PB89-193320
Application Guide for Source PM10
Measurement with Constant Sampling Rate
Southern Research Inst., Birmingham, AL
Prepared for:

Environmental Protection Agency, Research Triangle Park, NC
May 89

-------
THIS PAGE INTENTIONALLY BLANK

-------
                                              PB39-193320


                                            EPA/600/3-88/057
                                            May 1989
     APPLICATION GUIDE FOR SOURCE PM1Q  MEASUREMENT
              WITH CONSTANT SAMPLING RATE
                          by

                  William E. Farthing
                    Sherry S. Dawes
              Southern Research Institute.
            Birmingham, Alabama 35255-5305
                Contract No. 68-02-4442
                    Project Officer

                    Thomas E. Ward
              Quality Assurance Division
Atmospheric Research and Exposure Assessment Laboratory
          Office of Research and Development
     Research Triangle Park, North Carolina 27711
ATMOSPHERIC RESEARCH AND EXPOSURE ASSESSMENT LABORATORY
          OFFICE OF RESEARCH AND DEVELOPMENT
         U.S. ENVIRONMENTAL PROTECTION AGENCY
     RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711

-------
                                   TECHNICAL REPORT DATA
                            (ftease read Instructions on the reverse before completing)
1. REPORT NO.
  FPA/600/3-88/057
                              2.
             3. RECIPIENT'S ACCESSION NO.
                PB89   193320/AS
4. TITLE AND SUBTITLE
 Application Guide for Source  PM-10 Measurement with
 Constant Sampling Rate
             S. REPORT DATE

                Mav 1989
             6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
                                                           8. PERFORMING ORGANIZATION REPORT NO.
 William E.  Farthing, S.S. Dawes
9. PERFORMING ORGANIZATION NAME AND ADDRESS

  Southern Research Institute
  Birmingham, AL 35255
                                                            10. PROGRAM ELEMENT NO.
             11. CONTRACT/GRANT NO.
                                                               68-02-4442
12. SPONSORING AGENCY NAME AND ADDRESS
                                                            13. TYPE OF REPORT AND PERIOD COVERED
 Atmospheric Research and Exposure  Assessment Laboratory
 Office  of Research and Development
 U.S.  Environmental Protection Agency
 Research Triangle Park. NC 27711	
             14. SPONSORING AGENCY CODE
               EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
       This manual presents a method,  Constant Sampling Rate (CSR), which allows
 determination of stationary source PM-10 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 essentially 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 determina-
 tion  of stationary source PM-10  emissions.  Material presented in this manual includes
 calibration of sampling train components, pretest setup  calculations, sample
 recovery, test data reduction, and routine equipment maintenance.
 7.
                                KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS  C. COSATI Field/Croup
IB. DISTRIBUTION STATEMENT

 Release  to Public
IB. SECURITY CLASS (This Report/
 Unclassified
21. NO. OF PAGES

     7*7
20. SECURITY CLASS (Tills page I
 Unclassified
                           22. PRICE
EPA F«m 2220-1 {«•». 4-77)   PREVIOUS COITION i» OB»OLETE
                                             1

-------
                       DISCLAIMER AND PEER REVIEW NOTICE
     The information in this document has been funded wholly or  in part  by
the United States Environmental Protection Agency under contract 68-02-4442
to Southern Research Institute.  It has been subjected to the Agency's peer
and administrative review, and it has been approved' for publication  as an  EPA
document.  Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
                                     ii

-------
                                  FOREWORD
     Measurement and monitoring research efforts ..re designed to  anticipate
environmental problems, to support regulatory actions by developing  an
in-depth understanding of the nature and processes that impact health and  the
ecology, to provide innovative means of monitoring compliance with regula-
tions, and to evaluate the effectiveness of health and environmental protec-
tion efforts through the monitoring of long-term trends.  The Atmospheric
Research and Exposure Assessment Laboratory, Research Triangle Park, North
Carolina, has responsibility for:  assessment of environmental monitoring
technology and systems for air, implementation of agency-wide quality assur-
ance programs for air pollution measurement systems, and supplying technical
support to other groups in the Agency including the Office of Air and
Radiation, the Office of Toxic Substances, and the Office of Solid Waste.

     The environmental effects of PM10 particulate matter are of concern to
the Agency.  Acceptable measurement methodology is critical for proper
assessment of the impact on the environment of these emissions from  station-
ary sources.  Preparation of a manual which specifies measurement procedures
is a key component for assuring reliable test data.  This manual was prepared
to describe the maintenance and operating procedures for the Constant
Sampling Rate approach for measurement of PM1Q emissions from stationary
sources.
                     Gary J. Foley
                     Acting Director
                     Atmospheric Research and Exposure Assessment Laboratory
                     Research Triangle Park, North Carolina 27711
                                      iii

-------
                                  ABSTRACT
     This manual presents a method, Constant Sampling Rate  (CSR), which
allows determination of stationary source EM10 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 varia-
tion of particle size and anisokinetic sampling are kept within  predetermined
limits.  If the range of duct velocities would cause the limit  on anisokinet-
ic sampling error to be exceeded by a full traverse, the traverse is broken
into two or more subtraverses with different sampling nozzles.   The number of
traverse points is selected to reduce errors due to spatial variation of
emissions to acceptable levels while allowing sufficient time for necessary
operator decisions between sampling points.  In order to provide proper
averaging of emissions, the dwell time at each traverse point is proportional
to the local duct velocity.  Current specifications were designed to limit
error due to spatial variations to 10%.  The maximum allowable  error due to
anisokinetic sampling is ±20% for 10 urn particles; in essentially all
sampling situations, cancellation of sampling error, and much smaller
contributions for particles smaller than 10 urn will limit overall anisoki-
netic sampling error to much less than this value.

     The sampling device described in this manual is Cyclone I  of the SRI/EPA
five-stage series cyclone.  This device provides a 10-um size cut at a flow
rate of approximately 0.5 dscfm; the precise flow rate depends on local stack
conditions.
                                      IV

-------
                                   CONTENTS
Disclaimer	•	ii
Foreword	   iii
Abstract	   iv
Figures	viii
Tables	ix
Symbols 	   x

   1.  Introduction	    1
   2.  Operating Principles 	    3
       2.1  Error due to anisokinetic sampling. ............    3
            2.1.1  Measurement of FM10 emission rate	    4
            2.1.2  Measurement of PM, Q fraction	    6
       2.2  Error due to spatial variation	    6
       2.3  Sampling hardware 	 	    7
            2.3.1  Performance determination for PM^Q
                   samplers—cyclones 	   10
            2.3.2  Performance specifications for PMj0
                   samplers—cyclones 	   15
            2.3.3  Performance determination for PM1Q samplers—
                   cascade impactors	  .   16
            2.3.4  Performance specifications for PM1Q samplers—
                   cascade impactors. .	18
            2.3.5  SRI/EPA Cyclone 1	19
            2.3.6  Sample nozzles 	  ......   21
   3.  Calibration	24
       3.1  Flow metering system	  .   24
       3.2  Pitot tube	24
       3.3  Sampling nozzles	26
       3.4  Thermocouples	26
       3.5  Magnehelic gauges 	   26
       3.6  Support equipment	  .   26

   4.  Presampling Activities 	   27
       4.1  Equipment calibration and checks	27
       4.2  Preparation of sampling  reagents	28
   5.  Sampling Parameters	30
       5.1  Preliminary measurements	31
            5.1.1  Method 1	31
            5.1.2  Methods 2, 3, and 4	31
       5.2  Sample flow rate and nozzle  selection for a single stage
            cyclone sampler 	   31
            5.2.1  Sample flow rate	31
            5.2.2  Nozzle selection  	   33

                                       V

-------
                         CONTENTS  (Continued)
    5.3  Sample flow rate anc" nozzle selection for cascade  impactot
         samplers	   35
         5.3.1  Sample flow rate	35
         5.3.2  Nozzle selection 	   35
    5.4  Sample orifice AH	  .  .  .   36
    5.5  Dwell time	36
    5.f  Performing the calculations 	   37
6.  Taking the Sample	42
    6.1  Field assembly	42
    6.2  Leak test	42
    6.3  Pretest equipment warm-up 	   43
    6.4  System start-up	43
    6.5  Traversing	44
    6.6  Shutdown orientation	  .   44
    6.7  Data logging	44
7.  Sample Retrieval 	   46
    7.1  Recovery of the participate mass	46
    7.2  Moisture determination	47
8.  Post sampling Checks	,	48
    8.1  Equipment calibration checks	48
    8.2  Sample analysis	48
9.  Data Analysis. . .•	50
    9.1  Velocity-weighted average 	   50
    9.2  Dry gas meter volume	50
    9.3  Volume of water vapor ..... 	  	   50
    9.4  Percent moisture content	51
    9.5  Stack gas molecular weight	51
    9.6  Stack gas viscosity	51
    9.7  Sample flow rate	51
    9.8  Sampler Dgg . .		52
    9.9  Stack gas velocity	52
    9.10 Nozzle velocity	52
    9.11 Nozzle AP limits	52
    9.12 Percent isokinetic sampling 	   53
    9.13 Concentration	 . .	53
    9.14 Acceptance criteria	53
10. Maintenance	54
    10.1  Vacuum system. 	 .................   54
    10.2  Vacuum pump	54
    10.3  Magnehelic differential pressure gauges	   55
          10.3.1  Zero adjustment	55
          10.3.2  Calibration check	55
          10.3.3  Recalibration	55
                                  VI

-------
                            CONTENTS  (Continued)
       10.4  Dual manometer	56
       10.5  Pitot tube	57
       10.6  Nozzles	57
       10.7  Thermocouples	.  .  .   57
       10.8  Sampling probe	57
             10.8.1  Probe cleaning  ......  	   57
             10.8.2  Probe heater check 	   58
       10.9  Condensing system	58
   11. Auditing Procedures	59
   12. Recommended Standards for Establishing Traceability.  ......   60

References	61
Glossary of Symbols	  .   63
                                     VII

-------
                                   FIGURES
Number

 2-1   Velocity ratio R versus duct velocity  giving aspiration
       coefficient of 1.2 and 0.8 for particles with aerodynamic
       diameter of 10 um at various gas viscosities	
 2-2   PM.Q particulate sampling train with out-of-stack filter
        (analogous to Method 5)	    8

 2-3   PMin particulate sampling train with in-stack filter
        (analogous to Method 17)	    9

 2-4   Calibration system for heated aerosols 	   11

 2-5   Efficiency envelope for PM1Q sampler (cyclones)	16

 2-6   Efficiency envelope for PM1Q sampler (cascade impactors)  ....   19

 2-7   Cyclone I dimensions	20

 2-8   PM1Q nozzle design . .	   22

 2-9   Optimum inside cone angle, 8, for various nozzle inlet diameters   23

 5-1   Minimum number of traverse points for CSR traverse 	   32

 5-2   CSR worksheet I - sample flow rate and orifice AH calculations  .   38

 5-3   CSR worksheet II - nozzle selection	   39

 5-4   CSR worksheet III - dwell time calculations	40

 6-1   CSR field data sheet	45
                                   viii

-------
L
                                                 TABLES
              Number                                                                  Pags


               2-1   Particle sizes and nominal gas velocities for efficiency
                     performance tests of cyclones	15

                                                                          \
               2-2   Performance specifications for source PM,. samplers—cyclones.  .   15


               2-3   CSR nozzle diameters and velocity limits 	   21


               2-4   PM10 nozzle geometries	23
                                                   IX

-------
                                     SYMBOLS
A            Aspiration coefficient  (measured  concentration/actual
             concentration)
AJ           Total cross-sectional area of  the jet{s)  of the PM10  st-.age
AJJ           Cross-sectional  area of  sampling  nozzle,  ft2
B             (2 + 0.617/RJK
B            Water fraction of  stack  gas
C            Cunningham slip  factor
C            Concentration of particulate,  ag/dNra3
C'm          Concentration of particulate matter, gr/dscf
C_           Pitot tube coefficient  for the CSR probe
C^           Pitot tube coefficient  for Method 2 probe
d            Nozzle diameter
d.           Diameter or width  of one jet of the PM, »  stage
d^           Diameter of nozzle n, in.
D            Particle aerodynamic diameter, yra
D50          Cyclone cut diameter, ym
IT            Velocity-weighted  average of generic data terra, D
D-           Average value of generic data  term, D,  for  each nozzle size
D,,           Stack differential pressure, in.  H20
 ^stack                                         *
^           Collection efficiency for PM10  stage n
Es           Average Egara
Esam         Sampler-only collection  efficiency
Et           Average Etot
Efcot         Total (nozzle and  sampler) collection efficiency
f10          PM10 fraction
fc           Fraction C02
f            Fraction O2
I            Percent isokinetic sampling
K            Particle Stokes number with respect to  the  nozzle, iv/d
             85.48 ft/s (lb/mole-*R)
             Dry molecular weight of  stack  gas,  Ib/mole
Met          Mass of dye collected in the sampler exit tube
Mf^         Mass of dye collected in the backup filter
Mjj Q         Molecular weight of water, 18  Ib/roole
M^2          Mass of dye collected on PM10  stage n
Mn           Mass of dye collected on all surfaces downstream of PM10  stage n
M            Mass of dye collected in the sampling nozzle
             Mass of collected  particulate  (either per stage or total), mg
 sam         Mass of dye collected in the sampler body
MW           Wet molecular weight of  stack  gas,  Ib/mole
n            Number of component changes
P            Ambient pressure,  in. Hg

-------
 ST
5S
Sr
R
Re
R
 u
S
Stk
Stk50
 M
 ST
u
v
V.
 rain
 max
 M
 MS
 WS
Absolute stack pressure, in. Hg
Absolute pressure at standard conditions, 29.92  in. Hg
Flow rate through «-he sampler  (sampler conditions) , acfra
Sample flow rate  (standard conditions), ft3/rain
Ratio of stream velocity to nozzle velocity
Reynolds number
Ideal gas constant, 21.83 in. Hg-ft3/mole-*R
Standard deviation
Stokes number, fv/d, dimensionless
Stokes number giving 50% collection efficiency
Dwell time at the first traverse point, min
Run time for each nozzle size
Absolute gas meter temperature, *R
Absolute stack gas temperature, *R
Absolute temperature at standard conditions, 528  *R
Nozzle velocity, ft/s
Stream velocity, ft/s
Total volume of liquid collected in impingers or  condenser/silica
gel, mL
Minimum allowable stack velocity to be sampled with a given  nozzle
diameter, ft/s
Maximum allowable stack velocity to be sampled with a given  nozzle
diameter, ft/s
Volume of gas sample flow through the dry gas meter (meter
conditions), ft3
Volume of gas sample flow through the dry gas meter (standard
conditions), ft3
Stack gas velocity, ft/s
Volume of water vapor in the gas sample (standard conditions),  ft3
                                      XI

-------
v            Gas meter calibration constant
AH           Orifice pressure drop, in. H20
AH           Orifice pressure differential for a flow rate of 0.75 cfm at
             standard conditions, in. H2O
AP           Velocity pressure head, in. H2O
&?!          The velocity AP at the first traverse point, in. H2O
 (/AT)        Average square root of velocity APs, used to calculate average
     avg
             stack velocity
AP           Maximum allowable velocity pressure to be sampled with a given
  a ax
             nozzle diameter, in. H2O
AP .          Minimum allowable velocity pressure to be sampled with a given
  Bin
             nozzle diameter, in.
AP           The velocity AP at traverse point n, in. H,0
  n                                                    *
AP           Velocity pressure drop of standard pi tot tube, in. H. O
9            Total run time, rain
M            Gas viscosity, micropoise
j            Density of water, 1 g/mL
 Ho O
a            Gecsetric standard deviation of lognormal distribution
 9
T            Particle relaxation time CD2/18u
*            Outside taper or cone half-angle for a sampling nozzle
$            Inside taper or cone half-angle for a sampling nozzle
            0.5 Stk
*50          °'5 Stk50
                                      XI 1

-------
                                  SECTION•1

                                 INTRODUCTION
     To ensure 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 *10 Urn (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 hardware such as exhaust gas recycle  (EGR),  a PM10
sampling method must then become a compromise between the conflicting
requirements of inertial particle sizing and representative sampling.

