&EPA
         United States
         Environmental Protection
         Agency
          Industrial Environmental Research  EPA-600/7-79-020
          Laboratory          January 1979
          Research Triangle Park NC 27711
Evaluation of Granular
Bed Filters for
High-temperature/
High-pressure
Particulate Control

Interagency
Energy/Environment
R&D Program Report

-------
                 RESEARCH REPORTING SERIES


Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional  grouping  was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:

    1. Environmental Health Effects Research

    2. Environmental Protection Technology

    3. Ecological Research

    4. Environmental Monitoring

    5. Socioeconomic Environmental Studies

    6. Scientific and Technical Assessment Reports  (STAR)

    7. Interagency Energy-Environment Research and Development

    8. "Special" Reports

    9. Miscellaneous Reports

This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort funded  under  the  17-agency  Federal  Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from  adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments  of,  and development of, control technologies  for energy
systems;  and integrated assessments of a wide range of energy-related environ-
mental issues.
                        EPA REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved
for publication. Approval does not signify that the contents necessarily reflect
the views and policies of the Government, nor does mention of trade names or
commercial products  constitute endorsement or recommendation  for use.

This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                                  EPA-600/7-79-020

                                       January 1979
  Evaluation of  Granular Bed
Filters  for High-temperature/
   High-pressure Paniculate
                 Control
                      by

              Shui-Chow Yung, Ronald Patterson,
              Richard Parker, and Seymour Calvert

                Air Pollution Technology, Inc.
              4901 Morena Boulevard, Suite 402
                San Diego, California 92117
                 Contract No. §8-02-2183
               Program Element No. 1NE624
             EPA Project Officer: Dennis C. Drehmel

           Industrial Environmental Research Laboratory
             Office of Energy, Minerals, and Industry
              Research Triangle Park, NC 27711
                    Prepared for

           U.S. ENVIRONMENTAL PROTECTION AGENCY
              Office of Research and Development
                 Washington, DC 20460

-------
                            ABSTRACT

     The status and potential of granular bed filter (GBF) tech-
nology for fine particulate control has been critically reviewed
and evaluated with emphasis on high temperature and pressure (HTP)
applications.
     Available theroretical models and experimental data have been
evaluated and found to be inadequate for predicting the performance
of industrial GBF systems.  Additional experimental data were ob-
tained with  a bench scale GBF.  These data were used as the basis
for a clean  bed performance model based on inertial impaction as
the primary  collection mechanism.  Predictions were in good agree-
ment with data from industrial GBF systems.
     The performance and economics of fixed, continuously moving,
and intermittently moving GBF systems have been evaluated for HTP
applications.  At the present stage of development, GBF performance
is neither efficient enough nor sufficiently reliable to satisfy
HTP particulate control requirements.  Therefore, cost estimates
are highly speculative and serve mainly to indicate areas where
further development work might substantially reduce capital and
operating costs.
     This study shows that GBF systems have potential for HTP
applications but further development and design improvements
will be necessary before these systems can be considered adequate
HTP particulate control technologies.  Recommendations are made
for improving the efficiency and reliability of GBF systems.
                               111

-------
                            CONTENTS
                                                           Page
Abstract	
Figures	vi
Tables	   xii
Abbreviations and Symbols 	   xiv
Acknowledgement 	   xviii

Sections
1.  Summary and Conclusions 	   1
       Current Technology Evaluation	1
       Engineering Design Equations 	   13
       Potential for HTP Applications	16
       Conclusions	17
2.  Introduction	20
3.  Literature Review 	   22
       Patents	22
       GBF Studies	27
       Theory	63
4.  Comparison Between Model Prediction and Data	105
       Data Reported by McCain	105
       Data Reported by Hood	110
       Data Reported by Knettig and Beeckmans	Ill
       Data by Paretsky, et al	116
       Data by Gebhart, et al	116
       Conclusions	118
5.  Experiment	119
       Apparatus	119
       Data Analysis	122
       Pressure Drop Data	131
                               IV

-------
                      CONTENTS (continued)
6.  Design Model	135
       Mathematical Modeling	135
       Comparing Model Predictions with Performance Data. .  . 141
7.  Present Technology Evaluation 	 145
       Granular Bed Filter Systems	145
       Industrial Users and Performances	161
       Evaluation	175
8.  Potential for HTP Applications	185
       Cleanup Requirements 	 . 	 185
       Predicted GBF Performance	188
       Gaseous Pollutants	192
       Preliminary Cost Estimate	193
       Permissible Costs	20°
       Summary and Conclusions	203
9.  Recommendations fur Future Research and Development . .  .205
       Efficiency Improvement	205
       Bed Cleaning Methods 	 208
       Other Research Recommendations 	 209
       Recommended Research Program	209
References
                                                             212

-------
                           FIGURES


Number                                                        Page

  1    Filtration of aerosols through bed packed with  lead
       shot (Thomas and Yoder data)	30

  2    Time vs. efficiency - 0.033 m deep granular bed
       filtering fly ash (Taub's data) 	  33

  3    Penetration of 1.1 ym aerosol as a function of
       superficial gas velocity: 10-14 mesh sand (Paretsky,
       et al. data)	36

  4    Penetration of 1.1 \im aerosol as a function of
       superficial gas velocity: 20-30 mesh sand (Paretsky,
       et al. data)	36

  5    Experimental results obtained by Gebhart, et al.:
       Penetration vs. particle diameter for different flow
       velocities and granule size	37

  6    Experimenta results obtained by Gebhart, et al:
       Penetration vs. particle diameter for different flow
       velocities and granule size	37

  7    Experimental results obtained by Gebhart, et al. :
       Penetration vs. particle diameter for different flow
       velocities and granule size	38

  8    Experimental results obtained by Gebhart, et al:
       Penetration vs. particle diameter for different flow
       velocities and granule size	38

  9    Gravity  effect in granular beds measured with an
       upward and downward directed aerosol stream (Gebhart
       et al. data)	'  40

  10    Gravity  effect in granular beds measured with an
       upward and downward directed aerosol stream (Gebhart
       et al. data)	'  40

  11    Capture  efficiency  (transfer units) vs. bed height
       for  the  grid supported fixed bed (Knettig and Beeck-
       mans' data)	      41
                               VI

-------
                      FIGURES (continued)

Number                                                        Paj

  12    Capture efficiency (transfer units) vs. bed height
        for the screen supported fixed bed (Knettig and
        Beeckmans1 data) ...................

  13    Effect of mean gravel diameter and particulate load
        on granular bed collection efficiency  (Miyamoto and
        Bonn's data) ..................... 43

  14    Effect of gravel layer thickness and particulate
        load on granular bed collection efficiency (Miyamoto
        and Bonn's data) ................... 43
  15    Effect of superficial gas velocity and particulate
        load on granular bed collection efficiency (Miyamoto
        and Bonn's data) ................... 43

  16    Penetration tests on 420 to 710 micrometers plastic
        beads by monodispersed polystyrene latex aerosol
        (downflow) (Figueroa and Licht's data) ........ 46
  17    Effect of bed height and bed granule (bead) size on
        the downflow penetration of 0.5 micrometers poly-
        styrene latex aerosol particles on plastic beads
        (Figueroa and Licht's data) .............

  18    Comparison of downflow penetration of polystyrene
        latex aerosol particle on 420 to 710 micrometers
        plastic beads and -25+40 mesh sand granules
        (Figueroa and Licht's data) ............. 4

  19    Effect of bed granule (bead) size and flow direction
        on the penetration of 0.5 micrometer polystyrene
        latex aerosol particles on plastic beads ....... 47
  20    Schematic of granular bed filter (Westinghouse setup) 4^

  21    Schematic of test equipment (Westinghouse) ...... ^0

  22    Electrified packed bed ................ 61

  23    Performance characteristics of electrostatically
        augmented packed bed (Research-Cottrell data) ....

  24    Dendrite initiation, growth and idealization of the
        dendrite configuration ......... .......
                                                              70
  25    Particle deposition on single collector .......

  26    Typical isotherms for Langmuir and BET adsorption     R_
        patterns ...............  ........

                               vii

-------
                        FIGURES (continued)
Number                                                        Pase
  27    Schematic representation of fly ash emission
        mechanisms ......................  y5
  28    Fraction of total fly ash emitted by various
        mechanisms as a function of deposit thickness  .  .  .  .  95
  29    A comprehensive plot of pressure drop in fixed
        beds  ........................  97
  30    Effect of mean granule diameter on pressure drop
        (Miyamoto and Bonn's data) .......  .  ......  yy
  31    Effect of bed depth on pressure drop  (Miyamoto  and
        Bonn's data) .....................  yy
  32    Effect of superficial gas velocity on pressure  drop
        (Miyamoto and Bohn's data) ..............
  33    Comparison of McCain's gravel bed particle collection
        data with design equation predictions ........
  34    Comparison of Hood's data with predictions by  avail-
        able design equations  ................
   35     Experimental  and predicted performance of  CPC  GBF
         (A.P.T.  data)  ....................  112
   36     Experimental  and predicted performance of  CPC  GBF
         (A.P.T.  data)  ....................  113
   37     Experimental  and predicted performance of  CPC  GBF
         (A.P.T.  data)  ....................  113
   38     Comparison of Paretsky,  et al. data with predictions
         by available  design  equation .............  117
   39    Comparison of  Gebhart,  et  al. data  and predictions
         by Paretsky, et  al.  equation.  Collection  is  in  the
         diffusion regime ............
   40    Schematic  diagram  of  the  experimental  apparatus  .  .  .  12°
   41    Particle size  distribution  for  iron  shot .......  121
   42    Experimental particle penetration  of a clean  granular
         bed filter ..................  .  ...  123
   43    Experimental particle penetration  of a clean  granular
         bed filter ......................  123
                                viii

-------
                       FIGURES (continued)

Number                                                       Page

  44    Experimental particle penetration of a clean
        granular bed filter ........ . ........  124

  45    Experimental particle penetration of a clean
        granular bed filter .................  124

  46    Experimental particle penetration of a clean
        granular bed filter .................  125

  47    Experimental particle penetration of a clean
        granular bed filter .................  125

  48    Experimental particle penetration of a clean
        granular bed filter .................  126

  49    Experimental particle penetration of a clean
        granular bed filter .................  126

  50    Experimental particle penetration of a clean
        granular bed filter .................  127

  51    Experimental particle penetration of a clean
        granular bed filter .................  127

  52    Experimental particle penetration of a clean
        granular bed filter ...........  ......  128

  53    Experimental particle penetration of a clean
        granular bed filter .................  128

  54    Experimental and predicted pressure drops across
        a clean granular bed filter .............  132

  55    Experimental and predicted pressure drops across
        a clean granular bed filter .............  132

  56    Experimental and predicted pressure drops across
        a clean granular bed filter .............
  57    Experimental and predicted pressure drops across
        a clean granular bed filter .............  133

  58    Experimental and predicted pressure drops across
        a clean granular bed filter .............  134

  59    Diagram of granular bed showing impaction concept.  .  136

  60    Kp vs . n  for round jet model for particle collection
        in a GBF ......................  139

                                ix

-------
                        FIGURES (continued)
Number                                                       Page
  61    Kp vs. n for round jet	140
  62    Efficiency vs. inertial impaction parameter for
        comparison	140
  63    Comparison of McCain's gravel bed particle collection
        data with design equation predictions  	  142
  64    Comparison of Hood's data with predictions by avail-
        able design equations 	  142
  65    Experimental and predicted performance of CPC GBF
        (A.P.T. data)	143
  66    Experimental and predicted performance of CPC GBF
        (A.P.T. data) 	  143
  67    Experimental and predicted performance of CPC GBF
        (A.P.T. data) 	  144
  68    Dorfan impinge filter 	  147
  69    Consolidation Coal Company filter  	  148
  70    Combustion Power Company "dry scrubber"  	  151
  71    The  integral cyclone model of the  "dry scrubber".  .  .  152
  72    Possible design for Squires panel bed filter	153
  73    Lurgi MB filter	155
  74    Rexnord gravel bed filter	157
  75    Ducon  granular bed filter	159
  76    Filter  element	160
  77    Collection cycle	162
  78    Cleaning cycle	162
  79    Experimental grade efficiency curve of a Rexnord
        gravel  bed filter (McCain, 1976)	165
  80    Experimental penetration curves for CPC dry scrubber
         (Hood,  1976)	168

-------
                       FIGURES (continued)

Number                                                        Page

  81    Continuous moving bed GBF ............... 169

  82    Experimental grade penetration  (CPC data for nominal
        configuration) .................... 172

  83    Experimental grade penetration  (CPC data for thick
        bed configurations) .................. 172

  84    Experimental grade penetration  (CPC data for thick
        bed configuration) .................. 173

  85    Experimental grade penetration  (CPC data for short bed
        configuration) .................... 173

  86    Experimental grade penetration  (CPC data for small
        granule configuration) ................ 174

  87    Fractional penetration curve for Ducon granular bed. .176

  88    Fractional penetration for Ducon GBF (from Bertrand,
        et al., 1977) ..................... 176

  89    Predicted GBF performance ............... 189

  90    Particle size distributions from Exxon miniplant
        fluidized bed coal combustors
  91    Fixed bed GBF ..................... 195

  92    Summary G.E. comparison cycles  (cost of electricity
        at an assumed capacity factor of 65%)
202
                                XI

-------
                             TABLES
Number                                                       Page
   1    Summary of experimental  investigations  of  GBFs.  ... 2
   2    GBF evaluation  summary ................ 9
   3    Available equations for  the prediction  of  particle
        collection in a granular bed ............. 14
                                                              OQ
   4    Results of Hanford sand  filter  tests ......... "°
                                                              32
   5    Summary of Zahradnik,  et al.  data ..........
   6    Slopes and intercepts  of fitted lines ........
   7    Experimentally  determined properties  of the granular 45
        materials (Figueroa,  1974) ..............
    8    Westinghouse GBF data,  series #2 ...........
    9    Westinghouse GBF data,  series #3 ........  .  .  . 52
   10    Westinghouse GBF data,  series #5 ........... 53
   11    Westinghouse GBF data,  series #6 ........... 54
   12    Westinghouse GBF data,  series #7 ........... 55
   13    Data obtained from Strauss and Thring granular bed
         filter study ..................... 57
   14    Available equations for the prediction of particle
         collection in a granular bed ............. 107
   15    Comparison of Knettig and Beeckmans'  data and
         predictions ..................... 115
   16    Pressure drop and collection efficiency for 1.09 ym
         diameter particles at a superficial gas velocity of
         50 cm/s ......... . .... 7 .....  .  . .129
   17    Pressure drop for 50% collection of 1.1 ym diameter
         particles ...................... 130
   18    Measured and calculated void fraction of the granular
         bed ......................... 131
                                 xii

-------
                       TABLES  (continued)

Number                                                      Page
"^^"^™"^^^"*^^^™                                                      _^^^«^b«v

 19    Typical operating characteristics of Consolidation
       Coal Company granular bed filters ........... 149
 20    Summary of Rexnord gravel bed users

 21    Design specifications of the system as tested ..... 166

 22    Test parameters for CPC moving bed filter  (from Wade,
       et al., 1978) ..................... 171

 23    CPC moving bed filter overall penetration  correlation. 171

 24    Granular bed  filter performance  (from Bertrand, et al.177
       1977) .........................

 25    Conditions for high temperature  and pressure
       particulate collection  ................ 186
 26    Summary of EGAS phase II performance and cost results.
                               Xlll

-------
                        ABBREVIATIONS AND SYMBOLS

   a = effective mass transfer area, cm2/cm3
   a = a constant, cm2/dyne
  A  = fraction of the projected area of a single collector
   p   particle which is available  for capturing the aerosol
       by settling, fraction
   b = stoichiometry constant, dimensionless
   B = a constant expressive of the energy of interaction with
       the surface, fc/mol
   c = concentration of adsorbate in gas phase, g/cm3
  c  = equilibrium concentration of sorbate in bulk gas phase,
   6   g/cm3
  c  = saturation concentration of the adsorbate, g/cm3
   s
  C  = empirical constant defined by equation (5), dimensionless
 C  . = inlet particle concentration, g/cm3
  C' = Cunningham slip correction factor, dimensionless
  &c = granule diameter, cm
  d. - jet diameter, cm
  d = initial capillary diameter, cm
  d = particle diameter, cm or urn
  D  = effective  diffusion coefficient of gaseous reactant  in
       the residue  layer, cm2/s
   Dg  =  gas phase  diffusivity, cm2/s
   D  = particle diffusivity, cm2/s
  D   -  effective  particle diffusion coefficient in granule
   p    layer,  cm2/s
Dpore  = pore  diffusivity, cm2/s
    f  =  fraction factor, dimensionless
   f1  =  ratio of collector diameter to initial capillary diameter,
        dimensionless
    F  = compression  stress, dyne/cm2
    g  =  gravitational acceleration, cm/s2
    j  = ratio of channel width to packing diameter, dimensionless
                                xiv

-------
                    ABBREVIATIONS AND SYMBOLS (cont.)

    k = Boltzmann's constant = 1.38 x 10"   erg/°K
    k = chemical reaction rate constant, fc/mol-s
   kg = mass transfer coefficient, cm/s
k  re = pore diffusion coefficient, cm2/s
    K = equilibrium constant, £/mol
    K = proportionality factor, dimensionless
   K^ = D'Arcy permeability, cm2
   K  = inertial parameter, dimensionless
   m,  = expected particle number in the k'th layer of a dendrite,
        number
    M = molar weight, g/mol
   Mm = granule recirculation rate, kg/kg of gas
    N = number of impaction stages, dimensionless
   Pt = overall penetration, fraction or percent
  Pt, = penetration for particles with diameter "d ", fraction
        or percent                                "
    q = amount of sorbate adsorbed per unit weight of sorbent, g/cm3
   q  = amount of sorbate adsorbed per unit weight of adsorbent in
        equilibrium with concentration "C", g/cm3
   Q° = number of moles of adsorbate adsorbed per unit weight of
        adsorbent in forming a complete monolayer on the surface
    r = average pore radius, cm
   r  = radius of unreacted core, cm
   rc = collector radius, cm
   rH = hydraulic radius, cm
    R = original radius of the reacting particle, cm
    R = universal gas constant, J/g mol-s
    t = time, seconds
    T = absolute temperature, °K
   UG = superficial gas velocity, cm/s
  urh = actual gas velocity in bed, cm/s
  UGI = interstitial gas velocity, cm/s
   u- = gas velocity in the jet, cm/s
   ut = terminal settling velocity of the particle, cm/s
    w
particulate load, g/cm2
                                xv

-------
                       ABBREVIATIONS AND SYMBOLS  (cont.)

   xfi • fraction of reactant "B" converted into product,  fraction
    X = dust deposit thickness, cm
   yA = mole fraction of reactant "A"  in gas phase,  fraction
y  .. = critical trajectory  of the  aerosol,  cm
    Z = bed depth, cm

   Greek
    a = rate of particles  approaching  a clean  fiber  per unit
        length,  cm2/s
   AP = pressure drop, dyne/cm2
   PG = gas density, g/cm3
   p  = particle density,  g/cm3
   p  = density of solid,  g/cm3
     n = overall single granule collection efficiency or single
        stage  collection efficiency, fraction
   nD = single granule collection efficiency by diffusion mechanism,
        fraction
   TIDI  = single granule collection efficiency by direct interception
        mechanism,  fraction
   nrc  = single granule collection efficiency by gravity settling,
        fraction
    UG =  gas viscosity, poise
     e = bed porosity,  fraction
    e-  =  initial void fraction, fraction
     $ =  relative  force  field,  dimensionless
     X =  tortuosity factor, dimensionless
     0 =  angular cylindrical  coordinate, measured  counterclockwise
         from the  downstream stagnation point,  radian
     0 = pressure  drop  function defined in equation  (93)
  0£k = rate of increase of "m, "by deposition on particles already
    '   occupying  the k'th layer due to the  radial  flow component,
         cm'1
  (r)
 k-1  k = rate  of increase of"mk"by deposition on particles occupying
    *   the (k-l)'st  layer dui to radial  flow  component,s~'
   f (-1 "^
  0k k = rate  °^ increase of'W"by deposition on particles already
    '   occupying  the k'th layer, due  to  the angular flow component,
         o

                                 xv i
0

-------
                     ABBREVIATIONS AND SYMBOLS (cont.)

_T  V = rate of increase of"mj"by deposition on particles occupying
  *    the (k-l)'st layer due to the angular flow component,s~'
   T = time for complete reaction, s
   £ = internal void fraction of the solid, fraction

 Dimensionless Numbers
 NN  = Nusselt number, dimensionless
 Np  = Peclet number, dimensionless
 ND  = Reynolds number, dimensionless
  K€
 NC  = Schmidt number, dimensionless
 N, = Sherwood number, dimensionless
                               xvn

-------
                       ACKNOWLEDGEMENT
      A.P.T., Inc. wishes to express its appreciation for
excellent technical coordination and for very helpful assist-
ance in support of our technical effort to Dr. Dennis C. Drehmel
of the U.S. Environmental Protection Agency.
                              XVlll

-------
                           SECTION 1
                    SUMMARY AND CONCLUSIONS

     A number of advanced energy conversion processes under
development require high temperature and/or high pressure removal
of particulate matter to protect critical system components.
These processes must also meet existing and anticipated particu-
late emission standards.  Granular bed filter (GBF) systems
have been proposed as suitable devices for removing fine particles
from high temperature and high pressure (HTP) gas streams.  How-
ever, theoretical and experimental performance data for GBFs are
sparsely scattered throughout the literature.  A thorough survey
and evaluation of current GBF technology has been performed.  The
results are presented in this report.
     The objectives of this program have been to:
     1. Assess current GBF technology for control of airborne
        particulate pollutants,
     2. Evaluate existing GBF systems,
     3. Develop engineering models and design equations to pre-
        dict filter performance,
     4. Survey present usage problems, and
     5. Evaluate the potential of GBFs for high temperature and
        pressure applications.

CURRENT TECHNOLOGY EVALUATION
Performance Characteristics
     The literature was searched to assess the current
state of GBF technology.  Table 1 summarizes the experimental
investigations reported in the literature.  Most of the studies
were laboratory scale experiments.  The results of these studies
show that the collection efficiency of a GBF is a function of
particle diameter, face velocity, bed depth, granule diameter,

-------
                                                    TABLE 1.  SUMMARY OF EXPERIMENTAL INVESTIGATIONS OF GBFs.
     Investigator
                          Granular Bed Configuration
                                       Test Conditions
                                                           Parameter Studied
                                                                                                          Bed Performance
                                                                                                               Comments
I.  Laboratory Studies
   A.
      Thomas and
        Yoder
        (1956)
 (a)  Bed  type:  Fixed  bed
 (b)  Bed  material:  Sand
 (c)  Average  sand grain diameters:
      0.161  cm, 0.071 cm,
      0.036  cm
 (d)  Bed  height: 3.6  and 7.6 cm
 Aerosol:  OOP
 Aerosol particle  size:
   0.1  - 1.0 um
 Face velocity:  0.11  -
   2.2  cm/s
Sand particle size,
  particle shape,
  face velocity
(a) Penetration varied
      with sand particle
      size - decreasing
      diameter of sand.
(b) Rough and irregular
      sand showed higher
      collection efficiency
      than smooth sand.
(c) .Penetration
      varied with sand par-
      ticle size, face velo-
      city and aerosol parti-
      cle size. Efficiencies
      ranged from 40 to 99.8%.
Experimental result
  demonstrated the
  existance of an
  aerosol size of
  maximum penetration
  of about 0.3 vim.
  Particle size of
  maximum penetration
  decreased with in-
  creasing face
  velocity.
  B. McFee and
       Sedlet
       (1968)
(a)  Bed type: Fixed bed
(b)  Bed material: Sand
(c)  Sand grain diameter:
      0.036 to 0.071 cm
(d)  Bed height: 15.2 to
      76.2 cm
Aerosol: Pu-U-Mo
Alloy fume
Aerosol particle size:
  Geometric mean =
  0.07 \m, standard
  deviation » 2.7.
(Discrete sizes ranged
  from  0.02 to 4 pm)
Face velocity: 0.5-68
  cm/s
Aerosol grain loading:
  approximately 0.11
  g/m3
Bed height,  face
  velocity
(a) Penetration through 15.2
      cm of sand varied from
      0.08 to 0.57% over
      range of face veloci-
      ties from 0.5 to f>'.5
      cm/s.  Maximum penetra-
      tion of 0.57% occurred
      in range of 20 to 40
      cm/s.
(b) Penetration through 76.2
      cm of sand varied from
      0.004% at 14.2 cm/s to
      0.019% at 25 cm/s.
(c) Penetration decreased
      with increase in bed
      depth up to depth of
      30.5 to 45.7 cm, but
      relatively small im-
      provement occurred for
      beds of greater depth.
Experimental results
  indicate the per-
  centage of penetra-
  tion varies with
  aerosol particle
  size, face velocity,
  bed depth and degree
  of packing of bed.
  Experimental obser-
  vations confirm and
  extend finding of
  Thomas and Yoder.
                                                                                                                                         continued

-------
                                             TABLE  1.  (CONTINUED)
Investigator Granular Bed Configuration Test Conditions
Parameter Studied Bed Performance
Comments
(d) Maximum penetration
occurs at lower face
velocity with in-
creasing bed depth.
(e) Pressure drop through
sand varied with bed
dspth and face velo-
city and ranged from
1.4 to 259 cm W.C.
C. Ducon Bed type: Fixed bed with re- Aerosol:
Company verse gas flow cleaning. a) Iron oxide from
(Avco.Inc.) oxygen-lanced
(1969) electric arc
furnace
b) Iron oxide from
oxygen- lanced
open hearth
furance
c) Nickel ore
d) Fly ash
e) Talc dust
f) Plastic dust
Face velocity:
26 cm/s
D. Kovach and (a) Bed type: Fixed bed Aerosol: OOP, fly
Hannan (b) Bed material : Carbon ash
(1970) (c) Granule diameter: Aerosol particle
Aerosol type, inlet (a) Collection efficiency
grain loading ranged from 98 to
(4.6 to 11.4 99.9%.
g/ra3) (b) Pressure drop: 10.1 to
IS. 2 cm W.C.
Aerosol type, (a) Penetration varied
face velocity, with bed granule
granule size size-decreasing
Collection efficiency was
found to be higher for
the finer sized iron
oxide aerosols than for
coarse fly ash. Indi-
cates that physical
characteristics of aero-
sol are an important
factor in performance of
filter-possibly with
regard to agglomeration
behavior.
Penetration of fly ash
aerosol was less than
that for OOP aerosol .
      0.4 to 0.16 cm
(d)  Bed height:  38 cm
  size: fly ash -
  5 to 100 um
Face velocity:
  5 to 102 cm/s
      with decreasing
      granule size
(b)  Penetration  varied
      with face  velocity
However, particle size
of fly ash varied over
a wide range, and effect
may be due to both
changes in particle
size and aerosol type.
                                                                                                              continued

-------
                                                                   TABLE 1.  (CONTINUED)
    Investigator
    Granular Bed Configuration
                                                         Test Conditions
                                                                               Parameters Studied
                                                                                                        Bed Performance
                                                                                                                                   Comments
  E.  Zahradnik,        (a)  Bed type:  Shaft  filter (bed
       et al.                of slowly falling
       (1970)                granules through which
                            gas stream passes
                            horizontally)
                      (b)  Bed material:  Alkalized
                            alumina  (1/16  in.
                            beads)
 	        (c)  Bed width: 30.5  cm
                                    Aerosol:  Fly ash
                                    Aerosol  particle
                                      size:  11% 13.7 urn
                                    Gas velocity:
                                      Up to  28 cra/s
                                    Fly ash  loading:
                                      4.6 to 6.9 g/ra3
                                                                                Gas mass velocity
                                             (a) Collection efficiency:
                                                   99% at flow rate of
                                                   0.88 m'/rain and gas
                                                   temperature of 400 F.
                                             (b) Pressure drop: S.I cm
                                                   W.C.
                                                  Primary  intent  of  study
                                                   was  to investigate
                                                   simultaneous  removal
                                                   of fly ash and S02
                                                   by shaft-filter  system.
 F. Squires
      (1970)
 (a)  Bed type:  Fixed bed with
       intermittent movement
       of solids
 (b)  Bed material:  Sand
 (c)  Bed width:  3.5 cm
 (d)  Sand particle  diameter:
	approximately 0.05 cm
Aerosol:  Fly ash
Face velocity:  6 cm/s
Fly ash loading: 4.6
  to 9.2  g/m3
Puffback cleaning
(a) Collection efficiency:
    > 99%
(b) Pressure drop: 1.3
      cm W.C.
High collection effi-
  ciency may be due in
  part to preagglomeration
  of fly ash before enter-
  ing filter
 G. Paretsky          (a) Bed type: Fixed bed
     (1972)           (b) Bed material: Sand
                      (c) Sand particle diameter:
                           10 to 14 and 20 to
                           30 mesh (approximately
                           0.15 cm and 0.07 cm)
	(d) Bed height: 3.7 to 19.2 cm
                                    Aerosol:  Polystyrene
                                    Aerosol particle
                                      size: 1.1  urn
                                    Face  velocity: 0.3
                                      to  80 cm/s
                        Face velocity,  bed
                          height,  granule
                     (a) Penetration at a given
                           face velocity de-
                           creased as the gran-
                           ule size decreased
                            Results in agreement
                              with those reported
                              by other investiga-
                              tors.
 H.  Gebhart.-          (a) Bed type: Fixed
      et al.          (b) Bed material: glass beads
      (1973)          (c) Bead diameter: 0.4, 0.16,
                           0.05, and 0.0185 cm
                     (d) Bed height: 10.2, 20.5!
                           and 41 cm
                                   Aerosol: Polystyrene
                                      latex
                                   Aerosol diameter:
                                      0.1 to 2 pm
                                   Face velocity: 0.29
                                      to 6.6 cm/s
                        Granule diameter,
                        aerosol diameter,
                        face velocity,
                        bed height,  and
                        direction  of gas
                        flow
                                                   practical interest.
(a)  Downward gas flow gives  Gas velocity is too
      lower penetration than  low to be of any
      upward gas flow.
(b)  M¥£asional collection
      and gravity settling
      decrease with increas-
      ing gas velocity
(c)  Smaller granules give
      higher efficiency
(d)  Deeper bed gives
      higher efficiency
(e)  There exists a maximum
      penetration which is
      shifted to larger par-
      ticles with decreasing
      bead diameter
                                                                                                                                       continued

-------
                                                                  TABLE 1. (CONTINUED)
  Investigator
  Granular  Bed Configuration
  Test Conditions
Parameters Studied
                                                Bed Performance
                                                                                                             Comments
I. Knettig and
Beeckmans
(1974).
(a)
(b)
(c)
(d)
(e)
Bed type: Fixed and fluidized
Bed material: Glass beads
Bead diameter: 0.043 cm
Bed height: 1-12 cm
Bed support: Screen or grid
Aerosol: Uranine
methyl ene blue
Aerosol diameter
0.8, 1.6 and 2
Face velocity: 8
and 11.2 cm/s
and
.9 um
.2
Aerosol diameter,
support and face
velocity
(a)
(b)
Linear relationship
between collection
efficiency and bed
height
Collection efficiency
per unit volume of
                                                                                                          bed increased with
                                                                                                          both aerosol diameter
                                                                                                          and superficial gas
                                                                                                          velocity.	
J. Miyamoto and
     Bonn (1975)
(a)  Bed type:  Fixed
(b)  Bed material:  Gravel
Aerosol: Ammonium
  chloride
Aerosol diameter:
  0.1-3 um
Particulate load,
on collection
efficiency
(a)  Collection efficiency in- Results show the trend
      creased with increasing
      particulate load.
(b)  Thicker bed has higher
      initial collection
      efficiency but little
      effect when there  is
      filter cake.
(c)  Smaller granules have
      higher collection
      efficiency and have
      sharper increase in
      collection efficiency
      with particulate load.
(d)  Higher-face velocity
      decreases the rate
      of increase in effi-
      ciency with particulate
      load.
of cake filtration.
                                                                                                                                    continued

-------
                                                                   TABLE 1. (CONTINUED)
    Investigator
  Granular Bed Configuration
                                     Test Conditions
                                                                               Parameters Studied
                                                                                                      Bed Performance
                                                                                                             Comments
K. Figueroa
(1975)
L. Westinghouse
(Ciliberti,
1977)
(a) Bed type: Fixed and fluidized
(b) Bed material: Plastic beads,
sand
(c) Granule diameter, plastic
beads, 339 and 495 urn
diameter, sand - 702 pm
(a) Bed type: Fixed bed with
reverse gas flow
cleaning
(b) Bed material : Sand
(c) Sand particle size:
-15 +30 mesh
1 and 2 urn diameter
methylene blue, 0.5,
1.1 and 2.0 diameter
polystyrene latex
10 urn limestone
dust
Granule size,
bed height, face
velocity, flow direc-
tion, aerosol size
Face velocity
and bed depth
(a) Due to electrostatic
charges plastic beads
exhibited higher
collection effi-
ciency.
(b) Highest collection
efficiencies were
obtained on the
largest aerosol dia-
meter, with the deepest
bed of the finer
granules.
(a) Filter performance
improves as dust
accumulates in the bed.
Studied low gas
velocity region.
Experiment not well
controlled.
M.  Schmidt,
    et al. (1978)
(a) Bed type: Fixed bed
(b) Bed material:  Polystyrene
    beads,  quartz  gravel
(c) Granule diameter:
    polyethylene beads- 33 mm
    quartz  gravel  - 5  mm
Aerosol: DBP and
room aerosol

Face velocity: 15.2,
45.8, and 101 cm/s
Face velocity
granule size,
and aerosol size
(a)  At constant face velocity,
    the grade efficiency curve
    exhibits a minimum at a
    particular particle
    diameter.

-------
granule shape, aerosol type and packing density or bed porosity.
Other general results are listed below.
     1. At a constant face velocity, the grade penetration curve
exhibits a maximum penetration at a particular aerosol diameter.
The particle diameter corresponding to maximum penetration de-
creases with increasing face velocity and with increasing bed
depth.
     2. Particle penetration decreases as the granule diameter
is decreased, as packing density is increased, and as bed depth
is increased,
     3. Collection efficiency of a GBF composed of rough, irregu-
lar granules is higher than that of beds composed of smooth
granules.
     4. Aerosols which tend to agglomerate are more readily col-
lected.
     5. Surface and  internal filter cakes tend to improve (the.
collection efficiency of the GBF.
     6. There is a maximum particle loading that can be retained
in the bed.  Above this maximum loading, the collection efficiency
of the GBF declines.
     7. The collection efficiency of the GBF can be increased by
the use of augmenting forces.  The filtration efficiency of the
GBF can be enhanced  by imposing an electric field or a magnetic
field on the filtration medium.  A.P.T. is performing laboratory
experiments with charged and uncharged GBFs collecting charged and
uncharged particles. Preliminary results revealed that the collec-
tion efficiency of the GBF improved significantly without any in-
crease in pressure drop when the  bed  and/or particles were charged,
Present GBF Systems
     A granular bed  filter is any filtration system utilizing a
non-fluidized bed of discrete granules or particles as the fil-
tration medium.  A variety of types of GBFs are identified in the
literature.
     GBFs may be classified according to the bed structure or
according to the cleaning method.  With respect to the bed

-------
structure, GBFs may be classified as fixed bed, continuously moving
bed, or intermittently moving bed filters.  The advantages  and
disadvantages of each bed structure are summarized  in Table 2.
The following is a more detailed discussion.
Moving Bed Filters -
     The continuously moving bed filter is usually  arranged in  a
cross-flow configuration.  The bed is a vertical  layer of granular
material held in place by louvered walls.  The gas  passes hori-
zontally through the granular layer while the granules and  collected
dust move continuously downward and are removed from the bottom.
The dust is separated from the granules by mechanical vibration.
The cleaned granules are then returned to the overhead hopper and
the panel by a granule recirculation system.
     The Combustion Power Company's dry scrubber  is an example
of a continuously moving bed filter.  The performance of this
device has been reported by Wade, et al.  (1978).  They conducted
extensive cold flow tests to investigate  the effects of bed depth,
granule diameter and other parameters on  the collection efficiency.
     In general the CPC moving bed filter was found to be capable
of particulate removal efficiencies in excess of  98% for particles
in  the  1  to 10 ymA diameter range.  Submicron particles were
collected at an efficiency in excess of 90% in cases with high
velocities, high inlet particle loadings, and low granule rates.
Beds with larger thickness to granule diameter ratios were  most
effective in the capture and retention of particles in the  2 to
 5 ymA  diameter range.  Also, intermittent granule movement  was
 shown  to  improve efficiency by a few percent.
     High temperature tests of the moving bed filter are planned.
No  high temperature data are available at this time.
     The  major advantage of the moving bed filter design is that
 the bed granules are removed and cleaned  out of the primary gas
 stream.   This enables efficient cleaning  and relatively steady
 collection  efficiency.  Also it is not necessary  to isolate fil-
 ter units during cleaning so that the total filter  area open to
 gas flow  is available for filtration at any time.

-------
                                    TABLE 2.  GBF EVALUATION SUMMARY
GBF Type
          FIXED BED
        MOVING BED
       INTERMITTENT BED
Advantages
1. No granule recirculation.

2. Lower operating cost.
1. External separation
   of granule and dust.
1. External separation of
   granule and dust.
Disadvantages
1. Plugging of retaining grids
               2. Particle seepage through
                  bed during cleaning cycle.

               3. Fluidization redisperse
                  fine dust during cleaning.


               4. Ineffective bed cleaning
                  causes particle buildup
                  in bed.
               5. HTP valving required for
                  reverse air cleaning.
1. Erosion of retaining
   grids and transport
   system.

2. Particle reentrainment
   in moving bed.
3. Granule recirculation
   may cause high operating
   cost.

4. Difficult to form a
   filter cake in moving
   bed.
5. Needs a granule recir-
   culation and granule/
   dust separation system.
   No suitable mechanical
   device identified.  Trans
   port by pneumatic means
   needs large quantity of
   compressed air and  ener-
   gy to heat the transport
   air to bed temperature.
1. Low gas capacity can
   cause high capital cost.


2. Erosion of retaining grids
   and transport systems.

3. Granule recirculation may
   cause high operating cost.

4. Surface cake must be found
   to avoid bed plugging
   problem.
5. HTP three-way valving
   required for cleaning air.

