wEPA
United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park NC 27711
EMB Report 85-CCT-1
October 1986
Air
Cooling Towers
Drift Methods
Study
Chromium
Method
Development
And Evaluation
Report
Munters
Corporation
Fort Myers,
Florida
-------
METHOD DEVELOPMENT AND EVALUATION
FOR
CHROMIUM AIR EMISSIONS FROM COOLING TOWERS
MUNTERS CORPORATION
FORT MYERS, FLORIDA
ESED 85/2b
EMB Contract No. 68-02-4336
Work Assignment Nos. 3 and 5
Submitted by:
Entropy Environmentalists, Inc.
Research Triangle Park, North Carolina 27709
Submitted to:
United States Enviromental Protection Agency
Office of Air Quality Planning and Standards
Emissions Standards and Engineering Division
Research Triangle Park, North Carolina
Task Manager: Dan Bivins
DECEMBER 1986
-------
TABLE OF CONTENTS
SECTION NO. PAGE NO.
1.0 INTRODUCTION 1-1
2.0 TECHNICAL APPROACH TO METHOD DEVELOPMENT TEST 2-1
2.1 Phase I - Survey of Available Field Methods
and Selection of Best Available Field Method(s)
for Validation . 2-1
2.2 Phase II - Field Evaluation • 2-2
2.2.1 Approach 2-2
2.2.2 Cyclonic Flow Bias 2-3
2.2.3 Orientation of Sampling Train Nozzle 2-3
2.2.4 Field Experiment 2-4
2.2.5 Preliminary Evaluation 2-5
2.2.6 Experiment #1 2-6
2.2.7 Experiment #2 2-7
2.2.8 Experiment #3 2-7
2.2.9 Experiment #4 2-8
2.2.10 Experiment #5 2-8
2.2.11 Other Familiarization Tasks 2-8
2.3 References 2-9
3-0 SUMMARY AND DISCUSSION OF RESULTS 3-1
3-1 Propeller Anemometer Testing 3-1
3.1.1 Anemometer Sensor Tests 3-1
3-1.2 Propeller Calibration/Angle Tests 3-2
3.1.3 Preliminary Cooling Tower Tests 3-2
3-1.4 Use of the Propeller Anemometer
During Sampling 3-5
3-1.5 Conclusions for Propeller Anemometer 3-5
3.2 Method Development Test Experiments 3-5
3.2.1 Experiments 1 through 4 3-5
3.2.2 PMR vs. PMR for Large Particles 3-10
3-2.3 PMR vs. PMR for Small Particles 3-10
3.2.4 Further Analysis of Sample by NAA 3-10
3.2.5 Precision of Method 3~l4
3.2.6 Collection Efficiency of Sampling Train 3~l4
3-2.7 Surrogates for Drift Testing 3-18
3.2.8 Conclusions 3~l8
3.2.9 Other Measurements 3-21
4.0 PROCESS OPERATIONS 4-1
(continued)
ii
-------
TABLE OF CONTENTS (continued)
SECTION NO. PAGE NO.
5-0 SAMPLING LOCATIONS AND TEST METHODS 5-1
5-1 Cooling Tower Exhaust Stack (Sampling Location A) 5-1
5.2 Cooling Water Basin (Sampling Location B) 5-5
5-3 Ambient Meteorological Station
(Sampling Location C) 5-5
5-4 Velocity and Gas Temperature 5-5
5-5 Molecular Weight 5-5
5.6 Lithium and Bromine 5-5
6.0 QUALITY ASSURANCE 6-1
APPENDICES
A TEST RESULTS AND EXAMPLE CALCULATIONS A-l
Lithium Results A-3
Example Calculations A-7
Bromide Results A-186
Anemometer vs. 3-D Pitot Probe A-204
Hot-Wire Anemometer Results A-210
B RECOMMENDED DRAFT TEST METHODS B-l
Method - Direct Measurement of Ge.s Velocity and
Volumetric Flowrate Under Cyclonic Flow Conditions
(Propeller Anemometer) 3-3
Method - Determination of Chromium Emissions
From Cooling Towers B-ll
C FIELD DATA c_!
Preliminary Field Data C-3
Air Flow Rate Determinations C-4
3-Dimensional Velocity Traverse Data C-6
Lithium Field Data C-8
Propeller Anemometer Traverse Testing C-98
Hot-Wire Anemometer Data C-101
D ANALYTICAL DATA D-!
Filter and Backhalf Impinger Bromide Catch D-3
Analytical Results for Surrogates D-4
Molybdenum Recovery Tests D-8
NAA Results D_o
E CALIBRATION DATA E-l
Calibrations E~3
Train Component Calibrations E~5
Propeller Anemometer Sensor Calibration E-33
Propeller Anemometer Calibrations £-38
F SAMPLING PROGRAM PARTICIPANTS AND OBSERVERS F-l
G ADDITIONAL INFORMATION G-l
111
-------
LIST OF TABLES
TABLE NO. PAGE N0.
3-1 Comparison of Propeller Anemometer and 3-Dimensional
Pitot Results 3-3
3-2 Point-by-Point Comparisons of Flow Angle Differences 3-4
3-3 Testing Schedule at Munters Corporation Cooling Tower 3-7
3-4 Percent Sample Recovery By Train Component 3-8
3-5 Summary of PMR and PMR Results for Lithium Analyzed
By GFAA a c 3.9
3-6 Impinger-Type Train at Different Isokinetic Rates and
Nozzle Sizes 3-11
3-7 Impinger-Type Train at Different Isokinetic Rates and Nozzle
Sizes 3-12
3-8 Impinger-Type Train at 100# Isokinetic Rate and Different
Nozzle Sizes 3-13
3-9 Impinger Type-Train at Different Isokinetic Rates and
Nozzle Sizes, Analysis Conducted by NAA 3-15
3-10 Summary of PMR and PMR Results for Bromide Analyzed
by NAA a c 3_16
3-11 Calculated Precision of Impinger Train Results 3-17
3.12 Percent Recovery of Lithium by Sample Train Component 3-18
3.13 Analytical Techniques and Analytes Evaluated 3-19
4.1 Cooling Water and Air Temperatures in °F 4-2
5-1 Sampling Plan for Munters Corporation 5-3
6.1 Audit Report Analysis 6_2
IV
-------
LIST OF FIGURES
FIGURE NO. PAGE NO.
5-1 Air flow diagram of Hunters Corporation crossflow cooling
tower showing sampling locations. 5-2
5-2 Top view of crossflow cooling tower exhaust stack, Hunters
Corporation (Sampling Location A) 5-^
-------
1.0 INTRODUCTION
The U. S. Environmental Protection Agency has a program plan to provide
data to characterize chromium (Cr) emissions from existing cooling towers
nationwide. This includes emissions from cooling towers equipped with both
average and high-efficiency drift eliminators. Together with other
information, emission test data gathered will be used to develop a National
Emissions Standard for Hazardous Air Pollutants (NESHAP) for this source
category: cooling towers. At present, the EPA is considering an equipment
standard, an air emission limit standard, or combination of both. In response
to the program plan, the Emission Measurement Branch has been given the
responsibility to develop and evaluate a method for the sampling and analysis
of chromium emissions from cooling towers.
In developing standardized methods, EPA encourages, whenever possible, the
use of existing EPA Reference Methods and/or industry accepted or standardized
methods. In this case, however, none of the current EPA Reference Methods can
be used for cooling towers without some procedural modifications because of the
presence of cyclonic flow and large water droplets.
The primary cooling tower national organization or standardization group
was contacted to determine if they had a applicable test method for chromium
emissions. This group, the Cooling Tower Institute (CTI), does not currently
have any standardized emission measurement methods that could be used for
evaluating chromium emissions from cooling towers.
The method development and evaluation study concentrated on identifying a
source emissions measurement method that would provide results in terms of a
mass emission rate of chromium. Particle size distribution methods will be
determined during the standards setting test, but particle size is not likely
to be part of any proposed emissions standard. This study also considered
method accuracy, cost, and complexity. For any method to become an EPA
Reference Method, it must have acceptable accuracy, be cost effective, and be
useable by the typical emissions measurement organizations.
The initial methods development plan was discussed at the Cooling Tower
Institute Conference held January 27-29, 1986. The plan was presented at the
subcommittee meeting on drift measurement, and all comments received were
incorporated into the revised work plan.
1-1
-------
The data generated from the emission test program presented in this report
were used to develop a draft reference method for measurement of chromium from
cooling towers. This method includes the use of other compounds or elements as
surrogates for making measurements on cooling towers with low concentrations of
chromium and for sources that do not contain chromium. The methods development
and evaluation test program was conducted at The Hunters Corporation in Fort
Myers Florida during the period of March 3 through 8, 1986. The Hunters
facility was selected because:
1. The Munters Corporation operates and maintains pilot scale crossflow
and counterflow cooling towers that are equipped for product
performance testing purposes. The existing equipment configurations
would require only minor modifications to accomodate the testing
program;
2. The cooling towers are single cell units that could be modified easily
to simulate various cooling tower configurations;
3- Surrogate compounds could be added to the recirculating water;
4. The cooling towers could be operated at close to design conditions;
5. The cooling tower operating parameters could be altered to represent
the various conditions that were encountered during actual test
programs at other locations;
6. The operating parameters could be monitored easily when tests were
being performed;
7. Mild meteorological conditions were expected to be present in the Fort
Myers area during the test period; and
8. The Munters staff demonstrated a willingness to assist EPA in the
methods development program.
