Handout
Compliance Testing
Quality Assurance
Procedures Workshop
Selected Papers On
Paniculate Sampling In
Cyclonic Flow
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF ENFORCEMENT
OFFICE OF GENERAL ENFORCEMENT
SHINGTON, D.C. 20460
July 1980
-------
INTRODUCTION
In many cases, particulate stack testing is required at
sources where cyclonic (tangential) flow exists at the sampling
location. Where cyclonic flow conditions occur, the effluent
gas flows in a spiral motion up the stack. At each point in
the stack, the gases have both axial and tangential velocity
components. Cyclonic flow occurs most often after inertial
demisters following wet scrubbers, in stacks with tangential
inlets, and after axial fans.
Modification of the source to eliminate cyclonic flow or
to provide an alternate sampling location is often neither
feasible nor possible. Application of standard particulate
sampling methods can result in significant biases in emission
measurements. Thus, modifications to the sampling methodology
are necessary to provide accurate stack testing results.
This document is a compilation of six papers and articles
which address the significance of sampling errors associated
with velocity - volumetric flow rate measurements and particulate
concentration - emission rate measurements in cyclonic flow
situations. Modifications of standard sampling methodology to
provide more accurate emission measurements are also discussed.
The papers which are included are:
ANALYSIS OF SAMPLING REQUIREMENTS FOR CYCLONE OUTLETS;
Michael Durham and Dale Lundgren.
SAMPLING OF TANGENTIAL FLOW STREAMS; Dale A. Lundgren,
Michael D. Durham, and Kerry Wade Mason.
CHARACTERIZATION OF CYCLONIC FLOW AND ANALYSIS OF
PARTICULATE SAMPLING APPROACHES AT ASPHALT PLANTS;
James W. Peeler, Frank J. Phoenix, and D. James
Grove.
A METHOD FOR STACK SAMPLING CYCLONIC FLOW; Charles L.
Goerner, Fred H. Hartmann, and James B. Draper.
m
-------
Introduction - cont'd
5. CYCLONIC FLOW - CHARACTERIZATION AND RECOMMENDED
SAMPLING APPROACHES; Frank J. Phoenix and D.
James Grove.
6. FLOW VELOCITY STUDIES - D. P. Saari and H. A. Han-
son, from Section 6, "Wet Scrubber Sampling Tech-
niques," in "Effective Sampling Techniques for
Particulate Emissions from Atypical Stationary
Sources," January 1980, EPA-600/2-80-034.
These papers- present a variety of approaches toward de-
fining and resolving the problems associated with sampling in
cyclonic flow conditions. The information which is presented
should be of use to source operators, source testers, and con-
trol agency personnel in developing appropriate testing proce-
dures where cyclonic flow is encountered.
IV
-------
PAPER I
ANALYSIS OF SAMPLING REQUIREMENTS
FOR CYCLONE OUTLETS
Michael Durham
Dale Lundgren
-------
ANALYSIS OF SAMPLING REQUIREMENTS FOR CYCLONE OUTLETS
by
Michael Durham
DRI Electronics
University of Denver
Denver, Colorado 80208
(303)753-2241
and
Dale Lundgren
Dept. of Environmental Engineering
A.P. Black Hall
University of Florida
Gainesville, Florida 32611
(904)392-0846
A comprehensive analysis of intertial effects in aerosol
sampling was combined with a thorough study of swirling flow
patterns in a stack following the exit of a cyclone in order
to determine the errors involved in sampling particulate matter
from a tangential flow stream. Aerosol sampling bias was ana-
lyzed by comparing samples taken from a 10 cm wind tunnel at
duct velocities varying from 550 to 3600 cm/sec. Experiments
were performed at four sampling angles: 0, 30, 60 and 90 degrees
and for particles 1 to 19.9 micrometers in diameter. A mathe-
matical model was developed and tested which predicts the sampling
error when both nozzle misalignment and anisokinetic sampling
velocities occur simultaneously. A three-dimensional or five-
hole Pitot Tube was used to map cross-sectional and axial flow
patterns in a stack following the outlet of a cyclone. Using
information found in this study, a simulation model was developed
to determine the erros involved when making a Method 5 analysis
in a tangential flow stream.
1-3
-------
ANALYSIS OF SAMPLING REQUIREMENTS FOR CYCLONE OUTLETS
INTRODUCTION
Obtaining a representative sample of particulate matter from a
stack following the outlet of a cyclone poses a difficult problem when
standard sampling methodology is used. The swirling flow pattern pro-
duced by the cyclone is well maintained in a circular stack so that the
air flows in spiral or helical paths up the stack. Since the gas stream
flows at an angle to the stack axis, sampling errors are induced due to
the inertia of the particles and the limitations of the velocity measuring
instrument presently being used.
The analysis of sampling errors induced by cyclonic flow was ap-
proached from two directions in this study. One approach involved an
investigation of aerosol sampling bias due to anisokinetic sampling
velocities and misalignment of the nozzle with respect to the flow
stream as a function of particle and flow characteristics. The second
part of the study involved an accurate mapping of the flow patterns in a
tangential flow system. The information obtained in the two parts of
the study were then combined to simulate the errors that would be en-
countered when making an EPA Method 5 (1971.1977)1'2 analysis in a
tangential flow stream.
Review of the Literature on Anisokinetic Sampling
In order to obtain a representative sample of particulate natter
from a moving fluid it 1s necessary to sample isokinetically with the
inlet velocity equal to the free stream velocity and the nozzle aligned
parallel to the flow stream (Wi1cox,1957).3 Most of the early research
in this area has been concerned with the sampling errors when the ratio
of the free stream velocity to the inlet velocity is other than
unity. Several authors (Watson,1954; Badzioch,1959; Davies.1968; Lundgren
and Calvert,1967)4-7 found that the amount of error was a function of
the velocity ratio, particle inertia, and nozzle velocity and could be
best characterized by the dimensionless inertial impaction parameter or
Stokes number, K, defined as:
(1) K = CppV0Dp2/l8nD1
where
C = Cunningham's correction for slippage
p ~ particle density
VQ = particle velocity
D« = particle diameter
P
q = viscosity of the gas
D. = nozzle diameter
1-4
-------
Extensive experimental studies were performed by Belyaev and
Levin(1972,1974)8Be in which flush illumination photographic techniques
were ysed to itudy the trajectories of particles approaching and entering
a sampling nozzle. The results shown in Figure 1 and described by
equations 2 and 3 illustrate the relationship between Stokes number and
the aspiration coefficient (ratio of the sample concentration to the
true concentration) us a function of the velocity ratio (R=VQ/Vp.
(2)
(3)
. 617/R)K
The curves confirm the results ©f Dennis et al.(1957)10 and White! ey and
Reed(1959)1J that for small Stokes numbers the aspiration coefficient
approaches 1 for all velocity ratios (R)0 and for large Stokes numbers
it approaches R.
The sampling bias due t© sisalignment of the nozzle with the flow
stream is similar to that caused by superisokinetic sampling (R<1).
When the nozzle is at an angle to the flow stream , the projected area of
th$§ TOzzle is reduced by a factor equal t© the cosine of the angle.
Even if the nozzle velocity is equal to the flow stream velocity an
aspiration coefficient less than or equal to unity will be obtained
because ioroe of the larger particles will be unable to sake the turn
into the nozzle with the streamlines. May hood and Langs troth „ as re-
ported by Watson(1954)<1!) and Glauber®an(19S2)12 experimentally found
that the amount of sampling Grror increased proportional io the particle
size and th© angl© ©f o1s alignment. An ©quation derived by Lundgren et
al.(197S)13 describes th© stapling bias that would occur when both the
nozzle is oisaligned with th® flow gtreai and the velocity ratio is
other than unity:
(4) A = 1 * (Reos0=l)p'(K8R,e)
Where p' is a function ©f the velocity ratio. Stokes number, and the
angle of EaisaHgnraent. T© satisfy the boundary conditions, p* ®ust
approach zero for seuall Stokes numbers and must approach 1 for large
Stokes numbers. This aeans when the velocity ratio is equal to ls the
curve for the aspiration coefficient will approach cosQ at large values
of K.
Review of the Literature on Tangential Flow
The swirling flow 1n a itack following the outlet of a cyclone,
combines the characteristics of vortex aotion with axial motion along
the stack axis. Since this represents a developing flow field, the
swirl level decays and the velocity profiles and static pressure
distributions change with axial position along the stack (Baker and
Sayre,1974).14 Velocity vectors in tangential or vortex flows are
composed of axial, radial, and tangential or circumferential velocity
components (see Figure 2). The relative order of magnitude of the velocity
components varies across the flow field with the possibility of each one
of the components becoming dominant at particular points (Chigier,1974).1S
1-5
-------
1.0
2.0 3.0
STOKES NUMBER (K)
4.0
!5.0
Figure 1. Sampling efficiency as a function of Stokes number (K)
and velocltv ratio YR=V_/V.»}.8»9
Figure 2. Velocity components in a swirling flow field.
1-6
-------
The (istablish©d vortex flows are g@n@r@lly nxisyraetric but during
formation of the spiral ing flow the syanetry is often distorted.
Two distinctly different types of flow that are possible in a
swirling flow field are known as free vortex and forced vortex flows.
When the swirling component of flow is first created in the cyclone
exit, the tangential profile of the induced flow approahces that of a
forced vortex. As the forced vortex flow moves along the axis of the
stacks momentum transfer and losses occur at the wall which cause a
reduction in the tangential velocity and dissipation of angular momentum.
This loss of angular momentum id due to viscous action aided by unstable
flow and fluctuating components. Simultaneously, outside the laminar
sublayer at the wall where inertial forces are significant, the field
develops toward a state of constant angular momentum. This type of flow
field with constant angular momentum is classified as free vortex flow.
The angular aomentum and tangential velocities of the flow decay as the
gas stream flows up the stack. However, Baker and Sayre(1974) found
that even after 44 diameters the tangential velocity is still quite
significant when compared to the axial velocity. Therefore, satisfying
the EPA Method 5 requirement of sampling 8 stack diameters downstream of
the nearest upstream disturbances will not eliminate the effect of
sampling in tangential flow.
Types of errors that would be expected to be introduced by tangential
flow are nozzle misalignment, concentration gradients and invalid flow
tisasurisEjents. The sampling error caused by nozzle misalignment was
described previously. Concentration gradients occur because the rotational
flow causes the larger particles to move toward the walls of the stack,
causing higher concentrations in the outer regions. Masdn(1974)ls ran
tests at thu eutlut of a small industrial cyclone to determine the
magnitude of Grrors induced by cyclonic flow. He found that flow angles
as high us 70® existed in the stack and sampling with the nozzle parallel
to the stack wall produced an error of 52.7%. However, particle size
distribution tests showed no significant effect of a concentration
gradient across the traverse.
Th© errors in the aeasureraent of velocity and subsequent calculations
of flow rat© in tangential flow are due primarily to the erudeness of
the instruments used in source sampling. Because of the high particulate
loadings that exist in source sampling, standard pi tot tubes cannot be
used to seasure the velocity. Instead, the S-type pi tot tube aust be
used since it has large diameter pressure ports that will not plug.
Although th© S-type pi tot tube will give an accurate velocity ffleasurement,
it is somewhat insensitive to the direction of the flow (Hanson and
Saari,1977; Brooks and Williams, 1975; Grove and Ssiith,1973; Hanson et.
al.,1976, and Williams and DeJarnetter,1977).17=2i Although the S-type
pitot tube is very sensitive to pitch direction, the curve for yaw angle
(Figure 3) is symmetrical and somewhat flat for an angle of 45° in
either direction. Because of this insensitivity to direction of flow in
the yaw direction, the S-typ® pitot tube cannot be used in a tangential
flow situation to align the nozzle to th© direction of the flow, or to
accurately fasasur© the velocity in a particular direction.
The velocity in a rotational flow field can be broken up into three
components in the axial, radial, and tangential directions (see Figure 2).
