AIR POLLUTION CONTROL
    TECHNOLOGY SEMINAR
             Prepared For


U.S. ENVIRONMENTAL PROTECTION AGENCY
     ECONOMIC ANALYSIS DIVISION
            Washington, DC
THE
             P.O. Box 40284, Nashville,TN37204(61 5)794-01 10

-------
     AIR POLLUTION CONTROL
      TECHNOLOGY SEMINAR
               Prepared For

 U.S. ENVIRONMENTAL PROTECTION AGENCY
       ECONOMIC ANALYSIS DIVISION
              Washington, DC
               Presented by:

THE
             consultants in environmental manaoement
             P.O. Box 40284, Nashville, TN 37204 (615)794-0110

-------
                     AIR POLLUTION CONTROL TECHNOLOGY
                             SEMINAR SCHEDULE
                              June  25,  1981
  I.   Types and Sources of Air Contaminants


 II.   Meteorological  Aspects of Air Pollution Control

      A.   Ambient Sampling/Field Survey Design

      B.   Stack/Source Sampling


III.   Particulate Control

      A.   Gravity and Centrifugal Separators

      B.   Electrostatic Precipitation/Baghouses

      C.   Fugitive Controls

      D.   An Example  of a "Bubble"

      E.   Scrubbers

      F.   Filtration


 IV.   SOX

      A.   Flue Gas Desulfurization

      B.   Emerging Technologies Like Fluidized Bed and Limestone


  V.   NOX

      A.   Combustion  Modification

      B.   Thermal De-N0x

      C.   Selective Catalytic Reduction

      D.   Low NOX Burners

-------
                                     -2-
  VI.   Hydrocarbons



       A.   Storage Tanks



       B.   Floating Roofs



       C.   Vapor Recovery






 VII.   Mobile Source/CO



       A.   Catalytic Converter



       B.   Lead in fuel



       C.   Non-Point Source Projections





VIII.   Emission Control/Environmental Analysis



       A.   Toxics



       B.   Hazardous Uastes



       C.   Prediction (General Concepts)

-------
I.   TYPES AND SOURCES OF AIR CONTAMINANTS
           (ORAL  PRESENTATION)

-------
II.   METEOROLOGICAL ASPECTS  OF AIR POLLUTION  CONTROL

-------
       AIR QUALITY CONSIDERATIONS
               Prepared by

        F. G. Ziegler, Ph.D., P.E.
     Director of Resources Management
Associated Water and Air Resources Engineers,  Inc
              Prepared for

National  Hazardous Materials Training Course
            Vanderbilt University
            Nashville, Tennessee
            September 25-29, 1978

-------
HEIGHT
                                      Night
                             SPEED

-------
      Vertical Te^peracure  Scruc-ure of Che Acnosphera

      1.  Dry adiabatic  lapse  race (DALS.)
          a.  icuation or  scace
              Pv - RT  or  -  -  RT  (Ideal Gas)
                           P
          b.  Eydroscacic  Equation

                  dJL
                  dz

                  dP
                  P
                  ?a
             In?
             P.
                     RT
- - (-f=U
    \ XT/

=> ? (r.)  z.
                               =  surace
                         RTa
                   ?. a
                         -2-
                                   whera T =• T
                                                    =saa  raseratr
                             RTm

                       ?0  exp
                               _£
                               R
                                       C2
             can incsgraca LI I kr.ovn  as  a  riincticn. or z
c.  Adiabacic £:c_

        f1  -   (Y  -  1 )
T_
To
                                         (?rcn  Isc Lav)
                                         (7 - -j;r- =•  1.41 far dry air)
                                              v

-------
d.  Dry Adiabatic  Lase  Race
                         dz
                                                (T'  refers
                                                 displaced)
                                                           5iuE2 or  ai.r
    dT'  _/y - 1
         5 "  f - - (t)  f

    1C.) - I(c) - f
    Jor Air
                    1000 fc   ^

2.  Vertical Te=peracura Scruc'ura
                                   100
a.  Definiricns - assuae a parcel of air rising cr failias vill
    ;ollow adiabatic lapse race free poirit: of departure.  Therefore:

    Suparadiabacic - T asp era Curs decreases faster vich height Chan
                     adia'oacic raca - rising air parcel, ccoling
                     ac adiabatic race regains varner and less dense
                     Chan surrounding air - buoyancy forces cend
                     Co accelerate it upward - unstable equilibrium
                     buoyancy forces "and ca produce acceleration

      Subadiabacic - Tesperature decreases slower r-*ith height than
                     adia'oacic rate - rising air parcel becomes
                     cooler and sinks to its starting point -
                     stable - buoyancy forces send to restore
                     parcel to original position
         Adiabatic - Neither stable nor unstable - neutral

-------
b.  Stability conditions

    DAL3. - dry adiabatic lapse rate
      LS. =« actual lapse race
    Eeighc
                  Teoperacure
                 Superadiabatic
                   Unstable
Haighc
                                                      DALX - L3.
            Tesraerarura
              Adiabacic
               Neutral
      Height
                    Tesperacure

                   Subadiabatic
                      Stable
                                        Height
                                                       DALX
              Tenperacure

               Inversion
                Stable

-------
2.  Characteristic Joras of Smoke Pluses
      DALSL
    a.  Looping - Unstable

Successive sections of pluzie trav-
eling vit'n different inclinations

Light wind, sunshine varzs ground
surface, terrperature lapse high
degree of turbulence especially
convectiva or large scale aechan-
ical turbulence caused by obstruc-
tions .
    Tenroerature
\ \ fcf
\\LR
"1

1

v/

Rea
at
dis
loo

                  Reaches ground
                  distances tan
b.  Caning - slightly unstable o~
                Neutral

Successive sections follow trajectci
which are not videly different

Open country, moderate wind speed,
cloudy sky - or clear sky with
strong winds.  Diffusion equations
best
    Tesroeracure

-------
Height
                        may be sora
                        sinuous viaved
                        fron abova
             Tesoeracure
      c.  Fanning - Stable

Cross wind spread - meanders,
may be ainosi no vertical spraac

Vertical campanent suppressed
sore than horizontal, night  ti^ie,
clear skv, cooling of air near
ground, light wind, mechanical
turbulence susarassed
height
    d.  Lcfting - Transition
        Unstable to Stable

Late afternoon and early
evening

Replaced by  fanning as  inversion
deepens
             Terroerature

-------
               Stable Lavqr
                  E.   Funigation  -  Transition
                      Stable  co Unstable

              Jlornizg  folio-sing  a  stable night

              Clear skys,  light  vines, surfer,
              high concentrations  -more rapidly
              come to  ground, goad zixing
              below inversion laver
   3.  Diurnal Variation of Stability
                                Rural
Height
            Established
              Profile
           Inversion
             (Stable)
                                                  Established  Profile
                                                                     Condition
                                                                     r.Jhan  ta=p
                                                                     lapse height
                                                                     reaches
                                                                     stack height
                                                                     fuaigacicn
                                                                     will"occur
         Afternoon
evening
Late evening
Early norning
                                                             Momins

-------
                                          Urban-
  Eeighc
             (Unstable)
     ruaigaticn

     Heat from building*
       :reets, side-
       •alks, etc.
                          Inversion
                          (Stable)
                                                                        mry. gaticn
                      Aftaraoon     Evening  Late  evening  Co       Morning
                                             early morning
  Cone, of
Poliucant
                                       Urban Cone.
              Afternoon
Evening
Late evening
early coming

-------
                  CHAPTER 2




                Extracted From



"WORKBOOK OF ATMOSPHERIC DISPERSION  ESTIMATES'



              By D.  Bruce Turner

-------
                             -ESTIMATES  OF  ATMOSPHERIC DISPERSION
    This chapter outlines the basic procedure to
 be used in making dispersion  estimates  as sug-
 gested by Pasquill  (1961) and modified by Ginord
 (1961).

 COORDINATE  SYSTEM

    In the system considered  here  the origin is at
 ground level at  or  beneath the  point  of emission,
 with the X-axis extending horizontally in the  direc-
 tion of the  mean wind.  Ths y-azis is  in the hori-
 zontal plane perpendicular  to the  i-asis, and  the
 z-azis  extends vertically. The plume travels  along
 or parallel to the i-azis. Figure 3-1 illustrates  the
 coordinate system.

 DIFFUSION EQUATIONS

   The concentration, %, of gas  or  aerosols (para-
 des less than about 20 microns  diameter)  at r,y,z
 from a continuous source with an effective emission
 height, H, is given  by equation  3.1. The notation
 used  to  depict  this concentration  ia  x (^OWH).
 K is  the height of  the  plume csntsriine when it
becomes essentially  level, and is  the  sum  of  the
physical stack height,  h, and the  plume rise, iK.
The following assumptions  are  made:  the  plume
spread has a Gaussian distribution (see Appendix
2) in both the horizontal and vertical planes, with
standard deviations of plume concentration  distri-
bution in the horizoncal and vertical of o> and a,,
respectively;  the mean wind speed affecting  the
plume is u; the uniform emission rate of pollutants
is Q;  and total reflection of  the  plume takes placs
at the earth's surface, Le.,  there is no deposition
or reaction at the surface (see problem 9).
                                           1
                                           2

                                           (3.1)
'Note: MO —a/b = r-V11 where e is Sis Sase of natural logarithms
     and is approximately »qual to 2.7133.
                                                                                   U-y,0)
          Figure      Coordinate system showing Gaussian distributions in the horizontal  and vertical.
Estimates
                                                 10

-------
Any consistent set of units may be used.  The most
common is:

    X (g ra"1) or, for radioactivity (curies m"1)
    Q (g sec""1) or (curies sec~:)
    u (m sec"1)
     and <*. are evaluated in terms
of the downwind distance, r.
   Eq.  (3.1) is valid where  diffusion in the direc-
tion  of the plume travel can be neglected,  that is,
no diffusion in  the z direction.

This may be assumed if the  release is continuous
or if the duration of  release  is equal to or greater
than the travel time  (i/u) from the source to the
location of interest.

   For  concentrations calculated at ground level,
Le., z — 0, (see problem 3) the equation simplifies
to:
     (riy.o;H)
       IJ'
                                            (3.2)
   Where  the concentration is to be  calculated
along the centerline of the  plume (y —  0),  (see
problem  2)  further simplification results:
                                             3.3)

   For a ground-level source with no effective plume
rise (H — 0), (see problem 1):
     (r,0,0;0) -
                     Q
                        U
                                            (3.4)
EFFECTS  OF  STABILITY
   The values of  and   5

Strong
A
A-8
8
C
C

Mod tret:
A-8
3
B-C
W)
0

Slight
B
C
C
D
D
Night
fninly Overcast

^/SLowClot-d

E
D
D
D

-=3/3

Cloua

F
E
0
D
The neutral ciass, D, should be assumed for overcast nnditisns during
day or night.

   "Strong" incoming solar radiation corresponds
to a solar altitude greater rhf?n  60° with dear skies;
"slight" insolation  corresponds to a  solar altitude
from 15° to 35° with clear skies. Table 170, Solar
Altitude  and Azimuth, in  the Smithsonian  Mete-
orological Tables (List, 1951) can be  used in deter-
mining the solar altitude. Cloudiness will decrease
incoming solar radiation and should  be  considered
along with solar altitude is determining solar radia-
tion.  Incoming radiation   that  would  be strong
with clear skies can, be ezpected to be reduced  to
moderate with  broken (% to  %  cloud cover) mid-
dle clouds  and to  slight with broken  low  clouds.
An objective system of classifying  stability  from
hourly meteorological observations  based  on the
above method has been suggested (Turner, 1961).

   These methods  will give representative indica-
tions of stability over open  country  or  rural areas,
but are less reliable  for urban areas. This  differ-
ence is due  primarily to the influence of the city's
Larger  surface  roughness and heat  island effects
upon  the  stability  regime  over urban  areas.  The
greatest difference occurs on ralrn clear nights; on
such nights conditions over rural areas are very
stable, but over urban areas they are slightly un-
stable or near neutral to a height several times the
average building height, with a stable layer above
(Duckworth and Sandberg, 1954; DeMarrais, 1961).
                                                  11
        ATMOSPHERIC
                                                                                          ESTIMATES

-------
    Some prpl?mtr.ary results of a. dispersion experi-
 ment in St.  Louis  (Pooler, 1S65) showed that the
 dispersion over the city during the daytime behaved
 somewhat like types B  and C; for one night experi-
 ment OT varied with distance between types D and 3.

 ESTIMATION OF VERTICAL  AND
 HORIZONTAL  DISPERSION

    Having determined the stability  class  from
 Table 3-1, one can evaluate the estimates of o> and
 
                                           (3.6)

     for any z from 0 to L
     for x >2fc, SL is where o-t — 0.47 L
     The value of <7lL — 0.8 L


 EVALUATION OF VTND 5PEZD

    For the wind  speed, u, a mean through the ver-
 tical  extent of  the  plume  should be used. This
 would be from  the height K —  2  are in general less than  those  of  a-,.
The ground-level centeriine concentrations for these
Estimates

-------
10,000
                                         DISTANCE  DOWNWIND, km
       Figure      Horizontal dispersion  coefficient as a function of downwind distance from the source.
                                                   13
                                                            ATMOSPHERIC DISPERSION ESTIMATES

-------
    1,000
                                        1                                10
                                             DISTANCE  DOWNWIND,  km
                                                 100
        Figurs      Vertical dispersion coefficient as a function  of  downwind distance  from the source.
Estimates
   3»-Ml O - «t - :
14

-------
                            123*5
                              CONC.
                              5 30m«l«r>
                        Variations  in concentration  in the vertical beneath  a more stable layer.
three cases (where c. ran be erpectsd to be within
a factor of 2) should be correct within a factor of 3,
including errors in <*T and u. The relative confidence
in the o-'s  (in decreasing order) is indicated by the
heavy lines and dashed lines in Figures 3-2 and 3-3.

   Estimates of K, the effective height of the plume,
may be  in error because of uncertainties in the esti-
mation  of AH. the  plume rise.  Also, for problems
that require estimates of concentration at a specific
coint, the difficulty of determining the mean  wind
over a  ziver.  time  interval and consequently the
location of the i-a^is can cause considerable un-
certainty.

GRAPHS  FOR ESTIMATES  OF  DIFFUSION

   To  avoid repetitious computations, Figure 3-5
(A through F) gives reladve  ground-level concen-
trations times wind speed (~ u/Q) against down-
wind distances for various effective neights of  emis-
sion and limits to the vertical rriring for each sta-
bility class (I figure  for each  stability)- Computa-
tions were made  from EC. (3.3), (3.4), and (3.5).
Estimates  of  actual  concentrations may be deter-
mined by multiplying ordinate values  by Q/u.
PLOTTTNG GROUISD-LJTVEL
CONCEiXTRATION ISOPLETH5

   Often  one wishes  to  determine  the  locations
where concentrations equal or exceed a given  mag-
nitude.  First, the arial position of the plume  mus:
be determined by the  mean  wind  direction.  ': o:
plotting isopieths  of  ground-level  concentrations
the  relationship  between  ground-level  csnteriin:
concentrations and ground-level on-aris ccncentra
tions car,  be used:
     (*.y.Q;K)   _ ro  f _ j
     (x,0,0;K)     ^  [      2
_x_u,
 X
(n r
o./
The y coordinate of i particular isopletri rrom tc
T-PT-9 can be determined at each  downwind dif
tance, s.  Suppose that  one wishes to know tn
on-ais distance to the 1CT5  g m~! isopieth  at an
of 600 m, under stability type B, where the  grouse
level centerline concentration at this distance  :
2.9 z  10~J g m-5.
                             x  (iA-.0:H)
                               (x,0,0;K)
    icr5
  2.9 s 1CT5
               0.345
                                                             ATMOSPHERIC  DISPERSION ESTEVIAT1

-------
                10'
                10'
                                                        OlSTANC-.k.
;ure        xu/Q  with distance for various heights  of emission (H) and limits to vertical dispersion (I), A stability.
'iimatea
                                                           16

-------
               10'
Figura  •;;•




Estimates
10
   o.i                           i

                                         OIS7AXC-. i.


xu/Q with distance  for  various  heights of emission  (H) and  limits to vertical dispersion (L), E stability.



                                               17

-------
From Table A-l  (Appendix 3) when ezp

               — 0.345, y/a, — 1.46
f	
I     2
      r ieure 3-2, for stability 3 and i — 500 m, 
-------
                                  — EFFECTIVE  HEIGHT OF  EMISSION"
 GENERAL CONSIDERATIONS

    In most problems one must estimate the elec-
 tive stack height, H,  at which the plane becomes
 essentially level Rarely will this height correspond
 to the physical heigh: of the stack, h.  If the plume
 is caught in the turbulent wake  of the stack or of
 buildings in the vicinity  of the stack, the effluent
 will be mixed rapidly downward toward the ground
 (aerodynamic downwash).  If the plume is emitted
 free of these turbulent zones, a number of emission
 factors and me:eoroiogical factors influence the rise
 or  the plume.  The  emission factors  are:  velocity
 of the eSuent at the top  of the stack, v,; tempera-
 ture of the efRuent at the top of  the stack, T,;~  and
 diameter of the stack opening,  d. The meteorolog-
 ical factors influencing plume rise axe wind speed,
 u;  temperature  of the air, T.; shear  of  the wind
 speed  with height,  du/oz;  and  atmospheric  sta-
 bility. No theory on  plume rise takes into account
 all of  these variables; even if such a  theory were
 available, measurements  of all of  the parameters
 would seldom be available. Most of the equations
 that have  been formulated for computing the ef-
 fective height of emission  are semi-empirical. For a
 recent review of equations for effective height of
 emission see Moses, Strom, and Carson (19S4).

    Moses and Strom (1951), having compared ac-
 tual and calculated plume heights by means of sis
 plume rise equations, report "There is no one  for-
 mula which is outstanding in  all respects." The
 formulas  of  Davidson-Bryant   (1949),  Holland
 (1953), Bcsanquet-Carey-Haiton  (1950), and 3o-
 sanquet  (1957)  all give generally satisfactory  re-
 sults in the test situations. The  experiments con-
 ducted by  Moses and Strom involved plume rise
 from a stack of less than 0.5 meter diameter, stack
 gas exit velocities less than 15 m sec"1-, and effluent
 tempera rare not mors  than  35 "C  higher  than that
 of the  ambient air.

   The equation of  Holland was developed with
 erperimental  data from larger sources rb^-rt those
 of Moses and Strom (stack diameters  from L7 to
 4.3  me:ers and stack  temperatures  from  32  to
 2Q43C); Holland's equation is used in the solution
of the problems given in cHi.s workbook. This equa-
tion frequently underestimates the efective height
of emission; therefore its use often provides a slight
"safety" factor.

   Holland's equation is:
    v, — stack gas exit velocity, m sec"1
    d =— the inside stack diameter, m
    u =- wind speed, m sec"1-
    p — atmospheric pressure, mb
    T. — stack gas temperature, °K
    T» — air temperature, aK
and 2.53 z 10~s is a constant having units of mb~:
m~l.

    Holland (1953)  suggests that a value between
1.1 and 1J2 times  the  ^H from the equation should
be  used for unstable  conditions;  a value between
0.8 and 0.9 times  the  dH from the equation should
be used for stable conditions.
    Since the piume rise rrcm  a stack  occurs over
some distance downwind, EG.  (4.1) should not be
applied  within the first few hundred meters of the
stack.
   ~~( 1.5 -r 2.53 s 10" p    =    d) (4.1)
where:
   ;lH — the rise of the piume above the stack, m
E5frctiv» Height

-------
           Extracted From



 "GUIDELINE ON AIR QUALITY MODELS"



U.  S.  Environmental Protection Agency



     Research: Triangle Park,  NC

-------
 CLASSES OF MODELS
      Air quality models  discussed in  this  guideline  can  be  categorized
 into four generic classes:   Gaussian,  numerical,  statistical  or
 empirical, and  physical.   Within  some  of these  classes a  large number
 of  individual  "computational  algorithms" exists,  each with  its own
 specific applications.   While each  of  these  algorithms may  have  the
 same generic basis,  e.g.,  Gaussian, it is  accpeted practice to refer
 to  them individually as models.   For example, the Climatological
 Dispersion Model,  the Air  Quality Display  Model and  the Texas
 Climatological  Model  are commonly referred to as  individual models.
 In  fact,  they are  all variations of a  basic  Gaussian model.   In many
 cases  the  only  real  differences between models  is the degree of detail
 considered in the  inupt or output data.
     Gaussian models  are generally state-of-the-art  techniques for
 estimating  the  impact of nonreactive pollutants.  Numerical  models
 are more developmental in nature than  are Gaussian models and are not
 as widely  applied.   However,  they are more appropriate than Gaussian
models for multi-source applications which  involve reactive pollutants.
Statistical or empirical  techniques are frequently employed  in situations
where  incomplete scientific understanding of the physical  and  chemical
processes make the use of a Gaussian or numerical  model  impractical .
Various specific models  of these three generic types  are  recommended
in this guideline.
     Physical  models, the fourth generic type, are used  in wind  tunnel
or other fluid modeling  facilities.   Such models may  be  very useful
in evaluating  the air quality impact of a source or group of sources
                                   21

-------
 in a geographic area limited to a few square kilometers.   Where
 physical modeling is available and applicable (e.g., stable pollutants,
 a narrow range of meteorological  conditions' and a few sources)  it is
 recommended.   However,  physical  modeling is  frequently not a viable
 option due to limitations  associated with  (1)  the need to  simulate  a
 wide range of atmospheric  and source conditions,  (2)  availability
 of physical modeling facilities,  and (3)  requirements  for  technical
 expertise.  Thus,  due to these limitations,  physical modeling is  not
 discussed further  in this  guideline.
      In  addition  to  the four classes  of  models, this guideline  identi-
 fies  two hierarchies of models.   The  first hierarchy consists of
 general  estimation techniques  that  provide conservative estimates of
 the  air  quality impact of  a  specific  source, or source category.  The
 purpose  of such techniques is  to  eliminate from further consideration
 those  sources  that will  not  cause or  contribute to ambient concen-
 trations  in excess of MAAQS  or allowable concentration increments.
 Conversely these techniques  can be used to identify those control
 strategies that have  the potential to meet NAAQS and allowable increments
 The second hierarchy of models are those analytical techniques which
 provide more detailed treatment of physical  and chemical  atmospheric
 processes, require more  detailed and precise input data,  and provide
more specialized concentration estimates.  As' a result they provide
a more refined and, at least theoretically,  a more accurate estimate
of source impact and  the effectiveness of control  strategies.
     In some cases, the  first hierarchy of models  may be  equated
with screening techniques  to  determine if a  second or more  refined
                                    22

-------
 analysis  is  required.   However, while  the use of  screening

 techniques followed  by  more  refined analysis is desirable,

 there are situations where the screening techniques are practically

 and technically the  only models available.  Thus, at times a more

 refined analysis is  not possible and the screening techniques are

 the only viable option  for estimating source impact and evaluating

 control strategies.


 RECOMMENDED MODELS

    To meet the need for consistency identified in Section 2,

 selected point source and multi-source models applicable to specific

 pollutants and averaging times are recommended in this subsection.

 It is further recommended that the set of plume rise equations given

 by Briggs be used to calculate the effective stack height that

 is input to these models.

    Ideally,  air quality models that are recommended should meet

 prescribed standards of performance for particular applications and

 should be subjected to specific validation procedures.   However, there

 are no generally accepted standards of performance and validation pro-

cedures.  The models recommended in this guideline are simply

 those which are (1) representative of the state-of-the-art for atmospheric

 simulation models and (2) those most readily available to air pollution

control  agencies.


Point Source  Models for Sulfur Dioxide and Particulate Matter
 (All  Averaging Times)

    Gaussian  models are considered to be state-of-the-art for sulfur

dioxide (SO-) and particulate matter (PM).   They are the best choice

                                   23

-------
 for  most  point  source  evaluations.   For all  point  sources  two  hierarchies
 of models  are recorrcnended  for use.   The first hierarchy  is composed
 of models  which can  provide a preliminary'estimate of concentrations.
 It is  recomnended  that such a screening process be applied to  all
 major  sources.  If it  is found from  the screening  process  that the
 source will cause  a  concentration  that is more than one-half of the
 NAAQS  or other allowable air quality increment, then that source should
 be subjected to a  more refined analysis.
     For flat terrain situations that have no significant meteorological
 complexities, there  are several standard publications and computerized
 models that can be used for screening.  Screening should also be
 applied in situations where more complex conditions exist, e.g.,
 inversion-breakup  fumigation, lake/sea breezes, aerodynamic down-
 wash, long-range transport (greater than 50 kilometers), and complex
 terrain.  Volume 10 of the Guidelines for Air Quality Maintenance
 Planning and Analysis, "Reviewing New Stationary Sources," has
 summarized techniques applicable to both flat terrain and more
 complex situations.  The techniques are presented in a useful format
 for the purpose of screening and are recommended for use.
    In those cases where a siore refined analysis is required  and
 there are no significant meteorological terrain complexities, the
 Single Source (CRSTER) Model  is  recommended for use.   However, if
 the screening process indicates  that the meteorological  or terrain
complexities cause serious  uncertainties,  then  a model  that is more
detailed or more suitable than  the Single  Source (CRSTER) Model should
                                   24

-------
 be  applied.   No  refined,  widely  available models  which  are  applicable

 to  complex  situations  are identified  here.   It  is  recommended  that

 each  complex  situation  be treated on  a  case-by-case  basis with  the

 assistance  of expert advice.

     If  the  data  bases  required to apply the  Single Source (CRSTER)

 Model are unavailable,  or if more refined models  applicable to  a

 complex situation do not  exist,  then  it may  be  necessary to base

 estimates of  source impact and the evaluation of  control strategies

 on  only the preliminary estimates.  In  such  cases, every effort should

 be  made to  acquire or improve the necessary  data  bases  and to develop

 more  refined  analytical techniques.

    Models  specified here and in the  following  subsection are also

 applicable  to  stationary  sources of lead pollutants, provided the

 pollutants can be assumed to behave as  a gas.


 Multi-Source Models for Sulfur Dioxide  and Particulate Matter
 (Annual  Average)"

    Due to the complexity of most multi-source  situations and the

 wide acceptability of several  models,  a screening process is not

 generally conducted.   If a preliminary assessment of the adequacy

 of a control strategy is desired, the  Rollback Model  amy be  used.

 However, in most cases  such a  screening does not constitute  an

 adequate control  strategy demonstration.

    The  Climatological  Dispersion Model  (COM), the Air Quality

 Display  Model   (AQDM),  and the  Texas  Climatological Model (TCM)

 are recommended for evaluating multi-source  complexes.   iMore

detailed or suitable  models may be used, especially in regions  with


                                   25

-------
major meteorological or topographic complexities.  If the meteoro-

logical or topographic complexities of the region are such that the

use of any available air quality model is precluded, the model used

for strategy evaluation may be limited to a Rollback model.


Multi-Source Models for Sulfur Dioxide and Particulate Matter
(Short-Term Averages)

    As noted in the previous subsection, a Rollback Method may be

used for the preliminary assessment of a control strategy.  The Real-

Time Air-Quality-Simulation Model (RAH) is recommended for eval-

uating the impact of multi-source complexes on air quality averaged

over short-term periods.   The Texas Episodic Model (TEM) may be used

if the data bases required to apply RAM are unavailable.  Also, if the

resources required to operate RAM or TEM are not available, the COM,

AQOM or TCM may be used to estimate short-term concentrations of SCL

and particulate matter.  These latter models should be used with pro-

cedures, such as that discussed by Larsen, to convert 3-hour and 24-

hour average concentrations from annual average concentration estimates.

Such statistical techniques are valid only in urban, multi-source areas

and should not be used in situations dominated by single point sources.

    A more detailed or more suitable model may be used,  especially

in a region which has major meteorological or topographic complexities.

If the meteorological or topographic complexities of the Region are such

that the use of any available air quality model is precluded, the model

used for control strategy evaluation may be limited to a Rollback Model.
                                  26

-------
Point, Line and Multi-Source Models for Carbon Monoxide
    The recommendations for point source screening procedures and
models are also applicable to evaluate point sources of carbon monoxide
(CO).  The models, procedures and requirements described in Volume 9 of
the Guidelines for Air Quality Maintenance Planning and Analysis,
"Guidelines for Review of the Impact of Indirect Sources on Ambient Air
Quality," are recommended for screening all mobile sources of CO which
fulfill the definition of an indirect source.  The indirect source
guideline is based on the use of HIWAY and other simple dispersion
techniques.  It is acceptable to apply these latter techniques, e.g.
HIWAY, independently of the indirect source guideline if it is found
that the guideline does not adequately consider .a wide enough set of
circumstances.   The indirect source guideline and associated tech-
niques are also applicable as a screening technique for large numbers
of mobile sources.  For example, the HIWAY and Kanna-Gifford models may
be used jointly to screen a large urban area.
    Specific refined modeling techniques  are not recommended here.
Situations that require more refined techniques should be considered
on a case-by-case basis with the use of expert consultation.  If a
suitable model  is available and the data  and technical  competence
required for this model are available, it may be used.
    There are many situations where an individual  indirect source
or a group of sources is well  defined so  that the indirect source
guideline is adequate to assess source impact and to evaluate control
strategies.   However, if a region-wide analysis is  necessary and there
                                  27

-------
 is no available or appropriate model,  the model  used for strategy
 evaluation in multi-source complexes (mobile and stationary)  may be
 limited to a Rollback Model.

 Point and Multi-Source Models  for Nitrogen Dioxide
     The recommendations for point source  screening  procedures  and
 models are also  applicable to  evaluate  point sources  of  nitrogen oxides
 (NOX)  under limited  circumstances.   The circumstances  require  an
 assumption that  all  NO  is emitted  in the  form of NO.  and  that HO  is
                       x                             ^             2
 a  non-reactive  pollutant.
     For sources  located where  atmospheric  photochemical  reactions are
 significant,  a  Rollback Model  may be used  as  a preliminary assessment
 to evaluate the  control  strategies  for  multiple  sources  (mobile  and
 stationary)  of NO^.   Another acceptable screening procedure for  multiple
 sources  is  to make an  assumption  similar to  that required for  point
 sources  and  then to use  a  model for non-reactive pollutants, such as COM.
    Specific  refined modeling  techniques are  not recommended here.
 Situations  that  require more refined techniques should be considered on
 a case-by-case basis with  the use of expert consultation.  If a  suitable
model  is available and  the data and technical competence required for
 this model  are available,  it may be used to estimate average concentrations
of N02-  However, if a region-wide analysis is necessary and there is no
available or appropriate model, the model  used for strategy evaluation in
multi-source complexes  (mobile and stationary sources) may be limited
to a Rollback Model.
                                   28

-------
Multi-Source Models  for  Photochemical Oxidants



    To estimate maximum  oxidant concentrations and to evaluate control



strategies, it is recommended that the ozone isopleth technique be



used  in all initial  assessments.  Other techniques (e.g., statistical



relationships) that  are  applicable at specific sites may be used provided



that  prior approval  is obtained from the appropriate EPA Regional Admin-



istrator.  The Rollback  Model may only be used as a means for demon-



strating that attainment of the standard will require at least all



reasonably available control technology.  Appendix J is no longer



considered an acceptable technique and should not be used in demon-



strating the effectiveness of control stragegy revisions.



    In order to obtain a more refined estimate and to calculate the



temporal  and spatial distributions of oxidants, it is essential that a



numerical model be used.  Numerical  models have the ability to mathe-



matically simulate the transport, diffusion and chemical reaction



processes that occur in  the atmosphere.   Specific refined modeling



techniques are not recommended here;  however, situations that require



more refined techniques should be considered on a case-by-case basis



with the use of expert consultation.   If a suitable model is available



and the data and  technical  competence required for the model are



available, it may be used.



    In many cases, the more refined  techniques  may not be practical



for routine application or may have  data requirements that cannot be met.



In such cases  the first assessment with  the ozone isopleth technique may



be considered  adequate for control strategy evaluation until such time
                                   29

-------
as the more refined estimates are available.   Generally all  acceptable



techniques which are applicable to oxidants are still  undergoing testing



and evaluation and have not been widely applied.   Consequently,  these



techniques must be used judiciously.
                                  30

-------
STACKS-
                 -How
                                                            F. W. THOMAS, Assistant Chief, Occupational Health Branch.
                                                  S. B. CARPENTER, Public Health Engineer, Occupational Health Branch,
                                                                      and F. E. GARTRELL, Assistant Director of Health,
                                                                Division of Health and Safety, Tennessee  Valley Authority
         Sc
         Jo many excellent papers have
been presented on the role of stacks for
the abatement and control  of air  pollu-
tion emissions, that a significant new
contribution   is   problematical.   How-
ever,  because  of  the important role of
stacks in air pollution control, a review
of TVA experience  in  the  performance
of stacks  for  dispersal and dilution of
power plant wastes  may be of special
interest.
   Principles for guidance in the design
of stacks for the modern  power  plants
were  aptly, implied  by R. S. Scorer1
when he concluded a short paper titled
"Plumes  from Tall  Chimneys"   with
the following:  "The gaseous products
of combustion are only harmful  when
they are at the ground.  The objective
should  be,  therefore,  to  get them as
high into the  air as possible by  means
of a very few  tall wide chimneys  and a
certain amount  of  buoyancy."  The
extremely  large  single generating units
now being built  with  capacities  up to
1000  mw offer a  promising opportunity
to apply these principles.
   Tall stacks cannot remove all vestiges
of air  pollution. However,  in  many
instances they can  limit  pollution at
ground  level  so that  no  harmful or
damaging  effects are produced.  Fre-
quently  the  minimal  pollution  from
power plants  is  considerably less than
that which would result from individual
home coal or oil heating  plants,  which
are replaced  by space  heating   when
low cost electric power becomes avail-
able.

 Basic Slack Height Criteria
   Air pollution  control considerations
are the principal determinant for stack
 heights in  modern power plants.  While
 fly ash may be effectively  removed from
 flue gas by mechanical collectors and
 electrostatic prccipitators, no  practical
 means other than  stacks  have boi-ii
 found acceptable for disposition  of  the
 large quantity of sulfur dioxide funned
in coal combustion.   Numerous factors,
many of whicli are not common for any
two plants, influence the choice of stack
height.
  A  basic  minimum  requirement  is
provision  of sufficient stack  height  to
prevent downwash   during  periods  of
high  wind velocity.  With stacks less
than twice the height of the main power-
house structure,  experience has  demon-
strated that, during high velocity  wind,
fumes  may be caught in the turbulent
vortex sheath and  brought to ground
level in relatively high  concentrations
very near the plant and sometimes re-
enter the  building  air supply.  Exten-
sive  wind tunnel tests and field experi-
ence have demonstrated that downwash
does  not  pose  a problem  where  the
stack height is  at  least  21/: times the
height  of the  powerhouse  or  other
nearby structures and appropriate eiflux
velocities  are  provided.   Thus with  a
I00-ft-high powerhouse,  a 250-ft stuck
should  provide  adequate limitation of
downwash.   In   commenting  on  the
value  of  this simple  criterion.  R. S.
Scorer- writes:  "The well known  21/!
times  rule concerning chimney heights
is commendable because it is  compre-
hensible as a working  rule,  it has no
precise theoretical  justification, and  if
experience proved  to be inadequate  it
could be changed by Act of Parliament.''
   For the large modern plants  which
may  emit SOO-1200' tons of sulfur  di-
oxide per day, higher stacks than speci-
fied  by the 2l/2 times rule are usually
required to limit ground concentrations
of sulfur  dioxide resulting from normal
atmospheric  diffusion   of  the  smoke
plume.  Environmental  ar.d operational
factors requiring consideration include
the  number,  size,  and  separation  of
units  or  stacks;  the heat and  sulfur
dioxide   emission   rates;  population
di'ii.Mty and urban development:  topog-
raphy  and  terrain; general  regional
meteorology and macroini'teorology  of
the  immediate area;  and land  use ,-uch
as forest  types  and agricultural prac-
tices.

Stack Height Estimation

Empirical Extrapolation from  Data
on Existing Plants

  TVA's  choice  of  stack  heights  has
been based largely on (/) a consideration
of environmental,  design, and opera-
tional factors for each site. (2)  an accu-
rate definition of dispersion  from  each
of its  previous  and usually  somewhat
smaller plants, and (o) extrapolation of
these data to a larger new plant with
selection of a stack height which  will
effectively limit  the  magnitude  ami
frequency  of maximum concentration:'.
A combination  of  complete data  on
plant   operations,   data  from   well-
equipped   meteorological  stations  :ii
each  steam  plant, and  sulfur dioxide
data  from 17 autometcrs  operated an
aggregate  of 111  autometer-years at
selected sites 0.5  to 20  miles  from .-is
steam  plants  with from one to  I"
generating units, has provided volumi-
nous background information on plumr
dispersion from  TVA  steam  plant.-.
This information, along with other d:it:i
from extensive  plume height  observa-
tions,   mobile  ground  level  sampling
with  a Titrilog in an automobile.  :""'
extensive dispersion measurements  in 3
helicopter, has  provided the basis !»•'
empirical procedures used in establish-
ing stack  heights for  new or  prupv"l''-
 for new and  usually larger plant.-;.  '•
 170-ft  stacks initially provided f"r 'j'
 first large steam plant, at Joluis"iiM;'-
Tennessee, in 1952, proved inadi'M"1"'
 193
                                                                                    Journal of Ihc Air Pollulion Conlrol

-------
  u
  UJ
  u.
  I
  >-
  I

  UJ
  *
  u
UU
00


00




uu



00
o
-





Ol OJ

^CONSTRUCTION STASTEO^ 2 «
-

to
A

Ol



—


	



UJ

IANOSN
0






UNITS 1-6 7-10
5^^
ot



O













UI
UJ
c
u
o
o



^
01

UJ
UJ
z
X
V)
1-4 5-6 7-8 I-D












^\o >n
01 01



K w
01 01





z
o
V)













t-
vu
CD
O
u

>o



cc.
UJ
V)
0










LATIN
t
0


















u
in
o
c=
s
60O'»


















Z
cc
-1
3
CO




—



-

1-4 5-9 1-4 5 1-4 1-4 1-2 1
          JOHNSONVILLE


          WIDOWS CREEK
                 •

          SHAWNEE
          KINGSTON
             MEGA-
             WATTS
UNITS 1-6   750
       7-IO   600
  '   '1-6   750
       7-8  1000
       I-10  1500
       1-4   600
       5-9  1000
                      COLBERT
                          •

                      JOHN SEVIEFt
                      GALLATIN
                          •

                      PARADISE
                      BULL RUN
           MEGA-
           WATTS
UNITS  1-4   800
        5    500
       1-4   80O
       1-2   500
       3-4   550
       I-'2  I30O
        I     900
 f[mvever, the possibility of this  inadc-
 i|ii:icy  was  recognized in  the  design
 .•tuge, and steel was sized for support of
 :i   100-ft  extension.  This  extension
 added  in 1955  relieved  objectionable
 downwash and recirculation conditions.
 Experience at the Galhtin Steam Plant
 i>l7ers a good example  of application of
 empirical extrapolations in establishing
 .-tack height  for a new plant.  On the
 l>:i-is of data from exi.-iting  plants,  two
 500-ft stack-s,  one  serving two  250-mw
 units and one serving two 275-mw units.
 wvre built.  Data indicated that with
 .-t;icks of this  height, ground level con-
 n-ntrations of SO- (30-min avg)  in the
 direction of most frequent plume  travel
 «'<>u!d equal or exceed 0.5 ppm only two
 tunes per year,  a  frequency of 0.01%.
 In 2l/4 yr of operating experience since
 tlii.s generating station reached design
 capacity, the frequency of 0.4, 0.5,  and
 1-6, 30-min avg concentrations as  meas-
 ured by  three autometers sited in  this
 Direction has  been 0.032,  0.016,  and
 0-007%.  All  concentrations   at  the
 ^••5 and 0.6 levels occurred within  a 3-hr
 Period on one day  when the wind  speed
 R'3* zero and  a  stagnant  high-pressure
 •\vitcm was  centered  over the general
 arca.  A similar  check  will  soon  be
 afforded  the  600-ft   stacks provided
 at the larger 1-100 mw Paradise  plant.

Mathematical Diffusion  Analyses
  This empirical approach is attractive
"> establishing power plant stack heights
                Fig. 1.  SfocV heighli, TVA iteam planri.

              in  the relatively restricted TVA area
              where a mass of meteorological and air
              pollution data has beca  compiled at  a
              scries  of  power  plants.   However,
              mathematical  analyses  of  dispersion
              offer a preferable  approach to general
              industrial  plants.    \Ve   believe  that
              recent advances in the measurement anil
              definition of  vertical and  horizontal
              dispersion rates  in all ranges of atmos-
              pheric  stability  present  the basic in-
              formation necessary for confident esti-
              mate of stack height  for any problem
              from general dispersion equations, pro-
              vided a realistic  estimate of plume rise
              is available.  Data and procedures for
                      such  mathematical analyses h:\ve been
                      effectively compiled and presented by
                      Gifford.1 Cramer.4 Pasquill.1 and others.
                         It  ia of  interest  to note  the  close
                      agreement in estimates of stack height
                      by several  methods of  analysis for a
                      power plant with assumed SO-; emission
                      rate of approximately SOO tons per day.
                      These are summarized in Table I.

                      Relative  Effectiveness of Stacks  in
                      Principal Dispersion  Models
                        The common objective in establishing
                      stack heights is  limitation of maximum
                      ground concentrations during all meteor-
                      ological conditions to a level and a ire-
                                                 Table I
                                      Method
                                        For Level Terrain
                                     Estimated Stack Height
                                    Feet Above Ground Level
              1. Empirical extrapolation  baaed on  normalized data  from
                existing TVA pbnts

              2. Generalized  diffusion equation  for maximum  ground level
                concentrations, i.e.
                                              TOO
                                       r     i  /^:
                                — exp    - T I — ;
                                (T,!f  H L     2  W
                and using dispersion coefficients presented by Clifford1

              3. Generalized diffusion equation  for maximum  ground level
                concentration
                       X =
                             rC,C.
b-[-T
                and using diffusion coefficients based on full-scale dispersion
                studies* of smoke plumes from TVA steam plants
                                              6911
    1943 / Volume 13, No. 5
                                                                                         199

-------
                        INVERSION -INVERSION BREAKUP
  z
  o
  z
  iu
  o
  O
  u
CO
 I
_J
UJ
UJ
  UJ
  >
<
_J
UJ
<=
                                                                             r-
                                                                             Z
      O.S
                                      MILES

                      CONING - STREAMLINE
                      -.V-^ RELATIVE
                            - ^CONCENTRATION
                            -a	
                               500

                               400

                               200
                        I
                                    2            3
                                       MILES
                        LOOPING
u
_l
UJ
o
z
3
O
fr
o
                                                                               O
                                                                               m
                                                                               uj
                                                                               iu
                                   2           3

                                       MILES

                            Fig. 2.  Ptume dispersion modeli.
 qiiency which will not adversely affect
 the  environment.   In  considering  this
 objective  it  is  pertinent  to  examine
 relative   pollution  levels   which  are
 expected   during   the   meteorological
 conditions of principal  interest; also to
 consider the relation of peak and time
 average concentrations.  Such data for
 one  large  steam plant for the coning.
 looping,  and  inversion plume  reported
 in an earlier paper are illustrated in
 Fig. 2.  Here it is of  interest to note
 the limited variation in the  magnitude
 of maximum 30-inin avg concentrations
 found to  have  the  following relative
 values:  coning  or  streamline  plume
 typical of neutral stability.  1.0; looping
 jihime present with strong thermal in-
 stability,  0.4:  and inversion breakup,
 0.0.   Peak concentrations as measured
 by the autoinetcr which has an averag-
 ing period of 2-3 min were two to three
 times the value of 30-min avg concen-
 trations.   As  indicated  by data graphi-
 cally  illustrated in  Fig.  2, the provision
of adequate  dilution  for  control  of
 pollution  during  high  velocity  wind
ne.utral  conditions  should  generallv
                                                                          other
 satisfy    requirements  during
 meteorological conditions.

 Neu/ro/ Condition—Coning Plume

   While  stack   height   requirements
 have been  primarily  based on  control
 during the high wind  neutral condition.
 effectiveness  during  other  dispersion
 models  should be  appraised.  For  a
 neutral condition  the extent of  ground
 fumigation is largely a function of effec-
 tive stack  height (actual  height  plus
 rise  of  the heated  plume)  and wind
 speed  for the critical condition.   Be-
 cause of  the   rapid  plume  ventilation
 aftorded   by  relatively  high velocity
 wind, the buoyancy due  to heat emis-
 sion is quickly dissipated and has  less
 opportunity to reduce ground fumiga-
 tions.  This obviously accounts  for the
 relatively high fumigation levels  associ-
 ated with this condition.   However, it
 is  of interest .to examine  the buovanev
effect  during  inversion  and  looping
conditions, especially for  the extremely
 large generating  units being built  for
current plants.
 Inversion Conditions—Fanning Plume
    \\ hili' the condition where  temper-.
 tua- increases with elevation, de.vril,,.,j
 as inversion, is of profound  important
 in the appraisal of air pollution  di.-p(..v
 >ion. our  experience  suggests  that j,
 may have been  unduly maligned. p>|i,..
 cially with respect to power plants.  Al.
 though inver.-ion has been described :i-
 the scapegoat in general  urban air p(,|.
 lution  for years,  the equally important
 role  of low velocity wind  is now  receiv.
 ing  recognition.   Also,  review  of  tin-
 literature suggests that a lack  of di-
 crimination may exist between ground-
 based radiation inversions and elevate,!
 subsidence-type  inversions.  \Vith  the
 exception of infrequent passage of warm
 high-pressure  systems,  subsidi>nce-ty|v
 inversions, which chronically  accentu-
 ate air  pollution problem? in southern
 California, are considered to be  a rela-
 lively minor factor with  respect  to air
 pollution  from steam  power plant? in
 the Tennessee Valley.
   Surface  radiation inversions  do, of
 course,  affect  the dispersion of  steam
 plant plumes.   However,  with  a few
 exceptions, we have considered this as a
 beneficial effect.   Duo to   large heat
 omission, steam  plant plume? generally
 risVarid  stratify SQO-1200  ft  above
 ground level where they arc diluted ami
 innocuously conveyed great distances by
 winds aloft.  AVith the heating  of the
 ground and establishment of convection.
 breakup and attendant fumigation occur
 in  midmormng.    However,  breakup
 fumigations have  been low  level and
 transient except  in one area  where tin*
 plume  traverses   a  plateau approxi-
 mately 1000 ft above the plant elevation.

 Beneficial Effect of Heat Emission

 Inversion CordiHont

   The  increase in generator size from
 about 100 mw to 1000 mw within thela.-t
 10 yrs.  with attendant increase in heat
 emission, may present a  practical 0|>-
 portunity   for effective   reduction  of
 inversion fumigations.  Instead of serv-
 ing  as  ceiling  which  holds  pollution
 near ground level,  use may be made uf
 inversion as a shield or isolation medium
 to  prevent  pollution   from  reaching
 ground  level.   The  proposed Bull Run
 Steam Plant near  Oak Ridge. Teniic."
 see, may  be  used to  illustrate  sue!'
 possibilities.  Initially a single  900-niw
 unit will be constructed at this location.
Subsequently one or more units,  pos.-:-
 bly of  even larger capacity, may  I"'
added.  To provide assurance  of adi1-
quute dispersal from the ultimate  plant.
the initial unit will be sened  by a his:'1
stack  dc.-igncd  to prevent  significant
fumigation  during  neutral   coiulitinii-
whon  critical  wind  speeds exist.   Tin1
height of stack to be provided is ii»"
under studv.
200
                                                                                      Journal of the Air Pollution Control Aiiocia''°"

-------
   1500
   1000
    500
                                 FALL  1949
    SFC
        46       50       54       58        62        66       70
                                TEMP. - °F
                            FALL  1950
    SFC *
        56       60       64       68        72        76
                           TEMP -  °F
             Pig. 3.  Vertical temperature profilsi—Oak Ridge, Tenneiiee.
.  1000
                                          INVERSION  CONDITIONS
          5        6         7        8        9        10       II

                               WIND SPEED - MPH

                       Fig. 4.  Plume riie, Colbert Stjom Plcnt.

May 1943 / Volum- 1 3. No. 5
                                                                        12
   Fortunately  tin1  temperature profile
in this area up to elevation  1500 ft has
been  well defined  by tin.- Meteorology
Section attached to Al-'C :it  Oak Ridv;e.
Figure  3 illustrates  average  seasonal
temperature  profiles taken at  OWO.
0900, 1500, and 2100 in the fall of 19-10-
50.!   Extensive  data    reveal   that
temperature inversions rarely extender!
above 500 ft at OoOO and 2100.   In the
500-  to  1000-ft zone the temperature
gradient varied  from  isothermal  to
some less than adiabatic.  Above  1000
ft, the gradient approximated the refer-
ence  adiabatic  slope  in 1949 and  \vas
essentially  isothermal in  1950.   These
data  on  variation of  temperature  with
elevation along with  information  on a
plume rise  during full-scale disptT.-ion
studies  at   the  Colbert  Steam Plant
provide a frame in which the benefit of
plume rise due to heat emission  may he
examined.
  Figure 4 illustrates  the  range  of
plume rise  observed  during inversion
conditions on eight sample days at the
Colbert plant with three 200-rnw  units
in operation, each  with  a  300-ft stack.
\Vith a 6-mph  wind speed  at the plume
centerline,  the  rise  was approximately
600  ft.  A  comparable  rise above a
much  higher  Bull  Run  stack  should
carry the plume to an elevation where
temperature gradient Ls neur aciiahatic.
Actually the rise  from  tlin Kull  Run
stack should be appreciably  greater be-
cause (/) buoyant force is not dissipated
in penetrating the  intense  inversion
below 500  ft.  and  0?)  heat emission
from  the single  900-mw unit  is approxi-
mately  four times that from a single
200-nuv unit at the Colbert plant.
  Actual temperature profiles and plume
elevations for two of  the days on Fig. 4
illustrate the penetration  of a  surface
inversion by a steam plant plume, Fig. 5.
Plume measurements and temperature
profiles  were  taken   from  a specially
instrumented helicopter.   On Septem-
ber 24. the  top of the plume  was strati-
fied at about 500 ft and did not penetrate
the inverted temperature region.   On
October S,  the more  intense  surface in-
version  was penetrated and  the top  of
the plume stratified at about elevation
1600  where  the lapse rate was slightly
less than ariia'oatic.   It seems  reason-
ably  certain  that  a  five-fold  increase
of the unit heat rate, as estimated for
200-  and 1000-mw  units  and  a stack
height of, say,  500-700 ft. would  have
carried the  plume ivcll above the inver-
sion  and isothermal  strata into a  rela-
tively unstable near adinbatic   area
where upward dispersion would be en-
hanced and downward difTu.-ion blocked.
As a  matter of interest, a 500-ft stack
has been provided recently  for a  new
500-mw unit at this 'plant.

                                 201

-------

?snn
i_ onnn
bJ
u.
z
— i**nn
H
X
UJ
x tnnn

sno

^Pr
5









0 5
9-2




PLUME
\NN
'//
• 	 ^
5 6
»-57



\
TOP>, %
/ }
f/////)s.
\
08<0 — '
0 6




06O3
V
{


1 PLUME
^BOTTOM
»
5 7
TEMP









0 5
. -°F



PLUM



'/*
060.^
0 5
IO-a-57

o.«
£ TOP^V
S\\\\\\^


'//**r-?7/,
PLUME
BOTTOV
5 6




\

x»
X
/ X^XX/
/I
o
0 6
         Fig. 5.   Vertical temperature profile and plume elevation, Colbert Steam Plant.
Unsfab/e ConaWons—
"Looping" Plume

  According to the data  from one  of
our  large  steam plants,  illustrated  in
Fig.  2.  30-min  avg  ground  concentra-
tions during unstable low wind  condi-
tions when a looping  plume exists are
lower than those  in cither  neutral  or
inversion  breakup fumigations.  These
data  were taken at a plant with 250-
and  300-ft stacks.   In the analysis  of
similar  data from another plant with
two  500-ft  stacks,  a  point  of special
interest is that no significant difference
exists in maximum 30-min avg concen-
trations for neutral  and looping  condi-
tions.  This  suggests  that  as  stack
heights  are increased  the dilution  ef-
fected in neutral high velocity wind con-
ditions was no greater than that associ-
ated with thermal;- in looping conditions.
  The above description of fumigations
as looping according to Church' is not
considered fully appropriate to  plume
patterns from large steam plants.  Due
to the large heat emission, the  plume is
rarely brought to ground level  by small
thermals.    Fumigations  described   as
looping  are associated with large ther-
mals. loops of  several miles,  and are
not particularly transient.  The magni-
tude  of observed circulation  induced
by thcrninls  acting  on relatively small
power plant plumes  suggests that even
greater  dilution may  be  provided  by
the larger  plants now under construc-
tion  or being planned.
  When thermal convection is present,
the heat emission from  a large  power
plant or  similar  energy  source  may
initiate  thermal.-' and,  in fact,  establish
a semi-independent  local  atmospheric
circulation.   In this circumstance  the
plume may rise several thousand  feet
 and  often  produces  cumulus clouds.
 During the release of neutral constant
 volume tetroons at the Colbert Steam
 Plant, ascent in thermals  up  to 5000
 ft was observed.
  While a heat  source equivalent to 50
 mw over a 500-meter-square  area  has
 been  described  as  adequate  for  the
 initiation of thermals,  the  equivalent
 of  200  mw  of  heat may  be  released
 from  a single 1000-nw  generating unit.
 Obviously this four-fold greater energy
 source  should   produce  stronger  anil
 larger thermals—and result in further
 reduction of pollution  at ground level
 during unstable  conditions.

 Mathematical Formulation of Plume
 Rise

 Implications of Common Formulae
  Numerous formulae  for  calculation
 of plume rise have been  presented which
 attempt to take into account the effects
 of heat load, wind velocity, atmospheric
 stability,  and  other variables.  Vari-
 able  temperature  gradients,  such   as
 illustrated in Fig. 5, suggest that such
 formulae  are  at best approximations.
 Holland11 has presented  one of the more
 popular formulae for plume rise in aver-
 age conditions  which is attractive  be-
 cause of its simplicity.   In this formula,
h =  (1.5 rd +  3  x 10-' Qtf)/i/,,  the
 plume rise is inversely  proportional  to
 the wind speed  and, in  part,  propor-
 tional to the heat emission  rate.   In
 England,  Lucas, Moore, and  Spurr10
 have  conducted  careful and  extensive
studies leading  to the following expres-
sion for  plume  rise:  h  =  A.V2"4.i//i'.
 where K was found to have values  rang-
 ing  from 3000 to G200 for different prr.ver
stations.   These  studies in   England
suggest that the plume rise is much more
 dependent on wind speed than on |,,...
 content.
   It is of interest to note the o.-l.-ui\.
 values of plume  rise given  by tin-.
 formulae for, say, 100- and  lOOD-n,,.
 units.  The formulae developed at t|.
 Earlcy Station in England suggest th,-
 the plume rise from a 1000-iuw unitt
 about 1.7S times that for a 100-mw u,,\.
 According to  the  Holland  formula. :,':
 eight-fold increase in plume rise wi,(i;.-
 occur.
   An intriguing prospect of these plun,..
 rise  formulae is  their  relation  to  th.
 maximum ground  level concentratimi-
 For example, Fig.  G illustrates the rc\-,..
 live  maximum ground concentnuii,,,.
 estimated for  100-  to  1000-nw unii-
 with actual  stack  heights of 300. 5
 preciable  increase in ground concentra-
 tion with increase  in unit size.  These
 data  suggest  that  plume  rise  due t<>
 heat emission may be less than generally
 estimated.

 Estimate from TVA Data
   Data compiled  from  meteorological
 stations  and  networks  of  autometor-
 at five TVA steam  plants may be uso!
 for indirect approximation of the etTn't
 of heat emission on plume rise.   Exten-
 sive data have confirmed  that an  in-
 crease in  station size  as based on tin-
 number of  generating  units  does  ii"i
 result  in  a  linear  or proportional  in-
 crease in  maximum ground concentra-
 tions of 30.;.   These data, illustrated in
 Fig. 7, indicate that  an  increase  ii'
 number of units from one to ten h:is re-
sulted  in  a 5.3, say 5.0, rather than :1
 ten-fold increase in ground conrc-ntrn-
 tion.  This  nonlinearity  is  attributi1-!
 to (1) line source effect due to an avenip"
separation of about  SO ft between stack-
and  (;?) increase in plume  rise due '"
greater heat emission.   This relat'n'ii i-
cssentiallv unchanged for wind clireriii'11
202
                                                                                     Journal of the Air Pollution Control AliOt-ial"'rt

-------
       1.0
      0.8
      0.6
      0.4
0.2
o
z
o
o
UJ
>
u       o
rz
      0.6
      0.4
      0.2
                  Fig. 6.  Effect of plume rile

imrmal to or in line with the stacks.
^iggesting that the larger buoyant ef-
fect  for  in  lino winds approximately
compensates for line source eft'ect with
wind normal to the line of stack.?.  As
a first  approximation, the assumption
may be made that the nonlinearity re-
sults solely from greater buoyancy and
tlius higher effective stack height.  Ac-
tvpting the premise that the maximum
Around concentration is inversely pro-
portional to the effective  stack height.
(lie plume rise attributed  to heat emis-
sion may be calculated for. say,  a one
and  a  10-unit  plant, according  to the
relation C =  M/H"
   One-Unit Plant    Ten-Unit Plant
       1
                200       400      600      800       IOOO

                 CAPACITY  -  MEGA-WATTS
                                  on ground concentrations.

                                   where C  equals  the  relative  maximum
                                   ground concentration, and //  equals the
                                   effective   stack  height   (actual  stack
                                   height  plus plume rise).
                                     The  typical plant from  which these
                                   data were derived has an actual stack
                                   height  of  about  250  ft.  and plume rise
                                   in  a  15-mph wind  representative  of
                                   maximum  recorded  fumigation condi-
                                   tions has  been established  by observa-
                                   tion at about 150 ft.  Thus, //i Defec-
                                   tive stack height) is 250  4- 150 or 400 ft.
                                   For .a  ten-unit  plant and  a  five-fold
                                   increase in gruund concentration. Hv> =
                                   (l.-ilXr/t) = 564 ft; and the plume rise
                                   is 564  - 250 =  314 ft.  This represents
                                   about  a 2.1-fold increase in  plume rise
                                   from  a  one- to  ten-unit  plant.   Ex-
                                   pressed as an exponential,  the increase
                                   in plume rise is proportional to the heat
                             1-41
    1963 / Volume 13. No. 5
emission, calories per second,  raised tu
0.30 power.  Since  a  part of the ob-
served  nonlineaiity  in  i-onecntrit.ion
increase must be due  to separation of
the stacks,  this analysis sui:u;e.-ts that
a plume rise  proportional to the 0.2~>
power of heat emission observed  in the
Uritish power stations  may very closely
fit our experience at gcncratiiij: stations
with one to ten units of 125- to 250-rnw
capacity.
  Where the increase  in  heat emission
is due to larger units rather than more
units, it is reasonable  to expect  that a
greater  plume  rise  will result and  an
exponential some greater than 0.25 may
be  applicable.  While  the studies  of
plume rise at I3riti>h  and TV'A  power
stations demonstrate that increases in
heat  emission  have much less  than a
linear relation to plume rise, the resi-
due to  buoyancy is an important ele-
ment which  needs  better definition  for
proper recognition in design studies.

Efflux Velocity
  A  trend  for the  use  of increasingly
greater  flue gas exit velocity  is evident
in the design of steam plants over recent.
years.  In some circumstances, such as
legal restriction  of stack heights  for
airways control, there  may be no alter-
native to the use of higher exit velocities
for obtaining necessary plume dilution.
In  other  instances existing   structures
may be incapable of supporting  higher
stacks.
   For large  steam plants where  stacks
greater than 21/-  times  the  height of
powerhouse are provided for dilution of
the plume, the advantage and the effec-
tiveness of relatively high cfthix  veloci-
ties are considered questionable.  As a
test of higher exit velocities for increas-
ing effective stack  height, a nozzle was
installed  atop the stack of   a 150-mw
TV'A unit increasing the velocity from
45 to 90 fps.  Observations  and com-
parison  with  adjacent  units without
nozzles  revealed  no  appreciable  in-
crease in  plume rise-during inversion or
high velocity  wind conditions.   \\ Ink-
some benefit was  detectable  with me-
dium velocity winds, 5--S mph. dispor-iim
during  this  meteorological  model  is
generally of less, concern than others.
   For heat to make its greatest  contri-
 bution  to plume rise,  dynamic  mixing
 of  the plume1 with ambient air .-hoiild It:
 delayed as long  as possible.  Tim- in-
 creasingly higher efflux velocities which
 accelerate  the mixing of a   hot  plume
 with relatively  cool  ambient air may
 actually  reduce  the  rise of a  heated
 plume.   A velocity of 50 to  t>0  fps has
 been conventionally  accepted  as ade-
 quate to prevent  significant  downwa-h
 in  the  lee  of  the  stack.  Also,  for
 velocities in  this  order  the available
 draft in  the  stark.-  is approximately
 equivalent  to the  draft  loss  in  the
 stacks.   Higher velocities. >ay 00 to rJO

                                   203

-------
6.0
§ 5.0
UJ
5 4.0
UJ
rr
S 3.0
rr
** Z.O
2
3 1.0
0
C












X




t





S
S




X*




S





+'**






















'" 23456789 10
UNITS ON LINE
 Rj. 7.  Mo«imum overage r«lative ground concentration or 0.01 % frsquency at o large power plant.
 fps,  will normally increase  the  fan,
 and  motor size and costs, and chimney
 maintenance  costs, and  may  inhibit
 plume  rise due to buoyancy.

             Nomenclature
 Symbol
              Meaning
  Units
  X   Concentration
  Q    Source output

  o-.   Horizontal diffusion
         coefficients
  »,    Vertical diffusion
         coefficients
  u    Wind speed
  //    Plume height  above
         ground  level
  K,   Stability parameter
  K.   Stability parameter
  K    K, 4- K.
  r    Horizontal distance
         from source
  A    Plume rise above
         stack top
  r     Stack gas  exit
         velocity
  rf    Stack diameter
  QJI   Heat emission rate
  HI    Wind speed
  /v'i    Constant based on
         station parameters
  Q.\r   Heat emission  rate
  .)/    SO; emission
                              ppm
                              cu ft 30;
                                X 10'
                              ft

                              ft

                              fps
                              ft
ft

ft

mph

ft
cal/sec
mph
nuv
tons/day
    rruting Atmospheric Dispersion," A'u-
    cleir Safety, 2: 4 (1061).
 4. H. E. Cramer,  "A Practical Method
    for Estimating the Dispersal of Atmos-
    pheric _ Contaminants,"  Proceedings,
    First .Vational Conference on Applied
    .Meteorology,   Hartford,  Connecticut
    (1957).
 5. F. Pas_quill, Atmospheric Diffusion, D.
    Van Xostrand Co., Ltd. (1062)
 6. F. E. Gartrcll,  F. \V. Thomas,  and
    S. B. Carpenter, "An Interim Report
    on Full-Scale  Study of Dispersion of
    Stack Cases,"  J.'Air  Poll. Control
    .•U»oc., 11:2,00-65(1961).
 7. F. E. Gartrell,  F. W. Thomas,  and
    S. B.  Carpenter,  "Transport of  SO;
    in the Atmosphere  from a  Single
    Source," Monograph .\'o.  3, American
    Geophysical Union (1959).
 S. J. 7.. Holland, et al., "A Meteorological
    Survey of  the  Oak  Ridee Area,"
    ORO-90.
 9. P. E. Church  and  C.  A. Cosline,  Jr.,
    U. S. Atomic  Energy  Commission,
    Document MDDC-73.
10. D. H. Lucas, D.  J.  .Moore,  and G.
    Spurr, Central  Electricity Research
    Laboratories,  Leatherliead,   L'npub-
    lished Report.
            REFERENCES
I.  R.  S.  Scorer,  "Plumes  from  Tall
  • Chimneys."  Weather  10,  No. 4,  10(5
   (1954). '
'-'.  R. S. PcofL-r, "The Behavior of Chim-
   ney  Plumes."  Int. J.  Air  foil.,  1:
3.  F.  A.  CilTord,   "L'sc  of   Routine
   Meteorological Olijftrviitions fur Kgii-
                                               HEMEON ASSOCIATES
                                               Air Pollution Research Engineers
                                            APPRAISALS
                                              Control Equipment Perloimancs
                                                Flue Gas and Air Cleaning
                                              Stack  Emission Inventories
                                              Community Surveys . Tracer Studies

                                            ENGINEERING      	

                                              Oust. Fume, Odor Control
                                              Incineration . Catalytic Oxidation
                                              Activated Carton Applications
                                              Scrubbing . Filtration

                                                     W. C. I. Hemeon
                                                          Director
                                            121  f.loyran Ave., Pittsburgh 13, Pa.
                                                                                      filOMITORlMG  PROGRAM
                                                                                              (Continued from p. 10?)
                                                                                         Table  I!—Yearly Averages Of
                                                                                      Airborne  Radioactivity Since 1953
                                                                                      (Beta Emitters, Picociiries per Cw&i'r \l.,,.
1953
1954
1955
1956
1957
0.2
0.6
1.1
2.0
4.3
19.53
1059
1900
1961

— : —
•i it
0 •>
•j ,-

  levels since 1953  based  upon  tin- i
  weekly sample collection.
    As was stated in the introduction (.
  this paper, the  interpretation of tin- •..
  diation  levels  as   they  influence  i1'.
  general health of the people of Cincim,',-
  is the responsibility of the Health Di'|,;,-.
  ment.  Monthly reports released to t:..
  press by  the Bureau of  Air  Polluii..'-.
  Control, which carry the high radiom-u-..
  fallout levels, are accompanied by ?\;-.\..
  ments  from  the  Health  Dcpartr.n-t,:
  This procedure worked out well in Cnn-i:..
  nati as the press was interested in t!..
  high fallout levels but there was no pul.:.
  hysteria developed  because of the  hi-
  ratio  of  increase  (nearly 1000  lin.r.
  from  periods of  low normal backgrou:..
  to the high levels of fallout expericniv.;
  briefly  following   the  Soviet   to.-ii.-..-
  program.

  Summary
   In a number of communities  throu-;!.-
 out  the  country,   the  Air  Polluti...-.
 Control Officer has inherited the ra«li--
 logical  protection   assignment.  TL-
 appears to be a logical mission for an :..-
 pollution control agency since one nf ::•
 peace time missions is the asses.-mrnt •
 contamination   of    the   atmospli'-:-
 Radiological fallout  is atmospheric I-K.-
 [animation  in   its   truest  sense.  A-
 public concern over  fallout either (>'•:
 .veapons testing  or  from an emero-ri'.
 situation grows,  demand for morr :>:. •
 more information concerning radi"!'-:
 cal fallout and its effect  upon the (•••:
 munity will arise.   In the future ni"~
 and more  APC  agencies  thus will '•••
 asked to take on the radiation inonit-iri:.:
 and civil defense RADEF assignim-:-.:

 V. P. GEOPCEVIC  RESIGNS
   On January 15, 1963, the officer- :•..' :
 members  of the  Incinerator In.-tif.'.
 of .America  accepted with regn-t.  : '
 reasons of  health,  the  resignation  •
 V.  P.  Geopccvic \vho  has scrvi-'l  •'•'
Secretary-Treasurer since 19oO.
  The  Executive  Committee  i-i' '•'
Incinerator Institute of America a! :;
same  time  announced the appoint!"-''
of  Organization  Service  Corp'ir.ic-
as nianagpinent  for  the  affairs »f '•'
Institute,  with Donald V.  Pavd.  I1'1 '
idcnt  of OSC, acting in the c:iF':'
of Soi.-retary-Treasurer of the I:irin''r:-:-
In.-tituteof America.
204
                                                                                     Journal of Ihe Air Pollution Conlrol Allot'-'

-------
III.   PARTICULATE CONTROL

-------
        REMOVAL  OF  PARTICULATE MATTER  FROM
                             GASEOUS  WASTES

        GRAVITY,  INERTIAL, SONIC,  AND THERMAL COLLECTORS
1.00   GRAVITY SETTLING  CHAMBERS

  Probably the oldest device for the removal of
particles  from gases is the  gravity  settling
chamber.5"' "• '*•i:- -3  The device, in principle,
is simply an enlargement in the duct carrying
the gas where the gas slows down and remains
for a sufficient length of time to let the particles
settle out. Because of its simplicity, the cham-
ber has a small pressure drop, resulting mainly
from  the entrance and exit losses.
  The lateral gas velocity distribution must be
uniform and eddy currents must be minimized
to  achieve the maximum collection efficiency.
To produce such a uniform gas distribution,
gradual   inlet  transitions  (diffusers),  guide
vanes, and distribution screens or perforated
plates are used.3  Eddy currents can be reduced
by means of  curtains, rods, or screens sus-
pended in the gas stream.  Maintaining the gas
velocity  below 10 fps also helps to reduce eddy
currents  and  reduces  the possibility  of re-
entrainment of the particles which have settled
out.
  Since gravity is the only force acting to settle
out the  particles, the  simple gravity settling
chamber becomes impractically large for par-
ticles finer than approximately 40 microns. If
horizontal, equally spaced shelves are installed
in the chamber,  the collection efficiency is in-
creased  markedly because the dust does not
have  to  settle so far.  This  arrangement is
known as the Howard dust chamber'  and can
be used for particles as  fine as 10 microns.  The
Howard dust chamber suffers from two  disad-
vantages: elevated temperatures tend to warp
the shelves, and  the close shelf spacing  makes
cleaning difficult, thus limiting the practical
inlet dust load to 1 grain per cubic foot, or less.
   Except for some applications to natural draft
exhausts from furnaces and kilns,  gravity set-
tling chambers are seldom used. There is no
such  apparatus  currently on  the market ;-2
therefore, if one is to  be used, it must be spe-

  • Figures refer to REFERENCES on p. 12.
cially designed for the particular application.
  The theory of gravity settling  is relatively
simple. It is based on the idea that if a particle
is to be completely removed in the chamber, it
must be capable of settling the distance H, the
height of the chamber, in the time that it takes
the gas to flow through the chamber.  The usual
assumptions made in the theory are: that the
velocity profile is constant over a lateral cross-
section,  that there  are no  eddy currents  to
disturb the particle  as it settles, and that the
particles are not re-entrained after settling out.
  Based on these assumptions (and Fig. 1), the
time required for a  particle  to settle out is
                                      (1)
Where:
   B = breadth of chamber.
   H= height of chamber.
   L = length of  chamber.
   q= volumetric flow rate  of the gas at  the
         actual temperature and pressure.
    t= settling time.
   u0= terminal settling velocity of the particle.

Solving  for the particle terminal settling  ve-
locity,
. q
•LB
                                       (2)
       FIG. 1	Gravity Collecting Chamber.
SOURCE :   Removal  of Particulate  Matter  From Gaseous  '-Jastes--Gravi ty,  Inertia!,
          Sonic,  and Thermal Collectors,  Engineering  Report  Prepared  for
          American Petroleum Institute,  New York, NY  (1961).

-------
                              AMERICAN PETROLEUM INSTITUTE
  The particles which are most difficult to set-
tle out are the finer  ones, and they normally
obey Stokes' law of settling; therefore,
            1, _ Q  _Dpm;(pl.-p)g
             °~LB~     Ify
                                         (3)
Where:
  D
   pm= diameter of  the  smallest size particle
         that is completely settled  out.
   pp=particle density.
    p = fluid density.
    g= local gravitational constant.
    u = absolute viscosity.
Solving for Dfn,
                      LB(Pt-p)g
                                         (4)
  The collection  efficiency,  i;, for particles of
size  D,m will  be  100 per cent.  For particles
finer than £>„„,, the collection efficiency  will be
proportional to the terminal velocity ua; there-
fore, by application of Stokes'  law,

                                         (5)
                   'I- T)  5
                      JL/pm

  The collection  efficiency of a  Howard dust
chamber is greater than that of a simple cham-
ber because  of the reduced  settling distance.
If N  is the number of equally spaced  shelves
(including the bottom surface) installed in the
chamber, then
                t =
                    H   LBH
Solving for u0,
                  U°~NLB
                                         (6)
                                         (7)
Again, by applying Stokes' law and solving for
Dfm, it is found that
           Dpm"y NI
                    NLB(pp-p)g
Thus, for example, by making N equal to  16,
the minimum particle  size,  Dfm, is  reduced  by
a factor of 4.

     2.00  INERTIAL COLLECTORS

2.10  Introduction

  As a class, inertial collectors depend on the
fact that it is easier to change the direction of
motion of fluid streams than of particles being
carried by that stream. Thus, if the fluid must
change its direction to avoid an obstacle, the
particles will tend to strike the obstacle, adhere
to it, and as a result be separated from the fluid
stream. The most important inertial collector
is the  cyclone dust  collector, which  has been
discussed  in  great detail in an earlier publica-
tion  of the American Petroleum  Institute by
A. C. Stern, et  al.-°  Because of this, cyclone
collectors  will not be discussed  in this report.

2.20  Skimmers

  A  skimmer is an  elementary  inertial  collec-
tor.  It consists of a chamber in which the gas
is forced to reverse its path  (see Fig. 2).  The
result  is  that the  dust particles tend  to be
thrown out of the gas stream. In the skimmer
shown  in  Fig. 2, part of the gas is  drawn off
with the dust particles to help carry them into
the  dust hopper.  From the dust hopper this
gas is  vented to the outlet of the skimmer.
  From Fig. 2  it can be seen  that skimmers
operate somewhat like cyclone  separators  ex-
cept  that  the gas  stays  in the device  for a
much shorter time;  also,  they are  relatively
large in diameter compared with high-efficiency
cyclones.  Accordingly, they cannot be expected
to act effectively in removing fine particles from
gases.  They are effective for relatively  coarse
dusts and are used principally for:
1. Reduction of  dust loading to more efficient
collectors.
2. Primary collectors for relatively coarse dusts
when air pollution is not a problem.
                                                      AIR
                                                      IN
                                                 Courtesy of American Air Filter Company.
                                                                FIG. 2—Skimmer.

-------
                  REMOVAL OF  PARTICULATE  MATTER FROM GASEOUS WASTES
        FIG. 3—Rectangular Louver Dust Collector.

  From the theory of cyclone  separators, one
would expect skimmez's to have high efficiencies
for particles coarser  than 50  microns.   For
particles finer than 50 microns, the  collection
efficiency falls off rapidly.

2.30  Louver-Type Dust Separators

  The louver-type dust separator is a relatively
recent  development.  There are two  basic de-
signs for this collector. One consists of a trun-
cated cone which  is approximately  five  times
as long as its maximum diameter.  The mini-
mum diameter is about a fifth of the maximum
diameter.  The conical surface is made up of a
large number  of louvers.  The louvered  cones
are  commercially  available  in  sizes  ranging
from 9 in. to 25 in. at the maximum  diameter.
The other design consists of a "Venetian blind"
structure enclosed in a box  (see Fig. 3).
  The  principle of  operation  of  the cone  is
illustrated in Fig. 4.  The  dusty gas enters at
point A, flowing down the  cone at a substan-
tially constant  velocity.  Approximately  5 per
cent of the gas, the  blowdown,  issues from
point B.  The  remainder of the  gas passes
through the louvers.  In  order to do  so,  the
gas must  change its direction abruptly.  The
inertia of the dust particles prevents them from
following  the  gas  so  that  the  dust  particles
tend to stay inside the cone and  finally issue
from the cone  with the blowdown gas. Thus
the louver-type dust separator acts as a con-
centrator rather than as a collector. The blow-
down stream  must be treated further by  a
secondary collector to  complete  the separation
of the dust from the gas stream. A small, high-
efficiency cyclone collector is normally used as
the secondary collector. The cleaned blowdown
gas from the cyclone is then recirculated to the
louver  separator  to recover residual  dust not
removed in the cyclone collector.  Some  dust
collects  on the louvers.  This  is  removed  by
rapping the cones every 8 hr or so.
  A small  amount  of  dust collection  efficiency
data has   been publishedl  for  the  conical,
louver-type separator combined with  a cyclone
secondary  collector. These data are shown in
Fig. 5. The data indicate that louver-type sep-
arators have efficiencies comparable to medium-
to high-efficiency cyclones.
  It has been  reported that the collection  effi-
ciency  is a function of the Stokes'  law settling
velocity of a particle  rather than  the particle
diameter alone.  Thus, to  correct Fig. 5 to the
conditions  of a different gas viscosity, density,
EFFICIENCY V.
_Ijo.»i><»-««'<>£
oooooooooc





/
/
/
/
/


/
/
/





x
/
00
Sf
SP
^



r 4lfl AT 70"
MEHICiL PiKTlCLtS"
CCIfIC GRAVITY - 2
OUST CONCEMTSariON -
3 Gf>^















Courtesy of  Research-CottreU, Inc.

FIG. 4—Operating Principle of Louver Dujl Collector.
     0        5       10      15      2C

             PARTICLE  SIZE IN MICRONS

Courtesy of Research-CottreU,  Inc.

   FIG.  5—Louver Dust Collector Efficiencv Data.