     This manual presents a method. Constant Sampling Rate  (CSR),  which
allows determination of PM10 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% for 10 urn particles; in essentially all
sampling situations, cancellation of sampling error  and much  smaller
contributions for particles smaller than 10 um will  limit overall
anisokinetic sampling error to much less than this value.  This  method has
been identified in some previous EPA documents as  the Simulated  Method 5
(SIM-5) or Constant Flow Rate (CFR) method.

     Fundamental to a method of this type is the device used  to  perform the
particle sizing required for PMio collection.  The device used in  development
of .this method was Cyclone I of the SRI/EPA five-stage series cyclone

                                    1

-------
 (Smith et al., 1979).  For this reason, in some portions of this manual,  such
as Section 7, Sample Retrieval, instructions are given specifically  for this
device.  Another single stage cyclone could be used if its performance has
been adequately characterized as specified in Section 2.

     This does not mean the use of other devices, such as cascade impactors,
should be discouraged.  In recognition of the fact that cascade impactors may
often be used as the PM.Q device, key sections of this manual address the use
of these devices.  Specific instructions concerning performance determination
and specification for cascade impactors are given in Section 2.  In Section
5, instructions for choosing the sample flow rate and nozzle(s) are given.
Once these parameters are fixed, sampling procedures are the same as with a
single-stage sampler.  This manual does not address sample retrieval, stage
size cutpoint determination, or interpolation of data for multiple stages to
determine PM10.  Acceptance criteria for size outpoints, once they are
determined, are given in Section 9.  It is beyond the scope of this manual to
provide complete operating instructions for impactors.  Numerous manuals  are
available which provide information in this area.  Unfortunately, these
manuals also, reflect a varying degree of completeness.  The impactor manual
chosen for use in conjunction with this CSR manual should include discussions
of how anomalous behavior, such as particle bounce, may be avoided, the
effects of jet-to-plate spacing, and the variation of size cut with Reynolds
number.  A recommended manual is the "Procedures Manual for the Recommended
ARB Particle Size Distribution Method (Cascade Impactors)," by McCain et  al.
 (1986).

     This manual is organized so that the user is guided step-by-step through
initial use of the method in a field situation.  Section 2 describes the
basic principles from which CSR was developed and the specifications for  the
critical sampling hardware.  Section 3 describes the procedures by which
various components of a CSR system may be calibrated.  Activities which are
required or recommended prior to field use of the method are outlined in
Section 4.  Section 5 describes the calculations required to establish the
sampling parameters prior to sampling, and Section 6 outlines the steps to
follow during operation of a sampling system.   Retrieval of the collected
sample is described in Section 7.  Sections 8 and 9 describe required
postsampling checks and analysis of the field data.  Routine maintenance of a
CSR system is discussed in Section 10.  Auditing procedures and recommended
standards are described in Sections 11 and 12.

-------
                                  SECTION  2

                            OPERATING PRINCIPLES
     Many different approaches to resolving the dilemma posed by  the
conflicting requirements of fixed flow rate and representative  sampling  may
be suggested.  One approach used frequently in the past for control device
evaluation involved operating the sampler at a sufficient number  of points  in
the sampling plane such that the dominant error source would be anisokinetic
sampling bias.  Because the concentrations of the fine particles, which  were
the particles of greatest interest, could be measured correctly without
sampling isokinetically, this mode of operation was acceptable  for the
purpose.  However, sampling in this manner did not ensure any level of
accuracy was to be obtained in the measurement of overall size  distributions.
Another approach was to operate the sampler at near-isokinetic  conditions at
a single traverse point.  In this case, the errors due to spatial variation
in the particle size distribution would be large unless a full  traverse  of
the sampling plane was synthesized from a number of these single-point
measurements.  This would be very costly in terms of on-site sampling time.

     In developing the CSR strategy, a compromise between the options
described above was sought.  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  fluctua-
tions in source emissions or basic measurement inaccuracy).  Finally,
measurements would be made to provide an average representative of emission
rates rather than concentration.

2.1  ERROR DUE TO ANISOKINETIC SAMPLING

     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  selec-
tively depleted.  Conversely, if the gas velocity in the nozzle is lower than
the local duct velocity, the flux of large particles through 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.
                                     3

-------
     For any given particle diameter,  the  anisokinetic  sampling  error may be
expressed as the aspiration coefficient, which  is defined  as  the ratio of
measured concentration to  actual concentration.  Belyaev  and  Levin (1974)
developed a semi-empirical relationship for aspiration  coefficient,  A, in
terms of the particle Stokes number, K, and the  ratio of  stream  velocity,  v,
to nozzle velocity, u.

                            A = 1 +  (R - 1) B *  1                       (2-1)

where R  » velocity ratio v/u
      B  »•  (2 + 0.617/R)K
      K  = particle Stokes number with respect  to the nozzle,  TV/<3n
      T  = particle relaxation time CD^/ISu, seconds
      C  * Cunningham slip factor
      D  * particle aerodynamic diameter
      u  a gas viscosity, poise
      d  * nozzle diameter
       n

     Equation 2-1 shows that, for given stack conditions  and  particle size,
limiting the ani&okinetic sampling error becomes a question of limiting the
velocity ratio, R.  In other words, for a given  limit on error due to aniso-
kinetic sampling, maximum and minimum  values of  R  (\iax and R^^  will yield
results within the stated limits.  It  may  also be noted from  equation 2-1
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 EMjo 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.

2.1.1  Measurement of PM1Q Emission Rate

     For the purposes of PM1(., limits  of ±20% on error  due to  anisokinetic
sampling were chosen to specify the limits on the velocity ratios, R_in and
Rmax.  Solving equation 2-1 for R as a function  of stream  velocity,  v,
viscosity, u, sample flow rate, Q, and particles with aerodynamic diameter of
10 urn, the series of curves shown in Figure 2-1  are defined.   The  upper
curves give Rmax and correspond to the upper limit on A (1.2).  The  lower
curves give RJnin and correspond to the lower limit on A (0.8).  As can be
seen from the figure, the limits on R  are broad  at the  smaller stream
velocities and approach limits of ±20% at high stream velocities.

     Actual sampling error for most sources will be less than  the  ±20% limit
foe two reasons.  First, point-by-point R values will usually  be something
less than the limits, Rmjn and Rfflaxr and cancellation of errors  will occur

-------
  0.5
3    4    56789 10
              VELOCITY, m/sec
                                                            20
30   40   50 60
Figure 2-1. Velocity ratio R, duct velocity/nozzle velocity, versus duct velocity giving aspiration
           coefficient of 1.2 (upper curves) and 0.8 (lower curves) for particles with aerodynamic
           diameter of 10 n/n and gas viscosities ranging from 160 to 280 x 1Q-6 poise. Sampling
           error of 10 fim particles is less than ± 20% for R-values between an upper and a lower
          curve.

-------
because R-1 values will be negative  at  some points  and  positive at  others.
Second, the curves shown in Figure 2-1  *ere determined  by  assuming  a raono-
disperse sample of 10-Ura particles.   In actuality,  the  FM10  sample  is
composed of particles with aerodynamic  diameters less than 10 vim.

2.1.2  Measurement of PH,n Fraction

     Limits on the velocity ratio, R, for measurement of the PM, Q fraction  of
total emissions are determined  in a  slightly different  manner than  those  for
the PM.g emission rate.  This is necessary because of two  important differ-
ences between the two parameters.  First, because anisokinetic  sampling
introduces both positive and negative errors to the two size classifications
of interest (particles less than 10  urn  in diameter  and  particles greater  than
10 um in diameter), there is partial cancellation of the error  when the ratio
is taken to calculate the PM10  fraction.  Second, because  there is  no upper
limit on particle size for the  portion  greater than 10  urn,  the  ratio B/(B+1)
in equation 2-1 must be assumed to be unity for this size  fraction  when
determining the limits on sampling error.

     For a particular source, the limit on percent  error in  measurement of
PMjg fraction (f10) is approximately the quantity

                            (R  - 1)(1 - f1Q) x 100

where R is the minimum or maximum value of velocity ratio  experienced during
the sampling run.  Limits for acceptable values of  R are determined by
setting the error to ±10%.  The limits  then become
                  *min - 1 - T-T~ -* "max "  ' + TT^            I2'2'

Determination of these limits requires an estimate of the PM1Q  fraction,  so
it may be necessary to repeat some runs because  of poor  estimates  of  the
limits as a result of incorrect values for f.Q in conjunction with substan-
tial variation of velocity across the sample plane.

2. 2  ERROR DUB TO SPATIAL VARIATION

     The goal of a traversing strategy should be to reduce  error due  to
spatial variation so that it is not the dominant source  of  error but  is
comparable to or less than other sources of error.  Extensive analyses  have
been made of the uncertainty in measurements of  total particulate  emission
caused by spatial variation in process streams.  The analysis presented by
Shigehara (1982) suggests that 8 to 12 traverse points provide  a large  reduc-
tion in error from one point, whereas additional points  provide little  addi-
tional decrease.

     In general, the PM1Q fraction is expected to be less stratified  than
total particulate emissions.  The lower inertia of the m10 fraction  would
cause less deviation from gas flow in bends and  faster damping  from turbulent
diffusion once stratification is produced.  On the basis of available FM10
profiles. Farthing (1983) concluded that Shigehara's data may be used as  a

-------
conservative estimate of error due  to  spatial  variation  in  PMjn  measurements.
Using the 8 to 12 traverse points specified  in the CSR protocol  is  expected
to reduce this type of error to less than ±10%.

2.3  SAMPLING HARDWARE

     As stated previously, one of the  objectives during  the development  of
the CSR technique was to make use of existing  sampling equipment.   Therefore,
the hardware changes required to operate a CSR sampling  system have been kept
to a minimum.  Like a standard total particulate sampling train,  a  CSR system
nay be operated with an out-of-stack filter  (analogous to Method  5)  or an
in-stack filter  (analogous to Method 17).  Block diagrams of  the  systems are
shown in Figures 2-2 and 2-3, respectively.  If the CSR  train is  operated
with an in-stack filter, a heated probe is not required, but  steps  should be
taken to prevent condensed moisture from running back into  the sampling
device and contaminating the sample.

     As can be seen from Figures 2-2 and 2-3,  the apparent differences
between CSR and standard total emissions hardware are confined to the  front
end of the probe.  There are some less obvious differences, however.   The
sampling flow rate at which a CSR train is operated is set by the requirement
that the sample be segregated into two or more specified size fractions.  In
addition to stack conditions, this flow rate is dependent on  the  geometry and
characteristics of the sampling device.  In other words, different  sampling
devices may have significantly different sampling flow rates.  In the  case of
SRI/EPA Cyclone I, described later in  this section, the  flow  rate which
provides a 10-vm cut is approximately  0.50 dscfm.  This  is considerably  lower
than the Method 5 or Method 17 nominal flow rate of 0.75 dscfm.   To control
this flow rate with a reasonable degree of accuracy, a smaller orifice than
that typically used for Method 5 (0.180 in.) will be required.  Experience
has shown 0.130 in. is a practical diameter for the sample orifice  when  a CSR
train is operated with a SRI/EPA Cyclone I.  However, this may not  be  an
appropriate size for a different sampling device for the reasons  given
previously.  To minimize uncertainty in choosing the sample orifice size or
to prepare for operating the sampling  train with a number of  sampling
devices, a set of orifices should be acquired  with the following  diameters:
0.180 in./ 0.130 in., and 0.093 in.

     As Figures 2-2 and 2-3 show, the primary  difference between  CSR and
Methods 5 or 17 hardware is the PMjq sampling  device.  Although  a number of
particle sizing devices may be considered for  use as a PM,, 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
reentrainment keep this from being an  acceptable choice.  The multistage
device recommended for use with this method is a Cascade impactor.   Although
a series cyclone such as the SRI/EPA five-stage series cyclone provides
particle 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.

-------
                        HEATED PROBE
                    SIZE
                 SEPARATOR
                                                                IMPINGER TRAIN OPTIONAL:
                                                                MAY BE REPLACED BY AN
                                                                EQUIVALENT CONDENSER
                                                  HEATED
                                                  AREA    FILTER HOLDER
                                                                         THERMOMETER
                                        ORIFICE
                                                   THERMOMETERS     BYPASS
                                                                     VALVE
CHECK
VALVE
                                                                 IMPINGERS IN ICE BATH
                                                                   O<>r*-l>O-
                                                                        MAIN      VACUUM LINE
                                                                        VALVE
                                    MANOMETER     DRY TEST METER   AIRTIGHT PUMP
                                                                                         4111-C4SB
                       Figure 2-2. PM-JQ paniculate sampling train with out-of-stack filter (analogous to Method 5).
l_

-------
                                                IMPINGER TRAIN OPTIONAL:
                                                MAY BE REPLACED BY AN
                                                EQUIVALENT CONDENSER
   PM10
   SIZE
SEPARATOR
                                                                          CHECK
                                                                          VALVE
                                                 IMPINQERS IN ICE BATH
                                  THERMOMETERS      BYPASS    	            '
                                   ^        _       VALVE   ' /^\ VACUUM
                       ORIFICE     O	  Q    T     T  ^ GAUGE
                                                                            '
                                                        MAIfl      VACUUM LINE
                                                        VALVE
                   MANOMETER     DRY TEST METER   AIRTIGHT PUMP
                                                                         UI1-445A
        Figure 2-3. PM JQ paniculate sampling train with in-stack filter (analogous to Method 17).

-------
     Before any particle sizing device  is  used  as  a  PM10  sampler,  it must be
shown to meet some specific performance  requirements.  The  procedures by
which this itay be done  are discussed  in  the  following  sections  for  both
single-stage and multistage devices  (cyclones and  cascade impactors, respec-
tively) .  The specific  performance requirements  for  each  of these devices are
also given.

2.3.1  Performance Determination for  PM,n  Samplers—Cyclones

     To determine that  a given cyclone meets the requirements for use as a
PMig device, the performance determination procedures  outlined  in the follow-
ing text must be performed.  The objectives of these procedures  are twofold:
(1) to calibrate the cyclone  (i.e., establish the  relationship  between
collection efficiency,  flow rate, gas viscosity, and gas  density for the
given device) and (2) to determine that  cyclone performance satisfies the
performance specifications with the sampling nozzles used in practice.