6. It requires large quantities
   of cleaning air and energy
   to heat the cleaning air
   to bed temperature.

-------
     The moving bed design also has some limiting operating charac-
teristics.  Particle reentrainment caused by the relative motion
of the granules limits the granule flow rate and affects the over-
all collection efficiency.  Erosion of the retaining grids, louvers,
and transport system components may be a problem, especially in
high temperature and pressure systems.  The collected dust par-
ticles cannot form a filter cake so that the operating efficiency
will be essentially that of a clean bed.
     No suitable HTP mechanical granule recirculation system and
granule/dust separation system were identified in the literature.
CPC used a pneumatic method.  Granule transport and granule/dust
separation by the pneumatic method add significantly to the opera-
ting cost because large quantities of compressed air are required.
In order to maintain the high gas temperature at the gas turbine
inlet, it is necessary to heat the transport air to bed tempera-
ture to minimize the heat energy loss from the recirculating gran-
ules.  This  can  be  an  important  factor  in  the operating cost.
     It may be possible to resolve most of these problems through
further development and testing.  Performance data at high tem-
peratures and high pressures will be important in identifying
the most  serious operational problems.
Intermittently Moving Bed -
      In the  late 1950s, Squires modified the continuously moving
bed  design  to obtain a fixed bed device with an intermittent move-
ment  of granular solids.  The bed is stationary during filtration.
The  accumulated  filter cake and the surface layer of granules
 are  ejected  from the panel by a sharp backwash pulse and fall to
 the  bottom  of the filter vessel.  The expelled granules are immed-
 iately replaced  by  downward movement of fresh granules from the
 overhead  hoppers.
      The  principal  advantage of this type of bed structure is the
 capability  for external granule/dust separation with minimum dis-
 turbance  to  the  rooting cake.  A rooting cake is the foundation
 for  the formation of a surface cake.  After cleaning, the surface
 cake  is formed readily without disturbing the rooting cake and
 filtration  efficiencies can be much higher.
                               10

-------
     The intermittently moving bed also has the advantage that
granule cleaning is off-line and potentially more effective.
     The major disadvantage is that the gas capacity is lower than
for other granular bed filter designs and this results in high
capital costs for a given installation.  The face velocity is
about one third the velocity used in the fixed bed and continuously
moving bed GBFs.  Thus, more filtration area is required.
     Bed plugging also can be a problem if the surface cake is not
formed properly.  Erosion of the retaining grids, louvers, and
other components may be a problem.
     For HTP applications, Squire's design has another disadvantage.
It requires large quantities of compressed air for cleaning.  Com-
pressed air requirements could be as much as II of the gas treated.
The cleaning air is mixed with the gas being treated.  In order
to maintain the HTP conditions at gas turbine inlets, it is neces-
sary to preheat the cleaning air.  Cost to heat the cleaning air
can be an important factor  in the operating cost.
Fixed Bed Filters -
     Fixed bed filters operate in two modes; the filtration mode
and the cleaning mode.  During filtration the bed is stationary.
The gas passes through the bed and collected particles are depo-
sited within the bed and on the bed surface.  During cleaning the
bed is isolated from the main flow and agitated mechanically or
pneumatically by a reverse flow of gas.
     There are two fixed bed devices currently being developed;
the Rexnord gravel bed filter and the Ducon granular bed filter.
The Rexnord filter uses a rake-shaped stirring device to agitate
the bed during cleaning.  This loosens the filter cake which is
then removed by a reverse flow of clean air.
     The Ducon granular bed filter cleans the bed by a reverse
flow of gas which fluidizes the bed and elutriates the collected
particles.
     The Rexnord filters have been used successfully to control
emissions from clinker coolers in the cement industry.  They
operate in the range of 100 to 200°C  and near atmospheric

                               11

-------
pressure.  No Rexnord filters have been tested at high temperature
and pressure.
     The Ducon filter was tested on the effluent from a  fluid  bed
catalytic cracking unit regenerator at an oil refinery.   The gas
was at 370°C to 480°C and 1 to 1.5 atm.  A collection efficiency
of 85 to 98% was obtained on dust with a mass median diameter  of
35 urn and a geometric standard deviation of  about 4.
     Various high temperature and pressure designs  of the Ducon
filter were tested at the Exxon miniplant  (Hoke, et al.,  1978).
A number of operating problems were encountered during these tests.
     The lowest demonstrated particulate outlet concentration  was
68.6 mg/Nm3  (0.03 gr/SCF) which was considered to be too large to
protect  a gas turbine and borderline  for meeting current emissions
regulations.  The use of  smaller filter  granules could be expected
to  improve  efficiency.  However, at times  the filtration efficiency
was  very poor and the outlet particulate concentrations  were as high
 as  700  to  1,200  mg/Nm3  (0.3 to  0.5  gr/SCF).   It was also observed
 that the efficiency  decreased with  time  in the  longer  runs, drop-
 ping from 90%  initially to  about  50%  later in the  run.   Loss of
 filter  medium during blow back  was  another recurring problem.
 Further attempts were made  to use  50  mesh  retaining screens but
 they failed because  of  plugging.  Additional tests  made  with 10
 mesh screens also  resulted  in significant  screen plugging.
      A significant buildup  of particles  in  the filter beds was
 also observed amounting to  about  30%  of  the  weight  of the filter
 medium.  A possible  steady  long term  increase in filter  pressure
 drop may result  because of  this.  However, no significant increase
 in filter pressure drop was noted during any of the shakedown  runs.
      It was also observed that  the particles were not only building
 up in the beds  but were uniformly mixed with the filter  medium.
 It is  possible  that the  buildup and  mixing  of particles in the
 bed could be responsible  for the increase  in the particle concen-
 tration in the  outlet gas with  time.
      Another potential  problem with the current design was its
 vulnerability to upsets.  When upsets occurred, such as  bed

                               12

-------
plugging or loss of filter medium, the operating problems caused
by such upsets required shutdown of the  system.  Another problem
which may be unique to the miniplant was the  interaction of the
granular bed filter with the rest of the FBC  system during the
blow back cycle.  An increase  in system pressure was noted during
blow back resulting in problems with the coal feed system which
is controlled by the differential pressure between the coal feed
vessel and combustor.  This required modifications to the coal
feed control system to minimize the effects.
   The Exxon's granular bed filter test program was suspended in
November, 1977.  In all runs in which more than one outlet concen-
tration was measured, it was observed that the outlet concentration
increased with time.  They were not able to demonstrate that the
current EPA emission standard  (0.1 lb/106 Btu or 0.05 gr/SCF)
could be met for more than a few hours of operation.  In no runs
was the anticipated new standard (0.03 lb/106 Btu or approximately
0.05  gr/SCF) satisfied.

ENGINEERING DESIGN EQUATIONS
     The literature search provided information concerning
theoretical and empirical design and performance equations for
GBFs.  Table 3 is a summary of the available  equations.  Most of
the theoretical work considers the flow  through a packed granular
bed to be similar to the flow  around single granules.  This
assumption appears to be inadequate for  describing the efficiency
and pressure drop of actual granular bed filters.  These equations
do not adequately predict the  collection efficiency of a GBF at
face velocities normally encountered in  practice.
     In this study, a performance model  was developed which pre-
dicts collection efficiency for granular bed  filters.  The
model is based on the assumption that the GBF is equivalent
to a large number of impaction stages connected  in series.
                               13

-------
                  TABLE  3.   AVAILABLE EQUATIONS FOR THE PREDICTION OF
                            PARTICLE COLLECTION IN A GRANULAR BED
Investigator
  Equation
                                                                    Notes
 Jackson  and  Calvert
 (1968)
Paretsky et al.
(1971)
Miyamoto and Bohn
(1974)
Gebhart et al.
(1973)
Bohm and Jordan
(1976)
                       Pt   =  exp
                           = exp
 "c  z-il
 .  ' -c  "J



[- I 1=1 2- nl
L  2  e  d    I
                                 ['
                                              N
                                               pe
                                  -6.
                       Pt
                                                    36 U
                                  18e
                                     UG\  4 f Z

                                        '  yG dc
                                                                Impaction  only
                              Collection by
                              diffusion only,
                              Collection by
                              diffusion only
                              Collection by
                              diffusion.
                              Gravity settling
                              and impaction.
                                                                 Continued

-------
                         TABLE  3.   (continued)
    Investigator
                 Equation
                                    Notes
    Goren
    (1977)
Pt  = exp
  i (1-e)
  2
/

\
1250 K
                                z-zs
Collection by impaction,
gravity settling, and
diffusion
Westinghouse
(Ciliberti, 1977)
Ptj = exp
                          -3.75
- 5.8
                                        dc
                                     D
             Collection by impaction,
             interception, and diffusion
Schmidt, et al .
(1978)
Ptd =
     dc       I NPe
                            2.038 N
                                      1/6
                      5/H
                                   Re
                          + 1.45
                + 3.97 K  + —
                        p
             Semi-empirical equation,
             collection by diffusion,
             interception, impaction,
             and gravity settling.

-------
     The model is based on the collection of particles in a clean
granular bed.   The collection efficiency should be higher than pre-
dicted if there is a significant filter cake on the bed surface
and within the bed.  The presence of a surface cake, however, has
not been observed by investigators working with large-scale fil-
ters.  The clean bed model predicts a conservative estimate of
the efficiency attainable by granular filtration and is a satis-
factory model for filters which operate primarily without the
presence of a filter cake.
     This model has been used to predict the performance of indus-
trial GBF systems.  The particle collection efficiency predicted
with this model compares favorably with available field data.

POTENTIAL FOR HTP APPLICATIONS
     The feasibility of advanced energy processes, such as pres-
surized fluidized bed coal combustion, depends on the availability
of a very efficient HTP particulate cleanup device.  The parti-
culate control equipment should be capable of operating at a gas
temperature up to 9SO°C and a gas pressure up to 20 atm.
     The suitability of GBFs for controlling particulate emissions
from advanced energy processes is not limited by the gas tempera-
ture and pressure.  By properly selecting adequate granules and
structural materials, the granular bed filters could be capable
of operating at  any temperatures and pressures encountered in
advanced energy processes.  The Ducon GBF, Combustion Power Com-
pany's moving bed filter, and the CCNY panel bed filter, all can
be designed to operate at HTP.
     The use of GBFs for HTP applications is limited by the par-
ticulate removal efficiencies and operating difficulties. Required
efficiencies depend on the future emissions standards and on the
tolerance of gas turbines for fine particles.  Turbine require-
ments are not well established at this time.
     At  this time granular bed filters have not been demonstrated
to be efficient enough to perform as the final cleanup stage in
high  temperature and pressure gas cleanup systems.

                               16

-------
     There are several methods that may be used to increase the
collection efficiency.  One method is to use a deep bed of fine
granules.  This is not a good approach as the pressure drop would
be very high.  Other methods such as electrostatic augmentation
and cake filtration should be more effective.
     Quantitative data on the costs of HTP granular bed filters
are difficult to find.  However, we have completed cost comparisons
for various types of GBFs.  The cost estimates are based on the
cleanup requirement for a combined cycle steam turbine-gas turbine
power plant.
     The estimated capital costs of moving beds and intermittently
moving beds are about  141% and 247% higher than those of fixed
beds,respectively.  With regard to operating cost (not including
depreciation), moving beds are about 7.4 times higher than fixed
beds and intermittently moving beds are 4 times higher than fixed
beds.
     The Energy Conversion Alternatives Study (EGAS)  reported
estimates that the difference in the overall cost of electricity
between a pressurized fluidized bed boiler without cleanup and a
conventional steam power plant with stack gas cleaning is about
14 mills/kWhe.  Thus, a gas cleanup cost on the order of a few
mills/kWhe should allow sufficient economic advantages to warrant
continued development of the PFB boiler process.
     Based on the preliminary cost analysis, all three of the GBF
systems appear to be economically competitive.  However, at the
present stage of development, GBF performance is neither efficient
enough nor sufficiently reliable to satisfy HTP particulate con-
trol requirements.  Therefore, relative cost estimates must be
considered highly speculative and serve mainly to indicate areas
where further development work might substantially reduce costs.

CONCLUSIONS
     The principal objectives of this study were achieved and
the following conclusions may be drawn.
                               17

-------
Present Technology
     1. GBFs of present design are used successfully to control
emissions from clinker coolers in the cement industry and on hog-
fuel boilers in the forest products industry.  The particles
emitted from these sources are relatively large and present GBFs
are capable of cleaning the gas sufficiently to meet the emission
standards.
     2. GBF collection efficiency may be increased by using smal-
ler granules and deeper beds.  It may also be increased signifi-
cantly by imposing an electrostatic field on the bed and by
building a good filter cake on the bed surface.
     3. Further research work is required to increase the reli-
ability of the bed cleaning methods and granular solids transport
systems.
Design Equations
     1. None of the design equations reported in the literature
are adequate to predict the collection efficiency of a GBF at
face velocities normally encountered in the field.  A performance
model  is presented which predicts collection efficiency for GBFs.
The particle collection predicted by this model compares favorably
wtih available field data.
     2. The pressure drop across a clean bed may be predicted by
Ergun's equation for a packed bed.
HTP Potential
     1. Granular bed filters may be designed to operate in high
temperature and high pressure environments.  The available GBF
designs show potential for cleaning the gas to meet the current
New Source Performance Standard if fine granules (<500 pm in dia-
meter) are used as bed material.
     2. Limited research work has been conducted to demonstrate
the feasibility of simultaneous removal of particulates and gaseous
pollutants with GBFs.  It has been shown that packed beds of dolo-
mite are capable of removing sulfur dioxide and particles simul-
taneously.  Celatom MP-91 diatomaceous earth, Burgess No. 10
pigment and activated bauxite appear to be possible candidates

                               18

-------
as packed bed sorbents for removing alkali metal vapor from hot
combustion gases.
     3. The capital and operating costs of HTP GBF systems appear
to be within the acceptable range.  A combined-cycle power plant
with GBFs for HTP participate cleanup is economically competitive
when compared to a conventional power plant.
     4. Although granular bed filters appear to have the potential
for controlling particulate at high temperature and pressure in an
economically acceptable manner, they are far from a proven, state-
of-the-art technology.
     There are many operational problems and uncertainties which
need to be resolved before HTP granular bed filters can be con-
sidered sufficiently reliable for commercial application.  These
problems include the need to:
     a. Prevent particle seepage through the bed (during cleaning
        or filtration.
     b. Reduce temperature losses (especially during cleaning).
     c. Improve the efficiency and reduce the cost of granule
        regeneration and recirculation.
     d. Prevent attrition of granules causing particle reentrain-
        ment.
     e. Prevent sintering of granules.
     f. Prevent plugging of retaining grids.
     g. Reduce pressure drop across the bed.
     h. Improve primary and overall fine particle collection
        efficiency.
                               19

-------
                          SECTION 2
                         INTRODUCTION

     High temperature and pressure gas streams are encountered
in the development of advanced energy processes such as coal
gasification and pressurized fluidized bed combustion.  To in-
crease the thermal efficiency of these processes it has been
proposed that a combined gas turbine and steam turbine cycle be
used for the generation of power.   However, the presence of
particulate matter in the gas at the gas turbine can damage
the turbine blades and render them inoperative.  Thus it is
necessary to remove the particulate matter to a level which
will meet the turbine requirement for gas cleanliness with
minimal loss of gas temperature and pressure.
     There are a number of particulate removal systems under
development that are capable of operating at high gas temperature
and pressure.  The granular bed filter is one of these systems.
     Names like "granular bed filters," "gravel bed filters,"
"panel bed filters," "sand filters," "moving bed filters," and
"loose surface filters" are used by various researchers to des-
cribe the type of air filtration equipment which consists of
a bed of graded sand or gravel.  In this report the general
term "granular bed filters" is used and it is defined as any
filtration system comprised of a stationary or slowly moving
bed of separate, relatively close packed granules or particles
as the filtration medium.  In order to prevent the collected
particulate matter from plugging the interstices between the
granules and causing excessive pressure drops, the device should
embody some means for either periodic or continuous removal
of the collected particles from the collecting surfaces.  This
description  then excludes fluidized or dispersed beds where
granular particles are kept in motion by the  gas being  treated.
It does  include fixed bed or closely packed moving bed  systems.

                               20

-------
     This report reviews and evaluates the status and potential
of GBF technology for air pollution control with emphasis on high
temperature and high pressure applications.  The principles of
operation, design performance of granular bed filters for removing
fine particles from gas streams, and the effects of physical
parameters such as granule diameter, bed depth and the design
of bed containment structures are discussed.  Filtration mechan-
isms are described, as are several alternative procedures for
regenerating the granular bed filters.
     Current practices in the application of granular bed filters
for air pollution control and existing granular bed filter systems
are critically reviewed and evaluated.  Usage problems are iden-
tified.
     The potential of granular bed filters for control of parti-
culate emissions from advanced energy processes such as fluidized
combustion and combined cycle power plants is evaluated.  For
each potential application, the energy costs of high temperature
and high pressure particulate cleanup are compared with conven-
tional power plants.
                               21

-------
                           SECTION  3
                       LITERATURE REVIEW

PATENTS
     Granular bed devices and processes for the removal of
particles from gas streams have been reported as far back as
the late 1800's.  The following is a brief description of
principal patents issued since then.
     Solvay  (1889) patented a filter to remove dusts and vapors
from gases.   It consisted of a cylindrical granular bed of
sand (or alternatively a fibrous bed of asbestos) arranged  in
layers of increasing fineness upward from a foundation bed  of
coarse gravel or pebbles.  A steam jacket was added to the
vessel if a  condensible vapor was to be removed.  An internal
rake or scraper on a central vertical shaft provided cleaning
during operation.  This pioneering patent contained many of
the ideas later incorporated into the design of large coke  beds
for the removal of sulfuric acid mists and large sand beds  for
the removal  of radioactive particles.
     The early decades of the twentieth century produced a
series of diverse patents involving granular filters.  Fiechter
 (1919) patented a device in which the gaseous medium to be
cleaned is introduced at the top of a rotating disc of perforated
metal  (or a  screen mesh) covered with a layer of sand or other
granular filter material.  The gas is purified as it is passed
downward through the sand layer and is drawn off at the bottom
by a suction fan.  Removal and purification of filter media
may be done  continuously or intermittently using a revolving
screw  conveyor, mounted over the disc, and a vertically adjus-
table  scraper bar  to spread the  cleaned filter medium  to  the
desired layer thickness.  The method of cleaning the spent
sand or filter medium is not specified.
                                22

-------
      Another patent by Fiechter  (1922)  involved the use of a
moving sieve on an endless belt,  carrying  a horizontal  layer of
sand through which a gaseous medium  can  be filtered downward
under suction or pressure.  A  pair of  inner guide walls prevent
sand from slipping off the belt,  which is  slightly  lower at one
end, allowing spent sand  to trickle  into a cleaning device and
be returned via an elevator to the hopper  above  the opposite or
higher end of the belt.
      Klarding (1921) patented  a  filter  device for constantly
 purifying hot blast-furnace and  generator gases, containing
 large quantities of dust, without reduction in the temperature
 of the gases.  Granular  filtering material flows continuously
 from an upper hopper, downward through  a  main filter chamber,
 through a lower funnel onto a  vibrating sieve for dust removal
 and then onto a conveyor  that returns  the clean filter material
 to the upper feed hopper.  Dusty gases  are introduced at the
 side above the main filter bed,  flow  downward through the bed,
 and exit at the opposite side.   Part  of the cleaned gas is
 directed against the dust-laden  filter  material as it is
 shaken on the sieve to aid in  removing  the dust, which falls
 through the sieve.
      Nordstrom (1922, 1924) devised an  improved means of sepa-
 rating dust, smoke, and  the like from gases in a cement-burning
 process, a process for manufacturing  chloride of lime, and the
 copper smelting process.  The  gases are passed through a gran-
 ular filtering material  in a  filter tower.  A separate current
 of atmospheric air moves dust  from  the  filtering material as
 it falls through a step-like  bottom chamber; cleaned filter
 material is conveyed back to  the top  of the filter; the removed
 dusty matter may be returned  to  the original process or used
 for another purpose.  The filter tower, located between a
 furnace and a chimney, consists  of  two  concentric perforated
 walls with the filter material contained  between them.  Gases
                                 23

-------
from the furnace enter the inner chamber and pass outward
through the filter wall to an outer chamber connected  to the
chimney and are discharged to the atmosphere in a purified
state.
      Thomson and Nisbet (1924)  invented a filter for  cleaning
dust-laden gases from a blast furnace by allowing the  gases to
be drawn horizontally through a vertical downward moving screen
of suitable ballast material.  Ballast is continuously fed into
a V-shaped  hopper at top, slips downward over metal slots
arranged in louver fashion, and is conveyed by a worm  extractor
down a chute to a sloping metal screen at the bottom.  The screen
can be agitated to facilitate separation of dust from  ballast,
which is then conveyed to the top hopper by an elevator mechanism.
Suggested ballast material includes granulated or coarsely
powdered quartz, flint, or metallic fragments.
     Lynch (1930) patented a filter designed to handle large
volumes of air or gas at high temperatures.  It consists of a
thick bed of granular filter material falling downward into
piles in separate chambers.  Gas flows through the bed at a
slow rate, not exceeding 3 m/s (10 ft/sec), and discharges in
a direction approximately opposite to the direction of filter
material flow after passing downward and then upward through
each chamber.  Filter material is continuously cleaned, being
carried by conveyor to a rattler or other device for dust
removal, and then is hoisted back to the feed, chutes by an
elevator with bucket conveyor.
      Lynch  (1936) described a granular filter consisting of a
 bed of gravel 1.2 to 2.5 cm  (0.5 to 1 inch) in diameter that is
 continuously withdrawn from the bottom of the filter, passed
 over a screen for dust removal, and returned to the top of the
 bed.  Superficial gas velocity was approximately 91 cm/s (3 ft/s)
 for beds 30 to 122 cm (1 to 4 ft) deep and pressure drop was
 about 2.5 cm (1 inch) W.C.  Units of steel, high chromium steel,
 and brick have been used to filter gases having temperatures up
 to 454'C (850'F), 816°C  (1,500°F), and 1,093°C (2,000'F), respec-
 tively.
                              24

-------
     Fournier  (1936) designed an  apparatus for filtering gases
by means of a  filtering material  such as sand falling over
horizontal slats that may be vibrated on a combined system of
slots and sieves.  The gas passes transversely through the
layer of filtering material and between the slots of each series
Hammer vibrators are suggested as a means of increasing the
filter surface by facilitating the flow of filter material
and partially  cleaning it.  An endless chain of buckets or
rakes continuously feeds clean sand into the upper hoppers and
transports soiled sand flows by gravity down a sinuous channel
against an upflow of cleaning gas that removes the dust to a
cyclone dust separator.
     Berry and Fournier  (1939) presented a more limited version
of the above as a German patent.  Dust, soot, etc. are removed
from gases and vapors by passing  the gas transversely through
a granular filter of material falling in piles with natural
angles of repose over a series of horizontal slats.  The angle
of the slats may be varied, and filter material may be removed
at the bottom  of the apparatus, cleaned, and replaced at
the top.
     Carney (1944) devised an apparatus for separating carbon-
black dust entrained in a stream  of gas or air by passing the
gas upward through a bed of carbon-black granules contained
in a rotating  cylinder.  A spiral conduit, connected at the
 bottom of the  cylinder and wrapped around  it,  rotates  with  the
 cylinder and  lifts  the granules  to the  top of the cylinder
 while  the carbon dust  is agglomerated to the granules
     Mercier  and Ehlinger (1950)  developed a filter using
 sand or other  granular material  to remove  dust from hot gases
 issuing from  a boiler  firebox or from a boiler heated by gases
 under  pressure.   The filter is  designed to handle gases of
 any  temperature or  pressure.   In principle,  a sand of suitable
 size and quality is  arranged to  provide small volumes, small
 depth  of mass, and  considerable  surface area so that gases  may
 pass through  the filter without  excessive pressure loss.  Fine
                                25

-------
sand is held by walls formed by cone frustrums vertically
aligned and concentrically disposed approximately opposite one
another with conical surfaces tapering in opposite directions.
For coarse sand (> 4 mm), vertical layers are held between two
grids or between two perforated cylindrical metal sheets.  The
filter unit is radially divided into cells and is rotated.
     Gases to be cleaned enter at the top of a container  de-
signed to withstand a high pressure (such as 150 kg/cm ), zig-
zag  through the sand filter, and exit at the bottom during one-
half revolution of the  filter.  During the other half-revolution,
sand and dust are emptied down pipe to a rotating screen.
Dust falls through the screen while the sand passes into a
screw conveyor or similar device to be raised to the top of
the  unit for recycling.  Continuous cleaning and reuse of the
sand permits uninterrupted operation without loss of the heat
the  sand has acquir-ed through contact with the hot gases.
     Veron (1951) designed a sand filter for removing entrained
dust from gases produced by the combustion of pulverized coal
at high temperature and pressure and destined to feed a turbine.
The  filter is contained in a cylindrical body having a domed
cap  and conical bottom  - a form suitable for high gas pressures.
An inner lining able to withstand high temperature forms a
cooling jacket, through which compressed air is circulated.
Gas  enters at the bottom of the filter and passes upward through
stepped tiers of sand in multiple trays through a network of
interconnecting channels and pipes.  Clean gas then flows
through a separate system of inner chambers between the trays,
exiting at the bottom of the filter on the opposite side of
the  filter.
     When the sand becomes heavily dust laden, as indicated by
an  increase in pressure drop, slide valves are opened indi-
vidually, tier by tier, releasing the sand for cleaning  by any
suitable method, and then the sand is returned to the top of the
filter.  Sand was chosen as a readily available filter medium
that can stand high temperatures without damage or diminished
filtering power.
                              26

-------
     Various designs of granular bed filters for removal of
aerosols from industrial air and gas streams continued to appear
in the 1950's and 1960's.  Among those are Dorfan Impingo Filter
(U.S. Patent 2,604,187) Squires (U.S. Patent 3,296,775), Berz
and Berz (U.S. Patent  3,090,187), Kalen  (U.S. Patent 3,798,882)
and Zenz (U.S. Patent  3,410,055; 3,880,508).  All these designs
were tested at least in pilot scale and  some were introduced
commercially.  These designs will be discussed in a later section.

 GBF  STUDIES
 Low  Temperature  Filtration
     Among the earliest of granular bed filter  tests were  those
 carried out in 1948-1949  at the General Electric  Company,  Han-
 ford Works.   The granular bed  filters  were used  to  remove  radio-
 active  particles from air ventilation  systems.
     The results of the tests  were  reported by  Lapple  (1948).
 The  test chambers consisted of vertical cylinders,  25  to 30 cm
 (10  to  12  inches)  in diameter  and 91 to 153  cm  (3 to 5 ft) long. The
 filter  sand was  supported on a 8 mesh  screen welded  to the  sides
 of the  tank upon which was placed  successively  finer grades of
 sand.   Gas from  the main  ventilation system was  passed through
 the  bed in an upflow manner for the tests.
       Total concentration of suspended particles  at  the inlet
 was  on  the order of 2.46  mg/m3  (0.001  gr/ft3)  of  which less than
 0.025 mg/m3 was  radioactive.  Particle diameter  was  0.5 to  2.0  ym,
 Effects of bed depth,  granule  diameter, types  of  sand, and  gas
 velocity on collection efficiency were studied.   Collection effi-
 ciencies were based on the decrease of measured  radioactivity  of
 the  gas passing  through the bed rather than on  a decrease  of
 total particle concentration.
     Table 4  shows  a summary of the test results.   At  low  gas
 velocities (<5 cm/s),  laboratory tests indicated that  collection
 efficiency increased with increased bed depth  and finer sand
 grain size.   Irregular grained sands gave better collection effi-
 ciencies;  increasing gas  velocity  lowered collection efficiency.
 The  following approximation was derived:
                                27

-------
        TABLE 4.   RESULTS OF HANFORD SAND FILTER TESTS
Effect of Depth of Sand
              Depth of Sand (cm)
                     30.5
                     61
                     91.5
                    122
              Collection Efficiency,
                        91.6
                        99.5
                        99.8
                        99.9
Effect of Sand Size
    Sand Type
Hanford-16 + 20 mesh
Hanford-20+40 mesh
Hanford-16 + 20 mesh
Hanford-20 + 40 mesh
Ottawa-20+30 mesh
Ottawa-30 + 40 mesh
Ottawa-20+30 mesh
Ottawa-30 + 40 mesh
dc (cm)
   0.1
   0.056
   0.1
   0.056
   0.071
   0.048
   0.071
   0.048
(cm/s)
1.93
1.93
5.1
5.1
2.6
2.6
5.1
5.1
 Collection
Efficiency,1
    96.3
    99.98
    89.5
    99.88
    96.2
    98.6
    92.9
    98.2
Effect of Type of Sand
               Type of Sand
             Hanford  -20+40 mesh
             A.G.S. Flint-30+40 mesh
             Monterey Type G
             Ean Claire Type G

Effect of Gas Velocity
               ur  (cm/s)
                  0.91
                  3.1
                  5.1
                  9.3
                 20.3
              Collection Efficiency,
                        99.93
                        99.84
                        99.61
                        99.38
                               28
            Collection Efficiency,
                     99.97
                     99.83
                     98.9
                     92.7
                     97

-------
              Pt = exp
                        u  0.3 3  j  1.3 3
                         G     C
(D
where   Pt = overall particle penetration  on  an  activity basis,
         K = proportionality factor, dimensionless       fraction
         Z = bed depth,  cm
        UG = superficial  gas velocity,  cm/s
        d  = granule  diameter,  cm

     On the basis of these preliminary  tests,  large  scale sand
filters were designed and set up  at  Hanford with a capacity of
14.2 m3/s (30,000 CFM) each.  The  design face  velocity was
3 cm/s (6 ft/min), and the granular  layer  consisted  of a 61
cm (2 ft) depth of-20+40  mesh Hanford sand.   Seven layers of
coarser sand were also used in  the bed  to  aid  in flow distribu-
tion.  Overall pressure  drops were in the  range  of 10.2 to 17.8
cm W.C.  (4 to 7"W.C.),  and the  average collection efficiency was
about 99.7%.  The potential lifetime of these  sand beds was
estimated at about 5 years based  on  estimated  accumulation of
particles.
     Thomas and Yoder  (1956 a,  b)  investigated the filtration
of aerosols through  a  column of lead shot.  Their data are shown
in Figure 1.  The experiments were conducted  at  low  flow rates
and with large diameter  granules  so  that the  effects of inter-
ception and inertia  must have been negligible; the efficiency
was very low.    Collection was primarily  by  diffusion and
and sedimentation, the former prevailing in the  ascending
branches of the curves and the  later in descending.  Hence,
collection efficiency  decreased rapidly with  an  increase in
flow rate.  Sedimentation caused  the collection  efficiency to
differ for upwards and downwards  flow as shown in Figure 1.
Similar curves were  obtained with columns  of  sand in which
collection efficiency  diminished  appreciably  with increasing
flow rate and grain  size; grains  of  irregular shape  gave higher
efficiency than round  ones.

                               29

-------
   50
z
o
H
w
z
W
CU
   10
J	I
I	I	I	I
    1



  0.5
 Figure 1.  Filtration of aerosols through bed packed with
            lead shot (Thomas and Yoder data).
      0   0.1  0.2  0.3  0.4  0.5  0.6  0.7  0.8   0.91.0

                       PARTICLE RADIUS,
               D Flow up column 1.49 cm/s

               • Flow down column 1.49 cm/s

               O Flow up column 0.745  cm/s

               A Flow down column 0.745  cm/s

                 Column height 89 cm

                 Column area 11.17 cm

                 Lead shot 0.15 cm dia.

                                30

-------
    Zahradnik et al.  (1970)  carried  out  a  study on the simultan-
eous removal of fly ash  and  sulfur dioxide using  a granular bed.
The granular bed was  a square  cross-section  shaft, total length
224 cm (8 ft); the top 214 cm  (7  ft)  served  as the storage area,
the remaining 30 cm (1 ft) was  the actual  filtering section.
The cross-sectional area was 930  cm2  (1  ft2).  The filtering
medium was a slowly moving bed  of 0.16 to 0.32 cm (1/16 to 1/8 in)
diameter alkalized alumina particles  with  the solids rate con-
trolled by a vibratory feeder  at  the  bottom.  The gas, fly ash,
and sulfur dioxide flowed horizontally  across the filter.
    Data on the simultaneous removal  runs  are shown in Table 5.
The S02 removal efficiency was  100%  as determined by standard
iodometric titration  of  inlet  and outlet gas samples.  The fil-
tration efficiency was determined by  passing the  gas through a
Cuno filter which collected  all particles  larger  than 1 ym in
diameter.  For the size  distribution  of  fly  ash listed in Table
6, the overall filtration efficiency  was about 99%.
    Taub (1970) studied  the  transient behavior of granular bed
filters while collecting dispersed fly ash.  His  results show
high efficiencies are possible  with  clean  filters, but perfor-
mance deteriorates as the dust  content of  the filter increases.
    When a clean bed  is  put  into  operation,  particles are col-
lected in the interstices of the  filter  near the  surface.  As
the dust deposit builds  up,  the saturation zone or dust may
work its way through  from the  dirty  to the clean  side of the
filter.  This results from the  drag  force  exerted on the particle
deposits by the gas flow.  As  the deposit  extends through
the bed, the bed is saturated  with dust  and  reentrainment occurs
causing the collection efficiency to  decrease.  The result of
Taub's work (Figure 2) shows this trend.  The collection effi-
ciency remains initially constant with time.  As  the saturation
zone extends through  the bed,  the collection efficiency declines.
The work of Taub shows that  a  3.3 cm deep  bed of  1,500  ym par-
ticles has a capacity of approximately  3,000 g/m2(0.61  lb/ft2)
before the collection efficiency  begins  to fall.
                               31

-------
    TABLE 5.  SUMMARY OF ZAHRADNIK ET AL.'S DATA

Experimental Parameter           Run #6        Run #7
Gas flow rate, Nm3/min            0.83          0.84
Nominal space velocity, hr"      3,500         3,500
Sorbent rate, kg/hr               4.76          3.89
Temperature, °C                   204           204
Inlet S02 mole fraction          0.006         0.006
Outlet sorbent loading, g/100 g    19            10
Fly ash loading, g/m3             6.45          7.09
Filtration efficiency, %          99.3          98.7
S02 removal, I                    100           100
Pressure drop, cm W.C.            5.1           5.1
Sorbent diameter, cm              0.16          0.16

             Size Distribution of Fly Ash
       Particle diameter, urn            % less than
               104.8                       100
                82.2                        75.5
                71.6                        50.0
                60.8                        40.0
                54                          23.6
                39.6                        17.0
                32                          13.9
                27.4                        11.0
                            32

-------
sz:
w
PL,
PL,
                         0.033 m deep bed, 1.49 mm spheres,  60 cm/s gas velocity
           Figure   2  .  Time  vs efficiency  -  0.033 m deep  granular bed filtering
                        fly  ash, (Taub's  data) .

