EPA-650/2-73-036
October 1973
Environmental  Protection Technology Series








                                                                         1;
                                                                         I
                                                                         m
                                                                        I

                                                                        i
                                                                        .v.v.

                                                                        m

-------
                                      EPA-650/2-73-036
                FEASIBILITY
   OF FLUX  FORCE/CONDENSATION
                 SCRUBBING
FOR FINE PARTICULATE  COLLECTION
                       by
            Seymour Calvert, Jhuda Coldshmid,
            David Lcith, and Nikhil C  Jhaveri

                   A P T , Inc
                    P  0 Box 71
               Riverside, California  92502
                Contract No  68-02-0256
               Program Element No.lABOI2
                  ROAP No  2IADL
           EPA Project Officer Leslie E  Sparks

               Control Systems Laboratory
           National Environmental Research Center
        Research Triangle Park, North Carolina 27711
                    Prepared for

          OFFICE OF RESEARCH AND DEVELOPMENT
         U. S. ENVIRONMENTAL PROTECTION AGENCY
               WASHINGTON, D.C.  20160
                    October 1973

-------
This report has been reviewed by the Environmental Protection Agency and




approved for publication.  Approval does not signify that the contents




necessarily reflect the views and policies of the Agency, nor does




mention of trade names or commercial products constitute endorsement




or recommendation for use.

-------
                              FOREWORD
              This report, "Feasibility Of Flux Force/Condensation
        Scrubbing For Fine Particle Collection", is the final
        report submitted to the Control Systems Laboratory for
        E.P.A. Contract No, 68-02-0256.

              The principal objective of this experimental and
        theoretical program was to determine the feasibility of
        flux force/condensation (FF/C) scrubbing.  The main activi-
        ties under the scope of work were:

              1.  An extension of the existing theoretical treat-
                  ment of FF/C scrubbing, starting with a review
                  and analysis of past work and proceeding to
                  engineering design methods suitable for eval-
                  uating practical systems.

              2.  An exploratory experimental study limited to
                  the evaluation of key features of FF/C
                  scrubbing.

              3.  Preliminary engineering and cost analyses of
                  promising practical cases.

              4.  Recommendations  for future development of
                  FF/C scrubbing systems.

              Dr. Leslie E. Sparks,  of the  Control Systems Labora-
        tory,  National Environmental Research Center, Environmental
        Protection Agency was the  Project Officer for this program.

              Dr. Seymour Calvert, of A.P.T., Inc., was the Project
        Director.
                                    11
A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
                                ABSTRACT
              This report presents the results of a feasibility
         study of flux force/condensation (FF/C)  scrubbing for
         fine particle control.   FF/C scrubbing includes the
         effects of diffusiophoresis, Stephan flow, thermophoresis,
         and particle growth due to the condensation of water vapor,
         and is not restricted to any specific scrubber configura-
         tion.  Fine particles are those smaller than 2.0 micrometer
         diameter.

              The program purpose was to determine whether a devel-
         opment program to maximize FF/C effects is warranted and,
         if so, to detail the recommended program.  The scope of
         analytical and experimental work was limited to the
         exploration of the most essential features.

              Experimental determination of sub-micron particle
         collection efficiency in a bench-scale,  sieve plate type
         FF/C scrubber validated the mathematical model which had
         been developed in the analytical phase of this program.
         Process designs and cost estimates for the application
         of FF/C scrubbing to two fine particle pollution sources
         demonstrated its technical and economic potential feasibil-
         ity.  It is concluded that FF/C scrubbing is an attractive
         control method for fine particles when high efficiency
         is required and when the gas is hot enough to evaporate
         the necessary water vapor.  Further development of FF/C
         scrubbing is clearly warranted.

              This report was submitted in fulfillment of Contract
         No. 68-02-0256 by A.P.T., Inc. under the sponsorship of
         the Environmental Protection Agency.  Work was completed
         as of February 17, 1973.
                                    111
A. P. T. InC.                                POST OFFICE BOX 71,  RIVERSIDE, CA. 92502

-------
                               CONTENTS

                                                            Page

         Abstract                                            iii
         List of Tables                                      v
         List of Figures                                     vi
         Acknowledgements                                    viii

         Sections
         Summary and Conclusions                              1
         Introduction                                         7
         Engineering Design                                  22
         Theoretical Background                              27
         Mathematical Model,  Sprays                          44
         Mathematical Model,  Plates                          59
         Mathematical Model,  Impinging Jet                   77
         Mathematical Model,  Liquid Sheets                   83
         Experimental                                        92
         Economic Feasibility                               118
         Future  Research Recommendation                     133
         References                                          138
         Glossary                                           144
                                     IV
A. P. T. IOC.                                 POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
                                TABLES


        No.

          1    Selected  References  On  FF/C  Applications            15

          2    Collection  Efficiency Of  Particles By Drops         54
              For  K   =1.1 And  Gr  = 0.1 At Various
                  P             F
              Boundary  Layer  Thickness

          3    Summary Of  Plate  Computations With Pene-            74
              tration Theory  Predicted  Coefficients

          4    Results Of  Computer  Predictions  Of                  88
              Collection  By Sheets

          5    Experimental Conditions And  Results                111

          6    Transfer  Coefficients Determined By                Il6
              Mickley's Method

          7    B.O.F.  Control  System Costs                         126
A. r. T. Inc.                                 POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
                                FIGURES

         Np_._                                                   Page

          1    Schematic Representation For FF/C Scrubbing        23
              On  A Sieve Plate.

          2    Critical  Saturation Ratio For Homogeneous          33
              Nucleation Of Water (After Amelin (1967))

          3    Critical  Saturation Ratio For Water Upon A         34
              Plane Substrate  Of Given Contact Angle -
              Water At  20°C (After Fletcher 1966) .

          4    Critical  Saturation Ratio For Nucleation Of        35
              Water Droplets Upon A Particle Of Given
              Radius With cosa At 20°C As Parameter -
              (After Fletcher  1966) .

          5    Equilibrium Supersaturation As A Function          35
              Of  Droplet Radius  With  (im) As A Parameter
              (After Howell 1949).

          6    Dimensionless Concentration Or Temperature         40
              Vs.  Dimensionless  Distance

          7    Saturation Ratio Vs.  Dimensionless Distance        43

          8    Efficiency Of Single  Drop Versus Inertia           51
              Parameter At NR  ,  = 9.6  With N™ As  Parameter


          9    Efficiency Of A  Single Drop, n,  Versus NFn         52
              As  A Parameter (NR ,  = 9.6)

         10    Collection Efficiency For 0.5 ym Diameter          57
              Particles In A 1 Meter Spray Column

         11    Computed  Prediction Run  #3 For Sieve  Plate         71

         12    Predictions Of Effect Of Water Vapor  Concen-        71
              tration On Particle Radius

         13    Computed  Predictions  Run #26                       72

         14    Computed  Predictions  Run HIS                       72

         15    Computed  Predictions  Run #20                       73
                                    vi


A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
                                FIGURES

        No.                                                     Page

        16   Computed  Prediction  For  Sheets,  Run *1               89

        17   Computed  Prediction  For  Sheets,  Run #2               89

        18   Computed  Prediction  For  Sheets,  Run #3               90

        19   Computed  Prediction  For  Sheets,  Run #4               90

        20   Experimental Apparatus                               93

        21   Operating Characteristics  Of The Sieve-Plate        95

        22   Particle  Generator Assembly                         97

        23   Sampling  System                                     100

        24   Particle  Penetration Vs. Water Vapor               113
             Condensed

        25   Theoretical Vs.  Experimental Penetration           115

        26   Flow  Scheme For  B.O.F. Gas Cleaning                123'

        27   Flowsheet For  FF/C Scrubber On Kraft Liquor        129
             Recovery  Furnace
                                    VJ.1
A. P. T. InC.                                 POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
                               ACKNOWLEDGEMENTS







                   A.P.T.,  Inc. wishes  to  express  its appreciation



              for excellent technical coordination and for very



              helpful assistance  in support  of  our technical effort



              to Dr. Leslie Sparks, E.P.A.,  Project  Officer, and



              Mr, Robert Lorentz, E.P.A.
                                     Vlll
A. r. I. Inc.                                 POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
                         SUMMARY AND CONCLUSIONS
             Flux force and water vapor condensation effects have
        the potential to cause high efficiency collection of fine
        particles by scrubbers.  These effects can result from the
        cooling of a hot, humid gas by contact with cold water.
        The temperature gradient causes thermophoresis, which
        drives the particles from the hot gas to the cold water
        and the water vapor concentration gradient causes diffusio-
        phoresis, which also drives the particles toward the water
        surface.  Condensation of water vapor on the particles will
        cause their mass to increase and the particles will then be
        easier to collect by inertial impaction.

             The object of the research reported here was to
        determine the feasibility of using flux force/condensation
        (FF/C) effects for fine particle (i.e., particles with
        diameter smaller than 2 microns) collection and, if it
        appears feasible, to specify the nature of a development
        program which would be required to provide a proper basis
        for designing practical systems.  The results of the study
        have established the feasibility of FF/C scrubbing and
        have indicated the crucial design questions.  At the time
        of writing, development is continuing in order to extend
        and strengthen the base of engineering design knowledge.

             At the beginning of this research it was known from
        the literature and previous experience that FF/C scrubbing
        "works".  The phenomena of thermophoresis, diffusiophoresis
        and condensation on particles had been observed and there
        were reasonably accurate quantitative theories to describe
        each of them when it occured in a simple system.  On an
        empirical level, there were reports of several scrubber
        systems which employed flux forces and/or condensation
        either as the result of deliberate planning or happenstance
        Missing, however, was an integration of theory of flux
        force deposition and particle growth by condensation into
        an overall quantitative model which would account simultan-
        eously for these phenomena.  The model would also have to
        include heat and mass transfer, particle collection by
        other mechanisms, and the balances for energy, water vapor,
        and particles.

             The first stage of the research was to put all of the
        pieces together for the prediction of particle collection
A. P. T. InC.                                POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
        in scrubbers of several representative types.  Literature
        search and a review of theory led to the selection of the
        elements which were incorporated into the overall mathe-
        matical models.  These models have been, and are still
        being, refined through a series of evolutionary steps and
        have reached the point for some scrubber types where they
        are quite comprehensive and appear to be capable of realis-
        tic and accurate representation of particle collection so
        far as the available experimental data show.

             Models for plate scrubbers (as represented by sieve
        plates) and packed scrubbers have been developed to the
        point where they appear to do a good job of showing the
        simultaneous effects of heat and mass transfer for gas/
        particles and gas/liquid, diffusiophoresis, thermophoresis,
        Brownian diffusion, inertial impaction, and changes in
        particle concentration, humidity, and temperature as the
        gas flows through the scrubber.

             The model for spray scrubbers required appreciable
        effort on the computation of particle trajectories for
        the combined effects of flux forces and inertial collection
        by liquid drops.  The influence of flux forces on collection
        by drops was found to be defined by a new dimensionless
        parameter which is the ratio of flux force deposition
        velocity to the gas velocity and which we call the "flux
        deposition number".  If the flux deposition number is
        greater than 0.1, particle collection efficiency will be
        good and if it is less than 0.01, efficiency will be poor.
        The model for collection from an impinging jet showed that
        negligible benefit is given by flux force deposition because
        of the very short contact time.

             Theoretical analysis showed that, in general, the
        particle deposition velocity has to be on the order of
        0.1 cm/sec or larger for appreciable collection efficiency
        to occur.  Diffusiophoresis can produce deposition veloci-
        ties this high under the heat and mass transfer conditions
        of a realistic scrubber, but thermophoresis generally
        cannot.  If particle growth due to condensation occurs,
        inertial impaction of the resulting two or three micron
        diameter particles can give sufficiently high deposition
        velocities.  Particle growth has the advantage that the
        high deposition velocity can persist after the heat and
        mass fluxes are dissipated, because once the particles are
        enlarged, they can be collected to an increasing degree at
A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
        the  cost  of  relatively  little  additional  pressure  drop.

              Experiments  were  designed  to  explore  the  regions  of
         FF/C  scrubbing  where the  mathematical  model  was most depend-
         ent on the  several  relationships,  coefficients, and assump-
         tions which had been used.  A test aerosol of  condensed
         dibutyl  phthalate smoke particles  0.7  ym diameter was
         scrubbed from humid air by  a  10 cm diameter  sieve plate.
         Particle size and concentration, gas and liquid flow rates,
         temperatures, and other significant parameters were meas-
         ured.

              Comparison of  the experimental data with  the model
         required that first the heat  and mass  transfer coefficients
         used  in  the model be computed from experimental measure-
         ments of heat and mass transfer.   Predictions  of  particle
         collection  efficiency  based on  the experimental transfer
         coefficients were then compared with the experimental
         efficiencies.  The  predictions  were sufficiently  close  to
         the experimental  data  to  validate  the  mathematical model
         as  a  basis  for  exploration  of FF/C scrubbing potentiality.
         A number of details remain  to be resolved  by further effort
         despite  the fact  that  the model will give  a  realistic  pre-
         diction  of  flux force  effects for  the  experimental system.

         Economic Evaluation

              The exploration of the economics  of FF/C  scrubbing was
         started  as  soon as  an  apparently realistic mathematical
         model was available.   Two candidate processes  (Basic oxygen
         furnace  and a Kraft liquor  recovery furnace) were selected
         for study on the  basis of their meeting  the  criterion that
         the gas  should  ideally be hot enough to  evaporate the
         necessary water.

              This criterion develops  from  the  fact that  steam pur-
         chased at a cost  of $1.32 or  more  per  1,000  Kg steam ($0.60
         or -more  per thousand pounds)  is expensive  and  its use  must
         compete  with high-energy  scrubber  (or  electrostatic precipi
         tators or fabric  filters) on  the basis of  operating costs.
         For illustration, if 0.15 Kg  steam were  used per  Kg of dry
         gas (this is a  fairly  low ratio for FF/C scrubbing), the
         steam cost  would  be about $0.24/1,000  m3 dry gas
         ($0.20/1,000 Kg dry gas).  Cooling water for this case
         might cost  from about  $0.01/1,000  m3 to  $0.12/1,000 m3  dry
         gas,  with $0.03/1,000  m3  being  a probable  minimum for
         cooling  tower water.
A. P. T. InC.                                POST OFFICE BOX 71,  RIVERSIDE, CA. 92502

-------
             The FF/C steam and water costs may be compared with
        the power cost for a high-energy scrubber which is capable
        of 80% collection efficiency on 0.5 jjm diameter particles
        of density 1.0 g/cm3.  A venturi scrubber would require a
        pressure drop of about 450 cm W.C., which would mean a
        power cost of $0.24/1,000 m3 of gas if electricity cost
        l.Ot/K.W.H.  One can see that at this performance level an
        FF/C system could purchase steam and be nearly competitive,
        and that it would become more attractive as efficiency re-
        quirement becomes more stringent.   Generally, it would be
        advantageous to FF/C scrubbing if  the gas were hot or
        moderately hot and humid so that steam need not be purch-
        ased.

             A preliminary design and cost estimation for the use
        of a 99.9% efficient FF/C spray scrubber on a basic oxygen
        furnace of 250 ton/heat capacity resulted in a capital
        investment of about $1,700,000 and annual operating costs
        of about $790,000.  Cost given by  others for electrostatic
        precipitator, fabric filter, and high-energy wet scrubber
        systems ranged from about $3,200,000 to $6,000,000 for
        capital investment and from $1,200,000 to $2,400,000 for
        annual costs.  Since the gas is very hot and must be cooled
        (usually by water sprays) before it can be cleaned by any
        means, this situation js very favorable to FF/C scrubbing.
        This design should be revised in the light of the additional
        understanding of FF/C scrubbers which has been acquired
        since the time when the estimate was made.

             For another example of the possibilities for FF/C
        scrubbing, a system for up-grading an existing Kraft pulping
        liquor recovery furnace air pollution control installation
        was compared to two other alternatives.  In each of the
        three revision alternatives the original 80% efficient
        venturi scrubber was replaced with a low efficiency venturi
        evaporator and a high efficiency particulate collector was
        added as a second stage.

             If one added an electrostatic precipitator such that
        overall efficiency would be 99% the capital investment would
        be $1,930,000 and the annual cost  $420,000.  A venturi in
        series with the first such that overall efficiency would be
        97% would require a $710,000 investment and an annual cost
        of $32,000.  An FF/C system consisting of a condenser
        (tubular heat exchanger) and a spray scrubber capable of an
A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
        overall 971 efficiency would require a $610,000 capital
        investment and an annual cost of $24,000.

             At the level of precision of a preliminary estimate,
        there is no appreciable difference between adding a high
        efficiency venturi and adding an FF/C system.  The FF/C
        system has the great advantage that if higher efficiency
        than 97% is required, as would be likely, it can be up-
        graded for relatively low cost.  For example, the overall
        efficiency would be increased to 99% for an additional
        investment of $150,000 and an annual cost of about $40,000.
        This is possible because once the particle size has been
        increased by condensation, additional collection efficiency
        is much easier to obtain than it would be for the original
        particle size.

        Conclusions

             The results of the preliminary study show that FF/C
        scrubbing is capable of high particle collection efficiency
        on fine particles and that performance is "accountable"
        through the use of mathematical models.  Experimental data
        served the purpose of verifying the general content and form
        of mathematical models and, while the efficiencies measured
        for one-plate scrubbers are not very high, the predictions
        for multi-plate scrubbers can now be viewed with more
        confidence.

             Of the several particle collection mechanisms involved
        in FF/C scrubbers, diffusiophoresis and inertial impaction
        enhancement by particle growth are the two most important.
        Preliminary indications are that particle growth is the more
        important of the two.

             Economic considerations define the most favorable area
        of application for FF/C scrubbing as those situations in
        which the enthalpy of vaporization is available from the gas
        to be cleaned, although for high efficiency collection of
        fine particles the purchase of steam can be justified.  If
        an existing scrubber need be up-graded for the removal of
        sub-micron fume it could pay to introduce some steam, even
        if it had to be purchased.  Another criterion is that the
        smaller the particles, the better FF/C scrubbing will look,
        because its efficiency is relatively unaffected by particle
        size.  Cooling water costs can be significant although, de-
        pending on the system, charges as high as 4tf/m3 (15(/:/l,000
        gallons) can be accomodated.
A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
             The most  suitable  types  of  scrubber  for  FF/C  applica-
         tion appears to be multi-stage or  continuous  contact  type.
         This kind  of apparatus  can be readily  adapted to provide
         different  conditions and geometry  on different  plates  to
         accomodate  changing flow rates and particle concentrations.
         From the pressure drop  standpoint, there  should be  not  much
         difference  between this and other  scrubber types for  the
         same performance.

         Development Required

             The research which has been performed up to this point
         has established the feasibility  of FF/C scrubbing  and  our
         ability to  describe the process with engineering design
         equations.  It has'also shown that the specific details of
         heat and mass  transfer, the nucleation of condensation, and
         other matters  have very significant influence on FF/C
         mechanisms.

             In order  to resolve the  areas of uncertainty  mentioned
         above and  to further refine the engineering design  methods,
         a pilot plant  development effort is necessary.  The pilot
         plant scale is important because of the necessity  for having
         realistic  heat and mass transfer conditions.   Some  addition-
         al bench-scale work is  also needed to provide a means  for
         rapid scanning of a variety of conditions.  The importance
         of particle properties  and concentration  also dictates  that
         the effects of their variation be  studied.

             Next,  there must be a pilot plant demonstration  test
         under the  actual conditions of the specific application in
         order to ensure the performance and economics of FF/C
         scrubbing.  Ultimately, we will have to identify the
         specific characteristics which are of most significance
         and develop test methods for measuring them so  that perform-
         ance can be predicted and FF/C scrubbers designed without
         pilot plant testing.
A. P. T. InC                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
                             INTRODUCTION
               Preliminary theoretical  predictions  which were
          carried out as  a part  of the  "Wet Scrubber System
          Study" and reported by A.P.T. in the "Scrubber Handbook"
          (1972), indicated that flux forces (i.e., forces due to
          temperature and composition gradients)  and particle
          growth due to moisture condensation offered much promise
          as  a means for  separating fine particles  from gases.  In
          order to follow-up and evaluate that lead, Contract No.
          68-02-0256 was  established to determine the feasibility
          of  flux force/condensation (FF/C)  scrubbers.   This report
          is  the final report for that  feasibility  study.

               The major  efforts of the study were  to make experi-
          mental and theoretical explorations of  a  key region in
          which flux forces would act to separate particles.  It
          was anticipated that the experimental results would enable
          the validation  of mathematical models which could be used
          with some confidence to predict the technical and economic
          potential of FF/C scrubbing.   The object  was  to determine
          whether a development  program to maximize such forces is
          warranted and if so, to detail the recommended program.

               FF/C scrubbers were shown in the system study to have
          the potential for filling a gap in present day scrubber
          technology.  Whereas scrubbers as commonly used will collect
          fine particles  poorly  unless  high energy  is expended in
          accelerating the gas,  this is not the case for FF/C scrubbers.
          By  applying forces other than inertia on  the particles and/or
          by  causing the  particles to grow so that  inertial deposition
          is  more effective, the FF/C scrubber can  give performance
          on  fine particles which is nearly independent of the size
          of  the particles entering the scrubber.

               FF/C scrubbing involves  those forces which result from
          temperature, composition, or  electric fields  in the gas
          phase.  These include  thermophoretic, diffusiophoretic,
          photophoretic and electrophoretic forces  and  the Stefan
          flow.  The present study is limited only  to thermophoresis
          and diffusiophoresis (which we define to  include both diffu-
          siophoretic and Stefan flow forces).  Accordingly, we consider
          only those FF/C scrubbers where particle  removal from the gas
          is  aided by a temperature gradient, a vapor concentration
          gradient, vapor condensation, or combinations of the three.
          It  is not necessary that all  three forces act simultaneously,
          but in most actual cases they do.
A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
               For the case of air and water contacting it can
          readily be shown that the force resulting from the mass
          transfer of water vapor is the dominant force and a FF/C
          scrubber is distinguished by vapor condensing inside the
          scrubber.

               Flux forces may stem from the difference in momentum
          imparted to the particle on opposite sides by the molecules
          colliding with it or desorbing from it.  In the case of a
          temperature gradient, hotter (and thus faster) molecules
          colliding with the particle will impart a higher momentum
          to the particle than the cooler (slower)  molecules.  In a
          concentration gradient, which is accompanied by diffusion
          but not necessarily by net motion of the  gas phase, the
          heavier molecules will again impart a higher momentum than
          the lighter molecules.

               Another kind of flux force is due to Stefan flow
          which occurs when there is a net motion of the gas phase.
          The mass transfer can be brought about by evaporation,
          condensation or a chemical reaction.  It  was first suggested
          by Stefan in 1881 that near the surface of an evaporating
          or condensing body there must exist a hydrodynamic flow of
          the medium (directed away from the evaporating and towards
          the condensing surface).

               It should be noted here that we refer to the combina-
          tion of forces due to Stefan flow and to  the differential
          bombardment caused by a gas composition,  or molecular
          weight gradient, as diffusiophoretic force.

               In case of condensation of vapor molecules which are
          heavier than air, then both forces act in the same direc-
          tion; toward the condensing surface.  If  the condensing
          molecules are lighter than air (as in the case with water
          vapor), the Stefan flow force acts towards the condensing
          surface while the molecular weight gradient force acts
          away from it.  The net force for the condensation of water
          from air or flue gas is toward the cold surface because of
          the overwhelming effect of Stefan flow.

               Flux force effects on particles have been known for
          many years and the background is reviewed and discussed in
          depth by authors such as Waldman and Schmitt (1966) ,
          Goldsmith and May (1966), and Hidy and Brock (1970).  The
          studies reported by these authors include both theoretical
          and experimental work.  The experimental  systems were de-
          signed so as to be readily definable and  were much simpler


                                      8
A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
        than a scrubber in terms of the number of phenomena and
        unsteady conditions involved.  For example, we find
        experimental studies of particle deposition from gas
        flowing between flat parallel surfaces under the influence
        of a constant heat or mass flux between the surfaces.
        Details of pertinent previous studies will be discussed
        later in this report.

        PRACTICAL APPLICATIONS

            Over a period of many years there have been numerous
        instances of the use of scrubbers in which FF/C effects
        occured.  In most of the early pertinent literature, such
        as patents issued around the beginning of this century,
        there was no deliberate or conscious design to use FF/C
        but we can reasonably surmise from the descriptions given
        that FF/C effects were active.  For example, U.S. patent
        no. 1,039,008 (1912) employed sprays to clean blast furn-
        ace top gas and we would expect that the hot gas first
        became saturated with water vapor and subsequently was
        de-humidified by further contact with spray water.  Con-
        sequently, particle collection could have been enhanced
        by FF/C effects.                        ^

            More recent literature reveals a growing awareness
        of the potentialities of FF/C scrubbing; often as a means
        of explaining anomalously high particle collection
        efficiency.  Schauer (1951) reported high removal of OOP
        smoke when steam was introduced into a venturi scrubber.
        Lapple and Kamack (1955) reported that the addition of
        steam reduced dust loss at a given air pressure drop.
        Semrau et al. (1958) reported that the observed perform-
        ance differences between particle collection by a pipeline
        scrubber and a venturi-cyclonic spray scrubber were
        probably due to differences in the scrubbing liquid temp-
        erature .

            Description of other cases where particle collection
        was enhanced by steam addition can be found in articles
        by Demshin et al. (1965), Litvinov (1964) and many others.
        However, most of these authors observed the phenomena but
        did not study them in detail.  Increased collection efficien-
        cy due to steam addition was usually attributed to two groups
        of phenomena:  1)  Flux forces.  2)  Particle growth.  In
        most studies no attempt was made to isolate the various phen-
        omena and determine relationships between operating condi-
        tions and particle collection efficiency that will permit the
        engineering design of a FF/C scrubber.
A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
               More  detailed  studies  of the  deliberate use of
          FF/C effects  can  be found in  engineering articles after
          about 1965 and in the  patent  literature.  These studies
          generally  fall into one  of  four  categories:

                   1.   Studies of  FF/C  scrubbers,  per  se.

                   2.   Studies in  other fields  such as meteorology
                       and  atomic  reactor  accidents where particle
                       growth and  phoretic forces  were analyzed.

                   3.   Attempts  at analyzing the  FF/C  scrubber.

                   4.   Patent literature.

          Studies  of Flux Force/Condensation Scrubbers

               Rozen and Kostin  (1967)  studied  the collection of
          fine oil mist (average particle  diameter of  0.3 ym) in  a
          perforated plate  column. They built  their column in such
          a way that the gas  passed first  through  a plate with warm
          water where  it became  saturated  with  water vapor and then
          through  a  cold water plate  where condensation took place.
          Four such  plate pairs  were  in their column and collection
          efficiency was studied as a function  of temperature of
          the  plates.

               Collection efficiency  increased  with the temperature
          difference between  the hot  and cold plates.   The further
          downstream the plates  were, the  higher  was the collection
          efficiency.   Thus,  for example,  for a 50°C temperature
          difference collection  efficiency for  the first pair of
          hot  and  cold plates was  45%,  for the  second  pair it rose
          to 57.41,  for the third  pair  of  plates  it reached 74%,  and
          in the fourth pair  collection efficiency came to 90%.   They
          found that their  results could be  represented by an empiri-
          cal  equation relating  the particle penetration, "Pt", with
          the  mass of steam condensed per  gram  of  inlet particles,
          "q"
                                        -0-56
                             Pt  =  12.5  q                         (1)
               While  these  results  are  encouraging,  it is difficult
          if not impossible to extrapolate  the  Rozen and Kostin data
          to other installations  mainly because of the lack of a
          theoretical analysis.   Rozen  and  Kostin presented only their
          experimental results and   attributed  the high efficiencies
          to particle growth.   Our  theoretical  analysis of their data
          showed that their experimental results could be due to flux

                                     10


A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
          forces  as well  as  growth.  There  is  also  some  uncertainty
          about  their  results  at  high  temperatures  because  of the
          possibility  of  particle loss  by evaporation  of the  oil.