Mr. Rodney Gibson of Midwest Research Institute (MRI) monitored the process
operations. Mr. Dan Bivins (EPA Task Manager) of the Emission Measurement
Branch (EMB) observed the test program. Mr. Steve Adams served as the Munters
Corporation contact.
This report is organized into several sections addressing various aspects
of the testing program. Immediately following this introduction is the
"Technical Approach To Method Development Test" which describes the approach
1-2
-------
that EPA used to develop the test plan. Next the "Summary and Discussion of
Results" section presents a discussion of the experiments conducted, the
results of each experiment and a discussion on how this impacted the
recommended draft test method. Following this is the "Process Operation"
section which includes a discussion of the process and the operational
conditions during testing. The next section, "Sampling Location and Test
Methods" describes and illustrates the sampling location for the methods
evaluation testing and then explains the sampling strategies used. The
"Quality Assurance" section explains the procedures used to ensure the
integrity of the sampling program. The Appendices present the complete Test
Results and Example Calculations (Appendix A); Sampling and Analytical
Procedures (Appendix B); Field Data (Appendix C); Analytical Data (Appendix D);
Calibration Data (Appendix E); Test Participants and Observers (Appendix F);
and Additional Information (Appendix G). Appendix B will include the
"Recommended Draft Test Methods" which describe the test methods that will be
used in subsequent testing programs to measure emissions from cooling towers.
1-3
-------
2.0 TECHNICAL APPROACH TO METHOD DEVELOPMENT TEST
In general, method development and evaluation studies employ a three phase
approach involving (1) a survey of available methods and selection of the best
for validation, (2) laboratory evaluation of the method(s), and (3) field
validation of the method(s). In this case, extensive methods development and
evaluation of chromium emission measurement methods for several other indus-
tries has already provided a wealth of supporting laboratory data. For
example, the EPA currently has a draft test method for measurement of
hexavalent chromium emissions which is presented in Appendix B. Thus, the
selection and evaluation of the method for determining chromium emission from
cooling towers was be conducted in only two phases. These two phases are:
Phase I - Survey of available field methods and selection of best available
field method(s) for validation.
Phase II - Field validation.
2.1 PHASE I - SURVEY OF AVAILABLE FIELD METHODS AND SELECTION OF BEST
AVAILABLE FIELD METHOD(S) FOR VALIDATION
This first phase, involved the collection of any information relevant to
the measurement of chromium emissions from cooling towers. Numerous articles,
publications, studies, and test reports were collected and reviewed.
Discussions were held with industrial personnel responsible for assessing and
controlling cooling tower operations and with key emission measurement groups
that are currently performing cooling tower drift measurements and other
cooling tower evaluations.
Because chromium emissions require laboratory analysis, a remote sensing
device cannot be used in their measurement. With manual methods, chromium
emissions had, in the past for many source categories, been satisfactorily
collected using either a dry collection or wet collection technique. The dry
technique generally includes (1) a precollector (of the heated glass beads or
cyclone type) to collect the majority of the drift, followed by (2) a filter to
ensure near complete collection. The wet technique typically involves
collection of the drift in a liquid (in impingers), which may or may not be
followed by a filter.
2-1
-------
The use of the analytical techniques described in the attached EPA draft
test method for hexavalent chromium and the use of Neutron Activation Analysis
(NAA) should give sufficient accuracy within the working range for use in an
EPA Reference Method. Collection efficiency of the sampling train used to
collect the drift and an accurate analytical technique to measure the amount of
chromium collected were acceptable for other source categories but would have
to be verified again for this source category.
The single technical aspect of the method that needed to be resolved was
how to collect a representative sample from the cooling tower in the presence
of cyclonic flow.
2.2 PHASE II - FIELD EVALUATION
The initial method development and evaluation test was performed on a
mechanical cooling tower. Therefore, the following dicussions regarding the
field evaluation tests of the methods assume their use on mechanical cooling
towers with cyclonic flow.
As previously mentioned, both dry and wet collection techniques have been
previously used for drift measurement. Also, EPA has successfully used both
dry and wet collection techniques for studies of chromium emissions from other
industrial categories.
For these methods evaluation tests, EPA had initially selected a dry
collection technique using a cyclone prior to the filter. This approach has
been selected to reduce the overall weight and size of the sample collection
train which would be suspended above the fan. The wet technique would require
sample concentration prior to analysis to obtain a quantifiable chromium
concentration.
2.2.1 Approach
There are two methods of calculating the mass emission rate of measured
pollutants collected isokinetically: one is on a concentration basis and the
other on a ratio of areas basis. For its Reference Methods, the EPA currently
uses the concentration basis. To determine the mass emission rate, the con-
centration of pollutant measured by the sampling train is multiplied by the
flow rate measured by the pitot tube. This approach is more accurate for is-
okinetic testing involving emissions composed of small particles. The second
approach (using the ratio of areas) calculates the mass emissions collected by
2-2
-------
the sampling train divided by the time for collection and then multiplies this
by the ratio of the area of the stack divided by the area of the nozzle. This
approach is more accurate for a isokinetic sampling of large particles.
2.2.2 Cyclonic Flow Bias
Mechanical cooling towers have vane axial fans. These fans create cyclonic
flow in the discharge stack. This cyclonic flow creates significant problems
with respect to both measurement of the flow rate and sample collection. A
pitot tube, when oriented parallel to the tower wall, will give measured values
higher than true values. The type "S" pitot tube can be rotated to compensate
for flow misalignment in the yaw direction. However, it cannot compensate for
flow misalignment in the pitch angle.
An accurate flow determination can be made under cyclonic flow conditions
using a 3-dimensional pitot tube. These pitot tubes determines the flow
misalignment in both the yaw and the pitch angle directions. Because of the
overall weight of the 3-dimensional probe, the time required in making the
measurements, the complexity of its use, and the small sensor holes that are
prone to plugging, the 3-dimensional pitot tube would not be practical for
making flow measurements on cooling towers.
Based on discussions with Environmental Systems Corporation (ESC), which
has done much of the developmental work on cooling tower drift testing, the
most practical and most accurate instrument to measure flow rate in cooling
towers is the propeller anemometer. When the propeller anemometer is
positioned parallel to the axis of the fan, it very nearly gives the true value
for the gas velocity in the vertical direction.
2.2.3 Orientation of Sampling Train Nozzle
Since cooling towers are characterized by having emissions composed of very
large particles (as explained later), it should be possible to conduct accurate
testing using the ratio of areas for calculation. For this technique, the
probe nozzle would be oriented in the normal manner (parallel to the fan
axis). EPA intended to show that, in contrast to the use of the alignment
method, there is no need to align the nozzle in the actual direction of flow
at each sampling point.
Because the nozzle would also be oriented parallel to the axis of the fan,
there was some concern that the drift could impact on the sides of the nozzle
2-3
-------
and bias the results. Therefore, standard nozzles were to be checked to
determine if impacting drift would present a problem. In addition, nozzles
were redesigned to minimize the impact area and brought to the field in the
event that the standard nozzles showed either an apparent bias from the
impacting drift or a high degree of imprecision.
2.2.4 Field Experiment
The method development test was conducted at a pilot scale crossflow tower
at the Munters Corporation in Fort Myers, Florida. The field experiment was
designed to provide analytical data within 2k hours of sampling to ensure that
the majority of the tentative results of the study would be available to EPA
prior to leaving the field.
The field validation was designed with five experiments as follows:
1) Smaller particles with an average angle of flow misalignment -
standard nozzle
• 50% Isokinetic (2 trains) vs. 100% Isokinetic (2 trains)
vs. 150?; Isokinetic (2 trains)
2) Large particles with a greater than average flow misalignment -
standard nozzle
• 50# Isokinetic (2 trains) vs. 100% Isokinetic (2 trains)
vs. 150?; Isokinetic (2 trains)
3) Large particles with greater than average flow misalignment -
redesiged nozzle
• 50% Isokinetic (2 trains) vs. 100% Isokinetic (2 trains)
vs. 150# Isokinetic (2 trains)
k) Large particles with an average flow misalignment - Impinger train vs.
heated cyclone and filter
5) Analysis of pollutant concentration with respect to particle size
distribution
Each experiment was conducted at a single point within the tower. Six
sampling trains were used to provide the necessary data for statistical
evaluation. Each of the 6 nozzles were located within approximately 4 inches
of each other. Two velocity measurement devices were used, the propeller
anemometer and the 3~dimensional pitot tube. The isokinetic sampling rates for
all six trains were set based on the readings of the propeller anemometer. The
3-dimensional pitot tube readings were only used as reference values to
evaluate the propeller anemometer.
2-4
-------
Each of the comparisons were to be run 3 or 4 times depending on the
average sampling time required. This resulted in 3 or 4 pairs of results for
use in the statisical evaluation. If the first two experiments produced the
desired response, then the third experiment with the redesigned nozzle would
not be conducted as explained below.
2.2.5 Preliminary Evaluation
A preliminary evaluation of the propeller anemometer and characterization
of the cooling tower flow discharge was made. This included establishing the
proper equal area sampling point locations and then conducting two complete
traverses, each using the propeller anemometer and the 3-dimensional pitot
tube.
The cooling tower water was spiked with lithium bromide. The spiking of
the water was conducted to provide two measurable elements in the water that do
not typically exist in the ambient air. The use of a spike or surrogate
compound may be necessary to provide accurate data on other drift measurements
for chromium when the chromium concentration in the water is extremely low
(less than k ppm).
