1096 STACK SAMPLING
this line is called a traverse, and the individual sampling
points termed traverse points. The configuration of the ports
and the traverse points is generally chosen according to Test
Method 1.
Three additional considerations must be addressed in
order for the sample to be considered representative. First,
the particulate-laden gas stream entering the nozzle must be
typical of the stream flowing by it. Second, the makeup of
the gas stream leaving the probe and entering the rest of the
sampling train must be substantially the same as it was when
it entered the nozzle. Third, sampling must be conducted at a
time and for sufficient duration to cover any inconsistencies
in the pollution emission rate.
The first of these conditions may sound like overkill.
However, the previous work was to ensure the representa-
tiveness of the location. This part concerns the representa-
tiveness of the gas collected there. This is ensured by careful
design of the nozzle and control o the side stream flow rate.
The opening of the nozzle is designed with sharply tapered
edges. The nozzle itself is shaped to minimize deposition of
particulates on the inside walls as the stream turns 90°.
The sampling stream flow rate is extremely important
because of the difference in aerodynamics and inertial effects
of particles. Very small particles tend to behave like gas mol-
ecules and tend to follow gas flow stream lines. For these
particles, sample flow rates are not critical. Large particles,
however, do not necessarily follow the gas flow streamlines.
Instead, their flow is controlled more by their inertia. In
other words, they tend to keep going in straight lines. Thus,
if the flow rate into the sampling nozzles is different than
the local gas flow rate, the gas itself and the fine particles
will be skewed, either into or out of the nozzle, depend-
ing on the relative rates. The large particles, however, will
continue along their straight paths. Those, and only those,
in direct line with the nozzle face will enter. This can have
a significant effect on the measured particulate concentra-
tions, depending on the degrees of error in the nozzle flow
rate and on the fraction of particulate mass attributable to the
large particles.
Sampling at exactly the right flow rate is termed iso-
kinetic sampling. Sampling at too great a velocity is called
superisokinetic, while sampling at too low a velocity is
called subisokinetic. Generally, superisokinetic sampling
results in an underestimation of the actual particulate con-
centration (termed a low bias), while subisokinetic sam-
pling results in an overestimation (high bias). Test Method
5 contains instructions for choosing the appropriate nozzle
size and sampling flow rate to ensure isokinetic sampling.
Sample flow rate and nozzle size are based on the volume
of sample gas that needs to be collected and on the flow
rate in the stack gas. The needed sample volume is based
on the amount of particulate needed for the physical or
chemical analyses to be conducted, and will be discussed in
the analysis section. The stack gas flow rate is determined
according to procedures described in Test Method 2. The
procedure involves the measurement of linear flow rate by
means of the relationship between the static and dynamic
pressure in the gas stream. The static pressure is the pressureof the gas stream, as measured by a pressure tap perpendicu-
lar to the flow. The dynamic pressure is the pressure exerted
by the flowing stream and is indicative of the flow velocity.
It is measured by a pressure tap facing directly into the flow
stream. In practice, a device called an “S-type pitot tube” is
used to measure static and dynamic pressure at a single loca-
tion. Standard calculations are then used to compute the flow
rate. By moving the pitot tube across a stack cross section,
the flow rate at each point can be determined. All of this is
described in detail in Test Method 2.
Next, it is necessary to assure that the sample gas stream
does not change substantially between the time it enters the
nozzle and leaves the probe for the collection part of the sam-
pling train. This is accomplished by ensuring that the construc-
tion and operation of the probe do not interfere physically or
chemically with the flowing sample stream. The nozzle and
probe must be made of materials, such as stainless steel, glass,
or teflon, that are smooth and do not react with the stream. In
addition, the probe may be heated to ensure that vapors in the
stream do not condense on the walls of the probe. A stainless
steel nozzle and a glass-lined probe heated to 120 14ºC
(248 25ºF) will suffice for most situations. However, spe-
cial construction materials and/or probe temperature settings
may be required for sampling exhaust streams from certain
source types of containing certain contaminants. The appro-
priate sections of this chapter should be consulted in detail
before any decisions are made.
The third and final consideration for ensuring represen-
tative particulate sampling is that the sampling is conducted
at a time and over a sufficient time period to account for
variabilities in the exhaust steam. In general, it is desirable
to measure the maximum possible emissions, so as to deter-
mine compliance under so-called worst case conditions.
This is accomplished by first determining the stage or stages
in the plant process mot likely to produce the greatest emis-
sion rate. This might require preliminary testing, or it may be
specified in the applicable regulations.
Once the time for testing is selected, it is necessary
to decide the duration of each sample run and the number
of runs to be performed. Generally, this is specified in the
applicable regulation or in the Test Methods. However, in
some cases, it may be necessary to select different sampling
periods. The basic reason for sampling for any given time
is to account for temporal variability in the emissions. Very
few processes produce emission streams that are truly con-
stant over more than a few minutes at a time. In most cases,
though, an hour or two-hour sampling period will be suf-
ficient to smooth out any inherent variation in the exhaust
stream. In practical terms, this is accomplished by pumping
the sample stream through the filter, solutions, or sorbents
for the full sampling period. This effectively averages the
collected sample over the entire time period. Of course, it is
possible that variations in the emission rate with time could
mean that higher concentrations are measured at one point in
a traverse, such as near one stack wall, etc. However, in most
cases, these variations are not significant.
The individual regulations or Test methods usually state
the required sampling duration. There are two reasons thatC019_003_r03.indd 1096C019_003_r03.indd 1096 11/18/2005 11:07:14 AM11/18/2005 11:07:14 AM