DEVELOPMENT OF A NUCLEONIC
PARTIGULATE EMISSION GAUGE
FINAL REPORT
ENVIRONMENTAL PROTECTION AGENCY
CONTRACT NO. 68-02-0210
Federal Systems Division
EPA LIBRARY SERVICES RTP NC
APTD-1150
TECHNICAL DOCUMENT COLLECTION K|RMAN ROAD/ COLUMBUS, OHIO 43202
(614) 267-6351
dustrial
Vuclconics
-------
FINAL REPORT
DEVELOPMENT OF A
NUCLEONIC PARTICULATE
EMISSION GAUGE
Prepared by
Charles R. Duke
Boong Y Cho
Industrial Nucleonics Corporation
650 Ackerman Road
Columbus, Ohio 43202
For
Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Contract No. 68-02-0210
25 February 1972
-------
This report was furnished to the Environmental Protection
Agency by Industrial Nucleonics Corporation in fulfillment of Contract
No. 68-02-0210. The contents of this report are reproduced herein
as received from the contractor. The opinions, findings, and con-
clusions expressed are those of the authors and not necessarily those
of the Environmental Protection Agency. Mention of company or
product names does not constitute endorsement by the Environmental
Protection Agency.
-------
ACKNOWLEDGEMENT
The authors wish to acknowledge the IN program staff without
whose total dedication to purpose this program would not have been
possible: Orval Utt, Electrical Engineer; Boong Cho, Physicist;
Juan Crawford, Mechanical Engineer; Dan Landis, Designer;
Gerald Blessing, Technician. Special appreciation is offered to
Robert Herling of the Environmental Protection Agency and Project
Officer on this contract, who lended much expertise and guidance in
the field of sampling and filter systems.
-------
TABLE OF CONTENTS
Section Page
Number Number
1.0 INTRODUCTION 1-1
1. 1 Principal Components of the Gauge 1-3
1.2 Specifications 1-6
1.2.1 Major Specifications 1-7
2.0 TECHNICAL DISCUSSION 2-1
2. 1 Principles of Operations 2-1
2. 1. 1 Beta Gauging 2-1
2.1.2 Cassette Transfer Mechanism 2-4
2.2 Design Criteria for Selecting Filter Medium 2-9
2.2.1 The Air Gap Effects 2-9
2.2.2 Filter Variation Effects 2-10
2. 2. 3 Effect of Filter Thickness on 2-16
Counting Time
2. 3 General Comments on Filter Selection 2-21
2.4 Selection of Carbon- 14/PMT Geometry 2-24
2. 4. 1 Determination of Source Strength 2-25
Requirement
2.4.2 Detector Selection 2-28
3. 0 FIELD TESTS
3. 1 The Cincinnati Series 3-1
3.1.1 Clean Air Tests 3-3
3.1.2 Cassette Effects 3-4
3.1.3 AGC Delay Effects 3-4
3.1.4 Stability Tests 3-5
3.1.5 Stability Tests -- Using Filters 3-6
3. 1.6 Stability Tests -- Pallflex Medium 3-6
3.2 The Dow Series 3-7
3.2.1 Tests on Leaded Fuel (3. 0 cc/gal 2-10
TEL)
3.2.2 Tests on Non-Leaded Fuel 3-10
3.2.3 Tests on Low Leaded Fuel (0. 5 cc/ 3-12
gal TEL)
3.3 Conclusions Regarding the Field Tests 3-16
4.0 RECOMMENDATIONS 4-1
-------
LIST OF FIGURES
Figure Page
Number Number
1-1 Vehicle Exhaust Particulate Emission Gauge 1-2
1-2 Cabinet and Principal Components 1-5
2-1 Filter Cassette 2-5
2-2 Schematic Layout of the Three Stations and the 2-6
Transporting Mechanism
2-3 The Effect of Moisture Absorption on Dry Filters 2-11
2-4 The Effect of Moisture Desorption on Moisture Laden 2-13
Filters
2-5 Variation of Filter Basis V'eight as a Function of 2-15
Moisture Absorption and Desorption (Normalized)
2-6 Required Beta Sampling Time vs. Filter Basis 2-20
Weight for Four CFM
2-7 Required Beta Sampling Time vs. Filter Basis 2-22
Weight for Two CFM
2-8 Typical Beta Spectrum 2-29
11
-------
1. 0 INTRODUCTION
This report is the final report to Environmental Protection Agency
Contract No. 68-02-0210. Upon acceptance by the EPA, this report
represents completion of contractor-required delivered items.
In June of 1971, the EPA contracted with Industrial Nucleonics (IN)
to develop a beta gauge and filter sampler for measurement of particulate
emissions from automobile exhaust. The gauge was to have a minimum
sensitivity of 125 micrograms per cubic meter (|j.g/m ) with no specified
upper limit. Further, the filter collection system was to have an efficiency
using a dioctyl phthalate aerosol of at least 99. 9%. A gauge to meet these
requirements has been designed, built, tested, and delivered and is as
shown in Figure 1-1,
A Hewlett-Packard Computing Counter has been used to perform
the calculations and switching functions. In addition, a Fischer and Porter
Swirlmeter has been provided for reliable determination of gas flow. The
inclusion of these two units allows high accuracy and repeatability with
respect to computation. This was deemed necessary in order to fulfill
the wide variety of applications against which the unit was to be tested.
The major thrust of the program has been to design a research tool
rather than a prototype so that evaluation of the unit could be made on
the basis of the suitability of beta gauging for the various applications,
not merely IN's ability to produce such a gauge. Certainly, for many
specific applications, a considerable reduction in complexity and cost
is possible.
1-1
-------
VEHICLE EXHAUST
PARTICULATE EMISSION GAUGE
Figure
-------
As part of the contract, this unit was tested recently at one of
the EPA's Cincinnati facilities and at the Dow Chemical Company in
Midland, Michigan. The Cincinnati series was primarily intended for
unit de-bug and checkout. The tests at Dow were performed on automobiles
using leaded (3cc/gal tetraethyl lead), low-leaded (0. 5 cc/gal TEL) and
non-leaded (0. 0 cc/gal TEL) fuels. Good results were obtained using
leaded and low-leaded fuels when compared to gravimetric measure-
ments. However, less desirable results were obtained with no-leaded
fuels. No conclusion may be drawn from the non-leaded tests since
correlation between two different size (4 inch and 2 inch filters)
gravimetric measurements could not be achieved, thereby leaving no
standard against which to compare the gauge.
The details of these tests, and the details surrounding IN's
selection of this particular design constitute the text of this report.
While reference will occasionally be made to operation of the unit,
the reader is referred to the Operation Manual accompanying the unit
for specific operation, maintenance and safety procedures to be
followed.
1. 1 Principal Components of the Gauge
This gauge has been designed primarily as a research tool.
There has been little attention devoted to minimizing the size and weight
and maximizing the portability as efforts of this type were considered
to be beyond the scope of this contract. The device has been fitted
1-3
-------
with wheels, however, to assist in moving the unit from place to place.
Also the Computing Counter and Systems Programmer are quite easily
removed (See Operation Manual) for further ease of transport. A
complete list of the principal components and their model numbers
is as follows (See Figure 1-2):
1. Electronic circuitry (printed circuit boards). Access
to this circuitry is obtained by removing the side panel
2. Welded cabinet brace
3. Printer. Hewlett-Packard Model No. 5055A
4. Flowmeter preamplifier
5. Vacuum pump. CAST Model No. 0822
6. Flowmeter. Fischer and Porter "Swirlmeter"
7. Flowmeter supports
8. Flowmeter electronics
9. Swivel caster (four identical double-wheel units)
10. Control panel with switches and indicator lights
11. Systems programmer, Hewlett-Packard Model No. 5376A
12. Computing counter, Hewlett-Packard Model No. 5360A
13. Flow rate display meter
14. Particulate Collection station. Station at which gas is
drawn through the filter and particles are collected
15. Nucleonic source enclosure and filter locating mechanism
16. Magazine to hold 20 filters
1-4
-------
Z5-I/Z
37-1/4
L
"
-'
--
-
-
1
I
1
1
2,
CD
I a)
0 I !
oo o ob opj
SIDE
FRONT
CABINET AND PRINCIPAL COMPONENTS
(SIDE PANEL AND FRONT DOOR REMOVED)
Figure. I ~ Z
-------
17. Filter transporting mechanism actuator, Cliftronics
R28S1 32 rotary solenoids
18. Radiation detector bousing. Station at which filter
and collected particulate masses are detected.