After the water was spiked, the concentration of the bromine was measured
on-site with a specific ion electrode to verify that the proper concentration
had been reached. The specific ion electrode was to be used to measure the
concentration of both the cooling tower water and samples. This approach was
designed to ensure that all samples were within the proper quantifiable limits.
It had been EPA's intent to spike the cooling water at the Hunters facility
with chromate (in addition to lithium bromide) and to measure the emissions of
chromium, too. In pursuit of this, EPA requested that the State of Florida
allow a one-time chromate spiking of the cooling water and was denied the
request. The absence of chromium in the cooling water diminished the
evaluation of the test to some degree. However, since the analytical
techniques for chromium have already been thoroughly documented as both
accurate and precise, the evaluation using bromine and lithium would still be
suitable for demonstrating viability of the sampling method. The use of the
bromine as a surrogate for low chromium sources will have to be evaluated at
other sources during the standard setting program.
2-5
-------
2.2.6 Experiment #1
The first experiment was designed to show that cyclonic flow involving
small particles can be sampled with the nozzle in the normal orientation when
using the ratio of areas calculation.
Initially, the particle size distribution was set at the lower end
(preponderance of small particles) which is more characteristic of high
efficiency drift eliminators. The mass median particle size was believed to
be about 100 urn. An in-situ hot wire particle size measurement device was used
to estimate the particle size distribution.
A point within the duct was selected that had a typical misalignment flow
angle (30 to 60 ). This experiment was conducted first to minimize any
possible bias of the drift impacting on the sides of the nozzle.
Three runs were to be conducted; one pair of trains were to be set to
sample at 50% of isokinetic (I), one pair of trains were to be set to sample at
100# of isokinetic. and one pair of trains were to be set to sample at 150% of
isokinetic.
Initially, all samples collected were air freighted back to Research
Triangle Park, NC and analyzed the next morning by RTI (Entropy's subcontractor
under the EMB contract). RTI used the inductively-coupled argon plasmography
(ICAP) technique for sample analysis. This analytical technique can provide
analytical results for calcium, magnesium, manganese, sodium, and lithium. The
results were then relayed by telephone to enable emission calculations to be
completed within 24 hours of testing. In addition, all samples were analyzed
on-site for bromine concentration prior to shipment.
The mass emission rate for each test run was calculated using both the
ratio of areas technique and the concentration technique as shown below:
Pollutant Mass Rate (PMR ) - Ratio of Areas
a
PMR = mn As 1 = mg/hr
8.
e A 1000
n
where
m = mass of pollutant, ug
6 = sampling time, hr
2
A = area of stack, ft
s p
A = area of nozzle, ft
2-6
-------
Pollutant Mass Rate (PMR ) - Concentration Basis
PMR = % vs .. As = mg/hr
c std
V 1000
mstd
where
m = mass of pollutant, ug
V = sample volume metered, dscf
mstd
v = velocity of stack, dsfh
Sstd 2
A = area of stack, ft
s
Theoretically, the results of this experiment should have demonstrated that
PMR is the same at $0% vs. 100? I and at 100% vs. 150% I. In addition, for
3.
each of the evaluations, PMR should have equaled PMR at 100% isokinetic.
& C
2.2.7 Experiment #2
The second experiment was to assess any possible bias caused by the drift
impacting on the sides (outside and inside) of the nozzle. The particle size
distribution of the particles was changed to larger particles which are more
characteristic of medium efficiency drift eliminators. The mass median
diameter of the particles should have been approximately 300 to 400 urn. The
point in the tower selected for sampling was to have a greater than typical
misalignment flow angle (60 to SO ).
Experiment #2 involved the same runs and comparisons as Experiment #1. If
the comparisons again showed no bias, then there would be no need to run
Experiment #3. If, however, it was apparent that the impact of the drift on
the standard nozzle was creating either a high or low bias, then Experiment #3
was to be run using the redesigned nozzle.
2.2.8 Experiment #3
Experiment #3 was only to be conducted if Experiment #2 indicated that the
impacting drift on the nozzle created a bias. Experiment #3 would be conducted
in the same manner as Experiment #2 using the redesigned nozzle which would
have a shorter nozzle tip. The shorter tip would minimize the area on which
the drift could impact. Experiment #3 was not conducted since no bias was
apparent in Experiment #2.
2-7
-------
2.2.9 Experiment M
Experiment M was to be conducted using an impinger train with a particle
size distribution towards the larger particles (a condition expected to
generate the most testing problems) and the flow at a typical misalignment
angle (30 to 60°). The purpose of this experiment was to aid in designing a
system that can be used at cooling towers in refineries and chemical plants
where explosive atmospheres may exist. If an unheated train could be used, it
would eliminate the need for electrical components on the portion of sampling
train in contact with the cooling tower discharge.
Experiment #4 was to involve the same runs and comparisons as
Experiment #1. If the experiment demonstrated that there were no problems,
then the unheated system could be utilized at plants with potentially explosive
emissions.
2.2.10 Experiment #5
Experiment #5 was conducted to determine if the pollutant concentration
measured in the drift droplets varied with the particle size distribution of
the effluent. The current literature does not document changes in the
pollutant concentration with respect to particle size distribution in cooling
towers. If chemical analyses of specific compounds or elements in the drift
are to be an accurate indicator of the actual amount of drift, then the
concentration of compounds or elements measured in the drift would have to be
the same as the concentration in the cooling water over the entire range of
particle size distributions of the drift.
To determine the actual concentration of the drift, the bromine concentra-
tion was to be measured. This was to be accomplished using two cyclones. The
first cyclone would collect all particles greater than approximately 20 urn.
The second cyclone would collect particles from approximately 20 urn down to
approximately 5 urn. The concentration of bromine was to be analyzed in the
drift collected in each cyclone. The concentration of bromine measured in the
drift collected from the different particle size ranges would be compared with
the concentration of bromine in the cooling water.
2.2.11 Other Familiarization Tasks
Several other instruments which will not be part of the EPA Reference
Method were used to make measurements at this test. This was done to
2-8
-------
investigate instrument operations and calibrations and to provide sampling
personnel a chance to become familiar with the instrument operation.
2.3 REFERENCES
1. Cox, X. B., R. L. Linton, and F. E. Butler, "Determination of Chromium
Specialization in Environmental Particles; Multitechnique Study of
Ferrochrome Smelter Dust," ES&T, Vol. 19, No. 4, April 1985.
2. Peeler, J. W., F. J. Phoenix, and D. J. Grove, "Characterization of
Cylonic Flow and Analysis of Particulate Sampling Approaches at Asphalt
Plants," Entropy Environmentalists, Inc.
3. "Development of Droplet Sizing for the Evaluation of Scrubbing
Systems," United States Environmental Protection Agency, Research
Triangle Park, North Carolina, EPA-600/7-79-166.
2-9
-------
3.0 SUMMARY AND DISCUSSION OF RESULTS
The experiments described in Section 2.0 were to be conducted at the
crossflow cooling tower at the Hunters Corporation, however, later tests were
modified in response to results of earlier runs. Only the outlet location of
the Hunters Corporation cooling tower was tested. A flow diagram for the tower
is shown in Section 5.0 (Figure 5-1). No sample ports were installed on the
tower; samples were collected at the exit of the fan stack. In keeping with
this, all standards setting tests in the future will be conducted at the exit
of the cooling tower stack(s). A complete description and diagram of the stack
exit can be found in Section 5.0. The results for and any modifications in
each experiment in the methods development test are presented and discussed
below.
3-1 PROPELLER ANEMOMETER TESTING
The initial experiment was designed to confirm the propeller anemometer as
the flow measurement method of choice. Several pretest calibrations were
performed in a wind tunnel and then, during the cooling tower test program,
the propeller anemometer was evaluated by comparing its measured value at each
test point to the results obtained with a three dimensional pitot tube.
3-1.1 Anemometer Sensor Tests
The anemometer produces a voltage that is directly proportional to the
rotational speed of the propeller. Five propeller anemometers were tested to
determine a conversion factor for each of the sensors. The consistency of the
results indicated that a conversion factor of 3-60 rpm/mV should be used for
all of the sensors.
Initial Bearing Torque tests were conducted prior to the field test to
ensure that the performance of the bearings in each sensor was within the
limits established by the manufacturer. After the field testing was
completed, a final Bearing Torque test was conducted for the one sensor used
in the test program. No deterioration in the bearings of this sensor was
indicated.
The procedures and results for the anemometer sensor tests are presented in
Appendix E.
3-1
-------
3-1.2 Propeller Calibration/Angle Tests
The calibration of each of six propellers was evaluated using a wind tunnel
at air velocities of 8.1 and 4.8 m/s. The propellers tested included two
constructed of polypropylene (18 cm dia.), two constructed of polystyrene (23
cm dia.)t and one constructed of polystyrene (19 cm dia.).
For one polystyrene and one polypropylene propeller, the angle of the
propeller was varied 180° in increments of 10 in order to generate curves
showing wind angle versus percent response.
The resulting calibration and wind angle curves were found to agree with
those supplied by the manufacturer of the propellers. Since the test results
provided by the manufacturer were found to be within 2% of the results for the
wind tunnel tests, it was determined that the manufacturer's calibration
results should be used in the test program. Detailed descriptions of the
propeller calibrations are presented in Appendix E.
