Environmental Protection Technology Series
FILTRATION CHARACTERISTICS  OF
    GLASS  FIBER FILTER MEDIA  AT
         ELEVATED TEMPERATURES
                  Environmental Sciences Research Laboratory
                      Office of Research and Development
                     U.S. Environmental Protection Agency
                Research Triangle Park, North Carolina 27711

-------
                RESEARCH REPORTING SERIES

Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have  been grouped  into five series. These five  broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
     1.    Environmental Health Effects Research
     2.    Environmental Protection Technology
     3.    Ecological Research
     4.    Environmental Monitoring
     5.    Socioeconomic Environmental Studies

This report  has been  assigned  to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to develop and
demonstrate  instrumentation, equipment, and methodology to repair or prevent
environmental degradation from point and non-point sources of pollution. This
work provides  the new  or improved technology required for the control and
treatment of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                                               EPA-600/2-76-192
                                               July 1976
     FILTRATION CHARACTERISTICS OF GLASS FIBER
       FILTER MEDIA AT ELEVATED TEMPERATURES
                        by
     Dale A. Lundgren and Thomas C. Gunderson
 Department of Environmental Engineering Sciences
               University of Florida
            Gainesville, Florida  32611
               Grant No. 803126-01-0
                  Project Officer

               Dr. Kenneth T. Knapp
Emissions Measurement and Characterization Division
    Environmental Sciences Research Laboratory
         Research Triangle Park, NC  27711
    ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
        OFFICE OF RESEARCH AND DEVELOPMENT
       U.  S.  ENVIRONMENTAL PROTECTION AGENCY
         RESEARCH TRIANGLE PARK, NC  27711

-------
                            DISCLAIMER






     This report has been reviewed by the Environmental Sciences



Research Laboratory, U.S. Environmental Protection Agency, and



approved for publication.  Approval does not signify that the



contents necessarily reflect the views and policies of the U.S.



Environmental Protection Agency, nor does mention of trade names



or commerical products constitute endorsement or recommendation



for use.
                                li

-------
                            ABSTRACT

     Particle collection characteristics of a newly developed,
high-purity "Microquartz" fiber filter media and a Gelman Type A
glass fiber filter media were evaluated over a range of tempera-
tures 20°C to 540°C (68°F to 1004°F), particle sizes 0.05 ym to
26 ym, gas velocities 0.5 cm/sec to 51 cm/sec, and particle vola-
tilities.  Both types of high efficiency filters proved adequate
(>99.9% efficiency) for sampling nonvolatile particles over the
above variable ranges.  Nonvolatile particle penetration decreas-
ed with increasing temperature and increasing filter loading.

     The effect elevated temperature had on particle collection
characteristics was not a determining factor in application of
high efficiency filters.  The main problems encountered in the
high temperature environment were filter holder leakage and vola-
tilization of gas-borne particles which passed through the filter
media.
                               111

-------
                       CONTENTS

Abstract	i i i
Tables	vii
Figures	v i i i

    I.  General Introduction	1
   II.  Literature Review:  Filtration of Nonvolatile
        Aerosol s	3
        A. Theory	3
        B. Experimental  Data of Others	10
  III.  Literature Review: Filtration of H2$04	13
        A. Theory	13
   IV.  Description of Experimental Appartus and
        Methods	20
        A. General  Experimental Arrangement	20
        B. Aerosol  Generation	22
        C. Determination of Aerosol Size Distributions.24
        D. Temperature and Velocity Measurements	25
        E. Techniques to Determine Filter Efficiency...26
        F. Filters and Filter Holders	29
    V.  Results and Analysis	31
        A. Aerosol  Size  Distributions	31
        B. Loading Effects on Filtration Efficiency....31

-------
         C. Velocity Effects on Filtration Efficiency...34
         D. Paricle Size Effects on Filtration
            Efficiency	35
         E. Temperature Effects on Filtration           35
            Efficiency	35
         F. Filter Holder Leakage	36
         G. HUSO. Filtration at Elevated Temperature....37
         H. Pinhole Effects on Filtration Efficiency	38
         I. Comparison of Theory with Experimental Data.38
         J. Experimental Errors	39
   VI.   Conclusions	41

Symbol s	43
References	45
Appendices
    A. Conversion of SO. to H2SO.	50
    B. "Microquartz" filter at 5000x magnification
       (scanning electron microscope photo)	52
    C. Gelman Type A filter at 5000x magnification
       (scanning electron microscope photo)	53
                          VI

-------
                        TABLES
Number                                                Page
   1       Typical  exhaust-gas  composition  from
          coal-fired boiler.                             54
   2       Amount of H2S04 found in  particulate
          matter by various  stock  sampling
          methods.                                       55
   3       Size distribution  of aerosols  produced
          in Collison atomizer with impactor
          (water solvent).                               56
   4       Lui's data on  performance of Collison
          atomizer  with  impactor (water  solvent).        56
   5       Size distribution  of experimental  aero-
          sols.                                         57
   6       Effect of temperature on  filtration
          efficiency.                                   58
   7       H2S04 distribution  in sampling train.          59
   8       Theoretical and experimental  penetra-
          tions of  "Microquartz" filter  media.          60
                          vii

-------
                            FIGURES

Number                                                       Page
   1   Structure of "tree"  on a  metal  fiber.                     61
   2   Effect of temperature on  penetration  (theoretical).       62
   3   Effect of temperature on  penetration  (experimental).      63
   4   Effect of temperature on  collection mechanisms  of
      impaction,  interception and  diffusion.                    64
   5   Equilibrium conversion of S02 to  $63.                     65
   6   Equilibrium conversion of S03 to  H2S04  at  8.0
      vol% H20 in flue  gas.                                    66
   7   Dewpoint and condensate composition for vapor mix-
      tures of H20 and  H2S04 at 760 mm  Hg total  pressure.       67
   8   H2S04 dewpoint  for  typical flue gas moisture con-
      centrations.                                              68
   9   Variation of dewpoint with H2S04  content for gases
       having different H20 contents.                           69
  10   H2S04 dewpoint  obtained by various  investigators.         70
  11   Dewpoint as a function of H2S04 concentration.            71
  12   Relation of dewpoint  and  S03  content  of combustion
      gases to sulfur content of oil.                           72
  13   General arrangement  of test  apparatus.                    73
  14   Clamp assembly  for  glass  filter holder.                   74
  15   Effect of filter  loading  on  penetration.                  75
  16   Method to determine  penetration at  7.0  yg/cm^ loading.   76
  17   Effect of velocity  on penetration
      ("Microquartz"  Filter).                                  77

-------
Number                                                     Page
  18  Effect of velocity on  pentration
      (Gelman Type  A Filter).                                 78
  19  Pressure drop  vs., velocity.                            79
  20  Effect of particle size  on  penetration
      ("Microquartz" Filter).                                 80
  21  Effect of particle size  on  penetration
      (Gelman Type  A Filter).                                 81
  22  Effect of temperature  on penetration                   82
  23  Pressure drop  vs. velocity at elevated tempera-
      tures.                                                 83
  24  Effect of pinholes on  penetration.                      84
                          IX

-------
                 I.  GENERAL INTRODUCTION

        Glass fiber filter media is widely used for the
collection of participate matter from stack effluents.
This filtration media is both inexpensive and highly effi-
cient.  It does, however, have a fairly high and variable
background of extractable impurities, which often inter-
feres with chemical analysis of collected particulate
matter.  To overcome this problem of a high extractable
background, EPA supported development of a high-purity
filter media made from Oohns-Manvi11e 99.2% "Microquartz"
fibers.  This development project, performed by A.  D.
Little, Inc. , produced a filter with a low extractable
background suitable for stack gas sampling at temperatures
in excess of 500°C (932°F).
        When hot stack  gases  are sampled  with  EPA Method 5
              2
sampling train , a significant fraction of what may be
considered particulate matter is often found in the impin-
gers downstream from the filter.  The efficiency of glass
fiber filter media has, therefore, been questioned.  Further-
more, doubts about the effect of high temperature on filter
performance and about filter efficiency for volatile parti-
cles have arisen.   To answer these questions EPA supported
this program to evaluate glass fiber filters,  particularly
the "Microquartz"  filter, over a reasonable range of tempera-
tures, velocities,  particle  sizes, and particle volatilities:
                         1

-------
      Temperature Range:   20°C to 540°C (68°F to 1004°F)
         Velocity Range:   0.5 cm/sec to 51  cm/sec
                          (=1  ft/min to 100 ft/min)
Particle Diameter Range:   0.05 ym to 26 ym
   Particle Composition:   Volatile and Nonvolatile

-------
II.  LITERATURE REVIEW:  FILTRATION OF NONVOLATILE AEROSOLS

A.  Theory

1  .  Introduction

       The three factors which affect nonvolatile aero-
sol filtration are the dispersed particles, dispersing
gas, and fibrous filter.  Characterizing the dispersed
particles are their size or size distribution, shape,
mass, electrical charge, and concentration.  Charac-
terizing the dispersing gas are its velocity, density,
absolute temperature, pressure, viscosity, moisture
content, and composition.  Finally, the fibrous filter
is characterized by its surface area, thickness, fiber
size or distribution of fiber sizes,  filter  porosity,
specific fiber surface area, fiber composition, electri-
cal charge, and surface characteristics.
       Two phases may be distinguished in the filtration
process—the primary phase and the secondary phase.   In
the primary phase, aerodynamic capture of particles occurs
in a clean filter,  so the following assumptions are usu-
ally made for models of this phase:
       a)  Deposition of individual particles does not
           influence filter efficiency.
       b)  Filter efficiency is time independent.
       c)  Any particle that touches a filter fiber is
           retained.

-------
       d)  Particles are spherical.
       e)  All fibers have the same diameter.
       f)  Filter porosity is uniform.
       g)  Electrostatic, thermal, and gravitational
           effects are negligible.
       h)  No phase transitions occur in the aerosol
           during temperature changes.
       In the secondary filtration phase,structural changes
occur as a result of particle deposition; therefore efficiency is
                                              3
time dependent.  So-called "secondary effects"  appear,
consisting of deposition of particles upon one another,
dendrite formation, fusion and flushing of drops on the
fiber surfaces, capillary effects, loss of electrical
charge, clogging,and so on.
       The solution to the basic problem of predicting
filtration efficiency is easier when regarded from the point
of view of the first phase.  Relatively good results have
been obtained using primary'phase assumptions, while pub-
lished work including "secondary effects" has been empirical
in nature.   The section which follows discusses the basic
collection mechanisms from the primary phase viewpoint
approach.

