "ct Keat
Er.tc; 732-3
2. r.'et 0=r.cr2-or
Cutru-i; 72i-£95
Fig. 2. Steam Availability and Fuel Cost Allocation Diagram of tt Conventional Power Plant
-------
fuel
C2,f * C2,a
Fuel Cost Balance,
Fuel Cost of Unit Availability,
(a). Boiler (or n
inlet
Fuel Cost Balance, C
Fuel Cost of Unit Availability,
(b). TurMno (or Purin)
hot fluid
cold fluid
-A/VAAA/WX-
hot air
f
Y/ater
\-t
I
4
WWWvW
»
a
stcan _
C2,s ~ C2,w
pqwer
P
exit
- C2,e * C2,p
- C
1,o
—v^VVW^
?.
Fuel Cost Balance, C0 * - C~ 0 = C0 ,~ C0
C.f I >-*£• ry;er (or ConclcnGer)
Fig. 3. Major Elements and Cost Balance Equations
615
-------
V : now Tits (L3/>
P Prorsxsrc (?c:.\)
'
4235 P
Set Generator Output: 7751''5 (K1,.')
Fig. 4. Heat Balance Diagram of a Co-qeneration Power Plant
-------
CT-
M
-J
65.53 A(-.12?)
0.111J2 (1
I
169.: A
0.23J3 C
**^w.. j ji
0.71&? C 131515 V 15^29" '»,' 1?'<."G/>
£35.57 A ' |23f..9 A 259.?3 A
/•p. p.5005 C iC.477 C Oi4C02 C
g f -•;_
5 A !j>^ i?o.77A L~^ I17.39AI r^ jT.r.S A
0.07C h^-r,~''b.o;;
1.35115 A
0.0023 I C
-*~ |J.017C t,. ~ »^, i O.C023C
~0.75? A yX I."?..772 A / I 10.7fi? A / | f'0?'? W
0.1505C CG.3
0.
62.1:7 A
C.112 C
0.0514 C 0.0ii9 C
3.'.2-'i A
O.C055 C
Ket C-ner=tor Output : 775145 (XV)
Fig. 5. Steam Availability and Fuel Cost Allocation Diagram of a Co-generation Power Plant
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
. REPORT NO.
EPA-600/9-79-031a
2.
3. RECIPIENT'S ACCESSION-NO.
4. T,TLE AND SUBTITLE Proceedings. Second Conference on
Waste Heat Management and Utilization (December
1978, Miami Beach, FL), Volume 1
5. REPORT DATE
August 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
S.S.Lee and Subrata Sengupta, Compilers
9. PERFORMING ORGANIZATION NAME AND ADDRESS
University of Miami
Department of Mechanical Engineering
Coral Gables, Florida 33124
10. PROGRAM ELEMENT NO.
EHE624A
11. CONTRACT/GRANT HO.
EPA Purchase Order
DA86256J
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development*
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Proceedings: 12/78
14. SPONSORING AGENCY CODE
EPA/600/13
is. SUPPLEMENTARY NOTES IERL-RTP project officer is Theodore G. Brna, MD-61, 919/541-
2683. Cosponsors are: EPRI, Florida Power and Light Co. , Univ. of Miami, U.S.
DoE, U.S. EPA, and U.S. Nuclear Regulatory Commission. ____^_
16. ABSTRACT
proceedmgS document most presentations made during the Second Con-
ference on Waste Heat Management and Utilization, held December 4-6, 1978, at
Miami Beach, FL. Presentations were grouped by areas of concern: general, utili-
zation, mathematical modeling, ecological effects , cooling tower plumes, cooling
towers, cogeneration, cooling systems , cooling lakes , recovery systems , aquatic
thermal discharges, and atmospheric effects. Causes, effects, prediction, monit-
oring, utilization, and abatement of thermal discharges were represented. Utiliza-
tion was of prime importance because of increased awareness that waste heat is a
valuable resource. Cogeneration and recovery systems were added to reflect this
emphasis.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
COSATI Field/Group
Pollution
Heat Recovery
Management
Utilization
Mathematical Models
Ecology
Cooling Towers
Plumes
Pollution Control
Stationary Sources
Cogeneration
Cooling Lakes
Thermal Discharges
Atmospheric Effects
13B 07A,13I
20M,13A 21B
05A
14B
12A
06F
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
632
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
618
------- |