19. Flow regulating valve.
1. 2 Specifications
This gauge has been overdesigned for many specific applications.
This is a direct result of emphasis being placed on accuracy over a wide
range and flexibility. The beta counting times and pumping times may
seem astronomical and unnecessary. The magnitude of these numbers
is, however, an added bonus received by inclusion of the HP Computing
Counter and Systems Programmer. It is also correct to say that the
computational errors are infinitesimal by comparison to the errors in
the sensing heads (flowmeter and beta counting). This is as it should
be for an instrument of this nature. The limitation of the gauge is
contained solely in the limitations of the sensors used, namely the
state-of-the-art in stabilizing photomultiplier tubes, the state-of-the-
art in measuring absolute flow, and process repeatibility. This necessarily
implies that improvement in any one of these three areas will result in
improved measurement accuracy automatically.
The major specifications follow, but some prefacing comments
are necessary. First, it is difficult to state a range of the gauge in terms
of stream concentration unless some limit is placed on botv> flow rate and
time. This is a result of there being both a minimum and a maximum
1-6
-------
total mass which can be detected with the current design. The minimum
and maximum mass determinations are stated with the assumptions on
flow rate and time clarifying the stream concentration.
Second, probably the greatest limiting factor on accuracy at
high concentrations is the ability to measure the total sample volume
correctly. The Swirlmeter measures volumetric flow and therefore
to achieve a measurement in terms of an "Actual Cubic Foot", use must
be made of the suggested correction factor in the Operations Manual.
For most applications where this factor is not included, the flow accuracy
is ±2%, excluding corrections to "Standard Cubic Foot."
1. 2. 1 Major Specifications (assumes zero process error)
Specification
1. Effective Filter Area
2. Flow Rate
3. Minimum Concentration
4. Maximum Concentration
5. Beta Counting Range
6. Beta Counting Range
7. Pumping Time
8. Pumping Time
9. Minimum Mass Resolution
Value
3 cm
20-120 1/min
80 ± 16
Remarks
(2o-)
3 x 106 ±2% >ig/m3
(assumes no flow-
meter correction)
10" sec to 2 x 10 1
sec.
30 sec. to 300 sec.
-6 11
10 sec to 2 x 10
sec.
30 sec to 3 x 10
sec.
±60 ng (2
-------
10.
11.
Maximum Mass Determination 30,000 ±600 (o.g (2cr)
Flow Absolute Accuracy
12. Flow Absolute Accuracy
13. Weight
14. Power
15. Source
16. Source Activity
±0. 75% ACFM
±2. 0% ACFM
350 Ibs.
1500 watts
Carbon-14
100 microcuries
Manufacturer's specifi-
cation (using correction
factors)
no correction factors
flow >40 1/min
while pumping
1-8
-------
2.0 TECHNICAL DISCUSSION
2. 1 Principles of Operation
2.1.1 Beta Gauging
The Vehicle Exhaust Particulate Emission Gauge utilizes a
filter "cassette" on which to collect the particles. Detection of
particles on the filter is accomplished by an isotopic radioactive
source (Carbon- 14) and a scintillation detector (Photomultiplier Tube).
The attenuation of beta particles through a medium follows an
exponential law:
N = Noe (])
where N = detected counts
N = detected counts in absence of medium
o
[i = attenuation coefficient of medium
px = basis weight of medium.
In this application, the "medium" discussed consists of several separate
media: 1) the air gap between source and detector; 2) the clean filter;
3) the detector window, and 4) the particles collected. Therefore
equation (1) can be rewritten as:
-(upx)a -(npx)f -(M-PX)W -(HP*)0
N = N e e e e H (2)
where N = detected count
N = counts emitted from surface of source in
o
direction of detector
, (HPX) a = attenuation exponent for the air
2-1
-------
(}j.px) = attenuation exponent for clean filter
((j.px) = attenuation exponent for detector window
(fj.px) = attenuation exponent for particles.
Obviously, the only factor of interest in Equation (2) is the last
one referring to the particles. Every effort has been made to either
control or minimize the effect of these other variables. The effects
of variations in the density of the air has been minimized by reducing
the air gap between the source and detector to about one centimeter.
In order to account for variations between filters and to allow deter-
mination of the reduction in total collected counts due to detector
attenuation, a standardization measurement is made and retained for
later comparison. Equation (2) may then be written as:
N = N e
Lt 1
where N = detected counts after collection of particles
(Measurement)
N. = detected counts before collection of particles
(Standardization)
Taking the natural logarithm of both sides:
Ln
or
then px = Ln - (3)
2-2
-------
Equation (3) is the expression to determine the mass per unit
area collected. To determine total mass collected, multiply by total
filter sample collections area A.
/NA
Mass = Apx = Ln I 1. (4)
Notice that by making an initial standardization measurement,
all reference to the filter itself has been removed. This is indeed one
of the advantages of the beta gauge. Almost any filter medium may be
used, the only requirement being that the medium is not modified during
the sampling process. The "cassette" approach accentuates this advantage,
allowing use of filters not suitable for use as a tape.
The total gas flow through the unit is determined by use of a
Fischer and Porter Swirlmeter (1" size). This unit creates a precessing
vortex in a region containing a thermistor. The thermistor responds to
the different heating afforded by the center of the vortex and, through
pulse shaping, produces a pulse output. One pulse is representative
of a specific volume of gas. To determine the total flow, the gauge simply
collects the pulses from the flowmeter and multiplies by a constant. A.
more complete description is presented in the Operation Manual
accompanying the gauge. The volume then becomes:
Volume = K N (5)
To determine concentration, Equation (4) is divided by Equation (5):
2-3
-------
Concentration = = K ~^ Ln I I (6)
Volume 2 i i < >
where K is a constant relating the various units involved. Equations
C*
(4), (5), and (6) represent the outputs required by contract. These
along with the counts N , N , and N are printed on paper to provide
a permanent record of the test. By virtue of the Hewlett-Packard
computer, all calculations, including the natural logarithm, are done
digitally, affording extremely high accuracy of computation. The only
analog circuitry in the gauge is the pulse gathering and shaping circuitry
and the automatic gain control.
2.1.2 Cassette Transfer Mechanism
It has been previously mentioned that this gauge uses a filter
cassette on which to collect the particles (See Figure 2-1). The cassette
is constructed of 1/16" aluminum and is irregularly shaped to avoid
improper loading into the magazine. Installation of the filter on the
cassette is accomplished by cutting a square piece of filter paper
(nominally 1" square) and taping the filter to the cassette. Notice also
that the cassette has two holes on opposite corners to precisely locate
the cassette at the beta measurement station. This is to insure that
the source is irradiating the same ares after collection on which standardi-
zation measurement was made before collection.
During operation each cassette is transported automatically
through the gauge. The cassette transfer mechanism has three stations
(Figure 2-2). The first is the loading station, where the lowest cassette
2-4
-------
UPPER SIDE
Adhesive tope
Rlter medium
UNDER SIDE
FILTER CASSETTE
Figure 2-1
2- 5
-------
SEALS
GAS OUT
TO
SUCTION
PUMP
LOADING
STATION
c
(
^ j j'Hri-,
(n Uki ! ^
'
^4>
^
TOP VIEW OF TRANSPORTING SLIDE
180
ROTARY
SOLENOIDS
DETECTOR
COLLECTION
STATION
SOURCE-DETECTOR LOADING
STATION
STATION
ro
i
SCHEMATIC LAYOUT OF THE THREE STATIONS d THE TRANSPORTING MECHANISM
Figure 2-2
-------
in the magazine drops into a pocket in the transporting slide. At the
second station the cassette is positioned between the source and the
detector, so that the beam of radiation passes through the filter. At
the third station the cassette is positioned so that the gas to be monitored
may be drawn through the filter, and the particles collected.