The magnitude of the radial and tangential components relative to the
axial components will determine the degree of error induced by the
tangential flow. Neither the radial nor the tangential components of
. 1-7
-------
-40
-20
Q 4.57 a/sec
Q 15.24 re/sec
A 9.14 m/sec
20 40 g\ 60
Yaw Angle, degrees
-10*
.-201
Percent
Velocity
Error
. -30t
Figure 3. Velocity error vs. yaw angle for an S-type pitot'tube.
20
I J -.1 ..I .1
s 10 is
Ye* Angle, decrees
Pressure
Dirrerential.
Figure 4. Five-hole pitot tube
sensitivity to yaw
angle.
20
1-8
-------
velocity affect the flew rate through th© stack, but both affect th©
velocity Measurement raade by the S°type pi tot tube because it lucks
directional sensitivity.
Two eoMon types of pressure probes capable of ®
-------
EXPERIMENTAL DESIGN
Since cyclonic flow leads to probe misalignment it was necessary to
first determine the relationship between sampling bias and the angle of
the nozzle to the flow stream. This was accomplished by sampling BOno-
disperse particles, 1-20 micrometers in diameter, flowing through a
10 cm. wind tunnel shown in Figure 5. Two simultaneous samples were
taken, one para1 ">el to the duct and the other at an angle of 30, 60, or
90 degrees to the axis of the duct. The sampling rates were identical
so that the concentration difference between the two represented the
inertial sampling bias. Preliminary tests with both nozzles sampling
isokinetically and parallel to the flow proved that the concentration at
the two traverse positions was identical. Particle diameter, duct
velocity, and nozzle diameter were varied to produce a range of Stokes
numbers from 0.007 to 2.97. In order to determine the effect of simul-
taneous probe misalignment and anisokinetic sampling rates, additional
tests were performed in which the control nozzle sampled at an isokinetic
rate parallel with the flow stream while the test nozzle sampled at an
anisokinetic sampling rate and at an angle to the duct axis.
The system used to map the flow pattern in a tangential flow stream
is shown in Figure 6. It consisted of a 34,000 liters per minute In-
dustrial blower, a section of 15 cm. PVC pipe containing straightening
vanes, a small industrial cyclone collector, followed by a 6.1 meter
length of 20 cm. PVC pipe. The 150 cm. long cyclone was laid on its
side so that the stack was horizontal and could be conveniently traversed
at several points along its length. A change in flow through this
system was produced by supplying a restriction to the inlet of the
blower.
To measure the velocity in the stack, a United Sensor type DA
3-dimensional directional pitot tube was used. The probe is 0.32 cm. in
diameter and is capable of measuring yaw and pitch angles of the fluid
flow as well as total and static pressure. The yaw angle is a measure
of the flow perpendicular to the axis of the stack and tangent to the
stack walls, while the pitch angle is a measure of the flow perpendicular
to the axis of the stack and perpendicular to the stack walls.
RESULTS
Analysis of the Inertial Sampling Bias
The results of the tests to determine sampling bias as a function
of angle of misalignment and Stokes number are shown in Figure 7. For
all three angles, the curves approach a theoretical limit of cos8.13
However, the Stokes numbers where the curves closely approach their
limits decrease with increasing angles. This is due to an effective
decrease in nozzle diameter produced by the angle of misalignment. An
equation was developed to account for this and produces an "adjusted
Stokes number" (K') defined by:
(5) K' = Ke0-0229
1-10
-------
TO PUMPS AND GAS METERS
DILUTION AIR
I
AEROSOL
GENERATOR
MIXING
CHAMBER
TEST
SECTION
s
ABSOLUTE
FILTER
;TRAIGHTENING
VANES
ORIFICE
PLATE
BY-PASS
^
TO BLOWER
Figure 5. Experimental system to determine inertial sampling bias.
1-11
-------
IS cm. ID
(oj
Blower
Straightening vanes
•2.4
Cyclone
6.1 m
Figure 6.
Experimental system for measuring
cross sectional flow patterns in
a swirling flow stream.
o
>-•
N
O
M
3
00
•Jt
n
n
20cm
1-12
-------
Using this correction an equation was empirically derived for $' to be
used in equation 4 for R=l:
(6) p;(K',e,R=l) = 1 - 1 f
1 * 0.55K'e°'25t('
This equation is plotted along with the experimental data in Figure 7.
Further testing was performed to develop an equation to describe
the sampling bias for R^l and 6^0. The equation would incorporate the
work of Belyaev and Levin(197281974)8'8 for 6=0 and R^l and the results
for R=18 8^0 described by equation 6. The following equation was found
to best fit the data:
(7) p'(K,R,8) :
where
(8) P(K'.R) is defined by equation 3
t*s^l) is defined by equation 3 evaluated at
p*(K'8©88=l) is defined by equation 6
Equation 7 is plotted along with the data for 8=0.5 and 2.0, and ©=60e
in Figure 8.
Analysis of Cyclonic Flow Patterns
Eight traverse points for the velocity measurements were selected
according to EPA Method 1. Measurements were lade using the 5-hole
pi tot tube at two flow rates and at fivt axial distances froa the
inlet°=lDs 20, 40, 8D and 16D8 where 0 is the inner diameter of the
duct. At each point in the traverse8 the pitot tube was rotated until
the pressure differential between the yaw pressure taps was zero. This
angle was recorded as the yaw angle and the pressure readings frora all
five pressure taps were recorded for later calculation of total and
static pr@ssure8 and pitch angle.
During the initial traverse, a eor® area was discovered in the
center of the duct where the direction of the flow could not be determined
with the pi tot tube. The core area was characterized by negative readings
at all five pressure taps which did not vary nueh with rotation of the
probe. The location of the core area was sseasured at Qach location
along the duct axis and recorded.
Table X shows an example of the calculated results of the velocity
raeasureraents at eight diaraeters downstreaa of the cyclone for the lower
flow rate. The angle 0 represents the angle of the flow relative to th©
axis of the duct. The Reynolds number of the system calculated on a
basis of average axial flow rates of 11„260 and 15,500 liters per minute
were 80,000 and 111,000 for the low and high flow rates respectively.
The velocity measurements at the other traverse points for both flow
1-13
-------
30'
60*
90*
EQUATIONS 4 & £
0.1 1.0
STOKES NUMBER (K)
10.0
Figure 7. Aspiration coefficient vs. Stokes number—empirical equation
and experimental data for 30, 60, and 90 degrees.
1.0
° EXPERINERTAL DATA, R-2, 8»€0*
o EXPERIMENTAL DATA, IKS, 0-60*
EQUATIONS 4(7
STOKES NUMBER (K)
Figure 8. Sampling efficiency vs. Stokes number at 60° misalignment
for R - 2.0 and 0.5.
1-14
-------
pates and all five axial distances showed approximately the saae character-
isties. The pitch angle increased from the care area to the duet wall.
The yaw angle and the combined angle $ decreased from the core area to
the walls. At the inlet and up to eight diameters downstream,, ungles as
high us 70® wire found near the core area of the flow field. The total
velocity, axial velocity, and the tangential velocity all showed the
same eross°seetional flow pattern. The velocities were ainifaura at the
core, increased with radius and then slightly decreased near the walls.
These patterns are similar to those found in the swirling flow generated
with fixed vanes (Baker and Sayre,1974).14
In order to observe the changes in the flow as a function of axial
distance from the inlet, the cross-sectional averages of the angle 0 and
the core area were calculated and presented in Figure 9. Both parameters
are indicators of tangential flow and show a very gradual decay as was
expected from the reported tests (Baker and Sayre,1974).14 The curves
have the same shape for both flow rates.
Plotted in Figure 10 is the location of the core area with respect
to the duct center. It can be seen that the swirling flow is indeed not
axisymmetric and the location of the core ,area changes with axial distance.
A similar flow pattern was found for both 7low rates.
EPA Method 5 Simulation Model
A model was developed and tested which describes particle collection
efficiency as a function of particle characteristics, angle of aisalignssent,
and velocity ratio. Together with the measurement of velocity components
1n a swirling flow field it was possible to analyze emission rate ©rrors
that would occur when performing a Method 5 analysis of the ©ffluent
stream following a cyclone.
For this simulation analysis, the volumetric flow rate and isokinetic
sampling velocities were calculated from velocity measurements obtained
at the eight diameter sampling location using an S-type pitot tube. The
angle $, velocity ratio, and particle velocity were determined from
velocity rasasureraents ®ade at the same location using the S-hole pitot
tube. The particle characteristics were obtained from partiel© size
distribution tests made by Mason(1974)ls on basically the same system.
From a particle distribution with a 3.0 micrometer W3D and geometric
standard deviation of 2.13, 10 particle diameters were selQeted which
each represent the mid-points of 10% of the mass of the aerosol. The
density of the particles was assumed to be 2.7 g/em3. Tfie nozzle diameter
was selected using standard criteria to be 0.635 oa (k inch). In the
model it was assumed that the nozzle would be aligned parallel with th@
axis of the stack, and therefore, ©=$. Using these parameters, the
average aspiration coefficients were determined at each traverse point
using the ten particle diameters. Since the sampling velocity would
determine the volume of air sampled at each traverse point, the total
aspiration coefficient for each flow rate was determined by taking an
average weighted according to sample velocity.
The total aspiration coefficients calculated in this manner for the
low and high flow rates wer© 0.937 and 0.906 respectively. There ar©
two reasons for the relatively low amounts of concentration error found
in this analysis. One reason is that the two mechanisms causing sampling
error, nozzle raisalignraent and anisokinetic sampling velocities, caused
1-15
-------
601-
50»
Arts
0 High f km ratt
50cm*
40cm* .
w
O
o
30cm*
6 8 O
Oiom»f«r« downstream
12
14
16
Figure 9. Decay of the angle 6 and core area along the axis of the duct.
6 8 10
Diameters downstream
12
14
16
Figure 10.
Location of the negative pressure region as a function of
distance downstream from the cyclone.
1-16
-------
errors in the opposite direction. The S°type pitot tub© detected a
velocity less than or ©qua! to the actual velocity which would lead to
subisokinetic sampling producing an increased concentration. The nozzle
erisalignment when sampling parallel to the stack wall would produce a
decreased concentration. So each of these errors huve a tendency of
reducing the other error.
Another reason for the small errors was the small size ef the
aerosol. The Stokes numbers for over 50% of the particles were less
than 0.2 and 0.3 for the low and high flow rates respectively. These
values lead to small sampling errors, even when isokinetic sampling
conditions are not maintained. .
In order to see how rauch greater the error would be for larger
particles, a similar analysis was performed using a distribution with a
10 micrometer mass mean diameter and a 2.3 geometric standard deviation.
This was the distribution obtained at the outlet of a cyclone in a
hot-mix asphalt plant (Danielson.1973).22 Because of the larger dimeter
particles, the sampling efficiency was reduced to 0.799 for the high
flow condition.
The volumetric flow rates determined from the S=type pitot tube
measurements were compared with the flow rates calculated fro® §°hole
pitot tube measurements. The axial flow rates using the §~hole pitot
tube data is calculated by muHiplying the average axial velocity by the
inner duct area minus the core area. The flow rates using the S°type
pitot tube data were determined using two different aethods varying in
how the negative velocity at port 4 was handled.
In the first aethod, the negative velocity was not used to deturain©
the average axial velocity. The volumetric flow rate was'calculated by
aultiplying the average axial velocity by 7/Sth of the inner cross-sQCtienal
area. In the second method, the negative value was used in the deter-
si nation of the average velocity and the entire inner duet was used to
determine the flow rate.
The errors for both sampling efficiency and flow rate d^terraination
are presented in Table II for the three simulated conditions. The
sampling errors and flow rate errors are in opposite directions so that
when the two values are combined to detersain© ©mission rate, the
iffeet is reduced.