2.3.1.1  Particle Generation—
     The particle generating system used for the performance determination of
the sampler must be capable of producing solid, raonodisperse dye particles
with mass median aerodynamic diameters ranging  from  5  to  20 urn.  The geomet-
ric standard deviation  (a ) for each  particle size should not exceed 1.1.
Furthermore, the proportion of raultiplets  and satellites  should  not exceed
10% by mass.

     The size of the solid dye particles delivered to  the test  section of the
wind tunnel should be established by  using the operating  parameters of the
particle generating system.  This should be verified during the  tests by
microscopic examination of samples of the  particles collected on a  membrane
filter.  The precision of the particle size verification  technique  should be
0.5 um or better, and the particle size  determined in  this  manner should not
differ by more than 10% from that established by the system operating parame-
ters.

     The monodispersity of the particles should be verified for  each test by
either microscopic inspection of particles collected on filters  or  monitoring
techniques such as an optical particle counter followed by  a multichannel
pulse height analyzer:  It is preferable that verification  of acceptable
particle size distribution be performed  on an integrated.sample  obtained
during the sampling period of each test.   As an alternative, samples obtained
before and after each test may be used to  verify the size distribution.

     To determine cyclone behavior as a  function of yas conditions,  the
system must be operated at a range of temperatures.  The  dye particles must
withstand temperatures  from 22 *C (70 *F)  to 200 *C  (400  *F) without signifi-
cant change in size, density, or spectral  properties in solution (associated
with measurement of collected particulate  mass).   Ammonium  fluorescein
(available from a number of sources)   has been shown to meet these require-
ments for temperatures up to S50 *F and  Pontamine  Fast Turquoise 8GLP (avail-
able from B.I. DuPont de Nemours and  Company) has  been shown to  meet them up
to 400 "F  (Smith et al., 1979).  However,  the thermal  integrity  of  each dye
batch should be verified.

                                     10

-------
     The  requirements of the apparatus  for heating the monodisperse dye
aerosol axe  illustrated in Figure 2-4.   A pump with an orifice  or  other flow
aeter  is  used  to obtain the test  flow  rate through the cyclone.   The combina-
tion of absolute filter/bleed valve  allows excess aerosol  to  escape or addi-
tional air to  enter  as needed.  The  aerosol stream from the generator passes
through a copper tube heated to attain  the test temperature.  The  heat trans-
fer rate  and uniformity of heating should be sufficient for the aerosol to
attair. the t.*st  temperature but should  not cause the temperature of any
interior  surfaces to rise above the  teroerature used for verifying the
integrity of the dye.  The inlet tube  to the cyclone must  have  th»» same
inside diameter  as the inlet diameter of the cyclone.  This tube must be
cleaned between  runs and blanks performed to check for possible effects of
reentrainment  of particles which accumulate on its interior.  The  sample port
is necessary to  collect and examine  heated -particles for correct size, color,
and shape for  each measurement run.
 AEROSOL STREAM
 CROM VIBRATIMG
 ORIFICE AEROSOL
 GENERATOR
ABSOLUTE FILTER
                                         OVEN
                                         KEPT AT
                                         AEROSOL TEMPERATURE
                                              U
                                              il
                                                     MEAT |
                                                     EXCHANGER
        THERMOCOUPLE
                      SAMPLING
                      PORT
                                                         MERCURY    WATER
                                                         MANOMETER  MANOMETER
                                                                        4181 92
                     figure 2-4. Calibration system for heated aerosols.
                                     11

-------
2.3.1.2  Wind Tunnel--
     A portion of the collection efficiency  tests  must  be  performed under
isokinetic sampiirvg conditions  in a wind tunnel or  similar  apparatus so that
the  effect of the sampling nozzle on  the cyclone performance  may be deter-
mined.  This apparatus must be  capable of establishing  and  maintaining
(within 10%) velocities ranging from  7 to 25 ra/s.

     The velocity of the wind tunnel  gas stream in  the  vicinity  of  the
sampling nozzle should be measured by using  an appropriate  technique capable
of a precision of 5% or better  and of a spatial resolution  of  1  cm  or less
(e.g., hot wire aneraometry or miniature pitot tubes).   The  velocity should be
constant within 10% over the inlet area of the largest  sample  nozzle to be
used with the PM,0 sampler.  If the sampler obstructs more  than  10% of the
wind tunnel cross-sectional area, the velocity uniformity must be demon-
strated by velocity measurements with the sampler  in position  and operating.

     Poc each efficiency test,  the gas stream velocity  should  be determined
at the beginning of each test and maintained within 10% of  the set  value by
using a suitable monitoring technique with precision better than 5%.

2.3.1.3  Cyclone Calibration —
     To achieve the first objective of the performance  determination (estab-
lish the relationship between collection efficiency, flow  rate,  and gas
conditions) the following procedures  should  be followed.

     The operator should establish operation of the particle generator and
verify particle size microscopically.  If monodispersity is to be verified by
measurements at the beginning and end of the measurement run  rather than by
an integrated sample, these measurements should be  performed  at  this time.
Plow should be initiated through the cyclone at the test value after stable,
correct operation of the generator is established.  The operator should
sample long enough to obtain ±5% precision on total collected  mass  as deter-
mined by precision and sensitivity of the measuring technique.  Immediately
after completion of sampling, the size of the aerosol particles  should be
verified microscopically.

     The sampled particulate mass is determined by  a suitable  technique
(fluorittetry or absorption spectrophotometry for ammonium  fluorescein).  The
mass collected in the nozzle, PMJO sampler body, PMJO sampler  exit  tube, and
backup filter (*noz> MSiB' Met' Mfil' resP«ctively) ls  determined separately.
2ach separate surface must be rinsed  with an adequate amount of  an  appro-
priate solvent to dissolve the collected dye particles,  and care must be
taken not to contaminate the rinses with dye from other  surfaces.   Sufficient
solvent must be added to each rinse until the rinse volume  is  suitable for
Measurement or calculation.  The mass of dye in the rinses  is  to be deter-
mined from the spectroscopic measurement by using appropriate  blank and
standard solutions for reference and quality control.
                                    12

-------
The total  (nozzle  and  sampler)  and  sampler-only collection efficiencies (Etot
and Egam) may be calculated  from the  following  equations:

           •tot "  100S x  (Mnoz  + Msam>/       <2'3'


                   Esam "  100%  x Msam/               <2"*>

At least two replicates of the  above  steps should  be performed.

The average efficiency should be calculated and recorded as

                         n                      n
                         I   Etot(i)            I  Esam(i)
where Efc fc(i) an^ Esam^ represent  individual Efc  fc  and E     values and n
equals the number of replicates.

The standard deviation  (S) for the replicate measurements  should  be calcu-
lated and recorded as
                                                     0.5

                           .T   E2(i) -  ( I E(i))2/n'
                                    n - 1

where E(i) represents Etot(i) and Egam(i)
                                                                      (2-6)
For n « 2, S »  [E(1) - E(2)]//7.   If the value of S  for  Etofc exceeds 10% of
Et, the test run must be repeated.

     The size cut, D^Q, of the cyclone  is established by either  of two sets
of measurements.  In one set, operating conditions are  adjusted  to obtain a
collection efficiency, Eg, of SO ±5% for a single particle  size.   Three
replicate runs should be performed with the  actual particle size for each run
within ±5% of the desired value.   In the other set,  E   is measured with at
least three particle sizes at the  same operating conditions, and linear
interpolation in log-probability space  is used to determine the  05 0-  In the
latter set, the measured Eg valu ----- - •-- " " ----- ""    " """    "
values both below and above 50%'.
latter set, the measured E_ values must be between 20 and 80% and  include
                          3
2.3.1.4  PM1Q Plow Rate-
     To determine the empirical relationship between  FN^Q  flow  rate  and gas
conditions, the D^0 determination described above must be  performed  for at
least three temperatures.  The DJQ'S must be between  5 and 15 urn and measured
at temperatures within 60  *C  (108 *F) of the temperature at  which the cyclone
will be used.  In addition, one of the measured DSQ'S must be  10 ± 0.5 urn.
                                     13

-------
     Linear  regression  analysis  is  used  to  determine  the relationship between
the dinensionless parameters *SQ  (=  0.5  Stkcg)  and  Re,  where Stk50  is the
Stokes number giving  50% collection  and  Re  is  the Reynolds number of the gas
entering the cylcone.
and
where Q » gas flow rate through the cyclone  at the  inlet  conditions
      u » gas viscosity
      d * diameter of the cyclone  inlet
      p * gas density at the cyclone  inlet

Vfith the substitution of D5Q =• 10  urn  into the resulting relationship, the
flow rate for PM10 measurements is predicted as  a  function of gas conditions.

2.3.1.5  Determination of Cyclone/Nozzle Collection Efficiency —
     Because the cyclone and sampling  nozzles are  used  as a unit in actual
sampling situations, it is necessary  to establish  that  the nozzles do not
perturb the particle sizing characteristics  of the cyclone as determined by
the calibration procedures discussed  previously.   To do this, collection
efficiency tests should be performed  for the cyclone/nozzle unit by using the
particle diameters and gas velocities  shown  in Table 2-1.   For the appro-
priate PH10 sampler flow rate, the operator  should determine the nozzle size
appropriate for isokinetic sampling in each  of the three  velocity ranges
shown in the table.  If more than  one  nozzle is  suitable  for a range, the
larger nozzle may be chosen.

     After the three nozzle sizes  have been  determined, the first airstream
velocity to be tested in the wind  tunnel should  be established and verified
as described previously.  The particle generating  system  should then be
started and the particle size distribution verified.  The particle size, as
determined by the system operating conditions, must be within the tolerances
specified in the table.  The operator  should begin sampling by establishing
the flow rate required for a lO-ym D   in the cyclone.
     At the completion of the runs, the total and sampler-only collection
efficiencies  (Etot and Bg—B) should be determined from  equations 2-3  and 2-4.
Pot each of the three gas stream velocities tested, the average E^ and Bg
should be plotted as functions of particle size  (0) on  semi log paper.   Smooth
curves should be drawn through all sizes used.  The 050  for Ea should  be
defined as the diameter at which the Eg curve crosses 50% efficiency.
                                    14

-------
           TAPLE 2-1.  PARTICLE SIZES AND NOMINAL GAS VELOCITIES  FOR
                       EFFICIENCY PERFORMANCE TESTS OF CYCLONES
                                      Target uas velocity (a/a;	
           Particle Size  (um)a     7±1.0      1511.5      25 ±2.5
5 ±
7 ±
10 ±
14 ±
0.5
0.5
0.5
1.0
               20 ± 1.0
           aMass median aerodynamic diameter.
            Number of test points  (minimum of two replicates  for each
            combination of gas velocity and particle size) :   30.
2.3.2  Performance Specifications for PM   Samplers - - Cyclones
     The performance specifications for a PM1Q cyclone are shown  in
Table 2-2.  To be acceptable for use, the %0 for the sampler  (determined
from the Eg curve as described previously) must be 10 ± 1 urn.   In addition,
all data points used to determine the Et curves for each of the gas  stream
velocities tested must fall within the banded region shown in  Figure 2-5.
The portion of the acceptance envelope corresponding to large  particles  is
bounded by a vertical line at 12 yra, a horizontal line at 90%  efficiency, and
a log normal function (oblique line) with geometric standard deviation (o ) of
1.7 and 50% efficiency at 12 urn.  The boundary at small particle  sizes has a
vertical line at 8 urn and horizontal lines and lognormal functions which vary
between three ranges of gas stream velocity.  These horizongal  lines are at
10, 20, and 30% efficiency, respectively, with increasing gas  velocity.  At
the lowest range of velocity the lognormal function has a  of  1.7 and 50%
efficiency at 8 urn.  For the two higher velocity ranges, the lognormal func-
tions have 55% efficiency at 8 urn and values of a  of 2 and 2.9 respectively,
with increasing gas velocity.
  TABLE 2-2.  PERFORMANCE SPECIFICATIONS FOR SOURCE PM10 SAMPLERS—CYCLONES


          Parameter              Units              Specification

  1. Collection Efficiency         %        Such that collection efficiency
                                            falls within envelope specified
                                            in Figure 2-5

  2. Sampler 50% cut point        urn        10 ± 1 urn aerodynamic diameter
     «>50>

                                    15

-------
          95


      8  "
      0  80
      2
      0  7°
      I  60
      Z  5°
      ?  40
       0
       0
          30

          20

          10

           5
                  17 < v < 27 m/s
                   9 < v < 17 m/s
                      v < 9 m/s
                       I
                                 I      III
                       2          4     6    8  10         20         40

                                AEROOWWflC OWOER. fn               „„.,,

                      2-5. Efficiency envelope for PM JQ sampler (cyclones).
2.3.3  Performance Determination for
                                           Samplers— Cascade Impactors
     Like those foe cyclones,  the  performance determination procedures  for
cascade impactors have  two objectives:   (1)  to calibrate the cascade impactor
(i.e., determine the ?tk5Q for each of  the  three PH1Q  stages)  and  (2) deter-
mine the relationship between  Stk50 and nozzle inlet diameter for  the first
     stage.
     The particle generating system  and  wind tunnel used to perform these
tests should be the same  as those  described previously for cyclones.  There
fore, further description of this  test equipment  will not be given here.

2.3.3.1  Impactor Configuration  for  Performance Tests —
     It is expected, but not necessary,  that the  complete impactor hardware
assembly will be used  in each of the sampling runs of these calibration and
performance determinations.  ?or these measurements, the first FMio stage
                                     16

-------
must be calibrated with  the nozzles  sampling  isokinetically in the wind
tunnel.  The  first P^ Q  stage consists of  the collection  substrate and all
upstream surfaces, up to and including the nozzle.   This  includes all preced-
ing impactor  stages which are not designated as  PM1Q  stages.

     The second and third PM1Q stages consist of each respective collection
substrate  and all upstream surfaces  up to  (but excluding)  the  collection
substrate  of  the preceding PM10 stage.  This includes intervening impactor
stages, which are not designated as  PM10 stages.  These stages must be cali-
brated with the collection substrate (impaction  surface)  of the preceding
PMig stage in place so that gas flow patterns existing in field operc*-ion
will be simulated.

     Each  of  the PM^ stages should  be calibrated with the type of collection
substrate, viscid material  (such as  grease) or glass fiber used in FM10
measurements.  It should be noted that most materials used as  substrates at
elevated temperatures are not viscid at normal laboratory conditions.  The
substrate  material used  for calibrations should  minimize  particle bounce yet
be viscous enough to withstand erosion or  deformation by  the  impactor jets
and not interfere with the procedure for measuring  the collected particulate
mass.

2.3.3.2  Impactor Calibration—
     Calibration is begun by sampling monodisperse  aerosols of sizes within
±67% of the cut size utilized in field PMjQ measurements.   Procedures
analogous  to  those described previously for cyclones ace  used.  However,
potential  interference of substrate  material  (grease or glass  fiber)
entrained  in  the rinsing process must be eliminated.