2.  Filtration Efficiency Equations

       Aerosol filtration by  fiber filters involves the
three main  collection mechanisms of diffusion, direct inter-
ception, and inertia!  impaction.    Diffusive capture of

-------
particles is described by the dimensionless parameter N^:

                        ND = D/DV                     [1]

where  ND = Dimensionless diffusion parameter
       "D  = Diffusivity
       D  = Fiber diameter
       V  = Filtration velocity
Particle diffusivity is so related to absolute tempera-
ture and several other temperature dependent terms, that
diffusive capture increases with increasing temperature
(particularly for submicron particles).  The following
equations represent these relations:
                    D" = CkT/Sirnd                      [2]

where  I) = Diffusivity
       C = Cunningham correction factor
       k = Boltzmann's constant
       T = Absolute temperature
       n = Gas viscosity
       d = Particle diameter

        C = 1  + (X/d)[2.514 + (0.8)exp(-0.55d/X)]     [3]

where  C = Cunningham correction factor
       X = Mean free path of gas molecules
       d = Particle diameter

-------
                     X = n/(0.499pGc)                   [4]

where   A  = Mean  free path of gas molecules
        n  = Gas viscosity
        PG = Gas density
        c  = Average  velocity of  gas molecules
         n = n0[(273.1 + G)/(T + GJKT/Z/S.l)'        [5]
where  n  = Gas viscosity  at temperature T
       n0 = Gas viscosity at temperature 0°C
       T  = Absolute temperature,  °K
       G  = Constant = 114 for air at  1 atm
         PG = 1.293 X 10"3/[(1 + 3.67  X 10"3)H]        [6]
where  p~ = Air density,  gm/cm
       H = Temperature, °C
                     c = (8RT/TTM)0'5                   [7]

where  c = Average velocity of gas molecules
       R = Universal gas constant
       T = Absolute temperature
       M = Gas molecular weight
Direct interception is defined by the  temperature indepen-
dent parameter NR:

                        NR = d/D                       [8]
where NR = Dimensionless  interception parameter

-------
       d  = Particle diameter
       D  = Fiber diameter
 Inertial impaction is described by the Stokes1 number:

                    STK = Cppd2V/18nD                 [9]

where  STK = Stokes1  number
       C   = Cunningham correction factor
       p
        p  = Particle density
       d   = Particle diameter
       V   = Filtration velocity
       n   = Gas viscosity
       D   = Filter diameter
Gas viscosity is the factor most affected by increasing
temperature, rising with it.  Hence, effectiveness of
inertial  impaction as a collection mechanism decreases
with increasing temperature.
        Efficiency of a filter mat is a function of NR, ND> and STK
and is usually calculated from collection efficiency of the
individual  fibers comprising the mat.  Many individual
equations have been developed and recently were reviewed
       7                                      8
by Yen.  One representative equation by Davies  is:
Es = (0.16  + 10.9a -  17a2)[NR + (0.5 + 0.8NR)(NR + STK)
                  - 0.105NR(ND  + STKT]               [TO]

where  E   = Single fiber efficiency for diffusion,
             interception and inertial  impaction
       NR  = Dimensionless parameter for interception
       Nn  = Dimensionless parameter for diffusion

-------
       STK = Stokes'  number
       a   = Packing  density = W/Lp_
                                   F
       W   = Mass per filter face area
       L   = Filter thickness
       p,.,  = Fiber density
        r
Filter mat efficiency can then be related to individual
                                          Q
fiber efficiency by the following equation :
               EM = l-exp[-4EsW/TrpFD]                 [11]
where E^ = Filter mat efficiency
      E  = Single fiber efficiency for diffusion,
           interception, and inertial impaction [1 0]
      W  = Mass per filter face area
      PF = Fiber density
      D  = Fiber diameter

3.  Secondary Processes in Aerosol Filtration
       Filtration efficiency theories so far discussed are
based on particle capture by clean fibers.  Actually, a
deposit builds up which may reduce the filter pore size and
increase filter pressure drop.  Often, deposited particles do
not distribute themselves evenly over the surface of the
fibers, but form chain aggregates which act as collection
bodies and may capture particles more effectively than the
filter itself.  This  process has been described by Watson
and Leers   and is illustrated in Figure 1.
       Time variation of penetration and pressure drop of
                          8

-------
a filter during use depends upon the filter structure, fiber
shape, and nature of the aerosol.   Several  investigators,
                       3
including Radushkevich,   have suggested that penetration
descreases exponentially with time according to the equation
                     P = P0e-bt                       [12]
where P  = Penetration at time t=t
      PQ = Penetration at time t=0
      b  = Constant under given conditions

4.  Summary
       The filtration theory presented explains the tempera-
ture effect on filter efficiency, which is defined as
efficiency dependence on temperature of the filtered gas
while all other conditions remain constant.  However, the
theory rests on very limiting assumptions (Section IIA1),
which do not always hold, and does not allow for any "secon-
dary effects."  Nevertheless, filtration theory does illustrate
both qualitatively and quantitatively that with only tempera-
ture increasing Figure 2):
       1.  Inertial deposition decreases due to the increase
           in gas viscosity.
       2.  Direct  interception  is essentially
           indenpendent of temperature.
       3.  Diffuse deposition increases due to greater
           particle diffusivity  at higher temperatures
           (particularly for submicron particles).

-------
B.  Experimental Data of Others
     First e_t aK   measured the efficiency of ceramic fiber fil-
ters [capable of withstanding temperatures up to 1093°C (2000°F)]
at temperatures of 21°C (70°F) and 760°C (1400°F).  Values of
EM = 85% at 21°C (70°F) and EM = 82% at 760°C (1400°F) for fibers
with a D = 20 ym were determined.  For D = 8 ym, the correspond-
ing values were EM = 98% and EM =91%.  For all  experiments, the
                 M            rl
                                                        3
same conditions [d = 1 ym, V = 178 cm/sec, p  =  6.4 g/cm ] were
maintained to keep the inertial mechanism of particle deposition
dominant.  The efficiency decreased for the three filters with
increasing temperatures as predicted by theory.
     Pich and Binek   reported measurements of temperature char-
acteristics of a filter with D =1.2 ym from 20°C (70°F) to 200°C
(392°F)(Figure 3).  NaCl cubic particles with each edge 0.2 ym
and p  = 2.16 g/cm  at V = 0.6 cm/sec were used.  Under these
conditions, diffusion deposition predominated, and efficiency
increased from 94% [20°C (70°F)] to 98.5% [200°C (392°F)], re-
flecting filtration theory.
     Dyment   reported results of some glass fiber filter tests
with NaCl aerosol at temperatures up to 500°C (932°F).  He
found glass paper shrinks  at these  temperatures, so he
                               10

-------
preheated the samples to avoid cracking.  His paper does
not mention D, d, p  , and V.  However, at ambient tempera-
ture, he found penetration too low (<0.001%) to be regis-
tered.  The temperature was gradually raised to 440°C
(824°F) when a penetration of 0.001% was measured.  At
520°C (968°F) this increased to 0.09%.  His conclusions
were as follows:
       "In practice the effect of temperature and pressure
on filtration mechanism and performance has not been found
a determining factor in the application of high efficiency
filters.   The main problems are the physical and chemical
effects of a high temperature environment on the materials
of construction of the filter which are manifested by
reduced mechanical strength and resilience of loss of
adhesion, leading to mechanical leakage and loss in
efficiency."
       More recent high temperature filter efficiency tests
were reported by First.    Heat-shrunk quartz fiber filters
were tested at temperatures up to 510°C (950°F) with a
polydisperse NaCl aerosol  of 0.14ym mass median diameter
(mmd)  at  velocities (V) of 15 cm/sec          to 30 cm/sec.
Average penetrations ranged from 0.032% to
0.078% with no consistent  trend with respect to temperature.
Individual readings of NaCl penetration varied as much as
+ 50%  about the mean during each test series on an identical
filter, so a trend would be difficult to detect.
                        11

-------
       Thring and Strauss   carried out filter efficiency
calculations for a filter with D = lOym, V = 25 cm/sec,
and a temperature range of 0°C (32°F) to 1600°C (2912°F).
Results for d = O.Olym, d = O.lym, d = 5pm (Figure 4)
show effect of temperature on inertial impaction, inter-
ception, and diffusion collection mechanisms.
                        12

-------
  III.  LITERATURE REVIEW:  FILTRATION OF' H2S04 AEROSOLS

A.  Theory

1.  Introduction

        A major problem in stack sampling is how to clas-
sify pases at duct condition which condense and/or react
in the filtering train and medium to form what may be con-
sidered particulate matter.  One substance which falls into
this category is sulfur trioxide (SO^).formed in equilib-
rium with sulfur dioxide (SO^) when sulfur-containing
fossil fuels are burned.  Up to 5% of the total sulfur in
the fuel is converted to S03, yielding from 5 to 50 ppm
                    1819
S03 in the flue gas.   '    The S03 is in equilibrium with
water (H^O) vapor in the flue gas and, depending on tempera-
ture and gas component concentrations, various amounts of
sulfuric acid (H^SO.) vapor will be formed.   This H2SO.
can be collected on filters and weighed as particulate.
        The importance of establishing whether or not con-
densed SOo/HpSO. is to be considered particulate matter is
pointed out by a 36% average contribution of this material
to total measured particulate grain loading (oil-fired
                                            20
boiler emissions) as  reported  by Jaworowski.     As  fly ash
emission levels are reduced by air pollution control  equip-
ment,  this amount of condensed SO^/hLSO. may equal  or
surpass dry particulate contribution and could prevent
compliance under existing regulations.
                         13

-------
2.  SOX, HgO and HgSO^ Equilibria in Flue Gas

        Most sulfur in power plant flue gases appears as
S02 (Table I),21 with typical S03 levels ranging from 1.0% to
2.5% of the S02-  However, as Figure 5  shows, the equilib-
rium constant for the reaction:
                       S02+l/2 02JS03
strongly favors the formation of SO., at temperatures below
about 540°C (1000°F). This graph was calculated from data
                22
cited by Hedley.    Kinetics of the reaction are unfavor-
able in  the absence of a catalyst, but thermodynamical ly
the  SO-   concentrations  could    exist at levels much
greater than those normally encountered.  Ratios of SO., to
                                      23
SOo as high as 0.1 have been reported.    Since formation
of SO., is controlled by catalytic effects as well as amount
of excess air present, concentration of S03 resulting from
combustion of a particular fuel can only be estimated in
absence of direct measurements.
        Reaction between H^O vapor and SO, is given by:
        Figure 6 shows equilibrium conversion of S03 to
H^SO, as a function of temperature for a typical flue gas
H20 vapor concentration of 8 vol%.  Appendix A illustrates
the calculation method used in obtaining this curve.  At
temperatures below 204°C (400°F), essentially all S03
present is converted to H2SO. at equilibrium.  In contrast
                         14

-------
  to formation of SO-, formation of H2SOa occurs rapidly in



  the thermodynamically feasible temperature range.


                    \


  3.  Determination of H2$0. Dewpoint





          Fly ash particles can influence the apparent dew-



  point (saturation temperature of H^O^ in flue gas), but



  one commits practically no error by neglecting the presence


                                                           25
  of other gases and considering only the HpSO.-HpO  system.



  Thermodynamic analysis  of the H^SO.-H^O-flue gas system,



  ignoring fly ash effects, provides a theoretical basis for



  predicting acid dewpoints and condensate composition from



  vapor-liquid equilibria data.


              26
          Abel   was the  first to derive a relationship ena-



  bling calculation of H2S04,  H20 and S03 partial  pressures



  from enthalpy, entropy, free energy, and heat capacity



  values.   From  his H2SO«  partial  pressures and Greenwalt's



  H20 partial pressures over H2$04 solutions, H2$04-H20



  dewpoint charts were prepared, Figures 7 and 8 .  The range



  of uncertainty indicated  by  Abel  is on the order of 5°C



  (9°F) at 10 vol% H20 vapor.