Each cassette is transported between the stations by a slide of
the same thickness as the cassette. The slide is moved between stations
by two 180-degree rotary solenoids. In Figure 2-2 the slide is shown in its
rightmost position, ready to pick up a cassette from the magazine. The
next operation is for the right-hand solenoid to rotate 180 degrees. The
cam roller on the end of the solenoid arm will follow the path indicated,
causing the slide to travel to the left and transport the cassette to the
source-detector station. At this station two tapered pins are pushed
down into the two small holes in the cassette. At this time the standardizing
measurement is made to determine a background count in the absence of
particulate matter. The locating pins are then withdrawn and the left-hand
solenoid is rotated 180 degrees. The cam roller on the end of the solenoid
arm will follow the path indicated, causing the slide to travel further to the
left and transport the cassette to the collection station. At this station the
cassette is clamped between two elastomeric seals which prevent leakage
of gas into or out of the stream passing through the filter. The cassette
is then clamped, the vacuum pump is turned on, and the particles
2-7
-------
are collected on the filter. At the completion of the collection cycle,
the sealing clamp is released and the left-hand solenoid is rotated
180 degrees in the opposite direction to transport the cassette back
to the source-detector station. Here the locating pins are reintroduced
into the small holes in the cassette.
After the measurement of the material deposited on the filter
is made, the right-hand solenoid is rotated 180 degrees in the opposite
direction, and the slide is returned to its initial position. As it does
so the cassette in its rightward travel encounters the stripper springs,
two inclined flat springs which guide the cassette downwards through
an opening in the base plate and out of the transporting slide pocket.
The used cassette drops into a receiving container and the slide pocket
is left vacant, ready to pick up another cassette from the magazine.
Originally, it was proposed that the transfer mechanism be a
wheel in which holes had been cut to contain the cassette. After the
program started, a more complete examination of this approach revealed
certain problems. The problems, though not insurmountable, presented
complex switching and sensing conditiongs which were considered to be
simplified by using a linearly translating slide. If the filter manufacturing
quality control were such that the basis weight from filter to filter did not
2
vary by more than 10 jig/cm , a considerable reduction in complexity
could have been accomplished. Current state-of-the-art in filter manu-
facturing does not permit this, however.
2-8
-------
2. 2 Design Criteria for Selecting the Filter Medium
Several filter media were evaluated during the course of the
program. Before discussing the various filters, however, it is worth-
while to discuss the criteria for evaluation.
The contract required determination of the particulate concen-
tration in a stream having at least 125 micrograms per cubic meter.
3
This determination was to an accuracy of ±25 |j,g/m at a 95% confidence
level and an efficiency of collection 99. 9%. These requirements place
a rather severe constraint on the sampling head repeatibility and the
electronic system stability. In determining the amount of material
which must be collected to allow this determination, consideration must
be given to the magnitude of such error sources as the change in air
gap conditions and filter deformation.
2.2.1 The Air Gap Effects
The air gap between source and detector represents a "background"
attenuation for the beta particles. It is an attenuation which cannot be
avoided and any variation in the air gap to cause a change in air density
between standardization and measurement represents an error in measure-
ment. To minimize the effect of air density variations, the air gap itself
must be reduced as far as it practically feasible. In this gauge the air gap
is about one centimeter. This represents an effective background basis
weight of about ] . 3 mg/cm . If the variation in air density can be held
to within 1% (about 5 F) between standardization and measurement, then
2-9
-------
one may consider ± 0. 5% or ± 7 fig/cm fluctuation (]
-------
Rgure 2-3
2-11
-------
for extremely high structural strength. It has a basis weight of approxi-
2
mately 3.2 mg/cm . The sample chosen was a strip 31.8 cm x 110 cm.
(This is equivalent in area to a circular filter approximately 210 mm in
diameter. While admittedly large, an approximate number for any other
size filter may be obtained by scaling the areas, as the variation on a
basis weight consideration should be identical, regardless of filter size. )
The test procedure was to weigh the filter as delivered from the
manufacturer, dry the filter at 110 C for 10 minutes (to obtain the actual
"bone dry" weight would require much longer times such as 24 or 48
hours at 110 C), and then weigh the filter at periodic intervals after
exposure to 72 F and 50% relative humidity. As can be seen, a weight
loss of 49 milligrams (140 |j.g/cm ) occurred upon drying. Much of this
loss is recovered rapidly, so that after 10 minutes only 14 mg (40
remains unrecovered. At this point, however, the recovery becomes quite
slow, so that after 30 minutes, 11 mg (31. 5 |o.g/cm ) still remain lost, and
after 4 hours, 40 minutes, 9 mg (25. 7 [jig/cm ) of moisture still had not
been regained. For a typical filter size used (47 mm) for comparison,
this would represent an error of 445 micrograms due solely to moisture
absorption effects.
The effects of moisture desorption after having been exposed to
100% relative humidity at 72 F are even more dramatic (See Figure 2-4).
After exposure of the filter to this environment for 25 minutes, a weight
gain of 76 mg (217 (j.g/cm ) as observed. After 10 minutes of
2-12
-------
Figure 2-4
2-13
-------
stabilization, 22 mg (63 fig/cm ) were still retained, after 30 minutes,
2
12.8 mg (36.6 |j.g/cm ), and after four hours, 40 minutes, there was
2
still 9 mg (26. 7 p.g/cm ) of excess moisture.
If the data is normalized to initial filter weight to present it
on a percentage variation basis, Figure 2-5 is the result and illustrates
clearly the "hysteresis" effect of moisture on fibrous materials. This
hysteresis would represent an uncertainty in the amount of particulate mass
collected of as much as ± 1 % of the initial filter basis weight unless the
complete moisture history is known. For the Pallflex medium, this is
an error of approximately ± 32 jjig/cm due solely to moisture effects.
In addition, if the measurement is made immediately after sampling (as
it is in the case of this gauge), even higher errors can result.
From the above data, it is recommended that the Pallflex medium
not be used in this gauge. There are, of course, other filter media which
may be used which exhibit less sensitivity to either moisture absorption
or desorption.
The GE Nuclepore, for example, showed a weight loss of only
6. 4 |j.g/cm after drying and recovered to its initial weight (no apparent
hysteresis) after only three minutes of stabilization. When exposed to
2
high humidity, it gained only 1. 3 |j.g/cm and has regained its initial
weight within one minute. Unfortunately, this material does exhibit
the highest pressure drop for a given flow rate of almost any other type
of medium, and has shown a tendency to clog up.
2-14
-------
I
2-15
-------
The Gelman Type A filter avoids the problem of high pressure
drop for a given flow rate and does not clog easily. Upon drying as
2
previously described, this medium showed a loss of only 9. 7 (j.g/cm
and has recovered half this amount in 1 5 minutes. When exposed to
high humidity, however, this medium showed a weight gain of 221 |j.g/
2
cm , but had regained its start weight within 10 minutes. It is also
quite thick (approximately 7 mg/cm ) and will require a longer counting
time (see the next section).
A further consideration is that these errors were produced under
static conditions (i. e. , no forced flow through the filter) and may not
be reproducible under dynamic conditions with which the EPA is involved.
2.2.3 Effect of Filter Thickness on Counting Time
The following discussion concerns only stream concentrations
of 125 fjLg/m as far as the flow rates are concerned. It is equally
applicable when the required particulate basis weight resolution is of
the same order.