-------
TABLE 6.  SLOPES AND INTERCEPTS OF FITTED LINES
Support
Screen
Support



Grid
Support



Superficial
Velocity
(cm/sec)
8.2

11.2


8.2

11.2


Particle
Diameter
(ym)
0.8
1.6
2.9
0.8
1.6
2.9
0.8
1.6
2.9
0.8
1.6
2.9
Slope
(Transfer
unit /cm)
0.050
0.076
0.27
0.054
0.080
0.306
0.042
0.058
0.336
0.044
0.074
0.362
Intercept
(Transfer units)
(Percent)
0.017 (1.7)
0.12 (11.0)
0.66 (48.0)
0.017 (1.7)
0.16 (15.0)
0.56 (43.0)
0.46 (37.0)
0.81 (55.0)
1.04 (65.0)
0.46 (37.0)
0.81 (55.0)
1.04 (65.0)
                        34

-------
      Paretsky,et al. (1971) studied the filtration of dilute
 aerosols  by beds of sand.  They studied a bed of-10+14 mesh
 angular  sand at superficial velocities between 0.3 and 80 cm/s
 in  a  5.1  cm diameter bed at bed heights of 3.7, 8.2, and 19.2
 cm.   Flow directions studied were vertically upward, vertically
 downward, and horizontal.  Figure 3 gives data obtained at a bed
 height of 19.2 cm.
      At  superficial velocities less than about 20 cm/s, upward
 flow  exhibited higher penetration than downward flow, the differ-
 ence  in  penetration being greater at lower velocity.  Horizontal
 gas flow through the sand bed resulted in penetrations between
 those for upward and downward flows.
      Figure 4 gives data obtained at 8.2 cm height for finer
 sand  (-20 + 30 mesh).   The relationship between upward and downward
 flow  data is approximately the same for the coarse sand.
     Gebhart, et al.  (1973) published an extensive experimental
study  on  the collection of aerosol particles in packed beds con-
sisting of uniform glass  spheres.  Monodisperse aerosols were
produced  by atomizing a diluted suspension of polystyrene parti-
cles and  drying the  spray with clean air.  Aerosol particles had
diameters in the  0.1 to 2 \im range.  Concentration measurements
in front  and behind  the packed bed were carried out with a Laser
Aerosol Spectrometer.
     The  bed filter  consisted of a glass cylinder with an inside
diameter  of 8 cm.  The cylinder can be filled with glass beads
to a maximum height  of 41 cm.  The inlet and outlet of the filter
were funnel shaped to give uniform flow distribution over the
filter.  Bead sizes  investigated were 0.4, 0.16, 0.05 and 0.0185
cm in diameter.   The bed  was a horizontal bed and gas flow was
in  a  downward direction.
     The  particle penetrations are plotted against particle dia-
meter for four different  bead sizes  in Figures  5-8. The parameter
in each of the diagrams is  the mean  air velocity inside the fil-
ter; i.e., interstitial velocity.   Interstitial  gas velocity is
related to superficial gas velocity  by:

                                35

-------
  100
   80
   60
o
fr-
<
   40
z
UJ
o.
    20

                                    UPWARD FLOW
                                 Q  DOWNWARD FLOW
                                  -10+14 MESH
                                   SAND BED THICKNESS

                                   i  i i i i
                                                                 19.2 cm
      0.1  0.2    0.4
                       2        4      7    10      20
                     SUPERFICIAL VELOCITY, cm/s
                                                                 40
100
   Figure  3.   Penetration of  1.1 urn aerosol as a function of superficial  gas
              velocity:  10-14 mesh sand  (Paretsky et al. data).
   100
                                          UPWARD FLOW
                                          DOWNWARD FLOW
                                         - 20+30 MESH
                                           BED THICKNESS
      0.1  0.2    0.4
                                2       4      7   10       20
                              SUPERFICIAL  GAS  VELOCITY,  cm/s
                                                                  100
  Figure
4.   Penetration of 1.1  urn  aerosol  as  a  function of superficial gas
    velocity:  20-30 mesh sand  (Paretsky et  al.  data).
                                     36

-------
  100
O
i—i
H
<
OS
H
PJ
2
UJ
   20 L
   10
      0.1
      Figure  5.
   0.2
                        0.5
                                       dp,
Eixperimental results  obtained  by Gebhart  et  al :  Penetration
versus particle diameter  for different flow  velocities
and granule size.
           ;s
           o
           I—I
           H
           E-
           U4
           Z
           w
           D-,
                                           Z=20.5 cm   '0.72
                  0.1
           0.2
                                         d  ,
               Figure 6.   Experimental  results  obtained by C^
                          et al: Penetration  versus  particle  diameter
                          for different  flow  velocities and granule

                          size.
                                        37

-------
  100
§
1-^
i
   0.1
     0.1
      Figure 7.
                            0.2
                                     dp,
Experimental results obtained by Gebhart  et  al:  Penetration
versus particle diameter for different flow  velocities and
granule size.
               10
               101

           ~   »•
           g"   1(>-1

           I"-
           "•   10-'
               10-"
                   0.1
                                                       0.72
           0.2
                                         0.5
                                       dp,
               Figure  8.  Experimental results obtained by Gebhart
                         et al: Penetration versus particle diameter
                         for different flow velocities and granule
                         size.
                                    38

-------
                           ur
                     "Gi ' ~                            C2)

The porosity of the bed, e, was found to be  0.385  for  all  bead
sizes used.
     The diffusion and the sedimentation branches  of the pene-
tration curves are clearly distinguishable.   In between there
exists a penetration maximum.  The position  of this maximum  is
shifted to larger particles with decreasing  bead diameter.   The
exact position of maximum penetration is a characteristic  feature
of the size and shape of the bed material.
     There is a difference in measured penetrations for downward
and upward flow, as shown in Figures  9 and  10.  This is an  indi-
cation that gravity is important.  Downward  flow gives lower
penetration because the vectors of the flow  velocity and settling
velocity are complimentary.  For aerosols with diameter less  than 0.3
ym, the gravity effect is negligible.
     Knettig and Beeckmans (1974) studied the capture of mono-
disperse aerosol particles in the size range 0.8-2.9 ym in a
screen-supported and in a grid-supported fixed bed of 425 ym
glass beads.  Monodispersed aerosol particles (1:2 weight ratio
of uranine and methylene blue) were obtained from a spinning
disk aerosol generator.  The charges on the  aerosol particles
were neutralized by a radioactive source in  the aerosol genera-
tor.  The aerosol particles were sampled isokinetically upstream
and downstream of the test bed, and were analyzed quantitatively
by fluorometry.  All experiments were performed at ambient tem-
perature and pressure.  The bed was a horizontal bed with a dia-
meter of 12.7 cm.  The gas flow was in an upward direction.
     Figures 11 and 12 are experimental results.  Figure 11 shows
collection efficiency, expressed in transfer units (number of
transfer units, NTU = -InPt), plotted against bed height,  for
the bed supported on a screen, at a superficial velocity of 8.2
cm/s.  The screen was a 100-mesh screen (open area 30.3%).  Figure
12 shows a similar plot for a perforated aluminum plate, 1.6 mm
thick.  The plate had 144 holes spaced at 10 mm from center to

                               39

-------
                                                                    lo
      100
     *. 50
     2
     ta
     O.
       10
Figure  9
             I  I  I  I  I  I
                                  0.727 urn
0        0.5        1        1-5

  SUPERSTITIAL GAS VELOCITY,  cm/s


  Gravity effect in granular beds measured
  with an upward and downward directed aerosol
  stream (Gebhart et al. data).
                                                       o
                                                       I—I
                                                       t-
                                                       tu
                                                       D.
                                                                    10
                                                                    10
                                                                    10
                                                                       0.1
                                                                                                                  UP
                                                                                                       0.05 cm

                                                                                                       2.87 cm/s

                                                                                                       41 cm
  0.5
PARTICLE DIAMETER, pm
                                                                    Figure  10.   Gravity effect  in granular beds measured
                                                                                with  an upward  and downward directed aerosol
                                                                                stream (Gebhart et al.  data).

-------
  en
  H
  C/]
  1
  •**  i
  U  Z
  Z
  w
  UU
  u.
  LU

  w
  OS
               d  « 2.9
                        = 8.2 cm/s
                   5            10

                BF.D HF.IGHT, cm
Figure  11  .   Capture efficiency  (transfer units)
              versus bed height for  the grid supported
              fixed bed (Knettig  and Beeckmans data).
      CO
      {-
      t—(
      Z
      =3

      Of
      UJ
      u.
      10
                                                                        u
                                                                        I—(
                                                                        u.
      O!
      o:
                                  d  = 0.8 pm
                                                                                     0    ,
           o           5            in

                       BI:D nr.iniiT,  cm


Figure  12  . Capture efficiency  (transfer units)
            versus  bed  height for  the  screen
            supported fixed bed  (Knettig 6 Beeckman)

-------
center.  The diameter of the holes was 0.794 ram, resulting in an
open area of 0.5691.
     All of the curves gave a linear relationship between collec-
tion efficiency, expressed in transfer units, and bed height.
Straight lines were fitted to the data by least squares.  Table 6
shows the results for other superficial gas velocities.  A linear
relationship implies that collection efficiency per unit volume of
bed is independent of bed height.  Impaction appears to have been
the primary collection mechanism in the body of the bed because
collection efficiency per unit volume of bed increased with both
superficial gas velocity and aerosol particle size.
     Substantial aerosol capture occurred at the bed supports,
especially the grid support.  This is because each grid hole
acted as a jet and caused the aerosol particles to be collected
on the glass beads by inertial impaction.  However, the slopes
of the least square fitted lines for screen-supported and grid-
supported beds are the same.
     Miyamoto and Bohn (1975) studied the effect of particle
loading on collection efficiency of granular bed filters.  They
used water-washed river gravel for granules and ammonium chloride
fume for particles.  The particle diameter was 0.1 to 3 ym.
     The relationship between granular bed collection efficiency
and particle loading on the filter is presented for different
gravel diameter (Figure 13), bed depth (Figure 14) and superficial
gas velocity (Figure 15).  Particle loading is defined as the
weight of particles collected per unit bed area.  From Figures
13 through 15 the following observations may be made:
     1. The collection efficiency increased with increasing
particle loading.  The collected particles decrease bed porosity
and thereby increase filtering efficiency.
     2. Smaller granule diameters result in higher collection
efficiency and a sharper increase of collection efficiency with
particle loading.
     3. Thicker gravel layers had higher initial collection
efficiencies but little influence on collection efficiency at
                                42

-------
                   1.5mm
                   3.0mm
                   5. 2mm
                   7. 2mm
  0         0.1        0.2

  PARTICULATE LOAD, kg/m2
 Figure  13.
 Effect  of mean  gravel diameter
 and  particulate load on
 granular  bed  collection
 efficiency  (Miyamoto and
 Bonn's  data).
U
tu
I—I
U
PU
E- U.
a
J
o
U
             Z=10cm
              =3 Ocm
              = 5 Ocm
              =7 Ocm
              =90cm
Figure 14.
Effect of gravel layer thickne;
and particulate load on granuls
bed collection efficiency
(Miyamoto and Bohn's data).
 PARTICULATE  LOAD,  kg/m"
              =2.0 cm/s

              =3.0 cm/s

              =4.0 cm/s
"0          0.1         0.2

PARTICULATE LOAD, kg/m2
 Figure  15.
 Effect  of  superficial gas
 velocity and particulate
 load  on granular bed
 collection efficiency
 (Miyamoto  and  Bohn's data).
                            43

-------
higher particle loading because the collection is preferentially
near the surface where collected particles tend to form a cake.
     4. Higher superficial gas velocities markedly decreased the
rate of increase in efficiency with particle loading, probably
because filter cake is less stable at higher velocities.
     Figueroa (1974) and Figueroa and Licht (1976) conducted an
experimental program to determine the aerosol filtration effi-
ciency of a bed of granular solids.  The bed was a 10.2 cm (4
inch)  I.D. column packed with plastic beads and sand to various
depths.  The bed support was made of 325 mesh stainless steel
screen.
     Two different size fractions of plastic beads (polyacrylo-
nitrile) and one grade of sand were used as granular bed solids.
The properties of the granular materials are listed in Table 7.
     Two test aerosols were used:  monodispersed methylene blue
(MB) particles 1 ym and 2 ym in diameter and monodispersed poly-
styrene latex microspheres  0.50, 1.10, and 2.02 ym in diameter.
Aerosol number concentrations before and after the bed were
measured with an optical counter.
     Bed penetration was measured as a function of aerosol par-
ticle  diameter, granule diameter, bed depth, direction of flow,
and superficial gas velocity.  Figures 16 through 19 summarize
their  data.  For the low gas velocity range (20 cm/s), they
observed, for a fixed bed, that particle collection efficiency
always increases with decreasing granule diameter and increasing
bed depth.  Downward flow gives higher collection efficiency than
upward flow.  The collection efficiency of a granular material
with inherent electrostatic charging properties (e.g., plastic
beads) is higher than for a granular material without those proper-
ties (e.g., sand) .
        i
     Westinghouse Research Laboratories (Ciliberti, 1977) carried
out an experimental program to investigate efficiency and opera-
bility of granular bed filters.  The experimental bed was designed
with a 233 cm   (0.25 ft) cross section.  The bed was cleaned
by fluidization.  A distributor consisting of several independent
drilled tubes was used to control the fluidizing gas flow.  The

                                44

-------
TABLE 7.  EXPERIMENTALLY DETERMINED PROPERTIES OF THE GRANULAR MATERIALS
PROPERTY

Number
Median
Diameter

Sauter
Diameter

Geometric
Standard
Deviation

Solid
Granule
Density

Bulk
Density

Porosity or
Voidage
Fraction
          (Figueroa, 1974)

        SYMBOL        UNITS
          cN
          3 2
          g
micrometers



micrometers


micrometers



    g/cm3



    g/cm3
  POLYACRYLONITRILE
                                      495
1.10



0.67


0. 39
                  305
                      1.12



                      0.66


                      0.38
                                                                               SAND
-25 + 40 mesh    -40 + 60 mesh    -25 + 40 mesh


                                      680
514
1.13
339
1.23
702
1.12
                                                                        3.10




                                                                        1.85


                                                                        0.41

-------
  1.0
  0.1
  10'
|io-J
  10"
  10
           d  • 420  to  710  um  PLASTIC  BEADS
            Z • 3 on
        2.0 vm
                                     Lt i i I A i i I i i i I  i i i   0
                                                           12
                                                           10
                                                              3=
                                                              e
                                                              Q


                                                              £
    0         4         8         12         16        20

                SUPERFICIAL GAS VELOCITY, cm/s

Figure 16.  Penetration tests on 420 to 710  micrometers plastic
            beads by monodispersed polystyrene latex aerosol
            (downflow) (Figueroa and Licht's data).
                                                                         o
                                                                         o
                                                                         Q!
                                                                             1.0
                                                                             0.1
                                                                            10"
                                                                          10"
                                                                         £  10
                                                                            10"
                                                                            10"
                                                                                             d  -  420  to  710  \im PLASTIC
                                                                                                        d  « 250 to 420 vim PLASTIC
                                                                                                         c  BEADS
                                                                                    9 cm
                                                                                       d  - 0.5 urn PSL
                                                                                           I
                                                                                           4         8          12         16

                                                                                            SUPERFICIAL GAS VELOCITY,  cra/s
                                                                                                                                    20
                                                                            Figure  17.  Effect of bed height and bed  granule  (bead)  size
                                                                                        on the downflow penetration of 0.5 micrometers
                                                                                        polystyrene latex aerosol  particles on plastic
                                                                                        beads  (Figueroa and Licht's data).

-------
m      d - -25 +40 MESH SAND
10
   0         4         8         12          16

                SUPERFICIAL GAS VELOCITY, cm/s

Figure 18.  Comparison of downflow penetration  of polystyrene
            latex aerosol particle on 420 to 710 micrometers
            plastic beads and  -25 +40 mesh  sand granules
            (Figueroa and Licht's data).
                                                                          1.0
                                                                          0.1
                                                                          10"
                                                                        o
                                                                        H
                                                                        CH
                                                                        t-
                                                                        5
                                                                          10
                                                                          10
                                                                                                  I'M
                                                                                         d  = 420 to  710  ym

                                                                                          PLASTIC BEADS    •
                                                                               - UPWARD GAS
                                                                                  F
                                                                                              ^DOWNWARD
                                                                                               GAS FLOW
                                                                            d  = 0.5 m PSL

                                                                            Z = 3 on
                                                                                        4          8          12        16

                                                                                            SUPERFICIAL  GAS VELOCITY, cm/s
                                                                                                                          20
                                                                           Figure 19 .
                                                                               F.ffect of bed granule  (bead)  size  and flow
                                                                               direction on the penetration  of  0.5  micrometer
                                                                               polystyrene latex aerosol particles  on plastic
                                                                               beads.  Bed weight = 150 grams,  bed  height = 3 cm
                                                                               (Figueroa and Licht's  data).

-------
arrangement of the bed is illustrated in Figure 20.
     Bed depth ranged from 7.6 cm to 15.2 cm (3 to 6 in.) based
on requirements to minimize pressure drop while maintaining a
bed that can be successfully fluidized for cleanup.  Granule
size was in the range of 15 to 30 mesh to achieve efficient dust
collection with small bed depth and reasonable pressure drop.
Superficial gas velocity ranged from 9 to 30 cm/s.  The test
dust was finely ground (<10 ym) limestone dispersed in a high
velocity air jet.  The layout of the experimental equipment is
shown in Figure 21.
     Test results are summarized in Tables 8 through 12.  Particle
size and mass data were taken with cascade impactors.  Collection
efficiency for submicron particles was high.  The grade effi-
ciency curves were flat and showed high collection efficiency
for all particle sizes.  This was consistent with the observed
formation of a filter cake.
     To investigate dust accumulation in the bed, Westinghouse
made a series of five runs.  The initial clean bed material was
sampled, then after five consecutive runs bed samples were taken
at four levels through the bed.
     The clean bed contained 0.5 wt I fine dust.   After five
runs the dust content of the bed was 1.3 wt $.  At the end of the
filtration cycle, the dust level at the bed surface was 10 wt %.
However, at levels 2.5 cm below the surface and greater the dust
level was uniform at 1.3 wt %.
     A second test over ten cycles showed dust accumulation of
1.0 to 1.8 wt % in the bed.
     Another filter, 0.65 m2 (7 ft2), was tested  at temperature
of 1,100°F at atmospheric pressure.  Because of operating pro-
blems tests were discontinued.
     Combustion Power Company has conducted extensive  cold  flow
tests on a moving bed granular bed filter.  These results will
be discussed later.
High Temperature Filtration
     Dennis et al. (I960), in designing an incinerator for dis-
posal of low-level radioactive wastes from hospitals or biological

                               48

-------
                  FLUIDIZING  GAS  OUT
                                  DIRTY  GAS  IN
FLUIDIZING
  GAS IN
    A - DEPTH OF SAND BED

    B - DEPTH OF COARSE GRAVEL USED AS BED SUPPORT
                                                        FILTERED
                                                        GAS OUT
                Figure 20.  Schematic of granular bed filter
                            (Westinghouse setup).
                              49

-------
01
o
    OUTLET
    SAMPLE-
    PORT
                 INLET  SAMPLE
                          PORT
t
*
                              '»• • * • ;','.?•••'*•
                                                                               DUST
                                                                              HOPPER
                                                                    COMPRESSED
                                                                         AIR
                                Figure 21.  Schematic  of test  equipment  (Westinghouse).

-------
                 TABLE  8. WESTINGHOUSE  GBF  DATA

             Series #2    Gas Velocity:  17.8 cm/s  (+_ 10%)
                          Bed Material-20+30 mesh  Ottawa  sand
                          Bed Depth:  7.6 cm  (3  in.)

Particle Size      Inlet Dust      Outlet  Dust    Collection
                   (Typical)       (Typical)       Efficiency,
                    mg/m3           mg/m3
0-0.3
0.3-0.45
0.45-0.75
0.75-1.5
1.5-2.3
2.3-3.3
3.3-5.0
5.0-8.0
8.0 +
8.0 +
30.6
71.8
120.6
132.4
59.0
10.5
12.3
10.5
10.0
14.1
0.93
1.81
2.6
1.5
0.3
0.25
0.25
0.05
0.1
0.1
96.9
97.5
97.8
98.9
99.4
97.6
98.0
99.5
99.0
99.2
 Run No.               Inlet Loading g/m3      Overall Efficiency,
   2.7                        0.28                    98.8
   2.8                        0.47                    98.3
   2.9                        2.5                     97.9
   2.10                       1.5                     97-5
                                 51

-------
                TABLE 9  .  WESTINGHOUSE GBF DATA
            Series #3
Gas Velocity: 10.2 cm/s (+_ 10%)
Bed Material-20+30 mesh Ottawa sand
Bed Depth: 7.6 cm
Particle Size
  0-0.3
  0.3-0.45
  0.45-0.75
  0.75-1.5
  1.5-2.3
  2.3-3.3
  3.3-5.0
  5.0-8.0
  8.0 +
  8.0 +
Inlet Dust
(Typical)
  mg/m3
  12.9
  43.5
 102.0
 194.0
 203.0
 105.0
 108.0
  59.0
 111.0
 230.0
Outlet Dust
 (Typical)
   mg/m3
  1.1
  2.7
  3.6
  2.5
  0.5
  0.1
  0.05
                       Collection
                       Efficiency, %
                           91.5
                           93.8
                           96.5
                           98.7
                           99.8
                           99.9
                           99.95
                          100.0
                          100.0
                          100.0
  Run No.
   3.11
   3,12
   3.13
   3.14
   3.15
  Inlet Loading g/ms
         0.6
         1.16
         0.72
         0.72
         0.78
                 Overall Efficiency,
                  97.2
                  99.0
                  99.2
                  97.7
                  99.4
                      (95.5)
                      (98.8)
                      (98.5)
                      (96.2)
                      (99.4)*
  *Increased gas rate to 17.8 cm/s
                                52

-------
                TABLE 10.  WESTINGHOUSE GBF DATA

            Series #5     Gas Velocity: 25.4 cm/s  (+_ 10%)
                          Bed Material-20+30 mesh  Ottawa sand
                          Bed Depth:  7.6 cm

Particle Size     Inlet  Dust     Outlet Dust     Collection
(ym)
0-0.3
0.3-0.45
0.45-0.75
0.75-1.5
1.5-2.3
2.3-3.3
3.3-5.0
5.0-8.0
8.0 +
8.0 +
Run No.
5-22
5-23
mg/m3
52
66
100
138
79
37
41
34
73
221
Inlet


mg/m3

0.1
0.14
0.2






Loading mg/m3
1.0
0.8
Efficiency, %
100.0
99.8
99.8
99.8
100.0
100.0
100.0
100.0
100.0
100.0
Overall Efficiency,
99.96
99.95
                                 53

-------
                TABLE 11.   WESTINGHOUSE GBF DATA
            Series  #6
        Gas Velocity:  25.4 cm/s (^ 10%)
        Bed Material-16+20 mesh Ottawa sand
        Bed Depth:  10.2 cm
Particle Size
Inlet Dust
   mg/m3
Outlet Dust
    mg/m3
Collection
Efficiency,
0-0.3
0.3-0.45
0.45-0.75
0.75-1.5
1.5-2.3
2.3-3.3
3.3-5.0
5.0-8.0
8.0
8.0
32
20
33
53
47
32
53
46
108
513
0.39
0.08
0.12
0.08
0.04
0.04
0.04


_ _
98.8
99.6
99.6
99.8
99.9
99.9
99.9
100.0
100.0
100.0
  Run No.
   6-24
   6-25
   6-26
   Inlet Loading mg/m3
           5.8
           0.94
           1.17
           Overall Efficiency,
            99.96     (99.92)
            99.92     (99.8)
            99.90     (99.7)
                                54

-------
         TABLE 12.  WESTINGHOUSE GBF DATA
    Series #7
      Gas Velocity: 17.8 cm/s (+_ 10%)
      Bed Material-16+20 mesh sand
      Bed Depth: 10.2 cm
Particle Size
    Cum)

  0-0.3
  0.3-0.45
  0.45-0.75
  0.75-1.5
  1.5-2.3
  2.3-3.3
  3.3-5.0
  5-0-8.0
  8.0+
  8.0 +
         Inlet Dust
         Run 7-27
           (mg/m3)
            9.4
            25.9
            51.8
            56.5
            16.5
             1.2
             2.4

   TOTAL   163.5
Inlet Dust
 Run 7-28
  (mg/m3)
   11.7
   42.4
   73.0
   88.3
   41.2
    7.1
    2.4
    1.2
                                        266.0
   Overall  Efficiency
   7-27
   7-28
99.55%
99.58%
                         55

-------
laboratories, used a 20 cm (8 in.) layer of 0.64 cm  (0.25  in.)
gravel to screen out coarse particles before the gas passed
through a 5.1 cm (2 in.) bed of slag wool.  This filter unit,
with a 2,600 cm2 (2.8 ft2) filter area, was housed in half of
a 55-gal drum located approximately 244 cm (8 ft) downstream from
the incinerator.  Gases exiting from the incinerator at 870 to
982°C (1,600 to l,800°F) were passed at negative pressure through
a water-cooled condenser so that filtration gas temperatures
were 93 to 427°C (200 to 800°F) with a pressure drop of about 2.5 cm
W.C. (1 in. W.C.).  In one series of tests in which 408 kg
(900 Ib) of sawdust was burned, the pressure drop increased from
1.3 cm W.C. to 1.8 cm W.C. (0.5 to 0.7 in. W.C.) with a 90-98%
filter collection efficiency on a weight basis.
     Strauss and Thring (1960) carried out studies on filtration
of submicron fumes from open hearth furnace gases using a  granu-
lar bed.  The bed was 5.1 cm (2 inches) in diameter and consisted
of 0.79  (5/16 in.) crushed high temperature insulating bricks.
Experiments were run with variety of bed thicknesses and gas flow
rates.  Collection efficiency tests were carried out on cold and
preheated  beds.
     Table 13 shows data obtained from Strauss and Thring's granu-
lar bed filter study.
     Collection efficiencies of 59.3 to 96.3% were obtained
with a bed depth of 25  to 26.7 cm (1 to 10.5 in.), average gas
velocities of 36.3 to 102.2 cm/s  (1.2 to 3.4 ft/s), and maximum
pressure drops of 0.97  to 12.4 cm W.C. (0.38 to 4.9 in W.C.)
at gas temperatures from  230 to 520°C.
     Further theoretical studies by Thring and Strauss (1963)
considered the effect of high temperature on particle collection
mechanisms.  The controlling mechanism in these tests might be
inertial impaction, but they emphasized the importance of  the
effect of inlet dust concentration on efficiency.  It was  main-
tained that dust particle agglomeration in the bed occurring
in the tortuous paths between the collecting granules plays
a highly significant role in collection.  Increased mass flow
                              56

-------
                                    TABLE  13. DATA OBTAINED FROM STRAUSS AND THRING GRANULAR  BED  FILTER STUDY
cn
Test Furnace
No. Operation
3
4
5
8
9
10
12
14
15
32
36
37
41
45
49
56
71
. 80
R*
R
C
C
R
0
R
A
T
M
M
R
R
R
F
R
R
R
Test Time
(min)
13
15
19
4.88
13.13
6.95
15.1
14.4
12.7
24.9
20.4
10.3
20.6
13.0
10.4
15.1
22.2
IS. 9
Gas Mass
Flowrate
kg/m2-hr
1,103
996
1,098
839
708
683
1,044
1,035
996
615
786
976
1,650
1,005
1,547
1,249
1,532
1,728
UG
cm/s
43.6
36.6
44.8
36.3
29.3
24.1
50.9
47.0
46.4
28.1
38.4
45.8
102.2
39.3
89.7
63.7
76.6
70.2
T~ Inlet Fume
or Concentration
L g/Nra3
230
230
275
290
270
237
345
303
317
305
352
300
520
225
460
370
360
455
1
4
0
6
3
6
0
0
0
0
0
1
1
0
0
0
1
4
.63
.14
.123
.26
.83
.24
.93
.54
.60
.127
.163
.24
.73
.595
.232
.573
.18
.94
Fe203in
Fume
Wt *

--
--
--
--
78
64
20
23
36
44.2
55.2
--
--
--
--
--
--
Z
(cm)
26.7
26.7
26.7
26.7
26.7
22.9
22.9
22.9
22.9
22.9
22.9
22.9
7.6
7.6
7.6
2.5
2.5
Z.5
Collection
Efficiency
Wt *
90.5
96.3
87.3
87.6
94.7
85.0
90.2
84.0
87.5
84.6
74.3
94.2
85.3
65.5
82.7
59.3
87.0
88.2
Collection Pressure
Efficiency of Drop
FezOj, Wt t cm W.C.
5.6
5.6
6.6
8.9
10.2
87.3 11.9
96.0 5.6
97.1 . 6.1
98.1 6.4
85.7 12.4
74.1 9.1
9.9
5.3
1.8
6.0
0.97
2.2
1.3
                      * These letters  refer to  the  phase  of  the steelmaking cycle in progress at the time of the test.
                      0 = Oxygen lancing     M " Melting
                      R - Refining           c = Charging

-------
rate  of gas and increased temperature also increased collection,
but to a lesser extent.  When the temperature of the gases
differs greatly from that of bed, thermal precipitation plays
a role, but other mechanisms are more important.
     Goldman (1964) reported that gravel bed filters have been
used for several years in Germany as large-pore filters that are
wear resistant in high temperature applications (to 350°C) . The
theory of the collection process is given briefly.  Prior to
adoption of these filters, tests were made with various dusts,
including coke dust in the off-gases from a coke-drying operation,
phosphate dust, dust in the fumes from a carbide furnace, and
dust in the waste gases from a mixture of phosphorescents.
 Gases  contained  0.5 to 3  g/Nm3 of dust; the outlet gases con-
 tained 10  to 95 mg/Nm3.
     Dust  removal was  in  the range of 93 to 97%, with pressure drops
 of 11  to  20 cm W.C. and  flow rates of 4,000 to  7,000 m3/hr. When
the pressure drop became too high, the gravel was  washed and
the clean gravel returned to the bed for reuse.
      The  U.S.  Bureau  of  Mines, Morgantown  Coal  Research  Center
 tested a  GBF of  Squires'  design  at high temperature  (540°C,
 1,000°F)  (Wu,  1977).   The bed  is  a vertical  layer  of  sand held
 in place  by louvered  walls.  The  filtering surface of  the bed
 was 7.6 cm (3  inches)  wide and  30.5  cm  (12 inches) tall.   Several
 grades of sand were used as bed  material.
      The  test  dust was derived  directly from  a  boiler furnace.
 Pulverized coal  with  70% passing  through a 200  mesh  sieve was
 burned.   The coal  was  fed by a  screw feeder driven by  a  variable
 speed  motor into a combustor with a  capacity  of 2.3  kg/hr  (5  lb/
 hr).   The temperature  inside the  combustor was  about  l,093°-to
 1,315°C (2,000 to 2,400°F).   Natural gas  was  also burned during
 the coal  combustion  in order  to  insure  complete combustion
 because of a coal  feed rate problem.
      At a face velocity  of 10  cm/s  (20  ft/min), the  overall
 collection efficiency was 98%.   No particle  size  distribution
 data were given.
                               58

-------
     The City College of New York  tested  the  same  design  of  the
GBF at room temperature.   Redispersed  coal  fly  ash was used  as the
test dust.  The overall collection efficiency was  99.99%.  The par-
ticle size distributions might  not be  the same  for the two tests,
we cannot be sure whether  the lower efficiency  was a  result  of
operating at high temperature.
     One observation regarding  the difference in the  filter  cake
was recorded.  In the Bureau of Mines' high temperature tests, no
filter cake over the sand  surface  was  observed.  At near  room tem-
perature, a good surface filter cake was  observed.  Squires  (Wu,
1977) attributed this difference in filter  cake to the decrease
of adhesive and autohesive forces  of fly  ash  at high  temperatures.
Adhesion is defined as the interaction of particles with  a solid
surface and autohesion is  defined  as the  interaction  of particles
among themselves.
     A high temperature and pressure design of  the Ducon  granular
bed filter was tested at the miniplant of Exxon Research  and•
Engineering Company (Hoke, et al.,  1978).   Exxon's experience with
this filter will be discussed in a later  section.
GBF With Flux Forces
     Anderson and Silverman (1957,  1958)  reported  on  a study con-
ducted at the Harvard University Air Cleaning Laboratory  to inves-
tigate electrostatic filtration in fixed  and  fluidized granular
beds.  They observed that  triboelectrification or  friction char-
ging of fibrous and granular filter media can improve collection
efficiency with no increase in  flow resistance.  Similar results
have recently been reported by  Figueroa and Licht  (1976).
     Fuchs and Kirsch (1965) conducted tests  to determine the
effect of vapor condensation on the collection efficiency of gran-
ular beds.  Monodispersed  aerosols  of selenium  in  nitrogen (par-
ticle diameter of 0.2 and  0.4 ym)  were passed through a column
of silica gel, both directly and upon addition of  some ether vapor.
It was found that with the addition of ether  vapor, the collection
efficiency of the granular bed  increased.   Condensation of the
vapor and deposition of the particles occur simultaneously in the

                                59

-------
bed.  The vapor molecules diffusing towards the granules  sweep
the particles along with them.
     A study of the collection of submicron aerosol particles in
an electrified granular bed was carried out by Research-Cottrell,
Inc., Bound Brook, New Jersey.  One such device is described in
U.S. Patent No. 2,990,912, July 4, 1961.  Collection of particles
occurred in a bed of 3 to 6 mm diameter glass beads held  between
two screens maintained at different electrical potentials  (Figure
22) .  The aerosol particles were electrically charged upstream
from the bed.  Data taken on collection of a 0.5 to 0.7 ym dia-
meter methylene blue particles are shown in Figure 23.  It is noted
that a 99.5% efficiency was obtained in a. bed under 5.1 cm (2 in.)
in thickness with a pressure drop of 0.5 cm W.C. (0.2 in W.C.)
and a a face velocity of 61 cm/s (2 ft/s).
     Sharapov  (1975) tested a high gradient magnetic filter with
a bed of 8  mm steel balls for the removal of dust from open-
hearth furnace.  The capacity of the filter was 60 m3/min.  At
the optimum .voltage of 80 to 120 kV/m, the collection efficiency
was 80 to 901, and the energy consumption was 0.05 kWh/1,000 m3
of gas.  Without the magnetic field the collection efficiency was
25 to 30%.
Gaseous Pollutants
     Only limited work has been done in this area.  Zahradnik et
al. (1970) and Squires and Graft (1971) have demonstrated the fea-
sibility of simultaneous   removal  of fly ash and S02 from gas
streams using a granular bed filter packed with alkalized alumina
and half calcined dolomite.  Swift et al. (1977) are studying the
removal of alkali metal (sodium and potassium) compounds  from the
combustion gases at high temperature with a packed bed.  NaCl is
passed through beds packed with alundum, Celatom MP-91 diatomaceous
earth, Burgess No. 10 pigment (kaoline clay), attapulgus  clay
(magnesium aluminum silicate), and activated bauxite.  The tem-
peratures of the vapor and bed are maintained at about 870°C. The
gas velocity passing through the bed is 7.5 cm/s (3 in./s).  Pre-
liminary test results revealed that Celatom MP-91 diatomaceous
                               60

-------
o\
                                                                Electrified
                                                                   Beads
                                    Charging Zone  -=.
                               Figure   22  .  Electrified packed bed.

-------
     60
W
_J
u
t— (
0
Di
U
 t
w
I— 1
u
I— I
tL,
UH
 o
 w
 pa
 (X
     80
     90
95
99
     99.5
     9-9.9
           T	1	1	1	1	

                  FACE VELOCITY =2.44 m/s  (8.0  ft/s)-
              .3)
                                        1.22  m/s
                                           (4.0  ft/s)
                                                     12.3).
                                      =  0.61  m/s (2.0 ft/s)

                                     (0.46)
                                                      fO.51.
             NOTE

             EFFICIENCIES DO NOT  INCLUDE  CHARGING SECTION
             COLLECTION OF APPROX.  50%

             FIGURES IN PARENTHESIS ()  DENOTE PRESSURE
             DROPS  (cm W.C.) ACROSS BED
            I
I
1
I
        0.6     0.8     1.0      1.2       1.4     1.6     1.8

             PACKED BED THICKNESS, 6.0  mm SPHERES (INCHES)


    Figure  23.  Performance  characteristics of electrostatically
                augmented packed bed  (Research-Cottrell data).
                                 62

-------
earth, Burgess No. 10 pigment,  and  activated bauxite are promising
candidates as hot packed bed  sorbents  to remove  alkali vapor from
hot coal combustion gases.  Among the  three, activated bauxite,
which is thermally treated high-alumina-content natural bauxite
ore, is the most active for removing NaCl vapor under the test
conditions.