-------
                               AMERICAN  PETROLEUM INSTITUTE
and particle density, it is necessary to multiply
the  particle sizes shown on the figure by the
ratio
                  , /V PP—p
                 V 7 ,,„'-/
where the unprimed quantities are based on the
conditions  given in  the figure and the primed
quantities  are  based on the new set of condi-
tions to which the figure is being corrected.
   Uzhov :: has reported some performance data
for  Russian designs of a  rectangular  louver
collector.  His  data  indicate that the Russian
design is  of  much  lower efficiency  than the
conical collector.
   Contradictory results have  been reported on
the  effect of the inlet gas velocity to the col-
lector. Smith and Goglia:i found approximately
a  1-per-cent  decrease  in  collection  efficiency
with a threefold increase in  inlet  velocity.
Zverev,16 on the other  hand, found  a 30-per-
cent increase in collection efficiency for a three-
fold increase in inlet velocity.
   Smith and  Goglia =3  and Zverev:" were in
agreement  that the  ratio of the blowdown ve-
locity to the inlet velocity had a marked effect
on the collection efficiency.  If the ratio is re-
duced  below one, the efficiency drops off very
rapidly.  If the ratio is increased above one,
the efficiency  increases slowly.  The  moderate
increase can be accounted  for largely by the
fact that more of the gas  is  leaving as blow-
down.
   Both the foregoing sources claim that  an
important  mechanism involved in the separa-
tion of the  dust from the gas is  that the par-
ticles strike the louver and  then rebound back
into the dusty gas  stream.  Smith and Goglia
observed these particle  trajectories by shining
a  strong light  at right angles to  the viewing
angle,  causing the particle  trajectories to ap-
pear as bright streaks of light. Furthermore,
they noted  that the  louver  surface became
brightly polished as  a result cf the particle im-
pacts.
   Despite the fact  mentioned previously that
the particle impacts polish the louvers,  it is
claimed that the cone-type louver collectors are
resistant to erosion.   One user reports that no
erosion of the cone was observed after a year
and a half in fly ash recovery  service.  In addi-
tion, only  moderate  amounts of erosion of the
cyclone and the fan which returns the gas from
the cyclone to the cone were observed.  No other
maintenance problems were experienced  with
the unit.
   Pressure-drop  data for the conical, louver-
 type  separator have been  reported l  and  are
 shown in Fig. 6.
   No serious attempt at a  theoretical analysis
 of the performance of these collectors  has been
 reported. One of the difficulties involved is that
 it would be necessary to compute the  gas  flow
 lines  first in order to determine their  effect on
 the motion  of  the particles, the fine  particles
 being particularly sensitive to the motion of
 the gas.  The gas flow lines are quite difficult
 to compute  because  of the complicated  flow
 boundary and because  of the turbulent eddies
 generated at the downstream edge of each of
 the louvers.  Thus, performance characteristics
 of louver separators must be measured experi-
 mentally.
   In  summary, the louver-type separator has a
 collection efficiency comparable to medium- to
 high-efficiency cyclones at pressure drops some-
 what lower than  those reported for  cyclones.
 They act only as dust concentrators and must
 be used in conjunction with a secondary collec-
 tor such  as a small, high-efficiency cyclone.

 2.40   Centrifugal  Impeller Collectors

   There  are a few collectors which combine the
 functions of a dust remover and of a  fan.  An
 example  of such a collector  is shown in Fig. 7.
 Dusty air is  drawn into the collector at a high
 velocity  and flows radially outward  between
 specially designed impeller  blades to the outlet
 scroll. As the gas passes  through the  collector
 it turns  through  an  arc  of a little less than
   2000          2000          4000          iC

            INLET VELOCITY  FT/KIN »T 70'F.

Courtesy of Research-Cottrell, Inc.

   FIG.  6—Louver  Dust Collector Pressure Drop.

-------
                  REMOVAL OF PARTICULATE MATTER FROM  GASEOUS  WASTES
     -.
  •  . • •
    -'
Courtesy of American Air Filter Company.
       FIG. 7—Centrifugal Impeller Collector.

180 deg. As  a result,  the particles tend to be
thrown against the  impeller wheel and move
outward by centrifugal force to the outer edge
of the impeller wheel.  At this location the par-
ticles are projected through the narrow opening
between the impeller and the scroll into the sur-
rounding dust chamber. The tips  of the im-
peller  extend  into this chamber, setting  up a
secondary air circulation which carries the par-
ticles into the airtight dust hopper located be-
low the collector. The secondary air is  slowed
down in the hopper, letting the dust particles
fall out, and then is drawn back into the return-
air port to be recirculated through  the dust
chamber.
   Dalla Valle 3 has  said that since  these units
must be designed both for dust collection and
air blowing one cannot expect quite as  high a
blower efficiency from them as from a straight
air blower; but these units are compact, pres-
sure losses are smaller because fewer ducts are
required, they have been found to perform sat-
isfactorily  in capacities up to 20,000 cfm, and
they are widely used.
   The collection efficiency of such units is com-
parable  to that of  medium-efficiency cyclone
dust collectors. For particles smaller than ap-
proximately 10 microns the efficiency falls off
rapidly.  Dalla Valle states that there is theo-
retically no reason why smaller particles should
not be collectible with high efficiency, but prac-
tically it appeai-s  to be impossible.

2.50 Mechanical  Gas Centrifuges
  A second type of motor-operated inertial col-
lector is the mechanical gas centrifuge.  The
first design of this type was reported by Verho-
turov:s in 1935.  He proposed  using a large
rotating drum with short, radial vanes mounted
parallel to the axis  of  the drum.  This drum
rotated  inside a stationary  perforated shell.
The dusty gas flowed up between the drum and
the shell, and as a result of the  rotary  motion
the dust was centrifuged out of the gas  and
thrown  through  the perforations  into catch
pockets mounted outside  the  shell. The pro-
posed unit was very large. The rotating drum
was approximately 11 ft in  diameter  by ap-
proximately 25 ft in height. The drum  rotated
at 290 rpm inside a stationary  shell approxi-
mately 13 ft in diameter. The unit was to have
a capacity of 40,000 cfm of gas  and would
require  a 50-hp-drive motor. In its entirety the
unit would have weighed approximately 70 tons.
In short, it would be a massive machine when
compared with a modern multiple-cyclone in-
stallation of the same  capacity  which would
weigh approximately 6 to 7  tons and  require
approximately 35 hp to overcome the  gas pres-
sure drop.
  More recently, another mechanical gas cen-
trifuge  has been reported u  and is shown in
Fig. 8.  This centrifuge  is relatively small
compared with  Verhoturov's.  This  is so, in
part, because it is  designed for  a capacity
of only 1,000 cfm.  For higher  capacities the
appropriate number  of these  centrifuges is op-
erated in parallel. Suggested systems handling
up  to 250,000 cfm have been  reported.1'
  This  centrifuge  differs significantly  from
Verhoturov's in the following ways. First, be-
cause  the  rotating  drum operates  at much
higher  speeds, 1,500 to 2,500 rpm, the collec-
tion efficiency will be much higher. Second,  it
is not designed to operate continuously.  During
the  dedusting operation all the dust removed
from the gas stream remains in the centrifuge
because there is no means for disposing of the
dust  while  the  centrifuge is operating.  The
centrifuge is then shut down after some suit-
able period of operation  and rapped to make

-------
                              AMERICAN  PETROLEUM INSTITUTE
     DUST LADEN

       GAS
                      OUST
Courtesy of The Fly Ash Arrester Corporation.
           FIG. 8	Modern Centrifuge.

the collected dust  fall  into  a hopper  located
below  the centrifuge.  The maximum time of
operation per cycle depends  upon how  rapidly
dust accumulates in the  centrifuge and the
maximum amount of dust that can be allowed
to accumulate before dust disposal is required.
   Such batch operation  has an important ad-
vantage from the maintenance viewpoint. Once
the particle strikes the inner walls of the centri-
fuge, the particle comes  to rest.  Thus there is
no scrubbing effect of the particle against the
metal  of the equipment  and therefore  no ero-
sion.  As partial, experimental proof of this a
sample of dust was passed repeatedly through
the  centrifugeIS and  no degradation  of the
particles  was  observed,  which indicates  that
the particle-metal contacts are gentle. Further-
more,  in a full-scale experimental model of the
unit it was  observed after  several  months  of
operation with an erosive dust that the mill
scale of the carbon steel  sheets from  which the
unit had been constructed had  not  been worn
away.
   A~second advantage of batch operation is that
since the units are dampered off while they are
being rapped down there is no opportunity for
re-entrainment of the dust during rapping.
  The obvious disadvantage of batch operation
is that  even though the units are capable of
cleaning on the order of 1,000 cfm of gas while
operating,  their real capacity is less than this
because they are shut down part of the time
for cleaning.
  The draft loss is quite low  (i  in. to H in.
of water)  compared with the 2 in. to 4 in. of
water loss  that is normal for cyclone collectors.
Operating  tests have shown that the collection
efficiency decreases with increasing grain load-
ing. The  collection efficiency  decreased  from
99.7 per cent at 10 grains of dust per cubic foot
of gas to 99 per cent at 60 grains of dust per
cubic foot of gas in one series of tests.
  The development of  design  equations for a
centrifuge  collector is relatively simple because
its  principle of operation is simple. Derivation
of the design equation is much the same as for
the gravity collecting chamber except that the
gravitational force is replaced by a centrifugal
force.
   If it is assumed that the gas in the centrifuge
is not turbulent, that the gas is stationary with
respect to  the rotating vanes of the centrifuge
except  for motion parallel to the axis  of the
vanes, that the gas velocity is  independent  of
its position in  the centrifuge, and that a modi-
fied form  of Stokes' law applies, the smallest
particle that can be completely collected by the
centrifuge can be computed from
           D
pra — /!/ '
                      18/iV.
(9)
                    L(pf — p)-   RI

   By  means of this equation it is possible to
 calculate Dpa, for the centrifuge shown in Fig. 8.
 R, and R, are 6.5 in. and 4.0 in.,  respectively,
 and L,  the length  of the zone  of centrifugal
 action, is approximately 29 in. If the collecting
 cylinder is operated at 2,500 rpm, the gas is
 air-delivered at 1,000 cfm at 70 F and atmos-
 pheric pressure,  and the dust  has  a specific
 gravity of 2.0, then Dfm is 4 microns.  If  tur-
 bulence or gas  circulation   patterns exist in
 the gas passing  through the centrifuge, the
 minimum-size particle  completely collected is
 larger.  There have not been  enough data pub-
 lished to investigate this effect.
   The operating time between cleaning periods
 can be computed  readily from the quantity of
 dust  that  can be  retained  on the  collecting
 cylinder between the airfoil  stop rings.  The

-------
                  REMOVAL OF PARTICULATE MATTER FROM GASEOUS WASTES
available  volume  is approximately 0.6 cu ft.
If the dust packs to a concentration of approxi-
mately 60 per cent by volume, the dust load is
10 grains per cubic foot, and the dust density
is 2.0 grams per cubic centimeter, the operating
time will be approximately 30 min.
  The derivation of equation  (9)  is as follows.
The most difficult  particles to collect are  those
which enter the centrifuge near the center tube
because they are  farthest from the collecting
cylinder  shell.   Of  these  particles,  the  most
difficult to collect are those having the smallest
diameter. Consider a particle entering the cen-
trifuge whose trajectory is such that it enters
near  the  center tube  and  finally strikes the
collecting cylinder just before the gas stream
leaves the  cylinder.  This particle  will be of
the smallest size  that  can be  removed  com-
pletely from the gas stream. Any coarser par-
ticles will  also  be removed completely;  finer
particles will be only partially removed.
  Assuming that  gas-eddying has a  negligible
effect on the particle's  motion, the motion of
the particle can be computed from  Newton's

second law of motion, F=— (see  Glossary).

For this problem,  Newton's equation becomes:

                         irDpm3  Pp d-r _ rD,,,n:i
                           6    gc dt      6
For problems of this type it is  usually safe to
assume  that  even though the velocity of the
        dr
particle, -TT, is changing, the inertia term may
        etc
be neglected.  Assuming this and after algebraic
manipulation, the  equation becomes
           dr   Dp,,.1
           dt
                              _ -.
                                        (11)
                                                 2.60  Impingement Separators

                                                 2.61  Body Collectors

                                                   If a dust-laden  gas stream is forced to flow
                                                 around an  obstacle, a  fraction  of the  dust
                                                 particles  will strike the obstacle because  of
                                                 their inertia  and adhere to it, and will thus  be
                                                 removed from the gas  stream.  The most im-
                                                 portant collectors using this principle are scrub-
                                                 bers  and  fibrous  filters,  which are discussed
                                                 by  Gilbert' and Licht,13 and  impingement sep-
                                                 arators.
                                                   Impingement  separators  usually  consist  of
                                                 a large number  of vertical elements suspended
                                                 in the path of the dusty gas stream.  The ele-
                                                 ments are normally flat strips or cylinders, or,
                                                 if they are not, they may usually be treated  as
                                                 though they were for estimating their perform-
                                                 ance. The width of the elements  (perpendicu-
                                                 lar  to the  flow path) is on the order of 1 in.  or
                                                 more.
                                                   Impingement  separators  are  normally de-
                                                 signed for pressure drops on the order of 0.1 in.
                                                 to 1.5  in. of water.'* Their collection efficiency
                                                 drops  off  rapidly  for particles  finer than ap-
                                                 proximately  20  microns.  Therefore, they are
                                                 used only  for removing coarse particles.  Rap-
                                             (PP-
   o,
p)  go
                                                             go
               n. dr
                 dt
(10)
With the initial condition that at t = 0 the par-
ticle's position  r  is  R{,  the  solution  of  this
equation is
  Settling time, t, for the particle most difficult
to collect is equal to the length, L, of the centri-
fuging zone divided by the velocity, v0, of the

gas flowing through it. Replacing t bv — and
                                    "  v<>
solving  for Dfm  yields the size that  is  most
difficult to collect completely, i.e.,
                     18Mv.
                   L(p,, — p)w'
                               •TT-       O)
pers are frequently- used to remove the collected
dust from the impingement elements.   If the
dust is sticky, either because of elevated tem-
peratures or because of moisture condensation,
a circulating water film may be used to clean
the elements.  The  chief  advantage of these
units is their adaptability to  existing flues.
  The capture of dust particles by impingement
collectors can  be treated in  terms  of "target"
efficiencies.  Target efficiency, *;,, is defined as
the fraction of the particles  in the  gas  volume
swept by the collector element that impinge on
the element. This  efficiency is equal to the dust
collection efficiency  if  all the  particles that
strike the element adhere to  the element.  For
very fine particles (less than 1  to  5 microns)
this assumption  is probably  safe because such
particles   adhere  strongly to  solid  surfaces.
Coarse particles  (greater  than  40 to  50  mi-
crons)  will not  adhere  strongly  and will tend
to rebound from  the collecting  surface after
striking  it.  Thus  the collection  efficiency  will
be lower than the target efficiency unless cor-
rective measures are taken.  One method is to
wet  the  collector  surface  with  a  water  film

-------
                             AMERICAN PETROLEUM  INSTITUTE
                                       S£P»B»TIO» NUUBCR  N,

                Courtesy of Langmuir and Blodgett.12

                             FIG. 9—Target Efficiencies for Cylinders.
which prevents particle rebound and also keeps
the surface washed clean. A second method is
the use  of a hollow collecting element with an
open front.  The opportunity for a particle to
rebound out of the collector is reduced by this
arrangement.   Furthermore,  by  continuously
withdrawing a small  amount of gas from the
bottom  of  the hollow collecting  element,  the
dust can be concentrated in a small amount of
gas.
  The work of Langmuir and Blodgett'-  may
be used  for estimating the performance of  such
collectors. They found that the target efficiency
is  a function of two dimensionless groups and
that this function can be computed for a few-
simple geometric  shapes.  For the flow around
an  obstacle at a  high Reynolds number,  they
assumed the  gas  could  be treated as a  non-
viscous  fluid in order to  compute its flow  lines
around  the  obstacle.  This is only an approxi-
mation  but  it is a reasonable one. The motion
of the particles carried by the gas stream was
then computed from  the gas flow lines, from
Newton's second  law of  motion, and from the
correlation of the drag force acting on a sphere
moving relative to a gas stream. Langmuir's
and Blodgett's results are shown in  Fig.  9
and 10  for cylinders and flat ribbons.
   As an example of the use of  their results,
consider a dusty gas flowing around a flat  strip
2  in. in width. Suppose  that the gas is air at
60 F, 1 atm  pressure, and is  flowing  with a
velocity of  40 fps. Assume that the dust has
Courtesy of Langmuir and Elodgett.i:
    FIG. 10—Target Efficiencies  for Flat Strips.

-------
                  REMOVAL OF  PARTICULAR MATTER FROM GASEOUS WASTES
a density of 130 Ib per cu ft and that the col-
lection efficiency is equal to the target efficiency.
The  size particle that can be collected  with a
target efficiency of 50 per cent is calculated as
follows.
  First calculate $ as shown:
            WP
       ~ [ (0.0177) (0.672) (10-3) ] (130)
       = 445

  For 9 = 445 and 7/t = 0.50 the separation num-
ber,  Nt,  from Fig.  10 is 0.88; solving for  D,
from
N,=ppDp'V"
                                        (14)
  J/(IS) [(0.0177) (0.672) (10-')] (^(0.88)

    V                (130)(40)
  = V (0.604) (10-3)
  = (0.78) (10-) ft
  = 24 microns
  From this calculation it is apparent that in
order to achieve high collection efficiencies of
fine  particles it  is  necessary  to  use small-
diameter obstacles. The equation indicates that
increasing the gas velocity should also increase
the efficiency. Unfortunately, re-entrainment of
the deposited dust also increases, making it
impractical  to use gas  velocities greater  than
100 to 150 fps.


2.62  Impingement Separators Using  Jets

  In an interesting modification of the impinge-
ment  separator, the gas stream  is formed into
a jet  before it impinges  upon a flat collector
plate located from  one to three jet widths away
from  the jet outlet (see Fig. 11).  If small jet
widths (3^ in. to £ in.) are used, high collection
efficiencies of  fine  particles can  be  achieved.
  The jet can be formed by either a circular or
a slit-shaped orifice.  The experimental meas-
urements  of Ranz and Wong " (see Fig.  12)
indicate that circular orifices produce a higher
target efficiency than do slit-shaped orifices. The
difference is not  great enough to be the only
                                                            ORIFICE
                                                             PLATE
                                                                         TARGET
                                                                         PLATE
                                                 AIR
                                                  IN
                                                                                      AIR
                                                                                      OUT
                                                                      SECTION  A
                                                                PARTICLE  TRAJECTORY
                                                                           FLUID
                                                                           STREAM
                                                                           LINES
                                                            SECTION  A

                                                       FIG. 11—Jet Impingement Separator.

                                                 criterion for choosing an optimum impingement
                                                 separator design, however. Other factors such
                                                 as pressure drop, simplicity  of  construction,
                                                 and re-entrainment  of  the  collected  particles
                                                 also must  be  considered  before  choosing the
                                                 final design.
                                                   As with  the  body collectors, re-entrainment
                                                 is an important consideration.  Dry, solid par-
                                                 ticles are more likely to rebound  from the col-
                                                 lecting surface than to adhere to  it u unless
                                                 they are very small (under 1 micron in size)
                                                 or unless the collecting surface is coated with

-------
                  REMOVAL  OF PARTICULATE MATTER  FROM  GASEOUS  WASTES
                                          11
a film of flowing water to trap and carry away
the collected particles.


     3.00  SONIC AGGLOMERATION

  In 1860, Kundt discovered that a sound wave
passed through  an aerosol causes the  aerosol
particles to agglomerate.  Practical application
of this discovery has been made to agglomerate
fine  particles  that are difficult  to  collect  by
conventional  apparatus such  as  cyclone sepa-
rators.- 4':i
  The  mechanism  of sonic  agglomeration  is
complex  and  includes the following  factors:

1. Fine particles are  acted upon by Brownian
movement forces which, when the particles are
in high  concentration,  cause them  to collide
frequently with one another.
2. Because of  cohesive  forces,  fine particles
tend to adhere to one another after collision.
3. Radiation pressure of a sound wave causes
the  particles to accumulate at the antinodes of
the  wave. The increased particle  concentration
at the antinodes increases the frequency of par-
ticle collisions.
4. Viscous forces tend to couple  the motion of
the  particles to the oscillatory  motion of the
gas.  Because of the particle inertia,  the ampli-
tude of  oscillation of large  particles  is  less
than that of small particles.  The difference in
velocity of the particles increases  the frequency
of their collisions.
5. Bernoullian forces  cause particles  whose line
of centers is at right  angles to the direction of
propagation of the sound wave to attract one
another.  This, too, increases  the  frequency of
collisions.

  Sonic agglomeration is only effective for aero-
sol concentrations greater than  approximately
1 grain per cubic foot. For smaller  concentra-
tions  it  is necessary to increase the particle
concentration by, for  example, the injection of
steam.  The useful sound frequency range  is
from approximately 1 kc to 10 kc. The sound
intensity  required is  150 db or greater.  This
sound intensity causes serious  problems because
of its very high level.  For people who are close
to the  apparatus,  the sound level is above the
threshold of pain. For those  at a distance the
high-pitched  sound is a nuisance.  The high-
intensity sound  can also have a  serious effect
on equipment—the vibration being able to pro-
duce fatigue failures of the equipment.
  Several types of sound  generators are avail-
able but currently only the  gas  siren genera-
tor l° is  capable of producing high-intensity
sound at efficiencies which make sonic agglom-
eration economically practical.
  There have  been installed, in recent years, a
small number of sonic agglomerators operated
in series with cyclone collectors to recover ma-
terials such as carbon black,  sulfuric acid mist.
and soda ash fume.  The  number of industrial
applications is still  small because the use of
sonic agglomeration has  been recent; and a
considerable amount of research and  develop-
ment remains to be done to satisfactorily estab-
lish suitable applications, optimum designs, and
adequate design methods.

    4.00  THERMAL  PRECIPITATION

  In 1870, Tyndall observed that if a dust-laden
gas is brought into contact with  a heated sur-
face, a  dust-free space develops  between the
surface and the cloud of dust particles. A theo-
retical analysis by Epstein *  and experimental
investigations by Rosenblatt  and La Mer -° and
Saxton and Ranz -- have  shown that  this phe-
nomenon is a result of a force  on the particles
caused by  a creeping motion of  the gas  from
the cooler regions  to the heated regions of the
particles' surfaces.  This force  causes  the par-
ticles to be repelled by heated  surfaces and to
be attracted by cooled  surfaces.
  Epstein's theory shows that  the force is di-
rectly proportional to the  gas temperature gra-
dient and to the particle diameter.  Since the
motion of  the particles is resisted by viscous
drag forces which are also directly proportional
to the particle diameter, the  velocity of a par-
ticle moving through a gas temperature gradi-
ent should  be independent of particle diameter.
  Thus,  a device  consisting of  two  parallel
plates—one heated,  the  other  cooled—should
act as a dust precipitator which is just as effi-
cient for fine particles as for coarse particles.
Such devices, called thermal precipitators, are
used as instruments for atmospheric dust sam-
pling.  Thermal precipitation has had almost
no industrial applications because of the high
cost of maintaining  the  required temperature
gradient and because of its large space require-
ments.

-------
                                CYCLONE DUST COLLECTORS
                                                                         I  (-OUTLET LENCTH
                                                                         '  I (HEIGHT or ANNULUS )
                  CYCLONE   DUST  COLLECTORS

         (REPORT  ON  REMOVAL  OF  PARTICULATE MATTER  FROM

                                 GASEOUS  WASTES)

          1.00  INTRODUCTION
 1.10  Definition
   A cyclone is a structure without moving parts
 which separates particulate matter from gas by
 transforming the velocity of an inlet gas stream
 into a vortex confined within  the structure.  A
 vortex is a helical form of rotation of fluid about
 an axis.  In  nature, the axis of a vortex is fre-
 quently curved, and  although the axis of the
 confining structure  might likewise be  curved,
 it is in  practice a straight line.  Similarly, al-
 though  the  ideal structure would  be one gen-
 erated by  a curved line revolved about the
 cyclone  axis, practical cyclones employ confin-
 ing structures generated by the revolution of
 straight lines about  the axis, i.e., combinations
 of cylinders and cones (Fig. 1).

 1.11   Types  of Cyclones
   Structurally, a  cyclone must have an axial
 gas outlet, a dust discharge, and a means for
 gas inlet which will  produce  the gas rotation
 necessary to create the vortex. These three ele-
 ments may  be combined  in a number of differ-
 ent ways. Rotation  may be produced  by tan-
 gential gas entrance  or  by axial gas entrance
 through a set of swirl vanes. Separated dust
 may be  removed  either axially or  tangentially
 from  the periphery.  Dust  may   be removed
 from  either  the end opposite  to  the  axial gas
 outlet or from the same end. There may  be
 either one or a multiplicity of tangential inlets.
 The cyclone body may be  completely cylindrical,
 completely conical, or made of both cylinders
 and cones. The gas outlet may be either cylin-
 drical or conical.
  Cyclones in common use may be  classified as
follows:
1.  Cyclones with tangential inlet and axial dust
discharge.
    la.  Larger-diameter  ("conventional") cy-
    clones.
    \b.  Small-diameter ("high efficiency") cy-
    clones.
                                                                         OUST MOPPES
                                                             Typical Cyclone.
                                                                FIG.  1

                                              2.  Cyclones with tangential inlet and peripheral
                                              dust discharge.
                                              3.  Cyclones with axial inlet and axial dust dis-
                                              charge.
                                              4.  Cyclones with axial inlet and peripheral dust
                                              discharge.

                                                Type la is the most commonly used  type in
                                              which the total gas volume is handled through
                                              one cyclone body, although sometimes such cy-
                                              clones are arranged in parallel (Fig. 1). The
                                              other types are usually of small  capacity indi-
                                              vidually, and practical gas volumes are obtained
                                              by operating large numbers of individual units
                                              in  parallel.
SOURCE:   Cyclone Dust  Collectors, Engineering  Report Prepared  for American
          Petroleum Institute,  Washington, DC  (1956).

-------
                              AMERICAN PETROLEUM  INSTITUTE
1.20  Mechanism of Cyclone Operation

  The cyclone separates particles from the gas
by means of centrifugal force exerted on the
particles in vortex flow, tending to drive the
particles to the  wall of the cyclone body.  The
particles at the wall move toward the  dust dis-
charge by virtue of the axial component of the
vortex flow, aided, if the  axis is vertical, by
gravity.  The magnitude  of the radial  forces
acting on the particles depends on the nature
of the vortex flow in  the different sections of
the cyclone.  Counteracting forces, such as ra-
dial gas velocities, tend to offset the separating
forces.

1.21   Nature of Vortex Flow

   The inlet of a cyclone transforms linear flow
into vortex flow. After the gas leaves the inlet
annulus, it establishes a  vortex with  an axial
component in the direction of the dust discharge.

   1.211   FLOW  IN  THE  MAIN VORTEX:  The
characteristic of the main vortex is that the tan-
gential velocity at radius r (V,) increases as r
decreases from the radius of the confining struc-
ture 7-p to a maximum V,,n at some intermediate
radius rm.  As  r further decreases from rm to
zero at the axis, V, decreases from its maximum
 Vtm at rm to zero at the axis. The zone between
 the axis and rm is called the core of the vortex.
 In the outer zone of the  vortex  where  V, in-
 creases as r decreases, the  flow obeys  the equa-
 tion:
                                          (1)
 Where:
    Vt = tangential  velocity at  radius r, in feet
          per second.
   Vtp=tangential  velocity at rf, in feet per sec-
          ond.
     r,,= radius confining structure, in feet.
     r = radius, in feet.
     n = exponent, dimensionless.

   If —-r,, the ratio of the radius of rotation
      rp
 to the radius of the confining structure, equa-
 tion (1) reduces to:

                    V« = |a                (2)

   In an  ideal gas, n would equal 1. Real values
 of n are less than 1, and at ordinary  tempera-
 tures are greater than 0.5.
  Alexander-1 presents an equation evaluating
n for air at 10 C, as follows:

                   (24r.,)l>-M
                     2.5
                                         (3)
  For variations  in gas temperature, he  found
n to vary as:

              ^W
Where:
   n-r = value of n at T Kelvin.
    n0= value of n at 283 K  (10 C).
   Tk = temperature, in degrees Kelvin.
  Values of n computed from equations (3) and
 (4) are presented in  Fig. 2. Individual  values
of n reported by other investigators are:
    Shepherd and Lapple i::	 0.5
    Stairmand-'-" 	 0.5
    Ter Linden-1  	 0.52
    Prockat37  	 0.7
    First"	 0.88
   There is  radial gas flow  in the cuter  vortex
 zone. There has  been much theory propounded
 on  this point but little experimental measure-
 ment.  The  simplest theory, that of Feifel" for
 a simple  cyclone with  no  gas-outlet extension
 into the body, assumes i;sink" flow radially in-
 ward from  inlet  to outlet superimposed on tan-
 gential vortex flow.   Unfortunately, however.
 the use of an outlet-duct extension into  the cy-
 clone  body, and the presence of space  below
 and above  the inlet  into which  gas can  flow,
 causes radial velocity to vary in a manner pres-
 ently defying mathematical statement.
   Radial velocity has been measured in a con-
 ventional cyclone with air and in a hydraulic
 cyclone  with water.  In the hydraulic, cyclone
 the customary  underflow  (downflow through
 bottom of cone) was  suppressed.  Typical radial
 velocity profiles based on  these  findings show-
 that from rf to some intermediate radius there
 is  inward  radial velocity,  but from the inter-
 mediate  radius  to the axis there  is outward
 radial velocity (Fig.  4).
    1.212  FLOW  IN VORTEX CORE: Flow in the
 vortex core follows the equation :
                   V, = Kr"               (5)
  Where:
    K = proportionality  constant  varying with
          VP. rm. and n.
     q = velocity exponent.

   • Figures refer to REFERENCES on p. 66.

-------
                                   CYCLONE DUST COLLECTORS
                        3    456   8   10
    20     30  40  50 6O  80  100
                                                                                           200
                                 Values of Kj  [See Equation (17)].

                                             FIG. 6
 and its helical pitch about 10.5 ra.  In extremely
 long cyclones, the eccentricity has been observed
 up to 50 r0 before disappearing.
   As will be shown later  (Sect.  2.31), cyclones
, of common design have a "natural length" on
 the order  of 10 to 15 r0.
   At the  vortex base there is a region of con-
 siderable turbulence and reversal in direction of
 axial flow, \vith a likelihood that the core eccen-
 tricity will bring this region into close  prox-
 imity with, or in actual contact with,.already-
 separated dust at the cyclone wall. It is for this
 reason that much attention should be devoted
 to the design of dust discharge,  purge, and the
 like.

 1.22  Separation of Particles

   Particulate matter is separated from gas in  a
 cyclone  by  centrifugal force, or  radial force,
 tending to drive the particles (against the re-
 sistance of motion by the gas)  to the cyclone
 wall.  The radial force imparted  to the parti-
 cle is:
                       mnVp-
                                           (6)
 Where:
    F, = radial separating force, in pounds.
    mp= particle mass, in pounds.
    V,, = tangential velocity of particle, in feet
          per second.
    g = gravitational constant = 32.2 ft per sec
         per sec.
  The  tangential velocity of the particle is theo-
retically lower than that of the gas in which it
is suspended; but, in considering fine particles,
the two velocities are equal for all practical pur-
poses and VP = V..  By combining equations (2)
and  (6) :
which has a maximum value:

                ri     mnVr.n2
                                          (S)
Where:
   rfm= radius ratio at ?•„,, dimensionless.
  V,m = maximum tangential velocity (at »•„,), in
         feet per second.
  The  value ra? in the preceding equations is
proportional to the cube of the particle diame-
ter, thus:
                 mpZ^Dp1               O)
Where:
    P = volume shape factor, dimensionless.
   Pf = particle  density, in  pounds per cubic
         foot.
   Dp = particle diameter, in feet.
The  separating force [equation (7)] becomes:
                      grr,-
                                         (10)

-------
                              AMERICAN  PETROLEUM INSTITUTE
100
                                   rp/r0  (STAIRMANDK9
                                                     H/r0 (TERLINDEN  AND  BROERH5)
                                        45        6       7
                                   NUMBER OF OUTLET  RADII (To)

                                 Dimension Ratios and Efficiency.
                                            FIG. 7
  It can be shown 2" that the force  ("Stoke's
Law  force")  resisting  motion of  a particle
through a gas in the range of particle diameters
from 3 to 100 microns is:
                                        (11)
Where:
   Fr = frictional resistance to flow, in pounds.
   K: = proportionality constant, dimensionless.
    u = particle velocity with respect to gas, in
         feet per second.
    p. — gas viscosity, in pounds per second  per
         foot.
  Comparison of equations (10) and (11) shows
that the  separating  force varies with the cube
of the particle diameter, whereas the resistance
to flow varies only linearly.  Thus large-diame-
ter particles have a greater ratio of separating
force to opposing force and are, therefore, more
readily separated in a cyclone.
  The velocity u of the particle with respect to
the gas includes the radial as well as tangential
components of gas flow (Sect. 1.211).  The as-
sumption of Feifel allows the computation of a
"critical particle size" (Der)  for any radius of
rotation. Drr may be considered  the largest size
of particle  which is not separated from the gas
stream. On  the basis of this  simplifying as-
sumption, Drr could be computed for r0, and all
particles larger than Dc, would  be collected
while all smaller particles would be lost into the
outlet duct.

-------
                                  CYCLONE DUST COLLECTORS
V .•
X
/


L..
i
t
i
UJ


            Axial
   Helical

Cyclone Inlets.
   FIG. 8
Involute
  In fact, however, the radius at which a parti-
cle may be considered lost to the  exit gases is
not constant because the radial gas velocities
vary along the axis of the cyclone.  Since  the
radial velocity pattern has not yet been mathe-
matically  stated,  there can be no reliable com-
putation of critical particle size.  In actual prac-
tice, the presence of the gas-outlet extension,
the finite width of the entering gas stream,  the
random distribution of particles of various sizes
across this width, the presence of major eddy
currents,  etc., all further obviate  the value of
any such theoretical computation.
  Even were such a computation possible, it
would  be  found that not one, but a series of
critical sizes  would be obtained for various in-
crements  of cyclone length and from one side
to the other of the inlet.  Therefore,  it follows
        that a cyclone must be, and is, a poor classifier
        and does not make a  sharp-size cut between
        particles separated  and rejected. The cyclone
        has a curve  rather  than a straight line  as  its
        particle  size-efficiency characteristic.  Qualita-
        tively, the  predominant requirements for high
        collection efficiency  of  fine particles  are that
        r,m and rm be low and Vtm high.
           A number of authors have derived equations
        for Dcr, as shown in Table I. No critical evalua-
        tion has been made which would show one equa-
        tion more  nearly correct than  another.  The
        main value is to demonstrate the parameters in-
        volved and their qualitative relationships.

        1.23 Discharge of Separated Dust
           The separation of the dust from the gas
        stream is, in practical terms, only the  first step

-------
10
                              AMERICAN PETROLEUM INSTITUTE
       Reference
                                          TABLE I

                              Formulae for Critical Particle Size

                                     Formiil;i
Gardiner *f
Lapple"
                              D  =.'_^.-«
                               "•'  i12;Nv;(~p,,-p)~
                                 -J^^
Shepherd -
Muhlrad »•

                           D,, = ;  -         T--'
         Special L'nits: Remarks
a,Uin = particle size  above which effi-
       ciency will be 100  per cent
       in  absence  of  re-entrain-
       ment.
  P = viscosity, in  pounds  per sec-
       ond per foot.

  /« = viscosity, in  pounds  per sec-
       ond per foot.
 Drp=cut size, collected at 50-per-
       cent efficiency.
 N, = "effective" number of turns in
       cyclone = 5 to. 10 for typical
       cyclone  (Sect. 2.31).
  n= viscosity, in pounds per second
       per foot.
       For N, see reference.
     do
  w = -
Rosin, Rammler, and
   Intelmann::< 	
Tar j an-"
                               d   -J  9&A'r-
                               UIMIII — \~ti'V~ir"
                                            --
        All units metric—see refer-
          ence.

  Y= — -r = instantaneous distance

        of particle  from  cyclone
        wall. All units metric—see
        reference.
d,,,i,, = particle diameter at "blocking
        radius r"
   r = "blocking radius."
 in a series intended to result in solids or liquid
 at rest in a container, and a gas stream cleaned
 of particulates to the desired degree. The con-
 centrated dust layer swirling slowly (compared
 to the vortex)  down the walls of the cyclone
 body must be conducted into the dust hopper, or
 its equivalent, with a minimum of re-entrain-
 ment into the base of the vortex. Factors affect-
 ing re-entrainment are:
 a. The length, eccentricity, and diameter of the
 vortex core and its relation to the cyclone body
 walls.
 b. The presence or absence of purge flow out-
 ward  from the dust outlet.
 c. Recirculation of gas,  or infiltration of air,
 into the  dust outlet.
 d. Presence  or absence of an air-lock material
 discharge valve or dip-leg.
                                                  e. Smoothness of the inner walls of the cyclone.
                                                    The two major types of discharge are axially
                                                  through a core and peripherally. Both methods
                                                  may be used with or without purge gas, or with
                                                  various types of mechanical accessories. These
                                                  methods, and their effect on efficiency,  will be
                                                  discussed in later sections.
                                                    The dust in the gas stream which escapes into
                                                  the outlet duct tends to be concentrated  toward
                                                  the walls of the duct. Therefore, various de-
                                                  vices for skimming  off the outer layers of this
                                                  vortex have an effect also on the  overall per-
                                                  formance of a cyclone.

                                                  1.30  Scope of the Cyclone in Field Use
                                                  1.31  Typical Operating Ranges
                                                    Cyclone collectors, as a class, are regarded as
                                                  the lowest in first cost of equipment convention-

-------
CYCLONE DUST COLLECTORS
                                                                                           11
ally used for dust or droplet collection. They
also, as a  class,  provide  lower collection  effi-
ciency than other types of equipment, although
there are, of course, exceptions to this general-
ity. They are suitable for medium and coarse
dusts  and are unsuited for very fine dusts or
metallurgical fumes.   They  have advantages,
however, in that there are  no moving  parts,
choice of materials of construction is wide, and
maintenance costs are low.  Power requirements
(if reasonable efficiency  on  fine dust  is  re-
quired) are as high as. or higher than, most
other  types of equipment, although some types
of wet scrubbers have even  higher  require-
ments.

1.32  Efficiency in Field Use
   A clarification  in terminology regarding effi-
ciency will aid in this and later general discus-
sions.  Efficiency brackets are arbitrarily set as
follows:
                                .Miinimr Passing
  Cnllrcrinn Efficiency     Ktliri'-npy        follflctor
  il'erOnt by Welcht)       R:iir-'i-         iI>rOnc)
Above 50 to 80	Low       Above 20 to 50
Above 80 to 95 ... .Medium    Above  5 to 20
Above 95 to 99	High       Above  1 to 5
Above 99 to 99.99.. Very high  Above 0.01 to 1
Above 99.99	Ultra high Less than 0.01

   On this basis, most cyclones as applied oper-
ate in the low and medium efficiency ranges. A
well-designed  but  otherwise conventional  cy-
clone may be expected to provide high efficiency
for coarse particulate matter of  the 40- to 50-
micron diameter range:  and  small-diameter
 (less  than  1-ft)  "high efficiency" types  may
extend the high-efficiency range to particles as
small as  15- to  20-micron  sizes.  Typical effi-
ciency ranges for various particle size  ranges
are as follows:
    r.-irllrlP Slxi:     •Tnnvpntlon.il"    "IIIcli EfiVlTicjr"
      lC:iifji>         f'yi'lon*» F.rfi-       I'yrlonc Effl-
     i Mil-run.1        rll-iicy Runt"       Pii-ncy Range
Less than  5	  —             Low
  5 to 20	  Low           Medium
 15 to 50	  Medium        High
Greater  than  40..  High           High

   For various practical reasons, efficiencies in
the very  high or ultra-high  range are  almost
 never achieved on  industrial aerosols  by cy-
clones.

 1.33  Pressure Drop in Field Use
   The loss in pressure of the gas stream flowing
 through a cyclone depends on a number of vari-
               ables but, in practice, is usually found to be on
               the order of 1 to 4 inlet velocity heads.  With
               velocities commonly  used for handling  dusty
               air, the velocity head will range from i to 3 in.,
               water.gage. Thus the range of resistance would
               be from J to 8 in., water gage. Where ordinary
               centrifugal  blowers are used as the source of
               motive power for the air, a practical upper limit
               of resistance for the cyclone plus other system
               components is in the range of 15 to 16 in., water
               gage,  although  with other  types of motive
               equipment this upper limit may be considerably
               higher. Efficiency increases with inlet velocity,
               as will be shown later, but at a lower rate than
               the pressure drop increases.  Furthermore, for
               any given cyclone and dust combination, there
               is an  inlet  velocity  beyond which turbulence
               increases more  rapidly than  separation,  and
               efficiency begins to decrease.  Thus, even aside
               from power cost considerations,  there is little
               purpose in further increases in resistance. Most
               cyclones will be found to have from 2- to 7-in.,
               water gage, resistance as operated.

               1.34  Dust Loading in Field  Use
                  There appears to be almost no practical upper
               limit of dust loadings to a  properly designed
               cyclone. A  cyclone can be designed to handle
               any amount of material whi'ch can be moved by-
               gas flow and, in general, cyclone efficiency in-
               creases  with  increasing dust load. Since  this
               ability is not possessed by many other types of
               collectors of  inherently  higher efficiency, cy-
               clones  are  frequently used as precleaners for
               other types when dust loadings arc too  heavy
               for the secondary collector.
                  Conversely, cyclone efficiency decreases with
               decreasing  dust load, and  other types of  col-
               lectors are  well able to handle the lighter loads
               alone at higher efficiencies.  As a result,  cy-
               clones  are  usually applied  to dust loadings of
                10 grains per cu ft or higher,  and seldom are
                used below  1 grain per cu ft, unless the diiat is
                unusually coarse or  other factors of tempera-
                ture, corrosion,  and the  like are overriding in
                importance.

                1.35   Ran*e of  Capacities
                  Commercial cyclones are built in sizes having
                individual capacities of 30 to 30,000 cu  ft per
                min.  In general,  the  smaller  units  are  more
                efficient but must be used in parallel  to accom-
                modate practical gas volumes.

-------
 12
                               AMERICAN PETROLEUM  INSTITUTE
 1.36  Range of Physical  and Chemical Condi'
         tions
   Cyclones, as a class, can be designed to handle
 a wider range of chemical  and physical condi-
 tions of operation than any other type of collec-
 tion equipment. Any conditions of temperature,
 pressure, and corrosion  for  which  structural
 materials are available can  be met by a cyclone.'
   Erosion is sometimes troublesome, and some
 materials of  construction  necessary  for  tem-
 perature  or corrosion conditions  may not be
 especially desirable in erosion resistance.
   The  main  limiting factor in conditions of
 cyclone application is the physical characteris-
 tic: of the dust. A sticky, hygroscopic, or similar
 material may  give difficulty in cyclone opera-
 tion, and such materials are sometimes handled
 in cyclones  operated wet.  or in "wet collectors."
 Dusts which  have a  bad tendency to bridge,
 hang up, or cake may give difficulty at the dust
 outlet of small-diameter cyclones.

   2.00  EFFICIENCY Ars7D  PRESSURE
                   DROP
 2.10  Genera]  Considerations  of  Efficiency
   For  high efficiency, the separating forces
 should be high, the number of  gas revolutions
 large, the tangential velocity high, and the dust
 removal effective so that  separated dust is not
 re-entrained.
   Considerations governing the critical particle
 size D„ have been presented previously.  Since
 the particle size-efficiency performance of a cy-
 clone is a curve rather  than  a straight  line,
 there is no direct mathematical relation between
 efficiency i\  and Der.   However, in a qualitative
 sense, those parameters tending to increase Dcr
 should decrease 77 for  a given dust size distribu-
 tion, and vice versa.
   By inspection of Table  I, it may be seen that
 efficiency will increase with an increase in the
 following:
         Dust-particle density pf.
         Inlet velocity Vf.
         Cyclone-body length H.
         Number of gas revolutions N:

         Ratio 2--
               do
         Dust-particle size Dt.
  Similarly, efficiency will decrease with an in-
crease in the following:
         Gas viscosity p.
         Cyclone diameter D.
         Gas-outlet-duct diameter d0.
         Increase in inlet width W{.
         Inlet area A{.
         Gas density p.

   Efficiency is also  affected by various acces-
sory devices not included in the mathematical
relationships in Table I.  It is obviously desir-
able to have smooth inner surfaces in the cy-
clone body in order to  minimize projection of
separated dust out into the air stream.

2.20 General Considerations of Pressure Drop
   For  a given cyclone handling clean air,  the
resistance varies with the square of the air vol-
ume Q, or similarly with the square of the inlet
velocity V{. Since the expression for velocity
pressure ht of a gas is :
  = 0.0030^,
                                        (12)
hv similarly varies with the square of the veloc-
ity.  It therefore becomes convenient to express
cyclone pressure drop in terms of number of in-
let velocity heads, hn, thereby making such equa-
tions independent of gas volume and inlet  ve-
locity.
   It will be apparent from later discussion that
a  satisfactory method for  predicting pressure
drop from cyclone dimensions over a wide range
of construction has  not yet. been  developed.
Pressure drop should preferably be determined
experimentally on a geometrically similar pro-
totype.
   Most investigators agree that cyclone resist-
ance is a function of gas-inlet area  and outlet
area of the form :
nn=:
                                        (13)
Where:
    X = proportionality constant, dimensionless.
   H! = inlet height, in feet.
   W, = inlet width, in feet.
    d0 = outlet diameter, in feet.
    z = exponent, dimensionless.

  Values for X are given as follows:
a.  Ter Linden6" states that for common cyclones
with straight tangential  inlets without inlet
vanes 2 = 1 and X is between 25 and 32, but that
for involute inlets both 2 and X are lower.  For
a 180-deg involute inlet his data show .Y=21.7
if z = l (Fig. 10).
b.  Lapple"  gives  a value for A" of 18.4 for a
360-deg involute inlet.

-------
                       CYCLONE DUST COLLECTORS
                                                         13
                                                              ENTRY  WIDTH
                                                              ENTRY  HEIGHT
0.6 0.7 0.80.9 1.0
            2          3
ENTRY  DIMENSIONS-INCHES
 Entry Dimension Vs. Efficiency.
           FIG. 9
                                                                      7   8  9  10
c.  For a cyclone having dimension ratios -~•=
                                        TO

2 to 4, —= 8, — =6, Shepherd  and Lapple"-*'

reported°2 = l and X = IQ  without inlet vane,
and  7.5  for an inlet vane which  neither -ex-
panded nor  contracted the entering gas and did
not touch the outlet (a  "neutral" vane). They

also  found that variations  in — from 6 to zero

had  no effect on A".

d.  First" found that a  reduction in — from 2
                                    0
to  zero  caused  a  10-per-cent  decrease in X

 ^2=2  and— = 5.8 J .  In cyclones with -^ = 2
 T,         T,      I                   TO
     1              v/
and  — =2 but with — variable,  First concluded

that 2 = 1, and derived an expression for X, as
follows:
                       12

               •A. ~ f r \ i f 1.r\ I            \-l^/
                                       Where:
                                          Y = 0.5 for no inlet vane;
                                           = 1.0 for neutral inlet vane;
                                           = 2.0 for inlet vane that expands entering
                                                gas stream and touches outlet duct.
                                          L = cyclone-cylinder length, in feet.
                                          M = cyclone-cone length,' in feet.
                                          D = cyclone-body diameter, in feet.
                                         Alexander5 concluded that z = I, and derived
                                       an expression for X,  as follows :
                wjhere / varies with n as follows :
                       .. 0      0.2   - 0.4    0.6
                  n
                  f ..... 1.90   1.94   2.04   2.21
                                                                             0.8
                                                                             2.40
                                         Values of Alexander's  function  for  X  are
                                       plotted in Fig. 5.  For values of n, see Fig. 2.
                                         Stairmand has  derived a theoretical expres-
                                       sion for hn, as follows:
                                                                /A  V •*
                                                                              (16)

-------
                                         PETROLEUM INSTITUTE
                             o- NON EXPANDING
                             b - EXPANDING
                             c -THROTTLING
                 Inlet Vanes.

                   FIG.  10
where:
   See Table II for computed values of -^-, and

Fig. 6 for plots of equation  (17).
  The effect on A" of variations in -^ and — was
                                TO     i o
investigated by Ter  Linden  and  reported by
Broer* as the following multiplying factors for
correcting X:
     Tf
     r~
     1.5.
     2.0.
     2.5.
     3.0.
Factor
0.91
1.00
1.08
1.19
 H_
 To
 4.
 6.
 8.
10.
Fnctor
1.17
1.08
1.00
0.96
  A test of the results of these various equa-
tions has shown those based on equation  (13)
to have a higher order of reliability than those
of equation (16) ; and, further,  indicates that
the exponent  2 in equation (16)  is  probably
too high.
  A summary of factors for application to equa-
tion (13) is presented in Table III.
  An  attempt has been  made to analyze the
                                                                  TABLE  II
                                                                      •y
                                                  Calculated Values of -=¥• from Equation (20)
                                                                       ' i
\Kot.o 25
KatlcNv
Tc ^V
1 	
2 	
5. .
10.. .
20 	
0.10
1.21
1.53
1Q *N
2.24
2.47
0.20
1.08
1.31
1.58
1.75
1.88
0.00
Calculi;
0.87
1.00
1.13
1.21
1.27
1.0
itecl-^
0.71
0.78
0.85
0.89
0.92
2.0
0.55
0.59
0.63
0.65
0.67
3.0
038
040
04?
0.43
0.43
validity of equations (14) and (15),  and the
values cited  by Shepherd and Lapple,"'"  by
comparing calculated values  of X with  meas-
ured values of A".  Three sets of data were avail-
able and are presented  in Table IV. Group A
was reported  by Stairmand,<:> Group  B  by
Stern,'1 and  Group  C represents  field survey
data previously unpublished.  Since the three
groups of data probably involve differences in
the point and method of measuring static pres-
sure, and it is known that in Group C it was not
possible to determine the presence or absence of
internal accessories, the three groups were ana-
lyzed separately.
  Calculated X shows a large deviation from
measured X  in all three groups of data. The
constant 16 proposed by Lapple gives less devia-
tion for Groups A and  C, while the Alexander
equation gives lowest deviation for Group B.
The data were further subjected to analysis of
variance ratio,:u with the result that, for  Group
A, all three methods of computation were found
equally reliable; for Groups B and C, the Alex-
ander method and the Lapple constant were
equally reliable, while the First method  was less
reliable.  For Groups A, B, and C as a whole, the
Alexander and Lapple values were equally re-
liable, while the First method was less  reliable.
The results of this analysis may be summarized
as follows:
  Data
  Group          Rest Method
   A	 Lapple
   B	 Alexander
   C	 Lapple
A, B, C	 Lapple
   Equally Good
     Methods
First, Alexander
Lapple
Alexander
Alexander
                                    This analysis shows a constant value for A' to
                                  be equally as good as, if not better than, equa-
                                  tions for computing A' based on cyclone-dimen-
                                  sion ratios. It seems only reasonable to assume,
                                  however, that dimension ratios  will affect X;
                                  in fact, this  is proposed by the same authors
                                  who advocate a "constant" value (Table  III).

-------
Source •£ -||
[ 2 8
Ter Linden"". . J 9 H
I
Lapple1" 	 1.G2 to 3.25 —
Shepherd and r 2 to 4 8
Lapple «•" ..J
1 2 to 4 8
Broer" 	 1.5 to 3.0 4 to 10
First" 	 2 Variable

Alexander2.... — —
Lapple" 	 2 8

TABLE III
Pressure-Drop Factors for Equation (13)
-^ Special Conditions x X
— No inlet vane 1 25 to 32
— 180-Deg involute 1 21.7
inlet (Assumed)
1 360-Deg involute — 18.4
(Approxi- inlet, step-
mately) tapered body
G No inlet vane 1 16
6 Neutral inlet vane 1 7.5
— No inlet vane — See text for X correction factors
2 Y = 0.5 no inlet 1 12
vane; v Y
= 1.0 neutral in- /L\J/M\1
let vane; ( ~|)J ( U J
= 2.0 expanding
inlet vane
(Fig. 5)
1.25 No inlet vane — 8

0
n
s
z
M
O
c
l/l
-i
O
P
r«l
3
o
C/l

K«

-------

















Analysis

TABLE IV
of Calculation Methods for


X



Calculated Valuea of X
•4
« v —
III
£'a~-

2.00
2.00
5.00
5.00
3.00
2.8'l
O.G7
O.G7
O.G7
O.G7


0.98
1.00
O.G7
1.33
1.58
0.75
1.10

•o
III
5Q "

1.00
1.00
2.50
2.50
1.50
1.00
0.21
0.33
0.21
0.33


0..'J8
0.50
o.:;8
O.7.",
O.SJJ
0.43
0.62

«j a
i o
9i S
en n (5.
ow~

3.00
3.00
7.r.o
7.50
	
o.oo
0.33
0.33
o..°.:;
O.J53


0.83
1.17
	
0.81
1.00
O.G7
1.00


||-f
o* ~

4.00
4.00
10.00
10.00
G.OO
'1.50
1.08
1.08
0.75
0.75


O..IG
1.33
0.4 'J
1.17
0.02
0.'.)8
1.50

^
10 V
w c w
5" "

'J.OO
<1.00
10.00
10.00
G.OO
	
	
	
	
	


1.4G
1.85
0.97
0.0
2.1G
1.-1S
1.50

fc
JS "** Ui
"c S —•

0.50
1.00
3.00
2.50
1.50
3.50
0.33
0.33
O.G7
O.G7


0.25
D.GJi
0.25
0.33
0.50
0.50
0.67

£ H
Jf 1
->£ a c
« "" H h
Group A
0.25 27.0 15
0.50 11.0 15
1.25 17.0 15
1.25 12.5 15
0.75 16.6 15
0.25 22.0 17.5
0.10 13.5 to 16.6 17.5
0.10 lG.Gtol7.4 17.5
0.17 1C to 20 22
0.17 14.0 22

Group B
	 11.5 27
0.23 G.5 18
	 8.1 24
	 19.8 26
	 12.3 29
0.25 9.1 18
0.35 11.3 20

-
Ai
C
(d
Of


14
14
16
1G
15
17.5
13
12.5
13
12.5
Mean

14
13
12
12.5
13.5
12
12.5
Mean

*
9
o»
1

1C
16
16
1C
16
16
16
16
16
1C


1G
1C
16
16
16
16
16








Devlatloni* from

*
n
ki
E

12.0
4.0
2.0
2.5
1.6
4.5
2.5
0.5
4.0
8.0
4.2

15.5
11.5
15.9
6.2
16.7
8.9
8.1
11.8
A.
M
V
•o
a
d
s


13.0
3.0
1.0
3.5
1.6
4.5
2.0
4.5
5.0
1.5
4.0

2.5
G.5
3.9
7.3
1.2
2.9
0.6
3.6



Measured X

*
0>
"o.
c.
3

11.0
5.0
1.0
3.5
0.6
6.0
1.0
1.0
2.0
2.0
3.3

4.5
9.5
7.9
3.8
3.7
G.9
4.1
5.8
M
m







>
g
2

^
T*
d
0
K
c
3;
1— «
J5
G
3








-------
                               CYCLONE DUST COLLECTORS
                                                                                        17
   oq oq ca n TT o oq c-; cj o a • «     CM -J a: oi « N >-< •-<
   c^ ifl o oo o eo r-[ o o -3; T^
   co i-« —< —•< CM -ti -w cvi "3 evi csi
   O CO OO in ^ O  (NMOOOOf-
   OOU3OJOJU  TTMOCJoa
   iG r-( r-J i-H rH r-i  CO CJ TT OJ C>j P-!
   coooooo  oowooo
   oq iq o o in o  in in oo ia 1-1 o
   eo w •"? TT •* ^  od o' oo LO eo in
It can only be concluded that a really satisfac-
tory  method for predicting  cyclone pressure
drop  from  cyclone  dimensions over  a  wide
range of construction has not yet  been pre-
sented.
   At the  present state  of  the art, the best
method of determining the pressure drop of a
proposed cyclone is to  depend on the measured
pressure drop of a geometrically similar proto-
type, rather than to rely on a computed value.

2.30  Design  Factors Affecting  Pressure Drop
        and Efficiency

2.31  Diameter and Dimension Ratios
   As has been described generally, the cyclone-
diameter, inlet, and  gas-outlet dimensions are
of primary importance in  efficiency  and pres-
sure-drop consideration.
   A study has been made of 23 families of geo-
metrically   similar  cyclones.   The  consensus
design—the  "average"  dimension  ratios  of
these families—is presented not as having any
particular  virtue, but  as  representing typical
present practice, and thus serving as  a starting
point for discussion of variations in design.  A
widely recommended typical cyclone13 has the
following similar proportion ratios:
                                                                            Ratio:
                                                                                 Element
                                                   Cyclone Element        Consensus  •  Typlcul"
                                                Body diameter D	  2.0        2.0
                                                Cylinder length L	  1.5        4.0
                                                Cone length M	  2.5        4.0
                                                Total length H	  4.0        8.0
                                                Outlet length 1	  1.25       1.25
                                                Inlet height Ht	  0.9        1.0
                                                Inlet width W,	  0.4        0.5

                                              A  higher-efficiency  (and  higher-pressure-
                                            drop) design would have a larger length ratio,
                                            a smaller inlet-width ratio, and a larger body-
                                            diameter ratio with smaller body diameter.
                                              The length of the cyclone body is of impor-
                                            tance. Gas can be considered as flowing to  the
                                            gas outlet from  the main  vortex zone through
                                            an imaginary cylindrical extension of the gas
                                            outlet down to the bottom of the cyclone, or to
                                            its intersection  with  a conical bottom.  The
                                            area  of this cylin'der increases with the length
                                            of the main vortex zone. The mean inward gas
                                            velocity  past it decreases with such increasing
                                            length. Since this mean inward gas velocity is
                                            one force opposing particle motion to the  pe-
                                            riphery, it follows that increasing length of  the

-------
62
AMERICAN PETROLEUM INSTITUTE
                                        TABLE XVII

                               Costs of Cyclone Dust Collectors.
                                                          Cost
                                                (Cents Per Cubic Foot Per Minute)
Size
(Cubic Feet
Per Minute)
1,000 	
5,000 	
10,000 	
50,000 	
100,000 	
Knne"
Standard
	 20
	 10
	 9
9

(1952)
High-Efficiency
37 to 53
20 to 35
15 to 32
12 to 30
12 to 30
Dallavalle
Light-Doty
15
g
7
7

•M1953)
Seavy-Duty •
50
30
25
25

Dorfan n
Simple
10
g
7
z.

(1350)
Multiple
I!)
14

7

    * Includes multitube units. Manifolding banks of multituhe units may raise the cost to 60 cents per cubic foot
per minute.80
efficiency collection is economical at dust load-
ings on the  order  of 1.0  grains per cu ft or
greater.  Cyclone collectors do not  fall  in this
category, except for very coarse materials. They
may still be  used on  collection of valuable ma-
terials, however, if the operating conditions are
so rigorous as to  preclude use  of other, more
efficient types.
  Power costs will be rated on a pressure drop
of approximately 0.25 hp per inch, water gage,
per 1,000 cu  ft per min.
  No generalized data are available on  special
alloy  construction  or refractory,  erosion-,  or
corrosion-resistant  linings.   Thin   refractory
and erosion linings are frequently estimated at
$1 to §2 per  square foot.

 12.00  PRACTICAL CYCLONE  DESIGN
              OR  SELECTION

12.10  Steps  in Design or Selection

  1. Determine the following:
a. Volume rate  of flow of gas and its chemical
and physical properties.
/«. Dust loading, dust particle size and density,
and handling characteristics of dust.
c. Desired efficiency of collection.
(/. Special  problems  of erosion, corrosion,  or
operational problems.

  2. Determine from Sect.  1.32 what general
type of cyclone  is  required.
ft. If  expected efficiency is lower than  desired
efficiency, consider other types of equipment if
those types will  withstand  operating conditions.
l>. If 01 her types of equipment cannot be used,
consider use  of  series operation, bottom purge.
outlet  skimmers,  wet operation, etc., for  im-
proving efficiency.
                     3.  Determine  whether a commercial product
                   is to be  purchased or a cyclone designed and
                   fabricated by the user. Multitube, small-diame-
                   ter,  "high efficiency" cyclones  are  generally
                   purchased items.

                   a. If a  commercial product  is desired,  obtain
                   performance and efficiency data from the fol-
                   lowing:
                              Pilot test.
                              Laboratory test.
                              Similar experience data.

                   Correct  as  required  to  anticipated  operating
                   conditions.
                   b. If a cyclone is to be designed, calculate pres-
                   sure drop  and efficiency  for a "typical"  unit
                   (Sect. 2.31), with an assumed inlet velocity  of
                   50 ft per sec (normal range,  20 to  70  ft per
                   sec).  Efficiency  may be  evaluated  from the

                   curve of Fig. 36 showing efficiency vs. ratio -j,—
                                                            Uff
                   for  this cyclone, and  from a calculation of Dpe
                   (Table I).
                        (1)  If anticipated  efficiency  results are
                   better than desired, a lower pressure  drop may
                   be obtained by reducing inlet velocity (enlarging
                   the cyclone).  Recalculate Dpr from new dimen-
                   sions and  check against Fig. 36  for  desired
                   efficiency.
                        (2)  If anticipated  efficiency  results are
                   poorer than desired,  increase inlet velocity (re-
                   duce cyclone size), provide larger length ratio.
                   smaller inlet-width ratio, larger body-diameter
                   ratio with smaller body diameter, etc.  Recalcu-
                   late D,,.. from new dimensions and check against
                   Fig.  36  for desired efficiency.  Consider the use
                   of the following accessories  or  arrangements
                   for  improving performance:
                         Parallel operation of smaller units.
                         Series operation.
                         Purge from outlet.
                         Wet operation.

-------
                                  CYCLONE  DUST COLLECTORS
                                           63
FRACTIONAL EFFICIENCY, %
— — rorocj jkUio> -gmtoo
oo 01 o 01 o o o o oooo












































/









/








t
/








t










/









,










'








/
/








^^"








x-**









s=-









-*• — :
































^•^










«^










^•B










tmf











1 02 0.3 0.4 0.50.6 OS 1.0 1.5 2.0 3jO 4.0 5.06.0 8.0 10.
                         RATIO PARTICLE  SIZE  TO CUT SIZE Dp/Dpc

                      Fractional Efficiency of "Typical" Cyclone (See Sect. 2.31).
                                            FIG. 36
     (3)  If efficiency is satisfactory but pres-
sure drop too high, consider an outlet drum, etc.,
for pressure recovery.
  4. Make adequate  provision for preventing
gas flow into dust outlet, positive removal  of
dust from cone, and prevention of hopper over-
filling. In  multitube or parallel operation, de-
sign  to  prevent  recirculation  between  tube
outlets.
  5. If cyclone is to be operated wet, take nec-
essary special precautions.
  6. If erosion or fouling is likely, special de-
sign is needed. Consider special linings, scour-
ing, etc.
  7. If corrosive gases or materials are han-
dled, consider special alloys, linings, wet opera-
tion, or prevention of condensation, as may be
indicated.

1220  Field Tests on Fluid Catalytic Crackers

  Results of testing full-scale cyclones in fluid
catalytic-cracker installations are tabulated in
Table XVIII.  The  operating conditions  and
lack of access to the cyclones make such testing
difficult and account for the missing values in
the table for many tests.
  Collection efficiency in  such service  is high,
primarily on account of the high dust loadings
and large particle size. Note that even  with
such efficiencies, losses are on the order of tons
per day.  Cyclones in ordinary service do not
provide these high efficiencies.

-------
                                                                                                                                                                         en
                                                                       TABLE  XVIII

                                                Field Tests  on  Full-Scale  Fluid  Catalytic Crackers •
Hem  No.:  	       1              2              8              4             D
Service   	  llcgcncrntor    Ilegciicrotor   Regenerator     Itcaetor       Reactor
Number o( Beta  In  parallel	       0              8            13              0             P
Number In series	       1              2              2              1             2
Cyclone diameter,  ft:                                                                                  ,
    Primary   	   Footnote '     Footnote*         3.0        Footnote1     Footnote '
    Secondary 	   Footnote'     Footnote'         3.0        Footnote'     Footnote'
Outlet diameter,  ft:                                                                            _       ,
    Primary                   .  .   Footnote'     Footnote'         1.07       Footnote'     Footnote'
    Secondary".	   Footnote"     Footnote'         1.07       Footnote'     Footnote"
Inlet  conditions:
    Dunt land, grains per cu  ft at
      condition* 	      —             —          2.800             —            —
    Mass median slue, microns '..      CO»           37'          37*           *3"              . ..
                                         1.0 •          1.0*           1.0*          1.5*           1.8 •
    Inlet Telocity, ft per gee	      _             —            80             —            —
Ded conditions:
    Density,  Ib prr cu ft	      30.6           27.7          31.9           32.2           40.1
    Vaporvcloclly.ftper.ee	       170          2.28           2.33           1.70           1.05
    Disengaging height, ft	      20             22.3          20             20            80
    Temperature, degrees fahren-
      heit  	      —            —          ».°yo             —            —
    Pressure, palg  	      —             —            13-°           —            —
Outlet conditions:
    Load, grains per cu ft STP	      —            —              0.85          —            —
    Mass nicdlnn glee, microns*...       4'           33            —            14            11
                                         3.2«          1.7          —             1.0            1.8
     Efficiency, uer cent by weight..      —            —            08.08          —            —
     1,088,  tons  per day	      09            12.D            8.8           40              0.8
Ons volume, cu  ft per sec STP	      —            —            007             —            —
Remarks  	  Lined, Hn.     Unllned, {-In.   Primary:      Lined, )-ln.    Mncd, 1-ln.
                                     carbon slcrl   carbon stacl    lined.  |-ln.     stainless       carbon steel
                                                                  carbon steel   steel
                                                                 Secondary:
                                                                  unllned, J-ln.
                                                                  carbon stMl

     • Standard geometric deviation.
     'Catalyst In bed.
     • Includes Cottrell-prcclpllator return Ones.
     ' Dilute  phase 18 ft above dense bed.                             .      . _ ...
     • Courtesy Pan American  lleflnlng Corp.. The Pure Oil Co.. Union Oil Co.  of California.
     1 Dimensions requested  of vendors;  to be  forwarded  when  received.
     0
 Itoconerntor
     a
     3

 Footnote r
     3.0

 Footnote '
 Footnote '
 2.000
    43 b
     1.5*
    00

    36
  1,115
     8.6

     0.01
    12
     2.3
    09.070
     3.2
   070
Third-stage
 cyclone an me
 nn Bcoond-
 slnee
 Regenerator
     0
     R

  Footnote '
  Footnote '

  Footnote '
  Foolnoln '
  4.370
    42 b
      1.0*
    78

    32
  1.116
     8.4

     O.GO
    ID
     2.0
    00.804
     3.0
   000
Third-singe
 cyclone pnmc
 as sccond-
 Btago
     8
Itcecnerator
     0
     2

 Footnote '
 Footnote '

     0.83
     0.83
   0-7
 1.110
   22.3

     0.78
   15
     1.8

     0.5
 1.850
     0
Itcecncrator
     0
     2

 Footnote '
 Footnote '

     0.83
     0.83
                  84
 1.110
    10.8

     0.67
     6.0
 1,170
                                                             to
                                                             O
                                                             §
a
K
                55
                3

-------
                                  CLONE DUST COLLECTORS
                  CYCLONE  DUST  COLLECTORS
         (REPORT  ON REMOVAL OF  PARTICULATE MATTER  FROM
                                 GASEOUS  WASTES)
          1.00  INTRODUCTION
 1.10  Definition
   A cyclone is a structure without moving parts
 which separates particulate matter from gas by
 transforming the velocity of an inlet gas stream
 into a vortex confined within the structure. A
 vortex is a helical form of rotation of fluid about
 an axis. In nature, the axis of a vortex is  fre-
 quently  curved, and  although the  axis of the
 confining structure might  likewise be curved,
 it is in  practice a straight line.  Similarly, al-
 though  the ideal structure would be one gen-
 erated by a  curved  line revolved  about  the
 cyclone  axis,  practical cyclones employ confin-
 ing structures generated by the revolution of
 straight lines about the axis, i.e., combinations
 of cylinders and cones (Fig. 1).

 1.11   Types of Cyclones
   Structurally,  a cyclone  must have an axial
 gas outlet, a  dust discharge, and a means for
 gas inlet which will  produce the gas  rotation
 necessary to create the vortex.  These three ele-
 ments may be combined in a number of differ-
 ent ways.  Rotation may be produced by tan-
 gential gas entrance  or by axial gas entrance
 through a set of swirl  vanes.  Separated dust
 may be  removed either axially or tangentially
 from  the  periphery.   Dust  may  be  removed
 from  either the end opposite to  the axial  gas
 outlet or from the same end. There  may be
 either one or a multiplicity of tangential inlets.
 The cyclone body may be completely cylindrical,
 completely conical, or made of both cylinders
 and cones. The gas outlet may be either cylin-
 drical or conical.
  Cyclones in common use may be classified as
 follows:
 1.  Cyclones with tangential inlet and axial dust
discharge.
    la.  Larger-diameter ("conventional")  cy-
    clones.
    Ib.  Small-diameter ("high efficiency")  cy-
    clones.
                         Vic
                             (HEIGHT OF «n*ULUS I
 ANNULUS-"!	
       900Y... 	
                              KOPPE1
               Typical Cvclone.
                  FIG. 1

2.  Cyclones with tangential inlet and peripheral
dust discharge.
3.  Cyclones with axial inlet and axial dust dis-
charge.
4.  Cyclones with axial inlet and peripheral dust
discharge.

  Type la is the most commonly used type in
which the total gas volume is handled through
one cyclone body, although sometimes such cy-
clones are arranged in parallel  (Fig. 1). The
other types are  usually of small capacity indi-
vidually, and practical gas volumes are obtained
by operating large numbers of individual units
in  parallel.
SOURCE:  Cyclone Dust  Collectors, Engineering Report Prepared for American
          Petroleum  Institute,  Washington,  DC (1956).

-------
AMERICAN  PETROLEUM INSTITUTE
1.20  Mechanism of Cyclone Operation

  The cyclone separates particles from the gas
by means of centrifugal force exerted  on the
particles in vortex flow, tending to drive the
particles to the  wall of the cyclone body.  The
particles at the wall move toward the dust dis-
charge by virtue of the axial component of the
vortex flow, aided, if the  axis  is vertical, by
gravity.  The magnitude  of the radial  forces
acting on the particles depends on the nature
of the vortex flow in  the different sections of
the cyclone. Counteracting forces, such as ra-
dial gas  velocities, tend to offset the separating
forces.

1.21  Nature of Vortex Flow
   The inlet of a cyclone transforms linear flow
into vortex flow. After the gas leaves the inlet
annulus, it establishes a  vortex with an axial
component in the direction of the dust discharge.

   1.211   FLOW  IN THE  MAIN VORTEX:  The
 characteristic of the main vortex is that the tan-
 gential velocity at radius r (V,) increases as r
 decreases from the radius of the confining struc-
 ture r, to a maximum Vlm  at some intermediate
 radius rm.  As  r further decreases from rm to
 zero at the axis, V, decreases from its maximum
 Vtm at rn to zero at the axis.  The zone  between
 the axis and rm is called the core of the vortex.
 In the  outer zone of the vortex where  V, in-
 creases as  r decreases, the flow obeys the equa-
 tion :

                 Vt=v,p(^)"             (1)

 Where:
    Vt= tangential velocity at radius r, in feet
           per second.
    Vtp= tangential velocity at rr, in feet per sec-
           ond.
     rp = radius confining structure, in feet.
     r = radius, in feet.
     n = exponent, dimensionless.

    If  — = r,, the ratio of the radius  of rotation
      rp
  to the  radius of the confining structure, equa-
  tion (1) reduces to:

                    V, = Xffi                (2)

    In an ideal gas, n would equal 1. Real values
  of n are  less than 1, and at ordinary  tempera-
  tures are greater than 0.5.
                    Alexander-" presents an equation evaluating
                  n for air at 10 C, as follows:
                                   n=(^T''             (3)

                    For variations in gas temperature, he found
                  n to vary as:
                   Where:
                      n-r = value of n at T Kelvin.
                      n0= value of « at 283 K (10 C).
                      Tk = temperature, in degrees Kelvin.
                     Values of n computed from equations (3) and
                   (4) are presented in Fig.  2. Individual  values
                   of n reported by other investigators  are:
                       Shepherd  and Lapple  '•"• ........ 0.5
                       Stairmand™ ................. °-5
                       Ter Linden'"  ................ 0.52
                       Prockat37  ................... 0.7
                       First10  ...................... °-88
                      There is radial gas flow in  the outer  vortex
                   zone. There has been much theory propounded
                   on this point  but little experimental measure-
                   ment.  The simplest theory, that of  Feifel" for
                   a  simple cyclone with no gas-outlet extension
                   into the body, assumes "sink" flow  radially in-
                   ward from inlet to  outlet superimposed on tan-
                   gential vortex flow.  Unfortunately, however.
                   the use of an  outlet-duct extension  into  the cy-
                   clone body, and the  presence of space  below
                   and above the inlet into  which gas can flow,
                   causes radial velocity to vary in a manner pres-
                   ently defying mathematical statement.
                      Radial velocity has been measured in  a con-
                   ventional cyclone  with air and in  a hydraulic
                   cyclone with  water.  In the hydraulic,  cyclone
                   the customary  underflow (downflow  through
                   bottom of cone) was suppressed. Typical radial
                   velocity profiles based on these findings show
                   that from rf to some intermediate  radius there
                    is inward  radial velocity, but from the inter-
                    mediate radius to the axis  there  is  outward
                    radial velocity (Fig. 4).
                      1.212  FLOW IN VORTEX CORE:  Flow in  the
                    vortex core follows the equation :
                                                             (5)
                    Where:
                        K = proportionality constant varying  with
                             Vf, rm, and n.
                        q = velocity exponent.

                      • Figures refer to REFERENCES on p. 66.

-------
                                  CYCLONE DUST COLLECTORS
   10

   8

   6

   5

   4
<
 £ 3
                        = 2.0.
Lo/1

                       3   456    8   10
                     20    30   40 50 60   80  IOO
                                                                                           200
                                Values of Ki [See Equation (17)].

                                            FIC. 6
and its helical pitch about 10.5 ra.  In extremely
long cyclones, the eccentricity has been observed
up to 50 r0 before disappearing.
  As will be shown later  (Sect.  2.31), cyclones
of common design have a "natural length" on
the order of 10 to 15 r0.
  At the  vortex base there is a region of con-
siderable turbulence and reversal in direction of
axial flow, with a likelihood that the core eccen-
tricity will bring this region into close  prox-
imity  with, or in actual contact with,. already-
separated dust at the cyclone wall. It is for this
reason that much attention should be  devoted
to the design of dust discharge, purge,  and the
like.

1.22  Separation of Particles

  Particulate matter is separated from gas in  a
cyclone by centrifugal force, or radial  force,
tending to drive  the particles (against the re-
sistance  of motion by the gas)  to the cyclone
wall.  The radial force imparted  to the  parti-
cle is:
                 F.. = -Hl               (6)
Where:
   F, = radial separating force, in pounds.
   m,, = particle mass, in pounds.
   V,, = tangential velocity of particle, in feet
          per second.
    g = gravitational constant = 32.2 ft per sec
         per sec.
  The tangential velocity of the particle is theo-
retically lower than that of the gas in which it
is suspended;  but, in considering fine particles,
the two velocities are equal for all practical pur-
poses and V, = Vt.  By combining equations (2)
and (6):
                 T-,   mpV,p:
                 r , = -	
                                       gr rr11
                  which has a maximum value:
                                       TnpVt,n:
                                  F,ra=-
                                                           (7)
                                          (S)
                                       grmrfm-n
                  Where:
                    rfin = radius ratio at r,,,, dimensionless.
                   Vtm = maximum tangential velocity (at ?•„,), in
                           feet per second.
                   The value m? in the  preceding equations is
                  proportional to the cube of the particle diame-
                  ter, thus:
                                   mp=/3/0pDp1               (9)
                  Where:
                     /? = volume shape factor, dimensionless.
                     Pf = particle density, in  pounds per  cubic
                           foot.
                    Dp = particle diameter, in feet.
                  The separating force  [equation  (7)] becomes:

-------
                             AMERICAN PETROLEUM  INSTITUTE
100.
                                  rp/re (STAIRMANDX9
                                                  (TERLINDEN)
                                                    H/To (TERLINDEN  AND BROER)(5)
                           rpAo (TERLINDEN)
                                       4567
                                  NUMBER  OF OUTLET  RADII (Co)

                                 Dimension Ratios and Efficiency.
                                           FIG. 7
  It can be shown =• that  the force  ("Stoke's
Law force")  resisting motion  of a particle
through a gas in the range of particle diameters
from 3 to 100 microns is:
                 Fr=KjuDp/i             (11)
Where:
   Fr=frictional resistance to flow, in pounds.
   K:=proportionality constant, dimensionless.
    u = particle velocity with respect to gas, in
         feet per second.
    /i = gas viscosity,  in pounds per second per
         foot.
  Comparison of equations (10) and (11) shows
that the  separating force  varies with the cube
of the particle diameter, whereas the resistance
                                                to flow varies only linearly.  Thus large-diame-
                                                ter particles have a greater ratio of separating
                                                force to opposing force and are, therefore, more
                                                readily separated in a cyclone.
                                                  The velocity u of the particle with respect to
                                                the gas includes the radial as well as tangential
                                                components of gas flow  (Sect.  1.211). The as-
                                                sumption of Feifel allows the computation of a
                                                "critical particle size" (Der) for any radius of
                                                rotation. Der may be considered the  largest size
                                                of particle which is not separated from the  gas
                                                stream. On the basis of this simplifying  as-
                                                sumption, D„ could be computed for r0, and all
                                                particles larger than Der  would be  collected
                                                while all smaller particles would be lost into the
                                                outlet duct.

-------
                                  CYCLONE DUST COLLECTORS
           Axinl
   Helical

Cyclone Inlets.

   FIG. 8
                                                                             Involute
  In fact, however, the radius at which a parti-
cle may be considered lost to the exit gases is
not constant because the radial gas velocities
vary along the axis of the  cyclone.  Since  the
radial velocity pattern has not yet been mathe-
matically  stated,  there can be no reliable com-
putation of critical particle size.  In actual prac-
tice,  the presence of the gas-outlet extension,
the finite width of the entering gas stream,  the
random distribution of particles of various sizes
across this width, the presence of major eddy
currents,  etc., all further obviate the value of
any such theoretical computation.
  Even were  such a computation  possible, it
would  be  found that not one, but  a series of
critical sizes would be obtained for various in-
crements  of cyclone length  and from one side
to the other of the inlet.  Therefore,  it follows
        that a cyclone must be, and is, a poor classifier
        and does not make a  sharp-size cut between
        particles separated  and rejected. The cyclone
        has a  curve  rather  than a straight line  as  its
        particle  size-efficiency characteristic.  Qualita-
        tively, the predominant requirements for high
        collection efficiency  of fine particles  are that
        rlm and rm be low and Vtm high.
           A number of authors have derived equations
        for Der, as shown in Table I. No critical evalua-
        tion has been made which would show one equa-
        tion more nearly correct than  another.  The
        main value is to demonstrate the parameters  in-
        volved and their qualitative relationships.