     The particulate mass collected  by PM. Q stage n,  N ,  and particulate mass
on all surfaces downstream of stage  n, Mn  , should  be determined.  Particu-
late mass  on  the upstream side of a  jet plate should be included with the
substrate  downstream of  it.  However, agglomerates  of particles that obvious-
ly were removed or reentrained from  an upstream  surface should be included
with the preceding substrate.  Collection  efficiency for  stage n, E , is
calculated as
                          Bn - 100%  x V^n * V*                     {2"9)

Equation 2-10 demonstrates the calculation of th*> diraensionless parameter *
(=0.5 Stk, where Stk is the Stokes number) for each measurement.
                               0.5 Stk - D2 - 2
(2-10)
where  0 * particle diameter
       Q » gas flow rate through the PM10 stage at inlet conditions
       u » gas viscosity
      Aj » total cross-sectional area of the jet(s) of the  PMic  stage
      dj • diameter or width of one jet of the PMio stage
                                      17

-------
        should be measured while varying Y  to empirically determine  ¥50*  the
value for 50% collection efficiency.  Particle bounce can cause  efficiency to
decrease at high, values of ¥ and thus pass  through 50%  at multiple values of
Y.  Therefore, the calibration data should  clearly indicate  the  value  of
¥50 for minimum particle bounce.  Impactor  efficiency versus ¥ with  minimal
particle bounce is characterized by a monotonically increasing function with
constant or increasing slope as V increases.

2.3.3.3  Determination of *50 versus Nozzle Diameter—'
     The *50 of the first PM10 stage can potentially decrease with decreasing
nozzle size.  Therefore, this stage must be calibrated with the  nozzle
sampling isokinetically in a wind tunnel or similar test apparatus.  These
nozzles should have diameters ranging from  the smallest to the largest likely
to be used in FM10 measurements.  Calibrations should be performed with
enough nozzle sizes to provide a measured value within ±25% of any nozzle
size used in M10  measurements, thus providing a set of f5Q-values for the
first PMjg  stage, needed for determining particle size cut in PMJO
measurements.  The ^50 value for any nozzle size is determined by
interpolation between calibration values.

     In addition to measuring collection efficiency versus Hf for determining
?5Q, calibration measurements should be performed for the first  stage  at
values of ? low enough that the efficiency of stage one, EL , is  between 20
and 30%.  This data point and all other data (efficiency and ¥)  available for
any nozzle diameter are relevant to the performance specification in the  next
section.

2.3.4  Performance Specifications for PM^ Samplers—Cascade Impactors

     As can be seen from Figure 2-6, the acceptance envelope for a candidate
MIO impactor stage is analogous to that for cyclones with the exception  that
the abscissa has been changed from aerodynamic particle diameter to  values of
square root of ^/*§o«  To determine if a given impactor stage meets  the
performance criteria, efficiency versus the square root of I/Ten should be
plotted.  For the first stage, the data plotted should be those  obtained  with
each of the nozzle diameters tested.  All data must fall within  the  banded
region of Figure 2-6 if the stage is to be  accepted.
                                    18

-------
        2
        z
        0
            95
            90
            80
        n   70
60
50
40
30
        0   **
        0   20
            10

             5
                   17 < v < 27 m/s

                    9 < v < 17 m/8
                       v < 9 m/a
             0.1
           0.2
0.4    0.6  OJ  1
                                                                  (211-11
              Figure 2-6.  Efficiency envelope for PM JQ sampler (cascade impactor).
2.3.5  SRI/EPA Cyclone I

     One sampling  device known to meet the performance  specifications given
previously  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 standardized
to the original design parameters, as can be seen in Figure 2-7.   Laboratory
calibrations  (Farthing and Williamson, 1985) have shown Cyclone I produces a
10-um E^0 at a flow rate of approximately 0.5 dscfra; the  precise  flow rate
will depend on local stack conditions.
                                     19

-------
                               CYCLONE I DIMENSIONS
                                     BOTTOM EXIT
                                             cup
                                    DIMENSIONS
         /em ± 0.02 \
         \in.*0.
-------
 2.3.6   Sample Nozzles
     The  anisokinetic error  limits for CSR are bro-
nozzles  for  a single traverse will seldom be necea^  enough that multiple
sources,  the temporal or  spatial variation of streAy.   However, at some
"to  require more than one  nozzle.  The frequency of: velocity is sufficient
size is not  sufficient for the complete traverse i^*sts in which one nozzle
of  nozzle sizes available.   Small increments for t^ Delated to the increments
the likelihood that  a nozzle size will be availably  nozzle sizes increase
near the  mean stream velocity.  In this case, |R - to give a nozzle velocity
possible  value; moreover,  the likelihood is then 1% I  is at its lowest
will exceed  the limits for the chosen nozzle.  Thi^ * that rtream velocities
increments simplifies the  raeas 'ements and improve^ ^3 ing snaller nozzle
                                                    ^he  accuracy.
     The  set of nozzle diameters found to provide ^
increments is shown  in Table 2-3.  The nozzle velo^uff icient number of
diameter  for the Cyclone  I PM1Q flow rate at the q(^y corresponding to each
the limits on stream velocity for acceptable use of *n stack conditions and
shown.  The  minimum  and maximum velocity variation^ *ach nozzle are also
when using each nozzle are given in the last two dv*hich may be encountered
full range of nozzle diameters shown in the table wWs of the table.  The
most instances. In  other  words, sampling situation U not be required in
tered which  require  the smallest or largest nozzle '"ill seldom be uncoun-
table.                                              ^ameters shown in the
     TABLE  2-3.  CSR  NOZZLE  DIAMETERS (in.)  AND VE
                                                     tTY LIMITS (ft/s)a
Nozzle
Diameter

0. 136
0.150
0.164
0.180
0.197
0.215
0.233
0.264
0.300
0.342
0.390

*Por dry

79
For gas
Nozzle
Velocity

100.9
82.9
69.4
57.6
48.1
40.4
34.4
26.8
20.7
16.0
12.3

Min.
Stream Min.
vel. vel.
(A-0.8) ratio

76.4
61.5
50.1
40.0
31.6
24.4
18.1
13.4
10.4
8.0
6.1

gas composition (%
°2 CO2
4 16
conditions
CO
1
(*R, in.

0.76
0.74
0.72
0.70
0.66
0.60
0.53
0.50
0.50
0.50
0.50

volume) and
H,0
0 06
H.O, and in
"V
Stream
vel. Max.
(A»1.2) vel.

-
102.8
86.8
73.0
61.8
52.9
46.0
37.3
30.5
23.9
18.4

.23
.24
.25
.27
.29
.31
.34
.39
.47
.50
.50

Action:

. Hg):


Velocity
Variation
* t f

0.15
0.15
0.17
0.19
0.21
0.25
0.31
0.35
0.39
0.40
0.40




•• \

0.24
0.25
0.26
0.29
0.31
0.35
0.41
0.45
0.49
0.50
0.50




760
                    -2
                                     21

-------
     Because of  the  impact the nozzle geometry may have on performance of the
   j. device, specification of nozzle inlet diameters is not sufficient to
guarantee  acceptability.   Numerous laboratory tests with SRI/EPA Cyclone  I
have resulted  in  the development of the nozzle design shown in the
Figure 2-8.  The  critical  geometric parameters shown in the figure  are speci-
fied in Table  2-4 for  the  range of nozzle diameters discussed previously.
The inside taper  or cone  angle, *, for each of the nozzle diameters was
determined from Figure 2-9.   This figure gives the optimum cone angle/  as
determined from  laboratory tests, for any given nozzle inlet diameter.  The
outside taper, *, should be  15* and the inlets sharp-edged so that  the
criteria of Belyaev and Levin (1972)  for negligible particle bounce from  the
inlet will be met.  The length of straight section at the nozzle inlet, I,
should not exceed 0.05 in.   Longer inlet sections may result in the develop-
ment of velocity  profiles  in the nozzle which may seriously perturb perfor-
mance of the EM10  device or  contribute to excessive particle losses in the
nozzle.  The total length,  L, for the nozzle diameters (dn) given in
Table 2-4 range from 2.653 to 1.45 in.  The minimum length of 1.45  in.  was
chosen to position the nozzle inlet a sufficient distance away from the body
of the PM,0 sampler so that  the flowstreara would not be disrupted.  For
nozzles with larger inlet  diameters,  a straight section at the nozzle  outlet
(after the expansion from  the inlet diameter to the outlet dimension,
0.50 in.) will be necessary  to meet this minimum length requirement.
    Figure 2-8. PM10 nozzle design. L = total length of nozzle; dn = nozzle inlet diameter;
             C = straight inlet length; 0 = inside cone angle;  = outside taper angle.
                                     22

-------
             TABLE 2-4.  PM10 NOZZLE GEOMETRIES
Nozzle Cone
Diameter, dfl Angle,. 
(inches) (degrees)
0.136 4
0.150 4
0.164 5
0.180 6
0.197 6
0.215 6
0.233 6
0.264 5
0.300 4
0.342 4
0.390 3

i
£ 7 -
-e-
•
O
5s
b
y ,
G O -
— o _
2 •
UJ
z
04
1 •
O
n 1 1 	
Outside
Taper , $
(degrees)
15
15
15
15
15
15
15
15
15
15
15

i

1




!
Straight Inlet
Length, i
(inches)
<0.05
<0.05
<0.05
<0.05
<0.05
<0.05
<0.05
<0.05
<0.05
<0.05
<0.05

i i






t i
Total Length,
L
(inches)
2.653±0.05
2.553±0.05
.970±0.05
.572±0.05
.491±0.05
.45 ±0.05
.45 ±0.05
.45 ±0.05
1.48 ±0.05
1.45 ±0.05
1.45 ±0.05









            0.1        0.2        0.3        0.4
               NOZZLE INLET DIAMETER  (in.)
0.5
Figure 2-9. Optimum inside cone angle, , for various nozzle inlet diameters.
                             23

-------
                                  SECTION  3

                                 CALIBRATION
     As for Method 5 or Method 17, calibration of specific components  of a
CSR sampling system is required.  It is recommended a notebook or  other
record be kept of all calibration data pertinent to each system.   This record
should contain the complete calibration history of each device requiring such
service (i.e.r flow metering devices, pitot tubes, nozzles, thermocouples,
and Magnehelic* gauges).  A separate record may be desirable  for  support
equipment, such as balances and field barometers, which are not directly
connected with che system.

3.1  PLOW METERING SYSTEM

     The best calibration of a flow metering device is achieved when the
calibration data are restricted to the range of expected use.  The defined
Method 5 flow rate of 0.75 dscfm provides a finite range of flow  rates over
which the flow metering system should be calibrated.  However, the PM10  flow
rate is dependent on the characteristics of the sampler used  and  cannot  be
easily defined as a single value.  This makes defining a calibration range
for the flow metering devices difficult at best.  For the purposes of  this
manual, the W10 flow rate at typical meter box conditions is assumed  to be
0.50 acfm.  It is up to the reader to determine the applicability of this
flow rate to his particular system.

     The flow metering devices of a CSR system  (dry gas meter and  sample
orifice) should undergo a stringent laboratory calibration before  field  use.
In addition to this initial acceptance testing, the calibration should be
checked after each test series.  This calibration check procedure  ensures
that the calibration parameters assigned to the flow metering devices  are
still valid without requiring the level of effort of the initial  calibration.
Should the results of the calibration check fall outside acceptable limits,
the flow metering device in question should be recalibrated according  to the
initial calibration procedure.

     Before any c .ibration, a leak check of the metering system  should  be
performed.  This should be done for both the vacuum  (negative pressure)  and
positive pressure portions of the train as prescribed in Method 5  (U.S.  EPA,
1977).

     A calibrated wet test meter or any other such calibration standard
should be used to calibrate the flow metering system.  The outlet  of the
calibration standard should be connected to the sample inlet of the control
console.  Before starting the calibration, the operator should run the system
vacuum pump for 15 min. with the orifice meter pressure drop  (AH)  set  at
approximately 0.5 in. H^O to allow the pump to warm up and to wet  the
                                    24

-------
interior  surface of  the wet  test meter.  The  calibration should be performed
as required  in EPA Reference Method  5  for  a range of  orifice AH settings
(0.5,  1.0,  1.5, 2.0,  3.0, and  4.0  in.  H20).   The minimum wet test meter
volume for each run  should be  5.0  ft3, except for the two lowest flow rates,
for which a  volume of  3.O ft3  is acceptable.   The data should be recorded as
requested in Method  5.

     A calibration check should be performed  on all of the flow metering
devices after each field test  series.  The posttest calibration check shoul'.
consist of three calibration runs  at a single orifice setting.   This orifice
setting should be representative of  the orifice settings used during the
field test.  The run procedure should  be the  same as  described  for the
initial calibration.

     If the dry gas meter calibration  factor,  Y, deviates by less than 5%
from the  initial calibration factor, then  the calibration constant assigned
to the meter is still  valid.   If the posttest calibration check yields a
calibration  factor outside this limit, the gas meter  should be  recalibrated
by using the initial calibration procedure.

     The averaae AH recorded during  the calibration check should be compared
with that obtained from the calibration equation for  the device.  To calcu-
late the AH  from the calibration equation, the wet test  meter flow rate
corrected for temperature and  pressure should be used.   If the  recorded
pressure drop varies from the calculated value by more  than 10%, the orifice
should be recalibrated.

     If the dry gas meter requires recalibration, for the purposes of data
reduction, use whichever coefficient  (initial or recalibrated)  yields the
lower gas meter volume.

3.2  PITOT TUBS

     The construction, configuration,  and  calibration specifications outlined
in EPA Reference Method 2 (U.S. EPA, 1977) should be  applied to the CSR pitot
tube.  The pitot tube  should be located at the side of  the sampling nozzle
furthest  from the sampling device axis.  To check the pitot tube for leaks,
one end of the tube should be plugged  and  a positive  pressure applied at the
opposite end.  If the  tube will not maintain  pressure,  a soap solution can be
used tc identify the location of any leaks.   If the CSR  pitot tube requires
calibration, the velocity pressure drop, .*P,  should be measured with a
standard pitot tube and the S-type CSR pitot  tube at  the same point within a
cross-section of a straight run of ductwork for a desired range of gas
velocities.  The CSR pitot tube should be  calibrated  as  used; that is,  the
complete sampler assembly should be used in pitot tube calibration determina-
tions.  The CSR pitot  tube should be calibrated twice, as recommended in
Method 2, reversing the direction of the legs  during  the second calibration.
For each velocity, a pitot tube coefficient is determined as

                                       A P     1/2
                            C? - °-99
                                      25

-------
where    Cp =» S-type pitot  tube  coefficient,  dimensionless
       0.99 = Cp value  for  standard  pitot  tube,  dimensionless
       "?ST = velocity  pressure  drop of  standard pitot  tube,  inches of water
      -?CSR a velocity  pressure  drop of  CSR S-type  pitot  tube,  inches of
               water

The average value of C_  for each direction over  the range of  velocities used
should be calculated.

3.3  SAMPLING NOZZLES

     A micrometer should be used to  measure the  inside  diameter  of the nozzle
to the nearest 0.001 in.  Three  separate measurements should  be  made/ each
based on a different diameter, and the average of the measurements should be
calculated.  The largest deviation from  the average should not exceed
0.004 in.  If the variation is more  than 0.004 in.,  the nozzle should be
reshaped and recalibrated.