Flow patterns found in a stack following the exit of a cyclone are
such a nature that Makes it extremely difficult to obtain a representative
sample with the present EPA r
-------
Table I. Five-hole pitot tube measurements made at 8 diameters
downstream of the cyclone.
Point
1
2
3
4
5
6
7
8
i
***
6J..O
70.3
•H-f
63.9
53.0
47.0
46.4
Total
Velocity
cm/sec
***
1414
1436
+++
1396
1326
1289
1231
Axial
Velocity
en/sec
***
685
484
•H-+
614
798
879
849
Tangential
Velocity
cm/sec
***
1212
1346
•M-f
1250
1019
818
758
*** Point No. 1 was too close to the wall to allow insertion of all five
pressure taps.
+++ Point lies inside the negative pressure section.
Table II. Results of the cyclone outlet simulation model for three
conditions.
Particle Size Flow Concentration, Flow Rate*, Flow Rate*, Eaission Rate3, Emission Rateb,
Distribution Condition Heasured/True Measured/True Measured/True Measured/True Measured/True
VtV 3 u«
og = 2. 13
M> 3 un
og = 2.13
MM) 10 \i»
og = 2. 3
Low
High
High
0.937
0.906
0.799
1.31
1.34
1.34
1.19
1.22
1.22
*
1.23
1.21
1.07
1.11
1.10
0.975
1 Negative velocity was not used in the calculation of average velocity.
Negative velocity was used in the calculation of average velocity.
1-18
-------
are useful tools in determining the velocity components in a tangential
flow field. The 5-hole pi tot tube has the advantage of giving pitch
information as well as yaw angle. However, in a cyclonic flow stream,
the yaw angle is of much greater magnitude than the pitch angle and
therefore, the pitch angle can be ignored with small error. In the
situation modeled, if pitch angle were ignored, the calculated flow rate
would be in error by less than 6%.
RECOMMENDATIONS
EPA recommends that if the average angle of flow relative to the
axis of the stack is greater than 10 degrees, then EPA Method 5 should
not be performed. Since the maximum error in particle sampling has been
found to be Jl-Rcos0|, the 10 degrees requirement is unduly restrictive
and a 20 degrees limitation would be more appropriate. For a 20 degrees
angle the velocity measured by the S-type pitot tube would be approxi-
mately the same as the true velocity (i.e., R=l). Therefore, the maxi-
mum error would be l-cos20° or 6% for a very large aerosol.
When cyclonic flow does exist in a stack, EPA recommends either
straightening the flow or moving to another location. Because of the
physical limitations of these suggestions a better approach would be to
modify Method 5 so that it could be used in a tangential flow stream.
By replacing the S-type pitot tube with a 3-hole pitot tube, the direc-
tion of the flow could be accurately determined by aligning the nozzle
and the velocity components could be measured for a correct calculation
of volumetric flow rate. The sampling rate would be calculated on a
basis of the total velocity of the flow. However, the volumetric flow
rate through the stack would be calculated on a basis of only the axial
component of velocity (i.e. V = \Lcos8). In addition to the 3-hole
pitot tube, the modification \oula have to include a protractor to
measure the flow angle, and a method of rotating the probe without
rotating the entire impinger box.
ACKNOWLEDGEMENT
This research was partially supported by a grant (Grant No.
R802692-01) from the Environmental Protection Agency (EPA), and was
monitored by EPA's Project Officer Kenneth T. Knapp.
1-19
-------
REFERENCES
1. "Test Methods and Procedures. Method 5 - Determination of Particu-
late Emissions from Stationary Sources." Federal Regulations, 40
CFR 60.85.
2. Revision to Reference Method 1.8. Federal Register, Volume 42,
Number 160, Thursday, August 18 (1977).
3. J. D. Wilcox, "Isokinetic Flow and Sampling of Airborne Particu-
lates." Artificial Stimulation of Rain, p. 177 (1957).
4. H. H. Watson, "Erros Due to Anisokinetic Sampling of Aerosols."
Am. Ind. Hyg. Assoc. Quart.. 15;1 (1954).
5. S. Badzioch, "Collection of Gas-Borne Dust Particles by Means of an
Aspirated Sampling Nozzle." Br. J. Appl. Phys.. 10:26 (1959).
6. C. N. Davies, "The Entry of Aerosols into Sampling Tubes and
Heads." Br. J. Appl Phys.. Ser. 2, 1:921 (1968).
7. D. A. Lundgren and S. Calvert, "Aerosol Sampling with a Side Port
Probe." Am. Ind. Hyg. Assoc. J.. 28(3):208 (1967).
8. S. P. Belyaev and L. M. Levin, "Investigation of Aerosol Aspiration
by Photographing Particle Tracks Under Flash Illumination."
Aerosol Science. 3:127 (1972).
9. S. P. Belyaev and L. M. Levin, "Techniques for Collection of Repre-
sentative Aerosol Samples." Aerosol Sc1.. 5:325 (1974).
10. R. Dennis, W. R. Samples, D. M. Anderson and L. Silver-man, "Iso-
kinetic Sampling Probes." Ind. Eng. Chem.. 49:294 (1957).
11. A. B. Whiteley and L. E. Reed, "The Effect of Probe Shape on the
Accuracy of Sampling Flue Gases for Dust Content." J. Inst. Fuel.
32:316 (1959).
12. H. Glauberman, "The Directional Dependence of Air Samplers."
Am. Ind. Hyg. Assoc. J.. 23:235 (1962).
13. 0. A. Lundgren, M. D. Durham and K. W. Mason, "Sampling of Tan-
gential Flow Streams." Am. Ind. Hyg. Assoc.'J.. 39:640 (1978).
14. D. W. Baker and C. L. Sayre, "Decay of Swirling Turbulent Flow of
Incompressible Fluids 1n Long Pipes." Flow: Its Measurement and
Control in Science and Industryt Volume 1. Part 1, Flow
Characteristics. Instrument Society of America (1974).
15. N. A. Chigier, "Velocity Measurement in Vortex Flows." Flow; Its
Measurement and Control in Science and Industry. Volume 1. Part 1.
Flow Characteristics. Instrument Society of America (1974).
1-20
-------
16. K. W. Mason, Location of the Sampling Nozzle In Tangential Flow. H.
S. Thesis, University of Florida, Gainesville, Florida (1974).
17. H. A. Hanson and D. P. Saari, "Effective Sampling Techniques for
Participate Emissions from Atypical Stationary Sources." EPA-600/
2-77-036, U.S. Environmental Protection Agency, Research Triangle
Park, N.C. (1977).
18. E. F. Brooks and R. L. Williams, "Process Stream Volumetric Flow
Measurement and Gas Sample Extraction Methodology." TRW Document
No. 24916-6028-RU-00, TRW Systems Groub, Redondo Beach, California
(1975).
19. D. J. Grove and W. S. Smith, "Pilot Tube Errors Due to Misalignment
and Nonstreamlined Flow." Stack Sampling News. November (1973).
20. H. A. Hanson, R. J. Davini, J. K. Morgan and A. A. Iversen,
"Particulate Sampling Strategies for Large Power Plants Including
Nonuniform Flow." EPA-600/2-76-170, U.S. Environmental Protection
Agency, Research Triangle Park, N.C. (1976).
21. F. C. Williams and F. R. DeJarnette, "A Study on the Accuracy of
Type S Pitot Tube." EPA 600/4-77-030, U.S. Environmental Protec-
tion Agency, Research Triangle Park, N.C. (1977).
22. J. A. Daniel son, "Air Pollution Engineering Manual." Environmental
Protection Agency, OAQPS.AP40, Research Triangle Park, N.C. (1973).
1-21
-------
PAPER II
SAMPLING OF TANGENTIAL FLOW STREAMS
Dale A. Lundgren
Michael D. Durham
Kerry Wade Mason
-------
Reprinted from American Industrial Hygiene Association
Journal, August 1978. Study funded by Environmental
Science and Research Laboratory, Environmental Protection
Agency, Research Triangle Park, N.C. 27711.
The causes and characteristics of tangential flow in industrial stacks are described. Errors
induced by tangential flow in the determination of volumetric flow rate and particulate
concentration are analyzed. Experiments were conducted at the outlet of a cyclone
collector in order to investigate the effect of tangential flow on the determination of
emission rates. Straightening vanes were found to be useful in the reduction of error in
flow rate measurements.
Sampling of tangential flow streams
DALE A. LUNDGREN'. MICHAEL D. DURHAM', and KERRY WADE MASON'
'Dept. of Environmental Engineering Sciences. University of Florida, Gainesville.
Florida 32611; 2South Carolina Dept. of Health and Environmental Control, Columbia,
South Carolina 29201
sources of tangential flow
Tangential flow is the nonrandom flow in a
direction other than that parallel to the duct
center line direction. It is often encountered in
industrial stacks and provides a difficult
situation for obtaining a representative
particulate sample and for accurate determina-
tion of flow rate. In an air pollution control
device, whenever centrifugal force is used as the
primary particle collecting mechanism,
tangential flow will occur. Gas flowing from the
outlet of a cyclone is a classic example of
tangential flow and a well recognized problem
area for accurate particulate sampling.
Tangential flow can also be caused by flow
changes induced by ducting. If the duct work
introduces the gas stream into the stack .
tangentially, a helical flow will occur. Even if the
flow stream enters the center of the stack, if the
horizontal velocity is high compared to the
upward gas velocity, a double vortex flow
pattern will occur, in all of these cases, in a
cylindrical stack the flow will be characterized
by one or two primary vortices spiraling up the
stack. Since any other eddies produced in the
stack will be of a much smaller magnitude, there
will be very little interference and dissipation of
the primary vortex and thus, the spiraling flow
can be maintained the entire length of the stack.
Therefore, satisfying the requirement of
sampling 8 stack diameters downstream from a
flow disturbance will not eliminate the
tangential flow sampling problem.
errors caused by tangential flow
Types of errors that are introduced by tangential
flow are particle concentration gradients, nozzle
misalignment, and invalid flow rate and
concentration measurements. Concentration
gradients occur because the rotational flow in
the stack acts somewhat as a cyclone. The
centrifugal force causes the larger particles to
move toward the walls of the stack, causing
higher concentrations in the outer regions.
The bias due to misalignment of the probe is
similar to that caused by superisokinetic
sampling. In this situation the nozzle velocity is
greater than the flow stream velocity and
therefore the sampled area will be greater than
the nozzle area. As the flow stream converges
into the nozzle some of the larger particles,
because of their inertia, will be unable to make
the turn and will not be collected. Therefore, the
particle concentration in the gas that is collected
will be less than the actual concentration. When
the nozzle is at an angle to the flow stream, the
projected area of the nozzle is reduced by a
factor equal to the cosine of the angle between
the flow direction and the nozzle axis. Even
though the nozzle velocity is equal to the flow
stream velocity, a reduced concentration will be
obtained because some of the larger particles will
be unable to make the turn in the nozzle.
Therefore, whenever the nozzle is misaligned,
the large particle concentration will always be
less than when the nozzle is aligned.
Previous investigations relating to
anisokinetic sampling have primarily dealt with
the bias induced when free stream velocity was
not equal to the suction velocity. This bias. A, is
defined as:
A = C,/Co f)
where Ci = volumetric particulate concentra-
640
II-3
Urn. Ind- Hyg. Assoc J (39)
August. 1978
-------
tion in the nozzle, and
Co = volumetric concentration in the gas
stream.
It has been determined from experimentation
that the anisokinetic sampling bias, A, is a
function of two parameters: the inertial
impaction parameter, K, and the velocity ratio,
R. The parameters K and R can be defined by:
(2)
where C = Cunningham correction factor,
Pf = particle density,
V0 = duct velocity,
Dp = particle diameter,
ft = gas viscosity, and
Di = nozzle diameter.
R = V./V, (3)
where Vi = probe inlet velocity.