In the process of a quantitive determination of the collected mass,
the filter interferes with the measurements. For example, if the filter
weighs 1,000 to 10,000 times the weight of the minimum desired resolution
of the collected particulate, the total weight of the filter/particle system
has to be determined to 0. 1 to 0. 01% accuracy. Even though it is feasible
to determine the weight of certain objects to these accuracies, extreme
precaution is required. The weight of an object like a paper or glass
fibre filter is "'unstable" in the sense that its weight will change with the
2-16
-------
temperature and relative humidity of the environment and furthermore
a hysteresis effect is observed in the adsorption and desorption of water
vapor (see the previous section for a more complete discussion). Since
the gain and loss of the weight with the environment and its history will
be proportional to the weight of the filter, it is highly desirable to use
as thin a filter as possible for a sample collection medium.
In a nucleonic method of weight determination, a further considera-
tion is required in the sense that the filter will attenuate the radiation
without contributing to the signal. This attenuation is not detrimental
except for the possible physical deformation of the filter due to pumping,
if no restriction is imposed on the required source strength. However,
if the source strength is limited to a given value, the loss of radiation
intensity has to be made up by a long counting time to obtain a desired
statistical accuracy.
Consider a case of "zero" filter thickness. Then, with a 100 |j.Ci
C-14 source and the geometry of IN design, a count rate of 50, 000 cps is
obtained. If a two sigma resolution of ±20 |j.g/cm (corresponding to a
stream resolution of ±25 ^.g/m with a 4 CFM sampling flow rate, sampling
for 20 minutes from an aerosol stream containing 125 |J.g/m of particulate)
or ± 10 jig/cm one sigma resolution is required, the required counting
time is determined as follows: The attenuation of radiation is given by
N = N e rr (7)
2-17
-------
where N and N are the detected total counts with and without the sample;
and ^JL, p , and x are the mass attenuation coefficient, density and thick-
ness of the sample, respectively. By differentiation, one obtains
dN
or
(8)
For C-14 beta rays, (j, is approximately 250 cm /gm and the
minimum required resolution of (px) is ±10 |j.g/cm^ (one sigma). There-
fore, the minimum required resolution for (|j.px) is 250 x 10 x 10 =0. 0025.
This means, from equation (8), that the nuclear signal, N, has to be
resolved to an accuracy of ± 0. 25% (one sigma). It is well known that the
nuclear signal fluctuates with one sigma value of ± ^N when the total
count is N. Therefore,
= ">-2
or
N = 1. 6 x 105. (10)
2-18
-------
4
Since the count rate with no sample, n , is 5 x 10 cps, the counting
time to accumulate 1. 6 x 10 counts is
1.6 x 105 1.6 x 105 , , , Ml.
T = = =3.2 seconds. (11)
n° 5x10
Now, if the filter is assumed not to contribute further error due
to its unstable weight (absorption and desorption of moisture) and physical
deformation, its effect on the nucleonic measurement of collected particles
is to attenuate the radiation. Therefore, the count rate, n , will be
o
reduced according to
n = n e
o
where the subscript, F, represent the value for the filter. Substituting
n for n in equation (11), one obtains the required counting time for
various filter thicknesses from
T = ' = 3.2e seconds. (13)
n
This equation is plotted in Figure 2-6 labeled with "no electronic error."
An additional error is caused in the measurement of N(for the
purposes of discussion, this is called electronic error, but also includes
all other errors besides statistics) in equation (7). Thus, 0.25% total
allowed error in equation (9) should be allotted between the electronic
error and the statistical fluctuation. Assuming an RMS averaging,
one gets
Ee2 + -^j- = (2.5xlO~3) (14)
2-19
-------
4 5
Figure 2-6
8
2-20
-------
or
N = - -
(2. 5 x 10 ) - E
Following the same procedure as in the case of "no electronic
error," one obtains the counting time from
N
5 x 104 [(2.5xlO~3) -E2J seconds.
This equation is plotted in Figure 2-6 with E = 0. 05%, 0. 075%,
G
0.1%, 0.15%, 0.2%, and 0.23%.
If, however, the same stream resolution (±25 ng/m ) is required,
but only 2 cfm (56 liters/min) flow rate is used, the basis weight resolu-
tion is ± 5 fj.g/cm and Figure 2-7 is the result with E plotted for 0. 025%,
0. 05 %, 0. 1%, and 0. 12%. As can be seen, the effect is rather dramatic
on the counting times, with resolution of this magnitude impossible at
4 cfm if all other errors amount to 0.25%, and at 2 cfm if all other errors
amount to 0. 125%.
2. 3 General Comments on Filter Selection
While the advance copy of the Operation Manual delivered with the
instrument recommended that the Pallflex E70/2075W medium be used,
later test results show this medium to be undesirable from a moisture
absorption and desorption standpoint. It is IN's opinion that the
"optimum" filter medium has not yet been found. Much work remains
to be done in this area. In selecting a medium to use, however, several
conclusions can be drawn from the preceding sections of this report.
2-21
-------
4-
1
1
oth
Q
ifci
1ft
IQJ
4
Figure
2-7
7 e
2-22
-------
First, the filter should have as low a basis weight as possible
for two reasons. One, to reduce the amount of time required for beta
measurement, and two, to reduce the effect of any slight distortion
caused in the filter by the pumping process. From the previous
section, the beta collection time may vary as much as four to one,
depending on the basis weight of the filter. Also a 1% basis weight
2 22
variation for 1 mg/cm filters is 10 [o,g/cm , but for 8 mg/cm , a
1% basis weight variation is 80 p.g/cm .
Second, the property of paper-based filters to absorb and lose
moisture during the process casts doubt on results obtained using
these materials. In fact, even the Gelman Type A filters yield
dubious results after exposure to high humidity, which unfortunately
happens to be the situation with automotive exhaust. This does, of
course, mean that there is currently no standard against which to
reliably measure results, either with the beta gauge or with gravi-»
metric comparisons.
Third, the problem of moisture is not confined to beta gauging.
In fact, it isn't even confined to moisture. Any volatile material would
be subject to identical fluctuations. Again, automotive exhaust contains
a high percentage of volatile material, at least in the cases of low-leaded
and non-leaded fuels.
Finally, it is conceivable that a different type of filter material
may be required from filter manufacturers. Little consideration has
2-23
-------
been given by this industry to such properties as low moisture sensi-
tivity. This is presumably due to primary application areas being
chemical analysis of collected particulate or simply to clean the air
passing through the filter.
2. 4 Selection of Carbon-14/PMT Geometry
Several sources were evaluated prior to contract: Kr-85, C-14,
Ni-63, and Pm-147. A complete discussion of the tradeoffs involved in
selection is beyond the scope of this report and only the major reasons
will be presented here. Krypton-85 was eliminated on the basis of poor
sensitivity. Nickel-63 was eliminated on the basis of low modal energy.
Promethium-147 was eliminated on the basis of short half-life. This
left Carbon-14 as the optimum selection.
IN had proposed that the source to use was in the form of a
carbide, where there is a chemical bond between the carbon and
surrounding atoms. In so doing, an extremely high integrity source
could be achieved, capable of surviving almost any environmental
condition, up to and including fire. IN still maintains that this is the
most desirable source to have. Unfortunately a complete survey of
the source manufacturing field disclosed no vendor willing to undertake
the development of a carbide source. While the technology appears to
be available to produce such a source, neither time nor money would
permit the development under this contract.
2-24
-------
The vendor eventually selected was New England Nuclear,
Billerica, Massachusetts. Originally a source was received from
them which was a C-14/Kapton mixture deposited on a small alumi-
num disc. This proved to be unsatisfactory from a uniformity
standpoint. Several iterations were required before achieving the
physical integrity, uniformity, and emission efficiency we desired.
IN believes that the source which is now in the gauge is one of the
finest non-carbide sources made. It is constructed by depositing the
C-14 on an aluminum planchet, laying a thin plastic film over this
and spraying the active area with Kapton. This source has the highest
percentage yield of any other in IN's experience and made it possible
to reduce the activity to such a level as to prove the feasibility of
making this gauge eventually license free.