              Litvinov  (1964  A), (1964 B) ,  (1965),  (1967)  and
          (1972)  did extensive experimental  laboratory and  industrial
          scrubber work on the effect  of condensation  on particle
          collection efficiency in venturi  scrubbers followed by a
          cyclone or a tray  column for  mist  elimination.   In  his
          1967 paper he reported  on  the collection  of  fine  carbon
          black  dust (=1  ym  in diameter) and apatite dust in  a
          venturi scrubber followed  by  a two stage  tray  column.   In
          this paper he presented engineering  data  which permit  the
          reproduction of his  work.  However,  his analysis  of the
          results is very limited and  somewhat peculiar.   He  specu-
          lates  that neither particle  growth nor diffusiophoresis
          should  have  appreciable effect on  the collection  efficiency
          and he  suggests that "condensation produces  favorable
          conditions for  coagulation",  without even  describing a
          mechanism for this.

              In his  1972 paper  Litvinov experimentally studied
          three  industrial size combinations of dust removal  equip-
          ment.   In all three  combinations  the first unit is  a
          precleaner for  which he used  cyclones or  a combination
          cyclone and  electrostatic  precipitator.   Next  came  a
          venturi followed by  a one  or  two  stage tray  column. The
          dust used had a mean diameter of  =0.8 ym.  Here again  no
          quantitative prediction of collection efficiency  is given,
          only a  description of Litvinov's  experiments.   This time
          the increase in collection efficiency is  attributed to a
          change  in the particle  surface properties  caused  by water
          condensation.

              Terebenin  and Bykov (1972) describe  an  experiment where
          an aerosol of unrefined tin  with  an  arithmetic mean diameter
          of 0.128 ym  and saturated  with steam was  passed through  a
          series  of wetted wall rectangular  ducts.   They analyzed  their
          scrubber for flux  forces and  increased sedimentation due to
          particle growth.   Their conclusion is that the aerosol part-
          icles  grew through condensation and  then  settled, largely
          due to  the action  of flux  forces.  Here again  no  attempt was
          made at generalizing the equation  into a  design method.

              A study of steam injection  into a laboratory scale  wet
          scrubber was carried out by  Lancaster and Strauss (1971) .

                                     11
A. P. T. InC.                                POST OFFICE BOX 71,  RIVERSIDE. CA 92502

-------
          They used  redispersed  dry  agglomerates  of  ZnO  whose  size
          is  defined only  as  having  a  mean  diameter  of 1 ym at low
          concentrations and  a larger  diameter  at higher concentra-
          tions.   The aerosol was  injected  into a 2" x 4" duct 6'
          long to  which  first steam  and  then  water were  injected.
          The water  drops  were then  separated by  a cyclone.  They
          measured an increase in  particle  collection efficiency
          which was  in direct proportion to the amount of steam
          injected rather  than the amount condensed  and  they conclude
          that the increase in collection efficiency is  due to parti-
          cle growth.  The use of  hot  water (same temperature  as the
          temperature of the  saturated air) rather than  cold water
          in  the sprays  gave  a slightly  higher  efficiency.   This
          indicates  that diffusio  -  and  thermophoresis were ineffect-
          ive in the spray section.

               The peculiar dependence of collection efficiency on
          the amount of  steam introduced rather than that condensed
          is  attributed  by the authors to the mechanism  of  steam
          condensation by  self-nucleation on  dust particles when
          the steam  is quenched  more by  colder  gas.   The possibility
          of  additional  collection in  the cyclone separator due to
          FF/C effects is  not discussed, nor  is sufficient  information
          given so that  it can be  checked.  One must conclude  that the
          results  of this  experiment are valid  only  for  direct steam
          injection  with the  specific  geometry  and operating condi-
          tions studied.   No  general design methods  are  presented  in
          this paper.

          Studies  In Other Fields

               The studies grouped under this category are  more basic
          and more systematic than those in the first group.   Very
          thorough analytical studies  were  carried out by Horst (1968)f
          Hales et al. (1970), Hales et  al. (1971),  Hales and  Schwen-
          diman (1971).  These studies were concerned with  a reactor
          accident and explored  the  possibilities for aerosol  removal
          in  an enclosed environment at  a condensing steam  boundary
          layer.   Thus,  their analyses were confined to  liquid sheets
          with only  natural convection as the air moving mechanism,  a
          situation  which  is  very  unlikely  in a scrubber.  These stud-
          ies are  further  limited  by the lack of  experimental  confirm-
          ation .

               Goldsmith and May (1966)  reported  another very  important
          study.   They analyzed  the  removal of  a  radioactively tagged
          nickel-chromium  aerosol  0.02 to 0.2 urn  in  diameter.   The

                                     12
A. P. T. Inc.                                POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
          aerosol  moved  through  a  vapor  gradient  box  and  its
          deposition  velocity  was  measured.   They proved  experi-
          mentally the validity  of the Waldman  equation for diffus-
          iophoresis.  In  another  experiment  they showed  that  when
          the  vapor pressure gradient was  not closely controlled
          and  supersaturation  was  permitted,  collection was due  to
          particle growth  as well  as flux  forces.


          Attempts At Analyzing  The Flux Force/Condensation Scrubber

              Only a few  articles were  found in  the  literature where
          the  feasibility  of a flux force  or  a  FF/C scrubber was
          analyzed.   These articles were also the ones that came
          closest  to  presenting  design methods  and design equations.

              Sparks and  Pilat  (1970) presented  an analysis of a
          spray  column with condensation taking place on  the drops.
          They showed that for example,  when  K  =0.6, where "K " is

          the  Stokes  number, the collection efficiency in a spray
          column,  where  the vapor  pressure gradient is 10s  mbar/cm,
          was  1001 as compared to  an efficiency of 0% with  no  vapor
          gradient present.  While their equations can serve as a
          starting point for the design  of a  spray column they defin-
          itely  need  refinement  and experimental  verification.

              Davis  and Truitt  [1972) present  a  rough economic
          analysis of a  FF/C scrubber.   However,  a review of their
          work revealed  sufficient errors  in  their theoretical
          analysis and experimental techniques  that not much can  be
          obtained from  their  work to facilitate  evaluation and design.


          Patent Literature

              The patent  literature is  more  obscure  than any  of  the
          other  groups.  It does not describe the experimental work
          on which the patent  is based and in some cases  presents an
          idea that was  never  tried on a large  scale. Patents usually
          contain  much irrelevant  information and their descriptive
          part usually centers on  the equipment rather than on the
          design method.   Some examples  of the  patents available  which
          describe FF/C  scrubbers  are listed  here. Japanese patent
          No.  Sho  41-41184, Kazuo  Matsuzaki - inventor, describes a
          FF/C venturi scrubber  where the  steam is injected tangentially
          upstream of the  throat.   No relationships are given  which
          permit the  design of such a scrubber.

                                     13
A. P. T. InC.                                POST OFFICE BOX 71,  RIVERSIDE, CA. 92502

-------
               An article by Takashi Mashita (1971) describes a
          system based on the Solivore scrubber.  This system is
          based on several venturi's in parallel with water sprays
          saturating the air upstream of the venturi.  Examples of
          actual installations are given which permit a rough
          estimation by extrapolation of the scrubber performance
          from the operating conditions.  Other variants of the
          Solivore scrubber concept involve a single stage of satura-
          tion plus venturi and four stages in series, as described
          by Strauss (1966) .

               Nucleation and growth by condensation are claimed to
          be effective in the crossflow scrubber patented by Teller
          where saturated gas enters a packed section and is cooled
          with water.   This device has been successfully applied to
          phosphate fertilizer plant particulate collection.

          Background Summation

               The studies discussed above are examples of the avail-
          able background information on FF/C scrubbing.  Table I,
          Selected References on FF/C Applications, summarizes the
          references and their high points is meant to be illustra-
          tive of significant studies rather than a complete listing
          of the literature.  What can be distilled from a survey of
          the background information is a concept of where things
          stood at the onset of this feasibility study.

               To see  this status of knowledge with clarity it is
          helpful to consider first the question of what has to be
          known in order to design FF/C scrubbers so that their per-
          formance and costs can be predicted.  In the next section,
          therefore, we will turn briefly to discussing what we need
          for design.
                                     14
A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
H
HH

P
      AUTHORS          YEAR

      Semrau K.,       1958
      Marynowski C. ,
      Lande K.,
      Lapple C.

      Lapple C. ,       1955
      Kamack H.J.

      Schauer P.J.     1951
      Rozen A.M.,      1967
      Kostin V.M.
o
m
03
O
X
m
3
(a
O
m
O
TABLE I - Selected References on FF/C Applications

  HIGH POINTS OF STUDY                     REFERENCE
  Attributed increase in collection
  efficiency to scrubbing liquid
  temperature.


  Addition of steam reduced dust
  loss at a given air pressure drop.

  Addition of steam into a venturi
  scrubber increase removal of OOP
  smoke.

  Studied experimentally the collec-
  tion of fine  oil mist in a plate
  column.  They found that:
  1. Collection efficiency increased
     with the quantity of steam con-
     densed and proposed the empiri-
     cal  correlation
     Pt = 12.5  q -°-56 »Pt" is
     penetration and "q" = g steam
     condensed to g inlet particles.
  2. Collection in each pair of plates
     is higher than in the preceding
     pair.  (They attributed the in-
     crease in collection efficiency
     to particle growth.)
Ind. Eng. Chem.
50, 1615 C1958
Chem. Eng. Prog.
Sl_, 110, (1955)

Ind. Eng. Chem.
43, 1532, (1951)
Intern. Chem. Eng.,
7, 464 (1967)
CO
ro
01
o
10

-------
      AUTHORS
YEAR
         HIGH POINTS OF STUDY
REFERENCE
      Litvinov A.T,
s
CO
o
m
oo
O
X
m
33
CO
1964A    His studies  concentrated  on  venturi
1964B    scrubbers followed by  cyclone  or
1965     sieve tray for entrainment  separa-
1967     tion. In  his 1967 and  1972  papers
1972     he reaches the following  conclusions:
         1. Phoretic  forces are not  important
            for particles  larger than 0.1 uro
            in diameter.
         2. Condensation  of water  vapor in  a
            venturi and a  sieve tray column
            increases particle  removal  ef-
            ficiency.
         3. He gives  a design equation  for  the
            venturi based  on an experimental
            optimum liquid film thickness,
            which  permits  calculation of the
            "optimum" quantity  of  condensed
            steam.
         4. He gives  the  equation  NNu=0.78  NJ^65

            for the heat  transfer  coefficient from
            the gas to the liquid  drops.
         5. Particle  growth does not affect col-
            lection efficiency.
         6. Tray columns  with condensation  are
            more efficient than venturi with con-
            densation in  energy consumption.
                                                  Khim. prom.
                                                  8_, (1964)

                                                  Vestn. tekhn ,
                                                  i ekonom.  inform
                                                  5, (1964)
                                                                         Stal1,  7,  (1965)
                                                                          Zhurn,  Priklad,
                                                                          Khim,  40,  353
                                                                          (1967) —
«D
10
en
o
to

-------
ff
P
      AUTHORS

      Terebenin A.N
      Bykov, A.P.
YEAR

1972
      Lancaster B.W.
      Strauss W.
1971
o
m
oo
3
-»i
m
3J
en
O
m
O
CO
ro
S
10
HIGH POINTS OF STUDY

Analyzed the collection of particles
0.128 v>ro in diameter in cluster of
wetted wall rectangular ducts. They
concluded that in the presence of
steam particle removal is attributed
to growth and flux forces. They give
reference to particle growth equation,
the results of which do not agree with
their experimental findings of particle
diameter. No attempt to calculate pene-
tration was made.
     REFERENCE

     Zh.  prikl.  khim.
     45,  1012,  (1972)
Studied the collection of ZnO agglom-
erates with a mean diameter of 1 ym
in a 5 cm x 10 cm rectangular duct
1.85 m long. In different experiments
steam was injected upstream or down-
stream of the aerosol injection port,
and it was assumed that when steam was
injected upstream of the aerosol it
condensed on the aerosol particles. They
concluded:
1. Particle build-up was the major mech-
   anism responsible for improved scrubber
   performance.
2. Flux forces were not important.
3. For their system the dust penetration
   could be correlated with the rate of
   steam injection by:
            7"n  -1  where n and i
     Ind.  Eng.  Chem.
     Fund.,  10, 362
     (1971)  —
            Q = 0.2
are

-------
BT
n
AUTHORS
      Lohs W,
   00
s
tn
O
m
CO
O
X
-Nl
3J

m
31
CO
YEAR     HIGH POINTS OF STUDY

            the scrubber collection effic-
            iency with and without steam
            addition and "Q" is the steam
            injection rate Ib steam/lb air.
         4. Steam was used inefficiently in
            this particular scrubber.

1969     Fine particle removal efficiency was
         studied in a spray column. Na-SO, and
         polystyrene aerosol with median part-
         icle diameter varying from 0.43 to
         0.8 pm and from 0.4 to 1.3 ym respect-
         ively. Collection efficiency was im-
         proved by steam addition for both
         aerosols. The increase in collection
         efficiency was higher for the soluble
         Na2SO. aerosol. (0.3 - 0.5 vim particles
         can be removed at 60% efficiency)  The
         separation of hydrophobic fine dust is
         also increased particularly if the part-
         icle surface is rendered hydrophilic by
         means of a wetting agent.
         Increased efficiency is attributed to
         particle growth only though  Stefan
         flow is mentioned.  No equations,
         correlations or design method  are
         attempted.
REFERENCE
                                                                   Staub  29,  43,
                                                                   (1969)
to
ro
en

-------
 a
 o
AUTHORS          YEAR

Horst T.W.        1968

Hales J.M.,       1970
Horst.T.W.,
Schwendiman
      Hales J.M.,      1971
      Schwendiman L.C.
      Horst T.W.
CO

O
m
DO
CO
HIGH POINTS OF STUDY                     REFERENCE

This series of studies carried out
at Battelle Northwest was concerned
with the transport and deposition of
a radioactive aerosol expected to be
generated by fuel overheating follow-
ing an accident.   Their solutions to
the case of aerosol deposition through
laminar naturally-convected boundry
layer are more rigorous than for the
parallel turbulent case. Their con-
clusions were:
1. In a laminar boundry layer the
   dominant mechanism is diffusio-
   phoresis. In the turbulent case
   turbulent deposition may rival
   diffusiophoresis.
2. Relationships  can be derived be-
   tween steam consumption and part-
   icle deposition,
3. The rate of deposition of particles
   0.5 - 2 ym in  diameter in a laminar
   boundry layer is independent on part-
   icle diameter and depends only on the
   operating conditions within the air
   steam boundry layer.
4. When particle  deposition is assumed
   at the mass average velocity of the
   fluid, values  20-60% lower than in
   the case of flux forces are calculated
   for a laminar boundry layer.
                                                                         Battelle Northwest
                                                                         reports No. BNWL
                                                                         848 (1968) , BNWL
                                                                         1125 (1970) BNWL
                                                                         SA-3592 (1971)
                                                                         BNBL-SA-3734  (1971)
CO
M
O1
O
to

-------
D
P
AUTHORS

Goldsmith P. ,
Delafield H.J. ,
Cox L.C.

Goldsmith P. ,
May F.G.
                       1966
s
en
o
m
CO
O
X
m
U
CO
O
m

O
      Sparks, L.E
      Pilat M.J.
                 1970
HIGH POINTS OF STUDY

Experiments with radioactively tagged
nickel-chromium aerosol CParticles
0.02 to 0.2 urn in diameter)  gave
deposition velocities close  to those
predicted from the Waldman and Baka-
nov et al. equations. Comparison
of thermophoretic velocities with
theoretical predictions shows that
for r > A(r  the particle radius,
A - the mean free path under the
given conditions) they vary over a wider
range. Derjaguin equation predicted
a velocity 16% lower and Brock equa-
tion 40% higher than measured. Diffu-
siophoretic and thermophoretic forces
are additive. Experiments of particle
deposition efficiency were run in a
Liebig condenser. The results show that
collection efficiency could be plotted
vs. the rate of steam condensation g/min

Single droplet target efficiencies for
particle collection by the combined
mechanism of inertial impaction and
diffusiophoresis were calculated. These
values were used to calculate overall
collection efficiency in a spray tower.
It was found that condensation can
greatly improve particle collection.
REFERENCE

Quart. J. of
the Reg. Meteor.
Soc. 89, 43,
(1963F"
Chapt. VII in
Aerosol Science,
C.N. Davies ed
(1966)
Atmos . Env.  4_,
1,  C1970)
(O
10
in

-------
 0     AUTHORS          YEAR     HIGH POINTS OF STUDY                      REFERENCE

       Davis  R.J.       1972     They concluded that diffusio and          Instrum.  and
       Truitt J.                 thermophoresis would be too expen-        Control Systems
                                sive for use in scrubbers. Particle       pp.  68-70,
                                growth due to condensation followed       (Nov 1972)
                                by turbulent agglomeration is the
                                best way to increase scrubber
                                efficiency.

       Matsuzaki K.     1970     The invention describes a venturi         Japanese  patent
                                scrubber where steam is added             No.  Sho 41-41184
                                tangentially upstream of the throat.

       Mashita T.       1971     Describes  the Solivore scrubber which     Indus.  Public
   t^                            is composed of Venturis with water        Nuisance  7_, 573,
   "-1                            sprays upstream and downstream of         (1971)    ~~
                                the throat. Several operating condi-
                                tions are described.

s
CO
O
-n
T]
O
m
CD
8
m
CO
O
m
O
CO
ro
CJI
O
ro

-------
          ENGINEERING DESIGN

               For an engineering design of an FF/C scrubber one
          needs design equations which tie the operating conditions
          to scrubber performance.  Each scrubber type may require
          different equations.  (See for example Chapter 5.3 of the
          Scrubber Handbook.)  Thus, equations which describe a
          venturi scrubber would not be adequate for a tray column,
          etc.

               To predict particle penetration, which is the basic
          measure of scrubber performance, one has to know the
          particle deposition velocity and the deposition surface
          area per unit of gas flow rate.  The relevant deposition
          velocities are those due to the flux forces, inertial,
          gravitational and Brownian diffusional effects.  Since
          the flux force deposition velocities are functions of
          the temperature and vapor pressure gradients, knowledge
          of the magnitude of these gradients at various distances
          Cor residence times) along the gas path through the
          scrubber is needed.

               Simultaneously, the inertial and gravitational deposi-
          tion velocities are functions of the particle size and
          density which in turn change due to any vapor condensation
          on the particle.  The critical vapor supersaturation re-
          quired for nucleation depends upon the particle properties
          and the operating conditions.  Therefore,  to predict
          particle growth one must first know the vapor composition
          as a function of distance (or time)  or travel inside the
          scrubber.   Vapor composition depends on heat and mass
          transfer between gas/particles and gas/liquid.   Thus, the
          computation of mole fraction water vapor in the gas requires
          knowledge  of particle size and concentration, gas/liquid
          transfer area, heat and mass transfer coefficients, gas
          flow rate, and the critical saturation ratio required to
          nucleate condensation on the particles.  All of these para-
          meters change as the gas proceeds through the scrubber.

               Figure 1 is a pictorial and verbal representation of
          the changes which take place as the gas moves through the
          liquid (foam) on a sieve plate, as an example.   Comparable
          diagrams could be drawn for other types of scrubbers with-
          out much change in the major features which are shown.
          Also shown are the computations which are  needed in order
          to compute the changes of magnitude for the variables shown.

                                     22
A. P. T. Inc.                                POST OFFICE BOX 71.  RIVERSIDE. CA 92502

-------
       Plots of Variables
    Description
Computations Needed*
        .Direction of Gas
                Flow
Gas  temperature  drops
rapidly

Liquid temperature  is
fairly uniform.
 Ht. Tr. G/L, G/p
 Mass Tr. G/L,
 G/p
 Enthalpy Bal.
 Mass Bal.
                              Mol  fraction water
                              vapor in  gas reaches
                              equilibrium rapidly.
                         Same  as  above.
                              Particle  radius  grows
                              rapidly while  satura-
                              tion  ratio  is  high.
                         Mass Tr.  G/p  -
                         (NeedTG, TL, Mpi
                              Particle  concentration
                              drops  abruptly  during
                              bubble  formation,  then
                              fairly  uniformly.
                         Deposition Rate
                   PC
Diffusiophoretic vel.
rapid when "y" high.
Thermophoretic vel.
drops quickly.
Brownian diffusion is
low for r  > 0.05 \im.
Inertial  (centrifugal)
dep. increases with
"r ".
  P
 Temp,  gradient
 Composition grad.
 Slip  correction
 Tnertial  parameter
 Centrif.  force.
        *Mass Tr. G/L = Mass transfer, gas  to  liquid,  G/p  =  gas  to
         particles, other symbols are as defined in  "Nomenclature",
               Figure 1 - Schematic representation for  FF/C
                          scrubbing on a sieve plate.
A. P. T. Inc.
                                     23
                  POST OFFICE BOX 71.  RIVERSIDE. CA 92502

-------
          Because of the rapid changes in conditions and the compe-
          tition between the particles and the liquid surface for
          the condensing water vapor,  any realistic design method
          must consider the point-to-point conditions.

          Availability Of Design Information

               In order to carry out the design computations dis-
          cussed above, several items  of information are required.
          On the following page we discuss the various  design com-
          putations and the state of knowledge regarding them, as
          of the time this research was started.

          1.  The particle growth equation

               All the equations which are presented in the litera-
          ture are based on a wide and varied set of assumptions
          which have to be checked experimentally in order to deter-
          mine their applicability to  the specific case at hand.
          First one must know what saturation ratio is  required to
          cause nucleation of condensation on a particle of some
          specific substance (\vhich probably has  not been studied
          in this respect).  Additional complications occur for
          particles 0.5 to 2 ym in diameter, which are  in the slip
          flow regime under normal pressure and temperature, and
          for high vapor concentrations, such as  are probable in an
          FF/C scrubber.  Further, there are probably regions of
          higher saturation ratio than exists in  the bulk of the
          gas phase, and condensation  could be initiated there.  It
          was, therefore, imperative to check the validity and
          applicability of the particle growth equation chosen.

          2.  The overall energy balance for the  liquid interface.

               This equation used to determine the interface tempera-
          ture requires the knowledge  of the gas/liquid heat and mass
          transfer coefficients, the liquid phase heat  transfer co-
          efficient, the nature of the interfacial area, and the flow
          patterns of the two major phases.  The  coefficients can be
          estimated from the two-film  theory, the penetration theory,
          or empirical correlations.  Experimental coefficients can
          be found in the literature but they play a major role in
          the prediction of FF/C scrubber performance (particle pene-
          trationj and even small errors in the coefficients can cause
          a large error in the predicted performance.


                                     24
A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
                In  the  case  of  nonconventional  designs,  experimental
           coefficients were unavailable.   The  geometry  of the inter-
           face  must  be idealized  if  predictive computations  are  to
           be  made.   We are  aware,  for  instance,  from  the  voluminous
           research on  mass  transfer  in  plate type  equipment  for
           distillation and  gas  absorption  that simple models  depict-
           ing the  gas  phase as  spherical bubbles are  not  quite right
           but may yield some useful  answers.   Local gas velocity and
           the overall  gas flow  pattern  will influence both the local
           transfer coefficients and  the bulk conditions for  the
           phases in  contact in  any region.  Liquid phase  hydrodynamics
           has similar  effects  and  is likewise  unpredictable.

           3.  Rate of  vapor condensation.

                Vapor may condense on the particles and  on the collect-
           or  liquid  surface.  These  two phenomena  are governed by the
           same  equations and coefficients  discussed earlier  and  the
           uncertainties which applied there apply  also  to this case.
           Because of the competition and the effect on  the driving
           force for mass transfer, the  two rates must be  evaluated
           simultaneously.

           4.  Rate of  change of the  gas temperature.

                The gas  temperature changes due  to  the latent  heat of
           condensation  on the liquid and on the particles, the heat
           transferred  either to or from the liquid, and heat  losses.
           All three effects  are governed by the equations relating
           to  particle  growth, energy balance for the  liquid  interface
           and heat losses.  The uncertainties  and  limitations of these
           equations have been discussed earlier.

           5.  Rate of  particle  removal.

               Many uncertainties are piled one on top of the other
           in  the equations  used to calculate the particle concentra-
           tion.  We started by  assuming that the particle fluxes  due
           to  the flux  and centrifugal forces are additive.   In addition
           we  assumed that when particle collection is due to  two  con-
           secutive mechanisms,  the overall penetration is the product
           of  the individual penetrations.  Both these assumptions  are
           accepted assumptions  and are  sound theoretically, however,
           they had not been proven experimentally.

               The accuracy of  the deposition velocity equations  was
           questionable  in that there was little experimental  evidence
           in  the range of particle sizes considered in this work.   On


                                     25


A. P. T. InC.                                POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
          top  of this,  there  was  the  uncertainty as to the gas
          heat and mass transfer  coefficients  which in turn
          determine the gradients necessary for the calculation
          of the flux  deposition  velocities.   Interwoven with
          this was the  effect of  unknown  interfacial area for
          transfer and  deposition.  With  regard to the centrifugal
          deposition there  were questions concerning the gas flow
          radius of curvature and local velocity, especially when
          condensing vapor  is added.

               For plate type scrubbers,  an additional removal
          mechanism is  particle collection during bubble formation,
          which becomes especially important  after particle growth
          takes place.   Penetration during bubble formation is a
          function of  foam  density as well as  particle properties,
          plate geometry, and gas velocity under normal operating
          conditions.   However, foam  characteristics can be different
          when condensing vapor is present and there were no experi-
          mental data  on collection during bubble formation when
          vapor is condensing.

               One faces the  same kind of questions with other
          collection geometries,  such as  drops, packed bed and
          jets.  No equations were available  which would permit
          the  calculation of  drop target  efficiency in the presence
          of flux forces and  the  same also applied to jets.
                                     26
A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
                           THEORETICAL BACKGROUND
               General background on flux force deposition and
          particle growth by condensation of water vapor will be
          discussed in this section of the report before proceeding
          to the use of this knowledge in the development of FF/C
          scrubber models.  As a starting point it is important to
          note that fine particles are in about the same size range
          as the mean free path of the gas molecules.  Consequently,
          the interactions between the particles and the gas are not
          the same as they would be for plane surfaces and the gas.

               A measure of the departure of the transport processes
          in the gas-particle system from the laws of continuum mech-
          anics is provided by the Knudsen number, N  .


                              N   = —                            (21
                              NKn   rp                            W


          where "X" is the mean free path of the gas molecules (about
          0.07 urn for air at one atmosphere pressure and 40°C) and
          "r " is the particle radius.  Only in the limit, when
          ^Kn "* ^ can t'ie £as surrounding a particle be regarded as
          a Bontinuum.  The other extreme, when N,,  •* °°, is usually
          referred to as the free molecule regime.  The "slip flow"
          regime, which applies roughly when 0.25 < N.,  < 10.0, is
          of importance in this program because for particles of
          0.1 urn to 2.0 jjm diameter, the Knudsen number runs from
          about 1.0 to 15.  Maxwell (1890) suggested a procedure for
          treating the case of slip flow in which the transport
          processes in the bulk of the gas are described by the con-
          tinuum equations, but in a very thin boundary layer of gas
          next to a surface the non-continuum effects are accounted
          for.  Thus, the boundary conditions for slip flow allow for
          slip of the gas relative to the surface and analagous
          "jumps" in gas temperature and concentration between the
          surface and the gas.