3.4  THERMOCOUPLES

     The thermocouples  used to measure the various  temperatures  within the
sanpling train should be checked to  ensure they  are properly  calibrated
before they are installed in the system.  A two-point calibration  check using
an ice bath and a boiling water  bath should be performed  as outlined in EPA
Reference Method 2.  If any individual thermocouple does  not  produce a read-
ing within ±1.5% of the absolute reference temperature, it should  be replaced
with another thermocouple of the same type.   If  all thermocouples  show a
bias, the readout should be adjusted or  recalibrated according to  the
manufacturer's procedure.

3.5  MAGHEHELIC* GAUGES

     The calibration of the Magnehelic*  differential pressure gauges should
be checked prior to field use and periodically thereafter  to  prevent invali-
dation of test data.  Before calibration, the Magnehelic*  gauge  with no
applied pressure should be adjusted  to read zero using  the external
zero-adjust screw.  To perform a calibration check,  AP  values, as  read from
the Magnehelic* gauge,  should be compared with those from  an  inclined
•anoaeter at a minimum of three points.  The 4P  values  lead from the
Magnehelic* gauge should not deviate fro* the inclined  manometer readings by
sore than 5% at any point.

3.6  SOPPORT EQUIPMENT

     The field barometer should be adjusted initially and  before each test
series to agree with a standard  (a raercury-in-glass  barometer or the pressure
reported by a nearby National Weather Service station)   to  t 0.1  in.  Hg.

     The calibration of all balances to  be used  during  a test series should
be checked initially with class-S weights.  Triple  beam balances should  be
within to.5 g of the standard.  Analytical balances  should agree to ±2 mg of
the standard.  Balances which fail to meet these  criteria  should be adjusted
or returned to the manufacturer.

                                     26

-------
                                   SECTION  4

                           PRESAMPLING ACTIVITIES
     Preparation of the CSR  sampling system  for  field  use requires much the
same effort as preparation for Method  5 or Method  17 sampling (U.S. EPA,
1977).  Some type of pretest calibration or  operation  check  is necessary for
most components of the system.  This is also true  of supporting equipment
such as analytical balances.  Sampling reagents  also require preparation
prior to field use.  In several cases, the presampling activities described
below are similar to or the  same a& those required  for Method 5.

4.1  EQUIPMENT CALIBRATION AND CHECKS

     All sampling nozzles should be inspected  for  damage  and repaired where
necessary.  The nozzles should be cleaned with tap water, deionized water,
and finally acetone.  The inside diameter of each  nozzle  should be measured
to the nearest 0.001 in. as  described  in Section 3 of  this manual.  The knife
edge of the nozzle should be protected during  shipment by serum caps or
similar covers.

     The PM1(, sampler and filter holder should be  ultrasonically cleaned with
tap water and rinsed with deionized water.   After  a final rinse with acetone,
the sampler assembly should  be allowed to air dry.  An adequate supply of
Viton or silicone rubber O-rings should be available for  replacement of worn
0-rings in these devices.

     The openings of the pitot tube should be visually inspected for damage
such as dents or nicks.  A check should also be  made for  proper alignment.
The two legs of the pitot should be in a straight  line so that the opening of
one leg is directed 180* from the other.  If damage or misalignment is
evident, the pitot tube should be repaired or replaced.   If  repairs are made,
the pitot tube should be recalibration, as described in Section 3.

     The lines of the sampling probe,  including  the pitot lines,  should be
cleaned before field use.  The lines should  be cleaned internally by rinsing,
first with tap water, then deionized water,  and  then acetone.   The lines
should be rinsed a final time with acetone and allowed to air  dry.  If a
heated probe is used, the heating system should  be  checked for proper opera-
tion.   (If problems are encountered, see Section 10 of this  manual for
troubleshooting guidelines.)

     The water dropout system  (condenser and drying column)  should be checked
for leaks.  The condenser should be cleaned  with deionized water  and rinsed
with acetone.  The condenser should be inverted  to  ensure total drainage and
allowed to air dry.
                                   27

-------
     The system flow metering devices  (dry gas  meter  and  orifice)  should have
appropriate calibration factors assigned  to  them.  A  pretest calibration
check, performed according to the procedure  outlined  for  posttest  calibration
checks in Section 3, is recommended to ensure the calibrations are still
valid.  Although pretest calibration checks  of  the flow metering devices are
not required and do not take the place of the posttest  checks, they are
useful for detecting problems prior to field use.  Such a pretest  calibration
check is strongly recommended if the system  has not been  used for  some time.

     It is recommended, but not required, that  the calibration of  the
Magnehelic* pressure differential gauges be  checked prior to field sampling.
As with the flow metering system, this is strongly suggested if the system
has not been used for an extended period of  time.

     All system temperature sensors (thermocouples, temperature gauges, etc.)
should be checked against a mercury-in-glass thermometer  at  ambient tempera-
ture.

     Finally, it is recommended that a leak  check of  the  complete  system be
performed before it is shipped.  The system  should be assembled from tne
probe (it is not necessary to include the nozzle or PM1Q  sampler)  to the
control console.  The probe should be capped and the  system  leak-checked at
15 in. Bg vacuum.  Leak rates in excess of 0.02 cfm should be corrected.
Bach leg of the pitot tube should be leak-checked, including the umbilical
lines and the differential pressure gauge.

4.2  PREPARATION OF SAMPLING REAGENTS

     Osed silica gel should be regenerated by drying  at 350  *F for 2 h.  New
silica gel may be used as received.  Several 200- to  300-g portions may be
weighed in airtight containers to the nearest 0.5 g.  The total weight for
each container should be recorded.  As an alternative,  the silica  gel may be
weighed in the drying column at the test site.

     Quartz fiber is the recommended filter  material  because it is more
resistant to chemical reactions when exposed to stack gases.   Filters should
be desiccated and weighed as required for EPA Reference Method 5.   To prevent
the loss of filter cake, it is recommended that aluminum  foil envelopes be
made to enclose the filter.  If used, these  envelopes should be desiccated
and weighed with the filter.

     Aljainoa foil envelopes can also be used to collect  the PM Q  cyclone
catch.  These foil envelopes should be uniquely identified,  desiccated, and
weighed in the same manner as the filters.

     Aluminum weighing dishes or small glass beakers  may  be  used to collect
and evaporate the rinses from the sampling device.  These containers should
be cleaned thoroughly, labeled, desiccated,  and weighed prior  to use.   If the.
sampling train is operated in a Method 5 configuration, a larger glass beaker
or similar container will be needed for probe rinses.
                                    28

-------
     Acetone for sample recovery should be reagent grade with less than
0.001% residue.  Acetone blank determinations to ensure residue levels are
acceptable may be made before field use or as part of sample recovery.
                                    29

-------
                                   SECTION  5

                             SAMPLING PARAMETERS
     The sampling parameters which must be dealt with  in  the  CSR procedure
are  the same  as those  in EPA Methods  5 or 17  (U.S. EPA,  1977),  with  the
addition of those for  particulate mass for each size fraction.   However,  the
two  types of  procedures are substantially different in  terms  of control of
sampling parameters.   In the CSR procedure, sample flow rate  is determined by
the  requirement that the size  separation device maintain  fixed  size  cut(s)  at
or near 10 um (aerodynamic).   For a single-stage cyclone  sampler,  this flow
rate is determined uniquely by the gas composition and  temperature of the
process stream.  For a cascade impactor, the gas composition  and temperature
determine a range of permitted flow rates from which to choose  before
sampling is initiated.

     Because  flow rate cannot  be adjusted during a sampling run to match
nozzle velocity with the velocity of  the gas stream, thus eliminating errors
associated with anisokinetic sampling, the nozzle(s) must be  carefully
selected.  First the sample flow rate is determined, then the range  of stream
velocities at which sampling is to be performed determines the  nozzle(s)  to
be used.  For each nozzle diameter, there is a range of stream  velocities
which correspond to a  permissible level of anisokinetic sampling error.
Usually, the  stream velocity in a sampling plane will be  uniform enough that
one nozzle size will be acceptable for the entire range of stream velocities.
However, for  process streams with wide variation of velocity, it may be
necessary to  use more  than one nozzle for a complete traverse of the sample
plane.

     A third  difference between CSR and other methods for  particulate emis-
sions involves the sample duration, or dwell time, at each traverse  point.
In Method 5 or 17, the dwell time is  tha same for all traverse  points.   A
velocity-weighted average of the measured concentration,  or emission rate,  is
obtained when the sample flow  rate is adjusted proportionally to the local
stack velocity, as described previously.  Because the sample  flow rate cannot
be adjusted when the CSR method is used, a velocity-weighted  concentration  is
obtained by adjusting  the dwell time  at each point to be  proportional to the
local stack velocity.

     These differences between the methods manifest themselves  in  the setup
calculations  required  before a sampling run.  This section presents  the
governing equations and describes the calculations required to  determine  the
CSR sampling  parameters.  In some cases these calculations are  much  the same
as those required for Methods  5 or 17.  References to the  appropriate EPA
method will be cited in these  situations.  Only those equations and  calcula-
tions which are unique to CSR  will be discussed in detail  here.

                                    30

-------
5. 1  PRELIMINARY MEASUREMENTS

5.1.1  Method  1

     Before the appropriate sampling parameters can be calculated,  some
preliminary determinations must be made.  First, the  number  of  traverse
points required for CSR should be determined from Figure  5-1.   The  procedures
and requirements for this determination are the same  as Method  1.   For
rectangular stacks, the equivalent duct diameter should be calculated as in
Method 1  (U.S. EPA, 1977).  The traverse points should be located within the
duct as described in Method 1.

5.1.2  Methods 2, 3, and 4

     When the number of traverse points and their locations  within  the  duct
have been decided upon, the stack temperature  and velocity at each  of these
points should be determined by Method 2 (U.S.  EPA, 1977).

     The concentration of the primary stack gas constituents (i.e., oxygen,
carbon dioxide, nitrogen, carbon monoxide, and water  vapor)  may be  determined
from Method 3  (U.S. EPA, 1977) and Method 4 (U.S. EPA, 1977) data.   With this
information, the gas dry and wet molecular weights may be determined from the
following equations:

                   Md - 32 (f0) + 44 (fc) +28(1 - f Q - fc)               (5-1)

                                                                        (5-2)
It may be necessary to perform additional testing according  to  standard
methods in instances where the stack gas composition  is  influenced by gases
other than those listed above.  For example, when gases  from a  high-sulfur
source are sampled, Method 6  (U.S. EPA,  1977) testing should be performed to
determine the percentage of SO2 present  in the process gas,  and this  percent
SO2 should be taken into account when determining the dry molecular weight.

5.2 SAMPLE FLOW RATE AND NOZZLE SELECTION FOR A SINGLE-STAGE CYCLONE  SAMPLER

5.2.1  Sample Flow Rate

     If a single-stage size separation sampler is to be  used, the  sample  flow
rate must be that which provides a 10-um size cut.  This flow rate will be
determined by the gas composition and temperature of the process gas.   The
relationship between flow rate and gas parameters specific to the  sampling
device must be determined by calibration measurements (described in
Section 2) prior to field use.  For SRI/EPA Cyclone I, this  relationship
takes the form
                                        M p  -0. 29<»9
                        Q_ - 0.002837 (_ !L1)       u.               (5-3)
                                         Ts
                                      31

-------
DUCT DIAMETERS UPSTREAM FROM FLOW DISTURBANCE (DISTANCE A)
0.5 1.0 1.5 2.0 2.
SO
Z 40
O

AVERSE
w
O
oc
o
c
IU
| 20
3
Z
1
2
z 10
2™*
0
I I . I
» HIGHER NUMBER IS
FOR RECTANGULAR STACKS
OR DUCTS






—
J
t

B


•



-

—
I

1
7
DISTURBANCE
MEASUREMEN

I SITE


DISTURBANCE
~^M«wn««

T






—


12

"™"
I I III

8 OR 9" ~
I
 3         4        5         67         8         9
DUCT DIAMETERS DOWNSTREAM FROM FLOW DISTURBANCE (DISTANCE B)


   Figure 5- J. Minimum number of traverse points for CSR  traverse.
    10
57«5-€J
                       32

-------
     The viscosity of the  flue gas  can  be  determined  by the equation
(Williamson et al., 1983)

                    .. 5* (*  4- f* T  * P T 2  * I"1 B    A <"• F
                    u » G! + c2Ts + c3Ts^  + c^ws  + C5fo

where u is in micropoise,  T in "C,  and

          Cj - 160.62
          C2 » 0.42952
          C3 = 1.0483 x
          G, - -74.143
          Cg » 53.147

or for T in *R

          Cx » 51.05
          Cj - 0.207
          C3 » 3.24 x 10-5
          Q, » -74.143
          C5 - 53.147

     This equation fits data  (with  a standard error of 0.98 micropoise)  for
combustion gas of arbitrary composition in the range  0-350  *C,  0-70%
moisture.  This equation was generated  from large  "data banks"  of viscosities
calculated by the more rigorous algorithm  of Wilke (1950).

5.2.2  Nozzle Selection

     After the sample flow rate has been determined,  the corresponding nozzle
velocity (u) for any given nozzle may be determined from the following equa-
tion:

                                u = —i—-  Qs                         (5-5)
                                     wan

     For each nozzle velocity, a range  of  stack velocities, v ^  to v   ,  for
which the anisokinetic sampling errors  are within  acceptable limits must then
be determined.  As discussed in Section 2, this range of stream velocities is
governed by the semi-empirical relationship developed by Belyaev and Levin
(1974)
                            A - 1 + (R  - 1) .T4-T-                    (2-D
where A  * measured concentration/actual concentration
      R  » isokinetic ratio, v/u
      B  - (2 + 0.617/R)K
      K  » particle Stokes number with respect to the nozzle, tv/d
      T  « particle relaxation time CD2/T8u, seconds
      C  - Cunningham slip factor
      D  » particle aerodynamic diameter
      u  a gas viscosity, poise
      dn * nozzle diameter
                                      33

-------
Defining  vrain  as  tne  stack  velocity  at  which  A equals  0.8 and vraax the stack
velocity  at  which A equals  1.2, we obtain  equations  for  vmin and vmax such
that

                         ,          /0.2603 Qa1/2y,
                vmin  ' u [0-2457  + / 0.3072	--f	]            (5-6)
                         r          /0.2603 Qa1/2y
                       u [0.4457 + / 0.5690 +  	--f	
                                                    n J / i
vmax " u [°-4457 + ^ 0.5690 + 	3^2	J            (5-7)
where the terms in brackets are equivalent  to  the velocity  ratios',  ^-m^n and
Rmax, respectively, and have additional limits of 1.0 ±  0.5 if  more
restrictive.

     Although the vmin and vmax for a given nozzle  are useful,  the  local
stack velocity is actually measured as velocity pressure differentials during
both Method 2 and CSR traverses.  Therefore, it is  ioore  practical to convert
vmi« an^ vma» *° tne cor responding pressure differentials by using  the
 mxn      nicUm
following equations:


                     AP .  » 1.3686 x 10-*  -S-^  (-r^5-)                 (5-8)
                         .
                                             s
                                                                        (5-9)
where C_ is the pitot coefficient of the CSR sampling probe,  and  all other
variables are as defined previously.