2.4.1 Determination of Source Strength Requirement
The license free aspect of the gauge was paramount in IN's
philosophy of design. Carbon-14 occurs naturally and as a result has
one of the highest allowed quantities to qualify for National AEC license
exempt status. This quantity is 100 microcuries. It has already been
discussed (Section 2.2. 3) how much collection time is involved, pro-
4
viding a count rate of 5 x 10 pulses per second can be achieved in the
absence of a filter. It will now be shown that this count rate is in fact
achieved in this gauge.
2-25
-------
The effective yield of the 100 microcurie source delivered with
the gauge is estimated by New England Nuclear at 15%. The number of
counts actually emitted from the source in all directions is
, ,~10 counts ir,~4^ ir cc m5 counts
3. 7 x 10 ; x 10 Curies x . 15= 5. 5 x 10
sec. Curie ' second
If the source is place 1 cm from a disc- shaped detector of 1 inch
diameter, the geometrical factor is 1/2 (1 - cos 0) ~ 18% where 0 is
the half -angle subtended by the detector. Thus, in the absence of air
and sample 5. 55 x 10 x . 18 = 10 counts per second reach the
detector.
The effective attenuation coefficient (,) has been estimated at
2
250 cm /gm. The desired count rate is in the absence of the filter
but the attenuation due to air and the detector window must still be
considered. It was previously discussed that beta radiation attenuation
obeys a negative exponential law:
N = N e
o
This can be expressed differently as
where
N = N (
o
N = 10 counts per second (previously derived)
= attenuation exponent due to air
((j.px)w = attenuation exponent due to the detector window.
2-26
-------
For air, the density is equal to approximately 1.2 mg/cm and
the distance is 1 cm. The attenuation factor (e ^ -A) due to air then
becomes 0. 74. The detector window has a basis weight (px) of approxi-
mately 0.9 mg/cm (manufacturer's specification). Therefore the
detector window attenuation factor is 0.8. The total detected counts
5 4
then becomes 10 x.74x.8 = 5.92x10 counts per second. Allowing
for detector efficiency and inclusion of a threshold detector to exclude
PMT characteristic noise, it can be seen that the detected counts will
4
not vary significantly from the 5x10 counts per second required. In
fact, the count rate achieved in the absence of a filter is approximately
44, 000 cps so that good agreement with theory is achieved.
This count rate allows the use of the gauge over a wide range of
filter area densities without inordinately increasing the beta measure-
ment time (See Section 2. 2. 3). This was deemed desirable by IN in
order to increase its flexibility as a research tool. In Figure 2-6, for
example, it can be seen that using a filter of basis weight 7 mg/cm
(such as the Gelman Type A) and a process repeatibility of 0. 1% (estimated
reliability for the IN gauge) a counting time of 220 seconds is required to
achieve ±10 ^.g/cm resolution. If the count rate were one-tenth
our estimated rate (this would yield an estimated detected count rate
of 4. 4 x 10 counts per second, typical of maximum rates available from
Geiger-Mueller detectors), this time would be 2,200 seconds to achieve
the same accuracy.
2-27
-------
2.3. 2 Detector Selection
The deficiencies of a Geiger-Mueller tube in the rapid determina-
tion of filter basis weight has been discussed. There are three other
types of detectors which may be used for low energy beta radiation
detection: solid state detectors, proportional counters, and a PMT/
scintillation crystal combination. The current state-of-the-art in solid
state detectors are such that they must be used in a nearly constant
temperature environment. This was felt to be an unnecessary require-
ment to place on the detector station. Proportional counters and ionization
chambers were considered carefully, but the window thickness is such
as to prevent the use of license exempt quantities for C-14 sources.
That left only the PMT/scintillation crystal combination.
There are two basic scintillation crystals which could be used:
CaF_(Eu) or Nal(Tl). Since the modal energy of C-14 is so low, an
extremely thin window is required (on the order of 1. 0 mg/cm ). For
this reason and the fact that Nal is highly hygroscopic and would tend
to deteriorate with time, CaF was selected as the scintillator.
£t
Harshaw Chemical Company was then contracted to fabricate the
detector unit, using an Amperex XP-1011 tube, a CaF_ crystal, and
£j
aluminized Mylar as the detector window.
This detector was quite stable in gain in contrast to other
very similar detectors. As is the case with all PMT's, however, the
gain drift was significant enough to require the addition of an automatic
2-28
-------
gain control (AGC) circuit. A typical beta spectrum is as shown in
Figure 2-8. If two thresholds are established, L and H for low and
high, respectively, it is possible to select these thresholds such that
the difference in the two numbers remains relatively constant. For
any major changes which may occur, however, the ratio of the two
counts provides a control signal by which the gain can be adjusted to
keep the above condition true. Through this stabilization technique,
PMT stabilities on the order of 0. 1% are achieved and it is this
capability that allows the measurement to be made at the low end
of sensitivity.
g
o
H
Pulse Height ( Millivolts)
Figure 2-8
2-29
-------
3. 0 FIELD TESTS
Two series of tests were performed on the gauge. The first
series was performed during the week of 4 October 1971, at the EPA's
Cincinnati facility and was for the primary purpose of debugging the
unit. The second series was performed during the week of 25 October 1971,
at the Dow Chemical Company in Midland, Michigan, and was for the
primary purpose of acceptance tests on the unit. The results obtained
from both these series are the subject of this section of the report.
Prior to the field tests, the attenuation coefficient JJL was experimentally
2
determined to be 252 cm /gm by noting the effect on the count rate of
differ ring thickness of Mylar.
3. 1 The Cincinnati Series
Little data was obtained from automotive tests in this series due
to several problems encountered with the unit. As the purpose of the
tests was to reveal these problems, however, considerable progress
may be considered to have been made. Minor problems were encountered
with the locating solenoid binding up. Due to self-check microswitches
installed on the unit, no errors in data due to this malfunction are believed
to have occurred. A problem was encountered but not recognized until
after the Dow series with interference between the pressure fluctuations
in the air line due to the pump and the Swirlmeter. Apparently, these
fluctuations were viewed by the Swirlmeter as vortex precessions and
counted as a volume of flow. It is believed that this was only a problem
3-1
-------
at the low flow rates. To correct the problem, the flowmeter was
changed from its "downstream of the pump" location to its present
"upstream of the filter" location and the interference was no longer
observed. During the brief redesign period between the Cincinnati
and Dow series of tests, all encountered deficiencies were corrected.
One of the major thrusts of this series was to evaluate different
filter materials. There were four different media which were hopefully
to be tested: 1) the Gelman Type A; 2) GE Nuclepore (0. 4 \± and 1.0 u),
3) Pallflex No. E70/2075W; and 4) a paper filter of high efficiency and
low basis weight developed by Dr. Wendell Anderson, Naval Research
Lab. , Washington, D. C. Details of the tests follow, but a brief
discussion of these media is appropriate. Some deformation (tearing)
of Dr. Anderson's material was detected. This is believed to be a
result of the short time delay between the command to release the clamp
on the cassette at the sampling station and the command to transport the
cassette to the beta measurement station. After this, it was decided
to switch to the Pallflex medium, and as this filter seemed to perform
adequately at the time, no tests were performed on either the Gelman
or GE filters. Tests were later performed which showed the Pallflex
medium to be inadequate due to moisture interferences (See Section 2. 2. 2)
and it is now recommended that tests be performed by the EPA on the
Gelman and GE filters to determine their adequacy.
3-2
-------
This test series can be divided into six separate segments.
The included segments, reasons for their inclusion, and results
obtained are contained in the following sections. In this entire series,
volume collected is unimportant and will not be included in the data.
3.1.1 Clean Air Tests
These tests were performed on Dr. Anderson's medium and
were for the purpose of determining the clean air characteristics of
the unit. Clean air was obtained by inserting an in-line 47 mm Gelman
Type A filter in the air line at the entrance port to the unit. Only two
tests were performed. In both, the beta collection time for standardi-
zation and measurement was 90 seconds. In test 1 , the pumping time
was 23 minutes; in test 2, the pumping time was 5 minutes.