        1.23 Discharge of Separated Dust
           The separation of  the dust from the gas
        stream is, in practical' terms, only the  first step

-------
10
AMERICAN PETROLEUM INSTITUTE
       Reference

Da vies"	
            TABLE I

Formulae for Critical Particle Size

       Foninilii
u»h"
                         , = an:i,, =
        g             / r\ i "
    — "-p'i'Vjftl 1-\ ~r')  J
                       ''
Shepherd--


Muhlrad-1 .
D'-VsEVT
Rosin, Rammlcr, and
   Intelmann::? 	
           Y
Tarjan-":
         Spoci.il Units: Remarks
aMin = particle size above which effi-
       ciency will be 100 per cent
       in  absence  of  re-entrain-
       ment.
  n = viscosity, in  pounds per  sec-
       ond per foot.
  /i = viscosity, in  pounds per  sec-
       ond per foot.
 D,.,, = cut size, collected  at  50-per-
       cent efficiency.
 N, = "effective" number of turns in
       cyclone = 5 to 10 for typical
       cyclone  (Sect. 2.31).
  ju= viscosity, in pounds per second
       per foot.
       For .V, see reference.

  w = —
       All units metric—see refer-
          ence.

  Y= y -r = instantaneousdistance
       of  particle  from   cyclone
       wall. All units  metric—see
       reference.
 d,,,i,,= particle diameter at "blocking
       radius r."
   r = "blocking radius."
 in a series intended to result in solids or liquid
 at rest in a container, and a gas stream cleaned
 of particulates to the desired degree. The con-
 centrated dust layer swirling slowly (compared
 to the vortex)  down the walls of the cyclone
 body must be conducted into the dust hopper, or
 its equivalent, with a minimum of re-entrain-
 ment into the base of the vortex. Factors affect-
 ing re-entrainment are:
 a. The length, eccentricity, and diameter of the
 vortex core and its relation to the cyclone body
 walls.
 b. The presence or absence of purge flow out-
 ward  from the dust outlet.
 c. Recirculation of gas,  or infiltration of air,
 into the dust outlet.
 d. Presence  or  absence of an air-lock material
 discharge  valve or dip-leg.
                    e. Smoothness of the inner walls of the cyclone.
                      The two major types of discharge are axially
                    through a core and peripherally. Both methods
                    may be used with or without purge gas, or with
                    various types of mechanical  accessories. These
                    methods, and their effect on efficiency, will be
                    discussed in later sections.
                      The dust in the gas stream which escapes into
                    the outlet duct tends to be concentrated toward
                    the walls of the duct. Therefore, various de-
                    vices for skimming  off the outer layers of this
                    vortex have an effect also on  the overall per-
                    formance of a cyclone.

                    1.30  Scope of the Cyclone in Field Use
                    1.31  Typical Operating Ranges
                      Cyclone collectors, as a class, are regarded as
                    the lowest in first cost of equipment convention-

-------
                                 CYCLONE DUST COLLECTORS
                                                                         11
ally used for dust or droplet collection. They
also, as a  class,  provide lower collection  effi-
ciency than other types of equipment, although
there are, of course, exceptions to this general-
ity. They are suitable for medium and coarse
dusts  and are unsuited for very fine  dusts or
metallurgical fumes.  They  have advantages,
however, in that there  are  no moving  parts,
choice of materials of construction is wide, and
maintenance costs are low.  Power requirements
(if reasonable efficiency on  fine dust  ia  re-
quired) are as high as, or higher than, most
other types of equipment, although some types
of wet scrubbers have even  higher  require-
ments.

1.32  Efficiency in Field Use
   A clarification  in terminology regarding effi-
ciency will aid in this and  later general discus-
sions.  Efficiency brackets are arbitrarily set as
follows:
                                Ainniinr Passing
  rnlli-cfinn Efficiency     Ktlici'-nry        Collector
  il'erCcnt by W*icht)       Rair.-»        il'prOnc)
Above 50 to 80	Low        Above 20 to 50
Above 80 to 95	Medium     Above 5 to 20
Above 95 to 99	High       Above 1 to  5
Above 99 to 99.99.. Very high   Above 0.01 to 1
Above 99.99	Ultra high  Less than 0.01

   On this basis, most cyclones as applied oper-
ate in the low and medium efficiency ranges. A
well-designed  but  otherwise conventional  cy-
clone may be expected to provide high  efficiency
for coarse pavticulate matter of the 40- to 50-
micron diameter range:  and  small-diameter
 (less than  1-ft) "high efficiency" types  may
extend the high-efficiency  range to particles as
small  as  15- to 20-micron sizes. Typical  effi-
ciency ranges for various particle size  ranges
are as follows:
    I'll rli.-lp Six..-
"Cnnvpiirlnn.il"
 rydoni" Ertl-
 Less than 5	 —
  5 to 20	 Low
 15 to 50	 Medium
 Greater than 40. . High
"Ilich Efarl-ncy"
  Cyrlone Kffl-
  rii'iiny Range
  Low
  Medium
  High
  High
   For various practical reasons, efficiencies in
 the very high or ultra-high  range are almost
 never achieved on  industrial aerosols  by cy-
 clones.

 1.33  Pressure Drop in Field Use
   The loss in pressure of the gas stream flowing
 through a cyclone depends on a number of vari-
ables but, in practice, is usually found to be on
the order of 1 to 4 inlet velocity  heads.  With
velocities commonly used  for handling dusty
air, the velocity head will range from i to 3 in.,
water gage. Thus the range of resistance would
be from £ to 8 in., water gage.  Where ordinary
centrifugal blowers are used  as the source of
motive power for the air. a practical upper limit
of resistance for the cyclone plus  other system
components is in the range of 15 to 16 in., water
gage,  although  with  other  types  of motive
equipment this upper limit may be considerably
higher. Efficiency increases with inlet velocity,
as will be shown later, but at a lower rate than
the pressure drop  increases. Furthermore, for
any given cyclone  and dust combination, there
is an  inlet velocity beyond which  turbulence
increases more  rapidly  than  separation, and
efficiency begins to decrease.  Thus, even aside
from power cost considerations, there is little
purpose in further increases in resistance. Most
cyclones  will be found to have from 2- to 7-in.,
water gage, resistance as operated.

1.34  Dust Loading in Field Use
   There appears to be almost no practical upper
limit of  dust  loadings to a properly designed
cyclone.  A cyclone can be  designed to  handle
any amount of material which can be moved by-
gas flow and, in general, cyclone  efficiency in-
creases with increasing dust load.  Since this
ability is not possessed by  many other types of
collectors  of  inherently higher efficiency, cy-
clones are frequently used as  precleaners for
other  types when  dust loadings arc too heavy
for the secondary collector.
   Conversely, cyclone efficiency decreases with
decreasing dust load, and  other  types  of  col-
lectors are well  able to handle the lighter loads
alone  at higher efficiencies.  As  a result, cy-
clones are usually  applied  to dust  loading.-- of
10 grains per cu ft or higher,  and seldom are
used below 1 grain per cu  ft, unless the dust  is
unusually coarse or other factors of tempera-
ture, corrosion, and the like  are  overriding  in
importance.

 1.35  Ran*e of Capacities
   Commercial cyclones are built in sizes having
 individual capacities of 30 to 30,000 cu  ft per
min.  In general, the smaller  units  are more
 efficient but must be used  in  parallel to accom-
 modate  practical gas volumes.

-------
 12
AMERICAN PETROLEUM  INSTITUTE
1.36  Range of Physical and Chemical  Condi-
         tions
   Cyclones, as a class, can be designed to handle
a  wider range of chemical and physical condi-
tions of operation than any other type of collec-
tion equipment. Any conditions of temperature,
pressure, and corrosion for which  structural
materials are available can be met by a cyclone.'
   Erosion is sometimes troublesome, and some
materials of  construction necessary for  tem-
perature  or corrosion conditions may  not  be
especially desirable in erosion resistance.
   The  main  limiting factor in conditions  of
cyclone application is the physical characteris-
tic of the dust. A sticky, hygroscopic, or similar
material may  give difficulty in cyclone  opera-
tion, and such materials are sometimes handled
in cyclones operated wet, or in "wet collectors."
Dusts which  have a bad  tendency to bridge,
hang up, or cake may give difficulty at the dust
outlet of small-diameter cyclones.

   2.00  EFFICIENCY  AND  PRESSURE
                   DROP
2.10  General  Considerations  of  Efficiency
   For  high efficiency,  the  separating  forces
should be high, the  number  of gas revolutions
large, the tangential velocity high, and the dust
removal effective so that separated dust is not
re-entrained.
   Considerations governing the critical particle
size Dn have  been presented previously.  Since
the particle size-efficiency performance of a cy-
clone is a curve  rather than  a straight  line,
there is no direct mathematical relation between
efficiency 17 and Dcr.   However, in  a qualitative
sense, those parameters tending to increase Der
should decrease ^ for a given dust  size distribu-
tion, and vice versa.
   By inspection of Table I, it may be seen that
efficiency will increase with  an increase in the
following:
         Dust-particle density pf.
         Inlet velocity V{.
         Cyclone-body length H.
         Number of gas revolutions N.'

         Ratio •£-•
               do
         Dust-particle size Df.
   Similarly, efficiency will  decrease with an in-
crease in the following:
         Gas viscosity /*.
         Cyclone diameter D.
                           Gas-outlet-duct diameter d0.
                           Increase in inlet width Wt.
                           Inlet area A{.
                           Gas density />.

                     Efficiency is also affected by various acces-
                   sory  devices not included  in the mathematical
                   relationships in Table I.  It is obviously desir-
                   able  to have  smooth inner surfaces in the cy-
                   clone body  in order to minimize projection of
                   separated dust out into the air stream.

                   2.20  General Considerations of Pressure Drop
                     For a given cyclone handling clean air, the
                   resistance varies with the square of the air vol-
                   ume Q, or similarly with the square of the inlet
                   velocity Vf.  Since the expression  for velocity
                   pressure hv of a gas is:
                                  ^=0.0030^,            (12)

                   he similarly varies with the square of the veloc-
                   ity.  It therefore becomes convenient to express
                   cyclone pressure drop in terms of number of in-
                   let velocity heads,  hn, thereby making such equa-
                   tions independent of gas volume and inlet ve-
                   locity.
                     It will be apparent from later discussion that
                   a  satisfactory method for predicting pressure
                   drop  from cyclone dimensions over a wide range
                   of  construction has not  yet  been developed.
                   Pressure drop should preferably be determined
                   experimentally on a geometrically similar pro-
                   totype.
                     Most investigators agree that cyclone resist-
                   ance  is a function of gas-inlet area and outlet
                   area of the form :
                   Where:
                      X = proportionality constant, dimensionless.
                     HI = inlet height, in feet.
                     W, = inlet width, in feet.
                      d0=outlet diameter, in feet.
                       z = exponent, dimensionless.

                     Values for X are given as follows :
                   a. Ter Linden M states that for common cyclones
                   with straight  tangential inlets  without inlet
                   vanes z = l and A" is between 25 and 32, but that
                   for involute inlets both r and X are lower. For
                   a  180-deg involute inlet his data  show X = 21.7
                   if s = l (Fig. 10).
                   6. Lapple67' gives  a  value for X of 18.4 for a
                   360-deg involute inlet.

-------
CYCLONE DUST COLLECTORS
                                         13
                                       ENTRY  WIDTH
                                       ENTRY  HEIGHT
            2           3
ENTRY  DIMENSIONS-INCHES
 Entry Dimension  Vs. Efficiency.
           FIG. 9
                                               7   8   9 10
c.  For a cvclone having dimension  ratios — =
                                        TO

2 to 4, —= 8, j-=Q,  Shepherd and  Lapple""1

reported "2 = 1 and A" = 16 without  inlet vane,
and  7.5 for  an  inlet vane  which neither ex-
panded nor contracted the entering gas and did
not touch the outlet  (a "neutral" vane). They

also  found that variations in — from 6  to zero
                            TO
had  no effect on A".

d.  First17 found that a reduction in —  from 2
                                    9
to  zero caused  a 10-per-cent  decrease in X

(^=2  and— =5.8 ) . In cyclones with-*= 2
\T.         r.       )

and  — =2 but with — variable, First concluded
     *o              o
that 2 = 1, and derived  an expression for X, as
follows:
                      12
                       Y
                Where:
                   Y = 0.5 for no inlet vane;
                     = 1.0 for neutral inlet vane;
                     = 2.0 for inlet vane that expands entering
                         gas stream and touches outlet duct.
                   L = cyclone-cylinder length, in feet.
                   M = cyclone-cone length, in feet.
                   D = cyclone-body diameter, in feet.
                  Alexander1 concluded that  * = 1,  and derived
                an expression for X, as follows :
where / varies with n as follows :
  n ..... 0  •    0.2    0.4    0.6
  f ..... 1.90    1.94   2.04   2.21
                                                      0.8
                                                      2.40
                  Values of  Alexander's  function  for X are
                plotted in Fig. 5.  For values of n, see Fig. 2.
                  Stairmand has derived  a  theoretical expres-
                sion for /!„, as follows:
                                         /  A  \ n
                                                        (16)

-------
                              :..»::EICAN PETROLEUM  INSTITUTE
                             a- NON EXPANDING
                             b- EXPANDING
                             e-THROTTLING
                 Inlet Vanes.

                   FIG. 10
where:

   See Table II for computed values of -^, and
Fig. 6 for plots of equation  (17).
  The effect on A" of variations in -^ and — was
                                1 „     ia
investigated by Ter  Linden and  reported by
Broer * as the  following multiplying factors for
correcting X:
     r.
     1.5.
     2.0.
     2.5.
     3.0.
Factor
0.91
1.00
1.08
1.19
 r.
 4.
 6.
 8.
10.
Factor
1.17
1.08
1.00
0.96
  A test of the results of these various equa-
tions has shown those based on equation (13)
to have a higher order of reliability than those
of equation (16) ; and, further,  indicates that
the exponent 2 in equation (16)  is probably
too high.
  A summary of factors for application to equa-
tion (13) is presented in Table III.
  An  attempt has been  made to  analyze  the
                                                                 TABLE II
                                                                      y
                                                  Calculated Values of -=£ from Equation (20)
                                                                      "
0.10
1.21
1.53
1.95
2.24
2.47
0.2(1
1.08
1.31
1.58
1.75
1.88
0.30 1.0 2.0
Calculated^
0.87
1.00
1.13
1.21
1.27
0.71
0.78
0.85
0.89
0.92
0.55
0.59
0.63
0.65
0.67
3.0
0.38
0.40
0.42
0.43
0.43
validity of equations (14) and  (15),  and the
values cited  by Shepherd and Lapple,'3'"  by
comparing calculated values  of A" with  meas-
ured values of A".  Three sets of data were avail-
able and are presented  in Table IV. Group A
was reported  by Stairmand,"  Group  B  by
Stern,31 and  Group  C represents  field survey
data previously unpublished.  Since the three
groups of data probably involve differences in
the point and method of measuring static pres-
sure, and it is known that in Group C it was not
possible to determine the presence or absence of
internal accessories, the three groups were ana-
lyzed separately.
  Calculated X shows a large deviation from
measured X  in all three groups of  data. The
constant 16 proposed by Lapple gives less devia-
tion for Groups A and  C, while the  Alexander
equation gives lowest deviation for Group B.
The data were  further subjected to analysis of
variance ratio,:u with the result that, for  Group
A, all three methods of computation were found
equally reliable; for Groups B and C; the Alex-
ander  method  and the Lapple constant were
equally reliable, while the First method was less
reliable.  For Groups A, B, and C as a whole, the
Alexander and Lapple values were equally  re-
liable, while the First method was less reliable.
The results of this analysis may be summarized
as follows:
  Dnt a
  Group          Host Method
   A	 Lapple
   B	 Alexander
   C	 Lapple
A, B, C	 Lapple
   Equally Good
First, Alexander
Lapple
Alexander
Alexander
                                    This analysis shows a constant value for A" to
                                  be equally as good as, if not better than, equa-
                                  tions for computing A" based on cyclone-dimen-
                                  sion ratios. It seems only reasonable to assume,
                                  however, that dimension ratios  will affect A";
                                  in fact, this is proposed by the same authors
                                  who advocate a "constant" value (Table  III).

-------
Sourco
Ter Linden011.
Lapple"7 . . . .

Shepherd and
Lapple0- "
Broer" 	
First"
J l
..j 2
1.G2 to 3.25

f 2 to 4
"i 2 to 4
1.5 to 3.0
2
8
8

8
8
4 to 10
Variable
i
r.
1
(Approxi-
mately)
6
6
2
Alexander'
Lapple"
                                                     TABLE III

                                        Pressure-Drop Factors for Equation (13)

                                                  Special Condition!         x
                                                No inlet vane          1
                                                180-Deg involute        1
                                                  inlet             (Assumed)

                                                360-Deg involute        —
                                                  inlet, slep-
                                                  tapered body

                                                No inlet vane          1
                                                Neutral inlet vane      1

                                                No inlet vane          —

                                                Y = 0.5 no inlet         1
                                                     vane;
                                                  = 1.0 neutral in-
                                                     let vane;
                                                  = 2.0 expanding
                                                     inlet vane
                                                                                                 X
                                                                                              25 to 32

                                                                                                21.7


                                                                                                18.4
                                                                                                16
                                                                                                 .7.5

                                                                                  See text for X correction factors

                                                                                                   12
                                                                                                   Y
                                                                                  	A co c
                                                                                  — i.b£—
                                                                                       rn
                                                                                              /L\J/M\i
                                                                                              I'D)  ljj)
                                        1.25    No inlet vane
         n
(Fig. 5)

   8
                                                                                                                       o
                         •2.
                         M
                         rj
                         I
                         o
                         o
                         S
                         H
                         O
                                                                                                                       en

-------











Analysis
TABLE IV

of Calculation Methods for

X


Calculated Values of X
q
fc*
a> *"* *—
lIs
u ^ k*
O'P —

2.00
2.00
5.00
5.00
3.00
2.84
O.G7
O.G7
O.G7
O.G7



0.98
1.00
O.G7
1.33
1.58
0.75
1.10

•a
*jj M
~ *•• — .
3 c «-•
O B 01
« •- U*
<3Q ""

1.00
1.00
2.50
2.50
1.50
J.OO
0.21
0.33
0.21
0.33



0.38
0.50
0.38
0.7:"!
().S:5
0.4:!
0.62

*•
*• a
Of O

9 S H
Vt "* U<
5 w ~

3.00
3.00
7.50
7.50
	
O.GO
0.33
0.33
0.33
0.33



0.83
1.17
	
0.31
1.00
O.G7
1.00



.j
|||

CK*~

4.00
4.00
10.00
10.00
G.OO
•1.50
1.08
1.08
0.75
0.7.1



O.-IG
1.33
0.4 '1
1.17
0.62
0.!)8
1.50


£s
_
r.s
W C 4"
eSu,
JO i_i — '

4.00
4.00
10.00
10.00
G.OO
	
	
	
	
	



1.4G
1.85
0.97
0.0
2.16
1.18
1. 50


tc
w
~1?S
M £ U«
C +• —

0.50
1.00
3.00
2.50
1.50
3.50
0.33
0.33
0.67
O.G7



0.25
O.G2
0.25
0.33
0.50
0.50
O.G7

M
u g
!a— i
^3 " "
~ fa tl ^

Group A
0.25 27.0
0.50 11.0
1.25 17.0
1.25 12.5
0.75 16.6
0.25 22.0
0.10 13.5 to 16.6
0.10 16.6 to 17.4
0.17 16 to 20
0.17 14.0


Group 13
	 11.5
0.23 6.5
	 8.1
	 19.8
	 12.3
0.25 9.1
0.35 11.9

•


s
C
ta

15
15
15
15
15
17.5
17.5
17.5
22
22



27
18
24
26
29
18
20


ti
ft*
C
a
K
Ol


14
14
1G
1C
15
17.5
13
12.5
13
12.5

Mean

14
13
12
12.5
13.5
12
12.5
Mean

*
i
9
j£
a
3

1C
16
16
16
16
16
16
16
16
16



16
16
16
16
16
16
16






Dcvln(lnnR frnm

* I

«J
ki
E

12.0
4.0
2.0
2.5
1.6
4.5
2.5
0.5
4.0
8.0

4 2

15.5
11.5
15.9
G.2
16.7
8.9
8.1
11.8
A
^
01
*O
a
a)
H
4)


13.0
3.0
1.0
3.5
1.6
4.5
2.0
4.5
5.0
1.6

4.0

2.5
6.5
3.9
7.3
1.2
2.9
0.6
3.6
i-
c

Measured X



a>
c.
3

11.0 g
5.0 S
1.0 3
3.5 |
0.6 ^
6.0 g
1.0 o
1.0 S
2.0 S
2.0 _
y.
3.3 ^
c
4.5 ^
9.5
7.9
3.8
3.7
6.9
4.1
5.8

-------
                             CYCLONE DUST COLLECTORS
                                            17
oo oq eq so TT o  oo
ci TT c- so r-i o  -«r
                                 w- r^
                      O eo STJ rr   O  O
00 CO O CO ^* O  SO£~CMOU
o'«3<9inTreo  CM o i-J w CM
oq oq eo rq *? o  CM t> CM o «a
oo o eo CD (*•• o  c3 CM CM o ^3*
i-«CM   CMCMOJ  i-HCMCMCsJ
O O O C3 O O  O ;3 O 53 :3 S3
1-1 !-l !—I I—I I-l I—I  I—(I—I r-t J-l F-l I-H
                                 ^ r-
                                 -i  C-
                                 9  «
                                 00 UH
C- OO C- OO
                .
              00 CT f- 00 « O
                                 c   c
                                 d   J«
                                 i   i
incMOCMCMCM-^CMaoeM-«CM       £-1
CM co eo co co co co co CM co CM co
   CMCMOC-;OC) OOSOOOr-iOTr
   so ,-i co" CM TT' o o ci id C4 in o
O   rH CO        CM       CM CO
 a.
 o

^3 inCMOCMCMO CM^OOOOO
   t^^inTTpm ooomiom
   O'OOOOC) OOi-4OOO
  rt _; r-< ,_! I-H v  c\j co c .-! i-
to eg
IOCOSMP7OM C~(MO!MtOC-
c- iq « oq o eo 1-1 CT o •* TT •-;
tt r* —< -H CM W •^•CMWCMCJCM
ooooomrHO CMeooooot-
O O UO OJ CJ O -VCOOOOO
M i-J i-H -H i-i i-I JO W TT CM CM' r-i
CO O O O O O OOIOOOO
eq iq o ca iq o ua us so >q ^ es
o so TT TT -c1 in eo o od in co' in
 It can only be concluded that a really satisfac-
 tory method for  predicting  cyclone pressure
 drop from  cyclone dimensions over  a  wide
 range of  construction  has not  yet been  pre-
 sented.
   At the  present  state of  the art, the  best
 method of determining the pressure drop  of a
 proposed cyclone is to depend on the measured
 pressure drop of a geometrically similar proto-
 type, rather than to rely on a computed value.

 2.30  Design  Factors Affecting  Pressure Drop
        and Efficiency

 2.31  Diameter and Dimension Ratios

   As has been described generally, the cyclone-
 diameter,  inlet,  and gas-outlet dimensions are
 of primary importance in efficiency and pres-
 sure-drop consideration.
   A study has been made of 23 families  of  geo-
 metrically  similar  cyclones.   The  consensus
 design—the   "average"  dimension  ratios  of
 these families—is presented not  as having any
 particular virtue,  but as representing  typical
 present practice, and thus serving as a starting
 point for discussion of variations in  design. A
 widely recommended typical cyclone13 has the
 following similar proportion ratios:
                                                                          Ratio:
                                                                              Element
        Cyclone Element        Consensus  ' Typical"
     Body diameter D	 2.0        2.0
     Cylinder length L	 1.5        4.0
     Cone length M	 2.5        4.0
     Total length H	 4.0        8.0
     Outlet length 1	 1.25       1.25
     Inlet height H,	 0.9        1.0
     Inlet width W<	 0.4        0.5

  A  higher-efficiency   (and  higher-pressure-
drop)  design would have a larger length ratio,
a smaller inlet-width ratio, and a larger body-
diameter ratio with smaller body diameter.
  The length of the cyclone body is of impor-
tance.  Gas can be  considered as flowing to the
gas outlet from the main vortex zone  through
an imaginary  cylindrical extension  of  the  gas
outlet  down to the bottom of the cyclone, or to
its  intersection  with  a  conical  bottom.  The
area of this cylirider increases  with the length
of the  main vortex zone.  The mean inward  gas
velocity past it decreases with  such increasing
length. Since this  mean inward gas velocity is
one force opposing particle motion to  the  pe-
riphery, it follows that increasing length of the

-------
62
AMERICAN PETROLEUM INSTITUTE
                                        TABLE XVII

                               Costs of Cyclone Dust Collectors.
                                                          Coat
                                                (Cents Per Cubic Foot Per Minute)
Size
(Cubic Feet
Per Minute)
1,000 	
5,000 	
10,000 	
50,000 	
100,000 	
Knne=' (1952)
Standard
	 20
	 10
	 9
	 9

High-Efficiency
37 to 53
20 to 35
15 to 32
12 to 30
12 to 30
Dallaratle»<1953)
Light-Duty
15
8
7
7
Heavy-Duty •
50
30
25
25
Dorfan" US30)
Simple
10
8
7
5
Multiple
15
14
11
7
    " Includes multitube units. Manifolding banks of multitube units may raise the cost to 60 cents per cubic foot
per minute.50
efficiency collection is economical at  dust load-
ings on the  order of 1.0  grains per cu ft or
greater.  Cyclone  collectors do not fall  in this
category, except for very coarse materials. They
may still be  used  on  collection of valuable ma-
terials, however, if the operating conditions are
so rigorous as to preclude use of other, more
efficient types.
  Power costs will be rated on a pressure drop
of approximately  0.25 hp per inch, water gage,
per 1,000 cu  ft per min.
  No generalized  data are available  on  special
alloy  construction or refractory,  erosion-,  or
corrosion-resistant linings.   Thin   refractory
and erosion linings are frequently estimated at
$1 to $2 per  square foot.

 12.00  PRACTICAL CYCLONE DESIGN
              OR SELECTION

12.10   Steps  in Design or Selection

  1. Determine the following:
a.  Volume  rate  of flow of gas and its chemical
and physical properties.
/>.  Dust loading, dust particle size and density,
and handling characteristics of dust.
c.  Desired  efficiency of collection.
(1.  Special  problems  of erosion, corrosion,  or
operational problems.

  2. Determine from Sect.  1.32 what general
type of cyclone  is required.
«.  If  expected efficiency is lower than desired
efficiency, consider other types of equipment  if
those types will  withstand  operating conditions.
l>.  If oiher types of equipment cannot be used,
consider use  of  series operation, bottom purge.
outlet  skimmers,  wet operation, etc., for  im-
proving efficiency.
                     3.  Determine whether a commercial product
                   is to be purchased or a cyclone designed and
                   fabricated by the user. Multitube, small-diame-
                   ter,  "high efficiency" cyclones are  generally
                   purchased items.

                   a. If a commercial product  is desired, obtain
                   performance  and efficiency data from the fol-
                   lowing:
                              Pilot test.
                              Laboratory test.
                              Similar experience data.

                   Correct as required  to  anticipated  operating
                   conditions.
                   b. If a cyclone is  to be designed, calculate pres-
                   sure  drop and  efficiency  for a "typical"  unit
                   (Sect. 2.31),  with an assumed inlet velocity of
                   50 ft per  sec  (normal range,  20  to  70 ft per
                   sec).  Efficiency  may be  evaluated  from the

                   curve of Fig.  36 showing efficiency vs. ratio -n—
                                                            UfC
                   for this cyclone, and  from a  calculation of Dpe
                   (Table I).
                        (1) If anticipated  efficiency results are
                   better than desired, a lower pressure  drop may
                   be obtained by reducing inlet velocity (enlarging
                   the cyclone).  Recalculate Dfr from new dimen-
                   sions  and check  against Fig. 36 for dosireci
                   efficiency.
                        (2) If  anticipated  efficiency  results are
                   poorer than desired, increase inlet velocity (re-
                   duce cyclone size), provide larger length  ratio,
                   smaller inlet-width ratio, larger body-diameter
                   ratio with smaller body diameter, etc. Recalcu-
                   late Dp.- from  new dimensions and check against
                   Fig. 36 for desired efficiency.  Consider the use
                   of the following  accessories or arrangements
                   for improving performance:
                         Parallel operation of smaller units.
                         Series  operation.
                         Purge  from outlet.
                         Wet operation.

-------
                                   CYCLONE DUST COLLECTORS
                                                                                      63
lf\f\
IUU
Ort
yu
flA
ou
2? 7n
« f O
y fin
o «°
^ 50
llJ vU
5 10
u. 40
u.
UJ
•a/-v
_J 30
"* 95
2 '3
O
r" 9^
H tvJ
o
<
OC ic
U. l5
10















































/










/









f
/










/










/










^











X









x^
/









^x^*"










<^-











.-*•










, 	 •











p. —


















































































O.I
02     0.3  0.4 0.50.6   OJB  1.0     1.5  2.0     3£)  4.0 5.06.0

      RATIO PARTICLE  SIZE TO  CUT SIZE  Dp/Dpc
   Fractional Efficiency of "Typical" Cyclone  (See Sect. 2.31).
                         FIG.  36
                                                                                       8.0  10.
     (3) If efficiency  is satisfactory but pres-
sure drop too high, consider an outlet drum, etc.,
for pressure recovery.
  4. Make  adequate provision  for preventing
gas flow into dus.t outlet, positive removal of
dust from cone, and prevention of hopper over-
filling.  In  multitube or parallel  operation, de-
sign  to prevent  recirculation  between  tube
outlets.
  5. If cyclone is to be operated wet, take nec-
essary special precautions.
  6. If erosion or fouling is likely, special de-
sign is needed. Consider special  linings, scour-
ing, etc.
  7. If corrosive gases  or  materials are han-
dled, consider special alloys, linings, wet opera-
                                           tion, or prevention of condensation, as may be
                                           indicated.

                                           1220   Field Tests on Fluid Catalytic  Crackers

                                             Results of testing full-scale cyclones in fluid
                                           catalytic-cracker installations are tabulated in
                                           Table  XVIII.  The operating  conditions  and
                                           lack of access to the cyclones make such testing
                                           difficult and account for the missing  values in
                                           the table for many tests.
                                             Collection efficiency in  such service is high,
                                           primarily on account of the high dust loadings
                                           and large particle size. Note that even  with
                                           such efficiencies, losses are on the order of tons
                                           per day.  Cyclones in  ordinary service do not
                                           provide these high efficiencies.

-------
                                                                       TABLE XVIII
                                                Field Tests on Full-Scole Fluid Catolytic Crackers •
Item  No.:  	       1              2             8              *             0
Service   	  Hi-generator   Itegcncrnlor    Regenerator     Itcnctor       Reactor
Number of eels  In  parallel	       0              8            13              fi             0
Number In sfrlca 	       1              2             2              1             2
Cyclone diameter,  ft:                                                                         _
    Primary        .     ...     Footnote '     Footnote '         8.0        Footnote '     Footnote •
    Secondary  	   Footnote'     Footnote '         8.0        Footnote «     Footnote '
Outlet diameter,  (t:                                                                                    .
    rrlmnry                         Footnote'     Footnote1         1.07       Footnote'     Footnote1
    Secondary'!	   Foolnolr- '     Footnote'         1.07       Footnote'     Footnote'
Inlet  conditions:
    Dual load, grains per cu  ft nt
      condition* 	     —             —          2,800             —            —
    Mass median size, microns '..     GO •           37*           37'           43 »           41*
                                         1.0*          1.0*          1.0'           1.8*          1.8*
    Inlet velocity, ft per sec	     —             —            "
Ded conditions:                                                                     .....           ,- .
    Density. Ib  prr cu ft	     30.6           27.7           31.0           32.2           40.1
    Vauor velocity, ft per ice	       1.10          2.28          2.33           1.70          1.05
    Dlaongnglng height, ft	     20             22.3           20             26            80
    Temperature, degrees fahren-                                                   	            	

    Pr ensure, piilg  	     —             —            13-°
Outlet conditional
    Load, grolna per cu  ft 3TP....     —             —             0.83          —            —
    Mass median site, microns*...       4'           33            —             14            11
                                         3.2 *          1.7           —              1-0            1-8
    Efficiency, per cent by weight..     —             —            BO-OS          —            —
    I.oa..  ton. per day	     03             12.0            8.5           40             0.8
Oaa volume, cu ft per BCC 8TP	     —             —            »"             —            —
Remarks  	  Lined. 1-ln.    Unllned.  |-ln.   Primary:      Lined, {-In.    Lined. J-ln.
                                     carbonated    carbonated    lined,  Hn.     stainless      en r lion steel
                                                                  carbon steel   steel
                                                                 Secondary:
                                                                  unllned, Hn.
                                                                  carbon steel

     • Standard  geometric deviation.
     * Catalyst In bed.
     • Includes Cottrell-preclpltstor return Ones.
     ' Dilute phase  ID ft above dense bed.                            -      .„,...
     •Courtesy Pan Amerlcon  Ileflnlng Corp.. Tlie Pure OH Co.. Union Oil Co. of California.
     'Dimensions requested  of  vendors; to be  forwarded  when received.
 KcRcncrator
     0
     8

  Footnote '
     3.0

  Footnote '
  Footnote '
  2.000
    43*
     1.5*
    00

    35
  1.115
     8.0

     0.01
    12
     2.3
    00.070
     3.2
   070
Third-atone
 cyclone same
 ns sccond-
 slnce
 Regenerator
     0
  Footnote'
  Footnote '

  Footnote '
  Footnote '
  4,370
    42 b
      1.0*
    78

    32
  1,115
     8.4

     O.DO
    10
     2.0
    00.094
     3.0
   000
Third-singe
 cyclone same
 as Recond-
 stage
     8
Ilcgnnerator
     0
     2

 Footnote '
 Footnote '

     0.83
     0.83
    B-7
 1.110
   22.3

    0.78
   10
    1.8

    6.6
 1.890
Itcgcncrntor
     0
     2

 Footnote '
 Footnote '

     0.83
     0.83
                  84
 1.110
   10.8

    0.67
    B.O
 1.170
                               I
                               5
                                                             o

                                                             G
21
in

H
G
S

-------
         REMOVAL  OF  PARTICULATE  MATTER FROM
                              GASEOUS WASTES
                        ELECTROSTATIC  PRECIPITATORS
          1.00  INTRODUCTION

  The function of an electrostatic precipitator
is to  remove particles  (solid or liquid)  from
gaseous streams.  This is done by passing the
gas between a pair  of electrodes—a discharge
electrode at a high potential and an electrically
grounded collecting  electrode.  The  potential
difference must be great enough so that a corona
discharge surrounds the discharge electrode^
Under the action of the electrical field, gas ions
formed in the corona move rapidly toward the
collecting electrode  and transfer their charge
to the particles  by  collision with them.  The
electrical field interacting with the charge on
the particles then causes them to drift toward,
and be deposited on, the collecting electrode.
  When the particles are liquid droplets, the col-
lected droplets coalesce on the collecting  elec-
trode  and drip off the bottom of that  electrode
into a collecting sump. When the particles are
solid, the dust layer that forms on the collecting
electrode is removed by intermittent rapping to
cause  the deposit to  break loose from the  elec-
trode.  In effect, this returns the dust to the gas
stream, but not  in  its original finely divided
state.  As a result of cohesive forces developed
between the particles deposited on the electrode,
the dust is returned  as agglomerates which are
large  enough so  that gravity will cause  them
to fall into dust  hoppers  below the electrodes.
Reduced to its essentials,  the electrostatic pre-
cipitator  acts as  a particle agglomerator com-
bined  with a gravity settling chamber.
  The  electrical mechanisms for precipitation
of particles are  supplying:  1,  an electrical
charge to the particles; and, 2, the electrostatic
force that causes the charged particles to drift
toward the collecting electrode.  In the usual
industrial electrostatic precipitators,  both are
supplied  simultaneously  and the precipitator
acts as a single-stage unit.  Typical industrial
precipitators are shown in  Fig.  1  and 2. In

  * Figures refer to REFERENCES on p. 42.
the case of air conditioning applications and a
few industrial applications/"' -: however, a two-
stage precipitator is  used  in  which the two
mechanisms are separated. One set of electrodes
supplies the electrical charge to the particles,
and a second set supplies the electrostatic force
which precipitates the charged particles.
  Another interesting difference that  exists
between industrial and air conditioning  precip-
itators is  the polarity of the  discharge  elec-
Courtesy of Research-Cottrell, Inc.

       FIG. 1—A Plate-Type Precipitator.
SOURCE :   Removal  of Particulate Matter  From Gaseous Wastes—Electrostatic
          Precipitators, Engineering  Report Prepared for American Petroleum
          Institute, New York, NY (1961).

-------
                                AMERICAN PETROLEUM INSTITUTE
trodes. The industrial  precipitator almost in-
variably employs a negative polarity discharge
electrode,  because a higher  voltage drop can
be used before the disruptive effect of sparking
between the electrodes occurs. The higher volt-
age makes it possible to charge the particles to
a higher level and also makes the electrical field
strength greater for precipitation.  A positive
polarity discharge electrode is used in air condi-
tioning,  however,  because less  ozone  is  pro-
duced. Since the amount of ozone which can be
tolerated  in air  conditioning  applications  is
limited, design efficiency is  sacrificed to  some
extent for  reduced ozone production in such
applications.
   The essential components of the electrode sys-
 tem are a fine  "wire" discharge electrode and
 an extended-surface  collecting electrode.  Al-
 though hereinafter referred to as a wire, the
 discharge electrode may take forms other than
 finely  drawn wire (see page 6).  The  usual
 electrode arrangements are:
 1. Wires suspended along the centerline of ver-
 tical, tubular collecting electrodes.  The  tubes
 range from 6 in. ID by 6 ft in length to 1? in
 ID by 15 ft in length.
2. Vertical  wires suspended midway between
parallel-plate collecting electrodes.  The plates
are spaced 6 in. to 12 in. apart and are up to 20
ft in height.
                                               High Voltage Insulator

                                                    Compartment
                      Support Insulator
                      Steam Call
                    High Tension
                   Supper!  Frame	
                     Collecting
                    Electrode Pipes
                        Shell
                     High Tendon
                      Electrode
                     Electrode
                      Weight
                    Courtesy of Western  Precipitation Corporation.

                                FIG. 2—A Tube-Type Precipitator.

-------
                   REMOVAL OP  PARTICULATE MATTER FROM GASEOUS WASTES
   The wires are suspended  from an insulated
 hanger at the top and kept under tension  by
 weights attached to their lower ends. To reduce
 swinging  of  the wires,  groups  of the  wire
 weights are usually tied together by a frame to
 increase the effective mass of the system.
   For the wire-and-tube  electrode system, the
 gas flow path is down around the outside of the
 tubes and then up through the inside of the
 tubes. For the wire-and-plate electrode system,
 the gas flow can also be vertically upward be-
 tween the plates, but is usually horizontal be-
 tween the plates.  Both the collecting electrode
 surfaces and the wires must be vertical where
 gravity collection of liquid droplets or agglom-
 erated dust is employed.  Even if gravity re-
 moval of collected solid or  liquid were not a
 consideration, vertical orientation of the wires
 would be desirable to simplify maintenance of
 wire location and wire tension.
   The electrostatic precipitator is only one of
 many  types   of  particulate-collecting  devices
 available. However, one of the electrostatic pre-
 cipitator's striking characteristics is its  high
 collection efficiency for fine dusts.  In the petro-
 leum industry, collection  efficiencies of  over
 99.5 per cent (based upon the catalyst lost from
 the cyclone separators) are not uncommon for
 the recovery of fluid cracking catalyst.
   Briefly stated, the  precipitator's advantages
 are:
 1. It can handle large volumes of high-tempera-
 ture gases.
 2. It  has an almost  negligibly  small pressure
 drop.
 3. It has a high collection efficiency for a wide
 range of particle sizes and concentrations.
 4. Its operating and maintenance costs are low.
   Its disadvantages are: 1, the large space re-
 quired for its installation;  2,  its  high initial
 cost; and, 3, the possible explosion hazard when
 the gas or particles collected are combustible.
   The usual applications for dust collectors  are:
 1. Recovery of a valuable material such as the
catalyst in a fluid catalytic cracking process.
2. Purification of gases to be used in processes,
such as blast furnace gases to be used for power
or heating.
3. Prevention  of  air pollution by  dust-laden
gases vented to the atmosphere.
4. Cleaning of ventilation air.
   The justification for choosing  a particular
dust collector—e.g., an electrostatic precipitator
—depends upon what  must be removed from
the gas. The precipitator is not justified if the
gas  contains  a  relatively coarse dust, because
a cyclone can do the job at  a lower cost.  If the
gas contains a vapor which has to be removed
as well as dust, a scrubber is probably called
for, because the precipitator will, not remove
the vapor. But, if the dust is fine and must be
removed  with a high collection  efficiency and
low pressure drop, very often the precipitator is
the only collector which  can do the job. That
this situation is common is apparent  from the
large number of precipitators which have been
installed in this country.

       2.00  ELECTRICAL THEORY

2.10  Basic Mechanisms

   Several electrical processes are  important in
the operation of an  electrostatic  precipitator.
These are:
1. lonization of the gas by means of  a corona
discharge.
2. Charging of the dust particles  by means  of
ion bombardment of the particles and ion diffu-
sion to the particles.
3. Drift of the charged  dust particles  to the
collecting  electrode caused  by the  electrical
field existing between  the  electrodes and  re-
sisted  by the fluid drag on the particles.
4. Removal of the electrical charge  from the
dust particles  on the  collecting  electrode by
conduction through the dust layer.
5. Momentum transfer from the ions streaming
from the  discharge  electrode to the collecting
electrode.  This produces a lateral motion  of
the gas stream called the  electric wind. At one
time this  was thought  to be the basis for the
operation of the precipitator, but now it is be-
lieved  that the electric wind does nothing more
than increase the turbulence in the gas stream.

2.20  The Corona

   In order for particles to  be precipitated out
of the gas stream, it is essential  that they be
electrified so that they will be acted  upon by
the electrical  field.  Most particles entering a
precipitator already have a small amount  of
charge previously  accumulated by frictional
electrification and sometimes by flame ioniza-
tion, but this charge is not large enough for
economical operation of an electrostatic precipi-
tator.  Some additional method of electrification
is required.  The method that is used in all com-

-------
                               AMERICAN  PETROLEUM INSTITUTE
mercial operations is electrification by corona
discharge.
  The corona is generated by applying a direct-
current electrical  field between a pair of elec-
trodes, one having a small radius of curvature
(the discharge electrode), and the other having
a large radius of curvature  (the collecting elec-
trode).  As  the field strength  is increased  be-
tween the electrodes, no current passes between
them until the   corona  starting  voltage  is
reached. At this point an electrical  breakdown
(ionization) of the gas occurs, causing a hissing
sound and a  blue glow around the discharge
electrode. As  the  voltage is raised further,  the
current  passing   between  the electrodes  in-
creases  rapidly and the corona glow about  the
discharge electrode (or electrodes) increases in
brightness and volume until sparkover voltage
is reached.  A corona can also be formed by an
alternating-current voltage, but this  does  not
produce a satisfactory field for  particle pre-
cipitation.
  The  distinguishing  characteristic  between
corona  discharge and sparking  discharge is
that the corona discharge corresponds  to an in-
tense electrical breakdown of the gas in a small
space around  the discharge electrode,  whereas,
in sparking discharge,  the gas  breaks  down
electrically in narrowly confined paths extend-
ing from the discharge electrode to  the collect-
ing electrode.
  The field strength,  E,  of the electrical field
between the electrodes is defined as  the deriva-
tive of the voltage, V, with respect to  distance,
r, i.e., the voltage gradient.
  Expressed mathematically, the field strength
is:

                  E = -f

From this it follows that, if the voltage differ-
ence between  the electrodes is  V, then
                       /•Rs
                  --/E
                       •/B,
                        Edr              (2)
                       Ri
Where:

  r, Ru and R. = distances  measured from the
                  centerline  of the discharge
                  electrode,  with  Rt  repre-
                  senting the position  of the
                  outer surface  of  the  dis-
                  charge electrode and R< rep-
                  resenting the position of the
                  inner surface of the collect-
                  ing electrode.

-------
                  FUGITIVE DUST EMISSIONS AND CONTROL
                         George E. Weant, III*
                            B. H. Carpenter
                      Research Triangle Institute
              Research Triangle Park, North Carolina  27709
ABSTRACT

     Many Air Quality Control Regions (AQCR's) do not meet ambient air
quality standards for particulates.  In a majority of these AQCR's
(92 percent), the emissions from fugitive dust sources are higher than
those from nonfugitive sources.  In most cases, unpaved roads are the
largest source of fugitive dust, while agricultural tilling and construc-
tion also contribute substantial amounts.  The reentrainment of particles
from paved roads can also provide large quantities in urban areas.

     Present control techniques for fugitive dust are inadequate.  Esti-
mates of the relative effectiveness of control techniques are presented
in this paper.  If the emissions were controlled to the levels reported
in the literature, 87 percent of the AQCR's would still show fugitive
sources as greater contributors than other sources.
INTRODUCTION

     Many Air Quality Control Regions (AQCR's) do not meet the primary
and/or secondary standards for total suspended particulates (TSP).   This
study provides an estimate of the impact of fugitive dust emissions
(i.e., nonducted emissions) on the TSP in these AQCR's.   In making  this
estimate, the relationships between fugitive dust emissions and emissions
from other sources were examined in each AQCR.  The relationship of
emissions to ambient concentrations is not explored except in a general
fashion by examining published information on this relationship.
*Presently with the Phillips Petroleum Company.


                                   63

-------
     Existing control technologies for fugitive dust sources are examined
for effectiveness when applied to various sources.  Available data on
fractional efficiency of control are presented.
FUGITIVE EMISSIONS ASSESSMENT

Area Sources

     The 150 AQCR's that do not meet the total suspended particulate
standards were identified from a published report.*  Emission source
categories in each were then examined for relative importance.  The
categories examined were those over which man  has some control, and
included fugitive and point sources.  Classified as area sources under
the National Emissions Data System (NEDS), fugitive emissions sources
include dirt roads, landings and take offs from dirt airstrips, agri-
cultural tilling, construction, open burning, slash fires, and coal
refuse fires.  For the first four sources, emissions data were taken
from an updated NEDS card file that used countywide emission factors
and activity levels.  These data are thought to be more accurate than
the data derived from nationwide emissions factors because they are
corrected by local silt content, precipitation/evaporation indexes, and
dry days per year.  Of course, other area sources such as paved roads
should also be considered.  However, data on these types of emissions
could not be obtained on a nationwide basis.

     The data for coal refuse piles were generated using an emission
factor of 10 kg/m  (17 Ib/yd ) and assuming that .5 percent of the pile
burns each year.2  The data for the dirt road emissions were calculated
by multiplying the number of vehicle miles traveled by a local emission
factor.  The vehicle miles traveled were based on a nationwide popula-
tion extrapolation of data from Kansas where an accurate accounting has
been made.  A check of these data with results from St. Louis indicated
that rural data were within approximately 50 percent while urban data
could be as much as two orders of magnitude high.3  In view of this, the
computed values for dirt roads were arbitrarily reduced by one order of
magnitude.

     This analysis provides annual area emissions levels for the seven
source categories.  The results for the individual AQCR's are not pre-
sented here but can be found in the project report for the study.1*'
INDUSTRIAL SOURCES

     Some industrial sources have the potential to contribute signifi-
cant fugitive dust emissions.  A list, of these sources is given
in the project report. "*
                                   61*

-------
     Within these  industries, some of the most important sources of
 fugitive emissions include raw material and product piles, waste .heaps,
 materials handling equipment, comminution equipment, furnaces, and
 dryers.

     A search of the NEDS file provided data on the total emissions,
 number of industrial sources, and the percentage of the sources that
 are controlled in each of the AQCR's that do not meet the TSP standards,
 The data are presented in the project report.1*
PAVED ROADS

     Paved roads have been shown only recently to be significant
contributors to particulate emissions.  A study conducted in Seattle1s
Duwamish Valley compared dust emissions from paved and unpaved roads
to emissions from industry and other sources.   The results follow:
Source

Heavy industry
Vessels, trains, auto
 • tailpipes, home heating
Gravel roads (19 mi)
Paved roads (110 mi)
Emissions
(tons/yr)

  1200
  1000

  2100
   600
Percent of
Total Emissions

     24.5
     20.4

     42.9
     12.2
     Other data presented in this study showed that a car driven 12 ka
(7.5  mi) at 16 kia/h (10 mph) on a wet gravel road picked up 36 kg
(80 Ib) of mud.  Average pickup from dirt parking lots was 0.34 kg
(0.74 Ib) per vehicle.   Thus, mud pickup from unpaved roads and parking
lots can contribute significantly to the amount of material that is de-
posited on paved roads.

     Another study examined the contribution of various sources includ-
ing  paved roads to the total particulate emissions in three counties
in North Carolina.   The results are shown in Table 1.

     Tire wear debris has been examined as a source of airborne particu-
late.    Pierson and Brachaczek estimated that this debris was a rela-
tively minor source, accounting for 2 to 3 percent of the suspended par-
ticulate associated with vehicles.
                                   65

-------
TABLE 1 . COMPARISON OF THE CONTRIBUTION OF SOURCE CATEGORIES TO TOTAL
         EMISSIONS IN THREE COUNTIES -OF NORTH CAROLINA FOR 1973

                                 EMISSIONS BY SOURCE CATEGORY
                                 (% OF TOTAL EMISSIONS)

                         Mecklenburg       Forsyth       Guilford
Source                     County          County         County

Point                        74.0            8.7            9.6
Unpaved Roads                14.9           77.8           76.6
Paved Roads                   5.6            7.2            6.8
Other Area Sources            5.5            6.3            7.0
THE MAGNITUDE OF THE EMISSION PROBLEM

     To estimate the magnitude of the emission problem associated with
fugitive dust, the fugitive emissions were compared to the total point
source emissions for each of the AQCR's that did not meet either the
primary or the secondary TSP standards.  The results are shown below.
                                   Number of AQCR's      % of Total

Point Source >  Area  Source                 9                6.7
Area > Point                             139               92.0
    Area 5 times greater than point          97                 64.7
    Area 10 times greater than point         58                 38.7
Data Missing                               2                1.3
                    Totals               150              100.0
     As a further illustration of the magnitude of these area source
emissions, examine the effect of cutting them in half.   Even with
this, area source emissions in 129,  or 86 percent of the AQCR's are
still greater than the total point source emissions. With cutting them
to 0.1, 57, or 38 percent are still  greater.
IMPACT OF EMISSIONS ON TSP

     The evaluation of the impact of emissions on air quality requires
a complicated, site-specific calculation that is beyond the scope of
this report.   Therefore,  this report must concentrate on published re-
sults to evaluate fugitive dust imoacts.
                                   66

-------
      In a recent paper describing the impact of fugitive emissions on
TSP,  it was stated that in a large industrial city where the TS? loading
was influenced by fugitive emissions, the TSP on an annual basis averaged
25 ug/m  higher than industrial areas not influenced by fugitive emission
sources.8  This paper also stated that the results from 20 sites in five
heavily industrialized cities indicated that fugitive emissions increased
TSP by 10 ug/a  .

      In discussions with various EPA personnel, it was brought out that
fugitive dust emissions from dirt roads have a relatively minor effect
on TSP.  This statement was based on the assumption that the particle
size  of the dust from dirt roads is such that most particles fall out
of the air within short distances from the dirt roads and that most
dirt  roads are located in rural areas away from the sampling stations.
However, the study performed in Seattle showed that 27 percent of the
dust  from vehicles traveling over dirt roads at 32 kni/h (20 mph) was
suspendable (less than 10 urn in size), while 41 percent was suspendable
at 48 km/h (30 mph).5

      Further evidence of the substantial impact of fugitive dust emis-
sions comes from Massachusetts.  An item appearing in a weekly publi-
cation reported that the air of southeastern Massachusetts had been
declared a hazard to public health and that 80 percent of the particu-
lates came from windblown sand and road dirt.^   A followup discussion
with Massachusetts officials indicated that the episodes occurred during
the winter months and were the result of the reentrainment of sand that
was used for vehicle skid control.10

     A study of air quality maintenance areas in North Carolina found
that  the emissions inventories for particulate matter in several coun-
ties did not provide enough emissions to account for the ambient air
quality measurements obtain in urban areas.5  The study concluded that
paved roads contributed the substantial amount of emissions necessary
to make up the difference in TSP observations in urban sections.  To
account for this difference,  the emission factor that was based on the
Seattle study5 was raised by a factor of 2.3 for Mecklenburg County
where vacuum street cleaning is used and by 3.5 for Forsyth and
Guilford Counties where no vacuum street cleaning is used.   As a result,
an acceptable calibration of the Air Quality Display Model  (AQDM)  was
achieved.

     The data on both emission quantities and impact of emissions  on
TSP imply a strong relationship between fugitive dust emissions and
nonattainment in many AQCR's.   It must be emphasized that the precision
and reliability of the data base used for this analysis are unknown,
basically because of the inconsistencies in sampling,  testing,  and
recording methodologies used  throughout the network.   However,  ifEDS is
the only data base available  for a study of this type.   Therefore,  the
approach has  been to rely on  a comparison of relative magnitudes and
not on an exact quantification of each piece of data.
                                   67

-------
FUGITIVE DUST CONTROL TECHNIQUES

     Currently control of fugitive dusts from.area sources is predomi-
nantly by prevention rather than capture and separation.   Measures are
adopted to prevent the dust from becoming, airborne.  There are four
preventive technologies:  wetting, physical stabilization, chemical
stabilization, and vegetative stabilization.  Industrial  operations
such as mining and beneficiation, included in this study, tend to make
moderate-use of capture technologies. .

     Control technologies were examined to estimate their relative ef-
fectiveness and to comment on their limitations.  Their applications
were considered with respect to eight source categories:   agriculture,
transportation, materials handling, stockpiles and waste.heaps, mining
operations, beneficiation, construction, and miscellaneous sources
(e.g. open burning, incineration, cooling-tower drift).
 Wetting

     Wet suppression of dust using either water or water plus a wetting
 agent can be employed for temporary control of fugitive dust from some
 agricultural activity, cattle feedlots, unpaved roads, transport of raw
 materials or products, materials handling and beneficiation, stockpiles,
 waste heaps, and mining and construction activities.  The"temporary
 nature of "wet  suppression restricts its'usefulness.  In'cases when there
 is  continual activity at the source, the suppressive must be repeatedly
 applied  to be  useful.  This is due to the continual exposure of dry
 surfaces to climatic elements and is applicable to agricultural activity,
 unpaved  roads,  and  stockpiles.

     Water has proven to be a poor suppressive due to its high surface
 tension.  The  high  surface tension interferes with the wetting, spread-
 ing, and penetrating necessary for. effective suppression.

     Surface tension can be reduced by  the  addition of wettir^agents.
 These  agents increase the effectiveness of  wet suppression  by:

     1.  allowing particles to penetrate  the water droplet,  and thus
         exposing a larger water surface;

     2.  agglomerating  particles in  the droplet;

     3.   increasing the number of droplets  per unit volume,  the
          surface area,  and  the contact  potential  through increased
          efficiency of  atomization;  and

      4.   causing the liquid to wet faster and  deeper  and spread  farther.
                                   68

-------
     In. addition to being a temporary control measure,  wet dust suppres-
sion cannot be used where either the product or the next stage of pro-
cessing will not tolerate a wet product.  Examples of these instances
include grain processing and certain beneficiation processes that require
dry classification.  Drying steps can be taken but present additional
environmental problems as well as added costs.

     The wet suppression of dust is usually accomplished by spraying the
water either with or without a surfactant onto the surface of the exposed
material.  For many mining and construction roads and other surfaces,
this is usually done by a special truck equipped with a tank for the
liquid and a series of spray nozzles in the front and back.  For the
transport of products and raw materials, the carrier vehicle is usually
passed under a series of spray bars where the liquid is dispersed onto
the surface of the material.  For materials handling and beneficiation
sources, nozzles located at transfer points and at equipment intakes
spray the liquid on the material.  For stockpiles, nozzles spray the
liquid onto either the pile or the material as it is being transferred.
For feedlots, a spray system is also appropriate.

     The application of wet dust suppression to many fugitive dust
sources is not feasible.  These sources include some agricultural activ-
ity,  unpaved roads, and waste heaps.  Reasons for the infeasibility
include the potential shortages of water, magnitude of source, lack
of suitable equipment for transporting and dispersing water, and the
temporary nature of the control method.

     In recent years, a new wet dust suppression system has been intro-
duced.  The use of foam systems has become an important dust suppression
method.  Foam systems have been successfully applied to both hard rock
drilling operations and transfer points of conveyors.1-'12  These sys-
tems have advantages over untreated water in that they increase the
wettability, thus requiring a smaller supply of wetting fluid; and in
the case of drilling operations, they prevent overinjection of water
into the hole which in turn can cause collaring of the bit and decreased
penetration rates.

     Data on the control efficiency of wet dust suppression is minimal.
One reference cites as much as 80 percent control for cattle feedlots,
but this is very much dependent on soil conditions, local climatic
conditions, number of cattle, activity level, and many other things.1'4
This same reference reported efficiencies of 30 to 67 percent for highly
disturbed to nondisturbed storage piles, and efficiencies of 0 to 70
percent for construction site watering. 3

     Observations of several North Carolina granite quarries have shown
substantially reduced emissions from processing plants, haulage roads,
and drilling rigs using wet dust suppression with surfactants.11*

-------
Control efficiencies from drilling storage piles and construction sites
will depend on many factors, including type of material and percentage
of fines, local climatic conditions, type of equipment being used, mois-
ture, and activity rate.

     A recent study has examined the use of water sprays and foaia on
materials handling processes.    At a coal chain feeder-to-conveyor trans-
fer point with a 3-fc material drop, water controlled 70 percent of the
emissions while the foam spray controlled 91 percent.  These numbers rep-
resent  control with a spray under the belt as well as at the transfer
point.  Under-the-belt sprays were shown by this report to be effective
in controlling dust at conveyor transfer points when used in conjunction
with transfer point sprays.
Physical Stabilization

     Physical stabilization methods can be used for controlling fugitive
dust from inactive waste heaps, unpaved roads, and other sites.  Physical
stabilization requires the' covering of the exposed surface with a
material that prevents the wind from disturbing the surface particles.

     Common physical stabilizer materials for inactive waste heaps and
steep slopes include rock, soil, crushed or granulated slag, bark, wood
chips, and straw that are harrowed into the top few inches of the
material.    For dirt roads, paving is a common practice.  However,
paving is expensive and, in most cases, must be preceded by roadbed
buildup and improvement to prevent overdriving by vehicle operators.
Other methods of physical stabilization of these sources include covering
with elastomeric films, asphalt, wax, tar, oil, pitch, and other cover
materials.

     Very little information is available on the effectiveness of physi-
cal 'control methods.  One reference cites an 85-oercent control effi-
ciency with paving and right-of-way improvement on dirt roads.^  This
control efficiency is dependent on how inucn dirt is brought onto the
road and later reentrained by passing vehicles.
Chemical Stabilization

     Chemical stabilization requires the use of binding materials that,
upon drying, bind with surface particles to form a protective crust.  It
acts in much the same way as physical controls by isolating the surface
from climatic factors and is often used in combination with vegetative
stabilization.  Applications of chemical stabilization are found on
agricultural fields, unpaved roads, waste heaps, and excavation heaps.

     Evaluations of the suitability of various chemical stabilizing
materials have  been reported  in  the literature.   In one study evaluating
the cost and  effectiveness of  34  stabilizers,15  the evaluation  criteria
                                   70

-------
were cost, prevention of wind erosion, effect on plant germination and
 ;rowth of tomatoes and beans, and ease of application.  Those stabilizers
 :hat proved effective for reducing wind erosion from the piles for 180
days are ranked in Table 2.

    TABLE 2.  MATERIALS THAT REDUCED SOIL LOSS FOR 180 DAYS RANKED
              BY 1971 COST
                                 Nonerosion
                                 Rate           1971     Ranked
Product        Manufacturer      (per acre)    Cost ($)  Effectiveness

Elvanol 50-42  E. I. du Pont        13 Ib        8.20        6
Technical Pro-
tein Colloid
5-V            Swift & Co.         108 Ib       34.60        5
Geon 652       Goodrich Chemical    17 gal      51.20        8
Aquatain       Larutan Corp.        68 gal     172.50        7
ORTHO Soil
Mulch          Chevron Chemical    681 gal     242.20        1
Anionic Asphalt
Emulsion       Phillips Petro.    1226 gal     436.70        3
AGRI-MULCH     Douglas Oil         954 gal     445.70        4
Soil Erosion   Swift & Co.         571 gal    1159.90        2
Control Resin
Adhesive Z-3876
     A later report presented the results of the Bureau of Mines tests
on 70 different chemicals. *  Water and wind erosion tests were performed
in the laboratory on applications of these chemicals to various types of
mill tailings.  The more effective chemicals of those tested are listed
below in order of their relative effectiveness based upon the cost re-
quired to stabilize 1 yd~.12  Long-term effectiveness to wind erosion
was not measured.

     1. COHEREX - good wind resistance at coverage of 240 gal/acre
        at cost of ?65/acre, good water-jet resistance at cost of
        $650/acre.

     2.  Calcium, sodium, ammonium lignosulfonates - effective stabili-
         zers at coverage of 2400 Ib/acre at cost of §130 to $170/acre.

     3.  Compound SP-400, Soil Card, and DCA-70 - wind and water resis-
         tant surfaces at coverage of 55, 90, and 50 gal/acre, respec-
         tively.  Cost of about $130/acre.

     4.  Cement and milk of liine - effective stabilization at costs of
         about $190/acre.
                                    71

-------
     5.   Paracol  TC  1842  -  effective  stabilizer at cost of about $250/
         acre.

     6.   Pamak WTP - effective  at  cost of  $250/acre.

     7.   Petroset SB-1  -  effective at cost of  $250/acre.

     8.   Potassium silicate (Si02  to  K 0 ratio of 2.5) -  effective  at
         $450 to  §950/acre.

     9.   PB-4601  - effective at $500/acre.

    10.   Cationic neoprene  emulsion and Rezosol - effective  at  $500/
         acre.

    11.   Dresinol TC 1843 - effective at $500/acre.

    12.   Sodium silicates (Si09 to Na 0 ratios of 2.4 to  2.9 to 1)  -
         effective at about $200/acre, with calcium  chloride additive,
         amount of sodium silicate was reduced.

     One reference has estimated control efficiencies of  chemical  stabi-
lization on a number of sources.13  Examples of  these estimates are as
follows:

               Source                            Efficiency (%)

     Unpaved roads                                     50
     Construction -  completed cuts and  fills           80
     Agricultural fields                               40
     Tailings piles                                     80
     Continuous spray of aggregate as it  is piled      90
     Cattle feedlots                                   40

     The effectiveness of chemical stabilization of unpaved roads  would
seem to be extremely variable based on the amount of traffic.  Heavy
traffic would tend  to break up the surface crust,  pulverize particles,
and eject them into the atmosphere in much the same manner as if the
road were untreated.  Likewise, with cattle feedlots, the effectiveness
would seem to be heavily dependent on the activity'in the feedlot.  It
would seem that the  effectiveness  of  continuous spraying  of  aggregate as
it is piled could be highly variable  depending on such things -as the
quantity of fines in the mix, type of stone,  etc.   In addition, the
activity level of the storage pile is also important.
 Vegetative Stabilization

     Vegetation can be effectively used to stabilize a variety of ex-
 posed  surfaces.  In many cases, modifications must be made to the
                                  72

-------
surface or the surrounding terrain before effective stabilization can
occur (e.g., fertilization, pH modification, and slope reduction).

     Vegetative stablization for the control of fugitive dust is re.
stricted to inactive areas where the vegetation will-'not be iaechnically
disturbed once it is started.  These sources can include refuse piles
(coal and mineral) and road shoulders.

Coal Refuse Piles—

     Coal mining and preparation usually produce both fine and coarse
waste materials.  These materials consist of low grade coal, ash, car-
bonaceous and pyritic shale, slate, -clay, and sandstone.17

     The principal problems encountered in the vegetative stabilization
of coal refuse piles occur as a result of the acidic nature of the
wastes and  from the slopes of the piles' sides.  Thus, chemical or
physical treatment of the piles' components must be accomplished prior
to effective stabilization.  Chemical treatment usually involves the
addition of a soil neutralizing material such as agricultural limestone.
Other materials such as fly ash, mines phosphate rock, and treated
municipal sewage sludge have also been used.17-  Even with a neutraliza-
tion pretreattaent, it is  recommended  that acid-tolerant vegetative
species be  used for stabilization because the sulfide materials in the
waste can oxidize the acid sulfates and  thus lower the pH of the soils.

     Physical treatment of the piles  usually involves such things as
the burying of high pyritic materials, covering the piles  with a
layer of  topsoil, or grading to reduce slopes of the piles. 17  A good
premining restoration plan can be  effective for efficient physical
treatment methods.

     Many species of plants have been used  for  the stabilization of
coal mine refuse  piles:   grasses,  legumes,  trees,  shrubs, and vines.
For a  detailed discussion of these  plants and  their uses, refer to
Reference I7.

Mineral Refuse Piles—

     Mineral  mining and beneficiation produce  wastes  in  the  form of
overburden, gangue, and  tailings.   Overburden  and  gangue do  not  usually
present problems  to vegetative  stabilization.   However,  tailings  can
 present varied  and  extreme problems due  to  a deficiency  of  nutrients,
 saline or toxic  properties,  and variable pH.

     Most tailings  stabilization  is accomplished by  first covering  the
 waste with a layer  of  topsoil  and then  by establishing  a vegetative
 cover.  Without the topsoil cover, vegetation  usually requires  the
 assistance of other wind erosion  preventatives such as  mulches^ chemi-
 cal coatings, rapidly established plant covers, and watering.  •   How-
 ever,  even with these aids,  stabilization of  many  mineral wastes has
                                   73

-------
 not been effective.  Most species are very site-specific,  and  small
 changes in topography, climate, and tailings composition affect their
 growth success.

 Copper Tailings—

      The establishment of vegetation on copper tailings  is very site-
 specific.   Even with piles in the same general geographic  area,  it  is
 often difficult to.establish the same type of vegetation.

      In the western United States,  copper  tailings  have  been stabilized
 with vegetation.   In most cases,  maintenance in the form of liming,  fer-
 tilizing,  and irrigating after planting is required.   However,  at Magma,
 Utah, a form of permanent vegetative stabilization  seems to have been
 established with natural vegetation invading the pile.17

 Uranium Tailings—

      Uranium tailings  in Colorado have been stabilized using sweet  brome,
 sweetclover,  cereal rye,  barley,  alfalfa,  and various  wheat grasses.17
 There has  been very little invasion by natural species,  and continual
 maintenance is required.

 Iron Tailings—

      The vegetative stabilization of  iron  tailings  in  Pennsylvania and
 Minnesota  has  been relatively  successful.   Initial  stabilization with
 grasses and legumes  followed by the planting  of  woody  plants seems  to
 have been  successful.     Invasion by  native vegetation heightens the
 prospect of a  permanent,  maintenance-free  stabilization  site.

 Other Metallic  Tailings—

      In most cases, plants  tolerant  to  specific  conditions must be
 applied to  metallic tailings piles.   Some  success has been demonstrated
 with varieties  of  grasses on gold mining slimes  and sands; some species
 of grasses  have been found  to be  tolerant  to lead and zinc; but little
 long-term  success  has been demonstrated with rye on molybdenum tail-
 ings .!7

 Control Efficiency—

     The control efficiency of vegetative stabilization should  vary
 considerably with differences in the amount a'nd type of cover established
 on the tailings piles.  One report estimates a control efficiency of  from
 50 to 80 percent.    This estimate was made using the wind erosion
 equation and is not based on a measured efficiency.   This same  report
 estimates a 93-percent reduction in windblown emissions with a  combined
chemical/vegetative stabilization program.

-------
     It would seem reasonable to assume that these control efficiencies
could be achieved.   In fact,  efficiencies of 100 percent should be ap-
proached with complete vegetative covering on some sources.
OTHER CONTROL METHODS

     Numerous other control methods are available for various sources of
fugitive emissions.  Some of the most important include speed reduction
on unpaved roads, street cleaning of paved roads, reduction of fall dis-
tances for materials handling, and enclosure, hooding, and ducting.

Speed Reduction

     Reducing the speed of vehicles traveling over unpaved roads has
been shown to reduce the dust emissions from such travel.  A reduction
in vehicle speed reduces both the pulverization of road material and the
turbulent wake of the vehicle.  A well-quoted source has shown the
following results from vehicle travel at various speeds on dirt roads.
(Table3).u
           TABLE 3 .  EFFECT OF SPEED REDUCTION ON EMISSIONS

  Average Vehicle       Dust Emissions       Emissions Compared to
  Speed  (mph)           (Ib/vehicle mile)    Those at 40 mph (%)

     40                       2.50                100.0
     35                       1.47                 58.8
     25                       0.70                 28.0
     15                       0.48                 19.2
     In another report, the results of a study in Seattle's Duwamish
Valley have shown comparatively higher emissions.   In addition, this
study showed significant reductions in the quantities of suspendable
particulates with speed reduction.  The results are shown in Table 4.
                                   75

-------
 TABLE 4.  EFFECT OF SPEED REDUCTION ON EMISSIONS .IN SEATTLE'S DUWAMISH
           VALLEY
                                                   Total      Suspendable
                                                   Emissions  Emissions
                                                   Compared   Compared
                                 Suspendable       to Those   To Those
Vehicle       Total Emissions    Emissions         at 30 mph  at 30 mph
Speed  (mph)   (Ib/vehicle mile)   (Ib/vehicle mile)
   30                22.2                9            100.0     100.0
   20                 7.0                2             31.5      22.2
   10                 3.5                0.5           15.8       0.1
Street Cleaning

     With the recent interest on reentrained dust from paved roads as a
source of air pollution, attention has been focused on street cleaners
as dust control devices.  Essentially three types of cleaners are now
in use:  broom sweepers, flushers, and vacuum and regenerative air
sweepers.  Their effectiveness has not been overwhelmingly demonstrated.
Streetside samples have shown concentration reductions but regional
samplers have shown no reductions.

     Broom sweeping has been shown to reduce the average concentration
of dust in one study but has been shown to be ineffective in two others. 18
It has been estimated that this type of sweeper picks up 20 percent of the
material below 140 jin. 19  Also, while recovering this paltry amount of
material, the sweeper can actually generate air pollution by stirring
up the dust and by moving the material from the curbs into the middle
of the road where it can be reentrained by passing vehicles.

     Flushing showed significant particulate reduction in two studies
and no effect in two other studies.   In a fifth study, flushing showed
no reduction in the average monthlv concentration but did show reduction
on days when flushing took place.18

     Vacuum and regenerative air sweepers have been shown to be ineffec-
tive.13

     Two studies on mud carryover control showed substantial reductions
in particulate concentrations.18  These studies involved manual cleaning
at a construction site egress and strip paving and oiling of unpaved
parking lots, roads, and shoulders on an areawide basis.

Reduction of Fall Distances

     During the transfer of dusty materials from a conveyor or stacker
to another location such as another conveyor or a stockpile, the


                                    76

-------
separation of the fine taaterials from the large materials can be caused
by wind and/or the falling action of the material.   A simple method to
reduce dusting from these operations is to reduce the fall distances by
using hinged-boom conveyors, rock ladders, telescoping chutes, lowering
wells, or other devices.    The hinged-boom conveyor can raise or lower
the conveyor belt  and  thus  reduce the fall distance at the transfer
point.  Rock ladders allow the material to fall small distances in a
step-like fashion.  By reversing the direction of travel on successive
steps, the momentum that the material receives from the previous fall
and the dusting are reduced.

     Telescoping chutes carry the material from the discharge point to
the receiving point.  Thus, the material is not exposed.  Lowering wells,
or perforated pipes, allow material to flow out of  the pipe above the pile
surface.  The dusting from the impact of the falling material is retained
inside the pipe, and the material is protected from wind action.

Enclosure

     Simple enclosure of a fugitive dust source is  an effective control
method in some cases.  It has been applied to a number of sources includ-
ing storage of products, loading and unloading operations, product
bagging operations, and classification operations.   In process operations,
periodic cleaning is necessary and may preclude application.

     The enclosure of sources without providing adequate exhaust is not
applicable to sources where abrasive materials are  handled.  This is
especially true in hard rock processing plants where a high quartz cqn-
tent of the rock abrades the equipment.  Enclosure  is also not applicable
to sources whose dust would present the danger of explosion such as in
many grain handling operations.

Exhaust Systems

     Many process sources of fugitive dust emissions can be controlled
by the use of exhaust systems in combination with full or partial enclo-
sure or full- or partial-coverage hoods and the associated ducting.
Examples of sources amenable to this type of control include materials
handling (i.e., conveyors, elevators, feeders, loading and-unloading,
product bagging, and stockpiling), solids beneficiation (i.e., crushing,
screening, and other classifying), mining operations (i.e., drilling),
and others (i.e., furnaces and dryers).

     Complete enclosure of conveyors, elevators,, or feeders has been
practiced.  Another alternative is to enclose the transfer points.  Hoods
as well as enclosures can'be used on many loading and bagging operations.
     For solid beneficiation processes, both enclosures and hoods are
used.  For drilling operations, enclosure of the drill hole and ducting
to a baghouse mounted on the drill rig is used.
                                    77

-------
      Effectiveness of control is highly variable and dependent on many
 variables.  Efficiencies of 90 percent and greater are considered appro-
 priate.  For example, 90 percent efficiency is attainable on the enclo-
 sure of BOF furnaces.20

      No attempt has been made to provide detailed descriptions of ven-
 tilation practices.  However, several excellent references are available
 on this subject (see References 21 and 22).


 EFFECTIVENESS OF CONTROLS

      Based on the foregoing discussions and comparisons,  the relative
 effectiveness of the five types of control technologies as applied in
 the eight source categories is summarized in Table 5.   The comparative
 effectiveness is indicated by V? (very poor,  less than 20 percent effi-
 ciency);  P (poor,  about JO percent efficiency);  F (fair,  about  50 percent
 efficiency);  G (good,  perhaps up to 85 percent  efficiency).   These
 efficiencies  may be substantially less than indicated,  depending  upon
 the circumstances  of application.   Additional  symbols  used are:   NR (not
 rated,  usually because no application could be  cited);  VAR (variable)-
 UPR (unpaved  roads);  PR (paved roads).   More extensive  tables  showing
 ratings by individual  source  within each category are  given  in  the prol-
 ect report.4                                                        *   J


 TABLE 5.   RELATIVE  EFFECTIVENESS  OF'FUGITIVE DUST CONTROL TECHNOLOGIES
           BY  SOURCE                                                "   '
                                       Source Category

                                              Cfl
                                              e                 L,
                          c
                          o
                      CJ   4J
                      t   ca
                      s   u
                      "   i-
                     i-4   O
                      3   e.
 «                     a   en
 Control              -H   c
                      tt   _f .^
 Technology           ep   >-; c- ot
    "                 "^  E~* £3 C*

 Physical
 stabilization        NR    G  F   NR    F,G   NR   NR F,G
Wetting, wet
application          F    VP  P   FG    P,F  VP. P,F  VP,P
Chemical stabili-
zation               P    PFNR    P    NRNRNR

Vegetative stabi-
lization            VP      G  NR  FG    NR  NR  F,G




terials
3




eo
G
•H
r-(
•o
3




03
0)
p-l
—1
C,
^
O
o
u
CO




CO
c.
ra
_«
*••
u
u
CO
Q
3
O
•H
U
IS
4J
C.
0
cc
•H
e
r-l


C
o
neficiati

-------
TABLE 5 .  (Continued)

                                        Source Category
                                             cn
0)

3
4J
r-l
3
CJ
•H
\4
CO

C
o
•H
4J
(0
4J
J-
O
c.
CO
c
a ci
i^ p- ei
H =3 CU


ca
iH
ca
-H
u
(U
jj
eg




00
G •
•H
•H

G

~

cn
0)

•H
o.

CJ
o
u
cn
cn
CL
d
o
J2

o
u
cn
C3
5
ration:
0)
o.
o

CO
c
•H
C
•H
JE
0
•H
u
cd
TH
U
•H
IU4
(U
G
cu
C3
C
o
•H
ij
CJ
3
W '
u
cn
C
0
u
cn
3
O
CU
G
ca
•H
r*^
u
CJ
CO
•rl
2Z,
60 G
G O
•H -H
C 4J
M a
3 >->
& (U
C
G TH
CJ U
a. c
0 -H



tower
to
c
•H
rH
O
O
O



 Control
 Technology

 Special
   control device                                            **•*
   speed  reduction       Var
   enclosure                      F
   exhausting                   F,G         G  F,G
   reduction of fall            F-P
   wind screen                        VP
      F = fair     G = good        P  =  poor     VP  =  very  poor
       - 50%        " 85%         * 20%           < 20%

      UPR = unpaved roads    NR = not  rated      Var  = variable
      PR = caved roads
Conclusions

     While fair control of some agriculture emissions may be obtained by
wetting, this source lacks good controls.  Physical stabilization remains
practically untried, chemical stabilization is rated poor, and vegetative
stabilization very poor.

     Physical stabilization of dirt roads (chat is, paving) is the only
good control.  Paved roads benefit from vegetative stabilization of
shoulders.  Materials handling operations can attain fair to good con-
trol by wetting  when the techniques can be used.  Other technologies
remain unexploited at this time.

     Stockpiles benefit from physical and vegetative stabilization.

     Mining operations benefit from exhausting techniques, while wetting
is very poor and the stabilization technologies have not been rated, and
may not work well.
                                   79

-------
     Exhausting has been fair to good for control of beneficiation emis-
sions, while wetting has been poor to fair,  and stabilization unrated.

     Physical and vegetative stabilization  is  fair to good in construc-
tion operations, while wetting is poor, and chemical stabilization is
not rated.

     None of these technologies are effective in open burning or on
industrial cooling towers.  Incinerators, if properly designed, can be
effective as control devices.
EFFECT OF FUGITIVE EMISSION REDUCTION ON AQCR'S

     To examine the effects of fugitive dust emissions reduction on total
AQCR emissions, emissions from unpaved roads, agricultural tilling, and
construction were reduced by appropriate measures reported in the litera-
ture.  The reductions used were 50 percent for unpaved roads  and 40
percent for agricultural tilling (chemical stabilization effectiveness),
and 30 percent for construction (wetting).