The sampling bias is related to K and R by the
following expressions;"'11
A = I + (R-l) #K) (4)
where )3(K) is a function of both K and R.<2>
Sampling error associated with the nozzle
misalignment due to tangential flow has not
been adequately evaluated in past studies
because the sampled flow field was maintained
or assumed constant in velocity and parallel to
the duct axis. The studies that have been
performed on the effect of probe misalignment
do not provide enough quantitative information
to understand more than just the basic nature of
the problem. Results were produced through
investigations on the effect of the nozzle
misalignment on the collection efficiency of 4,
12, and 37 pm particles.13' In a study on the
directional dependence of air samplers,'4' it was
found that a sampler head facing into the
directional air stream collected the highest
concentration. Although these results coincide
with theoretical predictions (i.e., measured
concentration is less than or equal to actual
concentration and the concentration ratios
decrease as the particle size and the angle are
increased), the data are of little use since two
important parameters, free stream velocity and
nozzle diameter, are not included in the analysis.
Particles of 0.68,6.0 and 20 /zm diameter were
sampled'5' at wind speeds of 100, 200,400, and
700 cm/sec with the nozzle aligned over a range
of angles from 60 to 120 degrees. A
trigonometric function was then used to convert
equation (4) to the form:
See equation 5 below
This function only serves to invert the velocity
ratio between 0 and 90 degrees and does not
realistically represent the physical properties of
the flow stream. In fact, equation (5) becomes
unity at 45 degrees regardless of what the
velocity ratio or particle size is. This cannot be
true since it has been shown that the
concentration ratio will be less than or equal to
unity, and will decrease inversely with angle and
particle diameter.
A more representative function can be
determined in the following manner: Consider
the sampling velocity Vi to be greater than the
stack velocity V<>. Let a, be the cross sectional
area of the nozzle of diameter Di. The stream
tube approaching the nozzle will have a cross
sectional area ao such that:
a»V0 = ajVi {6)
If the nozzle is at an angle 6 to the flow stream,
the projected area perpendicular to the flow is an
ellipse with a major axis Di, minor axis Dicosd,
and area (Di2 7rcos0)/4. The projected area of the
nozzle would therefore be a,cos0. It can be seen
that all the particles contained in the volume
V0aicos0 will enter the nozzle. A fraction 0'(K) of
the particles in the volume (a« — 3iCos0)V0 will
leave the stream tube because of their inertia and
will not enter the nozzle. Therefore, with Co, the
actual concentration of the particles, the
measured concentration in the nozzle would be:
c __ Coa,co59V.[+ l-/3'(K)J(ao-a,co5fl)VeC.
Using equations (6) and (3), this may be
simplified to:
A=
oCos0)/(ViCos0 + Vosinfl)- I] (5)
American Industrial Hygiene Association JOURNAL (39) 8/78
641
II-4
-------
A = C/Co = 1 + /T(K) (Rcos0-l)
(8)
/3'(K) would be a function of the velocity ratio R
and the inertial impaction parameter K,'2' and
would also be a function of the angle 6 because as
the angle increases, the severity of the turn that
the particles must make to be collected is also
increased. For small angles the sampling
efficiency will be of the form:'6'
A = 1 - 4sin(7rK/fl) (9)
Errors in the measurement of tangential flow
velocity and subsequent calculations of flow rate
are due primarily to the crudeness of the
instruments used in source sampling. Because of
the high particulate loading that exists in source
sampling, standard pitot tubes cannot normally
be used to measure velocity. Instead, the S-type
pitot tube is used because it has large diameter
pressure ports that do not easily plug. This type
of pitot tube can give an accurate velocity
measurement, but is quite insensitive to flow
direction. It can be misaligned up to about 45
degrees in either direction of the flow and still
read approximately the same velocity head. This
means that the S-type pitot tube cannot be used
in a tangential flow situation to accurately
measure the velocity in a particular direction.
The velocity in a rotational flow field can be
broken up into an axial and radial component.
The magnitude of the radial component relative
to the axial component will determine the degree
of error induced by the tangential flow. The
radial velocity component does not affect the
stack gas flow rate but does affect the measured
velocity because the S-type pitot tube lacks
directional sensitivity. If the maximum velocity
head were used to calculate the stack velocity,
the resultant calculated flow rates would be off
by a factor of l/cos0. Aligning the probe parallel
to the stack centerline will reduce but not
eliminate this error because a large part of the
radial velocity component will still be detected.
Therefore, the actual stack gas flow rate cannot
normally be determined by an S-type pitot tube
in tangential flow because neither the radial
velocity Vr nor the axial velocity V§ can be
measured directly. Also, Vr increases in
magnitude as the probe is moved from the stack
center to the walls. Complicating analysis of the
subject is the fact that tangential flow is
sometimes accompanied by a reverse flow at the
642
stack center. One method to greatly reduce the
error in velocity measurement and flow rate
calculation is the use of in-stack flow
straighteners upstream from the sampling port.
These can eliminate the radial component of
velocity and allow a true flow rate to be
determined.
emission rate measurements obtained
from sampling the outlet of a cyclone
The outlet of a small industrial cyclone was
tested to determine the errors that arise from
sampling tangential flow.17' The previous
discussion suggests that sources of error induced
by tangential flow are: concentration gradients
across the stack, sampling bias due to
misalignment of the probe, and inaccurate
measurements of flow through the stack.
Experiments were also run to determine what
effect an in-stack flow straightener would have
on the measurement error.
experimental procedure
The major components of the experimental set
up included a dust feeder, fan, cyclone collector,
sampling equipment, and two stack extensions.
The test dust was a crushed gypsum rock with a
40 jum mass median diameter (MMD). A
standard design, high efficiency cyclone
collector with a body diameter of 45 cm was used
as the collector. Dust leaving the cyclone had a
MMD of 2.7 /xm. Two sampling trains were used
in the experiments: an Andersen cascade
impactor was used for particle size distribution,
and an EPA Method 5 train was used for particle
concentration and emission rate. The stack
extensions included a straight vertical stack
placed on the outlet of the cyclone; a second
extension turned the flow 90 degrees into a
horizontal duct section which contained a one-
foot long, cross type straightening vane.
Four types of tests were performed to
determine the errors involved in sampling
tangential flow: 1) velocity traverses at various
locations; 2) concentration measurements at
various probe angles; 3) emission rate
measurements at different locations; and 4)
particle size distribution measurements across
the dust traverse. Velocity traverses used for
determining volumetric flow rates were obtained
using an S-type pitot tube positioned parallel to
Am. Ind. Hyg. Assoc. J (39)
August. 1978
II-5
-------
the stack wall for one traverse and then rotated
to the point of maximum velocity head for
another traverse. A third velocity traverse was
performed in the section of duct following the
straightening vanes. To determine the effect of
sampling at various angles, four apparent
isokinetic samples were taken at 0,30,60, and 90
degrees with respect to the axis of the duct.
Emission rates were determined by sampling
downstream of the flow straighteners and
upstream in the straight stack extension. In the
straight stack extension, measurements were
made with the sampling nozzle aligned parallel
to the stack wall and also with the nozzle rotated
to the angle of maximum velocity head. Probes
were washed with acetone so wall losses were
included as collected particulate matter. Actual
emission rates were determined by subtracting
the collected dust in the cyclone from the dust
feed rate.
results
To determine the emission rate from a source, it
is necessary to determine the flow rate. Results
of flow rates determined at different locations of
the cyclone discharge indicated serious errors
can result in cases of tangential flow. A
maximum error of 212% (three times actual
flow) occurred when the pilot tube was rotated
to read a maximum velocity head. Sampling
parallel to the stack wall produced a flow rate
determination error of 74%. When sampling
downstream of the flow straightening vanes, the
flow rate error was reduced to 15%.
Dust concentration measurements were made
after the cyclone at a fixed sampling point, but at
different nozzle angles. Measured dust
concentration was lowest when the sampling
nozzle was located at an angle of 0 degrees or
parallel to the stack wall. The measured dust
concentration continued to increase at 30 and 60
degrees but then decreased at 90 degrees.
Equation (8) suggests that when sampling at an
angle under apparent isokinetic conditions (i.e.,
R = 1), the measured concentration will be less
than the true concentration by a factor inversely
proportional to cosO. A maximum concentra-
tion, which would be the true concentration, will
occur when the sampling nozzle axis is parallel
with the actual gas flow direction, which from
these data should lie at an angle between 60 and
Amencan Industrial Hygiene Association JOURNAL (39) S/78
90 degrees to the axis of the stack. This angle was
calculated from flow velocity data to be 72
degrees.
Dust concentration measurements obtained
by sampling with the nozzle parallel to the stack
wall produced an indicated concentration of
47% of the actual concentration. An even larger
error was expected because this position
represented approximately a 70 degree
misalignment with the actual flow. However, the
velocity determined by the S-type pitot tube,
which was used for the nozzle inlet or sampling
velocity, was much less than the actual approach
velocity and tended to counteract the effect of
the off-angle sampling. Sampling with the nozzle
aligned to the angle of maximum velocity head
reduced the error. Samples obtained from
sampling after the straightened flow produced
an indicated concentration of 64% of actual. It
was expected that sampling following
straightening vanes would produce accurate
results, but a significant amount of particulate
matter was impacted onto the straightening
vanes and was deposited in the horizontal
section of the duct.
Particle size distribution measurements made
at several points across the duct traverse show
particles greater than 6 Mm to be present in
higher concentration near the duct wall.
conclusions
To obtain a representative sample of particulate
matter in a tangential flow field, the nozzle
should not be aligned parallel to the stack wall,
but turned toward the direction of the maximum
velocity head. The sampling rate should be based
upon the maximum velocity. This will align the
probe with the correct direction and velocity of
the flow, and will produce a more representative
sample. Although an unbiased particulate
concentration may be determined by this
method, the overall emission rate will be
incorrect because an S-type pitot tube cannot
accurately determine the volumetric flow rate
through the stack when a tangential flow exists.
A velocity measuring device is required that is
sensitive to both direction and velocity of the
flow and is able to operate properly in the high
particulate environment of an industrial stack.
The use of in-stack flow straighteners is one
solution to the problem of tangential flow. This
643
II-6
-------
solution may not be feasible in large stacks
because of installation problems, cost, and
increased pressure drop created by the flow
straighteners.
acknowledgement
This research was funded in part by an
Environmental Protection Agency Research
Grant No. R803692.
references
1. Badzioch. S.: Collection of Gas-Borne Dust Particles
by Means of an Aspirated Sampling Nozzle. Br. J.
App. Phys. 70:26(1959).
2. Belyaev. S. P. and L. M. Levin: Techniques for
Collection of Representative Aerosol Samples. J.
Aerosol Sci. 5:325 (1974).
3. Watson. H. H.: Errors Due to Anioskinetic Sampling
of Aerosols. Am. Ind. Hyg. Assoc. J. 75:1 (1954).
4. Glauberman, H.: The Directional Dependence of Air
Samplers. Am. Ind. Hyg. Assoc. J. 75:1 (1954).
5. Raynor. G. S.: Variation in Entrance Efficiency of a
Filter Sampler with Air Speed. Flow Rate. Angle, and
Particle Size. Am. Ind. Hyg. Assoc. J. 37:294(1970)
6. Fuchs. N. A.: Sampling of Aerosols. Atmos. Envir.
3:697(1975).
7. Mason. K. W.: Location of the Sampling Nozzle in
Tangential Flow. M.S. Thesis, University of Florida.
Gainesville. Florida (1974).
Accepted February 15. 1978
644
Am. Ind. Hyg. Assoc. J (39)
August. 1978
II-7
-------
PAPER III
CHARACTERIZATION OF CYCLONIC FLOW
AND ANALYSIS OF PARTICULATE SAMPLING APPROACHES
AT ASPHALT PLANTS
James W. Peeler
Frank J. Phoenix
D. James Grove
-------
CHARACTERIZATION OF CYCLONIC FLOW
AND ANALYSIS OF PARTICULATE SAMPLING APPROACHES
AT ASPHALT PLANTS
James W. Peeler
M.S., Mechanical Engineering
Frank J. Phoenix
B.S., Engineering Science & Mechanics
D. James Grove, P.E.
M.E., Chemical Engineering
Entropy Environmentalists, Inc.