Test 1 indicated a total collected mass of 432. 8^.g while test 2
indicated a total collected mass of 169. 5p.g. Both of these values were
unacceptable since the gauge specification was ± 30p.g for 2 cfm flow
rate and ± 60)j.g for 4 cfm flow rate and this latter flow rate was the
maximum attainable with the unit. It was suggested that the cassettes
may introduce errors and the next group of tests were devoted to seeing
if that error was constant.
3-3
-------
3.1.2 Cassette Effects
Seven cassettes were labeled A through G and installed in the
unit with no filter attached. Here are the results:
Total Indicated Mass
Cassette No. (Micrograms)
A +3.9
B +393.1
C + 358. 3
D - 42.9
E +243.6
F +249.6
G +82.7
These tests showed that the error, if cassette introduced, was at least
not constant for all cassettes. It was then suggested that possibly the
temporary interruption of the beam by the cassette as it moved into
position may cause a disturbance of the AGC. To determine this, a
delay was introduced between the time the cassette was moved into the
source-detector station and the time the computer began accepting data.
3.1.3 AGC Delay Effects
Four of the lettered cassettes used in the previous test were
again tested with these various delays. Here are the results:
Cassette No.
B
C
D
G
Delay
(seconds)
60
15
15
60
Total Indicated Mass
(micrograms)
- 191. 3
- 44.4
- 88. 9
+ 125. 5
3-4
-------
While the grouping is tighter and generally of opposite sign from the
previous tests, these data were judged inconclusive and unacceptable.
It was then decided to investigate the performance of the gauge with-
out either filter or cassette in the transfer mechanism.
3.1.4 Stability Tests
These tests were performed by placing the unit in "automatic"
and allowing it to run with no cassettes in the loading magazine. During
test 9, it was noted that the count rate being observed on the Nixie dis-
play of the Computing Counter exhibited a sensitivity to whether or not
sunlight was shining on the floor of the test area. On tests 10, 11, and
12, precautions were taken to ensure that the external light reaching the
gauge was minimized. On tests 13, 14, and 15, the front door of the unit
was opened to allow light to shine on the cassette transfer mechanism.
On test 16, the light was again shut off. Here are the result:
Total Indicated Mass
Run No. (micrograms)
1 -1343.6
2 - 50.6
3 + 143.4
4 - 33. 7
5 + 100.8
6 - 26.6
7 - 183. 1
8 + 270. 7
9 + 131.4
10 - 59.6
11 + 68.1
12 + 49.6
13 + 131. 6
14 + 470. 0
15 - 301.4
16 - 21.1
3-5
-------
These tests confirmed conclusively that a light leak did exist in the
detector. In tests 10, 11, 12 and 16 the values obtained, while larger
than desirable, were considered adequate for this series provided flow
rates of approximately 4 cfm are used. It was then decided to use filters
in the transfer mechanism, taking care to see that external light
reaching the interior of the unit was minimized.
3.1.5 Stability Tests -- Using Filters
Tests were performed again using Dr. Anderson's medium on
the clean air previously discussed. Up until this point little attention
had been paid to deformation of the filter. On tests 5 and 6, filter
deformation in the form of tearing of the filter away from the cassette
was definitely observed.
Total Indicated Mass
Run No. (micrograms)
1 - 86.4
2 + 48. 1
3 - 3. 0
4 + 29. 0
5 +292.6
6 + 95. 9
On the basis of these data, the decision was made to change media and
use the tougher Pallflex media for the remainder of the tests. It was
decided to repeat the previous test using the Pallflex medium.
3.1.6 Stability Tests -- Pallflex Medium
Tests were again performed using a 47mm in-line filter to
obtain clean air. It had also been theorized at this point that some
error may be occurring due to either moisture loss in the medium
3-6
-------
or actual material loss from the filter itself. This was as a result
of data from tests at IN prior to the Cincinnati test series that showed
the no-cassette stability to be well within the ± 30|j.g total indicated
mass specification. Thus six tests were run using the same filter
cassette on all tests.
Total Indicated Mass
Run No. (micrograms)
1 -126.4
2 loss of record
3 - 83.2
4 - 58.6
5 - 31.4
6 - 27.4
On the basis of these data, it became fairly clear that some kind of
process error was occurring. It was not determined at this time
whether tbe error was due to moisture or material loss. Later studies
(Section 2.2.2 of this report) point tTie finger rather strongly towards
moisture, but no conclusion can be drawn from these studies regarding
the data from either the Cincinnati or Dow series of tests.
This concluded the Cincinnati series except for some tests on
a laboratory aerosol which were unsuccessful due to the shortness of
time available. Even though limited data had been obtained, the series
was considered successful in that the obvious deficiencies in design and
alignment had been pinpointed.
3. 2 The Dow Series
The tests at the Dow Chemical Company in Midland, Michigan,
during the week of 25-29 October, 1971, may be divided into three
3-7
-------
general areas: tests on 1) leaded fuels (3. Occ/gal TEL), 2) unleaded
fuels (0. 0 cc/gal TEL), and 3) low-leaded fuels (0. 5 cc/gal TEL).
Three different cars were used for the three different areas and will
be discussed later. Tests were performed with the automobiles operating
under steady state conditions (constant speed), the Federal LA-4 cycle,
and the older California Seven-Mode cycle.
The Dow facilities were quite well equipped and similar in
basic construction to the EPA's Cincinnati facilities. An isokinetic
mixing tube was used, in which the overall flow was 500 cubic feet
per minute. Two samples were taken by Dow for gravimetric com-
parison to the beta gauge. Initially these samples were two 143 mm
filters but later consisted of one 143 mm filter and one 2 inch filter.
Prior to the actual automotive tests, a test was run using a
cassette prepared with a filter as the medium (all tests but one used
Pallflex E70/2075W) was delivered from the manufacturer. This filter
showed a weight loss of 70. 6 micrograms. In order to test the stability
of the unit (± 30 micrograms was the specification) two tests were run
with no cassette at all in the instrument showing +10. 9fJ.g and +4. l|j.g,
respectively. Convinced that electronically the unit was well within
specification, several cassettes were prepared by baking at various
temperatures for various times and allowed to stabilize for various short
times in the ambient conditions. This baking discolored the filter
somewhat, possibly introducing contaminants (no discoloration had
3-8
-------
ever been noticed before at idential temperatures), and as a result
this conditioning was no longer employed after this series. The pump
was then allowed to draw clean air through the filter for a short time.
The results are as follows:
Bake Time Bake Temp Equalize Time Pump Time Weight Change
Test No. (min) ( ° F) (min) (min) (micrograms)
1
2
3
4
5
6
10
10
10
10
20
210
210
210
210
200
10
20
30
45
30
5
5
5
5
20
20
-54. 0
-93. 2
-83.2
-55. 4
-48.2
-77. 8
Test No. 6 is a repeat of the cassette used in test No. 5 with no inter-
mediate baking. That is to say, the cassette was immediately reprocessed
through the beta gauge to determine whether the same phenomena which
existed during the Cincinnati series (Section 3. 1. 6 of this report) could
be duplicated here. Obviously, it was not.
The average of the above six tests was -68. 6 micrograms, and
represented a "process" error (since all readings are negative and in
the 50 to 90 microgram category). Since no solution to this problem was
immediately apparent, it was decided to delay a more detailed study in
this area since the first automotive series would be run with leaded fuels
and the mass of collected particles was expected to be much higher than
this figure.
3-9
-------
3.2.1 Tests on Leaded Fuel (3. 0 cc/gal TEL)
On tests 1 through 8, the flowmeter was placed downstream of
the pump. After these 8 tests, it was determined that the pressure
fluctuations introduced into the air flow by the pump was interferring
with the vortex precession action of the Swirlmeter. For all subsequent
tests at Dow, the flowmeter was placed upstream of the filter.
A synopsis of this series is presented in Table 3-1. Little
conclusion can be drawn from tests 1 through 8 due to pump interference.