          Information contained in  Figure 7 can be used to



  predict  dewpoint temperature from an analysis of H2SO. and



  H20 vapor  content.   If  gas is cooled below its dewpoint,



1  condensate equilibrium  concentration and mass can  be ob-



  tained.   Condensate mass  predicted from use of the dewpoint



  chart is actually a prediction of the  amount avai Table for



  condensation.   The actual  amount  of condensate depositing



                           15

-------
on a fiber or metal surface may differ from the chart pre-
diction because of mass transfer considerations.
        As an example of the use of the chart, consider a
flue gas containing 10 ppm H^SO. and 10 vol% H^O vapor.
Condensation would occur at about 135°C (275°F), and conden-
sate composition at that point would be about 79 wt% H^SO
If the gas were cooled to 121°C (250°F), 85% of the H2$04
would be removed from the gas phase and an insignificant
amount of H20 vapor would also condense.  Condensation,
therefore, follows the 10 vol% water line, resulting in a
condensate which would be the equilibrium composition of the
condensate at 121°C (250°F), assuming the vapor phase is in
equilibrium with the total liquid condensed.  Composition
change of the liquid is small over the temperature interval
given in this example, ranging from 79 vol% at 135°C (275°F)
to 75 vol% at 121°C (250°F).
        Large changes in HpO vapor content of flue gases
cause only slight changes in acid dewpoint.  Variation of
dewpoint with HLSO. content of gases having different H^O
vapor concentrations is shown in Figure 9, where the range
from 0.5 vo~\% to 15 vol% H^O vapor changes dewpoint only
17°C (30°F) to 22°C (40°F) for the medium-to-high acid con-
centrations indicated.
        In addition to the procedure based on calculated
partial  pressure, a number of efforts have been made to
determine H^SO- dewpoints from  instrumental and chemical
procedures.  Figures 10 and 11 present results obtained for
flue gas dewpoints as a function of H2S04 content by various
                         16

-------
 investigators.  To make an exact comparison, all curves
 should be for a gas of the same vol% H20 vapor.  However,
 reference to Figure 7 will indicate that a variation in H20 vapor
 concentration from 7 vol% to 10 vol% can cause only about
 1°C  (2°F) to 2°C  (4°F) change in dewpoint.  Taylor's 29
 results were obtained from  an electrical dewpoint meter
 which is inaccurate at low acid partial pressures.  Lisle
                  30
 and  Sensenbaugh's   data were obtained with  a spiral con-
                                   •3-1        or     p c
 denser.  Dewpoint curves of Gmitro,   Muller,   Abel   and
          27
 Greenewalt   were based on calculated partial pressures.
        In view of the difficulties with calculation based
 on liquid phase thermodynamic properties and inaccuracy of
 dewpoint meters at low acid partial pressure, the most reli-
 able method of correlating H2$04 dewpoints with H^O and
 H?SO. vapor concentration is the experimental condensation
 method employed by Lisle and Sensenbaugh.  Their data corre-
 late.best with Muller's calculated dewpoints and are  the
 basis for ASME Power  Test  Code  19.10.
                          32
        Rendle and Wilsdon   have also published some data
 on relation of the SO^ content of combustion gas and of gas
 dewpoint to sulfur content of fuel oils> Figure 12 .  Results
 of several  other  investigators have also been plotted.   The
 type of oil, ash  content, and combustion conditions differ
 for the various sets of points.   Although the plot of SO-
 content shows considerable scatter, it is apparent that
with more than 0.5% sulfur in the oil, SO, content of the
 gas does not increase proportionately to the fuel  sulfur  %.
                         17

-------
        Figure 12 indicates the following:
        1.  There is a rapid initial rise in dewpoint
        with the first increment of sulfur in the fuel.
        For an estimated dewpoint of 38°C (100°F) with
        no sulfur, an increase to 127°C (260°) (H2$04
        dewpoint) is found with 1% sulfur.
        2.  There is a relatively small rise in dew-
        point as sulfur in the fuel oil increases
        from 1% to 6%.
B.  Experimental Data of Others
                          21
        Hillenbrand e_t a^L   sampled flue gas from a coal-
fired boiler with quartz filters maintained at 205°C (400°F)
and 138°C (280°F) and found that filter, temperature significantly
affected the amount of H2$0. found on the filter.  At 138°C
(280°F), 45% and 41% of the total H2S04 catch (total H2$04
mass in probe, filter, and impingers) were found on the
filter in two trials.  At 205°C (400°F), 24% and 8% of the
total t^SO. catch were found on the filter in two trials.
The greater amount of H2$04 found on the cooler filter was
interpreted by them to mean:
        1.  A considerable portion of H2SO, collected
        on the filter resulted from both condensation
        and reaction of particulate with the S02 and
        so3.
        2.  Condensation and consequent reaction is
        favored at lower temperature.
                  20
        Jaworowski    sampled flue gas from several  oil-fired
                         18

-------
boilers with  three different sampling methods:  EPA
Method 5 sampling train, ceramic thimble apparatus, and a
high-volume sampling system.  In all  three sampling meth-
ods, temperature of the filter was kept between 120°C (250°F)
and 150°C ( 300°F).  His results (Table 2) show the magni-
tude of HpSO^ contribution to total  particulate grain
loading ranged from 18% to 78%,  and  averaged 36% of the
total  measured emissions.
                        19

-------
IV.  DESCRIPTION OF EXPERIMENTAL APPARATUS AND METHODS

A.  General Experimental Arrangement

       The experimental apparatus provided capabilities
for:  controlled aerosol generation, filter efficiency meas-
urement (by fluorescent, gravimetric, atomic absorption,
titrametric or microscopic counting techniques) and aerosol
size distribution determination.
1  .  Nonacid Aerosols

       The general arrangement of the experimental equipment
is shown in Figure 13.  Air was supplied from the building's
compressed air system, filtered, and the stream split:  one
portion was used to operate an aerosol  generator and the
other to flow through an ion generator.  The ion generator
produced a high concentration of ions to accelerate attain-
ment of a Boltzmann charge distribution on the aerosol in
the conditioning chamber.  Part of the aerosol leaving the
conditioning chamber was exhausted and part heated in a
stainless steel coil located in an oven.  Aerosol temperature
was measured with an iron-constantan (Type J) thermocouple,
whose output was registered on a potentiometer (an ice point
reference juction was used with the thermocouple).  The
heated aerosol then was split into two equal streams.  The
reference stream was piped out of the oven and consecutively
through a coil immersed in cooling water, a high efficiency
                         20

-------
particulate filter, a volumetric gas meter, a flow regulating
valve, and a vacuum pump.  The test stream first passed
through the test filter, also in the oven, and then followed
a route identical to that of the reference stream.  Test
filter efficiency was determined by using fluorescent, gravi-
metric, titrametric, and microscopic counting or atomic
absorption techniques, depending on the type of test aerosol,
to compare collected aerosol masses on the reference filter
and backup filter following the test filter.

2.  H2S04 Aerosols

       A special sampling train was used to experiment with
hLSCL aerosol.   The train consisted of a heated box, to
keep a stainless steel coil and test filter at the desired
temperature; two Greenburg-Smith impingers filled with 100 ml of
80% isopropanol - 2Q% deionized water and immersed in an
ice bath (to collect any HpSO, that passed through the test
filter); a backup filter to catch any carry-over mist from
the impingers;  a gas volumetric meter; a flow regulating
valve; and a vacuum pump.  Details of how H^SO. was generated
are in the next section.
                          21

-------
B.  Aerosol Generation

1.  The Collison Atomizer

     A Collison atomizer, as described by Green   and Whitby
      37-39
e_t a_l_.     was used to generate aerosols in the 0.05 urn to
                                                          p
0.14 ym mmd range.  The atomizer was operated at 2.1 kg/cm
(30 psig) pressure, which produced an 11 &/min aerosol  stream
with =3 ym mmd particles.  Flow was increased to 76 &/min. upon
dilution by the charge neutralizing stream from the ion generator.
Placing an external impactor with a cut size of = 2 ym in the line
reduced particle size to about 1 ym mmd.
     Solute-solvent combinations of uranine in distilled water
and dinonyl phthalate in ethanol were used in the Collison atom-
izer.   If generated droplets contained a dissolved solute, then
upon evaporation,  the diameter of the aerosol is:
          d = C1/3Dd                                         [13]
     Whrre:   d = Particle diameter
             C = Ratio of solute volume to solvent volume
                 plus solute volume
             D. = Droplet original  diameter
                               22

-------
Thus.'an 8 vol % solution nebulized with a mean droplet dia-
meter of 1 ym would produce a mean particle diameter of
0.43 ym.
        Aerosol particles resulting from evaporation of the
liquid phase sometimes carried significant excess  static charge,
which was effectively and quickly reduced by contact with
a stream of bipolar ions. '""^  The ionizer used to produce
the ion stream was operated at 3000 volts A.C. with an air
flow of 61 £/min.   A conditioning chamber of approximately
60£ capacity allowed achievement of an equilibrium charge
distribution.  Effectiveness of charge reduction was easily
observed by measuring aerosol transmission loss in a short
section of 0.63 cm           O.D. Tygon tubing.  For 0.14 ym
mmd uranine aerosol flowing at 1 &/min              losses
amounted to about  20% per meter without the ionizer and less
than 5% per meter  with it.
2.  Spinning Disk  Aerosol Generator
        A spinning disk aerosol generator 36-38  was used
to generate aerosols in the 0.5 ym to:6.0 ym mmd range.
The disk's air motor speed, ionizer voltage, satellite air
flow, liquid feed  rate, and all other variables were held
constant to produce a main  droplet diameter of about 35 ym.
Equation 13 was again used  to predict aerosol  size. Differ-
ent size aerosols  were generated by varying only solute
                         23

-------
concentration.   Uranine ethanol  was the solute-solvent com-
bination for all  aerosol  sizes generated with the spinning
disk.
3.  ^$04 Aerosol Generation
              aerosol was generated by evaporating 0.10 N
H2S04  in a stainless steel coil located in a 370°C (618°F)
oven.  The ^$04 solution was metered at a constant rate
with  a 50 ml syringe driven by a constant speed syringe
drive.  A constant flow of ambient air was metered into
the coil  used for evaporation of the ^504.  It was essen-
tial to add the acid uniformly to prevent undue fluctuations
in the gas composition.
4.  Other Methods of Aerosol Generation
        For particles greater than 6.0 ym, Paper Mulberry
Pollen (= 13 vm mmd) and Ragweed Pollen (- 26 ym mmd) were
mechanically dispersed near the inlet to the filters with
a rubber squeeze bulb.  Fly ash with a 5.6 ym mmd  (ag =
2.09) was also dispersed with compressed air near  the inlet
to the filters.
C.  Determination of Aerosol Size Distribution
        An electrical aerosol size analyzer was used to
measure aerosol size distribution in the 0.01 ym to 0.50 ym
                         24

-------
diameter range.  Details of its operating theory and use
are given in two papers by Liu  ert aJK ,      and in Thermo
Systems Operating Service Manual. ^2  Since the electrical
aerosol size analyzer gave a number median diameter (nmd),
the mmd was calculated from the usual formula:
            Log(mmd) = Log(nmd) + 6.9 (log a )2       [14]

In this equation the standard geometric deviation (og) was
found by plotting particle size versus number data on loga
rithmic cumulative probability graph paper.  Then a   is
calculated from:

                  nmd at 84. 1% value
             a
              g   nmd at 50.0% value                   -'
        Aerosols larger than 0.50 pro diameter were sized
with  a                photomicroscope with incident illumi
nation, again  from the above two equations to obtain mmd's.

