EPA-650/2-73-022
September 1973
ENVIRONMENTAL PROTECTION TECHNOLOGY SERIES
-------
STATE OF THE ART: 1971
INSTRUMENTATION FOR
MEASUREMENT OF PARTICULATE EMISSIONS
FROM COMBUSTION SOURCES
VOLUME IV: EXPERIMENTS & FINAL REPORT
by
Gilmore J. Sem and John A. Borgos
Thermo-Systems, Inc.
1500 N. Cleveland Avenue
St. Paul, Minnesota 55113
Contract Number CPA 70-23
Program Element No. 1AA010
EPA Project Officer: John O. Burckle
Control Systems Laboratory
National Environmental Research Center
Research Triangle Park, N.C. 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
September 1973
THERMO-SYSTEMS INC. (Printed in U.S.A.)
-------
11
When U. S. Government drawings, specifications, or other data
are used for any purpose other than a definitely related
Government procurement operation, the Government thereby
incurs no responsibility nor any obligations whatsoever, and
the fact that the Government may have formulated, furnished,
or in any way supplied the said drawings, specifications,
or other data is not to be regarded by implication or other-
wise, or in any manner licensing the holder or any other
person or corporation, or conveying any rights or permission
to manufacture, use, or sell any patented invention that may
in any way be related thereto.
References to names commercial products in this report are
not to be considered in any sense as an endorsement of the
product by the Government.
This report has been reviewed by the Environmental Protection
Agency and approved for publication. Approval does not signify
that the contents necessarily reflect the views and policies of
the Agency, nor does mention of trade names or commercial
products constitue endorsement or recommendation for use.
THERMO-SYSTEMS INC.
-------
iii
TABLE OF CONTENTS
VOLUME IV
Page
FOREWORD V1
ABSTRACT viii
SECTION 1. EXECUTIVE SUMMARY: AUTOMATIC MONITORS OF PARTICULATE MASS
EMISSIONS FROM STATIONARY FOSSIL-FUEL COMBUSTION SOURCES 1
1.1 INTRODUCTION i
1.2 DEFINITION OF STACK PROPERTIES 2
1.3 NON-APPLICABLE SENSING METHODS 4
1.4 BETA RADIATION ATTENUATION 6
1.5 PIEZOELECTRIC MICROBALANCE 10
1.6 ELECTROSTATIC METHODS 14
1.6.a. ION CAPTURE 14
1.6.b. CONTACT CHARGING 18
1.7 LIGHT TRANSMISSION 20
1.8 REVIEW OF EXPERIMENTAL WORK 24
1.8.a. LABORATORY EXPERIMENTS 24
l.S.b. FIELD EXPERIMENT STATION 25
I.8.C. FIELD EVALUATION OF TWO PROTOTYPE BETA INSTRUMENTS . 26
1.9 CONCLUSIONS AND RECOMMENDATIONS 26
1.10 ACKNOWLEDGMENTS 27
1.11 BIBLIOGRAPHY 28
DETAILED REPORT:
SECTION 2. PRELIMINARY LABORATORY EXPERIMENTS 31
2.1 INTRODUCTION 31
2.2 CALIBRATION REPEATABILITY 31
2.3 EXPERIMENTS WITH PROMETHIUM-147 39
THERMO-SYSTEMS INC
-------
iv
TABLE OF CONTENTS (continued)
Page
2.4 RADIATION SOURCE CHARACTERISTICS 39
2.5 EXPERIMENTS WITH FLY ASH 47
2.6 SUMMARY 51
SECTION 3. DESIGN OF FIELD EXPERIMENT STATION 52
3.1 INTRODUCTION 52
3.2 NSP POWER PLANT 52
3.3 GENERAL LAYOUT 54
3.4 INSTRUMENTATION 54
3.5 SUMMARY 60
SECTION 4. CALIBRATION OF STACK FACILITY AT THE FIELD EXPERIMENT
STATION 61
4.1 PROCEDURE 61
4.2 RESULTS OF FILTER TESTS 62
4.2.a. INTRODUCTION 62
4.2.b. GAS FLOW MEASUREMENT 68
4.2.C. EFFECT OF SAMPLING TIME 71
4.2.d. PARTICLE LINE LOSSES 74
4.2.e. PARALLEL FILTER TESTS IN MAY 76
SECTION 5. FIELD EVALUATION OF TWO PROTOTYPE BETA INSTRUMENTS .... 78
5.1 PROCEDURE 78
5.2 RESULTS AND DISCUSSION 78
SECTION 6. EVALUATION OF TRANSMISSOMETER TECHNOLOGY 83
6.1 INTRODUCTION 83
6.2 RECOMMENDED DESIGN FEATURES 83
6.3 MINIMUM RECOMMENDATIONS GOVERNING INSTRUMENT INSTALLATION
AND OPERATION 84
THERMO SYSTEMS INC
-------
V
TABLE OF CONTENTS (continued)
Page
6.4 PRESENT COMMERCIAL INSTRUMENTS 85
6.5 MEANING OF THE MEASUREMENT 87
6. 6 CORRELATION OF THE MEASUREMENT WITH MASS 88
6.7 CONCLUSIONS 89
6.8 ACKNOWLEDGMENT 90
6.9 REFERENCES 90
SECTION 7. APPENDICES 91
APPENDLX A. COMPLETE DATA FOR THE AUGUST TESTS OF THE SAMPLING
SYSTEM 92
APPENDIX B. COMPLETE DATA FOR THE MAY TESTS OF TWO BETA
INSTRUMENTS "
THERMO-SYSTEMS INC
-------
FOREWORD
Airborne particulate matter is a major air pollutant having significant
effects on health, economics, ecology, visibility, and aesthetics. Effective
techniques and hardware systems for source emissions measurement are needed
for application to the various sources to achieve control of particulate
emissions and thus protect the environment.
This contract work, begun before the institution of the EPA, has been
sponsored to completion by the National Environmental Research Center at
Research Triangle Park, North Carolina for the purpose of the field evaluation
of two research prototype instrument systems based upon beta radiation atten-
uation as applied to the measurement of emissions from a coal-fired power plant.
This report constitutes the last of a four volume series which, taken to-
gether, comprise the final report for the contract effort. These volumes have
been issued separately as each phase of work was completed to make them avail-
able to the public as soon as possible. These volumes contain the following:
Volume I of this report was written for the engineer or planner
who needs to know a few basic facts about a particulate mass
measurement technique and wishes to minimize the time required
to obtain this information. Volume I is intended for use as a
quick reference guide. This volume is available from the National
Technical Information Service (NTIS), order number PB 202-665.
Volume II of this report is designed as a detailed in-depth report
on operating principles, techniques, historical data, and discussion
of the more viable techniques for particulate mass monitoring.
Volume II is designed for the plant engineer, abatement and control
officials, and others who may not be familiar with the detailed
technology of these areas. Included are sections on power plant
emissions properties and extraction sampling probes. This volume
is available from the NTIS, order number PB 202-666.
Volume III of this report is a comprehensive survey and critique
of particle sizing techniques which could possibly form the basis
for automatic particle size measurement. This volume is available
from the NTIS.
Volume IV of the report (this volume) provides a recapitulation of
the various sensing techniques and their applicability to mass monitor-
ing instrumentation and describes an experimental field evaluation of
two research prototype instruments based on beta radiation attenuation
as applied to a coal-fired power plant effluent. Problem areas requir-
ing further research and development efforts are identified for those
persons concerned with instrument development.
THERMO-SYSTEMS INC
-------
VII
This report is included in the Environmental Protection Technology
series - the series, devoted to new and improved technology required for
control and treatment of pollution sources to meet environmental quality
goals, includes reports of work dealing with research, development, and
demonstration of instrumentation, equipment, and methodology to repair
or prevent environmental degradation from point and non-point sources of
pollution.
John 0. Burckle
Project Officer
U.S. Environmental Protection Agency
Office of Research and Development
THERMO-SYSTEMS INC.
-------
viii
ABSTRACT
As this program began, no automatic particle mass concentration monitor
had been suitably developed for use in monitoring effluents from large coal-
fired combustion sources. It was not clear which of several measurement
techniques offered the greatest chance of success as a result of further
development programs. It was also not clear what operational problems would
be encountered in the use of such a monitor on an actual effluent stream.
The first phase of this program was a literature study and evaluation
of potential measurement techniques and a more precise definition of the
stack environment. Volumes I and II of this report series and Section 1
of this report contain the results of this study. The second phase was an
experimental evaluation of the most promising technique: beta radiation
attenuation. Sections 2-5 of this report discuss both laboratory and field
experiments performed with several beta radiation instruments.
Section 1 discusses various approaches for automatic, continuous measure-
ment of the rate of particulate mass emissions from smoke stacks on large
coal and oil combustion facilities. Reasons are given for rejecting a number
of particle sensing techniques. Inherent and practical strengths and weak-
nessess of the following methods are included: beta radiation attenuation,
piezoelectric microbalance, electrostatic, and light transmission. Comparative
evaluation shows that the beta radiation attenuation and piezoelectric micro-
balance techniques offer the best possibilities for accurate monitoring at the
present time. Although the electrostatic and light transmission techniques
offer several desirable operational features, their inability to sense the mass
of particles severely limits their usefulness for monitoring particulate mass
emissions. Section 1 includes a summary of Volumes I and II of this report
series.
Section 2 describes early laboratory tests performed with the sensing head
of a Gelman beta radiation instrument. The tests resulted in a set of calibra-
tion curves, each curve representing a different initial clean filter thickness.
The importance of knowing the initial clean filter thickness accurately is
illustrated. This section discusses the reasons for the different calibration
curves for different initial clean filter thicknesses. Tests were performed
with C-14 and Pm-147 as beta sources. Pm-147 is affected significantly less
than C-14 by variations in initial clean filter thickness. For C-14, a commonly-
encountered filter thickness variation of about 10% results in a deposit measure-
ment uncertainty of about 10%, even with the standard before-and-after radiation
counting technique.
Section 3 describes a sampling facility, designed specifically for the
evaluation of the instruments which measure particle mass concentration of
stack effluents, which was designed, constructed, evaluated, and operated on
a section of breeching of a coal-fired power plant. Identical parallel sampl-
ing systems supply nearly identical particulate samples to a high-efficiency
THERMO-SYSTEMS INC
-------
IX
filter and to the instrument under test. The sampling facility includes heated
sampling lines and boundary-layer diluters to reduce condensation and the result-
ing particle wall losses. The sampling line leading into the test instrument
can be adapted to a variety of lengths and configurations; and the sampling flow
rate, dilution ratio, and heater power can be adjusted. Each of the two parallel
sample lines contains a high-accuracy flowmeter and a valving system which allows
the operator to measure both dilution clean air flow rate and total diluted
aerosol flow rate with the single flowmeter. Thus, most flowmeter errors arc
eliminated resulting in more accurate measurement of sampling rates and dilution
ratios. The facility was not designed nor intended to obtain an optimum measure-
ment of effluent particle concentration, but rather to deliver representative and
identical effluent samples to the test instrument and its parallel filter.
Section 4 describes results of the calibration of the sampling facilities
in the field experiment station under several normal operating conditions with
identical high-efficiency filters in both parallel sampling lines. The ratio,
R.,, (instrument (beta) filter collected weight)/(parallel filter collected weight)
was near or somewhat less than 1.0 for nearly all runs.
Section 5 describes tests of two prototype instruments with beta radiation
particle mass sensors and with filter collectors which were tested for a short
time in the stack facility. Both were developed for auto exhaust particle
measurements under separate and parallel contracts funded by EPA. One was
developed by GCA Corporation (GCA)* and one by Industrial Nucleonics Corporation
(IN)**. The GCA instrument yielded highly consistent results and R^ was nearly
1.0. The IN instrument yielded considerably less consistent results with R_,
usually varying between 0.33 and 0.65. The sampling systems up to the two
instruments were plumbed and operated nearly identically, evidence that the error
of the IN instrument was instrument error and not sampling technique error. The
highly encouraging measurements with the GCA instrument indicate that beta
radiation sensing with filter collection of particles is a strong candidate for
the measurement of particle mass concentration in smoke stacks.
Section 6 presents candid comments regarding the state-of-the-art of
commercial transmissometers in January 1971. Recommendations regarding the
design, installation, and operation are included. The section briefly dis-
cusses what a transmissometer measurement means and how well the measurement
correlates with particle mass concentration.
*GCA Technology Division, Bedford, Massachussetts 01730
**Industrial Nucleonics Corporation, Columbus, Ohio 43202
THERMO SYSTEMS INC
-------
-1-
SECTION 1. EXECUTIVE SUMMARY: AUTOMATIC MONITORS OF PARTICUIATE MASS
EMISSIONS FROM STATIONARY FOSSIL-FUEL COMBUSTION SOURCES
1.1 INTRODUCTION
Measurement of the total mass flow of particulate emissions from smoke
stacks on large combustion facilities is a major problem facing stack owners,
pollution control officials, and emissions control equipment manufacturers.
Stack owners need such measurements to more carefully control the combustion
efficiency and emissions of the facility. Pollution control officials need
permanent records of stack mass emissions to aid in law enforcement. Emission
control equipment manufacturers need more efficient ways to evaluate the per-
formance of their equipment. As a result, instruments that automatically and
continuously record particulate mass emissions are now necessary.
Presently, particulate emissions measurements are made by sampling a
known volume of effluent gas through a filter, and weighing the filter before
and after sampling to find the particle mass concentration. By traversing the
effluent gas stream cross-section to measure gas velocity as well as to obtain
the filter samples, an estimate of the total mass of particles emitted per hour
is made. Such measurements require considerable equipment, labor, and time.
One series of measurements typically takes about four hours with perhaps three
man-days of labor for planning, equipment transport, setup, sampling, and data
reduction. Therefore, this method cannot economically be used for measuring
combustion efficiency, for monitoring pollutant emissions, or for extensively
evaluating pollution control equipment on a continuous basis.
There are several reasons for choosing mass as the measured particulate
emission parameter. Mass is a basic parameter of the particles which does not
depend on the instrument used to measure it. A familiar laboratory balance can
be used to check or calibrate mass sensing instruments. Since manual stack
sampling techniques measure the mass of particles, much of the existing data,
and most new regulations, are expressed in terms of particulate mass. As will
be seen below, mass is one of the easiest particle parameters to sense accurately
because most mass sensing techniques are not very sensitive to secondary par-
ticle parameters.
The primary aim of this discussion is not to defend the case for measuring
particulate mass, however. It is rather to explain the relative merits of
several particle sensing techniques with regard to the measurement of particulate
mass or mass concentration.
Particle sensing techniques have many monitoring applications ranging from
clean rooms to pneumatic conveying systems. Only a few of the techniques are
applicable to the sensing of particles in effluent gas streams from coal com-
bustion sources. Fewer yet are applicable to the sensing of particulate mass
THERMO-SYSTEMS INC
-------
-2-
in such streams. The measurement of primary interest from a pollution stand-
point is the particulate mass flow rate, i.e., the total mass of particles
leaving the stack per hour. Most particulate mass sensors, on the other hand,
measure particle concentration in terms of the mass of particles per unit of
gas volume. This discussion assumes that the measurement of particulate mass
concentration is sufficient to define particulate mass flow rate. The volu-
metric gas flow rate must be measured separately or estimated from the operat-
ing conditions of the process.
There is a question about the relative merits of integrated versus
point measurements of particle concentration. The most commonly used particle
monitors, light transmissometers, are integrating instruments. They measure
the average particle concentration along the measuring path. Most other potential
and existing particle monitoring instruments measure the particle concentration
at a point within the duct. If the gas velocity and particle concentration
profiles were relatively homogeneous across the duct, either integrated or point
sampling could yield good results with little trouble. However, such conditions
seldom exist. Not only is the gas velocity profile usually skewed, but the
particle concentration profile may be skewed in a different way. Thus, it is
not clear which sampling method has the advantage. One conclusion is certain,
however: the placement of any instrument within an effluent duct must be done
carefully so that a representative measurement is made.
1.2 DEFINITION OF STACK PROPERTIES
The operating environment strongly affects the design of any instrument for
measuring particulate emissions. Table 1 is a brief summary of typical effluent
gas stream conditions for large, modern coal combustion facilities with electro-
static precipitator control equipment^. The data was obtained through an exten-
sive survey of the open literature and private reports. Table 2 is a brief
summary of typical effluent gas stream conditions for large, modern oil combustion
facilities with no control equipment. Since the effluent from oil combustion
facilities is relatively free of particles and since only a small portion of the
electric power in the U.S. is generated by such plants, the remainder of the
discussion is directed primarily toward coal combustion facilities.
Particulate emissions from coal combustion facilities are usually measured
in the breeching, a section of rectangular duct between the collector equipment
and the vertical stack. Because the breeching is usually quite short and is
seldom straight, ideal sampling conditions seldom occur. The choice of the
sampling location is very important. Most manual sampling procedures specify
that the duct being sampled must be traversed in a specified manner. This makes
the choice of a sampling location even more critical. This problem should be
considered carefully whenever emissions measurements are made. In most cases,
the accuracy of the measurement depends as much on the representativeness of
the sample as on the accuracy of the sensor.
THERMO-SYSTEMS INC
-------
-3-
TABLE 1. TYPICAL COAL COMBUSTION EMISSIONS DATA
Particulate mass loading, after precipitator
before precipitator
Mass loading spatial variation at duct cross-section
Particle size, after precipitator
before precipitator
Extreme particle size range
Flue gas velocity
Flue gas temperature
Dew point
Moisture content of gas
Static pressure at sample ports
Turbulent flow fluctuations
Existing sampling port size
Traversing distance across duct from port
*2.29 gm/cu meter = 1 grain/cu foot
0.03 - 3.0 gm/cu meter*
0.2 - 12 gm/cu meter
± 5(^
Mass median diameter -5pm
95% <25ym (by mass)
Mass median diameter ~3-70ym
95% <100ym, 5%
-------
-4-
1.3 NON-APPLICABLE SENSING METHODS
The following discussion briefly describes many of the particle sensing
techniques which are not presently applicable to particulate mass monitoring
from coal combustion sources and lists reasons for rejecting each one. Later,
the applicable techniques will be covered in more detail, including the positive
and negative features of each.
Acoustic attenuation is the decrease in amplitude of a sound wave when par-
ticles are added to an acoustic field. Use of this technique has been strictly
limited to research^. Acoustic attenuation appears useful only after considerably
more research and then only for concentrations several orders-of-magnitude greatei
than the 0.1 gm/cu meter found in effluent gas streams.
The pressure drop across a_ nozzle increases if particles are added to the
carrier gas. This technique has been successfully used only for measuring par-
ticle flow in pneumatic conveying systems with particle concentrations three
orders-of-magnitude greater than the typical values found in effluent gas
streams-*. A cyclone has been used to enrich the particle concentration of
industrial dust emissions before it enters the sensing nozzle, but large fluc-
tuations in the enriched particle concentration leaving the cyclone severely
limited the instrument's usefulness^.
The pressure drop across a filter increases as the filter becomes loaded.
However, the rate of increase of pressure drop depends not only on particle con-
centration but also on particle size, shape, and stickiness, on particle penetra-
tion into the filter, and on filter characteristics. Because of the many variabl
the relationship between the particulate mass loading and pressure drop is too
erratic to make accurate measurements.
The unbalance of a. centrifuge measured by a displacement sensor as particle;
are deposited at one spot on the circumference is a direct measure of particle
mass-5. Another closely related method is the change in inertia of a rotating ma;
caused by the addition of particles, which can be sensed by measuring changes in
acceleration or deceleration of the rotating mass under the influence of a con-
stant torque. Both methods offer some promise for monitoring effluent gas streai
primarily because they sense mass directly. However, neither method has been
developed. A possible problem is lack of sensitivity.
An acoustic particle counter detects the audible click as individual par-
ticles greater than 5 - 10 Um pass through a laminar capillary**. This method,
about which little is known, requires low, clean-room concentrations, making it
unusable in effluent gas streams.
A hot-wire anemometer detects the number and size of liquid droplets which
impact on the surface of a hot cylindrical film^. The response of the instru-
ment to solid particles is not understood, so the method is unsuitable for stacl
monitoring.
THERMO SYSTEMS INC
-------
-5-
A vibrating band or wire stretched across a portion of an effluent gas
stream vibrates at a decreasing natural frequency as it becomes contaminated
with particles". The frequency change is directly proportional to the mass
of particles added to the band. However, the mechanism by which particles
reach the band depends on particle size, particle stickiness, and aerodynamic
factors. Correlation with particulate mass flow is questionable. Other methods
of particle deposition, such as impaction or electrostatic precipitation, could
make this method similar to the piezoelectric microbalance method discussed
below. Considerably more development is needed on this method.