3.1.3 Preliminary Cooling Tower Tests
A preliminary evaluation of the propeller anemometer was conducted at the
cooling tower discharge. The purpose of the testing was to characterize the
flow profile at the cooling tower discharge and to compare the velocity results
obtained from the propeller anemometer to those obtained from a 3~dimensional
pitot tube. The traverse point locations were determined using Reference
Method 5D. This method required that 24 points (12 points on each of two
diameters) be sampled at the cooling tower discharge. The four traverse points
closest to the sides of the exhaust duct could not be used in the anemometer
test because the center of the propeller anemometer could not be moved close
enough to the duct wall.
The results for the two runs on the cooling tower are summarized in Table
3.1. The mean velocities for Run 1 were 27.4 ft/s for the propeller anemometer
and 24.6 ft/s for the 3~diniensional pitot. For Run 2, the mean velocity for
the propeller anemometer was 27.1 ft/s and the mean velocity for the 3~dimen-
sional pitot was 24.2 ft/s. The mean differences for the two runs were 11% and
12%. Since the two methods were within 2% in the wind tunnel testing, these
differences were assumed to be measurement error and no corrections were made
to the propeller anemometer readings during testing.
The absolute angle of the flow in the cooling tower discharge was also
determined from the 3"dimensional pitot traverse. The results for the cooling
tower runs (presented in Table 3«2) show measured angles of 34° for Run 1 and
37 for Run 2. The angle of the flow was expected to cause a low bias
3-2
-------
TABLE 3.1. COMPARISON OF PROPELLER ANEMOMETER AND 3-DIMENSIONAL PITOT RESULTS
Traverse
Point
A-2
3
4
5
6
7
8
9
10
11
B-2
3
4
5
6
7
8
9
10
11
RITM 1_
riuii JL
Propeller
Anemometer
(ft/s)
38
38
32
21
12
11
25
33
37
35
34
36
32
28
10
8
17
28
35
38
3-D Pitot
(ft/s)
33
33
30
20
9
9
28
32
31
31
30
33
29
18
6
2
17
30
33
34
DT TM T
Propeller
Anemometer 3-D
(ft/s)
38
38
32
21
11
9
21
34
38
36
36
35
34
24
10
16
16
28
35
36
Pitot
(ft/s)
32
33
30
20
7
7
25
31
33
32
30
31
30
22
8
16
16
28
32
33
Mean Values: 21.
24.6
27.1
24.2
3-3
-------
TABLE 3.2. POINT-BY-POINT COMPARISONS OF FLOW ANGLE DIFFERENCES
(Angles in units of degrees)
Traverse Yaw Angle Pitch Angle Absolute Angle
Port Point Run 1 Run 2 Run 1 Run 2 Run 1 Run 2
A 1
2
3
4
5
6
7
8
9
10
11
12
B 1
2
3
4
5
6
7
8
9
10
11
12
-17
-11
-12
-19
-35
-64
60
24
22
17
20
24
-26
-19
-18
-20
-29
-65
83
47
24
15
13
14
-19
-13
-14
-17
-32
-65
62
33
20
20
20
22
-24
-21
-17
-16
-15
-60
76
49
27
16
14
14
8
11
17
18
24
23
-39
-27
-19
-14
-10
- 7
15
17
19
20
35
37
-29
-26
-17
-11
- 7
- 4
18
18
19
23
31
29
-44
-30
-23
-18
-13
-14
17
18
19
23
33
39
-36
-29
-22
-17
-15
-12
19
16
21
27
42
67
67
35
28
21
22
24
30 .
26
26
29
45
71
83
52
28
18
14
14
27
23
24
29
44
69
70
^3
30
26
23
25
30
28
26
28
36
68
78
54
34
23
20
18
Avg. Abs. 29 29 19 23 34 37
3-4
-------
in the propeller anemometer results, since the response of the propeller
anemometer is approximately proportional to the cosine of the flow angle.
Instead, the velocity measurements from the propeller anemometer were found to
be on the average, \2% higher than the 3~dimensional probe results. This effect
is believed to be caused by the pulsing effect of the exhaust and the inertia
associated with the anemometer propeller.
The data from the cooling tower discharge traverses are included in
Appendices A and C.
3.1.4 Use of the Propeller Anemometer During Sampling
During the sampling at the cooling tower discharge, the propeller anemom-
eter was used to determine the effluent velocity at the sampling point. Since
the sampling nozzles had to be located just upstream of the anemometer, there
was concern that the sampling nozzles would hinder the flow, causing a bias in
the anemometer results. It was found that the placement of six nozzles in front
of the anemometer caused a low bias of approximately 9% in the velocity results.
In order to account for this bias, the velocity was checked before and after
each run, without the sampling nozzles in place. The response was also checked
throughout the run to ensure that the velocity remained constant. The 9# dif-
ference was added to the value obtained for each run to account for the bias.
Since future implementation of this sampling method would require only one
nozzle, the propeller anemometer response was observed while one nozzle was
placed in its path. The addition of the nozzle produced no apparent change in
the anemometer response.
3.1.5 Conclusions for Propeller Anemometer
The propeller anemometer was cited by cooling tower testing experts as the
best method of determining the velocity and flow rate in cylonic flow. Both the
wind tunnel and field evaluations demonstrated the fact that the propeller
anemometer should be the flow measurement method of choice. As a result, it has
been chosen for use in cooling tower testing and is incorporated into the
recommended test methods presented in Appendix B.
3.2 METHODS DEVELOPMENT TEST EXPERIMENTS
3.2.1 Experiments 1 through 4
Experiment No. 1 was conducted using the EPA Method 5 (dry collection
technique) sampling train to collect the drift. It was designed to evaluate PMR
vs. PMR at different isokinetic rates using different nozzle diameters under
a O
small drift particle size conditions (respresenting use of high efficiency
3-5
-------
drift eliminators). As shown in Table 3-3 for Run 1 of Experiment No. 1 (with
the particle size distribution towards the smaller particles), six sample
trains were operated simultaneously, two at 50# of the isokinetic rate, two at
100% of the isokinetic rate, and two at 1$Q% of the isokinetic rate. The
attached nozzles had diameters of 0.25 inches. 0.3 inches and 0.375 inches
respectively. All were EPA Method 5-type trains which meant that they had
front-half filters. The filters were of the glass fiber type which is
typically used in EPA Method 5 testing. The only compound of interest in the
cooling water was lithium bromide.
Runs 2 and 3 duplicated Run 1 in Experiment No. 1. The initial results of
the front-half analysis indicated a problem with sample collection or the
analytical technique. The impinger contents (back-half) were then analyzed.
This analysis demonstrated that a significant amount of the compound was going
through the filter into the impingers as shown in Table 3-4. Although the
bromine analysis was not considered accurate enough for comparision of results
(as shown later in the discussion of Experiment Nos. 3 and 4), it did indicate
that more than half of the bromine was passing through the filter.
A Lithium Corporation representative was contacted and said that lithium
could react with the glass fiber filter and form an insoluble lithium
silicate. As a result of this possible reaction, the glass fiber filter was
replaced with a Teflon filter or no filter starting with Run 5 of Experiment
No. 1 and for all later runs.
Since the analytical results were not available until the afternoon
following the previous day's sampling. Run 4 was conducted in the morning using
a technique that proved unacceptable when the results of the previous day's
sampling were received. Run 4 was conducted using glass fiber filters with all
Trains at 100% isokinetic in an effort to determine the precision of sampling
and analytical technique.
After it was discovered that the lithium reacted with the glass fiber
filters. Run 1 of Experiment No. k (see Table 3.5) and Run 5 of Experiment No. 1
(see Table 3-5) were conducted using Teflon filters. It was not clear whether
the bromine was being caught in the impingers because of poor sample train
collection efficiency or because of the reaction of the lithium with the glass
fiber filter, therefore, Run 5 of Experiment No. 1 was conducted using only
front half Teflon filters and Run 4 of Experiment No. 1 was conducted using
three Method 5 trains and three impinger trains. The results from both sample
runs (see Table 3•**) clearly demonstrated that about half of many of the ions
present (Li, Br, Ca and Mg) was actually passing through the filter and being
3-6
-------
TABLE 3.3. TESTING SCHEDULE AT HUNTERS CORPORATION COOLING TOWER
Exper.
No.
1
1
1
1
'1
1
1
1
1
3
3
3
3
'I
'»
Run
No.
1
2
3
'1
1
5
6
7
8
1
2
3
'I
2
3
Date
(1986)
3/3
3/3
3/1
3/'l
3/5
3/5
3/6
3/6
3/6
3/6
3/6
3/7
3/7
3/8
3/8
Particle Size
Small
X
X
X
X
X
X
X
X
X
X
Large
X
X
X
X
X
No. of Trains
at Isokinetic
50%
2
2
2
2
2
2
2
2
2
2
2
100%
2
2
2
6
6
2
2
2
2
2
2
2
2
6
6
150J!