D.  Temperature and Velocity Measurements

        Air stream temperature was  measured with an iron-
constantan (Type J) thermocouple,  with an ice bath refer-
ence junction and a potentiometer to record the thermocou-
ple millivolt output.  Clean, filtered air was passed
through the system until the thermocouple reached equilib-
rium, so its response time was of minimal concern.
                         25

-------
        Gas velocity through the filter was determined by



recording trial time, gas volume, temperature, ambient at-



mospheric pressure, vacuum pressureof the two volumetric



gas meters,and pressure drops across all three filters.



Gas velocity at the elevated temperature was then calcu-



lated  from the following equation based on the Ideal Gas



Laws :



              VF = (QM Tp PM)/(AF TM PF)              [16]





where:   Vp = Velocity through filter at T



         Tp = Temperature of air stream



         TjY| = Temperature of volumetric gas meter



         PM - Atmospheric pressure - vacuum pressure



         PP = Atmospheric pressure + filter(s) pressure



              drop(s)



         Ap = Filter area



         Q|Y| = Volume at TM  registered by volumetric gas



              meter.





E.   Techniques to Determine Filter Efficiency





1.   F1uorescence




                             43 44
        Fluorometric analysis  '   was used for most of the



efficiency determinations, because it has excellent sensi-



tivity (* 0.5 yg/fc) and is easy to use.  The procedure



followed to determine the % penetration of a test filter



was:
                         26

-------
        a)   Wash in distilled water and dry several  Gelman
            Type A (47 mm) filters to reduce their fluores-
            cence background to an average of 0.5 yg/£.
        b)   Use these washed filters in the parallel  filter
            arrangement previously described.
        c)   Soak the uranine aerosol laden filters in 20 ml
            distilled water for 12 hours to insure all  the
            uranine goes into solution.
        d)   Analyze the uranine leachate in  a   fluorometer
            to determine the uranine concentration.
        e)   Calculate the % penetration:
          % Pcnctration-Uran'"ie mass on backuP  filter
                        Uranine mass on reference filter
                        x 100%                        [17]

        The fluorescence of the uranine decreased to about
75% of its  original value when exposed to 260°C(500°F).
However, since the decrease was the same for the reference
line and the test line, it did not affect the results.
Uranine was not used for temperatures over 260°C (500°F),
since its fluorescence  is greatly reduced at these temperatures

2.   Atomic  Absorption
        A Nad aerosol was used in place of uranine for
temperatures from 260°C (500°F) to 538°C (1000°F).  First45
has demonstrated that the vaporization of NaCl  at temperatures
                         27

-------
less than or about equal to 538°C (1000°F) does not repre-
sent a serious error in the determination of filter pene-
tration, especially since any error is on the conservative
side.
        The NaCl leachates from the filters were analyzed
with,  an atomic absorption spectrophotometer.  The cali-
bration and operating procedures for the instrument were
                                                          46 47
carefully followed as decribed in the instruction manuals.  '
For Na the practical working sensitivity proved to
be about 3.0 yg/£, much poorer  than the 0.5 yg/£ sensi-
tivity of the uranine fluorometric analysis.  Consequently,
longer test runs were needed to accumulate enough Na on the
backup filter for accurate analysis.  Teflon filters (5 ym
pore size) were used in the reference and backup filter
holder   because they had undetectable Na backgrounds.
Gelman Type A, Gelman Spectro Grade Type A, and Millipore
Membrane (0.8 ym pore size, white, 47 mm) filters all proved
to have Na backgrounds too high to be useful.
3.  Mass Measurement

        For the dinonyl phthalate     aerosol, penetrations
at elevated temperatures were high enough for the aerosol to be
detectedby its mass alone.  By weighing the filters before
and after the experimental runs, it was possible to determine
the mass of accumulated aerosol.  A microbalance with a sensi-
tivity of 0.0001 g was used for the weighings.
                         28

-------
4.  Microscopic Counting

        The larger pollen aerosols were easily visible under
a stereo microscope, so penetrations could be determined by
counting the individual pollen particles.   Millipore Membrane
Filters (0.8 ym pore size, black, 47 mm) were used as
backups to collect the few pollen particles that penetrated
the filter under test.
5.  Ti tration

       HpSO. masses in washings of the stainless steel  coil,
leachates  of the filters, and the impinger contents
were determined by titration with 0.01 NNaOH.
The    NaOH titrant was standardized against 0.01 Npotas-
sium acid  phthalate.  Phenolphthalein indicator (0.05 wt %
in deionized water) was used.

F.  Filters and Filter Holders

        Two types of glass fiber filters suitable for high
temperature sampling were used as test filters.  One was
the binderless Gelman Type A Glass Fiber Filter (47 mm),
manufactured .from microsize filaments of glass and treated
in a muffle furnace to remove trace amounts of organic
fiber content. This filter is acceptable for use in EPA
Method 5.   It withstood 538°C (1000°F) but became brittle
at that temperature.  The other  filter used       was a
                         29

-------
prototype "Microquartz"  filter  of 99.2% silica fibers,

by Johns-Manvi11e  Company.         It maintained its struc-

tural strength and flexibility even at 538°C (1000°F).

        Two filter holders with different effective filter-

ing areas enabled the coverage of a wide velocity range

of 0.50 cm/sec to 51.0 cm/sec.                  A  stainless
holder (Gelman No. 2220, 47 mm) was used for the higher
                                                      2
velocities.  Its  effective filtration area was 9.6  cm ,

           and its Viton 0-ring and Teflon captive thrust

ring proved capable of withstanding 260°C (500°F).  Remov-

ing these two parts for higher temperatures did not cause

the holder to leak.  Two holes were drilled and tapped into

the holder for threaded stainless steel capillary tubing

for         pressure     taps.  This holder style was also

used to hold the  reference and backup filters.

        A 10    cm          Pyrex glass filter holder with
                                        2
an effective filtration area of 64    cm             was

used for the lower velocity range.  The filter holder con-

sisted of two glass halves,  fritted glass filter backing,

and a two-piece metal clamp with four bolts.  The low melting

point of the Neoprene gasket prevented the filter holder

from being used above 260°C (500°F).  Also, expansion of

the bolts at 260°C (500°F) caused the holder to leak ex-

cessively (Section VF).  Therefore, small springs were

placed on each bolt between the locking nut and the metal

clamp (Figure 14) to prevent leakage.
                         30

-------
                 V.  RESULTS AND ANALYSIS


        Particle collection efficiency tests with  Gelman

Type A, Gelman Type E, Gelman Spectrograde Type A, Mine

Safety Appliance Type 1106BH, and "Microquartz" glass fiber

filters all produced very similar results at room tempera-

ture.  These filter media all have similar pressure drops

vs. flow rates     filter masses per unit area, and fiber size

distribution  (see  Appendices B and C).  Therefore, only the

Gelman type A filter was extensively tested for comparison

with the new  "Microquartz" filter at high temperatures.


A.  Aerosol Size Distribution


        The aerosols generated with the Collison atomizer

with impactor were sized with an electrical  aerosol size

analyzer.   Their distributions (Table 3) were slightly more
                             AQ
polydisperse than those of LiJ  (Table 4), who used a more

similar Col 1ison-Impactor generator.   The size  distributions

of the aerosols larger than 0.50 ym diameter were determined

with a photomicroscope using incident illumination (Table

5).


B.  Loading Effects on Filtration Efficiency


        Penetration of both the Gelman Type  A and Micro-
                         31

-------
quartz filters declined significantly with aerosol loading
of only several micrograms per square centimenter.  This
effect was noted at both 20°C (70°F) and 260°C (500°F)
(Figure 15) and was accompanied by a fi1ter-pressure-drop-
increase of less than 5%.  A test to determine if the uran-
ine aerosol had an excessive charge which might have contri-
buted to this phenomenon proved negative.  If the filter
was initially charged, the contribution to total  filter
efficiency (due to filter charge) should have
dissipated with time as the charge was lost, thus increas-
ing penetration.  This did not happen.  So, the penetration
decrease was attributed to plugging of microsized holes by
the first few micrograms of aerosol and/or tree-like branch-
es formed by the initial aerosol deposit serving as parti-
cle collection surfaces. The latter proposition is supported
            49
by Tomaides'   findings that particles can form a rather
sturdy bridge about 10 particle diameters long.
        A similar increase in efficiency well before an ap-
preciable filter-pressure-drop-increase was reported by Dorman.
He reported that a filter with an initial efficiency
of 99.50% on a heterodisperse aerosol of O.Gymmmd had a
final efficiency of 99.98%, while the filter pressure drop
changed from 2.50 cm            to 2.75 cm            HoO.
        The time variation of penetration of a filter dur-
ing use naturally depends on the filter structure, fiber
                         32

-------
material and the nature of the aerosol.  As previously dis-
cussed, Radushkevich suggested penetration decreases expon-
entially with time according to the equation:
             P = PQe-bt                               [18]
        Where:  P = Penetration at time t=t
                P = Penetration at time t=0
                b = Constant under given conditions
From  the data in Figure 15, the following b values were
calculated:
 Filter Type    t(min)           Temperature (°C)    b(min"')
   Type A        100               21 (70°F)         0.0233
   Type A        180              260 (500°F)        0.008
"Microquartz"    100               21 (70°F)         0.0240
"Microquartz"    180              260 (500°F)        0.005
        With these b values,   plots   on semi-log graph
paper of P=P e"bt between the  initial  penetration and the
final penetration measurements compared favorably with the
experimental data.

        To negate the decreasing penetration when comparing
filter penetrations, two consecutive tests were run on each
filter.  The two measured penetrations were then plotted on
the normal axis of semi-logarithmic graph paper (Figure 16),
while the respective average loadings were plotted on the
logarithmic axis.  Penetration at the arbitrary reference
                         33

-------
                 2
loading of 7yg/cm  was then found by interpolation between
the two test penetration values.

C.   Velocity Effects on Filtration Efficiency

        Penetration of submicron  aerosols was found to
increase with increasing velocity (Figures 17 and 18).
Similar     penetrations for the  0.09vimmmd and 0.14pmmmd
aerosolswere not expected but may indicate the presence of
microsized holes in the filter media.   Penetrations of less than
0.01% were found for aerosols greater  than about O.Symmmd
for all filtration velocities tested.   These penetrations
were so low that the aerosol measurement sensitivity was in-
adequate to determine larger particle  velocity effects on
filter efficiency.  This, in general,  was also true for all
particle sizes at filtration velocities of less than 5 cm^ec.

        The filter pressure drops increased linearly with
velocity (Figure 19), resembling  the findings of Benson,
et  al.. and obeying D'Arcy's Law:
                    Ap = XLnV                           [19]
        Where  Ap = Pressure drop
               X  = Permeability  coefficient (a constant)
               L = Filter thickness
               n = Gas viscosity
               V = Velocity
                         34

-------
D.  Effect of Particle Size on Filtration Efficiency

        Submicron aerosols were found to produce the great-
est filter penetrations (Figures 20 and 21).  Similar pene-
trations for particles less than 0.14ymmmd cannot be easily
explained except for the possible presence of microsized
holes.  Uranine aerosols greater than 0.8ym mmd, fly ash
aerosol of 1 .10ym mmd, a dinonyl phthalate oil aerosol  of
about l.Oyrommd, and pollen aerosols of 13.1ymand 26.7 ym
mmd's all evidence penetrations of less than 0.01% at am-
bient temperatures.