An automatic weighing instrument uses an automated gravimetric balance to
sense the mass of particles deposited by electrostatic precipitation, impaction,
or other collection methods^. The deposit surface must be cleaned or replaced
periodically, requiring complicated mechanisms. The cost of potential commercial
models appears high. Considerable development is needed to make this method
useful. Since it gravimetrically measures true particulate mass, further develop-
ment appears justified.
The tape spot photometer measures the blockage or attenuation of a light
beam shining through an indexing tape filter as the filter becomes loaded with
particleslO. The measurement is often called the soiling index. Correlation
of this measurement with particulate mass concentration is poor because the
attenuation of light does not depend on particle mass, but on particle size,
shape, refractive index, and surface characteristics, on filter variations, and
on light-beam wavelength, intensity, and geometry.
Light-scattering photometers, or nephelometers, measure the light scattered
by a cloud of suspended particles within a measuring chamberlljl^ The measure-
ment correlates poorly with particulate mass concentration because light scatter-
ing depends on particle size, shape, refractive index, and surface characteristics
and on light-beam wavelength, intensity, and geometry.
Lidar uses a laser light source and measures the light scattered from the
particles back toward the source!-3. Lidar measures particle concentrations
remotely (>5m) from the instrument, making installation in an effluent gas stream
difficult, but offering the possibility of monitoring emissions from several stacks
with one instrument located in some central location outside the stacks. Lidar
suffers all the problems of light-scattering photometers in correlating its
measurement with particulate mass. The method is presently expensive and requires
more development for use in the remote monitoring of stack emissions.
Single-particle light-scattering detects light scattered by individual par-
ticles as they pass single file through a light beam^^lS. This method requires
relatively low concentrations (<5,000 particles/cm3 all of which are >0.3 ym
diameter) in addition to the problems suffered by the light-scattering photometer.
THERMO-SYSTEMS INC
-------
-6-
Holography, a three-dimensional laser photographic technique, can be used
to remotely measure the total volume of a sample of suspended particles^, it
requires further development and is much too expensive for practical monitoring
purposes at this time. The primary expense is in reducing the data.
A particle bounce technique has been tested which consists of an electrical
condenser with a strong electric field between the plates of the condenser^-?.
When a particle enters the air gap between the plates, the electric field accel-
erates it toward one side or the other, depending on its initial charge. When
it strikes a plate, it exchanges its charge and is repelled toward the other
plate. This process continues until the particle is carried out from the con-
denser by the air flow. Unfortunately, such instruments appear to sense only
large (>10ym), non-adhesive particles, making correlations with mass highly
questionable.
The probe-in-nozzle technique consists of a cone-shaped, electrically-in-
sulated probe that is placed in the throat of a venturi nozzle-^. Particles
strike the probe and exchange electrical charge with it. The charge exchange
is sensed by an electrometer. This technique is sensitive only to dry, abrasive
particles and is dependent upon particle composition. The measurement does not
correlate well with particulate mass.
1.4 BETA RADIATION ATTENUATION
If beta particles (electrons) pass through a medium, some will be absorbed
and some reflected, resulting in a net reduction in the beam intensity. Such a
reduction is known as beta radiation attenuation and is a measure of the mass of
material through which the beam passes. The attenuation of beta particles de-
pends statistically on the number of electrons with which they interact. Corre-
lation of attenuation with particulate mass depends on the relationship between
the number of electrons per molecule (atomic number) and the mass of the molec-
ule (atomic number) and the mass of the molecular nucleus (atomic weight) of
the particulate material. This ratio is between 0.4 and 0.5 for all elements
except hydrogen, and is between 0.45 and 0.5 for nearly all elements normally
found in coal combustion particles. Figure 1 19 verifies that beta attenuation
is not significantly different for a wide variety of particulate materials.
The close correlation with particulate mass appears to be better than any other
known technique except automatic weighing and piezoelectric microbalance measure-
ments.
Instruments using this technique, shown schematically in Figure 2, have
recently measured the concentration of airborne particles in ambient air and
effluent gas streamsl9,20,21,22_ Carbon-14, with a half-life of 5568 years,
is a typical beta radiation source; thallium-204, cesium-137, and promethium-147
could also be used. Geiger-Muller (GM) counters are the most common detectors,
but proportional and scintillation counters have also been used. The particles
THERMO-SYSTEMS INC
-------
H
m
50
m
n
20 30 40 50
AREA DENSITY (mg./sq.cm.)
Figure 1. Calibration of a beta radiation attenuation
instrument for several dusts.
-------
K
1
THERMO- SYSTEMS INC.
/
/
N
ISOKINETIC
SAMPLING
PROBE
0.
1
FLUE
GAS
K.
/
^-DUCT WALL
s
SAMPLE
CONDITIONER
I I ._
n i i
i
O\1
FILTER TAPE-^
\ I
1 I
OUTPUT
INDICATOR
I DETECTOR
Jr
S ADI
^BETA SOURCE 60 L
r«^
^ 1 1
M
1 XI 1 1
^
VALVE FLOWMETER
RECORDER
3ROXIMATELY
ITERS/MINUTE
f^ | ^
i r
PUMP
Figure 2. Beta radiation attenuation instrument.
i
00
-------
9
from a known volume of effluent gas are usually first collected on a filter,
which is then placed between the beta source and GM detector. The difference
in count rate of the GM detector before and after the particles are collected
is a measure of the mass of particles on the filter. This method has been
automated by using an indexing tape filter as the collector and a recorder to
collect data. Strictly speaking, practical beta instruments are only quasi-
continuous. Sensitivity is high enough so that an average measurement, when
the sampling flow rate is 60 liters per minute (2 cu ft/min) through a glass-
fiber filter, can be made in a 1-to 15- minute period. This is sufficient for
most monitoring purposes.
Some measurement errors are associated with source, specimen, and detector
geometry. These errors can usually be minimized with good design, and the
system can be calibrated to reduce measurement uncertainty to an acceptable
value. The source and detector must be shaped and spaced such that beta radi-
ation scattered by the particle sample is not detected. The gas molecules with-
in the beam of beta particles must not become a significant portion of the total
mass through which the beam passes. Finally, the particles must be distributed
uniformly on the filter so that the measured filter mass loading is not dependent
on the position of the radiation beam.
The sampling system which brings the particle sample to the sensing head
can be the greatest source of error with this method. This error can also be
minimized by careful design, including simple heating of the probe and filter
holder to prevent condensation of vapors, or dilution of the effluent sample
to prevent condensation as the sample cools. Since the particle sensor is
normally separated from the particle collector, the radiation source and detec-
tor are not exposed to the high temperature gas stream.
Several particle collection techniques other than filters can be used with
beta radiation attenuation. A cyclone can collect particles with typically
higher gas flow rates providing better time resolution^. However, Table 1
shows that a signficiant portion of the mass of particles in coal combustion
emissions is below the particle size cutoff of practical cyclones (~lym), in-
troducing some uncertainty to measurements made with such instruments. An
impactor can collect particles larger than about 0.5 ym, giving a highly con-
centrated sample on a small deposit area^l. However, large particles may tend
to become reentrained in the air stream because of the high jet velocities
needed to collect small particles. Electrostatic precipitation has also been
used to collect particles for beta attenuation sensing23. Cyclones, impactors,
and electrostatic precipitators all offer advantages for application to certain
particle concentration measurements. Further investigation is necessary to
fully evaluate each collection method for coal combustion emissions.
The beta radiation attenuation particulate mass sensing technique, with one
of the several possible particle collection devices, is a promising contender
for monitoring the particulate mass concentration in stack emissions.
THERMO-SYSTEMS INC.
-------
-10-
1.5 PIEZOELECTRIC MICROBALANCE
Piezoelectricity is the property of certain crystals, including quartz,
which results in an electrical charge on certain surfaces of the crystal when
the crystal becomes mechanically stressed. Conversely, a piezoelectric mate-
rial becomes mechanically strained if an electrical charge is placed on cer-
tain of its crystal faces. A piezoelectric crystal, when placed in an appro-
priate electronic oscillating circuit, will cause the circuit to oscillate at
the natural vibrational frequency of the crystal. Some piezoelectric materials,
such as quartz, vibrate at very precise natural frequencies, so that frequency
changes of one part in 10 million are significant and easily detectable.
When foreign material adheres to the surface of a vibrating piezoelectric
crystal, the natural frequency of vibration of the crystal decreases. The
magnitude of the frequency change is directly proportional to the mass of
foreign material. This principle has been used recently to measure the mass
of ambient atmospheric and automobile exhaust particles deposited onto the
sensing surface by an electrostatic precipitator or an impactor24,25, 26^
The vibrational mode normally used in piezoelectric microbalances is one
in which the two parallel faces of a plate-like crystal move parallel to each
other. With this type of vibration, using type AT crystals as shown in Figure
3, particles will be weighed if they adhere to any point on the two electrodes.
A particle must be deposited on one of the two metal-film electrodes to be
sensed because only the portion of the crystal between the electrodes vibrates.
The relationship between added mass and the shift in natural vibrational fre-
quency for AT crystals is^:
Am = -- - Af (1)
o
where: 2
K = constant which depends only on crystal type (= 2.27 rr for
type AT crystals), ygm
Af = change in natural vibrational frequency, Hz,
f = natural vibrational frequency of the crystal MHz,
o
2
A = electrode (active) area of the crystal, cm , and
Am = mass added to electrode area, ygm.
THERMO-SYSTEMS INC
-------
-11-
.10cm DIA METALLIC
FILM ELECTRODES
0.33mm
SEE DETAIL BELOW
ELECTRICAL LEAD
1.90cm
NATURAL FREQUENCY=5MHZ
'//,'
/,' 0.33mm
Figure 3. Typical type AT quartz crystal for particulate
mass measurement showing the characteristic
thickness shear mode of vibration.
THERMO-SYSTEMS INC
-------
-12-
If a collection device, such as an electrostatic precipitator or impactor,
deposits particles onto the crystal electrode from a constant volume flow
rate Q during the sampling period At, then the average particle mass con-
centration C over the time interval At is:
ave
r = ^ (2)
ave Q At E k ;
where E is the particle mass collection efficiency of the collector. Combining
Equations 1 and 2, the particle mass concentration becomes:
A Af
2.27QEfo2 At
Thus, for a given electrode and a given collection device, particle mass con-
centration C is directly and nearly linearly proportional to the rate of
change of crystal frequency with time -TJ. Strictly speaking, Equation 3 is
somewhat non-linear since f appears in the denominator. However, since Af
is normally measured in tens or hundreds of Hz and f is normally 3-10 MHz,
any non-linearity is completely negligible.
Particles must adhere to the crystal electrode, i.e., the active, sensitive
portion of the crystal, if they are to be weighed. Thus the forces causing
particles to stick to the surface must be high enough with respect to the inertial
forces acting on the vibrating particles so that particles do not roll or slide
on the electrode surface. It appears that most particles smaller than about
10 pm diameter in ambient air adhere well enough to be weighed. Three modifi-
cations which may result in the weighing of larger effluent particles are:
choosing crystals with lower vibrating frequency, reducing vibrational amplitude,
and using various coating techniques to enhance the adherance of particles to the
crystal electrode.
A piezoelectric microbalance system for use in monitoring effluent particles
would consist of the components shown in Figure 4. Since the sensor samples
only a few liters per minute, an auxiliary vacuum system is necessary to remove
a representative sample from the effluent stream and deliver it to the sensor.
A method of cleaning or replacing loaded crystals is necessary because crystals
stop oscillating when over-loaded with particles. The output signal can be
conditioned so that it becomes directly proportional to particulate mass con-
centration. It then can be recorded by either digital or analog recorders.
The high temperature of effluent gas streams and the vapor condensation
which occurs while cooling the system are possible problems for piezoelectric
microbalances. Two possible solutions exist. The first is to find a type of
crystal which has a low temperature dependence and which operates at the gas
THERMO SYSTEMS INC
-------
N-
ISOKINETIC
SAMPLING
PROBE
/
/
H
m
73
2
m
/
FLUE
GAS
V
RECORDER
DKINETIC
WIPLING
PROBE
S. .
5
CON
FREQUENCY
MONITOR
AMPLE
DITIONER
OSCILLATOR
CIRCUIT
^"""""^^"^
_^-*
_ i *,
--^
-DUCT
WALL
PARTICLE
COLLECTION
REGION
PUMP
QUARTZ
CRYSTAL-
FLOWMETER
VALVE
PUMP
APPROXIMATELY
150 LITERS/MINUTE
APPROXIMATELY
I LITER/MINUTE
Figure 4. Piezoelectric microbalance instrument.
-------
-14-
stream temperature. The entire sampling probe and sensor could then be heated
to that temperature to prevent vapor condensation. The second solution is to
dilute the extracted sample with cool air. This would prevent condensation of
vapors and provide a more suitable environment for the crystals. Since this
closely simulates what actually happens when effluents reach the ambient air,
this method may be desirable to anyone wanting to measure the pollution hazard
of the stack gas. The piezoelectric microbalance has enough sensitivity to
measure typical effluent samples diluted by factors of 2 - 100 within a few
seconds.
Other than its ability to sense the true mass of particles, the most impor-
tant feature of this technique is its high sensitivity. Like the beta technique,
the piezoelectric technique is quasi-continuous. However, a piezoelectric micro-
balance can detect mass changes of less than 0.005 Ugm. Assuming an effluent
particle concentration of 0.1 grams/cu meter (approx. 0.05 grains/cu ft), a
piezoelectric microbalance sampling 1 liter/min can collect enough material for
a measurement accurate to within + 5% in well under 1 second. A beta radiation
attenuation instrument using a filter collector and sampling at 60 Uters/min
(2 cu ft/min) requires about 1-15 minutes.
Although the piezoelectric microbalance is relatively new and untried in
effluent gas streams, its desirable features, particularly its high sensitivity
to the direct sensing of particulare mass, make it a promising candidate for
automatic particulate mass monitoring. A more complete evaluation awaits
further development and testing.
1.6 ELECTROSTATIC METHODS
Electrostatic particle sensing methods include a number of instrument
designs using several distinctly different principles. The following dis-
cussion includes the three principles most applicable to particle monitoring
in effluent gas streams. The first two methods, variations of ion capture,
are quite similar and are discussed together. The third method, contact
charging, is a completely different technique.
1.6.a. Ion Capture
Imagine a constant supply of unipolar ions flowing perpendicularly across
a stream of airborne particles toward the grid of an electrometer. Some of
the ions will strike the particles and be carried away, thus reducing the ion
current as measured by the electrometer. The reduction in ion current, known
as the ion-current attenuation, is a measure of the particle flow rate.
27
Figure 5 shows a typical instrument design . A radioactive source, e.g.,
8 microcuries of cobalt-60, distributed evenly around the inner wall of the
outer tube, forms a convenient, constant ion supply. Ions of one polarity are
drawn across the aerosol stream and ions of the opposite polarity are repelled
THERMO-SYSTEMS INC
-------
H
m
JO
2
9
tin
c/i
m
2
t/5
_J\ K
ISOKINETIC
SAMPLING
PROBE
I
\r
HEATED
TUBE
PARTICLE CHARGING
AND COLLECTING REGION
COBALT-60
ELECTRODES
HIGH VOLTAGE
TRANSFORMER
RECORDER
ELECTROMETER
Figure 5. Ion-current attenuation instrument.
-------
-16-
by a low-intensity, radial electric field. An electrometer measures the ion
current reaching the central electrode. The difference between measured current
with particles passing through and the current with clean gas passing through is
a measure of particle concentration. A later design28 uses two identical chambers,
one with particles passing through and one with a high efficiency filter on the
inlet allowing only gas to pass through. The difference between the two measured
current levels is a measure of particle concentration. This technique has been
used successfully with low concentration dusts such as atmospheric aerosol.
Rather than measuring the portion of the total ion current which does not
attach itself to particles, as does the ion current attenuation method, the
second ion capture method measures the other portion of the total ion current -
the portion carried away by particles. Figure 6 shows one possible design using
electrostatic precipitation to collect the charged particles2^. This design
uses a periodic blast of clean scavenging air to clean the device. The effluent
gas stream aspirates the sample through the instrument. It therefore responds
directly to changes in gas velocity as well as particle concentration, yielding
a measurement of particle flow rate. The design shown in Figure 6 has been
operated in coal-fired effluent ducts. However, correlation of the measurement
with mass is questionable.
It appears that three major problems limit the usefulness of the two ion
capture techniques for measuring particulate mass in effluent gas streams.
First, the instruments do not sense particle mass or even volume, making
their correlation with mass poor in effluent gas streams where large fluctuations
in particle size probably occur. The ion current carried away by particles
depends primarily on particle size and number concentration. Thus, if particle
size and the charging mechanism remains constant, a calibrated instrument can
measure particle number concentration. Particle chargers can be designed, with-
in limits to place a constant, or saturation, charge on particles of a given
size. However, particle size does not remain constant within a given effluent
gas stream. Indeed, accurate measurements are most needed when something
changes in the process, and most such process changes cause particle size to
fluctuate. A look at the theory of particle charging30 shows that the saturation
charge level on a particle above 0.1 pm is proportional to D for diffusion
charging and D2 for field charging. Most instruments of thil type reported in
the literature appear to use field charging, causing the instrument to respond
most closely to changes in the product of D2 and particle number concentration,
in other words, to changes in total particli surface area. This makes ion
capture instruments extremely sensitive to fluctuations in the concentration
of submicron particles^!.
Second, as with any electrostatic method, the measurement of low current
levels (10-6 - l
-------
-17-
-Nr
r-SCAVENGING
\ AIR INLET
\ I
ISOKINETIC
INLET
NOZZLE
FLUE
GAS
HEATED
TUBE
PARTICLE
CHARGING
SECTION
PARTICLE
COLLECTING
SECTION
RECORDER
D.C. AMPLIFIER
HIGH VOLTAGE
TRANSFORMER
HEATER
Figure 6. Ion capture instrument.
THERMO-SYSTEMS INC
-------
-18-
Third, instruments become contaminated within a short time because the
electric field in the charging section causes particles to deposit on the
walls of that section. Not only does the charging efficiency change if con-
tamination becomes excessive but particles which collect in the charging
region rather than pass through the instrument represent a measurement error.
When monitoring dry, non-sticky particles, a periodic blast of air may be
sufficient to clean the contaminated surfaces.
The first problem is inherent with particle charging methods. No particle
charging technique exists in which the charge placed on a particle is pro-
portional to its mass. The other two problems are basically design problems
which may be eliminated by further development.
Ion capture instruments do pose several advantages which make these diffi-
culties less serious. As was mentioned above, the design shown in Figure 6 has
been used for monitoring particulate effluents from coal-fired sources. There-
fore, many basic operational problems have been solved. If the particle size
distribution and composition is known, the output of an ion capture instrument
can be correlated with mass concentration. Furthermore, such an instrument
does afford a truly continuous and automatic recording of the dust concen-
tration; beta radiation attenuation and piezoelectric microbalance, on the
other hand, are only quasi-continuous.
Although ion capture does not measure particulate mass, useful particle
monitoring instruments may develop from this technique. Particulate mass
measurements, however, will require other sensing techniques.
1.6.b. Contact Charging
When particles hit or slide along a surface, there is usually an electrical
charge transfer between the surface and the particles. This principle, a form
of contact charging, has been used in the design of the Konitest32 shown in
Figure 7. This instrument consists of an electrically-floating tube through
which airborne particles pass in a swirling, helical path. The entraice is a
tangential slot which gives the air and particles the helical motion. The
particles slide along the tube wall, causing an electrical charge transfer.
An electrometer measures current draining from the tube wall.
At first glance, the Konitest appears to share the same problems as the
previously discussed probe-in-nozzle technique, which is highly dependent on
the charge transfer characteristics of the particles. Indeed, data shows a
strong dependence on particle composition, and several reports indicate that
submicron particles adhere to the tube wall and change the calibration of the
instrument. However, several extensive experimental instrument evaluations
within coal-fired effluent gas streams report surprisingly good correlation
with gravimetric particulate mass concentration measurements^*22. in fact,
they also report few operational problems with the instrument, in contrast
to several other reports.
THERMO-SYSTEMS INC.
-------
-19-
r
FLUE
GAS"
ELECTRICALLY
FLOATING TUBE
RECORDER
D.C. AMPLIFIER
/ / / / 7^f^^
SECTION ~
Figure 7. Principle of the Konitest instrument.
THERMO-SYSTEMS INC.
-------
-20-
The Konitest cannot be thoroughly evaluated theoretically because the
theory of contact charging is not well understood. A partial explanation for
the apparently good correlation with particulate mass concentration may be
that the force causing particles to hit the tube wall, namely centrifugal force,
depends strongly on particle mass.
The Konitest offers nearly instantaneous reponse with direct analog output.