     The results of the emissions reduction are shown in the following
summary:
                                   Before Emissions
                                     Reduction
After Emissions
  Reduction
Total number of AQCR's not
  meeting TSP Standards                 150                150

Point > Area                              9                 17

Area > Point                            139                131

   Area 5x  > Point                         97                 68
   Area lOx > Point                        58                 38

Data Missing                              2                  2


CONCLUSIONS

     1.  Fugitive dust  sources  are  significant  emitters of particulates
         in a majority  of  the AQCR's.   Of the 150 AQCR's  that do not meet
         the TSP standards,  fugitive  dust emissions  exceed point source
         emissions  in 139  AQCR's, or  92 percent.  In  fact, fugitive emis-
         sions are  10 times  greater than point  source emissions in 58,
         or 39 percent,  of the  AQCR's.
                                   80

-------
      2.   In most cases, unpaved roads provide the largest source of
          particulate emissions in the AQCR's.  Agricultural tilling and
          construction sources are also very important and in some cases
          are the largest emitters.

      3.   The reentrainiaent of particles from paved roads is a source of
          large quantities of particulates in many AQCR's.

      4.   Industrial sources of fugitive emissions are plentiful and can
          have a substantial impact on surrounding areas.

      5.   Fugitive dust sources can contribute significantly to the TSP
          burden of an entire AQCR as well as have   an impact in a
          localized area:

      6.   The relationship between pollutant exposure and human health
          has been demonstrated.  Increased hospitalization rates have
          been observed with increased particulate pollutant exposure.

      7.  More attention should be given to the control of fugitive dust
          emissions because of their contribution to ambient dust
          loadings.

     8.  .Control effectiveness for fugitive sources is highly variable
         and depends on such things as type of control,  characteristics'
         of the source,  local climatic conditions,  and source activity.

     9.  Present control technology for unpaved  roads, agricultural
         tilling,  and construction activity is inadequate.   Reducing
         the emissions from these activities by  the amounts  reported in
         the literature  has only a small influence  on fugitive emissions
         in most AQCR's.
REFERENCES
     1.  Office of Air and Waste Management,  State Air poilution Imple_
         mentation Plan Progress Report,  Jaunary 1 to  June 30,  1976  "
         Office of Air Quality Planning and Standards, U.S.  EPA
         EPA-450/2-76-026, October 1976.

     2.  Personal  communication,  Mr.  Chuck Mann,  NADB, EPA,  Durham
         May 9,  1977.                                              '

     3.  Ibid.,  March  30,  1977.

     4.  Carpenter,  B.  H.,  and  G.  E.  Weant, III,  "Particulate Control
         for Fugitive  Dust," EPA  600/7-78-071, April 1978.
                                  81

-------
5.  Roberts, J. W., H. A. Watters, C. A. Mangold, and A. T. Rossano,
    "Cost and Benefits of Road Dust Control in Seattle's Industrial
    Valley." J. APCA. 25(9), September 1975, pp. 948-952.

6.  Haws, R. C. and H. L. Hamilton, Jr., "North Carolina Air Quality
    Maintenance Area Analysis, Vol. Ill:  TSP Dispersion Modeling
    and Analysis for Charlotte, Winston-Salem, and Greensboro AQMA's
    for 1973, 1975, 1980, 1985," RTI Final Report, EPA Contract
    68-02-1385, Task 15, April 1976.

7.  Pierson, W. R. and W. W. Brachaczek, "Note on In-Traffic
    Measurement of Airborne Tire-Wear Particulate Debris," J. APCA,
    25(4), April 1975.

8.  McCutchen, G., "Regulatory Aspects of Fugitive Emissions," paper
    in Symposium on Fugitive Emissions Measurement and Control,
    May 1976, Hartford, CT, EPA 600/2-76-246, September 1976.

9.  Air/Water Pollution Report. Business Publishers, Inc., Silver
  .  Spring, MD, May 2, 1977, p. 177.

10. Personal communication, Mr. Steve Dennis, Massachusetts Depart-
    ment of Environmental Quality Engineering, Boston, May 27, 1977.

11, Metzger, C. L., "Dust Duppression and Drilling with Foaming
    Agents," in Pit and Quarry Magazine. March 1976, pp. 132-133
    and 138.

12. Seibel, R. J., "Dust Control at a Transfer Point Using Foam
    and Water Sprays," U.S. Department of the Interior, Bureau of
    Mines, TPR 97, May 1976.

13. Jutze, G. and K.  Axetell,  "Investigation of Fugitive Dust,
    Vol.  1:  Sources,  Emissions,  and Control," EPA 450/3-74-036a,
    June 1974.

14.  Weant, G. E., Ill,  "Characterization of Particulate Emissions
     for the Stone-Processing  Industry," RTI Final Report,  Contract
     No.  68-02-02607,  Task 10,  U.S. EPA, Industrial Studies Branch,
     May 1975.

15.  Armbrust, D.  V.  and J. D.  Dickerson, "Temporary Wind Erosion
     Control:  Cost and Effectiveness of 34 Commercial Materials,"
     J.  of Soil and Water Conservation. .July 1971,  pp.  154-157.

16.  Dean, K. C.,  R.  Havens, and  M. W.  Glantz, "Methods and Costs
     for Stabilizing Fine-Mineral Wastes," U.S.  Department of the
     Interior, Bureau of Mines, RI 7894,  1974.
                             82

-------
17.  Donovan, R. P., R. M.  Felder,  and H.  H.  Rogers,  Vegetative
     Stabilization of Mineral Waste Heaps," EPA 600/2-76-087,
     April 1976.

18.  Axetell, K. and J. Zell, "Control of  Re-entrainment Dust  from
     Paved Streets," EPA 907/9-77-007, August 1977.

19.  Sartor, J. D., B. Boyd,  and W. H. VanHorn,  "How  Effective is
     Your Street Sweeping," APWA Reporter. 39(4),  1972,  p.  18.

20.  Nichols, A. G., "Fugitive Emission Control in the Steel
     Industry," Iron and Steel Engineer, July 1976, pp.  25-30.

21.  Committee on Industrial  Ventilation,  Industrial  Ventilation,
     A Manual of Recommended  Practice. 14th ed.,  2nd  Printing,
     American Conference of Governmental Industrial Hygienists,
     Lansing, Michigan, 1977.

22.  Environmental Control  Division,  Control  of Internal Foundry
     Environment, Vol. 1, American  Foundrymen's Society, Des
     Plaines, IL.
                             83

-------
                                                 EPA BUBBLE



                               STEEL  INDUSTRY EXAMPLE OF BEST USE OF BUBBLE

Iron Oxide Sources
Storage- Pile
Wind Loss
Blast-Furnace
Cast House
BOF Charging and
Tapping
Open-Hearth Charging
and Tapping
Emission Possible
(ton/yr) Controls
473 Spray System
309 Hoods and
Baghouse
326 Hoods and
Baghouse
140 Hoods and
Baghouse
Reduction Cost Cost per
% (ton/yr) $ Millions Annual Ton
60 (284) 0.2 $ 704
95 (294) 2.5 8,520
90 (293) 4.0 13,630
80 (112) 4.0 35,710

SOURCE:   Armco Steel  Corp.,  Middletown, Ohio.

-------
          REMOVAL  OF  PARTICULATE  MATTER  FROM
                               GASEOUS  WASTES
                                 WET COLLECTORS
          1.00  INTRODUCTION

 1.10  Definitions

   Wet collectors comprise all units in which a
 liquid, usually water, is employed to achieve or
 assist in the removal of dispersoids from gases.
 As used in this report, the  term  "dispersoid"
 will include dust, spray, fume, and mist.  Lap-
 pie " places  all dispersoids into  two general
 categories: "mechanical dispersoids" and "con-
 densed dispersoids." Mechanical dispersoids are
 formed by comminution, decrepitation, or dis-
 integration of larger masses of material, or by
 the grinding of solids or spraying of liquids. A
 wide range of particle sizes usually is involved.
 Mechanical dispersoids are further classified as
 "dust" and "spray," referring to dispersed sol-
 ids and liquids, respectively.
   Condensed dispersoids are formed by conden-
 sation of the vapor phase or as the product of a
 vapor phase  reaction.  Particles thus obtained
 usually are characterized by a  relatively nar-
 row size distribution.  Further classification of
 condensed dispersoids into "fume" and "mist"
 refers to solid and liquid dispersed phases, re-
 spectively.
   The sizes of dispersed particles generally are
 related to the manner in which they are formed.
 The following tabulation summarizes the classi-
 fication and presents representative size ranges
 of gas dispersoids:
                       Approximate Particle
                        Diameter (Microns)
                                > 1
                                < 1
    Dispersoids
    Mechanical:
      Dust
      Spray
    Condensed:
      Fume
      Mist
  The size ranges indicated in this tabulation
are representative of most typical cases; they
are not  intended to  be mutually exclusive nor
are they intended to serve as rigid definitions of
dispersoids which may be encountered in prac-
tice. For example, condensed  dispersoids gen-
     erally tend to agglomerate into particles of large
     size, which may actually exceed the size of many
     industrial dusts.  A fine atmospheric mist, com-
     monly called "fog," agglomerates into rather
     large rain drops.
       The term "aerosol" frequently is used in tech-
     nical literature to designate the  dispersion  of
     solid or liquid particles in a gas. The properties
     of an aerosol thus  involve the  properties  of
     dispersed particles or dispersoids as well  as
     those of the gas phase.  A summary of proper-
     ties of typical aerosols,  which is widely quoted
     and frequently reproduced by a number of au-
     thors, was  compiled by  C.  E.  Miller4J and  is
     reproduced  as Fig.  1.  Further  references  to
     Miller's  chart will be  made  throughout  this
     report.
       For reasons of simplicity and convenience
     herein, the  term "aerosol" will  be used to indi-
     cate  the  more  or  less  stable dispersions  of
     solids and  liquids in a gas; another term,
     "particulates," will be used to denote the dis-
     persed particles, both solid and liquid. In  con-
     sonance with this terminology, the present re-
     port will attempt to present information on the
     theory and  practice of  the  separation of  par-
     ticulates from aerosols by methods which utilize
     a liquid to achieve or assist in the separation.

     1.20  Classification of Wet Collectors

       The unit operation by which particulate mat-
     ter is removed from a stream of  gases may  be
     viewed as a sequence of three steps:
     1. Conditioning of particulate matter.
     2. Disengagement or separation of the particu-
     late phase from the carrier gas.
     3. Removal  of the separated particulates from
     the collector.
       In the  industrial equipment, step 1 may  be
     absent altogether;  may  be accomplished  sep-
     arately;  or  may be combined with disengage-
     ment, step 2.
       In wet  collectors, as defined in  Par. 1.10, the
     particulate  matter  leaves  the collector as  a
SOURCE:  Removal of  Particulate Matter From Gaseous  Uastes--Wet  Collectors
          Engineering  Report  Prepared
          New  York, NY (1961).
for  American  Petroleum Institute,

-------
                                         AMERICAN  PETROLEUM  INSTITUTE
 f MiC'I'l)
 Ctwpmini  ler
 Coiiiciif* 9*
  *ina*ol  of
'ariielti Iran a
                                                                     Gravity S«H'lnq  el  Soh«rtl  in
                                                                                  0«>  of SvHIing
                                                            Critical  Par lid* DiQm*r*r
                                                            ADO«I  wnich  Lo.  will
                                                                  Not  Apply
    100.000
H l» 1 10.000

    *>
                                                     >00,000
                                                     ,0.000
                           TT

                         Till.
                           II":-
                         1 1 i  :
                                     ilh
gi
- t-
itttrt — {
M
1 I 1 1 1
           1.000
 (IOO>IO.OI -
                              •-  i
                             an
                           A*i    •-
                             !
                             5    !
   jl-^
 n

                                        _ s
"•T
 I
 _L
                                                                   • 200.000 •
                                              «•*•»;.
                                                                                  Slckfi-CvKKinghoi" >e«
                                                                                                     Tt| Cd*-»«k«n {•r>ic«««
                                                                                                     $!•»••'>  (.•• ••••••I  !**•!
                                                                                                     fg* ••rlieltl il «l •»•••••

                                                                                                     ••• «••••  o.O i «ie**- i«'
                                                                                                                    .
                                                  ..........

                                                      C, > Oi»-i'«» •* ti"»»it«l MMrcU. M,

                                                                                         f l It.li  ••mil,. l«. IB«t«/(V. II.

                               FIC. 1—Summary of Properties of Some Typical Aerosols.

-------
                  REMOVAL  OF PARTICULATE MATTER FROM GASEOUS WASTES
slurry or in a dissolved state.  It is either dis-
carded to waste or is subjected to a separation
procedure—intended to recover the particulates,
if economically justified, or to permit recircula-
tion of the scrubbing liquid.
  Conditioning of the particulate matter, step 1,
and its disengagement  from the carrier  gas,
step 2, in wet collectors may involve  one or more
of several mechanisms. These mechanisms, dis-
cussed at greater length in  Sect. 2.00, may be
listed as follows:
1. Impingement  (impaction) and interception.
2. Brownian  motion diffusion.
3. Humidification of gas.
4. Condensation of liquid on particles.
5. Agglomeration of particles.
6. Electrostatic precipitation.
  Although it would  appear, at a first  glance,
that a classification of wet collectors based upon
the  preceding mechanisms  might be feasible,
in almost all instances  commercial  equipment
utilizes more  than one of the mechanisms listed
and thus  precludes such  a  classification.  The
wide variety of scrubbers and the many possible
combinations of  functional  mechanisms make
it difficult to  develop a comprehensive and  mu-
tually exclusive classification of types.
  A classification used  by Lapple  "•4-  enjoys
wide acceptance and is adopted in  the present
report.  A summary of this classification of wet
collectors, which is used as the basis for presen-
tation of material in Sect. 3.00, is presented as
follows:
1. Chamber scrubbers.
2. Cyclonic scrubbers.
3. Inertial scrubbers.
4. Mechanical scrubbers.
5. Packed scrubbers.
6. Film scrubbers.
7. Miscellaneous scrubbers.
  The several categories used in this classifica-
tion are briefly described  as follows:
  1. Chamber scrubbers are devices in which
the  dust-laden gas simply enters and leaves a
chamber where one or more spray  nozzles are
mounted.  The chamber  may be large in order
to slow down the gas flow (conventional spray
washer), or it may be in the form of a venturi
to speed up the gas flow.  The spray nozzles may
be operated at a high  pressure to produce a fog
or the liquid may be  introduced into the throat
of an ejector to provide the draft for the move-
ment of the gas.
  2. Cyclonic scrubbers contain structural char-
acteristics, which  impart centrifugal forces to
the aerosol. The centrifugal action may be in-
duced by a tangential entrance of the dust-laden
air, by forcing the gas through specially shaped
vanes, or by constraining the gas to a spiral-
shaped chamber. The scrubbing liquid is intro-
duced in various manners, and vanes and baffles
frequently are used to facilitate disengagement
of the collecting  liquid  from the gas  stream
before the latter leaves the scrubber.
  3. Inertial scrubber is the term applied to
those devices in which the energy of the dust-
laden gas stream  primarily is used  to  expand
the surface area of the scrubbing liquid  and
thus obtain liquid  contact.
  In a venturi scrubber, a popular and effective
scrubber in the inertial group, scrubbing liquor
enters the throat of a venturi at a relatively low
velocity but encounters a high-velocity  flow of
gas in the throat.  In another type of wet in-
ertial scrubber, gas jets produced by an orifice
plate  entrain the scrubbing liquor, which flows
over the orifice plate, and the mixed spray im-
pinges upon properly located target plates.  In
still another type of inertial collector, the scrub-
bing  liquor is  contacted  with dust-laden  gas
when  the  latter  is forced  through  suitably-
restrictive passages submerged in  the liquor.
Other specific examples of the inertial scrubber
are described in Sect. 3.00.
  4. Mechanical scrubbers are characterized by
the presence of some  mechanical devices for
producing  the spray. Thus, in one model of a
mechanical scrubber, widely  used in blast fur-
nace gas cleaning, the dust-laden gas and scrub-
bing liquid are passed outward through a series
of rotating and stationary  arms.  In another
group of devices of the same general type, dust-
laden gas is scrubbed with liquor dispersed by
means of various shaped rotors dipping into the
liquid.
  5. Packed scrubbers are conventional towers
packed with  raschig rings, berl saddles, grids,
and so forth operated with liquor loadings well
below flooding  point.  In these scrubbers the
liquor primarily serves  to wash the dust off
the surfaces and  to avoid  re-entrainment of
the dust.
  6. Film  scrubbers comprise those units in
which the scrubbing liquid is present only as a
film on collecting surfaces. In such units, except
for collection resulting from humidification and

-------
                               AMERICAN  PETROLEUM INSTITUTE
condensation, the liquid serves merely to keep
the collecting surface free of solids and to pre-
vent re-entrainment of collected dust.
   7. Miscellaneous scrubbers include a number
of wet collectors which  are  not conveniently
classified  under the preceding categories.

1.30  Range  of Application
   Wet collectors as a class provide a moderate
cost, high-efficiency method for removal of par-
ticulates.  Their use is indicated  as a potential
means for the collection of particulates4= -when:
1. Addition of liquid to the gas  stream is  not
objectionable.
2. Particulate  matter is  very  fine—diameters
predominantly ranging between 10 and 0.1  mi-
crons.
3. A  moderately  high collection efficiency is
required.
4. Gas must be cooled.
5. Vapors or gaseous contaminants must also be
removed from the air stream.
   The use of a wet collector is  not advisable
when a less expensive device, such as a cyclonic
scrubber,  will perform satisfactorily.  Wet  col-
lectors  normally  are  competitive with  cloth
collectors (bag filters), although  they  may be
used to handle dust-laden gases  under condi-
tions  where  bag filters are not applicable be-
cause of temperature, moisture conditions, ex-
plosion hazards, and so forth.
   The use of wet collectors presents a  number
of problems  under certain conditions.  Silver-
man <4 lists these conditions and resultant prob-
lems as follows:
1. Soluble particulates for recovery must be
recrystallized and  may  become  contaminated
during collection.
2. Collector requires means of disposal for the
sludges, e.g., sludge ponds, tailing  piles,  etc.,
some of which may create a source of secondary
contamination.
3. Recovery of insoluble product requires a de-
watering  step.
4. Low removal efficiency for particles smaller
than 1 micron  in diameter.
5. Removal  of  soluble  contaminants  by  the
scrubbing liquor may introduce corrosion prob-
lems.
6. Liquid  entrainment in the effluent gas stream
represents a source of contamination.
7. Freezing problems in cold weather.
  Wet collectors are used in many chemical and
process industries.

1.40  Comparison \vith Other Types of
        Collectors
  Kane " qualitatively compares the character-
istics of five principal types of dust collectors:
cyclones, high-efficiency cyclones, wet collectors,
fabric collectors,  and high-voltage electrostatic
precipitators. The  usual relationships between
the basic groups of collector designs are shown
in Appendix A; exceptions for specialized de-
signs are permissible.
  The use of the collectors, representative  of
the five basic groups in a number of industrial
processes, is summarized in Appendix B.  Al-
though the  ratings—usual, frequent, consider-
able,  occasional, etc.—are qualitative and may
be expected to reflect the personal opinions of
the compiler," the  tabulation nevertheless is
believed to be useful as a check list against con-
clusions reached by analysis of several factors
influencing collector selection.

2.00  REMOVAL OF SMALL  PARTICLES
        FROM MOVING GAS  STREAM
2.10  General Considerations
  As stated in Par. 1.20, as a class, wet collec-
tors are distinguished  by the employment  of
liquid to achieve or assist  in the removal  of
participate matter from gases. In most of the
wet collectors described in this report the liquid
is dispered through the  gas and the collection
of the particles is the result of their interaction
with dispersed droplets of the scrubbing liquid.
Thus, the effectiveness  of the removal of par-
ticles is the sum of interactions of all the drop-
lets with the particles of the aerosol.   These
interactions are involved in both particle con-
ditioning and particle precipitation.
  The several  mechanisms  of particle-droplet
interaction will be examined  quantitatively in
the following paragraphs.

2.20  Particle Flocculation
  The increase in the average size of the par-
ticles as a result of flocculation makes the sub-
sequent  collection  of particles  much  easier.
Flocculation  occurs when small  particles  en-
gaged in Brownian motion collide.  The colli-
sions  are inelastic and if the particles are solid
—e.g., fumes of lead, oxides of zinc, magnesium,
iron—chainlike aggregates are formed.  Liquid
particles coalesce into a  single drop.

-------
                  REMOVAL OF PARTICULATE MATTER FROM  GASEOUS WASTES
  Since  Brownian motion  is a  result of  the
impact  of molecules upon the  particles,  the
extent of this motion and, therefore, the impor-
tance of  flocculation in collection of participates
increases with the temperature of the aerosol.
However, this is significant only for those par-
ticles which are  sufficiently  small to be  in-
fluenced  by molecular impacts. Thus, the phe-
nomenon of flocculation is of importance with
submicron sizes only, 0.1 micron or less.  A high
concentration of  particles, by increasing  the
likelihood of collisions,  similarly increases  the
extent of flocculation.
  The flocculation rate is given quantitatively
by an expression for the rate of decrease of  the
particle concentration with time.  Junge30 pre-
sents the flocculation rate for a homogeneous
aerosol composed of spherical particles in still
air  as follows:
           dt
                           ""
                                         (1)
Where:
   dc
   dt:

   R':
   T =
  Km,=
   Dp =
    c =
       flocculation rate, in particles  per unit
         volume per unit time.
       gas  constant=8.3(10r), in dyne-centi-
         meters per (gram-mole) (degrees Kel-
         vin).
       absolute temperature, in degrees Kelvin.
       gas viscosity, in poises or dyne-seconds
         per square centimeter.
       Avogadro's number = 6.02 (10=3)  mole-
         cules per gram-mole.
       correction factor (1.72 for air).
       mean free  path of gas molecules, in mi-
         crons  (0.1 for air).
       diameter of particles, in microns.
       concentration, in particles per cubic cen-
         timeter.
  Flocculation is  of some  importance in the
conditioning of metallic fume particles before
removal in settling chambers  or by bag filters.
In the case of wet collectors, where the  usual
range  of applications is in the micron range
or higher,  Brownian diffusion and flocculation
do not contribute significantly to the collection
efficiency.  Moreover, the  interaction  of  par-
ticulates with droplets  of  the liquid predomi-
nates to such  an  extent that contribution of
collisions between particles themselves to the
phenomenon of conditioning is reduced further.
The occurrence of  flocculation, nevertheless,
must be kept in  mind when describing the
average  particle  size entering any collector, if
the process generating the aerosol gives rise to
a high concentration of submicron particles of
appropriate nature.
  A substantial amount of flocculation may be
produced if an aerosol is exposed to sound waves
of high frequency  such as from a siren.  The
process is not well understood nor is it possible
to calculate accurately the coagulation rate from
theory.30- *° In general, the particles must  be
smaller than 10 microns and their concentration
greater than 1 grain per cubic foot.
  The process  has been  tried on a  fair-sized
scale in the collection of carbon black 80 and for
the collection of  sulfuric acid,13 but has not as
yet been established on a firm design  basis.


2.30  Impaction of Particles on Liquid Droplets

  Fundamental  analysis  of wet collectors re-
quires an analysis  of impaction of small  par-
ticles upon droplets of the collecting liquid with
which the collector, or a portion of it, is filled.
Such an  analysis must include consideration of
all of the forces that operate between the  par-
ticles and the droplets. For particles of normal
density in the micron and submicron  ranges of
size, the  primary mechanisms  by which collec-
tion may take place are:

1.  Inertia.
2.  Interception.
3.  Settling.
4.  Brownian motion.
5.  Electrostatic attraction.

  In the case of inertia, a particle carried along
by the  gas stream, on approaching an obstruc-
tion (liquid droplet), tends to follow the stream
but may strike the obstruction because of its
inertia. In Fig. 2 solid lines represent the fluid
streamlines around  a droplet  of diameter Db
and the dotted lines represent the paths of  par-
ticles which initially followed the fluid stream-
lines.
  For a  flow around a spherical collector the
         V
quantity  i-, where X is the distance between
         •L>1>
limiting  streamlines A and B, represents the
fraction of particles initially present in a vol-
ume swept by the droplet which will be removed
by inertial impaction.

-------
                               AMERICAN  PETROLEUM  INSTITUTE
  In interception, the trajectory of the center
of a relatively large particle, although not inter-
secting the  collecting surface, may pass close
enough for the surface of the particle to touch
the collecting drop and be arrested by it. Also,
the particle may be encouraged to impact upon
the collecting surface by an actual or induced
electrostatic  force  between the  particle  and
collector.
surface provided by liquid  droplets, Ranz  and
Wong 5S defined the following parameters:

  1. Inertial parameter, ^=18D ^Q^Pp1

Where:
    C = empirical correction factor for resist-
         ance  of a gas to  movement of small
         particles, dimensionless =
                                             -0.44D
   0]
                                                        for
  In the case of settling, particles in a slowly
moving air stream may settle out on the collect-
ing surface under the influence of gravity. For
particles smaller than 0.1 micron,  Brownian
diffusion becomes significant.
  Impaction  of aerosol particles on body col-
lectors, such as cylinders and spheres, was stud-
ied theoretically by Albrecht,2 Landahl and
Herrmann,39  Langmuir and Blodgett,40 and  by
Sell," all of whom considered inertia as the only
mechanism of  collection.  Because of the com-
plicated nature of  the  problem, the results of
these authors were not in good agreement, par-
ticularly with regard to the question of whether
a  minimum  particle size exists  below which
impaction cannot occur.
  Ranz and Wong 50 developed  a mathematical
statement of the problem of impaction without
restricting their consideration to a single mech-
anism.  They carried out limiting solutions and
made  order-of-magnitude calculations, which
establish the nature and the importance of the
several mechanisms of impaction.
   Through evaluation of the forces which affect
the motion of  the particles and cause them to
move across the streamlines to the collecting


        	 FLUID  STREAMLINE

        	PARTICLE  PATH
    p = particle  density,  in grams  per cubic
         centimeter.
   v0= velocity  of  aerosol  stream, in  centi-
         meters per second.
   D,,=diameter of aerosol particle, in microns.
   Db = diameter of droplet, in microns.
    /i=viscosity of gas, in poises.

    A = ^£, mean free path of gas molecules, in
         centimeters.
    v = average  molecular velocity  of gas mole-
         cules,* in centimeters per second.
   Parameter ^ may be considered to be the ratio
of the force necessary to stop a particle initially

traveling at velocity v0  in  the  distance -£, to
                                        £i
the fluid resistance at a relative particle velocity
of v0.  It is also the ratio of the stopping dis-
tance—i.e., the distance a particle will penetrate
into still gas when given an initial velocity of
v0—to the  diameter  of the liquid  droplet.

   2. Interception parameter, R= -=p
   Parameter R is the ratio of the particle di-
ameter to  the diameter of the liquid droplet.
   3. Settling parameter,
                                                  Where:
                                                     <>,,= particle density, in gram mass per cubic
                                                           centimeter.
                                                     gL = absolute value  of local acceleration  of
                                                           gravity, in centimeters per second per
                                                           second.
                                                    Parameter F is the  ratio of the  force  of
                                                  gravity to the fluid resistance at a relative par-
                                                  ticle velocity of v0. It  is also the ratio of the
  FIG. 2—Inrrlial Impaction upon Single Droplet.
                                                    * v is the square root of the mean square velocity of
                                                  the molecular, the expression for which is given in all
                                                  standard treatments of kinetic theory; see for example,
                                                  Samuel Glasstone,  Textbook  of Physical  Chemistry,
                                                  D. Van Nostrand Co., Inc., New York, 247 (1940).

-------
                   REMOVAL  OF PARTICULATE  MATTER FROM  GASEOUS WASTES
free-settling  velocity of  the  particle to  the
stream velocity.
  4. Broivnian motion diffusion parameter,
              g_3Dr.M__  CR'T
                  v0Db  /tDbDpv0
Where:
  DDM = diffusivity, in square  centimeters  per
           second.
    R' = gas constant.
  Parameter 8  is the ratio  of diffusive force
caused by random thermal motion to the fluid
resistance.

  5. Electrostatic attraction.
  Two parameters  are  defined describing:  1,
the interaction of a positively charged particle
and droplet, KE; and 2, the interaction between
a charged droplet and a dielectric particle  on
which the droplet charge induces a charge, K,.
These parameters are:
  TABLE 1—Collection Parameters ut Venluri Throat

        (Gas velocity at throat = 500 fps)
                  Ammonium
                    Sulflte
 Dlhntyl
Phchnlate
            K,= =
Where:
   qp= electric charge on particle, in coulombs.
   qac=electric charge  on droplet, in coulombs
          per square centimeter.
    «„= permittivity of free space.
    nl«ss parameters are based on
the mean  diameters of  the  particles and water droplets. The
parameters for the electrostatic fon:i»3 are based on a maximum
charge density of (2.65)110-*) coulombs per square  centimeter
for charged surfaces in air.  Higher charges  will  leak away
through  air lonizatlon and corona discharge.  The dielectric
constant" for the ammonium salts was taken as 6.S nnd for the
dlbutyl phthalute as 6.
mechanisms  of collection.  At lower  velocities,
electrostatic  forces  may become important if
particles are charged artificially. Collection by
Brownian  diffusion is  negligible even  for the
0.27-micron ammonium chloride aerosol.  This
parameter becomes  important when  the parti-
cles are smaller than 0.1 micron and when the
relative velocity between the droplets and par-
ticles is small.
  The inertial  impaction  of aerosol particles
upon droplets of scrubbing liquid is discussed in
the following paragraphs.


2.40   Efficiency of Collection by Impaction on
       a Single Droplet

  The removal of aerosol particles  through im-
paction on droplets of collecting liquid has been
the subject of  a number of theoretical  and  ex-
perimental studies.3- =l--*• "• *°-:8' " Efficiency of
impaction, also termed "target  efficiency," TH,
may be considered as the ratio of the area of the
aerosol tube—from which all particles are re-
moved—to the projected  frontal  area  of  the
droplet.  In terms of Fig. 2, if X is the limiting
width of  the initial streamlines, in  which all
particles collide with and are impacted upon the
droplet of  diameter Dh, the efficiency of impac-
tion is:
  The several studies  (see  references  cited),
notably that  by Langmuir and Blodgett,40  es-
tablished that the target efficiency should be a
function of the dimensionless group
                     utV0                  (3)

-------
                  REMOVAL OF  PARTICIPATE MATTER FROM GASEOUS WASTES
                                          11
varies linearly with  the  specific  area of the
droplets.

2.60   Grade-Efficiency Curves for Collectors
  Although the material presented in the pre-
ceding paragraphs is  indicative of the effect of
particle size, droplet size,  specific surface area,
properties  of the gas phase, etc. on the collec-
tion efficiency, more dependable methods of pre-
dicting collector efficiencies must  be based  on
actual performance  data.  Unfortunately, the
collection of accurate performance data is time-
consuming, expensive, and is subject to a num-
ber of uncertainties involved in accurate char-
acterization of the particulates in the inlet and
outlet  of a collector.  Moreover,  isolated data
for plant performance on different duties are
difficult to  compare.
  Stairmand ar proposed a method  for practical
evaluation  of performance of collectors and in a
subsequent publication "8 cited a number of ex-
amples drawn from extensive  tests conducted
on large-scale industrial dust  collection equip-
ment.
  Stairmand describes  the performance  of a
number of dust collectors in terms of a "grade-
efficiency"  curve characteristic for each type of
equipment. In such curves, the collection effi-
ciency for  particles having an "effective den-
sity"  of 2.7 g per cu  cm is plotted against the
particle  size in  microns.   Fig. 6, reproduced
from Stairmand," is an example of such curves
and is based upon a large industrial spray tower
processing 70,000 cfm of gas.
  In addition, Stairmand "3 describes a method
for direct comparison of different collector types
           S       IO       19       20

                 OJBTICLE SIZE (MICRONS)
Courtesy of C. J. Stairmand and The Institute of Fuel
  (London).

  FIG. 6—Craclc-EfHciency  Curve for Spray Tower.
based on the use of their characteristic grade-
efficiency curves. This method consists in cal-
culating the overall collection efficiency of par-
ticulates  from  a standard-test  aerosol.   It  is
exemplified in Table 2, in which the overall col-
lection efficiency of a spray tower, with a grade-
efficiency curve as shown in Fig. 6, is computed
for three dusts which have the same grading as
shown in the first  two columns of the  table.
Table 2 also includes a prediction of the grading
of dust in the outlet gas.  It is assumed that the
spray tower operates under essentially the same
conditions  as  those for  which  the  grade-effi-
ciency curve was established.
  The calculations summarized in Table 2 illus-
trate the effect of particle-size distribution upon
the overall  collection efficiency of the scrubber.
The presence of sizable  particles results in  a
large  predicted overall removal  efficiency (ap-
proximately 92 per cent for minus 150-micron
dust), whereas  the fine  dust  can be removed
much less efficiently (77 per cent for minus 10-
micron dust).
  The method  described  by  Stairmand M  and
illustrated  in Table 2 can be used to compare
directly the performance  of different types of
collectors by computing in a similar manner the
expected  overall collection  efficiencies of the
standard dust by the different collectors.   It
should be noted that  the comparison  depends
upon  the availability  of  experimentally  deter-
mined relations of the grade efficiencies versus
particle size for conditions of gas flow, droplet
size, and so forth for which comparison is to be
made.


3.00 INDUSTRIAL  WET  COLLECTORS

3.10 Chamber  Scrubbers

  Chamber scrubbers represent a class of wet
collectors in which dust-laden gas simply  enters
and leaves a chamber  where one or more  spray
nozzles are mounted. There are many industrial
variations  of  this class  of collectors.  These
range from a stack spray arrangement, in  which
a single spray is installed in a stack carrying
exhaust gases from a furnace,  to much more
complicated, specially  designed towers, where a
series of baffles  is used to impart to the  gas a
tortuous path and provide for repeated contact
of the dust-laden gas with scrubbing liquor.
Several examples of chamber scrubbers are de-
scribed herein.

-------
12
                             AMERICAN PETROLEUM INSTITUTE
TABLE 2 — Prediction of Overall Efficiency of a Spray Tower and Grading
J 2 »' <" 5' ^
Efficiency
Per Cent by at Mean
Size of Weight In Size of
Oracle Grade nt Grade
(Microns) Inlet (Per Cent)
Dust A (Minus 150 Microns)
104 to 150
75 to 104
60 to 75
40 to 60
80 to 40
20 to 30
15 to 20
10 to 15
7.5 to 10.0
5.0 to 7.5
2.5 to 5.0
1.0 to 2.5
0.5 to 1.0
0.0 to 0.5

3
7
10
15
10
10
7
8
4
6
8
5
4
3
100
100
99.5
99.1
98.7
98.3
97.9
97.2
97.0
95.5
94.5
93.0
85.0
42.0
15.0

Overall
Collection
(Per Ceni)
3.00
6.97
9.91
14.81
9.83
9.79
6.80
7.76
3.82
5.67
7.44
4.25
1.68
0.45
92.18
of Exit Dust
C"
Grading of Exit Dust
As Per Cent
of Inlet
0.00
0.03
0.09
0.19
0.17
0.21
0.20
0.24
0.1S
0.33
0.56
0.75
2.32
2.55
7.82
Acrunl
I'erCent
0.0
0.4
1.1
2.4
2.2
2.7
2.6
3.1
2.3
4.2
7.2
9.6
29.6
32.6
100.0
Dust B (Minus 40 Microns)
30 to 40
20 to 30
15 to 20
10 to 15
7.5 to 10.0
5.0 to 7.5
2.5 to 5.0
1.0 to 2.5
0.5 to 1.0
0.0 to 0.5

15
15
11
12
6
10
12
&
6
5
100
9S.3
97.9
97.2
97.0
95.5
94.5
93.0
85.0
42.0
15.0

14.75
14.69
10.69
11.64
5.73
9.45
11.16
6.80
2.52
0.75
SS.18
0.25
0.31
0.31
0.36
0.27
0.55
0.84
1.20
3.48
4.25
11.82
2.1
2.6
2.6
3.0
2.3
4.7
7.1
10.2
29.4
36.0
100.0
Dust C (Minus 10 Microns)
7.5 to 10.0
5.0 to 7.5
2.5 to 5.0
1.0 to 2.5
0.5 to 1.0
0.0 to 0.5

• Olnmn •"•:

11 (.'(iliiinn 4 :
« Column 0:
J Column li:
13
21
25
17
13
11
100
Rp.-ul OH" frum Fig. G.
(ColnmnS) (Column 2)
100
Column 2-Coliiinn 4.
(Column S) (1001
Total oC Culiiinii u
95.5
94.5
93.0
85.0
42.0
15.0






12.42
19.85
23.25
14.45
5.46
1.65
77.07





0.58
1.15
1.75
2.55
7.54
9.35
22.92





2.5
5.0
7.6
11.1
33.0
40.8
100.0





 3.11  Gravity Spray Towers
    The use of sprays located on top of foundry
 cupolas for control of  cupola emission is de-
 scribed by Brechtelsbauer.'  The final arrange-
 ment recommended as a result of operating ex-
 perience is shown in Fig. 7.  Here the flue gas is
 deflected  by means of  a conical baffle  and is
 forced to flow through the annulus.  A single
 spray delivers water onto the top of the conical
baffle; the water flows across the top of the baffle
and contacts the gas in the  annular space be-
tween the edge of  the baffle  and the enclosing
shroud.  The flow of water over the conical baffle
also serves to cool the baffle and extends the life
of the cone. The  tests  indicate  that the dust
content  of the  outlet gas can be reduced to ap-
proximately 0.2 grain per cubic foot  of gas,
measured at standard conditions, with liquid

-------
                  REMOVAL  OF PARTICULATE MATTER FROM GASEOUS  WASTES
                                           13
consumption of approximately 6 gpm per sq ft
of cross-sectional stack area. The installation *
is inexpensive and  this degree of scrubbing is
stated to meet the  code requirements in many
areas of the country.
  Updegraff'- reports the results of a  labora-
tory study on collection of fly ash and dust from
boiler stacks during soot-blowing and fire-clean-
ing periods by means of a single nozzle installed
centrally in the stack.  An arrangement recom-
mended by  Bituminous Coal  Research, Inc./"'-
is illustrated in Fig.  8.   The spray  collection
efficiency was shown to be influenced primarily
by the water rate.  A rate of 0.4 gpm per sq ft
of stack area appeared to be optimum, giving
a collection of 60 to 70  per cent of the dust. The
selection of a proper nozzle is  based on water
pressure available; nozzles which deliver  the
recommended volume  of  water are easily se-
lected from  manufacturers' catalogs.
  An example of a considerably more  compli-
cated spray tower is described by Ashman,5 see
Fig. 9. The dust-laden gas is made to follow an
upward tortuous path  during which phase it is
contacted by liquid introduced at the top of the
tower and is cascaded downward over the baf-
fles. Primary sprays located near the gas inlet
are used to reduce the temperature of gas and to
facilitate dust removal. The mechanism of dust
collection  is principally the inertial impaction
of the dust  on wetted surfaces, although some
WATER
                          SUPPLf PIPE.
            SPRAY
                                     CLEAN  OUT
                                       DOOR
 Courtesy of Bituminous Coal Research, Inc.

       FIC. 8—Furnace Slack Spray Scrubber.
                          i UPPER
                             SPRAYS
                                                                             GAS
                                                                             OUTLET
                                                                           ^BAFFLES
Courtesy of American Foundrymen's Association.

          FIC. 7—Cupola Cas Scrubber.
                                                                             PRIMARY
                                                                           —I SPRAYS
                              I DRAIN

 Courtesy of R. Ashman and Institution of Mechanical
   Engineers (London).

        FIG. 9—Gravitational Spruv Scrubber.

-------
14
AMERICAN PETROLEUM  INSTITUTE
impaction on water droplets cascading from the
edges of baffles undoubtedly also occurs.
  The only data of sufficiently general interest
on  the collection  efficiency  of  gravity spray
towers is that  given by Stairmand;"  see  Par.
2.60 and Fig. 6. The data refer to performance
of a large industrial spray tower, handling ap-
proximately 70,000 cfm  of gas.  The tower  is
22  ft  in diameter by 66 ft high and  uses
75,000 gph of water, about IS gal per  1,000 cu
ft of gas.  The pressure  drop is stated to be
negligible, less than 1 in. of water. The efficien-
cies of removal of particulates in such a tower
range from 99 + per cent for particles 60 mi-
crons  or  larger to 55 per cent for 1-micron
particles, see Fig. 6. An overall efficiency of re-
moval of particles  in a test aerosol containing
solid particles of sizes up to 150  microns is ap-
proximately  92 per cent, decreasing to 77 per
cent for a test aerosol containing solid  particles
of sizes up to 10 microns only, see Table 2.
   Although the gravity spray tower is to some
extent tending to become obsolete because of its
relatively high cost, it is often used as a pre-
cooler where very  large  quantities of  gas are
involved, e.g.,  in blast furnace operation.  An-
other  advantage is that no very  fine clearances
are involved, thus it can handle relatively  high
concentrations of dust without chokage. As in-
dicated in Par. 2.40, because very small spray
droplet sizes are not called for, the spray nozzles
need not produce fine jets and reliability is im-
proved, with the added advantage that the spray
water can be recirculated until it contains  quite
a high concentration of suspended solids, which
simplifies the  effluent treatment and disposal
problems.
   The grade-efficiency curve for a spray tower,
see Fig. 6, shows that the collection efficiency of
a spray tower decreases rapidly for  particles
smaller than approximately 2 microns. The con-
sideration of the effect of droplet size on target
efficiency also  shows that there is little point in
decreasing the droplet size, even if it could be
accomplished  economically.   Thus, inherently,
 the use of spray towers is limited to cases where
 a high collection efficiency of particles in sizes
 below 2 microns is not required.

 3.12   Water-Jet Scrubbers

   In  an effort to increase the relative velocity
 between  the droplets of  the scrubbing  liquor
 and  the aerosol, the  scrubbing  liquor may be
 supplied in the form  of  a high-velocity jet  di-
 rected along the axis of a venturi nozzle.  In the
                   usual  industrial construction  the  water-jet is
                   utilized both as a means for moving the dust- or
                   fume-laden air and for removal of  particulates.
                   A typical water-jet scrubber is shown in Fig. 10.
                     Manufacturers' catalogs "• »• provide  perti-
                   nent data on the sizes of units and nozzles, water
                   consumption as a function of the water pres-
                   sure available, and  the amount of  gas  to be
                   passed through the unit. The literature M states
                   that standard  units  permit a reduction of 70 to
                   95 per cent of particulates with  a single unit
                   and a reduction of 80 to 99 per cent in a bank
                   of four units placed  in series for particles  larger
                   than 5 microns.  Where the particle size  is less
                   than 5 microns, much higher velocities and  tur-
                   bulences are required for adequate collection.
                   No  other information on  collection efficiencies
                   has been published.
                     As is shown in Fig. 10,  in addition to the jet,
                   a water-gas separating device, which may be
                   a  simple  gas reversal  chamber,  sometimes
                    Courtesy of Schutte and Koevting Company.

                           FIG. 10—Tvpical Water-Jet Scrubber.

-------
                  REMOVAL OF PARTICULATE MATTER  FROM  GASEOUS  WASTES
                                          15
equipped with baffles, must be provided on the
outlet side. Sizes ranging from 3 in. to 72 in.
in inlet diameter are available with capacities
up to 110,000 cfm (at  1-in draft, water pres-
sure  is 200 psig,  and  water flow through  the
nozzle is 10,000 gpm).

3.13   Chamber  Scrubbers   with   Mechanical
       Spray Generators

   Kristal, et al.,38 describe a wet chamber scrub-
ber consisting of  single or multiple collection
stages, each containing a venturi tube and two
mechanical spray generators.  An experimental
unit,  tested by the authors, consisting  of four
similar collection stages is diagramed in Fig. 11;
the individual stages are contained in quadrants
of a cylindrical housing, and not in  a simplified
structure as is schematically indicated.  In each
stage  the  aerosol encounters  a  spray  in  the
plenum chamber,  passes  through the  venturi
section and again  is contacted  with a spray in
a plenum chamber following the venturi section.
In multistage units, the spray generators  are
used as indicated in Fig. 11.
  The spray generators are of mechanical  de-
sign:  water jets produced by relatively large
orifices (TVm- diameter) impinge upon a motor-
driven, rapidly  rotating  disk  (3,300  rpm)
equipped with beveled vanes.  The  rotation of
the disk produces a fine spray of water drop-
lets with a linear velocity initially approaching
that  of the vane tips (approximately 150 fps).
The droplet size is of the order of magnitude of
100  to 400  microns and depends upon  the
amount of water  supplied, speed of the disk
rotation, orifice size, and so forth.
  In  operation,  dust-laden gas enters the  ple-
                       STJOE OUTLETS
          Q SPHAV oeNEftirOfl. 9 CAL./MIN. Q 7 V.S.I.

Courtesy of the Harvard Air Cleaning Laboratory and
  U.S. Atomic Energy Commission.

   FIG. 11—Experimental Four-Stage Wet Collector
    Equipped with Meclianicul Spray Generators.
num  chamber  of  the  first  stage where it  en-
counters the first-spray generator.  Collisions
between spray droplets and  particles in this
zone, followed  by inertial separation of large
drops (prior to entering the venturi tube), af-
ford a considerable degree of dust removal. The
aerosol then enters the first-stage venturi tube
in which additional collisions occur.  Here,  be-
cause of an increase in approach  velocity  be-
tween the  relatively large water droplets and
the small dust  particles in the accelerating air
stream, significant target efficiencies are  ob-
tained.  The velocity of small particles remains
essentially  the  same as that of the gas so that
they  pass  through a zone  of  relatively  slow-
moving, large  droplets, which increases  the
probability of  capture by  impaction.  A  small
temperature drop, produced  by rapid  (adia-
batic) expansion, occurs in the water-saturated
gas stream as it traverses the expanding portion
of the venturi.  Contrary to the expectation of
the designers, the  condensation of water vapor
on dust particles serving as nuclei, which would
"condition" the particles, is not an important
mechanism of  collection because  of the very
brief retention time.
  After leaving the venturi tube the aerosol  en-
ters a second plenum where it passes  over  the
second-spray generator. For single-stage opera-
tion the aerosol is then withdrawn through a
droplet eliminator. For multistage operation  the
cycle is repeated as shown in the schematic dia-
gram, Fig.  11.
  The collection efficiency of the device depends
upon the aerosol size, the number of stages, and
the rate of gas flow.  Significant improvement
was observed by using a second stage, but  use
of additional stages produced  only minor  im-
provement. Efficiencies determined for a single
stage varied from 99  per  cent by weight  for
resuspended fly ash  (median  diameter—count
basis, 0.6 micron;  mass basis, 14.3 microns) to
22 per  cent for iron oxide fume  (median  di-
ameter—count  basis, 0.03 micron;  mass basis,
0.6 micron).
  The device is characterized  by rather high
pressure loss, approximately 7 in. of water per
stage, and  large power and water requirements
compared  to other wet collectors  capable of
similar  performance.  The mechanical  spray
generator,  however, is nonclogging; its use may
eliminate the need for elaborate filtering  ap-
paratus and thus permit the economic recycling
of the spent spray water in regions of sparse
water supply.

-------
16
AMERICAN PETROLEUM  INSTITUTE
3.20  Cyclonic Scrubbers
  This group of wet collectors includes several
types of units in which scrubbing is combined
with cyclonic action.  In these units, the action
of water droplets conditions the aerosol parti-
cles by a combination of impaction, humidifica-
tion, and condensation.  The centrifugal motion
of the aerosol is utilized to separate the condi-
tioned particles from the gas phase by an in-
ertial  mechanism.  The  numerous industrial
modifications differ in the methods of introduc-
ing the scrubbing liquor, inducing centrifugal
motion  of the aerosol, arrangements for reduc-
ing entrainment, and  in details of removing the
sludge from the unit.
  Several typical industrial modifications of cy-
clonic scrubbers are discussed in the following
sections and  illustrated in schematic drawings.
All of these units can  be used to clean hot gases;
scrubbing  liquor usually is recirculated,  pro-
vided appropriate arrangements for removing
the collected solids are made; and  soluble con-
stituents of the gas phase are absorbed in the
scrubbing liquor.

3.21  Cyclonic Scruhhers Equipped with Vane
        Baffles
   These units are cylindrical  towers with a
tangential entry of the aerosol. Several typical
units are illustrated in Fig. 12 through 17.
   The  cyclonic wash scrubber, illustrated in
Fig. 12, accomplishes scrubbing "• *; of an aero-
sol by a combination of inertial and impaction
mechanism.  The unit is an arrangement of one
or  more washing stages  followed by  an elimi-
nator stage, as shown  in Fig. 12.  Centrifugal
motion is  imparted to  the gas  stream both by
the tangential inlet and by the vanes.  Washing
 liquid  usually is  introduced  through  nozzles
above the top washing stage. In a multistage ar-
 rangement the lower stages may be  supplied
 with additional liquid, this arrangement is rec-
 ommended for high  dust loadings to maintain
 desired slurry characteristics and also for pre-
 humidification where elevated temperatures are
 encountered.
   In operation,  the  dust-laden air enters the
 lower chamber.  The centrifugal motion aided
 by initial  wetting  removes the relatively  large
 particles from the air  stream.  The air stream
 then is accelerated by passing through the vanes
 of the washing stage where particulates impinge
 upon  wetted surfaces of the  vanes.  The ac-
 celerated, centrifugally moving air stream then
 contacts the washing liquid in the center spray
                                                          CLEAN  AIR  OUT
                                               WATER
                                                MIST.LADEN
                                                 AIR  IN
                              WATER  OUT

                   Courtesy of C. F. Montross and Chemical Engineering.

                           FIG.  12—Cyclonic \Fash Scrubber.
                                           GAS  OUT
                                                      OiRTY
                                                      GAS  IN
                              WATER OUT
                    Courtesy of C. F. Montross and Chemical Engineering.

                           FIG. 13—Cyclonic Baffle  Scrubber.

-------
                 REMOVAL OF PARTICIPATE  MATTER FROM GASEOUS WASTES
                                         17
                        CLEAN  AIR OUT
                                                                      GAS  OUT
   SEPARATOR
                                IMPINGEMENT
                                 PLATES
Courtesy of C. F. Montross and Chemical Engineering.
     FIG. 14—Cyclonic Multiple-Ruffle Scrubber.

chamber, where the particles are impacted upon
the liquid droplets and the latter are separated
from the gas stream by centrifugal motion.  A
shielded-cone baffle prevents formation of a vor-
tex in the air stream, and directs the liquid onto
the vanes and into  the centrifugal path of the
air stream.
  A set of vane baffles is provided at the top of
the tower to reduce the entrainment of liquid.
To  increase the efficiency of removal, two or
more washing stages of similar design may be
required.  For removal of submicron particles,
multivane designs are used; in these, two rows
of  vanes  are mounted  inversely  against  each
other and are spaced more closely to provide a
larger surface  area for impingement.
  Units  are designed with capacities ranging
from 500  cfm to 40,000  cfm of gas.  Normal
water requirements for units having a single
washing  stage  are approximately 2 gal per
1,000 cu.ft of gas.  The pressure drop  ranges
from 1£ in. to 3 in. of water. Units employing
more than one washing stage are characterized
by a higher water requirement and  pressure
drop.
                                                        ANTI-SPIN
                                                         VANES
                                                        CORE
                                                       BUSTER
                                                        DISK
                                                      DAMPER
                                                                           WATER
                                                                             IN
                                                   GAS  IN
                                                 Courtesy of C. F. Montross and Chemical Engineering.

                                                        FIC. 15—Cyclonic  Spray Scrubber I.
     CLEAN  AIR
     DISCHARGE
                            NOZZLE
                            ORIENTATION
CONTAMINATED
AIR INTAKE
                               BANKS OF
                               NOZZLES
PUMP
                     LIQUID  EFFLUENT
Courtesy of D. G. Hudson and Heating and Ventilating.

       FIG. 16—Cyclonic Spray Scrubber II.

-------
 18
AMERICAN PETROLEUM  INSTITUTE
                        CLEAN  AIR
                          OUTLET
  VERTICAL
   SPRAY
   RISERS
QUICK OPENING
   NOZZLE
   LATCHES
  .TOWER NOZZLES,
     DIRECTED
    CROSS-FLOW
 FLUSHING JETS,
DIRECTED  DOWNWARD
                                     RECTANGULAR
                                       INLET
                                    FRESH WATER
                                      SUPPLY
                        WASTE  OUTLET
 Courtesy of Buffalo Forge Company.
        FIG. 17—Cyclonic Spray Scrubber III.


   One manufacturer " states that collection ef-
 ficiencies of 98 to 99 per cent  by weight can be
 realized in  most cases.   Friedlander,  et al.,=-
 reports a  weight collection  efficiency of 74 per
 cent, when the effluent from a  dry-cyclone in-
 stallation  was washed in a hydraulic  scrubbing
 tower  of the design described herein. The two
 figures are not necessarily  contradictory.  The
 higher efficiency refers to collection of particu-
 lates from usual industrial effluents in which a
 fairly  wide  range  of particle sizes is  encoun-
 tered; the lower efficiency refers to  secondary
 treatment  of  an   effluent  with  low  loading
 (5.8 grains per cubic foot)  and  particle size of
 1.5 microns  and less.
   Performance efficiency data of wider applica-
 bility which would relate the collection efficiency
 with particle size, such as is illustrated in Fig. 6,
 is not available. It  would be expected, however,
 that a  properly designed unit of  this type would
 give a grade-efficiency curve  intermediate be-
 tween  that of a spray tower and that of a wet-
 impingement scrubber, approaching the latter.
As  is shown in Fig. 6 and  23, a  recovery of
98  per cent is not unlikely  for  the  hydraulic
scrubbing tower combining the cyclonic and
impingement effects.
  Essentially  the  same principles  are  utilized
and comparable effectiveness  of  operation  is
realized in other multistage towers with a some-
what different design of vane baffles, see Fig. 13
and 14.

3.22 Cyclonic  Scrubbers  with Radial Spray
       Injection

  Another widely  used wet collector, a cyclone
spray scrubber, based on  the" scrubbing of a
centrifugally moving gas stream with a liquid
spray,-'3' "•30 is illustrated  in Fig.  15.  In this
unit dust-laden gas  enters tangentially at the
bottom of a cylindrical tower  and pursues an
upward spiral path.  No baffles are used. Spray
is introduced into  rotating gas from an axially
located manifold in the lower part of the unit.
A core-buster disk is located usually  above the
sprays to prevent  formation of a vortex.  The
spray droplets sweep across the path of the gas
stream and, because of the centrifugal motion
imparted  to them  by the gas stream, impinge
upon the wall of the tower, run down and out
the bottom  of the unit together with the col-
lected particulates.
  The cyclone spray scrubber is an efficient de-
vice for  removing small particles  from  gases
when the particles are larger than 2 microns.
For smaller particles the efficiency of  the cy-
clone spray falls off rapidly as  the particle size
decreases, unless extremely high ratios of the
scrubbing liquid to gas are used.  This decrease
in efficiency of deposition of the small particles
has been ascribed  to streamlining  of particles
around the drops.30
  Kleinschmidt"  and Kleinschmidt and An-
thony 3C developed an equation for the overall
efficiency  of removal  of dust particles  by a
scrubber of the type  described as a function of:
    D = diameter of tower, in inches.
   W = effective volume of the scrubbing liquid,
         in cubic feet per minute.
    d = diameter of droplets formed,  in inches.
    G = volume of  gas scrubbed, in cubic feet
         per minute.
  For example, for an S-ft-diameter cyclone in
which  25,000 cfm of gas  tangentially  enter
through a 4-in. by 2-in. inlet, the centrifugal
force is approximately 27.5 times gravity and

-------
                   REMOVAL OF PARTICULATE MATTER  FROM  GASEOUS  WASTES
                                                                 19
 the target efficiencies of individual droplets of
 100-micron diameter are estimated to be:
      Particle Size
      (Microns)
          5
          2
          1
          0.5
          0.2
Target Efficiency
     0.98
     0.32
     0.11
     0.0
     0.0
   Equation (6), developed in Par. 2.50, relates
 the overall effectiveness of particle removal to
 the conditions  of  scrubbing by including the
 target efficiency in the exponent of e:
                             P'W
 Thus, the overall efficiency of a cyclonic scrub-
 ber is affected by all of the factors which deter-
 mine the value of 171 :
 1.  Particle size.
 2.  Relative velocity between particles and spray
 droplets.
 3.  Density  of particles.
 4.  Viscosity of gas.
 5.  Droplet size.
   In a cyclonic scrubber  the relative velocity
 between particles and spray droplets varies con-
 tinuously as the droplet traverses the path from
 the nozzle to the wall ; the distribution of drop-
 let diameters is determined by the nozzle design,
 pressure of liquid, rate of liquid,  and so forth.
 A theoretical construction of  a grade-efficiency
 curve thus becomes very difficult;  experimental
 data on  effectiveness of cyclonic scrubbers  are
 fragmentary and meager.  An empirical method
 of  scrubber design based on performance of a
 small scrubber, or a  large scrubber operating
 under  different conditions,  is  described   by
 Kleinschmidt.3*
  The overall  efficiencies typical  of a number
 of industrial installations are  assembled below ;
 these efficiencies apply only to specific installa-
 tions described in the references cited :
                             Particle
                            Size Range
      Dust                  (Microns)
Lime 	 2.0 to 40.0
Lime 	 1.0 to 25.0
Lead compounds	 0.5 to. 2.0
Iron ore, coke, etc	 0.5 to 20.0
Ammonium nitrate	(Unstable)
Chemical fume	 0.5 to  3.5
Chemical fume	 0.2 to  2.0
Boiler fly ash	 2.0 to  5.0
       Cyclonic spray scrubbers  require from 2 to
    10 gal of water per 1,000 cu ft of gas. The pres-
    sure drop is usually in the range of 1 in. to 4 in.
    of water. A variety of corrosion-resistant ma-
    terials is used in cyclonic  spray scrubber con-
    struction to permit its use  for the scrubbing of
    gases containing  soluble  and corrosive com-
    ponents.

    3.23  Cyclonic Scrubbers with Circumferential
           Spray Injection

       In another version of a cyclonic spray tower,
    described by Hudson =' and Thomas,70 the spray
    nozzles  are placed in  several  parallel  banks
    throughout the height of a vertical tower.  The
    nozzles are arranged to discharge at a common
    angle and are directed nearly tangentially to aid
    in imparting a centrifugal motion to the aerosol
    traversing the tower.  A spray  tower of this
    design is shown in Fig.  16.  The dust-laden
    gases enter at the top of the tower, spiral down-
    ward through  the  tower,  and exit through a
    central duct, concentric with the  tower.
       A spray tower of a similar design " features
    a  construction with easily accessible nozzles,
    permitting inspection  and  replacement  of  the
    nozzles while the unit is in operation.  In this
    design, shown in Fig. 17, the aerosol enters at
    the bottom of the tower through  a rectangular
    tangential port and exits from the top  of the
    tower.
       In both designs the scrubbing liquid is intro-
    duced through the  nozzles at  high pressures.
    400 psig to 600 psig, in an effort to improve the
    collection  of small  particles.  Information on
    droplet size in a spray  thus produced  is not
    available,  although  Silverman and Davidson81
    give indirect evidence that at  400 psig the drop-
    let size is less than 50 microns.  In a fog-like
    spray the droplet size is well below the optimum
    size needed for impaction. The droplets quickly
    approach the same velocity  as the suspended
    aerosol particles and thus high impaction  effi-
    ciencies cannot be realized.  The increased spe-

    Dust Loading
(Grains per Cubic Foot)
                   Inlet
                   9.2
                   7.7
               Exit
               O.OS
               0.25
               3.0 to 24.0     0.03  to 0.08
                  17.0
                   2.25
               0.25 to 2.5
               1.02
               0.79
           0.045 to 0.125
Removal
Efficiency
(Per Cent)
99
97
97
98
99 +
94
65
82 to 95

Reference
No.
51
51
51
51
51
36
36
36

-------
20
AMERICAN PETROLEUM  INSTITUTE
cific surface area of the fine droplets, however,
would tend to counteract this effect and facili-
tate impaction.  In addition, Brownian motion
diffusion would become  operative  and submi-
cron  particles  would collide with  and impact
upon the fog droplets. This consideration of
mechanisms  of collection  suggests that fog-
producing nozzles would add to the effectiveness
of  collection of  submicron  particles.  In the
units  diagramed  in Fig.  16 and 17,  larger
particles would be collected  by  inertial forces
induced by the centrifugal motion of the gas,
and the submicron particles  would be collected
by  combination of  diffusion and  inertial im-
paction.
  Collectors  of the type described herein some-
times are referred to as "fog niters." =*•:o The
fog filter,  because  of its  limitations and the
great diversity of engineering processes, must
be  engineered to the job. It appears to be more
effective on  aerosols containing particles less
than 10 microns in diameter; presence of a large
proportion of larger particles requires the use
of  a companion unit such as a dry cyclone. Per-
formance data for the fog filter on some indus-
trial gases given  by Thomas T0  indicate a 100-
per-cent removal of sulfuric acid mist, 90.0- to
99.4-per-cent removal of solids from  fertilizer
mixing plants, and approximately 90-per-cent
removal of low-density organic  solids, such as
phthalic and maleic anhydrides.
   A composite efficiency tabulation for a similar
 scrubber using  400 psig sprays, cited by  a
 manufacturer,"  is  as follows:
       Size Fraction                Efficiency
        (Microns)             (Per Cent by Weight)
       0.0 to 0.5                  78.6
       0.5 to 1.0                  33.4
       1.0 to 2.0                  89.7
       2.0 to 3.0                  94.0
       3.0 to 4.0                  96.4
       4.0 to  6.0                  97.5
       6.0 to  8.0                  98.6
       8.0 to 10.0                  99-2
 These data  are presented  also in the form of a
 grade-efficiency curve in Fig. 18 by plotting the
 efficiencies versus the average of size fractions.
    For scrubbers of this type., collection efficien-
 cies can be  brought to almost any desired level
 for many materials by the  following means: :o
 1. Increasing tower height.
 2. Compounding stages.
 3. Controlling gas volumes.
 4. Regulating liquid flow.
 5. Regulating spray pressures.
                     FIG. 18—Cra
-------
          REMOVAL  OF PARTICULATE MATTER  FROM GASEOUS WASTES
                                                                                                      41
                                 APPENDIX A

DUST COLLECTOR  CHARACTERISTICS  (ACCORDING TO KANE M)
                                           Hlsh-
                                         EtBclency
                                          Cyclones
                                                                     Wet
                                                                   Collectors
             Item                    Cyclones
Effect of Dust Variations

Efficiency, particles:
       < 1 micron 	     Poor           Poor        Poor to fair
     1 to 10 microns	     Poor        Poor to fair     Fair to good
    10 to 20 microns	     Poor           Good           Good
      > 20 microns	  Fair to good        Good           Good
Abrasion resistance	     Fair           Fair           Good
Ability to handle sticky, adhesive ma-
  terials  	     Fair           Poor        Poor to good
Bridging materials give  trouble	     Slight           Yes             No
Fire or explosion hazard minimized...     Fair           Fair           Good
Can handle hygroscopic materials	      Yes            Fair            Yes
Large foreign materials cause plug-
  ging 	    Seldom          Yes         Seldom to yes

Effect of Gas Stream Variations
Maximum temperature (degF),
  standard construction	      750            750           No limit
Troubles from condensed or entrained
  mists  or vapors	     Slight       Considerable       Slight
Corrosive  gases attack standard con-
  struction 	     Slight          Slight          Severe

Collector
Space 	     Large          Modest         Modest

Pressure drop	  1 in. to 2 in.      3 in. to 5 in.     3 in. to 6 in.
Reduced volume adversely  affects col-
  lection efficiency	      Yes        Yes with mosc    Depends on
                                                     designs         design
1 Fabric
Collectors
                                                                        Good
                                                                        Good
                                                                        Good
                                                                        Good
                                                                        Good

                                                                        Poor
                                                                         Yes
                                                                        Poor
                                                                      With care

                                                                        Seldom
                                                                      180 to 275

                                                                     Considerable

                                                                        Slight
                                                                       Modest to
                                                                        large
                                                                      2 in. to 6 in.

                                                                         No
   High-
  Voltage
Electrostatic
Prccipitators
               Good
               Good
               Good
               Good
               Good

               Poor
               Yes
               Poor
             With care

               Yes
                750

               Some

               Slight


               Large

            1 in. to 2 in.

                No

-------
                                           APPENDIX »
USUAL AIR CLEANER  SELECTIONS FOR  INDUSTRIAL PROCESSES (ACCORDING TO  KANE")
                                                               f'ulluirliit
                                                                         Usoil In Industry


C)|)i!rnlliiii
Ceramics
Haw product handling 	
Fettling 	

Refractory sizing 	
Glaze and vitreous enamel spray. . . .
Chemicals
Material handling 	

Crush ing, grinding 	

Pneumatic conveying 	

Roasters, kilns, coolers 	

Coal Mining uiid rower I'lunl
Material handling 	
Hunker ventilation 	
Ucdusling, air cleaning 	

Drying 	
l-'ly Ash
Coal -burn ing:
Chain grate 	
Stoker fired 	

I'ulvcri'/.cd fuel 	
Wood-burning 	
Foundry
Shake out 	

Sand handling 	

Tumbling mills 	

Abrasive cleaning 	



Ciincriitrnlliin

Light
Light

Heavy
Moderate

Light to
moderate
Moderate
to heavy
Very heavy

Heavy


Moderate
Moderate
Heavy

Moderate

Light
Moderate

Heavy
Varies

Light to
moderate
Moderate

Heavy

Moderate
to heavy

rnrtii-li!
Si/.fM

Fine
Fine to
medium
Coarse
Medium

Fine to
medium
Fine to
coarse
Fine to
coarse
Medium
to coarse

Medium
Fine
Medium
to coarse
Fine

Fine
Fine to
coarse
Fine
Ifeavy

Fine

Fine to
medium
Medium to
coarse
Fine to
medium


C'yclnniM

Rare
Rare

Seldom
No

Occasional

Often

Usual

Occasional


Rare
Occasional
Frequent

Hare

No
Rare

Rare
Occasional

Rare

Rare

No

No


ni^ii-
(•'.Illrli'licy
ryrlnlli-X

Seldom
Occasional

Occasional
No

Frequent

Frequent

Occasional

Usual


Occasional
Frequent
Frequent

Occasional

Ran:
Usual

Frequent
Occasional

Rare

Rare

No

Occasional


\Vrl
Cnllrrliir.s

Frequent
Frequent

Frequent
Usual

Frequent

Frequent

liarc

Usual


Frequent
Occasional
Occasional

Frequent

No
Nn

No
No

Usual

Usual

Frequent

Frequent


Knlirlc
Am-Mlcrs

Frequent
Frequent

Frequent
Occasional

Frequent

Frequent

Usual

Rare


Frequent
Frequent
Often

No

No
No

No
No

Rare

Rare

Frequent

Frequent

IliBh-
Kli-rlrtisliilta 1
rrri:l|>itiili>ra

No
No

No
No

Rare

No

No

Often


No
No
No

Nn

No
Rare

Frequent
No

No

No

No

No


inn'iar
NIL*

1
2

3

4

r.

G

7


8
il
10

11

12

33
31

10

1C

17

18


-------
Cruin Elevator, Flour ami Feed Mills
Grain handling	
Grain dryers 	
Flour dust	
Feed mill 	
Metal MMi-ng
Steel blast furnace.
Steel  open  hearth.
Stool  olcctric furnace....
ferrous cupola 	
Nonfcrrous rcvcrhcralory
Nonferrous crucible	
Metal Mining imdRuek I'rotlueta
Material handling	
Dryers, kilns	

Cement rock dryer.

Coment kiln	
Cement grinding	
Cement clinker cooler	

Melul Working
Froduclion grinding, scratch brushing,
  abrasive cutoff	
Portable and swing frame	
Tinning  	
Tool room 	
Cast iron machining	

I'liiirmiicenticnl mid Food Products

Mixers,  grinders, weighing, blending,
  bagging, packaging	
Coating  pans	
 /'/((sties
 Uaw material processing.
 1'laslic finish ing	
        1'roilitcts
 Mixers ...................
 liatchout rolls ............
 Talc dusting and deducting.
 Grinding .................
\VooituMrrkhnj

Woodworking machines .
Sanding  	
Waste conveying, hogs..
    • l>'or lU-iimrka act- p. 4-1.
  Light
  Light
Moderate
Moderate
 Heavy
Moderate

  Light
Moderate
 Varied
  Light
Moderate

Moderate
to heavy
Moderate

  Heavy

Moderate
Moderate
  Light
  Light
  Light
  Light
 Moderate
  Light
  Varied
 Light to
 moderate


 Moderate
  Light
 Moderate
 Moderate
                                       Moderate
                                       Moderate
                                         Heavy
 Medium
 Coarse
 Medium
 Medium
  Varied
  Fine to
  coarse
   Fine
  Varied
   Fine
   Fine
  Fine to
 medium
Medium to
  coarse
  Fine to
 medium
  Fine to
 medium
   Fine
  Coarse
  Coarse
  Varied
  Varied
   Fine
  Varied
 Medium
  Fine to
 medium
               Varied
   Fine
   Fine
 Medium
  Coarse
               Varied
                Fine
               Varied
  Usual      Occasional
   No           No
  Usual        Often
  Usual        Often
Frequent
   No

   No
   Rare
   No
   No
  Rare

 Frequent

  Karc
   Uarc
Occasional
 Frequent
 Frequent
 Frequent
 Frequent
   Rare
   Hare
   Uarc
   Hare
   No

   No
   Karc
   No
   No
Occasional

 Frequent

 Frequent

 Frequent

   Karc
Occasional
 Frequent
   Rare
   Rare
 Frequent
 Frequent
 Frequent
   Rare
Rare
No
Occasional
Occasional
Frequent
Doubtful
Considerable
Frequent
Hare
Rare
Usual
Frequent
Occasional
It a re
No
Considerable
Frequent
Frequent
Frequent
Considerable
Frequent
Frequent
Frequent
No
Frequent
Frequent
No
Possible
Frequent
Occasional
Occasional
Considerable
Rare
No
No
Frequent
Considerable
Rare
Rare
Frequent
Considerable
Frequent
Frequent
No
No
No
No
Frequent
Probable
Rare
Occasional

Occasional
Occasional
Considerable
Rare
No
No
No
No
No
No
                    (Sec commmls under cliemiciila)
              Frequent      Frequent     Frequent
    No
    No
    No
  Often
                Usual
              Frequent
                Usual
    No
    No
    No
   Often
              Occasional
              Occasional
                Rare
Frequent
  Usual
Frequent
Frequent
                 Rare
              Occasional
              Occasional
                                                                   Frequent
 Usual
Frequent
 Usual
 Often
              Frequent
              Frequent
             Occasional
                                                          No
No
No
No
No
                 No
                 No
                 No
                                                      10
                                                      20
                                                      21
                                                      22
                                        23
                                        24

                                        25
                                        2C
                                        27
                                        28
                                        20

                                        30

                                        31

                                        32

                                        33
                                        34



                                        35

                                        3(J
                                        37
                                        38
                                        30
                                        40
                                                                                              41
                                                                                              42
                                                                                                                                    43
                                                                                                                                    44
                                                                                                                                    45
                                                                                                                                    40
          47
          48
          40

-------
               REMOVAL  OF  PARTICULATE  MATTER
                         FROM  GASEOUS  WASTES
                                     FILTRATION
          1.00 INTRODUCTION
 1.10  Definition

   Filtration is an operation in which a stream
 of gas  carrying suspended particulate matter
 (dust, fume, fog, mist, or aerosol) is passed
 through a porous medium in such a way that
 the  particles impinge  on, and adhere to, the
 medium and are thereby removed from the gas
 which freely passes  on through the filter.  In
 many filters the deposit of dust so collected be-
 comes,  in turn, the filtering medium  for suc-
 ceeding particles.
   As the deposit accumulates, inevitably the
 porosity of the medium is reduced, eventually to
 the point where the flow of gas becomes so re-
 stricted as to require either: 1, removal of the
 deposit in order that the medium can be re-used;
 or, 2, discarding of the clogged filter  and re-
 placement with new medium.  It is an inherent
 feature of filtration that arrangements must be
 provided to accomplish one of these alternatives
 either periodically or continuously.

 1.20  Types of Filters

   A variety of filter arrangements and many
 kinds of porous media may be used.  Possible
 arrangements  include  passing the dust-laden
 gas through:
 1. A  flexible sheet, layer,  tube, or bag, as of
 woven fabric, felt, or paper.
 2. A semirigid supported fabric or nonwoven
 mat of fibrous material.
 3. A rigid porous solid.
 4. A fixed or packed  bed of dry granular  par-
 ticles.*
 5. A  fluidized  or  moving  bed of  granules or
 fibers.

   Fibers used to make  fabrics or mats may be
 of wool, cotton, metal  (e.g., steel wool), asbes-

  * The usual wet-packed  tower is considered a wet
 collector rather than a filter.
 tos,  cellulose,  fiberglas, Fiberfrax (ceramic),
 Orion, Nylon, or other synthetic polymers. They
 may be used in the natural (dry) condition or
 treated—as with resins, in order to give them
 an electrostatic charge; or with a viscous oil, in
 order to make them sticky. Rigid solids include
 metal screens, porous metals, or porous ceram-
 ics.  Fixed or moving beds may be composed of
 granules of sand, coke, slag wool, crushed rock,
 and other materials.
   The material and type of filter to be used in
 a  given application must be selected with due
 regard for the composition, temperature, and
 moisture content  of  the gas stream;  particle
 size and nature of the dust; corrosion and abra-
 sive  effects  which may be present; collection
 efficiency desired; and economic aspects of the
 operation, including  possible recovery of  the
 dust, re-use of the filter, or both.
   Filters may be classified in various ways ac-
 cording to type of media, arrangement of sup-
 port, cleaning mechanisms, etc.  A full discus-
 sion  of filter media and filter arrangements is
 given in  Sect.  3.00.  However,  it should be
 pointed out here that a very basic distinction
 is  made between  two general modes  of filter
 collection:
 1.  The medium itself  is the essential separation
 mechanism.
 2.  The medium acts  primarily  as support for
 the collected particles, which become the effec-
 tive filter.

   Auxiliary equipment,  such as dust removal
 and filter removal devices, blowers, ducts, and
 stacks, is usually  involved. There may also be
the need for coolers and other gas-conditioning
 devices before the  filtration proper.  Sometimes
 another type of particle-collecting equipment
 may also be used in series with a filter.

 1.30  Range of Application

  Dalla Valle has  stated: "It is  possible to se-
 cure  almost any kind  of filter for handling any
 type  of particle with any desired degree of effi-
SOURCE:   Removal of  Particulate Matter From Gaseous  Wastes—Filtration,
          Engineering Report  Prepared  for American  Petroleum Institute,
          New  York, NY (1961).

-------
                              AMERICAN PETROLEUM  INSTITUTE
ciency.  This can be said of no other method of
particle collection known."""
  This statement indicates the most important
advantage of filtration as a method of particle
collection—that it is capable of high-efficiency
collection of very small particles.  Its use is to
be considered  whenever a very  low dust load-
ing of the effluent gas stream is required, and
especially when the particles to be removed are
in the range of from 10 microns down to less
than 1 micron in size.
  On the other hand,  filtration  is a relatively
expensive gas-cleaning operation with  several
important limitations.   If the dust load of the
gas to be treated is high, it may be necessary
to use another type of collector as a preliminary
stage ahead of the filter. Certain filters  require
a rather low  lineal gas velocity for effective
operation, and this may, in turn, call for a very
large area  of filter surface. A relatively large
pressure difference is required to force  the gas
stream  through the  filter, and this involves
large energy consumption in the  blower  or com-
pressor. Cleaning and renewal  devices  involve
mechanical apparatus  requiring careful main-
tenance. Most filter media  must  be  used at
relatively low temperatures  and on relatively
dry gas in order to avoid excessive deteriora-
tion.  Many of them  have  rather limited re-
sistance to  corrosion and rather  low mechanical
strength, which limits the amount of handling
they can withstand.  Special media have been
developed to overcome some of these disadvan-
tages, but  their higher  cost may not be war-
ranted  in many applications.
  The following are examples of  situations in
which filters have been  used with satisfactory
results:
1. Removing  dust from atmospheric  air (as
used  in heating  or  ventilating systems)  by
panel filters made up  of mats of  glass, metal,
vegetable fibers, animal hair, etc.; often coated
with  some oily or adhesive substance or resin
(to impart electrostatic charge); usually dis-
carded  and replaced  when  dirty;  sometimes
washed, retreated, and re-used.tos
2. Collecting oxide, ash, and carbon fumes from
gray iron or nonferrous cupolas in Orion, wool,
or silicpne-treated glass \vool bag filters, cleaned
by periodic manual shaking.10'll1
3. Removing  sand and stone  dust from kiln
stack  gas  leaving a bituminous  mix  asphalt
plant, by filtration through dense wool felt bags
(following  a  preliminary  cyclone  collector),
cleaned continuously by a flow of air  from a

  1 Figures refer to REFERENCES  on p. 54.
"reverse jet" traveling over the obverse surface
of the bag."
4. Cleaning blast furnace gas  by bag  filters
made of a specially woven fabric  of asbestos
and glass fibers,  mounted in a  special way to
minimize the deterioration due  to the shaking
necessary for cleaning.70
5. Removing fine radioactive or other particles
with very high efficiency (99.99  per cent)  from
exhaust or supply ventilation air by a soft felt-
like asbestos-bearing cellulose paper formed in
thin pleated sheets, used until dirty, and then
discarded.91
6. Cleaning dust  (including  carbon particles
from  generator  brushes) from fresh and re-
circulated air  supplied to "motor rooms" by a
vertical endless-belt type of continuously travel-
ing screen, self-cleaning and oiled as it dips into
a pan of oil at the bottom of the  belt.1"
1. Removing  air-borne  bacteria  in  order to
produce sterile air in industrial fermentation
plants  (e.g.,  in  penicillin  manufacture),  by
filtration through thick beds of slag or  glass
wool, "cleaned" in situ by sterilization with dry
heat a large number of times before replace-
ment becomes necessary.18
8. Carrying fine  particles of  valuable catalyst
from a fluidized  bed and recovering them by
passing the exhaust gas through porous stain-
less  steel filters  made  by sintering powdered
metal, cleaned by automatic blowback.92
9. Removing  iron oxide dust  in stack  gases
from  an  open-hearth  furnace by  filtration
through a horizontally moving bed of fine-fiber
slag wool, continuously reclaimed by passing
through a  washing, drying, and bed-reforming
system.110
10. Removing coal  dust from synthesis gas
streams passed countercurrently upward to a
vertically moving bed of granular coke, peri-
odically  removed from the  bottom,  washed,
drained, and returned to the top."
11. Removing sulfuric acid  mist from air by
passing it through a fluidized bed of porous
silica  gel  or  alumina  particles revivified  by
washing with water and drying for re-use.38

   The foregoing examples  were selected only
to give a general idea of the possible applica-
tions of filtration methods of particle collection;
they are not intended to form a comprehensive
survey of  the field.  They do serve to indicate
two broad classes of  filter duty, which may be
defined roughly as:

1. The air-cleaning  range,  characterized by a
relatively low dust loading of the  carrier gas,
e.g., less than 1 grain per 1,000 cu  ft.