          DEPOSITION VELOCITIES

               Several equations have been derived which relate the
          flux force deposition velocity of the aerosol particle in
          the slip flow regime with the appropriate gradient.  For

                                     27
A. P. T. InC.                                 POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
          the influence  of a temperature  gradient on particle
          motion Epstein (1929)  derived:
                                       2k,
                    u
                     pT
                                               VT
                        C3)
                                      r
                                      G
          where  "ur"j  "Pr",  and "kr" are  respectively the gas viscos
          ity, density and thermal  conductivity,  "T" is the tempera-
          ture,  "7T" is the  temperature  gradient  and "k " is the
          particle  thermal conductivity.                "

              Brock (1962)  has corrected Epstein's equation by
          accounting not only for the thermal  slip but also the
          temperature  jump and the  friction slip.   His equation
          for the thermophoretic velocity is:
               u
                PT
3uG C'
kL ^* ^ 1.
/"• +• *T~" ".»»
u Li p
2>G T j1 + 3Cm ^)

2kG + kp * Ct T; kp)
                      7T  (4)
          where  "C"1  is  the  Cunningham slip correction factor and
          "C " and "C "  are  the  temperature jump coefficient and the
          isothermal  §lip coefficient  respectively.   The experimental
          values for  "C." and "C " are C^  = 2.3 and  C  =1.25.
                       t        m       t             m

               Bakanov and Derjaguin (1962) claimed  that the method of
          Epstein and Broek  is invalid.   They  calculated that Maxwell's
          expression  for thermal creep is  too  large  by a factor of 35
          and hence thermal  creep is too small  to account for thermo-
          phoresis.  Instead they considered a  term  in the equation
          for the heat flux  in the bulk oE the  gas which follows from
          the third order Chapman-Enskog theory.  Applying Onsager's
          principle of symmetry  of kinetic coefficients, they calcula-
          ted an expression  for  the thermophoretic velocity.  Derjaguin
          and Yalamov in (1965)  improved this  by accounting for veloc-
          ity slip and temperature jump and got
               u
                pT
      VT
                           2kr +  k  +  2C.  — k
                             G    p     t  rp  p
(5)
                                     28
A. P. T. Inc.
POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
          Diffusiophoresis

               Diffusiophoresis  in the  slip  flow  regime  was  anal-
          yzed by Schmitt and Waldmann  (I960)  as  a  hydrodynamic
          problem, accounting for a  finite gas  velocity  at  the
          particle surface due to diffusion  slip.   For the  case
          of vapor diffusing through a  resting  gas  they  got  for
          the diffusiophoretic velocity:
                        u
                         PD
                               yr
                            12  (1
(6)
          where "yr" an^ "X " are the mole  fractions  of  the  non-
          condensible gas and the water vapor,  respectively,  and
          "D" is the diffusivity of water vapor in  the gas.

               "a  " is a numerical constant called the  diffusion
                 i 2
          slip factor.  Kramer and Kistemaker  (1943)  calculated  it
          as :

                                M  - M^
                            1 2
                                M +
                                                                   (7)
          Combining equations 6 and  7 yields:
                     u
                      PD
                               y
                               7
                                      V
                                                  D
                                            v
                                                         (8)
               Derjaguin et al.  (1966) derived an equation  for  diffu-
          siophoresis of large particles  (r  > A) .  Their derivation
          is similar to the one  they used  for thermophoresis.   They
          give for the diffusiophoretic velocity near  a  condensing
          or evaporating drop:
where "p^" is the density of the inert gas and
density of the mixture.
                                                          "p  "  is  the
                                     29
A. P. T. Inc.
                                 POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
          Experimental Results

               Experiments by Schmitt  (1959) with oil drops  and
          by Schadt and Cadle (1961) with NaCl particles  and
          with tricresyl phosphate drops (high conductivity)
          0.05 < N..  < 1 agree within  =20% with  the predictions
          of Brock's theory.  Derjaguin, et al.  (1966)  on  the
          other hand, pointed to convective currents as an un-
          controlled source of error in the Millikan condenser
          apparatus.  They devised new experiments and  obtained
          data for 0.15 < N.,  < 0.5 which confirm their theory.
                           K.n

               Measurements of thermophoresis by Goldsmith and
          May (1966) with Nichrome aerosol (N.,   = 1) appear  to be
          bracketed by the opposing theories.  Fulford  et  al.  (1971)
          analyzed their experimental  results together  with  those
          obtained by Derjaguin et al. (1966), Keng and Orr  (1966),
          Brock (1967), Schmitt (1959),Schadt and Cadle (1961) and
          Calvert and Byers (1967).  Their conclusion is  that  for
          particles in the size range  of 1. ym "> r  ^ 0.1  ym the
          Derjaguin equation agrees quite well with the experimental
          results.  For larger particles they suggest an  empirical
          correlation based on their experimental data.

               Schmitt and Waldmann (1960) pointed out  that  equations
          (6) and (7) are accurate within ^9$ for the prediction  of
          diffusiophoretic deposition  velocity.  Goldsmith and May
          experimental results with Nichrome aerosol (N.,   =  1) com-
          pares quite well with the theoretical  results "predicted
          by equations (6) and (7).  Derjaguin et al. (1966) however,
          showed that Prokhorov and Leonov (1966) data  are in  better
          agreement with equation  (9).

          Condensation and Particle Growth

               The water vapor contained in a saturated or supersatu-
          rated aerosol stream may condense on the particles,  on  the
          cold walls of the conduit, or on water surfaces, when
          passing through a scrubber.  In our case the  aerosol parti-
          cles are =1 ym diameter, may be soluble, insoluble,  wettable
          or nonwettable in water, and flow at the same velocity  as
          the gas surrounding them.  Under these conditions  the  follow-
          ing assumptions apply:

               1.  The vapor pressure  increase due to particle
                   curvature can be neglected for particles  of
                   interest.

                                     30
A. P. T. InC.                                 POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
                2.  The  increase  in  heat  and  mass  transfer
                   caused by  the  relative  velocity  between
                   the  gas  and the  particles can  be
                   neglected.

                3.  There is no vapor pressure depression  due
                   to the colligative  properties  for  insoluble
                   particles .

                4.  Homogeneous nucleation  is neglected because
                   of the aerosol particles  present.

                Having limited our area of  interest, we will  look at
           the phenomena of  nucleation  and  growth  of water drops in
           a  saturated and supersaturated atmosphere.

           Nucleation  -  The  conditions  necessary for nucleation to
           occur on a  surface  may  be predicted from  equilibrium
           thermodynamic considerations for a system in stable
           equilibrium at a  constant temperature when  the  Helmholz
           free  energy is minimized.  This  fact is used to find the
           critical saturation ratio, S , for a surface exposed to the
           vapor.   (See  for  example  Abraham (1968),  Fletcher  (1966)
           and others.)

                The saturation ratio is defined as the ratio  of the
           vapor pressure at  the  point  in question "p  " to that of a
           plane water surface "p.".                 °
                                    S  =                             (10)..
                It  follows  from  the  Kelvin  equation  that for a given
           drop  diameter, water  droplets  grow  by  condensation if the
           saturation  ratio is larger  than  a critical  one.   For drops
           2  vim  in  diameter the  critical  saturation  ratio is 1.001 and
           for drops 0.2 urn in diameter  it  is  1.01.  The rate at which
           drops  in this size range  form  spontaneously from the vapor
           increases very steeply  with supersaturation.   For water at
           20°C  as  an  example, the rate  of  nucleation  will  increase
           from  1.2xlO-Vcm3 -sec  for  S  = 3.2  to  1.2xlO'2/cm3 -sec for
           S  = 3.6  (Amelin, 1967).  This  is important  since around the
           critical saturation ratio the  rate  of  nucleation can be
           increased substantially by  a  small  increase in the satura-
           tion  ratio.

                                     31
A. P. T. Inc.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
               The critical saturation ratio is also a  function  of
          temperature.  Figure 2 taken from Amelin  (1967)  shows  the
          change in the critical saturation ratio for water vapor as
          a  function of temperature.  The critical  saturation  ratio
          for homogeneous water vapor nucleation is always above 2.0,
          within our temperature range of interest.

          Nucleation by Insoluble Particles - Drops of  the critical
          radius will form on the surface of insoluble  particles at
          a  higher rate than for self nucleation.  Volmer  (1939)
          developed a modification of the original Volmer-Weber
          nucleation theory to include the case of  condensation  on
          a  plane surface.  He showed that the critical saturation
          ratio for insoluble surfaces is a function of the contact
          angle between the condensed vapor and the surface.
          Figure 3 shows the effect of contact angle on critical
          saturation ratio for water at 20°C.  The  contact angle is
          that which is measured between the solid  surface and the
          liquid surface at the line of contact, and which includes
          the liquid phase.

               This analysis was extended to condensation  on in-
          soluble particles by Fletcher (1958), and the results  of
          his theoretical calculations are plotted  in Figure 4 as
          the critical saturation ratio vs. drop radius with cos a,
          the cosine of the contact angle, as a parameter  for  water
          vapor at 20°C.  This plot shows that for  fine particles a
          saturation ratio of a little over 1.0 will suffice for
          highly wettable particles (a =0°, cos a = 1.0).  In  the
          case of non-wettable particles (a approaching 180°)  the
          critical saturation ratio required for nucleation to occur
          is equal to that required for homogeneous condensation.
          Thus, most particles will cause nucleation to occur  at
          lower saturation ratios than for homogeneous  nucleation.

          Nucleation By Soluble Particles - Nucleation  occurs  more
          readily on soluble particles than insoluble particles.
          Raoult's law states that the equilibrium vapor pressure
          over a solution is less than that over pure water by a
          factor equal to the mole fraction of water in the solution.
          Using the Van't Hoff factor, i, to take into  account the
          dissociation of inorganic salts, Howell (1969) derived an
          expression for the equilibrium saturation ratio  over a
          solution droplet containing "m" moles of a solute.   Fig-
          ure 5 is a plot of (p-Xp^ - 1) x 102 vs. droplet radius
          with "im" as a parameter, "p." is the vapor pressure above
          a  droplet of solution containing "m" moles of solute,  and
          "p^" is the vapor pressure above a plane  surface of  pure


                                     32


A. P. T. Inc.                                POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
                  i
                  2
                  O
                  n

                  1

                  I
                  1-1

                  u
                  I—I
                  H
                  I-H

                  U
                              10    20     30     40     50


                                        TEMPERATURE,  °C
60
70
                 Figure  2  -  Critical saturation  ratio for homogeneous
                             nucleation of water  (After Amelin  (1967)   )
                                       33
A. P. T. Inc.
                                             POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
                   o
                   I—I
                   E-
                   CO
                             20    40    60    80  100  120  140  160   180

                                   CONTACT  ANGLE,  DEGREES
                Figure 3 - Critical saturation ratio for water upon  a
                           plane substrate  of given contact angle  -
                           water at 20°C  (After Fletcher 1966).
A. P. T. Inc.
                                             POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
                            10
                10"7   I0~6   ID'S   lO'*4   10"3
               PARTICLE  RADIUS,  cm
                 CM

                 O
Figure 4 - Critical  saturation ratio for nucleation
           of water  droplets  upon a particle of
           given radius with  cos a at 20°C as
           parameter.   (After Fletcher 1966).
                    8
                   ex
                  H
                  2
                  OQ
                  —i
                  hJ
                       -0
                       -0
                         0.01
                                     DROPLET RADIUS, (pm)
                  Figure  5   Equilibrium supersaturation  as  a  function
                             of  droplet radius with (im)  as  a  parameter.
                             (After  Howell 1949) .

                                       35
A. P. T. Inc.
                          POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
          water.  If, for a given value of "im", the ambient satura-
          tion  is raised to a value below the peak of the equilibrium
          saturation curve on Figure 5, an equilibrium drop size will
          be formed.  If, however, the ambient saturation reaches  the
          peak  of the curve or higher, spontaneous nucleation, con-
          densation and growth will occur.

                The curves presented in Figure 5 apply only when the
          vapor concentration is higher than the concentration which
          is in equilibrium with saturated solution at this tempera-
          ture.  For example, a saturated solution of NaCI will
          reduce the equilibrium vapor pressure of the water by =221;
          thus  no condensation on the NaCI particles is expected
          unless the relative humidity is higher than 78%, which was
          found to be the case experimentally (Junge, 1952).   In the
          case  of NaCI, Figure 5 will apply only when the relative
          humidity is higher than 78%.

          Mixed Nuclei - In any real situation it is unlikely  that
          the natural condensation nuclei will be described completely
          by one of the types we have considered.  It is more  likely
          that  a mixture of soluble, insoluble, wettahle and non-
          wettable particles will be present.  In such a mixture we
          should consider two cases:

                1.  How the mixture would behave as condensation
                   nuclei.

                2.  How a particle formed by agglomeration of two
                   different type particles would behave as a
                   condensation nucleus.

                The first case is fairly simple because those particles
          which require lower saturation ratios to nucleate will
          nucleate first and grow, thus reducing the degree of super-
          saturation in the aerosol.  In order to find the exact
          proportion of nucleation on the various particles at a
          given supersaturation, one should examine the rate of
          nucleation and the rate of decrease of supersaturation.   In
          reality the situation may be complicated by the inhomogen-
          eity  of the saturation ratio within the aerosol stream,  as
          is discussed later.  However, when the saturation ratio  is
          below the critical value for each of the different cases,
          no nucleation and growth will take place.

                The second case, that of a mixed particle, is more
          difficult to analyze.  Even when all the components  of a
                                     36
A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
           coagulated  nucleus  are  similar particles,  it may possess
           properties  differing from those of a simple particle of
           the  same  size.   This is  due  to the fact  that its surface
           will  contain  cracks and  cavities between the particles,
           and  these are favoured  condensation sites.

                If the nucleus consists  of a mixture  of soluble and
           insoluble particles, the  nucleation behavior is  dominated
           by the  soluble  component.  Above the phase  transition of
           these components  the particle is enveloped  in a  film of
           solution, and behaves as  a solution droplet in the  way we
           have  already  considered.   The presence of  the insoluble
           components  enhances the  effectiveness of the soluble salts
           since less  water  is required  to make a droplet of a given
           size  and  the  resulting solution is more  concentrated.   The
           behavior  of such  mixed nuclei has been examined  in  some
           detail  by Junge  (1952).

           Particle  Growth  - After  a drop of critical  size  has been
           nucleated,it grows at a  rate determined by the  ambient
           environment and  conditions at its surface.   However, as
           vapor condenses  on  the drop,  the drop temperature rises
           and  hence the conditions  at  its surface  change.   Several
           equations can be  found in the literature describing the
           rate  of growth,  Fletcher  (1966),  Nuzhnyi et al. (1965)
           and  Fuchs (1959).   In general they all make  the  same
           assumptions and use the same  three  basis relations:
           rate  of mass  transfer, rate  of heat transfer and phase
           equilibrium.  Fletcher's  equation,  which accomodates soluble
           and  insoluble nuclei and  is  integrated to a comparatively
           simple  form was  used in  our  early modeling.   In  our later
           and  more  refined  models  for  particle growth and  collection,
           however,  we use  the basic heat and mass  transfer equations
           in order  to avoid some of Fletcher's approximations.

                Rate of  growth during a  condition of constant  satura-
           tion  ratio  is given by Fletcher as:
                       dr
                    r       =  G
                    p at      F
a  + b
r    PI ^"Re'"Sc
£CNDo,NqJ         (11)
           "r  "  is  the particle  radius  (cm)  at  time  "t",  "S"  is  the
           saturation ratio  defined  in  equation  10,  a/r   describes  the
           effect of surface  curvature  in  increasing the  vapor pressure
           of  small drops  and "b/r3" describes  the effect of  solute
                                     37
A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
           content on vapor pressure,   "a" and "b" can be approximated
           by
    3.3xlQ-5  /cm
a      T
                                       I
                         b = 4.3 i m
                                      cm -  gmol
                                           drop
                                                                    (12)
                                           (13)
           Where "im" is  the number of moles of solute in a drop and
           "G_"  is defined as:
             r

                       D
                  GF '
                                                                    (14)
           "Dr" is  the vapor diffusivity in air, "p " and "pj" are
           respectively the density of the vapor and the liquid, "L^"
           is  the latent heat of evaporation, "M " the molecular
           weight of the vapor,  "R" the gas constant = 1.987
           (cal/gmole - °K) and  "k '' is the gas thermal conductivity.
           f(NR »Ngc) and f(NRe»Npr) represent the increase in the
           heat and mass transfer coefficients due to the relative
           velocity between the  gas and the growing nuclei, which is
           negligible for fine particles.   The correction due to the
           surface  curvature of  the particle, "a/r ", can also be
           neglected and "b/r "  applies only in the case of soluble
           particles.  Thus Fletcher's equation reduces to:
                            dr
                          p dt
         GF I5'1 * 7T
                    P
           where:
DG Pv
PL
Dr LM
IA G M
R T2
p M
Hv v
kG
                                                 - 1
                                                                    (15)
                                                                    (16)
                Particles 1 ym in diameter are in the slip flow regime,
          and from Hidy  and Brock  (1970)  we  obtain:
DG pv
                1 +
                                               LM
                            M
                                                      v
                                             R
                                                        -1
                                                                   (17)
                                     38
A. P. T. Inc.
                   POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
                For high vapor concentrations a second correction is
           required due to the Stefan flow.  From Fuchs (1959) this
           correction is approximated by
Dp jl -
> Pi

+ PG
2P
pv
                   (1+1.2 N,
1+
1 Pi + Pf
D 1 •••
UG r 2?
LM pv Mv
R T2 k J
                                                                 -1
                                                                    (18)
           where "p.^" is the vapor pressure at the drop surface, "p "
           is the partial pressure in the gas stream and "k " is the
           thermal conductivity of the gas corrected for thi Stefan
           flow.

           Saturation Ratio in Boundary Layer - When considering the
           conditions under which particles will nucleate condensation,
           it is important to note that the saturation ratio can be
           higher in the boundary layer than it is in the bulk of the
           gas.   This can cause the onset of particle growth sooner
           than  would be expected based on bulk conditions.  Once the
           particle surface has been wetted it can serve as a site for
           further condensation so long as the saturation ratio ex-
           ceeds 1.0.  The particles grown by condensation will be
           more  susceptible to collection by mechanisms dependent on
           their mass.

                As an example, let us take the case of particle collec-
           tion  from the interiors of bubbles, and for simplicity we
           will  treat the boundary layer in terms of penetration
           theory.  We  will discuss the behavior of an element of gas
           which is undergoing simultaneous heat and mass transfer as
           it moves around the bubble due to gas circulation inside
           the bubble.   Both composition and temperature distributions
           in the gas contacting the water surface are assumed to be
           the same as  unsteady-state diffusion into (or out of) a
           semi-infinite plane sheet.   Crank's (Crank, 1956) solution
           for this case is shown in Figure 6, a plot of dimensionless
           concentration or dimensionless temperature versus dimen-
           sionless distance.   The dimensionless concentration is the
           difference in water partial pressure between any point and
           the gas-liquid interface divided by the difference in
           partial pressure between the bulk gas and the interface.
           Dimensionless temperature is defined in an analogous way
                Dimensionless  concentration  =
                                               P -
                                               PG
(19)
                                     39
A. P. T. Inc.
                                           POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
                           l.Or
                    H

                    i


                    H
f-

 i
 (J

f-   0.5
                      j-
                      o
                    a
                       ex
 e?
a
                                    0.4    0.8



                                            X =
                                                (4 Dt)1/2
                          1.0   1.4    1.8   2.0


                            x
                   Figure 6 - Dimensionless concentration  or  temperature

                              versus  dimensionless distance
                                       40
A. P. T. Inc.
                                             POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
               and
                                           T  - T.
               Dimensionless temperature E =	«—                 (20)
                                           !G     i

               Dimensionless distance = 	TT-T- = ^              ^   ^
                                        (4 Dt)1/*

          Where:

               p = partial pressure of transferring component  at  a
                   position "x", atm.

               T = temperature of gas at position "x", °K or °C

               x = distance from interface, cm

               D = diffusivity of mass or heat, cmz/sec

               t = penetration time, sec

               To get some idea of the magnitudes involved, consider
          the common case of a 0.4 cm diameter bubble rising at
          20 cm/sec and let us use representative values of the
          properties involved.  The penetration time for a circula-
          ting bubble is estimated from the time  it takes for  a
          1 dia. rise; or 0.02 sec.  The diffusivity of water  vapor
          in air is about 0.25 cm2/sec, so the value of "x" at a
          dimensionless distance, X, of 1.0 would be 0.14 cm.  The
          diffusivity of heat, (k/pC ), is approximately 0.2 cmz/sec
          and the value of "x" for "t" = 0.02 sec and a dimensionless
          distance of 1.0 would be 0.136 cm.

               From the above example, it can be  seen that the dimen-
          sionless composition and temperature distributions can  be
          considered identical on the same real distance scale.   To
          illustrate, we see that at X = 1.0  the  real distance to the
          interface is about the same (i.e.,  0.14 cm) for heat and
          mass transfer and that the values of dimensionless tempera-
          ture and composition are the same.  One can also note that
          most of the volume of a circulating bubble 0.2 cm radius
          would become involved in the boundary layer, which grows to
          about 0.14 cm at the bottom of a rising bubble.

               Some examples of gas and liquid conditions were chosen
          for illustrative purposes and the saturation ratios  which
          would occur because of the heat and mass transfer were  com-
          puted for various positions in the  boundary layer as given


                                     41


A. P. T. InC.                                POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
           by  Figure  6.   The  results  are  shown  in Figure 7, a plot
           of  saturation  ratio  vs.  dimensionless distance from the
           gas/liquid interface.   It  can  be seen that the curves
           for a  water interface  temperature of 20°C and bulk gas
           phase  conditions of  45°C,  S  =  1.0 and 60°C,  S = 1.0 have
           maxima where the saturation  ratio goes considerably above
           1.0 at a position  within the boundary layer.

               For the case  of a 20°C  liquid interface  temperature,
           gas phase  temperature  of 30°C,  and S = 1.4,  there is no
           maximum in the  curve.   It  can  be seen from these examples
           that it is possible  for  the  saturation ratio  in the
           boundary layer  to  be higher  than that in the  bulk of the
           gas phase, depending on  the  temperatures of  the gas and
           liquid.  Consequently, a substance which requires a
           critical saturation  ratio  of 1.5 in  order for water to
           condense on it  could nucleate  growth under the 60°C gas
           temperature condition, even  though the saturation ratio
           for the bulk gas phase is  only 1.0.   Once condensation
           starts on  a particle it  can  continue to grow  so long as
           the saturation  ratio is  1.0  or larger.
                                     42
A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
Curve No .
1
2
3
4
T °C
IG, L
60
45
30
45
T °f
L'
20
20
20
20
S
1.0
1.0
1.4
1.1
                                 0.5
1.0
1.5
2.0
                                     X  =
                                         (4 Dt)1/2
             Figure  7  -  Saturation ratio  versus dimensionless  distance
                                       43
A. P. T. Inc.
POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
                      MATHEMATICAL MODEL,  SPRAYS


              Mathematical models were  developed  for  three  principal
          types  of  scrubbers;  spray,  plate,  and  liquid sheets.   The
          models were  to be used  first for  the exploration of possible
          operating conditions and design variations in order to assess
          the  capability of each  scrubber type when operated in the
          FF/C mode and to locate an  approximately optimum operating
          range.  Next, the models would be  used for the correlation
          of experimental data and then  the  evaluation and possible
          modification of the  model in the  light of the data.

              The  models which were  developed early in this research
          were for  plates, as  represented by the unit  mechanism of
          collection from bubbles.  They did not allow for particle
          growth, simultaneous transfer  to  particles from liquid,
          and  other factors.   In  the  course  of the program the plate
          models were  refined; especially since  our experimental work
          was  on plates.  As it happened, the mathematical model for
          sprays was developed fairly early  in the program and was
          not  revised  at the time of  writing.  Because it is the
          oldest model, that for  sprays  will be  presented first, so
          that the  progressive development  of more realistic and com-
          plex models  during the  research will be  reported in order
          of our increasing understanding of FF/C  system modeling.

              Spray type scrubbers are  represented in fundamental
          terms  by  the unit mechanism of particle  collection by drops
          or spheres.  Collection of  particles by  spheres due to
          inertia alone is discussed  in  the  literature and target
          efficiencies have been  calculated  by many:   Langmuir and
          Blodget  (1946), Ranz and Wong  (1952),  Herne  (1960) and
          others.   When additional forces are involved,  the  equations
          of motion are different and new solutions are required.
          We adopted Sparks and Pilat (1970)  approach  and solved the
          equations of motion  of  the  particle in the "X" and "Y"
          direction after including the  additional flux force.

              A force balance on the particle in  the  "x" direction
          gives  for the case where Stokes law applies:

                   du      1             F

                   — =  -  (vx  * V + HT                     (22)
                   dt      T   x    x    p
                                     44



A. P. T. InC.                                POST OFFICE BOX 71,  RIVERSIDE, CA. 92502

-------
          and


                        Fx = 6" "G rp UF                          (23)

                                          2r° pD
          T  = particle relaxation time =  Q p—*-  (sec)
                                             VG

          v  = gas velocity in "x" direction, cm/sec
           X

          u  = particle velocity in "x" direction, cm/sec
           J\

          m  = particle mass, g

          F  = flux force acting on particle in "x" direction, dynes
           j\

          UF = particle velocity due to flux force, cm/sec


               The equations of motion can be put in dimensionless
          form by the use of the following new dimensionless para-
          meters and variables:

              x        Y    -    vy        vv   -   * vn
          X = —•   Y = —•   v  = — •  v  = —•  t = 	-
          A - r '       r '   VX - v '  VY - v '  C - ^FT
               d        a         o         o         d

               2 p  r" VQ
          K  =  g ^—p	 = inertial impaction parameter          (24)
                   G  d

          and


                               GF E K  v~                         (25)
                                     P  o

          Where:

          X = distance from center of drop in "x" direction, cm

          Y = distance from center of drop in "y" direction, cm

          VQ= velocity of undisturbed gas stream relative to sphere,
              cm/sec


          Substituting values of "v "  for potential flow, from Fuchs
          (1964),  we obtain the differential equations in their final
          form:
                                     45

A. P. T. InC.                                 POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
                         dX     2 X2 - Y2           GF  *      = 1_  (2g)
                       _

               dt2    K  dt   2 K  (X2 + Y2)25    (X2  +  Y2)  *    K
                       c         cr                              *
                       _      -      ,              .    .
                dt2   K  dt   2 K (X2 + Y2)25    (X2 +  Y2)1/2
                       P         P
                                                             =  0    (27)
          Equations  (26) and  (27) can be solved if boundary  conditions
          are specified and if an expression for  "G  " is available.


           FLUX  DEPOSITION

               Equations (26) and (27) show that  the two parameters,  K
          and G_, are required to define particle motion under  the  in-p
          fluence of inertial and flux forces.  It will be shown  later
          that the product of these parameters is a new parameter which
          defines the effect of adding flux forces to a system  involving
          only inertial force.  Equation (25) can be rearranged to  yield
          the product:


               v  r    UF _ Particle transport by flux forces   „
                D  P ~ v --- - - =  PD
                v       o   Particle transport by fluid flow
                                                                   (25)

               The dimensionless parameter "v /UF" is related to  the
          Peclet number since it is an indication of the relationship
          between convective and flux force transfer of particles
          diffusing to a boundary surface.   The Peclet number,  usually
          given N   = v d/D, is the characteristic ratio for convec-
          tive diffusion.  It represents the ratio of the rate  of
          transport by fluid motion to the rate of transport by molec-
          ular processes.  We will refer to the ratio "UF/V " as  the
          Flux Deposition number, N™.

          Significance of Flux Deposition Number

               To show the relationship of N   to Np , we can start
          with consideration of a diffusional procesl where the flux
          is given by:
                    NA = ~D    = UDM CA (9mol/cm2-sec)             (28)


                                     46
A. r. T. Inc.                                 POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
         where:

         U-™,  = deposition  velocity for mass transfer,  cm/sec
          DM
         c.   = concentration  of  transferring component ("A"),
               gmol/cm3

              If  the  concentration gradient is linear, there is no
         equilibrium  pressure above the interface (i.e.,  no "back-
         pressure") and  the boundary layer thickness,  "6", is pro-
         portional  to drop diameter,  "d":


                         dc    co   c                         (29)
                          dx      6      d


         Therefore:

                                      v
                 3 «  UDM,  and,  Npe  a  ^°-


         or:

                          Npe  =  -°-, if 6 =  d                    (30)


              The  form  of  the modificed Peclet number in equation (30)
         will apply to  any deposition process for which the flux can
         be described by  equation  (28).   Thus we should expect to
         find comparable  mathematical models and empirical correlations
         for deposition with  diffusion and  that with flux forces.