Unless the particles are sticky, it is self-cleaning. This is a major practical
advantage. The instrument has reportedly been used in several coal combustion
effluent streams with no particle adhesion problem. The Konitest has simple
construction and operation, and, therefore, low installation and operational
cost. If it truly does measure a parameter which correlates well with partic-
ulate mass, a question which has yet to be fully answered, it will be a very
useful instrument for mass emissions measurement.
1.7 LIGHT TRANSMISSION
When a beam of light is directed through a particle-laden gas stream, its
intensity is reduced. This attenuation is a function of many variables, in-
cluding: particle concentration, size, shape, refractive index, and surface
characteristics; light wavelength and orientation; and sensor geometry, orien-
tation, and sensitivity. Because there are so many variables, the physical
laws governing light attenuation are extremely complex.
Past development has been concerned primarily with optimizing instrument
design parameters, such as the light source and sensor, so that particle con-
centrations could be measured with a minimum of interference from the other
variables. However, development for the true measurement of particulate mass
concentration has either not been tried or has been unsuccessful.
Light transmission is presently the most popular method of monitoring par-
ticle loadings in effluent gas streams. Figure 8 shows the basic configuration
of the simplest type of light-transmission instrument. A light source, mounted
on one side of the duct, beams light across the duct to the light sensor. The
sensor, or photocell, collects only the light that is not obscured by the par-
ticles. Its output signal is calibrated to read Ringelmann number or equivalent
opacity. The portion of the beam through which particles are allowed to pass is
usually 1-3 meters long. The remainder of the beam is usually enclosed by a
pipe into the stack to protect the light source and detector from contamination.
The list of advantages of such an instrument is impressive. The measure-
ment is made entirely within the gas stream; no sample extraction with its
accompanying problems is necessary. The measurement is instantaneous and con-
tinuous in real time. The apparatus is simple and easy to understand, and
little maintenance is necessary. The electrical readout is easy to record con-
tinuously. Practical problems such as contamination of the source and sensor
windows can be easily solved.
THERMO SYSTEMS INC
-------
CONDENSER
LENS
m
TO
2
9
tin
oo
m
2
Z
o
PIN HOLE
LIGHT
SOURCE
COLLIMATOR
LENS
DUST PARTICLES-
PHOTOCELL
PINHOLE
RECORDER
AMPLIFIER
Figure 8. Principle of light-transmission measurements.
-------
-22-
Unfortunately, this measurement technique does not measure particle mass.
Rather, it measures the light opacity of the particle cloud. The instrument
can be calibrated to give particle mass for a given set of conditions of par-
ticle size, shape, refractive index, surface charactertistics, and density,
but if any of these characteristics changes, as often happens in any effluent
gas stream, the calibration is no longer valid. The time when accurate partic-
ulate measurements are most needed is precisely when such changes occur.
22
Figure 9 shows particulate mass concentration calibration curves for two
of the best light-transmission instruments for which such data exists. The two
instruments were mounted on a modern power generating plant with a pulverized
coal boiler and electrostatic precipitation particle collectors. Gravimetric
filter samples were used as the calibration standard. About 300 calibration
measurements were made with each instrument. Each measurement consisted of a
comparison of a five-minute average of the instrument reading with a filter
sample obtained during the corresponding five-minute period. Measurements were
made with three plant operating conditions: 1) with soot blowing, 2) without
soot blowing, and 3) with the plant operating at minimum load (about 50% of
capacity). Soot blowing is a common procedure in which blasts of steam or air
are used to clean soot off the heat exchanger, resulting in a substantial change
in effluent particulate properties. The heavy lines in the curves represent the
best fit calibration from data points. The numbered lines represent the best
fit calibration for each of the three plant operating conditions.
Figure 9 shows significantly different mass loading calibrations for the
three plant operating conditions. All methods which do not directly sense par-
ticle mass share this problem. Each instrument installation requires a separate
calibration for each different operating condition. The calibration also must
often be checked (every few months or so) to be sure no change has occurred.
The cost of these calibrations raises the total instrument cost above the potential
cost of direct mass sensing instruments such as beta radiation attenuation and
piezoelectric microbalance instruments. In addition, the instrument operator
must find which plant operating conditions are in use in order to know which
calibration is correct.
With light-transmission instruments, the only alternative to such high
maintenance and operations costs is to disregard the effects of the changing
plant operating conditions and accept the poor correlation of the instrument
with particulate mass concentration. The errors involved with typical mass
loadings (less than 200 milligrams per cubic meter) are shown to be quite
substantial even for the best instruments. Figure 9 represents the best partic-
ulate mass correlation found by the authors in the literature. The errors would
no doubt be worse with instruments of lesser quality. Therefore, light-trans-
mission instruments do not measure particulate mass concentration with acceptable
accuracy in their present state of development.
THERMO-SYSTEMS INC
-------
TRANSMISSOMETER A
TRANSMISSOMETER B
400
H
m
70
2
m
2
a:
Ul
f-
UJ
o
I 200
CO
tr
400
oc.
UJ
K-
Ul
o
Q
fe 200
CO
0 20
INSTRUMENT READING
(ARBITRARY UNITS)
40
20 40
INSTRUMENT READING
(ARBITRARY UNITS)
LEGEND: I-WITH SOOT BLOWING; 2-wiTHOUT SOOT BLOWING;3-MINIMUM LOAD
Figure 9. Results of an experimental calibration of two
light-transmission instruments in a modern coal-
fired power plant.
GJ
-------
-24-
Several possible variations to light-transmission techniques exist which
may improve the correlation with particulate mass concentration. One such
variation involves the use of several monochromatic light beams with differing
wavelengths. By proper data reduction, some particle size distribution informa-
tion could be obtained. The particle concentration measurement could then be
partially corrected for any size changes. Such changes, although easily auto-
mated, would significantly increase the cost of the instrument. The improvement
in particulate mass correlation with such an instrument cannot be estimated at
this time. However, correlation would still be poor in many cases since only
the error due to variations in particle size is reduced. Errors due to vari-
ations in particle shape, refractive index, surface characteristics, and density
are not reduced.
1.8 REVIEW OF EXPERIMENTAL WORK
Since one of the significant conclusions reached during the state-of-the-
art study early in this program was that beta radiation attenuation was the
most promising technique for measuring particulate mass concentration in stacks,
experiments to uncover problem areas and to evaluate two prototype beta instru-
ments were conducted. The emphasis was on evaluation of the beta technique
itself, not on the features of a particular instrument design.
l.S.a. Laboratory Experiments
In preliminary laboratory experiments using a Geiger-Muller detector, two
radiation sources, promethium 147 and carbon 14, were compared with each other
and with theory. Several important results emerged.
First, it has been generally assumed in the past that the relationship
between beta attenuation and particle loading can be expressed accurately by:
I/I = exp [-y X] (1)
o m
where:
I = intensity of beta radiation passing through a clean filter,
I = intensity of beta radiation passing through a loaded filter,
2
X = weight of particles on the filter, mg/cm , and
u = calibration constant, usually assumed to be independent of
m everything except radiation source.
Our experimental results, explained in Section 2 of this report, show that
I/I is not a simple linear exponential and significant errors (greater than + 102
can result from assuming Equation 4 to be correct.
THERMO-SYSTEMS INC.
-------
-25-
Secondly, we found experimentally that y in Equation 4 depends on I .
Stated differently, the calibration of a beta instrument depends on the initial
filter thickness. Details are shown later in Section 2 of this report. Un-
certainties in excess of + 5% can result (in addition to the errors caused by
the non-linear exponential relationship) using ordinary filter tapes with + 5%
variation in clean filter thickness, using C-14 as a beta source, and making the
initial beta count (I ) on the exact same spot of filter paper as the final
beta count (I).
Third, although Pm-147 is somewhat less sensitive than C-14, uncertainties
caused by variations in the initial filter thickness, using Pm-147 are about
half as great as for C-14. Thus, it appears that Pm-147 may be slightly superior
to C-14 as a beta radiation source from a technical standpoint. However, C-14
has less restrictive licensing laws making it easier to use and transport.
All three of the above-mentioned anomalies can be corrected on-line by using
data from our extensive calibration of the instrument in a small computer program.
Further work of this type is needed to identify other possible errors such as
changes in particulate composition. Such work could further illuminate the
magnitude of such errors in practical sampling applications.
l.S.b. Field Experiment Station
Because of the lack of a controlled facility to accurately simulate the
stack environment, we designed, constructed, and tested a particle sampling
facility on a 550-megawatt, coal-fired power plant owned by Northern State Power
Company and located at Bayport, Minnesota. Not enough information is known at
the present time about the stack environment to construct a simulation test
facility to obtain information about the practical ability of instruments to
obtain accurate particle mass concentration data. Although the samples reaching
the test instruments in our facility are not truly representative of the conditions
in the stack, the test samples are quite representative of samples obtained by
a typical sampling system.
The sampling system removes a heated stream of effluent from the stack,
splits the sample into 2 identical parts, and passes the 2 samples through
identical boundary-layer dilution coolers to the 2 test instruments. The
stream is heated to a temperature above the acid dew point to prevent con-
densation on the particles and on the sampling tubes. Since the instruments
were not capable of high temperature operation, boundary-layer dilution cool-
ing caused the effluent stream to cool while condensation particle losses
were prevented by keeping the particles from the walls of the system until
sufficient dilution had occurred. The 2 test instruments received identical
samples so that any differences in measurement could be attributed primarily
to instrument differences.
THERMO-SYSTEMS INC
-------
-26-
Tests of the sampling facility using a pair of identical 47-mm Nuclepore
filters in identical filter holders identified acceptable operating conditions
and showed a ratio R^ of nearly 1.0 where R^ is the ratio of (test instrument
filter collected weight)/(reference parallel filter collected weight). The
consistency of the results is good with the standard deviation of Rp for all
acceptable operating conditions being less than 0.05.
l.S.c. Field Evaluation of Two Prototype Beta Instruments
Two prototype beta instruments, developed by GCA Corporation and Industrial
Nucleonics, Inc., under EPA contracts, were evaluated in the field sampling
facility. Each was compared with a reference parallel 47-mm Nuclepore filter
in a modified Gelman filter holder. The particle mass concentration measured
by the GCA instrument agreed very well with the reference parallel filter with
R =0.98 with a standard deviation of 0.04. The particle mass concentration
measured by the Industrial Nucleonics instrument did not agree well with the
reference parallel filter with R generally between 0.33 and 0.65. Consistency
was also poor.
1.9 CONCLUSIONS AND RECOMMENDATIONS
1. The highly consistent accuracy of the GCA instrument in our limited
tests verifies the primary finding of the state-of-the-art study
conducted earlier under this contract. Beta radiation attenuation
is presently the best technique for automatically measuring effluent
particulate mass concentration with the present state of development.
Beta radiation attenuation actually senses a particle parameter very
closely related to mass. One measurement takes only 1-15 minutes.
Some of the other experimental results indicate that more experimental
evaluation and development remains if an accurate, reliable particle
mass concentration monitor is to result. Engineering design improve-
ment of present commercial models appears to be needed, especially
in the sample extraction probe, sensor geometry, and particle
collection technique. Beta radiation attenuation will probably
remain one of the most favorable particulate mass monitoring methods
for some time in all three potential use areas: continuous monitoring
by stack owners, by pollution abatement personnel, and by control
equipment evaluators.
2. The piezoelectric microbalance technique detects particle mass
directly. It could soon replace beta radiation attenuation as
the most favorable and most accurate automatic particulate mass
monitor for effluents, especially for measurement of low con-
centrations such as downstream of efficient control equipment
and such as for measurement of particle mass size distributions
THERMO SYSTEMS INC
-------
-27-
in the smaller size ranges. It remains to be proven whether the
instrument can be made to operate with coal combustion effluent
particles. The adherence of larger particles to the crystals
appears to be the major potential problem. This technique is
similar in many ways to the beta attenuation technique, but has
the additional feature of higher sensitivity, which should allow
significantly faster measurements (probably one measurement every
few seconds).
3. The Konitest electrostatic contact charge technique reportedly
shows surprisingly good correlation with gravimetric particulate
mass concentration in several extensive experimental tests per-
formed on coal combustion sources. It is known that particle
composition and instrument contamination strongly affect the
measurements. It is not clear whether these factors are problems
in coal combustion effluents. Since the theoretical basis for
the measurement is largely undefined, further testing is necessary
to evaluate the Konitest for particulate mass monitoring in any
coal combustion source. The instrument's primary features are its
nearly instantaneous response, continuous readout, and simple con-
struction.
4. Light-transmission and light-scattering techniques do not measure
particulate mass concentration. Although light-transmission instru-
ments are the most commonly used particulate emissions monitors to-
day, they only measure something related to the visual appearance
of the stack plume. However, since they are simple, inexpensive,
and easy to operate and maintain, they will probably remain popular
as emissions monitors even though they do not measure particle mass.
Although a number of particle parameters, such as size and shape,
affect the particle concentration measurement, light-transmission
and light-scattering instruments can be calibrated to give rough
measurements of particulate mass emissions under a given constant
set of conditions, but the calibration is no longer valid when
conditions change. Therefore, these instruments have serious
limitations in their use as monitors of particulate mass con-
centration in effluent gas streams. Several interesting variations
of light-transmission instruments have been suggested. Although
they may improve the correlation with particle mass, no way can be
seen to fully overcome the invalidation of mass calibration when
particle properties change.
1.10 ACKNOWLEDGEMENTS
The authors thank Dr. John G. Olin for his many helpful suggestions and for
critically reviewing an early manuscript of this section. The authors also thank
Professors K. T. Whitby and B.Y.H. Liu, J. P. Pilney, N. Barsic, and F. D. Dorman
for many helpful suggestions and much useful advice. Special thanks go to the
Project Officer, J. 0. Burckle for his continued interest in the work and for
his critical review of the manuscript of this report.
THERMO SYSTEMS INC
-------
-28-
1.11 BIBLIOGRAPHY
1. Sem, G. J., Borgos, J. A., Olin, J. G., Pilney, J. P., Liu, B.Y.H.,
Barsic, N., Whitby, K. T., and Dorman, F. D., "State of the Art:
1971 Instrumentation for Measurement of Particulate Emissions from
Combustion Sources, Volume I: Particulate Mass - Summary Report,
Volume II: Particulate Mass - Detail Report", Thermo-Systems Inc.,
St. Paul, Minn., report to EPA under Contract CPA 70-23 (1971).
2. Dobbins, R. A., and S. Tempkin, Journal of Colloid and Interface
Science, 25, 329 (1967).
3. Beck, M. S., and N. Wainwright, Powder Technology, 2, 189 (1968).
4. Duwel, L., Staub-Reinhalt. der Luft (English Translation), 28,
42 (March, 1968).
5. Whitby, K. T., private communication to the authors (1970).
6. Langer, G., Powder Technology, 2, 307 (1968-69).
7. Goldschmidt, V. W., and M. K. Householder, Atmospheric Environment,
3, 643 (1969).
8. Cast, T., Staub-Reinhalt. der Luft, 30, 235 (1970).
9. Cast, T., Staub-Reinhlat. der Luft, 21, 136 (1961).
10. Gruber, C. W., and C. E. Schumann, Journal of the Air Pollution
Control Association, 16, 272 (1966).
11. Sinclair, D., Journal of the Air Pollution Control Association,
17, 105 (1967).
12. Charlson, R. J., Environmental Science and Technology, 3, 913 (1969).
13. Barrett, E. W., and 0. Ben-Dov, Journal of Applied Meteorology, 6,
499 (1967).
14. Ogle, H. M., Journal of the Air Pollution Control Association, 18,
657 (1968).
15. Martens, A. E., and J. D. Keller, Journal of the American Industrial
Hygiene Assocation, 29, 257 (1968).
16. Belz, R. A., Clearinghouse No. AD 674 741 (1968).
17. Coenen, W., Staub-Reinhalt. der Luft (English Translation), 27,
32 (Dec., 1967).
18. Schutz, A., Staub-Reinhalt, der Luft (English Translation), 26,
18 (May, 1966).
THERMO-SYSTEMS INC
-------
-29-
19. Dresia, H., P. Fischotter, and G. Felden, VDI-Z, 106, 1191 (1964).
20. Jackson, M. R., A. Lieberman, L. B. Townsend, and W. Romanek,
Proceedings of the National Incinerator Conference, 182 (1970).
21. Lilienfeld, P., Conference of the American Industrial Hygiene
Assocation, Detroit (May, 1970).
22. Schnitzler, H., 0. Maier, and K. Jander, SchrReihe Ver. Wass.-
Boden-Lufthyg. Berlin-Dahlem, 33, 77 (1970).
23. Horn, W., Staub-Reinhalt. der Luft (English Translation), 28, 20
(Sept., 1968).
24. Olin, J. G., and G. J. Sem, Atmospheric Environment, Pergamon
Press, 5, (1971).
25. Olin, J. G., and G. J. Sem, and D. L. Christenson, American
Industrial Hygiene Association Journal, 32 (April, 1971).
26. Chuan, R. L., Journal of Aerosol Science, 1, 111 (1970).
27. Coenen, W., Staub-Reinhalt. der Luft, 24, 350 (1964).
28. Mohnen, V. A., and P. Holtz, Journal of the Air Pollution Control
Association, 18, 667 (1968).
29. Grindell, D. H., AEI Engineering, 2, 229 (1962).
30. Whitby, K. T., and B.Y.H. Liu, Aerosol Science, ed. by C.N. Davies,
p. 59, Academic Press, New York (1966).
31. Schutz, A., Staub-Reinhalt. der Luft (English Translation), 26,
1 (Oct., 1966).
32. Prochazka, R., Staub-Reinhalt. der Luft (English Translation), 26,
22 (May, 1966).
THERMO SYSTEMS INC
-------
-30-
DETAILED REPORT
Page
SECTION 2. PRELIMINARY LABORATORY EXPERIMENTS 31
SECTION 3. DESIGN OF FIELD EXPERIMENT STATION 52
SECTION 4. CALIBRATION OF STACK FACILITY AT THE FIELD EXPERIMENT
STATION 61
SECTION 5. FIELD EVALUATION OF TWO PROTOTYPE BETA INSTRUMENTS . 78
SECTION 6. EVALUATION OF TRANSMISSOMETER TECHNOLOGY 83
SECTION 7. APPENDICES 91
THERMO-SYSTEMS INC
-------
-31-
SECTION 2. PRELIMINARY LABORATORY EXPERIMENTS
2.1 INTRODUCTION
Since one of the significant conclusions reached from the state-of-the-
art study early in this program was that beta radiation attenuation was a
promising technique, an instrument was purchased for experimentation. A
Gelraan* Model 25000 was selected on the basis of cost and delivery schedule.
Most of the work with the Gelman instrument was performed in Thermo-Systems'
laboratory. Limited testing was also done with this instrument at the stack
facility described elsewhere in this report, but the instrument did not func-
tion well enough to produce any reliable data there.
This section discusses the results of various laboratory experiments per-
formed with portions of the Gelman instrument. We were not so much interested
in testing the Gelman design as the beta radiation attenuation technique itself.
We experimented with two radiation sources: promethium 147 and carbon 14. Our
data is compared with theory and to a limited extent with the data of other
investigators. A treatment of the theoretical aspects of the use of this tech-
nique is given in the section of Volume II of this report entitled "Beta Radiation
Attenuation."
2.2 CALIBRATION REPEATABILITY
It is frequently assumed, on the basis of several approximations, that the
calibration curve of an instrument using the beta technique is of the form**
where:
I/I - exp [-y X] (4)
o m
I is the intensity of the beta radiation passing through
° a clean filter,
I is the intensity of the beta radiation passing through
a dirty filter,
2
X is the weight (mg/cm ) of particles on the filter, and
y is the calibration constant, usually assumed to be
m independent of everything but the radiation source.
^Manufactured by Gelman Instrument Co., Ann Arbor, Michigan
**See Volume II, pp. 70-85.
THERMO-SYSTEMS INC.
-------
-32-
Such a calibration curve, shown in Figure 10, was provided with the
Gelman instrument. The radioactive source is 50 microcuries of C-14. The
radiation sensor is a Geiger-Muller tube, Amperex Type 18515.
We initially decided to verify whether the above calibration curve was
accurate. The electronics, which basically consisted of a timing circuit,
a counting circuit, and a digital printer, were found to be repeatable only
to within about + 5% in the range of interest. To determine this, we sub-
stituted an extra filter to simulate particle loading. This allowed nearly
exact reproduction of filter loadings, and the output varied + 5% from the
average for several runs with the same loading. This occurred over nearly
the entire portion of the calibration curve supplied with the instrument.
A longer counting time or a larger radiation source might have decreased
the variability, but we did not modify the instrument to check this. This
variability affects the accuracy of all our results with the Gelman instru-
ment.