-------
                   REMOVAL  OF PARTICULATE MATTER  FROM  GASEOUS WASTES
 2. The dust-collecting range, with loading of
 the  carrier gas ranging upward from 1 grain
 per 1,000 cu ft to as high as (5) (104) grains
 per 1,000 cu ft.
   In general, the nonrenewable mat, pad, paper
 or bed type of  filter will be used for air cleaning
 (examples 1, 5, 7), with a cleanable type used
 in some cases  (example 6).  The heavier load-
 ings involved in dust collecting usually require
 renewal of the filter by cleaning, either to re-
 cover valuable  material in the dust  (example 8)
 or to prevent  excessive coat of filter replace-
 ment where the  main purpose of  collection is
 to reduce  air pollution (examples  2,  3, 9, 11)
 or process gas  contamination (examples 4, 10).

 1.40  The Role of Filters in Collection

   The position of filtration in comparison with
 other methods  of particle collection is a ques-
 tion of relative effectiveness and relative cost.
 This is indicated  in a general way by the classi-
 fications given by Kane ••:
 1. High-efficiency, high-cost collectors:
   a. Electrostatic precipitators.
   6. Sonic agglomerators.
 2. High-efficiency, moderate-cost collectors:
   a. Fabric or fibrous filters.
   6. Wet collectors, packed towers, scrubbers,
      and centrifugals.
 3. Low-cost, lower efficiency designs:
   a. Cyclones and dry centrifugals.
   b. Dry dynamic.
   c. Inertial.
  A  chart presented by  Stairmand1!1  based
 upon actual  operating data generally  supports
the foregoing classifications, although  showing
 venturi scrubbers as being most costly  and elec-
 trostatic precipitators as  less efficient  than fil-
 ters  (see Sect.  4.00, Fig. 11).
  The important effect of particle size upon col-
 lector performance is illustrated by the charts
presented by Kane OB and McCabe." These indi-
 cate that filters retain their high-efficiency char-
acteristics down to finer particle sizes than any
other collectors except electrostatic precipita-
tors.
  These general results are supported  by many
tests  reported in  the literature. They  serve to
substantiate  the statement  that filtration is a
moderate- to high-cost method of collection, ca-
pable of the highest efficiency, and particularly
effective in dealing with fine particles.
  It should be noted that these statements refer
mainly to fabric and fibrous filters, inasmuch as
 the position of porous solid, packed-bed, and
 moving-bed filters is not indicated in these clas-
 sifications.  This is because these methods are
 of rather recent development and there is insuf-
 ficient  experience  available  from  commercial
 installations to judge their ultimate role.  Most
 of the pilot  plant  or developmental  type of
 studies which have been  reported,  however,
 indicate that such filters are all also capable of
 high efficiency, but probably at higher cost—
 which may be justifiable in special applications.

     2.00  THEORY OF  FILTRATION

   A comprehensive theory of the operation of
 filters should deal with the mechanism and effi-
 ciency of particle collection, the life  of the filter
 medium,  and  the pressure  drop through the
 filter. Each of these questions will be considered
 in turn.  For  the reader who does not wish to
 consider  theory in detail, the following para-
 graphs will give the highlights of the theoreti-
 cal approach and sources of the practical results
 obtained:  2.11, 2.127  (p. 12, 15),  2.133, and
 2.30.
   Theoretical developments  have for  the  most
 part been limited either to the performance of
 a "clean" filter, i.e., at the beginning of the col-
 lection  of a homogeneous dust, or to one on
 which a homogeneous  dust  has  collected  in  a
 uniform  manner.  That actual  filter  behavior
 may be more  complex than is contemplated in
 these^implified models  is pointed out in several
 places in the following discussion.

 2.10   Particle  Collection

   In order for a particle to be removed from
 the gas stream, it is first necessary  that it col-
 lide with  the surface of an element  (e.g.,  fiber
 or granules) of the filter  and then adhere to
 this element, at least until  it is desired to clean
 the filter.  The theory must therefore deal first
 with the  interaction between individual parti-
 cles  and  individual  filter  elements,  and  then
 with the  composite  effect  of all the elements
making up the filter medium.

2.11  Basic Mechanisms for Collision

  Stoppage of particles by direct sieving action
is seldom  an important  aspect of dust filtration.
as the spaces  between  the filter elements are
usually much larger than the particles collected.
If  this were not the case, clogging would occur

-------
4
AMERICAN PETROLEUM INSTITUTE
 rapidly, with an attendant severe rise in pres-
 sure drop across the filter.
1   Other mechanisms must be relied  upon to
 cause the particles to collide with the obstacle,
 which the  filter element represents,  in their
 path. These have been identified as:
 1. Direct interception or flow line interception.
 2. Inertial deposition or impaction.
 3. Diffusional deposition  or Brownian move-
 ment.
 4. Gravity settling.
 5. Electrostatic precipitation.
 6. Thermal precipitation.

   Direct interception, or flow line interception,
 occurs  whenever the  fluid streamline  along
 which a  particle approaches  a filter element
 passes within a distance from the element equal
 to one-half the particle diameter. If the particle
 has  a very  small mass,  although of finite size,
 it will not deviate from  the streamline as the
 latter curves around the obstacle and will there-
 fore collide  if the streamline passes sufficiently
 close. This  is illustrated by path A  in Fig. 1.
   Inertial impaction occurs when the mass of
 the particle is great  enough that it cannot fol-
 low  the streamline rapidly curving around the
 obstacle but tends to continue along a path of
 lesser curvature. This brings the particle closer
 to the  filter  element than  it  would  have ap-
 proached along the streamline.  Collisions may
 therefore occur due to this inertial effect, even
 when flow line interception would not take place
 (see path B in Fig. 1).
                     Brownian movement will be  superimposed
                   upon the flow motion of  very small particles.
                   This may cause a particle to diffuse toward, and
                   contact  the  surface  of, a filter  element as it
                   flows  by.- The particle must pass sufficiently
                   close to  the obstacle  for a long enough time in
                   order for the relatively slow diffusional velocity
                   to bring about a collision. The process is akin
                   to mass  transfer by molecular diffusion.
                     Gravity settling onto the filter surface may
                   result from vertical motion  of a particle due to
                   its weight as it passes through the filter.
                     Electrostatic precipitation will occur as a re-
                   sult of  electrostatic  forces drawing particle
                   and filter element together whenever either  or
                   both possess a static charge. These forces may
                   be either direct attraction, where both particle
                   and filter are  charged, or induced, if  only one
                   of them is charged.  Such charges are usually
                   not  present unless deliberately introduced dur-
                   ing  the manufacturing of  the filter. They must
                   be strong enough to  draw a particle out of its
                   flow path to the filter surface during  the time
                   the  particle passes nearby.
                     Thermal precipitation may occur whenever
                   there  is a temperature gradient between the
                   gas  stream and the filter surface.  Particles may
                   thus be caused to migrate toward a cold surface
                   or away from  a warm one.  Thermal gradients
                   do not exist in normal practice, and this princi-
                   ple  has not yet found industrial application.  It
                   will not be discussed further in this report.
                     All  of these mechanisms are not usually  in
                   effect at the  same time in  a given filtration sit-
                                                     Electrostatic  Attraction.
                                     b    (A) Direct  -Interception
                       FIG. 1—Streamlines and Particle Trajectories Approaching
                                         Filter Element.

-------
                   REMOVAL OF  PARTICULATE  MATTER FROM GASEOUS WASTES
 uation. Only one, or a combination of just two     adheres to it and therefore may be regarded as
 or three of them, may be involved. It is neces-     "collected,"  the single-element efficiency of col-
 sary to analyze each filter application to deter-     lection or "target" efficiency  is defined as
              _cross-sectional area of fluid stream from which particles are removed
             17   cross-sectional area of filter  element  projected  in direction of  flow
 mine the  controlling mechanisms present, in     or,
 order that the filter may be operated to the best                         _ b
 advantage. Which mechanisms are important                        1?=D7                 ^
 in a given case will be determined by such fac-
 tors as: size and density of particles, size and
 nature of filter elements, velocity and pattern of
 fluid flow, temperature, nature of gas, existence
 of electric fields, etc.
   Each mechanism may be characterized by a
 basic dimensionless  parameter which may be
 calculated  with reference to a single aerosol
 particle and an individual filter element. From
 the magnitude  of these parameters, the impor-
 tance of each mechanism may be judged in a
 given case and  its contribution to the efficiency
 of the filter estimated.   These parameters are
 discussed in detail hereinafter.
2.12  Collection Efficiency of Individual Filter
       Elements
  There are principally two kinds of elements
used to make up aerosol filters: fibers and gran-
ules.  The theoretical ideal model of a filter ele-
ment may therefore  be  regarded as either a
cylinder (representing a fibrous element) or a
sphere  (representing a  granular  element).
Granular filters  (e.g., deep-bed, fluidized-bed,
moving-bed) have not as yet come into wide-
spread use for various reasons; therefore, the
spherical model collecting element has been the
subject of little or no theoretical investigation
as applied to dry filters.   For wet collectors of
the spray type, the spherical model represents
the individual  drop of liquid as the collecting
element and, therefore, has been studied exten-
sively.
  The pattern of theoretical investigation of
the various  mechanisms is much the same for
cylindrical and spherical collectors, with  due
allowance for the consequences of the geometri-
cal differences. For the reasons just mentioned,
only the theory of cylindrical elements will be
presented here.  A review of  the theory  of
spherical elements may be found in the report
Wet Collectors by Gilbert, which is another in
this API series on dust collection.
  In theory, the idealized  aerosol particle is al-
ways regarded as a sphere. Assuming that each
such particle which collides with a filter element
 In Fig.  1, — is shown as the initial distance
 from the central streamline of particles which
 just graze the surface of the element, and D, is
 the diameter of that element. The value of b,
 and the resulting target efficiency, must be con-
 sidered for each mechanism separately and for
 various combinations of mechanisms  operating
 simultaneously.
   Since  any  particle passing  by  whatever

 mechanism within distance =£• of the element

 0=| will be collected,  the efficiency can also be
 calculated  as a flow  ratio:
                                         (2)
Here Q represents the volumetric rate of flow
per unit length of cylinder in the space within

a distance of-^of the surface at 0=|.  Q  is

found  from the  velocity  profile v£  at B = ~
                                 ~         &
according to
                                         (3)
  2.121  DESCRIPTION OF FLOW  PATTERN: A
description of the pattern of streamlines around
a filter element is evidently needed in  develop-
ing the theory of collection by any of the mecha-
nisms. For a cylinder, exact mathematical de-
scriptions taking into account the boundary
layer and separation  lines are  available  but
very complicated. It has been assumed that the
equations of streamlines relatively close to,  and
in front of, the cylinder are all that is needed.
On  this basis, two simplified descriptions have
been used: 1, ideal fluid; and,  2, viscous fluid.
  Ideal fluid:   At  high  Reynolds  numbers
(strictly, as  /VA.9-»co) the flow in front of a

-------
24
AMERICAN PETROLEUM  INSTITUTE
Df. At medium velocities,  direct interception
is important; and the smaller Dt, the higher the
filtration criterion.
3. For the same Df, Df, and ft: The filtration
criterion decreases at first, remains fairly con-
stant, and then increases as V increases.
4. For the same Df, V, and ft: At the high values
of V, the filtration criterion increases with in-
creasing Dr  At medium values of V, the filtra-
tion criterion remains essentially constant. At
high values of  V,  the filtration criterion in-
creases with decreasing Dr
   There are really no quantitative  theoretical
methods to predict  filter life or the optimum
time  of effective performance.   The  experi-
mental  data of Rowley and Jordan,107 for ex-
ample,  show that r;,  may  remain  essentially
constant for some time after Ap begins to in-
crease appreciably. When Ap is increasing more
rapidly than the penetration, the filtration cri-
terion begins to decrease rapidly.  At this point,
it is necessary to stop and either clean or renew
the filter.  However, this point can  be deter-
mined only by trial under operating conditions.

    3.00  INDUSTRIAL DUST FILTERS

   Industrial dust  filters may be classified ac-
cording to their construction and mode of op-
eration by the following scheme:

Cloth or Fabric Collectors

A. Intermittent—operation interrupted by
      cleaning:
1.  Bags, bag houses, tubes, hoses.
    a. Cleaned by shaking or rapping.
    6. Cleaned by reverse flow of air.
2.  Screen-supported envelopes.
    a. Cleaned by shaking or rapping.
    b. Cleaned by reverse flow of air.

B. Continuous—cleaning continuous during
    operation:
1.  Multiple sections of A-l and A-2.
2.  Reverse-jet filters.
3.  Nonshaking types.

Fixed Beds or Layers
A. Granular—deep  beds of coke, sand, etc.

B. Fibrous:
1.  Air filters—mats of fibers for air cleaning.
    a. Viscous  impingement—fibers  coated by
       fluid adhesive: 1, disposable; 2,  renew-
                        able;  3,  washable;  4,  automatic  self-
                        cleaning (see Moving Beds, B-l).
                      b. Dry:  1, disposable; 2, renewable;  3,
                        washable; 4, automatic self-cleaning (see
                        Moving Beds, B-l).
                   2.  Mats or pads for dust recovery.
                      a. Dry, through flow.
                      b. Radial flow, variable compression.
                      c. Treated—electrostatic, etc.
                   3.  Papers.
                      a. High efficiency: cellulose, asbestos, glass,
                        plastic.
                      b. Multiple plies.

                   C.  Rigid porous:
                   1.  Porous metal.
                   2.  Plastic.
                   3.  Porous ceramic.

                   Moving Beds

                   A. Granular:
                   1.  Gravity flow of collector granules.
                   2.  Fluidized bed  of collector granules.

                   B. Fibrous:
                   1.  Self-cleaning air filters.
                   2.  Traveling mat of collector fibers.

                     The construction, operation, applications, and
                   performance of each of the foregoing types of
                   filters will be described briefly.

                   3.10  Cloth  or Fabric Collectors

                     The  collecting medium is a woven or felted
                   fabric which first collects particles by the vari-
                   ous mechanisms  of impingement until  a layer
                   or floe  of dust is formed. This, in turn, acts as
                   the medium to collect additional  particles  by
                   sieving action. The collection becomes more effi-
                   cient, but the pressure drop increases ever more
                   rapidly as  the operation  proceeds.  Cleaning
                   removes most of the floe but leaves a basic layer
                   in the  fibers so that efficiency increases after
                   new cloth is installed, and the layer of dust is
                   an essential feature in obtaining high efficiency.
                   For  this reason,  best results  are obtained on
                   gas streams carrying a high load of dust. There
                   are two general designs  available:  the tube or
                   bag type, and  the cloth screen type.

                   3.11  Bags,  Bag Houses, and Tube Filters

                     The  filter fabric is in the form of a bag, tube,
                   or hose,  either cylindrical or  oblong, and sus-
                   pended vertically. Many such bags are operated

-------
                  REMOVAL OF PARTICULATE MATTER FROM GASEOUS WASTES
                                          25
in parallel, all confined in one structure called
a bag: house.  This is possibly the oldest type
of filter.
  In Bennett's description7 of one installed in
1906, it is stated that there were 1,920 bags,
each 18 in. in diameter by 28 ft 8 in. in length,
housed in  a reinforced concrete building ap-
proximately 90 ft by 126 ft by 60 ft high. These
bags were suspended from their closed top ends,
with the  open bottom  ends held  in  place  by
rings set  in a concrete floor.  Dust-laden hot
gases from a lead smelter were blown, at the
rate of 55,000 cfm, by an enormous fan into a
chamber  below the bag floor, where  much of
the coarser and heavier particles settled out.
The remaining finer particles were carried  up
inside the bags by the gas flow and trapped on
the inner surface, the clean gas passing through
the bags  and out to a  common flue and stack.
Periodically the bags were shaken—apparently
manually—to dislodge the collected dust, which
then fell  into the chamber below. With reduc-
tion in size, and improvements in capacity and
cleaning methods, this type of  filter is still in
use today.
  Present-day versions of the bag house ar-
rangement include one or more of the following
modifications:
1. A mechanical shaking device to agitate the
top hanger.
2. Subdivision  of the  bag  house  into 'units
which may be closed off individually for clean-
ing while the  remainder continues to operate,
thus giving essentially steady  continuous op-
eration.
3. Increase of gas pressure on the external side
during cleaning in order to cause a reverse flow
which aids in removing the dust  loosened  by
shaking.
4. Induced flow of gas by exhaust fans on the
clean side, where the dust cannot affect the fan
operation.
5. Automatic timing of the filtering and shak-
ing  periods;  and  automatic control  of the
switchover from one to the other, which may
be activated pneumatically  by the increase in
pressure  drop across  the  fabric as  the dust
collects.
6. A precoat of filter-aid on the inside of the
bag to promote the highest collection  efficiency
from the very beginning of the collection period.
7. Completely enclosed units for indoor use.
8. Envelope-  or  oblong-shaped  bags for more
compact utilization of space.
  Standard sizes  of bag  units  are  available
from a number of  manufacturers.48' ur Bags
are  usually made  with a length-to-diameter
ratio of less than 20  to 1, often in the neighbor-
hood of 16 to 1, a fairly common size being 6 in.
in diameter by 8 ft  long. Standard units con-
tain anywhere from just a few bags up to as
many as 1,300 bags and are capable of filtering
up to 60,000 cfm of gas at standard conditions.
A variety of fabrics is used, each having certain
desirable properties for certain applications.
Fabrics are discussed in detail in Par. 3.14.
  Operating conditions are fairly standardized
with regard to velocity and pressure drop. The
value of V, referred  to as the air-to-cloth ratio
or filter ratio (cfm flow per square foot of filter
surface), usually ranges from 2.0  to  5.0 fpm
and occasionally as high as  10 or 12 fpm. Pres-
sure drop,  \T>, is on  the order of 0.2 to 0.5  in.
of water for a clean filter,  and  2.0 to 5.0 in. of
water for a "dirty" one.  Operation is always
at or near atmospheric pressure, with allowable
temperatures depending  upon the fabric used.
Obviously,  the chemical nature of the dust and
gas must also be  considered in  the choice of
fabric.
  It is  generally considered =0- "• "• ur-ll3 that
the most appropriate range of dust loading for
bag filters  is for 0.1 
-------
                            AMERICAN PETROLEUM  INSTITUTE
                            .SOLENOID
                             VALVE
                                 DIRTY
                                 GAS
                  SOLIDS
        Courtesy of Chemical Engineering.

     FIG. 7	Bag Filters Cleaned by Air Jets.


the cleaning jets.  Standard sizes handling up
to 5,000 cfm per  unit are offered.
  Bag houses are used in industry for a wide
variety of nitration purposes. Their application
is limited chiefly by the  range of  temperature
and moisture content of the gas and by the
degree of shaking which available fabrics can
withstand.  For this reason,  gas  conditioning
 (cooling,  and reduction  of moisture  content)
prior to nitration is often necessary. Fabric
properties are discussed in Par. 3.14 and gas
conditioning in Par. 3.42.
   With proper preceding of the gas, bag houses
have been very  successfully used to clean the
 hot gases arising out of a variety of metallur-
 gical  operations such as  those performed
 gray iron cupolas,80 electric steel foundries,"
 open-hearth  steel  furnaces," lead blast  fur-
 naces9 copper-base  alloy smelters,112  and  the
 like When strict air pollution control was put
 in force in Los Angeles  County in the early
 1950's bag houses were the only answer to the
 dust  recovery problems of many foundries.20
 Summaries of such applications are given in a
 number  of articles."- >'•u4  A typical example
  is that of the Alhambra Foundry, described by
Siechert and Menardi,110  data for which are
tabulated as follows:
Particulate emission:  inert ash, silica, and iron
                     oxide—0.8  to 1.6 grams
                     per cubic  foot, 25 per
                     cent finer than 325-mesh.
Gas volume: 13,100 cfm  at 400 F.
Bag house: four compartments, 112 bags each
            —each bag 11 in. in diameter by
            15 ft long.
Fabric: silicone-treated glass wool—total area
         4,835 sq ft.
Filter ratio: 2.7 fpm.
Operating conditions: 400 F; Ap = 3 m. to 4 m.
                      of water.
 Shaking:   manual,  by  compartment,  every
            90 min..
 Collection efficiency: 99 4- per cent.
   Applications to nonmetallurgical operations,
 usually but not always involving gases at ordi-
 nary temperatures,  are also common.  Gold-
 field"  describes an enormous  bag  house
 48  compartments, each  containing 1,200 bags
 5 in. in diameter by 14 ft  long, for filtration
 of  90 tons per minute  of  air in an asbestos
 mill  Many uses are found in connection with
 crushing and grinding operations such as are
 associated with mining, materials preparation,
 and the various mineral industries."-
 handling equipment in  foundries, coal grind-
  ing " and similar operations produce dust often
  recovered by  bag houses. A typical example of
  a small-scale operation is cited by Silverman
  Particulate emission:  dust from hydrated lime
                       packing   operation — 7
                       grains per cubic foot;
                        100  per cent finer than
                        37 microns, 43  per cent
                        finer than 13 microns.
  Gas  volume:   approximately  1,200   cfm  to
                  1,500 cfm.
  Bags:  cotton fabric,  340 sq ft total surface.
  Filter ratio: 8 f pm to 5 f pm.
  Operating conditions:  room temperature; A}>
                         = 2  in. to  5.2 in.  of
                         water.
   Shaking:  once every 4 hr.
   Collection efficiency:  99+ per cent
     Aside from the limitations imposed by fabric
   properties, bag house filters also suffer from the
   disadvantages of low filter ratio which requires
   large filtering  surface and a  large amount of
   space for installation. Various other cloth t

-------
                  REMOVAL  OF  PARTICULATE MATTER FROM GASEOUS WASTES
                                          27
ter arrangements, which represent attempts to
overcome some of these problems, are discussed
in the succeeding sections.

3.12 Screen-Supported Envelopes

  In this arrangement the filtering surface con-
sists of a number of  bags, each shaped  as a
rectangular envelope open on one end.  Each
bag or envelope is slipped over a wire screen
or similar  supporting framework so that the
fabric is stretched taut. That end of  the screen
where the envelope is open is attached to a rack
which holds  a number of similar screens in
parallel.  The envelope is always mounted so as
to stand on edge, but its long dimension may be
either in a horizontal or a vertical direction.
Fig. 8 shows  a typical arrangement.
  Dust-laden gas flows inward around the out-
side of the envelopes,  and the dust collects on
their  outer surfaces.  The clean gas  passes
through the fabric, flows inside the screen sup-
port, then out the open end to a manifold  and
gas discharge duct. Therefore, the rack which
holds the screens  also serves as a connection to
the clean-gas discharge system.
  Cleaning of the bags is accomplished by shak-
ing the  screen supports, by a reverse flow of
clean air through them, or by a combination of
the two actions. If the bags are to be shaken, the
supporting rack is attached to  a motor-driven
vibrating system  or to a device which raps the
rack intermittently  with  a series  of sharp
blows.  If a reverse  flow  of air is to be used,
a valve system must shut off the flow of dirty
                                     CLEAN  AIR

                                     TO FAN
    CLEANING  AIR FROM  /
      ATMOSPHERE

  Courtesy of W. W. Sly Manufacturing Company.

           FIG. 8—Screen Cloth Filter.
air to a group of bags  and,  simultaneously,
admit clean air into the open end of the enve-
lopes.  Ingenious  arrangements  are available
for doing either of these things  continuously
and automatically over all of the bags in the
filter in sequence, thus giving an essentially un-
interrupted  steady filtering operation.  In one
make of filter, where a  suction fan is installed
in the clean-gas  duct to  induce the flow, the
same  suction is used to create the flow of the
cleaning air, as shown in Fig. 8.
  In  this type of  filter the volume of space
needed for  a given  area  of  filter  surface is
much less than in a bag house, because the flat
envelopes can be  arrayed  more compactly than
bags or tubes. The filter unit thus requires less
room  and less floor space, and may be installed
in locations where other  filters would not fit.
  Since the  fabric  remains taut and  is not
subjected to flexing (except for a slight balloon-
ing action in reverse-flow cleaning), it  need
not have as much mechanical strength as that
used in bags.  Certain fibers  may therefore be
used which  have excellent heat- or corrosion-
resistant properties, but which could not with-
stand the shaking action  in a bag house. The
shaking  mechanism  for  the  screen supports
would necessarily be  heavier and more power-
ful than for flexible bags. For this reason, the
reverse-flow cleaning is preferable, provided it
can be made to dislodge the dust as effectively.
  Operating conditions  for  bag houses  and
screen filters  are very similar.   A series of
comparative tests conducted by Dennis, et al.,so
showed essentially the  same  ranges of  filter
ratio, pressure drop, inlet dust loading,  and
collection efficiencies for both types. Standard
screen filters are available in capacities ranging
from  a few  hundred cubic feet per minute for
self-contained unit filters  of less than 10 bags
to upwards  of  60,000 cfm for filters of 600 to
700 bags.
  Despite the advantages cited  in the forego-
ing, the cloth screen seems not to  be as widely
used as the  bag house.  There are many more
articles in the literature describing specific bag
house installations than those describing  cloth
screen filters.  There also seem  to be more
manufacturers  of the bag type, according to
Silverman.117  The  reason apparently is that
many types of dust particles are particularly
difficult to dislodge on cleaning,  and the clean-
ing of the cloth screen is not effective enough
to prevent clogging in these cases.-

-------
 42
AMERICAN PETROLEUM  INSTITUTE
 reverse-jet filter (see Par. 3.13) the traveling
 jet mechanism  can be set in operation when-
 ever, and  only so long as, the pressure differ-
 ential is above this limit.  Bag-shaking devices
 may also be operated automatically in this way.
 Such control is especially desirable whenever
 the dust loading or volume of gas flowing fluctu-
 ates very  much. However, whenever a steady
 condition of rate of flow and dust concentration
 prevails, the cleaning cycle may be controlled
 simply by  a program  timer.  This will auto-
 matically,  and regularly,  cut  the  filter  to be
 cleaned  out of operation  for a  fixed period of
 time,  during which  the   cleaning mechanism
 operates.  In this way a  large  bag house, for
 example, can give overall continuous operation
 as different sections  of it are cleaned in turn.

 3.45  Safety Measures

  The principal hazard in connection with dust
 filters is that of explosion and fire whenever
 the dust consists  of a combustible material.
 Other hazards may arise in special cases de-
 pending upon the nature of the dust, e.g., health
 hazards  such as silicosis, radiation injury from
 radioactive substances, infectious diseases from
 air-borne bacteria, etc.  In these cases, the filter
 is itself the principal  safety device acting to re-
 move injurious substances from the air. There-
 fore,  the principal precaution to be taken is
 the elimination  of all leaks from the filtration
 system to avoid recontamination of the air. An
 electric-eye dust detector  may be installed in
 the clean-gas discharge ducts or stack to detect
 bag failures or leaks  automatically.
  Mumford, et al.,53 in discussing the applica-
 tion of cloth filters to  the collection of coal dust,
 give a good summary of safety measures to be
 taken against explosion and fire. These include:
 1. Flameproofing of filter fabric.
 2. Installation of explosion vents on each filter
 compartment.
 3. Elimination of horizontal runs of duct work
wherever possible.
 4. Provision of access doors and cleanout plugs
 in all  locations where dust might settle in the
system.
5. Elimination of all possible sources of sparks
or static electrical discharges.

  Jameson" points out the fire hazard inher-
ent in the  use of oil-film  air cleaners and in-
dicates that an oil or  adhesive substance of
 fire-resistant properties should  be used. The
                  automatic  self-cleaning type of continuous air
                  filter should be equipped with its own automatic
                  fire-extinguishing system.

                  3.50  Operating and Maintenance Problems

                     Throughout the  literature  the  theme  ex-
                  pressed in  the following quotation from Kane68
                  is found over and over again:
                  "Because dust collection equipment is not truly
                  production machinery,  it  has too  often been
                  installed in a place that is inaccessible with the
                  unfounded hopes that once installed it can be
                  forgotten.  Nothing  is further from the  truth
                  and the more effective the design, the  more
                  complicated will be the collector construction
                  and the more frequent need for  inspection,
                  servicing and preventative maintenance."
                     Accordingly, it is important to consider plans
                  by which dust filters may be kept  in smooth,
                  trouble-free operation.  Such planning should
                  be kept  in mind from  the very beginning of
                  the design of  the installation and should  be
                  based upon a  careful study of the filter manu-
                  facturer's  instructions  for installation opera-
                  tion and maintenance. This should result in the
                  establishment of a  servicing schedule setting
                  forth the operations to be performed and their
                  frequency.
                     Several lists of items to be included in such
                  a  servicing schedule have been  proposed  in
                  connection  with  installations in different  in-
                  dustries by Bolt,10 Kidder," Smith,1" Mumford,
                  et al.,80 Swift,"1 and Harris  and Mason,"  in
                  addition to the general  comments of Kane.84-66
                  The following list of problems and comments
                  represents  a composite digest  of these  refer-
                  ences :
                     Leakage  through tlie filter: This is perhaps
                  the most important service  problem.  Bag filters
                  must  be  regularly inspected for holes or  tears
                  in the fabric  and fiber filters for channeling
                  or leaks around the edges between the bed and
                  frames.  Regular  measurement of  the down-
                  stream dust concentration serves  as a check on
                  faulty filtration.  An electronic-eye dust  detec-
                  tor may be installed  in the outlet duct to warn
                  of an increase  in dust content of this stream.
                  A visual inspection on the clean-air  side of the
                  filter may  reveal staining  of  parts by leaked
                  dust.
                     Ordinarily, a set of bags should last several
                  years if the proper fabric has been selected for
                  the operating conditions. Labbe and Donoso T5
                  recommend a monthly test of the acid content
                  and tensile strength of a fabric sample  from

-------
                  REMOVAL OF PAETICULATE MATTER FROM  GASEOUS WASTES
                                         43
the bags in service. Howat80 suggests that bags
be washed every three  or four months.  Care
should be taken that the thread used to sew the
seams  in bags is also of the proper material;
otherwise,  splitting  of  the  seams  may  be a
source of trouble.
  Lubrication of moving parts:  Fan motors
and fan bearings, shaking mechanisms, reverse-
jet blow rings, valves,  dampers,  etc. must be
lubricated regularly and checked for wear. To
avoid serious delays and shutdowns, worn parts
should be replaced before they fail in service.
  Plugging of filter: This will be indicated by
abnormal  pressure drop across the  filter, re-
duced rate of flow, or both.  It may be due to
improper performance of the cleaning system,
too high humidity in the dust stream, foreign
substances carelessly added  to the gas, or a
sudden increase in dust  load beyond the design
capacity of the filter. Constant monitoring of
the pressure differential gage is important.
  Improper accumulation of dust: This may
occur in ducts, on fan blades, in hoppers, etc.—
in many locations where dust is not supposed
to accumulate. In time, this results in malfunc-
tion of the system, excessive wear, and possibly
a fire hazard. There should be provision for
frequent inspection of such locations, and clean-
ing whenever indicated. Sudden increases in
gas velocity may stir up such dust and  cause
overloading and  plugging of the filter.
   Wear of metal -parts: There should be regular
inspection of ducts,  hoods,  framework,  hous-
ings, etc. for signs of wear due  to corrosion,
erosion, excessive heat, excessive moisture, etc.
Leaks which  may develop in this  way may re-
lease hazardous dusts into the atmosphere and
defeat  the purpose  of  the filtering  system.
Regular painting of the metal surfaces  is in-
surance against  some of this sort of  trouble.
  Electrical overloading: The current through
motors should be checked regularly for indica-
tion  of overloading.  Temperature of motors
and bearings should likewise be  checked fre-
quently.
  Improper tension in belt devices: The tension
in belt  devices should be measured regularly
and adjusted before  belts fail in  service. Im-
proper speed of fans may result from incorrect
belt tension and  this, in turn, may cause an
incorrect  velocity of gas through the  filter.
Thus, the collection efficiency could be impaired
even though there was no obvious malfunction
of the equipment.
   Convenience of inspection and servicing: To
insure  that workmen assigned to maintenance
work will follow the schedule of inspections and
servicing,  it is desirable to make it as conven-
ient  as  possible.  Outdoor locations  should  be
sheltered,  service points  should  be readily ac-
cessible, and necessary tools and testing equip-
ment should be kept in good repair.
  Instrumentation checkup:   Pressure gages,
thermocouples,  flow  meters, and all other in-
struments must be constantly checked to insure
that  they  are  giving accurate readings. This
is basic to the use of instrumentation for  the
control of any operation.
  Precautions against fire:   Fire-fighting  ap-
paratus should always be kept in  working order
and  regularly  inspected and tested.  During
repair  work particular care should  be  taken
against sources  of ignition  such as  welding
torches. Routine  care against improper accu-
mulation of dust will also remove a  possible
source  of fire by spontaneous combustion.
  As an illustration of a service schedule,  the
following abbreviated version of  the preventive
maintenance  program followed   by Bolt"  is
given:
At beginning of each 8-hr shift:
  Empty collectors of accumulated dust.
Every 24 hr:
  Check temperature of motors  and bearings.
  Check all machinery for proper lubrication.
Weekly:
  Check current in motors, for  overloading.
  Check tension on belt drives.
  Clean out underground tunnels.
  Inspect  fire control equipment.
Every three weeks:
  Clean overhead duct work.
Every four weeks:
  Scrape blades of fans, and fan housings.
  Change lubricant in all bearings.
Every three months:
  Complete inspection of entire system.
Every  year (during summer vacation  shut-
down) :
  Overhaul entire system.

            4.00   COST DATA

  Specific  and up-to-date  cost   data  are,  of
course, not obtainable from the general techni-
cal literature.  The only way to  find the exact
cost  of a piece of equipment at any given time
is to solicit price quotations from the manu-
facturers.  The best that can  be  done  in a gen-
eral  survey of this kind is to indicate general
trends  in  costs, relative costs in comparison
with other types  of  dust collectors,  and those

-------
46
                              AMERICAN PETROLEUM  INSTITUTE
  Repair  materials  and replacement parts
     (bags, filter panels, etc.)
  Dust disposal
  Depreciation of equipment
  Other fixed charges—insurance, taxes, etc.
  The United States Atomic Energy Commis-
sion's Handbook on  Air Cleaning 48 states that
for metallurgical  bag filters total  operating
costs  are  divided roughly as follows:  38  per
cent for labor; 25 per cent for power; 12 per
cent for bag renewal; 7 per cent for supervi-
sion; and the remaining 18 per cent for fuel,
tools  and  supplies,  laundry, power shovel  and
engine, and car service. No mention  is made of
fixed  charges. This handbook  also  cites quite
a number of specific cost data for a variety of
collectors; however, inasmuch as it dates from
1946  to 1950, the figures are no  longer valid.
  A  useful source  of  current  information on
costs and  chemical  engineering  economics in
general is the annual  review, "Chemical Cost
and Profitability Estimation," published in In-
dustrial and Engineering Chemistry  in the May
or June issue. This is a survey and bibliography
of the literature which has appeared during the
previous calendar year,  with a subject index.
"Dust collectors" and "Filters,  air" are the
appropriate listings in the index.

5.00  PRACTICAL CONSIDERATIONS IN
           FILTER SELECTION

   In the approach  to any dust collection prob-
lem there  are always  two basic considerations
to be taken into account:
1. What is the character of the gas-dust stream
to be dealt with? This must include a complete
description of  all  its properties:   dust  size
distribution, dust concentration,  temperature,
moisture content, chemical composition of  dust
and gas, physical nature of dust,  etc.
2. What is to be accomplished by the collection
operation? The requirements to be  met by the
collection equipment must be specified in terms
of effluent dust loading; particle size, re-use, or
disposal of collected dust;  and quantity to be
handled.
   Answers to the two foregoing questions will
 generally  lead to a preliminary selection of the
 method of dust collection to be employed. In
 fact, consideration of only particle size, particle
 concentration, and  desired collection  efficiency
 is often sufficient to indicate the kind of equip-
 ment to be used, or at  least to narrow the choice
 down to between two or three types. Attention
may then be given to detailed  selection  of  a
specific unit.

5.10  Selection  of Collectors in General

  A number of aids are available  in under-
taking such preliminary selection studies. One
of these  is a chart prepared by Sylvan of the
American Air Filter Company, Inc., presented
by Kane,66 with a more detailed explanation of
its use given by Kayse.07 This chart shows the
collection efficiency to be  expected  from any
particular type of equipment operating upon a
stream of specified concentration (in grains per
cubic foot)  of  dust having  a given  mean par-
ticle size  (in microns). Typical ranges of par-
ticle size and  concentrations  encountered in
various practical situations are also shown.
  Stairmand"3 lists  the calculated  collection
efficiency of a number of devices on a standard
test dust, W.C.3 silica powder, which has about
the same size  distribution  as typical  fly ash
from a pulverized fuel  boiler.  A few values
selected from his table are listed in Table 11.
  Other   useful aids are the  check  list and
tables of Kane,64  the tabulation of First and
Silverinan,43 and  the general discussions by
Hedberg56 and Lapple.81 These indicate that
the following factors should all be  considered
in selecting the method  of  dust collection and
specific equipment to be used:

Mean  dust-particle size and range
Dust concentration or loading
Abrasive characteristics of  dust
Adhesive characteristics of dust
Bridging characteristics of dust
Fire or explosion hazard
Corrosive nature of gases
Volume of gases to be handled
Gas temperature and pressure
Condensable vapors present in gas
Possible  fluctuations  in  operating conditions
Collection efficiency required
Method  of  dust disposal or recovery
Collector size,  location, and space required
Collector cost,  initial and operating
Collector installation work required
Collector servicing and maintenance required
Need  for make-up air supply

 Kane's tables  summarize  the operating char-
 acteristics of different types of collecting equip-
 ment  with  regard  to  many of  these points.
 They  also show  the collector  types commonly
 used for a wide range of industrial processes.

-------
                  REMOVAL OF  PARTICULATE MATTER FROM  GASEOUS WASTES
                                                                                            47
TABLE 11—Collection Efficiency on W.C.3 Teat Dust l"
                   Overall    Efficiency   Efficiency
                   Efficiency  at 3 Microns  at 1 Micron
     Collector       (Per One)   (Per Cent)  (Per Cent)
Cyclone, medium
  efficiency  	 65.3       27         8
Cyclone, low-pressure
  drop  	 74.2       42         13
Electrostatic precipi-
  tator	 94.1       92         70
Fabric  filter 	 99.9       99.9        99
Spray tower 	 96.3       94         55
"enturi scrubber	 99.7       99.6        97
Sp
Ve
  A discussion of all of the factors listed in
the foregoing paragraph with reference to all
kinds of collectors is, of course, not within the
scope of this report. Such discussions for indi-
vidual types will be found in each of the reports
in this series prepared for the American Petro-
leum Institute.  The remainder of this section
will be devoted to the interpretation of such
problems in terms of the use of filters of vari-
ous kinds.

5.20  Selection of  Filters

  The general  place of filters in the  scheme
of dust collectors has been indicated in  Sect.
1.00.  They are characterized as high-efficiency,
moderate-cost devices capable of application to
a wide variety of problems by  virtue  of the
many different  kinds of filter  media available,
and especially useful on medium- to very small-
sized particles.  Assuming that it has been de-
cided to use a  filter, the  next question  to be
answered is:  What kind of filter shall it be?
The check list  given  in Par. 5.10 should be
repeated, this time with reference to the various
possibilities of filters. The order of importance
of these items  is  not necessarily  as they are
listed. Possibly, the first consideration  should
be given  to the range of dust concentration.
  There are two broad classifications of  filters
to be considered at the outset,  based upon dust
concentration.   If  this is less than,  roughly,
1 to 3 grains  per  cubic foot, an air filter such
as is discussed  in Par.  3.221  (stationary)  or
Par. 3.322 (movable) is indicated. If the con-
centration is  higher, a  "dust  collector"  filter
(industrial) is indicated.  This may  be  of the
fabric type (Par.  3.10), the fibrous bed type
(Par. 3.223 or 3.321), or the granular bed type
(Par. 3.21 or 3.31).
  The selection of air filters  is  discussed  by
May" and Rowe.103  The  small  unit or  panel
type, with either disposable or renewable media,
will be  preferred for  small  volumes  of  gas
where a minimum of attention is desirable, for
the collection of a dust which is not to  be re-
covered. For large volumes of air one  of the
automatic cleaning types will be more suitable,
especially if the filter has to be located in a
relatively inaccessible  place or  it is desirable
to reduce maintenance requirements to a mini-
mum.  Collection  efficiency is moderate, dust-
holding capacity high, and pressure drop low.
The dust-holding capacity may be  defined as
the weight  of dust accumulated  either  before
the pressure  drop across the filter reaches a
specified maximum or before its collection effi-
ciency drops below a specified minimum.  Rowe
recommends a maximum pressure drop  of 0.2
in., water gage, for  small  units (such as  a
home  furnace) and  0.5 in., water gage, for
heavy-duty installations (such as industrial fil-
ter banks).  Table 5 summarizes many  of the
selection criteria  involved.  As  the  name im-
plies, such filters are principally used in general
ventilation work for the  protection of  people
and machinery from air-borne contaminants.
  In the realm of dust collectors general discus-
sions of cloth filters are given by Stern l=8 and
Ebeling," and of porous  materials  by  Silver-
man UT. us  The general range of performance
to be expected from each type has  been cited
herein in the sections  dealing  with each one.
Extensive specific  performance test data are
cited for cloth bags, screens, and reverse jets
by  Dennis, et  al.M; and for unit collectors of
the screen,  bag, or air  filter type  by  Stern,
et aI.13T  For  other  types, less extensive data
are indicated in the references cited under each
type.  Such performance data must be at hand,
either  from the literature or from the  manu-
facturer, before a filter selection  can be  made.
  Often the choice between cloth filter and fiber
mat may be resolved by the ultimate destination
of the  dust.  It is almost impossible  to recover
dust out of a fiber mat for some re-use purpose.
Dust which has a re-use value must be collected
as a cake which can readily be freed from the
collecting medium.
  The  next step in filter selection is to  deter-.
mine the volume  of gas to be handled.  This
will, in turn, fix the total  area of filter needed
on the basis of the value  of lineal velocity V,
or filter ratio, appropriate to the type selected.
Typical values of  V  have been cited  in  the
description of each kind of filter  given herein.
As  consideration proceeds to the other  items
on the  check list, the size of filter and the space
it will  require can be kept in mind.

-------
48
AMERICAN PETROLEUM  INSTITUTE
  Particle size range  and collection efficiency
required to produce the desired effluent concen-
tration should be considered next. If legal fac-
tors such as air pollution code requirements are
involved, they must, of course, be investigated
and taken into account. If possible, an analysis
of the dust size distribution on a weight basis
should be obtained. Knowing the collection effi-
ciency as a function of particle size according
to the collection mechanisms involved, one can
then calculate the size distribution of the efflu-
ent  dust (or of that collected)  on  a weight
basis.  Stairmand l"'13S gives examples of such
calculations.  To do this, one must  also know
the  size  of fibers to be used in the filter  con-
templated.  The finer the dust,  the finer the
fibers  needed.
  If the size range is broad and includes an
appreciable amount of particles over  10 microns
in  diameter, the possibility of preceding the
filter by a primary collector of another  type
should not be overlooked. Coarser particles are
easily and efficiently removed by  less expensive
collectors, which will reduce the duty  of the
filter.  Two kinds of filters may also be used  in
series where the second is of a higher efficiency
than the first and is to be used for the ultimate
collection of  very fine particles (such as radio-
active dusts) which cannot, under any circum-
stances, be permitted to  escape  in the effluent
gas stream.
   Next, all the properties of the gas and dust
which will have an influence upon the filter
medium must be taken into account. Foremost
among  these is temperature.  The allowable
range of temperature for each of the various
fabrics,  fibers, and other media has been given
herein.  The  possible need for cooling apparatus
must be contemplated. Abrasive, corrosive, and
flammability characteristics will also influence
the choice of filter medium.
   Flexibility of operation may be very  impor-
tant in  some cases.  If  a plant is  to operate
 intermittently,  or  if there are likely to  be
surges in the dust or gas flow rates, provision
 must  be made to handle  the worst set of oper-
 ating conditions.  Filters which are primarily
 designed to  operate continuously under steady
 conditions on a regular cleaning cycle will ob-
 viously  not be appropriate.
   Consideration of all of the aforementioned
 factors  will  probably lead to a tentative selec-
 tion of  a  filter.  Before a  final  decision  is
 made, however, the warnings against overlook-
                   ing questions of location, accessibility, servicing
                   and maintenance  needs, installation problems,
                   and disposal methods must be heeded in order
                   that  costly mistakes be  avoided and smooth
                   operation  assured.  The size of the proposed
                   filter  installation,  including  auxiliaries, must
                   now be definite in order to plan on location and
                   layout.
                     Finally, the question of cost must be raised
                   and answered.  The  ratio of pressure  drop to
                   efficiency will  be  a useful criterion here as it
                   represents a ratio  of operating cost (in a gen-
                   eral sense) to unit of achievement.
                     There are many references in the literature
                   to individual problems connected with the selec-
                   tion and use of filters in specific industries, and
                   examples of how these  problems have  been met
                   in certain cases. Following is a representative
                   list:
                    Air-borne bacteria '"• "•8l
                    Coal dust °3
                    Fluid catalyst fines B2
                    Foundry and cupola fumes :o> 80
                    Hot gases "•"°
                    Metallurgical dusts *
                    Protection of electrical equipment •»•105'140'IM
                    Radioactive dusts •• «•ST


                     6.00  PROBLEMS IN PERFORMANCE
                                    TESTING

                     In testing the performance of a filter, there
                   are essentially three quantities to be checked:
                   pressure drop, gas flow rate, and collection effi-
                   ciency.  Pressure  drop  and flow rate are fairly
                   easy to measure by standard techniques. The
                   matter of collection efficiency, however, raises
                   certain problems.
                     A statement of the collection efficiency, or
                   per-cent penetration,  is meaningless without
                   an  accompanying description of the  method
                   of determination  employed.  Different methods
                   have been used, and they give different results
                   upon the same filter.  If, in the specifications
                   for  the  purchase of a filter, a statement or
                   guarantee  of collection performance  is to be
                   included, it will be very important to agree with
                   the  manufacturer upon  the method  by which
                   this performance is to  be rated. The statement
                   of the test  method  should  include a complete
                   description of the particle size distribution and
                   chemical composition of the  dust which will be
                   used for performance  testing.

-------
IV.  SO.

-------
Fluidized Bed Combustion Update



     The most  recent  research on fluidized  bed  combustion has developed



several possiblities for  improving  both  the  efficiency and emission con-



trol of coal-fired  power plants.   These  include primarily the method of



feeding the coal  to the  fluidized bed and the efficiency of the sulfate



sorbent.





Washed Coal Feed



     The standard fluidized bed combustion system has  a deep fluidized bed



and a tall  combustion  chamber  to ensure efficient combustion of finer coal



particles present  in  the crushed coal feed.   It has  recently been shown



that the  coal  does  not  have  to be  crushed  before  introduction  to  the



fluidized bed combustion chamber.



     The main  reason  for firing crushed  coal  is'the  need to maintain an



effective fluidized bed in the combustion  chamber.   The burning coal, ash,



and  SOp  sorbent must  remain  suspended  in the  upward airflow forming a



violently churning layer of particles.  However, during normal operation,



the fluidized bed contains only  a very small amount of coal (1-5 percent)



and consists mostly of ash and sorbent particles.



     To maintain effective fluidization,  particles in  the  bed  must be less



than approximately 5 mm  in size.  Coal, however, often contains rocks and



stones up to 2  in. in  diameter.   If these were introduced  to the combustion



chamber they would sink through the fluidized bed  and deposit on the  bottom



of the chamber.  The coal crushing process assures that the stone contained



in the coal remains fluidized.

-------
     If the coal  is washed,  however,  the more dense stones can be easily

separated from the coal.  Coal washing also removes much of the inorganic

sulfur contained in the coal, and reduces the amount of ash.  One-half to

1 in. size coal  can be  burned in  a fluidized bed as small as 15 cm high to

achieve a combustion efficiency of 97 percent.

     Fluidized bed combustion boilers  using  this  method of fuel feed are

currently in use in both the United States and Great Britain.

Sorbents

     Much current  research  is  being  devoted  to  the  improvement  of SOo

sorbent efficiencies and regeneration.  Currently,  limestone  is being used

as the SCU sorbent without  regeneration.  In this process,  the  limestone is

fed to the fluidized bed along with the feed coal.  The carbon dioxide is

quickly driven off or calcined,  leaving a porous CaO  structure (Figure 1).

The S02 -then  deposits  on the surface  of the remaining  stone  as calcium

sulfate (CaSO^).   Normally,  from 20 to  25  percent of  the calcium in the

limestone feed can be converted to calcium sulfate by this method.

     It recently  has  been   found  that the  limestone  efficiency  can be

improved  either  by adding  a small amount  of  sodium  chloride  with the

limestone or by hydrating  the calcium  sulfate in  a regeneration process.

Both of these processes increase  the pore diameter  and prevent sealing off

of  the  small micropores present in the nonmodified  calcined limestone.

(See Figure 1.)  Up to  50  percent  conversion  has  been  obtained by adding
               2
sodium chloride  to the limestone (see  Figure  2), while  85  percent calcium
                                                                 3
conversion has been obtained with successive cycles of hydration.

-------
                               -MACROPQHE3

                                  [-MICBOPORSS

                                    CALCIA GRAINS-:
                                 f- CALCIUM
                                   5ULFATE
                                  -CALCIA GRAINS

                            -MACROPOfiES             "-MLCIUM suLFATf
      Figure 1 Magnified cross-section comparison of limestone calcinalio.
      and sulfation, with and without salt (schematic)
        100
      u
      o
      h-
      a
      e
      U

      1
         20
Q RAW STONE

Q 0.3 .1 V. NaCI AOOmON

D 2.0 «1% NaCI ADDITION

           CiLCITE   ASL-9601 _ ANL.-950I

                •NL-9TOI
                          ANL-9ZO)  -ANL-S9OI - ANL-SIOl "
                         LIMESTONE aCSICKATICN

Figure -7, The effect of 0.5 and 2.0 wt % -NaCI treatment on limestone
subsequently precalcined and sulfeied at. 550 °C for 5 h-in 0.3% S02,
5% O2, 20% CO2. balance N2

           Q Bcriun  Titanste
           C Calcium Aluminate Cement
           S Conventional  Sorbent (Grove Limestone)
                                                        100
                                                                              2      J
                                                                         HYORATION CrCLC
                                                       Figure ?  Calcium conversion of the steady-stale product ax
                                                       a function o! hydration cycle. (Hydralion cycle 0 represent the
                                                       Initial sulfation).
 „   0.6
     0.5
     0.3
     0.2
                                                                                                    After
                                                                                                    Sulfotion
                                                                                                    After •-
                                                                                                    Regeneration
                                                 CYCLE
                          Figure '; Comparison of new regenerable sorbents and limestone

-------
     A major problem with  using  limestone as a sorbent in a regeneration
cycle, however, is a rapid  loss  or  attrition through erosion in the flu-
idized bed.   Also,  the  regeneration process  involves  high temperatures
which rapidly deactivate the limestone after only a few cycles.
     Two additional minerals have  shown  promise as possible SOo sorbents
                               4
in  the  fluidized  bed  process.     They  are  barium titanate  and  calcium
aluminate cement.   Both of  these  minerals  have  shown acceptable conversion
to sulfate as well  as erosion resistance.   Preliminary economic assessment
indicates that either of these may be feasible on a commercial basis.
     Regeneration of the SC^ sorbent  used in fluidized bed combustion  is
not currently  practiced.    However,  research indicates  that regeneration
will greatly reduce the solid waste  associated  with SO^ removal as well  as
possibly providing a commercially usable  sulfuric acid by-product.
    . Sci. & Tech.. Vol. 14, No. 3, pp. 270-288 (1980).
2Ibid., Vol. 14, No. 9, pp, 1113-1118 (1979).
3J.A.P.C.A.. Vol. 30, No. 6, pp. 684-688 (1980).
4Env. Sci. & Tech.. Vol. 13, No. 6, pp. 715-720 (1979).

-------
2.   The FGD System
 The R-C/Bahco facility operating
 at RAFB is a calcium-based
 throwaway FGD and paniculate
 removal system. Either pebble lime
 (CaO) or ground limestone (CaCOs)
 can be  used for S02 removal to
 produce mixtures containing
 calcium sulfite (CaSOa), gypsum
 (CaS04), and fly ash. The overall
 chemical reactions for the
 respective  reagents are shown in
 Table 1.

 The scrubbing system comprises
 the following major components:

 • Flue-gas-handling equipment
 • R-C/Bahco scrubber
 • Reagent-handling and -storage
   equipment
 • Sludge disposal equipment

 The entire  FGD system (Figure 1)
 is served by a centrally located
 control  room and is operated,
 part time, by heat plant personnel.


 Flue-Gas-Handling  Equipment

 The flue-gas-handling equipment
 includes a  flue gas header, bypass
 stack, mechanical collector, and
 booster fan. Flue gas from as
 many as eight stoker-fired hot
 water generators—up to 108,000
 actual ftVmin (51 mVs)—passes
 into the header and mechanical
 collector where coarse particulate
' matter is removed before it enters
 the booster fan and scrubber.
 Removing paniculate minimizes
 erosion of the fan and other
 scrubber components and reduces
 the amount of wet solids handled
by the scrubbing system. The ash
is disposed of via the existing
ash-handling system. A bypass
stack in the carbon steel flue gas
header serves two purposes: it
serves as a fail-safe emergency
bypass, and it permits air to enter
the system at low loads to
maintain gas velocity through the
mechanical collector and scrubber
to maximize collection efficiency.


Gas Flow

As shown in Figure 2, the
R-C/Bahco scrubber, which is
fabricated  from 316L stainless
steel, is a two-stage inverted
venturi unit specifically designed
to operate with slurries containing
calcium sulfite, calcium  sulfate
(gypsum), calcium carbonate,
calcium hydroxide, and fly ash. All
of the internal gas flow passages
are large, unobstructed, and well
irrigated with circulating  slurry or
makeup water to essentially
eliminate the possibility  of serious
plugging problems.

Hot flue gas from the booster fan
enters the first stage, where it
impinges on the surface of the
slurry, creating a cascade of
droplets that it carries into the
throat of the lower venturi. The
droplets, containing SC>2
scrubbing reagent, cool the gas to
its saturation temperature, absorb
sulfur dioxide, and trap particulate
                                 Table 1.

                                 Chemical Reactions for Lime and Limestone in SC"2 Removal
                                              Reagent
                                                                              Reaction
Lime:


Limestone:



Ca(OH)j * S02 — —
CaSO3 * ViO-2 	

faCn, *. I/,O, 	


	 *- CaSOj f H20




                 SOURCE:   "Capsule Report--Bahco  Flue  Gas Desulfurization and  Particulate
                            Removal  System,"  EPA 625/2-79-022  (July  1979).

-------
                                                                                          Thickener
  Reagent system
  module
  Reagent
  storage
   Reagent
   feeder and
   slaker
   Lime or
   limestone
   conveyor
   Unloading
   station —
                                                          Overflow
                                                          to lime-
                                                          dissolving
                                                          tank
Reagent-dissolving    Second stage
tank               PumP
Mill pump
Figure 1.

R-C/Bahco Scrubber System
matter. Above the first venturi, the
gas stream is turned downward
by the bottom of the pan in the
second stage venturi causing most
of the droplets to fall out. In the
second stage, or upper venturi, the
process of impinging the gas
stream on the  surface of a slurry
is repeated.  Here the gas/droplet
                   mixture passes up through the
                   throat of the upper venturi where
                   final SOi absorption and
                   paniculate removal are
                   accomplished. A cyclonic mist
                   eliminator above the upper venturi
                   imparts a spinning motion to the
                   gas stream, causing the droplets
                   to move toward the wall where
                   they coalesce and drain from the
                   scrubber. From the mist
                   eliminator, clean gas, which is not
                   reheated, enters the surrounding
                   atmosphere via the stack.
Slurry Flow

Two techniques of handling slurry
flow in the system are used to
eliminate or minimize the plugging
and erosion problems often
associated with calcium-based
FGD systems: maintaining
essentially constant slurry flow
rates through the scrubber, and
eliminating turndown in slurry
bleed streams by operating in an

-------
    Schedule:
        Gas flow
        Slurry flow
        Sludge
        removal
        Slowdown valve •* ^
Stack
                                                                       Manhole
                                                                                       Platform
                                                                                       Man door
      To reagent dissolver
                                                                                            Platform
                                                                                              Plarform
                                                                                        First stage drop collector
r
r
i
J «-r
                            Ground level
             £
Mill
pump
Figure 2.

R-C/Bahco Scrubber

-------
on-and-off mode with water
flushing after slurry flow is
interrupted.

Slurry flows by gravity from top to
bottom in the scrubber, counter-
current to the gas flow. Slurry
from the reagent dissolver {which
is also of 316L stainless steel)
contains makeup reagent—either
lime or ground limestone. The
slurry enters the pan in the upper
venturi. Slurry level in the pan
determines  the upper venturi
pressure drop; the level is set by
adjusting a  weir in the level tank
located outside the scrubber.

Slurry streams from the  mist
eliminator and the pan are
 combined in the level tank before
 flowing  by gravity to the mill
 under the lower venturi, where
 another level tank is used to set
 the pressure drop in the lower
 venturi. Part of  the slurry collected
 in the area between the upper and
 lower Venturis,  the part  that has
 contacted the gas stream twice,
 flows by gravity to the sludge
  disposal system. In the first-stage
  level tank this slurry is combined
  with overflow from the  mill and is
  returned to the reagent dissolver.
  More reagent is added in the
  dissolver before the slurry is
  recycled to the upper venturi. The
  fluid mill is powered by an
  external pump and is used to grind
  coarse limestone or other large
  particles in the system.
First-stage venturi, showing gas inlets and makeup water spray
manifolds
Reagent-Handling and -Storage
Equipment

The reagent system installed at
RAFB is capable of handling both
0.75-inch (1.9-cm) pebble lime
and 200-mesh ground limestone.
Primary components include
truck-unloading equipment, a steel
silo with 3 weeks' storage capacity
at winter load conditions, a weigh
belt feeder, and a lime slaker. The
silo, feeder, slaker, and reagent-
dissolving tank are integrated into
 a single module to minimize
 materials handling, supports, and
 space requirements. Lime or
 limestone drops directly out of the
 silo into the feeder-slaker and
 overflows into the reagent-dissolv-
 ing tank directly  under the slaker.
Sludge Disposal Equipment

Calcium sulfite, gypsum, and fly
ash collected in the scrubber are
concentrated from 10 percent to
approximately 40 percent solids
(by weight) in a thickener. The
overflow from the thickener is
returned by gravity to the
reagent-dissolving tank. The
underflow from the thickener is
pumped underground to a
hypalon-lined storage pond.

-------
3.   The Test  Program
The R-C/Bahco FGD system test
program, carried out at RAFB
between March 1976 and June
1977, incorporated the following
categories:

• Material balance
• Lime reagent process variable
• Lime reagent verification
• Particulate collection efficiency
• Limestone reagent process
  variable
• Sludge characterization
• Scrubber reliability monitoring

Material balance tests were
conducted to establish the range
of operating conditions over which
the R-C/Bahco scrubber could be
operated and to verify performance
at design conditions by completing
material balances. Maximum and
minimum gas flow rates, pressure
drops, and slurry circulation rates
were determined and preliminary
SOz and the paniculate
performance data at the limits  of
the system's capabilities were
obtained. The system was
operated at the design gas rate of
50,000 stdftVmin (25 normal
m3/s) and complete material
balances on calcium, sulfur, and
total solids were performed.

Statistically designed lime process
variable tests helped to establish
the quantitative effect of the
following process variables on  SC>2
removal: gas flow rate, first- and
second-stage pressure drops, mill
and second-stage slurry rates,
lime:S02 stoichiometric ratio,
slurry inventory, and slurry solids
concentration.

Lime reagent verification tests
were undertaken to verify the
results obtained in the lime
process variable tests, and to
determine the effect of very dilute
scrubber slurry (2 percent solids)   .
on system performance.
 Particulate collection efficiency
 tests were a continuation of the
 paniculate tests initiated during
 the earlier sampling phase.
 Relationships were determined
 between system variables,
 including  panicle size distribution
 and particulate removal efficiency.

 Limestone process variable tests
 were completed using the same
 statistically designed test plan
 used for lime. The effect of system
 variables on SOa removal
 efficiency  and reagent use was
 determined.

 Sludge samples generated at
 RAFB were tested to determine
 dewatering, transport, and
 disposal characteristics (sludge
 characterization).  Samples of
 sludge from lime  as well as
 limestone  scrubbing were tested.

 The R-C/Bahco system was
 monitored from March 1976 to
 June 1977, to document its
 operating  and maintenance history
 and to obtain data for a cost
 analysis. Data were gathered on
 reagent, coal, water, and power
 consumption as well as on
 operating  and maintenance labor
 requirements.

Throughout the test program
 samples were taken of slurry, flue
gas, lime,  limestone, and coal,
often in duplicate, for chemical
analyses, particulate loading, and
particle size distribution. A
field analytical laboratory was
established, and especially
developed  and highly efficient test
methods using thermogravimetric
analysis were employed
extensively.

-------
4.   Test  Results
 Capacity/Material Balance Tests

 Performance of the size 50
 R-C/Bahco scrubber  at RAFB is
 measured by its ability to  handle
 variations in system operating
 parameters while reducing S02
 and particulate  emissions to the
 limits allowed by the  applicable
 regulations, without exceeding the
 capacity of the system. Regulations
 applicable to RAFB limit S02
 emissions to 2.2 lb/106 Btu (3.96
 g/106 cal) and particulate to
 0.16 lb/106 Btu (0.29 g/106 cal).
 Table 2 lists maximum, minimum,
 and optimum operating levels
 determined for the system at
 RAFB. The cost of reducing
 emissions to meet requirements
 will be minimized at optimum
 operating levels.


 Lime Tests

The SO? removal capabilities of
the R-C/Bahco  system using
pebble lime were characterized in
two steps. First, a series of
screening tests determined the
effects of slurry rates, gas rate,
venturi pressure drops, slurry
density, system  volume, and lime
stoichiometry on S02 removal.
Tests results indicated that lime
stoichiometry—the ratio of lime
feed in the system to SOa in the
flue gas—was the only variable
controlling  SC>2 removal as long
as the system  was operated within
the limits outlined in Table 2.

A second group of tests, in which
the effects  of the gas flow, slurry
rates and slurry density were
determined, confirmed the initial
findings that stoichiometry alone
controlled SC>2 removal.

Results of these verification tests
are shown  in Figure 3. The figure
also shows that lime use is
essentially  100 percent—that is,
no excess lime is needed—up to
90 percent  S02 removal.  Figure  4
illustrates system performance
when S02  removal is above
90 percent—that is,  when S02
emissions at RAFB were reduced
below 0.6 lb/106 Btu (1.08 g/106
cal). The figure indicates that over
98 percent  of the S02
corresponding to 0.1 Ib S02/105
Btu (0.18 g  S02/106 cal). can be
                                 Table 2.

                                 R-C/Bahco Scrubber Operating Levels
Variable
Gas rate (actual ft3/min| 	
Slurry circulation rate (gal/min) . . .
Veniuri pressure drop for each
stage (inches H20) 	
Slurry concentration (wt %
solids) 	
ReagentSOj stoichiometry
(moles reagent: moles SO}.
based on inlet S02 levels):
Lime 	
Limestone 	
S02 removal efficiency (percent):
Lime 	
Limestone 	
S02 emission (lb/106 Btu):
Lime 	
Limestone 	
Particulate emission (lb/106
Blu) 	
Minimum
35,000
1.500

6

2



0.45
0.55

45
40

3.7
4.0

0.2-0.3
Maximum
55.000
3.000

12

25



1.05
1.2

98*
85

0.1
1.0

0.14
Optimum
40,000-50.000
2.300

7-10

10



0.7
0.75

70
70

2.0
2.0

0.16

-------
achieved with a stoichiometry of
1.1—that is, 10 percent excess
lime. The SC>2 emission rates
shown in Figure 4 are well below
the required 2.2 lb/106 Btu
(3.96 g/106 cal) and the guarantee
level of 1.0 lb/106 Btu (1.8
g/106  cal).

From the lime tests  it is concluded
that lime:SO2 stoichiometry is the
controlling factor in  determining
S02 removal efficiency. Virtually
any desired S02 removal
efficiency can be achieved when
lime is used in the R-C/Bahco
scrubber, simply by  adjusting the
lime:SC>2 stoichiometry. Lime use
approaches 100 percent at
stoichiometric ratios up to about
0.9. At stoichiometric ratios up
to 1.1,  producing up to 99 percent
removal,  lime use is above
90 percent. Because most S02
regulations for industrial boilers
permit  emissions in  the range of
1.0 to 2.0 lb/106 Btu (1.8 to
3.6 g/106 cal), lime, with its high
removal capabilities, can be used
to obtain offset credits in  a
nonattainment area  to apply
toward an expansion or new
facility. No further capital
expenditure need be made,
because the R-C/Bahco system
normally would  be designed to
handle lime as well  as limestone,
and switching from  limestone to
lime will  increase  the annual
operating costs  only by about
1 5 percent.


Limestone Tests

System performance with ground
limestone was determined in a
series of  screening tests very
similar to those  used for pebble
lime. These tests indicated that
slurry circulation in  addition to
limestone stoichiometry controls
S02 removal efficiency.
       100
    o
    LU
        20  -
                0.2     0.4     0.6      08     10     1.2

                 LIME STOICHIOMETRY (moles lime per mole (S02>
Figure 3.

SOj Removal Efficiency as a Function of Lime Stoichiometry
                                   Hypalon-lined storage pond

-------
Figure 5 shows the results of the
tests and gives limestone use
data. At a stoichiometry of 1.0 and
slurry circulation of 2.300 gal/min
(0.14  mVs), slightly over
80 percent S02 removal is
possible with 80 percent limestone
use. A practical limit for  limestone
is 80  percent S02 removal,
because higher removals result in
substantial reductions in limestone
use.

Operation with limestone at RAFB
produced sludge that contained
much more gypsum than did
operation with lime. That is, there
was more oxidation of CaSOs to
CaS04. Table 3 shows an average
gypsum (CaS04- 2H20) and
calcium sulfite (CaSOa- '/zH20)
content of 33 and 55 percent.
respectively, when lime was used.
The limestone slurry was almost
completely oxidized and contained
78 percent sulfate and less than
1 percent sulfite. The comparison
of average lime and limestone
slurry analyses during similar
boiler load periods listed in Table 3
indicates that the oxidation trend
is probably attributable to the
lower slurry pH encountered when
using limestone, because all other
operating conditions were
essentially the same.

Paniculate Removal Efficiency

Paniculate Removal Tests. Initial
paniculate removal tests on the
R-C/Bahco scrubber, performed in
March, April, and May of 1976,
revealed the presence of
substantial amounts of soot in the
stack gas. The average paniculate
emission rate for these tests was
0.23  lb/106 Btu (0.42 g/103 cal).
    cr

    2
    v>
    CO
    5
       2.5
    I  2.0
    §  1.5
    Ul
       0.5
    O
    in
                              RAFB EPA limit
                           Guarantee emission rate
                                        90% lime use
         0.7    0.8    0.9     1.0    1.1    1.2     1.3    1.4     1.5

                 LIME STOICHIOMETRY (molss lime per mole S02)-
Figure 4.

Relationship Between S02 Emission Rates and Lime:S02 Stoichiometry
Overall paniculate removal   '
averaged 93 to 94 percent. Ohio
emission standards require an
overall removal efficiency of
96 percent at a paniculate inlet
loading of 1.5 gr/stdft 3 dry
(3.4 g/normal m3 dry) to achieve
an emission rate of 0.16 lb/106
Btu (0.29 g/106 cal). Venturi
pressure drops were increased to
nearly double the design value of
7 inches (18 cm) H20 to reduce
these emissions. Below
approximately 18 inches (46 cm)
H20 total pressure drop,
particulate emissions increased
rapidly. The amount of soot
present in the flue gas at RAFB is
higher than in other stoker-fired
generators similar to the
Rickenbacker boiler.

The Air Force has undertaken an
extensive program to upgrade the
heat plant at RAFB.  Data  obtained
during this test program
contributed substantially to
information used to plan the
upgrading program, and so far the
following modifications have been
completed:

. Installation of a new 60-Btu/h
  (18-Watt) generator to replace
  the two old units
• Replacement of hot water
  distribution piping
• Installation of flue gas oxygen
  monitoring equipment
• Repair  of firing air distribution
  equipment and fire box pressure
  controls in the generators

-------
• Rebuilding mechanical collectors
  and induced draft fans on the
  generators
• Replacement of burned out
  ledge plates, which regulate
  combustion air flow around the
  grates
• Repair of traveling grates

The problem with soot at RAFB
points up a critical aspect of a
successful emission control
project—namely, that proper
operation of all equipment, boilers
as well  as the scrubber, is
essential to maintain satisfactory
emission levels. Inadequate
combustion or inadequate air can
be as detrimental to emission
control as improper scrubber
operation.

Slurry Entrainment and Gas
Bypassing. During the paniculate
tests, two phenomena were
observed when the system was
operated above its capacity limits.
The first, called entrainment,
occurs  at very low venturi
pressure drops—that is, under
6 inches (15 cm) HjO involves
small droplets of slurry carrying
through the second-stage mist
eliminator and out the stack. The
second, called bypassing, is
characterized by pulsations in the
gas flow through the scrubber; the
result is low collection efficiency
in all particle  size ranges. The
second  phenomenon takes place
when relatively high pressure
drops—that is. 12 inches (30 cm)
H20 or more in either venturi—are
coupled with slurry flows under
150 gal/min (0.01 m3/s) to the
scrubber.

Conclusion. The particulate
removal efficiency of the
R-C/Bahco scrubber is comparable
to that of low energy venturi
scrubbers for particles larger than
1 u.m, and appears to be better  for
particles smaller than  1pm. In  an
R-C/Bahco scrubber, the second
stage is the primary collector of
fine particles. Slurry carryover  and
     c
     3)
     u
    U
100


 90


 80


 70


 60
    t   50
         40
    O
    U)
         30
         20
         10
                0.2   0.4   0.6   0.8    1.0    1.2
                            STOICHIOMETRY RATIO
                                                 1.4
                                                            1.8
Figure 5.

862 Removal Efficiency as a Function of Limestone:S02 Stoichiometry
Table 3.

Lime and Limestone Slurry Analyses
Slurry solids
CaS04 2H,0 	
CaSOs-'/iHjO 	
CaCO3 	
MgCOi 	
Acid insolubles 	

Total 	

Lime
Slurry
(wt%)
	 33 4
	 54 5
	 3.7

	 4.6

	 96.2

Limestone
Slurry
(wt%)
77 5
1 0
17.3
0.8
3.4

100.0


-------
gas bypassing limit paniculate Table 4.
collection in an R-C/Bahco
scrubber operated outside the Dewatering Test Results
levels shown in Table 1 	 	
Particulate emissions from
stoker-fired coal-burning Tesl
equipment ran HP rorjurpd to
levels required by regulatory
formation is prevented.
Centrifuge 	
Sludge Characterization
Filter leaf 	
A series of scrubber sludge

Slurry
type
I Lime
\ Limestone9
I Lime
( Limestone
I Lime
I Limestone

Feed
solids
(wt%)
162
16 7
16.7
263
18 A
37.4
24.6
41 *i
37.4

Final
solids
(wt%)
44
CO
58
51
65
58
CQ
74

Rate at
35 percent
solids
22 Ib/d/ft2
578 Ib/d/ft2
70 lb/h/ft2
1 24 lb/h/ft2
64 Ib/h/ft2
 out at the Research-Cottrell
 laboratories to:

 • Determine scrubber sludge
  dewatering characteristics
 • Evaluate transportability of
  dewatered sludge
 • Determine physical/structural
  properties of dewatered sludge
 • Measure sludge leachate
  properties

 Slurry Dewatering. A series of
 settling, centrifuge, and filter leaf
 tests was run on lime and
 limestone slurry samples. The
 results are summarized in Table 4.

The settling tests showed that
limestone slurries settle more
rapidly and produce denser settled
layer than lime slurries.
Flocculation improved the settling
of limestone slurries, but not
that of lime slurries.
 aWith 5 ppm flocculant.



Table 5.

Sludge Leachate Analysis
Analysis
TDS(mg/l) 	
S0a[mg/l) 	
C00(mg/l) 	
Cl (mg/l) 	
Pb (ppb) 	
Cd (ppb) 	
Cr (ppb) 	
Hg (ppb) 	
Lime
leachate
9 Qfifi
i flin R
ft 4

^ino

-------
The centrifuge tests indicated that
final cake density increased as the
solids concentration in the feed
was increased, and that limestone
slurries produced higher cake
densities than did lime slurries.

Filter leaf tests showed that
limestone slurry filtration rates
were significantly lower than lime
slurry rates. However, limestone
again produced a denser cake.

Leachate  Tests. Leachate tests
were performed on  samples of
lime and limestone  sludges. The
results are listed in  Table 5.

Leachate compositions from lime
and limestone sludges are
essentially the same. Total
dissolved solids (TDS) in the range
of 2,500 to 3,000 mg/l and sulfate
levels of 1,600 to 1.800 mg/l
indicate that the leachates were
saturated  with respect to CaSCU.
Both sulfites in the  sludge and
organic matter in the fly ash
contribute to the chemical oxygen
demand (COD) levels observed.
Although the chloride level in the
lime leachate is somewhat higher
than the limestone leachate, the
other trace elements are present
in similar concentrations in both
leachates. The constituents found
in these leachates are similar in
type and concentration to those
reported in other studies. If a
disposal site is placed so as to
avoid infiltration of leachate into
ground water,  and if sludge and
soil cover  are placed properly to
avoid excessive contamination of
runoff, leachate from these
sludges will not present an
environmentally unacceptable
disposal problem.
First-stage level tank

-------
5.   Operating
Experience
  Since startup in March 1976, the
  R-C/Bahco system has performed
  well in all areas essential to
  successful FGD, including:

  •  S02 removal
  •  Paniculate removal
  •  Scrubber reliability
  •  Minimal routine maintenance
  •  Moderate operating costs
  •  Ease of operation

  During the test period of about
  11,000 hours, the scrubbing
  system operated for 6,194 hours.
  The operation is summarized in
  Figure 6. It is of interest that from
  December 1976 to February 1977
 (when severe winter weather was
 encountered), no outages  resulted
 from failure of auxiliary equipment.
 There were a few brief shutdowns
 caused by frozen air and water
 lines during this period, but
 system availability was over
 95 percent.

 Downtime is summarized in
 Table 6. This table shows the
 amount of time required to obtain
 parts as well as the actual time
 for repair work. Spare parts were
 not kept on hand during the test
 period, and this resulted in
 substantial unnecessary
 downtime. Since completion of the
 test program, a full supply of
 spares has been procured.  Table 6
 also shows that booster fan repair
 time accounted for 90 percent of
the downtime caused by repairs.
The booster fan operates on the
inlet side of the scrubber.
 downstream from the mechanical
 collector, and handles only hot dry
 flue gas with moderate amounts
 of fine fly ash. Modifications to the
 fan wheel and bearings, completed
 in May 1977, have eliminated the
 recurring failures associated with
 this piece of equipment.

 Total downtime, exclusive of fan
 repairs and procurement, was
 1,845 hours, or 17 percent of the
 test period.  During routine
 operation, the system availability
 should be over 95  percent, based
 on the factors observed during
 the test program.

 Scrubber  inspections were an
 integral part of the program to
 monitor scrubber performance. A
 thorough internal inspection was
 made in April 1976, approximately
 1 month after startup. A followup
 inspection was  made 2 months
 later, with subsequent inspections
 during outages  up to the end of
 the test program in June 1977.
These inspections confirmed the
effectiveness of the water makeup
system in keeping key areas of
the scrubber clean.

-------
Accumulations of solids were
detected at seven locations within
the scrubber (Figure 7).
Accumulations in four areas—1, 5,
6, and 7—had no impact on
scrubber performance. Problems
of solids buildup  in Areas 2, 3,
and 4 were easily corrected, as
follows:

In the first few months after
startup, the first-stage venturi
overflowed into the inlet manifold,
Area  2, resulting in an
accumulation of dried slurry in the
bottom of that area. Subsequent
investigation revealed that
operation of the first-stage at
pressure drops above 12 inches
(30 cm) H20 coupled with a
second-stage slurry pumping rate
more than 50 percent  higher than
the design rate of 2,600 gal/min
(0.16 mVs),  caused flooding when
the gas flow was reduced below
35,000 stdftVmin (17 normal
m3/s) or the booster fan was
shut down. This problem was
eliminated by decreasing the
speed of the slurry pump to
reduce the flow to design levels,
and by adding an interlock to
stop the pump when the booster
fan is shut down. The accumulated
material was removed  during
subsequent heat  plant  outages.

Areas 3 and 4 were affected twice
during the test by accumulations
of a coarse sandy material. The
first incident, which occurred
shortly after startup, was caused
by inadequate removal of grit from
the lime slaker. The material in
the pan was removed and the
slaker was readjusted to eliminate
the problem. The  second
accumulation took place during
the winter of 1976-77  when the
air lines, which activated the
blowdown valves on the first- and
second-stage level tanks, froze and
rendered these valves inoperative.
     Scrubber     Replace slurry  Scrubber   Replace slurry
     startup      pump lining    inspection  pump lining


'i^^g&H&&fa:?& I ;r^May-,:4/-.11
                             1976
                                                          Install
                                                          sludge I
                                                          thickener
                                                          rake
            *         t
          ReplaC8   Correct
          'orque    fan v,bratlon
          hmiter
                                                        Repair
                                                        sludge and
                                                        slurry lines
    ly^Uury;^!J'v^gAug. •»&&• { f
                                         "*£'r-i.
m
                                 1976
     Install improved
     booster fan bearing
                                            Replace
                                            blowdown
                                            valves
                              1976/1977
                 Inspection  Repair and
    water booster   and grit    modification
    pump bearing   removal    of fan wheel
                                      Replace
                                      slaker
                                      motor
                                   '* ••' X"M~V ' '  '-".^'^V^i-'
                                   '•&%§!::Mtif •• •$•&?*•
   Scrubber  F  1
   operabilitv &>$n


   Boiler      E^
   shutdown  I
                                 1977
Figure 6.