              It is instructive to  look for an estimate of the quan-
         titative  significance  of Np^.  Deposition rate on a sphere by
         any flux  or  diffusional mechanism  will follow the same form
         as the following equation

                    (4irrd2)NF  =  uFn  (4Tird2),  no./sec             (31)


         The rate at which particles approach the sphere is given by:

                    vQn  (nrd2) = convection  rate to sphere         (32)

         The fraction of  the  particles convected to the sphere which
         deposit on the sphere  (i.e., the collection  efficiency)  is:

                                      47

A. P. T. InC.                                 POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
                        4TT  r,2  up n   4 u
                    0  = __a_Z_P. —* = 4 NpD               (33,
                        irr*  v  c        v
                             o           o

               Thus,  for a  single drop if NFD is greater than 0.1,
          collection efficiency is good.   If Np[) number is less than
          0.01,  collection  efficiency will be only a few percent or
          less.   In situations  where there is collection on a series
          of  drops,  the efficiency of each drop need not be so high
          in  order  to  obtain a  suitable overall efficiency for the
          sequence  of  drop  contacts.  For example, the particle
          penetration  of a  counter-current spray tower with a water/
          gas ratio of 1.0J,/m3, drops 0.04 cm diameter, a height of
          3 meters,  and a gas velocity of 120 cm/sec and spray losses
          on  the walls equal to 80% of the inlet water is approximately:


                             Pt  = exp(-lO.On)                    (34)

               If a penetration of 0.1 (i.e. 90% collection efficiency)
          is  desired,  the collection efficiency for a single drop
          would have to be  0.23 (23%).  If the single sphere efficiency
          were 0.023,  the overall penetration would be about 0.8.
          Thus the  approximate  NFD range of 0.1 to 0.01 to distinguish
          good from poor performance for a single sphere is also about
          the right magnitude for a spray scrubber.

          SOLUTION  OF  THE EQUATIONS OF MOTION

               Let  us  now take  up the solution of equations (26) and
          (27)  for  particle collection under the influence of flux
          forces. The  velocity  and trajectory of the particle is
          determined by the resultant force and the air flow field
          around the particle.   In order to formulate this situation
          mathematically, the following assumptions are made:

               1.  All the  assumptions made by Langmuir and Blodgett
          (1946)  for collection due to inertia alone apply also in
          this case.

               2.  The time scale for the interaction between a single
          particle  and a single drop is small compared to the time
          scale of  changes  in the drop and gas temperature and humidity.
          Thus the  temperature  difference between the gas and the drop
          is  assumed constant.   This assumption was checked numeri-
          cally and found correct within 1%.

               3.  All the  resistance to heat transfer is assumed to
          be  in the thermal boundary layer.  The boundary layer thick-
          ness is estimated from the heat transfer coefficient ob-
          tained from the Whitaker (1972) equation:

                                      48
A. P. T. InC.                                 POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
                   NNU  =  2  +  °'4  V  "Pr°"

         A heat balance  around the drop gives:

                     kr
         Q =  h AT  =  -  AT                                        (96)
                                        h'  dd   dd
          Substituting  for  "h"  in N    = - -
          Other  equations  for  the  boundary layer thickness yield
          other  values  for "6".  These  values range from -0.5 r^
          to  -2  r^.

              4.  The  thermal boundary layer was assumed to be of
          a constant  thickness in  the  front half of the sphere.  The
          thickness varies with the  Reynolds number of the sphere
          as  can be seen from  the  previous assumption.  This is an
          approximation,  since the velocity boundary layer thickness
          approximately doubles from the stagnation point to an angle
          of  90°,  Schlichting  (1960).

              5.  The  temperature gradient in the thermal boundary
          layer  is assumed constant.   This was calculated to be ap-
          proximately the  case for a flat plate and N   * 0.6,
          Schlichting (1960).

              6.  The  thermophoretic  force acts on the particle only
          inside the  thermal boundary  layer.

              7.  Collection  in the back of the drop was neglected.
          This assumption  is correct for low drop Reynolds numbers.

              8.  G_ is constant  inside the boundary layer.
                   r

          Integration

              Equations (26)  and  (27)  were integrated for boundary
          conditions  which are consistent with the assumptions given
          above.   The boundary conditions are:
          X  = -»
          •V    «„
          Y  = Y
               o
          dX  _ ,
             = 0
          dt
at t = 0                                   (37)
                                     49


A. P. T. InC.                                 POST OFFICE BOX 71. RIVERSIDE, CA 92502

-------
                        G  = 0  for (X2  + Y2)>(1 + 6)
                         r
                        G-, = G-, for  (X2  + Y2)*(l 4- 6)2
                         t    r


                 These  equations were solved numerically on a digital
            computer  using a Runge-Kutta  method.   The solution was
            started at  x = -5.0.  Figure  8 is a plot of target ef-
            ficiency  vs.  "K " with Np^ as a parameter for NRgd =9.6.

                 To illustrate the significance of Figure 8, weQmight
            consider  a  case with initial  gas temperature of 100 C
            and  initial liquid temperature of 20  C,  for which we com-
            pute upT  =0.03 cm/sec.   It follows from Figure 8 that
            under these conditions the effect of  thermophoresis is
            very small.  If the drop diameter were,  say 200 urn, its
            settling  velocity would be about 74 cm/sec, and NFD would
            be about  2.5x10  , which is indicative of very poor
            collection  efficiency due to  flux force.
            DIFFUSIOPHORESIS  AND INERTIA

                 The  solutions  of equations (26)  and (27)  and the re-
            sulting Figure  8  are valid  for the case of diffusiophoresis
            and inertia  and the sum of  thermophoresis and  inertia.
            Figure  8  is  also  a  representation  of  collection efficiency
            when any  number of  forces are  acting  simultaneously;
            given that deposition velocity, upl is  defined as:


                 Up=u_+u+ any other deposition velocity    (38)


            and:
                                     u^
                                                                   (39)
                 Three  dimensionless  numbers,  K ,  N  ,  and NRe deter-
            mine  the  collection efficiency  of  particles by drops when
            the particles  are  subjected  to  the simultaneous action of
            inertia and flux  forces.   We have  reduced the role of the
            Reynolds  number by assuming  potential  flow around the drop.

                 From Figures  8 and 9, which is a  cross plot of Figure
            8, several  conclusions can be drawn:

                 1.   Collection efficiency  increases with the increase
            in "K " and "N " .
                 p        FD

                                     50
A. P. T. InC.                                POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
 I?
03


g
5

S
o
m
to
ro
en
s
                >~

                2

                i—i
                U
                i—i
                U-
                O
                M
                H
                U
                O
                U
                U4
                J
                O,
                O
                CA
                Q
                      5x10
                           - 2
             10'1                       1

                INERT IAL  PARAMETER, K
                                     P


Figure 8 - Efficiency  of single drop versus inertia
           parameter at  NRgd =9.6 with M   as parameter
10

-------
                                                                         1.2
             Figure  9 - Efficiency  of  a  single drop, n, versus N™ with
                        K  as a parameter (NR , = 9.6).
                                       52
A. P. T. Inc.
POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
              2.  The higher "K " is  (and the larger the particles),
         the smaller is the effect of the flux force on collection
         efficiency.

              3.  At low "Kp" values collection efficiency increases
         with decrease in "Kp".  This may be attributed to the  lower
         stream velocities and the longer time in which the particle
         is subjected to the flux forces.

              4.  Collection efficiency can increase above 100%
         for high values of "UD/U".  This stems from the fact that
         particles can be moved to the drop from positions outside
         the projected frontal area in the gas stream.

              We can make some general observations about the probable
         importance of flux force deposition mechanisms by estimating
         ranges of N D and then interpreting these values in the  light
         of Figures 8 and 9.  For example:

              1.  For poorly conducting particles in the fine size
         range and a temperature range of 0°C < AT < 80°C, flux
         deposition number is 0.04 > N D > 0.0007.  This means  that
         in this range thermophoresis is not important.

              2.  Diffusiophoresis can be very important for fine
         particles because with a water vapor partial pressure  range
         of 0.13 < pv < 0.65 atmospheres, the flux deposition number
         range is 2.5 > !!„„, > 0.01.
                         r D

              Collection efficiency as shown in Figures 8 and 9 is
         for NRe =9.6 which is the NRe of a free-falling 200 ym  dia-
         meter water drop.  The boundary thickness, as already  men-
         tioned, is a function of the Reynolds number and for the
         case presented in Figure 8 and 9 it was taken as 0.56(r^).
         Since these calculations are for potential flow, a change
         in the Reynolds number will change only the thickness  of
         the boundary layer.  To see the effect of changing the
         boundary layer thickness on the collection efficiency, Table
         II shows the change in

                      n vs. 6 = —, at K  = 1.1 and G,, = 0.1.
                                rxi      P            F
                                     53


A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
                               TABLE II

                 Collection Efficiency of Particles by

                  Drops for K  = 1.1 and Gp = 0.1 at

                   Various Boundary Layer Thickness
6
n
0
0.5
0.56
0.58
1
0.62
2.2
0.69
               We see that the thickness of the boundary layer has a
          small but noticeable effect on collection efficiency.


          PARTICLE GROWTH

               When humid hot air is passed through a spray column,
          cooling and dehumidification take place.   Transfer of heat
          and vapor from the gas to the drops occurs and particles
          are collected by the drops due to the combined effect of
          inertia, diffusiophoresis, and thermophoresis.  At the
          same time, if the saturation ratio is high enough, particle
          growth starts.   As shown in the previous  section, diffusio-
          phoresis is significant for the temperature range of
          interest, but thermophoresis can be neglected.

               A mathematical model for particle collection by dif-
          fusiophoresis and inertial impaction in a spray column
          was developed and solved as follows:

               1.  Computations were made for small increments of the
          spray column; ranging from 0.1 to 2.0 cm  high.  After the
          increment height was chosen the conditions calculated and
          compared to the conditions calculated from one half of
          this height.  If the difference was not significant the
          original increment height was used for further calculations.

               2.  Gas humidity and temperature, liquid temperature,
          and interfacial humidity and temperature  were calculated
          for each section in the method presented  in Brown (1950).
          These conditions were assumed constant throughout the
          section,  "hj."  an<* "^G"' tne neat transfer coefficients
          inside the drop and on the gas side, were estimated from
          penetration theory and equation (35).

               3.  Particle growth was computed by  means of Fletcher's
          equation, modified by the use of a factor "a", so that:
A. P. T. Inc.
                                     54
POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
                              Dr P<~ \

                              -^  
-------
        = gas flow rate, m3/sec

          terminal settling velocity of drop, cm/sec

          linear velocity of gas, cm/sec

          height of increment, cm

     n- = collection efficiency of drop under increment
      1   conditions, fractional

The overall penetration of the increments where cooling and
dehumidification take place was calculated from
              Ut =

              UG =

              AZ =
                  n

            Pt  = n Pt. = exp -
              C        -L
                                   4 QG  rd  (ut
                                                       n
                                                        (42)
         Penetration  for the rest of  the  column,  where  heat and mass
         transfer are small and affect  collection efficiency only
         slightly, was calculated from
                Pt  = exp -
                     3 QT u.  (Z-n AZ)
                        LJ   t

                     4 QG rd  (ut - UQ)
(43)
         where  n = nn-
               The  overall  collection  efficiency is :

                        E = 1  -  Pt  =  1 -  (Ptc Ptn)
                                                        (44)
               Figure  10  shows  Ee  = (1  -  Ptg)  and ET = (1 - PtT)  for
          a  1  meter  countercurrent spray  column as a function of "a"
          for  0.5  urn diameter particles of density 1 g/cm3 and inlet
          concentration of 106/cm3.  The  gas entered at 35°C and left
          at 33.2°C, while the  liquid entered at 20°C and left at
          25.3°C.  Gas inlet humidity,  "Y " was 0.155 g/g, corresponding
          to a saturation ratio of about  3.5 at 35°C; and the outlet
          was  0.077  g/g,  corresponding  to a saturation ratio of about
          2.1. Liquid  to  gas flow  ratio was 58,/m3, drop size 0.05 cm
          diameter,  and linear  gas velocity 60 cm/sec.
                                     56
A. P. T. Inc.
                                  POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
                                             overall  collection  —
                                             efficiency

                                             collection  efficiency
                                             in  section  where flux"
                                             forces  are  active
                         0.2        0.4       0.6        0.8

                           FRACTION OF PARTICLE GROWTH, a
                       1.0
              Figure  10  -  Collection efficiency for 0.5 ym diameter
                          particles in a 1 meter spray column
                                      57
A. P. T. Inc.
POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
              It can be seen that with no particle growth,  a  col-
          lection efficiency of 33% was calculated, while  for  full
          particle growth collection efficiency rose  to  92%.   It
          should be borne in mind that there are several points  of
          uncertainty in the computation so that one  should  view
          these results as being only approximate.  For  one  thing,
          the  liquid to gas ratio is rather high so the  wall losses
          and  agglomeration of drops would cause the  active  spray
          holdup and its efficiency to be less than assumed.   If,
          as might realistically be assumed, half of  the spray were
          lost to the walls, the efficiency would drop from  92%
          to 74%.

              Reducing the degree of supersaturation in the incoming
          gas  by raising its temperature, reduces collection effi-
          ciency markedly even if the humidity is not changed.   At
          the  time these calculations were made it seemed  reasonable
          to assume that as high a saturation ratio as 3.5 could be
          attained in the inlet gas.  Experimental experience  ob-
          tained later in this program leads us to have  serious
          doubt that saturation ratios much higher than  1.0  can  be
          attained in realistically dirty effluent gas streams.  For
          these and other reasons relating to computational  techniques,
          the  present spray model should be considered as  a  first
          exploratory step.
                                     58


A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
                      MATHEMATICAL MODEL, PLATES


              Plate type  scrubbers are  represented  to  a  first approxi-
         mation by the unit mechamism of transfer from bubbles.   Thus,
         while the model  which  is described  in  the  following section
         is defined in the specific  terms of a  sieve plate,  it is
         suitable for expressing the particle collection behavior
         of scrubbers in  which  bubble formation and rise through liquid
         are the principle mode of gas/liquid contacting.   In order to
         compute the simultaneous effects of transport and  particle
         deposition phenomena in plate  type  scrubbers, a set of  equa-
         tions was developed as described below.  The  final set  of
         equations describes the following phenomena:

              1.  Heat and mass transfer between bubbles and liquid.

              2.  Heat and mass transfer between bubbles and
                  particles suspended in bubbles.

              3.  Particle growth due to condensation.

              4.  Particle deposition by:

                  a.  Diffusiophoresis

                  b.  Thermophoresis

                  c.  Centrifugation during  bubble  rise

              The following general  assumptions were made:

              1.  Gas bubbles are spherical,  constant  diameter
          (d, =0.4 cm, usually), and perfectly  mixed internally
          (except for the  interfaces).

              2.  Gas properties are as for  air and water vapor.

              3.  Foam density  is constant throughout  the foam layer
         on the plate.

              4.  Liquid  bulk temperature is constant  throughout the
         foam, although the liquid-bubble interface temperature  can
         vary.

              5.  Particles are mono-disperse,  wettable, insoluble
         spheres.

              6.  Condensation  on particles  can occur  whenever the
         saturation ratio is 1.0 or  larger.


                                     59

A. P. T. InC                                POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
                7.  All  the  particles  are subjected to condensation
          and growth.
          PARTICLE GROWTH
                The rate of  change  of  particle radius is given by a
          mass  balance,
                        %  _ "'PC  'PG -  PPJ'   /,»_,
                        at          PM           Uec'
          Where :
                  2 DQ  P     /   gmol    \
          k" G  = - ( - 1  = mass transfer coefficient
            p    RT,, d  p_M \cm2-sec-atm/    particle to gas    ..,.
                   tj  P DM                                      [ *t O )
          PG = water vapor  partial pressure  in bulk  of gas bubble,  atra
          P  =  total pressure, atm
          p_M= mean partial pressure  of  non- transferring gas, atm
          d  = particle dia, cm
          TG = gas bulk temperature,  °K
          p  = molar density of water, gmol/cm3

          For air and water:
                                       1 75
                                   T
                        DG - °-256
          so
                          2.85xlO"7(TG)°-75
                   '    =
          It can be shown that "PDi" is close  to  "pG"  so that
          PBM =  d'Pg) for tne ca^e of particle growth.
                                     60
A. r. I. Inc.                                 POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
               Particle growth over a finite period is given by the
          integration of equation (45) after substitution of equation
          (46).
                             =

                                 RTG PM PBM


         where At = time of growth, sec.

              In our present computational programs, equation  (48)
         is not applied when p .  > pQ, because this would imply
         particle evaporation. P It would be possible to write a
         computer program which would allow the evaporation of only
         the water film on a particle, but this has not been necessary
         thus far.

              The vapor-liquid equilibrium relationship for water can
         be approximated within a few percent by:


              p „ = exp  (13.64 - 5-lxl°3 )  (atm)                 (49)
         p   = water vapor partial pressure in equlibrium at T °K

              Particle temperature can be computed from an energy
         balance:

                         /p  C   r \  d T .     ,
         h G(T i - TG) + (-E—2E—£) 	E- = k    LM  (pG - p .)   (50)
                         \   3     /  dt

         Where:

         h     2k a 7.5X10"5 /	cal	\ = particle to gas heat
          P     p      rp    \cm2-sec-°K/   transfer coefficient
         C   = heat capacity of particle, cal/g-°K


         k   = thermal conductivity of gas, cal/cm2-sec-°K/cm

         and

         LM - 10 "* [ - 5-J = latent heat of vaporization for water
                                     61

A. r. T. Inc.                                POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
              The second term on the left side of  equation (50)  re-
         presents the effect of particle heat capacity  and it is
         neglected in the computation.  It can be  shown that the
         product of particle mass and heat capacity  is  a very small
         percentage  (generally less than 1%) of the  product of gas
         mass and heat capacity.  Thus, for a given  change in tem-
         perature, most of the energy goes into increasing gas
         enthalpy.

              An explicit solution for particle surface temperature
         was provided by an alternate solution to  equation (50).
         This solution involves the approximation  that  the vapor-
         liquid equilibrium relationship is linear over a small  range
         of temperature or vapor pressure (see equation 54).   If we
         define a temperature, T*, which is the saturation (equili-
         brium) temperature corresponding to the partial pressure of
         water in the bulk gas, pr, we can show that:
                       Pr. * Pr
                          .
                        pi
                                          = C,
                                (52)
k pG LM
         and
T— T —
• *• i-> ~
pi G
r i 1
C2
1 + —
L m J
(T* - TQ)
                                                                  (53)
         Where:
                                  I"
         m = slope of eq. line =   (0.042) p_ + 5x10

         and
                     /atm
                     \°K
(54)
                         T* =
                                 5,100
                              13.64-ln p
                                (55)
                                        'G
                          .-"*
         Because C-^  4x10"   (atm/°K), the particle temperature,  T .,
         will be within a few degrees of T*, rather than  being  at ^>1
         approximately the gas temperature, TG.  For example, at
         PG = 0.1, m = 4.7xlO~3and C2/m = O.OH5.  Thus,
                       T .  - Tr * 0.925  (T* - T_)
                        pi    t>                G
                                     62
A. P. T. Inc.
                                            POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
         HEAT AND MASS TRANSFER  IN  BUBBLES

              The overall energy balance  for  a  bubble interface is
         given by


         k'bG ab LM(PG - PLi)AP  dz  =


             hbL Sb(TLi - TL)Ap  dZ+hbG ab(TLi -  TG)Ap dZ          (56)


         where :

                0 P               ^2
         . '      G     1.13P  /DG\      gmol               .     ,
         k bG = - - - j — ]    — 2 -   =  mass  transfer
                p_M   RT_ pOM \ 0D /    cm2-sec-atm     coefficient,  bubble
                 DM.     vj  BM x   o '                      . .   . ,
                                                      to liquid
                                                                  (57)

         a,   = surface area of bubbles/unit volume  of foam,  cm2 /cm3

         A   = area of plate, cm2
         0R  = surface renewal time, sec

                    B
         TL  = temperature of liquid bulk,  °K

              For a bubble diameter = 0.4 cm and bubble  rise  velocity
         20 cm/sec, the time to rise 1 diameter = 0.02 sec.   If  "0B"
         is evaluated as 0.02 sec for the penetration theory  models
         represented by equations (57) and  (58), the coefficients  for
         air and water properties are as follows:
         kG = l.exlO"*  gmol/Cm2-sec-atm                          (59)


         hbG ~ °-022  (TG*    cal/cm2-sec-°K                       (60)

              It should be noted that boundary layer theory would  pre-
         dict coefficients about half these and that the computation
         of dif fusiophoresis deposition from equation  (59) yields  a
         very high velocity.  Thus, we are cautioned to view  equations
                                      63


A. P. T. InC.                                 POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
         (59) and  (60) as rough approximations which may be used until
         experimental data are available.

              The heat transfer coefficient  from the interface to the
         bulk liquid is approximated through penetration theory as:
               'bL = 2 U ^ kX)   =  °-31  f	~	^             (61)
              Combining equations  (55),  (50),  (60),  and (61)  yields:


         1.6(pr - pT.
                       - 0.31  (T_. - TT)  -          (T_.  - T)  = 0
                                Ll    L        ^      Ll
           (1 - PG/2)                         TG                    (62)


         In equation  (62), p   is approximated  as  (1 -  PG/2)  because
         p. .  is usually very  low.

              Condensation of water  from  the gas bubble will  cause
         both a decrease in gas flow rate and composition,  as given
         by:


                        G dp_    dM     gmol H90 vapor
            -d(G y) = -- Z- = — H - = -      (63)
                         (1~PG)   dt   cm2  of plate  area-sec


         where G = molal gas flow rate per unit area of plate,
         gmol/cm2-sec

                   y = mol fraction  water vapor

              The mass of water vapor transferred is given  by:


                                      gmol H?0  vapor
                  dNT = dM  + dMb =  — - * -            (64)
                          F          cm  of plate area


              The transfer rate to particles is,

               dM    4Ti   (r* - r,3)
                                            V   dZ>  (^         (65)
             P dt      3(18) At        p



                                      64



A. P. T. Inc.                                 POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
               where:

               n  = number concentration of particles in gas, no./cm

                t = time interval,  sec

               V.  = volume fraction bubbles in froth, cm3/cm3
                b

          and the mass  transfer rate to the bubble interface is,

                  dMb
               »  /nr- = k Kr «H (Pr - Pf) A  dZ (^±)           (66)
                p dt      bG  b vrG   rLi   p    \ sec/


          where:   dZ = height of bubble rise in time "dt"

               In the difference calculation procedure used for the
          machine computations, equations (64), (65), and  (66) are
          combined to give:
                                                                  (65)

          The change of gas temperature due to sensible heat transfer
          is  defined by an energy balance as follows:


           G CPM ATG = hbG ab ^Li'V AZ + 6Mpx10' (  a**  )    <66>
          or

                      AQ.  + AQ,
                                                                  (67)
                         A0
          Where C M = molar heat capacity of gas, cal/mol-°K

          and  AQ =  heat  transferred in time A0, per cm2 of plate  area
          The  particle area per cm3 foam is:
          ap = np (41T rp2)  (1'F)  = np Vb (4 «P2)  c              (68}
                                     65


A. P. T. InC.                                 POST OFFICE BOX 71. RIVERSIDE. CA 92502

-------
         The bubble surface area per  cm3  foam is  assumed constant
         and given by:





         The change in gas flow rate  is approximated  by:


                       G = ^  /gaolJV                            Q)
                           At  \cm2


         PARTICLE DEPOSITION

              Particle deposition by  Brownian diffusion  is  not very
         effective on particles in the >0.1 pro dia. size  range,  and
         was not included, although it could  have been included with
         little difficulty.  The collection which occurs  during the
         period of bubble formation can be predicted  separately by
         means of equation (71) and it was not included  in  this
         model.
                    PtF = exp -  (40 F£2  Kp)                     (71)
         where
         F    = foam density, volume fraction  liquid

         Ptp  = particle penetration for collection during  bubble
                formation, fraction

         K    = inertial impaction parameter
          P   " 9 »G dh

         d    = aerodynamic diameter of particle
              It should be noted that collection  by  impaction during
         bubble formation is not very important for  one  tray and
                                     66

A. P. T. InC.                                 POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
          particle diameter smaller than 1 ym.  But if the particles
          grow during their stay in the first tray, this mechanism
          will become very important in collecting the particles
          on the second tray and following trays.

               For the present, we are looking at what happens after
          the initial collection by impaction.  The total particle
          flux from the gas to the liquid is defined as the sum of
          fluxes due to diffusiophoresis, thermophoresis, and centri-
          fugation.
M  - M  4. M  -u w  /no. particles\
Ns ~ ND + NT + NC f 	1	
                  1   cm -sec
                                                  J
          and the flux is related to the deposition velocity and
          particle concentration by:
                        Ns = ups np                               (73)


          For a spherical bubble of radius "r. ", the rate of change of
          particle concentration is:
                    dn~
                    _J£ = /£_\N  =l-^}u_                     (74)

                    dt    \rb/

          Where,

            u   = u   + u _ + UG = sum of deposition velocities  (75)


          and the penetration for a period = At:

                               3
                   Pt = exp -
              =r- (u  )  At                        (76)
                               rb
          Diffusiophoretic deposition velocity is:
        •/M, D
  _ __ ±  G _      /    v

                                 (sec/
                                              dy  /cm
                                              dr  (sec
                                     67


A. P. T. InC                                 POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
          UpD = Cl °G (l^y)
          where M,  = molecular weight of water, g/mol

               Mp = molecular weight of non-transferring  gas,  g/mol
          and for air and water it can be shown that:
          C,  - 0.85 (dimensionless) over the range  from y = 0.1 to 0.5


               The composition gradient can be related  to the mass
          transfer coefficient as:
                      dy _dp _ RTG k bG(PG-PLi)

                      dr   dr         D,
          Therefore, substitution from equations  (57),  (59)  and (78)
          into (77) yields:
                                     _
                u _ * 0.85 RT,, (1.6x10  )   fa  ljl -              (79)
                 P°          G            (1 - PG/2)

                             (p  - p . ) T
                u _ = 0.0112 — ^ - — — -  (cm/sec)                (80)
                 P°          (1 - PG/2)


          Thermophoretic deposition velocity is :
                    u   - -z'    G  dT
                    UPT ~  z  p-f  55F
          and the gradient is estimated from:


                              (TG - TLi> (EH)
dTG   hbG
                    dx     ]


          Using Brock's value for Z1 = 0.25 C', we have  for  air and
          water vapor:

                                      68

A. P. T. InC.                                 POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
               u T *  1.35x10'*  (C'T,^2  AT)  (cm/sec)              (83)
          Where:
                   AT- = (Tr - T_.) (°K)                            (84)
                     G     la    LI
               C1  =  (l  +  -*  — )  = Cunningham slip correction   (85)
                    \        P

          Centrifugal deposition velocity is:

                          2 r 2 p  C1  v. 2
                    u _ -- * - P - E. = K                      (86)
                     P       9y rb

          and  for  rfa =  0.2 cm, p =1.0 g/cm3, and vt = vb = 20 cm/sec ,


                          rn C<
                    u _ = -E= - (7.4xl08) (cm/sec)               (87)
          Note that "u  "  is dependent upon "v. 2" and is quite sensi-
          tive to the issumed values of "v, " and "r. ".


          PREDICTIONS

               A computer  program was written to solve the mathematical
          model comprising the equations given above, and some pre-
          dictions of performance were computed.  As a representative
          case, we took bubble diameter to be 0.4 cm, foam density 0.5
          (volume fraction gas) ,  and bubble rise velocity to be 20
          cm/sec.  The first computations were for the following
          conditions:

               particle radius =  0.5 ym

               particle concentration = IxlO6 particles/cm3

               gas molar flow rate = 0.003 gmol/cmz-sec

               foam height = 8.0  cm
                                     69



A. P. T. InC.                                 POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
               The predicted penetration for these conditions was much
          lower than experimental data such as Rosen and Kostin (see
          Figure 4.6.4-7 in Scrubber Handbook).   Because of this and
          the uncertainties about the transfer coefficients, the heat
          and mass transfer coefficients for the gas phase, bubble to
          liquid transfer were decreased to 1/2 the values given by
          equations (59)  and (60).   The lower values are in line with
          the predictions given by boundary layer theory and the
          penetration resulting from their use are of a more realistic
          magnitude.  The predictions which are discussed in the re-
          mainder of this section were made on the basis of 1/2 the
          values given by equations (59)  and (60)  until experimental
          data was available.

