-------
                                                                                   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
-------