THEORY
Particulate Collection
     The primary mechanisms for particulate collection in a bed
of granular solids are:
     1. Inertial impaction
     2. Flow-line interception
     3. Diffusional collection
     4. Gravity settling
Inertial Impaction -
     The inertial impaction mechanism  is based on the inertial
force of the particle.  The inertial force tends to move parti-
cles across the gas flow lines  toward  the collecting surface in
regions where the flow is diverging upstream of a fixed boundary
surface. Inertial forces increase with particle velocity and mass.
Diffusion  -
     Diffusion of the aerosol particle is based on the theory of
Brownian motion.  The diffusion mechanism becomes increasingly
effective with decreasing particle  diameter.
Gravity Settling -
     The gravitational force  can be significant for collection
of particles as small as 0.4  ym in  diameter.  This effect has
been demonstrated experimentally by the decrease in penetration
observed for downward flow compared with upward flow in sand
beds  (Thomas and Yoder, 1956b).  A  gravitational force in the
direction of the bulk flow will tend to increase collection due
to settling across flow lines.
                               63

-------
Interception -
     Interception is the mechanism whereby particles are collected
on surfaces while following the gas streamlines.  Collection by
this mechanism usually is negligible for a clean granular bed.
However, during a filtration cycle, particles will deposit in the
interstices of the bed to form an internal cake and on the surface
to form a surface cake.  As the cake builds up the bed porosity
and flow channels decrease and interception becomes an important
collection mechanism.  When the flow channels are too small to
allow particles to pass, collection is ensured, being referred
to as complete interception or sieving.
     In finely packed beds operated at low gas velocities, gravity
settling and diffusional deposition will predominate.  The collec-
tion efficiency will be expected to decrease as the gas velocity
increases.  Coarsely packed beds operating at higher velocities
(but still below fluidizing velocities) provide separation mainly
by inertial deposition and interception.  The collection efficiency
will increase with velocity provided that the gas velocity is not
so high as to reentrain collected material.
     The operation of granular bed filters is similar to fabric
filters even though granular bed filters have larger pore sizes
and deeper beds.  Payatakes (1977) classified the filtration cycle
into four successive stages.
     1. When the filter is new, particles deposit directly on the
granule surfaces.   This is referred to as clean bed filtration.
The collection efficiency for this stage of filtration depends
primarily upon the granule size and the depth of the bed.
     2. Particles deposit not only directly on granules but also
and preferentially on deposited particles, thereby forming par-
ticle dendrites.
     3. The dendrites grow to the extent that they intermesh with
their neighbors forming a particulate coating around each granule
which is non-uniform in thickness.
     4. If the granules of the bed are sufficiently small, particle
coatings of neighboring granules will bridge the gap to form an
internal cake.  Lee (1975) called the internal cake a rooting cake

                               64

-------
and it is the foundation which  supports the  formation of a surface
cake.  Once a surface cake  is formed,  filtration efficiency no
longer depends upon the depth of  the granular bed but rather on
the thickness and structure of  the  surface cake.  The cake filtra-
tion results in a much higher efficiency  than the original granular
bed, and particle collection by sieving becomes a more important
collection mechanism.
Mathematical Models -
     Currently there are several  mathematical models available
for the prediction of particle  collection in a granular bed.  All
models are for the prediction of  particle collection by clean
beds; i.e., stage 1 filtration.   Stage 1  filtration is very brief
compared to the total filtration  cycle.
     Model by Jackson and Calvert - Jackson  and Calvert (1966) and
Calvert  (1968) have developed a theoretical  relationship between
particle collection efficiency  and  packed bed operating parameters.
They assumed that the gas  (and  particle)  flow through the bed may
be modeled by the flow through  a  series of semicircular channels
and  that particles are collected  by centrifugal force on the out-
side channel walls as gas and particles pass through the channels.
Their equation for predicting the particle penetration for a packed
bed  is:
                    Ft, . erp I-  C, -± K I                  (3)
 where     Ptj  =  penetration for particles  with  diameter "dp",
                fraction
           C:  -  empirical  constant,  dimensionless
            Z  =  depth of the bed,  cm
           d  =  granule diameter,  cm
           K  =  inertial parameter,  dimensionless

                V  C<  S  UG
                 9 HG dc
                                65

-------
           d   = particle diameter, cm
            C1 = Cunningham slip factor, dimensionless
            p  = particle density, g/cm3
            Up = superficial gas velocity, cm/s
            yG = gas viscosity, g/cm-s

     The empirical constant, "Cy, is a function of bed porosity,
channel width and granule diameter.  It can be calculated by
the following formula:
                     C  =
                              IT                           , .x
                                                          (4)
                          — v j j j —
where   j = ratio of channel width to packing diameter, dimen-
            sionless
        e = bed porosity, dimensionless

     For a packed bed with a granule diameter of 1.27 cm (0.5 in),
the empirical constant was found to be 21.4.
     Model by Paretsky et al. -  Paretsky et al. (1971) proposed
the following equation (based on Happel's "free surface model")
for particle penetration through a granular bed:


               Ptd  =
                          -  2     dc   J
where      Ptd = penetration of particles with diameter "d ",
                 fraction
             e = bed porosity, fraction
             Z = bed depth, cm
            d  = granule diameter, cm
             n = overall single granule collection efficiency,
                 fraction
                               66

-------
     Single granule collection efficiency  includes the collec-
tion  by: (1) Brownian diffusion,  (2) direct interception, (3)
inertial impaction, and  (4) gravity settling.  The theoretical
equations using the "free surface" model for each of these four
collection mechanisms are:
                               4(N  )
     Brownian Diffusion:  nn - - Sh avg                 (6)
                                  NPe
                              =  5.04  f(er'/J Npe2/3
where   nD = single granule collection efficiency, fraction
         e = bed porosity, fraction
       NSh = Sfterwood number, dimensionless
       Np  = Peclet number, dimensionless

             2  - SCl-e)1 3+3(l-e)5/3-  2(l-e}2
                        1  - (1-e)5
                                              2
       Direct  Interception:   nnT = 	  I-£• I            (-8J
                              DI   ffel
where  nDI  =  single  granule  collection  efficiency by direct
              interception, fraction
        d   =  particle  diameter,  cm
         P
        d   =  collector diameter,  cm
      Inertial  Impaction:
                                    c
                                67

-------
where     TU = single granule collection efficiency by  impaction,
               fraction
       y  .   = critical trajectory of the aerosol, cm


            Gravity Settling:  r\   = -t  A                (10)
where    nGS = single granule collection efficiency by gravity
               settling, fraction
          u  = terminal settling velocity of the particle,
               cm/s
          UG = superficial gas velocity, cm/s
          A  = fraction of the projected area of a single
               collector particle which is available by capturing
               the aerosol by settling, fraction

"A  " can be taken as the minimum projected area available for
  p
flow between particles.  For a triangular packing it is 0.0377
and for a square packing it is 0.0871.
     Model by Miyamoto and Bohn -  Miyamoto and Bohn (1974) de-
rived the following equation for the particle collection in a
clean bed  (diffusion only) based on the expression of a single
sphere collection efficiency by Friedlander (1957) .
            Pt, = exp
6(l-e)Z  NNu
  dc     NPeJ
where     Pt,  = penetration for particles with diameter  "d  ",
                fraction
            e = bed porosity, fraction
            Z = bed depth, cm
           d  = single granule diameter, cm
          NNu = N11556!1 number, dimensionless
              = dckG
              "V

                               68

-------
          NPe  =  Peclet  number,  dimensionless
                 dc  uGb
                  Dpe
            k  =  mass  transfer coefficient,  cm/s
          Dpe  =  effective  particle diffusion coefficient in
                 granule  layer,  cm2/s
          uGb  =  actual 8as velocity in bed,  cm/s
     The effective  particle diffusion coefficient  in  granule
layer is related  to particle diffusivity by:

                Dpe '  Dp  e  * A*

where   D  = particle  diffusivity,  cm2/s
         * • relative  force field,  dimensionless
         A = tortuosity factor,  dimensionless

     Interstitial gas  velocity can  be calculated from superficial
gas velocity by:
                              A  UG
                        uGb = -7-                        (12)

     Model by Gebhart.et  al.  -   Gebhart  et al.  (1973) performed
an experimental study  on  the filtration  of aerosol particles in
the 0.1-2 ym size range by  a packed bed  of glass beads.  Based
on their data, they proposed an  empirical equation for aerosol
penetration through packed  beds  of  spheres under conditions at
which Brownian diffusion  dominates:
                              D  2/3  Z
            Ptd = exp
-6.39
      u  2/3 CO.Sd
                              *Gi
(13)
where      Pt_, = particle penetration for particles with diameter
                 "d ", fraction
            D  = particle diffusivity, cnf/s
             P
                                69

-------
            e =
          u
           Gi
           dc
            Z
bed porosity, fraction
interstial gas velocity
collector diameter, cm
bed depth, cm
                                               , cm/s
     Balasubramanian and Meisen (1975) showed that this  equation
may be derived independently from Wilson and Geankoplis1  (1966)
correlation for mass transfer coefficients.  Wilson and  Geankoplis
correlation is:

               kG = 1.09 (uG/e) Npe~2/3                    (14)

where k  = mass transfer coefficient, cm/s
       G
     Np  = Peclet number, dimensionless
         . dc "G
            DP
Equation (16) is valid for 0.35< e < 0.75, 0.0016 < NRe  <  55,
and 950  < N~   < 70,600.  The equation derived by Balasubramanian
and Meisen is:
           Pt, = exp
-2.06
(l-O  D  2/3
e uG2/3 (0.5dc)
                                          (15)
This equation is more general than the equation by  Gebhart  et  al.
After substituting e = 0.385  (bed porosity used by  Gebhart  et  al.)
equation  (17) reduces to equation (15).
     Model by Bohm and Jordan -  Bohm and Jordan  (1976) ,  using
capillary flow for describing the behavior of  sand  bed filters,
derived an expression for particle penetration.
    Pt , = exp  -
/2k T ef + V « pp e dc + '
\
\3Tr d UG 36 UG
T dp2 pp UG\ 4 f'Z~
. i
18e / yG dc2
                                                          (16)
                               70

-------
where      Ptd  =  particle penetration for particles with
                 diameter "d ",  fraction
             k  =  Boltzmann's constant
               =  1.38 x 10"16 erg/°K
             T  =  absolute temperature,  °K
            d  =  particle diameter,  cm
             e  =  bed porosity,  fraction
            UG  =  superficial gas velocity , cm/s
             g  =  gravitational  acceleration, cm/s2
            d  =  collector diameter, cm
             w
            p  =  particle density,  g/cm3
            y^  =  gas viscosity,  g/cm-s
             Z  »  bed depth, cm
            f  =  d /d  >  dimensionless
                  c  o
            d  =  initial capillary  diameter,  cm

     The first  part of the exponent stands for diffusional
deposition, the second for gravity settling and the third for
inertial impaction.  Interaction terms between the three col-
lection mechanisms are neglected.
     Model by Goren - Goren  (1977)  derived a semi-empirical
equation for granular bed collection efficiency.   The model
is based on collection by individual granules:
                  Ptd = exp   - -  (1-e)  —  n
 where:   Ptj  = particle penetration for particles with
               diameter "d ",  fraction
           e  = bed porosity,  fraction
           Z  = bed depth,  cm
          d  = granular diameter, cm
           n  = overall single granule collection efficiency,  fraction
                               71

-------
     Goren ran a small scale experiment with 2 mm diameter  gran-
ules as bed material.  He then derived an expression for "n"  from
data and equation (17).  Collection by diffusion, settling, and
impaction were considered.  The expressions are:
                                                          (18)


                                                          (19)

                                                          (20)
where    n~, rir-c* HT = single granule collection efficiency due
          U   bo   I
                       to diffusion, gravity settling, and impac-
                       tion, respectively, fraction
                 Np  = Peclet number, dimensionless
                  d  = granule diameter, cm
                  Up = gas velocity, cm/s
                  D  = particle diffusivity, cm2/s
                  u  = terminal settling velocity, cm/s
                  d  = particle diameter, cm
                   g = gravitational acceleration, cm/s2
                  \ir = gas viscosity, g/cm-s
                  C' = Cunningham slip factor, dimensionless
                  P  = particle density, g/cm
                  K  = inertial parameter, dimensionless

     Model by Westinghouse - Westinghouse Research Laboratories
 (Ciliberti, 1977) also developed a mathematical model.  They  took
 into account three collection mechanisms: impaction,  interception,
 and diffusion.  Their equation is:
                               72

-------
  Pt, = exp
              Kn Z        /d«\  7           D    Z
         -5.8 -2—  -3.75 (J£) -5.  - 3.752 —E	
               dc         VV dc          UT* V
                *~           C   C           (i    C
                                                           C21)
where   Ptd » penetration  for particle  diameter  "d  ",  fraction
         K  = inertial parameter,  dimensionless
         dc = collector  diameter,  cm
          A = bed depth, cm
         D  = particle diffusivity, cm2/s
         UG = superficial  gas velocity,  cm/s

     Model by Schmidt, et  al. -  Schmidt, et al.  (1978), used the
semi-empirical theories  of Johnstone  and Roberts  (1949) for
diffusion, Friedlander (1957) for  interception, Jackson and Cal-
vert (1966) for  inertial impaction, and Ranz  (1951)  for gravity
settling, and proposed the following  equation  for granular bed
collection efficiency.

   Ptd = exp  -7.5  (1-E) L-  (nD  +  nDI +  nz + nGS)l        (22)
                           c                      -I

where   Pt, = particle penetration for  particle diameter d ,
              fraction
         d  = collector  diameter,  cm
          Z = bed depth, cm
          e = bed porosity,  dimensionless
   nD » nCIt
            = single  granule collection efficiency  due to diffu-
              sion, direct interception, impaction,  and gravity
              settling,  respectively, fraction
   n
I ,  nGS
     Single granule collection efficiencies for various collection
mechanisms were obtained from:
                              2 038 N
                  n  = _8_ +        Re                   (23)
                   D   vt          -5/8
                       NPe     Npe/

                               73

-------
                                   d l2
                       n   = i 45 | _P_)                    (24)
                        DI   1'**\A
                                    c
                           = 3.97 Kp                      (25)

                           = ^                           (26)
                             UG
where   d  = granule diameter, ym
        d  = particle diameter, urn
        K  = impaction parameter, dimensionless
        u. = terminal settling velocity of particles, cm/s
        UG = gas velocity, cm/s
       Np  = Peclet number, dimensionless
       ND  = Reynolds number, dimensionless
        Ke

     Stage 2 Filtration - Stage 2 filtration has been observed
experimentally by several researchers.  Billings and Wilder
(1970) summarized the studies of many investigators concerned
with the cake formation process during the initial stage of
filtration.  They concluded that aerosol deposition occurs pri-
marily on previously deposited particles.
     Based on this observed dendrite-like growth, Payatakes
and Tien  (1976) proposed a model describing the dendrite growth
over the entire filtration period.  This model was expanded
and revised by Payatakes (1977).
     To express the growth process and to describe the dendrite
configuration, Payatakes (1977) divided the space adjacent to
the collector surface into layers of thickness "d " by planes
which are all parallel to a plane tangential to the collector
surface.  He numbered them in ascending order; i.e., the first
layer is immediately adjacent to the collector surface (Figure 24)
     The dendrite configuration is idealized with the convention
that if a particle of the dendrite structure has at least half
of its volume in the k'th layer it is assumed to lie entirely
                              74

-------
                             (a)  at  t  =  0
     (b) t = >0
in
                 A.

                 T
                                                                                       me =
                                                                                       m7 =
                         gm  ,

	        	m1 = i
                            (c)  Actual  dendrite
   (d)  Idealized  dendrite
                     Figure  24.   Dendrite  initiation, growth and  idealization  of  the  dendrite
                                 configurat ion.

-------
 in the k'th layer.   He also assumed that dendrite particles are
 of uniform size.
      By considering both the radial and angular contributions
 to deposition,  Payatakes (1977)  set,

            d m,     (0)      (0)     (r)       (r)
            -Ji= 9k-l,k + 0k,k + 0k-l,k  +0k,k
             dt

 where    m,  = expected particle number in the k'th layer of a
               dendrite, number
      0(0)   = rate of increase of mfc by deposition on particles
               occupying the (k-l)'st layer due to the angular
  flow component, number/s
=
             = rate of increase of m,  by deposition on particles
               already occupying the k'th layer, due to the
               angular flow component, number/s
       frl
      ®v i v = rate of increase of m,  by deposition on particles
       K .L ) K                        K
               occupying the (k-l)'st layer due to radial flow
               component, number/s
        £>y L = rate of increase of m,  by deposition on particles
         K , K                        K
               already occupying the k'th layer due to the radial
               flow component,   number/s
     In the  formulation, terms  involving addition of particles to
the kth layer by deposition onto particles occupying the (k+l)'st
layer are neglected.  The solution to equation  (27)  is»
   mk = f  Cki (9) exp [a bi (9) tJ' for k = 1, 2, 3, ...  (28)


where   mk = expected particle number in the kth layer of a
             dendrite
         t = time measured from the instant of deposition of
             the first particle of the dendrite, seconds
         a = rate of particles approaching a clean fiber per
             unit length = d  ur. n
                            C  Lil
                                76

-------
       0 • angular cylindrical coordinate, measured counter-

           clockwise from the down stream stagnation point

      d  » collector diameter, cm

     uGi = interstitial gas velocity,  cm/s

       n = particle number concentration, number/cm



"Cv- (Q)" and "b. (0)" are functions defined by the following
  & -L            J-
equations:



        C    =  1                                            (29)
        11


        c    =  __!^	                                     (30)
        2 1    fK, -K, 1
        C   = ——
               b,- b,
                 n      aj                                (32)

                 Jlk  b,-b3


                 ^i	     n     	^— , for k = 3,4..., (33)
                       2 
-------
where    p = maximum number of particles in the  (k+l)'st  layer
             which can be attached directly to the  same particle
             of the k'th layer
      4>?s-l = function defined so that "a 4>>S]M is the rate  of
       i > 3                                1»J
             increase of "m." by deposition on a particle in the
             i'th layer due to the flow component in the  "s"
             direction (s = r = radial direction, s =0=  angular
             direction).
        a  _ , (6)   +  , (r)
        ak ~ * k-l,k +  * k-l,k

      Equation (28) coupled with the assumed flow field (e.g.;
Happel's free surface model, Kuwabara flow field, etc.) can be
used to predict the increases in filtration efficiency and pres-
sure drop for the filter.  Payatakes (1976 a,b) presented some
sample calculations by applying the model to pure interception.
      Payatakes and lien's model described the rate of dendrite
growth.  It did not explain the reason for dendrite formation.
Wang  et al. (1977) and Tien  et al. (1977) proposed a concept,
the shadow effect, for dendrite growth.
      They hypothesized that once a particle is deposited,  it
creates a shadow area around itself on the collector surface,
within which no subsequent particle deposition may  take place.
This is represented by arc B'BB" for the deposited  particle "A"
shown in Figure 25.
      The creation of shadow areas by deposited particles has two
consequences.  Since there will be no deposition with any shadow
area, it means that particle collection takes place at a discrete
position along a collection surface, the deposited  dust cannot
be in the form of a smooth coating.
      The second consequence arises from the fact that with the
creation of shadow area, subsequent approaching particles which
would have deposited within the shadow area had there been  no
deposition, now attach themselves to the deposited  particle.  This
results in the formation and growth of chain-like particle
dendrites.
                               78

-------
                  Limiting Trajectory
                                                          Angle Covering shadow area
to
                          Figure 25.   Particle deposition on single collector

-------
      The magnitude of the shadow area is a function of many
variables including the location of its deposition, particle
diameter, and the inertial parameter.  They performed a simula-
tion by using this concept and the trajectories of random par-
ticles to verify this concept.
Gaseous Pollutants
      It is feasible to use various bed materials to remove metal
vapors and other gaseous pollutants.  The mechanisms expected
to be  important in removing gaseous pollutants with a granular
bed include:
      1. Adsorption
      2. Absorption
      3. Chemical reaction
      4. Condensation without nucleation
Adsorption  -
      The Chemical Engineers' Handbook gives a detailed treatment
of this subject.  The following is a brief discussion.  Adsorption
involves the interphase accumulation or concentration of sub-
stances at  a surface or interface. The process can occur at an
interface between any two phases, such as gas-solid, gas-liquid,
liquid-solid, or liquid-liquid interfaces.  The material being
concentrated or adsorbed is the adsorbate, the the adsorbing
phase is termed the adsorbent.
      There are three principal types of adsorption: electrical
attraction  of the solute to the adsorbent, van der Waals attrac-
tion and chemical adsorption.  Adsorption of the first type falls
within the  realm of ion-exchange and is often referred to as ex-
change adsorption.  Van der Waals attraction is generally termed
"physical"  adsorption, a term which has come to represent cases
in which the adsorbed molecule is not affixed to a specific
site at the surface but is free to undergo translational movement
within the  interface.  Adsorption of this type is sometimes
referred to also as "ideal" adsorption.  If the adsorbate under-
goes chemical interaction with the adsorbent, the phenomenon is
referred to as "chemical" adsorption, "activated" adsorption, or

                               80

-------
"chemisorption."  Chemically adsorbed molecules  are  considered
not to be free to move on the surface or within  the  interface.
      Performance of a solid sorbent depends upon four  factors.
      1. Stoichiometric capacity of the  solid. This  is  the ul-
timate capacity of the sorbent for the sorbate,  which may or may
not be fully utilized under actual process conditions.
      2. The phase equilibrium, which influences the efficiency
with which that capacity is reached, and in many cases  controls
the actual capacity of the surbent.
      3. The rate behavior including the mechanism and  resistance
controlling the mass transfer rate.
      4. The process arrangement and its effect  on the material
balance.
      Equilibrium Behavior - When the concentration of  the sorbate
remaining in the gas phase is in a dynamic equilibrium with that
at the surface of the solid, there is a definite distribution of
the sorbate between the gas and solid phases.  One form used to
depict this distribution is to express the quantity "q " as a
                                                      6
function of "c" at a fixed temperature.  The quantity "q " is
                                                        C
the amount of sorbate adsorbed per unit weight of sorbent and
"c" is the concentration of sorbate remaining in gas phase at
equilibrium.  An expression of this type is termed an adsorption
isotherm.  The adsorption isotherm is a functional expression
for the variation of adsorption with adsorbate concentration
in the gas at constant temperature.  Usually the amount of ad-
sorbed material per unit weight of adsorbent increases with
increasing concentration, but not in direct proportion  (Figure 26),
      Several types of isothermal adsorption relations may occur.
The most common relationship between "q " and "c" is obtained
for systems in which adsorption from the gas leads to the depo-
sition of an apparent single layer of sorbate molecules on the
surface of the solid.  Occasionally, multimolecular layers of
sorbate may be adsorbed.  The single layer adsorption can be
described by the Langmuir adsorption model and the multilayer
adsorption by the Brunauer, Emmett, Teller (BET) model.
                               81

-------
      a.   Langmuir
            c
            BET
Figure 26
Typical isotherms for
Langmuir and BET adsorption
patterns.
               82

-------
      The Langmuir treatment  is based on  the  assumption  that
maximum adsorption corresponds to  a  saturated monolayer  of  sor
bate molecules on the adsorbent surface,  that the  energy of ad-
sorption is constant, and  that there is no  transmigration of
adsorbate in the plane of  the surface.  Figure  26  is  a typical
isotherm for the Langmuir  pattern.   The Langmuir isotherm is:
                         qe -      -                      (36)
                          e    1  +  Kc
where   q  = the amount of  sorbate  adsorbed per unit weight of
         C
             adsorbent in equilibrium with concentration "c"> mol/g
        Q° = number  of moles of  adsorbate adsorbed per unit
             weight  of adsorbent in forming a complete monolayer
             on the  surface > mol/g
         K = equilibrium constant>  £/mol
         c = concentration  or partial pressure of adsorbate
             in gas  phase,  mol/fc or mm  Hg

      The BET model   assumes that a number of layers of adsorbate
molecules form at  the surface and that  the Langmuir equation
applies to each layer.  A further assumption of the BET model
is that a given layer need  not complete formation prior to the
initiation of subsequent layers.  For adsorption from the gas
phase with the additional assumption that layers beyond the
first have equal energies of adsorption, the BET equation takes
the  simplified form;
                      (c  -c)  [1  +  (B-l)(c/c  )]
                       s                  =
                                83

-------
where   c  = saturation concentration of the adsorbate, mol/g
         c = measured concentration in gas at equilibrium, mol/4
         B = a constant expressive of the energy of interaction
             with the surface, 2,/mol
        Q° = number of moles of adsorbate adsorbed per unit
             weight of adsorbent in forming a complete monolayer
             on the surface, mol/g
        q  = number of moles of adsorbate adsorbed per unit
             weight at concentration "c", mol/g

      One other question for isothermal adsorption, the Freundlich
or van Bemmelen equation, has been widely used for many years.
This equation is a special case for heterogeneous surface ener-
gies in which the energy term, "b", in the Langmuir equation
varies as a function of surface coverage, "q ", strictly due to
                                            "
variations in heat of adsorption.  The Freundlich equation  has
the general form:
                       = KF c
                              1/n
where  "Kp" and "n" are constants and n > 1.

      Rate Processes - There are essentially three consecutive
steps in the adsorption of materials from gas by porous adsorbents,
any one of which may be a rate determining step in a certain re-
gion of operating conditions.
      The first step is the transport of the adsorbate through
a surface film to the exterior of the adsorbent.  The transport
rate for adsorbate between the bulk  of the gas phase and the
outer surfaces of the sorbent granules is governed by the mole-
cular diffusivity and, in turbulent flow, by the eddy diffusivity
which controls the effective thickness of the boundary layer.
                               84

-------
One may assume that the concentration of sorbate at the point
in the hydrodynamic boundary layer immediately adjacent to the
external surface of the particle is in equilibrium with the average
solid-phase concentration on the internal surfaces.  This condition
may be stated algebraically as:
                   - kG a £-  (c  - ce)                    (39)

where   kg « mass-transfer coefficient, cm2/s
         a = effective area for mass transfer across the fluid
             film per unit volume of bed, cm2/cm3
         e « void fraction, fraction
        p  = density of the solid in the bed, g/cm3
         c « concentration of the sorbate in bulk gas phase,
             g/cm3
        c  = concentration of sorbate in bulk gas phase in
             equilibrium with the coexisting solid phase
             concentration, g/cm3
         q = concentration of sorbate on solid surface, g/cm3

      For packed bed, Wilke and Hougen  (1945) gave the following
equation for evaluating the mass-transfer coefficient:
                                     0.51      0.16
                10.9 ur U-E)  /  Dr \/Dr Pp\
         k  a = 	*	  ( 	—M-5—-)           (4°)
                     dc        ^dc UG/V »G   '

where   UG = superficial gas velocity,  cm/s
         e = void fraction ,  fraction
        d  = granule diameter, cm
        DG = gas phase diffusivity, cm2/s
        PG = gas phase density, g/cm3
           = gas phase viscosity, g/cm-s
                               85

-------
     The second of the  three  consecutive  steps  in  sorption by
porous adsorbents, with the exception  of  a  small amount of
adsorption that occurs  on the exterior surface  of  the adsorbate
after transport across  the exterior  film, is  the diffusion of
the adsorbate within the pores of  the  adsorbent.   The driving
force approximation for pore  diffusion is:
                                      - q)               (41)
where  k     = pore diffusion  coefficient, cm /s
           a = outer surface area  of particles per unit bed
               volume,  cm2 /cm
           5 = intraparticle void  ratio, dimensionless
           q = concentration of  sorbate on solid surface, g/cm3
          q  = local concentration of sorbate in the solid phase
           G
               that prevails at  the outer surface and is assumed
               to be in equilibrium with the coexisting gas phase
               at concentration  "c", g/cm3
     According to Vermenlen  and  Quilici  (1970) :

                          60
                                                       <42)
where   D ____  =  pore  diffusity, cm2/s
         po i c
           d  =  solid diameter, cm
            e  =  bed void  fraction, dimensionless
Pore diffusivity can  be expressed as:

                   = i  F-L-  /-^-IL\   + L.V
                      x  [4?  \2 RT;     DJ
                              86
              D

-------
where   D  _  « pore  diffusivity,  cm2/s
         pore
            5 = internal  void  fraction of the solid,  fraction
            r = average pore radius,  cm
            X = tortuosity factor  (usually between 2  and 6),
                dimensionless
            R = universal gas  constant,  J/gmol-°K
            T = absolute  temperature, °K
            M - molar weight,  g
           Dp = gas phase diffusivity, cm2/s

     The third and final  step  is the  adsorption of the adsorbate
on the interior surfaces  of  the adsorbent  (e.g. porous granules)
The following reaction schematic for  the  adsorption process is
considered:
     (sorbate)  + (sorbent)   -*•   (sorbate  • sorbent)
     If a Langmuir-type adsorption equilibrium is assumed, a
general expression for the rate of adsorption at the solid
surfaces may be given as:
 d  (sorbate-sorbent)  _
           	     (sorbate) (sorbent)  -
           dt
                                                (sorbate.sorbent)
                                                         (44)
 where  K = Langmuir equilibrium constant, A/mol
        k - rate constant for second-order surface reaction
            controlled kinetics, A/mol-s
      If the ultimate monolayer capacity of the adsorbent for  the
 adsorbate is designated by the term Q°, then (Q°-q)  represents
 the  unused capacity of the adsorbent.   Substitution  into
 equation (44")  yields:
                     If - k [c CQ°-q) - |]
                                87

-------
     The gas-solid heterogeneous reaction may be either a non-
catalytic or a catalytic reaction.  In this report, we will con-
sider the heterogeneous, noncatalytic reaction.  The treatment
of this type of reaction requires the consideration of mass trans-
fer between phases, the contacting patterns of the reacting phase,
and the reaction kinetics.  Levenspiel (1972) and the Chemical
Engineers' Handbook  present a detailed treatment of this sub-
ject.  The following is a brief summary of those discussions.
     A heterogeneous non-catalytic reaction may be represented
by:
           A (gas) + b B (solid) •* Products

The products may be gas, solid or both.  There are several models
available to describe the progress of the above reaction.  The
unreacted core model seems to work reasonably well in a wide
variety of situations.  This model considers that the reaction
occurs first at the outer skin of the solid.  It can leave behind
both completely reacted material, and inert solid material residue.
Thus, at any time there exists an unreacted core of material which
shrinks in size during reaction.  Two different cases may be
considered for this model.  The first assumes that the continuous
formation of solid product (inert residue) without its flaking
off would maintain the particle size unchanged.  In the second
case the particle size changes as the reaction progresses owing
to the formation of gaseous products, flaking off of the solids,
etc.
Unreacted Core Model for Spherical Particles of Unchanging Size
     This model visualizes the reaction occurring in three
successive steps.  Either one of the three steps may be rate
determining:
Step 1;  Diffusion of gaseous reactant "A" through the film sur-
rounding the solid to the exterior surface of the solid.
Step 2:  Penetration and diffusion of the reactant "A" through
the blanket of residue to the surface of the unreacted core.

-------
Step 3: Reaction of gaseous reactant with  solid at  the reaction
surface.
     If the resistance of  the  gas  film is  the  controlling  factor,
the reaction rate  is equal  to  the  diffusion rate  of the  gas  reac-
tant from the bulk phase to the  interface.   The chemical reaction
can be assumed to  be instantaneous.  Thus,  the concentration of
the gas at the solid surface  is  zero and  the concentration driving
force  for diffusion is constant  at all times during the  reaction
of the particle.   In terms  of  the  shrinking radius  of  the  unreac-
ted core, the reaction rate is:
                                    - b  kr  c              (45)
                                dt       G
 where    p   =  density of reactant "B",  g/cm3
          o
         r   =  radius  of unreacted core, cm
          R  «  original radius  of the reacting particle, cm
          b  =  stoichiometry constant,  dimensionless
         kg  =  mass  transfer coefficient,  cm2/s
          c  =  concentration of gas reactant "A"  in  the bulk  gas
              phase,  g/cm3
          t  =  time, s

     Rearranging and integrating, we  find  how the  unreacted
 core shrinks  with  time,

               t =    Ps R	  I i - i l£l  1                (-46)
                   3b kr. c
 Let the  time for complete reaction of a particle be "T".
 Then by  taking "r =0" in equation (46), we find,
                             P
                      T = 	
                          3b k
*	                      (47)
                                89

-------
The radius of the unreacted core expressed in  terms  of  fractional
time for complete conversion is obtained by combining equations:
                                                          (48)
                        ^R
where  XD = fraction of reactant "B" converted  into  product
        D
     By using the same approach, the integrated rate equations
under other conditions are:
     Diffusion through residue layer c<  itrols:


               ^ = 1-3 (l-xB)2/3 + 2 (l-xB)               (49)


                     PS R2
               T = 	5	                             (50)
                   6 b D  c
                        e
where   D  = effective diffusion coefficient of gaseous  reactant
             in the residue layer, cm2/s

     Chemical reaction controls:
              t             13
              - = 1-  d-xR)                              (51)
              T          B
                    P  R
                                                         (52)
                  b k
where   k  = first order rate constant for  the surface  reaction.

Unreacted Core Model for Shrinking Spherical  Particles
      When no adherent  residue forms,  the  reacting solid particle
shrinks during reaction,   finally disappearing.   For a  reaction
of  this kind we visualize  the following three steps  occurring  in
succession.
Step  1: Diffusion of reactant "A" from the  main  body of gas
through the gas film to the  surface  of the  solid.

                               90

-------
Step 2: Reaction on the surface between reactant  "A"  and  solid.
Step 3: Diffusion of reaction products from the surface of  the
solid through the gas film back into the main body of gas.
     As with constant size particles,  the  following  rate  ex-
pressions result when one or the other of  the resistances control
     Diffusion through gas film controls:
                 Small particle (Stokes regime)

                   £ = 1- d-xB)2/3                      (53)
                   T
                   T .
                       2 b DG c
where   y. = mole fraction of reactant  "A" in gas phase
        DG = gas diffusivity, cm2/s

                 Large particle  (UG = constant)

                   £ = 1-  (l-xB)°<5
                                    a/a
                   T = (constant) - —
     Chemical reaction  controls:  When  chemical reaction controls,
 the behavior is  identical  to  that of particles of unchanging size;
 therefore, equations (51) and  (52) will represent the time-conver-
 sion behavior of single particles, both shrinking and of constant
 size. With this  information on  reaction kinetics, we can determine
 the required granular bed  size  for various  gas-solid contacting
 schemes  (fixed bed batch process, continuous moving bed, etc.).
                                91

-------
Cake Filtration
     Granular bed filters usually have larger pore sizes and greater
thickness than the fibrous filter.  Whether internal cakes will
form depends largely on the granule size.  If the granules of the
bed are large, dendrites will not bridge to create an internal
cake.  On the other hand, if the grains are sufficiently small,
dendrites will bridge to form an internal cake.
     Lee (1975) called the internal cake a rooting cake and it
is the foundation to support the formation of a surface cake.
Once a surface cake is formed, filtration efficiency no longer
depends upon the depth of the granular bed but rather on the
thickness and structure of the surface cake.   The cake filtration
results in a much higher efficiency than the  original granular
bed.  This is because particle collection by sieving becomes
a more important collection mechanism.
     Leith, et al.  (1976) and Leith and First  (1977) studied
high velocity cake filtration of fabric filters.  Three mechanisms
were described by them by which particles can pass through a
fabric filter or a granular bed.  The three mechanisms are:
     1.  Straight through penetration
     2.  Seepage or bleeding penetration
     3.  Pinhole plug penetration
     In straight through penetration, particles pass through the
filter without stopping; i.e., they are not collected by the
filter.  Once a particle lands on or in the filter, it needs not
necessarily remain at its point of initial impact.  As the dust
deposit builds up, the dust may work its way through from the
dirty to the clean side of the filter. This may result from the
drag force exerted on the particle deposits by the gas moving
past.  Penetration of this sort is called seepage or bleeding.
The pinhole plug mechanism postulates a plug of deposited par-
ticles dislodged from the dust deposit and moves out of it,
leaving behind a pinhole.  Figure 27 is a schematic representa-
tion of penetration mechanisms.
     The size distribution of particles passing through the  fil-
ter  by the  straight through mechanism  should reflect a dependence
                                92

-------
 I
 I

 \
 I
6
 i
 t
O
 I
 I
       i
 I
O

                              ii
                                        Straight
                                        Through
                                       Seepage
                                       Pinhole
                                       Plugs
Figure 27. Schematic representation of fly ash emission
           mechanisms.
                          93

-------
on the forces causing particles to be collected there:  inertia,
interception, diffusion, gravity, etc.  However, the size distri-
bution of the particles which pass through by seepage or pin-
hole plugs should be the same as the size distribution of the
deposited dust;  that is, very close to the size distribution of
dust fed to the filter.
     Using a series of tagged dusts, the proportion of total
dust emitted which is accountable to each emission mechanism
was measured by Leith et al .  (1976) in relation to face velocity and
deposit thickness.  Deposit thickness is defined as:
                        W       C ,t
                              =  P1
                     pp (1-e)   Pp   -e

where   X = dust deposit thickness, um or cm
       p  = particle density, g/cm3
        W = particulate load, g/cm2
      C . = inlet particle concentration, g/cm3
       Up = superficial gas velocity, cm/ s
        t = time since last cleaning,  s
     They found significant trends in the dust penetration mech-
anism data.  The time-mechanism interaction was highly significant.
As time increases and the dust deposit thickens, the mechanisms
by which dust penetrates the filter change.  Straight through
penetration rapidly diminishes in importance although it is
important immediately after a filter cleaning cycle.  The emitted
dust accountable to the seepage mechanisms is relatively constant
during the entire filtration cycle.  The pinhole plug mechanism
rapidly rises in importance after cleaning, passes through a maxi-
mum, and then declines as the dust deposit becomes thicker and
pressure drop through the deposit increases.  Figure 28 is an
illustration of the trends.
     The velocity mechanism interaction was not significant.  At
any fixed time, the fraction of dust emitted by each penetration
mechanism is fairly constant at all velocities tested.
                               94

-------
       1.0
   O
   •H
   +J
   U
o
   CO
   CO
   CO
   ^5
   Q
   H
   u
       0.8
   o   0.6
       0.4
       0.2
                  Pinhole
                  Plugs
                       Straight
                       Through
                                          I
                      20           40

                       DEPOSIT THICKNESS, ym
                                            60
Figure 28.  Fraction of total fly ash emitted by various
            mechanisms as a function of deposit thickness.
                            95

-------
     Miyamoto and Bohn (1975) studied the effect of particle
loading on granular bed filter collection efficiency.  Their
results were presented in Figures 13, 14 and 15 in an earlier  sec-
tion of this report.
     Under contract with EPRI, Squires and his co-workers  at City
College of New York have experimentally determined the collection
efficiency of a granular bed with filter cake.  A final  report
has recently been published  (Lee  et al., 1977).
Pressure Drop
Clean Bed -
     Several investigators have shown that the flow through packed
beds can be described by:      * 7 „  n  ^
                               I L Uf-i L -1- ~ E J Pr1
                         -AP =	—           (58)
                                   dc  *'
where   AP = pressure drop across packed bed, g/cm-s
         f = friction factor, dimensionless
        UG = superficial gas velocity, cm/s
        PG = gas density, g/cm3
         e = bed porosity, fraction
        d  = bed particle diameter, cm
    Figure  29  is a plot of  friction factor versus Reynolds num-
ber for a fixed bed.  Ergun  (1952) has defined a Reynolds number
                             d  u,, p.-,
                      NRe ' f-^f                     <59>

     For laminar flow with NR < 1.0; and by analogy to flow in
many other systems, we can approximate "f" by a constant divided
by "NR  ."  An analysis of experimental data indicates that the
constant is 150.  Therefore, for  laminar  flow we have:

                          f  = P                        (60)

     This is referred to as  the Kozeny-Carman equation.  For a
given bed and fluid, it predicts that the flow rate is proportional
to the pressure drop, which  is D'Arcys law.

                               96

-------
<
 I
  a;
  o
  H
  U
  H
  U
                      I  I  I I I Illl   I  I
                                        IMI
                               BURKE    PLUMMER
                  10
20
30
40
                         N
                          Re

                          1-e


 Figure  29.  A comprehensive plot  of  pressure  drop
             in fixed beds.
                       97

-------
     For completely turbulent flow, it is reasoned that "£"
should approach a constant value and that all packed beds have
the same relative roughness.  The constant is found by experiment
to be 1.75, so we have:
                                  (-AP) d  e
                      £ = 1.75 = - - -         (61)
                                 Z UG* PG (1-0

This is called the Burke-Plummer equation.
     A consideration of flow at intermediate Reynolds numbers
led Ergun  (1952) to propose as a general equation:
                      £ = HO + 1>?5                     (62)
                          NRe
 Filter Cake  Resistance  to  Gas  Flow  -
     Miyamoto and  Bohn  (1975) studied  the  effect  of particulate
 load on pressure  drop.   Figures (30) through (32) show their results.
 The  pressure drop remained constant until the  particulate load
 of  the filters  reached  a threshold  load,  then  increased rapidly
 as  shown in  the figures.
     In granular bed  filters, the  flow resistance should change
 little so  long  as the large pores are open, but  will  increase
 when the large  pores are closed by  surface cake.   Depending  on
 whether the  compaction  effect  of  the  filter cake is present,
 the  increase of pressure drop  is  at a different  rate.
     No compaction effect - The pressure drop  across the surface
 cake is given by  D'Arcy's  law, i.e.,
                               G                         (63)
                              Kd
                                98

-------
u
  3

  SJ
 2
 O
 h- 1
 H
 U
 WH
                I I ill   I  I  I I I llll
               rm   i  r
i  i  i t ii ni
      BURKE § PLUMMER
i i ii ni	i  i  i 111 in   i  i
                 10
    20
                               30      40
                         N
                          Re

                         1-e

Figure  29. A  comprehensive plot  of  pressure drop
            in fixed beds.
                       97

-------
     For completely turbulent flow, it is reasoned that "£"
should approach a constant value and that all packed beds have
the same relative roughness.   The constant is found by experiment
to be 1.75, so we have:
                      f = 1.75 =
This is called the Burke -Plummer equation.
     A consideration of flow at intermediate Reynolds numbers
led Ergun (1952) to propose as a general equation:
                      £ . 15° + 1.75
                          NRe
Filter Cake  Resistance  to Gas  Flow  -
    Miyamoto  and Bohn  (1975) studied  the  effect  of particulate
load on pressure drop.  Figures (30) through (32) show their  results.
The pressure  drop remained constant until the  particulate  load
of the filters reached  a threshold  load,  then  increased rapidly
as shown in  the figures.
    In granular bed filters, the flow resistance should change
little so long as the large pores are open, but will  increase
when the large pores are closed by  surface cake.  Depending on
whether the compaction  effect  of the  filter cake is present,
the increase  of pressure drop  is at a different rate.
    No compaction effect  - The pressure  drop  across  the surface
cake is given by D'Arcy's law, i.e.,
                    - 4P .  GG                         (63)
                              Kd
                                98

-------
ft.
<
 g
Cu
<
      80
      60
      40
      20
                           0-1                0.2
               PARTICULATE LOAD, kg/mz
           Figure  30.  Effect of mean granule
                       diameter on pressure drop
                       (Miyamoto and Bohn's data),
     80
     60
     40
     20
                Z-10 cm
30
                                    » 2 cm/s

                                    « 3 mm
                                         70  90
         0                 0.1                 0.2
               PARTICULATE LOAD, kg/m2
           Figure 31.   Effect of bed depth on
                       pressure drop (Miyamoto and
                       Bohn's data).
     80
     60   -
     40   -
     20   -
         0                  0.1
             PARTICULATE  LOAD,  kg/ma

           Figure  32.  Effect of superficial  gas
                        velocity on pressure drop
                        (Miyamoto and Bohn's  data),
                         99

-------
where  Ap = pressure drop across packed bed,  g/cm-.s2
       yG = gas viscosity, g/cm-s
       u,, = superficial gas velocity, cm/s
        Z = bed depth, cm
       K, = D'Arcy permeability, cm2
     For laminar flow, it becomes

                       150 Z u^  (1-e) pc
where    e = bed porosity, fraction
        PG = gas density, g/cm3
        d  = granule diameter, cm
       NR  = Reynolds number ,  dimensionless
This is referred to as the Kozeny-Carman equation.