               As will be seen later,  under experimental work, the
          liquid phase heat transfer coefficient and the interfacial
          area for transfer seem to be quite different than those
          assumed.  The approach we have taken is to devise the best
          model possible based on existing information, devise and
          perform experiments  to test the predictions of the model,
          and then revise the  model to fit experimental findings.
          Because of that,  we  will  describe only the general features
          of the predictions based  on 1/2 the simple penetration
          theory transfer coefficients in this section.

               It is enlightening to explore the predicted effects
          of process variables as shown in Figures 11 through 15
          and Table III.   Table III summarizes the conditions and
          results for 20 computer runs.  Bear in mind that while the
          general trends are in line with experimental results, the
          predictions give  lower penetrations than experimentally
          observed,  apparently because of the liquid phase heat
          transfer coefficient used.   In all of these runs the total
          foam height was 6 cm;  so  the bubble residence time was
          0. 3 sec.

               In all figures,  the  "Y" axis ranges from 0. to 1.0 and
          the "X" axis denotes time,  with each increment of the
          digital computation,  At = 0.02 sec.  In order to fit all
          four curves into  the range from 0. to 1.0, we plotted the
          following:
               S'=   = - -               (88)
                        2[exp  (13.64)  -  5,100/T-)]
                                     70
A. P. T. InC.                                POST OFFICE BOX 71 .  RIVERSIDE. CA. 92502

-------
    w
    nj
    <
    u
    p
    w
    o
    2
S/2
              n  = 3.5xl03
                                          1.0
                                          0.5 ._
                  n  =  3. 5x103
                  0.1      0.2


                    TIME, sec
   0.3
0.1     0.2


  TIME, sec
0.3
                    Figure 11


               Computed prediction

             Run #3 for sieve plate
A. P. T. Inc.
                                      71
                         Figure  12


              Prediction of  effect of water

              vapor concentration  on particle

              radius.
            POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
                       0.1      0.2
                         TIME,  sec
0.3    0      0.1      0.2      0.3
                 TIME,  sec
                          Figure  13
                   Computed  predictions
                           Run  #  16
                  Figure 14
           Computed predictions
                   Run I 18
                                      72
A. P. T. Inc.
    POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
                             0.1 -
                                        0.1       0.2     0.3
                                           TIME,  sec

                                           Figure 15
                                      Computed predictions
                                              Run  ti20
A. P. T. Inc.
                                       73
POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
                                Table III

                SUMMARY  OF  PLATE COMPUTATIONS WITH PENETRATION
                         THEORY PREDICTED COEFFICIENTS
Run
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
TG
°C
30
45
45
30
45
45
45
45
45
45
45
45
30
30
45
60
60
60
60
60
PG
Atm
0.093
0.093
0.093
0.093
0.093
0.09
0.08
0.07
0.03
0.025
0.093
0.093
0.093
0.093
0.08
0.19
0.19
0.19
0.19
0.19
TL
°C
19
19
19
19
21
21
21
21
21
21
21
19
19
19
21
20
20
20
20
20
Initial
r . ym
P
0.33
0.33
0.33
0.33
0.33
0.33
0.33
0.33
0.33
0.33
0.33
0.33
0.33
0.33
0.33
0.25
0.25
0.25
0.25
0.25
#/cm3
3,500
3,500
3,500
3,500
3,500
3,500
3,500
3,500
3,500
3,500
3,500
10s
10s
106
10s
10 3
107
]06
10s
108
G
gmol/
cm2 -sec
0.003
0.003
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.009
0.006
0.006
0.006
0.006
0.003
0.003
0.003
0.003
0.003
Final
r , ym
P
4.7
3.0
2.8
6.6
2.55
2.35
1.3
0.36
0.33
0.33
2.55
2.55
4.7
3.3
1.32
4.2
1.4
3.1
4.0
0.68
Final
Pt
0.1
0.38
0.37
0.09
0.39
0.43
0.55
0.62
0.83
0.85
0.39
0.38
0.09
0.23
0.55
0.13
0.28
0.19
0.15
0.31
                                       74
A. P. T. Inc.
POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
               T,  = TG " TL   present AT                          (89)
                    T_  - TT   original AT
                     (jO    Li
               Pt = n /n   = Present particle cone.
                     p  po   original particle cone.


                1  _ _pg _ original particle radius
               rp    r    present particle radius


               Figure 11 shows the predicted results for computer run
          #3,  in which the air starts out saturated at 45°C and contacts
          water at a constant temperature of 19°C.  The saturation ratio
          rises above 1.0 so there is particle growth, as shown by the
          curve for "r1" (which decreases as particle size increases).
          Particle penetration decreases almost linearly with time (or
          bubble rise distance),  as does air temperature.

               Figure 12 shows only the predicted particle time para-
          meter for runs #3, 6, 7, and 8.  It can be seen that as the
          initial air saturation decreases (all other factors being
          the same for runs #6, 7, and 8) the amount of condensation
          decreases.  In run #8 there is a small amount of particle
          growth (about 5%) at the end of the 0.3 sec (corresponding
          to 6 cm of rise at 20 cm/sec for a 0.4 cm dia. bubble).

               Figures 13,  14, and 15 show the predictions for runs
          16,  18,  and 20, respectively.  The only output parameter
          changing in these runs is particle concentration, which is
          103 particles/cm3 for Figure 13, 106 for Figure 14, and 108
          for Figure 15.  One can see the effect of increasing particle
          concentration as it decreases the saturation ratio and
          particle growth,  while increasing the outlet air temperature
          and particle penetration.

               To get some perspective in accustomed engineering units,
          one might note that 106 particles/cm3 corresponds to about
          0.137 g/m3 (0.06 grains/ft3) if particle density is
          2.0 g/cm3 and particle diameter is 0.5 ym.  This would
          represent a few percent of the particulate mass if the
          concentration from a source were 2.3 to 4.6 g/m3 (1 or 2
          grains/ft3), so it would be on the order of a realistic
          fine particulate concentration.  Obviously, it would be
          advantageous to remove as much of the larger particles as
                                     75


A. P. T. InC.                                POST OFFICE BOX 71,  RIVERSIDE, CA. 92502

-------
          possible before raising the saturation ratio above 1.0.
          This last statement will be even more important for higher
          fine particle concentrations,  such as have been reported
          for ferroalloy furnace emissions.

               The plate scrubber model  and the computation procedure
          described in this section proved to be of proper form to
          describe FF/C scrubbing to the extent that theoretical and
          empirical information is available.  The various coefficients
          and quantities used may be adjusted to fit new experimental
          data as it becomes available.   As discussed previously, the
          development of a useful and reliable design method is an
          iterative procedure which can  continue to upgrade the method
          as  well as to increase our perception of experimental
          reality.
                                     76



A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
                    MATHEMATICAL MODEL, IMPINGING JET


               The approach to calculating the collection efficiency
          of particles from a jet impinging on a plate was similar
          to that taken for predicting collection efficiency by a
          drop.  The particle trajectory was calculated from its
          initial conditions and the forces acting on it.  In de-
          riving the equations of motion we followed the method out-
          lined by Mercer and Chow (1968)  adding an additional term
          to the equation which described the flux force.  To
          simplify the solution we assumed a constant flux force
          gradient in the space between the jet and the impact ion
          plate.

               A force balance leads to the following set of differen-
          tial equations in dimensionless form, comparable to equations
          (26)  and (27), where vapor gradient exists in the jet space:
                       -  - (vx - ux)  + KD                        (92)
                   dt2    K    *    X     U dX
                          P
                   — = — (VY - uy)  - Kn 21                     (93)
                   dt2   K         x       dY


          Where
                        9 yr (0.85) D
                   KD = 	«	                      (94)

                        2 pp V uo2  u - y>

          The  same  equations  will result when a thermal gradient is
          present,  with "KD"  replaced  by "KT"  and "dy" by "dTG".

              The  flow field assumed  by Mercer and Chow is


                      X "*        '  -(1 + AY)  < Y < (1 + AY)
                      3
                           for

                                   -3  < X < 0                     (95)
                                     77

A. r. T. Inc.                                 POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
                                       -(1 +  AY)  >  Y > (1 + AY)


               vx = 0         for



                                       -3 < X <  0





          The boundary conditions are:





                    vv = uv = 1 for X = -0
                     A    A
                    uy = 0 for t = 0
          The flux force is defined by:
             = const, and    = 0 for -d +  AY)  <  Y < (1 + AY)     (97)

          dX              dY
                £2 = dY. = o for -(1 + AY)  > Y  >  (1  + AY)           (98)

                dX   dY
          Where
               Y =
               v  = —
                X   V
"•   2X
X = — = diraensionless distance

    W


    2Y
  = — = dimensionless distance

    W


     v

   = — = dimensionless  velocity

      o



     vv
v  = — = dimensionless  velocity




~   2t vu    ,.     •
t = —rj—0- = dimensionless  time




W = jet width, cm



s = distance  from  jet nozzle to  plate,




    2s
g = — = dimensionless }et spacing
                                                       cm
                   W
                                      78



A. P. T. IDC                                 POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
                Y =  [Y  I, the ordinate  of a given streamline

                          at X = -8
                y = mol fraction water  vapor

          Note that the factor  0.85, which  accounts for molecular
          weight gradient effect, is as discussed previously, and
 2 P.
                                                                  (99)
                                          PG  TG
               The solution to equations  (92)  and  (93)  with the
          boundary conditions given by equation  (95)  in the region
          -(1 + AY) < Y <  (1 + AY) is:
          X = exp{-t/2K)
(K -1)  cos 1±- +
  U        2K
                sin
                                                               2K
                                                                  (100)
          where y =
fl) cxp h6-1**]
£' 2K
P J
0.5
4 K -$1

'
B J
e-1 	 [(£+!)
[2Kp
0.5
[3+4 K 1
P
1 s
i K dy
K P ^
D KD dX
                                                                  (101)
          Outside of this region:
              X - (X)1+y  = K
                        o    c
    dX
    dt
1 - exp
             K
(102)
               From these solutions target efficiency was  calculated as
          a function of "K " and "3" using a digital computer.   The
                                      79
A. P. T. Inc.
                                            POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
         maximum values of  "K  " and  "K  "  in  our  range  of interest were
         found  to be between 10-6 and 4x10"3.  These calculations showed
         that flux forces have practically no  effect on the impaction
         from a jet.  The same results  were  obtained for particle
         growth.  The reason for these  results lies in the very short
         residence time of  the particle in the collection region which
         is measured in KT11 to 10~3 seconds as  compared to =0.1
         seconds in the bubble and =0.25  seconds in the spray column.

              For impinging jets, flux  force/condensation scrubbing
         would  have its major effect where particle growth occurs
         prior  to impingement.  Thus, while  negligible collection
         would  be caused by phoretic forces, scrubbers incorporating
         impinging jets as  a unit mechanism  for  particle collection
         are still to be considered  as  a  possible alternative.   A com-
         posite scrubber involving condensation  and growth in a pri-
         mary stage and impingement  in  a  secondary stage might be
         desirable for some applications.
                                      80


A. P. T. InC.                                 POST OFFICE BOX 71,  RIVERSIDE, CA. 92502

-------
                   MATHEMATICAL MODEL,  LIQUID  SHEETS


             This  section  describes  the  development  of  a  mathematical
        model  for  particle collection  by liquid sheets.   Following
        the  conceptual  framework of  the  Scrubber Handbook,  liquid
        sheets  are the  basic  geometric element  involved in  the unit
        mechanism  which describes  collection  in packed  columns,
        wetted  wall columns,  wetted  plates, and similar devices.
        Included in the treatment  are  the following  effects:

                  1.  Diffusiophoresis

                  2.  Thermophoresis

                  3.  Particle  growth  by condensation

                  4.  Centrifugal  deposition

                  5.  Gravitational  settling

             Collection by Brownian  diffusion is not included be-
        cause of its  small effect  on particles  in the  size  range
        of immediate  interest.   Turbulent diffusion  is  also not
        included because the  gas  flow  in long,  straight geometries
        (such as wetted wall  columns and the  like) will generally
        be at Reynolds  numbers  of  only a few  thousand.   For systems
        involving  short curved  geometric elements (such as  packed
        columns) the  deposition is dominated  by centrifugal
        (inertial) impaction  and turbulence serves mainly to mix
        the  particles in the  gas  stream.

             The first  model  worked out  is for counter-current flow
        in fairly  straight systems such  as wetted wall (cylindrical
        or flat plates) columns.   As will be  seen,  the modification
        of a few terms  can adapt the model to packed beds and
        similar devices where the  contact or  surface renewal times
        for gas and liquid are  relatively short.  A  digital computer
        was  used to solve the simultaneous differential equations
        for heat  and  mass transfer and particle growth and collect-
        ion.
                                     81


A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
                                               P2
             The accompanying sketch
        illustrates the geometric
        relationships and the flow
        directions.  Gas enters the
        bottom of the column at a
        superficial molal flow rate
        (wet gas) of "G",
        gmol/cm2-sec, a temperature
        of "TG" °K, water vapor
        partial pressure of "p" atm,
        particle concentration of
        "n " particles/cm3, and
        particle radius of "r " cm.
        Liquid leaves the bottom of
        the column at a superficial
        molal flow rate of "L",
        gmol/cm2-sec and a temper-
        ature of "T, ".
                   Li
                                         'G2
                                 L2
r
G + dG xL+dL
. J Ji.
j
"ill"
G S L
i
J.
dZ Z
T
(
I -,
           Gl
Pi
                                   'Ll
             The heat balance about a differential element of the
        gas-liquid sheet interface is found, as for the case of
        bubbles, by equating the rate of latent heat evolution for
        condensation to 'the rates of heat transfer to the gas plus
        the rate of heat transfer to the liquid bulk.  We obtain:
nr
KT
                                          - Ti
TL - 0
               (103)
             The heat balance about the gas phase in an element of
        the column involves heat transfer from the liquid sheet and
        the particles (if there is condensation on the particles) ,
        temperature change of the particles, and temperature change
        of the gas.  It can be shown that the product of mass times
        heat capacity for the particles is a very small percentage
        (generally less than li) of "m C " for the gas; so particle
                                     82
A. P. T. Inc.
               POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
        heat capacity is neglected,  (i.e., the  last  term  on  the
        right of equation (104) is neglected).


            h, a (T. - TG) dZ * hp ap  (Tp  - TG) np e  ^ -


               Cp d[G TG) * mp Cpp RT  d  (np G Tp)               (104)


        where:

        a  = gas/liquid interfacial  area per unit volume  of  scrubber,
             cm2/cm3

        e  = volume fraction open space  in scrubber

        a  = surface area of particle, cm2

        h  = heat transfer coefficient, particle to  gas
         P   cal/cm2-sec-°K

             A heat balance for liquid flowing  through the differen-
        tial element is formed by equating the  rate  of heat  transfer
        to the temperature change of the liquid.


          -C .  d(L T.) = hT  a (T. -  TT)dZ  (cal/cm2 col.-sec)    (105)
            pL      L     L     i    L


             The change (decrease) in  liquid flow rate as one
        computes up the column is obtained from a mass balance  in
        which the condensation rate  equals the  rate  of gain  (or  loss)
        of liquid flow rate.

                s\
          -dL = kG a (pG - pi) dZ (g mol/cm2 col.-sec)          (106)

        Note that for co-current flow  the  signs of equations  (105)
        and (106) would be changed.

             The change in gas flow  rate is obtained  from a  mass
        balance which equates the rates  of condensation on particle
        plus sheets to the rate of loss of water vapor from  the
                                     83

A. P. T. IDC                                 POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
        gas phase.


        -d(G PG) = kG  a  (pG  - p.)  dZ  +  kp  ap np e (pG - Pp) dZ


                                         (gmol/cm2 col.-sec)    (107)

             The.rate  of  growth  of particles is computed, as for
        the case of bubbles, by  means of the following  equation:


                  2 Dr p(pr  - p  .)  At
        r* - r2  =	^	  (cm2) (also 4.3.4-19 in (108)
                   r !„  p.. p™.                Scrubber Handbook)
                       b   M   DM                	

        Note that equation (108)  is solved for the "bulk" value of
        partial  pressure, "PG".   It has been shown earlier in this
        report that growth can occur  in the boundary layer even
        though "Pp" is below saturation.   For the present, we
        neglect  this effect.

             Gas phase mass  transfer  coefficient is taken to be
        the conventional  one for pipe or channel flow:

               s\
               kr d
               -jr— RTr  = 0.023  ND °-83  Nc  OJ"'                  (109)
                 Dn     G          Re     Sc
                 u

             If  air and water properties are used as for the bubble
        case, we can reduce  equation  (109) to (110):

        yQ « 6xlO'7 T  (g/cm-sec  -  °K) x °K

        PG * 0.34/T g/cm3

        DG * 1.2x10-*  (T)1'75 cm2/sec

             0.023 (u)°-83 (D )0'56     75xlo-3   r°'83
        k  * 	G	  = 3.75x10   = G
         G   (vG)°-1( d0'17  RTG          d°'17        e


             If  we assume that the gas  phase heat and mass transfer
        coefficients can  be  related by  means of a penetration theory
        model, using equal contact times for both coefficients, we
                                      84

A. P. T. InC.                                 POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
       find:

                                1/2
            hG _ RT  PG CpG kG  ' „ 65.0
            J~ " RTG    Dp         (TG) ^
             G

       where:  k_ « 7.3xlQ-5  (cal . /cm2 -sec-°K/cm)
                b

            The liquid phase  heat transfer  coefficient  is  much more
       difficult to pin down  and, indeed, it will  be  greatly de-
       pendent upon the specific fluid flow system used.   As an
       approximation of a representative situation we have taken
       the heat transfer coefficient predicted  for a  wetted wall
       column and adjusted  it through  a process  of comparison
       with experimental mass transfer coefficients.

            Upon review of  the  experimental data for  mass  transfer
       coefficients in wetted wall and packed columns (e.g.,
       Norman, 1961, equation 6.34), it appears  that  "kL" , and
       therefore "h", for  packed columns depend upon liquid rate
       to  the 0.75 power.   This reflects the  influence  of  liquid
       flow rate upon  the fraction of  packing area which  is wetted.
       The  fraction of area wetted varies as  liquid rate  to the
       0.25 power and  the transfer rate varies  as  the 0.5  power,
       so  the total influence of liquid rate  depends  on the 0.75
       power.

            For the present purpose  of exploring the  general
       characteristics of collection by sheets,  we will assume
       that the liquid rate is  high  enough  and  the geometry such
       that the wetted surface  area  is constant.  Thus  an  equation
       such as the  following  should  describe  the influence of
       liquid rate  on  transfer  rate  only.
                 hL =  0.5 (Vg)1/2  =0.5 (2                  (112)

        where  VB =  volumetric liquid flow rate  per unit of sheet
                   width, cm3/sec-cm

        Deposition  Velocities

             In keeping with the  previous development for bubbles,
        the diffusiophoretic and  thermophoretic deposition velo-
        cities are  as  given below.   It will be  recalled that the
                                     85
A.P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
total particle flux can be computed from the  sum  of  the
individual deposition velocities.

               -  (Pr - P;)
u n * 0.83 RTr kr — 2 - — cm/sec                      (113)
 pD          G  G
                    PBM

           C1 yr  dT
ur>T = °'25 - ^      cm/sec                            (114)
 P         PG TG  dx

and, for air/water,


u T = 6.1xl(T3 C1 Tr hp (T. - Tr) cm/sec                (115)
 pi                Vi  Vj   1
                                        Li
             The centrifugal deposition  velocity  is  given by:
               2 r2  p  C' v?,
        u   ~  	P_J2	k (cm/sec)                           (116)
                9 u   R


        and for moist air and  p   =  1.0(g/cm3):


              3.7xlOs rl  C1 v'
        u r - 	*-	   cm/sec                         (117)
         P        T  R
                  TG  c

             In the present model we  do  not  include  this  effect.   For
        curved flows it will be used  and "R  " will be  an  effective
        radius of curvature which will depend on  the specific
        geometry.

             Gravitational  settling for  Stokes  law  is  defined  by:



                                                        cos a   (118)
m C1 cos a
P ~ t. fi^vi n*
611 u,, r
G p
~C' ^ + ro Pd3"
L rp TG J
        where:

        m  = mass of particle,  g

        r  = the original particle  radius  before  growth,  cm

        p, = the difference between  original  particle  density and
             density of water,  g/cm3


                                      86

A. P. T. InC.                                 POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
             The particle deposition velocity is
             Ups = > + UpT + UpG + UPC Cm/SeC
        Solution

             The relationships given above have been solved by a
        finite difference computation on a digital computer.  The
        results for four computer runs are shown in Table IV  and
        Figures 16 through 19.  All of these predictions were made
        for the case where centrifugal (inertial)  and gravitational
        separation are not active; as would be the case for vertical
        plane sheets.  Column height was taken as  300 cm for runs
        #1 and #2, and as 100 cm for runs #3 and #4.  For all runs
        the gas flow rate was 0.006 (gmol/cmz-sec) (which corres-
        ponds to an air velocity of about 5 ft/sec, depending on
        temperature), the liquid rate was 0.03 (gmol/cm2-sec), the
        particle radius was 0.33 (ym), and the gas/liquid inter-
        facial area was 2.0 (cm2/cm3).

             The method of solution was to start from the inlet gas
        and outlet liquid values (i.e., from the bottom of the
        column) and to compute values of the variables for small
        increments of length until the top end of the column is
        reached.  Figures 16 through 19 are plots  of the several
        parameters listed below, as a function of distance up the
        column.  Because different column lengths  are used, the
        length coordinate is given as fraction of total length.

             T_ and TT are given in °C; on "temperature scale"

             Pt  = fractional penetration; on "normalized scale"

             r1  = (r /r ); on normalized scale

             S/2 = 1/2 of saturation ratio; on normalized scale

             It can be seen that predicted penetrations for these
        few conditions are fairly high.  Inclusion of centrifugal
        deposition would give appreciable improvement in penetration
        because the particle size does increase quickly.  This
        indicates that packed columns would probably be superior to
        wetted wall configurations.  There would also be the bene-
        ficial effect on liquid phase heat transfer coefficient if
        packing were used.
                                     87

A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
 D
 o
                                            Table IV



                  RESULTS OF COMPUTER  PREDICTIONS OF COLLECTION  BY  SHEETS
TJ

O

Cfl
O
m

ra

O
X
m
31
CO

o
     oo

     oo

Run*
1
2
3
4
In]
TG(°C)
60.
45.
45.
45.
et Conditions
TL(°O
34.7
21.8
22 .5
22 .4
PG(atm)
0.2
0.092
0.092
0.092
n (#/
Pern3)
3,500
3,500
3,500
106
ZCcm)
300.
300.
100.
100.
Outlet Conditions
TG(°C)
34.8
21.9
26.1
26.4
TL(°C)
44.
27.
27.
27.
S
ratio
1.0
1.0
1.04
1.0
Pt(fr)
0.84
0.93
0.94
0.935

rp
(ym)
4.54
4.1
3.96
0.85
Note:  The above results were computed for the case  of vertical,  counter-current

       flow of air and water with no centrifugal deposition.   Other common

       conditions for all runs are as follows:



               G  = 0.006 (gmol/cm2-sec)


               L  = 0.03  (gmol/cm2-sec)


               r  =0.33  (um)


               a  =2.0   (cm2 contact area/cm3 column vol.)


               S,, = critical saturation ratio for condensation =1.0
8
en
o
to

-------
               1.0
           w
           u
           C/3

           P
           w
           o
           2
               0.5
                 0
                                     100
                                                   1.0
                                      100
                          Pt
                                       u
                                       o
                                     30
                                     20
                                     10
JO

 1
                   0     0.5

              FRACTION OF COLUMN LENGTH
            u
            j
            <
            u
            Q
            UU
            O
            2
                 0.5 h
 0
                                                               Pt
                     50
                                                                         40
                                      30
                                                                         20
                                      10
0
                        u
                        o
                        ia
                        en
                        3
                        t-
                                                                             w
                                                                             a.
   0      0.5       1

FRACTION OF COLUMN  LENGTH
                      Figure 16

                 Computed predictions
                 for sheets, Run  #1
                                        89
A. P. T. Inc.
                         Figure 17

                    Computed prediction
                    for sheets, Run #2
          POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
              1.0
          PJ
          ,-J
          <
          u
          UD

          0
          w
          Kl
0.5
                     100
                            Pt
              1.0
                        u
                        o
   w
   (*

ISO g
                        fr,
                        S

                     140 «
                                    50
                                    !0
                                    10
                                   JO
                  0               1.0

             FRACTION OF COLUMN LENGTH



                     Figure 18

               Computed prediction

               for sheets, Run #3
W



u



Q


N  0.5
h-H
»-J
                                               o
                                               z
                                     0
                                                Pt
                                              S/2
                         100
                                       u
                                       o
   (A

50 H


   n
   Du,


40 H
                                                         30
                                                         20
                                                         10
                                    0
                                       0               1.0

                                   FRACTION OF COLUMN LENGTH



                                          Figure  19

                                      Computed prediction

                                      for sheets,  Run #4
                                           90
A. P. T. Inc.
                               POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
             As in the case of transfer from bubbles, the effect
        of high particle concentration (see run #4, Figure  19)  is
        to reduce the water vapor concentration so that diffusio-
        phoretic velocity is lower and particle growth is less  than
        for low particle concentrations.  If condensation did not
        occur on the particles because the critical saturation
        ratio were not reached, the water vapor concentration would
        not be depleted and, therefore, the diffusiophoretic
        velocity would not be dependent on particle concentration.

             When condensation does occur, as in the examples shown
        in Table IV, the particle size may increase sufficiently
        that collection efficiency by inertial impaction can become
        significant.  In run #4 the particles grew to 1.7 pm dia-
        meter even though the inlet concentration was 106/cm3.
        Particles this size can be collected at moderate pressure
        drop.
                                      91


A. P. T. InC.                                POST OFFICE BOX 71,  RIVERSIDE. CA 92502

-------
                              EXPERIMENTAL
             Theoretical considerations had indicated that plate-
        type scrubbers should be good for FF/C scrubbing on  the
        basis of their assumed heat and mass transfer character-
        istics.  It was also clear that the transfer coefficients
        were very important in determining particle collection
        efficiency, so it was important to verify them by experi-
        mental measurements.  Thus, the kind of experiment needed
        to provide the crucial test of the preliminary theoretical
        analysis was defined.

             In this section of the report we will describe:

             1.  Details of the experimental apparatus

             2.  Experimental procedure

             3.  Analytical methods

             4.  Computational methods

             5.  Results

             6.  Comparison of experimental results with
                 predictions .


        EXPERIMENTAL APPARATUS

             A schematic flow diagram of the experimental apparatus
        is shown in Figure 20.  The major components were as follows

             Test Section:  The test section consisted of a  10.2 cm
        diameter glass column.  A sieve-plate was installed  in it,
        as the first configuration to be studied for FF/C effects
        scrubbing.

             The plate was made out of 1.6 mm aluminum sheet and had
        an overall diameter of 10.2 cm.  It had 30 perforations of
        4.8 mm diameter, adding up to 6.6$ free area on the  plate.
        It had two straight weirs, 8.1 cm long and 9.5 mm high, for
        the liquid inlet and outlet on the plate.  As the plate was
        to be used to study FF/C effects scrubbing, the additional
        design criteria used were:
                                     92
A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
     WATER
AIR
                                  AIR      GAS      WATER
Legend:

 1.  Valves
 2.  Pressure Regulator
 3.  Rotameters
 4.  Water Cooler
 5.  Air Prefliter
 6.  Air Blower
 7.  Air Cooler
 8.  Venturi-meter
 9.  Air Heater
10.  "Absolute" Filters
                                                             TO  DRAIN
            11.  Particle Charge Neutralizer
            12.  Gas Mixing Section
            13.  Boiler
            14.  Steam Entrainment Separator
            15.  Pressure Indicator
            16.  Air Filter
            17.  Pressure Indicator
            18.  Collison Atomizer
            19.  Impactor
            20.  Evaporation-Condensation Columns
            21.  Sieve-plate Scrubber
                 Figure 20  - Experimental apparatus
                                 93
A. P. T. Inc.
                                            POST OFFICE BOX 71. RIVERSIDE. CA 92502

-------
              1.  Very  low weir height.  Since  the major portion
                 of  the collection  is expected  to take place
                 during a bubble rise of a few  bubble diameters,
                 a large foam height is not needed.