Downtime Related to Auxiliary Equipment
Table 6.

Downtime Summary
                                           Hours
                                                             Percent

Booster fan 	
Thickener 	
Slurry pump ....
Water booster pump
Lime slaker 	
Modifications 	
Routine maintenance
Loss of utilities 	
Miscellaneous 	

Total
Procurement
	 • 514
471
252
190
	 122



	 56

1.605
Repairs
2 252
8
18
16
1 1



53

2.363
Downtime
2 766
479
270
206
133
388
139
116
1 14

4.611
period
25 1
4 4
2 5
1 9
1 2
3 5
1 3
1 1
09

41 9

-------
The results of this pan of the test
program demonstrated that there
are situations that can result in
deterioration of scrubber
performance, including:

• Infiltration of grit into the system
  through the lime slaker
• Inadequate operation of the
  scrubber blowdown valves
• Slow accumulation of solids in
  the straightening vanes in the
  stack

The infiltration of grit can be  kept
to a minimum by paying close
attention to the operation of the
lime slaker grit removal circuit.

The blowdown valves should be
operated two to four times a  shift,
depending on scrubber load,  to
avoid accumulations of solids in
the slurry outlets.

The straightening vanes at the
base of the  stack, which serve
only to minimize spin in the gas
stream  leaving  the scrubber,  may
accumulate some material and
should be checked twice a year.
The possibility of accumulations
taking place can be minimized by
operating the scrubber within the
limits outlined in Section 3 to
avoid slurry carryover.  Obviously,
elimination  of the vanes would
prevent the problem entirely, but
accurate outlet particulate
sampling  would then be difficult.

The RAFB operating experience
indicates  that there are no
significant problems related to
the accumulation of solids in the
R-C/Bahco system. The scrubber
can tolerate substantial
accumulations of solids resulting
from  external operating problems
before performance  is  adversely
affected. Any deterioration  in
performance that does occur is
gradual and can be rectified  at a
convenient  time.
      Stack
     Manhole
6
*-
             J
Platform
  Mist eliminator
    Removable
      cover
                  Man door
                       Platform
                         Platform
                   First stage drop collector
                                    Figure 7.

                                    R-C/Bahco Scrubber Module

-------
                                   E. C. McKemie,  WoodaU-Duckham Ltd., Babcock & Wilcox Croup
 Q Today, many plant operators must specify new boil-
 ers that will have a life expectancy of 30 years or more.
 Which fuels will be available for these plants, particu-
 larly in their later years of operation?  What environ-
 mental restrictions will be imposed on emissions?
   The trend is toward the prohibition of oil and natural
 gas  as boiler feeds. This  has come  about via either
 legislation or problems with  fuel availability and cost.
 Let  us, therefore, assume that the fuel for these boilers
 will be coal. We will first examine options on the design
 of a boiler  and its fuel-preparation and firing equip-
 ment, using the technology available  prior to  fluid-
 ized-bed combustion. Then we will compare this tech-
 nology to that of a  fluidized bed.
   Assume that the boiler output is greater than 400,000
 Ib/h of evaporation, so that we will  be considering a
 pulverized fuel-firing system.
   For a conventional boiler, we must  next decide what
 type of coal will be used, since a boiler can handle only
 a  limited  range of fuel.

 Bituminous-coal-fired  boilers
   Fig. 1  shows a sectional  elevation of a 300-MWe
 (megawatts, electrical) boiler designed for firing bitu-
 minous coal. The coal usually is ground into particles,
 70% of which are less than 75 fim, typically in a vertical
 spindle mill. The mill is swept with sufficient heated air
 to dry the coal and convey the ground material directly
 to the burners. Power consumed in this process is about
 18 kW/long ton of coal fired, i.e., the mill power plus
 the fan power.
   Furnace volume and shape are determined by the
 heat absorption required to meet  the  design-furnace
 exit  gas-temperature, if there is no concern about NOX
 emissions. This  temperature  is set below  the coal-ash
 melting point to prevent slagging on subsequent con-
 vection banks.
   The size and number of burners are chosen to provide
©  1973 Woodall-Duckham L:d	
 complete combustion within the furnace. A typical such
 pulverized-coal burner is shown  in F!g. 2. Primary air
 and pulverized coal enter a tube fitted concentrically in
 a cylindrical  secondary-air-rcgister. The amount  of
 swirl given to  the secondary air  is controlled by vanes
 on the periphery of the register. A smaller tube, located
 in the center of the coal/primary-air inlet, has an impel-
 ler that diverts the coal into the secondary-air stream.
' Inside this small tube is an oil burner for lighting up the
 coal flame, and a gas torch for igniting the oil burner.
 Both are withdrawn to a cool zone after they have been
 used.                                             ;;
   The impeller acts  also as a bluff-body stabilizer (a
 nonstreamlined object placed in the gas stream  to in-
 hibit flow) along with the flow-reversal mechanism set
 up by the swirled secondary air.  The burners are posi-
 tioned on the front wall (or on the front and rear wails
 for large boilers), so as to yield a  near-uniform heat
 release over the furnace  plan-area  and  minimize ash
 deposits on the walls.
   The turndown on a mill and its associated burners is
 limited to approximately 5Q"o by the need to (1)  main-
 tain velocity in the fuel pipes to  prevent fallout  of th;
 coal, and  hence plugging  and (2)  restrict maximum
 velocity so as to control erosion. Hence, the boiler load
 is controlled by varying the number of mills and  burn-
 ers in  service. For safety,  the oil burners arc lit usually
 during load changes or when mills are put in or out of
 service,  and are  maintained in  operation  when  the
 boiler is on low load.
   The spacing  of the platens is fixed to ensure that sb:
 accumulations  will not  cause blockages. (If the platens
 are too close together,  slag will  bridge the spaces be-
 tween them.)  The  velocity through the convection
 banks is limited by the abrasive properties of the 2$-"
 particles, in order to  prevent erosion of the cubes. Thi
 degree of abrasion with  any given ash is proportional :o
 the cube of the particle velocity. Normally,  the  boils-'
 can accept a wide range of ash properties in bituminous
116
                                       CHEMICAL E.NCINEEKLSC ALT.L'ST l+. I9i«

-------
                                                                         i	Secondary superheater
                                                                                   	Plaren
                                                                                    superheater
fter setting the stage with a review
'the operation of conventional
ial-fired boilers, this report discusses
e use of fluidized beds, including
pics such as  coal preparation, load
ntrol and emissions control.
                                              Conventional 300-MWe boiler that handles.,,
                                              bituminous coal affords complete combustion  • Fig. 1
s^ "-
PJL1
1
V
I

i-.^'r.iV.'-^v
_^"— ^ ^"^"
-• .-V
•..-"••-•^•x


-Secondary-air
**» -vanes-". •, . .'
. v. ^L. A, :«.•". ;
. »" J . ' *: •'. "."
-.. .*»-•.
••- "^ir"^::^-!*.".^'^ -V." /"- ^*' ^1?*'*-
1- .^•rj ^; "*^*"- ;"--AT>iT" -,?- • U-
^SSsgSK^SR^i ^ ;
I
I
V;
^
  ^•ner for firing pulverized bituminous coal can be withdrawn from furnace  :' , -        : -  :  *':?£&&£&  2
                              CHEMICAL ENGINEERING AUGUST 14, 1978
                                                                                 117

-------
                                                          FLU1DIZE BEDS
                        Secondary   Secondary
                       superheater    raheater
 Primary
superheater
r£ -x.*:.i. Designed for low-volatility, pulverized coaly':'- 'f^v-''l?5r3:A;
V:.^'-JP',;-300-MWe boiler requires high'miIf-poweK'.rVv-^y,f j9-.?.-'
coals. If a much different rank of fuel is used, such a
boiler and firing system are no longer suitable.

Low-volatility coal  boilers
  Figs.  3 and 4 show the boiler (also 300  MWJ and
burner systems for low-volatility coal. Here, one can no
longer use a turbulent burner. To assist ignition, the
coal is ground typically into particles, 85% of which are
less than 75 pm. Compared with  bituminous coal, the
mill power-requirement is increased by about 50%. The
burners are set in the roof of the combustion chamber.
Primary air and coal are fired through  rectangular sloes,
and the secondary air through adjacent slots. The rela-
tive velocities are designed so  that once ignition is es-
tablished, the secondary air mixes with the fuel at a rate
that supports combustion  but  does not quench it. The
absolute velocities are designed to yield long flames that
sweep down into the  lower part of the combustion
chamber and then  turn up to enter the center pass at
the convection banks.
  Mill and burner turndown are  the same  as  for bitu-
minous-coal  systems, but due to the relatively poor
reactivity of low-volatility coals, oil support is needed
earlier than it is with bituminous coal as the boiler load
is reduced from its normal maximum rating. These
supporting burners are  often placed in the side walls:
  Fig. 5 shows a 200-MWe boiler designed for Austra-
lian lignite. This fuel has  a moisture  content  of about
70%, but when dry is extremely reactive.  Usually, 99%
of the particles must be less than 1  mm, reqmring a mill
power input of  about 2kW/long ton. Also,  the coal
leaving the mill must be dried to about 20% moisture to
ensure stable ignition  in  the  boiler.  To achieve this,
gases from the furnace at  about 1,000°C pass through
the mill with'the raw-coal feed.                      -
  A hammer mill with a water-cooled shaft is the only
type of mill suited for these conditions  (see Fig. 6). Since
                Air supply for
                 oil burners'^
             Lighting up
                and
             stabilizing.*
             oil burner
          j
          i
          ! Flame detector-"
           ,Pulverized fuel and-
            primary air ports

           ^xSecondary-air
           '   connections

              xSight door
                                                                                      Primary air
                                                                                       and coal
                                         Sida elevation
                                                                                      Section A-A
           Nonturbulent burners for low-volatility coal. Units are set in chamber's roof
                                                                 Fig. 4
         118
                                                 CHEMICAL ENGINEERING AUGUST 14, 1973

-------
                                                                       Coal and heating
                                                                         flue-gas inlet
                                      m$£  m
                                      &Mfti -..--.,
          !i:;; •"!l-I, I'r't !'•!»'•':.
          i,;'':y , ':  !'• I !•!' -I'i •
          BllSogi
          ilijpV^.vvii1"
         WV. boiler for Australian li
i   *"5h is dried by flue gasMto.20%moistwir^V^5H^S
; Hammer mill with, water-cooled shaft is .>;v;v.i>^.r.
^usedto grind lignite for boiler in Fig. 5 . -S^^^FJg; 6
    relatively coarse product is acceptable, a simple classi-
     r can be used to return oversized panicles to the mill
     ses leaving the  mill are heavily  laden  with water
  vapor. The fuel and gases are partially separated in the
    ict carrying the coal to the burner, which  is a series of
    >nzontal  slots alternately  containing fuel-and-eas
  mixture and air.
   •A fuel-rich mixture is fired in the lower slot  of the
    irner, as shown on Fig. 7, to improve ignition stabil-
    - i he combustion chamber may be in the form of an
  octagon, with  one  burner,  served  by one  mill,  firing
    » of the  eight sides. The combustion  chamber is
    ge compared with chambers for higher-grade  coals
    Ihe above brief survey on pulverized-coaJ firing il-
  lustrates the variety of boiler, fuel-preparation and
    ng systems necessary  to suit the grade of the coal.
    nc of these systems can control  SO, emissions and
  "'is must be  done separately.
   Also, because of its highly turbulent burner and con-
    aently  high  flame temperature,  the  bituminous-
    1-nred boiler produces high NO, emissions. If NO  is
 » be controlled, whether by  two-stage  combustion
    -gas  regulation, or both, then the  combustion-
    mber volume becomes much greater and the boiler
                                     -->.*.?%Wu.iL^l*^te(.j^ii1i<^.^>^i^XJi*j|?
                                                Secondary-air
                                                 'nozzles
   Coal/flue-gas
 mixture from mill
Burner for separation firing of lignite   -,
alternates layers of air with coal/flue-gas
                                             Fig. 7
                                        CHEMICAL ENGINEERING AUGUST 14. 1978
                                                                                                           119

-------
                                               FLUIDtZE BKDS
more expensive. Anthracite and lignite boilers are likely
to produce much less NOX because of their lower flame
temperatures and, in the case of the lignite boiler, be-
cause of the high percentage of water  in the fuel.

Fluidized-bed combustors
   With fluidized-bed combustion, on the other hand, a
boiler and its fuel-preparation and Bring equipment
can be designed to suit any type of fuel, whether solid,
liquid or gaseous. This is provided that the fuel has a
net calorific value greater than that required  to heat
both  it and the makeup air to the bed temperature.
Obviously, fuel-handling systems must suit the various
forms of fuel, as well as have large enough capacities for
the lowest-caloriftc-value fuels foreseeable.
   The major advantage of fluidized beds, therefore, is
that  they allow the development  of standard-design
boilers and combustion systems to a degree not  possible
before. Fluidized beds also permit the use of low-grade
coal,  even at low loads, without the need for expensive
support fuels to ensure stable ignition. Another advan-
tage is that the relatively low combustion temperature
results in low NOZ emissions and provides optimum
conditions  for the retention of sulfur by limestone or
dolomite in the  bed.

Principles of  fluidized-bed combustion
   The bed, which  may be sand, firebrick, limestone or
coal ash, is fluidized by blowing evenly distributed air
through it. Both bed material and the coal fired  to it are
crushed to a size compatible with the chosen fluidizing
velocity, i.e., a size that minimizes  loss by elutriation.
The fluidizing velocity is calculated from the quantity
of air at bed  temperature and pressure, and from the
bed's superficial area:

                  W    Tb + 273   J_
             "' ~ Cpa  X    273    X /F

where of = fluidizing velocity, ft/s
      W = air quantity, Ib/s
     7"6 = bed temperature, "C
     P0 = air density  at 0"C, lb/ft3
      A = bed area, ft2
      C = bed pressure, atm
  At a given  excess-air level, heat-release rate per unit
area of the bed is a linear  function of the  fluidizing
velocity. The heat-absorption surface is immersed in the
bed and is matched to the heat input, so that the bed
temperature can  be controlled between chosen limits.
Since the  heat input is proportional to the  fluidizing
velociry, as the latter increases, the amount of installed
surface increases to keep within the design's maximum
bed-temperature.
  Furthermore, McLaren and Williams [/] have shown
that as the bed's mean particle size becomes greater to
suit higher fluidizing velocities, the heat-transfer coef-
ficient between the bed and the immersed surface de-
creases.  Hence, two factors determine the area of heat-
absorbing  surface required  as the fluidizing velocity
increases.  Since, for a  given tube arrangement, addi-
tional surface can only be provided by deepening the
bed, an optimization study has to be made between the
capital cost of the bed area and the operating cost of
higher fan power needed for the deeper bed.  Although
the combustion-air fan power requirement is considera-
   ."'fe^S^H?^-.^?.fl^^ed-bedcombustion      ...   'S';^
  *  ^. •.' -V'': •""£"""•.' ;   ..•   -.,. :, •   -.:  .- -

 ;  Companies involved with'this knowhow must choose among a variety of options; such as: Should the bed's
operating pressure be atmospheric or-elevated? Technological controversies now exist, and are examined here in an
excerpt from-a new report,  "Fluidized-Bed Energy Technology: Coming to a Boil," by Waller C. Patterson and
'Richard Griffin.. Issued by Inform, 25 Broad Street, New York,  N. Y. 10004, a nonprofit research organization,
the report  is available for  $45.  .     . '.'•.•""..  '  ' •• - ••"'   ~    .   -•   .'  .     =..   •  : .-•   _•-  . ',...   •..  .-
Q Energy planners all over the. world are looking to
coal as the one essential fuel..    •      • '
   Actually, all but the most single-minded coal enthu-
siasts will concede that, given the choice, oil and natural
gas are more-satisfactory fuels than coal, and that solar
energy is even better in some cases. Interest in advanced
coal-technologies stems from  the premise that oil  or
natural gas is no longer reliably available, that nuclear
power carries high environmental risks and  provokes
strong public opposition, and that solar technology is
not available or economical for certain important  ap-
plications, particularly those requiring intense heat.
  Thus, an increasing use of coal, especially over the
next few years, seems inevitable. If so, the potential
advantages of fluidized-bed systems  warrant  serious
study by those involved in energy-use and -supply deci-
sion-making.
  Opinions  about  fluidized-bed  technology  differ
widely among specialists  but all  agree that the basic
concept is sound. Three U.S., one Norwegian, and three
British companies are offering fluidized-bed combustors
for small-scale  industrial use,  and a number  of test
systems are already functioning. A typical  capacity is
100,000 Ib/h  of steam. When it comes to developmen:
120
                                      CHEMICAL ENGINEERING AUCCST U. 1

-------
  ble, it is largely offset by savings in mill and primary-air
  ' n power, when compared with conventional pulver-
   ed bituminous or anthracite units for beds under 3 ft.
   Choosing the maximum, rated bed-temperature de-
  pends on several factors. First, the maximum tempera-
   ,re at which the bed can function is limited by the ash
   sion characteristics.  Obviously, when firing coal, no
  matter what the starting  bed material is, the bed will
  "Itimately consist only of coal ash, and this must not be
   lowed to sinter. So far, it has been found that trouble
   .11 not occur if the bed temperature is at least 200°C
  below the ash's  initial deformation-temperature. Oper-
   ing at  the maximum permissible bed temperature
   hieves the highest combustion efficiency within the
  _.d, the highest heat flux  to the immersed surface, and
  the  widest boiler-load control that can be practiced by
   d-temperature control.
   Combustion efficiency in the bed and the freeboard is
 iitten high enough that sufficient carbon is burned up so
 *« any carbon in the precipitators can be ignored. In
   ne cases, it may be necessary to refirc a selected part
   the solids carried over to achieve an acceptable un-
 burned loss. As fluidized-bed boilers become bigger, and
   iltiple beds are employed, there will  be increased
   :dom  in  choosing  bed  temperature and  fluidizing
  ..ocity for the refired grits.
   Boiler load is controlled  by bed temperature and the
   •aber of bed zones in operation. To reduce the load,
   bed temperature is lowered by reducing the amount
 3i incoming fuel  and air. This is limited by the onset of
 ' "igmficant  drop in combustion -efficiency.  At  that
   at, selected zones will have the fuel and air cut off so
   :  the zones slump.
     When this happens, the heat-transfer coefficient from
   bed  to  tubes virtually  instantaneously  falls to zero,
   resulting in a step change in load. The bed adjacent to
   the tubes cools to the temperature of the tubes, but only
   a small part of the bed is cooled, because of its insulat-
   ing properties. The heat loss from the zone is so small
   that after several hours' idleness, the zone may be reac-
   tivated simply by restoring the air and  fuel supply.
     The extent of the load change by temperature control
   is a function of the  temperature range through which
   the bed operates  and the  surface temperature of the
   tubes immersed  in the bed.
     For example, consider a  bed with a maximum load
   temperature of 950°C, a tube surface temperature of
   400°C,  and a minimum  bed temperature of 750"C at
   which combustion efficiency begins to deteriorate.  As-
   suming half the heat flux  is through convection and
   half through radiation, then reducing  bed temperature
   from 950°C to 750'C would achieve an about 50% cut
   in the work done in the bed. If, however, for any reason,
  such as  sulfur retention, it would be necessary to limit
  the maximum operating  temperature to 850"C, lower-
  ing the  temperature  to 750°C would reduce the work
  done in the bed by only 30%. Thereafter, load reduction
  would be by slumping selected zones. The  number of
  zones required is such that by a combination of temper-
  ature control and zone slumping, all of the stipulated
  load range  could be covered.
    The rate at which the load is changed by temperature
  control is very fast, as shown in Fig. 8. It will be seen
  that by turning off the fuel  and leaving on the fluidiz-
  ing air, the temperature drops from 850°C to 750°C in
  100 s, even  with a fluidizing velocity as low as 4 ft/s.
;or larger, more-efficient or more-sophisticated systems',
;.   aions differ."     "   "                    -•"-•.
j  JKe  most fundamental  difference is between those
i.whoi fed that others are proceeding too fast, and those
'.   ).feel that others are doing the opposite. In general,
t   :' companies feel that  European firms are moving
f --.-quickly, while Europeans.believe die U.S compa-
;nie?arc lagging behind.."                    '
I   nother disagreement, more technical In -nature, is"
i   ther to use. atmospheric  or. pressurized systems."
S*TO£ companies, such as Pope, Evans. & Robbins, Inc.'
;i-^..York Ciry), are concentrating entirely on'atmos- '
•j   ic systems;  others such  as  Curtiss-Wright Corp.'
,1   pd-Ridge, N.J.), are concentrating on pressurized-'
^•sterns:.Still others, such  as  Babcock  & Wilcox Ltd.'
',(   ;don), arc actively involved in  both lines  of dcvel-
,c   v":.    "   '•'•.:• :.:•: - '•    .  •    :  .
fi • ,!mpanics Pureuing both lines feel that atmospheric
;_uidized-bed combustion will  be preferred for indus-
^   sized boilers, and that  utilities will prefer pressur-
°   systems, which can be adapted to combined-cycle
^- for a large utility plant, about 200 MW, important
•^  D"al-cost savings will be  realized  by the pressurized
   *ed-bed combustor's  smaller  size  (for the same
 amount  of  output)." Also,   improved ' efficiency "is
 .achieved through combined cycles.       ' "••" -'.".  ' '-.' •
 ."  Much of the controversy on how fast to proceed stems
._from uncertainty as to scaleup and the usefulness of
 pilot-plant data. Some U.S. research laboratories carry
 •out exhaustive small-scale investigations into emissions-
 -and their control, sorbcnt "behavior, and. other fluid-!
^-ized-bed phenomena. Others,  however, especially those '
'in Europe, insist that small-scale results will not apply"
 to scaled-up  units. Such companies,. Including' Stal-
- Laval .iTurbln (Fir.spong, Sweden) =an'd  Babcock &'
 •Wilcox Ltd.",'fed that" enough data already'exist to"
 permit construction of major  prototype facilities. V.
 '.. A different opinion is held by those involved iri a
 pressurized  demonstration  plant  to. be  built  at
 Grimethorpe, England, by the  British  government's
 National  Coal Board. The 80-MW, (megawatts, ther-
 mal) unit will be of prototype size but will be used
 purely for research. It is expected to be more flexible
 than a commercial unit and should better lend itself to
 full-scale experimentation.    ••             •   •
   Researchers at Grimethorpe' doubt there  are enough
 relevant data available to successfully design and oper-
 ate  full-scale, pressurized,  fluidized-bed-combuscor/
                                     CHEMICAL ENGINEERING AUGUST U, 1973
                                                                                                      121

-------
                                                 FLCIDIZE BEOS
      900
      850
      800
      750
    CI
    •a
      700
      650
      600
            Fluidizing velocity. 4 ft/i. ji SSO'C










         '''?£:* \''-V.'. 1^." '""• '•'.".•'' "."7":-.T-rT:"'' '"-I*
               I
                      I
                           J_
  i       0      50     100    150    2GO   250    300
  1                       Time, s  •
In fluidized-bed combustion, temperature
dips rapidly as fuel is cut off     '   .  „•• t
                                             Fig. 8
Obviously, by modulating the fuel input rather than
cutting it ofT, the rate can  be made slower.
  Separately slumping each zone is equivalent to turn-
ing off a  burner group served by one mill in a conven-
tional pulverized-coal-fired boiler.  There is, therefore,
much similarity  in the boiler load-control systems,  in
that the same sort of functions have to be performed.  In
a pulverized-coal boiler, the removal of a mill  and
burner group is  dictated by  its turndown  capability,
which is  defined for  the operator.  In a fluidized-bed-
fired boiler, the  removal of coal feed and  air from a
zone is dictated by the lowest temperature that provides
good combustion and maintains sulfur retention. Flu-
idized-bed firing has the advantage  that once a bed is at
its operating temperature, an expensive fuel-oil support
system is  not needed during load changes or at low load
operation, even with very-low-grade  coal.
  The author has described [2] the first year's experi-
ence with a 40,000-Ib/h boiler that  was converted  to
fluidized-bed firing.  In  subsequent trials with a coal
containing 60% ash, the boiler load  was controlled from
46,000 Ib/h of evaporation to 12,000- Ib/h, with 'no
support fuel required. This is readily understandable, as
the coal in the bed is  less than 1% by weight of the bed,
even at maximum load.

Atmospheric-pollution control
  The main pollutants in  the stack gases from current
conventional high-temperature combustion systems are
ash, sulfur oxides and nitrogen oxides. There is. how-
ever, a growing  concern about the  emission of trace
elements  such as arsenic, antimony and mercury.
  Ash is  trapped with acceptable efficiency  by precipi-
'"gas-turbine systems. However, Stal-Laval Turbin and
 Babcock &-Wilcox, Ltd., among others, disagree. The
 Grtmethorpe team considers it premature to tackle both
;the problenvoEa pressurized fluidized-bed cocnbustor_
 and that of coupling such, a combustor to a gas turbine. •
  .On'the other hand, the.Stal-Laval/Babcock & Wil-
_cox team feels that the Grimethorpe unit, will be."rein-
 venting the wheel," carrying out research on a nohcom- .
 mercial.uhit that  might  perfectly  well  be done  by
 building  a. commercial  prototype and acquiring the
 desired experience in. actual service. Some U.S.'experts
 feel that the Stal-Laval/Babcock & Wilcox team  is
 overlooking the problems of gas-turbine-blade corrosion
 due to sodium and  potassium in the coal.
   One project, considered already obsolete by some, is a
 30-\IWe (megawatts, electrical) prototype atmospheric
 unit, built in  Rivesville, W.  Va., by Pope, Evans &
 Robbins for the former  U.S.  Office  of Coal Research,
 now part of the U.S. Dept. of Energy (DOE). The project
 was an attempt to develop a cheap coal-fired boiler for
 power generation. Contracted for in 1972, before the oil
 price-rise,  the  system was supposed  to compete with
 then-cheap oil.                 '  •  .     .
   The plant is a multi-cell fluidized-bed unit, consisting
 of four separate beds. The first bed boils the water, the
                                                      second heats- the resulting steam, the third superheats
                                                      the steam, and the fourth burns up unburned carbon.
                                                      Problems have been encountered,  particularly  with
                                                      coal-handling and  coal-feed systems. One persistent
                                                     ' problem is clogging of coal in feed lines, blocking the •
                                                      fuel supply.
                                                        Critics feel  these problems are  partly due  to tight'
                                                      specifications laid down when the objective was to build
                                                      a very inexpensive boiler.
                                                        Today, DOE is primarily concerned with sulfur-oxide
                                                      emissions control. Fluidized-bed combustors must com-
                                                      pete economically not with oil-fired  boilers but  with
                                                      conventional coal-fired boilers using  scrubbers.  Pope,
                                                      Evans &.  Robbins concedes that times and priorities
                                                      have  changed. However, the  company questions  a
                                                      wholesale switch to pressurized fluidized-bed combus-
                                                      tion. It believes that money saved by smaller-size units
                                                      for the same output will be offset by the additional cost
                                                      of coping with the higher pressures, especially for feed-
                                                      ing coal and removing solid waste. Atmospheric systems
                                                      are thought preferable for all sizes.
                                                        Stal-Laval believes  that pressurized systems may be
                                                      preferable For msdium-to large-scale applications. Ths
                                                      firm cites the advantage of being able to double  a
                                                      pressurized system's useful energy output by using com-
 122
                                       CHEMICAL ENCI.Sf.EKING AUGUST 11. 1973

-------
  tators. Sulfur oxides can be controlled by burning low-
  sulfur fuels, which command high  prices,  unless they
  are coals of the sub-bituminous  types, such as those
  found in the western U.S. and in  south Australia, that
  have a high alkali content and give rise to  severe slag-
  ging  and fouling.
     For high-sulfur fuels, expensive scrubbing systems
  may  be used.  A popular method  is to scrub the gases
  \rith  a limestone slurry. But this has  two disadvantages.
  First, it creates a residual pollution  problem in dispos-
  ing of the spent limestone, unless an expensive recovery
  plant is  installed to regenerate the limestone and  re-
  cover the sulfur, which  has a doubtful market  value.
  Second, scrubbing carries a penalty of at least 4%  on
  gross thermal  efficiency because,  after scrubbing, the
  gases have to  be reheated  before entering the  stack.
    Nitrogen oxides are generally controlled by interfer-
  ing with the combustion process, again with attendant
 .penalties.
  :. With fluidized-bed firing, when  there is no restriction
 -on sulfur oxides emission, the removal of paniculate
  matter from the gases is somewhat easier than it is for
  pulverized-coal  firing. This  is because, in  a fluidized
  bed, for a given coal less goes to the precipitators, and
  the mean particle size is much greater. Where  sulfur
  retention  is  practiced, adding limestone (or dolomite)
  makes the dust burden greater and changes the charac-
  ter of the dust to be caught. Opinions  among the mak-
  en of electrostatic precipitators appear  to be rather
  mixed. Some claim that enough is  known to permit the
  design of satisfactory precipitators, while others regard
  this area as  still developmental.
    Garner, Howe and  Dzierlenga  [3]  list proposals of
        100
         80
       * 60
       I
       3
       840
         20
                                    _X '
                        lim»ston« A
                                ^Predicted pomti from
                                '"laboratory SCale tttisX
                                   '-'•* *> *• T"" •? •"  . •     »
40.000 Ib/h fluidiztd-bcd boiltr
Sulfur content of coal. 5.5%
8«d wmperaturt, 850'C. ;'r'
     •''
                            234
                            Calcium: sulfur mol ratio
                  0.2     0.4     0.6      0.8      1.0
                         Ratio of limestone to coal by weight
                             1.2
     Better than 90% retention of sulfur is  '.   '•..- /••
     achieved even when using high-sulfur coal         Fig. 9
    v.' ---- •,     ..»;• .  :    -•.  --  "--••••;, Tll-*"j;-~
p'- .blried cycles or cogeneration. The company believes this
[ >T|n>e the most effective lever in persuading industries
t; JO convert their boilers to ones that use pressurized
K.fiuidized-bed coal  firing.  .          -.  '..;',';„.-;,::;
       .Fraas, a  consultant to the Oak Ridge National
            (Oak Ridge, Tenn.), points out that using a
             hot-air turbine system makes it possible to
          atmospheric  fluidizcd-bed  combustion and
           cycles or cogeneration.. -., '"-::""-..i.v:i'I'::;-.i;:"/
          , Freedman, -director  of. DOE'S ^fluidized-bed
          ^^ ^hat industrial-sized atmospheric fluid-
           ar« closest to becoming  commercial. These
           eatest advantage, he believes, is their fuel-use"
 .. Freedman sees utility systems as a longer-
  t?7lPfoposition. DOE hopes to have both large atmos^.
(:.P ^ ^nd large pressurized fluidized-bed plants online.
^- By;1984, so utilities can choose between the two systems.
§• . ?** advantages  in both approaches: atmospheric  is
^ *cipler and  more reliable,  while pressurized  has  a
   '5"Uy higher  efficiency, and produces  slightly fewer
    lfur-dioxide emissions.
   .Other differences of opinion exist. Battelle Memorial
 •.Institute  (Columbus, Ohio) and various U.S. research-
  •"!* are concerned  about  possible  corrosion of boiler
      . Others, especially those in Europe, see nothing at
   .all to suggest that such problems will arise! ;-
      Coal  feed has  caused  continuous trouble  at  the'
 :  -Rivesville unit,  and several U.S. firms,are also con-
• ' cerned about feed problems, while the'Europeans see no
   such troubles. The general European view seems to be
  ~ that coal can be fed from above the bed,, within it or
  'under it; and that the easiest way, from  above, is satis-
   factory. : Coal fed from  above remains 'within the  bed
  ;r"long enough to burn completely withoutTundue carry,-"
 ?.--over of unburned particles into  the flue"gases. Europe-
 : . ans feel that if a company uses a special carbon bumup"
   cell for unburned particles when moving to large-scale
"; ^.applications (as  is the case at RivesviJle),--thJs indicates
  -•.the design of the main  bed is hot optimum:.'   . "• - '.':
  -, '.' Sdll another difference of technical opinion concerns
   •unit size. Some favor cell and  bed  dimensions small
   enough to permit prefabrication in the manufacturers
   shop, while others favor larger units.  The smaller-scale
   (or modular) approach  offers the advantages of repli-
   cation, and  of stable working conditions in an  estab-
   lished shop. However,  Babcock  & Wilcox,  Ltd.,  and
   others believe the modular design may not be as flexible
   in actual operation as the custom-built large-scale bed,
   which may be subdivided into separate cells with sepa-
   rate air and fuel feed.             -;   : ••*.
                                         CHEMICAL ENGINEERING AUGUST 14, 1978
                                                                                                            123

-------
                                                 FLU1DIZE BEDS
                                   Raw-coal
                                     inlet
                                    ,Throat gap
                                         Primary-
                                         air inlet
  Mill modified to operate as a crusher. •
  Unit separates out pyrites as well             Fig. 10
     400
     350
   M
   C
   g
   | 300
   o
   o
   .H
   I 250
   §
   ic

   1200
   o


   I 150
   a.
    x
   O
     100
      50
                   . .-           ..-
          .        with 1.154 nitrog«r> _.'-.'• •'-••"-. -"^
         i,40.000-lb/hboiler ; 3  •••..;. • I-.-.V'."- .--,'•.'•""
 ^ir/5ivj;lvT.r
 •-':»rj «">-• ,; •.*•'- .
 .' .. ---;j.-^v^" -    ^ r.
 ^~'r-'..- . Vl  _ .-.  \
 -*v-.;v.  -"..v>-.—. .T

''.»""•'•''•>:?-'~ ' t' -" -.'- -=rl
sp^:.-:;
-',.v^v--;^
^.^^s®'
       650    700    750    800    850    900    950
                    Bed temperature, °C

  NO*, emissions level off above SOO°C to
  value below current U.S. legal maximum       Fig. IT
                 three U.S. boiler makers. Two propose to trap larger
                 particles in multitube cyclones and do the final cleanup
                 in a baghouse. The third proposes using  two hot-side
                 electrostatic precipitators, one to trap the  coarse mate-
                 rial  for  possible reinjection and  the other for final
                 cleaning. There is no doubt  that baghouses can now be
                 used to meet particulate-emission limits. Precipitators,
                 however, may turn out to be the most economical way
                 to control particle emissions.

                 Limestone efficiency
                    Fig. 9 shows the effect of the limestone/coal ratio on
                 sulfur retention for two different limestones. Although
                 these have the same chemical composition, their  effi-
                 ciency in reacting with sulfur is different, probably due
                 to differences in their porous structures. These results
                 show two other features:  (1) even with very-high-sulfur
                 coal, retentions better than  90% are achieved;  and (2)
                 small-scale tests can accurately predict the Ca:S ratio
                 required to yield the desired performance in a boiler.
                 Ehrlich  [4] shows  the effect of bed temperature  on
                 sulfur  capture  and concludes that  the optimum tem-
                 perature at atmospheric  pressure is near 800 °C. A flu-
                 idized-bed combustor operating between  850 ""C and
                 750°C would probably provide the optimum in sulfur
                 retention and boiler load-modulation by bed tempera-
                 ture control.
                    Sulfur occurs in coal in organic and inorganic form.
                 The inorganic is mainly pyrites, an iron sulfide having a
                 specific gravity of 5.0. Coal has  a specific gravity of
                 about  1.3, while that for coal ash or limestone is around
                 2.7. In low-sulfur coals, the organic form predominates.
                 As the total sulfur increases, more of it becomes pyrites.
                 This is separated out in  some types of mills.
                    Fig. 10 shows a sectional elevation of such a mill. The
                 air for drying and conveying the coal through the mill
                 enters  a plenum chamber  under  the yoke and then
                 passes  through slots in the throat ring into the mill. The.
                 velocity through these slots  keeps coal particles in the •
                 mill, while heavier particles (pyrites) fall  through the
                 slots. The rejected material is then swept by ploughs to
                 a reject box.  The high-sulfur coal used  in Fig. 9 may
                 have half of its sulfur as  pyrites. The more discrete the
                 distribution of the  pyrites, the greater the proportion
                 will be rejected and, accordingly, the greater the savings
                 in limestone needed to reach the acceptable SO, emis-
                 sion level.
                   'With coals having  a modest sulfur content, the cal-
                 cium oxide in the coal ash may be sufficient to reduce
                 SO, emissions without requiring the addition of lime-
                 stone to  the bed.
                    In Fig. 11,  NOX emissions  from a Babcock boiler that
                 is  firing a coal containing 1.1% nitrogen are shown as a
                 function of bed temperature. The analyses were done by"
                 the chemiluminescence method. The maximum effluent
                 level is well below the current limit of 525 ppm at 3%
                 excess  oxygen set by the U.S. Environmental Protection
                 Agency for new coal-fired plants. In fluidized-bed com-
                 bustion,  it is now generally accepted that most of the
                 NOX formed  comes from the nitrogen in  the fuel  and
                 not from the atmosphere. The maximum  figure of 325
                 ppm shown on Fig. 11 corresponds to the conversion of
                 about  30% of the nitrogen in the coal
124
CHEMICAL ENGINEERING AUGUST U. 1973

-------
       Saturated-steam pressure
        150ib/inZ
   3i-drum boiler that will be converted to fluidized-bedfiringVv/U'
 Fluidized-bed advantages
   ^rther advantages are apparent when  comparing
   idized-bed combustors with other furnaces:
   Because combustion is carried out at a relatively low
   iperature, the coal-ash particles are kept below their
   cmng  temperature,  preventing  slagging   and
 ---ided-deposit  formation. Hence, the need for soqr-
 blowing is reduced. On the 40,000-lb/h boiler, the soot-
   hers that were in place before conversion  have never
   tied to  be used in over 6,000 h of operation. This
 snould allow for more-compact convective-heat-transfer
  -ks, thereby increasing heat-transfer coefficients and
   umizing the surface area.
   cry-high heat-transfer coefficients are obtained be-
   en the bed material and the immersed surface, so
    although the temperature differential is lower than
    pulverized-coal- or oil-fired boiler, the overall heat
l«4* is greater. Therefore,  less surface is required. Fur-
ther,  there is relatively little variation in  the heat flux
    all the immersed surface, so there is small risk of
   : failure due to high local heat flux which can lead
 o departure from nucleate boiling in the tubes.
   The low combustion temperature leads to markedly
 reduced  vapor pressures of alkali  metal  sulfates  and
 chlorides. Thus, the greater part of these harmful con-
 stituents  remain in the bed. This is true also of vana-
 dium and sodium salts when heavy fuel-oil is fired  in a
 fluidized  bed. Cooke and Rogers [5] report the results of
 corrosion trials on a number  of  low-chromium  and
 austenitic steels. They conclude that all the materials
 would provide satisfactory service under normal fiuid-
 ized-bed  operating conditions, if used at the tempera-
 tures employed in a conventional plant. Even with that
 limitation, however,  the  fluidized-bed system has  an
 advantage. Because of the high rate of mixing in  the
 bed, all the tubes will be at the same temperature, so
 there is no need to add temperature margins in design.

 Current designs of fluidized  boilers
  In  the  U.K., fluidized-bed-fired  boilers  having  an
evaporation rate to 500,000 Ib/h are being offered. The
upper limit  of evaporation is fixed at a level that is
achieved economically in  a single  bed, at  fluidizing
velocities up to 10 ft/s. Above that size, multiple beds
                                      CHEMICAL ENGINEERING AUGUST 14, 1978
                                                                                                        125

-------
                                                FLUIDIZE BEDS
      Evaporation rate:
       355,000 Ib/h
                                                                    jj;	Primary superheater
                                                                                              	Air heater
               Mills
Primary-air
   fan
Forced-draft
    fan
 Two^pass; radiant fluidized-bed boiler uses convection banks for superheating                        xV'^-.rig.-.T.
  • ••.^^•ifS^uC^^^f-.ff^.f^.\&JX^£:±^Z3:^y?::.:^:-* ::&.&..:- •>;':•f.~/^.->',-'W.<•-~SA'.;V^j--'-.l'V?.'C^'.^-yy-.^--, .^,»=U.'av:f •^--.'a^
are used. Not enough  time has yet been  devoted to
seeking the optimum arrangements for multiple beds.
Garner, et  d. [3] describe conceptual designs for 500-
MW  boilers by three U.S. boiler makers in conjunction
with two architectural engineering firms, and sponsored
by the U.S. Dept. of Energy. In the U.K., an order has
been received for an 80,000-lb/h boiler for a chemical
process company. It is a standard bi-drum boiler with
conventional oil burners  in  its front wall. However,
provision  has been made for  conversion  within  five
years to coal firing with a fiuidized bed.
  Two other projects are illustrated. Fig. 12 shows an-
other  bi-drum boiler, which will be converted  from
spreader/stoker firing to  a fluidized bed.  Conversion
must be looked at carefully, since it may be more eco-
nomical to install a new boiler, depending  on the state
of the plant and the original fuel for which the plant
was designed. Conversion of spreader/stoker firing to  a
fluidized bed can be attractive, since the  combustion
rate per unit area and the freeboard shape and size can
be the same for the  two  systems.  Only three separate
                          slumped zones are required to meet the load changes
                          demanded by this system.                          ;.,-
                            Fig. 13 shows a design for the 30- to 40-MW, range.
                          In this unit, all the superheating is done in convection.
                          banks, so that only generating tubes are immersed  in
                          the  bed. Also, the boiler plan area is reduced  and di-
                          vided to provide the gas velocities  required  to  reach
                          adequate  convection-heat-transfer  coefficients. When
                          boilers demanding multiple tiered beds  are called for,
                          the  plan area will not need as much, if any, reduction.
                            Fluidized beds are used  in the marine field [6\. The
                          ability of an oil-fired fluidized bed to retain vanadium
                          and sodium is being applied  to raise superheat and
                          reheat temperatures to 600°C in steam turbines for ship
                          propulsion.  Hitherto, such  temperatures  have been for-
                          bidden  with conventional  oil firing by  high-tempera-
                          ture corrosion of superheater  and reheater tubes.

                          Pressurized fiuidized  beds
                            So far,  we have concentrated on  atmospheric fluid-
                          ized-bed boilers, principally because their development
 126
                                         CHEMICAL ENG1NF.F.KINC AUGUST I ». 1

-------
 is more  advanced.  There is, however, a great deal of
 work going on  in  pressurized fluidized  beds—which
 promise these further advantages:
   1. For a given duty, the bed area is inversely propor-
     tional to the absolute pressure; thus the plant can
     be  compact and comparatively lightweight.
   2. Combustion efficiency within the bed is such that
     good design should eliminate  the  need  for grit
     recycling.
   3. Using dolomite, sulfur retention is improved  with
     pressurization.  (Unfortunately,  limestone effec-
     tiveness decreases at increased pressures.)
   4. NOX  emissions are reduced further by pressure.
   Pressurized  combustion should permit  the develop-
 ment of  combined gas and steam-turbine cycles, which
 hitherto  have  been  limited  by the necessity to  employ
 clean fuels  for the  gas  turbine. A gas turbine should
 operate  for  long periods without significant fouling,
 corrosion or erosion, due to the non-erosive nature of
 the coal ash,  the ability of the  bed  to retain alkali
 metals and sulfur, and the effectiveness of mechanical
 •emoval  of paniculate matter  from the gases with an
 acceptable pressure drop. Roberts, et at. [7] describe the
 experimental work and  the  types of cycle that may be
 employed.  Thurlow [8] mentions some design studies
 :urrcntly in  progress, and describes the application of
 fluidized-bed combustion to boilers smaller than those
 discussed here.

 Conclusions
   It has  been  said that there would be no interest in
 fluidized-bed combustion if it were not for the stringent
 •egulations concerning the emission of sulfur oxides and
 nitrogen  oxides. This is extremely difficult  to believe.
While atmospheric pollution is important—and it has
 Decn amply  demonstrated that these beds can  control
 t—surely an important factor is the ability to operate a
 plant at  its  design  capacity, no matter how the  fuel
 supply may change during  its life.
   Another important  factor  is  energy  conservation.
oince the Industrial Revolution,  coal mining  has re-
 jected combustible material  consumers would not take,
 lither because consumers could not use it or because it
 *as uneconomical to do so. For similar, reasons, there
are large quantities of  high-ash coal  deposits in the
world that have been left in the ground. In the U.K.,
 argc amounts of the heat  content  of the total  coal
 nined has  been lost to pit spoil-heaps. [9].  The world
can no longer afford to squander or neglect its energy
 •esources that way.
   Cost comparisons between ffuidized-bed combustion
tnd present conventional units are not made here be-
cause the costs of the latter vary so much with the type
 >f coal being fired.  Thurlow gives the results of some
 tudies both  for atmospheric and pressurized applica-
tions. Generally, these favor fluidized beds, particularly
•f flue-gas cleaning is taken  into consideration.
   There is continuing development in atmospheric and
pressurized fluidized-bed combustors.  In atmospheric
units,  the emphasis is on increasing combustion  effi-
ciency in the bed to reduce the need  for refiring  car-
bon-loaded grits.  In pressurized designs, emphasis is on
the development of power turbines and control systems
that can be successfully integrated with the combustor.
   It is hoped  that this report has  shown that  atmos-
pheric fluidized-bed boilers are gaining acceptance; any
operator  with  a  new plant in mind should give the
system careful study.

Acknowledgment
The author acknowledges with thanks the permission of
the directors of Woodall-Duckham Ltd. to publish this
repon, and thanks his colleagues who have helped  and
advised him in its preparation.
References
1. McLaren. J. and Williams. D. F., "Combustion Efficiency, Sulphur Reten-
  tion and Heat Transfer in Pilot Plant Fluidised Bed Combuston," /• liut
  Fuel, Aug. 1969.
2. McKeruie. E. C. Fluidised Bed Firing in Boilers, Sfoct Httlna mi Air
  Condiiwunf Jumol, Mar. 1977.
3. Garner, D. N., Howe, W. C. and Dtierlenga, P. S.. A Companion of Selected
  Deign Aspects of Three Atmospheric Fluidised Bed Combustion Concep-
  tual Power Plant Designs, Fifth International Conference on FBC, Wash-
  ington, Dec. 1977.
4.. Ehrlich, S., A Coal Tired Fluidised Bed Boiler and Fluidised Combustion
  Conference organized by  the Inst. Fuel, London, Sept. 1975.
S. Cooke, M. J. and Rogers. E. A., "Investigations of Fireside Corrosion in
  Fluidised Combustion Systems," Inst. Fuel Conference, Sept. 1975.
8. Stal-Laval publication, "Very Advanced Propulsion."
7. Roberts, A. G.. Stanton, J. E., Wilkins. D. M., Beacham, B. and Hoy, H. R.,
  "Fluidised Combustion of Coal Si Oil Under Pressure," Insi. Fuel Confer-
  ence, Sept. 1975.
8. Thurlow, G. G., The Combustion of Coal in Fluidised Beds. Pro. of Inst.
  Mich. Engri., Vol. 192, No. 15, pp. 145-156.
9. Down, W. S. and Brown, A., "Colliery Spoil Heaps as a Source of Energy,"
  Conference on Energy Recovers- in Process Plants, L Mech. E., London, Jan.
      For  an  introduction to  coal,  its  selection,
    preparation and combustion, see the Feb., Mar.
    and April, 1974, issues of Power.
                         The  author
                         E. C. McKenzie acts as a full-time
                         consultant in fuel technology to
                         Woodall-Duckhara Ltd., Babcock 4
                         Vv'ilotw Group, Woodall-Duckham
                         House, The  Boulevard, Crawley, Sussex,
                         England RH10 1UX. He received his
                         training in fuel technology at the Fuel
                         Research Station, which was pan of Her
                         Majesty's Dept. of Scientific and
                         Industrial Research. In 1936, he joined
                         Babcock & VVilcox, and has since
                         worked on problems including fuel
                         preparation  and heat transfer.
      Reprints of this report will be available shortly. To order, check number 295 on the reprint order-form
      in the back of this or any subsequent issue. Price: S3.00 per copy.
      Please circle No.  305 on the Reader Service Card for a complete Catalog of Reprints.
                                         CHKMICAL ENGINEERING AUGUST U, 1978
                                                                                                              127

-------
V.  NO

-------
5.2    COMBUSTION PROCESS MODIFICATION TECHNOLOGY

       As a result of emission control regulations for new and existing
stationary sources, NOX control techniques have been developed and
implemented in the past 10 years.  Nearly all current NOX control applications
use combustion process modifications.  Other- approaches, such as modifying
or switching fuels, using alternate energy systems, and treating post-combustion
flue gas  as well as more advanced combustion process modifications  are being
evaluated for potential future use.  Experience has shown that the applicability
and effectiveness of combustion process modifications depend on  the  specific
equipment/fuel combination to be controlled, and on whetner the  control  is
to be applied to existing field equipment or new units.  Accordingly,  control
development is focusing on specific equipment categories and fuel types.
In general, the following sequence of control development  is oemg pursued
for each major equipment/fuel category:

        •   Minor operational  adjustments

        t   Minor retrofit mod.if i cat ions
  SOURCE:  "Environmental  Assessment  of  Stationary Source NOX Control
           Technologies:   First Annual Report," EPA-600/7-78-046, U.S. EPA,
           Research Triangle Park,  NC (March  1978).

-------
       •   Extensive hardware changes, either retrofit or factor y- MiU" i^J  of
           units, of conventional design

       •   Major redesign of new equipment

Progress made in this sequence varies with the importance of the source  in
local and national NOX regulatory strategies.

       Currently, modifying combustion process conditions is the most
effective and widely-used technique for achieving 20- to 70-percent reduction
in combustion-generated oxides of nitrogen.  These modifications include:

       •   Low excess air firing

       •   Flue gas recirculation

       •   Off-stoicnicmetric combustion

       o   Load reduction

       •   Burner modifications

       •   Water injection

       •   Reduced air preheat

       •   Ammonia injection

The following paragraphs summarize the status of each of these controls.


Low Excess Air Firing

       Changing the overall fuel-air ratio is a simple, feasible,  and
effective technique for controlling NOX emissions from all stationary sources
of combustion except gas turbines.  For some sources, such as utility
boilers, low excess air (LEA) firing is currently a routine operating
procedure and is incorporated in all new units.  Since it is energy efficient
and easy to  implement, LEA firing will be  increasingly used in other sources.
However, most sources will have to use other control methods, in conjunction
with LEA, to meet  NOX- emissions standards.   In such cases, the extent to
which excess air can be"lowered will depend  upon the other control techniques
employed.  Virtually all programs for developing advanced NOX controls  are
emphasizing  operating at minimum levels of excess air.  Thus, LEA  will  be an
integral part of nearly all combustion modification NOX controls,  both
current  and  emerging, to be assessed in the  NOX E/A.


Flue Gas Recirculation

       The primary near-term  application of  flue gas recirculation (FGR) is
in gas-  and  oil-fired utility boilers.  Future applications are  limited. FGR
may be used  in  industrial  boilers as a retrofit or  in new designs, but
                                      43

-------
                                       x                 ~sT6Tch foraetrTc"
   Lonoustion,  also  are being  evaluated and may  prove more  attractive
  Off-Stolchlometric Combustion
  , ity boners           ^"?  ac°TinS""ed •Ul'-"t « "" « '
                                  re aca*1<"'s "> "e considered Include
                               ^^^
  Load  Reduction
        used only  as a  last resort  to  acfneve  compliance with  standards.
 Burner Hodffications
                                               .              O








Water  Injection





«ter ejection «,„ be replaced b, advanced cc'Stor dL^'fn^T,^"

-------
  Reduced Air Preheat
         Reduced air preheat for gas  turbines  and for boilers is not a
  practical  way to  control  NOX  unless the energy in the exhaust  gases can b*
  used  effectively  for  other purposes,  such  as  in combined  gas-steam turbine"
  cycles.  Reduced  air  preheat  will  thus  be  accorded  low priority in the  NO,
  E/A because  of associated  efficiency  losses.                              x
  Ammonia  Injection
                 * !! ®C * ' °? d?e* not appear  to  hflve  significant  near-term
       O  v      "°x 5°nti;?Vn the U'S-   However'  ]t  shows Promise  f°r far-
      applications and will be given primary  emphasis  in  the N0y E/A for
 assessment of advanced concepts for the 1980 's and 1990 's.


 5.3    ALTERNATE CONTROL TECHNIQUES

        In addition to combustion modifications, NOX can be controlled by one
 or more of the following techniques:   flue gas treatment, fuel
 denitrification, fuel  additives,  alternate or mixed fuels, or advanced,  low-
 NOX combustion concepts.   Each of these is briefly discussed below.


 Flue Gas Treatment

        The dry flue  gas  treatment (FGT)  techniques used in Japan -  notably
 selective catalytic  reduction  with  ammonia - can probably be applied to
 gas- and oil-fired sources  in the U.S.   However,  more  pilot and full  scale
 demonstration  tests  are  needed  before  full application of dry processes  is

 f?rPrf  nM™heSUhSOUrhCeS'-i ?"y p:°cesses  have ** to  be demonstrated  on  coal-
 fired  sources   although pl lot-scale tests  are currently planned.   Wet  processes
 are  less  well  developed and more  costly  than  dry  FGT processes-  however  wet
 processes  have  the potential  to remove NOX and SOX simultaneously    Again
                                                  "
                                                  X
SJj^;5"16/"??:?  and  field  tests  are  needed  to"  determine  costs,  secondary
effects,  reliability,  and waste  disposal  problems.   Flue  gas  treatment  holds
scrne promise  as a  control technique  if  very  stringent  emissions  standards

win n nheMSSsry  ° ,9re!tly reduce N°*-   However'  even in  these  instances FGT
will probably be employed to supplement combustion  modification.


Fuel Denitrification

       Fuel denitrification of coal or heavy oils could,  in principl°  be
used to control the component of NOX emissions produced by  the conversion
of fuel-bound nitrogen   The most likely use of fuel denitrification would
De to supplement combustion modifications  that reduce thermal N0y.  Currently
denitrification occurs only as a side effect of pretreating fuel to remove
™ T\n  ' °P ?therl P°]lutant Precursors.  Preliminary data indicate that
?r t° 4°-P^ce nt reductions In fuel nitrogen result from oil desulfurization
Deference 22).   Since these processes produce low denitrification efficiencies
                                     45

-------
 SOURCE:   Technical Assessment of Thermal DeNOx Process—Interagency

          Energy/Environment R&D Program Report. EPA-600/7-79-117,

          U.S.  EPA, Research Triangle Park, NC (May 1979).
                                 SECTION 2



                              PAST EXPERIENCE






       The noncatalytic reduction of NO by the Thermal DeNO  Process was



discovered in August 1972 by Exxon Research and Engineering Co.  (ER&E).



Since then, numerous laboratory, pilot, and full-scale tests have further



investigated the Thermal DeNOx Process.  These tests have been designed



to understand the critical process parameters and how they can be used to



control N0x emissions from both stationary and mobile sources.



       ER&E, developer of the Process and patent holder, has done most of



the laboratory research and field studies.  However, Exxon's tests were



limited to gas- and oil-fired facilities, except for full-scale  tests on a



solid waste incinerator.  KVB Inc., under contract to ER&E and EPRI has



recently studied the use of Thermal DeNO  on a 3 x 106 Btu/hr
                                        A


coal-fired test boiler.  KVB has also conducted the only domestic



full-scale application of NH3 injection.  This was on a thermal oil



recovery steam boiler.



       Table 2-1 presents a summary of all commercial installations



utilizing noncatalytic decomposition of NO  by ammonia.  All of these



installations use Exxon's ammonia injection technology- except for one



Japanese source noted.  Detailed information on all installations using



Exxon's Thermal DeNOx Process is not available.   However, depending on



the source and its operation, NO  reductions as high as 70 percent have



been achieved.
                                    2-1

-------
                      TABLE 2-1.   SUMMARY OF  COMMERCIAL  APPLICATIONS OF EXXON'S  THERMAL DeNOx PROCESS
ro
i
ro
Source
Steam Boiler
45 MM heat Input
Incinerator
7 ton/hr
Crude heater
ISO x lO^bb I/day
Steam boiler
76 KU heat input
Utility boiler
275 HW heat input
Utility boiler
275 HW heat Input
Crude heater
150 x 103bbl/day
Thermal recovery
heater
Utility boiler
375 KW
• Fuel
Burned
Gas/oil

Waste

Gas/oil

Gas/oil

Gas/oil

Gas/oil

Gas/oil

Oil
Residual
oil
Location
Japan

Japan

Japan

Japan

Japan

Japan

Japan

USA-
California
Japan
Initial Nitric
Oxide Emissions
(ppn as measured)*
120-150

100-180

150

95-145

80-120

80-120

80-85

260
HA
OeNOx Rate
(Percent)
60

20-70

35-65

35-50

60

50-60

40-65

50-70
40
Addltlveb
Yes

MAC

Yes

HA

NA

NA

NA

NA
No
Comments
No reduction obtained at full load

Difficult source to retrofit because
of constant change In fuel

Best reductions achieved at high
Injrf
IU6O
No details of retrofit system are
aval lable

No details of retrofit system are
available

No details of retrofit system are
available

Best reductions achieved at high
loads

First commercial U.S.
Installation
Does not use Exxon N)l3 Injection
technology — NHj emission limited
to 10 ppm
                      •No'  indicates that no additive was used
                     CNA •  no data are available
                                                                 «H3 to obt ..... ported NO. reduction performances

-------
       Based on these full-scale results, Exxon has commercialized the
process and will license it upon request on gas- and oil-fired boilers.
Additionally, they are continuing to study the feasibility of full-scale
applications on coal-fired boilers.  These studies are aimed at maximizing
N0x reduction and cost effectiveness, and exploring and defining
potential operating problems.
       This section summarizes the status of Thermal DeNO  Process
                                                         A
developments.   Results from gas and oil combustion in laboratory and
pilot-scale studies are discussed in Section 2.1.  Section 2.2 presents
results from ammonia injection in a coal-fired pilot-scale facility.
Results from ER&E's full-scale commercial demonstration of the process are
described in Section 2.3.
2.1    SUBSCALE TESTING — GAS AND OIL
       ER&E discovered a new reaction which is the basis of the Process in
research done in a laboratory flow reactor.  Based on the observed
kinetics of this reaction Lyon proposed the following mechanism
(Reference 2-1):
                    NH2 + NO - »- N2 + H + OH
                          NO
                    H + 02   - >• OH + 0
                    0 + NH3  - >- OH + NH
                    OH
                    H
       To further explore this reaction mechanism, ER&E conducted tests on
a small 0.3 MW (10  Btu/hr commercial size boiler).  These tests
investigated how key operating variables such as reaction temperature,
                                    2-3

-------
M, injection rate and flue gas residence time influence the NO,
  
-------
       Lyon reported that the reaction temperature of about 955°C
(1750°F) results in the largest NO reduction.  At temperatures higher
than approximately 1100°C (2010°F), the oxygen present in the flue gas
oxidizes the injected ammonia producing a net increase in NO.  Below
900°C (1650°F), the reaction of NH3 with NO  is slowed considerably
causing less NO reduction and more NH3 emission from unreacted gases.
Figure 2-1 shows how the reaction temperature affects the performance of
the Thermal DeNO  Process.
                A
       This strong dependence of Thermal DeNO  performance on reaction
                                             A
temperature can limit the use on some systems.  For example, the required
temperature window in gas turbines and 1C engines is located where a very
short residence time is available for reaction with NH.J.  Thus, Thermal
DeNO  may not be suited to these sources.  In large steam generators,
    A
the optimum temperatures and residence times for  noncatalytic reduction of
NO are usually  accessible in the boiler convective section.  However,
cross-sectional temperature gradients as high as  200°C (360°F) often
exist making ammonia injection considerably  less  effective for some  areas
of the flue gas ducts.  Load variations also shift the temperature profile
causing the temperature window to move in the convective section.
       These problems could theoretically be solved by installing more
than one  injection  stage  to account for the  shift in the temperature
window.   However,  additives have  also been evaluated as methods  to
accommodate temperature  variations  and unfavorable  location  of the
temperature window.  For  example,  hydrogen additive  is a demonstrated
alternative for controlling or shifting the  inj'ection  points to  the
optimum temperature  location for  the  Thermal DeNOx  Process.   Additive
injection is discussed  in Section  2.1.3.
                                     2-5

-------
            1.0
            0.8
            0.6
            0.2
                  Excess oxygen:  4*


                  Initial NO:  300 ppm
                                                 (NH3)/(NO)
                                                 1.6
                   700      800   '   900


                              Temperature,  C
1000
1100
      Figure 2-1.   Effect of flue  gas  temperature on Thermal DeNOx

                    performance  (Reference 2-2).
2.1.2  Ammonia Injection Rate



       Ammonia efficiently  reduces  NO   because of its ability to react
                                      A


selectively with nitric oxide  regardless of the amount of oxygen present



in the treated gas.  Thus,  the amount  of ammonia required in the Thermal



DeNO  Process is on  the order  of the initial NO concentration.  Other
    A


additives, such as methane  and ethane also can be used to reduce NO in  hot



flue gases.  However,  because  these reagents are nonselective,  they do  not



react with nitric  oxide  alone.  That is, all the free oxygen present in



the hot gases must  first  be consumed by the reagents before  the NO can  be



reduced.  Therefore, more hydrocarbons  are necessary, thus  causing these



additives to  be economically unattractive.
                                     2-6

-------
       Experiments conducted by ER&E and KVB show that for  typical
conditions, an ammonia injection rate of 2.0 (molar ratio of  NH_  to
initial NO concentration, (NH./NO)) achieves the optimum maximum  process
efficiency.  Figure 2-2 shows that minimal  additional NO reduction  is
obtained by increasing the ammonia injection rate beyond the  NH,/NO
molar ratio of 2.0.
                                   23     456
                                       (NH3)/(NO)
        Figure 2-2.  Effect of NH3  injection  rate  on  NO  emissions
                     (Reference 2-2).
       Ammonia  injection rates  depend  on  the  initial  concentration of
nitric oxide.   Experimental  data  illustrated  in  Figure 2-3  show that lower
molar ratios of NH.,/NO  are needed  to achieve  a given  process  efficiency
when the  initial  NO concentration  is greater  than  400 ppm.  These
experimental data further  indicate  that  the percent  oxygen  in the flue gas
may also  have some effect  on  required  NH^ injection  rates.  Minimum
amount of dilution with excess  air  decreases  the volume of  flue gas to be
                                     2-7

-------
treated and increases  initial  NO  concentration thus possibly reducing  the

amount of NH., needed.
           1.0 o-
           0.8  -3
         1 0.6  -
                     I         I
                  EXCESS OXYGEN:  22
                  TEMPERATURE: 960°C (1760°F)
                  INITIAL NO LEVEL (PPM)   ~~
                  D 100
                  A 200
                  O 100              _
                  O 680
                  O 1050
       Figure  2-3.
  1 .        2        3        1-       5
              (NH3)/(NO)

Effect of initial nitric  oxide  concentrations on
reductions with ammonia  injection (Reference 2-2).
        The  ammonia injection rate, the reaction  temperature and the

residence  time are critical in maintaining ammonia  emissions at minimum

levels.  Test data depicted in Figure 2-4  indicate  that the level of

unreacted  ammonia at the injection temperature  of 965°C (1770°F)

increases  significantly only at NH3/NO ratios greater than 2.0, as

expected.   When the reaction temperature  is  lowered to 870°C (1600°F)

the  level  of ammonia carryover is substantially increased because of the

slower  chemical reaction.  Therefore, the  ammonia emission level can be

controlled by allowing the reaction  to occur at slightly higher

temperatures than the optimum 955°C  (1750°F).   In fact, NH3
                                      2-8

-------
injection, at temperatures above 1000°C  (1830°F) virtually  all  NH,
breakthrough is eliminated.
                   2400
                   2000 -
                                    (NH3)/(NO)
       Figure 2-4.  Effect of NH3 injection rate on NH3  carryover
                    emissions (Reference 2-2).
       Poor mixing of ammonia with the flue gas may  cause  ammonia
carryover to occur also at NH../NO molar ratios much  lower  than  2.0.   In
fact, large scale applications of the Thermal OeNO   Process  have shown
that NH-/NO ratios in nonideal gas conditions generally  should  be  lower
than 1.5 to maintain minimum NH_ emissions.  High  levels of  ammonia
breakthrough were caused by high ammonia  injection rates combined  with
ineffective mixing or low flue gas temperatures.
       In summary, NH_/NO molar ratios can vary from 1.0 to  2.0 in large
scale applications of the Thermal DeNO^ Process.  The  actual  injection
rate used will depend on the desired NOX  reduction and could be limited
by ammonia breakthrough.  Therefore, the  injection rate  is the  result of
                                    2-9

-------
an optimization performance  study taking into account flue gas  conditions



and source configurations.



2.1.3  Hydrogen and Other Additives



       The Thermal DeNO  Process can be applied over a greatly  widened
                       A


range of temperatures  if certain additives are injected with  the  ammonia.



Of the many additives  investigated, hydrogen is the most effective over



the temperature range  from  700 to 1010°C (1290 to 1850°F).  Figure 2-5



shows the shifting effect of hydrogen injection on optimum reaction



temperature measured  in  a commercial size firetube boiler.  The magnitude



of this shift depends  on the amount of H_ injected relative to  the



NH.,.  For example, at  H_/NH, molar ratios on the order of 2:1 selective
  O                    CO


noncatalytic reduction of N0x can be made to occur at 700°C (1290°F)-..



Thus, by carefully selecting the H. injection rate, flue gas  treatment can



be controlled over a  wide  temperature range.
         200












         150

       E

       a
       >






       £ 100

       e

       M
       V*


       &




          50
                          NO . w/0 H-

                           *  ,  Z
                          Injection
                                       NHJt with H-   ^  NH3' w/0 H2

                                       Injection     N^nject1on
           700
800             900


 Flue Gas Temperature, °C
1000
 Figure  2-5.   Thermal  DeNOx reaction products as  functions  of temperature

               with and without hydrogen injection (Reference 2-3).
                                     2-10

-------
        Exxon also investigated the use of combined additives.  A mixture

 of 50 percent hydrogen and 50 percent methane was found to be more

 effective than either hydrogen or methane alone.   However, the

 introduction of methane in combustion gases,  especially at low excess air

 levels,  increased cyanide emissions by a few  ppm.

        Additive injection can also be used to control  ammonia breakthrough

 emissions to concentrations lower than 10 ppm.   For example,  small  amounts

 of \\2 injection with  amnonia would lower the  optimum reaction temperature

 from 955°C (1750°F) to 945°C (1733°F).   This  10°C (18°F)  temperature

 differential is sufficient to deplete some excess NH3  which would otherwise
 exit from the stack.

        Because \\2 can control  the temperature of  the NH.-KO-O-  reaction,

 Thermal  DeNOx is  technically feasible for most  boilers provided that the

 hardware can be installed within  the  boiler configuration.  However,  the

 cost of  the  DeNOx Process  is  greatly  increased  because of  the large

 volumes  of hydrogen needed.   This is  especially the case for  very low

 temperature  such  as 760°C  (1400°F).

       Until  recently,  the  NH3  injection  was  limited to boiler

 cavities.  Thus  if the  optimum  reaction  temperature  of about  955°C did

 not  occur  in  an  isothermal  cavity,  NH^ was  injected  at  temperatures

 below this level.  This situation warranted the use of an  additive to

 maximize the  efficiency of  the  DeNO   Process.   The  in-tube bank
                                    A
 injection  of  NH3, recently  demonstrated by Exxon, has  diminished  the

 dependence of  the process efficiency  on the use of an  additive.   DeNO
                                                                      x
 rates of 60 to 70 percent were  achieved by  injecting NH  in tube  banks
                                                       <3
without the use of an additive.
                                    2-11

-------
2.1.4  Byproduct Emissions
       The Exxon Thermal DeNOx Process may form byproduct pollutants
directly or indirectly from the presence of NH3 in the combustion gas.
Potential byproduct emissions suggested by ER&E are NH3> CO, HCN and
N20 and, when sulfur-bearing fuel is burned, NH3 and S03 combine to
form ammonium bisulfate, NH^HSO^.
       Ammonium bisulfate is a viscous liquid from 147°C to about
450°C (300-840°F).  It has been known to cause corrosion of metal
surfaces.  Thus far, however, no increase  in metal corrosion attributable
to anmonium bisulfate has been identified  when Thermal DeNOx has been
used.  The formation of NH4HS04 can be controlled by limiting the
amount of NH, carryover.  This can be accomplished by NH3  injection at
a temperature slightly  higher than optimum or  by  using  an  H? additive.
In general, ammonium bisulfate  is considered the  most serious byproduct
and  one  which could effect  the use of the  Thermal DeNOx Process.
        Carbon monoxide  emissions may  also  be promoted by ammonia injection
because  the Thermal DeNO   reaction  inhibits the oxidation  of CO to
                         A
CO  .   Thus,  if  there  is  unburned CO at  the point  of  NH3 injection, the
CO may not  be  oxidized,  but will be discharged to the atmosphere.   Under
 normal  operating  conditions, CO  levels  are not usually  significant in
 steam generators.  Using hydrocarbons as additives  to control  the
 NH3-NO-02 reaction increases the concentration of CO in the flue gas.
 Exxon reported that as much as 50  percent of  the  hydrocarbons  may be
 oxidized to CO.  This  CO may then  be  emitted  to the atmosphere because  the
 ammonia inhibits the 02+CO-*~C02  reaction.
        HCN is formed only if hydrocarbons are present in the region  in
 which NH3 is injected.  Under normal  boiler operation,  gaseous
                                     2-12

-------
hydrocarbons are not present unless they are injected along with the



NH^.  KVB reported that for gas, oil and coal firing, HCN was present in



the untreated flue gas at 3 to 10 ppm concentration, depending on excess



air level.  Injection of NH. did not measurably affect the HCN level.



       The reduction of NO by NH3 and 0- forms N-O as a minor byproduct.



However, less than 2 moles are generated for every 100 moles of NO



reduced, according to ER&E experimental data.  All the available evidence



indicates that N?0 is relatively harmless at those levels, and does not



represent an environmental concern.



2.2    SUBSCALE TESTING -- COAL



       Recently, KVB has conducted a pilot-scale  investigation of the



Thermal DeNO  Process to reduce NO levels from combustion of coal
            A


(Reference 2-4).  The work was sponsored by the Electric Power Research



Institute (EPRI) and Exxon Research and Engineering  (ER&E).



      ' The major objective of this investigation  was to determine the



level of NO  reduction which is achievable in flue gas resulting from
           A


coal combustion.  The primary variables investigated were the injection



temperature, the NH.,/NO ratio, and the coal type.  Additionally, a



hydrogen additive was used to lower the temperature  range for NO removal.



Four different coals were investigated; three coals  were bituminous and



one subbituminous.  Byproduct emissions were also measured at different



NH3 injection rates.



       The combustion facility consisted of a 0.9 MW (3 x 10  Btu/hr)



firetube boiler equipped with a ring-type natural gas burner and a scaled



down version of a commercial coal burner presently used in utility boilers



firing Western coal.  The NH- injection system consisted of five
                                    2-13

-------
injectors located at the end of the firetube section distributing the
ammonia and nitrogen (carrier gas) counter-flow to the flue gas stream.
The injectors were designed to be movable so that they could be positioned
axially along the length of the firebox thus providing for evaluation of
the effectiveness of different temperature profiles.  The injection method
was a result of an optimization study in which the injection grid and
nozzles were designed to provide substantial NO  reductions that allowed
                                               A
a valid comparison between the various coal types and natural gas.  Since
the injection method directly affects the efficiency of the Thermal
DeNO  Process, the results achieved by KVB do not necessarily represent
    ^
the maximum NO  reductions achievable with coal combustion.
       This section discusses the  results of this investigation.  These
results can be used to  compare noncatalytic NO reductions and byproduct
emissions between coal  and the gaseous and  liquid fuels previously
investigated.  Key parameters considered here are again:
       o   Reaction temperature
       •   Ammonia  injection  rate
       »   Hydrogen and other  additive injection
       •   Byproduct emissions
2.2.1  Reaction  Temperature
       The temperature  at  which  ammonia  is  injected  into  the  flue  gas is
the primary  variable which determines  the  amount  of  NO  removed  with the
Thermal  DeNO   Process.   A  major  objective  of  the  KVB study  was  to
determine  whether the  additional  pollutants resulting from  combustion of
coal,  such  as  S02 and  particulates,  would  influence the temperature
 dependence of  the process  or reduce  the  process  efficiency.  Figure 2-6
 shows  the  effect of reaction temperature on NO  reduction  for the four

                                     2-14

-------
TVA
          U.S. Environmental
          Protection Agency
          Office of Research
          and Development
          Industrial Environmental Research
          Laboratory
          Research Triangle Park NC 27711
Tennessee
Valley
Authority
National Fertilizer Development
Center
Muscle Shoals AL 35660
                                   TVAY-134
          Impact of Ammonia
          Utilization by NOX Flue
          Gas Treatment Processes

          Interagency
          Energy/Environment
          R&D Program Report

     «85Sa«aeaa&ffigBa8s

                         ~j>ffXS£l££&'IS&9f3X

-------
                                 INTRODUCTION
     Of the five most common air pollutants released In the United States,
nitrogen oxides (NOX) are the only ones projected to increase significantly in
the future.  The total emission of the other four pollutants [particulate
matter, sulfur oxides (SOX), hydrocarbons, and carbon monoxide (CO)] has
decreased during the past 6 yr and is expected to decline further in the future
as more point sources are equipped with more efficient control systems.  With
most public attention focused on these other pollutants which are typically
emitted at higher rates than NOx, the development of equipment and techniques
for controlling NOX has lagged behind.  However, recently more attention has
been given to the possible health effects of NOx emissions.  Although the NOx
emission from each source appears to be small, the cumulative total is measured
in millions of tons annually and is increasing substantially from year to year
due to the increasing number of sources.

     NOX emissions are divided into two classes, mobile or stationary, depending
on the source.  The mobile class, as its name implies, includes sources such
as automobiles, trucks, buses, trains, and planes.  Attempts at controlling
the NOx emissions from these sources have been delayed due to both technical
and economic considerations.  Stationary sources, which release about 56% of
the total  amount of NC^ emitted in the United States (an estimated 11:15 Mtons
in 1975),  are split into three groups:  combustion sources, industrial processes,
and other  miscellaneous sources.  In 1975 combustion sources contributed approxi-
mately 93% of all stationary source emission  (an estimated 10.4 Mtons) of which
nearly two-thirds came from fossil fuel-fired boilers  (approximately 7.0 Mtons).
If these boilers are further divided according to size (heat rate of the boiler),
large boilers  [i.e., greater than 250 MBtu/hr (25 MW equiv)] release 45% of al
NOX emitted from stationary combustion sources.  Figure 1 and Table 1 show the
relative contributions from mobile sources and each type of stationary source
to the total NOX emissions  in the United States during 1975.  Since the mobile
source standards are not expected to become more stringent in the near future,
restricting the amount of  NOx emissions from  stationary sources, particularly
large  fossil fuel-fired boilers, may becone necessary  since these boilers are
the second largest source  of NOX emissions in the U.S.  The reason  for the
possibility of stricter NOX emission control  on  large  boilers are threefold.
The emissions  from these boilers are projected  to increase at about 4.9%  annually
over  the next  decade  (doubling  every  15 yr) assuming present control  levels.
Secondly,  assuming the application of  combustion modification which is now
considered the bast  available control  technology  (about 30% decrease  in  the
NOx formation),  the  amount of N0.{ emitted  by  large  boilers will  increase  from
4.7 Mtons  in  1975  to  6.6 Mtons  in 1985  (50).  And thirdly, although the  total
amount of  NOx  emitted by  these  boilers  is  only  approximately one-half  that of
mobile sources,  there are  orders-of-magnitude fewer boilers and  hence  each
boiler source  emits  significantly more  NO^ than  each  individual  mobile  source.

-------
Industrial processes (2.3%)
        Gas turbines (2.8%)
               Other (1.5%)
                                            1C
                                          engines
                                          (14.3%)
                                                  Other
                                                 boilers
                                                 (11.2%)
Mobile
sources
(44.3%)
                                           Large
                                          boilers
                                          (23.6%)
Figure 1.  Breakdown of the total U.S. NOX emissions by original source (45)

-------
 Therefore,  it is expected that based on each ton of HO* removed it will be
                            "
               TABLE  1.  NOX  EMISSIONS  BY SOURCE IN 1975 (45)


                                           Annual amount,
                                               ktons	

                Mobile sources                 3 850
                Stationary sources
                  Large boilers                4,728
                  Other boilers                2*238
                  1C engines                   2,'849
                  Gas turbines                   555
                  Incinerators                   39
                  Industrial processes           454
                  Field burning                  290

                     Total                   20,000
FORMATION OF NOX IN LARGE BOILERS

         vS  f?™cd  during hiSh-temperature combustion operation and can be
                                                                   t
 hP.?  ??  v 2  /    S   temperatures.   The  second  method  is  the oxidation of
 neaically bound nitrogen within  the  fuel and  is called "fuel  NO! "   i^oli
                                     y b°th — "-isms,  but in
POTENTIAL NOX CONTROL METHODS FOR LARGE BOILERS

              NO, e,isston regulation, the e^has   is on fLe gas trteent

-------
     Cotabus t ion jnodif icat ion techniques attempt to prevent the formation
of NOx by altering the reaction conditions inside the boiler.  Since  thermal
fixation is primarily a function of the temperature  (i.e., the rate of  fonaatic-
increases with temperature) and the concentration of 02 in the boiler,  reducin-
either of these parameters will decrease the amount of thermal NOx formed.  This;-
same boiler modifications which reduce the 02 concentration  in the boiler can
also reduce the amount of fuel NOx formed since the conversion of fuel-bound
nitrogen is a strong function of the 02 concentration.  Various combustion
modification techniques, which are currently undergoing development work in the
United States, are compared in Table 2 for the three types of fossil  fuel-
fired boilers.  Combustion modifications are shown to be  the most effective on
gas- and oil-fired boilers where a major portion of the NOx  is formed by the
thermal mechanism.  For coal-fired boilers where much of  the NOx comes  from
fuel-bound nitrogen, the available combustion modification techniques are
limited to about 40% reduction in
     Although the resulting reduction in the total NOx emissions from  large
boilers would appear to be significant, a recent study (45) has determined
that, even with combustion modifications on all new  large  fossil-fueled boilers,
the expected growth rate for new large boilers will  result  in a significant
increase in the total NO^ emissions from these boilers,  rising from  4.73 Mtons
in 1975 to 6.56 Mtons in 1985.  Unfortunately, combustion modifications also
have undesirable side effects of reducing boiler efficiency, increasing the
potential for flame instability, and increasing the  soot, CO, and particulate
loadings in the flue gas.  For these reasons, an alternative method  of NOx
control technology, FGT, has recently begun to receive more attention.  Although
probably aore expensive than combustion modification, FGT allows normal operatic:.
of the boiler and can remove at least 90-95% of the  NOx  zrcm the flue  gas.  Many
different types of FGT systems are currently undergoing  development  work in
Japan and the United States, but the most technically advanced type  for removing
NO^Jfrora power plant stack gas is now selective catalytic  reduction  (|SCR)  (30).
In chis type, ammonia CNH3) is injected into the flue gas ducts and  the mixture
of NH3 and flue gas passes through a catalytic reactor containing a  base-metal
catalyst.  The NOx *s selectively reduced to molecular N2  by the following
reactions:
              4NH3(g)  + 41IO(g)  + °2 *

                4NH3(g) + 2N02(8) + °2'*

      Although these SCR processes are still in early stages of development
 (most have not been tested either on coal-fired flue gas or on a large scale
 unit),  they are receiving considerable attention because of both the potentially
 low capital investment and the fact that they generate molecular ^ directly
 without further chemical processing.