     By combining equations  (63) and  (64) we obtain

                       150 y  Z u  (1-e)2

                                                          C65)
     Substituting equation (57XX = Z) into equation (65) gives:

                       150 y  C  -t u2  (l-e)
                                                          C66)
where  Ap = pressure drop across filter cake,  g/cm-s2
       V~ = gas viscosity,  g/cm-s
        t = time since last cleaning, s
       UG = superficial gas velocity, cm/s
        e = filter cake porosity, fraction
       d  = granule diameter, cm
        \*
       p  = particle density, g/cm3
      C • = inlet particulate concentration, g/cm3
                                100

-------
     Equation (66) predicts that at constant  inlet  conditions,  the
pressure drop across the surface cake varies  linearly with  time
as long as the porosity of the filter cake  remains  constant  (no
compaction effect present).  It also predicts  a linear relation-
ship between pressure drop and the  square of  the  superficial
gas velocity, after the same time period, for  equal  loadings  and
bulk density of  the filter cake.
Compaction Effects - Leith et al  (1976) also  have studied the
compaction of filter cake and its implications. They found that
Kozeny-Carman's  equation yields excellent results for powders
which are compressed to a specified porosity provided the particles
are isometric.
     Compaction  of the cake is a complicated phenomenon which is
related to the increased drag exerted by the gas on the cake.
Although the dust on the surface of the cake is under negligible
mechanical stress due to this drag, the dust below the surface
must support the drag experienced by the dust  layer above it.  At
the bottom of the cake, the compressive stress is the greatest
and equals the total pressure drop  across the  cake, plus the stress
due to the  impact of new particles.
     Limited studies have been done to determine the effects of
increased velocities on permeability.  Stephan et al. (1960)
observed a 60% change in this factor when velocity was increased
from 1.8 to 3.0  m/min with clean air, and the  change was noted
to take place in 5 discrete collapses.  The cake was initially
deposited at 1 m/min  (3.4 ft/min).  Borguardi et al  (1968)
reported a similar change with the  same dust  (fly ash).  These
studies provide  some insight into cake collapse.  However, if
the cake is formed at high velocity, and the velocity remains
constant, the compaction of the cake may proceed more gradually
due to tighter initial packing.
     Orr (1966)  indicated that compaction of cake can take place
by three mechanisms:
     1.  sliding,
     2.  elastic and plastic deformation,
and  3.  fragmentation.
                                101

-------
The latter is unlikely except for extremely fragile particles
due to the relatively small compressive stresses.  The most  likely
mechanism is sliding which is opposed by frictional and  cohesive
forces between particles.   Compaction effects are often
described by the following exponential expression:

              e = ei exp  (-a F)                          (67)

where   e = final void fraction, fraction
       e. = initial void fraction, fraction
        a = constant, cm-s2/g
        F = compression stress, g/cm-s2
Differentiating equation (67) gives:

              3F - - a e                                 (68)

Differentiating equation (57)  yields:
              dx -         + *_ +    e                   (69)
                        P    P    (
D'Arcy's equation is

              dp
Combining equations (69)  and (70)  we  obtain:

                 K             dW     W     de
                      dp =
From the Kozeny-Carman equation
              K  =
               d   150
                                                         (70)
                               102

-------
Combining equations (71) and (72) yields
              dp = -
                     150 »G U
                      dc2  pp
For a small increment of cake
de
(73)
              dF = - dP

Combining equations

              dW _ k e2 + W
              HT ~   (e-1)

where  k = constant, g/cm2
             d 2 p
              c  "p
           150 a y~ UG
Solving equation  (75)  with boundary condition]
              e =  ei at W  « 0, yields
W
  = e2-2e-2(l-e)£n(l-e)-(l-e)
                                2ei-ei2
                                                         (74)
                                                         (75)
                                                              (76)
Since the total compressive  stress  on  the  cake  is equal to "p",
equation  (67) can be  rewritten  as,
              e -  e^^ exp  [  -a  (Ap)]

Substituting equation  (76)  into  equation  (77) yields,
                                                         (77)
J
                    .  2
                      [
                                                         (78)
                                103

-------
Since         W = C  .ur t
                   pi G
We have
                          -2e. e        -2
in
                                                  1-e.
                                                     1     (80)
A plot of pressure drop versus time for cake buildup can be a
curve fitted to equation (80) and the constants  a,  e^, and k
evaluated by regression analysis.
                                104

-------
                 TABLE 14. AVAILABLE  EQUATIONS  FOR THE  PREDICTION OF
                           PARTICLE COLLECTION  IN A GRANULAR BED
Investigator
Equation
                                                                     Notes
Jackson and Calvert
(1968)
                        Pt, = exp

                                                                 Impact ion only.
Paretsky et al.
(1971)
                        Ptd = exp
                                      2  e
Miyamoto and Bohn
(1974)
                              exp
                                            Z NNu
                                       dc
                            Collection by
                            diffusion only
Gebhart  et al.
(1973)
                        Ptd = exp
      D 2/8
-6.39 -2	r
                            Collection by
                            diffusion only
 Bbhm  and  Jordan
 (1976)
                        Pt, = exp
                          d
                                                d
/2 k T e f +  ^> 8

\37T d  U
                   e d
                                                     36
                             . T V PP UG
                                   18e
     -)
                                             4  £' Z
                             Collection by
                             diffusion.
                             Gravity settling
                             and impact ion.
                                                                  Continued

-------
                               TABLE  14.   (continued)
           Investigator
       Equation
                                               Notes
           Goren
           (1977)
Pt  = exp
 -   (1-e)  -


O.75
l250 K
                                2'25
Collection by impaction,
gravity settling, and
diffusion
o
00
      Westinghouse
       (Ciliberti, 1977)
Pt  = exp
f
 - 5.
                                              K  Z
                                 -  3.75
                               3.572
            Collection by impaction,
            interception, and diffusion
     Schmidt, et al .
     (1978)
Pt,, =
                                         7.5 Z
                                * 2.038

                                        (d \
                                        -£)
                                         c/
                           Collection by diffusion,
                           interception, impaction,
                           and  gravity settling
                +  3.97  K  +
                  u  1
                  -^

                  U-I

-------
and»           Ptd "
where   n = single granule collection efficiency, fraction
      Pt^ = particle penetration,  fraction
and subscripts D, DI, I and GS refer to diffusion, direct inter-
            ception, and gravity settling, respectively

     As can be seen from Figure 33 for this particular granular
bed filter, their equation predicts that the collection efficiency
will be very low.  This is not in  agreement with McCain's data.
This discrepancy may result from Paretsky  et al . basing their
equation on collection by an isolated sphere.  At a gas velocity
of 80 cm/s  (superficial  gas velocity in the granular bed during Mc-
Cain's tests), the dominant collection mechanism is inertial im-
paction.  For particle collection by inertial impaction onto a
single spherical collector, it may be generally assumed that there
will be no collection if the inertial impaction parameter is below
a critical value.  The critical value of impaction parameter is
about 0.083 for an isolated spherical collector.  For the test
conditions in McCain's tests, this is equivalent to the impaction
parameter of a 10 umA diameter particle.  Therefore, for particles
with diameters smaller than 10 ymA, there should be no collection
by impaction.
     Bohm and Jordan (1976) derived their equation by visualizing
the granular bed as a system of parallel capillaries. The ratio
of granule diameter to initial capillary diameter, "f", changes
during the filtration period due to accumulation of collected
particles.  According to Bohm and Jordan,

                         6.5 < f  < 10
                               109

-------
     The predicted particle penetration for f = 6.5 and  f1  =  10
is   plotted in Figure 33.  By assuming there was no surface  cake
and the pressure drop across the bed was 80% of the overall  pres-
sure drop, the bed porosity was estimated to be 0.25 (by  Ergun's
equation).  With this value for bed porosity, Bohm and Jordan's
equation predicted too low a particle penetration.  If we assume
f  = 0.75 the prediction will match the data.  However, this
assumption is unrealistic because for f1 < 1, the capillary  dia-
meter will be greater than the granule diameter.
     The predicted penetration based on Goren's model for e  =  0.25
is higher than measured.
     As mentioned earlier, the dominant collection mechanism was
inertial impaction for the operating conditions of the filter
during McCain's tests.  Both Miyamoto and Bohm's equation and
Gebhart  et al.'s equation are for particle collection by diffu-
sion only.  Therefore, these two equations are not suitable  for
comparing with McCain's data.
     For collection by impaction, Westinghouse's model and Schmidt
et  al.'s model reduce to:
                                      7
       Westinghouse:  Pt, = exp (-5.8 — K )
                        a             d   p
       Schmidt  et al.: Pt, = exp [-29.85 (1-e) — K ]
                          a                     d   p
                                                 c
     Except for the constant, these equations are identical  to
that of Jackson and Calvert (1968) .  Predictions by these equa-
tions are compared with McCain's data in Figure 33 .  Bed  porosity
is  assumed to be 0.25.  As can be seen, Westinghouse's model
slightly overestimated the penetration and Schmidt  et al.'s
model underestimated particle penetration.

DATA REPORTED BY HOOD
     Hood  (1976) reported the evaluation of the Combustion Power
Company's moving gravel bed filter on the control of particulate
emissions from a hog-fuel fired boiler.  The gravel bed filter
                               110

-------
was a prototype unit with suggested  capacity  of  1,133 Am3/min
(40,000 ACFM).  The bed was packed with  an  intermediate  size
gravel which was retained on a  3.2 mm  (1/8  in.)  wire mesh  and
passed a 6.4 mm (1/4 in.) mesh  screen.   The bed  was a single down-
flowing annulus 2.6 m  (8.5 ft)  O.D.  and  1.8m(6 ft)  I.D.
     During sampling the unit was operated  at a  flow rate  of
1,558 m3/min (55,000 ACFM).  The gas temperature was 177°C  (350°F)
     Particle size distribution and  concentration were sampled
with cascade impactors.  Particle penetration was calculated from
the cascade impactor data.
     Figure 34 shows the comparison  between Hood's results and
predictions by available design equations.  The  bed porosity was
calculated to be 0.25  (from Ergun's  equation).   As can be seen
from Figure  34, none of the available  design equations agrees
with the measured performance.
     Under the sponsorship of ERDA,  Combustion Power Company
conducted experimental studies  on their  GBF system to correlate
the collection efficiency of the GBF with mechanical and process
parameters.  Parameters studied included superficial gas velocity,
dust loading, particle size and distribution, granule diameter,
granule circulation rate, and bed thickness.
     The GBF was a pilot unit and was  operated at ambient tempera-
ture.  Redispersed hydrated alumina  was  used as  test dust.
     A.P.T. has acquired some data with  cascade  impactors on this
GBF system.  Figures 35 through 37 show  part of  the data along
with predictions by available design equations.  The model pre-
dictions do not agree with the  data.

DATA BY KNETTIG AND BEECKMANS
     As mentioned earlier, all  of the  available  equations are
for the prediction of particle  collection by  clean beds.  The
granular bed filter data reported by McCain and  by Hood  are data
obtained on  industrial installations.  The  beds  will not be clean.
In the following sections, data obtained on laboratory scale clean
granular bed filters will be used to test the design equations.
                                Ill

-------
   1.0
   0.5
   0.1
  0.05
  0.01
           HOOD'
           DATA
          A. JACKSON & CALVERT
          B. PARETSKY, ET AL.
          C. BOHM & JORDAN, f'«10
                                      BO'HN  ft  JORDAN,  f "6.5
                                   E.  GOREN,  e= 0.25
                                   F.  WESTIN6HOUSE
                                   G.  SCHMIDT, e= 0.25
       0.1
                        0.5        1.0                   5
                     AERODYNAMIC  PARTICLE  DIAMETER,  umA
                                                                   10
  Figure 34.  Comparison of Hood's data with predictions by available
              design equations.
    1.0
    0.5
o
3
u.
 .   0.1
S
£  0.05
   0.01
                                         GOREN,  E=  0.35
                     WESTINGHOUSE
          UQ = 81 cm/s
          d  » 0.2 cm
           Z = 20.2 cm
          GRANULE CIRCULATION
          RATE = 0.8 kg/kg air
                                                  \
                                   SCHMIDT,  ET AL.  \
                                   e= 0.35
                                                    \
                                                     \
                                                         DATA
       0.1
                                                           1  i I
                                                                   10
                         0.5      1.0                  5
                       AERODYNAMIC PARTICLE DIAMETER, umA
Figure 35.   Experimental  and predicted performance of CPC GBF (A.P.T. data).
                                      112

-------
     1.0
     0.5
"•   o.i
i   '
1
£  0.05
    0.01
                WESTINGHOUSE
                 SCHMIDT,  ET AL
                     e = 0.35
                                               GOREN, e= 0.35
                                                 GRANULE CIRCULATION
                                                  X       RATE =
                                                  "X    1.6 kg/kg
                                                      xx       air
                                                         N.

                                                     A.P.T.  DATA
            u. = 41  cm
            d  = 0.2 cm
             Z - 20.2 cm
                                                   \
                                GRANULE CIRCULATION  \
                                  RATE = 0.8 kg/kg air
        0.1                   0.5      1.0                  5        10

                      AERODYNAMIC PARTICLE DIAMETER,  umA

   Figure 36.   Experimental  and predicted performance of CPC GBF  (A.P.T. data).
     1.0
     0.5
o
5
S
     o.i
    0.05
    0.01
                                           GOREN, e= 0.35
                      WESTINGHOUSE
                  SCHMIDT, ET AL.

                   e = 0.35
            u. = 41  cm/s
            d  = 0.2 cm
             Z = 40.6 cm
        0.1
                                                                    10
                         0.5      1.0                  5
                 AERODYNAMIC PARTICLE DIAMETER, umA

Figure 37.  Experimental and predicted performance of CPC GBF (A.P.T.  data).
                                     113

-------
     Knettig  and Beeckmans  (1974)  studied the capture of mono-
 disperse  aerosol particles  in the size range of 0.8-2.9 urn in
 a screen  supported and in  a grid  supported fixed bed of 425 urn
 glass  beads.   Test results  showed a linear relationship between
 collection efficiency, expressed  in transfer units, and bed
 height.   Impaction appears  to have been the primary collection
 mechanism because collection efficiency increased with both
 superficial  gas velocity and aerosol particle size.  Transfer
 units  are related to penetration  by:

                      NTU =  - In Ptd                     (81)

     In terms of number of  transfer units,  various design equa-
 tions  become:

             Jackson and Calvert:     ^- =  - 2.         (82)
                                      Z     d
             Paretsky et al . :       »   (i-e)  IL_          (83)
                                Z    2       d_
            Goren:          = 1 (i-e)   Jl                 (84)
                        Z    2        dc
Bohm and Jordan:
                                2  TT  f •    Kp              (85)
                                  9  e    d
         Westinghouse:       =2.9-2-                     (86)
                         Z        d
                                   c

       Schmidt,  et al . :  — = 29.85 (1-e)  -2-             (87)
     Table 15 compares data with predictions.  None of the pre-
dictions agree with the data.   Jackson and Calvert's equation
(1965)  and Bohm and Jordan's (1976)  equations overpredict effi-
ciency.  Paretsky  et al.'s equation predicts no particle
                               114

-------
TABLE 15.  COMPARISON OF KNETTIG AND BEECKMAN'S DATA AND PREDICTIONS
Support Superficial
Gas Velocity
cm/s


Screen
Support




Grid
Support



8.2


11.2


8.2


11.2

Particle
Diameter
ym
0.8
1.6
2.9
0.8
1.6
2.9
0.8
1.6
1.9
0.8
1.6
2.9
(T)
experiment
0.05
0.076
0.27
0.054
0.08
0.306
0.042
0.058
0.336
0.044
0.074
0.362

Jackson §
Calvert
Ci = 10
0.28
1.03
3.23
0.38
1.4
4.4
0.28
1.03
3.23
0.38
1.4
4.4
(NTU/Z) Predicted
Paretsky
0
0
0
0
0
0
0
0
0
0
0
0
Bohm §
Jordan
f '=6.5
2.5
10.0
32.9
3.4
13.7
45.0
2.5
10.0
32.9
3.4
13.7
45.0


Goren
0.
0.
0.
0.
0.
r.
0.
0.
0.
0.
0.
1.
004
07
98
008
15
98
004
07
98
008
15
98


Westinghouse
0
0
0
0
0
0
0
0
0
0
0
0
.078
.14
.25
.106
.19
.34
.78
.14
.25
.106
.19
.34

Schmidt
0.48
0.88
1.54
0.66
1.20
2.10
0.48
0.88
1.54
0.66
1.20
2.10

-------
collection.  This does not agree with the experimental data.
Goren's equation  underestimates penetration for particles
smaller than 1.6 ym and predicts too low a penetration for
particles larger than 1.6 ym in diameter.

DATA BY PARETSKY, ET AL.
     Paretsky  et al. (1971) studied the filtration of dilute
aerosols by beds of sand.  Test conditions and data were reported
in an earlier section.  Their data obtained with a bed of 1,200
to 1,700 ym (-10+14 mesh) angular sand are compared with various
models in Figure 38.
     The agreement between Paretsky  et al.'s data and theory is
good for gas velocities less than 10 cm/s.  For higher gas flow
rates; i.e., in the region where particle collection by impaction
is dominant, the theory underestimates the collection efficiency.
     Agreement between Bohm and Jordan's equation and Paretsky
et al's data is poor in the high gas flow region.  In the diffu-
sional collection region the agreement is fair.
     Predictions by other models do not agree with Paretskyfs
data.

DATA BY GEBHART, ET AL.
     Gebhart  et al. (1973) published an extensive experimental
study on the collection of aerosol particles by diffusion in
packed beds consisting of uniform glass spheres.  They derived
an experimental correlation to predict the diffusional collection
in a granular bed.
     Figure 39 shows the predicted diffusional collection in a
granular bed filter by the equation proposed by Paretsky  et al.
along with the Gebhart  et al. data.  The agreement between theory
and data is fair.
     Both Goren's equation and B8hm and Jordan's equation predict
much too low a penetration when compared with the Gebhart  et al.
data.
                               116

-------
§
1
 »
§
^H
i
               PARETSKY ET AL.  DATA
               UPWARD FLOW
               DOWNWARD FLOW
       BED DEPTH « 19.2 cm
       AEROSOL = 1.1  urn 01A.
                       PSL
          - A.  JACKSON AND CALVERT. C
          - B.  PARETSKY, ET AL.
          _ C.  BOHM AND JORDAN
            0.  GOREN
          - E.  WESTINGHOUSE
            F.  SCHMIDT, ET AL.
            G.  GEBHART, ET AL.
                                                                                                   1.0
0.05
     0.01
                                     5       10
                          SUPERFICIAL GAS VELOCITY, cm/s
     Figure  38.  Comparison of Paretsky, et al.  data with predictions by
                available design equation.
                                                                                               o
o
2
u.
o
g
01
tu
D.
   0.1
                                                                                                 0.01
                                                                                                 0.1
                                                                                                                	GEBHART
                                                                                                                    DATA
  PARETSKY
  ET AL.
PREDICTION^
                                                                                                                             1.0
                                                                                                                 dp,  urn
                                                                                       Figure 39. Comparison of Gebhart  et al. data
                                                                                                   and  predictions by Paretsky et  al.
                                                                                                   equation.  Collection  is in the
                                                                                                   diffusion regime.

-------
CONCLUSIONS
     Based on the comparisons between theory and data presented
above the following conclusions may be drawn:
     1. For superficial gas velocities less than 10 cm/s,
        Paretsky's equation can be used to predict granular
        bed filter performance.
     2. For superficial gas velocities greater than 10 cm/s, the
        primary collection mechanism is impaction.  The avail-
        able models have not been shown to be  satisfactory for
        the prediction of granular bed collection efficiencies.
                             118

-------
                          SECTION  5
                         EXPERIMENT
APPARATUS
     The need for some experimental work became  apparent in the
course of the present research.  A practical  granular bed filter
should be operated at a high gas flow  rate where  inertial impac-
tion is the principal particle collection mechanism.  Available
design equations were not adequate for predicting granular bed
collection efficiencies in the inertial impaction regime.
     To obtain further information on  the mechanism of particle
collection by impaction and to generate additional clean bed per-
formance data, the experimental apparatus shown  in Figure 40  was
constructed.  Filtered room air was used for  the study and all
flow rates were monitored with rotameters.  Monodisperse poly-
styrene latex aerosol was generated using a Collison atomizer.
The aerosol mist from the generator mixed with a stream of dilu-
tion air and passed through a dryer to vaporize the water.   Sta-
tic charges were removed by passing the aerosol through a charge
neutralizing section.  The charge neutralizing section consisted
of a Krypton-85 charge neutralizer.
     Following the neutralizing section the aerosol was further
diluted with filtered room air and then passed into the granular
bed test section.  Gas flow through the granular bed was controlled
by using a bypass vent.
     The granular bed test section was made of 10.2 cm (4 in.)
I.D. glass pipe and the filter was a bed packed with either iron
shot or sand.  The aerosol concentrations before and after the
bed were measured with an optical counter.  Pressure drop was
monitored with calibrated gauges.
     Five grades of iron shot and one grade of sand were used as
bed materials.  The iron shots were SAE S-110, S-170, S-230,
S-280,  and S-330.  Figure 41 shows the size distributions of
                                119

-------
                                                    VENT
                                               ROTAMETER
                      OPTICAL
                      COUNTER
                           GRANULAR
                                BED
                       OPTICAL
                       COUNTER
                                              PRESSURE
                                                 TAP
          DILUTION AIR
VENT
ROTAMETER=
   FILTER
                                                  CHARGE
                                               NEUTRALIZER
                                  DRYER
                       COLLISON
                       ATOMIZER
           ROTAMETER


          FILTER
      COMPRESSED
          AIR
    AIR
                                            BLOWER
   Figure 40.   Schematic  diagram of  the  experimental  apparatus.
                               120

-------
   3,000
I  1,000
o
E
C/3
      300
            T     I    T


           O s-no

          _/\ S-170

               S-230

               S-330
                      I
I
              I
I
10    20   30  40  50 60  70   80    90


 PERCENT BY WEIGHT UNDERSIZE,  I
                                                          95
                                    98
            Figure 41 .  Particle size distribution for iron shot.

-------
these shots as measured with sieves.  The mass median diameters
are 490 urn, 620 ym, 730 ym, 790 ym,  and 860 ym for S-110, S-170,
S-230, S-280, and S-330 shots, respectively.  The sand was Agsco
#2 quartz which was obtained from Exxon  Research and Engineering
Company.  This sand is the same as Exxon used in their Ducon
granular bed filter.  The granule size of the sand is -30 +50
mesh.  The median diameter is 400 ym.
    The experiments were conducted with 0.5 ym, 0.76 ym, and 1.1
ym diameter polystyrene latex monodisperse particles.   Figures  42
through 53 show the data.  In general, collection efficiency
increases with decreasing granule size, increasing bed depth,
and increasing superficial gas velocity.

DATA ANALYSIS
    Table 16 is a list of pressure drops and collection efficien-
cies of 1.1 ym diameter PSL at a superficial gas velocity of 50
cm/s.  It reveals that less pressure drop is required for par-
ticle collection with small granules as bed material  and shallow
beds instead of deep beds.
    For a granular bed with a bed depth of 3.2 cm and operated
at a superficial gas velocity of 50 cm/s, the collection effi-
ciencies for 1.1 ym diameter particles are 22% and 531,  respec-
tively, for 620 ym and 490 ym diameter iron shot.   The pressure
drop increases from 10 cm W.C. for 620 ym diameter granules to
21 cm W.C. for 490 ym granules.   The increase in pressure drop
is 110%.  However, by using the finer grade of granules, the
increase in efficiency is 140%.
    Table 17 is a list of pressure drops for various beds whose
collection efficiencies are 50% for 1.1 ym diameter particles.
As can be seen the pressure drops for shallow beds are less than
for deep beds.
    For a shallow bed to have the same collection efficiency as
a deep bed, it has to run at a high superficial gas velocity.
Therefore, the gas flow capacity of a shallow bed is  higher than
that of a deep bed.  However, there is a limit on how high a gas
                               122

-------
   1.0
   0.9

   0.8

S  0.7

|  0.6

S
                                                                                         i     i
                            BED DEPTH = 3.1  cm
S
    0.5
    0.4
            BED MATERIAL:  IRON SHOT
            GRANULE DIAMETER:  490 \m
            AEROSOL: 0.5 urn DIA. POLYSTYRENE LATEX
    0.3
       10
                                  20
60
                                   30         40       50
                        SUPERFICIAL GAS VELOCITY, cm/s
Figure 42.  Experimental particle penetration of a clean granular  bed filter.
70   80   90  100
    1.0

    0.9

    0.8

g   0.7
t—•

§   0.6

§
t   0.5
    0.4
                             BED  DEPTH =  3.2 cm
                                          6.2 cm —
           BED MATERIAL:  IRON SHOT
           GRANULE DIAMETER:  620 pm
           AEROSOL:  0.5 pm DIA. POLYSTYRENE LATEX
    0.3
                                                                      i
                                                                             1
                                                                                   1
                                                                                        i
       10                         20             30          40      50     60    70    80   90  100
                                     SUPERFICIAL GAS VELOCITY, cm/s
              Figure 43.  Experimental particle penetration of a clean granular  bed filter.
                                                123

-------
    1.0



    0.9




    0.8





2   °'7
t—



£   0.6



o


i   0.5

h—
UJ
z
UJ
ex

    0.4










    0.3
                    BED DEPTft =3.2 cm

                                 6.2 cm

                                 9.2 cm


                                12.2 cm
    BED MATERIAL:   IRON SHOT


    GRANULE DIAMETER:  790 ,.m


    AEROSOL:  0.5 ..m DIA. POLYSTYRENE LATEX
 10
                                  20              30         40       50


                                      SUPERFICIAL GAS VELOCITY, cm/s
                                                               60
                                                                             70     80   90   100
         Figure 44.  Experimental particle penetration of a clean granular  bed  filter.
    1.0



    0.9




    0.8




    0.7





    0.6






    0.5







    0.4










    0.3
                   BED DEPTH   6.2 cm
    BED MATERIAL:  IRON SHOT


    GRANULE DIAMETER:  620 i.m


    AEROSOL:  0.76 ;.m DIA. POLYSTYRENE LATEX
10
                                                                              60
                                                                     70   80   90  100





Figure 45.   Experimental particle penetration of a clean granular bed filter.
20              30          40       50


    SUPERFICIAL GAS VELOCITY, cm/s
                                         124

-------
   1.0
   0.9
   0.8

S  0.7
M

3  0.6
2  0.5
tu
    0.4
    0.3
                BED DEPTH = 9.2 cm

                           12.2 cm
BED MATERIAL: IRON SHOT
GRANULE DIAMETER:  790 urn
AEROSOL:  0.76 \an DIA. POLYSTYRENE LATEX
        10
                       20              30         40       50
                            SUPERFICIAL GAS VELOCITY,  cm/s
                                                                            60    70    80   90  100
                 Figure 46.   Experimental  particle  penetration of a clean granular bed filter.
                 1.0
                 0.9

                 0.8

                 0.7

                 0.6

                 0.5


                 0.4



                 0.3
                 0.2
               9.2 cm
              12.2 cm
             BED MATERIAL: IRON SHOT
             GRANULE DIAMETER: 620 pm
             AEROSOL: 1.1 pm DIA.
                      POLYSTYRENE LATEX
                     20             30         40       50     60    70   80   90   100
                                      SUPERFICIAL GAS VELOCITY, cm/s
                       Figure 47.  Experimental particle penetration of a clean
                                   granular bed filter.
                                                125

-------
      1.0

      0.9

      0.8


      0.7


      0.6
       0.4
      0.3
      0.2
  BED DEPTH =
        3.2 cm
       12.2 cm
   BED MATERIAL:   IRON  SHOT

   GRANULE DIAMETER:  730 urn
   AEROSOL:   1.1  ym DIA. POLYSTYRENE
             LATEX
          20             30          40      50     60     70   80   90  100

                          SUPERFICIAL GAS VELOCITY,  cm/s

             Figure 48.  Experimental particle penetration  of a  clean granular
                         bed filter.
     1.0

     0.9

     0.8

     0.7


     0.6
P   0.5
2    0.4
     0.3
    0.2
 BED  DEPTH =
       3.2 cm
       6.2 cm
       9.2 cm


     12.2 era
  BED MATERIAL:  IRON SHOT
  GRANULE DIAMETER:  860 \m

- AEROSOL:  1.1 urn DIA. POLYSTYRENE LATEX
          20
              30          40       50     60    70    80  90  100
                 SUPERFICIAL GAS VELOCITY, cm/s
           Figure 49.  Experimental particle penetration of a clean granular
                       bed filter.
                                    126

-------
                                                                                                   1.0
IS)
    1.0

    0.9

    0.8

    0.7


§   0.6
i—*
(J
U_
 „  0.5
         <£
         ce.
             0.4
             0.3
             0.2
                   BED DEPTH =
                       3.2 cm

                       6.2 cm

                       9.2 cm

                      12.2 cm
                    BED MATERIAL:  IRON SHOT
                    GRANULE DIAMETER:  790 urn
                    AEROSOL:  1.1 urn DIA. POLYSTYRENE LATEX
                20               30    '     40       CO     60    70    80  90  100
                                      SUPERFICIAL GAS VELOCITY, cm/s
                  Figure  50.   Experimental particle penetration of a clean granular
                               bed filter.
§
1
     0.5
     0.4
                                                                                                   0.3
                                                                                                   0.2
                                                                                                   0.1
     0.05
                                                                                                                      9.2 cm
             BED  MATERIAL:   IRON
             GRANULE  DIAMETER:  490
             AEROSOL:  1.1 um DIA
                                                                                                                     SHOT
                                                                                                                       \im
                                                                                                                      PSL
                                      I
                                                    I
         10
                                                             I   I
                                                                                                               20         30     40    50
                                                                                                                SUPERFICIAL GAS VELOCITY, cm/s
100
                                                                                                   Figure 51.  Experimental  particle  penetration of a clean granular
                                                                                                              bed filter.

-------
            1.0
is}
oo
            0.5
            0.2
            0.1
           0.05
           0.04
           0.03
                                            1IIIITT
                                       0.5 urn DIA.  PSL
                 —  BED MATERIAL: AGSCO SAND

                 —  GRANULAR DIAMETER: -30+50
                                       MESH

                 -  BED DEPTH:  1.7 cm
                                  I
                I  	I
               10
    20         30      40   50

SUPERFICIAL GAS VELOCITY, cm/s
                                                                           100
             Figure 52.  Experimental particle penetration of a clean granular
                        bed filter.
                                                                                              1.0
                                                                                              0.5
                                                                                              0.1
                                                            o
                                                            r 0.05
                                                                                             0.01
                                                                                             0.05
0.03
                                                                                                                                        I     I    I  I  I  I  I <-
                                                                                                                                     0.5 pro DIA. PSL
                                                                                                                                     1.1 pm DIA. PSL
                                                                        BED MATERIAL:  AGSCO SAND

                                                                        GRANULE DIAMETER: -30+50 MESH

                                                                        BED DEPTH:  2.6 cm
                                                                                                          I      III  JIM
                                                                                                           I     I    I   I  I  I I  I
                         5        10

                   SUPERFICIAL GAS VELOCITY,  cm/s
                                                                                                                                                   50        100
                                                                 Figure 53.  Experimental penetration of a clean granular bed
                                                                             filter.

-------
                       TABLE 16.  PRESSURE DROP AND COLLECTION EFFICIENCY FOR 1.09 urn
                                  DIAMETER PARTICLES AT A SUPERFICIAL GAS VELOCITY OF 50 cm/s

Granule
Diameter
(ym)
490
620
730
790
860
Bed
%
Coll

53
22
14
12
10
Depth = 3.2 cm
AP
(cm W.C.)

21
10
7
65
6
Bed
%
Coll

83
38
27
30
19
Depth = 6.2 cm
AP
(cm W.C.)

38
20
14
12
12
Bed
%
Coll

93
55
35
44
27
Depth = 9.2 cm
AP
(cm W.C.)

_-
28
21
18
28
Bed
%
Coll

_-
68
53
55
38
Depth » 12.2 cm
AP
. (cm W.C.)

—
37
27
23
23
to

-------
TABLE 17.  PRESSURE  DROP  FOR  50% COLLECTION
          OF 1.1 urn DIAMETER PARTICLES
Granule
Di ameter
(ym)
490
620
730
790
860
Pressure Drop (cm W.C.)
Bed Depth
3.2 cm
19
17
16
15
15
Bed Depth
6.2 cm
--
24
23
20.5
23
Bed Depth
9.2 cm
--
26
27
22
28
Bed Depth
12.2 cm
—
22
25
21
30
                    130

-------
velocity  the  bed can be safely operated without the danger of
causing particle reentrainment.

PRESSURE  DROP DATA
      The  pressure drop data obtained with the iron shot bed mate-
rial  were analyzed.  Figures 54 through 58 show the experimental
and predicted pressure drops.   The prediction was based on
Ergun's  equation.  As can be seen, the predicted pressure drop
is lower  than that measured.
    The pressure drop prediction based on Ergun's equation is
very  sensitive to the bed porosity.   The difference in  the pre-
dicted and measured pressure drops might be caused by an  error
in bed porosity determination.   The bed porosity was calculated
from  the  measured weight of a  bed of known volume.   It  is  very
difficult to  accurately determine bed porosity by this  method.
    By fitting Ergun's equation to the pressure drop data,  the
iroid fraction  of the bed was obtained.  Table 18 shows the  result
along  with the measured void fractions.  The measured void  frac-
tion is very close to calculated void fraction.
        TABLE 18.   MEASURED AND CALCULATED VOID FRACTION OF
                   THE GRANULAR BED
Shot No.
S-110
S-170
S-230
S-280
S-330
Average Diameter
Vim
490
620
730
790
860
Measured
Void Fraction
0.39
0.39
0.41
0.41
0.40
Void Fraction From
Pressure Drop Data
0.37
0.39
0.40
0.40
0.39
                               131

-------
                                                                                             100
to
           100
            50
§ 40
a.
i
° 30
ae
            20
             10
                                   I          I        \
                               PREDICTED FOR e  = 0.39
                                      I    I   I   II

                                        BED DEPTH =
                                          9.2 cm
BED MATERIAL:  IRON SHOT
GRANULE DIAMETER: 490
                                  I
                                          I    I    I   I
                     20  •       30     40    50                100
                 SUPERFICIAL GAS VELOCITY, cm/s
Figure 54.   Experimental  and predicted pressure drops across  a
            clean granular bed filter.
                                                                                           o
                                                                                           on
                                                                                           o
                                                                                              50
                                                                                              40
                                                                                               30
                                                                                              20
                                                                                               10
                                                                                                                   T
                                                                                                                 I
                                                                                                                                   I     I    I    I   I
                                                                                              	  PREDICTED FOR e = 0.39
                                                                                              BED MATERIAL:  IRON SHOT
                                                                                              GRANULE DIAMETER:  620 urn
                                                                                                  10
                                                                                            BED DEPTH
                                                                                              12.2 cm~*

                                                                                               9.2 or-
                                                                                                                   6.2 cm-
                                                                                                                    3.2 cm
                                                                                                                    I	
                                                                                                                       I    I   I   I   I
                                                                                               20          30      40     50
                                                                                                SUPERFICIAL GAS  VELOCITY,  cm/s
                                                                                                                                                    100
                                                                                               Figure 55.  Experimental and predicted pressure drops across a
                                                                                                           clean granular bed filter.

-------
CM
            50

            40  I—


            30



            20
             ,o
            1          I
	 PREDICTED FOR e = 0.41
BED MATERIAL:  IRON SHOT
GRANULE DIAMETER:  730 urn
                 10
                              BED DEPTH
                               12.2 cm
                                 9.2 cm
                                 5.2  cm —'
                                 3.2 cm ^*
                                   I
            20         30      40   50
              SUPERFICIAL GAS VELOCITY, cm/s
                                                                           100
               Figure 56.   Experimental  and  predicted pressure drops of a clean
                           granular bed  filter.
                                                                                          O
                                                                                          a:
                                                                                          a
                                                                    in
                                                                    UJ
                                                                    ce.
50

40

30



20
                                                                                               10
              I          I
	  PREDICTED FOR
BED MATERIAL:  IRON SHOT
GRANULE DIAMETER:  790 MI
                                                                                                                                                      I   I
                                                                                                                              6 = 0.41
                                                                                                  10
                                                                                              BED DEPTH
                                                                                               12.2 cm-
                                                                                                                      3.2 cm
                                                                                                                    I
                                                                                                         I
                                       I
                                           I    I    I   I
                    20         30     40    50
                     SUPERFICIAL GAS VELOCITY, cm/s
                                                       100
                                                                           Figure  57.   Experimental and predicted pressure drops across  a
                                                                                       clean  granular bed filter.

-------
   40
                                          T     I     I    I   I   I
                     PREDICTED FOR e = 0.40
   30
BED MATERIAL:   IRON SHOT

GRANULE DIAMETER: 860 ym
   20
o
CL.
O
UJ
   10
                                          I      I     I    I   I   I
     10                20         30     40    50                100

                     SUPERFICIAL GAS VELOCITY, cm/s .