              2.  Large perforation  diameter and low  gas velocity
                 in  the perforations, since it  was desired  for
                 experimental purposes to maintain a low collection
                 efficiency during  bubble formation.

              3.  Large weir length, since it will ensure good
                 liquid distribution and thus  a constant liquid
                 temperature on the plate.

              The operating characteristics of  the plate are shown  in
         Figure 21.

              The water was introduced £rom the  top to the sieve-
         plate behind the inlet weir, through a  downcomer tube.  An
         outlet downcomer tube located behind the outlet weir re-
         moved the water from the plate.  A sampling valve was pro-
         vided in the outlet water flow.  The flow rate of water
         through the  test section was measured by a set of rotameters
         on the inlet line.  Water temperature was maintained by
         distributing the inlet water flow between a cooler  and a
         bypass line before it was introduced in  the test section.
         The inlet and outlet water  temperatures were measured by
         iron-constantan thermocouples located  in the respective
         downcomers.

             The air stream, under  controlled conditions of tempera-
         ture, humidity and particle concentration flowed upwards
         through the perforations in the sieve-plate.  The inlet and
         outlet temperature and humidity were measured by wet and dry
         bulb  thermometers located in the respective lines.  Sampling
         probes for particle concentration and size distribution
         measurements, together with static pressure taps for measur-
         ing the pressure drops, were also provided in the air lines
         at the inlet and outlet to  the test section.  A variable
         height probe was provided in the test section.  It  could be
         adjusted to obtain air samples and static pressure  readings
         within the test section, at any height  over the sieve-plate.
         Ample height was provided over the sieve-plate in the test
         section, to prevent entrainment of water drops in the outlet
         air stream.
                                    94


A. P. T. InC.                                POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
          £
          u
              2
                  O	L=0.985,liters/min
                  Q	L=1.9,  liters/min
                   Weep
                  O
              0.2             0.5        0.9
                   GAS  FLOWRATE,  CmVmin)*
e
u
                                                ex,
           —L =  2.50,  liters/min
           — L =  2.96,  liters/min
      0.3
                     aP
                                                                        Weep
                                                                        I
                             I
      0.5              0.8
GAS FLOWRATE,  (ms/min)*
                   20°C,  1  atm
                               Figure  21  -  Operating  characteristics of the
                                            sieve  plate.
A. P. T. Inc.
95
                                                                     POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
         Air  Flow  System

              A  1/2  HP  Rotron  "Cyclonair"  blower  supplied  the  air
         flow, which was  prefiltered.   The air  flow-rate was
         measured  by a  calibrated venturi  meter.   A  finned-tube
         heat exchanger in  the  air  line, cooled the  air  flow by
         means of  ice water to  18°C and a  heater  using three
         electrical  elements could  heat the  air if desired.  The
         air  was then passed through an "Ultra  Aire" absolute
         filter  made by Mine Safety Appliances  Company.  The
         filter  is rated  at 99.97%  efficiency of  retention for
         0.3  ym  particles by the manufacturer,  and was regularly
         checked by  measuring  the particle concentration in the
         filtered  air.

              A  controlled  flow of  aerosol from the  particle
         generator was  introduced into  the air  stream.  The air
         stream  then passed through a charge-neutralizer utilizing
         ten  Polonium 210 elements, with radioactivity of  500  pc
         each, mounted  on the wall  of a 76.2 cm section of 3.8 cm
         diameter  glass pipe.   At the maximum air flow-rate of
         0.71 (m3/min)  with particle concentration of 10s  particles/
         cm3, the  residence time of 0.1 sec  in  the charge  neutral-
         izer is considered sufficient, due  to^the high energy of
         disintegration [5.4 Mev) for the  Pozl  alpha particles.

              A  controlled  quantity of  steam was  then added to the
         air  stream  to  obtain  the desired  moisture content in  the
         air.  The steam passed through an entrainment separator
         column packed  with bronze  wool, and then through  an
         absolute  filter  similar to the filter  used  in the air
         line, before it was introduced into the  air stream.   The
         humidified  aerosol then entered the sieve-plate column
         following a mixing section.

         Particle Generator

              The major components  of the  particle generator assembly
         were:   (1)  a  3-orifice Collison  atomizer (CA).   (2)  an
         impactor with  a  cut-off diameter  of 2  um, and   (3)  an
         evaporation-condensation (E-C) aerosol generator.  The
         particle generator set-up  is shown  in  Figure 22.

              Dibutyl-phthalate (DBP),  with  0.01% by weight of anthra-
         cene as nucleating agent,  was  used  as  aerosol material  in
         the  CA.  Filtered  air  at 40 psig  was fed to the CA for  gene-
         rating a DBP spray.  Particles larger  than  2 vim   diameter
                                    96


A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
                 Compressed air

                    @  40 psig
                           Filter
                                r~\
                    3-Orifice
                    Collision
                    atomizer
L\
                               Impactor
                             2  ym cut-off
                                             (\
                                     Charge Neutralizer
                                                               Vent
                                                                ®
                Figure  22  -  Particle generator assembly
                                   97
P. T. Inc.
            POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
       were  then  removed by the  impactor  following  the  CA.

             The particles  then entered  the bottom of  the  first  E-C
       column, a  2.54  cm O.D. glass  tube, 94  cm  long  with heat-ing
       tapes wrapped around the  tube  at top and  bottom  to serve as
       evaporation  zones.  Axial baffles  were provided  near  the
       entrance to  the  tube to promote  mixing and better  heat
       transfer area in the first evaporating zone.   The  particles
       then  flowed  into the top  of the  second E-C column.  The
       upper section of the column was  a  48 cm long,  2.2  cm  O.D.
       Vigeraux column  and the lower  section  was plain  pyrex tubing
       30.5  cm long and 2.2 cm O.D.   The  second  E-C column had  one
       evaporation  zone located  at the  top near  the entrance to
       the column.

             The particle characteristics  were studied for various
       values of  voltages  applied to  the  three heating  tapes and
       various pressures in the  compressed air line.  The values
       of 40 psig pressure in the air line and 65V  applied to each
       of the heating  tapes were found  to give particles  with a
       mass  mean  diameter, d   = 0.7  pm and a geometric standard
       deviation, a  =  1.215, measured  by the eight stage, non-
       viable Anderson  sampler.  As  this  aerosol was  suitable for
       our purposes, the particle generator was  routinely operated
       at these settings.

       Particulate  Sampling System

             This  system consisted of  three major components:

             (1)   Filter sampling system to measure  the  particle
                  concentration in the inlet and  outlet  air
                  streams.

             (2)   The Climet CI-201 particle analyzer  to measure
                  the particle concentration and  size  distribution
                  in the inlet and outlet  air  streams, and

             (3)   The eight stage, non-viable  Anderson sampler to
                  measure the particle size distribution in the
                  inlet  and outlet air streams.

             Both  the inlet and outlet sampling lines  consisted  of
       insulated  6.4 mm aluminum tubing.  The sampling  probes were
       located near the center of the air pipes, bent to  face the
       air stream in the direction of flow.   Since  the  particle
                                     98


A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
        sizes in the air stream were small  (d   = 0-7 vim, a  =  1.215),
        no consideration need be given for  isoKinetic sampling.  Two
        lines were drawn from each of the sampling probes, one  going
        to- the Climet particle analyzer and the other to the Anderson
        sampler and the filter sampling systems.  The sampling  sys-
        tem was so designed, as to keep the sampling lines as short
        as possible.  A schematic diagram of the sampling system is
        shown in Figure 23.


        EXPERIMENTAL PROCEDURE

             The experiment was started by  bringing the particle
        generator to the operating conditions.  When an aerosol
        stream containing particles in the  range of 0.4 urn <_ d  <  2ym
        is illuminated by a beam of white light, the light scattering
        by the particles produce angular spectra, known as the  high-
        er-order Tyndall spectra.  The spectra exhibits different
        colors depending on the angle of observation.  The satura-
        tion and brightness of the colors increase with uniformity
        of particle size.  This phenomenon was used to determine
        whether the particle generator was producing nearly mono-
        disperse aerosol.  After an iron-constantan thermocouple
        located in the E-C column indicated that the temperature
        had reached a steady state, the aerosol was observed with
        forward-scattered light from a flashlight until a bright
        orange color was seen due to the higher order Tyndall
        Spectra.  This normally took 10 to  15 minutes and indicated
        that the generator was ready.

             Before the particles were introduced into the air
        stream, the air and water flow-rates were brought to the
        operating values.  When a saturated or unsaturated air
        flow was desired, the temperature of the air was set by
        either operating the heater or the  cooler, the particle
        concentration was measured in the filtered air using the
        particle analyzer and then steam was added to reach the
        operating values of temperature and humidity.  Supersaturated
        air flow was attained by first adding the required amount
        of steam into heated air flow so that the gas remained  un-
        saturated.  The heat was then gradually reduced to reach
        the desired values of temperature and supersaturation in
        gas.

             After the steady state was reached, which took from
        thirty to sixty minutes, the flow-rate conditions, foam
        height, pressure drop across the plate, the inlet and outlet
        air wet and dry bulb temperatures and the water inlet and
        outlet temperatures were noted.  When the air was


                                     99

A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
 D
 o
s
CO
o
m
03
O
X
m
CO
O
CO
rvj
en
O
ro
      o
      o
                    Water
                     in
Thermometers
Dryer
Cold trap
Climet particle  analyzer
with dilution system
                                    Water to drain.
                                       Figure 23 - Sampling system.

-------
        supersaturated,  a sample  was  drawn from the air line and
        heated to  determine  the humidity with wet and dry bulb
        thermometers.   This  enabled the  determination of the
        amount of  steam lost due  to condensation on the walls when
        the  air temperature  was reduced.

             The particles were then  introduced into the air
        stream.  The  particle concentration was controlled to
        obtain approximately 5 mg of  sample on the outlet filter
        in 30 minutes.   In the meantime, the Anderson sampler and
        the  filters were prepared and heated in the oven to the
        inlet gas  temperature. The sample lines were heated by
        drawing air through  them.   The filter holders were then
        connected  in  the respective lines and the sampling lines
        were checked  for possible leakage.

             The inlet  and outlet filter sampling was started.  The
        sampling time was approximately  30 minutes and the flow rate
        was  approximately 5.1 litres/min (@ 20°C, 1 atm) .  All the
        operating  conditions were rechecked while the filter samp-
        ling was going  on.

             Upon  completion of the filter sampling, sampling for
        the  particle  size distribution in the inlet gas was started
        using the  Anderson sampler.  Again, the same precautions
        were taken against condensation  and leakages in the
        sampling lines  as taken earlier  for filter sampling.  The
        sampling time was approximately  45 minutes and the flow rate
        28.3 (litres/min).

             After the  sampling using the Anderson sampler was com-
        pleted,  the particle size distribution and concentration in
        the  inlet  and outlet air  lines were measured using the
        particle analyzer.  This  was  done by connecting the sampling
        lines to the  particle analyzer and reading the particle con-
        centrations for successively  larger particle sizes, using
        the  potentiometer connected to the analyzer.

             The inlet  valve from the particle generator to the air
        line was then shut off.   The  experimental conditions of gas
        flow rate, etc.  were then recorded without the particles, as
        done earlier, prior  to the introduction of the particles.

             The filter papers and the Anderson sampler plates were
        weighed immediately  upon  the  completion of sampling on an
        analytical balance (Sartorius Model 2443; ±0.05 mg precision)
                                     101


A. P. T. InC.                                POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
        These were then left to dry in a desiccator and reweighed
        after at least 20 hours.   At this point the run is com-
        pleted.

             The experimental conditions were found to stay very
        stable,  once the steady state was reached.  The conditions
        were checked at the start of each sampling run for the
        filter sample and the Anderson Sampler.  They were rechecked
        at least once while the sampling was going on and then
        again at the end of the sampling run.  The water and steam
        flow-rates were significantly affected by variations in the
        inlet pressures.  Pressure regulators and throttle valves
        on these lines enabled the maintenance of constant flow-
        rates.  The particle generator had been repeatedly tested
        for consistency in particle size distribution and particle
        concentration in its output over a period of time.  These
        parameters were found to remain significantly constant at
        least for a period of up to four hours.  For all the experi-
        mental runs reported, the temperature conditions for the
        experiment varied within ±0.5°C during the experimental
        period.


        METHODS  OF ANALYSIS AND CALCULATION

             As  mentioned previously, three sampling methods were
        used to  determine the particle concentration and size
        distribution.  These were:  (a) filter sampling,  (b)
        Climet CI-201 particle analyzer and (c) Anderson sampler.
        Details  of each of these methods are given below.

             (a)  Filter Sampling

                  47 mm glass fibre filter papers (Type E, made by
        Gelman Instruments Company) were used in the filter sampling
        system.   The sampling flow-rates were measured by rotameters
        on each  line and were controlled by needle valves.  Upstream
        from the rotameter the line pressures were measured by Dwyer
        "Magnehelic" pressure gauges and the temperature by iron-
        constantan thermocouples.  Cold traps and dryers were used
        in the sampling lines to prevent condensation in the rota-
        meters .

             The DBF used to generate aerosol was rated by the manu-
        facturer to have a boiling point of 172°C-174°C @ 5mm of Hg.
        A substantial change in the particle concentration was ob-
        served when the gas temperature was changed.  This is
        attributed to the evaporation of some of the aerosol material,
        DBF.  It was decided to maintain the inlet gas temperature


                                     102

A. P. T. InC                                POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
        below 50°C in order to keep the fraction of DBF  in the
        vapor phase at a low, controllable level.  Also, the
        temperature of both the inlet and outlet filter  holders
        were maintained at the inlet gas temperature, by placing
        them in an oven.  This eliminated any biased effect of
        gas temperature (due to vaporization) on the particle con-
        centration during the sampling.  The inlet and outlet
        filter sampling were done simultaneously under identical
        conditions.

        The particle mass concentration was calculated as:


           _ (weight gain on the filter, at inlet gas temperature)

         P   (sampling flow-rate,Std. conditions) x (sampling time)

                                                      (g/cm3)

        where, c   = particle mass concentration.  Note that the
                P   particle mass concentration can change with the
                    gas temperature due to the evaporation loss of
                    DBF.  The concentration expressed here is at
                    the inlet gas temperature.

             The overall penetration (Ft) in the test section was
        calculated as:

                                     c
                                Pt = -E°
                                     °pi

        where "c   " and "c ." are calculated for the same conditions
        of temperature and pressure.

             As the inlet and outlet filter samples were collected
        with the sample gas heated to the inlet gas temperature in
        both the lines, and as the line pressures in both the lines
        were almost same, the maximum difference being less than
        12 cm W.G.; "Pt" can be expressed as:

               pt  _ weight gain on the outlet filter
                    weight gain on the inlet filter

             At least two samples were taken for each set of condi-
        tions to test for reproducibility.

             (b)   Climet CI-201 Particle Analyzer

                   The Climet CI-201 particle analyzer operates on


                                     103

A. P. T. InC.                                POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
         the  total  light-scattering principle, which  permits  direct
         characterization  of  the particles  as  to  their  size distri-
         bution  and concentration.  The  instrument  measures the
         particle concentrations in the  range  from  3.5X101* to
         3.5xl09 particles/m3 with the help of the  100/1  dilution
         system  (Model  0294-1) .  It was  modified  with a potentio-
         meter so that  the  counting size threshold  could  be varied
         continuously in the  range from  0.3 ym to 3 ym  diameter
         instead of being  restricted  to  several fixed sizes.   A
         dryer was  placed  in  the line connecting  the  particle
         analyzer to the diluter system  to  prevent  condensation
         in the  optical sensor.

             To determine  the particle  number concentration
         (n , #/cm3) , the  potentiometer  and the particle  size
         selector switch were set to  read the concentration of
         particles  greater  than 0.3 ym diameter.  Because particle
         size distribution  measured with the Anderson sampler had
         shown that 1001 particles were  >0.3 ym,  this setting on
         the  instrument indicated the total number  of particles per
         unit sample volume.  The inlet  and outlet  sample lines were
         connected,  in  turn,  to the instrument and  the  concentrations
         were read  off  the  dial.

             To determine  the particle  size distribution, the sample
         line was connected to the instrument and the potentiometer
         was  set at successively larger  values of particle diameters.
         From these data,  the geometric  standard  deviation (a ) and
         the number mean diameter (d  )  were determined.

             (c)   The Anderson Sampler

                   The Anderson Sampler  is  an eight-stage cascade
         impactor.   The perforations  in  each stage  are  designed to
         provide successively smaller cut-off diameters.  Thus, when
         an aerosol  is passed through it, successively  smaller dia-
         meter particles are  collected on the stages  and  the  particle
         size distribution  is determined on a weight  basis, by
         measuring  the weight gain on each  stage.   The  cut-off
         diameters  depend  on  the flow rate  through  the  sampler and
         a preliminary check  of the manufacturer's  rated  values
         using an optical microscope was  found to be  approximately
         correct.   An absolute filter was used after  the  final stage
         of the sampler to  capture all the  particles  that escape
         collection  in the  sampler.
                                    104


A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
             The flow rate through the sampler was maintained  at
        approximately 28.3 (liters/min) and measured by a cali-
        brated dry gas meter.  The sample flow line pressure was
        measured by a mercury U-tube manometer and the line temp-
        erature was measured by an iron-constantan thermocouple
        before it entered the dry gas meter.  As with the filter
        sampling system, the temperature of the Anderson sampler
        was maintained at the inlet gas temperature in the oven.

             The plates were weighed before and immediately after
        the sampling, and again after they were dried in a desic-
        cator for over 20 hours, on the Sartorius analytical bal-
        ance.  They were then washed with soap and rinsed success-
        ively with de-mineralized water, ethanol and acetone.
        Then they were placed back in the sampler and the sampler
        sealed off until the next run.

             The geometric standard deviation (0 ) and the mass
        mean diameter (dpg) of the aerosol were Determined from
        the weight of particles collected on the plates in
        different stages with known cut-off diameters.  The
        particle mass concentration was also determined as:

             (weight gain on sampler and filter,at inlet gas temp.)
        C  "" ~LI
         p   (sampling flow-rate,std. conditions) x (sampling  time)

                                                       (g/cm3)

        This value was then checked against the value determined
        from filter sampling.

        Air Supersaturation

             A direct determination of the saturation ratio (S)
        using the wet and dry bulb thermometers is not possible when
        the air is supersaturated.  Since it was desired to obtain
        data for different values of Tdi < 50°C, the inlet air was
        cooled prior to the introduction of steam.  When steam was
        introduced in the air flow, it was found that a part of it
        condensed in the line.  Assuming that the air line was well
        insulated and there was no heat loss through the pipe, we
        calculated "S" in the following way:

             (a)  The air was heated to a temperature higher than
                  T^ and steam was^introduced at a rate such
                  that the desired y was attained, under unsatura-
                  ted conditions.
                                     105

A. P.T. Inc.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
              (b)  The heaters were turned off and the air
                  temperature was allowed to reach a lower
                  steady state value, Tdg, keeping the same
                  rate of introduction of steam as in (a).

              (c)  The mole ratio, and hence S, were calculated
                  from an enthalpy balance.  First, the  increase
                  in air temperature due to the sensible heat of
                  steam was calculated and then the amount of
                  steam condensed in step (b) was determined.
                  Thus, the partial pressure of vapor, pfi, was
                  determined and S was calculated from:
                  S =
                      Saturated vapor pressure at T^


        Transfer Coefficients

             The transfer coefficients needed for predicting  theoret-
        ical values of penetration at the experimental conditions
        are: h, , k,,, and hr.  The values of these transfer  coeffi-
               L   u       b
        cients were determined individually for each run.

             The conditions for the gas and liquid  flow  rates,
        temperatures and moisture contents were such that the net
        heat and mass transfer were always from the gas  phase to
        the liquid phase.  Under such conditions, the  transfer
        coefficients can be determined using Mickley's method,
        described by A. S. Foust et al. (I960) and  many  others.
        However, the conventional graphical approach is  not suffi-
        ciently accurate due to the curvature in the enthalpy-
        temperature saturation curve in this range  of  gas and liquid
        conditions.

             A computer program was developed to evaluate coeffic-
        ients  for the experimental runs.  The approach used was
        essentially the same as the Mickley's method,  except  that
        heat and mass transfer in the gas-phase were used separately
        instead of being combined into gas-phase enthalpy transfer.

             Also,  instead of setting the Lewis number;

                                  k'  C
                            N   =  bG  Pm
                                "
                                     106

A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
         equal  to  one, which  is  not  valid  when  the  water vapor con-
         centration  is high,  the ratio  ^G/\G  was  assumed to remain
         constant  for a  run.   This ratio was  calculated from the
         penetration theory correlation,

                                           V*

                                                                (121)
kbG 1 1
hxr R Tr \P,
bG G ' i
DG
G« f ™-r*
pG G
                                                       cal
        where,  k   =  thermal  conductivity  of  gas,   	
                b                                  sec-cm-°K;'


        For  the experimental  conditions,  this  ratio  changes  by a
        maximum of 5.26%  (from  0.144  to 0.152  for  Run  No.'s  7 and 8)
        for  the inlet  and  exit  conditions  respectively.   Thus, an
        average value  can  be  used  for the  ratio during a  run, with-
        out  appreciable error.

             After the value  of k'/h.  is  determined  from  the  method
                                 ti LI
        discussed  above, h.  is  computed from a heat  balance  on the
                           LI
        liquid  bulk.   "kup"  and "h, .-," were then determined from

        the  values of  h, ,  k/r/hT ,  and k'  /h/r.

             This  approach incorporates the  following  assumptions:

             1.  The gas bubbles are  spherical and are of a
                 constant  diameter throughout  the  foam.

             2.  The bubbles  are perfectly mixed  internally.

             3.  The foam  density  is  constant  throughout  the
                 foam  layer  on  the plate.

             4.  The liquid  bulk temperature is constant  throughout
                 the foam, and  is  taken as the average temperature
                 between the  liquid inlet  and  outlet stream.

             5.  All of the  bubble surface area is available  for
                 transfer  mechanisms  and  this  area lends  itself
                 equally for  both, heat and  mass  transfer.

             6.  The penetration theory adequately predicts  the
                 ratio of  the gas-phase mass transfer  coefficient
                 to  the heat  transfer coefficient.

             Note  that the ratio is independent of the surface
        renewal time.

                                    107

A. P. T. InC.                                POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
         Accuracy of Measurements



              1.   Determining Overall Penetration:



                  The overall penetration, Pt was determined as:



              rr  _ weight gain on the outlet filter              n??1


                   weight gain on the inlet filter



              The analytical balance used has a precision of ±0.05 mg

              W

         Pt  -  °
         Pt  ~ W7
               i



         where, W is the weight gain on the filter.


           d_Ft _ d WQ   d w.



            Pt     W      W.
                    o      i



              As  the absolute values for the error are small compared

         to  the actual weights, we can say that,



          A_P~t _ A WQ   A W.



            Ft    w       w.
                   o       i



              For calculating maximum error, the error terms are

         considered to be additive.   Thus,

                        _     AW    AW.

         Maximum  error,  Pt =	 + 	-

                               W      W.
                                o      i



         The  precision of the  balance is  ±0.05  mg.


            the maximum values for AW  =  A W.  = 0.1 me
                                      o      i         6

         Minimum  values  for W   = W.  = 5 mg.



           Maximum error,  Pt  = ± ^-^- = ±4%.
                                   O



              In  the range  of  experimental conditions, the maximum

         error in determining  Pt is  ±4$.
                                    108





A. P. T. InC.                                 POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
             2.  Determining Particle Characteristics:

                 The particle characteristics determined were:
        (a)  particle concentration,  (b)  particle size distribu-
        tion.

             (a)  The particle concentrations were measured from
                  the weight gain during filter sampling.  The
                  analytical balance used to weigh the filter
                  paper had a precision of ±0.05 mg.  Most runs
                  lasted 30 minutes out of which the end effects
                  did not take more than 0.08 minutes, or 0.27%
                  of the sampling time.  The rotameters are
                  rated for accuracy in flow measurement at
                  ±5% the maximum error expected in determining
                  the particle mass concentration is ±12.27% for
                  a 5.0 mg sample on the filter.

             (b)  During the experimentation, the size distri-
                  bution measured was; d   = 0.7 ym, a  = 1.22
                                        pg            S
                  by the Anderson sampler.  An analysis of the
                  Anderson sampler performance for determining
                  size distribution of nearly monodisperse
                  aerosols has shown that the "a " value indi-
                  cated by the sampler would be higher than the
                  actual "a " of the aerosol.  The actual "a "
                           8                                g
                  value es expected to be between 1.0 and 1.22.

             Particle size data taken with the Climet instrument are
        not included in this report because they were not suffic-
        iently reproducible.  Operating problems with the counter
        and the sample dilution apparatus required that the instru-
        ment be sent back to the manufacturer several times for
        repairs.  During the time the instrument was operable the
        data obtained were not consistent.  Therefore, major re-
        liance had to be placed on sizing with the cascade impactor.

             3.  Air Flow Rate:

                 The air flow rate was determined by a venturi-
        meter.   The venturi-meter was calibrated using a pitot tube.
        The maximum error in the flow rate calibration is expected
        to be ±5% as rated by the manufacturer.

             4.  Water Flow Rate:

                 The water flow rate was measured by rotameters in


                                     109


A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
         the  line.  These were  calibrated  in  the  laboratory and the
         maximum error  is expected  to be II of  the  maximum flow
         rate.

         EXPERIMENTAL RESULTS

             The experimental  conditions  and results  are  shown in
         Table V.   It was desired to use a high liquid to  gas  flow-
         rate ratio (L/G) so as  to  maintain a low liquid  temperature
         on the sieve plate which permits  a high  driving  force for
         vapor condensation on  the  bubble  wall.   An average (L/G)
         ratio of 7.7 (£/m3) was maintained for runs  1 to  6.   This
         was  the highest  (L/G)  value that  could be  employed without
         entering the weeping and raining  regions of  the  sieve-
         plate.  The  (L/G) ratio was decreased  to an  average value
         of 3.53 (Jl/m3) for runs 7  and  8.

             Two sampling runs  were made  for each  set of  experi-
         mental conditions.  The air was maintained unsaturated
         for  runs 1 and 2.  The  air temperature was then  decreased
         keeping the humidity unchanged for runs  3  and 4.   The
         effect of  increasing the humidity at the original air
         temperature was  studied in runs 5 and  6.   Runs 7  and  8
         were for a different set of experimental conditions.

         Discussion

             Collection Mechanisms:  The  experimental Ft  values are
         the  ratios of outlet to inlet  particle mass  concentrations
         as determined from weight  gains on sampling  filters,  as
         explained earlier.  For the experimental conditions,  the
         collection can be attributed to the  combined  effects  of:
         impaction during bubble formation, diffusiophoresis,  thermo-
         phoresis, diffusion, and particle growth due  to  condensa-
         tion resulting in a higher centrifugal force.  Thus,  Pt
         can  be expressed as:

                  Ft = Ftp x PtD x PtT x  PtBD  x  Ptc             (123)


         For  runs 1 to 6, where  the flow conditions are approximately
         identical, the following observations  can  be  made:

             1.  Ptp will be the same  for these  runs  and  was
                 calculated at  963.

             2.  For these runs, the temperature gradients were
                 not substantially different and were quite small.
                 Thus Pt™ was  high and nearly  the  same for all
                 the runs.