P. O. Box 12291
Research Triangle Park, N. C.
27709
III-3
INTROPY
^•••i
iNVIRONMKNTAUaTB, INC
-------
CHARACTERIZATION OF CYCLONIC FLOW AND ANALYSIS OF PARTICULATE SAMPLING APPROACHES
AT ASPHALT PLANTS, J. W. Peeler, F. J. Phoenix, and D. J. Grove, Entropy Environ-
mentalists, Inc., P. 0. Box 12291, Research Triangle Park, N. C. 27709.
The information presented in this paper is directed to those individuals re-
sponbible for conducting particulate emission tests at asphalt plants where cyc-
lonic flow conditions occur in the stack. Characterization of cyclonic flow has
been' accomplished tvrough axial and tangential velocity measurements in several
stacks following inertial demisters. The axial and tangential velocities reach
maximums at 82% and 43% of the stack radius, respectively. The angle between the
flow direction and the stack axis varies from 41° to 66° across the stack. A
theoretical formulation of particle behavior in cyclonic flow, incorporating the
results of flow measurements, and a numerical solution to the resulting equations
is employed to determine radial particle velocities. Results indicate that 25
micron diameter and larger particles will reach the stack wall before the samp-
ling site. The radial velocity of 10 micron particles is small compared to axial
and tangential velocities and these particles will be encountered at sampling
sites located less than 8 diameters downstream of the starting point. The appli-
cation of three sampling methods to cyclonic flow are discussed. The Blind Man's
Approach should not be used. The Compensation Approach and Alignment Approach
satisfy the requirements of proportional sampling, isokinetic sampling, and ac-
curate volumetric flow rate measurements.
III-4
-------
Introduction
Cyclonic flow conditions often occur following inertial demisters in
stacks at asphalt plants equipped with wet scrubber control systems. Where
cyclonic flow conditions exist, the effluent gas flows in a spiral motion up
the stack. At each point in the stack, the gases have both axial and tangen-
tial velocity components.
Measurements of volumetric flow and mass concentration are considerably
more difficult under cyclonic flow conditions than under fully developed flow
conditions. EPA Reference Methods 2 and 5 are not applicable to sampling cy-
clonic flow systems, since the average angle of the free stream velocity is
inclined at greater than 10° to the stack axis. Thus, either an alternative
sampling location, modifications to the source to eliminate the cyclonic flow,
or alternate sampling methodology is necessitated.
This paper presents through empirical measurement a characterization of
cyclonic flow systems typically encountered at asphalt plants. Particle velo-
cities in a cyclonic flow system are determined through theoretical considera-
tions. A description of several particulate sampling approaches for cyclonic
flow systems and a discussion of some of the errors associated with these
methods are provided.
Characterization of Cyclonic Flow
Flow measurements were taken following inertial demisters in stacks at
three asphalt plants considered to be representative of the industry. These
measurements were made at sampling points as specified by EPA Reference Method
1, with a United Sensor Corporation, three dimensional, type DA probe. Two
24-point traverses were conducted; however, no useful data were obtained at
points 1, 2, 23, and 24, because there was an excessive amount of particulate
and water droplets near the wall. The measurements made on the near side of
the stack diameter were more stable and are considered more reliable than the
measurements obtained on the far side of the stack diameter. Axial and tan-
gential velocity components were determined as a function of stack radius, and
are shown in Figure 1. Pitch angle measurements, corresponding to radial velo-
city components, were found to vary randomly across the stack, and were highly
unstable with time. Since the pitch angles were small in comparison with the
yaw angles, the cyclonic flow system is considered to have no radial velocity
components.
Characterization of Particulate Behavior in a Cyclonic Flow System
The behavior of particles in cyclonic flow is of fundamental importance
in developing particulate sampling methodology. Large particles are of primary
concern here, since these particles present the greatest difficulty in obtain-
ing representative measurements due to the effects of the inertial forces im-
posed on these particles by the cyclonic flow. Small particles will behave
much like gas molecules and are easily and accurately sampled.
The position of a particle in cylindrical co-ordinates at time t is given
by R, 0, Z, and the radial velocity component (ur) and tangential velocity com-
ponent (ut) of the particle are:
ur = f* (1) ut = R dO (2)
dt dt
III-5
-------
50
10
V.
^-
u.
30
20
o
o
.J
UJ
10
UT (TANGENTIAL VELOCITY)
U (AXIAL VELOCITY)
50
100
r x 100
Figure 1. Axial and tangential fluid velocities along stack
radius for cyclonic flow.
III-6
-------
The radial and tangential accelerations of the particle are defined by the
material derivative of the velocities,
(3)
Multiplying the component accelerations by the particle mass provides the iner-
tial forces acting on the particle. These forces are resisted by viscous forc-
es due to the relative motion of the particle moving through the surrounding
fluid. The viscous forces can be described by Stokes law in terms of the vis-
cosity of the fluid (y) , particle diameter (d) , and relative velocity between
the particle and the fluid (urei) :
F = 3irydurel
(4)
Thus, following the development of Strauss2, force balances in the radial and
tangential directions are
3iryd
d2R
dt2
- R
dj3
dt
m
3iryd
= -ur =-
r dt
= 0
=£ (5)
(6)
where it is assumed that the tangential particle velocity is equal to the tan-
gential fluid velocity. Since large particles following an inertial demister
will be essentially particulate laden water droplets, the particles may be as-
sumed to be spherical, and the particulate mass may be expressed in terms of
the density (p) and the particle diameter. Employing this simplification, and
expressing the above equations in dimensionless form, yields,
d
- r
dt
and
dT<
2 ££.
dT
d<3
dT
(7)
(8)
where:
= d2puto
T = ~
r =
u = ur/uto
T = tuto/Ro
= 'tangential velocity at r
1 (R = R0)
III-7
-------
Equation 8 can be rewritten as
which requires
ufcr = constant (10)
The above equation is the free vortex equation. It can be> seen from the tan-
gential velocity measurements that the tangential velocity of the fluid is ac-
curately described by the free vortex relation over the range 0.7 i r i 1.0.
Thus , the above formulation is an accurate description of the particle behavior
over this range. The previous equations cannot be applied for values of r less
than 70% of the stack radius; however, this is not a serious constraint since
the larger particles will be in the outer portion of the stack if the demister
is functioning at all.
Substituting equation 10 in equation 7 leads to Strauss ' result^ :
I dr . r-3 = o (11)
dT2 T dT •
The above equation is non-linear due to the r~^ term. For our study, this
equation was solved numerically by employing the Taylor Method and retaining
fifth order terms. The Taylor Method was employed because it provides the
particle's position, radial velocity, and radial acceleration directly as func-
tions of time. Solutions were obtained for several values of T corresponding
to particle diameters of 10, 25, 50, and 100 microns. For each case, it was
assumed that the particle was initially located at r = .7 and had no initial
radial velocity. These initial conditions represent worst case conditions in
terms of the time required for the particles to reach the stack wall. The
radial velocities of several particles as functions of the stack radius are
presented in Figure 2.
From the results illustrated in Figure' 2 , it can be seen that the radial
velocity component is small compared to the axial and tangential velocity com-
ponents for particles with diameters equal to or less than 10 microns. Thus,
particles less than 10 microns in diameter essentially move with the cyclonic
gas stream. Also, by considering both the radial and axial particle velocities,
for particles starting at r = .7 and with no initial radial drift, it can be
shown that one stack diameter downstream of the starting point, all of the 100
and 50 micron particles will have reached the stack wall and .that the 25 micron
particles will have traveled outward to r = .94. In contrast, the 10 micron
particles will have drifted outward from ro = .7 to only r = .7455. The 10
micron particles will have reached r = .84 four stack diameters downstream from
the starting point and will not have reached the wall even after traveling ax-
ially for more than eight stack diameters. Thus, considering the typical length
of stacks on inertial demisters in the asphalt industry, particles at least as
large as 10 microns will be encountered at the sampling location.
The previous analysis assumes that the particles always have the same ax-
ial and tangential velocity components as the gas stream. These assumptions
are approximately true for r greater than .7, since only in this region can the
tangential velocity of the effluent stream be accurately described by the free
vortex equation, and since only in this region is the axial velocity fairly con-
stant. To determine the particle behavior for large particles initially locat-
ed within the inner portion of the stack, it is necessary to consider relative
III-8
-------
70
100
r x 100 (% RADIUS)
Figure 2. Radial particle velocities in cyclonic flow for particles
starting at 70% of stack radius.
-------
velocities between the particles and the fluid in both the tangential and axial
directions. Such a formulation leads to a set of three simultaneous, non-linear,
partial differential equations which describe the particle behavior. Solutions
to these equations have not yet been obtained.
Sampling Methods for Cyclonic Flow Systems
Stack sampling is usually conducted at asphalt plants to determine the
particulate concentration of the effluent, and, in some cases, the volumetric
flow rate, in order to compute mass emission rates. Several papers have been
written which discuss various methods for sampling cyclonic flow systems.1/3
Three general methods have been discussed: (1) Blind Man's Approach, (2) Com-
pensation Approach, and (3) Alignment Approach. These methods are outlined
below and some of the factors influencing these methods are discussed.
Blind Man's Approach
In the Blind Man's Approach, standard sampling methodology i& applied and
the cyclonic flow is simply ignored. The sampling results obtained through
this procedure are subject to multiple biasing effects. This approach is not
recommended; it is only discussed here because of the frequency with which it
is actually employed.
In the Blind Man's Approach, the sampling rate is based on the measured
velocity which is neither the total velocity nor the axial velocity. For the
axial and tangential velocities presented in Figure 1, the angle between the
flow direction and the stack axis ($) varies from a maximum of 66° at the in-
nermost sampling point to a minimum of 41° at the outermost point. According
to the data presented by Smith and Grove4, the total velocity will be underes-
timated by about 20% at the innermost point and overestimated by about 2% at
the outermost point. If the total velocity was accurately measured, the sam-
pling rate would be overisokinetic, since the nozzle is not aligned with the
flow and the effective area of the nozzle opening is reduced by cos $. The
combined biases resulting from velocity measurement errors and from the reduc-
tion in effective nozzle area will lead to overisokinetic sampling, where the
sampling rate is too great by approximately 30% at the outermost point and by
more than 100% at the innermost point. A low bias in the mass concentration
will result from overisokinetic sampling.
To satisfy the constraints of proportional sampling, the sampling rate
must be directly proportional to the axial velocity in the stack. In the
Blind Man's Approach, the sampling rate is proportional to the indicated velo-
city; therefore, the requirements of proportional sampling are not satisfied.
Since both the particulate concentration and axial velocity increase with in-
creasing distance from the center of the stack, a low bias is expected in the
measured mass concentration.
Compensation Approach
The Compensation Approach requires determination of the magnitude and
direction of flow at each sampling point. This may be accomplished through
the use of a three dimensional pitot sensor or similar device, or by rotating
an S-type pitot to obtain a null reading perpendicular to the flow direction,
and then rotating the pitot 90° to measure the flow velocity. In the Compen-
.sation Approach, the sampling nozzle is aligned with the stack axis. Since
the nozzle is not aligned with the flow direction, the effective nozzle area
111-10
-------
is reduced by cos $ where * is the angle between the flow direction and the
stack axis. Due to the reduction in the nozzle effective area, the sampling
rate must be reduced by cos to maintain isokinetic sampling conditions. If
isokinetic sampling conditions are maintained, the Compensation Approach satis-
fies the contraints of proportioned sampling, since the sampling rate is direct-
ly proportional to the axial velocity at each sampling point.