   Figure 58.  Experimental and predicted pressure drops across a
               clean granular bed filter.
                                 134

-------
                            SECTION 6
                           DESIGN MODEL
MATHEMATICAL MODELING
     The granular bed can be  envisioned  as  a great number of
impaction stages connected  in series  (Figure 59).  Particle
collection is by impaction  as in  a  cascade  impactor.  The jet
openings are the pores  in each layer  of  granules.  It is assumed
that the jet diameters  in the granular bed  are of uniform size
with a diameter equal to the  hydraulic diameter of the void
space.  The gas velocity in the jet is the  average interstitial
gas velocity.
     If 'n* is the collection efficiency of one impaction stage,
the particle penetration for  the  granular bed will be

                        Ptd  - (l-n)N                   (88)

     where           Ft, = penetration for particles with
                          diameter  d_, fraction
                      n = single  stage collection efficiency, fraction
                      N = number  of impaction stages, number
     As in some cascade impactors,  each  layer of granules served
both as the jet plate and as  the  collection plate.  Therefore,
each layer of granules  is an  impaction stage and "N" is equal to
the number of granular  layers  in  a  bed.  For a randomly packed
bed

                         N  =  1 1                         C89)
                              2dc

     where              Z- bed  depth,  cm
                     d  = granule diameter, cm
                                135

-------
                         jet formed by
                         granule cluster
                      granule
                      target
Figure 59.  Diagram of granular bed showing
            impaction concept.
                 136

-------
      and

                         Ptd -  U-n)  '   dc"                 (90)


      The impaction collection  efficiency,  "n"  is  a  function of
 "Kp", the inertial impaction parameter.  The impaction parameter
 is defined as

                     v    C' Pp dp   uj
                      P      Q 11  A	L                     (91}
                      c      y y^ a.
                               u   i
      where           c' =  Cunningham slip factor, dimensionless
                      Pp =  particle density, g/cm3
                      dp =  particle diameter, cm
                      u. =  jet velocity, cm/s
                      UG »  gas viscosity, g/cm-s
                      d. =  jet diameter, cm

      Since            U;j = uQi = ^                      (92)

      and      d.  = 4rH = | ^- dc                        (93)

      where          uGi = average interstitial  gas velocity,
                           cm/s
                       e = bed  porosity,  fraction
                      rH = hydraulic  radius,  cm
                      d   = granule  diameter,  cm

     we have    Kp  - |  if  C9PgdP  "G                  (94)
                        £         bC

     The relation  between "n"  and  "K "  can  be evaluated once the
flow field is  defined.   Flow fields  reported in the  literature for
inertial impaction, e.g.,  Ranz  and Wong  (1952)  and Marple (1970),
are adequate for K  > 0.15. For  Kp  < 0.15,  there  is no suitable
flow field reported in  the  literature.   Therefore, the relationship
between "n" and "K " could  not  be  calculated analytically.

                                 137

-------
     The relationship was back-calculated from equation   (90)  and
experimental data.  Figure 60 shows the results.  The curve  can
be approximated by the following equation:

                 n = 10.0 K 3'23 exp [0.27 (In Kp)2]       (95)

for  3 x 10"3 <  K  < 0.15
                  P -

     There is scatter in the lower end of the curve.  For K  <  10" ,
"n" is very sensitive to experimental data.   A few percent scatter
in the data will cause "n" to fluctuate greatly.  Figure 61  com-
pares the experimentally determined "n" versus "K " curve with
those reported by Ranz and Wong (1952), Stern, et al.  (1962),
and Mercer and Stafford (1969).  All reported curves are for
K  > 0.15.  As can be seen, the curve calculated in the present
study matches other researchers' results.  The curve determined
in this study is a continuation of other researchers'  curves.
     Paretsky  et al. (1971) and Knettig and Beeckmans (1974)
studied the collection of monodispersed aerosol particles in
granular bed filters.  Their data were transformed into "K "
versus "n" plots as shown in Figure 62.
     Knettig and Beeckmans used 425 ym glass beads as  granular
material.  Bed porosity was 0.38.  Aerosol particles were 0.8,
1.6, and 2.9 ym in diameter.  As can be seen from Figure 62,
their data are close to the results of present study.
     Paretsky  et al. (1971) studied the filtration of 1.1 ym
diameter polystyrene latex aerosols by beds of sand.  They studied
a bed of -10+14 mesh (1,200 to 1,700 ym) angular sand and a  bed
of -20+30 mesh (500 to 850 ym) sand at superficial gas velocities
between 0.3 and 80 cm/s.  Bed porosities were 0.41 and 0.43,
respectively.  Single stage collection efficiencies were calcu-
lated from their data.  In the calculation, the granule diameters
were assumed to be the arithmetical mean of the smallest and the
largest granule size in the bed.  The results are plotted in
Figure  62.  For a given inertial parameter, Paretksy  et al.

                                138

-------
      0.1
§
hH
H
U
u,
      10
        -2
      10
        -3
      10
        -4
                         1.09  vim
                         0.76  ym
                         0.5 urn
                               3*2 3               2
                      n = 10 Kp   exp [0.27 (In K ) ]
          2x10
    10
0.1
                             K ,  DIMENSIONLESS
           Figure  60.
K  versus n for  round  jet  model for
particle collection  in a GBF.
                             139

-------
§
CJ
      i.o
     0.5
     0.1
     o.os
     0.01
     0.005
     0.001
          10
                                    STERN
                                    ET AL,
                                    (1962)
                                              RANZ §
                                              WONG
                                              (1952)
                                            MERCER §
                                            STAFFORD
                                             (1969)
                               —-  PRESENT  STUDY
O.OS
0.1
0,5
                       K , DIMENSIONLESS
                                                                      0.05
                                                                      10
                                                                   LJ
                                             10"
                                                                      10
                                                                                I   I  I  I  I '
                                                       o
                                                KNr.TTlC.  f,
                                                Bl:.r.i:KMANS
                                                 (1974)
                                                                    PARHTSKY I:.T AL./

                                                                     20-30 MESR,
                                                   . PARMTSKY F.T AL.
                                                      11971)        X
                                                    10-14 MESH    X-
                                                                                            PRESENT
                                                                                            STUDY
                                                                                          i i i  i
                                                                                                                i  i i  i i i
                                                                          0.002
 0.01

K   DIMENSIONLESS
 P.
                                                                                                                      0.1
                                                                                                     0.2
        Figure 61,  'K ' versus  'n' for round  jet.
                                                Figure 62.  Efficiency vs. inertial  impaction
                                                            parameter for comparison.

-------
data give a higher collection efficiency than reported  in this
study.  Their data would be close to that of present study if
the mean diameter were equal to the smallest granule diameter.

COMPARING MODEL PREDICTIONS WITH PERFORMANCE DATA
     The design equation is for the prediction of particle col-
lection by a clean bed.  If no filter cake is formed and
the collected particles are uniformly distributed in the bed, then
the equation is still applicable.  The design equation has been
used to predict the performance of Rexnord gravel bed filters
and the Combustion Power Company "dry scrubber."
     Figure 63 shows the comparison between the data reported by
McCain  (1976) for a Rexnord gravel bed filter and the prediction by
the present model.  The prediction by the present model is close
to the data.
     Figure 64 shows the comparison of Hood's data with predic-
tions.  The predicted penetration is higher than that measured.
     Figures 65 through 67 show the comparison of A.P.T.'s data
for CPC GBF with predictions.  The present model predictions
agree with data for some runs but not for all runs.  The present
model is very sensitive to bed porosity.  In the calculations,
a bed porosity of 0.35 was used.  If a bed porosity of 0.3 is
used, the model would give a better fit with the data.
                                141

-------
     1.0
     0.5
§   0.1
I
2
£  0.05
    0.01
                            I   I
A. JACKSON & CALVERT
B. PARETSKY, ET AL.
C. BbHM & JORDAN,
   f = 10
D. B'OHM AND JORDAN,
   f = 6.5
E. GOREN.e = 0.25
F. WESTINGHOUSE
6. SCHMIDT, ET AL.e = 0.25
                         D
      I     I    I   I  I I  I  I !
       0.1
                 0.5       1
        AERODYNAMIC  PARTICLE  DIAMETER, umA
                                                                     10
    Figure 63. Comparison of McCain's gravel bed particle collection data
               with design equation predictions.
                                                                                         1.0
                                                                                         0.5
2
i
                                                                                       0.05
                                                                                       0.01
                                                                                         i     i    i   r
                                                                                                HOOD'S
                                                                                                 DATA —
                                                                                                A. JACKSON & CALVERT
                                                                                                B. PARETSKY, ET AL.
                                                                                                C. BOHM & JORDAN,
                                                                                                   f'= 10
                                                                                                D. BOHM & JORDAN, f=6.5
                                                                                                      I     I    I   I  I  I I  I I
                                          E. GOREN.e  = 0.25
                                          F. WESTINGHOUSE
                                          G. SCHMIDT, e  =  0.25
                                                 I     I    I   I  I  I  I  I
                                                                                           0.1
                              0.5       1
                               PARTICLE DIAMETER,  urn
                                                                                                                                                         10
                                                                           Figure 64.   Comparison  of Hood's data with predictions by available
                                                                                       design equations.

-------
data give a higher collection efficiency than reported  in this
study.  Their data would be close to that of present study if
the mean diameter were equal to the smallest granule diameter.

COMPARING MODEL PREDICTIONS WITH PERFORMANCE DATA
     The design equation is for the prediction of particle col-
lection by a clean bed.  If no filter cake is formed and
the collected particles are uniformly distributed in the bed, then
the equation is still applicable.  The design equation has been
used to predict the performance of Rexnord gravel bed filters
and the Combustion Power Company "dry scrubber."
     Figure 63 shows the comparison between the data reported by
McCain (1976) for a Rexnord gravel bed filter and the prediction by
the present model.  The prediction by the present model is close
to the data.
     Figure 64 shows the comparison of Hood's data with predic-
tions.  The predicted penetration is higher than that measured.
     Figures 65 through 67 show the comparison of A.P.T.'s data
for CPC GBF with predictions.  The present model predictions
agree with data for some runs but not for all runs.  The present
model is very sensitive to bed porosity.  In the calculations,
a bed porosity of 0.35 was used.  If a bed porosity of 0.3 is
used, the model would give a better fit with the data.
                                141

-------
IX)
             1.0
             0.5
        g
        i
0.1
            0.05
            0.01
                          T     I    Tl
— A. JACKSON & CALVERT
- B. PARETSKY, ET AL.
"" C. BbHM & JORDAN,
      f  = 10
   D. B'OHM AND JORDAN,
      f  = 6.5
- E. SOREN.e = 0.25
   F. WESTINGHOUSE
   G. SCHMIDT, ET AL.e = 0.25
                            D
         I     I    I   I  I I  I  I
                                                              I    I   i  I  I I  J
                0.1
                         0.5       1
                AERODYNAMIC  PARTICLE  DIAMETER, umA
                                                           10
             Figure 63.  Comparison  of McCain's gravel bed particle collection data
                        with  design equation  predictions.
                                                                                    1.0
                                                                                                 0.5
                                                                                    0.1
                                                                                               0.05
                                                                                   0.01
                                                                                           A. JACKSON & CALVERT
                                                                                           B. PARETSKY, ET AL.
                                                                                           C. BOHM  & JORDAN,
                                                                                              f = 10
                                                                                           D. BOHM  & JORDAN,  f=6.5
                                                                                                 I     I    I   I I  I
                                                                                                                                      E. GOREN.e = 0.25
                                                                                                                                      F. WESTINGHOUSE
                                                                                                                                      G. SCHMIDT, e  = 0.25
I I
           I     I
I  I  I  I  I
                                                                                                    o.i
                                                                                                        0.5      1
                                                                                                         PARTICLE DIAMETER,
                                                                                                                                                                 10
                                                                                   Figure 64.   Comparison of Hood's data with predictions by available
                                                                                               design equations.

-------
w
         1.0
         0.5
          0.1
      UJ
      z
      LU
         0.05
         0.01
UQ = 81 cm/s
dc = 0.2 cm
 Z = 20.2 cm
GRANULE CIRCULATION
RATE = 0.8 kg/kg AIR
                                                               A.P.T.  DATA
                                             SCHMIDT. ET AL.   .
                                             e = 0.35         \
                            J    I   I  I  I  I I I
                                         I    I   I  I I I  I
             0.1
                 0.5        1
            AERODYNAMIC  PARTICLE  DIAMETER,
                                                                          10
           Figure 65.  Experimental and predicted performance of CPC GBF
                       (A.P.T.  data).
                                                                            1.0
                                                                                              0.5
                                                                                           ,   0.1
                                                                                          fe  0.05
                                                                           0.01
                                                                                                           1     I
                                                                                                      PRESENT MODEL,
                                                                                                      E = 0.35
                                                                                                      WESTINGHOUSE
                                                                                                 0.1
                                                                                                            SCHMIDT, ET AL
                                                                                                                t. = 0.35
UG = 41 cm/s
dc = 0.2 cm
 A = 20.2 cm
                                                                                                                             I   I   I  I  I  I
                                                                                                                          GOREN, E  = 0.35   _|
                                                                                                                          GRANULE CIRCULATIOirl
                                                                                                                          RATE  = 1.6 kg/kg  -\
                                                                                                                             /        AIR
                                                                                                                              A.P.T.  DATA
                                                                                                                            GRANULE RECIRCULATION \
                                                                                                                            RATE = 0.8 kg/kg AIR
    I     I    I   I   I  I I I  I
I     I
I  I
                0.5      1

      AERODYNAMIC PARTICLE  DIAMETER,
                                                                                                                                                              10
                                                                             Figure 66.  Experimental and predicted performance of CPC GPF
                                                                                        (A.P.T. data).

-------
    1.0
    0.5
I     I    I   I   I  I  I I  I

        PRESENT MODEL
           e = 0.35

    WESTINGHOUSE-
o
i—
o
2
oi
LU
LU
Q-
   0.05
   SCHMIDT, ET AL.
       e = 0.35
             UG =
         I"  dc =
              I =
41 cm/s
0.2 cm
40.6 cm

                                                 I     I    I   I  I  I  I
GOREN, e = 0.35    _
   0.01
                   I     I    I   I  I  I  I I  I
                              \l     I    I   I  I I I I
        0.1                  0.5        1                    5
                     AERODYNAMIC PARTICLE DIAMETER, ymA
    Figure 67.  Experimental and predicted performance of CPC GBF
                (A.P.T. data).
                                                    10
                                  144

-------
                          SECTION 7
                 PRESENT TECHNOLOGY EVALUATION

GRANULAR BED FILTER SYSTEMS
     GBFs have been used commercially for over 30 years, with
several designs commercially available.  Granular bed filters
may be classified either according to bed structure or according
to bed cleaning method.  The first classification method is used
in this report.
     With respect to the bed structure, granular bed filters
may be classified as continuously moving, intermittently moving,
and fixed bed filters.
Continuously Moving Bed Filters
     The continuously moving bed filter is usually arranged in a
cross-flow configuration.  The bed is a vertical layer of granular
material held in place by louvered walls.  The gas passes hori-
zontally through the granular layer while the granules and col-
lected dust continuously move downward and are removed from the
bottom.  The dust and granules are separated by vibration.  The
cleaned granules are then returned to the overhead hopper by a
granule circulation system.
     Several commercial designs fall into this category.  They
include the Dorfan Impingo filter, the Consolidation Coal Company
filter, and the "dry scrubber" of the Combustion Power Company.
The "dry scrubber" is the only one that is presently marketed.
Dorfan Impingo Filter -
     The Dorfan Impingo filter was invented by Morton Dorfan and
was offered commercially by Mechanical Industries, Inc. in the
early 1950's.  The device is a vertical panel filter in which the
granular materials continuously fall through the panel.  The
granule flow rate is controlled by the setting of a rotary valve

                               145

-------
located near the bottom of the panel.  The dust-laden gas is
filtered by blowing the gas through the panel horizontally.
Filtered dust is carried downward with the granules (Figure 68).
     The opposite walls of the panel are not parallel but are
slightly offset from the vertical so that the panel is tapered
with its narrowest point at the top and widest at the bottom.
The granules thus travel downward in a panel of constantly in-
creasing cross-sectional area.  The rationale is that the tapered
construction acts to prevent hangups caused by size increases of
the granules as the dust loading builds upon their surfaces.
     Four units, each of 8 m3/s (17,000 CFM) gas capacity and
consisting of two cells, were installed in a plant to collect
asbestos rock dust from a stream of flue gas coming from a direct
fired dryer in which the rock was dried prior to milling.  The
granular bed was a 30 cm (1 ft) thick panel with 2.74 m x 4.27 m
(9 ft x 14 ft) filtering area.  The granules were raw asbestos
rock ranging from 1.3 cm to 3.8 cm (0.5 to 1.5 in.) in diameter.
     The dust was 100% finer than 100 mesh and 601 finer than
10 microns.  The concentration entering the collector was approxi-
mately 14.8 g/m3 (6 g/ft3) and that leaving about 0.49 g/m3  (0.2
g/ft3).
     The Dorfan Impingo filter installations are no longer in
use and the equipment is not presently marketed.
Consolidation Coal Company Filter -
     A granular bed filter was studied first on a pilot scale
and then with a full-size installation by the Consolidation Coal
Company in the period 1950-1952.  This equipment was used for
collecting dust from hot gases leaving coal drying operations.
Lump coal  [0.95 cm to 3.2 cm (3/8 in. to 1.25 in.)in size] was
used as the bed granules.  The design concept for the full size
installation is shown in Figure 69.
     The granules were fed from an overhead bin to two separate
panels.  The panel at each side consists of six separate cells
facing the inlet gas which flowed through the beds from a central
dust and exited via ducts at either side of the installation.
Typical operating characteristics are given in Table 19.

                               146

-------
                              Filter
                              Particles
                              Rotary Valve
Figure 68.  Dorfan Impinge Filter.
                 147

-------
                                  Granule Feed
            Granular Discharge
Figure 69.  Consolidation Coal Company filter






                     148

-------
     TABLE  19.  TYPICAL OPERATING CHARACTERISTICS OF
               CONSOLIDATION COAL COMPANY GRANULAR
               BED FILTERS
Type of service - recovery of dust from coal drying operation,
Size - Individual cell was 127 cm wide x 122 cm high x 51 cm
       thick (4'2" wide x 4' high x 20" thick).  In the unit
       the filtering surfaces were divided into two banks of
       six cells each in parallel.
Capacity - 0.5 - 1.5 m3/s per square meter of filter area
           (100 - 300 CFM/ft2)
Total filter area - 18.6 m2  (200 ft2)
Inlet dust loading - 7.4 - 17.2 g/m3 (3 - 7 gr/SCF)
Size of dust - 86% < 30 ym and 22% < 10 pm
Granular material - Coal, 0.95 cm to 3.2 cm (3/8" to 1-1/4")
Granule flow rate - 22 metric tons/hr
Operating temperature - 130°F
Pressure drop  - 5.8 cm W.C.
Collection efficiency - 89 - 99%
                               149

-------
Combustion Power Company "Dry Scrubber" -
     The "Dry Scrubber" is currently offered commercially by Com-
bustion Power Company.  It is similar to the Dorfan Impingo Filter.
It consists of a vertical bed of granules held in place by louvers.
It is operated in cross-flow configuration.  Gas flows horizontally
through the bed and the bed continuously moves downward.  Clean
granules are introduced at the top and a mixture of dust and
granules is removed from the bottom.  Dusts and granules are
separated in a shaking device.
     Two models are produced by Combustion Power Company.  Figure
70 shows the regular  "Dry Scrubber."  It is used when the parti-
culate  loading is low.  For very high inlet particle loadings, the
"Integral Cyclone Model" is recommended.  It has a low energy
cyclone wrapped around the outer shell of the standard "Dry Scrub-
ber" (Figure 71].
Intermittently Moving Bed Filters
     In the late 1950s, Squires modified the continuously moving
bed design of the Dorfan filter to obtain a fixed bed device with
intermittent movement of granular solids.  The design is called
the LS  (Loose Surface) filter.  It uses a finer grade granule than
the Dorfan filter and the bed is stationary during filtration.
The accumulated filter cake is removed by moving just the surface
layer of granules.
     Figure 72 shows one possible arrangement.  The granular bed
is a narrow vertical bed of granules (-50+60 mesh) held between a
panel of louvers and a fine mesh screen.  In some applications,
a relatively coarse grade granule is used on the gas exit side
to prevent blow-through of the smaller collecting granules.  During
operation the dusty gas flows horizontally through the panel bed.
The particles collected by the bed build up a cake on the exposed
bed surfaces and to some extent penetrate to the interstices.
When the resistance of the cake has reached an undesirable level,
the clean gas outlet valve is closed and a short pulse of compressed
air is blasted in reverse flow through the granular bed.  In con-
tinuous use, the valves operate on a timed cycle.
     The blow-back pulse is sufficient to physically lift the

                                150

-------
GAS
FLOW
CLEAN
MEDIA
CONVEYOR
                           TO DUST STORAGE
   Figure 70.  Combustion Power Company "Dry Scrubber."
                151

-------
                                                           CLEAN
                                                           MEDIA
                                                           CONVEYOR
                                   TO  DUST STORAGE
Figure 71. The intergral  cyclone model  of  the  "Dry Scrubber."
                                  152

-------
  DIRTY GAS
 )IRTYI GAS
CROSS-SECTION
  OF COLUMN
      CLEAN GAS,,
         OUT
                                       BLOWBACK
                                        VALVE
                                        SAND BED
DIRTY
 GAS
CLEAN
 GAS
                                               CROSS-SECTION
                                               OF FILTER  BED
                                     SAND REMOVAL CHUTE
   Figure 72. Possible design for Squires Panel  Bed Filter.
                            153

-------
sand beds as a mass, with minimum inter-particle movement, so that
a surface layer of granules between each pair of louvers is physi-
cally ejected from the panel and falls to the bottom of the filter
vessel along with the collected filter cake.  The expelled granule
is immediately replaced by the downward movement of a fresh granule
from the overhead hoppers.
     The development of the LS filter has continued for the past
ten years at the City College of the City University of New York
with financial support from EPA and EPRI.  This development work
resulted in minor modifications of louver configurations (wish-
bone type louvers) and of the puff-back cleaning technique.
Fixed Bed Granular Filters
     As opposed to the continuous and intermittent moving beds,
fixed bed granular filters require no granule circulation.   Col-
lected particles in the bed are removed either mechanically or
penumatically.  There are three fixed bed devices.  The "Lurgi-
MB Filter" and the "Rex-Gravel Bed Filter" clean the bed mechani-
cally.  The "Ducon Granular Bed Filter" uses a reversed gas flow
to clean the bed.
The Lurgi MB Filter -
     Max and Wolfgang Berz designed a granular bed that requires
no granule circulation.  The cleaning is carried out by flowing
a reverse flow of gas through the bed while subjecting it to
mechanical vibration of sufficient magnitude to cause the inter-
granule movement necessary for removal of entrapped particles.
     The granular bed is a layer of loosely packed material
such as gravel held on a horizontal sieve plate.   Gravel sizes
range from 1 to 6 mm in diameter.  Figure 73 is a sketch of the
system.
     In operation, the dusty gas flows upward through the gravel
bed and vents through the clean gas duct on top.  The dust will
be retained and gradually will buildup in the bed.  This increases
the pressure drop.  As soon as the pressure drop of the filter
exceeds a pre-determined level, the gas flow is stopped and the
bed is cleaned.  For some applications it is possible to stack two
                                154

-------
              -i
                         Inlet for dirty gas
                         Gravel layer
                         Shaking motor
                         Shut-off flap
                         Scavenging air flap
                         Clean gas outlet
                         Hopper
                         Screw-type discharger
Figure 73. Lurgi MB filter.
              155

-------
or more filter beds, one over the other, in a given device, with
the bottom-most layer consisting of relatively coarse material
to act as a pre-filter.
     This granular bed filter was marketed commercially by Lurgi
Appareteban Gesellschaft M.B.H. of Frankfurt, Germany.  It was
mostly used in cement plants.  It was redesigned in 1968 and was
withdrawn from the market at about the end of 1969.  Its short-
coming lies in the strain imposed on the necessary flexing membranes
associated with the vibrating technique.  It is conceivable that
at low temperatures where rubber membranes and spring-supported
bed mounts are feasible, such a filter could operate with a rea-
sonable life.  However, at low temperatures it could not compete
economically with baghouses.  At high temperatures, metal bellows
would be needed for the flexing membrane, and their life expectancy
in the hot and dusty environment is too short for practical
application.
Rex Gravel Bed Filter -
     Berz designed an alternate arrangement of the "Lurgi-MB
Filter" and marketed it through Gesellschaft fur Enstaubungsau-
lagen (GfE) of Munich, Germany.  In the U.S. it is built and
marketed by Rexnord in accordance with an exclusive license agree-
ment with GfE. It is built in modules for the treatment of large
gas volumes, with several modules arranged in parallel through a
common raw gas duct and a common clean gas duct.
     Each gravel bed filter module consists of a filter top
section containing two horizontal beds connected in parallel,
and a cyclone pre-cleaner located underneath.  The operation of
the system is illustrated in Figure 74.  The raw gas enters the
filter through an inlet chamber where immediate separation (set-
tling) of very coarse materials takes place.  From there, the gas
enters the cyclone separator where more coarse dust is separated
and removed through the discharge airlock at the outlet.
     The gas then rises from the cyclone through the vortex tube
and enters the filter chambers.  It passes from the top of the
horizontal filter beds to the bottom, so that the remaining fine

                                156

-------
            OPERATING PHASE
BACKFLUSH PHASE
12
                             14
                                      BACKFLUSH
                                         AIR
                      1.  INLET CHAMBER
                      2.  PRIMARY COLLECTOR (CYCLONE)
                      3.  DOUBLE TIPPING GATE  (DUST DISCHARGE)
                      4.  VORTEX TUBE
                      5.  FILTER CHAMBER
                      6.  GRAVEL BED
                      7.  SCREEN SUPPORT FOR BED
                      8.  CLEAN GAS COLLECTION CHAMBER
                      9.  EXHAUST PORT
                     10.  BACKWASH CONTROL VALVE
                     11.  BACKWASH AIR INLET
                     12.  VALVE CYLINDER
                     13.  STIRRING RAKE
                     14.  STIRRING RAKE MOTOR/REDUCERS
              Figure "74. Rexnord gravel  bed filter.
                                 157

-------
dust is deposited on the quartz grains and in the interstices of
the bed.  The cleaned gas flows through the clean gas collection
chamber and passes through the 3-way valve into the clean gas
duct.
     Cleaning of the filter unit can be initiated by means of a
pre-set sequencer or by automatically monitoring the pressure
differential across the filter bed.  During the cleaning cycle,
the unit is isolated from the gas stream by the 3-way valve.  Then
backwash air is admitted to the filter chamber in a reverse flow
direction.  It is either forced in by using a backwash air blower
or sucked in by negative pressure.  The backwash air loosens the
filter bed.
     During the cleaning process the rake-shaped double arm stir-
ring device is rotated by the geared motor.  This helps the dust
to be removed from the gravel and entrained by the backwash cleaning
operation.  The large agglomerated dust particles are carried
by the backwash air via the vortex tube into the precipitator,
where the velocity is reduced and the gas stream deflected so
that a large percentage of the dust is settled out.   The backwash
air, containing the remaining dust, mixes with the dust-laden
air in the raw gas duct and is then subjected to cleaning in the
remaining units of the filter.
Ducon Granular Bed Filter -
     The Ducon granular bed filter (Figure 75) consists of multiple
beds of sand stacked vertically within two perforated concentric
metal tubes.  The beds of sand grains rest on slotted inner screen
supports having opening dimensions slightly smaller than the dia-
meter of sand grains.  Above each bed there is an annular space
with an outer screen similar to the inner screen.  Thus, each
sand bed contained within the two concentric cylinders is supported
and caged by slotted metal retaining screens.  Figure 76 shows a
typical filter element.  The elements are supported from a clean
gas plenum within a housing.  The basic arrangement is exemplified
in the cutaway sketch of a typical 4-element unit shown in Figure 75.
                                158

-------
                          FILTER
                          ELEMENT
 INLET
CLEAN
GAS
OUTLET
                COLLECTED
                DUST
                OUTLET
   Figure 75.  Ducon granular bed filter.
                    159

-------
OUTER
SCREEN
                                          GRANULAR
                                          BED
                                          INNER SCREEN
          Figure 76.   Filter  element.
                          160

-------
     In operation, particle laden gas entering the dusty side of
the filter housing passes through the element's outer retaining
screens, flows downward through the inner cylinder into the clean
gas plenum to the vessel outlet nozzle.  The particles are deposited
on the surface and within the interstices of the sand beds.  To
clean the element, a small volume of compressed air is introduced
in a reverse pulse which induces blowback gas flow from the clean
gas plenum, sufficient to momentarily fluidize all the sand beds
within the element simultaneously. This blow-back gas flexes the
bed in fluidized expansion, expelling interstitially deposited
dusts, as well as any dusts on the bed surfaces, through the outer
screen.  The dust then falls into the collector hopper for eventual
removal.  Figures 77 and 78 illustrate the collection and cleaning
cycles.

INDUSTRIAL USERS AND PERFORMANCES
     Granular bed filters have been used in recent years on selected
sources.  While the use of granular beds has not been directed
at the control of fine particulates,  granular beds have been used
successfully on cement and lime kilns, asphalt dryers, and clinker
coolers.  Of the three commercially available granular bed filter
systems, only the Rexnord granular bed filter and the Combustion
Power Company's moving bed granular bed filter have industrial
installations.  Ducon granular bed filters have been used in some
pilot studies.
     We surveyed industrial granular bed filter users to obtain
performance data and to identify operating problems.   The following
is a summary of the results of this survey.
Rexnord Granular Bed Filter
     There are over thirty industrial users of the Rexnord granu-
lar bed filters.  Most of the installations are in Portland cement
plants to control particulate emissions from clinker coolers.
The average size of the dust particles from clinker coolers is
relatively large, and the granular bed adequately meets the emis-
sion standards.  However plugging of the granule retaining screens
has occasionally caused problems.
                                161

-------
 DUST
 LADEN
 GAS

          — ,
CLEAN
 GAS
  Figure  77.   Collection cycle.
COLLECTED
DUST    4
           •• v
 A
 A
                     .
                     r-'i
              PURGE
               GAS
           LUIDIZED
          GRANULES
 Figure 78.   Cleaning Cycle
                 162

-------
     There is one unit which is installed in a steel  sinter plant
to control the emissions from a windbox exhaust.  The  installation
consists of 24 modules with a total capacity of 4,400  Nm3/min
(240,000 ACFM @ 300°F).  The bed is packed with 0.1 cm diameter
gravel to a depth of 9 cm.  The unit was started up in February
1976 and encountered several operational difficulties.  Design
modifications were required.  Among the operational difficulties
encountered were:
     1. The gravel medium developed growth problems;   i.e.,
collected dust adhered to granules and could not be removed.
     2. Downcomer valves for dirty gas had to be modified.
     3. Backflush fans were replaced by larger fans.
     4. Heavier gauge bed support screens had to be substituted
for the originals.
     5. Backflush air preheater burners have malfunctioned
repeatedly.
     Some users supplied cost data.  Table 20 summarizes the cost
data available.
     McCain (1976) conducted a performance test on a Rexnord
granular bed installed in a Portland cement plant.  Samples were
taken simultaneously at the filter inlet and outlet with cascade
impactors.  Particle size distributions and grade efficiencies
were calculated from the impactor data.
     The particles were found to have a mass median diameter of
about 200 ymA.  The overall collection efficiency was found to
be from 99.3 to 99.7%.  The system pressure drop ranged from 9.6
to 14 cm W.C. (3.8 to 5.5 in. W.C.).  The system energy usage
during the tests was approximately 1,780 joules/Nm3 (47.7 in W.C.).
Figure 79 shows the grade efficiency curves for three sampling
runs reported by McCain for three cascade impactor runs.  Table 21
lists the operational conditions for the granular bed filter
during the test period.
Combustion Power Company's Moving Bed Granular Bed Filter
     Four units have been on line.  All installations are on
hog-mill waste combustors.  A prototype unit was installed at

                                163

-------
TABLE 20- SUMMARY OF REXNORD GRAVEL BED USERS
Plant/
Date
Install.
A
6/73
B
4/75
C
1974
Gas
Capacity
Am /min
5,100
(180 M ACFM)
3,820
(135 M ACFM)
4,020
(142 M ACFM)
Gas
Temperature
°C
126
(258°F)
163
(325°F)
182
(360°F)
Pressure
Drop
cm W.C.
36
(14"W.C.)
19
(7.5"W.C.)
13
(5"W.C.)
Capital
Cost
$
750,000
2,500,000
1,400,000
Annual
Power
Cost
$
78,000
	
8,000
Annual
Maintenance
Cost
$
	
6,000
4,000

-------
     i.o r
o
I—I
H
U
2:

     o.i-
H
W
u
l-l
H
o;
    0.01
        0.1
                     I   I
I     I
I  i i 1 I
I    I    I  I  I  I I  1
     1.0
                  AERODYNAMIC PARTICLE DIAMETER, ymA
                                                           10
 Figure 79.  Experimental grade efficiency curve of a Rexnord

             gravel bed filter (McCain, 1976),
                               165

-------
TABLE 21.  DESIGN SPECIFICATIONS OF THE SYSTEM
           AS TESTED
Inlet Volume Flow:  2,266 ACM/min at 204°C
                    (80,000 ACFM at 400°F)
Backflush Volume Flow:  317 ACM/min at 66°C
                        (11,200 ACFM at 150°F)
Pressure Drop:  25.3 cm W.C. (10 in. W.C.)
Gravel Size: 4 mm (5/32 in.) x no. 6 mesh
Bed Depth:  11.4 cm (4 1/2 in.)
Bed Area:  3.72 m2/bed (40 £t2/bed)
           (For a total of 59.5 m2 of bed area with
            52 m2 actively filtering in normal operation)
                      166

-------
the Weyerhaeuser Company, Snoqualmie Falls, Washington  Plant.
This unit was single down-flowing sand annulus  2.6 m  (8.5  ft)
O.D. and about 1.8 m (6 ft) I.D. with an effective filtering
height of 4.9 cm (16 ft).  The surface area was calculated at
34 m2 (365 ft2).  The granules were angular in  shape  and ranged
from 3.2 mm to 6.4 mm (1/8 in. to 1/4 in.) in average diameter.
The granules moved downward through the annulus in gravity flow
at a bulk velocity of 61 to 122 cm/hr (2 to 4 ft/hr).   The  granule
inventory was of the order of 36 metric tons though only about
18 metric tons were in the region exposed to gas flow.  The
manufacturer suggested capacity was 1,133 Am3/min (40,000  ACFM).
     Hood (1976) reported the performance test data on  this unit.
The experimental penetration curves are shown in Figure 80  for
three cascade impactor sampling runs.  As can be seen,  the unit
performed at an efficiency of from 75 to 95% in the removal of
particulates of 2 vim in diameter, at an efficiency of 70 to 90%
for particulate removal of 1 ym in diameter, and at an  efficiency
of 65 to 80% for particles 0.5 ym in diameter.  The efficiency of
the unit was low for the removal of particles of 0.25 ym in dia-
meter.  The data indicated that in the removal of particles in
the lower size ranges the unit performed at an efficiency of
between 45 and 65%.
     Under the sponsorship of the Department of Energy, Combustion
Power Company conducted parametric studies on their GBF system
to correlate the collection efficiency of the GBF with mechanical
and process parameters. Parameters studied included superficial
gas velocity, dust loading, particle size distribution, granule
diameter, granule circulation rate, bed thickness and length of
bed.
     The GBF was a pilot unit and was operated at ambient  tempera-
ture.  Figure 81 shows the system.  The granular bed material
(alumina) flows downward between two concentric cylinders.  The
gas passes through the bed and is filtered by the granules.  The
granules are recycled pneumatically and the collected dust par-
ticles are disengaged from the granules and sent to a conventional
baghouse.
                                167

-------
  1.0
  0.1
OS
H
tu
2
W
O-.
  0.01
       0.1
                          1  I  I I
               1.0

      PARTICLE DIAMETER, ym
10
   Figure  80
Experimental penetration curves for CPC  dry
scrubber (Hood,  1976).
                           168

-------
GAS
IMLET
          GAS
        OUTLET
                                            DISENGAGEMENT
                                            VESSEL
                                             FLUIDIZED BED
                                             FLUIDIZING  AIR
                                             MEDIA  RETURN PIPE
FRONT PANEL
FILTER PANEL
OUTLET PANEL
                                              TRANSPORT PIPE
                                         .	MEDIA  OUTLET PIPE
                                                  TRANSPORT AIR
                           EJECTOR
                             AIR

           Figure 81.   Continuous moving bed GBF.
                               169

-------
     The performance of this device has been reported by Guillory
(1978), Wade, et al . (1978) and Wigton and Wade (1978).  Test
parameters for the nominal, thick bed, thin bed, short bed, and
small collector granule configurations are listed in Table 22.
Particle concentrations ranged from 0-46 to 4.6 g/Nm3 (0.2 to
2.0 gr/SCF) .   Different test dusts were used in order to vary
the mass median particle diameter from 3 to 10 ymA.  The super-
ficial gas velocity was varied from 20 to 80 cm/s (40 to 160 ft/min)
The granule flow rate was varied to 0.4 to 1.6 kg of granule/kg of
air.  Pressure drop ranged from 1.2 to 5.7 kPa (5 to 23 in W.C.).
     Fractional efficiency curves are shown in Figures 82 through
86.   The overall penetrations are correlated with the pressure
drop function and are tabulated in Table 23.  The pressure drop
function was defined as:
                             AP C
                        0 = - V.
                             ur M
                              G  m
where   0 = pressure drop function
       AP = pressure drop, cm W.C.
       UQ = superficial gas velocity, cm/s
       M  = media rate, kg granules/kg air

     In general the CPC moving bed filter was found to be capable
of particle removal efficiencies in excess of 98% for particles
in the 1 to 10 ymA diameter range.  Submicron particles were
collected at an efficiency in excess of 90% in cases with high
velocities, high loadings, and low media rates.  Beds with larger
thickness to granule diameter ratios were most effective in the
capture and retention of particles in the 2 to 5 ymA diameter
range.  Also, intermittent media movement was shown to improve
efficiency by a few percent.  No cost data were available.
Ducon Granular Bed Filter
     There is no current industrial user of the Ducon granular
bed filter.  However, one test unit was installed in a refinery
                                170

-------
             TABLE 22.  TEST PARAMETERS FOR CPC MOVING BED
                        FILTER  (from Wade, et al.,  1978)

                         Granule          Active Bed    Bed Thickness
Configuration         Diameter, mm        Length, cm    	cm	

Nominal                     2                134.6           20.3
Thick Bed                   2                134.6           40.6
Thin Bed                    2                134.6           10.2
Short Bed                   2                 67.3           20.3
Small Medium               0.8               134.6           20.3
             TABLE 23.  CPC MOVING BED FILTER OVERALL
                        PENETRATION CORRELATION
Configuration                     Correlation
Nominal                         Ft" = 0.06 0"
Thick Bed                       Ft = 0.0175  0~°'7
Thin Bed                        Ft = 0.091 0~°'*7
Short Bed                       Ft" = 0.0425 0~°>S
Small Medium                    Ft = 0.0259 0~°'59
                                171

-------
tsJ
             0.5
       o    0.1
            0.05
           0.01
                                    II1I   I  I I  \-
                   I  I  I I I
               0.5       1                    5        10

                    AERODYNAMIC PARTICLE DIAMETER, gmA

       Figure 82. Experimental grade penetration (CPC data
                  for nominal configuration).
                                                                               1.0
                                                                               0.5
                                                                               0.1
                                                                              0.05
                                                                               0.01
             TT
          I  I I 11
I     I    I   I  I  I  I  I
      0.5       1                     5       10

          AERODYNAMIC PARTICLE DIAMETER, umA

Figure 83. Experimental grade penetration (CPC data
            for thick bed configurations).