     Finally, given the most recent Method 2 data  for the  sampling  plane and
the AP limits for a given set of nozzles, the nozzle or  nozzles required to
perform the sample traverse may be determined.  In instances  where  the pitot
coefficients of the Method 2 probe and the CSR probe are different,  the
Method 2 data should be converted to equivalent values for the CSR  pitot.
This is accomplished by using the following equation:

                                               C
                        AP(CSR) = AP(Method 2) (	)2                   (5-10)
                                               S1


     The AP limits for the chosen nozzle should bracket the measured stream
values.  If more than one nozzle is acceptable, the one giving the  greatest
symmetry should be selected.  If the expected pitot AP for one or more points
is near a limit for the chosen nozzle, it may be outside the  limits  at the
time of the run due to temporal variations of the  process.
                                     34-

-------
5.3  SAMPLE PLOW RATE AND NOZZLE SELECTION FOR CASCADE  IMPACTOR SAMPLERS

     Several cascade impactocs  ace  available which can  be  used  for  PM1Q
measurement.  The relationships between particle  size cut,  sample flow rate/
gas coaposition, temperature, and nozzle diameter  (for  the  first PM..  stage)
must have been determined by calibration measurements  (described in Section
2) before field sampling.  Procedures for operating these devices,
calculating particle size cut for each stage, and analysis  of the results to
obtain cumulative mass concentration as a function of particle  size are
described by McCain et al. (1986).

5.3.1  Sample Flow Rate

     The multiple stages of a cascade impactor provide  the  ability  to  inter-
polate, thus permitting determination of PMjg without the requirement  that a
stage have a 10-yra size cut.   This  flexibility in flow  rate permits minimiza-
tion of error due to anisokinetic sampling by selection of  a nozzle and  flow
rate corresponding to the average stream velocity, v, of the process stream.
The sample flow rate is selected from the set, Q_, determined by

                                Qn  • v *dn2/4                          (5-11)

where Qn * flow rate for nozzle n at the average stream velocity
      dn = diameter of nozzle n.


One of this set of flow rates is selected subject to the additional criteria
below:                 .

     e  The impactor must include one stage cut at or greater than  10  urn and
        two succeeding size cuts less than 10 urn.

     o  The separation between these successive size cuts must  be no more
        than a factor of 2 and no less than a factor of 1.5.

     o  The product of jet velocity and size cut for each impactor  stage with
        cut above 2 urn should satisfy the criteria given by McCain  et  al.
        (1986) for the particulate  collection substrates utilized to limit
        particle bounce.

5.3.2  Nozzle Selection

     Calculation of minimum and maximum stream velocities, vm^n and vmax, for
a given nozzle depends upon the desired accuracy for particles  greater than
10 urn.  If the >10-um particulate fraction is of little concern,  the calcula-
tion of vmin and vmax given for single-stage cyclones (equations  5-6 and 5-7)
and similarly for APmin and APm?x (equations 5-8 and 5-9)  can be  used.   The
errors in cumulative concentration  versus particle size, due to anisokinetic
sampling, will be limited to no more than ±20% at 10 yra, decreasing sharply
                                      35

-------
with particle size.  The corresponding error for M10 will  be  much  less,
depending upon the size distribution of the <10-ura particles.

     If accurate measurement of particle sizes greater  than 10 urn is  sought,
then the more restrictive limits of McCain et al.  (1986), which  require
nozzle velocity to be within ±20% of the stream velocity at each traverse
point, are recommended.  In this case, the error limit  of cumulative  percent
data for arbitrary size distribution is given by the percent deviation  of
nozzle velocity from stream velocity.

5.4  SAMPLE ORIFICE AH

     After the sample flow rate and nozzle size have been determined, the
orifice pressure drop, AH, corresponding to this sample flow rate may then be
calculated from the following equation:

                         2S(1 * Bws)ps          1.083 AH@
                   AH = ( - T- - )2  TMMd ( - p— 1)             (5-12)
5.5  DWELL TIME

     When EPA Method 5 or 17 is used for sampling, the dwell  time  at  each
traverse point is the same.  The measured concentration  is  a  velocity-
weighted average for all points, as it should be for determination of emis-
sion rate, because the sampling rate is varied at each point  proportionally
to point velocity.  Because flow rate cannot be adjusted  from point to  point
with CSR, the dwell time at each point must be proportional to  the point
velocity to obtain a velocity-weighted sample.

     The dwell time for the first sample point is calculated  before the run
is started in the following manner:

                             /APT      (Total Run Time)
                     t.  - - i -- : -                (5-13)
                                              Points)
where       t^ *» dwell time at the first traverse point, rain
           Apj « the expected AP at the first traverse point
                 (from a previous velocity traverse) , in. H.O
       (/7TP) av  = the average square root of AP from a previous velocity
                 traverse

     At subsequent traverse, points, the velocity AP is measured and  the dwell
time calculated using the actual AP recorded at the first point such that
                                    36

-------
The number of minutes  sampled  at  each  point  should  be rounded to the nearest
1/4 rain.

     -For sampling situations in which  the  stack  temperature varies by more
than ±50 *F, the temperature must be factored  into  the dwell time calculation
to obtain a velocity-weighted  rumple.  The dwell  time equations become

                                          (Total  Run Time)
                                                                       (5-15)
                                            (No.  Points)
                                    avg
                         t,  /AP  T
                    t  »   *     n  fl.                                  (5-16)
                           /AP1 Tl
where   T^ = stack temperature at the  first  traverse  point,  *R
      T    = the average stack temperature  (from  a previous  velocity
             traverse), *R
        T  » stack temperature at traverse point  n,  *R

     The total sampling time given in  equation  5-13  and  5-15 should be
greater than or equal to the minimum total sampling  time specified in the
test procedures for the specific industry.   Further/  it  is required that (1)
the sampling time per point is not less  than 2  min (or some  greater time
interval if specified by the Administrator)  and (2)  the  sample volume taken
(corrected to standard conditions) will  exceed  the required  minimum total gas
sample volume.  The expected total volume of gas  to  be sampled during a
single run may be approximated by using  the  sample flow  rate and average run
time.

     In some circumstances (e.g., batch  cycles) it may be necessary to sample
for shorter times at the traverse points and to obtain smaller gas sample
volumes.  In these cases,  the approval of the Administrator  should be
obtained.

5.6  PERFORMING THE CALCULATIONS

     Calculation with the  equations  presented in  the  previous subsections may
be performed in a number of ways.  A set of  worksheets to be used for hand
calculation of the setup equations has been  prepared.  These worksheets  are
shown in Figures 5-2, 5-3, and 5-4.

     If the stack gas temperature changes such  that  it is outside ±50 *F of
the average used for the setup calculations,  the  setup parameters (sample
orifice AH and nozzle AP limits)  should  be recalculated.  To save time during
the run, it is recommended that the  setup calculations be performed prior to
sampling for a range of average stack  temperatures.   These temperatures
should be in increments of 50 degrees  and span  the range of  expected stack
temperatures.  This limits errors in the sample flow  rate and nozzle AP
limits to approximately ±10%.  A table is included in Figure 5-4 to facili-
tate this approach.
                                     37

-------
                                CSR WORKSHEET I

                        SAMPLE FLOW RATE AND ORIFICE AH
Barometric  Pressure,  Pfl/ in. Hg
Stack Differential Pressure, Dp stacjc»  in»

Average Stack  Temperature, Tg, 'R -

Meter Temperature, TM, *R » _

Orifice AHg, in.  H20 » _
Gas Analysis:
          C02  fraction, f
          02  fraction, fQ •
          Water  fraction, Bwg
Md - 44(fc)  + 32(f0)  + 28(1 - fc - fo) -


«  * M(1  *  B>  * 18(BJ  -
     p    PP  stack  .
 s    a     13.6
V » 51.05  +  0.207 Ts + 3.2355 x 1Q-5 Tg2 +  53.147(f )
    - 74.143(Bwg)  -    ;	
               ,M__ P,,-0 .
   - 0.002837      8
-QS (1 " Bws)Ps-
                                   ,  1.083
                              « «a I — ir
                  S-i
           f igure 5-2. CSR worksheet I • sample flow rate and orifice AA/ calculations.
                                      38

-------
                               CSR WORKSHEET II


                               NOZZLE SELECTION
Barometric Pressure  ?a,  in.  Hg »
Stack Differential  Pressure,  Dp gtackf in H20



Average Stack Temperature,  TS,  "R » 	



Meter Temperature,  TJJ,  *R *           •



Sample Flowrate, Q_,  acfin - 	•
Gas Analysis:
          CO.  fraction,  f
          0-  fraction,  f
          Water  fraction,  BWS



CSR Pitot Coefficient,  C.
                        "P
Method 2 Pitot Coefficient,  C'
     44 (fc> +  32 (f0)  + 28(1
     M(1 - B)  * 18(B)  »
p.p. DP  stack
 s    a      TT75
j » 51.05  *  0.207  Ts + 3.2355 x 10~5 Tfi2 * 53.147(fQ)


    - 74.143 B._ -
     3.056 Q.
u -  	,	i  *
                    Figure 5-3. CSR worksheet II • nozzle selection.
                                      39

-------
           °-2457  * '0.3072  -
                               0.2603 QC1/2 u,
                                      ,3/2
vaax " u
                    /         0.2603 Q_1/2 u
                  * /0.5690  +

iP»in '  1'3686 x  10"
-?aax *  1-3686 x  10
                       P  M     v
                    -!,  3   w  :•   max > 2
Nozzle ID
dn, in.
u, fps
v»in' «?«
Va«' f^
AP«in' in- H2°
Apmajt, in. 820




























Velocity Traverse Data:
iP(CSR) « AP (Method  2)
                        - C
            Port
          Port
Port
                                             Port
Point t
Soz ID     iP    Soz ID     iP     Noz ID
                                                                        Noz ID
  1
  2
  3
  4
                                   5-3. (Continued).

-------
                               CSR WORKSHEET III

                                  DWELL TIME
Previous Method  2 data:
                                  Total Run time, rain.

                                  No.  Traverse points
     avg
                           Average T_ "R *
                    ('AP)
                              Total run time
                               No. points
                                             'A P.
                         avg
Note: If T  varies by  more than ±50*F from the average, T  must be included
in the dwell  time calculations as shown in Equations 5-15 and 5-16, which
include the factors /T1/Tavg and rTjy/tj in the equations  for  t.  and tn,
respectively.
Ta, 'F
AH, in. HjO
dn, in.






Apmin APmax






min max






Apmin Apmax






APmin APmax



^
Point t

  1

  2

  3

  4
Port

  Ap
Port

  Ap
Port

  Ap
                                                                Port
t        Ap        t
                Figure 5-4. CSR worksheet III - dwell time calculations.

-------
                                  SECTION  6

                              TAKING THE SAMPLE
6.1  FIELD ASSEMBLY

     After the sampling system has arrived at  the  test  site,  it should be
visually inspected to determine if any damage  was  incurred  during transport.
To check for internal probe damage, a positive-pressure leak  check on each of
the lines running through the probe is recommended.   To do  this, one end of
each line should be blocked and positive pressure  applied at  the other.  The
pressure on the line should be monitored with  a manometer connected in
parallel.  Failure to hold pressure indicates  an internal probe leak, which
should be found and repaired before proceeding. '

     When it has been determined the system  is in  proper operating condition,
the operator should begin assembling the system.   The particle-sizing device
should be assembled as the operating manual  dictates.   The  sampling nozzle
should then be attached to the sampler and the complete device mounted on the
probe.  With the proper length of extension  tubes, the  pitot  head should then
be mounted on the probe.  The sampling nozzle  and  one leg of  the pitot must
face the same direction while all the tubing unions  are fully tightened.  A
combined umbilical or individual tubing can  then be  used to connect the probe
to the control console.  The sample line should be attached to the inlet of
the water dropout system  (condenser and silica gel column), which is, in
turn, attached to the sample inlet of the control  console.  If available, an
umbilical which encloses the sample line and thermocouple extensions in a
single sheath is preferred.

     Because the amount of water collected in  the  condensing  system must be
known, all components of this system should  be clean and free of any foreign
material.  If silica gel columns are used, a preweight  of the column and
silica gel should be obtained before any testing.  Then the column must be
sealed until testing commences to avoid any  accidental  uptake of moisture.
After sampling, the column should be weighed again to determine the amount of
water uptake.  If a condenser is used, it should be  placed  in an ice chest
and ice added to the chest until the condenser is  sufficiently covered.

6.2  L2AK TEST

     When the system has been completely assembled,  the control console pump
may be used to leak-check the vacuum system  as in  Method 5:

     e    Plug the sampling nozzle and turn  on the pump to  produce a vacuum
          across the desired test section.

     o    Use the fine-adjust valve to set the system vacuum  to 15 in.  Hg.

                                      U2

-------
If the required vacuum reading on the console-mounted  vacuum gauge cannot be
achieved or if the gas meter  indicates a  leak  rate greater  than 0.02 cfra, the
system is not sealed, and  the leak(s) must be  located  and  fixed.

     The positive pressure portion of the control console can be  tested for
leaks by using the procedure described in Method 5.

6.3  PRETEST EQUIPMENT WARM-UP

     Because most flue streams, to be tested are not  at ambient temperatures,
the CSR sampling device must be heated to stack conditions.   This helps
ensure isokinetic sampling and significantly reduces the chance of acid
deposition within the sample line.  The sampler should be heated  in the gas
stream long enough to equilibrate with the temperature of the surrounding
stack gases.  Typically, the PMiQ sampler should remain in  the flue at least
15-20 nin to ensure thermal equilibrium.  The  nozzle,  if uncapped, should not
point into the flow field during preheating*   If possible,  the nozzle should
be capped or plugged during preheating, and the cap  or plug  should be removed
immediately before sampling.

     If the absolute pressure of the gas  stream to be  sampled is  significant-
ly lower than ambient pressure, care must be taken to  prevent rupturing the
filter when the sampler is inserted into  the duct.   In Method 5,  this problem
may be avoided by starting the pump as the probe is  inserted through the
port.  Because of the preheat 'required with CSR, this  method cannot be used.

     If an in-stack filter configuration  is used, the  sample line should be
closed at the back end of  the probe.  However, because enough gas volume
remains in the probe to rupture the filter, steps must be taken to slowly
insert the probe into the  stack.  This allows  the probe to  gradually adjust
to the duct vacuum without rupturing the  filter.

     If the system is operated in a Method 5 configuration,  a heated probe
should be used.  The heating system should be  adjusted to maintain a probe
temperature slightly greater than the gas dew  point.   If an  in-stack filter
is used, a heated probe is not necessary.  However,  care should be used to
prevent condensed vapors from running back into the  sampler.   One solution is
to keep the probe inclined during operation.

6.4  SYSTEM START-UP
                                                                 !*•

     After the sampling device has preheated a sufficient length  of time, the
nozzle cap (if one is used) should be removed  and the  nozzle turned into the
flow.  The vacuum pump should be started  and the sample line opened simulta-
neously.  The sample orifice AH should then be adjusted to  the appropriate
setting.  The dwell time at the first traverse point should  be that calcu-
lated from equation 5-13.  The dwell time at each subsequent point should be
calculated during the run from equation 5-14.