As a first check, we used Whatman #4* filter paper and AN-5000** mem-
brane filter to simulate particle loadings. The Gelman instrument uses
Whatman #4 in the form of a filter tape as its standard for collecting par-
ticles. These tests revealed a significant error in the calibration curve
as shown by the data points in Figure 11. The value of V^ (see Equation 4)
assumed by Gelman is 0.272 cm2/mg. It would require a large adjustment of
this coefficient to fit the data to the Gelman-supplied curve.***
We found that the consistency of the thickness of filter material (partic-
ularly Whatman #4 filter paper) is not very good. The paper contained varia-
tions ranging from about 8 to about 9 mg/cm2. Extensive testing was done in
our home laboratory to determine the practical effect of filter thickness
variations on the calibration of a beta instrument.
The results of the first series of these tests was summarized in Figure
12. Again, the procedure used was to add filter thicknesses of known density
to simulate increased filter loadings. Each curve in Figure 12 has a differ-
ent value of X , which is the initial unloaded filter weight used to measure
the I radiation intensity. The value of X in these tests varied from zero
to 11?3 mg/cm2. This covers the thicknesses of most practical filter materials.
*Sold by H. Reeve Angel & Co., Inc., Clifton, New Jersey.
**Sold by Gelman Instrument Co., Ann Arbor, Michigan.
***See Volume II, p. 76 for typical values of y«
THERMO-SYSTEMS INC.
-------
1.0
0.9
0.8
0.7
ADDED MASS DEPOSIT THICKNESS , MGM/CM'
I 23456
a 0.5
UJ
0.4
w
<
oc
h-
O 0-3
K
-------
-34-
LU
J-
1.0
0.9
0.8
0.7
0.6
0.5
0.4
5 0.3
a:
<
~I 1 1
FILTER TYPE:
A AN-5000
O WHATMAN N0.4
-1.0
OD
z
o
GELMAN-SUPPLIED
CALIBRATION CURVE
0.2
O.I
0.0
-0.5
-1.5
-2.0
8
10
SIMULATED FILTER LOADING, MGM/CM'
Figure 11. Comparison of Gelman-supplied calibration curve with actual
data obtained using Whatman #4 filter paper and Gelman AN-5000
membrane filters to simulate deposit loadings.
THERMO-SYSTEMS INC
-------
-35-
0.0 MGM/CM2
.15 MGM/CM2
2.30 MGM/CM2
3.45 MGM/CM2
5.65 MGM/CM2
.30 MGM/CM2
XAIR= 2.25MGM/CM
0.0
---0.5
-1.0
--I.5
-2.0
2 4 6 8 10 12
SIMULATED FILTER LOADING ,X , MGM/CM z
Figure 12. Results of tests using the standard Gelman sensor con-
figuration, C-14, and with six different initial clean
filter thicknesses. The filter loading was simulated by
other filters (Millipore Type HA and Nuclepore 5 ym pore
size). Note that variations in initial clean filter
thickness significantly affects the calibration curve.
THERMO-SYSTEMS INC
-------
-36-
The filters used in this experiment were Millipore* cellulose membrane
filters (Type HA, 0.45 urn pore size, thickness = 5.65 mg/cm2) and Nuclepore
filters** (5 ym pore size, thickness = 1.15 mg/cm2) . It should be noted
that the value of X does not represent all the mass between the radiation
source and detector during the I count. An air gap of 1.6 cm (X . - 2.25
mg/cm ) was present, along with the window of the GM counter tublir(approx.
thickness of 1.5 - 2.0 mg/cm2).
These curves illustrate that the variation in thickness of filter
material has a significant effect on the calibration of an instrument such
as the Gelman instrument. Their non-linearity (on a semi-logarithmic scale)
also suggests that the calibration curve cannot be assumed to be a true
exponential function of the form shown in Equation 4. As the value of X
increases, the curves seem to become more linear.
The next step was to move the radiation source and detector closer to-
gether and thus reduce the thickness of the air gap between them to 0.2 cm
(X ~ 0.36 mg/cm2). We hoped to separate out any effects of the geometry
of the apparatus in this way. These results are shown in Figure 13. Notice
that the same general trends in the data are present as in Figure 12. From
the results of these two experiments, we suggest the possibility that,
independent of the geometrical configuration, there is a very significant
error in assuming the calibration curve of a beta instrument to be of the
form shown in Equation 4.
To point out the magnitude of this error, Figure 14 presents some of
the data from Figure 13 in a different way. Figure 14 shows the effect of
the filter thickness (XQ) on the determination of the unknown (X) for a
constant output (output = I/I ). For example, it was stated earlier that
Whatman #4 filter paper varies between 8 and 9 mg/cm2. 'Looking at Figure 14,
we see that if our output (I/I ) was measured to be 0.6, and if we did not
know the filter paper thickness (8
-------
-37-
X0 = 0.0 MGM/CM2
X0 = 2.30MGM/CM2
X0 = 5.65 MGM/CM2
X0 = 7.95 MGM/CM2
X0 = 9-10 MGM/CM2
X0 = 11.80 MGM/CM2
XAIR= 0.36 MGM/CM
0.0
--0.5
o
h-l
-1.0 £
z
-1.5
--2.0
2 4 6 6 10 12
SIMULATED FILTER LOADING, X , MGM/CM2
Figure 13. Results of C-14 tests run identically to those in Figure
12 except with a smaller air gap between beta source and
detector. The results are similar.
THERMO-SYSTEMS INC
-------
-38-
0
SIMULATED FILTER LOADING ,X , MGM / CM'
Figure 14. C-14 data of Figure 13 plotted to show that, for a constant
beta instrument output (I/IO = 0.6), the measured filter
loading depends significantly on the initial clean filter
thickness.
THERMO SYSTEMS INC
-------
-39-
2.3 EXPERIMENTS WITH PROMETHIUM-147
In order to establish criteria with which to evaluate radiation sources
for use in beta instruments, a small amount of Pm-147 was purchased. Pm-147
has a short half-life (2.26 years as opposed to 5568 years for C-14) and the
radiation is more energetic (E = 0.229 mev as opposed to 0.155 mev for
C-14)*. The maximum amount that can be used without an AEC license is 10
microcuries, as opposed to 50 microcuries for C-14. Our experiments there-
fore were done with only 10 yc of Pm-147.
The data was taken in the same manner as that of Figure 12 and is pre-
sented in Figure 15. These curves cover the same range of filter thicknesses
as those for C-14. Two characteristic differences should be emphasized. The
first is the nonlinearity of the curves. The curves for Pm-147 are more
nearly linear than those for C-14, and therefore more closely approximate
Equation 4. Second, as the value of X changes, the curves are not so greatly
shifted. As will be explained later, these two differences result from a
single fundamental difference in the radiation energy spectra.
The shift in the curves as the value of X changes is portrayed in Figure
16, which is analogous to Figure 14. These two curves can provide a rough
comparison of the errors one might expect if the filter weight varied from
8 to 9 mg/cm2 (e.g., Whatman #4 filter paper). From Figure 16, if the instru-
ment output (I/IQ) was 0.6, and if 8
-------
-40-
1.0
0.9
08
07
0.6
0.5
o
- 0.4
o
to 0.3
2
(A
Z
(T
H
<
UJ
CD
0.2
O.I
0.0
o
A
D
O
Xn =
AIR
I I
0.0 MGM/CM2
3.45MGM/CM2
5.65MGM/CM2
7.95MGM/CM2
9.IOMGM/CM2
= I0.25MGM/CM2
= 4.6 MGM/CM8
-2.0
2 4 6 8 10 12
SIMULATED FILTER LOADING,X,MGM/CM2
14
Figure 15. Results of simulated filter loading tests using Pm-147
and six initial clean filter thicknesses, analogous to
Figure 12.
THERMO-SYSTEMS INC
-------
-41.-
12
10
(VI
2
O
^
2
O
O
X
to"
CO
UJ
O
X
I-
tr
UJ
UJ
8
Pm-147
= 0.6
8
SIMULATED FILTER LOADlNG,X,MGM/CM2
Figure 16. Pm-147 data of Figure 15 plotted to show that, for a constant
beta instrument output (I/I =0.6), the measured filter load-
ing depends less strongly on the initial clean filter thick-
ness than does C-14 shown in Figure 14.
THERMO-SYSTEMS INC
-------
-42-
A calibration curve, such as those in Figure 13, is really a measure of the
amount of radiation that is absorbed (or "stopped") by the deposit on a filter.
A prediction of such a curve could easily be made if we knew how much radiation
would be absorbed by the filter itself, the air gap between beta source and
detector, and the detector window, and also knew how much more would be absorbed
if a certain mass of particles was added on the filter. Two more pieces of in-
formation are needed to predict the calibration curve. First, we must know the
relationship of range versus energy for beta radiation (i.e., how many mg/cm^
will a beta particle of known energy penetrate on the average). Second, we
must know the cumulative distribution of beta particles versus their energy.
This will indicate the fraction of particles emitted with energy greater than
a specified energy level.
The relationship of range versus energy has been determined by Friedlander,
et al,* and is shown in Figure 17. Notice that at typical energy levels, the
relation is not linear.** The energy distribution curve for a number of beta
radiation sources have been theoretically calculated by Hogan, et al,*** in-
cluding the spectra for C-14 and Pm-147. These are shown in Figures 18 and 19.
v
o
ir
UJ
0.1 0.2 0.5 IO 2.0 5.0 10 20 50 IOO 200 500 IK 2K 5K IOK
RANGE IN ALUMINUM, MGM/CM2
Figure 17. Relationship of the energy required for the beta
particle to penetrate a given thickness 'of aluminum.
*Friedlander, G., Kennedy, J.W., and Miller, J.M., Nuclear and Radiochemistry,
2nd Ed. Wiley, N.Y., N.Y. (1964).
**See Volume II, p. 72.
***Hogan, O.K., Zigman, P.E., Mackin, J.L., "Beta Spectra: II. Spectra of
Individual Negatron Emitters", U.S. Naval Radiological Defense Laboratory,
Report USNRDL-TR-802, 1964.
THERMO-SYSTEMS INC
-------
-43-
I I I I I I I
I I I I I
I I I I I I I 1 I 1 I I
0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16
ENERGY, M«V
Figure 18. Energy distribution spectrum for a C-14 beta source.
25
O
o
z
=>
u.
15
10
ffi
*
CO
o
ai 5
to
I I I I I
Pm-147
I I I I I I
0.03 0.06 0.09 0.12
0.15
ENERGY,M«V
0.18 O.2I 0.24
Figure 19. Energy distribution spectrum for a Pm-14/ beta source.
THERMO SYSTEMS INC
-------
-44-
These energy spectra must be plotted as the cumulative energy distribution
(i.e., fraction of particles with energy greater than E versus E) in order to
be meaningful. Figure 20 shows this cumulative energy distribution for C-14.
The two curves, Figure 17 and Figure 20, provide the last two pieces of in-
formation needed to predict a calibration curve for C-14.
The information in these curves is more useful if the cumulative range
distribution is plotted (Figure 21). This is because it is the range of the
beta particles (i.e., how much mass in mg/cm^ is required to stop the particles),
not the energy of the particles, which is of interest.
Figure 21 is a form of calibration curve for a C-14 source, though it is
somewhat cumbersome to use. First, one would have to measure the thickness in
mg/cm^ of the air gap, the detector window, and the unloaded filter by some
other means. This would then tell us the range required of particles to make
the zero count. For instance, suppose this combination was 15 rng/cm^ thick,
including a reasonable value for Whatman #4 filter paper. Referring to Figure
21, this corresponds for C-14 to an intensity ratio of 0.0645, which means
that 6.45% of the beta radiation emitted from the source and headed in the
direction of the detector will actually reach the detector and be counted. The
other 93.55% will be absorbed, by the air gap, the filter, and the detector
window.
The next step is to make an "I " count; e.g., count beta particles for
10 seconds and note the total. A sample of particulate matter can then be
collected on the filter and an "I" count can be made in the same manner as
the "I " count was made. The value of An(I/I) = (An I - An I ) is then
o o
calculated. We assumed the value of An I = An 0.0645 = -2.741. We also know
o
the difference, An I - An I . So in our hypothetical case the value of An I =
(An I - An IQ) + An I = An (I/I ) -2.741. Suppose we measured I/I = 0.5.
(This ratio will, of course, always be less than unity.) Then An (I/I ) =
An 0.5 = -0.693 and An I = -0.693 - 2.721 = -3.434. Referring to Figure 21,
this corresponds to a range of 18.1 mg/cm^. So we can say that only beta
particles with ranges greater than 18.1 mg/cm2 are getting through the air
gap, the filter, the detector window, and the particulate matter deposited on
the filter. This is about 3.2% of the radiation. More importantly, we can
say that the thickness of the deposit of particulate matter on the filter
is 18.1 - 15.0 =3.1 mg/cm^. This number multiplied by the filter area
would yield the total mass on the filter, assuming a uniform deposit.
The nonlinearity of the curve in Figure 21 explains the nonlinearity of
the previous calibration curves (Figures 12 and 13). Notice too that at the
high end of the curve (i.e., range >12 mg/cm^) the nonlinearity becomes greater.
In other words, it is important to know where one is operating on the curve.
It is important to know the value of the filter thickness, the air gap, and
the detector window.
THERMO-SYSTEMS INC
-------
-45-
0.01
60 80
ENERGY,KeV
Figure 20. Cumulative energy distribution for a C-14 beta source.
THERMO-SYSTEMS INC
-------
-46-
20 25 30 35
0.01
5 10 15
RANGE IN ALUMINUM, MGM/CM'
Figure 21. Cumulative range distribution for a C-14 beta source.
THERMO SYSTEMS INC
-------
-47-
To illustrate this even more clearly, recall from the above example that
for an intensity ratio of I/I = 0.5, we calculated the thickness of the
particulate matter to be 3.1 mg/cm^. If the thickness of the filter plus air
gap plus detector window would have been 16 mg/cm^ instead of 15 mg/cm^, and
if we had measured the intensity ratio I/I to be 0.5 as above, we would have
found from Figure 21 that the total thickness of air gap plus filter plus
detector window plus particulate matter deposited on the filter was 18.85 mg/cm .
This results in a deposit of particulate matter of 2.85 mg/cm^. An instrument
that did not take into account the nonlinearity of Figure 21 would possess an
error of about 8%, assuming that the unknown changes would be limited to 1 mg/cm ,
Using Figure 19, the cumulative energy distribution of Pm-147 radiation
has also been plotted (Figure 22). From Figure 17 and 22 we get the cumulative
range distribution of Pm-147 radiation (Figure 23).
Referring back to the examples shown, if we again had a case where the
air gap plus filter plus detector window amounted to 15 mg/cm^ and measured
I/I to be 0.5 with Pm-147, the thickness of particulate deposit would be
6.8 mg/cm^. If the air gap plus filter plus detector window amounted to
16 mg/cm2 and I/I was measured to be 0.5, the thickness of the particulate
deposit would be 6.7 mg/cm^. So an instrument that did not take into account
the nonlinearity of Figure 23 would possess an error of about 1.5%, assuming
that unknown changes would be limited to 1 mg/cm^.
The examples used have demonstrated that, given a range of around 15 -
20 mg/cm2, Pm-147 radiation more closely follows the exponential law of
Equation 4 and therefore provides better accuracy if the Equation 4 relation-
ship is assumed. It is clear from Figures 21 and 23 that neither curve is
linear throughout the spectrum. When considering the application of one of
these sources, one has to work with the proper portion of the curve. These
curves were not calculated for any other radiation source.
2.5 EXPERIMENTS WITH FLY ASH
An attempt was made to operate the Gelman instrument in our facility at
the smoke stack. We found, however, that the deposits of particulate matter
on the filter paper were very nonuniform due to the design of the sampling
head. Furthermore the seal around the filter leaked. We did not find it
feasible to redesign the sampling head during this program to correct these
problems.
In order to get some data with real fly ash, another approach was followed.
We used our parallel filter system to collect identical samples on two filters
(identical, that is, to within + 10%). We weighed one filter deposit with a
microbalance, and the other we "weighed" with the Gelman instrument. The results
of some of these runs are displayed in Figure 24. The broken line is a linear
approximation from Figure 13 for X - 7 mg/cm^, which was the thickness of the
filters used for these runs.
THERMO SYSTEMS INC
-------
-48-
UJ
0.01
20
40
60 80 100
ENERGY, KeV
120
140
ISO
Figure 22. Cumulative energy distribution for a Pm-147 beta source.
THERMO-SYSTEMS INC
-------
-49-
0.01
10 15 20 25 30
RANGE IN ALUMINUM, MGM/CM2
35
Figure 23. Cumulative range distribution for a Pro-147 beta source.
THERMO-SYSTEMS INC
-------
-50-
1.0
09
0.8
0.7
0.6
0.5
0.4
0.3
o
U5
V)
z
cr
LU
QQ 0.2
O.I
o
o
o
LN(I/I0) =-0.269 X
O
O
--I.O
0.0
1.5
2.0
0123
ADDED MASS DEPOSIT THICKNESS, X, MGM/CM!
Figure 24. Results of tests with the standard Gelman unit (C-14) using fly
ash collected in a separate filter holder as the filter loading.
THERMO SYSTEMS INC
-------
-51-
The data points fall consistently below the broken line. This is
explained by the fact that the particulate deposits on the filters were
not uniform. There was a somewhat heavier concentration near the center
than around the edge. Since the central portion of the filter was "weighed"
by the Gelman instrument, one would expect the data to deviate in the
direction shown. This data is therefore inconclusive, as it was not
possible to measure the variations in deposit thickness on a single filter.
No other experiments were performed with the Gelman instrument on fly
ash.
2.6 SUMMARY
The Gelman instrument was first tested in TSI's laboratories to determine
the parameters that affect the calibration. It was found that the variations
in filter tape thickness and the radiation source characteristics both signi-
ficantly affect the calibration, even if I is measured on the same section
of tape as is I. This should be considered carefully when advanced instru-
mentation is designed and tested. Some experiments were performed on fly
ash aerosol from TSI's stack facility, but the data from these experiments
was inconclusive.
THERMO-SYSTEMS INC
-------
-52-
SECTION 3. DESIGN OF FIELD EXPERIMENT STATION
3.1 INTRODUCTION
This section describes the field laboratory that was constructed and
instrumented by TSI as a part of this contract. Most of the experimental
work on the contract was done in this field laboratory, so an understanding
of its features and capabilities is important.
The field laboratory is located in the stack of the Northern States Power
Company's Allen S. King electrical power generating plant at Bayport, Minnesota.
The laboratory is both heated and air conditioned, and therefore can be used
at all times of the year. Access to the stack laboratory facility can be gained
only through the proper authorities at the King plant.
3.2 NSP POWER PLANT
The King plant, shown in Figure 25, has a rated power output of 550 mega-
watts. Its location is 25 miles east of TSI on the St. Croix River at the
Minnesota-Wisconsin border. The principal fuel is southern Illinois and
Eastern Kentucky coal, which has a sulfur content of about 3%. The coal is
crushed and burned in 12 cyclone furnaces which are all connected to a single
heat exchanger. The plant is equipped with electrostatic precipitators that
remove about 99% of the mass of particulate emissions from the flue gas. The
stack is 785 feet high. It consists of a central steel chimney which contains
the flue gas and an outer concrete supporting shell with an annular space
between them. The sampling done under this contract was from the horizontal
breeching within the annular space at a point about 10 feet before the effluent
reached the vertical steel chimney. The effluent turns an 80 horizontal bend
about 25 feet upstream of the sampling point. The internal dimensions of the
breeching at the sampling point are 12 feet across and 27 feet high. The
vertical steel chimney is 26 feet in diameter. Another identical horizontal
breeching empties into the stack directly opposite the sampled breeching.
The choice of the King plant as a test location for instrument development
was based on several considerations. It is within convenient driving distance
from TSI. It is typical of large, modern, and coal-fired combustion sources.
Since it is a base-load plant, operating conditions are relatively constant,
allowing one to repeat a test under reasonably constant conditions to locate a
problem.
The decision to do experimental work in a smoke stack rather than a simulated
stack atmosphere was made because of the difficulty in simulating the inside of
a stack. The lack of good measurements of both gas flow and particle flux char-
acteristics in stacks coupled with the lack of an experimental facility for such
simulation indicated strongly that instrument testing and development at the
time of this work would be profitable only in an actual stack.
THERMO-SYSTEMS INC
-------
-53-
Figure 25. Northern States Power Company's Allen S. King
electrical power generating plant, located at
Bayport, Minnesota, where the field experiment
station was constructed.
THERMO-SYSTEMS INC
-------
-54-
3.3 GENERAL LAYOUT
The location of the stack laboratory is about 70 feet above ground level.
This is the level at which the flue gas from the electrostatic precipitators
enters the vertical stack, as shown in Figure 26. The construction of the
stack itself is illustrated in Figure 27, which is a cutaway section of the
area where the breeching enters the stack. The laboratory is constructed on
a heavy steel floor between the outer concrete shell and the inner steel stack,
as shown in Figure 28.