2
2
2
2
2
2
2
2
2
2
2
No. of Trains
at Nozzle Size
0.25
2
2
2
2
2
2
2
2
2
2
2
2
2
2
0.3
2
2
2
2
6
2
2
2
2
2
2
2
2
2
2
0-375
2
2
2
2
2
2
2
2
2
2
2
2
2
2
Filter Location
Front
6
6
6
6
3
6
Back
3
3
3
3
3
3
3
3
3
3
None
3
3
3
3
3
3
3
3
3
Compounds in Water
Li Br Na Cl Mo
X X
X X
X X
X X
X X
X X X X
X X X X
X X X X
X X X X
X X X X X
X X X X X
X X X X X
X X X X X
X X X X X
X X X X X
Filter Type
Glass
X
X
X
X
Teflon
X
X
X
X
X
X
X
X
X
X
X
uo
-------
TABLE 3.4. PERCENT SAMPLE RECOVERY BY TRAIN COMPONENT
Series of
Run Numbers
l-A+F-3
l-A+F-4
l-A+F-5
l-A+F-5
4-A.C.F-l
3-A+F-2.3+4
3-A+F-3
Train
Type
Method 5
Method 5
Method 5
Method 5
Method 5
Impinger
Train
Impinger
Train
Compound
Li
Li
Ca
Mg
Li
Li*
Br*
Percent Recovery (of Total Catch)
in Particular Train Compartment
91-52 in front-half, but reaction with
glass fiber filter
87.12 in front-half, but reaction with
glass fiber filter
69.92 in front-half
63.5% in front-half
49.3 in front-half. Teflon filter used
752 in first two impingers, 8.7% in
third impinger, 16.3/U on Teflon filter
98.92 in first two impingers, 0.72 in
third impinger, 0.42 on Teflon filter
*Efficiency results are not shown for collection of Ca and Mg in the impinger
train since the plant's distilled water that was used in the impinger reagent
contained high levels of Ca, Mg, and Na.
3-8
-------
TABLE 3.5. SUMMARY OF PMR and PMR RESULTS FOR LITHIUM ANALYZED BY GFAA
EL C
Series
of Run
Numbers
Sample
Train
Probe Avg.
Length (ft) AH
Nozzle Isokinetic
dia (in.) Rate (%)
Method 5 type train with Teflon filters at different isokinetic
l-A+F-5
4-A+F-l
4-A+F-2
4-A+F-3
l-A+F-6
l-A+F-7 ,
l-A+F-8
3-A+F-l
3-A+F-2
3-A+F-3
3-A+F-4
!A+B
C+D
E+F
Impinger
(A+B
C+D
E+F
Impinger
(A+B
C+D
E+F
Impinger
(A+B
C+D
E+F
Impinger
!A+B
C+D
E+F
2
4
6
type train
2
4
6
type train
2
4
6
type train
2
4
6
type train
6
4
2
1
3
2
at
2
3
2
at
5
3
1
at
1
3
2
.6
• 3
.8
100%
.8
.2
• 9
100%
.8
.0
.2
0.
0.
0.
37
31
25
isokinetic rate
0.
0.
0.
31
31
30
isokinetic rate
0.
0.
0.
37
30
25
different isokinetic
.6
.4
.9
0.
0.
0.
37
31
25
at different isokinetic
1
3
2
.6
.1
.8
0.
0.
0.
37
30
25
50
100
150
with same
100
100
100
Pollutant
PMR
a
Mass Rate*
PMR
c
rates and nozzle sizes
105
100*
182
nozzle size
82
100*
99
with different nozzle
100
100
100
rates and
50
100
150
rates and
50
100
150
87
100*
378
nozzle sizes
101
100*
136
nozzle sizes
107
100*
193
210
105
123
78
103
103
sizes
89
103
375
204
100
91
212
102
130
*Pollutant mass rate averages presented have been normalized to the average PMR 's
for the runs conducted at 100% isokinetic, which have been set equal to 100. a
The high and low values for all groups with more than one run have been dropped in
calculating the averages.
3-9
-------
collected in the impingers. These results clearly demonstrated that the front
filter was not acceptable for quantitively removing the compounds of interest.
3.2.2 PMRQ vs. PMRg for Large Particles
To increase the emissions in an effort to reduce the sampling time and to
evaluate PMRQ vs. PMRc for the larger particles, the particle size distribution
was shifted toward the larger particles. This was accomplished by partially
removing one of the mist eliminator panels.
Runs 6, 7 and 8 of Experiment No. 1 were then conducted using impinger
trains (see Table 3-6). The results for these runs again showed that the PMR
o
for the run conducted at 150% isokinetic was biased 36% high, having a
normalized value of 136.
3.2.3 PMRQ vs PMRc for Small Particles
After the test program, Munters Corporation had planned to add sodium
molybdate to the cooling water to assist in their corrosion problems. To
provide an additional element for analysis, molybdenum, the sodium molybdate
was added prior to the last day of testing. The final series of experiments,
Runs 2, 3, and l» of Experiment No. 3 and all of Experiment No. 4, was conducted
with a smaller particle size distribution. Since it was observed that the runs
showing a high bias (all conducted at 150% isokinetic) had all used the longer
probes and quarter-inch nozzles, the probes were switched for Runs 2, 3 and 4
of Experiment No. 3 to determine if the bias was a result of the longer probe.
Runs 2 and 3 of Experiment No. 4 used the probes in the orginal order to, if
possible, quantify the bias. The results shown in Table 3.7 indicate that the
trains run at 150% isokinetic with the shorter probes and quarter-inch nozzles
showed a high bias again. The results for Runs 2 and 3 of Experiment No. 4
(Table 3-8) showed a similar bias for trains run at 100% isokinetic with the
longer probes and quarter-inch nozzles. The resulting implication was that the
nozzles (illogically) were causing the bias. The results of the analysis for
molybdenum are not shown since the results indicated that a large portion of
the molybdenum was lost during the analysis.
3-2.4 Further Analysis of Sample by NAA
Since the bias of the results from runs conducted at 150% isokinetic could
not be explained, further analysis for bromine of three sets of samples were
conducted using Neutron Activation Analysis (NAA). This was done to determine
if the bias was due to the lithium concentration being so close to the detection
3-10
-------
TABLE 3.6. IMPINGER TYPE-TRAIN AT DIFFERENT ISOKINETIC RATES AND NOZZLE SIZES
Run
Number
l-A-6
l-B-6
l-C-6
l-D-6
l-E-6
l-F-6
l-A-7
l-B-7
l-C-7
l-D-7
l-E-7
l-F-7
l-A-8
l-B-8
l-C-8
l-D-8
l-E-8
l-F-8
Date
(1986)
3/6
3/6
3/6
3/6
3/6
3/6
3/6
3/6
3/6
3/6
3/6
3/6
3/6
3/6
3/6
3/6
3/6
3/6
Filter
Type
None
Teflon
None
Teflon
None
Teflon
None
Teflon
None
Teflon
None
Teflon
None
Teflon
None
Teflon
None
Teflon
Filter
Location
N/A
Back
N/A
Back
N/A
Back
N/A
Back
N/A
Back
N/A
Back
N/A
Back
N/A
Back
N/A
Back
Probe
Length (ft)
2
2
4
4
6
6
2
2
4
4
6
6
2
2
4
4
6
6
Avg.
AH
1.59
1.58
3.40
3.42
2.86
2.86
1.57
1.55
3-34
3-33
2.82
2.85
1.61
1.59
3-41
3-39
2.87
2.88
Nozzle
Dia (in.)
0.374
0.374
0.305
0.310
0.244
0.245
0.374
0.374
0-305
0.310
0.244
0.245
0.374
0-374
0.305
0.310
0.244
0.245
Isokinetic
Rate (%)
50.5
50.5
102.9
97.1
150.4
153-3
50.5
50.2
102.6
96.8
149.8
149-9
50.7
50.0
102.8
97-1
149-3
148.2
Pollutant Mass Rate of Lithium
PMR
Ib/ftr
x 10'6
161.0
220.6
238.4
99-9
224.6
160.0
246.0
131-6
83.6
149-9
430.1
95.2
138.9
115-2
200.3
133.2
291.8
270.4
PMR
Ib/hr
x 10"6
318.2
436.3
231.3
102.7
149.2
104.2
486.7
261.9
81.4
154.6
286.8
63.4
273-8
230.1
194.7
137-0
195-2
182.2
PMR *
normal .
95
130
141
59
133
95
211
113
72
128
368
82
83
69
120
80
175
162
PMR *
normal .
188
258
137
61
88
62
417
224
70
132
230
54
167
138
117
82
117
109
(JO
I
'Pollutant mass rate averages presented here have been normalized to the average of the PMR 's for the C and D
trains for each run, which was set equal to 100. a
-------
TABLE 3.7. IMPINGER-TYPE TRAIN AT DIFFERENT ISOKINETIC RATES AND NOZZLE SIZES
Run
Number
3-A-2
3-B-2
3-C-2
3-D-2
3-E-2
3-F-2
3-A-3
3-B-3
3-C-3
3-D-3
3-E-3
3-F-3
3-A-1
3-B-1
3-C-1
3-D-1
3-E-1
3-F-1
Date
(1986)
3/7
3/7
3/7
3/7
3/7
3/7
3/7
3/7
3/7
3/7
3/7
3/7
3/7
3/7
3/7
3/7
3/7
3/7
Filter
Type
None
Teflon
None
Teflon
None
Teflon
None
Teflon
None
Teflon
Teflon
None
None
Teflon
None
Teflon
None
Teflon
Filter
Location
N/A
Back
N/A
Back
N/A
Back
N/A
Back
N/A
Back
Back
N/A
N/A
Back
N/A
Back
N/A
Back
Probe
Length (ft)
6
6
1
1
2
6
6
4
4
2
2
6
6
1
4
2
2
Avg.
& H
1.53
1.51
3-01
3-02
2.74
2\ 75
1-57
1.51
3.07
3.06
2.77
2.79
1.61
1.58
3-17
3.16
2.78
2.77
Nozzle
Dia (in.)