E.  Effect of Temperature on Filtration Efficiency

        The effect of increased temperature was to decrease
aerosol penetration for  nonvolatile submicron particles
whose penetration could be measured (Table 6 and Figure
22).  Since submicron particles are collected mainly by dif-
fusion, which is more effective with increasing temperature,
the experiemental results qualitatively confirm the theory
previously outlined.
        Volatile particles were not collected as effectively
at elevated temperatures as were the nonvolatile particles.
For example, a  Gelman Type A filter was tested at 21°C
(70°F), 150°C (374°F), and 260°C (500°F) with a dinonyl
phthalate (DNP) aerosol, which vaporizes at 232°C (450°F).
The results at  a 51 cm/sec           filtration velocity
                         35

-------
were:
              Le_!HE_.llCl             ^Penetration
                 20                    <  0.01
                150                    - 25
                260                    =100
Obviously,at 260°C (500°F) the DNP vaporized, passed through

the filter, and condensed upon being cooled below its dew-

point.  DNP was found both in the cooling coil following the

test filter and on the backup filter.

        These results imply that the definition of particu-

late matter is a function of the sampling method (i.e., the

temperature at which the sample is collected).  Consequently,

if compliance tests are to be compared from source to source,

the same sampl ing method (i.e., sampling temperature) must

be employed if volatiles or condensables are involved.

        Filter pressure drop also increased with increasing

temperature, due to the increasing air viscosity values at

elevated temperatures.  This effect (Figure 23) reflects

D'Arcy's Law (Equation 19) in which pressure drop is direct-

ly proportional to both viscosity and  velocity.


F.  Filter Holder Leakage


        A stainless steel filter holder (Gelman No. 2220,

47 mm)was used for most of the experiments and was air-tight

at temperatures to 538°C (1000°F).  However, a Pyrex glass

filter holder held together by a two-piece metal clamp  with
                         36

-------
four bolts leaked significantly at 260°C (500°F).  The metal
bolts expanded,  deforming the clamping device and creating
a relatively large air leak.   This resulted in an apparent
50% to 60% aerosol penetration.  This leakage problem was
corrected by spring loading the bolts of the metal clamp.
                       2
        In EPA Method 5, a leakage check of the sampling
train, including the filter holder, is usually done with the
system at ambient temperature.  From our findings, it would
seem necessary to bring the filter unit to the recommended
temperature of 120°C before making the leakage check.  This
precaution would become even  more necessary if the proposed
Method 5 filter box temperature maximum of 160°C (320°F)
is accepted.

G.  H^SO. Filtration Efficiency at Elevated Temperatures

        Experiments with  FUSO. aerosol  were conducted at
two temperatures [ 120°C (248°F), 205°C (401°F)],25 cm/sec
         filtration velocity, 8.5 vol % HgO vapor,and 140
ppm (jJ5 ppm) F^SOa.  At these concentrations of H^SO. and
H20 vapor, the acid dewpoint  is about 170°C (338°F).   The
results in Table 7 show at 120°C (248°F),  below the H SO
dewpoint, most of the f^SO^was found in the coil and  on
the test filter.  At 205°C (401°F), above  the H2SO. dew-
point, most of the H?SO. was  found after the test filter
in the impinger contents and  on the backup filter.  Simple
calculations based on typical stack sampling data from oil-
                         37

-------
fired boilers and on this experimental  data indicate HUSO,
could account for more than 50% of the  total particulate
catch at a 120°C (248°F) sampling temperature (or at a tem-
perature below the HUSO, dewpoint in the stack gas), but
only for about 9% at a 205°C (401°F) sampling temperature
(or at a temperature above the I^SO^ dewpoint in the stack
gas).
H.  Pinhole Effects on Filter Efficiency

        Figure 24 shows how the penetration of a high ef-
ficiency glass fiber filter was affected by punching two
0.75 mm diameter pinholes through the filter mat. Note that
penetration is greater for the small aerosol than for the
large aerosol.  The pinholes were clearly visible when
the filter was examined against an illuminated background,
so it is doubtful whether defective filters with pinholes
as large as these would pass unnoticed  and be used for
sampling.   These penetrations are small enough not to sig-
nificantly affect the outcome of a stack sampling test.
I.  Comparison of Theory with Experimental Data
        Equations 10 and 11 were used to calculate the
theoretical efficiency of the "Microquartz" filter, and
the results are in Table 8.  Both the theory and data show
penetration to decrease with rising temperature.  The
theoretical penetrations are only within two orders of
                         38

-------
 magnitude of the experimental penetrations, being very de-
 pendent on the filter fiber diameter and particle size and/
 or size distribution.  Work is presently being done by us
 to improve the accuracy of several theoretical filtration
 equations.

 J.  Experimental Errors

         The largest variables in the experiments were the
 changes in penetration from filter to filter.  As an illus-
 tration, the penetrations of ten Gelman Type A filters were
 tested under identical conditions [T=21°C (70°F),
 v=18 cm/sec,  d=0.14ym mmd] using fluorometric analysis.
                                                         o
 The penetrations ranged from 0.02% to 0.05% at 7.0 yg/cm
 loading; the average penetration was 0.03% +_ 0.01%.  Al-
 lowing for the sensitivity of the fluorometric analysis
(0.5 ng/£ or 0.005% penetration for these trials), the fil-
 ter-to-filter penetrations varied as much as 67% from the
 average penetration for the ten trials.
         The atomic absorption analytic method which was
 used for the analysis of NaCl was not as sensitive as the
 fluorometric technique.  The sensitivity of the former was
3.0 yg/£, while that of the latter was 0.5 jjg/fc.' Hence, more
 mass had to be accumulated on the test filter so that enough
 aerosol penetrated to be detected accurately.  The accuracy
 of the atomic absorption method was j^ 0.01% penetration,
 which was adequate, considering the Targe variability from
                          39

-------
filter to filter.
        The smallest detectable H SO  mass, with titration
of   0.01 NNaOH was about 45 yg.  Since the H SO. mass col-
lected in the impingers, coil, or filters was about 6,000yg,
the error due to the titration was less than about 0.75%.
However, completely flushing all the H SO  from inside the
coil  with deionized water was difficult, so about a +_ 10%
error in the ^SO^ masses determined for the coil can be
assumed.
        The accuracies of temperature and velocity measure-
ments were +_10% and +5%  respectively.  The sensitivity of
the mass determinations, using a microbalance, was +_0.0002 g.
This translates into an accuracy in penetration of +0.67%.
The mass method of determining filter efficiency was used
only for the dinonyl pthalate     aerosol, which penetrated
100% at some high temperature conditions.  Thus,     +_0.67%
penetration was sufficient accuracy.
        Due to the inherent errors in particle size measure-
ment by electrical mobility, the mmd's determined with an
electrical aerosol size analyzer were not better than +_20%.
About the same accuracy can be ascribed to the size distri-
butions made by counting the particles with  a light micro-
scope.  The size of the aerosols generated with the Collison
atomizer with impactor compared favorably with the size of
those generated by others (See Tables 3 and 4).
                         40

-------
                      VI  CONCLUSIONS

        It was experimentally demonstrated that Gelman
Type A and "Microquartz" high efficiency glass fiber
filters are adequate for sampling nonvolatile particles
at temperatures to 538°C (1000°).   Submicron particles pen-
etrated more than did larger particles, and they penetrated
the most at the highest filtration velocity tested (51 cm/sec)
In all tests, however,  the aerosol penetration was never
more than about 0.10%.  Nonvolatile particles penetrated
less with increasing temperature  and increasing filter
1oadi ng.
        Particles with  vaporization points below the samp-
ling temperature, including H^SO., can vaporize, pass
through the glass fiber filters,  and then recondense when
cooled below their dewpoints.  Therefore, the definition
of "particulate matter" must be based upon a prescribed
temperature.   Hot stack gases  sampled at different filter
temperatures   should not necessarily be comparable.  Partic-
ulate emission standards must involve a suitable reference
temperature to allow proper enforcement.
        Filtration efficiencies calculated by theoretical
equations change dramatically with small  changes in the
assumed average filter fiber diameter and/or particle size
(or size distribution) used in the calculations.
                         41

-------
        Pinholes not visible to the naked eye do not
appear to affect the penetration of glass fiber filters
enough to significantly alter stack sampling results.
        The effect of temperature on filtration of non-
volatile particles was an increase in the col-
lection of submicron particles with increasing tempera-
ture.   The main problems encountered at elevated tempera-
tures  were vaporization of volatile particles and mechani
cal  leakage of the filter holder.
                         42

-------
                          SYMBOLS



Ap   =  Filter area

b    =  Constant (In Equation 12)

C    =  Ratio of solute volume to solvent volume plus
        solute volume

C    =  Cunningham correction factor

c    =  Average velocity of gas molecules

d    =  Particle diameter

D,   =  Droplet original diameter

D    =  Fiber diameter

"D    =  Diffusivity

E»   =  Filter mat efficiency

E    =  Single fiber efficiency for diffusion, inter-
        ception and inertia! impaction

G    =  Constant (In Equation 5)

H    =  Temperature, °C

k    =  Boltzmann's constant

L    =  Filter thickness

M    =  Molecular weight of gas

N    =  Dimension!ess interception parameter
 R
ND   =  Dimensionless diffusion parameter

P       Penetration at time t = t

PQ   =  Penetration at time t = 0

QM   =  Volume at TM registered by volumetric gas meter

R    =  Universal gas constant
                         43

-------
STK  =  Stokes1 number
T..   =  Temperature of volumetric gas meter
T    =  Absolute temperature
Tp   =  Temperature of air stream
V    =  Filtration velocity
Vp   =  Velocity through filter at Tp
M    =  Mass per filter face area
X    =  Permeability coefficient
a    =  Packing density
Ap   =  Pressure drop
Pp   =  Fiber density
PQ   =  Gas density
Pp   =  Particle density
n    =  Gas viscosity
n0   =  Gas viscosity at 0°C
A    =  Mean free path of gas molecules
PM   =  Atmospheric pressure - vacuum pressure
Pp   =  Atmospheric pressure  +  filter(s) pressure drop(s)
                         44

-------
                            References
 1.  Benson, A. L., Levins, P. L., Massucco, A. A., and
     Valentine, J. R., Development of a High-Purity Filter
     for High Temperature Participate Sampling and Analysis.
     EPA-650/2-73-032, by Arthur D. Little, Inc., Cambridge,
     Mass., Nov. 1973.

 2.  "EPA Standards of Performance for New Sources,"
     Federal Register, 36. (247):24876  (1971).

 3.  Radushkevich, L. V., Izv. Akad. Nauk SSSR. ser. khim.
     nauk, 3.:407 (1963).

 4.  Chen, C. Y., "Filtration of Aerosols by Fibrous Media,"
     Chem. Rev., 55^:595-623 (1955).

 5.  Perry, J.  H., and Chilton, C. H., Chem. Engineer's
     Handbook,  5th Ed., McGraw Hill Book Co., Inc., New York,
     p. 3-248 (1973).