                                    110

A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
                                           Table  V

                              EXPERIMENTAL CONDITIONS AND  RESULTS
Run
Mo.
1
2
3
4
S
6
7
8
Air*
Flow-
Rate
(m'/min)
x 10
3.60
3.60
3.66
3.66
3.66
3.66
5.10
5.10
Water
Flow-
Rate
(liters/
mm)
2.92
2.92
2.88
2.88
2.62
2.62
2.35
2.46
Particle
Concen-
tration
xlO'5
(»/cc)
2.3
2.4
1.9
2.3
4 .6
4.4
4.7
4.0
Gas Conditions
Inlet
T ,.
di
CC)
41
41
33
33
41
41
47
42
*i
0.059
0.064
0.057
0.059
0.095
0.095
0.111
0.111
Exit
Tde
(°C)
24
25
24
24
28
28
29
29
ye
0.031
0.032
0.031
0.031
0.039
0.039
0.042
0.042
Water Temps .
TLi
(8C)
17.9
17.9
17.9
17.9
17.8
17.8
16.0
16.8
TLO
(°C3
19.1
19.1
19.0
19.0
20.0
20.0
21.5
19.0
S
0.73
0.78
1.09
1.12
1.13
1.13
0.93
1.21
y . -y
7 i ' e
0.028
0.032
0.026
0.028
0.056
0.056
0.069
0.069
Pt
exp
(*}
89.0
86.6
86.4
82.2
72.9
69.1
84.4
81.9
Pttheo
(%)
94.0
90.8
88.3
87.0
84.7
84.7
84.8
74.2
hexp
Ccm)
5.0
5.0
4.7
4.7
4.8
4.8
6.4
7.6
htheo
Ccm]
7.6
9.2
7.2
7.6
8.4
8.1
10.2
9.7
         * a 20°C, 1 atm.
           _ g-mole vapor
             g-mole dry air
A. P. T. Inc.
                                               111
                                                                       POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
             3.  For the particle s:i_ze used,  in the range  of
                 0.5 ym to 1.0 urn, PtgD was insignificant  and
                 can be neelected.
can be neglected.
             4.  It is possible that Ftp was nearly  the  same
                 for all the runs and quite high, as  the DBF
                 particles are not highly wettable and would
                 require a saturation ratio larger than  per-
                 haps 1.9 in order for water to condense on
                 them.  The saturation ratio in the  bulk of
                 the gas phase did not go much higher than
                 1.0.

             The results indicate that diffusiophoresis  is the  major
        contributor to particle collection efficiency for a single
        p_late.  The contributions of the rest of the phenomena  to
        Ft will remain approximately the same for all the runs.
        Ptn depends on the amount of vapor condensed, which can  be
        expressed as (y. - y ), or the gmole of vapor condensed
        per gmole of dry air.  Figure 24 is a plot_of Ft vs.
        (y. - y ) for runs 1 to 6, and also shows Ftp due to  im-
        paction during bubble formation, under the experimental
        conditions .

             Comparing Ft    values for runs 1 and 2 with runs  3
        and 4, we see that for approximately the same water vapor
        concentrations, the penetration is lower for  the lower  gas
        temperatures.  This is also observed by comparing run 7
        with run 8.  Comparing runs 1 and 2 with runs 5  and 6 shows
        that the penetration is lower when the water vapor concentra-
        tion is increased at the  same temperature.   It is also
        evident from Table V that the penetration decreases as  the
        saturation ratio, S, increases.

             However, further experimental studies under wider
        ranges of operating conditions are needed before the  trend
        of Ft with any of the parameters can be generalized.

        Comparison of results with predictions

             The model for collection in a sieve plate scrubber,
        described earlier was used to predict the theoretical
        penetration, Pttueo> for  the experimental conditions.   It
        was found that the model  will satisfactorily predict  the
                                   112


A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
                     0.01
0.02
0.03
0.04
0.05
0.06
                                       g-mol vapor condensed
                                     '  g-mol dry air
                    Figure  24  -  Particle penetration versus  water
                                 vapor condensed.
A. P. T. Inc.
                                     113
            POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
         general form and magnitude of experimental results  if  the
         proper heat and mass transfer coefficients are used.   The
         heat and mass transfer coefficients, h., k/G and h,-,,  were
         determined individually for each run.  The calculation pro-
         cedure is shown in an earlier section and the resulting
         coefficients are shown in Table VI.

              The results for runs 2 to 8 are comparable to  the
         values predicted by the penetration theory for k/r  and
         hbG; 1.6xlO-VpBM and 2.2xlO-2//TG respectively.    The
         values for "h," are approximately 10~2 times the corres-
         ponding value predicted by the penetration theory,  0.31.
         This indicates that the assumptions regarding the bubble
         area available for transfer, the bubble diameter, liquid
         temperature, liquid surface renewal time, and the consis-
         tency of the foam density may be in error.  The resolution
         of this question will require additional experimental  and
         computational work.

              Pt ,    values were then calculated for the experimen-
         tal runs, using the corresponding experimental values  of
         transfer coefficients.   The Pt .    values, together with
         the foam height, h ,   , indicated by the model to reach
         the exit gas temperature, are listed in Table V with the
         corresponding experimental conditions.  The difference  in
         ^exo ant* ^theo va-'-ues indicate that the assumptions re-
         garding the bubble diameter area available for transfer
         and the consistency of the foam density may be in error.
         These questions could only be resolved by experimentation
         under highly controlled conditions as indicated earlier.

              Pttheo is Plotted against Pt    on Figure 25.  Except
         for run 8,  Pit .    was higher than Pt
                      theo       &           exp
              To check if the  difference between Pt.,,    and Ft
                                                   theo       exp
         was only due to statistical scatter in the data,  the follow-
         ing assumptions were  made:

              1.   Pttheo values  correctly and uniquely represent
                  the penetration for any set of experimental
                  conditions examined.   The experimental conditions
                  dp not change  during an experimental run.  Thus,
                  Pttheo rePresent th-e mean of Pt    values for any
                  set of experimental conditions.
                                     114

A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
               1.0
               0.9
               0.8
           u
           HH

           E-
           O
           w
               0.6
               0.5
                                       00
                  0.5        0.6       0.7        0.8        0.9        1.0


                                  EXPERIMENTAL Pt
                       Figure  25 - Theoretical versus  experimental
                                   penetration.
                                     115
A. P. T. Inc.
POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
                                   Table VI

                     TRANSFER COEFFICIENTS  DETERMINED BY
                               MICKLEY'S METHOD
Run
No.
1
2
3
4
5
6
7
8
h.xlO3
L
1.84
2.09
2.56
2.52
2.54
2.54
3.57
3.57
kbGXl°"
14.72
2.09
2.04
2.02
2.54
2.54
2.50
2.50
hbG/T^x
16.72
2.48
2.32
2.29
2.95
2.95
3.00
3.00
                                      116


A. P. T. InC.                                  POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
             2.  Pt    values are normally  distributed  about
                   C A L/        ^^_
                 the mean  (or pttheo) for any  set  of  experi-

                 mental conditions.

             Thus, the standard error between  Pttheo  and
        6.6%.  As this error is higher than  the maximum experimental
        error, 4%, our assumption that Pt,,     is the  true  mean  of
        Pt    values is incorrect.
          exp

             Noting that Pt    values are normally distributed,  the
                           GXp
        probability of obtaining less than  two observations on  the
        negative side of the mean,  in a  sample of  eight observations,
        is 0.035.  The lowjjrobability indicates that there is  a
        positive bias for  Pt    values,  when compared to P~ttu
        More experimental  data are  needed if the statistical  rela-
        tion between Pt.,    and Pt    is to  be studied  more
                       theo       exp
        rigorously.

             Overall, the  comparison of  experimental  data  with
        predictions based  on our mathematical  model shows  that  the
        model is capable of giving  useful and  realistic results.
        While additional refinement, as  discussed  above, is needed,
        the model can account for what is observed experimentally
        and gives us a good tool for interpreting  what  we  see and
        utilizing this knowledge for engineering design of practical
        equipment.
                                    117


A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
                        ECONOMIC  FEASIBILITY
              The  technological  feasibility  of  FF/C  scrubbing has
         been  discussed  in previous  sections and  it  is  clear  that
         fine  particles  can  be collected  with high efficiency by
         this  means.   Both diffusiophoresis  and particle  growth by
         condensation  are practically  insensitive to particle size
         and they  are  the most important  factors  in  FF/C  scrubbing.
         What  remains  to be  shown  is whether the  cost is  low  enough.
         The following section deals with the question  of economic
         feasibility,  with the object  of  determining some broad
         outlines  of the cost picture.  Two  specific cases of in-
         dustrial  applications for FF/C scrubbing were  studied and
         preliminary process designs and  cost estimates are des-
         cribed .

              Before getting into  the  details of  case studies it is
         instructive to  consider some  general features  of FF/C
         scrubbing.  Both the theoretical and experimental results
         indicate  that it should require  a minimum of about 0.1 or
         0.15  g water  condensed/g  dry  gas in order to grow particles
         to about  2.0  ym diameter  if the  concentration  is about
         107/cm3.  The experimental  data  show a penetration of about
         70% at a  steam  condensation of about 0.035  g/g for
         n  -  105  entering one sieve plate.

              The  computer runs  show the  effect of steam  condensing
         on the liquid as well as  the  particles,  which  raises the
         steam consumption for a given increase in particle size.
         These runs  indicate that  steam condensation of about
         0.06  g/g  would  cause growth to r = 0.85 urn for  n = 106/cm3
         (see  run  #4 for sheets, Table IVj and  about 0.14 g/g would
         cause growth  to r    1.4  um for  n  = 107/cm3 (see run #17
         for plates, Table III).   In contrast,  we may note that if
         all the steam condensed on  the particles, it would take
         0.033 g/g condensed to  grow 107  particles/cm3  to r = 1.0  ym.

              For  illustrative purposes,  we  may take 0.15 g/g as rep-
         resenting a low practical steam  condensation rate.  If steam
         costs $1.32/1,000 Kg ($0.60/1,000 Ib), FF/C scrubbing would
         cost  $0.20/1,000 Kg ($0.24/1,000 m3) of  dry gas  for  steam
         alone.  A venturi scrubber which had a performance capabil-
         ity of 80%  efficiency on  0.5  ym  diameter particles with
                                    118


A. P. T. InC                                POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
        density of 1.0 g/cm3 would require a pressure drop of about
        450 cm W.C. (see S.H.B. Figure 5.3.6-9).  The scrubbing
        power cost for this pressure drop would be about
        $0.24/1,000 m3 if electricity cost l.Of/K.W.H.  Thus, the
        costs for FF/C and high-energy scrubbing are fairly close
        for this example and will be more favorable for FF/C scrub-
        bing as the collection efficiency requirement becomes more
        stringent.

             It is clear from the above example that FF/C scrubbing
        might or might not afford to pay for steam, depending on the
        particle size and concentration, and the required perform-
        ance.  The next question is whether cooling water costs are
        acceptable.  There is a wide range of cooling water costs
        given in the literature; from about 0.25^/m3 to 3.9
-------
        in this report.  Heat and mass transfer coefficient  and  the
        critical saturation values required for nucleation were
        either based on experimental data or calculated and  checked
        against similar coefficients for similar cases.

             The capital and especially the operating costs  of the
        FF/C scrubber vary according to geographical location,
        availability of cold water and climatic conditions.  Our
        estimations are based on the availability of water at 20°C.

             The calculations were made for large industrial plants
        and this assumes that one could satisfactorily scale-up
        processes which have been studied only in the laboratory.
        Accepted scale-up methods were used to extrapolate the
        laboratory results to the industrial scale.

        BASIC OXYGEN FURNACE EMISSION CONTROL

             In recent years the use of the Basic Oxygen  Furnace
        (B.O.F.) is increasing among the various steelmaking pro-
        cesses.  The major reason for its wide acceptance is its
        short cycle which lasts about 50 minutes.  The Basic Oxygen
        Furnaces vary in size from less than 50 ton/heat  to  over 400
        ton/heat with 200-250 ton/heat being a common size.  During
        about twenty minutes of each cycle oxygen is blown into  the
        bath and that is when the flue gas reaches a temperature of
        over 1,480°C.  The reason for this high temperature  lies in
        the high CO content of the flue gas.

             Because of the low fuel value and hazardous  properties
        of CO, in most U.S. plants air is permitted to leak  into the
        systems at the hood inlet in order to burn the CO to CO-.
        In most foreign countries the CO is used either for  its  heat
        value or for chemical synthesis.  This is done in one of two
        ways:  1.  The gases containing the CO are cleaned and then
        transferred for further use.  2.  The CO is burned to C02  as
        in most U.S. processes but the heat recovered in  a waste
        heat boiler.

             The gas stream leaving the B.O.F. during the first  part
        of the cycle contain less dust, mainly kish, and  is  at a
        lower temperature.  During the period when oxygen is blown
        the flue gas is hot and its dust load is as high  as  45-90
        grams per standard cubic meter of dry gas.  The dust con-
        sists mainly of iron oxides, slag and other metallic oxide
        fumes.
                                    120

A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
        Present  Control  Methods

             Venturi  scrubbers  and  electrostatic precipitators are
        presently used to control B.O.F.  emission.   Both have their
        advantages and disadvantages  and  one  is  not preferred over
        the  other. When electrostatic precipitators are used the
        hot  gases are first cooled  to 320°C in an evaporation
        chamber  or a  spark box.  The  evaporation chamber is fol-
        lowed by a drop-out box  where the larger dust particles
        settle.   In the  spark box excess  water is sprayed and the
        large dust particles settle together  with the excess
        water and are removed from  the system.  The dirty gases
        now  at 320°C  enter the  electrostatic  precipitator, are
        drawn by the  fan and sent out through the stack.

             When high energy venturi scrubbers  are used to clean
        the  gas, the  gas is also first cooled by water injection.
        Concurrently  the large  particles  are  removed from the gas
        and  the  water sludge is  drained.   The gas then enters a
        venturi  scrubber followed by  an entrainment separator.  In
        most cases the gas is further cooled  in  a cooling section
        to increase its  density and as a  result  the fan efficiency.
        The  cold gases enter the fan  that emit them through a stack
        to the atmosphere.

        Background Data

             Information regarding  the B.O.F. process, dust load
        and  particle  size distribution were collected from the
        following sources:  Battelle  report to NAPCA on the Iron and
        Steel Industry (1969),  an  article by H.  C.  Henschen (1968),
        M.R.I, report on particulate  emissions (1971) and SRI Manual
        of Electrostatic Precipitators Technology (1970).

             The dust load coming  out of the B.O.F. ranges from
        45-90 grams/standard dry cubic meter of gas.  Most air pol-
        lution control agencies require that the dust concentration
        in the outlet gas not exceed 0.115 gram/standard dry cubic
        meter of gas  (which is  practically invisible).  Thus a col-
        lection efficiency of 99.741  to 99.87$ is required.  The
        dust particle size distribution is reported by M.R.I, to
        have a mass median diameter of 0.095 ym and a geometric
        standard deviation of a  =  2.3.  The dust composition and

        properties were also taken  from the M.R.I,  report.
                                    121

A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
              For design purposes we shall consider a shop which con-
         sists of two 250 ton vessels,  one of which is assumed to be
         in operation while the other is assumed to be shut down for
         relining.   Such a shop makes about 30 heats of steel a day
         or one every 48 minutes.  During about 20 minutes of each
         cycle oxygen is blown into the bath at a rate of
         710.0 mVmin (25,000 CFM) .   About 3.75 ton of dust, mostly
         oxides of iron, are discharged from the vessel during the
         blowing period.  During the majority of the blowing period
         the gas discharging from the furnace mouth is at =1,500°C
         and contains 90% CO and 10$ CO,.  Upon combustion with the
         theoretical quantity of air a  temperature in excess of
         2,000°C can be developed at the furnace mouth.  For calcu-
         lation purposes we assumed the gas temperature when enter-
         ing quenchers to be constant at =1,500°C during the hot
         part of the cycle.  (We shall  consider later the beginning
         and end of the cycle when  the  gas temperature is lower.)  At
         this temperature the gas flow  rate is 29,400 actual cubic
         meter per minute (1.04xl06  ACFM).

              Steam addition is beneficial for efficient operation of
         the FF/C scrubber if it can increase the saturation ratio
         above 1.0.   For this example we will assume that it can be
         done .

              An alternative approach would be to saturate the gas
         by direct contact with water sprays and dispense with the
         steam generation and storage apparatus.  This may be
         superior economically and  it might also be that the specific
         nature of the particulates  makes it impossible to sustain a
         saturation ratio higher than 1,0 so that the indirect gener-
         ation of steam for addition to the gas would have no benefit,

              The sensible heat of  the  flue gases will be utilized in
         three ways:   a.  A steam production in a waste heat boiler.
         b.  Air reheat at the tail  end of the process to prevent
         steam plume and give the gas additional buoyancy,  c.  Gas
         saturation with water vapor by spraying water into the gas
         in a saturation chamber.

              From these operating  conditions and with the use of
         material and energy balances the process flowsheet illustra-
         ted schematically in Figure 26 was calculated.  The calcu-
         lation method we have developed for a FF/C spray tower
         scrubber could now be used to  calculate the scrubber perform-
         ance.  Particle growth was  calculated based on a critical
         supersaturation of one.
                                    122

A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
                   I
                        WATER
WATER
WATER
TO STACK
                                         I
FROM
B.O.F.
t . tv
1
V
••^•^•B
o
o
o
_y
S
>
. rv
7 L
\ V
SATURATOR



e
u
c
»-
^~
O
L
f
*
7
STEAM
ACCUMULATOR

9
OUJT JUJ\'
SATURATOR
i

-------
              A heat  balance  shows  that  enough steam is  produced
         during the high  temperature  part  of the  cycle to supply the
         needs of the whole  cycle.  Steam  accumulators are required
         to  store the excess  steam  which is  generated during the hot
         period.   The clean,  cool gas now  enters  the fan which trans-
         fers  it  through  a reheater and  a  stack to the atmosphere.
         The reheater was calculated  based on the TVA experiments
         that  warming the flue  gases  by  50°F is enough to prevent
         visible  steam condensation when the ambient temperature
         ranges between 50°F  and 80°F.

              Since the purpose of  this  study is  to show how FF/C
         scrubber compares with other control equipment  a common
         ground for comparison was  sought.   No attempt was made to
         optimize the system  and the  only  changes made were those
         necessary for the FF/C scrubber to  work  adequately.

              Capital cost information came  from  the following
         sources:  Battelle  report  (1969), Swindell and  Dressier
         study (1969),  A.P.T. Scrubber Handbook (1972),  Cost
         Engineering  i.n the  Process Industries (1960)  and Modern
         Cost  Engineering Techniques  (1970) .   The method used to
         calculate the  capital costs  is  the  one described in:  "Data
         Techniques for Preliminary Capital  Cost  Estimating"  by
         K.  M.  Guthrie, Modern Cost Engineering Techniques!   The
         method is based on  calculating  the  F.O.B.  equipment  cost
         and the   multiplying by various factors  for materials,
         labor,  indirect costs, contingency,  and  fee.  The capital
         cost  was  then  adjusted to  a  common  time  base.   The Chemical
         Engineering  Plant Cost Index was  used to account for cost
         changes  with time.

         Cost  of  FF/C System

              Equipment for the system shown  schematically in
         Figure  27 and  discussed above was selected and  sized for
         the purpose  of cost estimation.   A  summary of the results
         of-  this  process design and cost estimation are  as
         follows:
                                    124


A. P. T. InC.                                POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
              Capital  Cost:
                   Tubular Membrane  Hood (estimated from
                       the cost of waste heat boilers)      $690,000

                   Steam Accumulators,  3 vessels,  1.0 m
                       dia.  x 25.0 m long                   513,000

                   Saturator 4.6m dia x 30.0m long           53,000

                   Spray Tower (3.0  m high)                  218,000

                   Reheater  (300  m2  area)                     96,000

                   Blower (1,300  HP)                         100,000
              Subtotal                                    $1,670,000

              Piping  8  91  of  total                           150,000
              Total                                       $1,820,000

              Annual Operating  Costs:

                  Maintenance  @4I  of  capital  cost        $    73,000

                  Capital  expenses  and  depreciation
                       @20% of  capital cost                  364,000

                  Plant  O.H. @  $0.10/ACFM-year               50,000

                  Power  @$0.011/hw-hr                       104,000

                  Water: 247 gpm @  $.40/Mgal                 47,500

                         2000  gpm  9 $.lS/Mgal               144,000

                  Labor: 8000  hrs  @ $5/hr                    40,000


              TOTAL                                       $   822,500


              There is  no charge for steam in  the operating  costs
         since the gas  sensible  heat is  used for  steam  production.
         The water and  boiler costs  are  included  in their proper
         places.
                                    125


A. P. T. Inc.                                POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
                                   TABLE  VII
                         B.O.F. CONTROL  SYSTEM COSTS
Item
Electrostatic
Precipitator
Capital Cost
Annual Cost
Fabric Filter
Capital Cost
Annual Cost
High Energy
Scrubber
Capital Cost
Annual Cost
M.R.I.
Report
$4,900,000
1,660,000
4,500,000
1,260,000
6,000,000
2,360,000
Swindell-
Dressier
Report
$4,150,000
1,520,000
3,200,000
1,160,000
3,800,000
1,850,000
S.R.I.
Report
$5,000,000




	
        FF/C  Scrubber Costs by A.P.T., Inc.
               Capital Cost       $1,820,000
               Annual  Cost        $  822,500
                                      126
A. P. T. Inc.
POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
             For comparison, the costs presented by Midwest  Research
         Institute, Swindell-Dressier and Southern Research Institute
         are given in Table VII.  The costs quoted can be  reduced by
         various manipulations, such as assuming two B.O.F. furnaces
         operating at staggered intervals, etc.  However,  to  obtain
         a comparison between the three alternatives the same condi-
         tions were assumed for all three control equipments, which
         would not necessarily lead to the lowest conceivable costs.

         KRAFT RECOVERY FURNACE CONTROL

             Characteristic air-pollution problems of the pulp  in-
         dustry are associated with the release of malodorous sulfur
         compounds and particulate matter.  Pulp is made by either
         the sulfate (kraft), sulfite, semichemical, soda, or by a
         mechanical process.  Most of the pulp produced in the
         United States is made by the kraft process.  In the  follow-
         ing we shall examine the economics of using a FF/C scrubber
         to control the fine particles emitted from the recovery
         furnace in the kraft process.

             The chemical pulping process known as Kraft pulping,
         employs a cooking liquor whose main ingredients are  sodium
         sulfide and sodium hydroxide in solution.

             The spent cooking liquor is black from the lignin,
         waste fibers, and dissolved sulfide salts--hence  called
         "black liquor".  Vitally important to the economics  of  the
         kraft process is the recovery and recycle of inorganic
         chemicals in the black liquor.  To accomplish chemical
         recovery, black liquor is concentrated by evaporation and
         burned in recovery furnaces.  Most of the organic and in-
         organic sulfur is reduced in the lower oxygen-poor region
         of the furnace to form an ash or smelt of molten  chemicals,
         primarily sodium sulfide and carbonate.

             Recovery furnaces also produce valuable process steam
         from the heat of the burning black liquor.  Hot gases from
         the combustion zone relinquish most of their heat energy in
         passing over boiler tubes and heat economizers.   Steam  may
         be used elsewhere in the pulp-making process or sent to
         turbines for electrical power generation.  Additional
         utilization of the furnace heat is accomplished by the  use
         of a direct contact evaporator.  Such an evaporator  util-
         izes the heat of the flue gases to further evaporate black
         liquor just prior to its firing in the recovery furnace.
                                    127


A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
         Direct  contact evaporation has one  serious  drawback,
         however,  from an  air-pollution standpoint--the  stripping
         of hydrogen  sulfide which occurs when  acidic  flue  gases
         contact the  black liquor.  Following the  direct contact
         evaporator,  furnace gases pass through  collectors  (such as
         electrostatic precipitators  and scrubbers)  to remove  par-
         ticulate  matter consisting of sodium salts  and  carbon
         particles.   Exhaust gases containing the  remaining par-
         ticulates, plus the malodorous sulfur  compounds, then pass
         to the  atmosphere.

             When the black liquor is burned in the furnace,  an
         appreciable  quantity  of particulate is  liberated.   Chemical
         content of the solids entrained in  the  recovery furnace
         flue gases is a function of  furnace operating conditions
         and feed  liquor composition.  Usually Na2SO., Na-CO..,  and
         NaCl in the  fume  can  vary greatly from  mill to  mill.   For
         example,  NaCl is  present in  measurable  quantities  when logs
         have been stored  in salt water before  chipping;  inland mills
         have less NaCl passing through the  recovery system.   The
         SRI Manual of Electrostatic  Precipitator  Technology (1972) •
         presents  size distribution data obtained  from electron
         microscope pictures.  The mass median particle  diameter
         is 1.94 urn and the geometric standard  deviation is 2.06.

             All  the particles emitted from the recovery furnace
         are soluble  and do not require supersaturation  for growth.
         They will even grow when the saturation ratio is less than
         1.0 (=80% of the  particles are Na^SO. which will start grow-
         ing at  a  relative humidity of =901).

             To use  a condensation scrubber advantageously,  the size
         of the waste heat boiler is  increased  and the exit tempera-
         ture of the  flue  gases is reduced to =120°C.  The  gases then
         flow through a waste  heat recovery  venturi  evaporator and
         into the  condensation scrubber system  as  shown  in  Figure  27.
         A surface condenser (heat exchanger) was  selected  to  con-
         dense most of the water vapor from  the  gas  stream  leaving
         the cyclone  separator following the venturi evaporator.  The
         purpose of this condenser is to isolate the condensate liquid
         which contains dissolved salts so that  it can be recycled to
         the process  without requiring concentration and so that the
         cooling water can be  kept clean.  An alternative approach
         would be  to  use a direct contact condenser  and  circulate  the
         condenser coolant liquid through the tubes  of an evaporative
         type cooler.
                                    128



A. P. T. InC.                                POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
                                                              28.2 m3/min

                                                                Cool ing

                                                                 Water
TJ

O
CO


O
-n
Tl

O
X
-j

Reco ve ry
Furnace
Gas
Stream Am3
1
^-
i



Venturi
k
1 4
fc

^>s
Cyclone
^

Concentrated V
Liquid J
/mm T °C
1 14,000 121
2 12,700 74
3 9,900 60
4 9 , 800 60


Condenser

^
J
-^

-*•
1
Spray
Scrubber
3
r
+

Sump
1
r

-O

4 To


                                                        Re cycle

                                                         Liquid
                                       Recycle

                                        Liquid
m
3)
CA

O
m

O
Figure 27 - Flowsheet for FF/C scrubber on

            Kraft  liquor recovery  furnace.
to
ro
in
o

-------
             Spray scrubber liquid would recirculate at nearly
        constant temperature and with makeup liquid flow  rate
        being equal to  the rate of blow-down liquid flow  back to
        the process.  Particle collection efficiency in the  spray
        scrubber would  be 95% for the design shown; in order to
        be comparable to one of the alternatives given by Hendrick
        son et al. (see below).  Because the major collection mech-
        anism active in the scrubber would be  inertial impaction
        of the grown particles, the scrubber penetration  could be
        decreased by increasing liquid rate or contact length, or
        by appropriately changing other factors.  It would be
        possible to decrease penetration from  5.01 to 0.25%  by
        doubling scrubber length, for example.

             Much information regarding emission control  in  the
        wood pulping industry is found in the  report "Control of
        Atmospheric Emissions in the Wood Pulping Industry",
        E. R. Hendrickson et al., (1970).  In  that study, they
        presented the economics for several processes for up-
        grading the performance of existing Kraft furnace air
        pollution control systems.  Costs for  two of these
        processes will  be compared with costs  for the FF/C scrub-
        bing system described above.  Inlet gas flow rates,
        temperatures, particulate loadings and other pulping
        process conditions are the same for the FF/C system  and
        the two alternatives listed below:

             1.  Increase the collection efficiency of an
                 existing system by converting an existing
                 venturi scrubber to a low efficiency
                 (about 50%) venturi/cyclone evaporator
                 and adding a 99% efficiency electrostatic
                 precipitator.