     If the stack temperature at any point in  the traverse  falls  outside
-50 "p of the average used to calculate the operating  parameters,  a new
temperature range should be chosen and the operating parameters (AH and

-------
nozzle AP limits) corresponding  to  this new  temperature  range should be used.
Dwell times should be calculated using equations 5-15  and  5-16.

6.5  TRAVERSING

     During traversing  (moving to a new point or new port),  all  motion should
be smooth and brief to  avoid bumping or vibrating the  sampler.   When removing
or inserting the sampler, care must be taken not to scrape the nozzle on the
port wall.  Also, the sampler should not be  allowed to bump  against the far
inside wall of the duct.

6.6  SHUTDOWN ORIENTATION

     The orientation of the sampler may require the operator  to  maintain an
appreciable flow rate while removing the sampler from  the  flue.   The flow
rate should be maintained until the sampler can be placed  in  a favorable
orientation (usually horizontal}.   This is particularly  true  when operating a
cyclone in a vertical orientation.  Otherwise, some dust might fall from one
stage of the sampler to another  and thus be measured where it was not
collected.  Special care must be exercised to prevent  contamination of the
sample by dust from the walls of the port.  After the  flow has been termi-
nated, the sampler can be transported to the laboratory.  The sampler should
be kept in a horizontal position with the nozzle plugged or covered to avoid
contamination or loss of sample.

6.7  DATA LOGGING

     The parameters of the test should be recorded in  a clear, concise format
like that shown in Figure 6-1.  Parameters that are likely to change, such as
stack temperature and velocity AP,  should be recorded  periodically.   After
sampling at the first point, move to the next and immediately compare the
pitot A.

-------
Run
Code
Saapler
ID
Filter
ID
Sampler
Orientation
Sampling
Location
Nozzle
Diameter -ID (in)
Operator (s)
Date
Start
Time
End
Time
Sampling
Duration (min)
OGM
(initial)
DGM
(final)
Sample
Volume (ft3)
Dual Manometer Leveled and Zeroed?
Magnehelics Zeroed?
Gas Composition
%C02 Klj %CO
Moisture Content
Bun
Time












Port No
Trav.Pt.












Pi tot Leak Check
(Pos) (Neg)

AP
Pi tot












AH
Sample












DGM
Volume












Stack
Temperature (*F)
Differential Stack
Pressure (in. HjO)
Ambient
Temperature (*F)
Ambient
Pressure (in. Hg)
Gas
Velocity
System Leak Check
Notes
T
Stack












T
Probe












T
DGM












Dwell
Time*





/






aDwell  Time,  t.
                          Figure 6-1. CSff field data sheet.
                                        45

-------
                                   SECTION  7

                              SAMPLE RETRIEVAL
     After the sampling system has been  allowed  to cool  to  a point  where it
can be safely handled, the collected  sample may  be carefully recovered.   The
recovery procedures described below are  specifically  for  the configuration of
SRI/EPA Cyclone I.  Different sampling devices may have  different sample
recovery procedures, as described in  the manufacturer's  operating manuals and
in the literature.

7.1  RECOVERY OP THE PARTICOIATE MASS

     Great care is needed during recovery of the collected  particles  from the
PMj0 sampler to ensure that all of the particulate matter is recovered  and
placed in the proper sample containers.  The sample can  be  effectively
recovered from both stages of the PM.Q sampler (cyclone  and filter) by  using
a combination of brushing and washing.

     The first step in recovery of the particulate matter from  the  nozzle and
cyclone is brushing the collected mass into the  appropriate foil  envelope.  A
clean no. 7 camel's hair brush or small  nylon bristle brush is  suggested for
this operation.  The brushed surfaces should then be  rinsed thoroughly  with
acetone, or similar solvent, to recover  any particles that  continue to  adhere
to the sampler.  These rinses should  be  collected in  a uniquely identified
sample container.  The brush used for recovery should also  be rinsed  into
this container.

     The filter should be recovered from the filter holder  and  returned  to
the appropriate container.  Any particulate matter or filter fibers adhering
to the filter holder surfaces or rubber  0-ring should be brushed  onto the
surface of the filter.  The interior  surfaces of the  filter holder  should
then be rinsed with the solvent as described above.

     As stated previously, assignment of the collected particulate  matter to
the appropriate sample container is very important.  Particulate  matter
collected on the inner surfaces of the nozzle, the cyclone  body,  collection
cup and cap are to be considered as collected by the cyclone. '  Furthermore,
any matter brushed or rinsed from the outside of the cyclone exit tube  is
also to be considered part of the cyclone catch.  The PM,0  fraction consists
of particulate matter collected from  the inner surface of the "turn-around"
on the cyclone cap, the inside wall of the exit  tube, the inner walls of the
filter holder (upstream of the filter),  and the  surface of  the  filter.

     Final weights for all particulate samples should be determined on site,
prior to shipment.  Recommended procedures are outlined  in  Section  8.

                                     46

-------
7.2  MOISTURE DETERMINATION

     The condenser should be drained of any collected moisture  and  the amount
of liquid determined either volumetrically  (to ±  1 mL) or  gravimetrically (to
± 0.5 g).  The liquid may be discarded after the weight or volume is  record-
ed.  The spent silica gel should be weighed in the appropriate  container
(such as the drying column or shipment container) to determine  the  moisture
uptake.

-------
                                  SECTION 8

                             POSTSAMPLING CHECKS
     Posttest activities for a CSV. system involve equipment calibration
checks, field sample analysis, and equipment maintenance.  The  first  two
items are discussed in this section.  The third item is discussed  in
Section 10 of this manual.

8.1  EQUIPMENT CALIBRATION CHECKS

     A posttest calibration check of the flow metering devices  is  required.
The posttest calibration check should be performed as described in Section 3
of this manual.  If the gas meter correction factor obtained  from  the cali-
bration check deviates from the initial calibration factor by more than 5%,
the meter should be recalibrated.  The posttest data reduction  should then be
performed with whichever calibration factor yields the lower  gas meter
volumes.

     Calibration checks should also be performed on the stack and  dry gas
meter thermocouples.  Ea*h of the above temperature sensors should be
compared with a mercury-in-glass thermometer at ambient temperature.   If  the
stack temperature thermocouple reading differs from the reference  by  more
than 1.5% of the absolute temperature, the thermocouple should  be  recali-
brated as described in Section 3.  The old and new calibrations should be
compared to determine the sign and magnitude of the correction  to  be  applied
to the average stack temperature.  If the dry gas meter thermocouple  reading
varies from the reference by more than 6 "C  (10.8 *F), the thermocouple
should be recalibrated.  For data reduction calculations, the calibrations
which give the higher gas meter temperature should be used.

     A posttest calibration check is also required for the system  Magnehelic*
gauges.  This calibration check should be performed as described in
Section 3.

8.2  SAMPLE ANALYSIS

     Analysis of the field samples is essentially the same as for  Method  5.
Filter and cyclone catches should desiccate for a minimum of  24 h  before  the
initial weighing.  Each sample should be weighed to a constant  weight,  which
is achieved when the difference between consecutive weighings is no more  than
0.5 mg or 1% of the total weight less tare weight, whichever  is greater;  at
least 6 h of desiccation time should be allowed between weighings.


                                    48

-------
     As an alternative, the samples may  be  oven  dried  at the average stack
temperature or 220  *F, whichever  is less, for  2  to  3 h,  cooled in a desicca-
tor, and weighed to a constant weight.   The tester  may also opt to oven dry
the samples as described above, weigh the samples and  use this as the final
weight.  Whichever option.is chosen, final  weights  of  all cyclone and filter
samples should be determined to the nearest 0.1  rag  on  site, before shipping.

     Acetone rinse and blank samples should be inspected to confirm that no
leakage has occurred.  If a noticeable amount of sample  has been lost through
leakage, the sample must be either declared void or corrected in the final
results with methods approved by  the sponsoring  agency.   The liquid should be
measured either volumetrically to ±1 mL  or  gravimetrically to ±0.5 g.  Each
sample should be evaporated to dryness at ambient temperature and pressure in
a tared 250-mL beaker or similar  container.  The evaporated samples should be
desiccated for 24 h and weighed to a constant weight.  Results should be
recorded to the nearest 0.1 rag.

     If the silica gel is not analyzed in the field, the spent silica gel
samples should be weighed in the  appropriate container to the nearest 0.5 g.
                                    49

-------
                                   SECTION  9

                                DATA ANALYSIS
9.1  VELOCITY-WEIGHTED AVERAGES

     If the velocity distribution at the sampling plane  is  such  that two or
more nozzle sizes are required to complete a  traverse  or  if the  stack
temperature variation exceeds ±50 *F so that  new setup parameters  are
required during the run, a velocity-weighted  average of  the stack  temperature
and particulate concentration must be calculated.  For a  generic data term,
D, this equation takes the form

                                      n

                               15 " I  I  Di fci                          (9'1)
                                   rj=1

where  T5 * the velocity-weighted average of D
      D. = average value of D for each nozzle size
       6 * total run time
      t. » run time for each nozzle size
       n = number of component changes

For the remainder of this section, the term "run" will refer  to  that portion
of a traverse performed without any component, changes.

9.2  DRY GAS METER VOLUME

     The sample volume measured by the dry gas meter can  be corrected to
standard conditions (68 *F, 29.92 in. Hg) by  using the following equation:
                                           /p  +  AH^
                                     My)  | a  ^ I
                          - 17.65  (vMy)  f  a  v™76]                  (9-2)

9.3  VOLUME OF WATER VAPOR

    The volume of the water vapor collected from flue gas is calculated as
follows:

                                      pH2o  VST
                           V   = Vo     *   _
                            ^        MH20   PST

                               - 0.04707 V^c                            (9-3)

                                      50

-------
9.4  PERCENT  MOISTURE  CONTENT

     The moisture  content  of the  stack (or  sample)  gas is calculated by the
equation

                                         VWS
                               BWS  =  „—,-g—                          (9-4)
                                ws    VMS *  VWS

9.5  STACK GAS MOIJEC0IAR WEIGHT

                   Md = 32 (f0)  + 44  (fc)  + 28  (1  - fQ - fc)               (5-1)


                        Mw = Md  (1  '  Bws} + 18 (Bws}                     (5-2>

9.6  STACK GAS VISCOSITY

     The viscosity of  the  stack gas may  be  calculated from the following
equation  (Williamson et al., 1983):

                   W - Cj  + C2TS  +  C3TS2  +  C,BWS + C5f0                 (5-4)

where u is in micropoise,  T in *C,  and

          Cj - 160.62
          Cfe-- 0.42952
          C3 - 1.0483  x 10"^
          Q, - -74.143
          Cg » 53.147

or for T in *R
          Cr = 51.05
          C2 - 0.207
          C3 - 3.24 x  10~5
          Q, - -74.143
          Cj .- 53.147

9.7  SAMPLE PLOW RATE

     The sample flow rate, at  standard conditions,  can be found by using the
following equation:

                                        VMS
                                  Q.   ' ' ~—                            (9~5)
                                  SST    8

     However, to calculate the particle cut diameter  of the inertial classi-
fier, it is necessary  to know  the flow rate through the sampler at the actual
sampler conditions.  This can  be  accomplished  by using the  following equa-
tion:
                                     51

-------
                             ST   1  -  Bws      TST
                        =  0.056  Qg    IT  1 B—)  l-p^J                    (9-6)


9.8  SAMPLER D5n

     The D^g or cut-point  of each  stage of the chosen  particle-sizing device
should be calculated  for accurate  determination  of  the particle size distri-
bution.  The cut-point is  primarily  a  function of the  actual  flow rate
through the sampler and the viscosity  and density of the  stack gas.

     The currently available data  concerning  calibration  of the SRI/EPA
Cyclone I show the behavior to be  described by the  equation
              05,3 = 0.15625  (j£) -0-2091 Qg-0-7091 U0.7091          (9_7)
If another inert ial sizing device  is used  in conjunction  with  CSR,  the D50's
for each of the stages should be calculated as described  in the vendor-
supplied operator's manual.

9.9  STACK GAS VELOCITY

     The stack gas velocity can be  found by using  the  following equation:
                                             T     1/2



9.10  NOZZLE VELOCITY

     The nozzle velocity  (u) , constant  for a specific  nozzle diameter (d) ,
nay be calculated from the following equation:
                                     3.056  Qg

                                U	—

9.11  NOZZLE £P LIMITS
(9-9)
     For each nozzle and stack gas temperature  range,  limits  on  acceptable
velocity pressure differentials, AP  .  and AP_   , were calculated  from equa-
                                   IBH1
tions 5-8 and 5-9 of the setup calculations.  To determine  if  the  run was
performed within these limits, the actual stack temperature  and  velocity AP
at each traverse point should be tabulated  and compared with the limits,
~pain an
-------
9.12  PERCENT ISOKINETIC SAMPLING

     The following equation should be used to determine  the  percentage of
isofcineti'c sampling:


                    U677 T8 [0.002.67 V£c +-(£-(Pa + ^5 ]
              1% -- „ ^ ps ^  • - - -         (9-10,

9.13  CONCENTRATION

     The concentration of the particulate matter caught  by each stage in
grains per standard cubic foot can be calculated by the  equation below

                                            Mp
                             C^, * 0.0154  (.y — )                        (9-11)
                                            nS

     The units can be converted to milligrams per dry normal cubic meter by
using the following:

                               C_ » 2293.2 C'                          (9-12)
                                mm

9.14  ACCEPTANCE CRITERIA

     To be acceptable, each run must meet the following  criteria:

     1)  The velocity at each traverse point must be within the  velocity
        limits for the nozzle used, or the velocity at one point may be below
        the minimum for the nozzle if the percent isokinetic  falls within
        100 - 20%.  This second specification allows for the acceptance of
        data with a positive bias but which are still within allowable error
        limits.
     2)  For single stage PM   samplers .
          o iVn must be 10 ± 1 urn.

        For multiple stage PM, 0 samplers

          o one stage cut must be at or greater than  10 urn and  two  succeeding
            size cuts must be le.°s than 10 urn
          o separation between successive size cuts must be no  more than  a
            factor of 2 and no less than a factor of  1.S
          o the product of jet velocity and size cut  for each stage with  cut
            above 2 urn should satisfy the criteria given by McCain  et  al.
            (1986).
                                     53

-------
                                  SECTION  10

                                  MAINTENANC2
     A notebook or other record of  all maintenance  procedures should be kept.
This will provide a definite  and current  record  of  all  information pertinent
to reliable operation of the  CSR sampling  system.   Maintenance of the system
should be performed as described previously  (Rom,  1972),  with the exceptions
noted below.

10.1  VACUUM SYSTEM

     The vacuum system may be checked as  follows:

     o    Insert a plugged 0.5-in.  male quick connect into  the sample inlet
          of the control console.

     o    Turn the pump switch to ON.

     o    Turn th<- coarse-adjust valve to  the ON position.

     o    Close fully the fine-adjust valve.

The vacuum gauge should read  about  25 in.  Hg when ambient barometric pressure
is near 30 in. Hg.  If this pressure cannot be achieved,  a  leak or sticking
pu»p vane should be suspected.  If  the leakage rate measured  by the dry gas
•eter exceeds 0.02 cfm, the leak or leaks  must be found  and corrected.   Parts
to check are the pump, vacuum gauge, metering valves, and tubing.