The Compensation Approach is subject to biases when the angle between the
nozzle and flow stream becomes sufficiently large. For large angles, the flow
around the nozzle will create aerodynamic interferences with isokinetic sam-
pling. In general, the degree of the bias increases as the velocity increases
and as the angle of misalignment increases. These effects are currently un-
quantified; however, these interferences will bias the measured mass concentra-
tion low. Also, for large angles between the flow direction and nozzle orienta-
tion, particulate agglomeration on both the inside and outside walls of the sam-
pling nozzle should be expected. Care must be exercised in moving the sampling
train to prevent accidental loss of material accumulated inside the nozzle, and
in sample recovery to prevent material on the outside of the nozzle from being
included in the particulate catch.
Alignment Approach .
The alignment approach involves determination of the flow direction at
each sampling point by means of a three dimensional pitot sensor, or by obtain-
ing a null reading normal to the flow direction with an S-type pitot. The sam-
pling nozzle and pitot are then aligned with the flow direction at each sampling
point.
In the alignment method, the sampling rate must be based on the total velo-
city at each sampling point in order to maintain isokinetic sampling conditions.
Since the angle between the flow direction and stack axis varies across the
stack, the sampling velocity is not weighted proportionally to the axial velo-
city component. Proportional sampling requirements can be satisfied by adjust-
ing the sampling time for each sampling point such that the volume of sample
collected at each point is related by a constant to the axial velocity compon-
ent at each point. Thus,
t2 = t! cos $ (12)
where
ti = nominal sampling time per point
t2 = actual sampling time per point
<1> = angle between flow direction and stack
axis
Care should be exercised in selecting the nominal sampling time per point to
ensure collection of sufficient sample volume to provide accurate mass concen-
tration measurements since the application of the above weighting procedure
will reduce the actual sampling time.
III-ll
-------
10
When sampling to determine a mass emission rate, the volumetric flow rate
should be determined as:
N
I (*
-------
11
References
1. F. J. Phoenix and D. J. Grove, "Cyclonic flow - characterization and recom-
mended sampling approaches," paper prepared by Entropy Environmentalists,
Inc. for the U. S. Environmental Protection Agency, EPA Contract # 68-01-
,4148, November 1977 (Draft), 14 p.
2. W. Strauss, Industrial Gas Cleaning, Pergammon Press, London, 1966,
pp. 122-125, 160-165.
3. J. W. Peeler, "Isokinetic particulate sampling in non-parallel flow sys-
tems - cyclonic flow," paper prepared by Entropy Environmentalists, Inc.
for the U. S. Environmental Protection Agency, EPA Contract # 68-01-4148,
1977 (Draft), 27 p.
4. D. J. Grove and W. S. Smith, "Pitot tube errors due to misalignment and
non-streamlined flow," Stack Sampling News, 1(5): 11 (1973).
111-13
-------
12
Acknowledgment
The work which has resulted in this paper was in part funded by the United
States Environmental Protection Agency, Division of Stationary Source Enforcement,
under Contract No. 68-01-4148, Tasks #20 and 37.
111-14
-------
PAPER IV
A METHOD FOR STACK SAMPLING CYCLONIC FLOW
Charles L. Goerner
Fred H. Hartmann
James B. Draper
-------
A METHOD FOR STACK SAMPLING CYCLONIC FLOW
Charles L. Goerner
Fred H. Hartmann
James B. Draper
Texas Air Control Board
Austin, Texas
Charles L. Goerner, B.S., M.S., P.E,
Engineer
Fred H. Hartmann, B.S., P.E.
Engineer
James B. Draper, B.S., P.E.
Engineer
Texas Air Control Board
8520 Shoal Creek Boulevard
Austin, Texas 78758
IV-3
-------
78-35.2
Abstract
This paper presents a method for particulate sampling in
stacks with cyclonic flow. Specific procedures and quantita-
tive adjustments to sampling parameters are described. Sam-
pling is performed isokinetically with the nozzle and pitot
tubes aligned parallel to the direction of flow and with sam-
pling time at each point weighted by the cosine of the flow
angle at that point. The.method is specifically applicable
to particles with tangential velocity components without con-
sideration of radial velocity components. Comments are made
concerning the behavior of particles with radial velocity com-
ponents as applicable to the accuracy of this method.
IV-4
-------
78-35.2 1
Introduction
Accurate sampling results cannot be obtained with conventional
sampling procedures from stacks with severe cyclonic flow,
i.e. flow with tangential velocity components. Cyclonic flow
may exist after cyclones, tangential stack inlets, or other
configurations that tend to induce swirling.
Several papers have been written describing and evaluating
various procedures for sampling cyclonic flow. This paper
presents a method that is currently being used by the Texas
Air Control Board staff. One inherent characteristic of this
method is that adjustments to the nozzle and pitot tube posi-
tion are made for tangential velocity components (yaw) but no
adjustments for radial velocity components (pitch) are made.
This fact and its possible effect on the accuracy of the meth-
od are discussed.
The generally accepted criteria for acceptable flow conditions
for stack sampling requires that the direction of flow be
within - 10° of the stack axis. If the flow direction is out-
side this range, special sampling procedures are needed to
obtain unbiased results. The angle between the longitudinal
axis of the stack and the plane of the pitot tubes when
aligned parallel to the flow direction is referred to as the
flow angle. It has the same magnitude as the angle between a
plane perpendicular to the stack axis and the plane of the
pitot tubes at the null (zero manometer reading) position.
The basic attempt of this paper is to describe the method as
applicable to determination of pollutant mass flow rates.
This requires determination of pollutant concentration as well
as volume flow rate. The procedure is not as complex if only
pollutant concentration is needed.
Particulate Sampling
A particulate stack sample must be extracted isokinetically at
each sampling point, and the volume extracted must be propor-
tional to the stack exit volume from each area increment.
If particulate sampling is performed with the nozzle and pitot
tubes in any position other than parallel to the flow stream,
various sources of bias are introduced. Distortions of nozzle
area and variations of pitot tube reading with flow angles
other than zero are sources of bias.l The method presented is
offered as a procedure to reduce biasing effects.
IV-5
-------
78-35.2 2
The volume extracted at a sampling point may be expressed as
Vn = (An) (vn) (t) (1)
where:
Vn = Nozzle volume extracted at the point
An = Area of the nozzle
vn = Nozzle velocity at the point
t = Sampling time at the point
Varying nozzle area (An) from point to point is not feasable,
and nozzle velocity must be equal to the velocity of the flow
stream. Therefore, sampling time at each point must be
adjusted so that the volume extracted at each sampling point
is proportional to the stack exit volume from each area incre-
ment. This is accomplished by weighting the sampling time at
each point according to the vertical component of velocity at
that point (cosine of the flow angle).
Suggested Procedure
Sampling parameters for cyclonic flow sampling are set up in
the same manner as for non-cyclonic flow. Preliminary velocity
traverse readings are taken with the pitot tubes aligned paral-
lel to the flow at each sampling point. The direction of flow
at each point is determined by locating the null position of
the pitot tubes (zero manometer reading) and then rotating the
pitot tubes 90° to obtain velocity measurements. The flow
angle at each sampling point is recorded during the preliminary
velocity traverse.
Isokinetic sampling is performed at each sampling point in the
normal manner except with the nozzle and pitot tubes aligned
parallel to the flow and with sampling time weighted according
to the cosine of the flow angle at each point. This may be
accomplished by selecting a basic sampling time for each point
which may be multiplied by the cosine of the previously mea-
sured flow angle for each point. Inspection of the planned
sampling times is necessary to insure that total sampling time
and volume are sufficient, and that the shortest sampling time
is long enough for accurate measurement and recording.
Calculations
Emission calculations on a concentration basis are
C = M/V (2)
where:
C = Particulate concentration
M = Mass of particulate caught
V = Volume of gas extracted
IV-6
-------
78-35.2
The results are directly applicable to stack emission concen-
tration since the mass of particulate caught (M) and the vol-
ume of gas extracted (V) have been weighted according to the
stack exit volume from each area increment.
Emission calculations on a pollutant mass rate basis are
P = (M/V) (vs) (As) (COS F) (3)
where:
P = Mass flow rate of particulate
M = Mass of particulate caught
V = Volume of gas extracted
vs = Average measured stack velocity
As = Area of the stack
COS F = Average of the cosines of the flow angles
The emission concentration (M/V) is weighted according to the
stack exit volume from each area increment, and the average
measured stack velocity (vs) is measured with the pitot tubes
aligned parallel to the flow at each sampling point. There-
fore, the average velocity must be multiplied by the average
of the cosines of the flow angle at each point to obtain the
exiting component.
Calculations of isokinetic variation are made in the normal
manner. Since sample volume becomes weighted when sampling
time is weighted, no additional adjustments are necessary, and
input values to the isokinetic calculation are directly used
as measured.
Accuracy Considerations
According to sampling terminology, a large particle is one
that is influenced more by its own inertial characteristics
than by the flow stream. Therefore, when the nozzle is paral-
lel to the flow direction of a cyclonic flow stream it may not
be parallel to the flow direction of large particles in the
stream. This problem is not necessarily peculiar to cyclonic
flow streams. The effect of particle paths not parallel to
the nozzle is a smaller effective nozzle area resulting in
high isokinetic variation which in turn tends to induce a low
bias to the sample.1 The effects of this type bias have not
been quantitated but this sampling method is an attempt to
keep such bias to a minimum.
The sampling method presented is limited to flow streams with
tangential components of flow. The following exercise shows
that adjustments for radial flow components are unnecessary if
the sampling ports are at least two stack diameters downstream
from the stack inlet or disturbance.
IV-7
-------
78-35.2 4
Consider a particle in a stack with a vertical velocity compo-
nent, v, a tangential velocity component, vt, and a radial
velocity component, vr, at a distance R from the center of the
stack.
The radial acceleration (Ar) of the particle due to centrifu-
gal effects of vt is
Ar = vt2/R (4)
If the particle starts from rest at the center of the stack
(most restrictive case) and accelerates at Ar, the time (t)
required to reach the position, R distance from the center, is
t = R/(ivr) (5)
and vr = (Arj (t) (6)
Substituting (4) and (5) into (6)
vr = (vt2/R)(R/ivr)
Simplifying vr2 = 2 v^2 (7)
At the initial occurence of cyclonic flow (flow 10° from axial)
Vt/v = tan 10°
or vt = v tan 10° (8)
Substituting (8) into (7)
Vr2 = v2(2 tan2 10°)
or vr = (0.25)v (9)
which shows that at the smallest flow angle at which cyclonic
flow exists, the radial velocity of a particle is one fourth
the vertical velocity. Therefore, if the sampling ports are
at least two diameters from the entrance to the stack, the
particle will reach the stack wall before reaching the ports
because it will travel half a diameter in a radial direction
while it travels two diameters in a vertical direction. If
the particle reaches the stack wall before reaching the ports,
no radial component of velocity is possible, and no pitch
adjustment of the probe is necessary. This is substantiated
by the cyclonic flow work described by Phoenix and Grove.
"Two 24-point traverses were chosen but, in most cases, points
1, 2, 23, and 24 were not sampled because of an excessive
amount of particulate and water droplets at the wall". If the
average flow angle in the stack is greater than 10°, the par-
ticle reaches the stack wall before travelling two diameters
IV-8
-------
78-35.2
vertically. If the average flow angle in the stack is less
than 10°, sampling is performed in the normal manner with no
adjustments necessary.
IV-9
-------
78-35.2
References
1. J. W. Peeler, "Isokinetic particulate sampling in non-
parallel flow systems - cyclonic flow", Entropy Environ-
mentalists, Inc., (1977) (Draft).
2. F. J. Phoenix and D. J. Grove, "Cyclonic flow - character-
ization and recommended sampling approaches", Entropy
Environmentalists, Inc., EPA Contract 68-01-4148, (Novem-
ber, 1977) (Draft).