Since it is inside the concrete shell, the stack laboratory is protected
from rain and snow. The temperature of the air surrounding the laboratory
approximates the outdoor temperature because of the natural ventilating effect
in the stack. An air conditioner and a heater have been installed so that the
facility can be used in nearly all weather conditions.
As Figures 27 and 28 illustrate, there are seven ports installed in the
breeching adjacent to the laboratory. These are 6 inch diameter ports and can
be made accessible for sampling from the breeching. Only one of the ports was
used on this program.
3.4 INSTRUMENTATION
The purpose of the stack laboratory is to facilitate the evaluation of
particle mass sensing instruments using genuine effluent aerosol directly from
a smoke stack. The method which we chose to evaluate instruments is to deliver
identical samples of stack aerosol to both the instrument and a filter operating
in parallel with the instrument. The instrument measurement can then be compared
directly to the measurement obtained by weighing the parallel filter. It should
be noted that, for the purpose of this work, it is not so important that a truly
representative stack aerosol sample reach the instrument and parallel filter. It
is much more important that the two samples be identical.
With this in mind, we fabricated the large sampling system shown in Figure 29.
It can draw up to 20 CFM of flue gas from the stack and present identical samples
to the individual parallel sampling systems as shown in Figure 30. The large
sampling system removes a reasonably representative sample of effluent aerosol
from the stack with approximately isokinetic conditions. All parts of the large
system which are outside of the stack are heated to prevent condensation on the
walls. An air ejector pump, operated by an air compressor located at ground
level, draws the flue gas through the system. The large sampling tube serves
three functions:
THERMO SYSTEMS INC
-------
-55-
,
Figure 26. A view of the twin electrostatic precipitators
and horizontal breeching of the power plant.
The field experiment station is located in the
right breeching just inside the outer wall of
the vertical stack. Effluent passes from
bottom to top of the photograph.
THERMO-SYSTEMSINC
-------
t
-56-
TOTAL STACK HEIGHT =785'
OPEN
ANNULAR
SPACE
STEEL PLATFORM
STEEL CHIMNEY
CONCRETE SUPPORT
STACK
PORT USED FOR
THIS PROGRAM
20'
27'
12'-
NORTH
BREECHING
GROUND
Figure 27. A side view of the horizontal breeching from which samples for
the experimental instrument evaluations were drawn.
THERMO-SYSTEMS INC
-------
t
-57-
TOTAL STACK HEIGHT =785'
OPEN
ANNULAR
SPACE
-STEEL CHIMNEY
CONCRETE SUPPORT
STACK
6" PORT (TOTAL OF7)
STACK LABORATORY
AIR CONDITIONER
STEEL PLATFORM
OPEN
STORAGE
SPACE
AGROUND
Figure 28. A cross-sectional view of the horizontal breeching showing the
location of the laboratory with respect to the breeching and stack.
THERMO SYSTEMS INC
-------
-58-
COMPRESSED
AIR
SAMPLING LINES TO
INSTRUMENT
FLO* SYSTEMS
Figure 29. The sampling system which draws a somewhat representative sample
of effluent aerosol from the breeching and delivers it to 4
parallel sampling systems serving as many as 4 separate instru-
ments simultaneously.
ISOXINETIC SAMPLING LINE
FROM
MAIN SAMPLING SYSTEM
BOUNDARY
LATER
OlLUTOR
BALL VALVE
TEST FILTER
OR
INSTRUMENT
SONIC ORtFICE
FILTERED
COMPRESSED
AIR
ABSOLUTE FILTER
Figure 30. A typical parallel sampling system which removes an aerosol sample
from the main sampling system shown in Figure 29, dilutes the sample
from 0 - 90%, and delivers the diluted aerosol to the instrument or
test filter. Four of these parallel systems can operate simulta-
neously. Two are designed for 20 - 50 LPM, and two for about 1 LPM,
through the instrument or test filter.
THERMO SYSTEMS INC
-------
-59-
1. It is physically large enough so that several identical, but
separate, samples can be drawn out vertically downward and
routed to several instruments simultaneously. The vertical
orientation minimizes the loss of large particles by gravi-
tational effects.
2. It lowers the velocity of the effluent aerosol so that instru-
ments with low sampling rates of a few liters per minute can
sample nearly isokinetically without having to use an un-
reasonably small inlet tube.
3. It is large enough and has enough capacity so that the flow
field at the entrance to the individual parallel sampling
systems is not distorted when an individual sampling tube is
removed or shut off.
The conical expansion section, which connects the 1-inch tube from the stack
with the 6-inch downstream tube, is used as a rough, but adequate, flow in-
dicator by measuring the pressure differential from one end to the other.
A pair of parallel sampling systems, as shown in Figure 30, was used for
all instrument tests at the stack. The aerosol travels downward with very
little change in tube size through the parallel systems. The aerosol first
passes through a boundary layer diluter which cools the sample and dilutes it
enough to prevent nearly all condensation on the instrument and parallel
filter, which are at room temperature. The dilution systems can operate with
0 to 90% clean air. The sampling tubes upstream of the diluters are heated.
A ball valve downstream of the diluter is the primary method of beginning
and ending a run. The instrument or parallel filter can be installed down-
stream of the ball valve. Sometimes, as in the beta instrument tests described
in Section 5, a right angle bend must be made in the sample line just upstream
of the instrument. In all tests on this program, if a bend was required for
the beta instrument, an identical bend was installed in the parallel filter
line.
The gas is drawn through the parallel systems by vacuum pumps. Dilution
air can be regulated and held very constant for operating periods of several
hours. Notice that the dilution air and total diluted aerosol flow (mixture
of clean air and flue gas) are measured with the same high-precision flow
meter. The dilution air is normally measured and adjusted before the start
of a run and aerosol flow is measured and adjusted at the beginning of each
run. As the filter loads, some adjustment of the total diluted aerosol flow
is usually necessary.
THERMO-SYSTEMS INC
-------
-60-
The accurate measurement of the dilution (clean) air and total diluted
aerosol flow is very important for accurate particle mass concentration
measurements and for the comparison of 2 parallel systems. Reasons for
the necessity of accurate flow measurements are discussed in Section 4.2.b.
We found that Thermo-Systems Inc. thermal mass flowmeters, using a hot-film
anemometer sensor in a venturi nozzle, proved satisfactory for this service.
The flowmeters, one for each parallel sampling system, are compensated for
temperature fluctuations and have a rated accuracy of about + 1% of reading.
An "absolute" filter (MSA Model 92706) just upstream of the flowmeter keeps
the sensor clean during several months of operation.
A Stauscheibe (S-type) pitot probe was available for making stack velocity
measurements.
3.5 SUMMARY
The stack laboratory constructed on a commercial power plant stack is a
versatile facility that can be used for many varied types of instrument test-
ing. Several ports are available for sampling from the stack. Pairs of
identical parallel sampling systems, using valves, diluters, and precision
flowmeters,are available for testing of instruments such as the tests described
in Section 5. Modifications can be easily made to accommodate other instruments
and sensors.
THERMO-SYSTEMS INC
-------
-61-
SECTION 4. CALIBRATION OF STACK FACILITY AT THE FIELD EXPERIMENT STATION
4.1 PROCEDURE
The ability of the sampling facility to deliver identical samples of efflu-
ent to each instrument had to be experimentally verified. This was done before,
during, and after each instrument test reported elsewhere in this report. Iden-
tical 47-mm Gelman Type A glass fiber filters without binder in Gelman stainless
steel filter holders were plumbed into the identical legs of the sampling system.
One of these filter holders was identified as the parallel filter. The other
filter holder, its filter sampler identified as beta filter, substituted for the
test instrument during these tests of the sampling system. The test setup during
these sampling tests is described in Section 3 of this report.
During all of these tests, the parallel filter was always operated at 50 LPM
total diluated aerosol flow rate and 75% dilution, the same condition that was
used for most instrument tests. The beta filter was operated at several conditions:
50, 30, and 20 LPM total diluted aerosol flow rate; and 75% and 87.5% dilution.
Both filters were always operated for identical sampling times of 2, 3, 5, or 10
minutes.
The procedure began a day or two before the test by weighing enough clean
filters to accommodate a single day's sampling on a Mettler micro analytical
balance, Model M5/SA. The weighed filters were placed in individual sealed
plastic petri dishes, Millipore Cat. No. PD1004700. The operator carried them
with him to the stack on the morning of test day.
The facility was always warmed up at least a half hour, and usually nearly
one hour before the first run began. Each run was preceeded by the cleaning
of both filter holders and the careful mounting of filters in both of them. The
dilution (clean) air flow was then adjusted. Referring to Figure 30, each run
began by opening the upper 3-way valve to allow clean air flow to the diluter,
opening the ball valve, and opening the lower 3-way valve to allow the pump to
suck a sample through the test filter. The total diluted aerosol flow rate was
monitored continuously and adjusted as needed to maintain a constant measured flow
rate. Little adjustment was needed during most runs. Each run ended by returning
the three valves to their standby position. The dilution air flow rate was checked
between each run and any significant changes (very rare) was noted. Each filter
was then carefully removed from its holder and placed in its sealed carrier. Pre-
paration for the next run then could begin.
The filters were carried to the balance at the end of the day. Usually, at
least one day passed before they could be weighed with each filter stored in its
individual, sealed petri dish. The weighing procedure began by opening the petri
dish and immediately placing the filter onto the microbalance tray. A static
eliminator, 3-M Model 204 with 5 millicuries of Po-210, was waved over the
THERMO-SYSTEMS INC.
-------
-62-
fliter just before actual weighing. Usually, the filters would immediately
begin to lose or gain weight in the desiccated microbalance. The earliest
possible weighing with the least filter weight change was used for all weights
in these tests. We chose the "unchanged" weight because that was most nearly
the condition sensed by the beta instruments.
An attempt was made to obtain a rough estimate of particle losses in the
vertical straight tube, the 90° elbow, and the horizontal straight tube
immediately upstream of the filter holders. The components were carefully
cleaned before the three days of runs during August, 1972. After all the runs
were completed the three components of each sampling line were removed and
carefully washed out with alcohol. The solid sediment was collected and
weighed. Although not a precise technique, this method does offer a rough
estimate of line losses in a portion of the sampling system.
4.2 RESULTS OF FILTER TESTS
4.2.a. Introduction
The primary set of parallel filter tests for evaluation of the stack
sampling facility were conducted on August 25, 30, and 31, 1972. Ideally,
these tests would have been completed immediately preceeding the evaluation
tests of the beta instruments which were performed on May 15 - 18 and May 22 -
24, 1972. However, scheduling difficulties made such a program impossible.
Instead, we performed enough parallel filter tests before and during the beta
instrument evaluation tests to assure acceptable operating conditions and
reserved the more thorough parallel filter tests for a later date.
The complete data for the August tests are found in Appendix A. Table 3
summarizes the measured stack concentrations (uncorrected for sampling line
losses) for each operating condition and for each run. The two filters are
designated beta filter for the one which takes the place of the beta test instru-
ment and parallel filter for the one which normally serves as a reference for the
beta test instrument. Since 75% dilution and 50 LPM total diluted aerosol flow
rate was the condition used most often in the May beta instruments tests, 5
repetitions were run as this condition for each sampling time. Three rep-
etitions were made for every other condition. The sampling systems leading to
both filters are intended to be identical.
Note that the operating conditions for the parallel filter on all test runs
was 75% dilution, 50 LPM total flow rate, and the same sample time as the^beta
filter. The beta filter operating conditions were adjusted as indicated in
Table 3. It is important to remember that this is not the situation during
actual instrument evaluation tests. During nearly all May test runs, both the
beta instrument and the parallel filter were operated at identical conditions.
Thus, except for the 75% dilution, 50 LPM total flow rate runs, the August runs
are not indicative of the expected ratio between test instrument and parallel
filter, but rather, are an indication of some of the characteristics and cap-
abilities of the test setup.
THERMO SYSTEMS INC
-------
Table 3. Measured stack concentration, mgm/m , for both the beta filter and the parallel filter during all August runs. The data is
uncorrected for line losses.
Dilution: 75% 87.5%
Diluted Aerosol
Flow Rate, LPM
Sample Time, Min.
Beta
Filter
Repetitions
Parallel
Filter
Repetitions
1
2
3
4
5
1
2
3
4
5
5
85.
119.
136.
138.
146.
95.
130.
138.
141.
144.
9
3
9
3
1
8
5
8
4
8
50
10
155.
142.
137.
141.
144.
158.
145.
135.
144.
144.
5
8
1
8
8
2
4
2
4
7
2
94.
65.
66.
64.
58.
93.
79.
57.
60.
60.
30 20 50 30
5U>2J±I02J>l02.j3lO
0 115.0 121.7 74.0 89.0 83. -9 28.7 59.4 74.2 31.7 54.6 69.4
1* 87.7 106.9 69.4 66.1 62.5 17.2 62.3 65.5 22.2 52.1 72.7
6 66.6 114.2 68.1 94.6** 62.8 21.4 60.6 68.5 28.8 51.8 70.4
7
4
2 125.3 130.8 77.2 111.9 97.5 27.4 . 67.7 81.8 35.7 61.8 76.7
8* 108.8 123.3 79.9 103.2 71.7 27.2 70.9 77.0 33.4 58.6 82.8
7 111.9 130.8 75.8 116.0* 70.9 26.0 67.9 82.4 33.7 55.9 84.6
0
6
20
2 J> 10 2
30.7 40.3 48.6 25.2
19.5 34.3 52.1 22.2
24.9 38.5 54.2 122.0
33.3 55.4 77.5 32.0
29.3 55.6 84.5 23.5
30.0 55.1 87.2 24.8
Note: All 75%, 50 LPM, 5 Min. repetitions were run first; all 75%, 50 LPM, 10 Min. repetitions were run next, etc., in the order shown above.
*Runs before this point run Aug. 25, 1972; runs after this point run Aug. 30, 1972.
**Runs before this point run Aug. 30, 1972; runs after this point run Aug. 31, 1972.
OJ
I
-------
Table 4. Ratio R for all August runs. R
r r
uncorrected for line losses. The
= (beta filter measured concentration/parallel filter measured concentration). The data is
average R and the standard deviation of R for each operating condition are also shown.
Dilution:
75%
87.5%
iluted Aerosol
low Rate, LPM
ample Time, Min.
1
2
3
4
5
Average :
Sigma, s:
_5
.897
.914
.986
.978
1.009
.9568
.0485
501
12
.983
.982
1.014
.982
1.001
.9924
.0146
2_
1.009
*
.816
1.155
1.077
.964
1.0042
.1279
_5
.918
.806
.595
-
_
.7730
.1645
302
10
.930
.867
.873
-
_
.8900
.0348
2 5
.959 .795
.869 .641
.898 .816**
-
_ _
.9087 .7507
.0459 .0956
202
10
.861
.872
.886
-
_
.8730
.0125
2
1.047
.632
.823
-
_
.834i
.207
Jj
.877
.879
.892
2
50
10
.909
.851
.831
_2 j>
.888 .883
.665 .889
.855 .927
2
30
10
.905
.878
.832
2
.922
.666
.830
_5
.727
.617
.699
2
20
10
.627
.617
.622
_2
.788
.945
4.919
(.8665)
2.37
(.078)
Note: All 75%, 50 LPM, 5 Min. repetitions were run first; all 75%, 50 LPM, 10 Min. repetitions were run next, etc., in the order shown above.
Both filters at same operating conditions
2
Beta filter at condition indicated, PF at 50 LPM, 75% dilution, and time indicated
*Before this, run on Aug. 25, 1972; after
**Before this, run on Aug. 30, 1927; after
this run on Aug. 30.
this run on Aug. 31.
-------
-65-
The primary criteria for evaluating the test results was that the ratio
of beta filter concentration to parallel filter concentration R,., be nearly 1.0
and, more importantly, be constant. Table 4 shows the ratio Rp for all August
runs. Also shown are the average and standard deviation for each operating
condition. Figure 31 shows the data points for all runs. Figure 32 shows the
standard deviations at various operating conditions. For many types of tests,
a reasonable criteria for judging identical parallel filter tests may be that
a standard deviation of greater than 0.05 indicates that the technique and
system yielded unacceptably inconsistent data.
Several conclusions can be made immediately. Two minute sampling times
yield inconsistent data under this mode of operation. Ten minute sampling
times are acceptably consistent under any operating conditions. The higher
dilution ratio (87.5%) was generally more consistent than 75% dilution, partic-
ularly with 5 minute sample time and 20 LPM total diluted aerosol flow rate.
Higher flow rates (50 LPM) generally give better consistency than lower flow
rates (30 or 20 LPM).
Table 5 identifies the operating conditions which were judged to yield
acceptably consistent results in this series of tests. The average ratio R
for 75% dilution, 50 LPM total diluted aerosol flow rate are between 0.95
and 1.00. The standard deviations are generally lower than for other con-
ditions. The 15 runs at these operating conditions are the ones most indicative
of the probable errors encountered during the May instrument tests.
Table 5. Operating conditions which yielded acceptably consistent results
during August runs. Also shown are the average ratio R.^ (top
number in each box) and the standard deviations (bottom number in
each box) of the ratio R^ for each operating condition.
Dilution:
Diluted LPM :
MIN
75%
50
30
10
0)
00
a
20
50
87.5%
30
.9924 .8900 .8730
YES YES YES
.0146 .0348 .0125
.9568 .7730 .7507
YES NO NO
.0485 .1645 .0956
1.0042 .9087 .8340
NO NO NO
.1279 .0459 .2075
20
.8637
YES
.0396
.8827
YES
.0082
.8027
NO
.1204
.8717
YES
.0369
.8997
YES
.0238
.8060
NO
.1292
.6220
OK
.0050
.6810
NO
.0572
.8665
NO
.078
THERMO-SYSTEMS INC
-------
BETA FILTER OPERATED AT:
75% DILUTION 87j/2% DILUTION
-66-
100 120 '40 160 0 20 4O 60 80
PARALLEL FILTER CONCENTRATION, mgm/ms
PARALLEL FILTER
DPtRATED AT
75% DILUTION, 5O LP«
O- 2 MINUTES
E> 5 MINUTES
A-10 MINUTES
Figure 31. Stack concentration as measured by the beta filter plotted against stack con-
centration as measured by the parallel filter for all August runs. The diago-
nal line on each graph represents a perfect (1:1) correlation. Note that the
parallel filter was always operated at 50 LPM diluted aerosol flow rate and
75% dilution while the beta filter was operated at the conditions shown.
point was rejected before calculating RAVG and S for thxs condition.
-------
m
73
5
m
o
i i
SAMPLE
TIME,
MINUTES
DILUTED
AEROSOL
FLOW RATE,
LPM
.20
2468
SAMPLE TIME, MINUTES
10 20 3O 4O 5O
DILUTED AEROSOL FLOW RATE, LPM
75% 87|/z%
DILUTION, %
Figure 32. Average standard deviation S of the ratio R as a function of sample
time, diluted aerosol flow rate, and dilution ratio for all August
runs. A low standard deviation denotes high reproducibility of the
ratio from run to run, but does not denote perfect (R = 1.0)
correlation.
-------
-68-
The average measured concentration ratio R^ for nearly all other accept-
able conditions ranges from 0.86 to 0.90. Thus, the reference parallel filter
usually measured higher stack concentrations. The primary reason for this
lack of perfect correlation is probably the inaccuracies of the flow measure-
ments. This argument is strengthened by the fact that the ratios R tend to
be even lower as total flow decreases and dilution ratio increases. We had
considerable trouble earlier in the test program because we were measuring
dilution (clean) air flow rate and total diluted aerosol flow rate with 2
separate flowmeters using 2 different measurement principles. We eliminated
most of the problem by using a single, highly-accurate, gaseous mass flowmeter
to measure both dilution air flow rate and total diluted aerosol flow rate.
However, the inaccuracies cannot be entirely eliminated.
All 2-minute samples are rejected because of inconsistent results. The
relatively good results shown in Table 5 for 75% dilution, 30 LPM total diluted
aerosol flow, and 2 minute sample time is not considered significant.