10.2  VACUCH POM?

     Host pump trouble can be corrected by flushing the  unit  according  to the
manufacturer's instructions rather  than disassembly.  A  noisy or  inefficient
pump is frequently caused by  nothing more  serious than a  vane stuck in  a
rocor slot because of foreign material in  the unit.  To  flush the unit,
follow the procedure given below:

     o    Separate the pump from the system.

     e    Slowly add several  teaspoons of  solvent at the  intake while the
          unit is running (recommended commercial solvents  include Loctite
          Safety Solvent, Inhibisol Safety Solvent, or Dow  Chemical
          Chlorothane).

     «    Lay the unit on its side with the outlet  downward so the solvent
          will work out again.

-------
10.3  MAGNEHELIC* DIFFERENTIAL  PRESSURE  GAUGES

     Magnehelic* differential pressure gauges are precision  instruments
assembled and precalibrated by  the manufacturer.  If  trained instrument
mechanics are not available, it is recommended that any  instruments  requiring
repair be returned to the factory.

     No lubrication or periodic servicing is required.   If the  interior is
protected from dust, dirt, and corrosive gases and fluids, years  of  trouble-
free service may be expected.

10.3.1  Zero Adjustment

     The indicating pointer should be set exactly on  che zero mark by  using
the external zero-adjust screw on the cover at the bottom.   The zero check or
adjustment can be made only if the high-pressure and  low-pressure taps are
both open to atmosphere.

10.3.2  Calibration Check

     For service requiring a high degree of continued accuracy, periodic
calibration checks are recommended.  In  general, the  Magnehelic*  calibration
should be checked by following the procedure below.

1.   As a comparison gauge, use a hook gauge, micromanometer, or  inclined
     gauge of known accuracy.

2.   Connect the Magnehelic* gauge and reference gauge with  two leads  from a
     "T." Connect rubber tubing to the third leg of the  "T," and  impose the
     .-fissure, slowly.

3.   Be certain no leaks exist in the system, and provide adequate time for
     comparison gauges to reach equilibrium, because  fluid drainage  and
     different dynamic characteristics can affect the reading.

10.3.3  Recalibration

1.   Remove the plastic cover.

2.   Remove the two screws holding the scale, and slide the  scale out,  using
     care not to damage the pointer.

3.   Loosen the two set screws in the range spring clamp (Dwyer part no.
     NUA-70B); move toward the helix to  increase the  range and  back  to
     decrease.  Secure the clamp with the set screws, replace the scale,
     check the gauge zero, and compare readings as in Section 10.3.2.

4.   Replace the cover.  The cover must  be tight and  leakproof  for accurate
     readings on the high-pressure side.  Observe the following procedure.

     a.  Place the cover in position with the notch engaged  and with the
         O-ring properly seated.

                                     55

-------
     b.  Jockey the zero-adjust screw into position so its hex end  is
         inserted in the socket set screw, which actuates the zero-adjusting
         mechanism.
     c.  Hold the cover in position and screw the bezel down snuggly.  The
         0-ring must take some squeeze to effect an airtight seal.
         Caution:  If the bezel binds because of galling action of  aluminum
         surfaces, lubricate sparingly with light oil or molybdenum sulfate
         compound.
     d.  Troubleshooting.
         1.  Gauge sluggish.
             - Leads may be plugged or leaking.
             - Cover may be loose or leaking.
             - Pointer may be touching scale.
             - Jewels supporting helix may be overtightened.
         2.  Gauge fails to indicate zero properly.
             - See comments above regarding sluggish readings.
             - Iron particles may be in a strong magnetic field between  helix
               and magnet.  If found, they may be removed by touching each
               particle and withdrawing it with a small screw driver.
             - Magnet may be shifted and touching helix.
         3.  Apparent inaccuracy.
             - See preceding comments.
             - Improper connections to pick up desired differential.
         4.  Consult factory for unusual conditions of temperature, pressure,
             etc., and the effect on gauge operation and accuracy.

10.4  DUAL MANOMETER

     The dual manometer may be checked as follows:

     o    Visually check the pitot and orifice manometer lines to ensure they
          are free of fluid.

     o    Check for leaks, especially around the fluid-level zeroing controls
          and drain screws.

     o    Wipe the dual manometer clean.  The back can be cleaned with
          compressed air, or the device can be removed from the control  panel
          and wiped clean.

     o    if the dual manometer is unusually dirty, clean as recommended on
          the instruction plate.

     o    Make sure that the manometer ports are open (1-1/2 turns  counter-
          clockwise from the seat) and the manometer lines are connected.

     o    Level the manometer and check the,fluid level.

     o    TO fill the manometer with fluid, remove the screw on the left
          side.  When the oil meniscus and the reflected image at zero are
          aligned, the fluid-level plunger (zeroing control)  should have
          about 1/4 to 1/2 in. travel inward.


                                     56

-------
          Note;  During rough shipment, the manometer  lines  should  be
          disconnected and the manometer ports closed  by  turning clockwise
          until sealed.

     o    If for any reason the manometer unit has been inverted, be  sure the
          floating check valves of the manometer have  returned  to their
          normal position.  These floating valves are  located under the
          manometer ports and must be in the normal position when the
          manometer is used.

10.5  PITOT TUBE

     The pitot tube should occasionally be inspected for  any deformation of
the pressure inlets, because this may change the pitot calibration  coeffi-
cient.  Any dents or nicks should be repaired or the pitot head should be
replaced if the damage warrants it.  Before each test  run, the  operator
should blow gently into each pitot inlet to check for  obstructions.   If  the
pitot tube lines are clear, the pitot tube gauge will  respond.  If  no
response is noted, the operator should blow out the pitot lines with
compressed air.  The pitot tube can be checked for leaks  by  plugging  one end
of the tube and applying a positive pressure at the opposite end.   If the
tube will not maintain pressure, a soap solution can be used to identify the
location of any leaks.

10.6 NOZZLES

     The sample nozzle should be visually inspected before any  testing.   If
repair is necessary, a plumb bob should be used for inside damage and emery
paper for outside damage.  After any nozzle repair, the nozzle  diameter
should be remeasured.  The knife edge of the nozzle should be covered with
serum caps or similar covers to avoid damage when the  nozzle is not in use.

10.7  THERMOCOUPLES

     The thermocouples throughout the system should occasionally be checked
against room temperature by using a mercury-in-glass thermometer as the
standard.  If any thermocouples do not read within ±5  "C, the thermocouples
or readout should be replaced or recalibrated.

10.8  SAMPLING PROBE

10.8.1  Probe Cleaning

     Before each field test, all lines of the sampling probe, including  the
pitot tube lines, should be cleaned according to the procedure  outlined
below.

     o    Clean the probe internally by rinsing, first with  tap water, then
          with distilled, deionized water, then with acetone or dichloro-
          me thane.
                                     57

-------
     o    Rinse the internal tubes with the chosen organic  solvent and allow
          them to air dry.

     o    visually inspect the probe for cleanliness,  and repeat the proce-
          dure if necessary.

     o    Rinse the pitot lines with water and blow them out  with compressed
          air.

10.8.2  Probe Heater Check

     The procedure below may be used to check the probe heater.

     o    Plug the probe heater line and controlling thermocouple into the
          control case, and turn the heat controller on.
          The indicator light on the controller should come on,  and the probe
          should become warm to the touch in a few minutes.  After a few
          minutes, the indicator light should begin to cycle  on  and off.

     o    If the probe does not heat, check the probe  for loose  connections.

     o    if the probe still does not heat, remove the probe  liner from the
          probe sheath for inspection of the heating element.

     o    After the probe lines have been removed, unwrap the insulation and
          visually inspect the probe heating element for shorts  or burned
          spots.  An ohmmeter can also be used to measure the resistance
          between leads  (approximately 17 ohms) and also to ground (infi-
          nite).  Deviations from these values indicate faulty wiring.

     o    After any electrical problem has been solved, rewrap the probe
          lines with insulating material, and reassemble the  probe.

10.9  CONDENSING SYSTEM

     If the CSR sampling train is operated in a Method 17 configuration, a
condenser and silica gel column may replace the Method 5 impinger tra:-.
assembly.  An impinger system will be required if the  "back-half" catch is to
be measured.  Whichever system is used for collection  of water vapor from the
sampled stack gas, it must be clean and free of leaks  before  being used.
Glass impingecs should be cleaned with distilled, deionized water and then
acetone and should then be allowed to air dry.  Stainless steel  condensers
should also be rinsed by the same procedure and allowed to  air dry, inverted
to ensure total drainage.  The drying can be speeded by blowing  out the
condenser with compressed air.  Silica gel columns  (along with condensers)
should be leak tested, along with the control box or separately,  by applying
positive pressure at the inlet and plugging the outlet.  Ideally, these
devices should maintain a pressure of at least 10 in.  Hg above absolute.
                                      58

-------
                                 SECTION  11

                             AUDITING PROCEDURES
     Routine quality assurance activities, such as equipment calibrations,
are essential to obtaining good data.  An assessment of  the quality  of  these
data may be made through an audit.  The audit must be performed with equip-
ment or standards independent of those used for actual measurements  to  ensure
that the tasks involved are being performed properly.

     The audits recommended for use in a program using the CSR method are
similar to those described for Method 5  (U.S. Environmental Protection
Agency, 1977, Section 3.4.8).  Two types of audits, performance and  system,
are commonly performed.

     Performance audits provide a quantitative evaluation of the quality of
data produced by a measurement system.  One type of performance audit recom-
mended for Method 5 assesses the accuracy of a system's  flow metering devices
with a critical flow orifice.  This is also recommended  for measurement pro-
grams using the CSR.

     A performance audit of data processing is also recommended.  As for
Method 5, an audit of this type can uncover and eliminate errors in  data
transfer, calculations, etc.  The flow of data from field data forms and
weight sheets to data reduction programs or hand calculations should be
traced for at least a portion of the data base.  Calculation of results for &
standard data set is another method by which data reduction procedures  may  be
audited.        .

     A system audit is a qualitative inspection and examination of the  proce-
dures and techniques used by the field team.  This type  of audit is  strongly
recommended .if the team is not familiar with the CSR method.

-------
r
                                                   SECTION 12

                               RECOMMENDED  STANDARDS  FOR ESTABLISHING TRACEABILITY
                       Although  the  use of  quality  control  checks  and  independent audits is
                  essential  to obtaining  data  of  the  desired  quality,  another  important consid-
                  eration  is the traceability  of  individual elements of  the measurement
                  process.   All  materials,  equipment,  and procedures used  should be traceable
                  to  a  standard  of reference.

                       Working calibration  standards  should be  traceable to primary or higher
                  level standards.   The system's  flow metering  devices should  be calibrated
                  against  a  wet  test meter  which  has  been verified as  required for Method 5
                  (U.S.  EPA,  1977).  The  performance  of  the analytical balance should be
                  checked  against class-S weights that are  traceable to  MBS standards.
                                                    60

-------
                             REFERENCES
Belyaev, S. P. and L. M. Levin.   Investigation of  Aerosol  Aspiration by
Photographing Particle Tracks under Flash  Illumination.  J.  Aerosol Sci.
3:127-140, 1972.

Belyaev, S. P. and L. M. Levin.   Techniques  for Collection of
Representative Samples.  J. Aerosol Sci. 5:325-338,  1974.

Farthing, W. E.  Evaluation and Recommendations of Protocols for  PM1Q  in
Process Streams:  Recommended Methods; Volume I.   SRI-EAS-83-1038,
Southern Research Institute, Birmingham, AL, 1983.

Farthing, W. E. and A. 0. Williamson.  "A  Unified  View of  Inertial
Impactors and Cyclones."  Presented at 1985  Annual Meeting of American
Association for Aerosol Research, Albuquerque, NM, 1985.

McCain, J. 0., S. S. Dawes, J. W. Ragland, A. D. Williamson.  Procedures
Manual for the Recommended ARB Particle Size Distribution  Method
(Cascade Impactors).  NTIS PB 86-218666 California Air Resources  Board,
1986.

Rom, J. J.  Maintenance, Calibration, and  Operation of Isokinetic Source
Sampling Equipment.   APTO-0576,  U.S. Environmental Protection Agency,
Research Triangle Park, NC, March 1972.

Shigehara, R. T.  Proposed Revisions to Reduce Number of Traverse Points
in Method 1 - Background Information Document.  EPA-450/3-82-016a,  O.S.
Environmental Protection Agency,  Research  Triangle Park, NC, 1982.

Smith, W. B., D. B. Harris, and R. R. Wilson, Jr.  A Five-Stage Cyclone
System for In Situ Sampling.  Environ. Sci.  Technol.  13(11): 1387-1392,
1979.

U.S. Environmental Protection Agency - EPA Method  1.  40 CFR Part 60,
Appendix A.

U.S. Environmental Protection Agency - EPA Method  2.  40 CFR Part 60,
Appendix A.

U.S. Environmental Protection Agency - EPA Method  3.  40 CFR Part 60,
Appendix A.

U.S. Environmental Protection. Agency - EPA Method  4.  40 CFR Part 60,
Appendix A.


                               61

-------
                      REFERENCES  (Continued)
U.S. Environmental Protection Agency - EPA Method  5.   40  CFR Part 60,
Appendix A.

U.S. Environmental Protection Agency - EPA Method  6.   40  CFR Part 60,
Appendix A.

U.S. Environmental Protection Agency - EPA Method  17.  40  CFR Part 60,
Appendix A.

U. S. Environmental Protection Agency.  Quality Assurance Handbook for
Air Pollution Measurement Systems, Volume III: Stationary Source
Specific Methods.  EPA-600/2-77-026, U.S. Environmental Protection
Agency, Research Triangle Park, NC, 1977.

Wilke, C. R.  A Viscosity Equation for Gas Mixtures. J. Chera.  Phys.
8:517, 1950.

Williamson, A. D., D. L. lozia, P. V. Bush, W. E.  Farthing,  J.  D.
McCain, and W. B. Smith.  Development, Application, and Support of
Particulate Sampling Procedures.  Third Annual Report  (1982).
SRI-EAS-83-348, Southern Research Institute, Birmingham,  AL,  1983.
                               62

-------
                                  GLOSSARY
Aerodynamic diameter:  The aerodynamic diameter of a particle is the diameter
     of a sphere of unit density whch has the same settling velocity in  the
     gas as the particle of interest.  For spherical particles with diameter
     Dp, larger than a few microns and gas conditions of interest for source
     PMjg, the aerodynamic diameter is essentially given by /p~~ D  where p
     is the particle density.                                 P  P


Cut-point:  The cut-point of a size classifier is the particle diameter  for
     which all particles of equal or greater diameter are captured and all
     particles with smaller diameters are not captured.  No real device
     actually has a sharp step function cut-point, but the theoretically
     defined D   of a stage is often called its cut-point.
Geometric standard deviation, o : A measure of dispersion in a lognormal
     distribution.  It can be calculated as the ratio of particle diameter
     corresponding to a cumulative percent of 50 to the diameter
     corresponding to a cumulative percent of 15.86.
                                    63

-------