-------
to control the particulate emission from the regenerator of a fluid
catalytic cracking (FCC) unit.  The unit had four filter elements;
each had 14 beds.  The bed was a 6.3 cm (2.5 in.) deep bed of 760
Vim sand.  The face velocity was between 15 and 45 cm/s (0.5 and
1.5 ft/s).  Gas temperature at the inlet was 371 to 482°C (700 to
900°F).
     The overall collection efficiency reported by Kalen and
Zenz (1973) was 85 to 98% for particles with a mass median diameter
of 35 ym and a geometric standard deviation of about 4.  No grade
penetration or efficiency curves were given by them.
     The grade penetration curve was calculated for the Ducon
granular bed filter based on the information provided by Kalen
and Zenz (1973).  Figure 87 shows the results.  The shaded area
is the range of calculated points.
     A high temperature and pressure design of the Ducon filter
was tested at the Exxon miniplant (Hoke, et al. (1978).  The
performance data for all runs through November, 1977 are listed
in Table 24.  The efficiencies are based on air inlet concentra-
tion of 2.3 g/Nm3 (1.0 gr/SCF) which is the average for the emis-
sions from the secondary cyclone.  The lowest demonstrated parti-
culate outlet concentration was 68.6 mg/Nm3 (0.03 gr/SCF)  however,
they were unable to maintain this level of performance for more
than a few hours of operation.  At times the filtration efficiency
was very poor and the outlet particulate concentration was as
high as 700 to 1,200 mg/Nm3 (0.3 to 0.5 gr/SCF).  Fractional
efficiency data are presented in Figure 88.

EVALUATION
     Granular bed filters perform in many respects in a manner
analogous to fiber filtration systems.  The major difference
appears to be the size differences between the fiber used in
fabric filters and the granules in granular bed filters.
                                175

-------
     1.0
    0.05
o
3
w
    0.1
    0.05
    0.01
0.5       1
                                  5         10

                          PARTICLE DIAMETER, u
                                                      i     i   i
                                                        15
     Figure  87 .   Fractional penetration curve  for Ducon
                   granular bed.
                                                                                   1.0
                                                                                  0.5
                                                                                  0.1
                                                                                  0.05
                                                                                  0.01
                                                                                      0.1
                                                                                  Figure
                                                                                       I    I    I   I  I I I  I I
                                                                                                                          I    I   I  I  I I  I
                                                                                                               UG = 45 cm/s
                                                                                                               dc = 0.04 cm
                                                                                                                Z = 3.8 cm
                                                                                               I     I   I   I  I  I I  II
                                                                                                                  A    I   I  I  1 J
0.5       1

PARTICLE DIAMETER,
                                                                                                                                     10
                                                                                    Fractional penetration for Ducon GBF (from Bertrand, et
                                                                                    al., 1977).

-------
           TABLE 24.   GRANULAR BED FILTER PERFORMANCE
                      (FROM BERTRAND, ET AL., 1977)
                   Outlet Concentration
Run Number

54
57
59 (Sample 1)
59 (Sample 2)
59 (Sample 3)
61
62.1
62.3
63 (Sample 1)
63 (Sample 2)
63 (Sample 3)
64 (Sample 1)
64 (Sample 2)
64 (Sample 3)
65 (Sample 1)
65 (Sample 2)
66
gr/SCF
0.69
0.04-0.08
0.08
0.28
0.54
0.46
0.03
0.21
0.05
0.07
0.12
0.28
0.29
0.27
0.05
0.06
0.06
g/m3
1.57
0.09-0.18
0.18
0.64
1.23
1.05
0.07
0.48
0.11
0.16
0.27
0.64
0.66
0.61
0.11
0.14
0.14
Collection Efficiency*

           31.0
        92.0-96.0
           92.0
           72.0
           46.0
           54.0
           97.0
           79.0
           95.0
           93.0
           88.0
           72.0
           71.0
           73.0
           95.0
           94.0
           94.0
* Based on a 2.3  g/Nm3(1.0  gr/SCF)  inlet  concentration
                                  177

-------
     Generally, for low to medium temperature applications, granu-
lar bed filters are not economically competitive with fabric fil-
ters even though the gas flow capacity of granular bed filters
is much higher than that of fabric filters.  Because granular
bed filters are not mass produced and the weight of the bed re-
quires special support, the cost of granular bed filters is much
higher than that of fabric filters.  Also, granular bed filters
generally have lower efficiencies than fabric filters.
     The best potential applications of granular beds as devices
to control particle emissions are in situations which require
the control of effluents under both corrosive and high tempera-
ture conditions.  GBFs are most successful in controlling emissions
of particulates that agglomerate easily as in the case of cement
dust.
Cleaning Methods
     During the operation of a granular bed filter, d.ust deposits
in the interstices of the bed and on the surface.  It is necessary
to clean the dust from the bed to prevent it from saturating
the bed and causing high pressure drop.
     Depending on bed structure, different cleaning methods are
used.  Fixed bed GBFs, such as Ducon GBF and Rexnord GBF,
apply a reverse gas flow to blow the dust out.  In the continu-
ously moving bed and intermittently moving bed GBFs, the bed
is cleaned by removing both the granules and the dust from the
filter and repacking the filter with clean granules.
     The advantages and disadvantages of present GBF systems
are discussed in the following paragraphs.
Fixed Bed GBFs
     The fixed bed GBFs are usually horizontal beds.  Beds are
cleaned by reverse gas flow.  The reverse flushing gas may be
used to fluidize  the bed and elutriate the dust deposits as
in the case of  the Ducon granular bed system.
     The advantage  of fixed bed GBFs is that they require
no granule recirculation system  and therefore have a lower opera-
ting cost.
                              178

-------
     The disadvantages are:
     1. Leakage or bypassing;
     2. Plugging of retaining grids, and loss of granules
        when retaining grids are not used;
     3. Efficiency decreases with time (particle seepage); and
     4. Ineffective bed cleaning.
Leakage or Bypassing -
     Dust loosened in the cleaning cycle may be carried directly
to the stack when the bed is put back on line.  This problem has
been described by Kalen and Zenz (1973) with the Ducon GBF and
by McCain (1976) with the Rexnord GBF.
Plugging of Retaining Grids -
     The collection efficiency of the granular bed can be in-
creased by using finer grades of granules.  However, grids with
openings smaller than the granules need to be used.  This in-
creases the tendency to accumulate dust deposits on the grid
and eventually causes plugging.  This problem was experienced by
Rexnord GBF users and by Exxon Research and Engineering Company
with the Ducon granular bed installed at their pressurized fluid-
ized bed combustor miniplant.  At Exxon, the inlet retaining grid
of the Ducon granular bed filter plugged quickly even when the
inlet grid was a 10 mesh screen.  The deposited dust on the screen
could not be blown out by reverse gas flow.  Shutdown of the plant
was required to remove the dust manually.
     In solving this plugging problem, Exxon has eliminated the
inlet retaining grid by increasing the free board above the bed
(Bertrand, et al., 1977).  However, this arrangement causes granules
to be blown out and lost from the filter beds.  In subsequent
tests Exxon lowered the blow back gas velocity.  Some indications
of particle buildup in the filter beds at the low blow back velo-
cities were noted.  In one run about 35% of the bed was fly ash
which was not blown out of the bed.  It was also found that if
the filter beds were overloaded with fine particles, the particles
tended to move to the top of the filter bed during blow back where
they prilled, forming larger spheres which could not be removed

                               179

-------
except at much higher blow back velocities. Exxon has  tried  using
heavier granules, however, some of the material was still lost
during blow back.
Efficiency Decreases with Time -
     For non-agglomerating dust, the blow back cleaning technique
will decrease the collection efficiency of the bed with each
cleaning cycle.  During blow back, collected dust redisperses
into the gas stream.  However, the fine particles will remain in
suspension.  This will increase the particle loading of fine par-
ticles in the gas stream.  Since the blow back gas is cleaned by
other beds and the collection efficiency of the bed is less than
100%, the fine particle penetration will progressively increase
with each cleaning cycle, (i.e., the efficiency of the bed goes
down).  Exxon (Bertrand, et al. 1977) has noticed this decrease
in efficiency with time.  In one test by Exxon, the filtration
efficiency was fairly high at the start of the run.  The outlet
particulate loading was about 0.12 g/m3 (0.05 g/SCF) .  The outlet
loading increased somewhat during the run to 0.17 g/m3 (0.07 g/SCF)
and finally to 0.25 g/m3 (0.1 g/SCF) at the end of the 12 hour test
Ineffective Bed Cleaning -
     Cleaning by reverse gas flow is not an effective cleaning
method especially when the cleaning duration is short.  The clean-
ing efficiency is even lower when particles are charged.
     Generally, an adhesive force exists between the collected
dust and the granules.  To separate the collected dust from
granules by reverse gas flow, the aerodynamic drag force must be
higher than the adhesive force between the dust and granules.  For
those particles which have a strong adhesive force, as in the case
of charged particles, dust will not be removed completely from the
granules by the reverse gas flow alone.  Thus, dust will gradually
accumulate in the bed.  In testing the Ducon GBF, .Exxon (Bertrand
et al., 1977) discovered a large accumulation of fly ash in the
bed, uniformly distributed through the filter medium.
                                180

-------
     The accumulation of dust in the filter bed has some signi-
ficant consequences.  The dust can approach the outlet section of
the filter and plug the retaining screens.  Once the dust reaches
the retaining screen, it could be entrained and blown out of the
filter bed during the filtration step.
Continuously Moving Bed GBFs
     This method is normally limited to vertical panel filters.
Dust and granules are continuously removed at the bottom of the
GBF.  The advantage of this type of bed structure is that the
collected dust and granules are separated outside the GBF.  This
method does not increase the particle loading in the gas stream
and the cleaning is more effective compared to reverse gas flow.
     The disadvantages are:
     1. Costly solid handling system
     2. Solids distribution
     3. Solid flow
     4. Particle reentrainment
     5. Erosion
     6. Heat loss
Solid Handling -
     A costly granule circulation system and dust/granule separa-
tion system is required.  Under low to medium gas temperature
environments,  collected dust and granules may be separated by
shaking.  Cleaned granules are transported to the top of the
panel by mechanical means.  In high temperature and pressure
(HTP) applications, these methods are not feasible.  Other
methods of transport and separation are required.
     Under HTP conditions, granules could be transported by
pneumatic means or some improved mechanical means.  Depending
on the mass flow rate ratio, pneumatic transport may require a
large quantity of compressed gas.  This will increase the capi-
tal cost and operating cost.
     Dust and granules under HTP conditions are difficult to
separate by mechanical shaking because of the requirement for a
suitable metal bellows or seal for the moving mechanism.  It

                               181

-------
may be possible to separate dust and granules by sending  them
through a baffled "rattler."  Dust would be shaken off the granules
when the granules hit the baffle.  The dust would then be elutriated
out of the rattler with air and later cleaned by means of a  conven-
tional separator such as a baghouse and cyclone at lower tempera-
ture.  This setup requires the investment of low temperature
secondary cleanup systems.
Solid Flow -
     The downward movement of the solids can create a dead zone
near the filter surfaces because some granules are retained by
the louvers.  These zones could eventually be saturated with dust
and lead to plugging of the louvers.
     Recently, Combustion Power Company made a design change to
alleviate this problem.  In the new design, gas flows from the
inside core radially outward through the bed.  The inlet louvers
are replaced with slotted panels.  During operation, solids flow
downward as well as spill through the slots.  This design eliminates
the dead zone.
Particle Reentrainment -
     The grinding of the granules due to the relative motion of
the filter granules can dislodge the collected particles and
allow them to be reentrained into the gas stream.  The reentrain-
ment rate depends on the granule recirculation rate and filtration
gas velocity.  At high recirculation rate, collected particles
are easier to dislodge.  At too low a recirculation rate, the bed
may be saturated with collected dust.  Depending on the inlet
particle loading, there exists an optimal granule recirculation
rate.  Under the sponsorship of the Department of Energy, CPC
conducted a parametric test (Wade, 1977, Guillory, 1977) on their
GBF.  The granule recirculation rate was one of the parameters
CPC studied.  They showed that low recirculation rate and inter-
mittent media movement improved collection efficiency by a few
percent.  This improvement could be attributed to a lower particle
reentrainment.
                               182

-------
Solids Distribution -
     Solids distribution to the filter panels may present some
difficulties.  It is not easy to distribute the solids evenly
to the panels.
Erosion -
     Solids retaining elements will be subjected to erosion by
the moving granular bed.  Selection of materials that can resist
erosion in HTP conditions becomes a problem.  Under the sponsor-
ship of ERDA, Combustion Power Company designed and constructed
a full scale filter to treat the total gas flow from the CPU-400
fluidized bed combustor at 1,000 kg/min (2,200 Ib/min).  The gas
temperature was at 704°C (1,300°F).  The retaining louvers were
made of RS 330 steel.  During a shakedown test in December 1975,
the louvers had a structural failure in about twenty hours of
operation.
Heat Loss-
     Granules are withdrawn from the bed for cleaning.  To keep
the granules hot, substantial energy may be required for HTP
conditions.  If the granules are recirculated by pneumatic means
it is necessary to pre-heat the transport air to minimize heat
loss from granules.
Intermittently Moving Bed
     Intermittent movement is normally limited to vertical panel
filters.  The granules are intermittently removed in a cross-
flow arrangement, as in CCNY's panel bed.  The advantages of this
type of bed structure are external granule/dust separation and
minimum disturbance to the rooting cake.  A rooting cake is the
foundation on which the surface cake is formed.  The surface cake
is formed readily without disturbing the rooting cake and filtra-
tion efficiency is higher.
     The disadvantages are the same as discussed earlier for the
continuously moving bed.  In addition,  the  intermittently moving
bed suffers the following disadvantages.
     1. Low gas capacity
     2. High operating cost and heat loss

                               183

-------
Low Gas Capacity -
     During cleaning, about two to three layers of granules are
removed from the bed.  To prevent the dust from being carried
deep into the bed by the gas, the filtration velocity should be
kept as low as possible to reduce the aerodynamic drag force.
CCNY usually operates the panel bed filter at about 15 cm/s
(30 ft/min).  This velocity is about one third the velocity used
in the fixed bed and continuously moving bed GBFs.  Thus, more
filtration area is required.
High Operating Cost and Heat Loss -
     To prevent the dust from penetrating deep into the bed, sur-
face layers are ejected frequently.  Depending on dust loading,
the surface layer has been removed as frequently as every 30
seconds in the operation of the CCNY panel bed GBF.  At 30
second "puff-back" frequency, the volume of the reverse gas flow
is about 1% of the total volume of gas treated.  The reverse
gas flow has a pressure of about 140 kPa (20 psi) higher than the
pressure of the gas to be treated.  The costs to operate the com-
pressor which supplies the reverse gas flow may be substantial
if the gas is to be treated at high pressure.
     This cleaning method has another drawback.  The "puff-back"
gas will have to be pre-heated in order to prevent significant
heat loss.  To maintain the gas temperature by pre-heating the
puff-back gas, the energy requirement is high.
                               184

-------
                             SECTION 8
                 POTENTIAL FOR HTP APPLICATIONS

     Fluidized bed coal combustion and low-BTU coal gasification
are among the advanced energy processes which require high tempera-
ture and high pressure (HTP) particulate cleanup. In addition, HTP
cleanup might be required for other high temperature and/or high
pressure processes as reported by Parker and Calvert (1977).
     The suitability of granular bed filters (GBF) for controlling
particulate emissions from advanced energy processes is not limited
by the gas temperature and pressure.  By properly selecting the
adequate granules and structural materials, the granular bed filters
should be capable of operating at any temperatures and pressures
encountered in advanced energy processes.

CLEANUP REQUIREMENTS
     In order to evaluate the potential of GBFs for HTP cleanup,
it is necessary to consider the following:
     1. The cleanup requirements.
     20 The performance characteristics of GBFs.
     30 The particulate size distribution and concentration in the
        inlet gas.
     4o Capital and operating cost estimates for GBFs.
     5o Permissible costs for HTP gas cleanup.
     The points listed above are discussed in this section,  A
summary and conclusions section is given at the end of this section.
     Parker and Calvert (1977) have reviewed and evaluated the
cleanup requirements for various processes.  The conditions for HTP
particle collection are summarized in Table 25.   As can be seen
from Table 25 gas temperatures range up to 1,100°C (2,000 F) and
pressures range up to 70 atm.  Granules, such as quartz sand and
ceramics', can handle these extreme conditions.
     The Ducon GBF, Combustion Power Company moving bed filter,
and the CCNY panel bed filter, all can be designed to be operated

                                185

-------
          TABLE 25.  CONDITIONS  FOR HIGH TEMPERATURE AND PRESSURE PARTICULATE COLLECTION
        PROCESS
TEMPERATURE
    °C
PRESSURE
  atm
TYPICAL GAS COMPOSITION
        mol %
EXPECTED
PARTICULATE
COMPOSITION
00
     Open cycle
     coal-fired
     gas turbine

     Fluidized
     bed coal
     combustion

     Coal gasifi-
     cation

         02 blown
         Air blown
     FCC regener-
     ator
     Metallurgical
     furnaces

     MHD power
     generation
 650-1,000
 800-900
 150-1,100
 300-800
 250-1,000
 300-800
  4-10
 -1-20
 -1-70
 -1-3
 ~1
831 N2, 15% C02 ,  2% 02, H20,
SOx, NOX, CO, and gaseous
hydrocarbons

80% N2, 10% C02>  6% 02,
4% H20, + S02, NO, CO
30% H2, 25% CO, 15% C02,
20% H20, 3% CH,,, H2S, N2

50% N2, 12% H2, 20% CO,
10% H20, 6% C02, + CHu,
H2S

68% N2, 5% CO, 31 02,
8% C02, 16% H20, + NOX,
SOX, NH3, HCN, aldehydes,
hydrocarbons

N2, C02, 02
coal ash, unburnt
carbon
60 wt % ash, 30%
unburnt carbon,
10% sorbent

ash, unburnt
carbon, sorbent,
possibly tar
catalyst dust
depends on cata-
lyst type, commonly
silica and alumina

very fine metal
fume

K2COs seed par-
ticles

-------
at HTP.  The Rexnord GBF can be operated  at  high  temperature
but not at high pressure.  Rexnord does not  recommend  that  their
GBF be used at HTP.  They have designed a new  type  of  filter  for
these conditions.  The new design is  similar to the CPC moving
bed GBF.
     The use of granular bed filters  for  HTP applications is
limited by the particulate and gaseous pollutant  removal effi-
ciencies.  Particulate cleanup requirements  for HTP processes
vary depending on the intended use of the gas.  If  it  is to be
vented, the gas must be cleaned sufficiently to meet the emission
standards.  Current new source performance standards are 43 mg/MJ
(0.1 lb/106 BTU), however, a stricter standard of 13 mg/MJ  (0.03
lb/106 BTU) has been proposed.
     If the hot gas is to be expanded through a gas turbine, then
the gas must meet the turbine requirement for cleanliness.  A gas
containing dust particles can severely erode and  corrode turbine
blades and other internal components.  Also, deposition of dust
particles on the turbine blades can impair the aerodynamic
performance of the turbine.
     A large number of research investigations have been reported
which deal with turbine blade erosion and deposition problems.
Much of this work was done in connection  with military gas tur-
bines for helicopter and tracked-ground vehicle engines.  Similar
research has also been conducted with industrial  gas turbines.
Generally, it is believed that large  particles (over 2-5 ym
diameter) cause severe erosion damage and must be removed.  Par-
ticles smaller than 1-2 ym diameter cause much less erosion
damage.  However, there is a scarcity of  data concerning the tol-
erance of turbines for fine particles.
     From the available data on turbine tolerances for particulate
matter, it appears that effectively all particles larger than
about 2 ym must be removed from the gas.  It has  been  suggested
(Westinghouse, 1974) that a mass loading  of  370 mg/Nm3  (0.15
gr/SCF) for particles smaller than 2  ym would allow a  satisfactory
turbine life.   The particulate removal requirements  imposed by
                                187

-------
the gas turbine limitations may not be as stringent as the emission
standard.  If there were a sufficient loading of particulate
smaller than 2 urn, it would be possible for the gas to be cleaned
sufficiently to protect the turbine while still exceeding the
emissions regulation of 0.1 lb/106 BTU (approximately 0.05 gr/SCF).
     Recently Sverdrup and Archer (1977)  proposed that to protect
the turbine, the particulate concentration should be no more than
5 mg/Nm3 (0.002 gr/SCF) and there should be no particles larger
than 6 ym in diameter.

PREDICTED GBF PERFORMANCE
     Figure 89 gives the predicted granular bed performance
for several conditions.  It shows the effects of temperature and
pressure on particle penetration through a granular bed filter.
The predictions were made for a bed packed with 400 urn diameter
granules to a depth of 3.8 cm (1.5 in.)-   Other assumptions were:
an approaching gas velocity of 45 cm/s, no surface cake
formed,  the collected dust was uniformly distributed in the
bed, the average void fraction of the bed between cleaning cycles
was 0.25, and the particle density was 1.5 g/cm3.
     As concluded in a report by Calvert  and Parker (1977), high
temperature and pressure particle collection is more difficult
than at low temperatures.  The predicted penetration for the gran-
ular bed filter at 875°C and 10 atm is much higher than that for
ambient conditions.
     If the particle size distribution of a source were known, it
would be possible to predict whether a granular bed filter would
be able to meet various cleanup requirements.  The equation relat-
ing the fractional penetration (for a specific particle diameter)
to the overall penetration is:
                    Pt =    Ptd £(dp)d(dp)
                               188

-------
    1.0
z
o
PL,

 •X


O
I—I
H
W
2
W
OH
1-J
u
I— I
H
OS
<
P-
     0. 5
     0.1
0.05
    0.01
        0.1
                                GRANULE DIAMETER: 400v»m

                                BED DEPTH:  3.8 cm

                                   = 45 cm/s

                                   = 1.5 g/cm3

                                   = 0.25
                                 10 a tin
                                 870°C
                                  1  atm
                                  870°C
                           iiii
                      0.5     1.0

                   PARTICLE DIAMETER, ym
                                                   I I  1
                Figure  89.   Predicted GBF performance
                                                           10
                                189

-------
where   Ft = overall penetration, percent or fraction
       Pt, = penetration for particles with diameter, d  , fraction
     f(d ) = particle size frequency distribution
        d  = particle diameter, ym or cm

Particle Size and Concentration
     Information on particle size distribution and mass  loading
of particles leaving HTP sources is scarce.  The best available
data are those reported by Exxon Research and Engineering Company
(Hoke, 1976, 1977).  Exxon has a pressurized fluidized bed coal
combustor miniplant.  The gas leaves the combustor at a temperature
of about 870°C (1,600°F) and a pressure of about 10 atm0  First it
passes through a primary cyclone which removes larger particles
(including unburnt carbon) and recycles these particles  to the
combustor.  The gas leaves the primary cyclone and passes through
a secondary cyclone„  This removes more large particles  and reduces
the mass loading to the order of 2.5 g/Nm3 (1 gr/SCF)„  A Ducon
GBF is connected to the outlet of the secondary cyclone  to further
clean the gas stream,
     Hoke (1976) reported the particle size distribution at the
secondary cyclone exit.  It was obtained by sieve and Coulter
counter analysis.  The particles have a mass mean diameter of 8 ym
and geometric standard deviation of 2.7 (Figure  90).
     Overall collection efficiency was calculated graphically from
equation (97)for this size distribution and for a granular bed
packed to a depth of 3,8 cm with 400 ym diameter granules„  The
approach velocity of the gas was assumed to be 45 cm/s0  The over-
all penetration was calculated to be 208% (collection efficiency
97.2%).  Therefore, the predicted emissions will be 0.07 g/Nm3
(0002& gr/SCF) and it is in compliance with the current  standard
for particulate emissions (001 lb/106 BTU or about 1014  g/Nm3)0
However, it will not meet the proposed new standard (0,03 lb/106
BTU or about 0.04 g/Nm3).
                                190

-------
    50
w
H
W
o
w
H
OS
    10
    1.0
   0.5
                                        JULY 1976
                                          DATA
                                      MAY 1977
                                        DATA

                   10      20    30   40   50   60    70

                       WEIGHT  PERCENT  UNDERSIZE,  %
                                        80
90
Figure  90
Particle size distributions from Exxon miniplant
fluidized bed coal combustors.  From monthly
report #77, July 1976 and report #86, May 1977.
                              191

-------
     As revealed in Figure 89 , the predicted penetration decreases
rapidly around 1 ym.  The predicted penetration for 2 ym particles
is effectively zero.  Therefore, there should be no particles big-
ger than 2 ym in the GBF exhaust.  The predicted loading of 0.07
g/Nm3 (0.028 gr/SCF) meets the turbine requirement suggested by
Westinghouse (1974); but, it does not meet the requirement proposed
by Sverdrup and Archer (1977).
     Hoke (1977) reported some new data on particle size distribu-
tion at the secondary cyclone exit.  The mass median diameter is
3.5 ym and the geometric standard deviation is 2.9.
     The overall penetration for this size distribution will be
4.99 (95.1% overall collection efficiency).  The emission will be
0.12 g/Nm3 (0.049 gr/SCF).  This still satisfies the current emis-
sion standard (approximately 0.06 gr/SCF) and also meets the tur-
bine requirement suggested by Westinghouse).
     From the above discussion, it appears that improvements in
GBF designs are required in order to improve  collection efficiency
and resolve operation problems.  However, theoretical collection
efficiencies appear to be sufficient to meet  the current emissions
regulations for particulates.  Performance may be satifactory for
protecting gas turbines, however, this will depend strongly on
the amount of submicron particles a turbine can tolerate.

GASEOUS POLLUTANTS
     In pressurized fluidized bed combustion, besides the emission
of particulates, gaseous pollutants such as S02 and alkali metal
compounds may also be present in the flue gas from the combustor.
The presence of S02 might not have any ill effect on the operation
of the gas turbine«,  The S02 may be removed from the gas stream
at HTP with a GBF with dolomite or alumina as granular material
or be removed by conventional means after the gas has been expanded
through the turbine.
     The presence of alkali metal compounds,  such as sodium and
potassium chlorides will contribute to hot corrosion of the gas
turbine.  Therefore, to protect the turbine,  the concentrations
                                192

-------
of alkali metal compounds in the flue gas have to be reduced to a
tolerable level.  A direct method for accomplishing this is by
controlling the combustion to minimize the evolution of alkali
metal compounds.  An alternative method  is to use GBFs with some
sort of sorbents as granular material to remove particulates and
alakli metal compounds simultaneously.
     Limited work has been done in this  area.  Swift  et al. (1977)
have shown that activated bauxite, which is a thermally treated
high alumina content natural bauxite ore, was efficient at removing
NaCl vapor at 900°C and atmospheric pressure.
     As with the case of particulate matter, the acceptable level
of alkali metal vapor in flue gas stream is not well established.
Therefore, it is impossible at this time to speculate whether the
GBFs are able to remove harmful vapors to acceptable levels.

PRELIMINARY COST ESTIMATE
     Granular bed filter technology is still in the developmental
stage and there are no proven designs to be used in HTP cleanup.
Many of the cost factors have not been established.  Therefore,
it is extremely difficult to do a detailed cost analysis.
     In the following section, the relative capital and operating
costs between a HTP fixed bed GBF, a HTP intermittently moving
bed, and a HTP moving bed system are presented.  The cost esti-
mates were performed for the GBF layouts presented in Figures 72,
81 and 91.    It should be noted that these designs might not be
the optimum.
Basis
     In this report, the estimate was based on the cleanup re-
quirement for one gas turbine.  The seventeen-stage axial flow
compressor is designed to develop a pressure ratio of approximately
10 to 1 while using 345 kg/s (761 Ib/s) air at IOS (International
Organization for Standardization) ambient conditions (298.33°K
dry bulb temperature with 60% relative humidity).  The four-stage
expander is designed to operate with a turbine inlet temperature
of 1,233°K (1,760°F), and under these conditions the net output
                                193

-------
has been calculated to be approximately 71 MW (Beecher, et al.
1976).  If the turbine inlet temperature were 1,143°K instead of
1,233°K and the turbine efficiency remained the same, the turbine
power output would be about 66 MW.  The flow rate of 760 Ib/s at
ambient conditions is approximately 6,700 m3/min at 870°C and 10
atm (237,000 ACFM at 1,600°F and 10 atm) and 7,220 m3/min at 960°C
and 10 atm (255,000 ACFM at 1,760°F and 10 atm).
     The following were used as the basis for the cost estimate:
     1. Inlet gas temperature:  870°C (1,600°F)
     2. Inlet gas pressure:  10 atm
     3. Outlet gas pressure:  1 atm
     4. Net turbine power output:  66 MW
     5. Gas volume flow rate: 6,700 Am3/min at 870°C
     6. Total power output for combined cycle plant:  355 MW

Capital Costs
     Since the GBF's are non-standard fabricated units, the esti-
mated fabricated cost was based on the costs of raw materials
and labor requirements.  Costs for design, administration, con-
tingency, engineering, etc. were not included.  Auxiliary equip-
ment was priced based on vendor's quotations.  Equipment cost was
bare module cost, not installed cost.  The cost for this study was
developed in terms of fourth quarter 1978 U.S. dollars.
Fixed Bed GBF -
     The fixed bed GBF is designed for a superficial gas velocity
of 40 cm/s (80 ft/min).  Total bed area required is 275 m2 (2,960
ft2).  Eight GBFs are required.  Each GBF has 48 filter elements
and each filter element has 16 beds.  Total bed area per GBF is
34.3m2(370 ft2).
     The GBF shell is a pressure vessel.  The dimensions are shown
in Figure 91 .  It is made of 3.8 cm (1.5 in.) thick carbon steel
lined with 15 cm (6 in.) thick refractory.  There is also an inner
of 0.32 cm  (1/8  in.)  thick  type-316  S.S.
     The filter  element consists  of  two perforated  concentric  tubes
made from type-316 S.S.  Beds of  granules are stacked vertically

                               194

-------
                           BACKWASH MANIFOLD
                           AND VALVES
                                 FILTER ELEMENT
Figure  91.     Fixed bed  GBF.
               195

-------
within the annulus.  The O.D. of the outer tube is 30 cm  (1  ft)
and the I.D. of the inner tube if 19 cm (7.5 in.).
     Material costs used in the estimates are: carbon steel
$1.00/kg ($0.45/lb); 316 S.S. $9.90/kg ($4.48/lb); castable  re-
fractory $0.74/kg  ($0.34/lb).  Fabrication labor costs are:
carbon steel, $1.00/kg ($0.45/lb); 316 S.S., $3.96/kg ($1.80/lb);
castable refractory, $0.74/kg ($0.34/lb).
     The fabricated costs of the GBF system are as follows:
     Pressure Vessel                         $960,000
     Filter Elements                          693,000
     Blowback Valves, Piping § Lock Hopper    925,000
     Ductwork (refractory lined)              247,000
     Compressor                               179,000
                                Total      $3,004,000

     The estimated GBF bare module cost is $448.36/Am3 ($12.68/AGFM).
     The compressor indicated is for the supply of reverse gas
flow during cleaning.  It has a capacity of about 102 m3/min
(3,600 SCFM) and the developed pressure is 1,388 kPa (200 psig).
Continuously Moving Bed GBF -
     The arrangement of the moving bed is shown in Figure 81.
The bed is a single downflowing annulus 2.6 m (8.5 ft) O.D.  and
2.3 m  (7.5 ft) I.D.  The height of the filtering surface  is  4.9
m  (16  ft).  The bed is placed in a 3.7 m (12 ft) diameter pressure
vessel.
     Each GBF can  handle a gas capacity of 1,130 Am3/min  (40,000
ACFM)  at a superficial gas velocity of 50 cm/s (100 ft/s).   There-
fore,  six GBFs are required to treat the gas for one gas  turbine.
     The pressure vessel consists of three layers.  The outside
shell  is 3.8 cm thick carbon steel.  The inner layer is 16 gauge
(1/8") 316 S.S.  In between there is a 15 cm (6 in.) thick layer
of refractory.  The material and labor costs to fabricate the
pressure vessel are the same as presented in the last section.
     The outer panel of the granular bed is a perforated wall and
the inner retainer is a slotted panel.  Since granules are
                                196

-------
continuously moving downward,  the bed  retaining  walls  should  be
made of erosion resisting materials; e.g.,  Hastelloy C,  RA  330,  and
RA 333.  The material cost used  in  this  cost  estimate  is $39.40/kg
($17.90/lb).  Fabrication labor  cost is  estimated  to be  $7.92/kg
($3.60/lb).
     Under HTP conditions, it  is impractical  to  separate dust and
granules by shaking apparatus  and to transport granules  by  mechani-
cal means.  Therefore, pneumatic transport  and solid/dust separa-
tion techniques were designed  into  the GBF  system.  It is assumed
that the weight of granules  recirculated  is equal  to the weight
of gas treated; i.e., granule  recirculation rate is 345  kg/s
(760 Ib/s) .  Granules are withdrawn from  the  bed by means of an
ejector and are transported  to a granule  holding tank above the
GBF.  The mass flow rate ratio of granules  to transport  air is
assumed to be 20 g/g.
     The granules flow from  the  granule holding tank by  gravity
into a baffled rattler section.  In the rattler section, the dust
is knocked off the granules  and  elutriated by an air stream which
is also used as the fluidizing air  to help the flow of granules.
The mass flow rate ratio of  granules to the fluidization air is
assumed to be 80 g/g.  All granule  transport  lines are carbon
steel pipe lined with refractory and erosion  resisting metal
liners.
     The fabricated cost of  the  GBF system is as follows:
     Pressure Vessel                       $  791,000
     Filter Retaining Walls  §  Stiffeners    2,156,000
     Hot Gas Piping                            52,000
     Granule Cleaning Ductwork               110,000
     Ductwork                                 185,000
     Compressor                               784,000
     Secondary Collector (incl. ductwork)     150,000
                                    Total  $4,228,000

     The estimated fabricated  cost  is about $631.04/Am3  ($17.84/
ACFM), which is 1411 higher  than that for the fixed bed  system.
     The cost of continuously  moving bed  systems can be  reduced
if a better method for granule recirculation  is found.   In  the
                                197

-------
estimate, low bulk density pneumatic transport is used for granule
transport.  Other transport methods such as dense-phase pneumatic
transport may be more cost efficient.
     Some testing is required to determine whether the dense-phase
vertical pneumatic transport of 1-2 mm solids is feasible.  If it
is feasible, the capital cost (compressor cost) and operating
cost will be lower than predicted in this estimate.
Intermittently Moving Bed GBF -
     One possible arrangement for the intermittently moving bed
GBF is shown in Figure  72.  Each square filtration module is 0.56
m x 0.56 m x 4.6 m (22 in. x 22 in. x 15 ft).  Active filtration
area per module is 5.57 m  (60 ft , each side is 1 ft x 15 ft).
     Filtration velocity is about 15 cm/s (30 ft/min).   Therefore,
total filtration area required is 21,970 m2 (236,500 ft2)  and the
number of filtration module required is 132.  The filter elements
are made of Hastelloy "C".  Since the panel is a louvered wall,
material requirement is 8 times the unit length of the  panel
(Wu, 1977).
     It is possible to place 17 modules in a pressure vessel 4.88
m (16 ft) in diameter.  Thus, eight GBFs are required for each gas
turbine.
     As in the case of the continuously moving bed GBF, granule
circulation and dust/granule separation are accomplished by pneu-
matic means.
     The cost data used in the estimate were the same as used
previously.  The estimated costs are as follows:
     Pressure Vessel                         $1,050,000
     Filter Elements                          5,186,000
     Ductwork                                   250,000
     Blowback Cleaning System                   132,000
     Compressor                                  '92,000
     Lock Hopper § Granule Cleaning System      720,000
                                Total        $7,430,000
                               198

-------
     The estimated  fabricated  cost  is  about $1,108.96/Am3  ($31.35/
ACFM) which  is  247%  higher  than that of the fixed bed system and
about 176% higher than  that of the  continuously  moving bed  system.
     The greatest cost  item is  that  of  the  filter element.   The
cost can be reduced  if  the  GBF  is run at  a  higher filtration velo-
city.  The filtration velocity  used  in  this  estimate  is that
recommended by Squires  (Wu,  1977).   Research work is  required
to determine whether it can be  run at higher filtration velocities.
Operating Power Costs
     The operating power costs  for the  GBF  systems described
above were estimated in this study and  include those  due to  the
following:
     1. Pressure drop across the bed.
     2. Power requirement to operate the  compressor which is used
either for cleaning  or  for  pneumatic transport of solids.
     3. Cost to heat the transport air  and cleaning air to the
bed temperature.
     4. Heat loss to the surroundings.
     Pressure drop and heat  losses cause  the gas temperature and
pressure at the turbine inlet to be  lower than it would be with-
out cleanup.  Lower  gas temperatures and pressures reduce  the tur-
bine power output which in  turn lowers  the revenue.
     The fixed bed operating power cost (not including deprecia-
tion) is the lowest  among the three  systems.  The operating power
costs of the continuously moving bed and  the intermittently moving
bed are about 7.4 times and  4 times  higher than the fixed  bed,
respectively.
     More than 60% of the moving bed operating cost is due to the
compressor for pneumatic transport.  The moving bed operating
cost can be lowered  if better solids transportation methods are
found.
     In the intermittently  moving bed system, the high operating
cost is also due to  the compressor.  However, the compressor is
not for solids transport as  in  the continuously moving bed system.
It is used for "puff-back"  air.  The "puff-back" air  requirement

                                199

-------
per puff is 0.28 m3 (10 ft3) for each square filter unit.  "Puff-
back" frequency depends on dust loading.  It is usually about one
puff every half minute.