 6.  Hodgman, C. D., Handbook of Chemistry and Physics,
     Chemical Rubber Publishing Co., Cleveland, Ohio, p.
     2205 (1962).

 7.  Yeh, H. C., A Fundamental Study of Aerosol Filtration
     by Fibrous Filters.  Univ. of Minn., Ph.D. Thesis (1972).

 8.  Davies, C. N., "The  Separation of Airborne Dust and
     Particles," Proc. Inst.  Mech. Engng., ]_B:185 (1952).

 9.  Whitby, K. T. and Lundgren, D. A., "Mechanics of Air
     Cleaning," Trans, of ASAE, 8(3):342-344 , 351-352 (1965).

10.  Watson, J. H. L., "Filmless Sample Mounting for the
     Electron Microscope," J. Appl . Phys. , ]_7:121-127 (1946).

11.  Leems, R., "Die Abscheidung Von Schwebstaffen in
     Fasernfiltern," Staub. 50:402-417 (1967).

12.  Davies, C. N., Aerosol Science, Academic Press, New
     York, p. 270 (1966).

13.  Pich, J.,  and Binek, B., "Temperature Characteristics
     of Fiber Filters," in Aerosols. Physical Chemistry and
     Applications, Proc.  First Nat. Conference on Aerosols,
     p. 257-264, Czech. Akad. Sciences, Prague (1965).
                           45

-------
 14.   First, M. W. , Graham, J. B., Butler, G. M., Walworth, C.
      B. , and Wanen,  R.  P., "High Temperature Dust Filtration,"
      Ind. and  Eng. Chem. . 48(4):696-720  (1956).

 15.   Dyment, J.,  "Assessment of Air  Filters at  Elevated
      Temperatures and Pressures," Filtration and Separation.
      p. 441-445,  July/August (1970~T

 16.   First, M. W., "Performance of Absolute Filters at Temp-
      eratures  from Ambient to 1000°F," 12th AEC Air Cleaning
      Conference,  p.  677-702.

 17.   Thring, M. W.,  and  Strauss, W.,  "The Effects of High
      Temperature  on  Particle Collection  Mechanisms, "Trans.
      Instn. Chem. Engrs.. 4J_:248-259  (1963).

 18.   Danielson, J. A.,  Air Pollution  Engineering Manual, Los
      Angeles County  Air  Pollution Control District, Los
      Angeles,  Cal.,  p.  536, 1967.

 19.   Hemeon, W. C. L.,  and Black, A.  W.  J., "Stack Dust Sam-
      pling:  In-Stack Filter or EPA  Train," J.  Air Pol 1.
      Control Assoc.. Vol. 22, No. 7,  p.  516, 1972.

 20.   Jaworowski,  R.  J.,  "Condensed Sulfur:  Trioxide Partic-
      ulate or  Vapor?" J. Air Poll. Control Assoc.. Vol. 23,
      No. 9, p. 791,  1973.

 21.   Hillenbrand, L. J., Engdahl , R.  B., and Barrett, R. E.,
      Chemical  Composition of Particulate Air Pollutants
      From Fossil-Fuel Combustion Sources, Battelle Columbus
      Laboratories, p. II-2, March 1,  1973.

 22.   Hedley, A. B.,  in  The Mechanism  of  Corrosion by Fuel
      Impurities (H.  R.  Johnson and D. L. Littler, editors),
      Butterworth, London, p. 204, 1963.

 23.   Cuffe, S. T., Gerstle, R. W., Orning, A. A. and Schwartz,
      C. H., J. Air Poll . Control Assoc. , Vol. 14, p. 353,
      1964.

 24.   Snowden,  P.  N.  and  Ryan, M. H.,  "Sulfuric  Acid Conden-
      sation from  Flue Gases Containing Sulfur Oxides," J.
      Inst. Fuel,  Vol. 42, p. 188, 1969.

 25.   Mueller,  P., "Study of the Influence of Sulfuric Acid on
      the Dew Point Temperature of the Flue Gas," Chemie-Ing.-
      Tech. , Vol 31,  p. 345, 1959.

26.  Abel,  E.,  "The  Vapor Phase  Above the System Sulfuric
     Acid-Water,"  J.  Phys.  Chem.,  Vol. 50,  p.  260,  1946.
                          46

-------
27.  Greenewalt, C.  H., "Partial  Pressure of Water Out of
     Aqueous Solutions of Sulfuric Acid," Ind.  and Eng.
     Chem.,  Vol. 17, pp.  522-523.

28.  Matty,  R.  E.,  and Diehl, E.  K.,  "New Methods for Deter-
     mining  S09 and  S03 in Flue Gas," Power Engineering,  Vol.
     57, p.  87, Dec. ,  T953.

29.  Taylor, A. A.,  "Relation Between Dew Point and the Con-
     centration of  Sulfuric  Acid  in Flue Gases," J. Inst.
     Fuel ,  Vol. 16,  p. 25, 194,2.

30.  Lisle,  E.  S.  and  Sensenbaugh, J. D., "The  Determination
     of Sulfur Trioxide and  Acid  Dew  Point in Flue Cases,"
     Combustion, Vol. 36,  No. 1, p. 12, 1965.

31.  Gmitro, J. I.,  and Vermuelen, T., "Vapor-Liquid Equili-
     bria  for Aqueous  Sulfuric Acid," Univ. of  Cal. Radiation
     Lab.  Report 10866, Berkeley,  Cal., June 24, 1963.

32.  Rendle, L. K.,  and Wilson, R. D., "The Prevention of
     Acid  Condensation in Oil-Fired Boilers," J. Inst. Fuel ,
     Vol.  29, pp.  372-380, 1956.

33.  Taylor, R. P.,  and Lewis, A., "Sulfur Trioxide Formation
     in Oil  Firing," Proc. Fourth  Inst. Congress on Industrial
     Heating, Group  II, Sec. 24,  No.  154, Paris, France,  1952.

34.  Flint,  D., Lindsay,  A.  W., and Littlejohn, R. F., "The
     Effect  of Metal Oxide Smokes  on  the $03 Content of Com-
     bustion Gases  from Fuel Oils," J. Inst. Fuel, Vol.  26,
     pp. 122-127,  1953.

35.  Corbett, P. F., and  Fireday,  F., "The S03  Content of the
     Combustion Gases  from an Oil-Fired Water-Tube Boiler,"
     J.  Inst.  Fuel, Vol. 26, pp.  92-106, 1953.

36.  Green,  H.  L.,  and Lane, W. R., Particulate Clouds:   Dusts
     Smokes  and Mists, E. &  F. M.  Spon. Ltd., London, 1957.

37.  Whitby, K. T.,  Lundgren, D.  A.,  and Jordan, R. C.,  "Hom-
     ogeneous Aerosol  Generators," Technical Report No.  13,
     Cooperative Research Project, Univ. of Minn., Dept.  of
     Mech.  Eng. and  USPHS (USPHS  Grant No. S-23 (C-4), Jan.,
     1961.

38.  Whitby, K. T.,  Lundgren, D.  A.,  and Peterson, C. M.,
     "Homogeneous  Aerosol Generators," J. Air and Water Pol-
     lution, Vol.  9, p. 263, 1965.
                          47

-------
39.   Whitby,  K.  T.,  Generator for Producing High Concentra-
     tions of Small  Ions," Rev.  Sci.  Inst. . Vol. 32,  p.
     1351, 1961.

40.   Liu, B.  Y.  H.,  and Pui,  D.  Y.  H.,  "On  the Performance
     of the Electrical  Aerosol  Analyzer,"  Particle Technology
     Lab. Publ.  No.  237, Particle Tech.  Lab., Mech.  Eng.
     Dept., Univ.  of Minn.,  Oct., 1974.

41.   Liu, B.  Y.  H.,  Whitby,  K.  T.,  and  Pui, D. Y.  H., "A
     Portable Electrical Analyzer for Size  Distribution  Meas-
     urement  of  Submicron Aerosols,"  J.  Air Pol 1.  Control
     Assoc. .  Vol 24. No. 11,  Nov.,  1974, p. 1067.

42.   TSI Model  3030  Electrical  Aerosol  Size Analyzer  Operat-
     ing and  Service Manual,  Thermo-Systems,  Inc., St.  Paul,
     Minn.

43.   Manual of Fluorometric  Clinical  Procedures, G.  K.  Turner
     Assoc. ,  Nov. ,  1971.

44.   Operating and  Service Manual Model  110 Fluorometer,  G.
     K. Turner Assoc.,  April, 1971.

45.   First, M.  W.,  "Performance of Absolute Filters  at  Tem-
     peratures from  Ambient  to  1000°F,"  12th  AEC Air  Clean-
     ing Conference, pp. 677-702.

46.   Instruction Manual for  Models  1100  and 1200 Atomic
     Absorption  Spectrophotometers.  Varian  Techtron  Pty.,
     Ltd., Melbourne, Australia, Sept.,  1973.

47.   Analytical  Methods for  Flame Spectroscopy.  Varian  Tech-
     tron Pty.,  Ltd., Melbourne, Australia, Sept., 1972.
48.  Liu, B.  Y.  H., "Methods of Generating Monodisperse Aero-
     sols," Pub.  No.  104,  Particle Technology Lab.,  Mech.
     Eng. Dept.,  Univ.  of  Minn., Feb.  8,  1967.

49.  Tomaides, M.,  U.  of Minn., Personal  Communication (1967)
     in:   Billings, C.  E., Wilder, J., Fabric Filter Systems
     Study. Vol.  I:  Handbook of Fabric Filter  Technology,
     GCA  Corporation,  Report No. NTIS  No.  PB 200-648,  1970,
     pp.  200-648,  1970, pp.  2-83.
                          48

-------
50.   Dorman, R. G. , "Filtration," in Aerosol Science by C.
     N. Davies, Academic Press, N.Y., p. 218, 1966.

51.   Federal Register. Vol. 39, No. 177, Sept. 11, 1974, p
     32853.

52.   Yost, D. M. and Russell, H.,Jr., Systematic Inorganic
     Chemistry. Prentice-Hall, Inc., N.Y., 336 pp. (1946).
                          49

-------
                           APPENDIX A


                   CONVERSION OF S03 TO
     To answer the question of how much of the  S03  in. flue  gas  is
present as H2SOi», consider the following equilibrium:52

                     H2SOi,(g)J S03(g)+H20(g)


                            *    0-ole. vapor/I)
                          + 0.75 logioT -  0.00057T +  4.086
                                 (over the  temperature range  598
                                 to 756*10


First calculate position of the above equilibrium at  the  highest
temperature of interest, 400 F  (477 °K) assuming the above expression
for logioK is valid:

            logioK=!0.480 + 2.009 - 0.272  +  4.086

                  = -4.657

so at 400"F      K=0.22 x HT" =("A°^S?3)
                                 (H2SOi»)

            at 1 atm, 0°C, ideal gas = -0  ... m/£ = 0.0446 m/i
            at 1 atm, 477°K ideal gas =  (477) (22  40) =  °'0256 m/Jl

In the typical exhaust-gas composition from a coal  fired boiler
(Table 1) , H20 = 4%,

so H20 = 0.04 x 0.0256 = 1.024 x 10~3 m/5,.

