United States
Environmental Protection
Agency
Atmospheric Research and Exposure v
Assessment Laboratory .3
Research Triangle Park NC 27711 -^
Research and Development
EPA/600/S3-88/057 Aug. 1989
v°/EPA Project Summary
Application Guide for Source
PM10 Measurement with
Constant Sampling Rate
Wffiam E. Farthing and Sherry S. Dawes
This manual presents a method,
Constant Sampling Rate (CSR), which
allows determination of stationary
source PM10 emissions with
hardware similar to that used for
Methods 5 or 17. The operating
principle of the method is to extract a
multipoint sample so that errors due
to spatial variation of particle size
and anisokinetic sampling are kept
within predetermined limits. Current
specifications were designed to limit
error due to spatial variations to 10%.
The maximum allowable error due to
anisokinetic sampling is ±20%; in es-
sentially all sampling situations,
cancellation of sampling error will
limit overall anisokinetic sampling
error to much less than this value.
This manual specifically addresses
the use of the CSR methodology for
determination of stationary source
PM10 emissions. Material presented
in this manual includes: calibration
of sampling train components, pre-
test setup calculations, sample rec-
overy, test data reduction, and rou-
tine equipment maintenance.
This Project Summary was
developed by EPA's Atmospheric
Research and Exposure Assessment
Laboratory, Research Triangle Park,
NC, to announce key findings of the
research project that is fully
documented In a separate report of
the same title (see Project Report
ordering information at back).
Introduction
To ensure that a representative sample
of particulate matter is obtained from a
flowing gas stream, three key factors
must be considered. First, the length of
the sampling period must be adequate to
formulate an appropriate temporal
average of stream conditions. Second,
the location and number of sampling
points must be chosen so that a spatial
average of emissions across the sampling
plane is obtained. Finally, sampling must
be performed isokinetically so that the
sample is not biased with respect to
particle size. These conditions are
addressed in the EPA methods for
measuring total particulate emissions
(Methods 5 and 17) by specifying
sampling periods which take into account
process cycles, traversing techniques
rather than single point sampling, and
adjustments in sample flow rate (i.e.,
nozzle velocity) to match local stream
velocity at each point of the traverse so
that isokinetic sampling is maintained.
However, a size-specific method, such
as one for measuring particulate matter of
aerodynamic diameter s 10 vim (PM10),
must combine the considerations of
obtaining a representative sample with
the need to segregate the sample into
two or more size fractions. Inertial sizing
devices such as cascade impactors and
sampling cyclones must be operated at a
constant flow rate to maintain constant
size cuts. For a fixed nozzle size, this
precludes any adjustment in nozzle
-------
velocity to maintain isokinetic sampling.
Without the use of new sampling hard-
ware, a PMio sampling method must
then become a compromise between the
conflicting requirements of inertial par-
ticle sizing and representative sampling.
Without the use of new sampling hard-
ware, a PMio sampling method must
then become a compromise between the
conflicting requirements of inertial par-
ticle sizing and representative sampling.
This manual presents a method,
Constant Sampling Rate (CSR), which
allows determination of PMio emissions
from stationary sources with hardware
similar to that used for Methods 5 or 17.
The operating principle of the method is
to extract a multipoint sample so that
errors due to spatial variation of particle
size and anisokinetic sampling are kept
within predetermined limits. Current
specifications were designed to limit
error due to spatial variations to 10%
The maximum allowable error due to
anisokinetic sampling is ±20%; in
essentially all sampling situations,
cancellation of sampling error will limit
overall anisokinetic sampling error to
much less than this value.
Operating Principles
In developing the CSR strategy,
several specific objectives shaped the
details of the method. First, the
technique was designed to minimize
changes in equipment from that used for
Methods 5 or 17. Second, the details of
the traversing strategy were selected to
limit errors from spatial variation and
anisokinetic sampling to the level of
more intrinsic errors (such as fluctuations
in source emissions or basic
measurement inaccuracy). Finally,
measurements would be made to
provide an average representative of
emission rates rather than concentration.
To obtain a sample which is unbiased
with respect to particle size, one must
sample isokinetically. That is, the gas
velocity of the sample stream entering
the sampling nozzle must match the local
gas velocity in the duct from which the
sample is being withdrawn. If the gas
velocity in the nozzle is greater than the
local duct velocity, the flux of large
particles through the nozzle cross-
section will be less than that for the free
stream; large particles are those which
do not follow flow streamlines because of
their inertia. As a result, the collected
mass of large particles will be selectively
depleted. Conversely, if the gas velocity
in the nozzle is lower than the local duct
velocity, the flux of large particles though
the nozzle cross-section will be greater
than that of the free stream. The
collected mass of large particles in this
instance will be selectively enriched.
The resulting concentration of very small
particles in the sample remains
unchanged from that in the duct in either
case. The resulting concentration of very
large particles in the sample approaches
the ratio of the duct velocity to the nozzle
velocity.