-------
        TABLE 2.  EFFECTS OF BOILER MODIFICATIONS  TO  REDUCE NOX

                   EMISSIONS BY FOSSIL FUEL TYPE  (30)
                                        Percent  decrease of NO^ formation
                                        	in boilers by fuel type
          modification
    Gas
                                                        Oil
                          Coal
Prevention of thermal NOX by:
  FLue gas recirculation
  Reduced air preheat
  Steam or water injection
Prevention of both  thermal and  fuel
 NOx  by:
  Staged combustion
  Low excess air
  Reduced heat release  rate
  Combination of stage  combustion,  low
   excess air, and  reduced heat release
   rate
Prevention of fuel  NOx  by:
  Change  to  fuel with lower  percent
   nitrogen
60
50
60
55
20
20
50
20  Not effective
40  Not competitive
40  Not competitive
40  40
20  20
20  20
35  40
Not applicable  40   20

-------
     A primary consideration in the widespread application of these SCR
processes in the United States is the potential availability and the cost of
NH3 in the future.  Since nearly all of the NH3 produced in the United States
is made from natural gas, there are serious concerns over the potential impact
of widespread use of NH3 in FGT applications.  The availability and cost of
both natural gas and NH3 may be affected.  Would this additional demand for
NH3 create shortages in NH3~based fertilizers and thus increase food costs?
Will there be sufficient natural  gas  available  for meeting  this demand  for
NH3?  If not, what other methods and feedstocks are available for generating
NH3?  This report will attempt to provide insights .into  these questions.

-------
                       NH3 REQUIREMENTS FOR NOx CONTROL


 NH3 REQUIREMENTS  FOR TYPICAL POWER PLANT APPLICATIONS

      Any evaluation of the impacts of the additional NH3 demand for power
 plant FGT systems  must begin with a determination of the magnitude of the
 additional demand.   In order to  calculate the NH3 requirements  for typical
 fossil fuel-fired  boilers,  it is necessary to presuppose future NO., standards
 for large stationary source boilers.

      Although  the  future  NOX emission limits  have not been  published,  for the
 purposes of this study the  following  scenario has been hypothesized.   Since
 the SCR and other  FGT processes  are still in  the  early stages of development
 in  the United  States,  strict Federal  NOX emission regulations of 90%  removal
 from large stationary sources will probably not be enforced until  1985.   Since
 Federal new-source  performance standards (NSPS) apply only  to new  sources,
 existing sources will not be required to meet these  standards but  existing
 sources converted  from gas  or oil  to  oil or coal  after 1985 will be.   Between
 the present and 1985,  the NOX emission  limits may be reduced but only  to  the
 extent  that combustion modifications  can be used  to  meet  the newer regulations.

      Two possible methods would  be available  for  meeting  these  eml&sinn regula-
 tions of 90% NOX removal,  (1) the  installation of  an 1CR  system  designed'for
 90% NOX removal or  (2)  the  use of  combustion  modi ft.c.a^ilQ.n.f*  tn cut  NOx  emissions
 from the boiler by  50% with  the  additional installation of  an SCR  system  designed
 for 80% NOX removal  to give  an overall NOX removal efficiency of "90%.  The
 impacts  of  each of  these  methods are  considered in later  sections.

     The typical fossil fuel-fired'boilers included  in  this study were 500-MW
 coal- and  oil-fired  boilers with the  flue gas  compositions  shown in Table 3.
 Gas-fired  boilers were not considered since they are  not  expected  to be built
 after 1985.  For the  two  alternative  cases, the uncontrolled conventional 500-MW
 coal-fired  boiler would release about 600 ppm NOX in  the  flue gas  (3009 Ib/hr),
would require an SCR system capable of a 90% NOx removal  efficiency, and would'
 release  only approximately 60 ppm NOx (300 Ib/hr) of NOX  after FGT control.
Through  the use of combustion modification techniques in  the second coal-fired
case, the boiler would emit 300 ppm NOX  (1504  Ib/hr) and  the additional SCR
system would remove 80% of the remaining NOX  to obtain a  total of 90% NO
removal.   In a similar manner, the uncontrolled 500-MW oil-fired boiler releasing
200 ppm NOX in the flue gas  (856 Ib/hr) would be required 'to have 90% NOX removal,
i.e., down  to 20 ppm  (86  Ib/hr) NOX.  Again this could be obtained either by an
SCR system designed for 90% removal or through combustion modification to lower
boiler emissions to 100 ppm NOX and then an 80% efficient SCR system.

-------
                   TABLE  3.   FLUE  GAS  COMPOSITIONS  FROM 500-MW

                         COAL- AND OIL-FIRED  BOILERS
           Constituent^

           N2
           °2
           C02
           S02
           303
Oil-fired
Vol, %
73.60 2
2.54
11.96
0.13
0.0013
0.02
^
11.75
boiler
Lb/hr
,929,000
115,400
747,900
12,060
151
856
-
300,800
Coal-fired
Vol, %
73.76 3
4.83
12.31
0.24
0.0024
0.06
0.01
8.79.
boiler
Lb/hr
,450,000
258,200
904,200
25 130
317
3,009
661
264,500
           HC1
           H20
                         100.00     4,106,000   100.00     4,906,000

           Fly.ash, gr/sft3  (wet)        0.032                   6i06
 lllTl^r*?'™ inSldf thlS rea"°r «e * '-?•» "•» =" "bout 3£Z£c









 the range of 1.0-1.1:1  for 90% NOX  rental  in a cLier^l sy««    "
NO   r                                     »ol  ratios  for  the 90% and  the 80%

p?an t"  rk   i^e  "l  0^1  S^ll'?  ^  10%  h±gher  th" ^ USed in  ^S '"«-






oil-fired boiler without  combustion modification, and (4) 500-MW oil-fired  boil!?
vith combustion modification.  The estimated annual NH3 consumptio^ I for each of  these

-------
JOO
0
         Figure 2.
               0.6        0.8         i.
                MOL N1I3/MOL NOX
NOX removal efficiency as a function  of
                                                                     1.2
                                                                    mol ratio.
                                                                                1.4

-------
10
                LeoencL
          500-MW conventional coal
          500-MW coal with CM
          500-MW conventional
          500-MM oil with CM
                         1.0
     1.5        2.0
MOL NH3/MOL NOX
                                    10

-------
four cases is listed in Table 5.  The values range from a high of 6266 tons
of NH3/yr for the 500-MW conventional coal-fired boiler without combustion
modifications to a low of 772 tons of NH3 annually for the 500-MW oil-fired
boiler with combustion modifications.
                  TABLE 4.  PREMISES FOR THE CALCULATION

                   OF ANNUAL NH3 CONSUMPTION FOR 500-MW

                           FOSSIL-FIRED BOILERS

Parameter
Heat rate, Btu/kWh
Heating value
Coal, Btu/lb
Oil, Btu/gal
Availability, hr/hr
Value
9,000
10 , 500
144,000
7,000

                    TABLE 5.'  NH3  CONSUMPTION  FOR TYPICAL

                      500-MW COAL- AND OIL-FIRED  BOILERS



Fuel
Coal
Coal
Oil
Oil
NOx
treatment
scheme
SCRa
CM & SCRb
SCR
CM & SCR
NOX
concentration
in flue gas, oom
600
300c
200
10 Oc
Mols NH3
per mol
NOx
1.05
0.91
1.05
0.91

NH3 consumption,
tons/yr
6,266
2,715
1,781
772
       a.   Selective catalytic reduction.
       b.   Combustion modification followed  by  selective  catalytic
           reduction.
       c.   Combustion modification was  assumed  to be 50%  efficient
           in controlling NOX emissions.



      The economics of present-day NH3  generating plants  dictate a  minimum capacity
 of 1,000 tons of NH3/day (330,000 tons/yr)  and thus local,  small NH3 plants at
 each boiler would be highly unlikely.   Even with large 2,500-MW coal-fired power
 plants,  the annual NH3 requirement of  31,000 tons would  not justify a captive
 NH3 plant.  Thus it is assumed that for all NOX FGT applications the NH3 would
 be purchased from existing fertilizer  plants already producing NH3 and shipped
 to the power plant and stored in large tanks onsite.
                                       11

-------
VI.  HYDROCARBONS

-------
                                HYDROCARBONS






SOURCES




     Hydrocarbons by definition contain the elements  hydrogen and carbon.   In




addition, naturally-occurring,  processed,  and synthetic hydrocarbon molecules




may contain other elements such as oxygen, nitrogen,  sulfur,  and halogens.




     The health effects of hydrocarbon and organic solvent emissions are of two




types, direct and indirect.  Direct effects are caused by the original,  unal-




tered emissions, and indirect effects are  caused by substances formed by photo-




chemical reactions of the original emissions with other substances in the




atmosphere.




     Generally, hydrocarbon materials at levels encountered in the ambient  air




have no direct health or welfare effects and the prime reason for controlling




hydrocarbon emissions is to prevent their  participation  in atmospheric  photo-




chemical reactions.  Olefins are regarded  as being the most reactive of  the




organic compounds in photochemical smog formation, although reactivity varies




widely with chemical structure.




     As in the case of nitrogen oxides and carbon monoxide, the largest  single




source of hydrocarbons is transportation.   Stationary sources of hydrocarbon




emissions include petroleum refining, gasoline distribution and marketing,




chemical manufacturing, coal coking, fuel  burning, waste disposal, and food




processing.  Sources of organic solvent emissions include manufacture and appli-




cation of protective coatings,  manufacture of rubber and plastic products,




degreasing and cleaning of metal parts, dry cleaning operations, printing,  and




manufacture of chemicals.  A summary of total nonmethane hydrocarbon emissions




by source is given in Table I.




     From the production of crude oil to the marketing of finished products, the




petroleum industry has the potential for emitting significant quantities of

-------
                                    - 2 -






hydrocarbon gases and vapors.   These emissions are often undesirable since they




may be precursors of photochemical smog.  Crude oil is first produced from the




ground.  Then the liquid hydrocarbons are separated from the gases, light




hydrocarbon vapors, and water.  Finally, the crude oil is stored until it is




removed to the refinery where it is converted to salable products.  In crude




oil production, most of the emissions are due to evaporation of hydrocarbons




from storage tanks.



     The design of a refinery depends on the kind of- crude oil it processes and




on the final products it manufactures.  Refinery operations are most easily




discussed, therefore, in terms of their similar functions.




     Since crude oil as it is produced has few uses, it  is processed to obtain




salable products, such as gasoline, kerosene, fuel oil,  petrochemical raw mate-




rials, waxes,  lubricating oils, and asphalt.  Processing involves  four major




steps:  separation, conversion, treattnent, and blending.




     The first refining step, separation by distillation within a  specific




temperature range, yields fractions,  the relative  volumes of which are deter-




mined  by the nature of the crude  oil.   These  fractions  are usually further




refined to meet  the demands  for the  various petroleum products.   These processes




are  outlined below.



      Conversion  by  cracking  is  employed to convert high-molecular-weight  hydro-




carbons  into products  of  lower  molecular weights.   For  example> cracking




partially  converts  heavy  gas  oil  to  gasoline.   If a catalyst  is used (the more




usual case),  it  is  called catalytic  cracking,  if  not,  it is  thermal cracking.




Thermal  cracking requires higher  temperatures and pressures  than  those  required




 to catalytic  cracking.



      Gasoline  yield and quality can be improved by several  other  processes.  In




 catalytic  reforming,  the  molecules of the  gasoline feed stock are rearranged




 and dehydrogenated to produce high-octane  gasoline blending stocks.

-------
                                    - 3 -

Isomerization rearranges molecules to increase the octane number;  it also
increases molecular branching, but it does not add to or remove anything from
the original material.  In still other conversion processes, liquid gasoline is
made from the hydrocarbon gases generated during cracking.  Polymerization joins
an olefin with a branched chain paraffin to yield a saturated hydrocarbon.
     Treatment steps are used to purify the material or to prevent an undesir-
able reaction vith an impurity.  For example, during selective, hydrogenation,
sulfur and nitrogen as impurities in the feed stocks are converted to hydrogen
sulfide and ammonia, respectively.  In. addition, olefins and aromatic compounds
may be hydrogenated to partial or complete saturation.  Three types of treat-
ment employed are acid treatment, "sweetening," and solvent extraction.  Petro-
leum fractions may be brought into contact with concentrated sulfuric acid to
remove sulfur, nitrogen, and undesirable unsaturated compounds and to improve
color and odor.  Sweetening converts mercaptans to disulfides and this improves
odor.  Sodium plumbite (doctor), lead sulfide, hyprochlorite, and copper chloride
are common sweetening agents.  In solvent extraction, solvents are used to remove
undesired contaminants or to concentrate desired components.
     Physical treatments such as absorption, air-bloving, electrical coalescence,
and filtration are used in intermediate refining processes to remove contaminants.
     Another commonplace activity at refineries is the blending of base stocks to
produce a wide variety of finished products.
     A list of equipment, facilities, and processes likely to produce organic
emissions in crude oil production and in refining includes:
                      1.  Storage.
                      2.  Pressure relief valves.
                      3-  Slowdown systems.
                      U.  Flares.
                      5-  Catalytic cracking units.
                      6.  Asphalt oxidation.

-------
                                    -  k -

                     7.   Chemical treatment.
                     8.   Loading facilities.
                     9-   Oil-water separators.
                    10.   Pumps.
                    11.   Valves.
REGULATIONS
     Hydrocarbon emission standards for refineries and related petroleum indus-

try 'facilities were not included in the performance standards for new stationary

sources which were issued in December, 1971> by the Federal Environmental Protec-
tion Agency.  The EPA has indicated, however, that petroleum refineries will be

a source classification in the next group of standards issued and that hydrocar-
bon emission standards will be included for all types of producing and refining

activities.  It is anticipated that proposed standards will be issued before

mid-1972.
     To assist the state regulatory agencies in preparation of implementation

plans, the EPA included extensive model regulations for control of hydrocarbon

emissions  in the proposed implementation plan guidelines which were issued in

April, 1971.  The model regulations, which vere patterned quite closely after

the Los Angeles County hydrocarbon  emission regulations, require that:
     1.  Storage facilities for volatile hydrocarbons (> 1-5 psia) larger than

Uo,000 gallons must be equipped with floating roofs, vapor recovery systems, or

other control devices to control evaporative emissions.  Further, the use of

floating roofs is limited to hydrocarbons with a vapor pressure of less than

11.0 psia.
     2.  Storage tanks for volatile hydrocarbons larger than 250 gallons must
be  equipped with a permanent submerged  fill pipe or other evaporative control

device to  reduce hydrocarbon emissions  during filling operations.
     3-  Loading facilities for volatile hydrocarbons must be equipped with a

vapor collection and  disposal or recovery system to control  losses during

-------
                                    -  5 -



loading operations.


     U.  Oil-water separators which receive more than 200 gallons per day of


volatile hydrocarbons must be equipped with a solid cover, floating roof, vapor


recovery system or other means to prevent hydrocarbon losses.


     5-  Pumps and compressors handling volatile organic materials must have


mechanical seals to keep leakage to a minimum.


     6.  Hydrocarbon gases from vapor blowdown systems must be burned in smoke-


less flares.


     7.  Emissions of organic solvents from numerous varied operations must be


controlled through the use of such methods as incineration, carbon adsorption,


and vapor recovery.


     In the final  implementation plan  guidelines issued by EPA in August, 1971,


the model hydrocarbon emission regulations were not significantly modified.  It


was emphasized, however,  in  the final  guidelines that the major  reason for con-


trolling hydrocarbons is  to  prevent photochemical  oxidant production  and that


control of  stationary hydrocarbon  sources would be necessary only  in  areas
  \

where  the anticipated reduction obtained from the  motor  vehicle  standards would


not be sufficient  to result  in attainment of  the oxidant ambient air  quality


standard.


      Virtually all states have proposed or  already adopted  some  hydrocarbon


emission  regulations primarily based  on the EPA model standards.  The petroleum


 industry, through the  state  petroleum councils  and individual  company state-


 ments, has  presented extensive  testimony at state  hearings  which have been  held


 on  the proposed hydrocarbon  regulations.   Principle  points  covered by the indus-


 try include requests that:


      1.  Applicability of any hydrocarbon regulations be limited to those areas


 where the photochemical oxidant standard is met or exceeded.   This would limit

-------
                                    - 6 -






imposition of costly control requirements to those major metropolitan areas




where there currently is a photochemical smog problem and would be consistent




with latest EPA recommendations.




     2.  The definition of volatile hydrocarbon be changed to mean any material




with a vapor pressure of over 2.5 psia instead of 1.5 psia.  This would except




JP-k jet fuel, heavy naphtha solvents and feed stocks and kerosene-type materi-




als from the requirement of being stored in floating roof tanks and being loaded




in facilities with vapor recovery systems.  This change could also make a dif-




ference in the number of oil-water separators which need to b-s covered.




     3.  The use of floating roofs be allowed for hydrocarbons with a vapor




pressure of up to 12-5 psia instead of 11.0 psia.  This would eliminate the




need to store materials such as LVN and LCN in pressure tanks during the summer.




     k.  The requirement for vapor recovery at loading facilities be limited to




larger facilities which load more than 20,000 gallons per day.  This would




except bulk plants which typically load only several truckloads per day.




     5-  The requirement for mechanical seals be limited to rotary and centrifu-




gal pumps and compressors  since mechanical seals are not applicable to recipro-




cal units.



     6.  A differentiation be  made between vapor blowdown  systems and  emergency




relief devices and  that the requirement  for smokeless flares be  limited  to blow-




down systems.  This change would preclude the need to connect  emergency  pressure




relief  valves, which  are  infrequently,  if ever, used, to  the blowdown  and  flare




system, which  is  normally  used during startups, shutdown,  and  minor upsets.




      Some of the  industry  recommendations have been  accepted by  some of  the




states.   In  others,  compromise versions  have  been  adopted or the application of




the  most  stringent  regulations has  been limited  to new  facilities.   It is  likely,




when all  the states have  completed adoption of  regulations and implementation

-------
                                    - 7 -






plans later this year, that the hydrocarbon emission regulations will vary con-




siderably from state to state and that these variations will be disconcerting




to multiple-facility companies attempting to comply with regulations in several




states.




STORAGE




     Storage is potentially the most important source of hydrocarbon emissions




in the petroleum industry.  Vapors can be emitted when storage tanks "breathe,"




when vapors are displaced during filling, and when liquids evaporate.  Tanks




"breathe" due to the expansion and contraction of their contents with the heat.




of the day and the cool of the night.  When the contents expand, air mixed with




hydrocarbon vapors is forced out of the tank.  Methods have been developed to




estimate losses and to minimize these losses from storage tanks.




     Even in the most modern petroleum refineries and petrochemical plants,




storage facilities must be provided for large volumes of liquids and gases.




These facilities can be classified as closed-storage or open-storage vessels.




Closed-storage vessels include fixed-roof tanks, pressure tanks, floating-roof




tanks, and conservation tanks.  Open-storage vessels include open tanks,




reservoirs, pits, and ponds.




     Closed-storage vessels are constructed in a variety of shapes, but most




commonly as cylinders, spheres, or spheroids.  Steel plate is the usual material




of construction though concrete, wood, and other materials are sometimes used.




Before modern welding methods, the sections of the tank shell were joined by




rivets or bolts.  Welded  joints are now used almost universally except for the




small bolted tank found in production fields.  Capacities of storage vessels




range from a few gallons  up to 500,000 barrels, but tanks with capacities in




excess of 150,000 barrels are relatively rare.




     Open-storage vessels are also found in a variety of shapes and materials




of construction.  Open tanks generally have cylindrical or rectangular shells

-------
                                    - 8 -






of steel, wood, or concrete.  Reservoirs, pits,  ponds,  and s-imps are usually




oval, circular, or rectangular depressions in the ground.   The sides and bottom




may be the earth itself or may be covered with an asphalt-like material or con-




crete.  Any roofs or covers are usually of vood with asphalt or tar protection.




Capacities of the larger reservoirs may be as much as 3 million barrels.




     Vapors, gases, aerosols, and odors are examples of air contaminants emitted




from storage facilities.  In most cases, practical and feasible air pollution




control measures are available to reduce the emissions.




Pressure Tanks and Fixed-Roof Tanks




     Pressure tanks and fixed-roof tanks are grouped together because, in a




sense, pressure tanks are special examples of fixed-roof tanks designed to




operate at greater than atmospheric pressure.  Horizontal, cylindrical (bullet)




pressure tanks are the most common pressure tanks.  Other types of pressure




tanks include spheres, plain and noded spheroids, and noded hemispheroids-




Maximum capacities of these pressure tanks are as much as 30,000 barrels for




spheres and hemispheroids, and 120,000 barrels for noded spheroids.  Spheres




can-be operated at pressures up to 217 psi; spheroids, up to 50 psi; noded




spheroids, up to 20 psi; and plain or noded hemispheroids, up to 15 and 2-1/?




psi,respectively.  Horizontal, cylindrical pressure tanks are constructed with




various capacities and pressures.




     Typical storage tanks are vertical, cylindrical, fixed-roof tanks.  This type




of storage facility operates at or within a few ounces of atmospheric pressure and




may have a flat, recessed flat, conical, or domed roof.  The term gastight, often




applied to welded tanks, is misleading.  Many of the roofs of the welded tanks




have free vents open to the atmosphere.  Others are equipped with conservation




vents that open at very slight positive pressures.  A tank also has many stan-




dard appurtenances including gaging  hatches, sample hatches, relief vents, and

-------
                                    - 9 -






foam mixers.  Any of these accessories may fail in service and result in vapor




leaks.




     The operating pressure of a tank is limited by the thickness (weight) of




the roof.  A cone roof tank may be operated at higher pressures, if necessary,




by structural reinforcement or weighting of the roof.  Safe operating pressures




up to U ounces can be realized by this added expense.  Use of unsupported dome-




shaped roofs is another method of'increasing the allowable operating pressure




of the fixed-roof tank.




Floating-Roof Tanks




     Floating-roof storage tanks are used for storing volatile material with




vapor pressures in the lower explosive range, to minimize potential fire or




explosion hazards.  These vessels also economically store volatile products that




do not boil at atmospheric pressures or less and at storage temperatures or




below.  These tanks are subclass ified by the type of floating-roof section as




pan, pontoon, or double-deck floating-roof tanks.




     Pan-type floating-roof tanks were placed in service more than 50 years ago.




These roofs require considerable support or trussing to prevent the flat metal




plate used as the roof from buckling.  These roofs are seldom used on new tanks




because extreme tilting and holes in the roof have caused more than one-fifth




of installed pan roofs to sink, and because their use results in high vaporiza-




tion losses.  Solar heat falling on the metal roof in contact with the liquid




surface results in higher than normal liquid surface temperatures,  hydrocarbons




boil away more rapidly at the higher temperatures and escape from the opening




around the periphery of the roof.




     To overcome these disadvantages, pontoon sections were added to the top of




the exposed deck.  Better stability of the roof was obtained, and a center drain




with hinged or flexible connections solved the drainage problem.  -Center-weighted

-------
                                   - 10 -






pontoons, double pontoons, and high- and low-deck pontoon floating-roof tanks




are available today.  Current practice is to use the pontoon roof on tanks with




very large diameters.  Included with some pontoon roof designs is a vapor trap




or dam installed on the underside of the roof.  This trap helps retain any




vapors formed as a result of localized boiling and converts the dead vapor




space into an insulation medium.  This dead vapor space tends to retard addi-




tional boiling.




     The more expensive double-deck floating roof was eventually introduced to




reduce the effect of solar boiling and to gain roof rigidity.  The final design




generally incorporates compartmented dead-air spaces more than 12 inches deep




over the entire liquid surface.  The top deck is generally sloped toward the




center or to a drainage area.  Any liquid forming or falling on the roof top is




drained away through a flexible roof drain to prevent the roof from sinking.




The bottom deck is normally coned upwards.  This traps under the roof any vapors




entrained with incoming liquid or any vapors that might form in storage.  A




vertical dam similar to those used on pan or pontoon floating roofs can also be




added to retain these vapors.




Conservation Tanks




     Storage vessels classified as conservation tanks include lifter-roof tanks




and tanks with internal, flexible diaphragms or internal, plastic, floating




blankets.  The lifter roof or, as more commonly known, gas holder, is used for




low-pressure gaseous products or for low-volatility liquids.  This type of




vessel can be employed as a vapor surge tank when manifolded to vapor spaces of




fixed-roof tanks.




     Two types of lifter-roof tanks are available.  One type has a dry seal con-




sisting of a gastight, flexible fabric; the other type employs a liquid seal.




The sealing liquid can be fuel oil, kerosene, or water.  Water should not be

-------
                                   - 11 -






employed as a sealing liquid where there is danger of freezing.




     The physical weight of the roof itself floating on vapor maintains e  slight




positive pressure in the lifter-roof tank.  When the roof has reached its  maxi-




mum height, the vapor is vented to prevent overpressure and damage to tank.




     The conservation tank classification also includes fixed-roof tanks with




an internal coated-fabric diaphragm.  The diaphragm is flexible and rises  and




falls to balance pressure changes.




     Two basic types of diaphragm tanks are the integrated tank, which stores




both liquid and vapor, and the separate tank, which stores only vapor.  Common




trade names for integrated tanks are "diaflote,"  "dialift," and "vapor-mizer"




tanks, or  they may be referred to as vapor spheres or vapor tanks.  The separate




type of tank offers more flexibility and does not require extensive alteration




of existing tanks.




Open-Top Tanks, Reservoirs, Pits, and Ponds




     The open-top tank is not used  as extensively as in the past.  Safety, con-




servation, and housekeeping are factors  affecting the elimination of open vessels.




Even tanks that require  full access can  and should be equipped with removable




covers.  The open vessels generally have  a cylindrical shell, but some have a




rectangular shell.



     Reservoirs were  devised to store  the large quantities of residual oils,




fuel oils, and, sometimes,-crude  oils  resulting from petroleum production and




refining.  Safety considerations,  larger fixed-roof  tanks, and  controlled crude




oil production have  reduced  the number  of reservoirs  in use  today.   Even when




covered,  reservoirs  have open  vents, which maintain  atmospheric pressures in the




reservoir.  Windbreaks  divert  the windflov pattern  over a  large roof  area and




prevent the  roof  from raising  and buckling.



     Open ponds  or  earthen pits were  created by diking low areas  or  by excava-



tion.   These  storage facilities served for holding  waste  products, refinery

-------
                                   - 12 -






effluent water, or inexpensive oil products for considerable periods  of time.




In these, oils "weathered" extensively, leaving viscous,  tar-like materials,  and




water seeped into the lower ground levels.   As the pond filled with solids  and




semisolids, the contents were removed by mechanical means,  covered in place,  or




the pond was simply abandoned.  The use of these ponds has  diminished, and  the




remaining ponds are usually reserved for emergency service.




     Smaller ponds or sumps were once used extensively in the crude oil produc-




tion fields.  This use was primarily for drilling muds though oil-water emulsions




and crude oil were also stored by this method.  Their use is gradually disappear-




ing because unattended or abandoned sumps cause nuisance problems to a community.




     Control of air pollution originating from storage vessels serves a three--




fold purpose:  (l) elimination or reduction of air contaminants, (2) elimination




or reduction of fire hazards, and (3)  economic savings through recovery of valu-




able products.  Methods of control  include use of floating roofs, plastic




blankets,  spheres, variable  vapor space  systems, various recovery systems, and




altered  pumping and  storage  operations.




Seals  for  Floating-Roof Tanks



     The principle by which  a  floating roof controls  emissions from  a  volatile




liquid is  that of eliminating  the  vapor  space so  that the  liquid cannot evapo-




rate and later be vented,  to  be successful the floating roof must completely




seal off the  liquid  surface  from the atmosphere.   The seal for the floating




roof  is  therefore very  important.   The floating section  is customarily con-




 structed about 8  inches  less in diameter than the tank shell. A sealing




 mechanism  must be provided for the remaining  open annular  gap.   The  seal also




 helps  keep the roof  centered.



      Conventional seals generally consist of  vertical metal plates or shoes




 connected  by braces  or  pantograph devices to  the  floating  roof-   The shoes are

-------
                                    - 13 -






suspended in such a way that they are forced outward against the inner tank wall.




An impervious fabric bridges the annular area between the tops of the shoes con-




tacting the tank wall and the circumference of the floating roof.  To reduce




emissions, a secondary seal or wiper blade has been added to the floating-roof




design by extending the fabric seal or by adding a second section of fabric.




This seal remains in contact with the tank wall.  Its flexibility allows it to




make contact even in rivet head areas of the inner shell or in places where the




shell might be slightly out of round.  This improvement lowers hydrocarbon emis-



sions further by reducing the effect of wetting and wicking associated with



floating-roof tanks.




     Recently, other types of sealing devices to close the annular gap which




have been marketed include filled tube seals.  These devices consist of a fabric




tube that rests on the surface of liquid exposed in the annular space.  The




fabric tube is filled with air, liquid or plastic material.  The pneumatic,




inflated seal is provided with uniform air pressure by means of a small expan-




sion chamber and control valves.  The sides of the tube remain in contact with




the roof and inner shell.  The liquid-filled tube holds a ribbed scuff band




against the tank wall.  The ribbed band acts as a series of wiper blades as well




as a closure.  All tubes are protected by some type of weather covering.




     A weather covering can also be added to protect the sealing fabric of the



conventional seals.  The covering includes flat metal sections held in place by




a metal band.  The metal protects the fabric seal from the elements.   When




floating-roof sections are added to older tanks constructed of riveted sections,



better contact of the shoes with the shell can be ensured by guniting or plastic




coating the inner shell.  The wetting condition of gunited walls may, however,



offset the gain of better contact.

-------
Floating Plastic Blankets




     A floating plastic blanket, operates on the same principle of control as a




floating roof.  The blanket is usually made of polyvinyl chloride but can be




made of other plastics such as polyvinyl alcohol, superpolyaraides, polyesters,




fluoride hydrocarbons, and so forth.  The blanket's underside is constructed of




a large number of floats of the same plastic material.  The blanket is custom




manufactured so that only a 1-inch gap remains around the periphery.   A verti-




cal raised skirt is provided at the edge of the blanket to serve as a vapor



seal over the annular area.  Once this area is saturated, further evaporation




diminishes.  The only remaining loss is gaseous diffusion.  The seal  is made as




effective as possible by using an elastic, Z-shaped skirt.




     Provisions are made in the blanket for openings fitted vith vertical sleeves




for measuring and sampling operations.  These openings have a crosscut, flexible




inner diaphragm to minimize exposure of the liquid surface.  Small holes vith




downspouts to effect a liquid seal are used to provide drainage of any conden-




sate from the top of the blanket.  Another feature includes a stainless steel




cable grid to prevent a buildup of static charges.  The grid is closely attached




Just under the blanket in parallel lines and connected to the tank shell by a




flexible conductor cable.  Installation of a plastic blanket is convenient for




both new and existing tanks.  The blanket is made in sections and can be intro-




duced into a tank through a manhole.



     A rigid foam-plastic cover constructed of polyisocyanate foam is also




available to equip small fixed-roof tanks with a floating cover.  The cover is




manufactured in radial sections, each equipped with a flexible neoprene seel




attached on the outer edge.  The sections are easily installed through roof man-




holes and assembled with slip-fit Joints.

-------
                                   - 15 -






Plastic Microspheres




     An outgrowth of application of plastic material provides another type of




control mechanism.  This type of control is also similar to the floating roof




and involves the use of a phenolic or urea resin in the shap-j of tiny, hollow,




spherical particles.  This material has the physical properties necessary to




form a foam covering over the denser petroleum products.  Tho fluidity of the




layer enables it to flow around any internal tank parts while keeping the liquid




surface sealed throughout any level changes.  These plastic spheres are known




under their trademark names of microballoons or microspheres.  These coverings




have proved to be effective controls for fixed-roof crude oil tanks.  Excessive




amounts of condensation or high turbulence should be avoided.  The plastic foam




has not proved as satisfactory for one-component liquid or gasoline products.




     A 1/2-inch layer of foam has been found sufficient for crude oil where




pumping rates do not exceed U,000 barrels per hour.  A layer 1 inch thick is




recommended for pumping rates up to 10,000 barrels per hour.  In order to over-




come wall holdup in smaller tanks, it is suggested that a 1-inch layer be used




regardless of pumping rates.  For tanks storing gasoline, the recommended foam




thickness is ? inches for tanks up to ^0 feet in diameter, and 1 inch for all




larger diameter vessels.



     Various methods can be used to put the foam covering on the crude oil.  One




method is to inject the plastic spheres with the crude oil as it is charged to




the tank.  Spheres are added by means of an aspirator and hopper similar to




equipment used in fire-fighting foam systems.  The spheres can also be added by




placing the desired quantity on the clean, dry floor of the tank just before the



crude oil is charged.  A wetting agent must be used when the foaa covering is to




be used on gasoline products.  This is accomplished by  slurrying the plastic




spheres, wetting agent, and gasoline in a  separate container.  The slurry is



then injected into the  tank.  Changes  in tank operation are not necessary except

-------
                                   - 16 -






for gaging or sampling.  A floating-type veil attached to a common-type gaging




tape allows accurate measurement of the tank's contents.   A sample thief with




a piercing-type bottom is needed for sampling.




     Protection against excessive loss of the plastic spheres is necessary




because of the relative value of the foam covering.   Precaution must be taken




against overfilling and pumping the tank too low.   Standard precautions against




air entrainment in pipelines normally safeguard against the latter.   Overfilling




can be prevented by automatic shutoff valves or preset shutoff operations.  Low-




level shutoff should prevent vortices created during tank emptying.   Other than




loss of the foam, no trouble should be encountered if the spheres escape into




process lines.  The plastic material is not as abrasive as the sand particles




normally found entrained in crude oil.  Excessive pressures crush the spheres




and the plastic settles in the water or sediment.   At high temperatures, the




thermo-setting resins soften, liquefy, and mix with the fuel oil, asphalt, or




coke.



     Plastic microspheres have proved to be effective for control of evaporative




losses from fixed-roof crude oil storage tanks but do not reduce emissions from




gasoline storage tanks as effectively as'other devices.




Vapor Balances Systems



     Variable vapor space or vapor balance systems are designed to contain the




vapors produced in storage.  They do not achieve as great a reduction in emissions




as an appropriately designed vapor recovery system does.  A well-planned unit




includes storage of similar or related products, and uses the advantage of in-




belance pumping situations.  Only the vapor space of the tanks  is manifolded




together in these systems.  Other systems include a vapor reservoir tank that is




either a lifter-roof type or a vessel with an internal diaphragm.  The latter




vessel can be an integrated vapor-liquid tank or a separate vaporsphere.

-------
                                   - 17 -



The manifold system includes various sizes of lightweight lines installed to


effect a balanced pressure drop in all the branches while not exceeding allowable


pressure drops.  Providing isolating valves for each tank so that each tank can


be removed from the vapor balance system during gaging or sampling operations is


also good practice.  Excessive vapors that exceed the capacity of the balance


system should be incinerated in a smokeless flare or used as fuel.


Vapor Recovery Systems


     The vapor recovery system is in many ways similar to and yet superior to a


vapor balance system in terms of emissions prevented.  The service of this type


of vapor recovery system is more flexible as to the number of tanks and products


being stored.  The recovery unit is designed to handle vapors originating from


filling operations as well as from breathing.  The recovered vapors are com-


pressed and charged to an absorption unit for recovery of condensable hydrocar-


bons.  Noncondensable vapors are piped to the fuel gas system or to a smokeless


flare.  When absorption of the condensable vapors is not practical from an


economic standpoint, these vapors, too, are sent directly to the fuel system or


incinerated in a smokeless flare.


     The recovery system, like the vapor balance system, includes vapor lines


interconnecting the vapor space of the tanks that the system serves.  Each tank


should be capable of being isolated from the system.  This enables the tanks to


be sampled or gaged without a resulting loss of vapors from the entire system.


The branches are usually isolated by providing a butterfly-type valve, a regula-

      /
tor, or a check valve-  Since the valves offer more line resistance, their use


is sometimes restricted.  Small vessels or knockout pots should be installed at


low points on the vapor manifold lines to remove any condensate.


     In some vapor recovery systems, certain tanks must be blanketed with an


inert atmosphere in order to prevent explosive mixtures and product contamina-


tion.  In other, larger systems, the entire manifolded section is maintained

-------
                                   - 18 -






under a vacuum.  Each tank is isolated by a regulator-control valve.   The valves




operate from pressure changes occuring in the tank vapor space.




     Because the vapor-gathering system is based upon positive net vapor flow to




the terminus (suction of compressors), the proper size of the vapor lines is




important.  Sizing of the line, as well as that of the compressors, absorption




unit, or flare, is based upon the anticipated amount of vapors.   These vapors




are the result of filling operations and breathing.  The distance through which




the vapors must be moved is also important.



Miscellaneous Control Measures



     Recent tests have shown that breathing emissions from fixed-roof tanks can




be reduced by increasing the storage pressure.  An increase of 1 ounce per




square inch was found to result in an 8 percent decrease in emissions due to




breathing.  Tanks operated at 2-1/2 psig or higher were found to have little or



no breathing emissions.  The pressure setting, however, should not exceed the




weight of the roof.



     Another method of reducing breathing losses is based upon the degree of




saturation in the vapor space.  A baffle located in a horizontal position




immediately below the vent directs entering atmospheric air into a stratified




layer next to the top of the tank.  Since this air is lighter, it tends to




remain in the top area; thus, there is less mixing of the free air and any of




the rich vapor  immediately above the  liquid surface.. The top stratified layer



is first expelled during the outbreathing cycle.  Test data indicate a reduced




surface evaporation of 25 to 50 percent.



     Hydrocarbon emissions can be minimized further by the proper selection of




paint for the  tank shell and roof.  The  protective coating applied to the out-




side of shell and roof influences the  vapor space and liquid tetnperatures.




Reflectivity and glossiness of a paint determine the quantity of heat a vessel

-------
                                   - 19 -






can receive via radiation.   A cooler roof and shell also allows  any heat retained




in the stored material to dissipate.  Weathering of the paint also influences its




effectiveness.  Vapor space temperature reductions of 60"F.  have been reported




with proper selection of paint color.  Similarly, liquid-surface temperature




reductions of 3 to 11 degrees have been achieved.  Data gathered by the American •




Petroleum Institute on hydrocarbon emissions indicate breathing emission reduc-




tions of 25 percent for aluminum over black paint and 25 percent for white over




aluminum paint.  All paints revert to  "black body" heat absorption media in a




corrosive or dirt-laden atmosphere.



     Insulation applied to the outside of the tank is one method of reducing the




heat energy normally conducted through the wall and roof of the vessel.  Another




method of controlling tank temperatures  is the use of water.  The water can be




sprayed or retained on the roof surface.  The evaporation of the water results



in cooling of  the  tank vapors.  Increased maintenance and corrosion problems may,




however, be encountered.



      Storage  temperatures may be  reduced by  external refrigeration or autorefrig-




eration.   External refrigeration  units require  the circulation  of the refrigerant




or of the  tank contents.  Autorefrigeration  is  practical  in  one-component liquid




hydrocarbon  storage where high  vapor pressure material  is involved.  The pres-




sure in the  tank is  reduced  by  removing  a portion of the  vapor.   Additional  vapor



 is immediately formed.   This flash vaporization results in  lowering  the tempera-




 ture of the  main liquid body.



      Routine operations  can  be  conducted in such a manner as to minimize other




 emissions associated with storage tanks. Use of remote-level reading  gages  and




 sampling devices reduces emissions by eliminating the  need  to open tank gage



 hatches.  Emissions  can be further reduced  by proper production scheduling  to




 (l) maintain a minimum of vapor space, (2)  pump liquid to the storage  tank  during




 cool hours and withdraw during hotter periods,  and (3) maintain short  periods

-------
                                   - 20 -
between pumping operations.




     Using vet scrubbers as control equipment for certain stored materials that




are sufficiently soluble in the scrubbing media employed is both possible and




practical.  The scrubbers can be located over the vent vhen the scrubbing medium,




for example, a water scrubber for aqua ammonia storage, can be tolerated in the




product.  In other cases, the vent of one or more tanks can be manifolded so




that any displaced gas is passed through a scrubbing unit before being discharged




to the atmosphere.  A typical example is a scrubber 'packed with plastic spirals




that serves ketone storage vessels.  The scrubbing liquid is water, which is




drained to a closed waste effluent disposal system.




     Properly designed condensers can be used to reduce the vapor load from tank




vents in order that smaller control devices can be employed.




Costs of Storage Vessels




     The installed costs of various types of hydrocarbon storage tanks are sum-




marized in Table I.  Included in these costs are standard tank accessories such




as manholes, vents, ladders, stairways, drains, gage hatches, and flanged connec-




tions.




WASTE-GAS DISPOSAL SYSTEMS



     Large volumes of hydrocarbon gases are produced in modern refinery and




petrochemical plants.  Generally, these gases are used as fuel or as raw material




for  further processing.  In the past, however, large quantities of these gases




were considered waste gases, and along with waste liquids, were dumped to open




pits and burned, producing large volumes of black smoke.  With modernization of




processing units,  this method of waste-gas disposal, even for emergency gas




releases, has become less  acceptable  to the  industry.  Moreover, many  local




governments have adopted or are contemplating ordinances limiting the opacity of




smoke  from  combustion processes.

-------
                                   - 21 -






     Nevertheless, petroleum refineries are still faced with the problem of safe




disposal of volatile liquids and gases resulting from scheduled shutdowns and




sudden or unexpected upsets in process units.   Emergencies that can cause the




sudden venting of excessive amounts of gases and vapors include:




     1.  Failure of the cooling water supply.




     2.  Failure of a reflux system.




     3.  Entrance of a more volatile fluid into the equipment.




     U.  Vapor generation due to fire exposure.




     5.  Excessive heat inputs other than from fire.




     6.  Accumulation of noncondensible gases.




     7.  Closed or plugged equipment outlets.




     8.  Failure of automatic flow, temperature, or pressure control equipment.




     9.  Internal explosions.




    10.  Uncontrolled chemical reactions.




    11.  Failure of heat exchanger  internals.




    12.  Power failure.




    13.  Thermal expansion.




    ll».  Compressor failure.




     A system for disposal of emergency  and waste refinery gases consists of a




 manifolded pressure-relieving or blowdown  system, and  a blowdown recovery system




 or  a system of flares for  the combustion of the  excess gases,  or both.   Many




 refineries, however, do not  operate blowdown  recovery  systems.  In addition to




 disposing  of  emergency and  excess  gas  flows,  these  systems are used in  the




 evacuation of units during shutdowns  and turarounds.   Normally a unit is shut




 down by depressuring  into  a  fuel gas  or  vapor recovery system with further




 depressuring  to  essentially atmospheric  pressure by venting  to a low-pressure




 flare  system. Thus,  overall emissions of  refinery  hydrocarbons are substan-




 tially reduced.

-------
                                   - 22 -






     Refinery pressure-relieving systems, commonly called blowdown systems,  are




used primarily to ensure the safety of personnel and protect equipment in the




event of emergencies such as process upset,  equipment failure,  and fire.   In




addition, a properly designed pressure relief system permits substantial  reduc-




tion of hydrocarbon emissions to the atmosphere.




     The equipment in a refinery can operate at pressures ranging from less  than




atmospheric to 1,000 psig and higher.  This  equipment must be designed to permit




safe disposal of excess gases and liquids in case operational difficulties or




fires occur.  These materials are usually removed from the process area by auto-




matic safety and relief valves, as well as by manually controlled valves, mani-




folded to a header that conducts the material avay from the unit involved.  The




preferred method of disposing of the waste gases that cannot be recovered in a




blowdown recovery system is by burning in a  smokeless flare.  Liquid blowdowns




are usually conducted to appropriately designed holding vessels and reclaimed.




     A blowdown or pressure-relieving system consists of relief valves, safety




valves, manual bypass valves, blovdown headers, knockout vessels, and holding




tanks.  A blowdown recovery system also includes compressors and vapor surge




vessels such as gas holders or vapor spheres.  Flares are usually considered as




part of the blowdown system in a modern refinery.




     The pressure-relieving system can be used for liquids or vapors or both.




For reasons of economy and safety, vessels and equipment discharging to blowdown




systems are usually segregated according to  their operating pressure.  In other




words, there is a high-pressure blowdown system for equipment working, for




example, above 100 psig, and low-pressure systems for those vessels with working




pressures below 100 psig.  Butane and propane are usually discharged to a sepa-




rate blowdown drum, which is operated above atmospheric pressure to increase




recovery of liquids.  Usually a direct-contact type of condenser is used to




permit recovery of as much hydrocarbon liquid as possible from the blowdown vapors.

-------
                                   - 23 -






The noncondensibles are burned in a flare.




Design of Pressure Refllef System




     The design of e pressure relief system is one of the most important problems




in the planning of a refinery or petrochemical plant.  The safety of personnel




and equipment depends upon the proper design and functioning of this type of




system.  The consequences of poor design can be disastrous.




     A pressure relief system can consist of one relief valve, safety valve, or




rupture disc, or of several relief devices manifolded to a common header.




Usually the systems are segregated according to the type of material handled,




that is, liquid or vapor, as well as to the operating pressures involved.




     The several factors that must be considered in designing a pressure relief




system are (l) the governing code, such as that of ASME (American Society of




Mechanical Engineers, 19^2); (2) characteristics of the pressure relief devices;




(3) the design pressure of the equipment protected by the pressure relief




devices, (U) line sizes and lengths, and (5) physical properties of the material




to be relieved to the system.




Safety Valves




     Nozzle-type safety valves are available in the conventional or balanced-




bellows configurations.  Backpressure in the piping downstream of the standard-




type valve affects its set pressure, but theoretically, this backpressure does




not affect the set pressure of the balanced-type valve.  Owing, however, to




imperfections in manufacture and limitations of practical design,  the balanced




valves available vary in relieving pressure when the backpressure reaches




approximately ^0 percent of the set pressure.   The actual accumulation depends




upon the manufacturer.




     Until the advent of balanced valves, the general practice in the industry




was to select safety valves that start relieving at the design pressure of the




vessel and reach full capacity at 3 to 10 percent above the design pressure.

-------
                                   - 2k -






This overpressure vas defined as accumulation.  With the balanced safety valves,




the allowable accumulation can be retained with smaller pipe size.




     Each safety valve installation is an individual problem.   The required




capacity of the valve depends upon the condition producing the overpressure.




Rupture Discs




     A rupture disc is an emergency relief device consisting of a thin metal




diaphragm carefully designed to rupture at a predetermined pressure.




     The obvious difference between a relief or safety valve and a rupture disc




is that the valve reseats and the disc does not.  Rupture discs may be installed




in parallel or series with a relief valve.  To prevent an incorrect pressure




differential from existing, the space between the disc and the valve must be




maintained at atmospheric pressure.  In some cases a rupture disc may be used




to supplement a relief or safety valve. -In an installation such as this, the




relief or safety valve is sized by conventional methods and the rupture disc is




usually designed to relieve at 1.5 times the maximum allowable working pressure




of the vessel.




     In determining the size of a disc, three important effects that must be




evaluated are low rupture pressure, elevated temperatures, and corrosion.  Manu-




facturers can supply discs that are guaranteed to burst at plus or minus 5 per-




cent of their rated pressures.




     The corrosive effects of a system determine the type of material used in a




disc.  Even a slight amount of corrosion can drastically shorten disc life.




Discs are available with plastic linings, or they can be made from pure carbon




materials.




FLARES




     The air pollution problem associated with the uncontrolled disposal of waste




gases is the venting of large volumes of hydrocarbons and other odorous gases

-------
                                    - 25 -






 and aerosols.   The preferred control method for excess gases and vapors is  to




 recover them in a blowdown recovery system and, failing that,  to incinerate them




 in an elevated-type flare.  Such flares introduce the  possibility of smoke  and




 other objectionable gases such as carbon monoxide,  sulfur dir.xide,  and nitrogen •




 oxides.  Flares have been further developed to  ensure  that this  combustion  is




 smokeless and  in some cast:s nonluminous.   Luminosity,  while r.ot  an  air pollution




 problem,  does  attract attention to the  refinery operation and  in certain  cases



 can cause bad  public relations.




 Smoke from Flares




      Smoke is  the result  of incomplete  combustion.  Smokeless combustion can be




 achieved  by:   (l) adequate heat  values  to obtain the minimum Theoretical  combus-




 tion temperatures,  (2)  adequate  combustion air,  and (3) adequate  mixing of  the



 air and fuel.




      An insufficient supply of air results  in a  smoky  flame.  Combustion begins



 around  the periphery of the gas  stream where the air and  fuel mix, and within this




 flame envelope  the  supply  of air  is limited.  Hydrocarbon  side reactions occur




 with  the  production of smoke.  In  this reducing atmosphere, hydrocarbons crack to




 elemental hydrogen  and carbon, or polymerize to form heavier hydrocarbons.




 Since the carbon  particles  are difficult  to burn, large volumes of carbon parti-




 cles appear as  smoke  upon  cooling.  Side  reactions become nore pronounced as




 molecular weight  and  unsaturation of the  fuel gas increase.  Olefins, diolefins,




 and aromatics characteristically burn with smoky, sooty flames as compared with



 paraffins and naphthenes .




     A smokeless  flame can be obtained when an adequate amount of combustion air




 is mixed sufficiently with the fuel so that it burns completely and rapidly




before any side reactions can take place.

-------
                                   - 26 -


Types of Flares

     There are, in general, three types of flares for the disposal of waste gases:

elevated flares, ground-level flares, and burning pits.

     The burning pits are reserved for extremely large gas flows caused by catas-

trophic emergencies in which the capacity of the primary smokeless.flares is

exceeded.  Ordinarily, the main gas header to the flare system has a water seal

bypass to a burning pit.  Excessive pressure in the header blows the water seal

and permits the vapors and gases to vent a burning pit where combustion occurs.

     The essential parts of a flare are the burner, stack, seal, liquid trap,
                                                              I
controls, pilot burner, and ignition system.  In some cases, vented gases flow

through chemical solutions to receive treatment before combustion.  As an exam-

ple, gases vented from an isomerization unit that may contain small amounts of

hydrochloric acid are scrubbed with caustic before being vented to the flare.

Elevated Flares

     Smokeless combustion can be obtained in an elevated flare by the injection

of an inert gas to the combustion zone to provide turbulence and inspirate air.

A mechanical air-mixing system would be ideal but is not economical in view of

the large volume of gases handled.   The most commonly enountered air-inspirating

material for an elevated flare is steam.  Three main types of steam-injected

elevated flares are in use.  These types vary in the manner in which the steam

is injected into the combustion zone.

     In the first type, there is a commercially available multiple nozzle which

consists of an alloy steel tip mounted on the top of an elevated stack.   Steam

injection is accomplished by several small Jets placed concentrically around the

flare tip.  These Jets are installed at an angle, causing the steam to discharge

in a converging pattern immediately above the flare'tip.

     A second type of elevated flare has a flare tip with no obstruction to flov,

that is, the flare tip is the same diameter as the stack.   The steam is  injected

-------
                                   - 27 -






by a single nozzle located concentrically within the burner tip.  In this type




of flare, the steam is prefixed with the gas before ignition and discharge.




     A third type of elevated flare is equipped with a flare tip constructed to




cause the gases to flow through several tangential openings -;o promote turbulence.




A steam ring at the top of the stack has numerous equally spaced holes about 1/B




inch in diameter for discharging steam into the gas stream.




     The injection of steam in flares may be automatically or manually controlled.




Most flares are instrumented to the extent that steam is automatically supplied




when there is a measurable gas flow.  In most cases, the steain is proportioned




automatically to the reate of gas flow; however, in some installations, the




steam is automatically supplied at maximum rates, and manual throttling of a




steam valve is required for adjusting the steam flow to the particular gas flow




rate.  There are many variations of instrumentation among various flares, some




designs being more desirable than others.  For economic reasons, all designs




attempt to proportion steam flow to the gas flow rate.




     Steam injection is generally believed to result in the following benefits:




(l) energy available at relatively low cost can be used to inspirate air and pro-




vide turbulence within the flame, (2) steam reacts with the fuel to form oxygen-




ated compounds that burn readily at relatively low temperatures, (3) water-gas




reactions also occur with this same end result, and (^) steam reduces the partial




pressure of the fuel and retards polymerization,  (inert gases such as nitrogen




have also been found effective for this purpose; however, the expense of provid-




ing a diluent such as this is prohibitive.)




Ground-Level Flares




     Ground-level flares are of four principal types:   horizontal venturi, water




injection, multijet, and vertical venturi.




     A horizontal venturi-type flare system utilizes groups of standard venturi




burners.  In this type of burner, the gas pressure inspirates combustion air for

-------
                                   - 28 -






smokeless operation.




     A water-injection flare consists of a single burner witli a water spray ring




around the burner nozzle.   The water spray inspirates air am  provides water




vapor for the smokeless combustion of gases.   Water is not aj  effective as  steam




for controlling smoke with high gas-flow rates,  unsaturated materials, or wet




gases.




     A multijet ground flare uses two sets of burners, one for normal gas release




rates and both for higher flaring rates.




     A vertical, venturi-type ground flare also uses commercial-type venturi




burners.  This type of flare is suitable for relatively small flows of gas  at a




constant rate-




     Ground-level flares are seldom used today in refineries because of space




limitations, the inability to safely dissipate heat generated, and the difficulty




of diffusing any vapors that may be emitted.




Effect of Steam Injection




     A flare installation that does not inspirate' an adequate amount of air or




does not mix the air and hydrocarbons properly emits dense, black clouds of




smoke that obscure the flame.  The injection of steam into the zone of combus-




tion causes a gradual decrease in the amount of smoke, and the flame becomes




more visible.  When trailing smoke has been eliminated, the flame is very




luminous and orange with a few wisps of black smoke around the periphery.  The




minimum amount of steam required produces a yellowish-orange, luminous flame




with no smoke.  Increasing the amount of steam injection further decreases  the




luminosity of the flame.  As the steam rate increases, the flame becomes color-




less and finally invisible during the day.  At night this flame appears blue.




     The injection of an excessive amount of steam causes the flame to disappear




completely and be replaced with a steam plume.  An excessive amount of steam may




extinguish the burning gases and permit unburned hydrocarbons to discharge to

-------
                                .   - 29 -






the atmosphere.   When the flame is out, there is a change in the sound of the




flare because a  steam hiss replaces the roar of combustion.   The commercially




available pilot  burners are usually not extinguished by excersive amounts of




steam, and the flame reappears as  the steam injection rate it: reduced.  As the




use of automatic instrumentation becomes more prevalent in fl.are installations,




the use of excessive amounts of steam and the emission of unturned hydrocarbons




decrease and greater steam economies can be achieved.  In evaluating flare




installations from an air pollution standpoint, controlling the volume of steam




is important.  Too little steam results in black smoke, which, obviously, is




objectionable.  Conversely, excessive use of steam produces a white steam plume




and an invisible emission of unburned hydrocarbons.  A condition such as this




can also be a serious air pollution problem.




Design of a Smokeless Flare




     The choice  of a flare is dictated by the particular requirements of the




installation.  The usual flare system  includes gas collection equipment, the




liquid knockout tank preceding the flare stack.  A water seal tank is usually




located between the knockout pot and the flare stack to prevent flashbacks into




the system.  Flame arrestors are sometimes used in place of or in conjunction




with a water seal pot.  The flare stack should be continuously purged with steam




or refinery gas to prevent the formation of a combustible mixture that could




cause an explosion in the stack.




     The preferred method of inspirating air is injecting stest?. either into the




stack or into the combustion zone.  Water has sometimes been used in ground




flares where there is an abundant supply.  There is, however, less assurance of




complete combustion when water is used, because the flare is limited in  its




operation by the type and composition  of gases  it can handle efficiently.




     The diameter of the flare stack depends upon the expected emergency gas




flow rate and the permissible backpressure  in the vapor relief manifold  system.

-------
                                   - 30 -






The stack diameter is usually the same or greater than that ->f the vapor header




discharging to the stack and should be the same diameter as or greater than




that of the burner section.  The velocity of the gas in the r tack should be as




high as possible to permit use of lower stack heights; promote turbulent flow




with resultant improved combustion, and prevent flashback.




     Adequate stack heights must be provided to permit safe dispersion of toxic




or combustible material in the event of pilot burner failure.




     The structural support of an elevated-flare stack over 1*0 to 50 feet high




requires the use of guy wires.  A self-supporting stack over f>0 feet high




requires a large and expensive foundation.  Stacks over 100 feet high are




usually supported by a steel structure.




     The amount of steam required for smokeless combustion varies according to




the maximum expected gas flow, the molecular weight, and the percent of unsatu-




rated hydrocarbons in the gas.  Actual tests should be run on the various mate-




rials to be flared in order to determine a suitable steam-to-hydrocarbon ratio.




In the typical refinery, the ratio of steam to hydrocarbon varies from 0.2 to




0-5 pound of steam per pound of hydrocarbon.




Pilot Ignition System




     The ignition of flare gases is normally accomplished with one of three pilot




burners.  A separate system must be provided for the ignition of the pilot




burner to safeguard against flame failure.  In this system, an easily ignited




flame with stable combustion and low fuel usage must be provided.  In addition,




the system must be protected from the weather.




     On elevated flares, the pilot flame is usually not visible, and an alarm




system to indicate flame failure is desirable.  This is usually accomplished by




installing thermocouples in the pilot burner flame.  In the event of flame fail-




ure, the temperature drops to a preset level, and an alarm sounds.

-------
                                   - 31 -






Instrumentation and Control of Steam and Gas




     For adequate prevention of smoke emission and possible '-iolations of air




pollution regulations, an elevated, smokeless flare should bf-- equipped to pro-




vide steam automatically and in proportion to the emergency ^as flow.




     Basically, the instrumentation required for a flare is i  flow-sensing




element, such as a pitot tube, and a flow transmitter that seids e signal




(usually pneumatic) to a control valve in the steam line.   Although the pitot




tube has been used extensively in flare systems, it Is limited by the  minimum




linear velocity required to produce a measurable velocity head.  Thus, small




gas flows will not actuate the steam control valves.  This problem is  usually




overcome by installing a small bypass valve to permit a constant flow  of steam




to the flame burner.  A more sensititve type of flow-measuring device  is the




inverted weir.  A variation of the inverted weir is the slotted orifice.




     The hot-wire flow meter has also been used in flare systems.  The sensing




element is basically a heat loss anemometer consisting of an electrically heated




wire exposed to the gas stream to measure the velocity.  The gas flow  is perpen-




dicular to the axis of the hot wire.  A conventional recorder is used  with this




probe, modified for the resistance bridge circuit of the gas flow meter.  As




the flow of gas past the probe varies, the heat loss from the hot wire varies




and causes an imbalance of the bridge circuit.  The recorder then adjusts for




the imbalance in the bridge and indicates the gas flow.  This type of  installa-




tion provides sensitivity at low velocities, and the gas flow measurement can be




made without causing an appreciable pressure drop.  This is an important advan-




tage in a system using constant backpressure-type relief valves.  The  hot-wire




flow meter can be used as a primary flow-sensing element or as a leak detector




in laterals connected to the main  flare header.

-------
                                   - 32 -






Maintenance of Flares




     Most refineries and petrochemical plants have a fixed schedule for inspec-




tion and maintenance of processing units and their auxiliaries.   The flare




system should not be exempted from this practice.   Removal ol' a  flare from




service for maintenance requires some type of standby equipment  to disperse




emergency gas vents during the shutdown.  A simple stack with pilot burner should




suffice for a standby.  Coordinating this inspection to take place at time when




the major processing units are also shut down is good practice.




     Flare instrumentation requires scheduled maintenance to ensure proper opera-




tion.  Most of the costs and problems of flare maintenance ar;'.se from the instru-




mentation.




     Maintenance expenses for flare burners can be reduced by constructing them




of chrome-nickel alloy.  Because of the inaccessibility of elevated flares, the




use of alloy construction is recommended.




PROCESS OPERATIONS




Catalytic Cracking Units




     Petroleum fractions are cracked to produce compounds of lower molecular




weight.  Catalysts in the form of powders or beads are utilized.  The catalyst




particles become coated with carbon and high-molecular-weight compounds.   These




materials must be burned off the catalyst in order to maintain its activity.




The catalyst continuously circulates from the reactor chamber zo the regenerator




chamber.  In the regenerator, a controlled amount of air is admitted to burn off




the coatings.  This causes the formation of CO and hydrocarbons.  Typical hydro-




carbon emissions from the regenerators of catalytic cracking units are estimated




to be 220 lbs./l,000 barrels of fresh feed for fluidized units and 8? lbs./l,000




barrels of fresh feed for moving bed units.




     Hydrocarbon emissions from regenerators of catalytic cracking units are




generally of secondary importance compared to the carbon monoxide emissions and

-------
                                   - 33 -






their control  is incidental to CO control.  A vaste heat or CO boiler will pro-




vide essentially a 100 percent reduction  in hydrocarbon emissions as veil as




controlling CO emissions.  The internal CO combustion technique also results in




significant reduction in hydrocarbon emissions but data are incomplete as to




the exact percentage reduction which can be obtained with thio technique.



Asphalt Oxidation




     Asphalt is a dark brcvn to black, solid or semisolid mat-.-rial found in




naturally occurring deposits or as a colloidal suspension in «.:rude oil.  Analyti-



cal methods have been used to separate asphalt into three conr:.)0nent groups--




asphaltenes, resins, and oils.  A particular grade of asphalt may be character-




ized by the amounts of each group it contains.  The asphaltenf particle provides




a nucleus about which the resin forms a protective coating.   The particles are




suspended in an oil that is usually paraffinic but can be naphthenic or naptheno-



aromatic.




     Over 90 percent of all asphalt used in the United States is recovered from




crude oil.  The method of recovery depends upon the type of crude oil being pro-




cessed.  Practically all types of crudes are first distilled at atmospheric




pressure to remove the lower boiling materials such as gasoline, kerosene,



diesel oil, and others.  Recovery of nondistillable asphalt from selected




topped crudes may then be accomplished by vacuum distillation, solvent extrac-



tion, or a combination of both.




     A vacuum distillation unit uses a heater, preflash tower, vacuum vessel,




and appurtenances for processing topped crudes.  Distillation o.f topped crude.




under a high vacuum removes oils and wax as distillate products, leaving the



asphalt as a residue.  The amount of oil distilled from the residue asphalt



controls its properties; the more oil and resin or oily constituents removed by




distillation, the harder the residual asphalt.  Residual asphalt can be used as

-------
paving material or it can be further refined "by airblowing.




     Asphalt is also produced as a secondary product in solvent extraction pro-




cesses.  This process separates the asphalt from remaining constituents of




topped crudes by differences in chemical types and molecular weights rather




than boiling points as in vacuum distillation processes.  The solvent, usually




a light hydrocarbon such as propane or butane, is used to retmve selectively a




gas-oil fraction from the asphalt residue.




     Economical removal of the gas-oil fraction from topped crude, leaving an




asphaltic product, is occasionally feasible only by airblowing the crude residue




at elevated temperatures.  Excellent paving-grade asphalts are produced by this




method.  Another important application of airblowing is in the production of




high-quality specialty asphalts for roofing, pipe coating, and similar uses.




These asphalts require certain plastic properties imparted by reacting with air.




     Airblowing is mainly a dehydrogenation process.  Oxygen in the air combines




with hydrogen in the oil molecules to form water vapor.  The progressive loss of




hydrogen results in polymerization or condensation of .the asphalt to the desired"




consistency.  Blowing is usually carried out batchwise in horizontal or vertical




stills equipped to blanket the charge with steam, but it may also be done con-




tinuously.  Vertical stills are more efficient because of longer air-asphalt




contact time.  The asphalt is heated by an internal fire-tube heater or by




circulating the charge material through a separate tubestill.  A temperature of




300° to ^00°F. is reached before the airblowing cycle begins.  Air quantities




used range from 5 to 20  cubic feet per minute per ton of charge.  Little addi-




tional heat is then needed since the reaction becomes exothermic.




     Effluents from the  asphalt airblowing stills include oxygen, nitrogen and




its compounds, water vapor, sulfur compounds, and hydrocarbons as gases, odors,




and aerosols.  Discharge of these vapors  directly to the atmosphere is objectionable

-------
                                   - 35 -






from an air pollution control standpoint.  The disagreeable c-dors and airborne




oil particles entrained with the gases result in nuisance complaints.  Disposal




methods are available that can satisfactorily eliminate the pollution potential




of the effluents.




     Control of effluent vapors from asphalt airbloving stilLs has been accom-




plished by scrubbing and incineration, singly or in combination.   Most installa-




tions use the combination.  For scrubbing alone to be effective,  a very high




water-to-gas ratio of about 100 gallons per 1,000 scf is necessary.



     Where removal of most of the potential air pollutants is not feasible by




scrubbing alone, the noncondensibles must be incinerated.  Essential to effec-




tive incineration is direct-flame contact with the vapors, a minimum retention




time of 0-3 second in the combustion zone, and maintenance of a minimum combus-




tion chamber temperature of 1,200°F.  Other desirable features include turbulent




mixing of vapors in the combustion chamber, tangential flame entry, and adequate




instrumentation.  Primary condensation of any steam or water vapor allows use




of smaller incinerators and results in fuel savings.  Some of the heat released




by incineration of the waste gases may be recovered and used for generation of




steam.



     Catalytic fume burners are not recommended for the disposal of vapors from




the airblowing of asphalt because the matter entrained in the vapors would




quickly clog the catalyst bed.




Chemical Treating Processes



     In acid treatment, emissions can be reduced by substituting continuous




mechanical mixing for batch-type agitators that employ airblowing for mixing.




Acid regeneration can also be used instead of the hydrolysis-concentration



method of acid recovery.  Gases emitted during acid-sludge recovery can be




vented to caustic scrubbers to remove sulfur dioxide and odorants.  Gases from




scrubbers can then be vented to a firebox or flare.  For new installations,

-------
                                   - 36 -

acid treatment can also be replaced by catalytic hydrogenation or by other pro-
cessing techniques that may prove to be more effective.
     In doctor treating, the doctor solution can be steam-stripped to recover
hydrocarbons prior to airblowing for regeneration.   The  effluent from airblowing
can then be incinerated to destroy hydrocarbon vapors.
     In the disposal of spent caustic, entrained hydrocarbons are often removed
by stripping with inert gases.  The vapors removed in this stripping operation
can be vented to a flare or to a furnace firebox.
     Whenever hydrocarbons are removed in air or gasbloving operations, the
effluent hydrocarbons can be destroyed by incineration.
LOADING FACILITIES
     Gasoline and other petroleum products are distributed from the manufacturing
facility to the consumer by a network of -pipelines, tank vehicle routes, railroad
tank cars, and oceangoing tankers.
     As integral parts of the network, intermediate storage and loading stations
receive products from refineries by either pipelines or tank vehicles.  If the
intermediate station is supplied by pipeline, it is called a bulk terminal, to
distinguish it from the station supplied by tank vehicle, which is called a bulk
plant.  Retail service stations fueling motor vehicles for the public are, as a
general rule, supplied by tank vehicle from bulk terminals or bulk plants.  Con-
sumer accounts, which are privately owned facilities operated, for example, to
fuel vehicles of a company fleet, are supplied by tank vehicles from intermediate
bulk installations or directly from refineries.
     Gasoline and other petroleum products are loaded into tank trucks, trailers,
or tank cars at bulk installations and refineries by means of loading racks.
Bulk products are also delivered into tankers at bulk marine terminals.
     When a compartment of a  tank vehicle or tanker is filled through an open
overhead hatch or bottom  connection,  the  incoming liquid  displaces the vapors

-------
                                   - 37 -






in the compartment to the atmosphere.   Except in rare instances,  vhere a tank




vehicle or tanker is free of hydrocarbon vapor,  as when being used for the first




time, the displaced vapors consist of a mixture  of air and hydrocarbon concen-




tration, depending upon the product being loaded, the temperature of the product




and of the tank compartment, and the type of loading.  Ordinarily, but not




always, when gasoline is loaded, the hydrocarbon concentration of the vapors is




from 30 to 50 percent by volume and consists of  gasoline fractions ranging from




propane through hexane.



     The volume of vapors produced during the loading operation,  as well as




their composition, is greatly influenced by the  type of loading or filling




employed.  The types in use throughout the industry may be classified under two




general headings, overhead loading and bottom loading.




     Overhead loading, presently the most widely used method, may be further




divided into splash and submerged filling.  In splash filling, the outlet of the




delivery tube is above the liquid surface during all or most of the loading.




In submerged filling the outlet of the delivery  tube is extended to within 6




inches of the bottom and is submerged beneath the liquid during most of the




loading.  Splash filling generates more turbulence and therefore more hydrocar-




bon vapors.




     Bottom loading has been introduced by a number of oil companies for new




•facilities.  The equipment required is simpler than that used for overhead load-.



ing.  Loading by this method is accomplished by connecting a swing-type loading




arm or hose at ground level to a matching fitting on the underside of the tank




vehicles.  Aircraft-type, quick-coupling valves  are used to ensure a fast, posi-




tive shutoff and prevent liquid spills.  Several companies experienced in aircraft-




fueling operations have developed fully automatic bottom-loading systems.  All the




loading is submerged and under a slight pressure; thus, turbulence and resultant

-------
                                   - 38 -






production of vapors are minimized.




     Loading losses for splash filling typically range from 0.2 to 0-35 percent




depending on loading rate, RVP and temperature of product, ar-.cl degree of satura-




tion of the displaced vapors.  Loading losses for "bottom loading and submerged




filling are roughly equivalent with values ranging from 0.06 to 0.23 percent




depending on conditions.




     The method employed for loading marine tankers is essentially a bottom-.




loading operation.  Liquid is delivered to the various compartments through




lines that discharge at the bottom of each compartment.  The vapors displaced




during loading are vented through a manifold line to the top of the ship's mast




for discharge to the atmosphere.



     In addition to the emissions resulting from the displacement of hydrocarbon




vapors from the tank vehicles, additional emissions during loading result from




evaporation of spillage, drainage, and leakage of product.




     An effective system for control of vapor emissions from loading must include




a device to collect the vapors at the tank vehicle hatch' and a means for disposal




of these vapors.




Overhead Loading



     Four types of vapor collectors have been developed for use during overhead




loading operations.  All are essentially plug-shaped devices that are inserted




into a fitting for the hatch opening.  Gasoline flows through a central channel




in the device into the tank vehicle compartment.  This central channel is sur-




rounded by an annular space  into which vapors enter through openings on the




bottom of the hatch fitting.  The annular space is in turn connected to a hose




or pipe leading to a vapor disposal system.




     The Mobil Oil Corporation device is connected to a vapor chamber with a



transparent section to allow the operator to see the calibrated capacity markers

-------
                                   - 39 -






located within the tank compartment.  This closure has adjustable height.  It




requires a constant downward force to keep it firmly in place during filling.




It is built to fit only hatches 8 to 10 inches in diameter.




     The Chiksan device incorporates the hatch closure, the vapor return line,




and the fill line into an assembled unit.  This unit has features to prevent




overfills, topping off, or filling unless the assembly is properly seated in




the tank hatch.



     The Greenwood vapor closure developed by the Vernon Tool Company, also



requires downward force during filling.  It ordinarily does not have a trans-




parent vapor chamber.  This closure has an adapter for hatches larger than 10




inches.



     The fourth device was developed by Standard Oil Company of California and




has a positive clamp for the hatch opening, which, when closed automatically




actuates the vapor chamber.  It also has a safety shutoff float that senses the




gas level and prevents overfilling.  These SOCO devices can be used with adapters




for hatches larger than 8 inches in diameter.



     The slide positioner of the Mobil Oil Corporation device can be a source of




vapor leaks and requires close attention by the operator during adjustments for




fitting and submerged loading.  The inner valves of the SOCO devices make them




considerably heavier than other types.  This device increases pressure drop and




slows the loading rates.  Mobil and Greenwood devices both require check valves



in the vapor-gathering lines to prevent the vapor from discharging back to the




atmosphere when the assembly is withdrawn.  In addition, these devices require




nearly vertical entry of the loading tube into the hatch opening in order to



provide a tight seal against vapor leaks.  An assembly is available to assure




that the Greenwood device maintains this vertical position.

-------
Bottom Loading



     Bottom loading permits easier collection of displaced vapors.   Because the




filling line and the vapor collection line are independent of each other,  col-




lection during bottom loading is relatively easy.  The vapor collection line




consists of a flexible hose or swing-type arm connected to a quick-acting  valve




fitting on the dome of the vehicle.  A check valve must, of course, be installed-




on the vapor collection line to prevent backflow of vapors to the atmosphere




when the connection to the tank is broken.



     In designing for complete vapor pickup at the tank vehicle hatch, several




factors, including tank settling, liquid drainage, and topping off must be con- .




sidered.



     The settling of a tank vehicle due to the weight of product being added




requires that provision be made for vertical travel of the loading arm to  follow  .




the. motion of the vehicle so that the vapor collector remains sealed in the tank




hatch during the entire loading cycle.  Two solutions to the problem of settling




have been used.  The first, applicable to pneumatically-operated arms, includes




the continuous application of air pressure to the piston in the air cylinder




acting on the  arm.   The  arm is thus forced to follow the motion of the vehicle




without need for clamping or fastening the vapor  collector to the  tank vehicle.




The second  solution, employed on counterweighted  and torsion spring loading




arms, provides for  locking the  vapor collector to the tank vehicle hatch.   The




arm then necessarily follows the motion of the vehicle.  The second solution  is




also  applicable  to  vapor  collection arms  or hoses that  are connected  to the top




of a  tank  vehicle during  bottom loading.



      The second  problem,  that of preventing considerable liquid drainage  from a




loading arm as  it  is withdrawn  after completion  of  filling operations, has been




adequately solved.  The air  valve  that operates  the air cylinder of pneumatically

-------
                                   - in -






6t>erated loading arms may be modified by addition of an orifice on the discharge




side of the valve.  The orifice allows 30 to k$ seconds to elapse before the




loading assembly clears the hatch compartment.  This time interval is sufficient




to permit complete draining of liquid into tank compartments from arms fitted




with loading valves located in an outboard position.  Loading arms with inboard




valves require additional drainage time and present the- problem of gasoline




retention in the horizontal section of the arm.  To prevent drainage the SOCO




vapor collection closure is fitted with an internal shutoff valve that is closed



before the loading arm is withdrawn from the tank hatch.  Providing for thermal



expansion has been found necessary when an inboard valve and a SOCO vapor closure




are used.  This has been accomplished by installing a small expansion chamber at




the normal position of the loading  arm's vacuum breaker.  In bottom loading, the




valve coupling  at the end of the loading arm or hose, as well as the mating por-




tion of the  valve on the trucks, is self-sealing to prevent drainage of product




when the connection  is made or broken.



     The third  factor to be considered  in the  design of  an effective  vapor col-




lection system  is topping off.  Topping  off  is the  term  applied  to the loading




operation  during  which the  liquid  level  is adjusted to the capacity marker




 inside  the  tank vehicle  compartment.   Since  the  loading  arm  is out of  the com-




partment hatch  during  the  topping  operation,  vapor  pickup by  the collector  is




 nil.   Metering  the  desired  volumes during loading  is one solution to  the problem.



 Metered loading must,  however,-be  restricted to  empty  trucks  or  to trucks pre-




 checked for loading volume  available.   Accuracy of  certain totalizing meters  or




 preset stop meters  is  satisfactory for loading without the  need  for  subsequent




 ooen topping.   An interlock device for the  pneumatic-type loading arms,  consist-




 ing of pneumatic  control or mechanical linkage,  prevents opening of  the  loading




 valve  unless the  air cylinder valve is in the down position.   Thus,  open topping



 is theoretically  impossible.

-------
     Topping off is not a problem vhen bottom loading is employed.   Metered




loading, or installation of a sensing device in the vehicle compartments  that




actuates a shutoff valve located either on the truck or the loading island,




eliminates the need for topping off.




Vapor Disposal



     The methods of disposing of vapors collected during loading operations may




be considered under three headings:  using the vapors as fuel, processing the




vapors for recovery of hydrocarbons, effecting a vapor balance system in conJune-'




tion with submerged loading, or simple incineration of generated vapors.




     The first method of disposal, using the vapors directly as fuel, may be




employed when the  loading facilities are located in or near a facility that



includes fired heaters or boilers.  In a typical disposal system, the displaced




vapors  flow  through a drip pot to  a small vapor holder that is gas blanketed to




prevent forming of explosive mixtures.  The vapors are drawn from the holder by




a  compressor and  are discharged to the fuel gas system.



     The  second method of  disposal uses equipment  designed 'to recover the  hydro-




carbon  vapors.  Vapors have  been  successfully absorbed  in a liquid  such  as gaso-




line or kerosene.  If  the  loading facility  is located near  a refinery or gas




absorption plant,  the  vapor  line  can be connected  from  the  loading  facility  to




an existing vapor recovery system through a regulator valve.