IV-10
-------
PAPER V
CYCLONIC FLOW - CHARACTERIZATION AND
RECOMMENDED SAMPLING APPROACHES
Frank J. Phoenix
D. James Grove
-------
CYCLONIC FLOW - CHARACTERIZATION AND
RECOMMENDED SAMPLING APPROACHES
BY
FRANK 0. PHOENIX & D. JAMES GROVE
ENTROPY ENVIRONMENTALISTS, INC.
PROJECT OFFICER
KIRK FOSTER
DIVISION OF STATIONARY SOURCE EMISSIONS
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
CONTRACT NO.
68-01-4148
EPA TASK 20
FEBRUARY, 1978
V-3
NTROPV %EC,AL,STS,N
MVIRONMENTALISTS, INC . /CAMPLING
-------
ABSTRACT
Presented are the results of a study addressing the
problems of particulate sampling in cyclonic flow. The
study characterized cyclonic flow in circular stacks
with tangential gas flow inlets. The flow direction and
magnitude of gas velocities are defined at specified
points on the stack cross section. Using this information,
the applicability of the three different approaches to
Method 5 particulate sampling are analyzed. In addition,
an evaluation of the accuracy of the Alignment Approach,
based on assumptions about particulate behavior, is made.
Finally, some general conclusions of interest to the
stack sampling community are presented, as well as some
recommendations on future research.
V-4
-i-
-------
TABLE OF CONTENTS
ABSTRACT i
TABLE OF CONTENTS 1
INTRODUCTION 2
GAS VELOCITY 4
Figure 1 Yaw Angle Across a Traverse 4
Figure 2 Frequency of Pitch Angles 6
Figure 3 Axial Velocity vs. Percent Diameter 7
Figure 4 Tangential Velocity vs. Percent Diameter 8
PARTICULATE SAMPLING 9
CONCLUSIONS AND RECOMMENDATIONS 13
V-5
-1
-------
INTRODUCTION
This report summarizes the work undertaken to characterize
cyclonic flow. Initial emphasis was placed on the measurement
of pathline directions and pressure differentials in stacks
with cyclonic flow.
Four asphalt plants were tested. In each case, the
process gases were introduced tangentially to the stack.
Three of the four, referred to as plants A, B, and C
showed similar flow patterns. The fourth, plant D, had
straightening vanes inside the stack, and showed a different
flow pattern. Plants A, B, C, and D had stack diameters
of 6', 5', 4', and 5', respectively. These plants were
chosen because they appeared representative of the
asphalt industry. This decision was based on our
experience gained from testing more than 100 asphalt plants.
In addition, an 18-inch diameter wind tunnel, designed
with turning vanes to create cyclonic flow, was tested.
Measurements were taken at sampling points determined
as specified by Standard EPA Procedures (Method 1) with a
United Sensor Corporation, three dimensional, type DA probe.
Two 24-point traverses were chosen, but in most cases,
points 1, 2, 23, and' 24 were not sampled because of an
excessive amount of particulate and water droplets at the wall.
Once cyclonic flow patterns were established, an evaluation
of various approaches to Method 5 sampling was made, culminating
in an analysis of the alignment approach.
-------
The text of this report includes a summary of the
analysis performed to characterize gas velocities,
i
a discussion of particulate sampling in cyclonic flow,
and some general conclusions and recommendations of
interest to the stack sampling community.
V-7
-3-
-------
GAS VELOCITY
Velocity directions were defined by measuring perpendicular
component parts, pitch and yaw, of the true flow direction. Gases
were found to travel in a spiral trajectory up the stack. Spiral
angle, hereafter called yaw angle, was measured from the vertical,
at specified points along two perpendicular traverse axes.
Yaw was measured in the vertical plane perpendicular to the
traverse axis. Data from numerous traverses at plants A, B, and
C and wind tunnel were compared. On each traverse, yaw angle
decreased with radial distance from the center. This trend is
illustrated in Figure 1 below which presents the average yaw
angle at each point for plants A, B, and C. Sampling points
are plotted as a percentage of the stack diameter.
30°.
Yaw
Anqle
50°--
70°--
90°-
Yaw Angle Across a Traverse
AT •
•40'
A •
A
A A
.
•
A
A
0
• Plant A
• B
A C
n Projected Yaw
-- 60'
80'
20",
40 *
60%
80'
Percent Diameter
Fiaure 1. Yaw Angle vs. Percent Diameter for Plants A,B, & C
-4-
V-8
-------
For all traverse performed, yaw angle averaged 39° at
the wall and 73° at 20" radius. In the center of the stack,
flow pressures were too low for accurate measurement of
yaw angle. Instead, we projected a continuation of the
curves in Figure 1, and found yaw to average 84° at the center.
The stack with straightening vanes showed a different
flow pattern. Yaw angle at the wall averaged 25° and decreased
to zero at the stack center. Axial velocity was almost
constant across the stack.
Pitch angle was measured in the vertical plane of the
traverse axis. Positive pitch was chosen toward the center
of the stack and negative pitch toward the stack wall.
No pitch angle pattern was apparent from the data,
rather, pitch angle direction varied randomly across
each traverse. In addition, the magnitude and sign of pitch
ateachpointvariedwithtime.
At each point, pitch angle magnitude was samll in
comparison to yaw angle. Taking each traverse separately,
absolute pitch angles were averaged. The worst traverse
encountered showed an average pitch angle of 10.6°. In
contrast, the smallest average pitch angle on a traverse
was 1.8°. In addition, all pitch angles measured on the
16 traverses tested were averaged. This yielded an average
pi tch angle of 6° .
Figure 2 is a graph of the frequency of occurance of
each pitch angle measured. On the ordinate, the number of
times a particular pitch angle was encountered is plotted
-5-
V-9
-------
vs. the pitch angle on the abscissa. For example, a 2°
pitch angle was measured 17 times and a 10° pitch was
measured 5 times. As shown, a 0° pitch angle was encountered
more than any other angle. At the other extreme, a 34°
angle was measured on-c.e. This was the largest angle measured
Frequency
of
Occurrance
36.
32-
28-
24-
20-
16-
12 •
8 -
4 -
0 •
* Frequency of Occurrance of Pitch
vs
t Pitch Angle
'
»
. * .*
* A A
1 1 1 1 1 1 1 1 1 1 —
3° 12° 16° 20° 24° 28° 32° 36°
Pitch Angle
Figure 2. Frequency of Occurrance of Pitch Angle
-d-
V-10
-------
Axial and tangential velocity components were calculated
from the pressure differentials measured in the flow direction.
Axial velocity increased with radial distance. T.he maximum
axial velocity averaged at 85>> radius. The mean axial
velocity averaged 52" radius. (see Figure 3) The graph
represents a typical axial velocity profile. It is an average
for all traverses performed at plants A, B, and C. The
velocity calculated from each point on the traverse was plotted
as a percentage of the maximum velocity measured on the
traverse.. Sampling points are plotted as a percentage of the
stack diameter.
Axial Velocity Across a Traverse
Axial
Velocity
Percent Diameter
Figure 3. Axial Velocity Components vs. Percent Diameter for Plants A,B, & C
V-ll
-------
Tangential velocity showed a slightly different trend.
(see Figure 4) The location on the traverse of the maximum
tangential velocity averaged 57% of the radial distance
from the center.
Tangential Velocity Across a Traverse
Tangential
Velocity
Percent Diameter
Figure 4. Tangential Velocity vs. Percent Diameter
-8-V-12
-------
PARTICULATE SAMPLING
Several papers have been written that outline different
possible approaches to stack sampling for participates in
cyclonic flow. They are: 1) Blind Man's Approach,
2) Alignment Approach, 3) Compensation Approach, and
4) source modifications. Source modifications are always
recommended where possible, but when impractical, most
suggest using the Alignment or the Compensation Approach.
None of the first three approaches compensate for the fact
that the particulates may be moving in some direction other
than that of the gas flow.
Several problems are encountered when using the
Blind Man's Approach, which involves pretending that cyclonic
flow does not exist. Until recently, this was the approach
most commonly used. Errors in.axial velocity measurements,
due to misalignment of the pitot tube with the flow direction,
as evidenced in Figures 1 and 2, range from 30 - 100% across
the stack. Thisin turn causes an anisokinetic sampling rate.
In addition, since the nozzle is not aligned with the flow
direction, the effective area of the nozzle opening is reduced.
The amount is a function of the average misalignment angle across
the stack. Assuming an average yaw angle from Figure 1,
1 Walter S. Smith & D. James Grove, "Pitot Tube Errors
Due to Misalignment and Non-Streamlined Flow"
Stack Sampling News, Vol. 1, No. 5, Nov. 1973
-9-
V-13
-------
Plant A of 65°, the effective nozzle area would be reduced
by a factor equal to cosine of 65° or 58%. Another problem
arises because the constraint of proportional sampling is
not met. All of these problems introduce biases of various
directions and quantities.
The Compensation Approach is similar to the Blind Man's
Approach above except that the nozzle size and sampling rates
are adjusted as per the yaw angle measured at each point.
The nozzle in this approach is still positioned parallel
to the stack walls. While it should give results which
are more accurate than the Blind Man's Approach, for large
angles of misalignment, like those found near the stack
center, the flow around the nozzle will create aerodynamic
interferences with the isokinetic sampling. In general,
these interferences will bias the concentration low, but
to what extent is unknown.
The Alignment Approach requires the determination of
yaw angle flow directions at specified points on the
traverse axes. It is assumed that pitch angles are small
enough to be ignored. Once yaw angles are measured, the
probe is aligned with the flow direction and sampling
performed as outlined. This approach will give valid
results if particulate flow directions do not deviate from
gas flow directions. But, this is unlikely. Thus, the
alignment of the probe with the flow direction is not enough
to ensure the collection of an accurate particulate sample.
V-14
-10-
-------
The degree of bias is dependent upon the size and
behavior of participates in the cyclonic flow stream.
Unfortunately, very little work has been done to characterize
particulates under these conditions. Therefore, to facilitate
the evaluation of the sample collected, as described above,
the following assumptions have been made about particulates:
1) Particles less than 1 micron in size behave like
gases in the flow stream and follow the same spiral
trajectory up the stack as the gases.
2) Particles between 1 and 10 microns in size behave
somewhat like gases, but begin to take on particle
properties as they travel up the stack.
3) Particles greater than 10 microns in size are affected
more by intertial forces than gas molecules. They
most likely deviate substantially from the spiral
trajectory of the gases as they travel up the stack.
The following discussion estimates the bias encountered,
based on the assumptions presented above when using the
Alignment Approach. Three simplified cases are examined. None
are meant to depict an actual situation, but rather span
the range of possibilities.
Case One: Consider a cyclonic flow stream in which all
entrained particles are less than one micron in size. Since
particles of this size behave like gases, the only misalignment
angles are caused by errors in yaw angle measurement and pitch
angle flow. At Plant A, in port A, yaw angle was measured
rll-
/-ii
-------
to within +2° and pitch angle averaged 5°. The worst
case of misalignment would result in a decrease in the
effective nozzle area by a factor equal to cosine 2°
times cosine 5° or 0.4%. Errors in velocity readings
taken from pitch and yaw calibration curves would range from
-2% to +4%.
Case Two: Let us assume an average particle size
(including particulate-1aden water droplets) of 50 microns
with a deviation angle of 40°. Velocity errors are
comparable to those in the previous case, but the effective
nozzle area is decreased by a factor equal to cosine 40°
or 23%. This will cause a particulate concentration result
which is too low by more than 23%, and an equally low bias
in the pollutant mass calculation.
Case Three: Consider the situation where the majority
of particles average 100 microns in size with a deviation
angle of 80°. Velocity errors will be similar to those in
Case One, but the effective nozzle area will decrease by
a factor as high as cosine 80° or 83%. Such a decrease
will result in a particulate catch which is too low by
83%.
v-i6
-12-
-------
CONCLUSIONS AND RECOMMENDATIONS
The results of this study were compared with two
2
similar studies on cyclonic flow. Kiochi linoya focuses
on IV'» 2", 3", and 4" diameter cyclone exit pipes.
o
Fluidyne Corporation examined a wind tunnel they designed
to create cyclonic flow. The results from all three
studies yield similar conclusions of interest to stack
samplers. These conclusions are summarized below.