4.2.b. Gas Flow Measurement
Since gas flow measurements play such an important role in these tests,
further discussion is justified. The measurement of stack particle concentra-
tion requires the measurement of the weight of particles within a known or
measured volume of air. If no dilution of the sample takes place, any errors
in the measurement of the air volume will result in a similar error in measured
concentration; i.e., a 5% error in air volume measurement results approximately
in a 5% error in measured concentration. However, dilution of the sample up-
stream of the particle collecting device complicates the problem. The measured
particle concentration at the collecting device C must be multiplied by the
dilution ratio R^ (R_ = total diluted aerosol flow rate Q ./undiluted aerosol
flow rate QTTA) to obtain the stack particle concentration. We concluded during
the system design phase that we could not tolerate any available flowmeter in
the sample stream upstream 'of the particle collecting device because of the
resulting loss of particles leading to a plugged flowmeter and a changed cali-
bration. Thus, we had to measure dilution (clean) air flow rate Q and total
diluted aerosol flow rate Q_. to obtain the undiluted aerosol flow rate Q^:
V = QDA - QCA (2)
The stack concentration C (uncorrected for particle sampling line losses) then
becomes :
QUA CD- QDA-QC CD
C (3)
C
For example, let us assume that we are using highly accurate flowmeters which
have an error of only + 1% of reading for the measurement of QDA and QCA,
THERMO-SYSTEMSINC
-------
-69-
Also assume that the dilution ratio which we attempt to obtain is IL = 8
(87.5% dilution). Now assume that Q is measuring 1% lower than the desired
50 LPM and Q^ is measuring 1% higher than the desired 43.75 LPM. Plugging
these values into Eq. (3);
c = C QpA = r 49.5 LPM
S, Error CD Q ~ Q D (49.5 - 44.1875) LPM
We were trying to adjust the flow measurements to obtain:
C = r 50
^S, Actual UCD 50 - 43.75 ~ LCD
Thus, the maximum error in measured stack concentration caused by the low + 1%
error in the flow measurement is:
- 8)
This amount of error could also have been negative. Thus, if one of the parallel
systems was positive and the other negative, the difference between the actual
particle concentrations entering the two parallel filters could be as much as
about 30%.
We chose to measure both clean air flow and total diluted aerosol flow
with a single flowmeter using a proper valving system to direct each flow
in turn through the flowmeter. This eliminated differences between 2 flow-
meters and 2 calibrations as an error. Also, the use of a single flowmeter
partially compensates for the need for increasingly accurate flow measurements
with increasing dilution because both measurements (Q and Q ) are made on
a short segment of the flowmeter calibration curve making the measurements more
accurate with respect to each other. Using a single flowmeter does not allow
the operator to monitor both flow rates during a run. He must adjust the
dilution (clean) air flow before a run, monitor and adjust the total diluted
aerosol flow during the run, and recheck the dilution air flow after the run.
We found very little change in dilution air during any run. It was necessary
to monitor and adjust the total diluted aerosol flow as the filter loaded dur-
ing most runs.
The flowmeters we used in both sample lines were Thermo-Sys terns' thermal
mass flowmeters (Model 1352-3) utilizing a hot-film anemometer sensor in a
venturi nozzle. Both flowmeters were calibrated in clean air at 70°F and the
sensors are temperature compensated. Although we feel this flowmeter was the
best choice available, the user should understand several facts before using
it in an effluent stream. The gas composition (more precisely, the heat transfer
characteristics of the gas) was not accurately known and probably did not remain
THERMO-SYSTEMS INC.
-------
-70-
completely constant. The clean dilution air and the contaminated total diluted
air streams could be expected to have somewhat different heat transfer character-
istics, although both streams are made up primarily of N_. Fluctuating temp-
erature or a temperature different from the calibration temperature could cause
measurement inaccuracies, particularly while the total diluted aerosol steam is
heating up the sampling line and filter holder at the beginning of a run. We
calibrated the flowmeters just before the May tests and again before the August
runs. We recommend recalibration against a highly stable reference every 2-3
weeks of tests such as these.
Summarizing our thoughts regarding flow measurement errors and their effects
on dilution ratio determination, we recommend avoiding dilutions greater than
87.5%. We also recommend operating both of the parallel sampling lines at the
same operating conditions. Flowmeters used in the sampling system must be highly
accurate and recalibrated often. Direct measurement of undiluted aerosol flow
rate probably results in excessive particle line losses. A single flowmeter
should measure both dilution air flow and total diluted aerosol flow. Even so,
any errors in flow measurement became magnified several-fold because direct
measurement of the undiluted aerosol cannot be tolerated.
Most of the lack of correlation between the parallel sampling lines in
the August data can be attributed to errors in flow measurements. During all
August runs except the 75% dilution, 50 LPM runs, the two parallel sampling
lines operated at different flow and dilution conditions. Measurements at all
sampling conditions except 75% dilution, 50 LPM resulted in a ratio R^ of less
than 0.91. However, the average ratio I* for the 75% dilution, 50 LPM runs at
all sampling times was 0.984. We suspect that the better correlation at that
condition and the correspondingly poorest correlation at 87.5% dilution, 20 LPM,
was primarily due to one or both of the following reasons:
1. Both sampling lines were operating at identical conditions for
the 75% dilution, 50 LPM runs, while operating conditions were
not identical on all other runs.
2. 75% dilution, 50 LPM is a preferred compromise operating con-
dition where dilution is sufficient to prevent excessive con-
densation while not extreme with the resulting flow measure-
ment problems.
Three more topics will be discussed before going on to the beta instru-
ment tests. First, the interesting major effect of sampling time on measured
stack concentration. Second, an experimental estimate of sampling line losses
in a portion of the line. Third, an experimental correlation between the
August runs and several May runs.
THERMO-SYSTEMS INC
-------
-71-
4.2.C. Effect of Sampling Time
Table 6 shows the average measured stack concentration for the runs of
August 31. Note the striking effect of sampling time on the measured con-
centration. With 87.5% dilution, the average stack concentration for all
2 minute samples is 30.7 mgm/m3, for all 5 minute samples is 61.0 mgm/m3,
and for all 10 minute samples is 81.6 mgm/m3. The 5 minute samples average
about twice the concentration measured in 2 minute samples and the 10 minute
samples average nearly triple the concentration measured in 2 minute samples.
Table 6. Average stack concentration (uncorrected for line
losses) measured by the parallel filter with 87.5%
dilution on August 31. Note the strong influence
of sampling time.
Sampling Time, Min.
O ^4
to K
0 iJ
0) .
-------
H
m
73
2
m
z
n
100
90
80
<9
O
70
2 60
u
o
o
50
40
§ 30
UJ
20
10
12:00
\.
SAMPLING TIME
10 MINUTES
5 MINUTES
2 MINUTES -
I
13100
14:00 is:oo
TIME OF DAY
ie:oo
17:00
Figure 33. Measured stack concentration (uncorrected for line losses) measured by the
parallel filter with 87.5% dilution on August 31. Note the strong in-
fluence of sampling time.
I
vj
NO
-------
-73-
2. Particles can build up in the sampling tube upstream of the
valve between runs, resulting in a sudden slug of particles
at the start of a run. Assuming that the time between runs
remains approximately constant, this would result in pro-
portionally extra particles for the 2 minute runs, an effect
directly opposite from our data.
3. Changes could be taking place on the filter media which depend
upon particle loading and/or time. Weight could be added to
the filter by adsorption or absorption of vapors or by chemi-
cal reactions at the media-gas phase or particle-gas phase
boundary resulting in the precipitation of a previously gaseous
material. Weight could be subtracted from the filter by the
reverse of those processes. For example, we have observed
major decreases in loaded filter weight when it was placed in
a low humidity environment for a few minutes. We have also
noticed major increases in loaded filter weight when it was
placed in a moderate-high humidity environment for a few minutes.
These changes were definitely greater than the equivalent change
of a clean filter. These observations are mentioned only to in-
dicate that vapors can cause major changes in maasured filter
loadings outside the stack environment. We suspect even greater
effects within the effluent stream, including not only water phase
changes, but also other chemical changes.
4. Condensation could occur on the cool sampling tubes between the
diluter and the filter holder at the beginning of each run. When
the tubes are warm later in the run, little or no condensation may
occur. The dilution system was included in the sampling system
specifically to reduce or eliminate condensation in the effluent
stream as the sample cools to room temperature. Yet, the sampling
line loss experiment reported later in this discussion indicates
very significant amounts of solid particles deposited on the in-
side of the tubes between the diluter and the filter holder. It
is difficult to explain such large line losses without condensation.
Thus, condensation remains one of the most likely explanations for
the lower concentration measurements with 2 minute samples.
5. Isokinetic sampling conditions may not be maintained at the point
where the two parallel samples remove their respective samples
from the main stream of effluent removed from the stack. However,
isokinetic sampling conditions do not vary with changes in sampling
time, but instead with changes in undiluted aerosol flow rate (i.e.,
diluted aerosol flow rate and dilution ratio). Since undiluted
aerosol flow rate of the parallel filter line did not change at
all during the August tests, nonisokinetic sampling would not have
any significant effect on the results.
THERMO-SYSTEMS INC
-------
-74-
The most probable cause of the lower stack concentrations measured with
the short sampling time is No. 4 and/or No. 3 above with No. 1 having a
possible secondary effect. No further conclusions can be drawn without
further experiments.
It is important to note that, although sampling time had such a large
influence on measured concentration, even when both parallel lines were
operated with identical conditions, both filters consistently collected
nearly the same weight of material. Thus, although the phenomena should
certainly be investigated further, the results of the May instrument evalu-
ation program are almost certainly not significantly affected.
4.2.d. Particle Line Losses
We attempted experimentally to obtain an estimate of particle loss in
the sampling lines. On each of the parallel lines, we washed the two straight
tubes and the elbow which connects the diluter with the filter holder before
the August tests. After the entire set of tests were completed, we again
washed the components with alcohol, carefully collecting the wash and filter-
ing out the particles onto a weighed glass fiber filter. After the filter
dried, we reweighed it, determining the weight of collected particles. The
sampling line components were identified as shown in Figure 34. The result-
BETA LINE
J
FLOW
NO. 3
FLOW
VERTICAL
PARALLEL LINE
LINE
ORIENTATION
NO. I HORIZONTAL
J
i,
NO. 6
NO. 4
NO. 2
NO. 5
Figure 34. Identification of sampling system components for line loss
tests. These 3 components connected the valve just down-
stream of the diluter with the beta filter or parallel filter.
THERMO-SYSTEMS INC
-------
-75-
ing weights are shown in Table 7. Also shown in Table 7 are the approximate
total particle weights collected on all filter samples during the entire
August test series.
Table 7. Measured weight of particulate material collected for
line loss measurement from sampling system components
identified in Figure 34.
Part No. Deposit Collected, mgm
1 4.652
2 35.657
3 32.869
Total on Parts 73.178
Total on Beta Filters 227.000
4 49.648
5 118.304
6 . 39.655
Total on Parts 207.607
Total on Parallel Filters 404.000
The sampling line loss in the beta filter line amounted to about 24% of
the total filter weights plus sampling line loss. The parallel filter line
loss was 34% of the total particle weight collected in that sampling line.
Although the technique may not be precise and most of the line loss may have
occurred during a small number of runs, the weighed amounts should be correct
to within better than a factor of 2. The line losses in both sampling lines
were greater than expected and may lend greater credibility to condensation
as a possible explanation for lower measured concentration with short sampling
time (Item 4 in earlier discussion). The parallel filter line was always
operated at 75% dilution and 50 LPM diluted aerosol flow rate while the beta
filter line was operated half of the time at 87.5% dilution. Thus, the beta
filter line carried about half as much particulate mass as the parallel filter line
during half of the runs. Although this may explain the greater measured line
loss in the parallel filter line, the line loss measurement technique is
probably not precise enough to make the difference significant.
The amount of line loss found in each of the 3 system components may offer
clues to the collection mechanism. In both sampling lines, the horizontal tube
collected the greatest amount, probably largely by gravity settling. The 2
vertical tubes collected similar amounts of material, possibly by condensation
and turbulent impaction. We cannot explain the much larger line loss in the
elbow of the parallel line, greater than 10 times the line loss in the elbow of
the beta filter line.
THERMO-SYSTEMS INC.
-------
-76-
We need more experimental data before we can draw meaningful conclusions
about particle line losses.
4.2.e. Parallel Filter Tests in May
Among the data covering the May beta instrument tests, which will be
presented in the next section, the reader will find a number of runs with
2 parallel filters in the sampling lines, just as in the 75% dilution, 50
LPM portion of the August runs. It must be understood that the first of
these runs were not meant to evaluate the system, but rather to train the
operating personnel and to work out any possible system problems which had
developed during the long period of non-use. Most of the other parallel
filter runs dispersed throughout the instrument tests were intended only
to verify that nothing in the system had gone grossly wrong. However, one
set of parallel filter runs was made for the purpose of evaluating the
system on May 24, the last day of beta instrument testing. The good
correlation between the stack particle concentrations measured by the two
filters can be seen in Figure 35. Notice that the sampling time for these
runs was only 3 minutes. The scatter in the data compares very well with
the scatter found in the August tests at 75% dilution and 50 LPM total
diluted aerosol flow rate. The average ratio of filter weights R-^ is just
over 1.0, similar to the August runs. This data provides experimental
evidence that the May and August operating conditions were comparable and
that there was no major equipment problem which developed between the 2
sets of tests. The May 24 data also provides additonal evidence that the
2 parallel systems correlate very well when operated with identical dilution
ratios, flow rates, and sampling times, conditions which did not exist dur-
ing most of the August runs.
THERMO-SYSTEMSINC
-------
-77-
BOTH FILTERS
OPERATED AT-
75% DILUTION
50LPM
3 MINUTES
MAY 24,1972
CD
8O 100 120 I4O 160 180
PARALLEL FILTER, MGM/M*
200 220 240
Figure 35. Stack concentration as measured by the beta filter
plotted against stack concentration as measured by
the parallel filter for May 24 runs. The diagonal
line represents perfect (1:1) correlation. Both
filters were operated at the same conditions.
THERMO-SYSTEMS INC
-------
-78-
SECTION 5. FIELD EVALUATION OF TWO PROTOTYPE BETA INSTRUMENTS
5.1 PROCEDURE
The Environmental Protection Agency furnished two prototype beta instruments,
developed under contract by GCA Corporation and Industrial Nucleonics Corp. for
field evaluation in the experimental stack facility.
The test set-up for the evaluation of the two beta instruments was identical
to the set-up for the parallel filter tests described in Section 4 with the test
instrument replacing the beta filter. In most runs, the TSI flowmeter plumbed
into the system downstream of the test instrument was used to measure and control
total diluted aerosol flow rate during a run. When the swirlmeter supplied with
the beta instruments was used to measure and control total diluted aerosol flow
rate, the data reported here will note that fact.
The two beta instruments were operated in the manual mode. Before every run,
the instrument made its zero beta count. The instrument was then prepared for the
sampling step and the run was begun. The total diluted aerosol flow rate was
monitored and maintained at a constant value throughout the run. Considerably more
adjustment was necessary to maintain constant flow through the beta instruments
than through the parallel filter, presumably because of the smaller filter face
area and subsequent faster loading in the beta instrument. After the valves were
closed ending the run, the instrument made its final beta count and printed out
its measured values of total particle loading (weight) and measured particle mass
concentration (diluted). Preparations for the next run could then begin.
The operating procedure for the parallel filter was identical with the
procedure used in the August sampling facility tests reported in Section 4.
5.2 RESULTS AND DISCUSSION
Tests were conducted with the Industrial Nucleonics (IN) beta instrument
on May 15 - 18 and May 23 - 24. Tests were conducted with the GCA Corporation
(GCA) beta instrument2 on May 22 - 23. The instruments were available only for
the 2-week period, limiting the tests and evaluation which could be performed.
However, we were able to fulfill the original objective of the experimental
portion of this contract: choose the most promising technique for particle mass
concentration measurement in the stack of large coal-fired combustion sources
(See Vol. I and II of this report) and experimentally prove that the technique
is feasible.
1. Duke, Charles R., and Cho, Boong Y., "Development of a Nucleonic Particulate
Emission Gauge", Final Report prepared for Environmental Protection Agency
under Contract No. 68-02-0210, Industrial Nucleonics Corp., 650 Ackerman Rd.,
Columbus, Ohio 43202, Feb. 1972.
2. Lilienfeld, P., and Dulchinos, J., "Vehicle Particulate Exhaust Mass Monitor",
Final Report prepared for Environmental Protection Agency under Contract No.
68-02-0209, GCA Corp., GCA Technology Div., Bedford, Mass. 01730, Mar. 1972.
THERMO-SYSTEMS INC
-------
-79-
Complete data for all runs conducted during May 10 - 24 are shown in
Appendix B. As mentioned in an earlier discussion, many parallel filter
runs were made to train operators, to make sure the facility was still
reliable after many months of idleness and modification, and to make sure
nothing had gone grossly wrong with the facility during these tests. How-
ever, this discussion will be limited to presenting the beta instrument
data and analyzing those results.
Figure 36 shows the comparison between stack concentration measured by
the GCA instrument and the stack concentration measured by the parallel filter.
We rejected the data from Runs 76 - 77 and 84 - 87 because, even though the GCA
instrument data appears acceptable, the concentration measured by the parallel
filter appeared significantly higher or lower than data obtained before or
after those runs. During Run 95, an operator error lead to the destruction
of the Geiger-Muller tube, disabling the instrument for the remainder of the
tests.
The 13 valid data points shown in Figure 36 display excellent correla-
_tion between the GCA instrument and the parallel filter. The average ratio
R (Rr/-A = stac^ particle concentration measured by GCA instrument ^ stack
particle concentration measured by parallel filter) was 0.982 and the standard
deviation of R was 0.0437. The correlation between 2 parallel filters in
the August tests were not significantly better than this.
Our only recommendations for improvement after 11/2 days of tests are
minor: 1) protect the G.M. tube from blowout by the mishandling of the vacuum
pump and 2) use an accurate flowmeter downstream of the particle collector.
One must guard against becoming too optimistic about an instrument which
displays such good correlation in only 1 1/2 days of tests at a rather con-
stant particle concentration level on a single stack. Somewhat poorer results
in earlier tests with artifically-generated uranine aerosol were reported to
the authors.3 Much more testing is necessary in a wide variety of stacks
operated at a variety of conditions before one could conclude that any instru-
ment is ready for duty as a continuous monitor of particle mass concentration
in a stack. However, the results of this test highly recommend the GCA beta
instrument for such further testing.
Figure 37 shows the comparison between the stack concentration measured by
the IN instrument and the stack concentration measured by the parallel filter.
A number of data points were rejected because the parallel filter concentration
was significantly higher or lower than runs just before or after the run, because
3. Herling, R. J., Environmental Protection Agency, private communication to
the authors, May 1972.
THERMO-SYSTEMS INC.
-------
-80-
BOTH UNITS
OPERATED AT:
75% DILUTION
50 LPM
©=2 MINUTES
ffi=5 MINUTES
A=7 MINUTES
MAY 2223,1972
RAVG=0.982
8=0.0437
8O 100 I2O 140 160 180
PARALLEL FILTER, MGM/M3
2OO 220 240
Figure 36. Stack concentration measured by the GCA beta instrument
compared with stack concentration measured by the parallel
filter. The diagonal line represents perfect (1:1) corre-
lation.
THERMO-SYSTEMS INC
-------
-81-
240
1 \ 1 1 1 1 1
20 40 60
MAY 16.1972
:|0 MINUTES
80 100 120 140 160 180
PARALLEL FILTER, MGM/M3
BOTH UNITS OPERATED AT
75% DILUTION. 50 LPM
^^^^ MAY 17,1972
200 220 240
MAY 23,1972
MINUTE
-a-r3.U5 MINUTES
@ = 5 MINUTES
MAY 24,1972
RAtf8=0.499
S=0.055I
-A-
A-
=2 MINUTES
:5 MINUTES
10 MINUTES
RAV6=O.609
S-O.0729
ALL 5 AND 10
=l MINUTE
- = 2 MINUTES
= 5 MINUTES
RAV6= 0.601
S=O.0663
MINUTE RUNS
R =0.509
8=0.0959
ABOVE!
Figure 37. Stack concentration measured by the IN beta instrument
compared with stack concentration measured by the parallel
filter. The solid diagonal line represents a perfect (1:1)
correlation and the dashed diagonal line represents a
parallel filter measurement of twice as high as the beta
instrument measurement.
THERMO-SYSTEMS INC
-------
-82-
it was the first run of the day, or because the IN instrument did not display
or print any data at the end of the run. The data points shown in Figure 37
are only from validated runs. Note that some of the higher data points are
from 1 or 2 minute runs. These could have been rejected but may offer some
clue as to why the correlation was not better.
The IN beta instrument measured stack Concentrations about half as high
as the parallel filter. The average ratio R for any single day of tests
ranged from 0.428 to 0.609. The standard deviation of R^N for any single
day of tests ranged from 0.0536 to 0.0729, all acceptable values. However,
the standard_deviation for all 5 and 10 minute samples shown in Figure 37 is
0.0959 with RTXT = 0.507.