PERMISSIBLE COSTS
     High temperature and pressure (HTP) particulate control is
an important factor in determining the economic feasibility of
many advanced energy processes.  The energy process which relies
most strongly on HTP gas cleanup is the pressurized fluidized bed
(PFB) combustion process.  The PFB process was described in detail
as part of the Energy Conversion Alternatives Study (EGAS) - a
cooperative effort of the Energy Research and Development Adminis-
tration (now the Department of Energy), the National Science
Foundation and the National Aeronautics and Space Administration.
     For the PFB process to be economically competitive it is
necessary to recover energy from the HTP effluent gas by expanding
it through a gas turbine.  To protect the turbine and not lose
energy, it is essential that the gas be cleaned with a minimum
loss of temperature and pressure.
     The General Electric and Westinghouse Phase II EGAS reports
(Brown  et al. 1976 and Beecher  et al. 1976) present detailed
designs and cost estimates for PFB boiler power plants.  The re-
sults are summarized in Table 26 (from Lewis Research Center, 1977)
The overall cost of electricity for the General Electric design
is presented in Figure 92  as a comparison with other advanced
processes considered in EGAS Phase II.  The reference steam cycle
is a conventional coal-fired boiler power plant with wet lime
stack gas scrubbers (Brown, 1976) .
     The advanced steam PFB designs assumed Ducon granular bed
filters would be suitable for HTP particle collection.  The EGAS
reports were not explicit as to what percentage of the cost of
electricity is attributable to the granular bed filters, although
about 20% of the capital investment was for the cleanup system.
Figure  92 shows that a cost difference of 5 to 6 mills/kWh exists
between the PFB with cleanup and the reference steam cycle.  We
                                200

-------
TABLE 26. SUMMARY OF EGAS PHASE II PERFORMANCE AND COST RESULTS
System and contractor
1 - AFB/steam
(General Electric)
2 - PFB/ steam
(General Electric)
3 - PFB/ steam
(Westinghouae)
4 - PFB/poLaeeium/eteam
(General Electric)
5 - AFB/closed-cycle gas
tu rbine/or ganic
(General Electric)
6 - Low-Btu gasifier/gas
turbine/ steam
(General Electric)
7 - Low-Btu gaslfler/gas
turbine/eteam
(Westinghouse)
8 - Semiclean-fuel-fired
gas turbine/steam
(Westinghouse)
9 - Scmiclean- fuel- fired
gas turbine/eteam
(General Electric)
10 - Coal/MHD/ steam
(General Electric)
11 - Low-Btu gaBifler/moltcn-
carbonate fuel cell/stoam
(United Technologies Corp.)
Net
power,
MW
814

904

679

996

476


585


786


874


847


1932

635


Efficiency, percent
Thermo-
dynamlc
43.9

41.3

42.3

47.8

50.1


44.2


48.5


53.6


52.7


54.0

53.6


Power-
plant
35.8

39.2

39.0

44.4

39.9


39.6


46.8


52.2


51.1


49.8

49.6


Over-
all
35.8

39.2

39.0

44.4

39.9


39.6


46.8


38.6


37.8


48.3

49.6


Capital cost,
$/kWe
EGAS
ground
rules
632

723

549

934

1232


771


614


329


418


720

593


Constant
mid- 1975
dollars
447

411

401

660

899


562


448


256


306


478

433


Cost of electricity,
mills/kW-hr
ECAS
ground
rules
31.7

34.1

28.1

39.9

49.3


35.1


29.1


26.0


29.5


31.8

28.9


Constant
mid- 197 5
dollars
25.8

27.4

23.5

31.2

38.6


28.6


23.9


23.7


25.9


24.1

23.9


                              201

-------
 Reference Steam Cycle
         COST OF ELECTRICITY (Mills/kWhl
       10      20      30       40
                               <-
                                                                         50
 Advanced Steam - AFB1814.3 MWe)
 Advanced Steam - PFB (903.8 MWe)
 Closed Gas Turbine:
    Organic Bottoming - AFB (474.1 MWe)
 Potassium Topping Cycle - PFB 1993.8 MWe>
 Open Cycle Gas Turbine Combined:
    Water Cooled - Semiclean Fuel (846.5 MWe)
 Open Cycle Gas Turbine Combined;
    Air Cooled - LBTU Gasifier (584.8 MWe)
 Own Cycle MHO (mi 2 MWe)

           Key (Contribution to Total):
               Capital I   I    Fuel
V////A    Maintenance and Operating I
Figure   92.  Summary  G.E.  Comparison  of Cycles  (Cost of
                Electricity at  an  assumed  capacity  factor
                of 65%).
                                    202

-------
estimate that the capital cost  for  cleanup  would  be  about 8.5
mills/kWh so that cost difference between the  PFB without cleanup
and the reference steam plant with  cleanup  is  about  14 mills/kWh.
Therefore, one can conservatively state  that a cost  on the order
of a few mills/kWh would make the PFB process  economically competi-
tive for further development.   There is  also the  advantage of fuel
conservation which is due to the higher  overall efficiency of the
PFB process as compared to a conventional power plant.

SUMMARY AND CONCLUSIONS
     Particulate matter removal from gas at HTP is desirable for
any process which involves the  use  of a  turbine on the HTP gas
stream.  The degree of particulate  cleanup  required  to protect
the turbine is uncertain.  Published estimates of turbine require-
ments range from a maximum 370  mg/Nm3 (0.15 gr/SCF)  for particles
smaller than 2 ym diameter (Westinghouse, 1974) to a maximum
loading of 5 mg/Nm3 (0.002 gr/SCF)  and no particles  larger  than
5 ym in diameter (Sverdrup and  Archer, 1977).
     Particle data obtained on  the  Exxon PFBC Miniplant range
from 2.5 g/Nm3 of fly ash with  d    = 9 ym and ag  = 2.7 to the
same loading with d   = 3.5 ym  and  a  =  2.9.  The A.P.T.  perfor-
mance model when applied to the larger particle size distribution
and the Ducon GBF in use on the Miniplant predicts outlet loadings
around 70 mg/Nm3 and no particles larger than  2 ym diameter,
which meets the Westinghouse turbine requirements and present
New Source Performance Standard (NSPS) (about  140 mg/Nm3) but
does not meet the Sverdrup and  Archer turbine  requirements or
the proposed NSPS (abour 40 mg/Nm3).
     For the smaller particle size  distribution the A.P.T. model
predicts that the Ducon GBF would give   an  outlet loading of
about 120 mg/Nm3, which is close to the present NSPS.
     A criterion for the permissible cost of HTP  particulate
removal can be estimated from the performance  and cost estimate
data available from the EGAS studies (Figure  92  ) .  The PFBC
plant is a good example because its feasibility may  depend on a

                               203

-------
satisfactory HTP cleanup system.  The power generating difference
between a PFBC and a conventional steam power plant is about
14 mills/kWh.  A gas cleanup cost on the order of a few mills/kWh
should be acceptable and even higher costs may be justifiable.
     Cost estimates were made for three GBF systems for HTP clean-
up of the combustion gas in a 355 MW combined cycle power plant.
The gas turbine provides 66 MW of the plant capacity.   Of the three
GBF systems, the fixed bed system has the lowest estimated capital
and operating costs.  The capital costs of the continuously moving
bed system and the intermittently moving bed system are predicted,
respectively, 141% and 2471 higher than the fixed bed system.
With regard to the operating power cost, the continuously moving
bed system and the intermittently moving bed system are about 7.4
times and 4 times higher than that of the fixed bed system.
     By assuming a cost factor of 4 in the estimation  of costs of
engineering, installation, site preparation, contractor's fee,
contingency, construction, and working capital cost, the capital
investment of a continuously moving bed system would be $16,912,000;
or $47.64/kWh of plant capacity.  The operating power  cost of the
continuously moving bed system is about 0.48 mills/kWh of plant
capacity.  For a 10 year life, depreciation charge is  about 0.55
mills/kWh of plant capacity.  Thus, the estimated total operating
cost would  be about  1 mill/kWh for the continuously  moving bed
system.  This cost figure is economically competitive  when com-
pared to the EGAS Phase II results.
     However, it should be noted that no granular bed  filter sys-
tems have been demonstrated to have high enough collection effi-
ciency and reliable performance to satisfy HTP gas cleanup re-
quirements.  More development and redesign are necessary.  Con-
sequently, any estimates of cost and process economics should be
considered highly speculative at this time.  Adequate  comparisons
between GBF system economics must be based on proven pilot plant
designs.
                               204

-------
                            SECTION 9
      RECOMMENDATIONS FOR FUTURE RESEARCH AND DEVELOPMENT

     The primary objective of evaluating the feasibility of granu-
lar bed filters (GBF) for the collection of particulates at high
temperature and pressure (HTP) has been achieved in this study.
It has been shown that GBFs are capable of operating at HTP.
Whether the GBF meets the cleanup requirements depends on the
application.  Present GBFs have the potential to meet the current
emission standards.  However, unless aided by other collection
mechanisms, the present designs are not likely to meet the proposed
NSPS for boilers or the turbine requirements proposed by Sverdrup
and Archer  (1977).  More research and development work is required
to improve the performance of GBFs.  Future research and develop-
ment work is needed in the following areas:
     1. Efficiency improvements.
     2. Bed cleaning methods
     3. Other potential problems, such as sintering of the
        granules, granule transport method, etc.

EFFICIENCY  IMPROVEMENT
     The present investigation has shown that existing GBFs may
not have a high enough collection efficiency for fine particles,
especially when operating at high temperature.  There are a few
studies reported in the literature which show that the collection
efficiency of the bed may be increased by:
     1. Electrostatic augmentation
     2. Cake filtration
                                205

-------
Electrostatic Augmentation
     The filtration efficiency can be enhanced by electro-
static augmentation.  If the filtration medium is immersed
in an electrostatic field, the dust particles will be driven
in a direction that tends to increase the probability of
impact between particles and the filter medium.  Limited
work has been done with electrostatically augmented granular
beds.  A.P.T. has carried out experimental work to study the
effect of charging the bed and/or particles on the collection
efficiency.  This work will be presented in a separate
report.
     These small-scale experiments have demonstrated that the
GBF collection efficiency can be improved greatly by charging
the bed or by charging both the bed and particles.  It is
possible for the GBF to achieve a 99+% collection efficiency
when an electrostatic field is imposed on the bed.
     For industrial applications, the capital cost of electro-
statically augmented GBFs will be higher than without augmenta-
tion.  However, the increase in capital outlay is compensated
by the reduction in operating cost.  The electrostatically
augmented GBF can be operated at much lower pressure drop by
using a shallow bed and the energy cost to maintain the electro-
static field is small  (5 x 10~6 mills/kWhe).
     The next steps in evaluating the electrostatically augmented
GBF should move toward a pilot-scale test on an actual source,
such as., a fluidized bed coal combustor.  More laboratory-scale
studies are required to refine the technology and establish
design criteria for the pilot plant.  Areas which need further
research include:
     1.  bed charging method.
     2.  electrode configurations.
     3.  bed structure.
     4.  HTP electrode insulation.
     5.  effects of polarity on efficiency.
     6.  cleaning methods

                               206

-------
     Most of these effects can be evaluated  in  the  laboratory
and the most promising combination can be tested  on  a pilot-
scale plant.
Cake Filtration
     There is good reason to believe that the GBF with a good
surface cake would have a much higher filtration  efficiency
than the original clean bed.  This would result from increased
particle collection by sieving.
     There is no published information on the efficiency of cake
filtration.  A small-scale laboratory research program could
generate the needed information on cake filtration.  One approach
could be to build a cake on the surface of the GBF and then
measure the collection efficiency using monodisperse particles.
The effects of cake thickness, pressure drop, particle size
distribution of the cake, and superficial gas velocity would be
studied.
     Further research is needed to determine the GBF operating
conditions which result in cake formation.  Whether or not there
will be a surface cake depends largely on whether a rooting cake
will be formed.  During a filtration cycle, particles will de-
posit in the interstices of the bed and will coat the granules.
If the granules of the bed are large, the coatings or dendrites
will not bridge to create the rooting cake.  On the other hand,
if the granules are sufficiently small, dendrites will bridge
to form an internal cake.
     The formation of surface cake has been observed at CCNY
(Wu, 1977) for ambient temperatures.  However, at high tem-
peratures, there was no indication that a surface cake was
formed.  Exxon Research and Engineering Company (Bertrand,
et al., 1977) reported the same experience.  No  surface cake
was observed in the Ducon GBF in their HTP fluidized bed coal
combustor miniplant.
                               207

-------
     Westinghouse Research Laboratories reported some collection
efficiency data for their laboratory scale GBF under ambient
conditions (Ciliberti, 1977).  The GBF utilized fluidized air
to clean the bed.  The reported overall collection efficiencies
were high and the grade efficiency curves calculated from cas-
cade impactor data were flat (i.e., almost independent of particle
diameter).  These phenomena indicated that a surface cake was
present.
     Surface cake formation at high temperature has been observed
for filtration on a ceramic cloth medium (Shackleton and Kennedy,
1977).  Therefore, the formation of surface cake appears to be
dependent upon the substrate, although it is influenced somewhat
by temperature.  Research work is required to study the cake
formation process at high temperatures.  We recommend a pilot-
scale study using actual combustion flyash.  Variables to be
studied should include bed granule diameter, bed structure, gas
temperature and pressure, and bed operating conditions.
     A small-scale laboratory study on cake filtration and the
pilot plant study on the cake formation process could be studied
concurrently.

BED CLEANING METHODS
     To prevent the bed from becoming saturated with collected
dust, it is necessary to clean the bed either periddically or
continuously.  The present bed cleaning methods were evaluated
in this report.  Major research and development work is needed
to improve the efficiency of the present cleaning methods and to
decrease the operating costs.
     Present bed cleaning methods share a common problem - des-
truction of rooting cake.  In the continuously moving bed system,
the relative motion of bed granules prevents the formation of
the internal cake.  In the reverse gas flow cleaning method,
the bed is either fluidized or stirred.  Thus internal cakes
                               208

-------
are destroyed or deteriorated.   The  rooting  cake  needs  to  be
rebuilt before a surface  cake  can  be formed.
     Since the rooting  cake  is  the foundation  of  surface cake
and the formation of  a  surface  cake  enhances the  collection effi
ciency of the GBF,  it is  desirable to preserve the  rooting cake
during the cleaning cycle.   Research work  is required to develop
cleaning methods that remove the surface cake  and preserve the
rooting cake.
OTHER RESEARCH  RECOMMENDATIONS
     There are many operational problems and uncertainties which
need to be resolved before HTP granular bed filters can be
considered sufficiently reliable and economical for commercial
application.  These problems  include:
     1. How to prevent particle seepage through the bed (during
        cleaning or filtration).
     2. How to reduce temperature losses (especially during
        cleaning).
     3. How to improve the efficiency and reduce the cost of
        granule regeneration  and recirculation.
     4. How to reduce pressure drop across bed.
     5. How to prevent attrition of granules causing particle
        reentrainment.
     Resolving these problems will not only help solve the
HTP particle collection problem, but will improve granular
bed filter technology for many other applications, especially
where hot, corrosive gases are encountered.

RECOMMENDED RESEARCH PROGRAM
     Of the above mentioned research and development recommen-
dations, electrostatic augmentation is presently the most
proven approach to improve the performance of  the GBFs.  It has
                               209

-------
been shown that the electrostatically augmented GBF has  the
potential to meet the most stringent cleanup requirements under
ambient conditions.  It is expected that the same statement will
hold for HTP conditions where higher electric field can  be imposed
on the beds.  Therefore, electrostatic augmentation is recommended
as the approach having the best probability of success,  based on
present  knowledge.
     Another potential advantage of the electrostatic GBF is that
it achieves high collection efficiency without the benefit of cake
filtration.  In actual applications, there might be filter cakes
and further research may show the way to promoting surface cake.
The presence of surface and rooting cakes will increase  the col-
lection efficiency of the GBF even more.
     The capital cost of an electrostatic GBF will be slightly
higher than that of regular GBFs.  The additional cost is due to
the requirement of a high voltage power supply and HTP insulation
for the electrodes.  However, the operating cost could be lower
due to a lower drop in pressure.  The estimated operating cost
for the Ducon GBF is about 31.8 mills/m3/min (0.9 mills/SCFM).
For an electrostatically augmented Ducon, GBF, the operating cost
would be 27.9 mills/mVmin (0.79 mills/SCFM) to maintain the same
efficiency.
     A detailed program to demonstrate the feasibility of using
electrostatic augmentation to improve GBFs for particulate con-
trol at high temperature is described below.  We recommend a
study of the electrostatically augmented GBF on a pilot  plant
scale of about 14.2 Am3/min (500 ACFM).  To duplicate actual
industrial applications, fresh test dust should be produced in-
stead of regenerated dust.  Since GBFs will be used in advanced
energy processes, it is desirable to test the electrostatically
augmented GBF on these processes.  A good approach would be to
use an atmospheric fluidized bed combustor.
     The GBF should be designed in such a way that it is easy to
change from one configuration to- another.  Bed cleaning  can be
achieved either by fluidization or by continuously withdrawing
granules and dust from the bed.
                              210

-------
     To aid in the design of the pilot plant, some small-scale
experimental work should be conducted concurrently,
     In outline, the objectives consist of the following tasks
     1. Conduct small-scale experiments to obtain design
        information.
     2. Design the pilot plant.
     3. Fabricate, install and start up the pilot plant.
     4. Prepare a detailed test plan describing:
       . a. The proposed test matrix,
        b. The measurement techniques to be used.
        c. The data handling methods.
     5. Conduct test programs.
     6. Analyze data,  conduct engineering and cost analyses
        of various configurations.
     7. Based on the above analyses, design and estimate the
        cost of a GBF  system for HTP applications.
     8. Recommend a test program to demonstrate a full-scale
        GBF system on  an HTP source.
                                211

-------
                          REFERENCES
Anderson, D.M. and L. Silverman, "Mechanisms in Electrostatic
     Filtration of Aerosols with Fixed and Fluidized Granules,"
     report by the Air Cleaning Laboratory, Harvard University
     to the U.S. Atomic Energy Commission, No. NYU-4615, 1958.
     See also, "Development of a Triboelectrified Fluidized Bed
     for Aerosol Filtration," 5th Atomic Energy Commission Air
     Cleaning Conference, p. 140, USAEC, TID-7551, 1957.

Avco, Inc. "Evaluation of Granular Bed Devices," Avco Applied
     Technology Division, AVATD-0107-69-RR, June 1969.

Balasubramanian, M. and A. Meisen, "A Note on the Diffusional
     Deposition of Aerosol Particles in Packed Bed," J. of Aerosol
     Science, £: 461-463  (1975).

Beecher, et al., "Energy Conversion Alternatives Study (EGAS)
     Westinghouse Phase II Final Report, Volume III - Summary and
     Advanced Steam Plant with Pressurized Fluidized Bed Boilers,"
     NSF/RA-760590, NTIS PB 268-558, October 1976.

Berry, F.J. arid J. Fournier, German Patent #676508 (1939).

Bertrand, et al., "A Regnerative Limestone Process for Fluidized
     Bed Coal Combustion and Desulfurization," Monthly Report
     no. 87, 1977.

Billings, C.E. and J. Wilder, "Handbook of Fabric Filter Tech-
     nology, Volume I," NTIS PB 200-648 (1970).

Bohm, L. and S. Jordan, "On the Filtration of Sodium Oxide
     Aerosols by Multilayer Sand Bed Filters," J. of Aerosol
     Science, I: 311-318  (1976).

Borguardi, R.H., R.E. Harrington, and P.W. Sparte, "Filtration
     Characteristics of Fly Ash," J. of APCA, 18: 387 (1968).

Brown, et al., "Energy Conversion Alternatives Study (EGAS)
     General Electric Phase II Final Report, Volume II - Advanced
     Energy Conversion Systems, Conceptual Designs," NASA-CR
     134949, NTIS PB 268-247, December 1976.

Brown, D.H., "Conceptual Design and Assessment of a Utility
     Steam Plant with Conventional Furnace and Wet Lime Stack
     Gas Scrubbers," NASA CR-134950, December 1976.
                               212

-------
Calvert, S. and R. Parker,  "Effects  of  Temperature  and  Pressure
Calvert, S., Chapter 46  in  "Air  Pollution," A.C. Stern, editor,
     3rd edition, Academic  Press,  New York, 1968.

Carney, S.D., U.S. Patent #2343401 (1944).

Ciliberti, D.F., Westinghouse  Research Laboratories, personal
     communication.

Gorman, J.C., and G.R. Fox,  "ECAS-General Electric Phase II
     Final Report, Volume I: Executive Summary," NASA-CR 134949,
     NTIS #PB 268-455, December  1976.

Dennis, R. , et al., "Special Incineration Studies - Institutional
     Design," 6th AEC Air Cleaning Conference, July 7-9, 1959.
     TID-7593, pg. 344-364  (1960).

Ergun, S., "Fluid Flow Through Packed  Columns," Chemical Engineering
     Progress, 4_£: 89-94  (1952).

Fiechter, L.B., British  Patent #135216 (1919).

Fiechter, L.B., British  Patent #163039 (1922).

Figueroa, A.R., "Aerosol Filtration by Fixed and Fluidized
     Granular Beds," Ph.D.  Dissertation, University of Cincinnati,
     (1975).

Figueroa, A.R., and W. Licht,  "Filtration of Submicron Particles
     by Fixed and Fluidized  Granular  Beds," Paper #33b presented
     at the 81st AIChE National  Meeting, Kansis City, Missouri,
     April 11-14, 1976.

Fournier, J., British Patent #450048  (1936).

Friedlander, S.K., "Mass and Heat  Transfer to Single Spheres
     and Cylinders at Low Reynolds Numbers," AIChE Journal,
     3: 43-48 (1957).

Fuchs, N. and A. Kirsch, "The^  Effect  of Condensation of a Vapor
     on the Grains and of Evaporation  from Their Surface on the
     Deposition of Aerosol  in  Granular Bed," Chem, Eng. Science,
     '20: 181-185 (1965).

Gebhart, J., C. Roth, and W. Stahlhofen, "Filtration Properties
     of Glass Bead Media for Aerosol  Particles in the 0.1-2 ym
     Size Range," J. of Aerosol  Science, 4_: 355-371 (1973).

Goldman, L., "Experiences in the Use  of Pebble-Bed Filters in
     Inorganic Chemical Plants," Staub,  2£: 449 (1964).


                               213

-------
Goren, S., in "Granular Bed Filter Development Program: Monthly
     Report for August 1977" by K.E. Phillips, prepared for the
     U.S. Energy Research and Development Administration under
     contract #EF-77-C-01-2579, October 1977.

Guillory, J.L, "Filtration Performance of a Moving Bed Granular
     Filter: Experimental Cold Flow Data," paper presented at
     the 5th International Conference on Fluidized Bed Combustion,
     Washington, D.C., December 12-14, 1977.

Guillory, J.L, "Granular Bed Filter Development Program,"
     prepared for the U.S. Dept. of Energy, FE-2579-15, 1978.

Hoke, R.C., et al., "A Regenerative Limestone Process for
     Fluidized Bed Coal Combustion and Desulfurization," Monthly
     Report no. 77, July 1976.

Hood, K.T., "Evaluation of the Combustion Power Company Moving
     Gravel Bed Dry Scrubber on the Control of Particulate
     Emissions from a Hog Fired Boiler," NCASI special report,
     September 1976.

Jackson, S. and S. Calvert, "Entrained Particle Collection in
     Packed Beds," AIChE Journal, L2: 1075-1078 (1966).

Kalen, B. and F.A. Zenz, "Filtering Effluent from a Cat Cracker,"
     Chem. Eng. Progress, 66: 67-71 (1973).

Klarding, N., British Patent #153263 (1921).

Knapsack-Griesham Aktiengesellschaft, Koln, Germany, British
     Patent #851776 (1960).

Knettig, P. and J.M. Beeckmans, "Capture of Monodispersed Aerosol
     Particles in a Fixed and in a Fluidized Bed," Canadian
     J. of Chemical Engineering, 52: 703-706 (1974).

Kovach, I.L., and D.G. Hanan, "Combined Dust, Aerosol and Vapor
     Removal Efficiency of Packed Activated Carbon Beds,"
     Staub, 3_0: (12), 35-39 (1970).

Lapple, C.E., "Stack Contamination - 200 Areas, Interim Report,"
     HDC-611, August 5-October 12, 1948.

Lee, K.C., "Filtration of Redispersed Power-Station Fly Ash by
     a Panel Bed Filter with Puffback," Ph.D. Dissertation,
     The City University of New York, (1975).

Lee, K.C., et al., "Panel Bed Filter," EPRI-AF-560, 1977.
                               214

-------
 Levenspiel, 0., "Chemical Reaction Engineering," Wiley 8 Sons,
      New York, 1972.

 Leith, D., S.N. Rudnick, and M.W. First, "High Velocity, High
      Efficiency Aerosol Filtration," NTIS PB 249-457, 1976.

 Leith,  D.  and M.  First,  "Performance  of a Pulse-Jet Filter  at
      High  Filtration Velocity,  I - Particle  Collection,"  J.
      of APCA 27_:  534-539 (1977).

 Levi,  J.P.  and J.  Blume, French Patent #763077  (1934).


 Lewis  Research Center,  "Evaluation of Phase  II  Conceptual De-
      signs  and Implementation Assessment Resulting  from the
      Energy Conversion  Alternatives  Study (EGAS),"  NASA TMX-
      73515, NTIS  PB 270-017,  April 1977.

 Lynch,  G.E.,  U.S.  Patent #1766221  (1930).

 Lynch,  G.E.,  Fuel  Economist,  12: 47 (October 1936).

 Marple, V.,  "The Fundamental  Study of  Inertial Impactors,"
     Ph.D.  Thesis,  University of Minnesota, 1970.

 McCain, J.D.,  "Evaluation of  Rexnord Gravel Bed Filter," EPA
     600/2-76-164,  NTIS  PB 225-095, June  1976.

 McFee,  D.R.,  and J.  Sedlet, "Plutonium-Uranium-Molybdenum Fume
     Characteristics  and Sand Filtration," J. of Nuclear Energy,
     22_i 641-650  (1968).

 Mercer, T.T.  and R.G. Stafford,  "Impaction from Round Jets,"
     Ann. Occupational Hygiene,  12: 41-48 (1969).

 Mercier, E.  and M.  Ehlinger, U.S. Patent  #2493356 (1950).

 Miyamoto, S.  and H.  Bonn, "Filtration of  Airborne Particulates
     by Gravel Filters:  I. Initial Collection Efficiency of a
     Gravel  Layer,  "  J. of APCA, 20; 1051-1054 (1974).

Miyamoto, S.  and H.  Bohn, "Filtration of  Airborne Particulates
     by Gravel Filters:  II. Collection Efficiency and Pressure
     Drop in  Filtering Fume," J. of APCA, 25_: 40-43 (1975).

Nordstrom,  0., Swedish Patent #56098  (1922).

Nordstrom,  0., British Patent #217113  (1924).

Orr, C., Jr.,  "Particulate Technology, "  p. 420, MacMillan Com-
     pany, New York  (1966).
                                 215

-------
 Parker,  R.  and  S.  Calvert,  "High Temperature and High Pressure
      Particulate  Control  Requirements,"  EPA 600/7-77-071,  July
      1977.

 Payatakes,  A.C.,  "Model of  Aerosol  Particle Deposition in  Fibrous
      Media  with Dendrite-Like  Pattern: Application to Pure Inter-
      ception  During  Period  of  Unhindered Growth," Filtration
      and Separation,  15;  602  (1976a).

Payatakes, A.C., "Model of the Dynamic Behavior of a Fibrous
     Filter; Application to Case of Pure  Interception During
     Period of Unhindered  Growth," Powder Technology, 14;
     267 (1976b).

Payatakes, A.C., "Model of Transient Aerosol Particle Deposition
     in Fibrous Media with Dendrite Pattern," AIChE Journal,
     23^: 192-202 (1977).

Payatakes, A.C. and C. Tien, "Particle Deposition in Fibrous Media
     with Dendrite-Like Pattern: A Preliminary Model, " J.  of
     Aerosol Science, 1_:  85-100 (1976).

Paretsky, L.,  L. Theodore, R.  Pfeffer, and A.M. Squires, "Panel
     Bed Filters for Simultaneous Removal of Fly Ash and Sulfur
     Dioxide:  II.  Filtration of Diluted Aerosol by Sand Beds,"
     J. of APCA, 21^:  204-209 (1971).

Paretsky, L.C., "Filtration of Aerosols by Granular Beds,"
     Ph.D. Thesis, City University of New York (1972).

Perry, J.H.  (editor)/'Chemical Engineer's Handbook,"McGraw-
     Hill, New York,  1973.

Phillips, K.E. "Granular Bed Filter Development Program: Monthly
     Report for August 1977,"  prepared for the U.S. Energy
     Research and Development  Administration under contract
     no. EF-77-C-01-2579,  October 1977.

Ranz, W.E. and J.B. Wong,  "Impaction of Dust and Smoke Particles,"
     Ind. Eng. Chem.  44:  1371-1381 (1952).

Research Cottrell, U.S. Patent No. 2,990,912, July 4, 1961.

Schmidt, E.W., J.  Gieseke, P.  Gelfand, T.W. Lugar, and D.A. Furlong,
     "Filtration Theory for Granular Bed," J.. of APCA, 28:   143-146,
     1978.                                             —

Shackleton,  M. and J. Kennedy, "Ceramic Fabric Filtration at
     High Temperatures and Pressure," paper presented at the
     EPA/ERDA Symposium on High Temperature/Pressure Particulate
     Control,  Washington,  D.C., September 20-21, 1977.
                                216

-------
 Sharapov, K.A., V.V. Leonov, I.L. Saklarnova, A.F. Skvartsov,
      and N.G. Braginets, "Study of a Multigradient Electro-
      magnetic Filter for the Dry Cleaning of Gases," Stal'
      963-964 (1975), Ref. Zh. Khim. GI775 (1976).

 Solvay, E., British Patent #18573 (1889).

 Squires, A.M. and R.A. Graff, "Panel Bed Filters for Simultaneous
      Removal of Fly Ash and Sulfur Dioxide: III. Reaction of
      Sulfur Dioxide with Half-Calcined Dolomite," J. of APCA,
      21.: 272-276 (1971).

 Squires,  A.M.  and R.  Pfeffer,  "Panel  Bed Filters for  Simultaneous
      Removal of Fly  Ash and  S07:  I.  Introduction,"  J. of APCA,
      2Q_:  523 (1970).           L

 Stephan,  D.G.,  G.W.  Walsh,  and  R.A. Herrick,  "Concepts  in  Fabric
      Air Filtration," AIHA Journal,  2±:  L (1960).

 Stern,  A.C., H.W.  Zeller,  and A.I.  Schekman,  "Collection Effi-
      ciency of Jet  Impactors  at  Reduced  Pressures,"  Ind. and
      Eng.  Fundamentals, !_:  273  (1962).

 Strauss,  W. , and M.W.  Thring,  "Studies  in High  Temperature Gas
      Collecting," J.  of the  Iron and  Steel Institute, 196:
      62 (1960).

 Sverdrup,  E.F.  and  D.H. Archer,  "The  Tolerance  of  Large Gas
      Turbines to Rocks, Dusts,  and  Chemical  Corrodants," pre-
      sented at EPA/ERDA Symposium on  High Temperature and  Pres-
      sure Particulate Control,  Washington, D.C.  September  1977.

 Swift,  M.W.. et al.   "Plans and Studies  on Flue Gas Cleaning
      and Particulate Monitoring in PFBC," Proceedings of the
      EPA/DOE Symposium on High  Temperature/ High Pressure  Parti-
      culate Control,  EPA 600/9-78-004,  September 1977.

Taub, R. , "Filtration Phenomena  in a Packed Bed  Filter," Ph.D.
     Dissertation, Carnegie-Mellon University, 1970.

Thomas, J.W. and R.E. Yoder, "Aerosol Size for Maximum  Penetra-
     tion Through Fiberglass and  Sand Filters,"  Archives  Ind.
     Health, 121: 545  (1956a) .

Thomas, J.W. and R.E. Yoder, "Aerosol Penetration Through a Leak
     Shot Column, A Method of Particle Size Evaluation," AMA
     Archives Ind. Health, 13^: 550  (1956b) .

Thomson, T. and N. Nisbet, British Patent  #216675  (1924).

Thring, M.W. and W. Strauss, "The Effect  of High Temperatures
     on Particle Collection Mechanisms,"  Trans.  Instru. Chem.
     Eng. 41: 248-253 (1963).
                                 217

-------
Tien, C., C.S. Wang and T.D. Barot, "Chain-Like Formation of
     Particle Deposition in Fluid Particle Separation," Science,
     196: 983-985, 1977.

Veron, M., U.S. Patent No.  2,564,316 (1951).

Wade, G.L., "Performance and Modeling of Moving Granular Bed
     Filter," paper presented at EPA/ERDA Symposium on High
     Temperature/Pressure Particulate Control, Washington, D.Ci,
     September, 1977.

Wade, G., H. Wigton, J. Guillory, G. Goldbach, and K. Phillips,
     "Granular Bed Filter Development," Final report prepared
     for U.S. Dept. of Energy, No. FE-2579-19, 1978.

Wang, C.S., M. Beizail, and C. Tien, "New Concepts of Particle
     Deposition from Suspensions Flowing Past a Collector,"
     paper #18b, presented at the AIChE 83rd national meeting
     Houston, Texas, March 1977.

Westinghouse Electric Corp., "Clean Power Generation from
     Coal," O.C.R., 84, NTIS: PB 234-188, April 1974.

Wigton, H. and G. Wade, "Granular Bed Filter Development Program,"
     report prepared for the U.S. Dept. of Energy, No. FE-2579-
     18  (1978).

Wilson, E.J. and C.J. Geankoplis, Ind.  Engineering Chemistry,
     5: 9 (1966).

Wu, M.S., Air Pollution Technology, Inc., Personal Communication,
     1977.

Zahradnik, R.L, J. Anyigbo, R.A. Steinberg, and H.L Toor,
     "Simultaneous Removal of Fly Ash and S02 from Gas Streams
     by a Shaft Filter Sorber," Env. Science and Tech., 4: 663-
     667 (1970).
                                218

-------
                          /»,     TECHNICAL REPORT DATA
                          (Please read Instructions on the reverse before completing)
1. REPORT NO.
 EPA-600/7-79-020
                           2.
                                                      3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
Evaluation of Granular Bed Filters for High-temper-
   ature/High-pressure Particulate Control
                                                      5. REPORT DATE
                                                      January 1979
                                                      6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Shui-Chow Yung, Ronald Patterson, Richard Parker,
   and Seymour Calvert
                                                      8. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Air Pollution Technology, Inc.
4901 Morena Boulevard, Suite 402
San Diego, California  92117
                                                      10. PROGRAM ELEMENT NO.
                                                      1NE624
                                                      11. CONTRACT/GRANT NO.

                                                      68-02-2183
 12. SPONSORING AGENCY NAME AND ADDRESS
 EPA, Office of Research and Development
 Industrial Environmental Research Laboratory
 Research Triangle Park, NC 27711
                                                      13. TYPE OF REPORT AND PERIOD COVERED
                                                      Final; 9/76 - 9/78	
                                                      14. SPONSORING AGENCY CODE
                                                       EPA/600/13
15. SUPPLEMENTARY NOTES JERL.RTp projeCt OfHC6r IS
2925.
                                                         Drehmel, MD-61, 919/541-
 is. ABSTRACT
               repOrt gives resuitg of a critical review and evaluation of the status
 of granular bed filter (GBF) technology. GBFs can operate at the high temperatures
 and pressures normally encountered in advanced energy processes. In theory, the
 filters can clean the gas to meet the current New Source  Performance Standard if
 certain operating difficulties are overcome. Use of a combined- cycle power plant
 with GBFs for particulate cleanup is economically competitive when compared to a
 conventional power plant. However, successful and reliable performance at high tem-
 peratures and pressures has not yet been demonstrated.  None of the design equations
 reported in the literature are adequate to predict the collection efficiency of a GBF
 at face  velocities normally encountered in the field. A performance model was
 developed in this  study to predict collection efficiency for GBFs.  A small scale
 experiment was performed to obtain additional performance data for GBFs.
17.
                             KEY WORDS AND DOCUMENT ANALYSIS
                 DESCRIPTORS
                                          b.lDENTIFIERS/OPEN ENDED TERMS
                                                                  c. COS AT I Field/Group
 Pollution             Dust
 Gas Scrubbing        Aerosols
 Gas Filters           Electric Power
 Granular Materials     Plants
 High-temperature Tests
 High-pressure Tests
                                          Pollution Control
                                          Stationary Sources
                                          Granular Bed Filters
                                          Particulate
13B
07A,13H
13K
11G
14B
11G
07D

10B
18. DISTRIBUTION STATEMENT

 Unlimited
                                          19. SECURITY CLASS (ThisReport)
                                          Unclassified
                                          20. SECURITY CLASS (Thispage)
                                          Unclassified
 !1. NO. OF PAGES

      231	
                                                                  22. PRICE
EPA Form 2220-1 (9-73)
                                         219

-------