0.371
0.371
0.305
0.301
0.211
0.215
0.371
0.371
0.305
0.301
0.211
0.215
0.371
0-371
0.305
0.301
0.211
0.215
Isokinetic
Rate (%)
50.5
19-8
98.1
95-8
119.6
H7-5
50.6
19-7
98.7
96.2
118.5
117.1
50.8
50.5
99.3
97-5
150.9
119.1
Pollutant Mass Rate of Lithium
PMR
Ib/nT
x 10"6
31-3
13-3
21.6
31.6
69.1
80.0
23.7
27.8
23.3
39-6
55-7
11.9
37-6
36.8
36.9
29.7
10.3
76.2
PMR
Ib/hr
x 10"6
67.9
86.9
21.9
36.1
16.1
51.1
16.8
55-8
23.6
11.1
37-1
28.5
71.0
72.7
37.1
30.1
26.7
51.0
PMR *
normal.
116
116
83
117
233
270
75
88
71
126
177
133
113
111
ill
89
121
229
PMR *
normal .
229
291
81
122
156
183
119
177
75
131
119
91
222
218
111
91
80
153
I
l-»
I\J
*Pollutant mass rate averages presented here have been normalized to the average of the PMR 's for the C and D trains
for each run, which was set equal to 100. a
-------
TABLE 3.8. IMPINGER-TYPE TRAIN AT 100% ISOKINETIC RATE AND DIFFERENT NOZZLE SIZES
Run
Number
4-A-2
4-B-2
4-C-2
4-D-2
4-E-2
4-F-2
4-A-3
4-B-3
4-C-3
4-D-3
4-E-3
4-F-3
Date
(1986)
3/8
3/8
3/8
3/8
3/8
3/8
3/8
3/8
3/8
3/8
3/8
3/8
Filter
Type
None
Teflon
None
Teflon
None
Teflon
None
Teflon
Teflon
None
None
Teflon
Filter
Location
N/A
Back
N/A
Back
N/A
Back
N/A
Back
Back
N/A
N/A
• Back
Probe
Length (ft)
2
2
4
4
6
6
2
2
4
4
6
6
Avg.
AH
5-73
6.15
3.05
3.05
1.19
1.19
5.6?
5.78
3.02
3-05
1.23
1.23
Nozzle
Dia (in.)
0.374
0.371*
0.305
0.304
0.244
0.245
0.374
0-374
0.305
0.304
0.244
0.245
Isokinetic
Rate (%)
100.7
98.0
102.8
96.1
103-6
97.9
99-9
96.2
98.9
96.6
102.7
99-2
Pollutant Mass Rate of Lithium
PMR
Ib/nV
x 10"6
19-6
18.0
40.5
7.4
128.6
70.5
19.6
18.8
17.2
23.5
94.1
45.7
PMR
Ib/hr
x 10"6
19.5
18.3
39.4
7-7
124.0
71.9
19-6
19-5
17.4
24.3
91.5
46.0
PMR *
normal.
82
75
169
31
537
294
96
92
85
115
462
225
PMR *
normal .
82
76
165
32
518
300
96
96
86
119
450
226
u>
H*
uo
•Pollutant mass rate averages presented here have been normalized to the average of the PMR
trains for each run, which was set equal to 100. a
s for the C and D
-------
limit of the analytical method. The complete results for the bromine analysis
by NAA are shown in Table 3-9- A summary of these results is presented in Table
3.10. The bromine results indicate that the bias was not due to any analytical
error as they agreed very well with the lithium results.
3.2.5 Precision of Method
The results expressed as PMR for runs conducted at 50% isokinetic and runs
a
at 100% isokinetic were very similar. Since it could not be determined why the
results (PMRQ) at 150% isokinetic were biased high, the decision was made to
determine the precision of the data and to determine if the calculated precision
was acceptable. Table 3.11 presents the percent differences for the paired
trains (Trains A and B, Trains C and D, and Trains E and F). The average
emission rates and precisions at a 95% confidence level were calculated for the
four trains (A, B, C and D) and for the six trains (A, B, C, D, E and F). These
results show that for the purpose of a compliance test which consists of three
runs conducted within 10 percent of the isokinetic rate (100 % isokinetic), the
average of the three runs should be within about 35% of the actual value
(determined by this reference method over numerous runs) with a 95% confidence
level. This precision of the sampling and analytical method does not account
for process variation. It has been demonstrated that the chromium test method
when applied to sources having greater chromium concentrations (i.e., hard
chrome plating operations) has the precision of EPA Method 5 (^10% with a 95%
confidence level).
3-2.6 Collection Efficiency of Sampling Train
For test series 3-2, 3-3, 3-4, 4-2, and 4-3, the impinger train collection
efficiency was checked. Analyses of the impinger samples were conducted for
several compounds; however, except in the case of lithium, the blank values were
too high to obtain reliable results. A summary of the analytical results for the
efficiency check is shown in Table 3.12. These results indicate that some
lithium was not collected by the sample train. The collection efficiency of
hexavalent chromium has been measured for the impinger train at other indus-
tries, and close to 100% collection efficiency was always observed. Based on
these results and the results of other studies conducted by EPA, the impinger
train should quanitatively remove all the chromium from the emissions sampled and
should be sufficient for collecting bromide (the recommended surrogate) which is
more soluble than lithium. Further checks will be made on future tests.
3-14
-------
TABLE 3.9. IMPINGER-TYPE TRAIN AT DIFFERENT ISOKINETIC RATES AND NOZZLE SIZES
ANALYSIS CONDUCTED BY NAA
Run
Number
3-A-l
3-B-l
3-c-l
3-D-l
3-E-l
3-F-l
3-A-3
3-B-3
3-C-3
3-D-3
3-E-3
3-F-3
4-A-3
4-B-3
4-C-3
4-D-3
4-E-3
4-F-3
Date
(1986)
3/6
3/6
3/6
3/6
3/6
3/6
3/7
3/7
3/7
3/7
3/7
3/7
3/8
3/8
3/8
3/8
3/8
3/8
Filter
Type
None
None
None
None
None
None
None
Teflon
None
None
Teflon
None
None
Teflon
Teflon
None
None
Teflon
Filter
Location
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Back
N/A
Back
Back
N/A
N/A
Back
Back
N/A
N/A
Back
Probe
Length (ft)
2
2
4
4
6
6
6
6
4
4
2
2
2
2
4
4
6
6
Avg.
AH
1.51
1.1*9
3-20
3-20
2.70
2.69
1.57
1.5*
3-07
3-06
2.77
2.79
5.67
5-78
3.02
3-05
1.23
1.23
Nozzle
Dia (in.)
0.374
0.371
0.305
0.310
0.244
0.245
0.374
0.374
0.305
0.304
0.244
0.245
0.374
0.374
0.305
0.304
0.244
0.245
Isokinetic
Rate (%)
50.3
50-3
102.1
95-5
148.5
147.0
50.6
49.7
98.7
96.2
148.5
147.1
99-9
96.2
98.9
96.6
102.7
99.2
Pollutant Mass Rate of Bromide
PMR
Ib/nV
x 10'6
780
973
584
865
1624
1194
203
519
388
453
1202
1082
436
452
554
653
1234
965
PMR
Ib/hr
x 10"6
1547
1945
571
905
1092
811
402
1043
393
470
808
735
435
469
560
675
1200
972
PMR *
normal .
108
134
81
119
224
165
123
92
108
286
257
72
75
92
108
204
160
PMR *
normal .
214
268
79
125
151
112
248
93
112
192
175
72
78
93
112
199
161
(JO
t->
ui
"Pollutant mass rate averages presented here have been normalized to the average of the PMR 's for the C and D trains
for each run, which was set equal to 100. a
-------
3-A+F-l
3-A+F-3
TABLE 3.10. SUMMARY OF PMR and PMR RESULTS FOR BROMIDE ANALYZED BY NAA
fl c
Series
of Run Sample Probe Avg. Nozzle
Numbers Train Length (ft) AH dia (in.)
Isokinetic
Rate (%)
Pollutant Mass Rate*
PMR PMR
a c
Impinger type train at different isokinetic rates and nozzle sizes
2
4
6
1.5
3.2
2.7
0.37
0.31
0.25
50
100
150
121
100*
195
102
132
Impinger type train at 1002 isokinetic rate and different nozzle sizes
6
4
2
1.5
3-1
2.8
0.37
0.30
0.25
50
100
150
123
100*
272
248
103
184
Impinger type train at 1QQ% isokinetic rate with different nozzle sizes
4-A+F-3 j
!A+B
OD
E+F
2
4
6
5-7
3-0
1.2
0.37
0.30
0.25
100
100
100
73
100*
182
75
103
180
"Pollutant mass rate averages presented have been normalized to the average PMR 's
for the runs conducted at 100* isokinetic, which have been set equal to 100. a
The high and low values for all groups with more than one run have been dropped in
calculating the averages.