             2.  Make the same change as in No. 1, except
                 add a  971 efficiency second stage venturi
                 rather than the electrostatic precipitator.

             The FF/C system is also based on  the change  to  a
        venturi/cyclone evaporator.  Costs for the FF/C system
        were computed to be as follows:
                                    130


A. P. T. Inc.                                POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
             Capital Investment
                Condenser (7,500 m2 area)              $400,000
                Spray Scrubber (4m high, concrete
                 lined)                                  160,000
                Pumps (4x7.5 m'/min)                     40,000
                Stack                                    11,600
                 (piping costs are included above)     	
                            Total                      $611,000

             Annual Costs
                Capital  cost and depreciation          $121,500
                Cooling  water @ 1.25
-------
             It can be seen that within the accuracy of  this  kind
        of estimate the venturi and the FF/C add-on systems  are
        about even.  If high efficiency were required, however,  the
        FF/C system could decidedly surpass either of the  other
        two systems because the major additional expense required
        would be for another scrubber increment.  Thus the addition-
        al capital investment would be about $160,000 and  the
        additional annual operating cost would be perhaps  $40,000.
                                     132


A. P. T. InC.                                 POST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
                       FUTURE RESEARCH RECOMMENDATION
             The primary objective of investigating the feasibility
        of FF/C scrubbing has been achieved with a clearly affirma-
        tive result.  It has been shown in the present program that
        FF/C scrubbing can remove fine particles at high efficiency
        and that, within some limitations, it is more economical
        than other means of particle collection.  Mathematical
        modeling of the many simultaneous phenomena taking place
        in an FF/C scrubber has been accomplished for some important
        unit mechanisms, although several coefficients remain to be
        fitted to experimental data.

             In accordance with the original objective of selecting
        and performing a brief exploratory experiment of significant
        nature, our experimental work extended over a narrow range
        of conditions.  Within this range the agreement between
        theoretical predictions and experimental results for parti-
        cle collection is fairly good, once the heat and mass
        transfer coefficients are evaluated from experimental data.
        However, despite the good agreement, theoretical predictions
        of collection efficiency for plates are consistently higher
        than the experimental results.  This, along with the dis-
        crepancy between computed and predicted liquid phase heat
        transfer coefficients, indicates that the mathematical
        model for bubbles should be revised.

             To achieve the objective of building a working FF/C
        scrubber, development work will be required in three main
        areas:

             1.  Complete the experimental evaluations and
                 development of the theoretical models for
                 bubbles, sheets, and drops.

             2.  Define and develop an optimal FF/C scrubber,
                 based on the improved design methods.

             3.  Build a pilot plant scale unit of the proposed
                 optimum FF/C scrubber.  Test the pilot unit in
                 the laboratory and under actual operating
                 conditions in the field.

             The rest of this section will be devoted to a detailed
        discussion of these three topics.
                                   133


A. P. T. InC.                                POST OFFICE BOX 71,  RIVERSIDE, CA. 92502

-------
        EVALUATION  OF THEORETICAL  MODELS

             Our present model  for hubbies  (i.e.,  plate type
        devices), when compared to experimental results, presents
        two major questions:

             A.   Why is the computed liquid phase  heat
                 transfer coefficient so low in comparison
                 to the predicted  coefficient?

             B.   Why is the predicted particle  penetration
                 higher than the experimental?

             Possible explanations to these questions may be:

             1.   Incorrect estimate of the  heat and mass
                 transfer area.

             2.   Low estimate of liquid surface temperature,
                 if co-current  flow of liquid with rising
                 bubbles is significant.

             3.   Unusual sieve  plate design and operating
                 conditions have caused a significant
                 departure from customary behavior.

             4,   Different bubble  shape and gas dynamics
                 cause different transfer rates and particle
                 deposition by  centrifugal  force.

             5.   Measurement problems may introduce errors.

             In  order to resolve the questions raised up to this
        point, it will be necessary to perform experiments which
        provide  the following features:

             1.   Particles used for tests should have a range of
                 surface properties and high enough vapor pressure
                 that gas temperature of 80°C will cause negligible
                 vaporization.

             2.   Wettable particles should be used in some runs
                 so that the effect of growth can be clearly
                 identified.  These runs should be at relatively
                 low saturations,  such as were used for D.B.P.
                 aerosol in the runs reported here.
                                    134


A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
             3.   Large water vapor concentration gradients
                 should be used under conditions such that
                 saturation is low enough to avoid condensation
                 on particles.  This could be achieved through
                 the use of high gas and liquid temperatures
                 with non-wettable particles.

             4.   Sieve plates with more customary dimensions
                 as for mass transfer, should be used and
                 operated at usual gas velocities.

             5.   Packed columns should be used to validate the
                 "liquid sheets" model.  The more defineable
                 transfer surface area and liquid renewal time
                 for packings as compared to plates will enable
                 a less equivocal interpretation of particle
                 penetration and transport rate data.

             We consider this extension of the work on bubbles and
        sheets important since it will test the assumptions made
        in deriving the theoretical model.  Once they are either
        proven or modified so as to be valid, the concepts from
        these models can be used to derive unit mechanism equations
        for other geometries.

        DEFINE AND DEVELOP AN OPTIMAL FF/C SCRUBBER

             To design an efficient and economical FF/C scrubber
        one should attempt to optimize the geometric configurations,
        the method by which the desired saturation ratio is achieved
        and the combination of collectors to achieve various stages
        of particle separation.

             In the work done to date, various unit mechanism equa-
        tions were derived.  The next step is to integrate these
        equations so that they describe a scrubber.  For example,
        the unit mechanism equation for the collection of particles
        by drops can be integrated into a venturi type design, a
        spray column, an ejector venturi, part of a wet cyclone, an
        impingement and entrainment scrubber or a moving bed scrub-
        ber.  Collection by sheets can be integrated into packed
        bed scrubbers, baffle and secondary flow scrubbers or
        similar designs.  Similarly, the bubble model is the basis
        for various plate type devices.  Devices which combine
        several unit mechanisms can also be explored through mathe-
        matical modeling.  Only after this is done can we say which
                                    135


A. P. T. InC.                                POST OFFICE BOX 71.  RIVERSIDE. CA. 92502

-------
        device best utilizes the FF/C effects and is most efficient
        and economical.

             It follows  from our present models that saturation and
        particle growth  are very important.  The economics of the
        scrubber may depend in the final analysis, on our ability
        to obtain the required saturation ratio in a large enough
        fraction of the  aerosol, to permit nucleation and growth
        of almost all the particles.

             Several ways to achieve the required degree of satura-
        tion have been described previously.  All possible ways
        should be checked for each device until the method most
        suitable is found.  It is quite probable that no single
        method would be  best in all cases and different methods will
        better suit different aerosols.  Approaches such as the
        alteration of particle wettability by means of a pre-
        treatment in which there is some absorption of a surface
        conditioning should be explored.

             Once a scrubber geometry most suitable for particle
        collection and the formation of the desired saturation
        ratio has been defined, it becomes a straight-forward
        matter to build  the scrubber.  This scrubber should then
        be tested in the laboratory under controlled conditions in
        order to determine whether theoretical predictions and
        experimental findings agree.  It could be expected that
        during these tests further modifications will make the
        scrubber a more  efficient one.

        FIELD TESTS OF PILOT SCALE FF/C SCRUBBER

             The efficiency and economics of FF/C scrubbers are
        much more sensitive to the properties of the dust being
        removed and to the presence of small condensation nuclei
        than most other  scrubbers.  For this reason field tests on
        various dusts emanating from different sources are necessary,
        A good way to obtain these results is to build a mobile FF/C
        scrubber and test it on various industrial sources.  These
        field tests will give us valuable information on the be-
        havior and economics of FF/C scrubbers when the dust is
        neither uniform nor monodisperse.  Further, the effects of
        trace components, process variations, and other practical
        complications will become known.

             An important adjunct of the field tests should be the
        development and proving-out of analytical methods for the
                                    136


A.P. T. InC.                                POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
        definition of all significant properties of  the  source.
        For example, it  is of special importance that we  be  able
        to measure the particle number concentration, the critical
        saturation ratio, and the number of condensation  nuclei;
        in addition to the usual size distribution,  etc.   The
        ultimate aim should be to eliminate the need for  pilot  test
        ing in order to  determine FF/C scrubbing feasibility.

             Another round of process design and economic evalua-
        tion should be performed after the pilot study information
        has been obtained and assimilated.  The accuracy  of  cost
        estimates is highly dependent on the amount  of process
        detail available and o'n the specificity of equipment defi-
        nition.  Items such as cooling towers, water treatment,
        and large (and cheap) heat exchangers are very significant
        in the FF/C process economics and should be  studied  in
        detail.

             It should be noted that the program reported here  did
        not stop with the termination of the contract period of
        performance.  Additional work to answer some of the  ques-
        tions discussed here has been funded under another contract,
                                    137

A. P. T. InC.                                POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
                              REFERENCES
            Abraham, F. F.   A Re-Examination of Homogeneous
            Nucleation Theory: Thermodynamic Aspects. J Atmos-
            pheric Sci. 25:47-53, 1968

            Air Pollution Manual. Part 2. American Industrial
            Hygiene Association. 1968.

            Amelin, A. G.  Theory of Fog Condensation. Second
            Edition. Jerusalem, Israel Program for Scientific
            Translations, 1967.  235 p.

            Barnes, T. M., and H. W. Lownie Jr.  A Cost Analysis
            of Air Pollution Controls in the Integrated Iron and
            Steel Industry. Battelle Memorial Institute. Columbus,
            Ohio. Contract No. PH-22-68-65. May 1969. 259 p.

            Brock, J. R.  On the Theory of Thermal Forces Acting
            on Aerosol Particles. J Coll Sci. 17:768-780, 1962.

            Brown, G. G. and Associates. Unit Operations. New York,
            John Wiley $ Sons, Inc., 1950.

            Calvert, S., J. Goldshmid, D. Leith, and D. Mehta.
            Scrubber Handbook. A.P.T., Inc. Riverside, California.
            EPA Contract No. CPA-70-95. August 1972.

            Chilton, C. H. Cost Engineering in the Process
            Industries. New York, McGraw-Hill, 1960. 475 p.

            Davies, J. T.  Turbulence Phenomena. New York and
            London, Academic Press, 1972. 412 p.

            Davis, R. J., and J. Truitt. Using Condensing Steam for
            Air Cleaning. Instruments and Control Systems, p 68-70,
            November 1972.

            Demshin, V. Ya., and G. V. Ermakov. Investigation of
            the Performance of Dust Removal Apparatus of the Foam
            Type with Preliminary Condensation of Water Vapor on
            the Dust Particles. Zhurnal Prikladno: Khimii. 38(3):
            691-693, March 1965.

            Derjaguin, B. V., and Y. I. Yalamov. Theory of Thermo-
            phoresis of Large Aerosol Particles. J Coll Sci. 20:
            555-570, 1965.           '
                                    138

A. P. T. InC                                POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
             Derjaguin, B.V., Y. I. Yalamov.  Theory of Thermo-
             phoresis of Large Aerosol Particles. J Coll Interface
             Sci. 22:117-225, 1966.

             Derjaguin, B.V., A. I. Storozhilova, and Y, I. Rabino-
             vich.  Experimental Verification of the Theory of
             Thermophoresis of Aerosol Particles. J Coll Interface
             Sci. 21:35-58, 1966.

             Derjaguin, B.V., and S. P. Bakanov. Dokl Akad Nauk
             SSSP (Phys Chem, USSR). 147:139, 1967.

             Epstein, P.S.  Zur Theorie des Radiometers. Z Phys
             54:537, 1929.

             Fletcher, N. H.  Size Effect in Heterogeneous Nuclea-
             tion. J Chem Phys. 29:572-576, 1958.

             Fletcher, N. H.  The Physics of Rain Clouds. London,
             Cambridge University Press, 1962. 389p.

             Foust, A. S., L. A. Wenzel, C. W. Clump, L. Mans, and
             L. B. Andersen.  Principles of Unit Operations.
             New York, John Wiley § Sons, Inc., 1960. 578p.

             Fuchs, N. A.  Evaporation and Droplet Growth in Gaseous
             Media. New York, Pergamon Press, 1959. 72p.

             Fuchs, N. A.  The Mechanics of Aerosols. New York,
             Pergamon Press, 1964. 408p.

             Goldsmith, P., and F. G. May. Diffusiophoresis and
             Thermophoresis in Water Vapor Systems.  In:  Aerosol
             Science, Davies, C.N. (ed.). New York, Academic Press,
             1966. p.163-194.

             Hales, J. M., T. W. Horst, and L. C. Schwendiman.
             Aerosol Transport in a Condensing-Steam Boundary Layer.
             Battelle Northwest, Richland, Washington. Publication
             Number BNWL-1125. June 1970. 63p.

             Hales, J. M., and L. C. Schwendiman.  Phoretic Trans-
             port of Aerosols in Laminar Boundary Layers. Battelle
             Northwest, Richland, Washington.  Publication Number
             BNWL-SA-3734.  February 1971.  14p.
                                    139


A. P. T. InC.                                POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
              Hales,  J.  M.,  L.  C.  Schweridiman,  and  T.  W.  Horst.
              Battelle Northwest,  Richland,  Washington.   Publication
              Number  BNWL-SA-3592.   1971.

              Handbook of Emissions,  Effluents  and  Control  Practices
              for  Stationary Particulate Pollution  Source.  Midwest
              Research Institute.  Report to  NAPCA,  Contract No.  CPA-
              22-69-104.   1970.

              Hendrickson, E. R.,  J.  E. Robertson,  and J. B.  Koogler.
              Control of  Atmospheric  Emissions  in the  Wood  Pulp
              Industry.  Environmental Engineering,  Inc. Gainesville,
              Florida.   Contract No.  PHS-CPA-22-69-18. March  1970.
              269  p.

              Henschen,  H. C.   Wet Vs. Dry Gas  Cleaning in  the Steel
              Industry.  Air  Poll Control Assoc  J 18:338-342,  1968.

              Herne,  H.   The  Classical Computation  of  the Aerodynamic
              Capture of  Particles by Spheres.  Intern  J of  Air and
              Water Poll.  3:26-34, 1960.

              Hidy, G. M., and  J.  R.  Brock.  The Dynamics of  Aero-
              colloidal  Systems. New  York, Pergamon Press,  1970,
              379  p.

              Horst,  T. W. A  Review of Particle Transport in  a
              Condensing  Steam  Environment.  Battelle  Northwest,
              Richland, Washington.   Publication Number BNWL-848.
              June 1968.  45p.

              Howell, W.  E.   The Growth of Cloud Drops in Uniformly
              Cooled  Air.  J  of  Met.,  6:134,  1949.

              Junge,  D.   Die  Konstitution des Atmospharischen
              Aerosols.   Ann  Met (Beiheft).  1952.

              Kinzer1, G.  D.,  and R. Gunn.  The Evaporation  Tempera-
              ture and Thermal  Relaxation Time  of Free Falling Water
              Drops.  J of Met. 8:71-82, 1951.

              Klauss, P.  R.,  P. L. Sieffert, and J. F. Skelly.   Costs
              and  Performance of Control Systems and Control  Equip-
              ment.   Swindell-Dressier Company.  Appendix C for A Cost
              Analysis of Air Pollution Controls in the Integrated
              Iron and Steel  Industry.  Battelle Memorial Institute
              Columbus, Ohio. Contract No. PH-22-68-65. May 1969. 122p.


                                    140

A. F. I.  InC                                pOST OFFICE BOX 71.  RIVERSIDE, CA. 92502

-------
              Lancaster,  B.  W.,  and W.  Strauss.   A Study of Steam
              Injection Into Wet Scrubbers.  Ind  Eng Chem Fundamentals.
              10(3):362-369, March 1971.

              Langmuir, J.,  and  K. Blodgett.   A  Mathematical Investi-
              gation of Water Droplet Trajectories.  Amer.  A.F.  Tech.
              Report 5418.   1946.

              Lapple,  C.  W., and H.  J.  Kamack.   Performance of Wet
              Dust Scrubbers.  Chem Eng Prog.  51(3) :110-121,
              March 1955.

              Litvinov, A. T.  Vestnik  tekhnicheskoy  i  ekonomicheskoy
              informatsii.   5:26-29,  1964.

              Litvinov, A. T.  Khimicheskayer  promyshlennost.   8:64-69.
              1964.

              Litvinov, A. T.  Stal1.   7:667-669,  1965.

              Litvinov, A. T.  Influence  of Condensation on the
              Effectiveness  of Capture  of Fine Particles During
              Cleaning of Gases  by the  Wet Method.  Zhurnal
              Prikladnoi  Khimii.   40(2)-.353-361,  February 1967.

              Litvinov, A. T.  Fine  Gas Scrubbing  to  Remove Highly
              Dispersed Hydrophobic  Particles  Using the  Condensa-
              tion Effect.   Stal1.   2:184-186, 1972.

              Mashita,  T. Industrial  Public Nuisance.   7:573,  1971.

              Mason, B. J.   The  Physics of Clouds.  Oxford  1957.

              Matsuzaki, K.  Japanese patent Number Sho  41-41184.
              1970.

              Maxwell,  J.  Collected  Scientific  Papers.   Cambridge
              1890.

              Mercer,  T. T., and H. Y. Chow.   Impaction  from
              Rectangular Jets.  J Coll Interface  Sci. 27:75-83, 1968.

              Control  Techniques for  Particulate Air  Pollutants. NAPCA
              Publication No.  AP-51, Washington, D.C. 1969.
                                    141


A. P. T. Inc.                                POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
             Norman, W. S.  Absorption, Distillation and Cooling
             Towers.  New York, John Wiley f, Sons, Inc., 1961. 477p.

             Nuzhnyi, V. M., Yu. I. Shimanskii, and G. K. Ivanitskii.
             Some Aspects of the Diffusion Theory of Evaporation of
             Drops of Volatile Liquids.  Coll. J. USSR. 27:494-498,
             1965.

             Oglesby, Jr. S., and G. B. Nichols.  A Manual of Electro-
             static Precipitator Technology.  Part II.  Southern
             Research Institute.  1970.

             Popper, H.   Modern Cost Engineering Techniques.
             New York, McGraw-Hill.  1970.

             Prokhorov, P. S., and L. F. Leonov.  Investigation of
             Long Range Diffusion Forces Between Water Droplets and
             Non-Volatile Particles.  Diss Faraday Soc.  30:124-129,
             1960.

             Ranz, IV. E., and J. B. Wong.  Impaction of Dust and
             Smoke Particles on Surface and Body Collectors.  Ind
             Eng Chem  44:1371-1380, 1952.

             Rozen, A. M., and V. M. Kostin.  Collection of Finely
             Dispersed Aerosols in Plate Columns by Condensation
             Enlargement.  Inter Chem Eng.  7:464-467, July 1967.

             Schadt, C. F., and R. D. Cadle.  Thermal Forces on
             Aerosol Particles.  J Phys Chem.  65:1689-1694, 1961.

             Schauer, P. J.  Removal of Submicron Aerosol Particles
             from Moving Gas Stream.  Tnd Eng Chem.  43(7):1532-1538 ,
             July 1951.

             Schlichting, H.  Boundary Layer Theory.  New York, McGraw-
             Hill, 1960.

             Schmitt, K. H.  Untersuchungen on Schwebstoffteilehen in
             Temperaturfeld.  Z. Naturforsch  14a:870-881, 1959.

             Schmitt, K. H., and L. Waldmann.  Untersuchungen on
             Schwebstoffteilehen im Diffundierenden Gasen.  Z.  Natur-
             forsch  15a:843-851, 1960.
                                    142


A. P. T. InC.                                POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
              Semrau,  K. T., C. W. Marynowski, K. E.  Lunde,  and
              C.  E.  Lapple.  Influence  of  Power  Input on  Efficiency
              of  Dust  Scrubber.   Ind Eng Chem.   50(11):1615-1620,
              November 1958.

              Sparks,  L. E., and  M, J.  Pilat.  Effect of  Diffusio-
              phoresis on Particle Collection by Wet  Scrubbersi
              Atmospheric Environment.  4:1-10, 1970.

              Taheri,  M., and S.  Calvert.  Removal of Small  Particles
              From Air by Foam in a Sieve-Plate Column.   J Air Poll
              Control  Assoc.  18:240-245,  1968.

              Terebenin, A. N., and A.  P.  Bykov. Zhurnal  Prikladnoi
              Khimii (USSR).  45:1012,  1972.

              Varga<, Jr. J., and  H. W.  Lownie, Jr.  A System Analysis
              of  the Integrated Iron and Steel Industry.  Report No.
              PB  184557.  1969.

              Volmer,  M.  Kinetik der Phasenbilding.   Dresden and
              Leipzig:  Steinkopff.  1939.

              Waldmann, L., and K. H. Schmitt.  Thermophoresis and
              Diffusiophoresis of Aerosols.  In:  Aerosol Science,
              Davies,  C. N. (ed.).  New York, Academic Press, 1966.
              p 137-161.

              Whitaker, S.  Forced Convection Heat Transfer Correlations
              for Flow  in Pipes,  Past Flat Plates, Single Cylinders,
              Single Spheres and  for Flow  in Packed Beds  and Tube
              Bundles.  AIChE J.  18:361-371, 1972.
                                    143


A. P. T. Inc.                                POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
                                           ,3
                   GLOSSARY



a   -  constant defined by equation (12)

a,   -  specific surface area of bubble, cm"1

A   -  area of plate, cm2

b   -  constant defined by equation (13)

B   -  dimensionless parameter defined by equation 4-38

c   -  mass concentration Kg/m3 or g/cm3

c.   -  molar concentration, gmol/cnr

CT   -  Cunningham correction factor, dimensionless

C   -  specific heat, kcal/kg-°C or cal/g-°C

C   -  temperature jump coefficient, = 2.3, dimensionless

C   -  isothermal slip coefficient, = 1.25, dimensionless

d   -  diameter, m or cm

D   -  diffusivity, cm2/sec

D   -  particle diffusivity for Brownian diffusion, cm2/sec

F   -  fraction of the total condensing vapor which condense
       on the drops.

F   -  f]ux force in "x" direction, dynes
 .A

F   -  foam density, ratio of clear liquid height to total
       foam height.

G   -  gas flow rate, gmol/cm2-sec

Gp   -  dimensionless quantity defined by equation (14),  (18)
       or (24)

h   -  heat transfer coefficient, cal/sec-cm2-°C

I   -  integral defined by equation 4-40

i   -  Van't Hoff factor
                                    144


A. P. T. InC.                                POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
          J   -   dimensionless  parameter defined by equation 4-45

          k.   -   thermal  conductivity of substance A,  cal/sec-cm-°C

          k1   -   mass  transfer  coefficient,  gmol/cm2-sec-atm

          KD   -   diffusiophoresis  constant defined by  equation (94)

          Kp   -   overall  gas  phase transfer  coefficient gmol/sec-atm-cm2

          K   -   inertia  parameter defined by  equation (24)

          K   -   particle flux  coefficient for Brownian diffusion,
           p      cm/sec

          KT   -   thermophoresis  constant defined by equation (99)

          L   -   liquid mass  flowrate,  kg/sec  or g/sec

          L,.   -   latent heat  of  vaporization cal/g

          m   -   mass, kg or  g

          m   -   moles of solute in a drop,  gmol/drop

          M   -   molecular weight, kg/kgmol  or g/gmol

          n   -   particle number concentration,  no./cm3

          N   -   mass  flux, kg/m2-sec or g/cm2-sec

          N   -   particle flux,  no./cm2-sec  or g/cm2-sec

          N B  -   particle flux due to Brownian diffusion,  no./cmz-sec

          p   -   partial  pressure  of solute, mbar or atm

          p M  -   mean  partial pressure  of non-transferring gas,  atm

          Pt   -   penetration  (one  minus  efficiency), fraction  or
                 percent

          Ptc, Ptn,  and PtT are  defined  in equations (42), (43)  and (44)

          q   -   mass  vapor condensed per mass particles,  g/g

          Q   -   volumetric flow rate, mVsec, cm3/sec or  JL/sec



                                     145


A. F. T. InC.                                 POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
         Q    -   heat  transferred  per  unit  cross-section area of

                column,  cal/cm2



         r    -   radius,  cm  or  ym



         r'   -   dimensionless  particle  radius  Ln  equation



         R    -   ideal  gas law  constant



         R    -   effective radius  of curvature
         c


         s    -   distance from  jet nozzle  to  plate,  cm



         S    -   saturation  ratio,  atm/atm



         t    -   time,  sec



         T    -   temperature, °K  (or °C, where  specified)



         T'   -   dimensionless  temperature  in equation (89)



         u    -   velocity, m/sec or cm/sec



         v    -   velocity, cm/sec



         v_   -   liquid flow rate  per  unit  width,  cm3/sec-cm



         V    -   volume, m3  or  cm3



         W    -   jet width,  cm



         W    -   weight gain on filter



         x    -   distance in "x" direction, cm



         X    -   dimensionless  distance  in  "x"  direction



         y    -   mole  fraction, gmol/gmol



         Y    -   distance in "Y" direction



         z    -   height of foam, cm



         Z    -   distance in "Z" direction



         Z1   -   thermophoretic parameter
                                    146






A. P. T. InC                                POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
         Dimensionless Numbers
         N      UF
          FD =  — ,  Flux deposition number

                 o



                h'd

         N^,  =   ,  y ,  Nusselt number
         Nc   =  1V/ Po D_,  Schmidt number
          oC     U
                     u
         Npr =   C  p/k,  Prandtl number




         Np  =   du/D,  Peclet number




         NRe =   pdu/y,  Reynolds number




         NKn E   */r >  Knudsen number







         Greek



         a    -   contact  angle of a drop on a solid surface, degrees



         a    -   fraction of  particle mass  increase as predicted

                theoretically,  used in equation (40)



         6    -   dimensionless jet spacing  = 2 s/w



         6    -   boundary layer  thickness,  cm or ym



         e    -   volume fraction voids  in scrubber



         0    -   contact  time, sec



         n    -   efficiency due  to a unit mechanism,  fraction or

                percent



         A    -   mean free path  of gas  molecules,  cm  or \im



         p    -   density, kg/m3  or g/cm3








                                     147




A. P. T. Inc.                                POST OFFICE BOX 71,  RIVERSIDE. CA. 92502

-------
        a    -   geometric standard deviation
         &

        o    -   diffusion slip factor in equation  6,  dimensionless


        T    -   relaxation time, sec


        p    -   viscosity, g/cm-sec


        v    -   kinematic viscosity = y/p, cm2/sec
Subscripts
b
BD -
C
d
D
d
e
exp -
F
F
F
G
i
i
i
L
m
M
P
bubble
Brownian diffusion
centrifugal
drop
diffusiophoretic
difference
exit
experimental
foam
impaction during bubble formation
flux force
gas phase
interface
in
increment
liquid phase
mixture
molal quantity
particle
                                     148


A. P. T. InC.                                 POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
        pa   -   aerodynamic



        pB   -   particle,  Brownian  diffusion



        pD   -   particle,  diffusiophoresis



        pn   -   particle,  number mean



        pT   -   particle,  thermophoresis



        pi   -   particle,  inertial



        s    -   surface



        s    -   sum  of



        t    -   terminal velocity



        tg   -   tangential



        theo  -  theoretical



        T    -   turbulent  layer



        T    -   thermophoretic



        v    -   vapor



        x    -   in the "x" direction



        y    -   in the "y" direction



        o    -   original condition



        oo    -   for  plane  surface





        Superscripts



        *    -   in equilibrium



             -   film



                dimensionless quantity




        —    -   overall or average









                                     149






A. P. T. InC.                                 POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
  BIBLIOGRAPHIC DATA
  SHEET
1. Report No.
 EPA-650/2-73-036
3. Recipient's Accession No.
 4. I ulc and Subtitle
 Feasibility of Flux Force/Condensation Scrubbing for Fine
      Particle Collection
                                            5' Report Date
                                               October 197-3
                                            6.
 7. Amhor(s)
 Seymour Calvert. Jhuda Goldshmid, David Leith. Nikhil Jhave
                                            &• Performing Organization Kept.
                                              No.
   Performing Organization Nnmc.' find A72

-------