IN
We have no explanation for the factor-of-2 low measurements by the IN
beta instruments. Nearly perfect correlation in earlier tests with arti-
ficially-generated uranine aerosol was reported to the authors.* Since the
GCA instrument correlated well with the parallel filter as did an identical
parallel filter, it would appear that the problem is within the IN instru-
ment and is not a sampling problem. Both the GCA and IN instrument used the
same filter media and both use a swirlmeter upstream of the filter for total
diluted aerosol flow rate measurement. However, a TSI mass flow meter was
used to monitor and control the flow rate with both instruments. The plumb-
ing for the aerosol sample inside the two instruments was similar, but not
identical. We suspect that the somewhat greater scatter of data points with
this instrument may be improved when the other low measurement problem is
corrected.
The IN instrument, as well as the GCA instrument, was not designed
specifically for stack measurements, but rather for auto exhaust measurements.
It was intended as a versatile laboratory tool, not as a rugged field instru-
ment. In the case of the IN instrument, we found that the electronics mal-
functioned often when the temperature around the instrument exceeded about
80°F. The malfunction was frustrating since the operator was not aware of
a malfunction until no data appeared at the end of a run. Such malfunctions
are not mentioned as a criticism of the beta radiation sensing technique, but
as a deficiency of this specific instrument, limiting its usefulness for stack
measurements.
In summary, we strongly recommend extensive stack testing with the GCA
instrument under a variety of conditions. We recommend further laboratory work
on the IN instrument to find the reason for the lack of correlation in these
tests. The discovery of the reason for this problem may result in better, more
trouble-free instruments in the future.
THERMO-SYSTEMS INC
-------
-83-
SECTION 6. EVALUATION OF TRANSMISSOMETER TECHNOLOGY
6.1 INTRODUCTION
The transmissometer is the only instrument which has been used to any
significant extent for monitoring the particle loading in effluents from
coal-combustion facilities. There is disagreement regarding what constitutes
a good instrument. There is even more disagreement regarding what trans-
missometers measure.
Transmissometers measure the optical density of the light path of the
instrument. The recommended list of features below should result in the
best instrument for measuring optical density of flue gases containing par-
ticles. The recommendations include nothing which cannot easily be done with
commercial technology available in January 1971. Correlation with any other
particulate parameter (such as particle surface area or mass concentration)
requires interpretation of the specific situation.
It is much more difficult to specify a transmissometer which correlates
well with average particulate mass concentration within the light path. The
authors feel that, without considerable experimental testing and development,
the preferred approach is to design an instrument which does a good job of
measuring the parameter which transmissometers are intended to measure: the
optical density of the light path.
6.2 RECOMMENDED DESIGN FEATURES
1. Use a stablized light source.
a. Choose a wavelength which avoids interference by infrared
sources within the duct.
2. Use a detector that yields an electrical output signal which is
linear with incident light intensity.
3. The combination of light source and detector should operate over
a well-defined wavelength spectrum which is constant with time.
4. Use pinhole aperatures (or small acceptance angles) to avoid
collecting scattered and extraneous light and to make the measure-
ment of optical density independent of path length.
5. Compensate automatically for dirty windows, light source aging,
line voltage fluctuations and electronic drift.
a. When the system is out of tolerance because of dirty windows,
etc., a warning should tell the operator to correct the situation.
THERMO-SYSTEMS INC
-------
-84-
b. Keep the source, detector, and all surfaces within the light
path clean; provide easy cleaning access to such surfaces
when they become dirty; do not install the instrument so that
particles can settle onto surfaces within the light path.
c. Provide an automatic zero and span check periodically during
operation.
d. Consider heating all surfaces in contact with flue gas to
prevent vapor condensation.
e. Consider heating any purge air to prevent condensation of
vapors on window and corrosive surfaces.
6. Design the source and detector to align with each other at all time,
even during plant shutdown and start up when severe duct distortion
may occur.
7. Make the instrument compatible with its environment:
a. All exterior housings must be weatherproof.
b. All seals between flue gas and sensitive apparatus must
be gas tight.
c. Entire assembly must withstand duct vibrations.
d. All components in contact with flue gas must be noncorrosive.
6.3 MINIMUM RECOMMENDATIONS GOVERNING INSTRUMENT INSTALLATION AND OPERATION
1. Install the light beam in a representative axis of the duct:
a. Consider particulate stratification in horizontal ducts.
b. Consider flue gas and particulate concentration profiles.
c. Consider bends, expansions, contractions, and obstructions
in the duct.
2. Check the instrument thoroughly after installation and:
a. Check zero and span every time the recorded measurement is
checked.
b. Check for dirty windows and other such malfunctions weekly.
c. Check all parts in contact with flue gas for corrosion every
six months.
THERMO-SYSTEMS INC
-------
-85-
3. If mass concentration or mass emissions rate correlation is
necessary, complete mass calibration is necessary:
a. After installation.
b. Whenever certain plant operating conditions change significantly.
c. Yearly.
6.4 PRESENT COMMERCIAL INSTRUMENTS
There are no commercial transmissometers available which use all the design
features recommended above. However, three West German companies make instruments
which appear to meet most requirements:
Irwin Sick (Model RM-3g)*
Optik Elektronic
Neuried, West Germany
(Licensed to: Intertech Corporation
262 Alexander Street
Princeton, N.J. 08540)*
Durag Apparatebau Gmbh (Model D-R 110)
2 Hamburg 61
Killanstrasse 105, West Germany
AEG (Model R 72)
(address unavailable)
The best of these three is probably the Sick instrument. Figure 38 shows the
principle of operation of the instrument.
Several comparative evaluations of these instruments in German power stations burn-
ing lignite fuel show that the Sick instrument yields the most reproducible measure
1189» 22i> 1188>
ments and is the most reliable for long-term operation.
The Sick instrument when operating perfectly, correlates reasonably well with
particulate mass concentration measured by manually-collected filter samples.
However, notice that, in the concentration range where modern, controlled stacks
must operate (below 150 milligrams per cubic meter), the accuracy quickly
deteriorates until estimated errors are larger than the measurement itself.
The authors suggest that an "everclear" window13'487 at the point where the
light beam enters and leaves the flue gas on both ends of the stack may help to
prevent contamination of mirror and window surfaces. However, Duwel1254 indicates
that this may not be a significant problem if a purge air system is used. Other
*As this report goes to press, a newer model, RM-4, has become available and is
now marketed by Lear Siegler, Inc., 32 Denver Technological Center, Englewood,
Colorado 80110.
THERMO-SYSTEMS INC.
-------
H
m
50
l/l
m
2
yi
z
o
P/ane mirrow
b) Modulator disk
c) Reflector
d) Iris diaphragm
I
u A
^J
I
^ \ / fc\
e) Semitransparent mirrow
f) Photocell
\
oo
T
Figure 38. Schematic of optical system on Sick Model RM 3g transmissometer.
-------
-87-
possible Improvements are the light source and detector. (Sick uses red light
and a photodiode detector.) There is no assurance that other choices would be
better. Duwel indicated to the authors that Sick has overcome the source and
reflector alignment problem in recent models. Improvements in the mechanical or
electrical assembly may be possible, but the authors can offer no suggestions
without testing the instrument. Duwel indicates that the Sick instrument is
now quite reliable once it is installed.
All other known commercial instruments, including all American models, place
the light source on one side of the flue gas duct and the detector on the opposite
side. Thus, no correction is possible for dirty windows and no automatic zero and
span checks are possible. The result is a poorly defined measurement which does
not correlate with opacity, optical density, Ringelmann number or particulate
mass concentration. No comparative tests of American models have been carried out,
but such tests would probably show poor reproducibility and even poorer accuracy.
Thus, the Sick Model RM 3g appears to be the best transmissometer available
in January 1971. However, all supporting data is from German lignite-fired
sources. Further comparative tests on coal-fired sources is strongly needed.
6.5 MEANING OF THE MEASUREMENT
The instrument specified in Section 6.2 senses the optical density of the light
path. The optical density (or opacity) is closely related to the total cross-
sectional area of the particles in the light beam if all the particles have diameters
greater than 1 or 2 microns (geometrical scattering regime). However, submicron
particles present in coal-combustion effluents affect the optical density of the
light beam more than an equal total cross-section area of large particles. There-
fore, fluctuations in the relatively small amount of submicron particulate material
present in any stack will introduce significant and unavoidable error into the
total cross-sectional area measurement and make the correlation of measured optical
density with any other parameter of the particulate cloud very difficult. Since
oil-combustion emissions contain a much higher proportion of submicron particles,
the difficulty in correlating optical density with other particulate properties
when oil is used as fuel is even greater than with coal.
The optical density measured by a transmissometer has no relationship with
the familiar Ringelmann number for several reasons:
1. Ringelmann number is based on dark-colored plumes while the
transmissometer measures optical density regardless of the color.
2. A transmissometer senses particles as they exist within the stack
(high temperature, non-atmospheric gaseous composition, etc.) while
Ringelmann number characterizes plumes in the atmosphere and is
affected by extraneous factors such as steam formation, condensation
of various vapors, agglomeration of particles, the color of the sky
behind the plume, and the direction from which the plume is illuminated.
THERMO-SYSTEMS INC
-------
-88-
Many transmissometer manufacturers claim correlation with Ringelmann number
and place Ringelmann markings on the readout meters. This is not a valid
practice. Nearly all emissions sources experience conditions when Ringelmann
number and optical density have entirely opposite trends. For example, the
plume from a coal-fired source operating in the winter with ambient air at
+ 10°F appears white because steam overwhelms the fly ash in the plume. The
Ringelmann number may be 0 or 1 indicating "clean" emissions while the optical
density within the stack may be very high indicating very "dirty" emissions.
Cement plant emissions are the classic example of this phenomena.
A transmissometer must be located in a representative portion of the
stack or duct. Because a transmissometer measures an average optical density
within the light beam, it has a potential advantage over instruments which
sample or measure concentration at a point. However, this advantage is can-
celled in many installations because of poor placement of the transmissometer.
The flue gas velocity and particulate concentration profiles are just as
important for locating a transmissometer as for locating point sampling instru-
ments.
A transmissometer has one very clear advantage over most other instruments.
It does not require a sample of flue gas to be removed from the duct. It can
measure the optical density of the particles as they exist in the duct without
disturbing them in any way.
6.6 CORRELATION OF THE MEASUREMENT WITH MASS
The optical density as measured with the instrument described in Part 2
of this section will correlate moderately with the mass of particles within
the light beam under certain conditions. The problem is that the operator
cannot easily distinguish the periods of good correlation from those with bad
correlation within any given stack. Thus, he is never sure of the accuracy of
the mass correlation.
Best mass correlation occurs when all particles have diameters greater
than 2 Um, when the volume-surface diameter remains constant, and when the
specific gravity of the particles is known and remains constant. As stated
earlier, the small mass of submicron particles present in coal-combustion
effluents affect the optical density of the light beam more than an equal
mass of large particles, making the transmissometer too sensitive to fluctu-
ations in submicron particle concentration. If the mean volume-surface dia-
meter decreases by a factor of two with all other parameters constant, the
indicated optical density increases by a factor of two. The particle size
distribution (and, thus, the mean volume-surface diameter) of particles in a
flue gas stream does not remain constant during normal power plant operating
conditions. Factors which can cause significant changes in particle size
distribution are:
1. The fraction of rated capacity at which the plant is operating,
THERMO-SYSTEMS INC
-------
-89-
2. Coal-burning efficiency,
3. Type and composition of coal used,
4. Sootblowing,
5. Type and efficiency of emissions control equipment, and
6. Operating conditions of control equipment including whether
rapping is occurring and whether all banks of an electro-
static precipitator are operating at optimum conditions.
Minor, unnoticed changes in several of these factors can cause major changes in
the particle size distribution which can, in turn, render any mass calibration
of a transmissometer invalid. These same factors can affect the specific gravity
of the particles. The particles in coal-combustion emissions reportedly vary
from hollow spheres (specific gravity under 0.5) to very dense solids (specific
gravity of about 10.0). The average specific gravity can change drastically with
minor unnoticed changes in fuel composition or combustion efficiency, making a
mass calibration useless. Perhaps the worst feature of all these problems is
that the operator usually does not know if the mass calibration is valid at a
specific time.
Every installation requires an extensive calibration for even rough
correlation with particulate mass concentration. Calibrations must be repeated
periodically, probably at least once every year. Because of the many effects
discussed above which result in degradation of the mass correlation, each plant
operating condition requires a different calibration. A calibration consists
of a complete mass concentration and gas velocity characterization of the stack.
The minimum procedure for one calibration consists of a traverse of the stack
at the transmissometer installation with a velocity probe and a manual filter-
sampling probe. An experienced crew of 3 people can normally calibrate one
installation in about 1-2 weeks. Any shortcut to a complete calibration such
as this results in severe degradation of the accuracy of the mass correlation.
6.7 CONCLUSIONS
A well-designed transmissometer measures the optical density of the light
path. Transmissometers cannot measure particulate mass concentration without
extensive, periodic calibration. When significant changes occur in the physical
properties of the particles, a new calibration must be performed for correlation
with particulate mass concentration.
THERMO-SYSTEMS INC
-------
-90-
6.8 ACKNOWLEDGMENT
The authors wish to thank Professor B.Y.H. Liu of the Mechanical Engineering
Dept., University of Minnesota, for many helpful suggestions and for valuable
critical review of an early manuscript of this section.
6.9 REFERENCES
487 Crosse, P. A., Lucas, D.H., and Snowsill, W. L., "Design of an
"Everclean Window" for the Observation of the Optical Density
of Flue Gas", Journal Inst. of Fuel, V. 34, no. 250, p. 503-505
(1961).
225 Duwel, L., "Latest State of Development of Control Instruments for
the Continuous Monitoring of Dust Emissions", Staub-Reinhalt der Luft
(Engl. Trans.), V. 28, no. 2, p. 42-53 (Mar 1968).
1254 Duwel, L., "Comparative Investigations of Different Measuring
Principles for Continuous Monitoring of Dust Emissions of Lignite
Fired Steam-Boiler", Paper presented at Second International Clean
Air Congress, Washington, D. C., Dec. 6 - 11, 1970.
13 Lucas, D.H., "I. Air Pollution Measurements", Phil. Trans. Royal
Society of London, V. A257, p. 143-151 (1969).
1189 Schnitzler, H., "Messtand fur die Prufung und Kalibrierung von
Registrierenden Staub - und Gasmessgeraten in einem Steinkohlengefeuerten
Kraftwerk", SchrReihe Ver. Wass. - Boden Lufthyg. Berlin-Dahlem, V. 33,
Stuttgart (1970).
1188 Schnitzler, H., "Untersuchungen uber die Eignung Registrierender
Cerate zur Messungdes Staubgehaltes in Abhasen", Gesundheit-Ingenieur,
V. 10, p. 307-309 (Jan 1970).
Sem, G. J., Borgos, J. A., Olin, J. G., Pilney, J.P., Liu, B. Y. H.,
Barsic, N., Whitby, K. T., and Dorman, F. D., "State of the Art: 1971
Instrumentation for Measurement of Particulate Emissions from Combustion
Sources, Volume II: Particulate Mass - Detail Report", Thermo-Systems
Inc., St. Paul, Minn., report to EPA under Contract CPA 70-23 (1971).
Anon, "Operational Handbook, Smoke Density Measuring System RM 3 g",
Intertech Corp., Princeton, N.J. (U.S. Licensee for Erwin Sick
transmissometers).
THERMO SYSTEMS INC
-------
-91-
SECTION 7. APPENDICES
THERMO-SYSTEMS INC
-------
APPENDIX A
COMPLETE DATA FOR THE AUGUST TESTS
OF THE SAMPLING SYSTEM
THERMO SYSTEMS INC.
-------
Table A. Complete data for the August tests of the sampling system.
Runs 1-60 plumbed in with same configuration as beta instruments in May tests (see text).
Micrograms
Run
No.
1
2
3
4
5
6
7
8
9
10
11
Filter
No.
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
Operate
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
Filter,
Beta or
Parallel
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
August
Date
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
Time
of Day
1119
1119
1149
1149
1159
1159
1209
1209
1222
1222
1233
1233
1247
1247
1301
1301
1316
1316
1330
1330
1344
1344
Min,
Elapsed
Time
5
5
5
5
5
5
5
5
5
5
10
10
10
10
10
10
10
10
10
10
2
2
LPM,
Diluted
Aerosol
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
Dilution
Air
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
Liters,
Diluted
Aerosol
250
250
250
250
250
250
250
250
250
250
500
500
500
500
500
500
500
500
500
500
100
100
Micrograms ,
Particle
Weight
5371
5989
7459
8155
8555
8678
8641
8839
9134
9051
19433
19780
17855
18180
17133
16895
17730
18055
18097
18087
2351
2331
Per M3,
Stack Con-
centration
85.9
95.8
119.3
130.5
136.9
138.8
138.3
141.4
146.1
144.8
155.5
158.2
142.8
145.4
137.1
135.2
141.8
144.4
144.8
144.7
94.0
93.2
Ratio:
B. Filter Cone.
P. Filter Cone.
.897
.914
.986
.978
1.009
.983
.982
1.014
.982
1.001
1.009
U>
-------
(continued)
Table A. Complete data for the August tests of the sampling system.
Runs 1-60 plumbed in with same configuration as beta instruments in May tests (see text).
M
o
Micrograms
Run
No.
12
13
14
15
16
17
18
19
20
21
22
Filter
No.
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
4J
03
M
a)
a
o
B
B
B
R
B
B
B
B
B
B
B
B
B
B
B
R
B
B
B
B
B
R
Filter,
Beta or
Parallel
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
August
Date
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
Time
of Day
1017
1017
1024
1024
1030
1030
1036
1036
1054
1054
1103
1103
1112
1112
1137
1137
1151
1151
1205
1205
1219
1219
Min,
Elapsed
Time
2
2
2
2
2
2
2
2
5
5
5
5
5
5
10
10
10
10
10
10
2
2
LPM,
Diluted
Aerosol
50
50
50
50
50
50
50
50
30
50
30
50
30
50
30
50
30
50
30
50
30
50
Dilution
Air
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
Liters,
Diluted
Aerosol
100
100
100
100
100
100
100
100
150
250
150
250
150
250
300
500
300
500
300
500
60
100
Micrograms,
Particle
Weight
1627
1995
1666
1443
1617
1501
1460
1515
4312
7832
3290
6798
2497
6991
9129
16347
8018
15417
8564
16350
1110
1929
Per MJ,
Stack Con-
centration
65.1
79.8
66.6
57.7
64.7
60.0
58.4
60.6
115.0
125.3
87.7
108.8
66.6
111.9
121.7
130.8
106.9
123.3
114.2
130.8
74.0
77.2
Ratio:
B. Filter Cone.
P. Filter Cone.
.816
1.155
1.077
.964
.918
.806
.595
.930
.867
.873
.959
-------
(continued)
Table A. Complete data for the August tests of the sampling system.
Runs 1-60 plumbed in with same configuration as beta instruments in May tests (see text).
Run
No.
23
24
25
26
27
28
29
30
31
32
33
Filter
No.
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
Operator
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
Filter,
Beta or
Parallel
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
August
Date
30
30
30
30
30
30
30
30
31
31
31
31
31
31
31
31
31
31
31
31
31
31
Time
of Day
1225
1225
1231
1231
1244
1244
1255
1255
1100
1100
1109
1109
1122
1122
1135
1135
1148
1148
1153
1153
1158
1158
Min,
Elapsed
Time
2
2
2
2
5
5
5
5
5
5
10
10
10
10
10
10
2
2
2
2
2
2
LPM,
Diluted
Aerosol
30
50
30
50
20
50
20
50
20
50
20
50
20
50
20
50
20
50
20
50
20
50
%
Dilution
Air
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
75
Liters,
Diluted
Aerosol
60
100
60
100
100
250
100
250
100
250
200
500
200
500
200
500
40
100
40
100
40
100
Micrograms,
Particle
Weight
1041
1998
1021
1894
2224
6996
1652
6450
2366
7251
4194
12189
3124
8966
3141
8863
287
684
172
681
214
649
Micrograms
Per M3,
Stack Con-
centration
69.4
79.9
68.1
75.8
89.0
111.9
66.1
103.2
94.6
116.0
83.9
97.5
62.5
71.7
62.8
70.9
28.7
27.4
17.2
27.2
21.4
26.0
Ratio:
B. Filter Cone.
P. Filter Cone.
.869
.898
.795
.641
.816
.861
.872
.886
1.047
.632
.823
i
yD
Ln
-------
(continued;
Table A. Complete data for the August tests of the sampling system.
Runs 1-60 plumbed in with same configuration as beta instruments in May tests (see text).
Run
No.
34
35
36
37
38
39
40
41
42
43
44
Filter
No.