3-16
-------
TABLE 3.11. CALCULATED PRECISION OF IMPINGER TRAIN RESULTS
Run Number and
Description
l-A+F-5
Different Isokinetic
Rates and Nozzle Sizes
l-A+F-6
Different Isokinetic
Rates and Nozzle Sizes
l-A+F-7
Different Isokinetic
Rates and Nozzle Sizes
l-A+F-8
Different Isokinetic
Rates and Nozzle Sizes
3-A+F-l
Different Isokinetic
Rates and Nozzles Sizes
3-A+F-2
Different Isokinetic
Rates and Nozzles Sizes
3-A+F-3
All 100% Isokinetic
Different Nozzle Sizes
3-A+F-4
All 1002 Isokinetic
Different Nozzle Sizes
4-A+F-l
All 100% Isokinetic
Different Nozzle Sizes
4-A+F-2
Different Isokinetic
Rates and Nozzle Sizes
4-A+F-3
Different Isokinetic
Rates and Nozzle Sizes
Relative Percent
Difference
A/B
C/D
E/F
A/B
C/D
E/F
A/B
C/D
E/F
A/B
C/D
E/F
A/B
C/D
E/F
A/B
C/D
E/F
A/B
C/D
E/F
A/B
C/D
E/F
A/C/E
B/D/F
A/B
C/D
E/F
A/B
C/D
E/F
21.1%
12.1%
20.4%
15-6%
40.9%
16.8%
30.3%
28.4%
63-8%
9-3%
20.1%
3-8%
23.2%
15-5%
19-0%
11.6%
16.9%
7-3%
8.0%
25-9%
14.1%
1.1%
10.8%
30.8%
23-5%
9-0%
3-4%
69-1%
29.2%
2.1%
15-5%
34.6%
Mean Values +/- Percent Range
at the 95% Confidence Level
(A+B+C+D)
(All six)
(A+B+C+D)
(All six)
(A+B+C+D)
(All six)
(A+B+C+D)
(All six)
(A+B+C+D)
(All six)
(A+B+C+D)
(All six)
(A+B+C+D)
(All six)
(A+B+C+D)
(All six)
(A+C+E)
(B+D+F)
(A+B+C+D)
(All six)
(A+B+C+D)
(All six)
74.3 +/- 19-96%
93-4 +/- 30.91%
180.0 +/- 34.20%
184.1 +/- 23.10%
152.8 +/- 43.72%
189.4 +/- 55-44%
146.9 +/- 24.69%
191.6 +/- 31.44%
63.6 +/- 22.32%
68.2 +/- 19.06%
34.2 +/- 21.89%
47-7 +/- 36.83%
28.6 +/- 26.08%
35-3 +/- 28.88%
35-3 +/- 10.33%
42.9 +/- 31.10%
62.2 +/- 32.56%
171.1 +/- 12.44%
21.5 +/- 63.18%
47-5 +/- 76.83%
19-8 +/- 13.26%
36.5 +/- 66.11%
3-17
-------
TABLE 3.12. PERCENT RECOVERY OF LITHIUM BY SAMPLE TRAIN COMPONENT
Run
No.
3-2
3-3
3-4
4-2
4-3
Percent of Total Sample Collected
by Sampling Train Component
Probe and first
two impingers
79%
78*
772
12%
63%
Third
impinge r
8%
5%
4%
16%
12%
Backup
filter
13%
17%
19%
12%
25%
3-2.7 Surrogates for Drift Testing
Several compounds and elements were analyzed in this testing program. Only
one was found to be acceptable to meet the criteria necessary for surrogate
analytes for cooling towers; this element was bromine. Based on a previous EPA
screening study, total chromium also appeared to be acceptable. Both should be
analyzed using Neutron Activation Analysis (NAA). NAA can detect chromium at
levels as low as 0.05 ug with an accuracy of +15% and bromide at levels of
0.0005 ug with an accuraccy of +10%. All other elements and analytical
techniques evaluated require a detection limit of approximately 1 ug of the
specific analyte in the sample and/or the element exists in sufficient
quantities in the ambient air to create a high background or blank level (which
would have to be subtracted from the emission rate). Table 3-13 presents the
elements and analytical methods evaluated during this sampling program and
recommendations for the use of each. When an element was not recommended as a
surrogate, the recommendation was based only on the results of this study. If
the tester has a more accurate analytical technique or is in an area that is
free of background contamination, an element may be acceptable.
3-2.8 Conclusions
The results obtained at Munters Corporation support the information
supplied by CTI and other experts in the field of cooling tower testing, that a
propeller anemometer should be used for measuring fan cell gas velocities and
that bromide would be a suitable surrogate for cooling tower drift emission
measurements. Based on a previous EPA screenig study, total chromium also
appeared to be suitable.
A recommended test method for measuring chromium emissions from cooling
towers (see Appendix B) has been developed based on testing the low level
3-18
-------
TABLE 3.13. ANALYTICAL TECHNIQUES AND ANALVTES EVALUATED
Element
Li
Li
Br
Br
Na
Ca
Mo
Analytical
Technique
Graphite
Furnace/Atomic
Absorption
Neutron
Activation
Analysis
Ion Chromato-
graphic
Neutron
Activation
Analysis
Inductively-
Coupled Argon
Plasmography
Inductively-
Coupled Argon
Plasmography
Inductively-
Coupled Argon
Plasmography
Quantifiable
Detection
Limit
0.1 ug
0.1 ug
1 ug
0.005 ug
1 ug
0.5 ug
0.05 ug
Present in
Ambient Air
Unlikely
Unlikely
Unlikely
Unlikely
Likely
Likely
Unlikely
Remarks
Lithium can react with glass to
form an insoluble compound
Sodium in cooling water makes NAA
of Li not acceptable
Other elements make low level
analysis difficult and detection
limit requires long sampling times
Bromine is one of elements better
analyzed by NAA
Both potential background levels
and high detection limit make
sodium a poor choice at low levels
Both potential background levels
and high detection limit make cal-
cium a poor choice at low levels
Molybdenum was lost in
sampling train or in sample
containers ; poor recovery
Recommendation
Not recommended due to
possible reactions
Unacceptable at low levels
Unacceptable at low levels
Acceptable
Unacceptable at low levels
Unacceptable at low levels
Unacceptable
UJ
(continued)
-------
TABLE 3.13. (Continued)
Element
Mo
Cr*
Cr
Analytical
Technique
Neutron
Activation
Analysis
Colorimetric
Neutron
Activation
Analysis
Quantifiable
Detection
Limit
0.05 ug
1 ug
0.05 ug
Present in
Ambient Air
Unlikely
Unlikely
Unlikely
Remarks
Molybdenum was lost in the
sample train or in the sample
containers; poor recovery
Detection limit requires a
sample collection time
which is too long
Chromium is one of the elements
better analyzed NAA
Recommendation
Unacceptable at low levels
Unacceptable at low levels
Acceptable
I
l\)
o
-------
drift emissions at the Munters Corporation and on numerous tests conducted at
other industries having chromium emissions. The draft method presented in
Appendix B utilizes a sampling train similar to that currently used by the
Cooling Tower Institute (CTI) for drift measurement.
The recommended method does not account for recirculation of the drift back
into the same tower or cross contamination by drift from surrounding towers.
Also, the background (ambient) contribution of chromium is not taken into
account since chromium should not be present in the ambient air. It would not
be practical to account for these cases, and each facility tested should use
its best judgement to minimize the same tower and cross-tower contamination.
3-2.9 Other Measurements
Ambient Measurements - A meteorological station equipped with a wind vane
and a cup anemometer was used to determine the wind direction and speed at the
cooling tower discharge. The relative humidity and ambient temperature were
determined from wet- and dry-bulb thermometers. A globe thermometer was used
to measure the mean radiant temperature. The barometric pressure was measured
by personnel at Page Field, an airport adjacent to the Munters facility.
DC-2 Droplet Counter - The DC-2 Hot Wire Droplet Counter is a prototype
instrument developed by KLD Associates, Inc. for the measurement of
concentration and size distribution of liquid droplets in a gas stream. This
instrument was used to collect data on the particulate size distribution in the
cooling tower discharge at the sampling location. The averages of all the runs
indicated that 85% by weight of droplets were >^50 microns and 97# by weight of
droplets were >28l microns in size. Problems were encountered with the
instrument during testing, and upon further discussion with KLD Associates it
was revealed that they did not consider the prototype reliable.
3-21
-------
4.0 PROCESS OPERATIONS
The Hunters Corporation evaluates the performance of components such as
fill and drift eliminators in simulated cooling tower configurations on a
contract basis and for product development in their crossflow and counterflow
cooling towers. A wide range of design specifications can be selected to
simulate operating conditions that exist in actual cooling tower
installations. The airflow and waterflow rates can be varied to achieve a wide
range of liquid-to-gas ratios, air velocities, and drift rates. Various types
and combinations of fill materials and drift eliminators can be installed to
mock various cooling tower configurations.
Testing was conducted on Hunters' cross-flow tower with high-efficiency
drift eliminators (D-15). The tower has only.a single cell as shown in Figure
5.1 (see Section 5.0). No blowdown was conducted during testing to conserve
the surrogate analytes. Makeup water was added between runs, as needed, to
keep the water at the proper level.
During the entire test series, the tower was operated at a recirculation
waterflow rate of about 250 gallons per minute and at a liquid to gas ratio of
approximately 0.66. The only process operating parameter that was not
representative of an actual high-efficiency cooling tower was the percent of
saturation. The Hunters pilot unit is designed for heating the water and
testing over shorter time intervals than were tested. As a result, the
discharge from the tower could not be kept near saturation. The gases exiting
the drift eliminator are computer monitored by the Hunters Corporation and a
summary of the monitoring results are presented in Table 4.1.
4-1
-------
TABLE 4.1. COOLING WATER AND AIR TEMPERATURES IN °F
Run
No.