In tests 5, 9, and 10, it is suspected that there was electrical noise
introduced into the power line, causing the computer to malfunction.
This problem did not occur for the remainder of the Dow series.
Tests 11 through 15, however, showed excellent agreement between
the beta gauge and a gravimetric comparison.
3. 2. 2 Tests on Non-Leaded Fuel
The next series of tests were performed using an automobile
which had been fueled solely with non-leaded fuel since its manufacture.
Considerable discrepancy existed in cases between the beta gauge and
the gravimetric comparison. This phenomenon is not clearly under-
stood as of the date of this report. To provide additional data, and since
good agreement was obtained between the two 143 mm gravimetric com-
parisons, it was decided to substitute a 2 inch diameter filter for one
of the 143 mm diameter filters for test 20 and all subsequent tests.
3-10
-------
Test
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Test
Mode
Moist. Chk.
30 mph
30 mph
30 mph
--
60 mph
LA-4(C.S. )
LA-4(H. S. )
LA-4(H. S. )
LA-4(H. S. )
60 mph
60 mph
60 mph
30 mph
1 5 mph
Pump Time
(minutes)
5
20
20
20
--
20
23
10
23
23
10
10
10
15
15
IN Mass
Gain ((jig)
-14
2980.
2491
2485
--
2366
3230
2151
--
--
3627
3870
1929
1502
1437
TABLE 3-1
LEADED FUEL TESTS
IN Total
Flow (liters)
595
2315
1935
1840
--
874
1022
787
--
--
453
795
417
1184
1382
Dow 143 mm
Gain (^g)
--
3650
3850
3600
--
13200
14000
4050
--
--
9150
5800
5750
2400
1750
IN Rate
(mg/min)
--
18.2
18.2
19.1
--
38. 3
45.8
39. 1
--
--
113. 3
68. 9
65.5
18. 0
14. 7
Dow Rate
(mg/min)
22.8
24. 1
22. 5
--
82. 5
76. 1
50.6
--
--
114.4
72. 5
71.9
20.0
14.8
% Dev.
IN - Dow
Dow
-20. 1
-24.2
-15. 1
--
-53.6
-39. 8
-22. 7
--
--
- 1.0
-5.0
- 8.9
-10. 0
- 0.6
Remarks
A
B
B
C
B
C
C
D
A
NOTES: 1. Dow Flow Rate for all tests - 4 CFM
2. Mass Flow Rates are from automobile
3. Fuel is 3. 0 cc/gal TEL
4. Vehicle is 1971 Chevrolet - 350 CID Engine
REMARKS: A. Flow not constant throughout test
B. Difficulty adjusting Flow - Pump
Interference Suspected
C. Run Aborted - Computer did not cycle
D. Flow meter upstream of filter for all
subsequent Dow tests.
-------
This approach would provide data as to whether the phenomenon
was area sensitive. This was indeed the case and there are some later
indications that the indicated mass gain of the filter may be a function
of different filter manufacturing lots. The data regarding this, however,
is inconclusive at this point. Tables 3-2 and 3-3 synopsize the data
from the non-leaded tests. A significant observation is that the per-
centage deviation between the 2 inch and 143 mm sample is of the same
order as the beta gauge (20 mm diameter sample) and the 143 mm sample.
This phenomenon is currently attributed to a higher moisture absorption
and absorption of high volatility organics by the larger filter. Inasmuch
as excellent data had been obtained with leaded fuels, it was decided
to discontinue the non-leaded fuels and switch to low-leaded fuels.
3.2.3 Tests on Low-Leaded (0. 5 cc/gal TEL) Fuels
This series was also run using a 1971 Chevrolet with 350 CID
engine. There was some attempt during this series to vary certain
parameters surrounding the sampling process and determine their effect
but little conclusion can be drawn from these attempts. Tables 3-4 and
3-5 present a synopsis of the low-leaded tests.
Here again the beta gauge agreed quite closely in several cases
with the 2 inch diameter filter. It is interesting to note that as the test
mode switched from the LA-4 cycle to steady-state operation (tests 26
and 27), the polarity of the percentage deviation between the two inch
filter and the 143 mm filter reversed, causing the beta gauge to give
3-12
-------
Test
No.
16
17
18
19
20
21
Test
Mode
LA-4(C. S.)
LA-4(H. S. )
LA-4(H. S. )
LA-4(H. S. )
LA-4(H. S.)
Calif. Cyc.
Pump Time
(minutes)
23
10
10
23
23
60
IN Mass
Gain (^g)
1049
156
33
230
197
359
TABLE 3-2
IN Total
Flow (liters)
2454
1208
736
2075
2378
4750
Dow 143 mm
Gain (jig)
4050
1350
1400
1400
1400
2200
Dow 2 in
(Gain (p.g)
--
--
--
--
200
600
IN Rate
(mg/min)
6. 05
1.83
.64
1. 57
1. 17
1.08
Dow 143 mm
Rate (mg/min)
22.01
16.87
7.61
7.61
7.61
5. 79
Dow 2 in
Rate (mg/min)
--
--
--
--
1. 32
1. 92
TABLE 3-3
Test
No.
16
17
18
19
20
21
% Dev.
IN -143 min
143 min
-72. 5
-89. 1
-91.6
-79. 4
-84.6
-81.0
% Dev.
IN -2 inch
2 inch
--
--
--
--
-11. 3
-43.8
% Dev.
2 inch- 143 min
143 min
--
--
--
--
-82. 9
-66.8
Remarks
Test terminated at 47. 5 min. (Automotive Failure)
NOTES: 1. Dow 143 mm flor - 4CFM
2. Dow 2 inch flow - 3. 3 CFM
3. Fuel is non-leaded
4. Vehicle is 1971 Chevrolet - 350 CID engine
5. 2 inch filter had 6. 5 in Hg differential pressure
6. 143 mm filter had negligible differential pressure
-------
TABLE 3-4
Test
No.
22
23
24
25
26
27
28
29
30
31
32
33
Test
Mode
LA-4(C.S.)
LA-4(H. S. )
LA-4(H. S. )
LA-4(H. S. )
LA-4(H. S. )
60 mph
30 mph
1 5 mph
Idle
30 mph
60 mph
60 mph
Pump Time
(minutes)
23
10
23
23
23
15
15
15
15
15
5
5
IN Mass
Gain ((ig)
2796
960
251
298
959
4119
350
332
248
510
2308
1605
IN Total
Flow (liters)
1784
1222
506
506
1840
993
996
1469
1488
1472
496
495
Dow 143 mm
Gain ((ig)
1700
1600
2300
2400
2500
7000
1200
700
500
1500
3000
2200
Dow 143 mm
Flow (liters)
2600
1130
2600
2600
2600
1400
1400
1400
1400
1700
566
566
Dow 2 in
Gain ((jig)
8200
900
1300
1400
1400
600
400
300
200
500
1600
1200
Dow 2 in
Flow (liters)
2140
931
2140
2140
2140
1400
1400
1400
1400
1400
465
465
IN Rate
(mg/min)
22. 2
11. 1
7. 0
8. 3
7.4
58. 7
5. 1
3.2
2.4
4.9
65. 5
45. 7
-------
TABLE 3-5
Test
No.
22
23
24
25
26
27
28
29
30
31
32
33
Rate (mg/min)
38.6
20. 0
12.5
13.0
13.6
70.9
12.2
7. 1
5. 1
12.5
75.0
55.0
Rate (mg/min)
54.2
13.7
8.6
9.3
. 9. 3
52.5
4. 1
3. 0
2.0
5. 1
48.6
36.5
% Dev.
IN - 143mm
143 mm
-42. 4
-44. 4
-43. 9
-36.3
-45.7
-17.2
-58.0
-55.0
-53.4
-60. 8
-12. 7
-16.8
% Dev.
IN - 2 in
2 in
-59.0
-18. 7
-18.4
-10. 3
-20. 3
+12. 0
+26.0
+ 5.0
+16. 4
- 3.2
+34.6
+25.4
% Dev.