Similarly, S03 = (3.0 x 10~5 x 2.56 x 10~2 - X) = (7.7  x 10~7 -  X) moles/*

HaSOit = X moles/liter

0.220 x 10- = d.02 x 10-3)(7.7 x IP"7  - X)
                             A

           X = 7.51 x 10~7; or 97.8% of  the S0  is  H$0.
                               50

-------
                 00
 Similarly,  at 300 F  (422 K):

          log  K = 11.846 + 1.970 - 0.241 + 4.086
             10

          log  K = -6.031
             10
                             -6
               K = 0.935 x 10
             o                        273
 at 1 atm, 422 K, ideal gas is 0.446 x 4T2" = 0.289
                                         -3
          H 0 = 0.40 x 0.0289 = 1.16 x 10   m/Jl
            2

          H SO  = X m/i
            24
                      -5                          -7
          SO  =3x10   x 0.0289 - X = (8.66  x 10   -  X)  m/l
            3

                         -3            -7
          -6   (1.16 x 10  )  (8.66 x 10   - X)
0.035 x 10                  X
           -6              -9             -3
 (0.935 x 10  )X = 1.04 x 10   - (1.16 x 10  )X

                        -9
              1 .004  x 10                -6
          X =          -3   = 0.866 x 10
              1.16x10
                          -7
at 300 F H SO  =  8.66 x 10  ;  or * 100% of the  SO  is H  SO  .
          24                                    324

     Hence,  SO   essentially exists as  H SO  over  the entire range
              3                        24
of conditions of  interest,  if the time is  sufficient to  allow

equilibrium of  the  hydration.
                               21
(Above from  Hillenbrand,  et  al.   )
                                 51

-------
                          1  ym





"Microquartz" filter at BOOOx magnification (scanning



electron microscope photo).



                           52

-------
                          1  ym
Gelman Type A filter at 5000x magnification (scanning



electron microscope photo).




                            53

-------
                               Concentration
Component
H20
CO 2
Fly ash before
precipitator
Fly ash after
Volume
percent
4.0
15.0


(a)
g/m3
30
273
9.16
(4 gr/ft3)
0.458
 precipitator

      NO

      S02

      SO 3


 Hydrocarbons
 0.050

 0.20

 0.0030
(30  ppm)

(0.0010)
(0.2  gr/ft3)

    0.63

    5.3

    0.10
(a)  At 21°C (70°F), 1 atmosphere.
Table 1.  Typical exhaust-gas composition from coal-fired
          boiler. (From Hillenbrand,  Engdahl,  and Barrett21)
                          54

-------
Location
Plant A







Plant A



Plant A








Plant A


Plant B





Plant C

Plant C

H2S04
Filter ppm
Thimble1 8.1
8.1
10.8
9.5
9.9
9.5
9.5
9.5
Hi-volume1 6.9
6.0
8.8
8.1
EPA/APCO1 14.9
13.8
7.5
11.6
6.1
9.5
9.5
8.8
8.8
EPA/APCO1 11.0
9.9
9.9
EPA/APCO2 5.0
5.0
6.7
5.7
5.9
5.4
Hi-volume1 4.7
2.8
EPA/APCO2 3.5
4.2
H2S04
gr/SCF
0.0147
0.0147
0.0196
0.0172
0.0180
0.0172
0.0172
0.0172
0.0125
0.0109
0.0147
0.0050
0.027
0.025
0.0135
0.021
0.011
0.0172
0.0172
0.0159
0.016
0.020
0.018
0.018
0.0092
0.0092
0.0123
0.0104
0.0104
0.0098
0.0085
0.0050
0.0064
0.0076
Total
gr/SCF
0.0694
0.0344
0.107
0.0366
0.0645
0.0329
0.0688
0.0405
0.0548
0.0540
0.0255
0.0292
0.151
0.0321
0.0308
0.0388
0.0242
0.0403
0.0643
0.0659
0.033
0.033
0.077
0.0757
0.022
0.023
0.036
0.030
0.031
0.033
0.0111
0.0076
0.0212
0.0200
H2SO«.
%of
total
21
43
18
47
28
52
25
42
23
20
62
50
18
78
44
54
45
43
27
24
48
61
23
24
42
40
34
35
34
30
77
66
30
38
]BaCls precipitation.
*NaOH titration.


Table 2.  Amount  of H SO  found in particulate  matter by
                      24                        20
various stock  sampling methods. CFrom Jaworowski   ).
                             55

-------
Solute Cone. (Wt.%)         NMD(ym)          MMD(ym)

1.00% Uranine                0.050            0.140           1.80
O'.l% Uranine                 '0.028            'o.089           1.86
0.01% Uranine                0,017            0.050           1.82
1.0% NaCl                    0.023            0.061           1.77

Table 3.  Size distribution of aerosols produced in Collison atomizer
          with impactor (water solvent).


Solute Cone. (Wt.%)         NMD(ym)          MMD(ym)

1.0% Uranine                 0.054            0.103           1.41
0.1% Uranine                 0.028            0.049           1.49
0.01% Uranine                0.016            0.028           1.45

Table 4.  Liu's data on performance of Collison atomizer with
          impactor (water solvent). (From Liu48)
                               56

-------
Aerosol Description    Generation Method     NMD(ym)  MMD(ym)

0.002% Uranine by        Spinning Disk        0.75     0.79   1.14
wt. in ethanol

0.005% Uranine by        Spinning Disk        1.06     1.10   1.12
wt. in ethanol

0.02% Uranine by         Spinning Disk        1.43     1.49   1.12
wt. in ethanol

0.1% Uranine by          Spinning Disk        2.68     2.84   1.15
wt. in ethanol

1.0% Uranine by          Spinning Disk        5.82     5.98   1.10
wt. in ethanol

Paper Mulberry            Mechanical         12.9     13.1    1.07
Pollen                    Dispersion

Ragweed Pollen            Mechanical         26.2     26.7    1.08
                          Dispersion

Fly Ash                   Mechanical          1.10     5.61   2.09
                          Dispersion

Table 5.  Size distributions of experimental aerosols.
                               57

-------
         Aerosol
         Uranine
en
oo
Uranine
Uranine
        Uranine
         NaCl
         NaCl
         NaCl
        Uranine
               mmd (ym)

                0.09
0.09
0.14
                0.14
                0.06
                0.06
                0.06
             0.79,1.10,
             1.49,2.84,
                5.98
          Velocity(cm/sec)

                51
                                        51
                                        18
                                         *
                                        18
                                         6
                                         6
                                         0.5
 0
51
51
18
18
 6
 6
 0
 0
51
                51
Temp(°C)

   21
  120
  260
   21
  120
  260
   21
  120
  260
   21
  120
  260
   21
  120
  260
   21
  260
   21
  260
   21
  260
   21
  260
   21
  260
  537
   21
                         % Pen. @ 7.0
                        yg/cm2 loading   Filter Type
 0.10
 0.05
 0.03
 0.14
 0.05
 0.02
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
 0.10
 0.01
 0.03
<0.01
<0.01
<0.01
<0.01
<0.01
 0.02
<0.01
<0.01
<0.01
<0.01
Microquartz (Q)
       Q
       Q
  Type A (A)
       A
       A
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
       Q
        Table  6.  Effect of temperature on filtration efficiency.

-------
vo
               S.S.  Coil at Test
               Temperature
               Filter at Test
               Temperature
Two Impingers +
Backup Filter
                                            120°C(1)
                                           % of Total
                                         H2SOi» Catch (2)
                                61
                                32
                                              100 Total
                                                       205°C(1)
                                                     % of Total
                                                   H2SCK Catch  (2)
11
                                                                        81
                                                        100  Total
               (1)  HaSO.* Dewpoint =170°C
               (2)  Average of two trials
            Table 7.
                                           distribution in sampling train.

-------
       Vel(cm/sec)


            5.1
           18.3
           51.0

            5.1
           18.3
           51.0
                  T(°C)

                   21
                   21
                   21

                  260
                  260
                  260
                                              Theoretical - % Penetration
                                                                               Experimental-
                                                                               % Penetration
D=0.4 (1)
d=0.14 (2)
a =1.0
g
<0.01
<0.01

-------
Figure 1.  Structure of "tree" on a metal fiber. (From
Davies  }.
                           61

-------
                                     direct  interception
Figure 2.   Effect of temperature  on  penetration  (theoretical)
          13
CFrom Rich  ).
                           62

-------
£%

 I wo
  95
           20     40  90
100
150
200
             Figure 3.   Effect of temperature  on  penetration  (experimental)
                       13
             CFrom Rich  ).                               "
                                        63

-------
   100


     50



     20


     10
 o  1-0
 g>

iH  0-5
•«-
UJ
    0-2
   0*1
   C-05
   0-02
                                 I
                  400
  800          1200

Temperature,   °C
1600
      — — Inertial  impaction
      	 Interception
           	 Diffusion


Figure  4.  Effect of temperature on  collection  mechanisms  of

impaction, interception and  diffusion.   (From Thring and
        17
Strauss   }.
                             64

-------
                                I     I     I     I     I     I
O
in
c
o
•H
a
H
o
O
u
W
100




 do




 80




 70




 60




 50




 40
*   30
    20
    10
       600
                            1     I	I	I	I	!	I	I
             700
800       900      1000       1100


    Temperature,  °F
12'
    Figure  5.   Equilibrium conversion of SO  to SO  .  (From

                                      21    2      3'
    Hillenbrand,  Engdahl ,  and Barrett  )
                                65

-------
  100
o
CO

B*
   90
   80
a 70
 c 60
5
 a
 8 so
 >
   40
I"
H

1, 20
M


   10
                                I     I     I     I
     200
                          j	I
                           j	I
300
 400       500


Taicperature, *?
600
700
  Figure  6.   Equilibrium  conversion of  $0   to H SO  at 8.0

                                        3     2  4

  vol%  H  0 in flue  gas.  (From  Hillenbrand,  Engdahl , and
       2 21

  Barrett   )
                           66

-------
 100
    JIMttiMJV!!!;i!ii'!!!:il\*?!l!i!!!'\l-^^^
                                     -Av. ^:.rr.!:v-..-:-t:::::..v^:/M.'i.\ ::p
                                     .;\^T-I .;!itXinriH;i-!i:\"/;^:aTr:V:!!

                             6   8  10        20
                            Water Vapor, Vol %
40
60  60  100
Figure  7.   Dewpoint and condensate  compositi'on for vapor

mixtures  of  H 0  and H  SO   at 760 mm Hg  total  pressure.
            26 2        2  4   27
[From Abel    and  Greenewalt  )
                                 67

-------
    100



     80



     60





     40
    20
 I  10
     "


«    6
o
CO
       220
         0|p

         <0
                                          7
                240
260      280


  Dew Point, °P

300
320
Figure 8.  H SO  dewpoint  for typical flue gas moisture

            24            26               27

concentrations.  (From  Abel    and Greenewalt  )
                            68

-------
     500
 c
'6
 o.