1-1
1-2
1-3
1-1
4-1
1-5
1-6
1-7
1-8
3-1
3-2
3-3
4-2
Water Temperature
Enter Exit
.61.5
57-5
82.2
79-2
88.4
77-5
88.9
77-5
71.4
92.4
82.5
58.5
56.5
70.0
69.6
74.5
68.7
73.6
68.3
66.4
75-8
71.6
Ambient Temperature
Wet bulb Dry bulb
56.9
56.3
55-2
60.3
56.2
60.3
58.2
59-7
60.9
60.1
61.6
.— — ______ MHT
WOT
67.0
64.5
63.2
69-5
67.3
70.5
65-5
68.7
68.5
67-7
74.2
DETTHjnim— __.
RrrnRriF.n
Discharge Temperature
Wet bulb Dry bulb
58.0
57.0
65.2
67-2
69-3
66.9
69.3
66.5
65-1
72.2
68.9
63-3
62.2
66.2
69.8
70.5
70.5
70.2
68.4
68.3
72.6
73.0
4-2
-------
5.0 SAMPLING LOCATIONS AND TEST METHODS
This section describes the sampling locations and test methods used to
characterize emissions from the crossflow cooling tower at the Munters
Corporation's test facility in Fort Myers, Florida. The cooling tower exhaust
stack was used as the sampling location for the testing to measure the drift
emissions using the surrogate analytes lithium and bromine (introduced into the
cooling water as LiBr), the drift size distribution, and the exhaust velocity.
In addition, cooling water samples were taken from the recirculation basin for
background analysis of the bromine content. A meteorological station was
located approximately twelve feet from the exhaust stack sampling location to
monitor ambient temperatures, wind speed, and wind direction. The relative
positions and the type of testing conducted at each location are shown in the
simplified process flow diagram (see Figure 5-1) and accompanying Table 5.1.
The subsections which follow further describe each sampling location and the
applicable test methods.
5-1 COOLING TOWER EXHAUST STACK (Sampling Location A)
Sampling for lithium and bromine, as well as the drift sizing and velocity
determination, was conducted at the crossflow cooling tower exhaust stack
(Sampling Location A) shown in Figure 5-2. Sampling probes were introduced
into the flow from the exit of the exhaust stack, and nozzles were located
approximately two inches below the exhaust plane. Based on a preliminary
three-dimensional pitot probe velocity traverse, a single point along the
east-west diameter of the exhaust stack was determined to have a typical
misaligned flow angle (30°-60°). All sampling probes were placed within one
inch of this point (except the particle sizing probe; it was located on the
same diameter roughly opposite this point). The axial component of flow was
continuously monitored using a propeller anemometer located approximately three
inches above the sampling point.
Method 5- and impinger-type trains were used for lithium and bromine sample
collection. Sampling was conducted at a single point for approximately 120
minutes per run. A total of six sampling probes and trains were used during
each run allowing different nozzle sizes, probe lengths, filtering techniques,
and isokinetic flow percentages to be used within a single run in order to
assess the relative merits of these sampling equipment variables.
5-1
-------
ARM
VATERFROM
BASIN AND PUMP
RECRCULATION
BASN
D-15DRFT
ELMHATOR
FIGURE 5 • 1 AIR FLOV DIAGRAM OF MUNTERS CORPORATION CROSSFLOV
COOLN6 TOVER SHOVING SAMPLING LOCATIONS.
5-2
-------
TABLE 5.1. SAMPLING PLAN FOR HUNTERS CORPORATION
Sample Type
Lithium
Bromine
Drift Size
Distribution
Drift Size
Distribution
Velocity
Determination
Cooling Water
Samples
Meteorological Data
Sampling
Location
A
A
A
A
A .
B
C
Number
of Runs
15
15
3
15
Single
Point
Continuous
15 Pair
Grab
Samples
Continuous
Methods
EPA Method 5- or Impinger-type
train with ICAP Analysis Off -site
EPA Method 5~ of Impinger-type
train with 1C and NAA Analysis
Off-site
Cyclone
Modified Hot-Wire Anemometer
Propeller Anemometer
Specific Ion
Electrode Bromine
Analysis On-site and ICAP Off-site
Cup Anemometer,
Directional
Anemometer, Globe
Thermometer, Wet & Dry Bulb
Thermometers
5-3
-------
VELOCITY TRAVERSE POIHTS
2 AXES
12 POINTS/AXIS
24 TOTAL POINTS
PARTICLE
SENG
ROOF
61 DIA.
MULTPLE
TRAIN
SAMPLING
TEST PLATFORM
?
FATION
FIGURE5-2 TOP VEV OF CROSSFLOV COOLING TOVER EXHAUST STACK,
MUNTERS CORPORATION (SAMPLING LOCATION A).
-------
5.2 COOLING WATER BASIN (Sampling Location B)
At the start and completion of each emissions test run, two cooling water
samples were taken from the recirculation water basin. These samples were
taken by hand and stored in 500 ml glass jars. Analysis of each sample for
bromine content was conducted on-site using a specific ion-electrode and
off-site using inductively-coupled argon plasmography (ICAP).
5.3 AMBIENT METEOROLOGICAL STATION (Sampling Location C)
A meteorological station was assembled and operated continuously 12 feet
from the exhaust stack in order to quantify ambient conditions at the time of
sampling. Wind speed and direction were monitored using a cup anemometer and
directional anemometer and recorded on a strip chart recorder. Ambient wet
bulb/dry bulb temperatures were obtained using a psychrometer and the effects
of radiant energy were quantified using a globe thermometer (black-body
thermometer).
5.4 VELOCITY AND GAS TEMPERATURE
A three-dimensional pitot tube and a bank of magnehelic gauges were used to
measure the gas velocity pressure (delta P) and both the yaw and pitch flow
angles. Velocity pressures and angles were measured at 12 sampling points
along each of the two traverse diameters (as per Method 5D) as shown in Figure
5.2, to determine an average flow velocity and angle at each point. The
temperature at each sampling point was measured using a thermocouple and
digital readout. In addition, a propeller anemometer was used to determine a
total flow velocity in the axial direction at each of these points; the results
were then compared to the results of the 3-D pitot probe traverse.
5-5 MOLECULAR WEIGHT
Flue gas composition was essentially that of the ambient air drawn into the
cooling tower via the vane axial fan. Therefore, the dry molecular weight and
composition of air was used.
5.6 LITHIUM AND BROMINE
Method 5 sampling procedures, as described in the Federal Register,* were
used with the Method 5- and impinger-type trains to measure lithium and bromine
*40 CFR 60, Appendix A, Reference Methods 2, 3, and 5, July 1, 1980.
5-5
-------
emissions at Sampling Location A. All tests were conducted at constant
percentages of isokinetic (50, 100, and 150) by regulating the sample flow rate
relative to the flue gas flow rate as measured by the propeller anemometer.
Sampling trains consisted of a heated, glass-lined probe, and either (1) a
3-inch nominal diameter glass fiber or Teflon filter and a series of
Greenburg-Smith impingers (two containing 100 ml of deionized-distilled water,
one empty and one with silica gel), or (2) a series of Greenburg-Smith
impingers (two containing 100 ml of deionized-distilled water, one empty and
with with silica gel) with a 3-inch nominal Teflon filter located between the
third and fourth impinger. A distilled water rinse of the nozzle, probe,
appropriate filter holder portions, and impingers of the sample train was made
at the end of each test. This rinse was added to the impinger and filter
sample. The entire sample was typically concentrated to approximately 10 ml.
The lithium content of the cooling tower exhaust samples was determined by RTI
(Entropy's subcontractor under the EMB contract) using the inductively-coupled
argon plasmography or the graphite frunace/atomic absorption technique for
sample analysis. Bromine content was measured by RTI using ion chromato-
graphy. See Appendix B for the detailed draft test methods.
5-6
-------
6.0 QUALITY ASSURANCE
Because the end product of testing is to produce representative emission
results, quality assurance is one of the main facets of stack sampling.
Quality assurance guidelines provide the detailed procedures and actions
necessary for defining and producing acceptable data. Two such documents were
used in this test program to ensure the collection of acceptable data and to
provide a definition of unacceptable data. These documents are the EPA Quality
Assurance Handbook Volume III, EPA-600/4-77-027 and Entropy's "Quality
Assurance Program Plan," which has been approved by the U. S. EPA, EMB.
Relative to this test program, the following steps were taken to ensure
that the testing and analytical procedures produce quality data.
• Calibration of field sampling equipment. (Appendix E describes
calibration guidelines in more detail.)
• Checks of train configuration and calculations.
• On-site quality assurance checks of sampling train components.
• Use of designated analytical equipment and sampling reagents.
Pre- and post-test calibrations were performed for each of the meter boxes
used for sampling. Calibrations were also performed for the temperature
sensing equipment, nozzles, anemometer sensor, and the entire propeller
anemometer apparatus. Appendix E includes the calibration data sheets for
each dry gas meter used for testing and data sheets for the calibrations of
the other sampling equipment mentioned.
Audit solutions were used to check the analytical procedures of the
laboratories conducting the lithium, bromine, molybdenum, calcium, and
magnesium analyses. Table 6.1 presents the results of these analytical
audits. The audit tests show that the analytical techniques were good.
The sampling equipment, reagents, and analytical procedures for this test
series were in compliance with all necessary guidelines set forth for accurate
test results as described in Volume III of the Quality Assurance Handbook.
6-1
-------
TABLE 6.1. AUDIT REPORT ANALYSIS
Plant:
Task No.:
Date Samples Received:
Samples Analyzed By: A^7\Z"
Reviewed By: P.
Date Analyzed:
/ C7T A fa-l.
EPA
A/AA
^0.7-
SRh
Mo
/V/W
SAM
Nkk
6-2
------- |