Yaw angle decreases with radial distance from the
center, is reproducible, and is independent of flow
quantity. Total velocity, in the flow direction, increases
with radial distance to its maximum between 60-65% radius
before gradually dropping off toward the wall. Axial
velocity increases with radial distance to its maximum
found between 85-95% radius. Radial velocity components
are small and accurate velocity measurements to within 4%
can be obtained by ignoring pitch.
Wherever possible, source modifications should be made
to eliminate the need for particulate sampling in cyclonic
flow. When source modifications are impractical, the
Alignment Approach provides the best alternative proposed
to date, but even it can have accuracy problems.
Koichi linoya, "Study on the Cyclone", Memoirs of
the Faculty of Engineering, Nagoya University,
Vol. 5, No. 2, Sept. 1953
David P. Saari, "Effective Sampling Techniques for
Particulate Emissions From Atypical Stationary Sources",
EPA Contract 68-02-1796.
-V-i7
-------
The main obstacle confronting stack sampling in
^f-
cyclonic flow is the lack of knowledge on participate behavior
Further research should focus on this problem. Information
on the size distribution of particles in typical pollution
sources having cyclonic flow stacks is also needed. Since
most of these stacks are used for dewatering after a scrubber,
the water droplets must also be studied.
-14-
V-18
-------
PAPER VI
FLOW VELOCITY STUDIES
D. P. Saari
H. A. Hanson
-------
EFFECTIVE SAMPLING TECHNIQUES
FOR PARTICULATE EMISSIONS
FROM ATYPICAL STATIONARY SOURCES
EPA-600/2-80-034
January 1980
Section 6.
"Wet Scrubber Sampling Techniques"
Flow Velocity Studies
D. P. Saari
H. A. Hanson
The main objective of the velocity studies in the scrubber
exhaust model was to assess the ability of various instruments
to determine the volumetric flow rate and the flow angularity
in the cyclonic flow field. Three instruments were used to
make velocity measurements in the model; a standard S-tube, a
three dimensional directional pitot probe, and a Fecheimer type
probe.
The S-tube, attached to a standard EPA Method 5 type
sampling probe, was first used to assess the error level which
would exist if this instrument were simply oriented parallel to
the duct centerline without accounting for the cyclonic nature
of the flow field. Several 12 point velocity surveys were made
with the S-tube, traversing the duct as if the flow were uni-
formly parallel to the duct centerline, and the volumetric flow
rate determined from these traverses was compared with the
actual volumetric flow rate through the model as determined
from pitot tube traverses in the horizontal air supply ductwork
upstream of the test section. The results of these surveys,
shown in Fig. 73, show that the error is on the order of 60% to
115% of the actual flow rate and that the error actually in-
creases from 2 to 6 duct diameters downstream of the mist
eliminator, after an initial decrease.
Attempts were made to determine the. yaw angle of the flow
by using the demonstrated directional sensitivity of an S-tube
as indicated in Fig. 56. However, the turbulent fluctuations
in the flow caused fluctuations in the S-tube output of such
magnitude as to completely overshadow the capability of this
instrument to locate the angle of maximum velocity. Thus, the
S-tube could not be used with any degree of reliability to
determine the yaw angle in cyclonic flow in this manner. A
potential method for use with the S-tube which was not evalua-
ted is the so called "null method," in which the S-tube is
rotated until the pressure sensed by the two pressure taps is
equal. The instrument should then be aligned at 90° to the
VI-3
-------
Figure 68. Scrubber exhaust model and test platform at
Medicine Lake Laboratory
110
VI-4
-------
Figure 69.
Droplet injection system for scrubber exhaust
model-external hardware
Figure 70.
Spray nozzles for droplet injection system (re-
moved from ductwork of scrubber exhaust model)
ft1-!
-------
Figure 71. Droplet injection system spray nozzles producing
low droplet density mist
Figure 72.
Droplet injection system spray nozzles producing
high droplet density mist
112
VI-6
-------
V = measured volumetric flow rate
m
V = actual volumetric flow rate
Q.
2.2
2.0
1.8
1.6
Vm 1-4-
v~
a i"-2
l.Q
0. 8
0.6-
0.4-
O.Z
0
•g
i i
o o
D °- a
o
0 - V = 115 m3/min
a
3
D - V = 68 m /min
a.
1 1 1 : 1 1 u.
Distance Downstream of Mist Eliminator,
(Duct Diameters)
Figure 73. Volumetric flow rate error for 12-point
velocity surveys•in scrubber exhaust
model with S-tube parallel to duct
centerline
113
VI-7
-------
velocity and can be rotated so as to be properly aligned for a .
velocity measurement.
Extensive surveys of the cyclonic flow field were made
with a directional velocity probe capable of accurately measur-
ing both pitch and yaw angles as well as the flow velocity.
This instrument, the United Sensor DC-125 directional velocity
probe, is shown in Figs. 74 and 75 together with a duct traverse
unit which was used to accurately position the probe in the
test section of the scrubber exhaust model. As can be seen in
Fig. 75, the pressure taps on this probe are very small, thus
making use of the instrument impractical in particulate-laden
streams. It was used in the model, however, to obtain an
accurate map of the flow field for evaluation of other velocity
instruments. This probe mounted in the scrubber exhaust model
is shown in Fig. 76. Manufacturer's calibration data were used
to reduce the probe measurements to velocites.
The results of the velocity measurements with the United
Sensor probe are shown in Figs. 77-80. The definition of the
coordinates used in presenting the data is shown in Fig. 67.
The actual flow rate, Q , used to nondimensionalize the velocity
terms in Figs. 77-80, was determined from pitot tube traverses
in the horizontal air supply ductwork upstream of the test
section. The curves indicated in Figs. 77-80 represent intui-
tive estimates of the "average" data and are not based on
curve-fitting or regression formulas. The most interesting
observation made from these velocity surveys is the fact that
the average radial velocity component is very small, particu-
larly as the distance downstream of the mist eliminator in-
creases.
Extensive velocity surveys were also made with a Fecheimer
type probe, shown in Fig. 81. The traverse unit shown in Figs.
74 and 76 was used to position the Fecheimer probe for these
measurements. The same points were measured with the Fecheimer
probe as with the United Sensor probe. The Fecheimer probe was
calibrated against a standard pitot tube and was found to have
a velocity coefficient of 1.06.
Velocity components determined from the yaw angle and
total velocity measured with the Fecheimer probe are shown in
Figs. 82-85. The same dimensionless presentation is used as
for the United Sensor probe. While the United Sensor probe
appears to provide somewhat more repeatable data, comparison of
the measurements indicates that the average velocity components
determined with the two instruments agree quite closely, with
the exception of the radial component which can not be measured
with the two-dimensional Fecheimer probe. Since the radial
components indicated in Figs. 77-80 are essentially negligible,
however, the approximation made in using the two-dimensional
instrument seems to be acceptable.
114
VI-8
-------
Figure 74.
United sensor DC-125 3-dimentional directional
probe and traverse unit
&$$&?.''' ' ':.*' ' v-'t , ' •'•
Figure 75. Sensing head of united sensor DC-125 directional
probe
115
VI-9
-------
INTERNAL
VIEW
EXTERNAL
VIEW
Figure 76. United sensor DC-125 directional probe mounted
in scrubber exhaust model for velocity traverses
116
VI-10
-------
Radial
Component
v* = v o
Pa
1
0
-1L
o e
a o
O e
o
90
180
A 6 = 270
-ft-
Circumferential
Component
V
Axial
Component
Total
Velocity
v*
Duct
Centerline
Figure 77.
Normalized velocity components measured 0.75 duct
diameters downstream of mist eliminator -
United sensor probe
117
VI-11
-------
Radial
Component
vr*
0
-1
O 0
D e
O e
A e
o
90
180
270
Circumferential
Component 3
V
Axial
Component
V
2 •
1 -
-1 L
Total
Velocity.
v*
A
3
1
0
0
Duct
Centerline
0.2
0.4 0.6 . 0.8
r/ro
1.0
Duct
Wall
Figure 78.
Normalized velocity components measured 2.25 duct
diameters downstream of mist eliminator -
United sensor probe
118
VI-12
-------
Radial
Component
v *
r
7rr
Q
2
o e
D e
O e
A 0
0
90
180
270
Circumferential
Component
V
Axial
Component
v *
z
Total
Velocity
v*
Figure 79,
0
Duct
Centerline
Normalized velocity components measured 3.75 duct
diameters downstream of mist eliminator -
United sensor probe
119
VI-13
-------
Radial
Component
« *
7* =
Ttr,
o o
a o
O e
A e
o
90
180
270
1
0
1
A
d
A A
y-y i>
a Q
A
Circumferential
Component
v *
ve
3
2
1
0
Axial
Component
v *
z
Total
Velocity
4 r
3
2
1
0
0
Duct
Centerline
0.2
0.4
0.6
r/r
0.8
1.0
Duct
Wall
Figure 80.
Normalized velocity components measured 5.25 duct
diameters downstream of mist eliminator -
United sensor probe
120
VI-14
-------
Figure 81.
Fecheimer probe used in velocity
surveys of scrubber exhaust model
VI-15
121
-------
Ttr
v* =
o
D
O
A
o
G
e
= o
= 90
= 180
0 = 270
Circumfer-
ential
Component
V
Axial
Component
v *
z
2
1
0
-1
Total
Velocity
A
3
1
0
0.2
Duct
Centerline
0.4
0.6
r/r
0.8
1.0
Duct
Wall
Figure 82.
Normalized velocity components measured 0.75 duct
diameters downstream of mist eliminator -
Fecheimer probe
122
VI-16
-------
v* *ro
o e
D e
O e
A Q
o
90
180
270
Ar-
Circumferential 3
Component
v * 2
1 -
0
Axial
Component
v *
Total
Velocity
A i-
2
1
0
0
0.2
Duct
Centerline
J_
I
0.4 0.6 , 0.8
r/r0
1.0
Duct
Wall
Figure 83.
Normalized velocity components measured 2.25 duct
diameters downstream of mist eliminator -
Fecheimer probe
123
VI-17
-------
o
D
O
e
e
e
o
90
180
e = 270
Circumferential
Component
v *
ve
Axial
Component
v *
z
Total
Velocity
v*
2 _
0
Duct
Centerline
O
o
0.2
0.4
0.6
0.8
O
1.0
Duct
Wall
Figure 84.
Normalized velocity components measured 3.75 duct
diameters downstream of mist eliminator -
Fecheimer probe
124
VI-18
-------
v*= oT" v
o
D
O
A
e
e
e
o
90
180
e = 270
Circumferential
Component
V
Axial
Component
v *
z
Total
Velocity
v*
0
0.2
Duct
Centerline
0.4
0.6
0.8
r/r
1.0
Duct
Wall
Figure 85.
Normalized velocity components measured 5.25 duct
diameters downstream of mist eliminator -
Fecheimer probe
125
VI-19
-------
The degree of accuracy of the Fecheimer probe in deter-
mining total volumetric flow rate was also examined. Compari-
sons of flow rates determined from the axial velocity component
in several 24-point surveys and the actual flow rate are shown
in Fig. 86. It can be seen that the reliability of the volu-
metric flow rate surveys increased greatly with distance down-
stream of the mist eliminator. At the maximum distance achiev-
ed in the scrubber exhaust model, 5.25 duct diameters, the 24-
point surveys provided accuracy within 5% of the actual value.
At locations closer to the mist eliminator, a greater number of
measurement points would be required to achieve sufficient
accuracy and repeatability.
On the basis of the information presented above, the
Fecheimer probe appears to be a suitable instrument for deter-
mination of both volumetric flow rate and flow angularity in
the cyclonic flow typical of wet scrubber outlets.
VI -20
------- |