I
     400
     300
Water -vapor -
                          content of. gases,
     200
       QOOO   0.005   QOIO    0.015

              H2  S04 in Dry Gas, %
 Figure 9.  Variation of dewpoint with H SO content for
                              24            28
 gases having different H 0 contents. (From Matty, and Diehl )
                   2
                      69

-------
Figure 10.  H SO  dewpoint obtained by various investigators
             2  4  26,27.        31        30        25
Abel  and Greenewalt     ,  Gmitro  , Lisle  ,  Muller  ,  and
      29
Taylor  .

                            70

-------
i
3fU
350
330
310
290
270
250
230
210
ISq,
— — — — A.
p&
r.
0 E)
1
t
•
•
-
•
»
^i
t i i i
A. Toyior dewpoinl
Muller calculated
cperimental partia
pressure measurei



,/•
K
• iii


X
o _x
X

1 III
t meter
1
1
Tients
i
/
/
*j . r
/
i
t
i
i
i
i
i
i
i . 1 1 I


•
f
4
•
-
•
-
|
).OI 0.1 1.0 10 100
             S03(H2S04) in Flue Gas, ppm
  Figure 11.  Dewpoint as a  function of H SO  concentration
            29           25         24
  CFrom Taylor  and Mueller )
                         71

-------
m U.UUD
0)
3
O
J? 0.004
o
19 0.003
O
*o
| 0.002
"c
o
O
g 0.001
0.000
<
300
U.
•§ 200
Q.
0>
O
100
O



X
x^
X'
y\
>
f
I
1



%
V xX*1
x^^x^
X ^

\x E



Xvx
><^^r
•^^^



H Cor bet t
® Flint
^ Taylor c
All othe
Wilsdc

X
X*^-
\


^-
and Fired
ind Lewis
r points R
>n
v ^
<
^^ ^
X X


A 1M--f
•LJV" ^
ay
endle and
<•
C
"x
<






                           234
                        Sulfur Content of Oil,%
    Figure 12. Relation of dewpoint  and  SO   content of combustion
                                          3                   35
    gases to sulfur content of oil.  (From Corbett  and  Fireday
            34                   33                        32
Flint et\aj_.  , Taylor and Lewis   ,  and  Rendle  and Wilson  )
                               72

-------
CO
        AEROSOL
       GENERATOR
                         r
                          COIL
:N
'""'"I
i
r \
FILTER}
^ n i
U 1
	 l
WATER
BATH
rcoTlTI
1 AA '
i VV i
i i
1 COIL ]
1 AA !
• VV i
i 	 i
REFERE
FILTEF
n >
U >
n >
u >
BACKUP
FILTER
    TO
GAS METERS
    AND
   PUMPS
         Figure  13.  General arrangement of test apparatus.

-------
                 TOP VIEW
 METAL t C
PLATES
                                   BOLTS
    NUTtC±=l
                                4 SPRING
               FRONT VIEW
 Figure 14. Clamp assembly for glass filter holder.

                     74

-------
en
   0.20
   0.18
   0.16
|  0.14
£  0.12
LJ
g  o.io
Q.
S5  0.08
   0.06
   0.04
   0.02
   0.00
                VELOCITY=5I cm/sec
                AEROSOL MMD = 0.
                URANINE DYE AEROSOL
                                         A- GELMAN TYPE A-2I°C
                                         B- GELMAN TYPE A-260°C
                                         C-MICROQUARTZ-2I°C
                                         D- MICROQUARTZ-260°C
TYPICAL ACCURACY FOR 95%
      CONFIDENCE INTERVAL
                                                                        I   i  I  I  I
                                               10
                                  FILTER LOADING - pg /cm2
                                                                            100
                   Figure  15.   Effect of filter loading on penetration.

-------
        O.I2r
        0.10
       o.oe
or

LU
Z
UJ
CL
       0.06
en
       0.04
       0.02
       0.00
                                TEST I
                                I	I	llli
                                                     TEST 2
                                          7     10

                                      FILTER LOADING  pg/cm2
                                                                                100
     Figure 16.   Method to determine penetration  at  7.0 pg/cm  loading

-------
<
oc.
LJ

Q_
   0.10
   0.08
   o.oe
   0.04
   0.02
   0.00
FILTER -MICROQUARTZ



TEMPER AT URE-21 °C


FILTER LOADING-7.0 pg/cm2





A-AEROSOL MMD^  0.09pm


B-AEROSOL MMD »  0.14 pm


C-AEROSOL MMD ^ 0.76 pm
                          TYPICAL 95% C.I
                                                                           100
                                    VELOCITY-cm/sec
 Ptgure 17.   Effect of  velocity  on  penetration(nMicroquartz"Filter)

-------
00
          o

          6
          or
          LJ
          Q.
             0.14
             0.12
             0.10
             0.08
             0.06
             0.04
             002
            0.00
FILTER-GELMAN TYPE A


TEMPERATURE-21°C


FILTER LOADING-7Opg/cm2





A-AEROSOL MMD* 009pm


B -AEROSOL M M D * 0.14 pm


C-AEROSOL M MD » 0.76pm
                                 TYPICAL95% C.I
            j	I
                 I
                              10


                       VELOCITY-cm/sec
I	I
i  i  i  i
       100
               Figure 18.   Effect of velocity on  penetration  (Gelman Type A  Filter)

-------
or
LU
E
o
    50
    45
   40
    35
   30
Q_
O
QC
0  25

LL)
QC
^
CO  2O
CO  ^U
LU
a:
CL

    15
    10
A - GELMAN TYPE A

B- MICROOUARTZ


C-GELMAN TYPE A (I POINT)

D-MSA-II06BH  (I POINT)
                                                      A
                                                      B
               TYPICAL 95% C.I. FOR

               MICROQUARTZ FILTER
               10        20        30

                    VELOCITY  cm/sec
                             40
_J

 50
       Figure  19.  Pressure  drop vs. velocity.
                                 79

-------
                            FILTER-MICROQUARTZ
                            TEMPERATURE- 21° C
                            FILTER  LOADING-7.0>ug/cm2
oo
o
                      0.10
                  O 0.08
                     0.06
LJ
LJ
CL
                     O.04
                     0.02
                     O.OO
          A-VELOCITY =51 cm/sec
          B- VELOCITY = 18 cm/sec
          C- VELOCITY =  5 cm/sec
          D- VELOCITY = 0.5 cm/sec
                         O.OI
                                                              TYPICAL 95% C.I
                                             0.10
                                    AEROSOL MMD-jum
                                                                                                    i i  I
I.O
                      Figure 20.  Effect of particle  size on  penetration  ("Microquartz'Tilter)

-------
00
I

UJ
z
UJ
Q.
            0.14
            0.12
            0.10
            0.08
            0.06
            0.04
            0.02
            0.00
            FILTER-TYPE A


            TEMPERATURE - 21 ° C


            FILTER LOADING - 7.0pg/cm2
A - VELOCITY = 51 cm/sec



B - VELOCITY = 18 cm/sec



C- VELOCITY = 5 cm/sec



D - VELOCITY = 0.5cm/sec
                                                               TYPICAL  95 % C.I.
                0.01
                             I	I	I
                                         0.10

                                AEROSOL  MMD-pm
                                                                1.0
            Ftgure 21.   Effect of particle size  on  penetration  (Gelman  Type  A  Filter)

-------
    01
   0.10
  0.09
  0.08
  0.07
a: o.oe

LJ
z
UJ
0_ 0.05
  0.04
   0.03
  0.02
FILTER - MICROOUARTZ


FILTER LOADING = 7.0 jug/cm2


AEROSOL MMD~0.09/im


URANINE DYE AEROSOL


A- VELOCITY" 51 cm/sec


8- VELOCITY =18 cm/sec


C- VELOCITY = 0.5 cm/sec
                                      TYPICAL 95% C.I.
                        100               200

                             TEMPERATURE °C
                                   300
      Figure 22.  Effect of temperature  on penetration
                                82

-------
    80
    70
    60
    50
  o:
  UJ
  E 4O
  o
  i

  CL
  O
  cr
  o

  iu 30
  a:
  CO
  LJ
  cr
     20
     10
FILTER-MICROQUARTZ


A-538°C


B- 260°C


C- 120° C


D- 20°C
                     TYPICAL 95%C.I
                10        20


                     VELOCITY
                     30


                    cm/sec
40
.  50
Figure 23.  Pressure drop vs.  velocity at elevated temperatures
                            83

-------
2.2

2.0
1.8
1.6
1.4
- A«l
>
—
—
_ O.IC
PEh
_ SAG
UJ
 1.2

 1-0

0.8

0.6

0.4

0.2

0.0
   0
                         Fl LTER - GELMAN TYPE A
                         TEMPERATURE-2I°C
                         2 HOLES-0.75mm  each
                         VELOCITY = 51 cm/sec
                         A-AEROSOL MMD*O.I4pm
                         B-AEROSOL MMD*2.6*jm
                              TYPICAL 95% C.I.
                                 t
       PENETRATION UNDER
       SAME CONDITIONS
       EXCEPT NO PINHOLES
                                            0.75mm
          
-------
                                   TECHNICAL REPORT DATA
                            (Please read Instructions on the reverse before completing)
 i. REPORT NO.
    EPA-600/2-76-192
                                                           3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
   FILTRATION CHARACTERISTICS OF GLASS FIBER
   FILTER MEDIA AT ELEVATED TEMPERATURES
                                                           5. REPORT DATE
                                                              July 1976
             6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)

   Dale A. Lundgren and  Thomas C.  Gunderson
                                                           8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
   Department of Environmental Engineering Sciences
   University of Florida
   Gainesville, Florida   32601
             10. PROGRAM ELEMENT NO.
                1AA010
             11. CONTRACT/GRANT NO.

                R803126-01
 12. SPONSORING AGENCY NAME AND ADDRESS
   Environmental  Sciences Research Laboratory
   Office of Research and Development
   United States  Environmental Protection Agency
            Triangle Park. North Carolina   27711
             13. TYPE OF REPORT AND PERIOD COVERED
                Final 7/74  -  7/75	
             14. SPONSORING AGENCY CODE

                EPA - ORD
15. SUPPLEMENTARY NOTES
16. ABSTRACT

        Particle collection  characteristics of a newly developed, high-purity  "Micro-
   quartz" fiber filter media and a Gelman Type A glass fiber filter media were  evalu-
   ated over a range of temperatures (20°C to 540°C),  particle sizes (0.05 ym  to 26 ym),
   gas  velocities  (0.5 cm/sec to 51 cm/sec), and particle volatilities.  Both  types of
   high efficiency filters proved adequate (>99.9%  efficiency) for sampling nonvolatile
   particles over the above  variable ranges.  Nonvolatile particle penetration decreased
   with increasing temperature and increasing filter  loading.

        The effect elevated  temperature had on particle collection characteristics  was
   not  a determining factor  in the application of high efficiency filters.  The  main
   problems encountered in the high temperature environment were filter holder leakage
   and  volatilization of gas-borne particles that passed through the filter media.
17.
                                KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS  C. COSATI Field/Group
   *Air pollution
   *Aerosols
   *Filtration
   Filter materials
   *Ceramic fibers
   *Temperature
                             13 B
                             07 D
                             13 K
                             11 E
18. DISTRIBUTION STATEMENT

   RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
  UNCLASSIFIED
21. NO. OF PAGES
  95
                                              20. SECURITY CLASS (Thispage)
                                                UNCLASSIFIED
                                                                         22. PRICE
EPA Form 2220-1 (9-73)
                                            85

-------