For any given particle diameter, the
anisokinetic sampling error may be
expressed as the aspiration coefficient,
A, which is defined as the ratio of
measured concentration to actual
concentration, in terms of the particle
Stokes number, K, and the ratio of
stream velocity, v, to nozzle velocity, u.
A = 1+ (R - I)
B
(B+l)
where R = velocity ratio v/u
B = (2 + 0.617/R)K
K = particle Stokes number
with respect to the
nozzle, iv/d
i = particle relaxation time
CD2/i8n, seconds
C = Cunningham slip factor
D = particle aerodynamic
diameter
p = gas viscosity, poise
d = nozzle diameter
The equation shows that, for given
stack conditions and particle size,
limiting the anisokinetic sampling error
becomes a question of limiting the
velocity ratio, R. In other words, for a
given limit on error due to anisokinetic
sampling, maximum and minimum
values of R (Rmax arid Rmin) will yield
results within the stated limits. It may
also be noted from the equation that B is
proportional to particle diameter squared.
This indicates that anisokinetic sampling
error decreases with decreasing particle
size. In other words, when sampling for
PM-io emissions, the velocity ratio, R,
could be outside the 0.9 to 1.1 range
specified for total emissions in Methods
5 and 17 and still retain equivalent
accuracy.
The choice of the limits on anisokinetic
sampling error for CSR is important. An
overly generous range could produce
data with an excessive amount of error.
At the other extreme, small limits would
restrict the velocity ratio to the point that
most sites would require multiple nozzle
sizes for a complete traverse, which
would increase the on-site sampling
effort. For the purposes of PM^, limits
of ±20% on error due to anisokinetic
sampling were chosen to specify the
limits on the velocity ratios, Rmm an
Rmax- Actual sampling error for moi
sources will be less than the ± 20% lim
for two reasons. First, point-by-poir
R values will usually be something les
than the limits, Rmtn and Rmax. an
cancellation of errors will occur becaus
R-1 values will be negative at som
points and positive at others. Seconc
the limit of ±20% was determined b
assuming a monodisperse sample c
10-nm particles. In actuality, the PM-\
sample is composed of particles wit
aerodynamic diameters less than 10 urn.
The goal of a traversing strateg
should be to reduce error due to spatie
variation so that it is not the dominan
source of error but is comparable to o
less than other sources of error. In gen
eral, the PM^ fraction is expected to bi
less stratified than total particulate emis
sions. The lower inertia of the PM-n
fraction would cause less deviation fron
gas flow in bends and faster dampmc
from turbulent diffusion once stratificatior
is produced. On the basis of available
PM-io profiles, using 8 to 12 traverse
points is expected to reduce this type o
error to less than ± 10%.
Sampling Hardware
As stated previously, one of the
objectives during the development of the
CSR technique was to make use o
existing sampling equipment. Therefore
the hardware changes required tc
operate a CSR sampling system have
been kept to a minimum. Like a standarc
total particulate sampling train, a CSF
system may be operated with an out-
of-stack filter (analogous to Method 5;
or an in-stack filter (analogous to Meth-
od 17).
The primary difference between CSP
and Methods 5 or 17 hardware is the
PM^o sampling device. Although z
number of particle sizing devices may be
considered for use as a PM-io device.
practical considerations eliminate several
of the choices. For the purposes of this
method, the only single-stage device
which should be considered is an in-
stack cyclone. Although a single stage of
a cascade impactor may provide the
appropriate size segregation, problems
such as particle bounce and re-
entrainment keep this from being an
acceptable choice. The multistage de-
vice recommended for use with this
method is a cascade impactor. Although
a series cyclone such as the SRI/EPA
five-stage series cyclone provides par-
ticle sizing similar to that of a cascade
impactor, the mass loading necessary for
adequate sample retrieval would require
-------
unacceptably long run times for most
outlet concentrations.
Before any particle sizing device is
used as a PM^ sampler, it must be
shown to meet some specific
performance requirements. These
requirements are designed to ensure that
the relationship between collection
efficiency, particle diameter, and
operating conditions is well defined and
that the performance of the sampling
device is not disturbed by the sampling
nozzles used in practice.
One sampling device known to meet
the performance specifications is the
commercially available version of
Cyclone I, the first stage of the SRI/EPA
five-stage series cyclone. The outer
dimensions and physical appearance of
the cyclone may vary, depending on the
specific commercial source. The critical
inner dimensions, however, are standard-
ized to the original design parameters.
Laboratory calibrations have shown
Cyclone I produces a 10-|im 050 at a
flow rate of approximately 0.5 dscfm; the
precise flow rate will depend on local
stack conditions.
-------
William E. Farthing and Sherry S. Dawes are with the Southern Research Institute,
Birmingham, AL 35255.
Thomas £ Ward is the EPA Project Officer (see below).
The complete report, entitled "Application Guide for Source PM10 Measurement
with Constant Sampling Rate," (Order No. PB 89-193 3201 AS; Cost: $15.95,
subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
United States Center for Environmental Research
Environmental Protection Information
Agency Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S3-88/057
------- |