      Vapors are recovered from loading installations distant  from  existing pro-



 cessing facilities by  use of package units.   One  such unit that, absorbs  hydro-




 carbon  vapors  in  gasoline has  been developed by the Superior  Tank  and  Construction




 Company.   This  unit  includes a vaporsphere  or tank equipped with flexible  membrane




 diaphragm, saturator,  absorber,  compressor, pumps, and instrumentation.   Units  are



 available to fit  any size operation at any desired loading location since  they  use




 the gasoline product as the absorbent.

-------
                                    . U3 - -






     Explosive mixtures must be prevented from existing in this unit.   This is




accomplished by passing the vapors displaced at the loading rack through a




saturator countercurrently to gasoline pumped from storage.  The saturated




vapors then flow to the vaporsphere.  Position of the diaphragm in the vapor-




sphere automatically actuates a compressor that draws the vapors from the sphere




and injects them at about 200 psig into the absorber.  Countercurrent flow of




stripped gasoline from the saturator or of fresh gasoline fror. storage is used




to absorb the hydrocarbon 'vapors.   Gasoline from the absorber bottoms is returned




to storage vhile the tail gases, essentially air, are released to the atmosphere




through a backpressure regulator.   Some difficulty 'has been experienced with air




entrained or dissolved in the sponge gasoline returning to storage.  Any air




released in the storage tank is discharged to the atmosphere saturated with




hydrocarbon vapors.  A considerable portion of the air can be removed by flash-




ing the liquid gasoline from the absorber in one or more additional vessels




operating at successively .lower pressures.




     Another type package unit is offered by Farker-Hannifin.  This unit does




not utilize a vaporsaver to level off the vapor flow rate to the compressor and,




therefore, requires a larger compressor and higher power usage.  The unit also




incorporates a refrigeration unit in conjunction with the absorber.  Costs of




the two systems are comparable since the savings from eliminating the vaporsaver




is  offset by the cost of th'e larger compressor and refrigeration unit.




     A third type of package unit adsorbs the hydrocarbon vapors on activated




carbon.  The application of this type of unit is presently restricted to loading




installations that have low throughputs of gasoline, since the adsorbing capacity




and the life of the carbon are limited.  Units of this type find application in




control of vapors resulting from fueling of jet aircraft.




     The vapors displaced during bottom filling are minimal.  Data indicate a




volume displacement ratio of vapor to liquid of nearly 1:1.  A closed system can

-------
                                   -  kk.

then be employed by returning all the displaced vapors  to a storage tank.   The
storage tank should be connected to a vapor recovery system.
     For medium-sized terminals loading less than about 1.8 million barrels per
year of gasoline, extensive recovery facilities such as those described above
cannot be economically Justified and the preferred emission control method is
incineration.  For small facilities such as bulk plants which handle only
several truck loads per day, even incineration cannot be justified and a require-
ment for emission controls would result in closing of the facility and reorgani-
zation of the gasoline distribution system.
OIL-WATER SEPARATORS
     A typical waste-water gathering system for a modern refinery usually includes
gathering lines, drain seals,  junction boxes, and pipes of vitrified clay or con-
crete for transmitting waste water from processing units to large basins or ponds
used as oil-water separators.  These basins are sized to receive all effluent
water, sometimes even rain runoff; they are constructed as earthen pits, concrete-
lined basins, and steel tanks.
     Liquid wastes discharging to these systems originate  at a wide variety of
sources such as  pump glands, accumulators,  spills, cleanouts, sampling lines,
and relief  valves.
     Organic compounds  can  escape  to  the  atmosphere  from openings  in the sewer
system, channels,  vessels,  and oil-water  separators.   The  large  exposed surface
area of these separators  can result  in large  hydrocarbon  emissions  to the  atmo-
sphere.
     The  most effective means  of control  of hydrocarbon  emissions  from  oil-water
separators  has  been the covering of  forebays  or primary  separator  sections.
 Either fixed roofs or floating roofs  are  acceptable  covers.   Separation and skim-
ming of over 80 percent of the floatable  oil  layer takes  place  in  the  covered
sections.   Thus, only a small  amount of oil is contained in the  effluent  water,

-------
which flows under concrete curtains to the open afterbays or secondary separator




sections.




     Satisfactory fixed roofs have been constructed by using vooden beams for




structural support and asbestos paper as a cover.  A mastic-type sealing compound




is then used to seal all joints and cracks.  Although this form of roof is accept-




able for the control of pollutants, in practice a completely vaportight roof is




difficult to achieve.  The resultant leakage of air into the vapor space, and




vapor leakage into the atmosphere are not desirable from standpoints of air pol-




lution or safety.




     The explosion hazard associated with fixed roofs is not present in a float-




ing-roof installation.  These roofs are similar to those developed for storage




tanks.  The floating covers are built to fit into bays with about 1 inch of




clearance around the perimeter.  Fabric or rubber may be used to seal the gap




between the roof edge and the separator wall.  The roofs are fitted with access




manholes, skimmers, gage hatches, and supporting legs.  In operation, skimmed




oil flows through lines from the skimmers to a covered tank or sump and then is




pumped to demulsifying processing facilities.




      A simpler type  of floating  cover can be provided by Foamglas slabs which




can be used to cover  the primary section of the  oil-water separator.  This mate-




rial  has been used to cover  the  oil-water separators at Wood River where it is




estimated  that evaporation losses  have been by 100 to 600 B/D depending on the




season and slop  oil  composition.   Estimated efficiency  is 80 to 90 percent.




     Any type of floating cover on an oil-water separator will interfere with




skimming operations  if the separator is equipped with flight skimmers.  This




problem can be overcome by increasing the weir level so that the flights are




completely submerged.  This will result in constant retention of a layer of oil




several inches deep  in the separator at all times.

-------
                                    - U6 -

                                                              o
     The installed cost of Foam glas slabs  is roughly $-50/ft.   compared to

$1.00-3.00/ft.^ for other types depending on the elaborateness  of the^cover.

     In addition to covering the separator  open sewer lines  that may carry

volatile products can be converted to closed, underground lines with water-seal-

type vents.  Junction boxes can also be vented to vapor recovery facilities,  and

steam can be used to blanket the sewer lines to inhibit formation of explosive

mixtures.

PUMPS

     Pumps are used in every phase of the petroleum industry.   Their applications

range from the lifting of crude oil from the depths of a well' to the dispensing

of fuel to automobile engines.  Leakage from pumps can cause  air pollution wher-

ever organic liquids are handled.

     Pumps are available in a wide variety  of models and sizes.  Their capaci-

ties may range from several milliliters per hour, required for some laboratory

pumps, to 3A million gallons per minute, required of each of the new pumps  at

Grand Coulee Dam.

     Materials used for construction of pumps are also many and varied.  All the

common machinable metals and alloys, as well as plastics, rubber, and ceramics,

are used.  Pumps may be classified under two general headings,  positive displace-

ment and centrifugal.

     Positive-displacement pumps have as their principle of operation the dis-

placement of the liquid from the pump case  by reciprocating action of a piston

or diaphragm, or rotating action of a gear, cam, vane, or screw.  The type of

action may be used to classify positive-displacement pumps as reciprocating or

rotary.  When a positive-displacement pump is stopped, it serves as a check valve

to prevent backflow.

     Centrifugal pumps operate by the principle of converting velocity pressure

generated by centrifugal force to static pressure.  Velocity is imparted to the

-------
fluid by an impeller that is rotated at high speeds.   The fluid enters  at.the




center of the impeller and is discharged from its periphery.   Unlike positive-




displacement pumps, when the centrifugal type of pump is stopped there  is  a




tendency for the fluid to backflow.




     Other specialized types of pumps are available,  but, generally, the pumps




used by the petroleum industry fall into the two categories discussed.




     Power for driving the various types of pumps is  usually derived from electric




motors, internal combustion engines, or steam drives..  Any one of these sources




may be adapted for use with either reciprocating pumps or centrifugal pumps.




Most rotary pumps are driven by electric motor.




     Operation of various pumps in the handling of fluids in petroleum process




units can result in the release of air contaminants.   Volatile materials such as




hydrocarbons, and odorous substances such as hydrogen sulfide or mercaptans are




of particular concern because of the large volumes handled.  Both reciprocating




and centrifugal pumps can be sources of emissions.




     The opening in the cylinder or fluid end through which the connecting rod




actuates the piston is the major potenial source of contaminants from a recipro-




cating pump.  In centrifugal pumps, normally the only potential source of leakage




occurs where the drive shaft passes through the  impeller casing.




     Several means have been devised for sealing the annular clearance between




pump shafts and fluid casings to retard leakage.  For most refinery applications,




packed seals and mechanical seals are widely used.




     Packed seals  can be used on both positive displacement and centrifugal type




pumps.  Typical packed seals consist of a stuffing box filled with sealing




material that encases the moving shaft.  The stuffing box  is fitted with a takeup




ring that  is made  to compress the packing and cause it to  tighten around the shaft.




Materials  used  for packing  vary with the product temperature, physical and chemi-




cal properties, pressure, and pump  type.  Some commonly used materials are metal,

-------
rubber, leather, wood, and plastics.




     Lubrication of the contact surfaces of the packing and shaft is  effected by




a controlled amount of product leakage to the atmosphere.   This feature makes




packing seals undesirable in applications where the product can cause a pollu-




tion problem.  The packing itself may also be saturated with some material such




as graphite or oil that acts as a lubricant.  In some cases cooling or quench




water  is used to cool the impeller shaft and,"the bearings.




     The second commonly used means of sealing is the mechanical seal, which was




developed over a period of years as a means of reducing leakage from pump glands.




This type of seal can be used only in pumps that have a rotary shaft motion.  A




simple mechanical seal consists of two rings with wearing surfaces at right




angles to the shaft.  One ring is stationary while the other is attached to the




shaft  and rotates with it.  A spring and the action of fluid pressure keep the




two faces in contact.  Lubrication of the wearing faces is effected by a thin




film of the material  being pumped.  The wearing faces are precisely finished to




ensure perfectly flat surfaces.  Materials  used in the manufacture of the sealing




rings  are many  and varied.  Choice of materials depends primarily upon properties




of fluid being  pumped, pressure, temperature, and speed of rotation.  The vast




majority of  rotating  faces  in commercial use are made of  carbon.




      Emissions  to  the atmosphere from centrifugal pumps may be controlled in




some  cases by use  of  the  described mechanical-type seals  instead of packing




glands.  For cases not feasible  to control  with mechanical seals,  specialized




types  of pumps,  such  as canned,  diaphragm,  or  electromagnetic, are required.




      The canned-type  pump is  totally  enclosed, with  its motor  built as  an integral




part  of the  pump.   Seals  and  attendant  leakage are eliminated.  The diaphragm




pump  is another type  devoid of  seals.   A diaphragm is  actuated hydraulically,




mechanically,  or pneumatically  to  effect a  pumping action.  The  electromagnetic




pumps use  an electric current passed  through the  fluid, which  is  in  the presence

-------
of a strong magnetic field, to cause motion.
     A pressure-seal-type application can reduce packing gland leakage.   A
liquid, less volatile or dangerous than the product being pumped, is introduced
between two sets of packing.  This sealing liquid must also be compatible with
the product.  Since this liquid is maintained at a higher pressure than the
product, some of it passes by the packing into the product,  '.Che pressure
differential prevents the product from leaking outward, and the sealing liquid
provides the necessary lubricant for the packing gland.  Some of the sealing
liquid passes the outer packing (hence the necessity of low volatility), and a
means should be provided for  its disposal.
     This  application is also adaptable to pumps with mechanical seals.  A dual
set of mechanical seals similar to  the two  sets of packing is used.
     Volatile vapors that  leak past a main  seal may be vented to vapor recovery
by using dual seals  and a  shaft housing.-               ,77
     Other than the  direct methods  used to  control leakage, operational  changes
may minimize  release of contaminants to the atmosphere.   One desireable  change
 is to  bleed off pump casings  during shutdown to the  fuel  gas system,  vapor
 recovery facilities, or  a  flare  instead of directly  to  the atmosphere.
 VALVES
      Valves are employed  in every phase  of the petroleum industry where petroleum
 or petroleum product is  transferred by piping from one point  to another.  There
 is a  great variety of  valve designs, but, generally,  valves may be classified by
 their application as flow control or pressure relief.
      Manual and automatic flow control valves are used to regulate the flow of
 fluids through a system.   Included under this classification ere the gate,  globe,
 angle, plug,  and other common types of valves.  These valves are subject to
 product leakage from the valve stem as a result of the action of vibration, heat,
 pressure, corrosion, or improper maintenance of valve stem packing.

-------
                                   -  50 -






     Pressure relief and safety valves  are used to prevent excessive  pressures




from developing in process vessels and  lines.   The relief valve designates  liquid




flow while the safety valve designates  vapor or gas flow.  These valves  may




develop leaks because of the corrosive  action of the product or because  of  fail-




ure of the valve to reseat properly after blowoff.  Rupture discs are sometimes




used in place of pressure reflief valves.   Their use is restricted to equipment




in batch-type processes.  The maintenance and operational difficulties caused




by the inaccessibility of many pressure relief valves may allow leakage  to




become substantial.



     Obviously, the controlling factor in preventing leakage from valves is




maintenance.  An effective schedule of inspection and preventive maintenance can




keep leakage at a minimum.  Minor leaks that might not be detected by casual




observation can be located and eliminated by thorough periodic inspections.  New




blind designs are being  incorporated in refinery pipeline systems in conjunction




with flow valves.  This  is done to ensure against normal leakage that can occur




through a closed valve.



     Emissions from pressure relief valves are sometimes controlled by manifold-




ing  to a  vapor control  device.  Normally, these disposal systems are not designed




exclusively  to collect  vapors  from relief valves.  The primary function of the




system may be  to collect off gases produced by a  process unit, or vapors released




from storage  facilities, or those released by  depressurizing equipment during




shutdowns.



     Another method  of  control to prevent  excessive  emissions  from relief  valve




leakage  is the use of a dual valve with a  shutoff interlock.   A  means of removing




and  repairing a  detected leaking  valve without waiting until the equipment can




be'taken out of  service is thus provided.   The practice  of  allowing  a valve  with




a minor  leak to  continue in service  without correction until the operating unit

-------
                                   -  51 -



is shut down for general Inspection is common in mny refineries.   This  practice




should be kept at a minimum.



     A rupture disc is sometimes used to protect against relief valve leakage




caused by excessive corrosion.   The disc is installed on the pressure side of




the relief valve.  The space between the rupture disc and relief valve seat




should be protected from pinhole leaks that could occur in the rupture disc.




Otherwise, an incorrect pressure differential could keep the rupture disc from




breaking at its specified pressure.  This,  in turn, could keep the relief valve




from opening, and excessive pressures could occur in the operating equipment.




     One method of ensuring against these small leaks in rupture discs is to




install a pressure gage and a small manually operated purge valve in the system.




The pressure gage would easily detect any pressure increases from even small




leaks.  In the event of leaks,  the vessel would be removed from service, and the




faulty rupture disc would then be replaced.  A second, but less satisfactory




method from an air pollution control standpoint, is to maintain the space at




atmospheric pressure by installing a small vent opening.  Any minute leaks would




then be vented directly to the atmosphere, and a pressure increase could not




exist.

-------
                             TABLE I

             SUMMARY OF HYDROCARBON EMISSIONS
         Source                                Percent of Total

Transportat ion                                       51-9

     Motor Vehicles             W-T
     Aircraft                    1-0
     Railroads                   1-0
     Vessels                     0.2
     Other                       1-0

Fuel Combustion                                       2-2

Industrial Processes                                 ^'^

Solid Waste  Disposal                                  5-0

Miscellaneous                                        2o"-5

     Forest  Fires                6-9
     Organic Solvent             9-7
     Gasoline  Marketing         3-8
     Agricultural  Burning       5-3
     Other                       0 -8                 _

                                                     100.0

Total  Emissions  -  32.0 X 10  Tons /Year

-------
                         TABLE II

          INSTALLED COSTS OF STORAGE TANKS (1972)
                                     Cost ($/B)	
         Type              20,000 BU0,000 BBo~COO B

Cone Roof                    2.20       1.85       1.60

Floating Roof
     a) Pontoon-Type         2.90       2.20       1.75
     b) Double.Deck          3-20       2.80       1-90

Hemispheroids (2.5psi)      ^.00       2-90

Spheroids (5 psi)            5-60       ^.00
         (15 psi)            6.kQ       U.90

Spheres (30 psi)             7-20       6.00
        (50 psi)            10.to       8.30

-------
      EVAPORATION LOSS IN THE PETROLEUM INDUSTRY—CAUSES
                                         AND CONTROL
                      CHAPTER 1—SOURCES OF  EVAPORATION LOSS
  Evaporation loss is the natural process whereby a
liquid is converted to a vapor  which subsequently is
lost to the atmosphere. The liquid may be unconfirmed
or may be enclosed in a container such as an oil-storage
tank.  By definition, evaporation loss occurs only when
the vapors reach the atmosphere.
  Evaporation loss is common  to all branches of the
petroleum industry. Because tanks are used similarly
throughout the industry, sources of loss from tanks are
discussed first. Other sources of evaporation loss asso-
ciated with operational features of each branch are then
considered. To further emphasize true sources of loss,
conditions which can falsely indicate  loss are  also dis-
cussed.

A. Loss in Storage

  Six kinds of evaporation loss occur from petroleum
in storage:  breathing loss,  standing-storage loss,  riling
loss, emptying loss, wetting loss, and boiling loss.
  Breathing Loss:  Vapors expelled  from a tank be-
cause of the  thermal  expansion of existing vapors,
and/or  expansion caused  by  barometric   pressure
changes, and/or an increase in  the amount of  vapor
from  added vaporization in the absence  of liquid-level
change, except  that which  results from, boiling,  is de-
fined  as breathing  loss. The term vapor is used herein
to denote any mixture of  hydrocarbon vapor and air.
The term hydrocarbon vapor refers to hydrocarbons in
the gaseous state independent of the presence or ab-
sence of air.
  Breathing loss takes place in most types of tanks and
occurs when  limits of pressure or volume change are
exceeded.  Fixed-roof tanks,  herein denoting  ordinary
storage  tanks designed for  only a few inches  of water
pressure or vacuum,  suffer relatively large  breathing
losses. Tanks protected from a loss or gain in heat by
reflective coatings, burying, insulation,  or shading ex-
perience less  breathing  loss.  Pressure  tanks  which
operate at  2| psig or higher, normally experience rela-
tively little or no breathing loss.  Variable-vapor-space-
tank  systems  also normally experience little or  no
breathing loss.  Floating-roof tanks  almost  eliminate
vapor spaces, and little or no breathing loss occurs past
the seals.
  Standing-Storage Loss: Vapor from tanks, which re-
sults  from causes  other than breathing or change in
liquid level, is defined as  standing-storage loss.  For
floating-roof tanks, the largest potential source of stand-
ing-storage loss is  attributed to an improper fit  of  the
seal and shoe to the shell. This condition exposes some
liquid surface to  the  atmosphere;  wind  affects  this
source of loss. Also, a small amount of vapor may per-
meate through  the flexible membrane  that  seals  the
space between the  shoes and the roof. The permeation
of flexible membranes,  or absorption in liquid seals,
may also be  a source of loss from variable-vapor-space
tanks. Other sources of standing-storage loss are vapor
escape from  open  hatches or  other openings,  glands,
valves, and fittings.
  Filling Loss:  Vapors expelled from a tank as a re-
sult of filling, irrespective  of the exact mechanism by
which the vapors are produced, is defined as filling loss.
This loss is  common to all types  of  tanks except the
floating-roof tank and  closed-system  pressure storage,
such as  for liquefied petroleum gas (LPG).  It occurs
when the pressure inside  the  tank exceeds  the relief
pressure.  For fixed-roof tanks, the relief pressure  is
low, therefore the  filling loss is relatively high.  Filling
loss  from pressure and variable-vapor-space tanks  is
somewhat less because  these tanks have added vapor-
storage-capacity. The pressure tank also promotes con-
densation of hydrocarbon vapors during filling.
  Emptying Loss:  Vapors expelled from a tank after
the liquid is  removed is defined as emptying  loss.  Be-
cause vaporization lags behind the expansion of the
vapor space during such withdrawal, the partial pres-
sure of the hydrocarbon vapor drops. Enough air enters
during the withdrawal to maintain total pressure at at-
mospheric pressure.  When vaporization into the  new
air reaches equilibrium, the vapor volume exceeds the
capacity of.  the vapor  space.  This  increase in vapor
volume causes the expulsion.
   Emptying loss is common to all types of  tanks ex-
cept the floating-roof tank and closed-system pressure
storage.  Fixed-roof tanks are  most vulnerable to this
loss. Pressure tanks and variable-vapor-space tanks are
less subject to this loss but will encounter it if the vapor-
storage capacity is exceeded.
   In the loading of transportation vessels the definition
of emptying  loss is restricted: The transporter con-
siders emptying loss to be only that portion which
evaporates into the vapor space of the tank during the
actual withdrawal,  that is, between  the opening  and
closing of the gages.
   Wetting Loss: Vaporization of liquid from a wetted-
tank wall, exposed when a floating roof is lowered by
withdrawal of liquid, is defined as wetting loss.  This
source of evaporation loss is small.

-------
10
EVAPORATION Loss—CAUSES AND CONTROL
  Boiling Loss: Vapors expelled from a tank as a re-
sult of boiling of the liquid is arbitrarily defined as boil-
ing loss. Boiling loss may occur from any  tank.  The
fixed-roof tank is more subject to this  loss than  the
pressure tank.  The earliest  floating-roof tank, the pan
type,  is  especially  vulnerable to boiling loss because
heat is readily conducted through the roof  directly to
the liquid and no vapor-storage capacity exists under
the deck.

B.  Loss in Production

  Production entails three operations which contribute
to evaporation loss:  gas-oil separation, emulsion treat*
ing, and lease-tank operation.
  in  gas-oil separation, the oil may be rich in light
components, which later are lost from the usual fixed-
roof lease  tank.  In  a  recovery system, butanes  and
pentanes may not be completely extracted from the gas
and may be lost.  A true evaporation loss occurs when
the gas is flared or vented.   In addition  to  the  loss in
crude-oil volume, the API gravity is decreased.
  In  emulsion treating, heat is applied and released
vapors may be vented. Also, the crude oil reaching the
lease  tank at elevated temperature contributes  to  the
evaporation loss.
  At the lease tank,  splashing may occur as oil is in-
troduced; in such cases vaporization and evaporation
loss  are accelerated.  Dark-colored  tanks  contribute
further to evaporation loss.

C. Loss in Refining

  Refining involves three operations which are sources
of evaporation loss:  treating  and blending in freely
vented vessels, such  as  an  agitator;  pressure systems
which may  leak; and sewers, ponds, and  open sepa-
rators.
  Use of air and agitation can result in  high-evapora-
tion loss from vessels which are not part of a closed
system.  Sweetening naphtha in agitators and blending
volatile components in a semiopen  vessel are potential
examples of this source of evaporation loss.
  Pressure systems, common to refineries and natural-
gasoline extraction plants, may have sources of evapora-
tion loss from leaking exchangers,  glands,  valves, and
fittings.  Hydrocarbon vapor may leak directly  to  the
atmosphere.  Also, liquid may leak and evaporate rap-
idly if volatile at  the operating temperature. Besides
outward leaks, inward leaks of air, such as at pump suc-
tions, are sources  of loss because this air  becomes at
least partially saturated before venting.
  Sewers, ponds,  and open  separators are  sources of
evaporation loss if volatile liquids are permitted to reach
them.  Such liquids usually  encounter high  turbulence
in sewers and collect in thin layers offering large  ex-
posures  for  evaporation.  The recovered  skimmings
from  ponds and separators  require demulsification in-
                        volving heat and constitute another source of loss from
                        the recovery equipment.

                        D. Loss in Transportation and Marketing

                          Transportation includes  pipeline shipments and the
                        loading, transit, and unloading of transport vessels from
                        which evaporation loss can occur.  Pipelines are subject
                        to loss from air eliminators used in metering systems
                        and from leaking glands,  valves and fittings, and cor-
                        roded pipes.
                          Filling and emptying losses occur from nonpressure
                        tankers, barges, tank cars,  and trucks in much the same
                        manner as from tanks. If tank cars and trucks are top-
                        loaded with a short filling pipe, the undue splashing not
                        only accelerates vaporization but also produces  small
                        droplets  which may be lost by entrainment.  Air  in-
                        spired during loading can  be an added source of evap-
                        oration loss because  such air becomes at least partially
                        saturated with hydrocarbons before it is vented; loosely
                        connected,  submerged loading spouts  are sources of
                        such inspiration.
                          In-transit losses from transport vessels are essentially
                        breathing losses.  Excessive heating  of crude  oil in
                        marine vessels is a potential source of evaporation loss
                        during transit.
                          Marketing operations entail many of the previously
                        discussed sources of loss, particularly those discussed in
                        Par. A, "Loss in  Storage."

                        E. False Indications of  Loss

                          Under certain conditions a loss appears to have oc-
                        curred which actually did not.  Being aware of  these
                        conditions  and being able to distinguish between an
                        "actual" and a "false" loss is important, otherwise cor-
                        rective efforts may be directed toward conditions where
                        no real improvement can result. Conversely, these con-
                        ditions may balance out and cover up an  actual loss,
                        with the result that necessary corrective action is  over-
                        looked. Five such conditions are:  inaccurate measure-
                        ment, gravitation between tanks,  Inaccurate volume of
                        supply lines, inaccurate calibration of meters, and physi-
                        cal changes in volume.
                          Inaccurate Measurement:  Apparent gains or losses
                        can result from inaccurate measurement, either of aver-
                        age liquid  temperature, height of liquid, or incorrect
                        calibration of containers. Other sources of error would
                        be the  failure to correct all volumes to  a common tem-
                        perature  base by  use of the unabridged Table 6, "Re-
                        duction of Volume  to  60 F Against API Gravity at
                        60 F," of the ASTM-IP Petroleum Measurement Tables
                        (1953)'.
                          Gravitation Between Tanks:   Any  leakage  past a
                        valve believed to close a line between two tanks results
                        in gravitation—a  loss of product in one tank results in
                        a gain in the other tank which may not be observed.
                          Inaccurate Volume of Supply Lines: If the volumes
                        of the supply lines are not known accurately, or if lines

-------
                                     SOURCES OF EVAPORATION  Loss
                                                11
are not in the full or empty condition which is assumed,
unrealistic losses or gains may be indicated.
  Inaccurate Calibration of Meters:  The  inaccurate
calibration  of meters can  result in apparent  losses,
which may be attributed incorrectly to evaporation—
or in apparent gains which may conceal actual evapora-
tion losses.
  Physical Changes in  Volume:  Certain  processing
operations, such as cracking, polymerization, and  the
blending and separation of light and heavy stocks, re-
sult in physical changes in volume even when full cor-
rection is made for changes in temperature.  For ex-
ample, in a cracking process, where small molecules are
produced from large ones, the products will occupy a
greater volume than the  charge.  In a polymerization
process, where large molecules are produced from small
ones, the product volume shrinks. With such volume
changes  API  gravity  always  changes  but the  total
weight, before and after the volume change, is the same.

-------
            CHAPTER 2-FACTORS  AFFECTING EVAPORATION LOSS  FROM TANKS
     The total amount of evaporation loss depends upon
   the rate of loss and the period of time involved. Primary
   factors affecting the rate  of loss are:  true vapor pres-
   sure of the liquid, temperature changes in the tank tank
   outage,  tank diameter, schedule  of tank filling and
   emptyings,  tank condition, and type of tank.  Satura-
   tion and diffusion effects are  only a part of the mecha-
   nism of the loss and are classed as dependent, or second-
   ary, variables.  Although  quantitative-loss relationships
   for the primary  factors are not yet available, a fair un-
   derstanding based on theory and practice can be gained
   by considering the mechanism  of loss from  fixed-roof
   tanks.   With such understanding,  the advantages  of
   floating-roof, variable-vapor-space, and pressure-tank
   systems will readily be apparent.

  A. True Vapor Pressure of the Liquid

    True vapor pressure affects the rate of loss because it
  is the basic  force causing  vaporization. It varies  with
  liquid composition and temperature. True vapor pres-
  sure at storage temperature is all important. For hydro-
  carbon mixtures, this pressure decreases with evapora-
  tion because of the change in liquid composition  True
  vapor pressure usually is determined from correlations
  relating it to Reid  vapor pressure (RVP).  Such rela-
  tjpnships  are illustrated in  nomograph form in Appen-

    The effect of true vapor pressure on rate of breathinc
  loss from  a fixed-roof tank involves at least two internal
  considerations—the  saturation  concentration  and  the
  diffusion and convection factor.  The maximum  con-
 centration of hydrocarbons  which can be present in ex-
 pelled vapor,  known as the saturation  concentration
 increases in direct proportion to true vapor pressure  It
 follows  that  if vented  vapors  were fully  saturated
 evaporation loss  would increase  rapidly as true vapor
 pressure approaches the tank relieving pressure (a boil-
 ing  condition).   However,  another mechanism—the
 diffusion and convection of hydrocarbon vapor from the
 hquid  surface through the vapor space  is too  slow to
 fully saturate  it.  Experience shows that  vapors vented
 during normal breathing are usually only  80 per cent to
 90 per cent saturated.  Thus, the  driving  force to over-
 come resistance  to  diffusion factors and convection
 through the vapor space is one of the controlling factors
 Such driving force can be  looked upon as bein<» the
 true vapor pressure of the liquid minus the partial pres-
 sure of hydrocarbons in the vapor space.  As true vapor
pressure  rises,  this driving force  would  rise in direct
proportion if percentage saturation in the vapor space
remains constant.  Thus, both the  saturation considera-
tion and the diffusion and convection consideration sug-
                                                     12
   gest that actual loss is at least directly proportional to
   rising true vapor pressure.
     Filling or emptying losses from fixed-roof tanks are
   directly proportional to increasing true vapor pressure
   because of the relationship between true vapor pressure
   ana saturation concentration.
     This concept does not apply when true vapor pres-
   sure exceeds the absolute tank pressure because boilin*
   occurs  and  losses may be large.  Then, the main con-
   trolling factor is heat input.
     In terms of total loss over a period of time, the effect
   of true vapor pressure depends upon the composition
   ot the stock.  For example,  two crude oils of identical
   true vapor pressure may weather at different rates. One
   crude oil may contain a relatively  high  per cent of
   volatile propane and ethane; for a specific starting loss
  rate, the vapor pressure will drop rapidly  and the loss
  rate  will drop shortly thereafter.  The other crude  oil
  may derive  vapor pressure  from relatively hi°h  con-
  centrations of less-volatile pentanes  and butanes-  for
  the same starting loss rate, the  vapor pressure will drop
  less rapidly but the loss rate will remain higher for a
  longer period.  This  consideration is  particularly si°-
  nificant  for newly  produced crude oils at leases  an°d
  pipeline  storage terminals where  true vapor pressures
  may be close to atmospheric pressure.

  B.  Temperature Changes in the Tank

   Internal temperature  changes, brought about by at-
  mospheric and solar heat, tend  to cause the tank vapor
  space to  breathe.  During the day, heat flowing through
  the roof  and  upper walls raises the vapor  temperature
 and expands  the volume. The pure thermal effect  is
 augmented by vaporization of hydrocarbons from  the
 tank contents during the same period. The heat input
 also may increase  the liquid-surface temperature and
 accelerate vaporization.   At  night, reverse processes
 shrink the vapor and cause an intake of air.
   Atmospheric and solar heat also cause forced con-
 vection in the vapor space which promotes evaporation
 from the liquid surface and aids  in the dispersion of the
 hydrocarbon vapor.
   Although efforts  have been made to develop more
 precise criteria, the average daily change in atmospheric
 temperature is still  the only accepted  way  to charac-
 terize  atmospheric   and  solar-heat effects.   Monthly
 meteorological data for various locations in the United
 States and Canada are presented  in Appendix VI.
   Theoretical  considerations do not permit good es-
 timation of how much loss will  increase with increas-
ing atmospheric temperature change; however, it prob-

-------
                                        EVAPORATION-LOSS FACTORS
                                                 13
ably will be somewhat less than directly proportional to
the increase in atmospheric temperature change.

C  Tank Outage
  The volume of most vapor spaces is directly propor-
tional to outage—the height of the vapor space. For a
fixed-roof tank,  higher outage  means greater loss be-
cause the  larger volume  will breathe  more. However,
when outage is increased heat input is not increased in
direct proportion.  Heat enters the vapor space through
the tank wall,  the area  of which  increases in direct
proportion, and through the tank roof,  the area of which
remains unchanged. Furthermore,  with added height
of vapor space,  resistance  to  transfer of hydrocarbon
vapors from  the Liquid surface to  the  vent increases.
Therefore,  the average concentration  of hydrocarbons
in vented vapor  should fall. Experience has confirmed
that loss will increase less than directly proportional to
increasing outage.

D.  Tank Diameter
  Tank diameter  influences the  volume of the vapor
space and  the liquid-surface conditions. Breathing  is
less than directly proportional  to increase in vapor vol-
ume because of the less  than proportional increase in
area for heat transfer  into the vapor space. Further-
more, increasing diameter should reduce the tempera-
ture rise of the  liquid surface because the rising hot
stock, in contact with the tank wall, must  spread in a
thinner  film over the surface area.  Assuming constant
tank height, total breathing loss, therefore,  increases at
a rate less than directly proportional to tank volume.

E.  Schedule of Tank Fillings and Emptyings
  Over a period of time, the  frequency of stock turn-
over and the average outage affect total loss.  Opera-
tions that promote high outages may result  in relatively
high breathing losses.  Fillings and emptyings scheduled
to compensate the dally temperature  changes may  re-
duce breathing loss. The time interval between empty-
ing and filling may have a significant effect on loss. For
a system of tanks connected with vapor lines,  simul-
taneously filling one tank while emptying another main-
tains the vapor-storage capacity relatively constant and
filling loss is reduced.

F.  Tank Condition

  Tank condition is another factor affecting loss rate;
however, quantitative effects cannot be predicted. Open
vents result in high loss when gusty or turbulent winds
cause rapid pressure changes In tanks in which volatile
Liquids  are stored. Rapid  breathing occurs  as short
puffs.  Any hole in a tank roof, diaphragm, seal, or  ac-
cessory results in the same type of loss.
  Where there are two or more openings in the tank,
loss is further increased.  Pressure differences,  which
result from wind or thermal effects, cause a  constant
flow of air through some openings into the vapor space
and an outflow of vapor through  other openings.

G.  Type of  Tank

  The type of tank or storage system will affect  the
evaporation loss experienced. The amount of loss de-
pends  upon the volume of the vapor space available
and the pressure limitations of the equipment.
  If  tanks have  their vapor  spaces interconnected,
vapor-space volume can be controlled to a Limited ex-
tent by scheduling fillings and emptyings, where feasible.
  If the vapor space  is allowed  to change  volume at
constant pressure, breathing loss can be  practically
eliminated and filling loss can be  reduced.  The  extent
of the  reduction in loss is dependent upon the amount
of variable vapor space provided.

-------
         CHAPTER 3-TANKS AND  EQUIPMENT TO CONTROL  EVAPORATION  LOSS
     The industry may choose from four  basic types of
   tanks for storage of petroleum and its products:  fixed-
   roof tanks,  floating-roof  tanks,  variable-vapor-space
   tanks, and pressure tanks.  Each type is designed for
   specific storage requirements, the actual storage  prob-
   lem should determine the type selected.  In many in-
   stances, the most economical tank can be selected only
   after a detailed study comparing loss from, and the cost
   of different tanks. For stocks having a low true vapor
   pressure, less than 2 psia, the fixed-roof tank generally
   will be the most economical selection.  For stocks of
   motor-gasoline range of  volatility  at high throuohputs
   the floating-roof tank generally will be the best choice'
   but at lower throughputs the variable-vapor-space tank
   generally will be better.  For stocks which boil at at-
  mospheric  pressure and storage temperature, pressure
  tanks are best generally; however, in some cases use of
  the  fixed-roof tanks in conjunction with a vapor-recov-
  ery  system  may offer more advantages.
    The tendency to boil in storage is a function of vapor
  pressure, altitude, barometric pressure, and liquid-sur-
  face  temperature.  Maximum  liquid-surface  tempera-
  tures  vary  throughout the  United  States.   For  the
  indicated temperatures, the maximum Reid vapor pres-
  sures (RVP)  of  stocks which can be stored at atmos-
  pheric pressure without general  boiling (but at the ex-
  pense of high-loss rates) are:

                                 Maximum
                                  Liquid-
                                  Surface
                                 Tempera-   Maximum
                                  ture •     Reid Vapor
                                 (Degrees   Pressure •
              A"3              Fahrenheit)  (Pounds)
 West  Coast (tempered by Pacific Ocean) .  80        18

 Gulf   Coast,  Atlantic  Seaboard,   and   °        15'5
   northern Middle West .............. IQO        13 j

 Mid-Continent area and arid Southwest. .115        n'
                                   120        10
                                   at
   Effective loss-control operation of each tank is de-
pendent upon certain accessory items, such as breather
valves and automatic gages. Continued effective opera-
tion is dependent upon a program to maintain the tank
and accessories in a gaslight condition.
   Choice of paint color may be  an  important factor in
reducing loss.  In special instances, loss may be reduced
by use of a floating plastic blanket; or, by  employin»
insulation,  a shading device, water  sprays, mechanical
cooling, or by burying the tank.  In  some cases, it may
be possible to  reduce loss further by specially schedul-
ing fillings and emptyings.
   A. Fixed-Roof  Tanks

     The minimum  accepted  standard for  stora°e  of
   volatile oils is the fixed-roof tank. It can sustain an in-
   ternal pressure, or vacuum, of only an ounce or two per
   square inch. Being susceptible  to sizable breathing and
   tiling losses, this type of tank  is used most frequently
   for services which  cannot economically  justify a con-
   servation tank.                             }
     Design of Tank:  The fixed-roof tank, the predecessor
   of conservation tanks, came into being durino the early
   days of the petroleum industry. Wooden barrels were
   used at first, but they could not keep up with the rising
   flood  of  oil that  poured from the Pennsylvania wells
   As production increased, open pits and diked areas were
   used, but they were hazardous.  Wooden tanks  caulked
  with oakum and held together with iron hoops,  first ap-
  peared in 1861.  The capacity  of these  tanks  ranged
  between 500 bbl and 1,000 bbl.                  °
    The first iron tank with a wooden, gravelled roof ap-
  peared in 1864;  it provided larger and safer storage
  Shortly after the  Civil War, bolted- and  riveted-stlel
  tanks  came into  use.   They  ranged in  size  up to
  35,000 bbl.  After 1915 capacities were increased and
  in 1919 the first 80,000-bbl tank was erected  The in-
  troduction of electric welding, in 1923, made possible
  the  fabrication of welded roofs and bottoms.  The
  welded tank was introduced in  1927.
    Fixed-roof  tanks built  today  usually  are  welded
  throughout, but many riveted  tanks are still  used and
  bolted  tanks are common in the smaller sizes.  Whereas
  the seams  of welded tanks are almost inherently <»as-
  tight, the seams of bolted and  riveted tanks frequently
 require additional maintenance. In some areas, particu-
 larly on leases  where corrosion is a problem, wooden
 roofs, which are seldom of gastieht construction  are
 still in use. Loss from these tanks is much  greater than
 from steel-roof tanks.
   If a fixed-roof tank is  found to be the best type for
 a particular  storage  problem,   careful consideration
 should be g1Vea to  the size  before the  final selection is
 made. Because the loss rate increases significantly with
 outage and tank diameter, the use of the smallest tank
 possible for the given storage requirement  results in a
 minimum loss. For further insight as to the outage and
 diameter effects, refer to Chapter  2, "Factors Affectino
 Evaporation Loss from Tanks."
   Maintenance of  Tank:  To maintain a gaslight con-
 dition, tanks should be inspected at  regular intervals
 and repaired as necessary.  The frequency of inspections
 usually  is  determined  by  experience.  Riveted-roof
tanks, because of their  greater tendency  to  develop
leaks, should be inspected  more frequently than welded-
roof tanks.
                                                     14

-------
                                        EVAPORATION-LOSS CONTROL
                                                 15
  When the tank is under pressure, five ways in which
leaks may be detected are:
1. Observation of heat-wave-like  trails of the escap-
ing vapors.
2. Hearing the hiss of escaping vapors.
3. Smelling the vapors.
4. Use of gas testers or "sniffers."
5. Applying soap solution or linseed oil to seams.
Also, stains on the painted surfaces frequently indicate
leaks.
  Design of Accessory Equipment: The fixed-roof tank
has several openings in roof for venting, gaging, and
sampling. To maintain a gaslight roof, accessory equip-
ment of  a gaslight design must  be provided for  these
openings.
  The  accessory for  the vent opening is called a
breather valve, pressure-vacuum relief valve, or con-
servation vent.  When operating properly, this device
prevents either the inflow of air or the escape of vapors
until some preset vacuum  or  pressure  is developed.
Most breather  valves,  especially the  metal-to-metal
types, allow some leakage below the pressure or vacuum
setting.  A tight breather valve is important in reducing
evaporation loss. The actual magnitude of the savings
will  depend  upon such factors  as vapor pressure of
stock stored, weather conditions, paint color, and al-
lowable working-pressure range. However, the savings
realized  usually  will pay for  the  installation.   The
breather valve also  contributes to safe operation by
keeping the tank vent closed to the atmosphere most of
the time.
  The pressure and vacuum settings of a breather valve
are dictated by the structural characteristics of the tank
and  should be within safe operating limits. A certain
amount of pressure and vacuum beyond these settings
is necessary to overcome pressure drop in order to  ob-
tain required flow. Proper size and settings can best be
determined by reference  to API  Sid 2000:  Venting
Atmospheric  and Low-Pressure Storage Tanks (1968)
and  to the manufacturer's tank data determined in ac-
cordance with this publication. The  pressure setting
for vent valves to be installed on large tanks constructed
in accordance  with API 12D: Specification for Large
Welded Production  Tanks (1957) usually is limited to
+ oz because roof  plates will start  to shift when  the
pressure rises much above 1 02. For small tanks, and all
tanks having  special  structural  features,  the  pressure
range can be increased in accordance with the manu-
facturer's recommendations.
  Breather valves should be designed to give:
1. High-flow  capacity at relatively  small  pressure or
vacuum  above the setting.
2. A gaslight seal.
3. Freedom from sticking or freezing.
4. Easy access to all  parts for inspection  and main-
tenance.
  Diaphragm  and liquid-seal valves have less  leakage
than  metal-to-metal types.  For  dependable  service,
diaphragms should be resistant to tank vapors.
  Open  vents of the mushroom or return-bend  type
should not be used on fixed-roof tanks storing volatile
oils as they permit high loss.  These  vents are merely
hooded openings equipped with protective screens.  The
opening is turned down to prevent any blockage by ice
or snow.
  Venting  accessories  sometimes used  are:  flame ar-
restors, flame snuffers, and flash screens. They usually
have  little  effect on vapor loss except  when they are
installed  between the tank and vent valve and must be
removed for cleaning.
  Some  vapor loss is  inherent in manual gaging and
sampling methods which necessitate opening  a  lank lo
Ihe atmosphere each time  these  operations are  per-
formed.  This loss can be minimized through the use of
automatic  gaging  equipmenl,  double-closure   gaging
locks, and a syslem of thermometers and sample valves
in the tank shell.
  Accessories which  help  to  reduce  evaporation loss
from  lease  tanks include:
1. Pressure-vacuum thief halch  and venl-line valve.
2. Automatic-closing valve in the  equalizer line which
closes when the gage hatch is opened.
3. A diagonal-slotled  downcomer type  of fill  line to
minimize free fall and splashing.
  Maintenance of Accessory Equipment: To maintain
accessories in a gastight condition they should be in-
spected and restored periodically.  Pallels of the metal-
to-metal  breather valves which become warped  in serv-
ice must  be machined to restore a gastight fit.  Defective
diaphragms of diaphragm  brealher  valves should be
replaced.
  Liquid-seal breather valves may be affected by  dilu-
tion or loss of liquid and may have lo be inspected and
mainlained more frequently as determined by experi-
ence. A  loose filling gage-hatch lid can  be made nearly
gastight  by replacing the gasket or by machining the
sealing surface.
  Rame arreslors and flash screens can become clogged
with dust, rust, and ice.  Such obslruclions in Ihe venl-
ing system can cause severe damage to  the tank from
excessive internal pressure  or vacuum. These acces-
sories should be inspected and cleaned frequently.
  Choice of  Paint: Tank painting is imporlant in re-
ducing evaporalion loss as well as  in preserving the
lank.   An  adequate  paint program,  using  reflective
paints, will minimize the heat inpul lo the lank by re-
ducing Ihe metal lemperalure of  Ihe lank.
  While painl is a simple  and effeclive means for re-
ducing evaporalion loss.  The  addilional low  cosl of
mainlaining a clean white surface on a  tank frequently
has an attractive economic return.  Recenl dala  indicale
that painting the tank roof and shell white, rather lhan

-------
16
                                 EVAPORATION  Loss—CAUSES  AND  CONTROL
gray,  reduces the  evaporation-loss rate by at least
20 per cent.
  Use of Insulation:  Insulation on the roof and shell
of a storage tank tends to reduce heat input and heat
loss; this tendency toward constant temperature reduces
breathing loss.  However, when  stock at higher than
normal storage  temperature  is added to an insulated
tank, the average tank temperature increases, which in
turn  increases true vapor  pressure  and promotes  a
higher  concentration of  hydrocarbons  in   the vapor
space.  This  condition may increase  filling  loss some-
what and breathing loss to a lesser degree.  Insulation
usually is expensive  to install and may involve con-
siderable maintenance.  Unless  moisture is prevented
from entering the insulation, loss of insulation effect as
well as corrosion of the tank shell may result.
   Use of Shade: Shading of a storage tank  from direct
sunlight reduces heat input, but generally does  not re-
duce normal  heat  loss.  Hence, compared to  a  bare
tank,  shading provides less  variation in internal  tem-
perature and, usually, results in a  lower average stock
temperature.  Although the installation of shades gen-
erally has not been considered economical, usually the
maintenance is not an expensive factor.
   Use of Flexible Blanket on Liquid Surface: Evapora-
 tion loss in fixed-roof tanks can be reduced with the use
of  flexible blankets which float  directly on the liquid
 surface.  The blanket  acts in  the same manner  as a
 floating roof. There are  two types of flexible blankets:
 one is a foam blanket  made up of plastic spheres; the
 other is a blanket or raft made from plastic sheeting.
 The  latter type has not been tested extensively in this
 country.
   The floating plastic-foam blanket  consists of micro-
 scopically small, hollow, plastic, gas-filled spheres.  This
 material has been used extensively on fixed-roof storage
 tanks in crude-oil  service.  Tests made in  this country
 and  Canada  have  indicated that under favorable  con-
 ditions evaporation loss on crude oil can  be  reduced
 from 50 per cent  to 70  per cent in working tanks and
 from 70 per cent to 90 per cent in static tanks. A  i-in.
 thickness usually is used on static tanks. A 1-in. thick-
 ness is used on working  tanks to avoid breaking up the
 complete foam layer during filling and emptying.   Gag-
 ing  difficulties  with  the  plastic foam have been mini-
 mized by the use of a portable type of gage well carried
 on a gage tape.
    Loss of the plastic spheres can  occur  if tanks are
 pumped  to  low  levels  at  high pumpout  rates.  The
 operation of mixers must be controlled to prevent the
 plastic spheres from being dispersed in the oil and then
 pumped out during emptyings.
    Not enough information is available to estimate  ac-
  curately the service life of the  spheres.  However,  re-
  ports received indicate  that the material has  operated
  satisfactorily for two and one half years.  The spheres
  should not be installed in tanks containing liquids sub-
  ject  to boiling; water  also may damage the plastic
spheres,  especially,  if agitation  promotes  contact  of
moisture with the spheres.

B.  Floating-Roof Tanks
  The floating-roof  tank is  an effective conservation
device for stocks of motor-gasoline volatility. The basic
design virtually eliminates  the vapor space, which re-
sults in low losses both from  breathing and filling. The
exceptionally low losses  from filling brought this tank
into widespread  use.  Other  advantages of this  tank
are excellent  fire protection  and corrosion  resistance.
Ignition  may  occur  only in  the  seal area; being con-
fined to this localized area, the fire normally is easy to
extinguish.  However, should the  stock boil, the fire
may  be difficult to extinguish.  Excellent  protection
from sour or corrosive  stocks is afforded  by floating
roofs which are in contact with the entire liquid surface.
   Design of Tank:  There are three  basic  designs of
floating-roof tanks in operation:  the pan type, pontoon
type, and double-deck type.
   The first successful floating-roof tank, the pan type,
was built in Gushing, Oklahoma, in 1922. A typical ex-
ample of a pan-type floating roof in operation today is
shown in  Fig. 1.  A  single deck  covers most of the
liquid surface and a seal is attached to  the rim of this
deck. The  deck slopes to the center for  drainage.
   This  roof  has three  disadvantages  which account
 for its limited use.  The single  deck is  exposed to the
 sun during the middle of the day.  Because the deck is
 held forcibly in  contact with the liquid, heat is trans-
 ferred directly to the liquid surface. The liquid-surface
 temperature rises appreciably.  Sometimes the product
 boils and losses of gasoline and similar  stocks may re-
 sult. The  pan roof  also may tip and sink under heavy
 loads of water or snow, or  from leaks  which may de-
 velop in the deck.
    The pontoon roof was developed in  1928 by adding
 pontoons  to  the pan-type roof  to give it greater sta-
 bility and  bouyancy.  Thus, the  pontoon  roof has  a
 single deck over only a part of the total  area, closed
 pontoons cover the  remaining area. Pontoon-type roofs
 are illustrated in Fig. 2 and  Fig. 3.  The pontoons are
 arranged  and compartmented to  provide  floating  sta-
 bility under  heavy loads of water or  snow.  Enough
 bouyancy normally is provided so that the roof will not
 sink when the single-deck area leaks or the drain fails.
 Properly  compartmented, the pontoons can  be par-
 tially flooded without endangering the roof structurally.
    A pontoon  roof having  a  single deck, which can
 rise -above the liquid surface when boiling starts, pro-
 vides an  insulating vapor space and reduces the heat
 transfer from the  sun  to the liquid surface.   Boiling
 losses usually will  not occur with stocks in  the motor-
 gasoline range o£ volatility.
    In the  mid 1940's the  double-deck roof was offered,
 which in effect made the entire roof  a series of pon-
  toons, see Fig. 4.   Circular  and radial bulkheads divide

-------
18
EVAPORATION Loss—CAUSES AND  CONTROL
  Courtesy: Graver Tank and Manufacturing Company, Inc.

                                     FIG. 3—Pontoon-Type Floating Roof.
the space between  the two decks into compartments;
this design can  give  good stability  and load-carrying
capacity  and provides an insulating air space over the
entire area.  Boiling losses usually will not occur  with
stocks in the motor-gasoline range of volatility.
  The usual seal in a floating-roof tank consists  of a
relatively thin-gage  shoe  or sealing ring   supported
against the tank shell around the edge of the floating
roof. The bottom of the sealing ring is below the liquid
surface, and  the top is a few inches above the top rim
of the roof.  A piece of flame-retardant rubberized cloth
closes the space between the sealing ring and the roof.
Another  type of seal consists of a flexible tube, fastened
to the roof  and occupying the annular space between
the roof  and shell.  The tube, rilled with a nonfreezing
liquid,  is held on  the liquid surface and completely'1
eliminates the vapor space. In tanks with riveted shells,
abnormally high losses of the more volatile stocks oc-
cur because  the rivet heads and overlapping steel plates
hold the  sealing ring  away from  the tank shell. Thus,
it is advisable to use a secondary seal on riveted tanks.
This type of seal consists of a strip or loop  of rubber
adapted  to  cover  the slot  at  the  top of  the sealing
ring. The value of a secondary seal for welded  tanks
is uncertain.
   Maintenance of Tank: Efficient  and safe  operation
of any mechanical  device which moves intermittently
requires  inspection  and  maintenance at regular  inter-
vals; the  floating-roof  tank is no exception.  Shoes must
                        fit well, seals must be in good condition, the roof should
                        be level at all times, and the breather valve and bleeder
                        vent must operate  satisfactorily.
                           Before the tank  is put into service  the opening be-
                        tween  tank shell and shoe should be minimized by ad-
                        justing shoe springs or hangers. If the liquid surface
                        is plainly visible between shoes  and shell, after all ad-
                        justments are  made, the tank  may be out-of-round
                        caused by faulty construction or uneven settling.  Be-
                        cause  this condition leads to large losses it  should be
                        corrected. Shoe fit should be checked periodically; at
                        the same time,  the  above-deck hangers  should be  serv-
                        iced to keep them in an operating condition.
                           The primary and secondary seals should be inspected
                        periodically  for tightness  and general condition.  Sec-
                        tions of  the primary seal which have deteriorated and
                        have weakened should be replaced.  Holes that appear
                        in sections of good material may be repaired by patch-
                        ing.  The secondary seals are subject  to considerable
                        abrasive  wear and  are not amenable to patching,  such
                        worn-out seals should be replaced.
                           The floating position (the level)  of  a roof depends
                        upon the weight of the load supported and how easily
                        the roof can move up and  down.  With riveted tanks,
                        shoes  occasionally  bear unevenly on  the  shell of the
                        tank. Inspection for this condition should be made peri-
                        odically  and shoes adjusted as necessary.  After every
                        rain, drainage  from the roof should be checked and
                        any  debris clogging the screened roof drains should be

-------
                                        EVAPORATION-LOSS CONTROL
                                                 19
                           ^
     • -/ •  • -  _^-T»^   ^-
'
  Courtesy: Hammond Iron Works.
                                     FIG. 4—Double-Deck Floating Roof.
removed.  Although a snow load is not serious for the
pontoon or  double-deck  roofs it  can be with the pan-
type roof.  When more than a foot of snow accumu-
lates on a roof of any type, it should be removed,  par-
ticularly if it drifts unevenly.
  Design  of Accessory  Equipment:  Two accessories
are necessary  to the operation of a floating-roof tank:
a breather valve for the rim space  and a bleeder vent for
the roof.  One breather valve, sometimes two,  is  pro-
vided at the outer  edges of all  types of floating-roof
tanks; it is similar in design to that used on fixed-roof
tanks.  The  bleeder vent for the  pontoon and double-
deck tanks allows air trapped under the roof to escape
before  the  roof floats  and prevents a vacuum as the
roof comes to  rest on its supports.
  Maintenance  of  Accessory  Equipment: The  fol-
lowing accessories should  be inspected regularly  and
repaired as  necessary:   breather valves (rim  vents),
bleeder  vents,  sample and gage hatches, and any other
openings from which vapors might escape.
  Choice  of Paint:  The value  of a highly  reflective
paint in reducing  evaporation loss from  floating-roof
tanks is questionable.  Although the floating-roof tank is
a very  efficient conservation  device, savings effected
through the use of a particular color of paint will be
less than for other types of tanks. Reflective paint on
the pan-type roof may  be justified because it will re-
duce the chance for boiling. It is not so important on
the pontoon and double-deck roofs because these roofs
are designed to provide insulating barriers to heat trans-
fer.  Reflective paint  on the shell of these tanks  may
be justified because it may reduce boiling in the seal
area and it will help maintain a lower liquid tempera-
ture throughout the tank. Such reductions are beneficial
particularly for the older riveted tanks where it is dif-
ficult  to maintain a good fit  between  the  shoes and
shell.

C. Variable-Vapor-Space Tanks

  The variable-vapor-space tank is  an effective con-
servation device particularly suited to reducing breath-
ing losses.  Expanding  vapors are stored temporarily
in a gasholder device and vented to the atmosphere only

-------
                                       EVAPORATION-LOSS CONTROL
                                                 25
E. Vapor-Recovery Systems

  Design  of System:  Vapor-recovery systems collect
vapor from storage tanks and send it to a gas-recovery
plant.  These systems  have sensitive  pressure-vacuum
controls and remove vapor as pressures build up dur-
ing pumping into  the tanks or during breathing.  The
vapor is collected, compressed, and then recovered  by
absorption or condensation; with  lease tanks, how-
ever, the  compressed gas normally is discharged into
an extraction-plant gathering system.   A properly de-
signed system should eliminate most of the evaporation
loss, but  because  of control difficulties  the  efficiency
actually is somewhat lower.
  Refinery  or natural gas is  sometimes used  for re-
pressuring the vapor spaces of tanks when air  or a
corrosive  atmosphere is  undesirable  in  the  vapor-re-
covery system.   The vapors  are  withdrawn  from the
tank as the internal pressure increases, and the repres-
suring gas is admitted to the  tank  when air normally
would be  drawn in.  Some provision must be made to
prevent collapse of  the  tanks when insufficient gas
evolves to maintain pressure in the tanks.
  Where  it is uneconomical to design  a tank or a stor-
age system  to  operate at pressures  high  enough  to
make evaporation loss  negligible,  various  vapor-re-
covery methods  can  be utilized. To recover vapor  by
condensation, one or a combination  of  four methods
may be used:
1. Absorption may be accomplished in a suitable liquid
of higher  molecular weight than that of the vapors be-
ing recovered. This  rich oil must be  reprocessed if it
is desired to separate the  absorbed vapors.  The liquid
from which  the  vapors originally escaped also can  be
used as the  absorption medium and then the enriched
liquid can be returned  to the storage  tank without
further processing. Vapor usually  is  absorbed under
pressure.
2. Compression  of  the  vapors,  under  suitable  tem-
perature  conditions,  will  condense  part or all of the
vapors.
3. Cooling,  alone or in combination with compression,
can return vapors to  the liquid state.
4. Adsorption in  suitable  material, such as activated
charcoal or silica gel, is a means of collecting the hydro-
carbon vapors if they have been mixed with noncon-
densables, such as air or other gases.  Further process-
ing by heat will remove the hydrocarbons  from the
adsorbent material; the vapors may then be condensed
to the liquid state by cooling, for return to the tank.
   Maintenance oj System: Vapor-recovery systems  re-
quire that all the  tanks be kept gaslight and that the
instrumentation  and  fittings be adequately  maintained.
   Vapor  lines should  be sloped to a low spot to col-
lect  condensate.  Condensed  vapors  and  moisture
should be drained periodically from each line.  If vapor
lines are underground, the low spot should be in a  pit
and it may be necessary to pump  the condensate.
F. Other Ways to Control Loss

  Special techniques that reduce heat input minimize
evaporation loss.  Before one of the techniques is used,
it should be considered in relation  to the specific prob-
lem; the  reduction in breathing loss anticipated should
be related to the cost of the technique adopted, i.e., that
an economic payout be  obtained.
  Water Sprays:  Water sprays cause cooling  due to
absorption of heat to vaporize the water.
  Mechanical  Cooling:  In mechanical cooling,  cooling
coils or refrigeration units are  used to  reduce  the  ef-
fect  of heat input.  This technique  is probably most
commonly used for  condensing vapors.
  Underground Storage: In underground storage, the
earth eliminates absorption or emission of radiant en-
ergy to or from the tank. Breathing effects  are, there-
fore, greatly minimized.  Accordingly,  where  under-
ground storage is  used for any reason, evaporation loss
is minimized.  Burying tanks near  hot lines should  be
avoided.
  Schedule o] Tank Fillings and  Emptyings: Breath-
ing loss sometimes can be minimized  where conditions
permit coordinating tank filling and tank emptying with
the daily breathing cycle.  This is accomplished by fill-
ing during a normal period of inbreathing and empty-
ing during a normal period of outbreaking.  The com-
pensating effects  of breathing will partially  cancel out
filling losses.  Inbreathing normally begins when the
tank starts to  cool in the afternoon; outbreaking nor-
mally begins when  the tank starts to warm up in the •
morning.
  Pumpings to fixed-roof tanks should be scheduled to
maintain  minimum average outage.  If  volatile stock
accumulates in a group of non-interconnected tanks,
only one of  which is a  conservation-type,  this  tank
might be used for  daily  accumulation;  enough stock
periodically would be transferred from it to  completely
fill  one  of the fixed-roof tanks.   Another  method to
minimize  average outage is to refill a tank  as soon as
possible after  emptying.  This procedure also tends to
reduce filling loss. Immediately after a tank is emptied
the vapor space is lean in hydrocarbon. Refilling within
a day expels vapor which is lean  in  hydrocarbon.  A
delay oE three days  may nullify most of the  advantage.
  Evaporation loss from a variable-vapor-space sys-
tem can be minimized by balancing fillings and empty-
ings. Filling into one tank should be scheduled when
vapor expansion is at a minimum or when another tank
is being emptied.
  Filling loss from pressure tanks  can be minimized by
controlling the fill rate in order to  avoid pressure build
up;  this  allows time  for condensation of  vapor  and
equilibrium is  maintained.  If  heat  of  condensation
cannot be dissipated as fast as  condensation occurs, a
rise  in internal temperature will result in a higher  in-
ternal pressure.

-------
                          SECTION 1.0
                         INTRODUCTION
The specific  objectives of the study were  to:
     1)   Assess  the  feasibility  of applying  vapor  control
          technology for benzene  transfer operations includinj
          tank:- cars, railcars, barges,  tankers,  stor-a-j-e • ta-nfcs-,
          and pipeline operations.
     2)   Determine the achievable  emission level and emission
          reduction for each vapor control alternative.
     3)   Determine any secondary  emissions  that would result
          from applying each vapor control alternative.
     U)   Quantify the capital  and  annualized costs  cf the
          control  alternatives.

Visits were made to the plants of  two  benzene producers  to gather
information on liquid benzene storage  and  transfer operations.  A
literature search was conducted  to obtain  data on  benzene
handling  and  storage,  as  well as to  investigate technological
aTter"na~ti'v'es  to control  emissions.  This  activity was  brief
because of the desire to evaluate  technolojies that could  readily
be applied to  industry. Equipment  manufacturers were consulted to
determine the  state-of-art of commercially  available equipment
and__ascertain  the  effectiveness, cost, and  operating history of
their treatment units.    Three technologies exhibited promise as
effective nethocs  to reduce benzene  emissions,  and were  selected
for further  study.  These were a  refri?eration and lean oil
absorption unit,  vacuum  regenerated carbon  adsorption, and
thermal incineration.

-------
Hypothetical models  were prepared to  represent  a  typical
current-day benzene producer, and two benzene consumers.  These
models serve as base cases  for  the  study.   Six control schemes
were developed and applied  to the base cases.   Four were applied
to the producer,  and two  to  the consumers.  Each of the three
control technologies discussed above were applied utilizing  their
respective achievable  emission levels  to the control schemes
resulting  in 16 case studies.   The cost effectiveness of each
case study was calculated,  and the technologies  rated.

-------
                          SECTION 2.0
                            SUMMARY
The three control  technologies evaluated were:

     1)   Condensation  of benzene vapors by  refrigeration
          followed by absorption of benzene vapors. .Ln-a-n -o-il
          absorbing/stripping systea.

     2)   Carbon adsorption beds regenerated  by vacuum.

     3)   Theraal  incineration using supplemental  fuel.

Other technologies  were considered,  but  dropped because  of lack
of design information and/or cocanercial  availability.

The control  technologies were  evaluated by applying then  in
various configrations to hypothetical models  which were  prepared
to represent  facilities and operations typical of  current-day
p-ro
-------
The claimed  reaoval efficiencies  of the  three technologies
studied are all  high.  The predicted  benzene emission concen-
tration levels that  are practical to achieve are:

          Refrigeration-absorption - 1000 ppm
          Carbon adsorption        -   10 ppm
          Thermal incineration     -   10 ppm

The technologies were evaluated using  the above emission levels.
The economic penality for installing  and  operating a thermal
incinerator at 10 rather than 1000 ppm  is small.  This is not the
case with carbon adsorption and a meaningful economic comparison
of this technology can only be made when it  and competing  tech-
nologies are  evaluated at the same emission concentration level.
Using  the above emission levels, refrigeration-absorption  has  a
cost effectiveness very close  to  that  of  thermal .incineration.
Averag.e  cost effectiveness of  the refrigeration-absorption
systems is $3.83/lb  reduction, while that of thermal incineration
is $3.78/lb reduction.  (Note:  Units  used in this report are the
same as used  by suppliers of raw data.   A metric conversion  chart
is contained  in Appendix A.) This is a  negligible difference.   A
slight rise in the value of benzene and/or  the cost of natural
gas relative  to electricity would make  refrigeration-absorption
the most cost effective.  Although there is  no single component
in_. the—system that  is unique;  i.e.,  closed loop refrigeration
vapor  scrubbing tower, gas-oil  separation by  distillation;  the
combination of these components into a  single package for remote
automatic  efficient operation  is  not  yet  demonstrated.   This
system is thought to need more control  and  fine tuning  than the
other  technologies to achieve efficient  operation.  A great deal
more operating experience would likely  be  required to make this
technology widely accepted.  What makes  refrigeration-absorption
particularly attractive is its  potential  to be the most cost
effective and its conservation of benzene.

-------
Thermal  incinerator technology has been used  aore  in the  control
of storage and  transfer  emissions than the other two  techno-
logies.  The transfer of sasoline handling knowledge  to  benzene
handlinj  is much sore direct  than that of the other technologies.
The state of the art for thernal incineration is at a high level,
and potential iaprovenents are possible with energy recovery  by
heat exchangers.  Advantage was-not taken for heat recovery  In
the case study models.  Also the  particular commercially
available thermal incinerators  investigated did  not  offer  heat
recovery  as a regular option.  If heat recovery is a possibility
for any  particular plant,  theraal incineration would be even more
cost effective.   Standard theraal  incineration  units are
available  as  "off  the shelf" items froa  at  least  two
manufacturers.

Vacuum  regenerated carbon  adsorption with 10 ppa emissions was
calculated to be the least cost  effective means  of controlling
benzene  emissions but at 1000 ppa  emissions aay be coapetitive
with other technologies.  On  a functional basis,  carbon  adsorp-
tion stands out as the aost attractive technology.  It has a very
high efficiency of  benzene  recovery and reaoval,  relatively
sinple  operation well suited  for automation, and wide turndown
ranges.  Experience with benzene  is presently  linited to extrapo-
lation_of results gained from gasoline service with gasoline con-
taining  benzene.  Substantial advanceaent in  the state of the art
is expected as aore experience is obtained.

Steaa regenerated  carbon systems have wide experience  in the
treatment and recovery of  solvents from solvent contaainated air
in extreaely dilute concentrations.  These  units are available
froa several manufacturers as standard package iteas.  However,
no experience was found pertaining to benzene, gasoline,  or  high
concentration hydrocarbon usage.   Ho  pricing  estimates for

-------
benzene applications  of  steam regenerated  systems were available.
Sone means for disposal of benzene  contaminated condensate  is
necessary for this  type  system.

Calculations revealed that there  is  considerably more benzene
lost as  a  result of loading and  storage (per unit of benzene
handled)  by producers than for consumers.  The emission factor
for the base case producer is 2.608  lb/10  gallons compared  to
.468 for the consumer case.  Floating  roof tanks represent a  high
level of control.   (Texas state regulations require floating  roof
tanks for the base  case.) Conversely if  a plant has cone  roof
tanks,  the  first  efforts should be  directed to reducing storage
losses  by conversion to either open  floating roof or internal
floating  cover depending on their  relative cost effectiveness.
Either  method is  highly cost effective.

When the implementation of carbon  adsorption technology  is
desired,  the most cost effective  design will incorporate  features
to reduce the capacity (in terms  of  benzene  loading and volu-
metric  flowrate)  of the individual  treatment units,  permit higher
ppm emissions,  and minimize the  number of  units required.  Capa-
city reducing features might include vapor holders to act as  flow
equalizers  and  displacenent of vapor  from  tank to tank or -carrier
to tank.  The  additional cost due  to  capacity reducing measures
will. b_e_jnore. .than offset by the  savings in capital  costs of the
carbon adsorption  units.   Capacity  reducing  measures do  not
provide  similar  cost effectiveness  gains  for refrigeration-
absorption  and  thermal incineration technologies.  The increased
cost of  the  capacity reduction measures outweighs the  cost
savings obtained  by reducing the  size and  number  of treatment
units.

-------
                          SECTION 3.0
                CONCLUSIONS  AND  RECOMMENDATIONS
Conclusions and recoraaendations are:

     1)  It is concluded  that  thermal incineration offers  the
        best aeans for  control of benzene  vapor to lev"! =_c.f.. 1_3
        ppo benzene.  The  risk  in  applying this technology  to
        benzene  service  is considered  to  be low.   Thermal
        incineration systeas  hava the distinct advantage  of
        being able to dispose of other pollutants.
    2}  Theraal  incineration at the level of 10 ppa  benzene
        aaission  and rsfr igeration-adsorpt ior. at 1000 pea are
        equal  in  cose effectiveness.
    3)  Carbon adsorption  is not as  cost  effective  as cheraal
        incineration when both are  compared at 10  pen.
    **)  The ccst  of carbon adsorption is sensitive  to  final
        benzene  eaission level and  a  true cost  comparison  to
        other technologies  can only be aade  when  all  tech-
      . .oologies are evaluated  at the  sane eaission level.
    5)   Benzene emission control efforts are more  cost effective
        in  producer racher than consumer  facilities.   Plants
        with cone  roof storage tanks  should receive  attention
        before those using  floating  roof tanks.   When  the
        producer plant  is  equipped with floating roof tanks, the
        priority shifts  to  controlling the  loading losses.
    6)   Modifications  to  carriers  to reduce transit lo'sses
        (defined  =s  breathing  losses during shipaer.t)  should
        receive the lowest  priority.  Modifications to carriers

-------
    should be limited  to those which are required  to reduce
    loading losses.
7)  Secondary emissions  for the control systems  evaluated
    were low,  and do not present a significant  problem.

8)  Air-benzene mixtures in pipe lines  to recovery  systems
    introduce significant  explosion hazards,  and  designs
    must incorporate equipment to avoid  this hazard.   (This
    was done  for designs evaluated in this report.)

-------
VII.  MOBILE SOURCE/CO



  (ORAL PRESENTATION)

-------
Oxidation of CO  in the Exhaust Gas

     Exhaust manifold air  injection, thermal reactors, and catalytic con-

verters all control  CO  emissions  by oxidizing CO  in  the exhaust to COp-

The gas temperature, oxygen concentration, catalyst parameters  and CO con-

centration are the important operating  variables.   Secondary air  injection

and temperature control are often required.  Two  kinds  of thermal reactors

have been developed for automotive (gasoline soark  ignition) engines:  the

Rich Thermal  Reactor  (RTR)  for   fuel  rich air/fuel ratios and  the Lean

Thermal Reactor (LTR) for lean ratios.   The thermal  reactor is a container

which,  by  its  size  and  configuration, increases  the  residence  time and

turbulence of  exhaust gases, thereby  providing  a  chamber for  the  high-

temperature oxidation reaction.   High  temperatures are maintained by the

exothermic oxidation  of  CO and HC  in  the exhaust  gas.  The  rich thermal

reactor operates at temperature from 870 to 1,040°C (1,600 to 1,900°F) and

is designed for fuel rich operation.  At rich air/fuel ratios of 11-12 to

1, NO  emissions are reduced to less than  6 g/kwhr  (4.5  g/hphr), but fuel
     A

consumption penalties are  incurred.  Secondary  air injection  is normally

injected into  the thermal reactor  for complete oxidation,  and construction

materials such as Inconel 601 are needed for the inner core,  bafflers and

port liners.   Temperature  control  devices  are  required to  protect  the

reactor construction materials against  overtemperature.
Reference:  Control Techniques for Carbon Monoxide Emissions, EPA-450/3-79-006
(June 1979).

-------
     The  lean  thermal  reactor operates  at  higher  air/fuel ratios  (17-19
to 1) and lower operating temperatures,  760-870°C (1,400  to 1,600°F), than
the rich thermal reactor.  Secondary air-injection  is  not usually required
and construction materials  have  less  severe durability requirements than
do the materials for rich thermal reactors.   Oxidation catalysts and 3-way
catalysts are being used extensively in  the control of CO  from automotive
engines.  This CO control strategy can  be equally effective in the control
of CO  from  stationary engine  sources.    Recent  literature  describes a
patented platinum catalyst on a ceramic  honey comb support  that has with-
stood 50,000 hours of  stationary engine  testing.   The catalytic converter
has also been used for small Diesel, LP  gas, and  gasoline engines in sizes
up to 13.1 litres (800 cu  in.) displacement  and is  applicable to 2- and 4-
cycle naturally  aspirated  or  turbocharged engines.  Applications include
Diesel  powered mining  and  tunneling  equipment, locomotives,  loaders, fork-
lift trucks operated  in  enclosed spaces, and electric generators located
near airconditioning intakes.  For oxidation catalysts to  be an effective
means of  controlliIng  CO and HC emissions, the engine  must  be properly
tuned and unleaded  fuel  must be used.   Also,  the control  system should
ideally be adjusted to preclude  the formation  of sulfate emissions which
can be  formed  in  the catalyst due to excess oxygen in the exhaust gases and
sulfur  content of the fuel.   Alternatively,  sulfur  can be removed from the
fuel.  In the case of 3-way catalysts,  rich mixtures are conducive to the
formation of HCN and ammonia.
     Air injection into the exhaust manifold can reduce CO emissions by a
factor  of 55 percent from baseline emissions on some engines with modifica-
tions to the air/fuel ratio, compression ratio,  and spark  ignition timing
schedule.

-------
Design Changes and Operating Adjustments



     The air/fuel ratio is the  operating  variable  that determines CO emis-



sions, and  it  has a  significant  effect on NO  emissions.   Operation at
                                              A


air/fuel ratios that produce low CO emissions can produce high or  low NO
                                                                        /\


emissions depending on the exact value of the air/fuel ratio used.  Since



NO  emissions  from  stationary  reciprocating  internal  combustion  engines



are considered more of a  problem  than CO emissions, design and operating



changes are expected to be made  in  these sources primarily for NO  control.
                                                               J\


Care  must  be  taken  to ensure  that  the  entire   emission  control  system



provides adequate control  of all  emissions  that  need  to  be controlled.



This sometimes leads to more sophisticated systems.

-------
VIII.  EMISSION CONTROL/ENVIRONMENTAL ANALYSIS

-------
                                  SECTION  1
                                  SUMMARY

       The Industrial Environmental Research Laboratory  is responsible  for
performing the research  and development required  to  assess the  impacts  of
pollution from a variety of industries and to evaluate the applicability
of various control technologies for these industries.  Pollution control
options must be evaluated for efficiency, reliability, economics,  and
energy consumption.  If secondary pollutants are  generated by the  cleanup
of the original pollutant, their  impact must also be assessed.  This
report on acrylonitrile plants addresses  these aspects of control
technology evaluation.
       The purpose of the report  is to provide data for making decisions
about control technology.  Control technologies are identified and ranked
in terms of efficiency,  cost,  and energy  requirements.  Control technology
demonstration opportunities in the acrylonitrile  industry are also
identified.
       There are six operating acrylonitrile plants in the U.S. Each has
several air pollutant emission sources.  The effluent streams addressed in
this report are:
       e   The absorber  vent gas stream
       •   The liquid waste streams that go to the holding ponds and
           deep-well  ponds

-------
       e   The HCN and acetonitrile incinerators and their off-gas streams
       0   The reactor startup emission streams
       The absorber vent gas stream,  when unabated, emits large quantities
of hydrocarbons.  Thermal incineration is used for abatement of this
stream at one acrylonitrile plant, and catalytic incineration is used at
another plant.  Data for these streams and their abatement by the
incineration processes were available from EPA contractors.  Using these
data, the effectiveness of catalytic and thermal incineration was
evaluated.  A quick review of the  literature showed other methods to be
unsuitable:  carbon adsorption because the pollutants are  too low in
molecular weight,  and hydrocarbon  absorption because the  stream.is too
dilute.   It was concluded that thermal incineration with  waste  heat :
recovery  is the best method for abatement of this  stream;  catalytic
incineration  has  a high  unburned-hydrocarbon passthrough  rate.
        High  levels of hydrocarbon  emission occur from the holding ponds.
There  are no  reasonable  pollution  control technologies  for open ponds,  but
there  are control  technologies for hydrocarbon removal  from waste water on
 its  way to the ponds.   A review of studies  and demonstration projects  on
solvent extraction of organic nitrogen containing  waste waters  was  made.
 In addition,  a patent for changing the acrylonitrile processing to
 eliminate water scrubbing of  the  product was  reviewed.   This would  also
eliminate most of the waste water production.   These methods are still in
 the  research and development  stage,  and conclusions about their efficacy
 cannot be drawn.
        All acrylonitrile plants  have HCN and acetonitrile thermal
 incinerators; the emissions data available (from other EPA contractors)
 for the exit streams  from these incinerators showed 0.6 percent conversion

                                     1-2

-------
of fuel nitrogen to NO .   A review of the combustion literature revealed
                      A
that 20 to 80 percent of conversion of fuel nitrogen to NO  could be
expected.   (This discrepancy should be resolved by further study.)
Catalytic incinerators were evaluated as replacements for the existing
incinerators; a literature review shows that similar levels of NO
production could be expected.
       When reactors at acrylonitrile plants are started up, the emissions
from these reactors are vented directly to the atmosphere.  To control
this intermittent pollution stream, which contains up to 10,000 Ibs of
acrylonitrile per reactor per emission, flares and carbon adsorption were
evaluated.  Flares (and other combustion methods) form unacceptable
amounts of NO .  Carbon adsorption, and wet scrubbing followed by carbon
             rt
adsorption, appear to be more effective.
       This report presents the following conclusions:
       •   Absorber vent stream:  Thermal incineration is an acceptable
           and efficient control method. .Thermal incinerators are
           currently in use, and no further development is required.
       e  . Holding pond:   Extraction of hydrocarbons from the waste water
           before it is sent to a holding pond is the most desirable
           control method..  Bench and pilot-plant scale research on carbon
           adsorption and hydrocarbon absorption (solvent extraction) is
           recommended.  A literature review of the waste water control
           methods in use in Europe (e.g., the Montecatini plant) is also
           recommended.
       e   Hydrogen cyam'de/acetonitrile incinerators:  Investigation of
           the NO  production of the existing  incinerators  is
           recommended.  Threre is a potential for high levels of N0x

                                    1-3

-------
    emissions from these incinerators.   A feasibility study of
    advanced incineration techniques — two-stage (low N0x)
    thermal and catalytic incinerators — is also recommended.
•   Startup emissions:  A study of the feasibility of routing
    startup emissions to the absorber tower for scrubbing  and a
    demonstration of a combined wet-scrubber and'carbon adsorption
    abatement technique are recommended.
                             . 1-4

-------