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
Operator
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
Filter,
Beta or
Parallel
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
August
Date
31
31
31
31
31
31
31
31
31
31
31
31
31
31
31
31
31
31
31
31
31
31
Time
of Day
1212
1212
1220
1220
1228
1228
1236
1236
1249
1249
1302
1302
1315
1315
1320
1320
1326
1326
1341
1341
1349
1349
Min,
Elapsed
Time
5
5
5
5
5
5
10
10
10
10
10
10
2
2
2
2
2
2
5
5
5
5
LPM,
Diluted
Aerosol
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
30
50
30
50
Dilution
Air
87.5
75
87.5
75
87.5
75
87.5
75
87.5
75
87.5
75
87.5
75
87.5
75
87.5
75
87.5
75
87.5
75
Liters,
Diluted
Aerosol
250
250
250
250
250
250
500
500
500
500
500
500
100
100
100
100
100
100
150
250
150
250
Micrograms,
Particle
Weight
1855
4234
1947
4433
1893
4244
4637
10199
4095
9619
4284
10300
396
889
278
835
360
843
1024
3865
977
3662
Micrograms
Per M3,
Stack Con-
centration
59.4
67.7
62.3
70.9
60.6
67.9
74.2
81.6
65.5
77.0
68.5
82.4
31.7
35.7
22.2
33.4
28.8
33.7
54.6
61.8
52.1
58.6
Ratio:
B. Filter Cone.
P. Filter Cone.
.877
.879
.892
.909
.851
.831
.888
.665
.855
.883
.889
-------
(continued)
Table A. Complete data for the August tests of the sampling system.
Runs 1-60 plumbed in with same configuration as beta instruments in May tests (see text).
Micrograms
Run
No.
45
46
47
48
49
50
51
52
53
54
55
Filter
No.
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
a
1-1
-------
(continued)
Table A. Complete data for the August tests of the sampling system.
Runs 1-60 plumbed in with same configuration as beta instruments in May tests (see text).
Run
No.
56
57
58
59
60
Filter
No.
295
296
297
298
299
300
301
302
303
304
Operatoi
B
B
B
B
B
B
B
B
B
B
Filter,
Beta or
Parallel
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
August
Date
31
31
31
31
31
31
31
31
31
31
Time
of Day
1539
1539
1552
1552
1605
1605
1610
1610
1615
1615
Min,
Elapsed
Time
10
10
10
10
2
2
2
2
2
2
LPM,
Diluted
Aerosol
20
50
20
50
20
50
20
50
20
50
% Liters,
Dilution Diluted
Air Aerosol
87.5
75
87.5
75
87.5
75
87.5
75
87.5
75
200
500
200
500
40
100
40
100
40
100
Micrograms
Micrograms, Per M ,
Particle Stack Con-
Weight centration
1302
10560
1354
10896
126
801
111
587
610
620
52.1
84.5
54.2
87.2
25.2
32.0
22.2
23.5
122.0
24.8
Ratio:
B. Filter Cone.
P. Filter Cone.
.617
.622
.788
.945
4.919
00
-------
-99-
APPENDIX B
COMPLETE DATA FOR THE MAY TESTS
OF TWO BETA INSTRUMENTS
THERMO-SYSTEMS INC
-------
Table B. Complete data for the May tests of two beta instruments.
Runs 1-23 plumbed in short and vertical downstream of diluter.
Runs 24 - 127 plumbed in with configuration necessary for beta instruments (see text).
o Beta**
« or Min, CFM
Run Filter
-------
(jonipj-c ce dut.ci ±or~ ttie May tests oi two beta ins Crunients
Runs 1-23 plumbed in short and vertical downstream of diluter.
Runs 24 - 127 plumbed in with configuration necessary for beta instruments (see text)
Run
No.
11
12
13
14
15
16
17
18
19
20
21
Filter
No.
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
Operatoi
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
Beta
or
Parallel
Filter
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
May
Date
11
11
11
11
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
15
15
Micro- Micrograms
Min, CFM % Liters grams, Per M3
Time Elapsed Diluted Dilution Diluted Particle Stack Con-
of Day Time Aerosol Air Aerosol Weight centration
1449 15 1.0
1513 30 1.0
1235 30 1.0
1321 20 1.0
1347 20 1.0
1412 20 1.0
1438 20 1.0
1502 20 1.0
1527 20 1.0
1552 20 1.0
1015 15 1.0
10088
50 425
9106
10900
75 1060
8285
15596
75 1060
12543
11811
75 707
9297
14820
75 707
12659
16645
75 707
14273
16332
75 707
14706
15790
75 707
14260
15049
75 707
13637
14333
75 707
12968
12735
75 530
14472
41.0
42.9
41.2
31.3
58.9
47.4
66.8
52.6
83.9
71.7
94.2
80.9
92.5
83.3
89.4
80.7
85.2
77.2
81.1
73.5
96.1
109.4
Ratio:
Beta Weight
P. Filter Wt. Comments
1.109
First run
1.316 @4:1
Ran 1 dry run
1.243 before this
1.270
1.170
1.167
1.111
1.108
1.104
1.105
(1.137) Trial Run. Reversed
0.880 flowmeters from
Run 20.
i
o
I
-------
(continued)
Table B. Complete data for the May tests of two beta instruments.
Runs 1-23 plumbed in short and vertical downstream of diluter.
Runs 24 - 127 plumbed in with configuration necessary for beta instruments (see text).
Run
No.
22
23
24
25
26
27
28
Filter
No.
43
44
45
46
47
48
49
50
51
n
o
4J
-------
Table B. Complete data for the May tests of two beta instruments.
Runs 1-23 plumbed in short and vertical downstream of diluter.
Runs 24 - 127 plumbed in with configuration necessary for beta instruments (see text).
o Beta
Micro- Micrograms
Run
No.
29
30
31
32
33
34
35
36
37
38
Filter
No.
52
53
54
55
56
57
59
58
60
61
62
63
64
65
66
67
td
p
0)
0
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
or
Parallel
Filter
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
IN
PF
IN
PF
IN
PF
IN
PF
May
Date
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
Time
of Daj/
1050
1110
1135
1155
1210
1225
1245
1300
1320
1335
Min,
Elapsed
Time
5
5
5
5
5
5
10
10
10
10
LPM,
Diluted
Aerosol
50
50
50
50
50
50
50
50
50
50
%
Dilution
Air
75
75
75
75
75
75
75
75
75
75
Liters
Diluted
Aerosol
250
250
250
250
250
250
500
500
500
500
grams,
Particle
Weight
13663
15387
10037
15982
11601
12690
11273
12551
11488
13612
10348
12390
9090
24950
12046
24801
12065
25438
8976
25390
Per M-3
Stack Con-
centration
219
246
160.6
256
185.8
203
180.6
201
184
218
165.6
198.4
72.7
199.7
96.5
198.4
96.5
203.4
71.8
203.1
Ratio:
Beta Weight
P. Filter Wt
.888
.628
.914
.898
.844
.835
.364
.488
.474
.354
Comments
IN plumbed
through swirl-
meter. Filter
tore, stuck to
screen.
Same as above
Stuck to screen,
soaked
Good run
Good run
Good run
IN plumbed
through swirl-
meter.
o
Co
-------
(continued^
Table B. Complete data for the May tests of two beta instruments.
Runs 1-23 plumbed in short and vertical downstream of diluter.
Runs 24 - 127 plumbed in with configuration necessary for beta instruments (see text).
Run
No.
39
40
41
42
43
44
45
46
47
48
49
Filter
No.
68
69
70
71
72
73
74
75
76
77
78
Operator
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
Beta
or
Parallel
Filter
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
May
Date
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
17
17
17
17
17
17
Time
of Day
1350
1405
1425
1440
1455
1515
1530
1545
1225
1240
1250
Min,
Elapsed
Time
10
10
10
10
10
10
10
10
5
5
5
LPM,
Diluted
Aerosol
50
50
50
50
50
50
50
50
50
50
50
%
Dilution
Air
75
75
75
75
75
75
75
75
75
75
75
Liters
Diluted
Aerosol
500
500
500
500
500
500
500
500
250
250
250
Micro-
grams,
Particle
Weight
12008
26533
11917
27605
11733
25847
12102
24613
11697
25757
9375
25499
11437
25349
8750
25015
6032
12300
6379
11679
6584
12258
Micrograms
Per M3
Stack Con-
centration
96.1
212
95.3
221
93.9
207
96.9
197
93.6
206
75
204
91.5
203
70
200
96.5
197
102.1
187.1
105.3
196.1
Ratio:
Beta Weight
P. Filter Wt
.453
.432
.454
.492
.454
.368
.451
.350
.490
.546
.538
Comments
Plumbed IN to
eliminate swirl-
meter
i
o
-JS
-------
(continued)
Table B. Complete data for the May tests of two beta instruments.
Runs 1-23 plumbed in short and vertical downstream of diluter.
Runs 24 - 127 plumbed in with configuration necessary for beta instruments (see text).
Run
No.
50
51
52
53
54
55
56
57
58
59
60
Filter
No.
79
80
91
92
93
94
95
96
97
113
114
Operatoi
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
B
B
B
B
Beta
or
Parallel
Filter
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
May
Date
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
18
18
18
18
Time
of Day
1300
1325
1350
1405
1420
1430
1440
1450
1505
1140
1150
Min, LPM,
Elapsed Diluted
Time Aerosol
5 50
5 50
5 50
5 50
3.115 50
5 50
5 50
5 50
5 50
5 50
5 50
%
Dilution
Air
75
75
75
75
75
75
75
75
75
75
75
Liters
Diluted
Aerosol
250
250
250
250
155.75
250
250
250
250
250
250
Micro-
grams,
Particle
Weight
6073
12623
6492
13515
6908
12411
4463
8022
5657
12908
5052
12561
12045
12608
5703
17100
6557
16979
Micrograms
Per M-* Ratio:
Stack Con- Beta Weight
centration P. Filter Wt . Comments
97.2
.481
202
103.9
.481
216
-
110.4
.556
198.6
114.5
.556
207
90.5
.438
207
80.9
.401
201
192.8
202
91.3 3 min. count time.
.333 power plant
274 normal
105 3 min. count time,
.386 power plant
272 normal
i
M
O
Ln
-------
(continued)
Table B. Complete data for the May tests of two beta instruments.
Runs 1-23 plumbed in short and vertical downstream of diluter.
Runs 24 - 127 plumbed in with configuration necessary for beta instruments (see text).
o Beta
Micro- Micrograms
Run
No.
61
62
63
64
65
66
67
68
69
70
Filter
No.
115
116
117
108
109
110
111
112
118
119
120
121
122
Cl)
M
-------
(continued)
Table B. Complete data for the May tests of two beta instruments.
Runs 1-23 plumbed in short and vertical downstream of diluter.
Runs 24 - 127 plumbed in with configuration necessary for beta instruments (see text).
Run
No.
71
72
73
74
75
76
77
78
79
80
Filter
No.
98
99
100
101
102
103
81
104
82
105
123
124
125
126
127
*GCA =
o Beta
4J
to or
cu Parallel
o Filter
B BF
B PF
B BF
B PF
B BF
B PF
B BF
B PF
B BF
B PF
G GCA*
G PF
G GCA
G PF
G GCA
G PF
G GCA
G PF
G GCA
G PF
May
Date
19
19
19
19
19
19
19
19
19
19
22
22
22
22
22
22
22
22
22
22
GCA Corporation beta
**Number shown in GCA
and all calculations
Time
of Day
1410
1422
1430
1450
1500
1145
1200
1220
1235
1410
Min,
Elapsed
Time
5
2 1/2
5
5
5
10
5
5
5
5
LPM,
%
Diluted Dilution
Aerosol
21.2
42.5
21.2
42.5
21.2
42.5
28.3
28.3
28.3
28.3
50
50
50
50
50
Air
75
75
75
75
75
75
75
75
75
75
Liters
Diluted
Aerosol
106
212.5
53
106.25
106
212.5
141.5
141.5
141.5
141.5
(568)**
500
(294)
250
(284)**
250
(282)**
250
(287)**
250
Micro-
grams,
Particle
Weight
2925
7622
1276
3680
3216
7853
4390
3464
3822
2541
25310
19560
13320
15862
11900
12795
11250
12060
12100
12547
Micrograms
Per M3
Stack Con-
centration
110.5
143.5
96.3
138.6
121.4
147.8
124
98
108
71.9
203
157
213
254
190
205
180
193
194
201
Ratio:
Beta Weight
P. Filter Wt . Comments
.770
.695
.822
1.27
1.50
1.295
0.840
0.930
0.933
NSP switched off
0.965 fly ash reinjec-
tion @- 1300
instrument
indicated flow
based
on TSI
, flow rate operated
flowmeter .
with TSI
flowmeter
i
0
i
-------
(continued)
Table B. Complete data for the May tests of two beta instruments.
Runs 1-23 plumbed in short and vertical downstream of diluter.
Runs 24 - 127 plumbed in with configuration necessary for beta instruments (see text).
Run
No.
81
82
83
84
85
86
87
88
89
90
91
Filter
No.
132
131
130
129
128
133
134
135
136
137
138
Operator
G
G
G
r,
G
G
G
G
G
r;
G
G
G
G
G
G
G
G
G
G
G
G
*See note
Beta
or
Parallel
Filter
GCA
PF
GCA
PF
GCA
PF
GCA
PF
GCA
PF
GCA
PF
GCA
PF
GCA
PF
GCA
PF
GCA
PF
GCA
PF
for run 76
May
Date
22
22
22
22
22
22
22
22
22
22
23
23
23
23
23
23
23
23
23
23
23
23
Time
of Day
1430
1445
1500
1520
1535
1015
1030
1045
1100
1120
1135
Min,
Elapsed
Time
5
5
5
5
5
5
5
2
5
7
5
LPM,
Diluted
Aerosol
50
50
50
50
50
50
50
50
50
50
50
%
Dilution
Air
75
75
75
75
75
75
75
75
75
75
75
Liters
Diluted
Aerosol
(290)*
250
(287)*
250
(289)*
250
(294)*
250
(288)*
250
(291)
250
(290)*
250
(118)*
100
(287)*
250
(407)*
350
(291)*
250
Micro-
grams ,
Particle
Weight
11430
11452
10930
11923
11100
11458
11930
8860
11300
15250
11800
15206
10490
11600
5400
5344
11480
12041
17020
16784
12520
12576
Micrograms
Per M3
Stack Con-
centration
183
183
175
191
178
183
191
142
181
244
189
244
167
186
216
214
184
193
194
192
200
201
Ratio:
Beta Weight
P. Filter Wt . Comments
0.997
0.918
0.969
1.348
0.741
Sampling system
0.777 still cool
Sampling system
0.905 nearly warm.
1.010
0.955
1.014
0.995
i
M
O
00
-------
Table B. Complete data for the May tests of two beta instruments.
Runs 1-23 plumbed in short and vertical downstream of diluter.
Runs 24 - 127 plumbed in with configuration necessary for beta instruments (see text).
o Beta
or
Run Filter cu Parallel
Min, LPM,
May Time Elapsed Diluted
% Liters
Dilution Diluted
Micro- Micrograms
grams, Per M-^ Ratio:
Particle Stack Con- Beta Weight
No.
92
93
94
95
96
97
98
99
100
101
No.
139
140
141
142
143
144
145
146
147
148
*See note
Hi
O
G
G
G
G
G
G
G
G
"
G
G
G
G
G
G
G
G
G
G
G
for
Filter
GCA
PF
GCA
PF
GCA
PF
GCA
PF
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
run 76
Date of Day Time Aerosol
23
1150 2 50
23
23
1205 5 50
23
23
1310 5 50
23
23
1325 5 50
23
1335 5 50
23
23
1410 5 50
23
23
1425 5 50
23
23
1435 2 50
23
23
1440 1 50
23
23
1450 5 50
23
Air Aerosol
(116)*
75
100
(290)
75
250
(289)*
75
250
-
75
250
75
250
(267)
75
250
(257)
75
250
(117)
75
100
(52)
75
50
(253)
75
250
Weight
5540
5567
12820
12888
14300
13213
-
2622
8866
7150
11931
7970
12913
3390
5400
1980
2566
6750
11914
centration
222
223
205
206
229
211
-
41.96
142
114
191
127
207
136
216
158
205
108
191
P. Filter Wt . Comments
0.995
0.994
NSP reduced
1.082 rapping inten-
sity on first 2
rows of wires
@1230 - 1300
GCA-GM tube
- blew out
IN with swirl-
n ,-nri meter in place.
0.600 TSI flowmeter for
calculations
0.617
0.628
0.772
0.567
i
0
i
-------
(continued)
Table B. Complete data for the May tests of two beta instruments.
Runs 1-23 plumbed In short and vertical downstream of diluter.
Runs 24 - 127 plumbed in with configuration necessary for beta instruments (see text).
Run
No.
102
103
104
105
106
107
108
109
110
111
112
Filter
No.
149
150
151
152
153
154
155
156
157
158
159
Operator
G
G
G
G
G
G
G
G
G
G
B
B
B
B
B
B
B
B
B
B
B
B
Beta
or
Parallel
Filter
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
IN
PF
May
Date
23
23
23
23
23
23
23
23
23
23
24
24
24
24
24
24
24
24
24
24
24
24
Time
of Day
1500
1508
1520
1534
1610
1035
1050
1100
1107
1116
1124
Min,
Elapsed
Time
1
1
10
1
5
5
5
1
2
?
5
LPM,
Diluted
Aerosol
50
50
50
50
50
50
50
50
50
50
50
%
Dilution
Air
75
75
75
75
75
75
75
75
75
75
75
Liters
Diluted
Aerosol
(52)
50
(50)
50
(538)
500
(53)
50
(258)
250
250
(257)
250
(53)
50
(105)
100
9
(264)
250
Micro-
grams,
Particle
Weight
1770
2565
1510
2696
11980
22141
1480
2709
6850
12113
12513
5770
10288
1220
1985
2310
3855
4907
6110
11639
Micrograms
Per M3
Stack Con-
centration
142
205
121
216
96
178
118
217
110
194
200.4
89.64
161.8
93.16
158.8
87.80
154.4
92.68
186.4
Ratio:
Beta Weight
P. Filter Wt . Comments
0.690
0.560
0.542
0.547
0.565
No data from
IN
0.561
0.615
0.599
IN stopped
too soon
0.525
o
-------
(continued)
Table B. Complete data for the May tests of two beta instruments.
Runs 1-23 plumbed in short and vertical downstream of diluter.
Runs 24 - 127 plumbed in with configuration necessary for beta instruments (see text).
Run
No.
113
114
115
116
117
118
119
120
J21
J22
]2'i
Filter
No.
160
161
163
162
164
166
165
167
168
169
170
171
172
173
174
175
176
177
178
Operatoi
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
Beta
or
Parallel
Filter
IN
PF
IN
PF
BF
PF
BF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
BF
PF
May
Date
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
Time
of Day
1135
1144
1238
1306
1321
1327
1334
1340
1348
1354
1400
Min,
Elapsed
Time
1
1
5
5
3
3
3
3
3
3
3
LPM,
Diluted
Aerosol
50
50
50
50
50
50
50
50
50
50
50
%
Dilution
Air
75
75
75
75
75
75
75
75
75
75
75
Liters
Diluted
Aerosol
50
(52)
50
250
250
150
150
150
150
150
150
150
Micro-
grams,
Particle
Weight
1970
1210
1716
17191
13495
11700
6654
6230
6856
6022
6197
5728
6018
6348
6837
6480
5570
5855
6078
5831
Micrograms
Per M3
Stack Con-
centration
157.6
92.88
137.2
275.2
216
187.2
177.6
166.4
182.8
160.8
165.2
152.8
160.4
169.2
182.4
172.8
148.4
156.4
162.4
155.6
Ratio:
Beta Weight
P. Filter Wt . Comments
No data from IN
0.705
1.274
-
Fly ash rein-
1.067 jection turned
off at 3315
1.138
1.082
0.948
1.055
0.952
Fly ash rein-
1.041 jection turned
on at 1400
-------
(continued)
Table B. Complete data for the May tests of two beta instruments.
Runs 1-23 plumbed in short and vertical downstream of diluter.
Runs 24 - 127 plumbed in with configuration necessary for beta instruments (see text) .
o Beta Micro- Micrograms
« or Min, LPM, % Liters grams, Per M3 Ratio:
Run Filter 01 Parallel May Time Elapsed Diluted Dilution Diluted Particle Stack Con- Beta Weight
No. No. Q Filter _ Date of Day Time Aerosol _ Air Aerosol Weight centration P. Filter Wt.
Comments
124
125
126
127
179 B
180 B
181 B
182 B
183 B
184 B
106 B
107 B
BF
PF
BF
PF
BF
PF
BF
PF
24
24
24
24
24
24
24
24
1406
1412
1418
1424
50
50
50
50
75
75
75
75
150
150
150
150
6070
6311
6219
6130
_
6207
6570
6570
161.6
168.4
166.0
163.6
__
165.6
17~572
175.2
0.962
1.014
0.995
1.000
------- |