2 in - 143 mm
143 mm
+54. 1
-31. 7
-31. 3
-29.1
-31. 9
-26.0
-66. 7
-57.1
-60. 0
-59. 5
-35. 1
-33. 7
Remarks
IN Flow/2in Flow
not constant
Flows steady
Face velocities
matched on IN and
and 2 in
Gelman Type A
filter used
143mm - 2 in flow
rates matched
143mm - 2 in flow
rates matched
143mm - 2 in flow
rates matched
143mm - 2 in flow
rates matched
143mm flow returned
to 4 cfm
Doors to facility open
Doors to facility
closed
-------
an indication which is between the two gravimetric measurements.
There was a very evident change in moisture content between the
two modes with some condensate appearing on the clear plastic tubing
which was used during the La-4 cycle and no visible condensate during
the steady state runs. The temperature of the sample flow did not
appear to change appreciably on any of the runs besides the 60 mph
steady-state runs. There is also a higher concentration of volatile
organics during the La-4 cycle and this may also be a major contributing
factor.
An experiment was also run by varying the temperature of the
room in which the automobile was located. This was accomplished
by running a test with the front and rear doors of the dynamometer
facility open, then duplicating the test with the door nearest the front
of the automobile closed. This produced a temperature variation of
30°F in the room and tests 32 and 33 are the results. Notice that there
is little difference between tests 27 and 32, both being run with the doors
open. Test 33, however, showed a significant reduction in particulate
emission resulting from raising the intake air temperature to the engine.
3. 3 Conclusions Regarding the Field Tests
The field tests at Dow and Cincinnati were sufficient to establish
the capability of beta gauging as a process monitor. The detector
stability achieved and the use of digital processing throughout assures
high computational accuracy and repeatability. Much, however, remains
3-16
-------
to be learned about the sampling process. The fact that poor correla-
tion is achieved between filters of different diameters must surely be
more fully explored and understood.
The data on moisture absorption by paper filters also indicates
this medium to be undesirable for most applications requiring resolu-
tion of 1 - 2% of filter basis weight. Such materials as the GE Nuclepore
or Gelman Type A do not exhibit as serious a problem in this area as the
Pallflex E70/2075W and it is understood that these materials are under
investigation by the EPA at the time of this writing. Other materials
must be investigated and the .filter industry must be called upon to
address this problem.
Variations of a percentage of initial filter basis weight would
affect both gravimetric and beta gauging identically. The larger filter
face velocity of the beta gauge, however, would tend to concentrate the
material and result in a larger basis weight change for the particles.
For example, the effective sample area for the beta gauge is approxi-
mately 3 cm while for the 2 inch gravimetric comparison, the effective
2
area is approximately 10 cm . At 4. 2 cfm for the beta gauge, this
3 2
represented a normalized flow rate of 1. 4 ft /(min cm ). At 3. 3 cfm
3 2
for the gravimetric unit, the normalized flow rate is 0. 33 ft /(min cm ).
Obviously, then, for the same stream concentration, more material will
be collected on the beta gauge's filter on a per-square-centimeter basis
and the moisture variance will not be as significant.
3-17
-------
It is not the intent of this document to attribute the inconsistent
data solely to moisture. Rather, this was an area in which IN has
considerable expertise and facilities to investigate (at least under static
conditions) from our previous experience in the paper industry. It is
the intent, however, to point out to the reader that moisture can (and
in many cases likely will) be a major cause of inconsistent data. The
effects of filter composition on "errors" in data are relatively unknown
to IN, and as a result, little positive or conclusive data can be presented
in this area.
3-18
-------
4.0 RECOMMENDATIONS
There are several areas of study and clarification revealed as
a result of this program. First, it is believed that some definition is
required by the Environmental Protection Agency as to what constitutes
"particles". This may eventually require an exact preparation of the
mixing tube air with respect to temperature and humidity. It may also
require withdrawing a sample from the tube at a reasonably fixed flow
rate (implying a constant pressure drop across the filter, barring any
filter clogging).
In order to do the above, however, the problem of moisture
interference must be solved. In IN's opinion, this is one of the major
problems facing particle sample today. It should be of concern
not only to those involved with mobile sources but also those concerned
with stationary source or ambient monitoring. It is not sufficient simply
to allow the gravimetric samples to stabilize in a controlled environment
room as Figure 2-5 of this report will verify. In effect, in order to
place any validity on data, the complete previous history of the filter
must be either known or standardized and the nature of each sample must
be known and taken into consideration. Solution may require either the
discovery of a medium already produced but not widely touted or the
manufacture of a new medium, using materials specifically designated
to ward off water.
4-1
-------
There is also limited data which shows that the composition
of the filter medium may introduce wide fluctuations in the data. This
is attributed to an affinity (or lack of same) for highly volatile constituents
in the sample which are condensed from the stream and which would
ordinarily remain in the gaseous state. A study should be initiated
to determine the total nature of vehicular exhaust under both steady state
and Federal Cycle operating conditions. This information could then be
used to specify (or eliminate) certain materials used in the manufacture
of the medium.
It is believed that extensive experimentation will be required by
the EPA to establish correct and reliable sampling procedures. The
proof of principle of beta gauging has been successful: a nucleonic gauge
can be built with the required sensitivity. Further development in this
area should be directed towards specific application areas rather than
general purpose application. A filter medium must be found, not only
for the beta gauge but for gravimetric comparison as well. The IN
design is excellently suited to this type of experimentation by virtue
of the cassette approach and should be able to yield valuable data in
this area.
Finally, the Federal Cycle Test as currently understood by
IN is at best cumbersome. It is believed that solution of the above two
problems will yield far more consistent results than available in the
past. It is also our belief that reliable data may also show that a
4-2
-------
definite correlation exists between a typical steady-state sample
(60 mph, for example) and the LA-4 Federal Cycle Test. If this
be the case, a considerable reduction in the complexity (and there-
fore cost) of an automobile particulate emission gauge is possible.
4-3
-------
BIBLIOGRAPHIC DATA
SHEET
1. Report No.
3. Recipient's Accession No.
4. Title and Subtitle
Development of a Nucleonic Particulate Emission Gauge
5. Report Date
March, 1972
6.
7. Author(s)
Charles R. Duke and Boong Y. Cho
8. Performing Organization Rept.
No.
9. Performing Organization Name and Address
Industrial Nucleonics Corporation
Federal Systems Division
650 Ackerman Road
Columbus, Ohio 43202
10. Project/Task/Work Unit No.
11. Contract/Grant No.
68-02-0210
12. Sponsoring Organization Name and Address
Environmental Protection Agency
Research Triangle Park, North Carolina 27711
13. Type of Report & Period
Covered
Final Report
14.
15. Supplementary Notes
16. Abstracts
A project was conducted to design, fabricate, and test a particulate
emission monitor utilizing a beta radiation attenuation and filter sampler
technique. The instrument was to have a minimum sensitivity of 125 ±25
micrograms per cubic meter and be capable of operation in either a manual
or automatic mode of operation. Carbon-14 was selected as the optimum
isotope and a filter "cassette" was used to collect the particles. The effect
of moisture absorption by various types of filter media were investigated
and reported. A series of tests were performed on automobiles using leaded,
low-leaded and non-leaded fuels. The results of these tests are reported as
well as recommendations for future effort.
17. Key Words and Document Analysis. 17a. Descriptors
Automotive Pollution
Beta Gauges
Filter Samplers
Filters
Air Pollution
Moisture Interference
Automotive Organics
Automatic Particulate Monitors
17b. Identifiers/Open-Ended Terms
17c. COSATI Field/Group
18. Availability Statement
Unlimited
19. Security Class (This
Report)
UNCLASSIFIED
~*.^~^..~^>. .^^,
20. Security Class (This
Page
UNCLASSIFIED
21. No. of Pages
sixty-two (62)
22. Price
FORM NTIS-35 (10-70)
USCOMM-DC 40329-P71
-------
|